diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,9134 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 650.0,
+  "eval_steps": 500,
+  "global_step": 1300,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.927849292755127,
+      "learning_rate": 0.0009999985400000594,
+      "loss": 3.6008,
+      "step": 1
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.966495990753174,
+      "learning_rate": 0.0009999941600087644,
+      "loss": 3.9178,
+      "step": 2
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 10.260856628417969,
+      "learning_rate": 0.0009999868600516937,
+      "loss": 3.7168,
+      "step": 3
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 26.86942481994629,
+      "learning_rate": 0.0009999766401714793,
+      "loss": 3.6664,
+      "step": 4
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 8.516195297241211,
+      "learning_rate": 0.0009999635004278053,
+      "loss": 3.5704,
+      "step": 5
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 1.8732255697250366,
+      "learning_rate": 0.000999947440897408,
+      "loss": 3.2983,
+      "step": 6
+    },
+    {
+      "epoch": 3.5,
+      "grad_norm": 2.523327350616455,
+      "learning_rate": 0.0009999284616740744,
+      "loss": 3.42,
+      "step": 7
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 1.8988794088363647,
+      "learning_rate": 0.0009999065628686437,
+      "loss": 3.3475,
+      "step": 8
+    },
+    {
+      "epoch": 4.5,
+      "grad_norm": 3.8478260040283203,
+      "learning_rate": 0.0009998817446090049,
+      "loss": 3.3901,
+      "step": 9
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 1.445601224899292,
+      "learning_rate": 0.0009998540070400965,
+      "loss": 3.1986,
+      "step": 10
+    },
+    {
+      "epoch": 5.5,
+      "grad_norm": 1.4163767099380493,
+      "learning_rate": 0.000999823350323906,
+      "loss": 3.2046,
+      "step": 11
+    },
+    {
+      "epoch": 6.0,
+      "grad_norm": 3.229616403579712,
+      "learning_rate": 0.0009997897746394685,
+      "loss": 3.2965,
+      "step": 12
+    },
+    {
+      "epoch": 6.5,
+      "grad_norm": 5.967291355133057,
+      "learning_rate": 0.0009997532801828658,
+      "loss": 3.2129,
+      "step": 13
+    },
+    {
+      "epoch": 7.0,
+      "grad_norm": 7.281788349151611,
+      "learning_rate": 0.000999713867167226,
+      "loss": 3.2,
+      "step": 14
+    },
+    {
+      "epoch": 7.5,
+      "grad_norm": 2.5564589500427246,
+      "learning_rate": 0.0009996715358227205,
+      "loss": 3.2944,
+      "step": 15
+    },
+    {
+      "epoch": 8.0,
+      "grad_norm": 4.69608211517334,
+      "learning_rate": 0.000999626286396565,
+      "loss": 2.9153,
+      "step": 16
+    },
+    {
+      "epoch": 8.5,
+      "grad_norm": 1.745740532875061,
+      "learning_rate": 0.0009995781191530157,
+      "loss": 3.1243,
+      "step": 17
+    },
+    {
+      "epoch": 9.0,
+      "grad_norm": 4.76589298248291,
+      "learning_rate": 0.0009995270343733697,
+      "loss": 3.2669,
+      "step": 18
+    },
+    {
+      "epoch": 9.5,
+      "grad_norm": 3.99920916557312,
+      "learning_rate": 0.0009994730323559615,
+      "loss": 3.2176,
+      "step": 19
+    },
+    {
+      "epoch": 10.0,
+      "grad_norm": 4.058233261108398,
+      "learning_rate": 0.0009994161134161633,
+      "loss": 3.366,
+      "step": 20
+    },
+    {
+      "epoch": 10.5,
+      "grad_norm": 4.022902965545654,
+      "learning_rate": 0.0009993562778863815,
+      "loss": 3.07,
+      "step": 21
+    },
+    {
+      "epoch": 11.0,
+      "grad_norm": 2.851071834564209,
+      "learning_rate": 0.000999293526116056,
+      "loss": 3.0395,
+      "step": 22
+    },
+    {
+      "epoch": 11.5,
+      "grad_norm": 2.5752713680267334,
+      "learning_rate": 0.0009992278584716563,
+      "loss": 2.9745,
+      "step": 23
+    },
+    {
+      "epoch": 12.0,
+      "grad_norm": 3.3147332668304443,
+      "learning_rate": 0.000999159275336682,
+      "loss": 3.2549,
+      "step": 24
+    },
+    {
+      "epoch": 12.5,
+      "grad_norm": 1.0202107429504395,
+      "learning_rate": 0.0009990877771116587,
+      "loss": 2.8853,
+      "step": 25
+    },
+    {
+      "epoch": 13.0,
+      "grad_norm": 2.582786798477173,
+      "learning_rate": 0.0009990133642141358,
+      "loss": 3.1888,
+      "step": 26
+    },
+    {
+      "epoch": 13.5,
+      "grad_norm": 1.762489676475525,
+      "learning_rate": 0.0009989360370786847,
+      "loss": 2.9965,
+      "step": 27
+    },
+    {
+      "epoch": 14.0,
+      "grad_norm": 1.6308578252792358,
+      "learning_rate": 0.0009988557961568955,
+      "loss": 2.9122,
+      "step": 28
+    },
+    {
+      "epoch": 14.5,
+      "grad_norm": 1.3620671033859253,
+      "learning_rate": 0.0009987726419173756,
+      "loss": 2.9703,
+      "step": 29
+    },
+    {
+      "epoch": 15.0,
+      "grad_norm": 1.8282300233840942,
+      "learning_rate": 0.0009986865748457456,
+      "loss": 2.8727,
+      "step": 30
+    },
+    {
+      "epoch": 15.5,
+      "grad_norm": 2.435718536376953,
+      "learning_rate": 0.0009985975954446371,
+      "loss": 2.8886,
+      "step": 31
+    },
+    {
+      "epoch": 16.0,
+      "grad_norm": 2.0487797260284424,
+      "learning_rate": 0.0009985057042336898,
+      "loss": 2.9508,
+      "step": 32
+    },
+    {
+      "epoch": 16.5,
+      "grad_norm": 1.3014992475509644,
+      "learning_rate": 0.0009984109017495484,
+      "loss": 2.855,
+      "step": 33
+    },
+    {
+      "epoch": 17.0,
+      "grad_norm": 2.105820894241333,
+      "learning_rate": 0.0009983131885458597,
+      "loss": 2.8369,
+      "step": 34
+    },
+    {
+      "epoch": 17.5,
+      "grad_norm": 1.8218109607696533,
+      "learning_rate": 0.0009982125651932681,
+      "loss": 2.8211,
+      "step": 35
+    },
+    {
+      "epoch": 18.0,
+      "grad_norm": 7.161676406860352,
+      "learning_rate": 0.0009981090322794144,
+      "loss": 2.653,
+      "step": 36
+    },
+    {
+      "epoch": 18.5,
+      "grad_norm": 1.682011365890503,
+      "learning_rate": 0.0009980025904089307,
+      "loss": 2.8251,
+      "step": 37
+    },
+    {
+      "epoch": 19.0,
+      "grad_norm": 2.0136234760284424,
+      "learning_rate": 0.0009978932402034376,
+      "loss": 2.6821,
+      "step": 38
+    },
+    {
+      "epoch": 19.5,
+      "grad_norm": 1.3691494464874268,
+      "learning_rate": 0.00099778098230154,
+      "loss": 2.713,
+      "step": 39
+    },
+    {
+      "epoch": 20.0,
+      "grad_norm": 5.579954624176025,
+      "learning_rate": 0.0009976658173588243,
+      "loss": 3.0606,
+      "step": 40
+    },
+    {
+      "epoch": 20.5,
+      "grad_norm": 1.5473341941833496,
+      "learning_rate": 0.0009975477460478538,
+      "loss": 2.7286,
+      "step": 41
+    },
+    {
+      "epoch": 21.0,
+      "grad_norm": 5.342884540557861,
+      "learning_rate": 0.0009974267690581644,
+      "loss": 2.9215,
+      "step": 42
+    },
+    {
+      "epoch": 21.5,
+      "grad_norm": 2.1975221633911133,
+      "learning_rate": 0.0009973028870962624,
+      "loss": 2.8504,
+      "step": 43
+    },
+    {
+      "epoch": 22.0,
+      "grad_norm": 1.9156906604766846,
+      "learning_rate": 0.000997176100885618,
+      "loss": 2.6226,
+      "step": 44
+    },
+    {
+      "epoch": 22.5,
+      "grad_norm": 2.1156492233276367,
+      "learning_rate": 0.0009970464111666627,
+      "loss": 2.7287,
+      "step": 45
+    },
+    {
+      "epoch": 23.0,
+      "grad_norm": 4.055300235748291,
+      "learning_rate": 0.0009969138186967843,
+      "loss": 2.8322,
+      "step": 46
+    },
+    {
+      "epoch": 23.5,
+      "grad_norm": 3.527277946472168,
+      "learning_rate": 0.0009967783242503232,
+      "loss": 2.6762,
+      "step": 47
+    },
+    {
+      "epoch": 24.0,
+      "grad_norm": 4.392818927764893,
+      "learning_rate": 0.0009966399286185665,
+      "loss": 2.8783,
+      "step": 48
+    },
+    {
+      "epoch": 24.5,
+      "grad_norm": 2.0216639041900635,
+      "learning_rate": 0.000996498632609745,
+      "loss": 2.7311,
+      "step": 49
+    },
+    {
+      "epoch": 25.0,
+      "grad_norm": 4.521383285522461,
+      "learning_rate": 0.000996354437049027,
+      "loss": 2.4351,
+      "step": 50
+    },
+    {
+      "epoch": 25.5,
+      "grad_norm": 2.5309815406799316,
+      "learning_rate": 0.0009962073427785148,
+      "loss": 2.5086,
+      "step": 51
+    },
+    {
+      "epoch": 26.0,
+      "grad_norm": 7.775767803192139,
+      "learning_rate": 0.000996057350657239,
+      "loss": 3.1966,
+      "step": 52
+    },
+    {
+      "epoch": 26.5,
+      "grad_norm": 3.7811288833618164,
+      "learning_rate": 0.0009959044615611532,
+      "loss": 2.64,
+      "step": 53
+    },
+    {
+      "epoch": 27.0,
+      "grad_norm": 2.6169111728668213,
+      "learning_rate": 0.00099574867638313,
+      "loss": 2.5458,
+      "step": 54
+    },
+    {
+      "epoch": 27.5,
+      "grad_norm": 1.930254578590393,
+      "learning_rate": 0.0009955899960329545,
+      "loss": 2.6784,
+      "step": 55
+    },
+    {
+      "epoch": 28.0,
+      "grad_norm": 3.1318228244781494,
+      "learning_rate": 0.0009954284214373204,
+      "loss": 2.524,
+      "step": 56
+    },
+    {
+      "epoch": 28.5,
+      "grad_norm": 2.571221351623535,
+      "learning_rate": 0.0009952639535398228,
+      "loss": 2.5219,
+      "step": 57
+    },
+    {
+      "epoch": 29.0,
+      "grad_norm": 2.3045191764831543,
+      "learning_rate": 0.0009950965933009544,
+      "loss": 2.6272,
+      "step": 58
+    },
+    {
+      "epoch": 29.5,
+      "grad_norm": 2.4024806022644043,
+      "learning_rate": 0.0009949263416980988,
+      "loss": 2.4242,
+      "step": 59
+    },
+    {
+      "epoch": 30.0,
+      "grad_norm": 2.5173654556274414,
+      "learning_rate": 0.0009947531997255255,
+      "loss": 2.7259,
+      "step": 60
+    },
+    {
+      "epoch": 30.5,
+      "grad_norm": 2.248589038848877,
+      "learning_rate": 0.0009945771683943837,
+      "loss": 2.5274,
+      "step": 61
+    },
+    {
+      "epoch": 31.0,
+      "grad_norm": 3.8984169960021973,
+      "learning_rate": 0.000994398248732696,
+      "loss": 2.4633,
+      "step": 62
+    },
+    {
+      "epoch": 31.5,
+      "grad_norm": 2.2693045139312744,
+      "learning_rate": 0.0009942164417853533,
+      "loss": 2.5301,
+      "step": 63
+    },
+    {
+      "epoch": 32.0,
+      "grad_norm": 7.542248725891113,
+      "learning_rate": 0.0009940317486141082,
+      "loss": 2.8032,
+      "step": 64
+    },
+    {
+      "epoch": 32.5,
+      "grad_norm": 4.379039287567139,
+      "learning_rate": 0.0009938441702975688,
+      "loss": 2.4167,
+      "step": 65
+    },
+    {
+      "epoch": 33.0,
+      "grad_norm": 4.136926651000977,
+      "learning_rate": 0.0009936537079311926,
+      "loss": 2.7075,
+      "step": 66
+    },
+    {
+      "epoch": 33.5,
+      "grad_norm": 4.302084445953369,
+      "learning_rate": 0.0009934603626272792,
+      "loss": 2.3672,
+      "step": 67
+    },
+    {
+      "epoch": 34.0,
+      "grad_norm": 2.3563308715820312,
+      "learning_rate": 0.0009932641355149653,
+      "loss": 2.5928,
+      "step": 68
+    },
+    {
+      "epoch": 34.5,
+      "grad_norm": 3.951860189437866,
+      "learning_rate": 0.0009930650277402177,
+      "loss": 2.3806,
+      "step": 69
+    },
+    {
+      "epoch": 35.0,
+      "grad_norm": 2.843623161315918,
+      "learning_rate": 0.0009928630404658254,
+      "loss": 2.6033,
+      "step": 70
+    },
+    {
+      "epoch": 35.5,
+      "grad_norm": 3.8206822872161865,
+      "learning_rate": 0.000992658174871394,
+      "loss": 2.4961,
+      "step": 71
+    },
+    {
+      "epoch": 36.0,
+      "grad_norm": 4.13731575012207,
+      "learning_rate": 0.0009924504321533387,
+      "loss": 2.3828,
+      "step": 72
+    },
+    {
+      "epoch": 36.5,
+      "grad_norm": 3.663447141647339,
+      "learning_rate": 0.0009922398135248766,
+      "loss": 2.4227,
+      "step": 73
+    },
+    {
+      "epoch": 37.0,
+      "grad_norm": 4.151423931121826,
+      "learning_rate": 0.0009920263202160206,
+      "loss": 2.5378,
+      "step": 74
+    },
+    {
+      "epoch": 37.5,
+      "grad_norm": 2.8569905757904053,
+      "learning_rate": 0.0009918099534735718,
+      "loss": 2.3476,
+      "step": 75
+    },
+    {
+      "epoch": 38.0,
+      "grad_norm": 3.730823516845703,
+      "learning_rate": 0.0009915907145611115,
+      "loss": 2.481,
+      "step": 76
+    },
+    {
+      "epoch": 38.5,
+      "grad_norm": 3.776073932647705,
+      "learning_rate": 0.0009913686047589953,
+      "loss": 2.2842,
+      "step": 77
+    },
+    {
+      "epoch": 39.0,
+      "grad_norm": 3.668161392211914,
+      "learning_rate": 0.0009911436253643444,
+      "loss": 2.44,
+      "step": 78
+    },
+    {
+      "epoch": 39.5,
+      "grad_norm": 3.550088882446289,
+      "learning_rate": 0.000990915777691038,
+      "loss": 2.3331,
+      "step": 79
+    },
+    {
+      "epoch": 40.0,
+      "grad_norm": 3.5155515670776367,
+      "learning_rate": 0.0009906850630697068,
+      "loss": 2.4583,
+      "step": 80
+    },
+    {
+      "epoch": 40.5,
+      "grad_norm": 3.3810510635375977,
+      "learning_rate": 0.0009904514828477239,
+      "loss": 2.4103,
+      "step": 81
+    },
+    {
+      "epoch": 41.0,
+      "grad_norm": 6.9539079666137695,
+      "learning_rate": 0.0009902150383891979,
+      "loss": 2.2329,
+      "step": 82
+    },
+    {
+      "epoch": 41.5,
+      "grad_norm": 3.1371703147888184,
+      "learning_rate": 0.0009899757310749644,
+      "loss": 2.3054,
+      "step": 83
+    },
+    {
+      "epoch": 42.0,
+      "grad_norm": 5.924234390258789,
+      "learning_rate": 0.000989733562302578,
+      "loss": 2.3634,
+      "step": 84
+    },
+    {
+      "epoch": 42.5,
+      "grad_norm": 3.6022372245788574,
+      "learning_rate": 0.0009894885334863044,
+      "loss": 2.3954,
+      "step": 85
+    },
+    {
+      "epoch": 43.0,
+      "grad_norm": 4.565064907073975,
+      "learning_rate": 0.0009892406460571114,
+      "loss": 2.3331,
+      "step": 86
+    },
+    {
+      "epoch": 43.5,
+      "grad_norm": 3.520937204360962,
+      "learning_rate": 0.000988989901462662,
+      "loss": 2.2789,
+      "step": 87
+    },
+    {
+      "epoch": 44.0,
+      "grad_norm": 3.50038480758667,
+      "learning_rate": 0.0009887363011673045,
+      "loss": 2.2498,
+      "step": 88
+    },
+    {
+      "epoch": 44.5,
+      "grad_norm": 4.377898693084717,
+      "learning_rate": 0.0009884798466520644,
+      "loss": 2.3036,
+      "step": 89
+    },
+    {
+      "epoch": 45.0,
+      "grad_norm": 10.908841133117676,
+      "learning_rate": 0.0009882205394146362,
+      "loss": 2.2828,
+      "step": 90
+    },
+    {
+      "epoch": 45.5,
+      "grad_norm": 3.9022789001464844,
+      "learning_rate": 0.0009879583809693738,
+      "loss": 2.3522,
+      "step": 91
+    },
+    {
+      "epoch": 46.0,
+      "grad_norm": 4.888319969177246,
+      "learning_rate": 0.0009876933728472826,
+      "loss": 2.2359,
+      "step": 92
+    },
+    {
+      "epoch": 46.5,
+      "grad_norm": 4.596319198608398,
+      "learning_rate": 0.0009874255165960102,
+      "loss": 2.2598,
+      "step": 93
+    },
+    {
+      "epoch": 47.0,
+      "grad_norm": 7.082674503326416,
+      "learning_rate": 0.0009871548137798368,
+      "loss": 2.24,
+      "step": 94
+    },
+    {
+      "epoch": 47.5,
+      "grad_norm": 4.798644065856934,
+      "learning_rate": 0.0009868812659796667,
+      "loss": 2.1508,
+      "step": 95
+    },
+    {
+      "epoch": 48.0,
+      "grad_norm": 5.70681095123291,
+      "learning_rate": 0.0009866048747930194,
+      "loss": 2.2962,
+      "step": 96
+    },
+    {
+      "epoch": 48.5,
+      "grad_norm": 3.807445526123047,
+      "learning_rate": 0.0009863256418340188,
+      "loss": 2.2555,
+      "step": 97
+    },
+    {
+      "epoch": 49.0,
+      "grad_norm": 8.064421653747559,
+      "learning_rate": 0.0009860435687333857,
+      "loss": 2.0541,
+      "step": 98
+    },
+    {
+      "epoch": 49.5,
+      "grad_norm": 6.223331928253174,
+      "learning_rate": 0.000985758657138427,
+      "loss": 2.1736,
+      "step": 99
+    },
+    {
+      "epoch": 50.0,
+      "grad_norm": 7.163537502288818,
+      "learning_rate": 0.000985470908713026,
+      "loss": 2.2025,
+      "step": 100
+    },
+    {
+      "epoch": 50.5,
+      "grad_norm": 5.331923484802246,
+      "learning_rate": 0.0009851803251376336,
+      "loss": 2.2715,
+      "step": 101
+    },
+    {
+      "epoch": 51.0,
+      "grad_norm": 7.094827175140381,
+      "learning_rate": 0.000984886908109258,
+      "loss": 2.0748,
+      "step": 102
+    },
+    {
+      "epoch": 51.5,
+      "grad_norm": 5.051886081695557,
+      "learning_rate": 0.0009845906593414545,
+      "loss": 2.0948,
+      "step": 103
+    },
+    {
+      "epoch": 52.0,
+      "grad_norm": 18.785734176635742,
+      "learning_rate": 0.0009842915805643156,
+      "loss": 2.44,
+      "step": 104
+    },
+    {
+      "epoch": 52.5,
+      "grad_norm": 7.797198295593262,
+      "learning_rate": 0.0009839896735244615,
+      "loss": 2.1405,
+      "step": 105
+    },
+    {
+      "epoch": 53.0,
+      "grad_norm": 19.246366500854492,
+      "learning_rate": 0.0009836849399850291,
+      "loss": 2.2466,
+      "step": 106
+    },
+    {
+      "epoch": 53.5,
+      "grad_norm": 5.801110744476318,
+      "learning_rate": 0.0009833773817256623,
+      "loss": 2.1511,
+      "step": 107
+    },
+    {
+      "epoch": 54.0,
+      "grad_norm": 4.608914852142334,
+      "learning_rate": 0.0009830670005425012,
+      "loss": 2.2047,
+      "step": 108
+    },
+    {
+      "epoch": 54.5,
+      "grad_norm": 6.6651506423950195,
+      "learning_rate": 0.000982753798248172,
+      "loss": 2.1357,
+      "step": 109
+    },
+    {
+      "epoch": 55.0,
+      "grad_norm": 5.379322052001953,
+      "learning_rate": 0.0009824377766717758,
+      "loss": 2.1322,
+      "step": 110
+    },
+    {
+      "epoch": 55.5,
+      "grad_norm": 7.068450927734375,
+      "learning_rate": 0.0009821189376588786,
+      "loss": 2.0161,
+      "step": 111
+    },
+    {
+      "epoch": 56.0,
+      "grad_norm": 5.703207969665527,
+      "learning_rate": 0.0009817972830715002,
+      "loss": 2.1359,
+      "step": 112
+    },
+    {
+      "epoch": 56.5,
+      "grad_norm": 7.250577926635742,
+      "learning_rate": 0.0009814728147881033,
+      "loss": 2.2323,
+      "step": 113
+    },
+    {
+      "epoch": 57.0,
+      "grad_norm": 6.518582344055176,
+      "learning_rate": 0.0009811455347035827,
+      "loss": 1.9895,
+      "step": 114
+    },
+    {
+      "epoch": 57.5,
+      "grad_norm": 8.122664451599121,
+      "learning_rate": 0.000980815444729254,
+      "loss": 2.0342,
+      "step": 115
+    },
+    {
+      "epoch": 58.0,
+      "grad_norm": 8.20573616027832,
+      "learning_rate": 0.0009804825467928423,
+      "loss": 2.0057,
+      "step": 116
+    },
+    {
+      "epoch": 58.5,
+      "grad_norm": 10.629237174987793,
+      "learning_rate": 0.0009801468428384716,
+      "loss": 2.1169,
+      "step": 117
+    },
+    {
+      "epoch": 59.0,
+      "grad_norm": 11.46422004699707,
+      "learning_rate": 0.000979808334826653,
+      "loss": 2.2132,
+      "step": 118
+    },
+    {
+      "epoch": 59.5,
+      "grad_norm": 6.267131805419922,
+      "learning_rate": 0.0009794670247342731,
+      "loss": 2.0814,
+      "step": 119
+    },
+    {
+      "epoch": 60.0,
+      "grad_norm": 10.882943153381348,
+      "learning_rate": 0.0009791229145545831,
+      "loss": 1.9246,
+      "step": 120
+    },
+    {
+      "epoch": 60.5,
+      "grad_norm": 6.599976539611816,
+      "learning_rate": 0.000978776006297186,
+      "loss": 1.995,
+      "step": 121
+    },
+    {
+      "epoch": 61.0,
+      "grad_norm": 5.407978534698486,
+      "learning_rate": 0.0009784263019880259,
+      "loss": 2.1412,
+      "step": 122
+    },
+    {
+      "epoch": 61.5,
+      "grad_norm": 5.978907585144043,
+      "learning_rate": 0.0009780738036693764,
+      "loss": 1.9382,
+      "step": 123
+    },
+    {
+      "epoch": 62.0,
+      "grad_norm": 8.217355728149414,
+      "learning_rate": 0.0009777185133998268,
+      "loss": 2.02,
+      "step": 124
+    },
+    {
+      "epoch": 62.5,
+      "grad_norm": 5.982725620269775,
+      "learning_rate": 0.0009773604332542728,
+      "loss": 1.954,
+      "step": 125
+    },
+    {
+      "epoch": 63.0,
+      "grad_norm": 5.391740322113037,
+      "learning_rate": 0.0009769995653239022,
+      "loss": 2.1362,
+      "step": 126
+    },
+    {
+      "epoch": 63.5,
+      "grad_norm": 7.327131271362305,
+      "learning_rate": 0.0009766359117161837,
+      "loss": 1.9135,
+      "step": 127
+    },
+    {
+      "epoch": 64.0,
+      "grad_norm": 6.565709114074707,
+      "learning_rate": 0.000976269474554854,
+      "loss": 2.0883,
+      "step": 128
+    },
+    {
+      "epoch": 64.5,
+      "grad_norm": 8.937880516052246,
+      "learning_rate": 0.0009759002559799061,
+      "loss": 1.9862,
+      "step": 129
+    },
+    {
+      "epoch": 65.0,
+      "grad_norm": 9.248127937316895,
+      "learning_rate": 0.0009755282581475768,
+      "loss": 2.0221,
+      "step": 130
+    },
+    {
+      "epoch": 65.5,
+      "grad_norm": 5.868675231933594,
+      "learning_rate": 0.000975153483230333,
+      "loss": 1.9801,
+      "step": 131
+    },
+    {
+      "epoch": 66.0,
+      "grad_norm": 16.19522476196289,
+      "learning_rate": 0.0009747759334168601,
+      "loss": 1.8436,
+      "step": 132
+    },
+    {
+      "epoch": 66.5,
+      "grad_norm": 6.120225429534912,
+      "learning_rate": 0.0009743956109120491,
+      "loss": 1.9597,
+      "step": 133
+    },
+    {
+      "epoch": 67.0,
+      "grad_norm": 6.745411396026611,
+      "learning_rate": 0.0009740125179369832,
+      "loss": 1.8723,
+      "step": 134
+    },
+    {
+      "epoch": 67.5,
+      "grad_norm": 7.258579730987549,
+      "learning_rate": 0.0009736266567289254,
+      "loss": 1.9936,
+      "step": 135
+    },
+    {
+      "epoch": 68.0,
+      "grad_norm": 8.306854248046875,
+      "learning_rate": 0.0009732380295413049,
+      "loss": 1.7594,
+      "step": 136
+    },
+    {
+      "epoch": 68.5,
+      "grad_norm": 4.320150852203369,
+      "learning_rate": 0.0009728466386437045,
+      "loss": 2.0804,
+      "step": 137
+    },
+    {
+      "epoch": 69.0,
+      "grad_norm": 9.941926956176758,
+      "learning_rate": 0.000972452486321847,
+      "loss": 1.6351,
+      "step": 138
+    },
+    {
+      "epoch": 69.5,
+      "grad_norm": 7.138210773468018,
+      "learning_rate": 0.0009720555748775818,
+      "loss": 1.7578,
+      "step": 139
+    },
+    {
+      "epoch": 70.0,
+      "grad_norm": 22.76557731628418,
+      "learning_rate": 0.0009716559066288715,
+      "loss": 2.1038,
+      "step": 140
+    },
+    {
+      "epoch": 70.5,
+      "grad_norm": 8.667755126953125,
+      "learning_rate": 0.0009712534839097788,
+      "loss": 1.969,
+      "step": 141
+    },
+    {
+      "epoch": 71.0,
+      "grad_norm": 7.31285285949707,
+      "learning_rate": 0.0009708483090704523,
+      "loss": 1.6195,
+      "step": 142
+    },
+    {
+      "epoch": 71.5,
+      "grad_norm": 12.262401580810547,
+      "learning_rate": 0.0009704403844771128,
+      "loss": 1.8415,
+      "step": 143
+    },
+    {
+      "epoch": 72.0,
+      "grad_norm": 11.3973388671875,
+      "learning_rate": 0.0009700297125120399,
+      "loss": 1.8513,
+      "step": 144
+    },
+    {
+      "epoch": 72.5,
+      "grad_norm": 11.92567253112793,
+      "learning_rate": 0.0009696162955735577,
+      "loss": 1.9143,
+      "step": 145
+    },
+    {
+      "epoch": 73.0,
+      "grad_norm": 10.612945556640625,
+      "learning_rate": 0.0009692001360760211,
+      "loss": 1.7683,
+      "step": 146
+    },
+    {
+      "epoch": 73.5,
+      "grad_norm": 13.262011528015137,
+      "learning_rate": 0.0009687812364498014,
+      "loss": 1.8646,
+      "step": 147
+    },
+    {
+      "epoch": 74.0,
+      "grad_norm": 9.530356407165527,
+      "learning_rate": 0.0009683595991412725,
+      "loss": 1.9153,
+      "step": 148
+    },
+    {
+      "epoch": 74.5,
+      "grad_norm": 11.099687576293945,
+      "learning_rate": 0.000967935226612796,
+      "loss": 1.7594,
+      "step": 149
+    },
+    {
+      "epoch": 75.0,
+      "grad_norm": 7.055493354797363,
+      "learning_rate": 0.0009675081213427075,
+      "loss": 2.0234,
+      "step": 150
+    },
+    {
+      "epoch": 75.5,
+      "grad_norm": 7.786091327667236,
+      "learning_rate": 0.0009670782858253015,
+      "loss": 1.8842,
+      "step": 151
+    },
+    {
+      "epoch": 76.0,
+      "grad_norm": 6.588303089141846,
+      "learning_rate": 0.0009666457225708174,
+      "loss": 1.9084,
+      "step": 152
+    },
+    {
+      "epoch": 76.5,
+      "grad_norm": 8.27043342590332,
+      "learning_rate": 0.0009662104341054246,
+      "loss": 1.8623,
+      "step": 153
+    },
+    {
+      "epoch": 77.0,
+      "grad_norm": 7.812788009643555,
+      "learning_rate": 0.0009657724229712075,
+      "loss": 1.7695,
+      "step": 154
+    },
+    {
+      "epoch": 77.5,
+      "grad_norm": 14.671862602233887,
+      "learning_rate": 0.0009653316917261511,
+      "loss": 1.8302,
+      "step": 155
+    },
+    {
+      "epoch": 78.0,
+      "grad_norm": 6.912378311157227,
+      "learning_rate": 0.0009648882429441257,
+      "loss": 1.7928,
+      "step": 156
+    },
+    {
+      "epoch": 78.5,
+      "grad_norm": 10.783658981323242,
+      "learning_rate": 0.0009644420792148722,
+      "loss": 1.7339,
+      "step": 157
+    },
+    {
+      "epoch": 79.0,
+      "grad_norm": 11.711565017700195,
+      "learning_rate": 0.0009639932031439866,
+      "loss": 1.8855,
+      "step": 158
+    },
+    {
+      "epoch": 79.5,
+      "grad_norm": 6.835365295410156,
+      "learning_rate": 0.000963541617352905,
+      "loss": 1.6674,
+      "step": 159
+    },
+    {
+      "epoch": 80.0,
+      "grad_norm": 7.767190933227539,
+      "learning_rate": 0.0009630873244788883,
+      "loss": 1.9275,
+      "step": 160
+    },
+    {
+      "epoch": 80.5,
+      "grad_norm": 9.744241714477539,
+      "learning_rate": 0.0009626303271750068,
+      "loss": 1.6804,
+      "step": 161
+    },
+    {
+      "epoch": 81.0,
+      "grad_norm": 10.34986686706543,
+      "learning_rate": 0.0009621706281101248,
+      "loss": 1.8927,
+      "step": 162
+    },
+    {
+      "epoch": 81.5,
+      "grad_norm": 8.027229309082031,
+      "learning_rate": 0.0009617082299688847,
+      "loss": 1.6489,
+      "step": 163
+    },
+    {
+      "epoch": 82.0,
+      "grad_norm": 8.991996765136719,
+      "learning_rate": 0.0009612431354516912,
+      "loss": 1.8506,
+      "step": 164
+    },
+    {
+      "epoch": 82.5,
+      "grad_norm": 12.033332824707031,
+      "learning_rate": 0.0009607753472746965,
+      "loss": 1.8361,
+      "step": 165
+    },
+    {
+      "epoch": 83.0,
+      "grad_norm": 15.585806846618652,
+      "learning_rate": 0.0009603048681697834,
+      "loss": 1.6737,
+      "step": 166
+    },
+    {
+      "epoch": 83.5,
+      "grad_norm": 18.806915283203125,
+      "learning_rate": 0.0009598317008845496,
+      "loss": 1.8409,
+      "step": 167
+    },
+    {
+      "epoch": 84.0,
+      "grad_norm": 12.116573333740234,
+      "learning_rate": 0.0009593558481822921,
+      "loss": 2.0167,
+      "step": 168
+    },
+    {
+      "epoch": 84.5,
+      "grad_norm": 14.075790405273438,
+      "learning_rate": 0.0009588773128419905,
+      "loss": 1.8515,
+      "step": 169
+    },
+    {
+      "epoch": 85.0,
+      "grad_norm": 10.268305778503418,
+      "learning_rate": 0.0009583960976582913,
+      "loss": 1.5902,
+      "step": 170
+    },
+    {
+      "epoch": 85.5,
+      "grad_norm": 17.725053787231445,
+      "learning_rate": 0.0009579122054414906,
+      "loss": 1.9089,
+      "step": 171
+    },
+    {
+      "epoch": 86.0,
+      "grad_norm": 11.047142028808594,
+      "learning_rate": 0.0009574256390175192,
+      "loss": 1.6296,
+      "step": 172
+    },
+    {
+      "epoch": 86.5,
+      "grad_norm": 11.587921142578125,
+      "learning_rate": 0.0009569364012279247,
+      "loss": 1.5714,
+      "step": 173
+    },
+    {
+      "epoch": 87.0,
+      "grad_norm": 7.835581302642822,
+      "learning_rate": 0.0009564444949298558,
+      "loss": 1.7596,
+      "step": 174
+    },
+    {
+      "epoch": 87.5,
+      "grad_norm": 13.838013648986816,
+      "learning_rate": 0.0009559499229960451,
+      "loss": 1.6376,
+      "step": 175
+    },
+    {
+      "epoch": 88.0,
+      "grad_norm": 12.516178131103516,
+      "learning_rate": 0.0009554526883147925,
+      "loss": 1.8259,
+      "step": 176
+    },
+    {
+      "epoch": 88.5,
+      "grad_norm": 6.196531772613525,
+      "learning_rate": 0.0009549527937899487,
+      "loss": 1.8276,
+      "step": 177
+    },
+    {
+      "epoch": 89.0,
+      "grad_norm": 8.698816299438477,
+      "learning_rate": 0.0009544502423408973,
+      "loss": 1.589,
+      "step": 178
+    },
+    {
+      "epoch": 89.5,
+      "grad_norm": 8.368842124938965,
+      "learning_rate": 0.0009539450369025387,
+      "loss": 1.6927,
+      "step": 179
+    },
+    {
+      "epoch": 90.0,
+      "grad_norm": 8.643660545349121,
+      "learning_rate": 0.0009534371804252727,
+      "loss": 1.7054,
+      "step": 180
+    },
+    {
+      "epoch": 90.5,
+      "grad_norm": 6.945327281951904,
+      "learning_rate": 0.000952926675874981,
+      "loss": 1.8919,
+      "step": 181
+    },
+    {
+      "epoch": 91.0,
+      "grad_norm": 14.514830589294434,
+      "learning_rate": 0.0009524135262330098,
+      "loss": 1.4765,
+      "step": 182
+    },
+    {
+      "epoch": 91.5,
+      "grad_norm": 10.830658912658691,
+      "learning_rate": 0.000951897734496153,
+      "loss": 1.6864,
+      "step": 183
+    },
+    {
+      "epoch": 92.0,
+      "grad_norm": 9.927618980407715,
+      "learning_rate": 0.0009513793036766345,
+      "loss": 1.6992,
+      "step": 184
+    },
+    {
+      "epoch": 92.5,
+      "grad_norm": 8.724226951599121,
+      "learning_rate": 0.0009508582368020897,
+      "loss": 1.553,
+      "step": 185
+    },
+    {
+      "epoch": 93.0,
+      "grad_norm": 9.168572425842285,
+      "learning_rate": 0.0009503345369155494,
+      "loss": 1.8587,
+      "step": 186
+    },
+    {
+      "epoch": 93.5,
+      "grad_norm": 11.56299114227295,
+      "learning_rate": 0.0009498082070754206,
+      "loss": 1.5631,
+      "step": 187
+    },
+    {
+      "epoch": 94.0,
+      "grad_norm": 11.164207458496094,
+      "learning_rate": 0.0009492792503554695,
+      "loss": 1.751,
+      "step": 188
+    },
+    {
+      "epoch": 94.5,
+      "grad_norm": 8.851701736450195,
+      "learning_rate": 0.0009487476698448032,
+      "loss": 1.6499,
+      "step": 189
+    },
+    {
+      "epoch": 95.0,
+      "grad_norm": 11.101888656616211,
+      "learning_rate": 0.0009482134686478518,
+      "loss": 1.7649,
+      "step": 190
+    },
+    {
+      "epoch": 95.5,
+      "grad_norm": 9.127381324768066,
+      "learning_rate": 0.0009476766498843503,
+      "loss": 1.7786,
+      "step": 191
+    },
+    {
+      "epoch": 96.0,
+      "grad_norm": 13.868022918701172,
+      "learning_rate": 0.0009471372166893198,
+      "loss": 1.4831,
+      "step": 192
+    },
+    {
+      "epoch": 96.5,
+      "grad_norm": 10.440984725952148,
+      "learning_rate": 0.0009465951722130504,
+      "loss": 1.6713,
+      "step": 193
+    },
+    {
+      "epoch": 97.0,
+      "grad_norm": 12.911642074584961,
+      "learning_rate": 0.0009460505196210813,
+      "loss": 1.5111,
+      "step": 194
+    },
+    {
+      "epoch": 97.5,
+      "grad_norm": 6.718067169189453,
+      "learning_rate": 0.0009455032620941839,
+      "loss": 1.776,
+      "step": 195
+    },
+    {
+      "epoch": 98.0,
+      "grad_norm": 13.30576229095459,
+      "learning_rate": 0.000944953402828342,
+      "loss": 1.5254,
+      "step": 196
+    },
+    {
+      "epoch": 98.5,
+      "grad_norm": 13.088066101074219,
+      "learning_rate": 0.0009444009450347333,
+      "loss": 1.5001,
+      "step": 197
+    },
+    {
+      "epoch": 99.0,
+      "grad_norm": 14.803836822509766,
+      "learning_rate": 0.0009438458919397112,
+      "loss": 1.7501,
+      "step": 198
+    },
+    {
+      "epoch": 99.5,
+      "grad_norm": 13.634467124938965,
+      "learning_rate": 0.000943288246784786,
+      "loss": 1.4281,
+      "step": 199
+    },
+    {
+      "epoch": 100.0,
+      "grad_norm": 21.91295051574707,
+      "learning_rate": 0.0009427280128266049,
+      "loss": 1.712,
+      "step": 200
+    },
+    {
+      "epoch": 100.5,
+      "grad_norm": 8.674274444580078,
+      "learning_rate": 0.0009421651933369344,
+      "loss": 1.6702,
+      "step": 201
+    },
+    {
+      "epoch": 101.0,
+      "grad_norm": 12.499960899353027,
+      "learning_rate": 0.00094159979160264,
+      "loss": 1.5329,
+      "step": 202
+    },
+    {
+      "epoch": 101.5,
+      "grad_norm": 15.862119674682617,
+      "learning_rate": 0.0009410318109256676,
+      "loss": 1.6178,
+      "step": 203
+    },
+    {
+      "epoch": 102.0,
+      "grad_norm": 10.31312370300293,
+      "learning_rate": 0.0009404612546230243,
+      "loss": 1.5641,
+      "step": 204
+    },
+    {
+      "epoch": 102.5,
+      "grad_norm": 14.866028785705566,
+      "learning_rate": 0.0009398881260267589,
+      "loss": 1.8177,
+      "step": 205
+    },
+    {
+      "epoch": 103.0,
+      "grad_norm": 24.74469566345215,
+      "learning_rate": 0.0009393124284839423,
+      "loss": 1.4349,
+      "step": 206
+    },
+    {
+      "epoch": 103.5,
+      "grad_norm": 16.636493682861328,
+      "learning_rate": 0.0009387341653566476,
+      "loss": 1.6918,
+      "step": 207
+    },
+    {
+      "epoch": 104.0,
+      "grad_norm": 12.825433731079102,
+      "learning_rate": 0.0009381533400219318,
+      "loss": 1.6022,
+      "step": 208
+    },
+    {
+      "epoch": 104.5,
+      "grad_norm": 21.460569381713867,
+      "learning_rate": 0.0009375699558718145,
+      "loss": 1.5298,
+      "step": 209
+    },
+    {
+      "epoch": 105.0,
+      "grad_norm": 16.756227493286133,
+      "learning_rate": 0.000936984016313259,
+      "loss": 1.7019,
+      "step": 210
+    },
+    {
+      "epoch": 105.5,
+      "grad_norm": 13.23095703125,
+      "learning_rate": 0.0009363955247681521,
+      "loss": 1.5358,
+      "step": 211
+    },
+    {
+      "epoch": 106.0,
+      "grad_norm": 13.423588752746582,
+      "learning_rate": 0.0009358044846732847,
+      "loss": 1.4743,
+      "step": 212
+    },
+    {
+      "epoch": 106.5,
+      "grad_norm": 7.051394939422607,
+      "learning_rate": 0.0009352108994803302,
+      "loss": 1.6436,
+      "step": 213
+    },
+    {
+      "epoch": 107.0,
+      "grad_norm": 17.12705421447754,
+      "learning_rate": 0.0009346147726558265,
+      "loss": 1.5267,
+      "step": 214
+    },
+    {
+      "epoch": 107.5,
+      "grad_norm": 9.214910507202148,
+      "learning_rate": 0.0009340161076811538,
+      "loss": 1.5217,
+      "step": 215
+    },
+    {
+      "epoch": 108.0,
+      "grad_norm": 17.610891342163086,
+      "learning_rate": 0.0009334149080525154,
+      "loss": 1.5891,
+      "step": 216
+    },
+    {
+      "epoch": 108.5,
+      "grad_norm": 20.65456199645996,
+      "learning_rate": 0.0009328111772809172,
+      "loss": 1.3947,
+      "step": 217
+    },
+    {
+      "epoch": 109.0,
+      "grad_norm": 16.141841888427734,
+      "learning_rate": 0.0009322049188921467,
+      "loss": 1.6089,
+      "step": 218
+    },
+    {
+      "epoch": 109.5,
+      "grad_norm": 12.685590744018555,
+      "learning_rate": 0.0009315961364267526,
+      "loss": 1.6089,
+      "step": 219
+    },
+    {
+      "epoch": 110.0,
+      "grad_norm": 8.32016658782959,
+      "learning_rate": 0.0009309848334400246,
+      "loss": 1.5337,
+      "step": 220
+    },
+    {
+      "epoch": 110.5,
+      "grad_norm": 11.81705093383789,
+      "learning_rate": 0.0009303710135019718,
+      "loss": 1.537,
+      "step": 221
+    },
+    {
+      "epoch": 111.0,
+      "grad_norm": 18.108293533325195,
+      "learning_rate": 0.0009297546801973025,
+      "loss": 1.2677,
+      "step": 222
+    },
+    {
+      "epoch": 111.5,
+      "grad_norm": 9.78410530090332,
+      "learning_rate": 0.0009291358371254033,
+      "loss": 1.2752,
+      "step": 223
+    },
+    {
+      "epoch": 112.0,
+      "grad_norm": 10.054555892944336,
+      "learning_rate": 0.0009285144879003172,
+      "loss": 1.7922,
+      "step": 224
+    },
+    {
+      "epoch": 112.5,
+      "grad_norm": 8.019747734069824,
+      "learning_rate": 0.0009278906361507238,
+      "loss": 1.4366,
+      "step": 225
+    },
+    {
+      "epoch": 113.0,
+      "grad_norm": 17.618715286254883,
+      "learning_rate": 0.0009272642855199171,
+      "loss": 1.5628,
+      "step": 226
+    },
+    {
+      "epoch": 113.5,
+      "grad_norm": 12.588883399963379,
+      "learning_rate": 0.0009266354396657846,
+      "loss": 1.292,
+      "step": 227
+    },
+    {
+      "epoch": 114.0,
+      "grad_norm": 9.3181791305542,
+      "learning_rate": 0.0009260041022607859,
+      "loss": 1.7196,
+      "step": 228
+    },
+    {
+      "epoch": 114.5,
+      "grad_norm": 8.130376815795898,
+      "learning_rate": 0.0009253702769919315,
+      "loss": 1.6181,
+      "step": 229
+    },
+    {
+      "epoch": 115.0,
+      "grad_norm": 16.976757049560547,
+      "learning_rate": 0.0009247339675607605,
+      "loss": 1.5381,
+      "step": 230
+    },
+    {
+      "epoch": 115.5,
+      "grad_norm": 22.368785858154297,
+      "learning_rate": 0.0009240951776833202,
+      "loss": 1.4554,
+      "step": 231
+    },
+    {
+      "epoch": 116.0,
+      "grad_norm": 33.194637298583984,
+      "learning_rate": 0.000923453911090143,
+      "loss": 2.4396,
+      "step": 232
+    },
+    {
+      "epoch": 116.5,
+      "grad_norm": 16.492259979248047,
+      "learning_rate": 0.0009228101715262258,
+      "loss": 1.733,
+      "step": 233
+    },
+    {
+      "epoch": 117.0,
+      "grad_norm": 14.9178466796875,
+      "learning_rate": 0.0009221639627510075,
+      "loss": 1.5963,
+      "step": 234
+    },
+    {
+      "epoch": 117.5,
+      "grad_norm": 10.215200424194336,
+      "learning_rate": 0.0009215152885383473,
+      "loss": 1.6351,
+      "step": 235
+    },
+    {
+      "epoch": 118.0,
+      "grad_norm": 12.47648811340332,
+      "learning_rate": 0.0009208641526765023,
+      "loss": 1.5152,
+      "step": 236
+    },
+    {
+      "epoch": 118.5,
+      "grad_norm": 10.15252685546875,
+      "learning_rate": 0.0009202105589681057,
+      "loss": 1.6814,
+      "step": 237
+    },
+    {
+      "epoch": 119.0,
+      "grad_norm": 12.799568176269531,
+      "learning_rate": 0.0009195545112301446,
+      "loss": 1.3236,
+      "step": 238
+    },
+    {
+      "epoch": 119.5,
+      "grad_norm": 14.984341621398926,
+      "learning_rate": 0.000918896013293938,
+      "loss": 1.5856,
+      "step": 239
+    },
+    {
+      "epoch": 120.0,
+      "grad_norm": 14.022126197814941,
+      "learning_rate": 0.0009182350690051134,
+      "loss": 1.3913,
+      "step": 240
+    },
+    {
+      "epoch": 120.5,
+      "grad_norm": 12.633641242980957,
+      "learning_rate": 0.0009175716822235853,
+      "loss": 1.6047,
+      "step": 241
+    },
+    {
+      "epoch": 121.0,
+      "grad_norm": 12.839395523071289,
+      "learning_rate": 0.0009169058568235323,
+      "loss": 1.4977,
+      "step": 242
+    },
+    {
+      "epoch": 121.5,
+      "grad_norm": 11.50823974609375,
+      "learning_rate": 0.0009162375966933747,
+      "loss": 1.3668,
+      "step": 243
+    },
+    {
+      "epoch": 122.0,
+      "grad_norm": 8.134180068969727,
+      "learning_rate": 0.0009155669057357514,
+      "loss": 1.5449,
+      "step": 244
+    },
+    {
+      "epoch": 122.5,
+      "grad_norm": 11.019954681396484,
+      "learning_rate": 0.0009148937878674975,
+      "loss": 1.4755,
+      "step": 245
+    },
+    {
+      "epoch": 123.0,
+      "grad_norm": 12.864787101745605,
+      "learning_rate": 0.0009142182470196212,
+      "loss": 1.4074,
+      "step": 246
+    },
+    {
+      "epoch": 123.5,
+      "grad_norm": 23.820556640625,
+      "learning_rate": 0.0009135402871372809,
+      "loss": 1.5804,
+      "step": 247
+    },
+    {
+      "epoch": 124.0,
+      "grad_norm": 13.41064167022705,
+      "learning_rate": 0.0009128599121797621,
+      "loss": 1.3697,
+      "step": 248
+    },
+    {
+      "epoch": 124.5,
+      "grad_norm": 14.276724815368652,
+      "learning_rate": 0.0009121771261204544,
+      "loss": 1.3194,
+      "step": 249
+    },
+    {
+      "epoch": 125.0,
+      "grad_norm": 12.61825180053711,
+      "learning_rate": 0.0009114919329468282,
+      "loss": 1.6941,
+      "step": 250
+    },
+    {
+      "epoch": 125.5,
+      "grad_norm": 18.70136260986328,
+      "learning_rate": 0.0009108043366604115,
+      "loss": 1.4425,
+      "step": 251
+    },
+    {
+      "epoch": 126.0,
+      "grad_norm": 13.025047302246094,
+      "learning_rate": 0.0009101143412767665,
+      "loss": 1.6107,
+      "step": 252
+    },
+    {
+      "epoch": 126.5,
+      "grad_norm": 14.64149284362793,
+      "learning_rate": 0.000909421950825466,
+      "loss": 1.4423,
+      "step": 253
+    },
+    {
+      "epoch": 127.0,
+      "grad_norm": 12.096929550170898,
+      "learning_rate": 0.00090872716935007,
+      "loss": 1.5401,
+      "step": 254
+    },
+    {
+      "epoch": 127.5,
+      "grad_norm": 19.043426513671875,
+      "learning_rate": 0.0009080300009081024,
+      "loss": 1.4329,
+      "step": 255
+    },
+    {
+      "epoch": 128.0,
+      "grad_norm": 20.608753204345703,
+      "learning_rate": 0.0009073304495710266,
+      "loss": 1.4398,
+      "step": 256
+    },
+    {
+      "epoch": 128.5,
+      "grad_norm": 11.185997009277344,
+      "learning_rate": 0.0009066285194242221,
+      "loss": 1.5492,
+      "step": 257
+    },
+    {
+      "epoch": 129.0,
+      "grad_norm": 20.962587356567383,
+      "learning_rate": 0.000905924214566961,
+      "loss": 1.4088,
+      "step": 258
+    },
+    {
+      "epoch": 129.5,
+      "grad_norm": 10.081209182739258,
+      "learning_rate": 0.0009052175391123834,
+      "loss": 1.3783,
+      "step": 259
+    },
+    {
+      "epoch": 130.0,
+      "grad_norm": 14.76672077178955,
+      "learning_rate": 0.0009045084971874737,
+      "loss": 1.3848,
+      "step": 260
+    },
+    {
+      "epoch": 130.5,
+      "grad_norm": 11.170145034790039,
+      "learning_rate": 0.0009037970929330368,
+      "loss": 1.4359,
+      "step": 261
+    },
+    {
+      "epoch": 131.0,
+      "grad_norm": 8.148204803466797,
+      "learning_rate": 0.0009030833305036732,
+      "loss": 1.5243,
+      "step": 262
+    },
+    {
+      "epoch": 131.5,
+      "grad_norm": 13.777543067932129,
+      "learning_rate": 0.0009023672140677554,
+      "loss": 1.3912,
+      "step": 263
+    },
+    {
+      "epoch": 132.0,
+      "grad_norm": 26.317293167114258,
+      "learning_rate": 0.0009016487478074031,
+      "loss": 1.2777,
+      "step": 264
+    },
+    {
+      "epoch": 132.5,
+      "grad_norm": 9.286993980407715,
+      "learning_rate": 0.0009009279359184592,
+      "loss": 1.5338,
+      "step": 265
+    },
+    {
+      "epoch": 133.0,
+      "grad_norm": 17.991010665893555,
+      "learning_rate": 0.0009002047826104651,
+      "loss": 1.3308,
+      "step": 266
+    },
+    {
+      "epoch": 133.5,
+      "grad_norm": 12.991069793701172,
+      "learning_rate": 0.0008994792921066355,
+      "loss": 1.4701,
+      "step": 267
+    },
+    {
+      "epoch": 134.0,
+      "grad_norm": 19.580888748168945,
+      "learning_rate": 0.0008987514686438353,
+      "loss": 1.2231,
+      "step": 268
+    },
+    {
+      "epoch": 134.5,
+      "grad_norm": 9.609387397766113,
+      "learning_rate": 0.0008980213164725531,
+      "loss": 1.3796,
+      "step": 269
+    },
+    {
+      "epoch": 135.0,
+      "grad_norm": 11.453824996948242,
+      "learning_rate": 0.0008972888398568772,
+      "loss": 1.1492,
+      "step": 270
+    },
+    {
+      "epoch": 135.5,
+      "grad_norm": 15.122506141662598,
+      "learning_rate": 0.0008965540430744712,
+      "loss": 1.2197,
+      "step": 271
+    },
+    {
+      "epoch": 136.0,
+      "grad_norm": 12.762795448303223,
+      "learning_rate": 0.0008958169304165479,
+      "loss": 1.533,
+      "step": 272
+    },
+    {
+      "epoch": 136.5,
+      "grad_norm": 24.887075424194336,
+      "learning_rate": 0.0008950775061878452,
+      "loss": 1.3602,
+      "step": 273
+    },
+    {
+      "epoch": 137.0,
+      "grad_norm": 18.015087127685547,
+      "learning_rate": 0.0008943357747066003,
+      "loss": 1.5373,
+      "step": 274
+    },
+    {
+      "epoch": 137.5,
+      "grad_norm": 13.313756942749023,
+      "learning_rate": 0.000893591740304525,
+      "loss": 1.4967,
+      "step": 275
+    },
+    {
+      "epoch": 138.0,
+      "grad_norm": 12.451852798461914,
+      "learning_rate": 0.0008928454073267801,
+      "loss": 1.3501,
+      "step": 276
+    },
+    {
+      "epoch": 138.5,
+      "grad_norm": 9.580190658569336,
+      "learning_rate": 0.0008920967801319497,
+      "loss": 1.4711,
+      "step": 277
+    },
+    {
+      "epoch": 139.0,
+      "grad_norm": 13.102746963500977,
+      "learning_rate": 0.0008913458630920168,
+      "loss": 1.1741,
+      "step": 278
+    },
+    {
+      "epoch": 139.5,
+      "grad_norm": 9.171207427978516,
+      "learning_rate": 0.0008905926605923364,
+      "loss": 1.3669,
+      "step": 279
+    },
+    {
+      "epoch": 140.0,
+      "grad_norm": 9.742522239685059,
+      "learning_rate": 0.0008898371770316111,
+      "loss": 1.3348,
+      "step": 280
+    },
+    {
+      "epoch": 140.5,
+      "grad_norm": 10.743603706359863,
+      "learning_rate": 0.0008890794168218649,
+      "loss": 1.3181,
+      "step": 281
+    },
+    {
+      "epoch": 141.0,
+      "grad_norm": 9.673843383789062,
+      "learning_rate": 0.0008883193843884169,
+      "loss": 1.4994,
+      "step": 282
+    },
+    {
+      "epoch": 141.5,
+      "grad_norm": 12.297232627868652,
+      "learning_rate": 0.0008875570841698564,
+      "loss": 1.5218,
+      "step": 283
+    },
+    {
+      "epoch": 142.0,
+      "grad_norm": 11.80233097076416,
+      "learning_rate": 0.0008867925206180165,
+      "loss": 1.3322,
+      "step": 284
+    },
+    {
+      "epoch": 142.5,
+      "grad_norm": 13.119105339050293,
+      "learning_rate": 0.0008860256981979484,
+      "loss": 1.3427,
+      "step": 285
+    },
+    {
+      "epoch": 143.0,
+      "grad_norm": 14.094135284423828,
+      "learning_rate": 0.0008852566213878947,
+      "loss": 1.3228,
+      "step": 286
+    },
+    {
+      "epoch": 143.5,
+      "grad_norm": 9.89586353302002,
+      "learning_rate": 0.0008844852946792636,
+      "loss": 1.3083,
+      "step": 287
+    },
+    {
+      "epoch": 144.0,
+      "grad_norm": 13.487672805786133,
+      "learning_rate": 0.0008837117225766032,
+      "loss": 1.4085,
+      "step": 288
+    },
+    {
+      "epoch": 144.5,
+      "grad_norm": 12.937167167663574,
+      "learning_rate": 0.0008829359095975741,
+      "loss": 1.4436,
+      "step": 289
+    },
+    {
+      "epoch": 145.0,
+      "grad_norm": 12.483976364135742,
+      "learning_rate": 0.0008821578602729241,
+      "loss": 1.236,
+      "step": 290
+    },
+    {
+      "epoch": 145.5,
+      "grad_norm": 17.987428665161133,
+      "learning_rate": 0.0008813775791464611,
+      "loss": 1.2803,
+      "step": 291
+    },
+    {
+      "epoch": 146.0,
+      "grad_norm": 11.369746208190918,
+      "learning_rate": 0.0008805950707750268,
+      "loss": 1.3146,
+      "step": 292
+    },
+    {
+      "epoch": 146.5,
+      "grad_norm": 10.420331001281738,
+      "learning_rate": 0.0008798103397284695,
+      "loss": 1.3973,
+      "step": 293
+    },
+    {
+      "epoch": 147.0,
+      "grad_norm": 11.98904800415039,
+      "learning_rate": 0.0008790233905896185,
+      "loss": 0.9791,
+      "step": 294
+    },
+    {
+      "epoch": 147.5,
+      "grad_norm": 8.22117805480957,
+      "learning_rate": 0.0008782342279542568,
+      "loss": 1.2734,
+      "step": 295
+    },
+    {
+      "epoch": 148.0,
+      "grad_norm": 16.36408233642578,
+      "learning_rate": 0.0008774428564310938,
+      "loss": 1.1829,
+      "step": 296
+    },
+    {
+      "epoch": 148.5,
+      "grad_norm": 15.284476280212402,
+      "learning_rate": 0.000876649280641739,
+      "loss": 1.3396,
+      "step": 297
+    },
+    {
+      "epoch": 149.0,
+      "grad_norm": 10.37116813659668,
+      "learning_rate": 0.0008758535052206749,
+      "loss": 1.3563,
+      "step": 298
+    },
+    {
+      "epoch": 149.5,
+      "grad_norm": 29.462068557739258,
+      "learning_rate": 0.0008750555348152298,
+      "loss": 1.2941,
+      "step": 299
+    },
+    {
+      "epoch": 150.0,
+      "grad_norm": 11.296568870544434,
+      "learning_rate": 0.0008742553740855505,
+      "loss": 1.3482,
+      "step": 300
+    },
+    {
+      "epoch": 150.5,
+      "grad_norm": 13.459062576293945,
+      "learning_rate": 0.0008734530277045758,
+      "loss": 1.227,
+      "step": 301
+    },
+    {
+      "epoch": 151.0,
+      "grad_norm": 9.609284400939941,
+      "learning_rate": 0.000872648500358008,
+      "loss": 1.3937,
+      "step": 302
+    },
+    {
+      "epoch": 151.5,
+      "grad_norm": 34.58816909790039,
+      "learning_rate": 0.0008718417967442868,
+      "loss": 1.0911,
+      "step": 303
+    },
+    {
+      "epoch": 152.0,
+      "grad_norm": 19.765283584594727,
+      "learning_rate": 0.0008710329215745611,
+      "loss": 1.4282,
+      "step": 304
+    },
+    {
+      "epoch": 152.5,
+      "grad_norm": 12.024175643920898,
+      "learning_rate": 0.0008702218795726618,
+      "loss": 1.5104,
+      "step": 305
+    },
+    {
+      "epoch": 153.0,
+      "grad_norm": 19.932125091552734,
+      "learning_rate": 0.0008694086754750737,
+      "loss": 1.0787,
+      "step": 306
+    },
+    {
+      "epoch": 153.5,
+      "grad_norm": 9.023565292358398,
+      "learning_rate": 0.0008685933140309088,
+      "loss": 1.29,
+      "step": 307
+    },
+    {
+      "epoch": 154.0,
+      "grad_norm": 26.47771453857422,
+      "learning_rate": 0.0008677758000018776,
+      "loss": 1.1245,
+      "step": 308
+    },
+    {
+      "epoch": 154.5,
+      "grad_norm": 11.213178634643555,
+      "learning_rate": 0.0008669561381622619,
+      "loss": 1.4043,
+      "step": 309
+    },
+    {
+      "epoch": 155.0,
+      "grad_norm": 17.272253036499023,
+      "learning_rate": 0.0008661343332988868,
+      "loss": 1.0728,
+      "step": 310
+    },
+    {
+      "epoch": 155.5,
+      "grad_norm": 17.470422744750977,
+      "learning_rate": 0.0008653103902110923,
+      "loss": 1.3769,
+      "step": 311
+    },
+    {
+      "epoch": 156.0,
+      "grad_norm": 13.005993843078613,
+      "learning_rate": 0.0008644843137107057,
+      "loss": 1.2039,
+      "step": 312
+    },
+    {
+      "epoch": 156.5,
+      "grad_norm": 18.14457130432129,
+      "learning_rate": 0.000863656108622014,
+      "loss": 1.3536,
+      "step": 313
+    },
+    {
+      "epoch": 157.0,
+      "grad_norm": 13.936728477478027,
+      "learning_rate": 0.0008628257797817344,
+      "loss": 1.3521,
+      "step": 314
+    },
+    {
+      "epoch": 157.5,
+      "grad_norm": 15.196125984191895,
+      "learning_rate": 0.0008619933320389873,
+      "loss": 1.2514,
+      "step": 315
+    },
+    {
+      "epoch": 158.0,
+      "grad_norm": 15.42523193359375,
+      "learning_rate": 0.000861158770255267,
+      "loss": 1.269,
+      "step": 316
+    },
+    {
+      "epoch": 158.5,
+      "grad_norm": 20.262516021728516,
+      "learning_rate": 0.0008603220993044144,
+      "loss": 1.1126,
+      "step": 317
+    },
+    {
+      "epoch": 159.0,
+      "grad_norm": 23.66640853881836,
+      "learning_rate": 0.0008594833240725876,
+      "loss": 1.3849,
+      "step": 318
+    },
+    {
+      "epoch": 159.5,
+      "grad_norm": 14.617342948913574,
+      "learning_rate": 0.0008586424494582339,
+      "loss": 1.0276,
+      "step": 319
+    },
+    {
+      "epoch": 160.0,
+      "grad_norm": 12.280439376831055,
+      "learning_rate": 0.0008577994803720606,
+      "loss": 1.3745,
+      "step": 320
+    },
+    {
+      "epoch": 160.5,
+      "grad_norm": 8.214456558227539,
+      "learning_rate": 0.0008569544217370071,
+      "loss": 1.3652,
+      "step": 321
+    },
+    {
+      "epoch": 161.0,
+      "grad_norm": 18.886180877685547,
+      "learning_rate": 0.0008561072784882155,
+      "loss": 1.1102,
+      "step": 322
+    },
+    {
+      "epoch": 161.5,
+      "grad_norm": 13.760377883911133,
+      "learning_rate": 0.0008552580555730023,
+      "loss": 1.2523,
+      "step": 323
+    },
+    {
+      "epoch": 162.0,
+      "grad_norm": 13.003059387207031,
+      "learning_rate": 0.0008544067579508291,
+      "loss": 1.1512,
+      "step": 324
+    },
+    {
+      "epoch": 162.5,
+      "grad_norm": 8.50968074798584,
+      "learning_rate": 0.0008535533905932737,
+      "loss": 1.4246,
+      "step": 325
+    },
+    {
+      "epoch": 163.0,
+      "grad_norm": 11.936058044433594,
+      "learning_rate": 0.0008526979584840015,
+      "loss": 1.0148,
+      "step": 326
+    },
+    {
+      "epoch": 163.5,
+      "grad_norm": 9.090569496154785,
+      "learning_rate": 0.0008518404666187356,
+      "loss": 1.3052,
+      "step": 327
+    },
+    {
+      "epoch": 164.0,
+      "grad_norm": 13.546277046203613,
+      "learning_rate": 0.0008509809200052286,
+      "loss": 1.1196,
+      "step": 328
+    },
+    {
+      "epoch": 164.5,
+      "grad_norm": 10.964517593383789,
+      "learning_rate": 0.0008501193236632318,
+      "loss": 1.219,
+      "step": 329
+    },
+    {
+      "epoch": 165.0,
+      "grad_norm": 13.495043754577637,
+      "learning_rate": 0.0008492556826244686,
+      "loss": 1.2563,
+      "step": 330
+    },
+    {
+      "epoch": 165.5,
+      "grad_norm": 14.502145767211914,
+      "learning_rate": 0.0008483900019326016,
+      "loss": 1.0771,
+      "step": 331
+    },
+    {
+      "epoch": 166.0,
+      "grad_norm": 11.228195190429688,
+      "learning_rate": 0.0008475222866432064,
+      "loss": 1.3367,
+      "step": 332
+    },
+    {
+      "epoch": 166.5,
+      "grad_norm": 16.335834503173828,
+      "learning_rate": 0.0008466525418237399,
+      "loss": 1.2087,
+      "step": 333
+    },
+    {
+      "epoch": 167.0,
+      "grad_norm": 10.01768684387207,
+      "learning_rate": 0.0008457807725535116,
+      "loss": 1.2481,
+      "step": 334
+    },
+    {
+      "epoch": 167.5,
+      "grad_norm": 16.788789749145508,
+      "learning_rate": 0.0008449069839236538,
+      "loss": 1.3254,
+      "step": 335
+    },
+    {
+      "epoch": 168.0,
+      "grad_norm": 12.021644592285156,
+      "learning_rate": 0.0008440311810370921,
+      "loss": 0.9685,
+      "step": 336
+    },
+    {
+      "epoch": 168.5,
+      "grad_norm": 7.517643928527832,
+      "learning_rate": 0.000843153369008515,
+      "loss": 1.2549,
+      "step": 337
+    },
+    {
+      "epoch": 169.0,
+      "grad_norm": 14.728811264038086,
+      "learning_rate": 0.0008422735529643444,
+      "loss": 1.0911,
+      "step": 338
+    },
+    {
+      "epoch": 169.5,
+      "grad_norm": 10.382369041442871,
+      "learning_rate": 0.000841391738042706,
+      "loss": 1.158,
+      "step": 339
+    },
+    {
+      "epoch": 170.0,
+      "grad_norm": 11.966410636901855,
+      "learning_rate": 0.0008405079293933986,
+      "loss": 1.1464,
+      "step": 340
+    },
+    {
+      "epoch": 170.5,
+      "grad_norm": 16.708616256713867,
+      "learning_rate": 0.0008396221321778645,
+      "loss": 1.136,
+      "step": 341
+    },
+    {
+      "epoch": 171.0,
+      "grad_norm": 17.043764114379883,
+      "learning_rate": 0.0008387343515691595,
+      "loss": 1.2479,
+      "step": 342
+    },
+    {
+      "epoch": 171.5,
+      "grad_norm": 20.736236572265625,
+      "learning_rate": 0.0008378445927519218,
+      "loss": 1.3321,
+      "step": 343
+    },
+    {
+      "epoch": 172.0,
+      "grad_norm": 23.245758056640625,
+      "learning_rate": 0.0008369528609223429,
+      "loss": 1.0732,
+      "step": 344
+    },
+    {
+      "epoch": 172.5,
+      "grad_norm": 18.068241119384766,
+      "learning_rate": 0.0008360591612881363,
+      "loss": 1.0567,
+      "step": 345
+    },
+    {
+      "epoch": 173.0,
+      "grad_norm": 18.97540283203125,
+      "learning_rate": 0.0008351634990685079,
+      "loss": 1.4729,
+      "step": 346
+    },
+    {
+      "epoch": 173.5,
+      "grad_norm": 22.397050857543945,
+      "learning_rate": 0.0008342658794941245,
+      "loss": 1.3259,
+      "step": 347
+    },
+    {
+      "epoch": 174.0,
+      "grad_norm": 19.437824249267578,
+      "learning_rate": 0.0008333663078070846,
+      "loss": 1.2383,
+      "step": 348
+    },
+    {
+      "epoch": 174.5,
+      "grad_norm": 24.972137451171875,
+      "learning_rate": 0.0008324647892608861,
+      "loss": 1.1771,
+      "step": 349
+    },
+    {
+      "epoch": 175.0,
+      "grad_norm": 27.837862014770508,
+      "learning_rate": 0.0008315613291203976,
+      "loss": 1.2783,
+      "step": 350
+    },
+    {
+      "epoch": 175.5,
+      "grad_norm": 13.65988826751709,
+      "learning_rate": 0.0008306559326618259,
+      "loss": 1.1396,
+      "step": 351
+    },
+    {
+      "epoch": 176.0,
+      "grad_norm": 18.38920021057129,
+      "learning_rate": 0.0008297486051726862,
+      "loss": 1.2326,
+      "step": 352
+    },
+    {
+      "epoch": 176.5,
+      "grad_norm": 16.228105545043945,
+      "learning_rate": 0.0008288393519517707,
+      "loss": 1.2494,
+      "step": 353
+    },
+    {
+      "epoch": 177.0,
+      "grad_norm": 18.1085147857666,
+      "learning_rate": 0.0008279281783091181,
+      "loss": 1.2178,
+      "step": 354
+    },
+    {
+      "epoch": 177.5,
+      "grad_norm": 24.951215744018555,
+      "learning_rate": 0.0008270150895659823,
+      "loss": 1.2215,
+      "step": 355
+    },
+    {
+      "epoch": 178.0,
+      "grad_norm": 16.52242088317871,
+      "learning_rate": 0.000826100091054801,
+      "loss": 1.115,
+      "step": 356
+    },
+    {
+      "epoch": 178.5,
+      "grad_norm": 15.718801498413086,
+      "learning_rate": 0.0008251831881191656,
+      "loss": 1.4289,
+      "step": 357
+    },
+    {
+      "epoch": 179.0,
+      "grad_norm": 20.691102981567383,
+      "learning_rate": 0.0008242643861137891,
+      "loss": 1.0356,
+      "step": 358
+    },
+    {
+      "epoch": 179.5,
+      "grad_norm": 9.117208480834961,
+      "learning_rate": 0.0008233436904044746,
+      "loss": 1.031,
+      "step": 359
+    },
+    {
+      "epoch": 180.0,
+      "grad_norm": 25.741392135620117,
+      "learning_rate": 0.0008224211063680853,
+      "loss": 1.279,
+      "step": 360
+    },
+    {
+      "epoch": 180.5,
+      "grad_norm": 14.685586929321289,
+      "learning_rate": 0.0008214966393925115,
+      "loss": 1.2295,
+      "step": 361
+    },
+    {
+      "epoch": 181.0,
+      "grad_norm": 19.009296417236328,
+      "learning_rate": 0.00082057029487664,
+      "loss": 1.0802,
+      "step": 362
+    },
+    {
+      "epoch": 181.5,
+      "grad_norm": 19.698888778686523,
+      "learning_rate": 0.0008196420782303229,
+      "loss": 1.3394,
+      "step": 363
+    },
+    {
+      "epoch": 182.0,
+      "grad_norm": 22.171974182128906,
+      "learning_rate": 0.0008187119948743449,
+      "loss": 1.1056,
+      "step": 364
+    },
+    {
+      "epoch": 182.5,
+      "grad_norm": 18.326642990112305,
+      "learning_rate": 0.0008177800502403927,
+      "loss": 1.0256,
+      "step": 365
+    },
+    {
+      "epoch": 183.0,
+      "grad_norm": 15.044474601745605,
+      "learning_rate": 0.0008168462497710226,
+      "loss": 1.4816,
+      "step": 366
+    },
+    {
+      "epoch": 183.5,
+      "grad_norm": 13.202553749084473,
+      "learning_rate": 0.0008159105989196294,
+      "loss": 1.2311,
+      "step": 367
+    },
+    {
+      "epoch": 184.0,
+      "grad_norm": 11.56424331665039,
+      "learning_rate": 0.0008149731031504135,
+      "loss": 1.2841,
+      "step": 368
+    },
+    {
+      "epoch": 184.5,
+      "grad_norm": 11.499375343322754,
+      "learning_rate": 0.0008140337679383502,
+      "loss": 1.1849,
+      "step": 369
+    },
+    {
+      "epoch": 185.0,
+      "grad_norm": 13.620736122131348,
+      "learning_rate": 0.0008130925987691568,
+      "loss": 1.1439,
+      "step": 370
+    },
+    {
+      "epoch": 185.5,
+      "grad_norm": 12.77872371673584,
+      "learning_rate": 0.0008121496011392613,
+      "loss": 0.9747,
+      "step": 371
+    },
+    {
+      "epoch": 186.0,
+      "grad_norm": 11.60251522064209,
+      "learning_rate": 0.0008112047805557692,
+      "loss": 1.1644,
+      "step": 372
+    },
+    {
+      "epoch": 186.5,
+      "grad_norm": 7.552931785583496,
+      "learning_rate": 0.0008102581425364329,
+      "loss": 1.0536,
+      "step": 373
+    },
+    {
+      "epoch": 187.0,
+      "grad_norm": 8.702902793884277,
+      "learning_rate": 0.0008093096926096177,
+      "loss": 1.3447,
+      "step": 374
+    },
+    {
+      "epoch": 187.5,
+      "grad_norm": 8.282918930053711,
+      "learning_rate": 0.0008083594363142716,
+      "loss": 1.0906,
+      "step": 375
+    },
+    {
+      "epoch": 188.0,
+      "grad_norm": 11.302000999450684,
+      "learning_rate": 0.0008074073791998906,
+      "loss": 1.1874,
+      "step": 376
+    },
+    {
+      "epoch": 188.5,
+      "grad_norm": 9.900726318359375,
+      "learning_rate": 0.0008064535268264883,
+      "loss": 1.1426,
+      "step": 377
+    },
+    {
+      "epoch": 189.0,
+      "grad_norm": 20.115741729736328,
+      "learning_rate": 0.0008054978847645622,
+      "loss": 0.9049,
+      "step": 378
+    },
+    {
+      "epoch": 189.5,
+      "grad_norm": 11.737709999084473,
+      "learning_rate": 0.0008045404585950617,
+      "loss": 1.0278,
+      "step": 379
+    },
+    {
+      "epoch": 190.0,
+      "grad_norm": 12.482132911682129,
+      "learning_rate": 0.0008035812539093556,
+      "loss": 1.1051,
+      "step": 380
+    },
+    {
+      "epoch": 190.5,
+      "grad_norm": 15.668073654174805,
+      "learning_rate": 0.000802620276309199,
+      "loss": 1.1258,
+      "step": 381
+    },
+    {
+      "epoch": 191.0,
+      "grad_norm": 10.978836059570312,
+      "learning_rate": 0.0008016575314067005,
+      "loss": 0.9933,
+      "step": 382
+    },
+    {
+      "epoch": 191.5,
+      "grad_norm": 13.691133499145508,
+      "learning_rate": 0.0008006930248242905,
+      "loss": 1.3695,
+      "step": 383
+    },
+    {
+      "epoch": 192.0,
+      "grad_norm": 18.32457160949707,
+      "learning_rate": 0.000799726762194687,
+      "loss": 0.8894,
+      "step": 384
+    },
+    {
+      "epoch": 192.5,
+      "grad_norm": 14.071953773498535,
+      "learning_rate": 0.0007987587491608637,
+      "loss": 1.1168,
+      "step": 385
+    },
+    {
+      "epoch": 193.0,
+      "grad_norm": 19.494342803955078,
+      "learning_rate": 0.0007977889913760163,
+      "loss": 0.9342,
+      "step": 386
+    },
+    {
+      "epoch": 193.5,
+      "grad_norm": 12.709486961364746,
+      "learning_rate": 0.0007968174945035301,
+      "loss": 1.1706,
+      "step": 387
+    },
+    {
+      "epoch": 194.0,
+      "grad_norm": 12.435921669006348,
+      "learning_rate": 0.0007958442642169468,
+      "loss": 1.2111,
+      "step": 388
+    },
+    {
+      "epoch": 194.5,
+      "grad_norm": 15.320987701416016,
+      "learning_rate": 0.0007948693061999306,
+      "loss": 1.2347,
+      "step": 389
+    },
+    {
+      "epoch": 195.0,
+      "grad_norm": 18.02007293701172,
+      "learning_rate": 0.0007938926261462366,
+      "loss": 0.8604,
+      "step": 390
+    },
+    {
+      "epoch": 195.5,
+      "grad_norm": 15.309648513793945,
+      "learning_rate": 0.0007929142297596756,
+      "loss": 0.8303,
+      "step": 391
+    },
+    {
+      "epoch": 196.0,
+      "grad_norm": 13.070034980773926,
+      "learning_rate": 0.0007919341227540828,
+      "loss": 1.2363,
+      "step": 392
+    },
+    {
+      "epoch": 196.5,
+      "grad_norm": 13.873756408691406,
+      "learning_rate": 0.0007909523108532824,
+      "loss": 1.1844,
+      "step": 393
+    },
+    {
+      "epoch": 197.0,
+      "grad_norm": 19.855224609375,
+      "learning_rate": 0.0007899687997910558,
+      "loss": 0.9823,
+      "step": 394
+    },
+    {
+      "epoch": 197.5,
+      "grad_norm": 9.651673316955566,
+      "learning_rate": 0.0007889835953111075,
+      "loss": 1.0085,
+      "step": 395
+    },
+    {
+      "epoch": 198.0,
+      "grad_norm": 16.118881225585938,
+      "learning_rate": 0.0007879967031670313,
+      "loss": 1.1873,
+      "step": 396
+    },
+    {
+      "epoch": 198.5,
+      "grad_norm": 12.0645170211792,
+      "learning_rate": 0.0007870081291222773,
+      "loss": 1.241,
+      "step": 397
+    },
+    {
+      "epoch": 199.0,
+      "grad_norm": 22.15360450744629,
+      "learning_rate": 0.0007860178789501172,
+      "loss": 0.8422,
+      "step": 398
+    },
+    {
+      "epoch": 199.5,
+      "grad_norm": 19.30936050415039,
+      "learning_rate": 0.0007850259584336124,
+      "loss": 0.8752,
+      "step": 399
+    },
+    {
+      "epoch": 200.0,
+      "grad_norm": 8.654556274414062,
+      "learning_rate": 0.0007840323733655779,
+      "loss": 1.3957,
+      "step": 400
+    },
+    {
+      "epoch": 200.5,
+      "grad_norm": 17.317325592041016,
+      "learning_rate": 0.0007830371295485507,
+      "loss": 1.2408,
+      "step": 401
+    },
+    {
+      "epoch": 201.0,
+      "grad_norm": 17.761459350585938,
+      "learning_rate": 0.0007820402327947542,
+      "loss": 0.8927,
+      "step": 402
+    },
+    {
+      "epoch": 201.5,
+      "grad_norm": 28.925222396850586,
+      "learning_rate": 0.0007810416889260653,
+      "loss": 1.0741,
+      "step": 403
+    },
+    {
+      "epoch": 202.0,
+      "grad_norm": 30.54914093017578,
+      "learning_rate": 0.0007800415037739801,
+      "loss": 1.0025,
+      "step": 404
+    },
+    {
+      "epoch": 202.5,
+      "grad_norm": 10.152922630310059,
+      "learning_rate": 0.0007790396831795791,
+      "loss": 0.9443,
+      "step": 405
+    },
+    {
+      "epoch": 203.0,
+      "grad_norm": 12.885197639465332,
+      "learning_rate": 0.0007780362329934951,
+      "loss": 1.2781,
+      "step": 406
+    },
+    {
+      "epoch": 203.5,
+      "grad_norm": 10.669059753417969,
+      "learning_rate": 0.0007770311590758765,
+      "loss": 1.121,
+      "step": 407
+    },
+    {
+      "epoch": 204.0,
+      "grad_norm": 25.019651412963867,
+      "learning_rate": 0.0007760244672963548,
+      "loss": 0.9944,
+      "step": 408
+    },
+    {
+      "epoch": 204.5,
+      "grad_norm": 13.973177909851074,
+      "learning_rate": 0.0007750161635340096,
+      "loss": 1.1843,
+      "step": 409
+    },
+    {
+      "epoch": 205.0,
+      "grad_norm": 20.85015106201172,
+      "learning_rate": 0.0007740062536773351,
+      "loss": 0.9116,
+      "step": 410
+    },
+    {
+      "epoch": 205.5,
+      "grad_norm": 13.778724670410156,
+      "learning_rate": 0.000772994743624204,
+      "loss": 1.1115,
+      "step": 411
+    },
+    {
+      "epoch": 206.0,
+      "grad_norm": 10.414752960205078,
+      "learning_rate": 0.0007719816392818353,
+      "loss": 0.8761,
+      "step": 412
+    },
+    {
+      "epoch": 206.5,
+      "grad_norm": 17.092060089111328,
+      "learning_rate": 0.0007709669465667578,
+      "loss": 1.109,
+      "step": 413
+    },
+    {
+      "epoch": 207.0,
+      "grad_norm": 13.398775100708008,
+      "learning_rate": 0.000769950671404777,
+      "loss": 0.895,
+      "step": 414
+    },
+    {
+      "epoch": 207.5,
+      "grad_norm": 13.47548770904541,
+      "learning_rate": 0.0007689328197309393,
+      "loss": 1.1908,
+      "step": 415
+    },
+    {
+      "epoch": 208.0,
+      "grad_norm": 10.889326095581055,
+      "learning_rate": 0.0007679133974894983,
+      "loss": 0.7725,
+      "step": 416
+    },
+    {
+      "epoch": 208.5,
+      "grad_norm": 28.72051239013672,
+      "learning_rate": 0.0007668924106338798,
+      "loss": 0.9402,
+      "step": 417
+    },
+    {
+      "epoch": 209.0,
+      "grad_norm": 20.7203311920166,
+      "learning_rate": 0.0007658698651266467,
+      "loss": 1.0503,
+      "step": 418
+    },
+    {
+      "epoch": 209.5,
+      "grad_norm": 12.22101879119873,
+      "learning_rate": 0.0007648457669394645,
+      "loss": 0.8768,
+      "step": 419
+    },
+    {
+      "epoch": 210.0,
+      "grad_norm": 23.04597282409668,
+      "learning_rate": 0.0007638201220530663,
+      "loss": 1.2628,
+      "step": 420
+    },
+    {
+      "epoch": 210.5,
+      "grad_norm": 10.055203437805176,
+      "learning_rate": 0.0007627929364572184,
+      "loss": 1.0497,
+      "step": 421
+    },
+    {
+      "epoch": 211.0,
+      "grad_norm": 18.847671508789062,
+      "learning_rate": 0.0007617642161506837,
+      "loss": 1.111,
+      "step": 422
+    },
+    {
+      "epoch": 211.5,
+      "grad_norm": 13.537520408630371,
+      "learning_rate": 0.0007607339671411893,
+      "loss": 0.9615,
+      "step": 423
+    },
+    {
+      "epoch": 212.0,
+      "grad_norm": 7.64473295211792,
+      "learning_rate": 0.0007597021954453886,
+      "loss": 1.0351,
+      "step": 424
+    },
+    {
+      "epoch": 212.5,
+      "grad_norm": 13.806032180786133,
+      "learning_rate": 0.0007586689070888284,
+      "loss": 1.0541,
+      "step": 425
+    },
+    {
+      "epoch": 213.0,
+      "grad_norm": 10.88693904876709,
+      "learning_rate": 0.0007576341081059123,
+      "loss": 0.9626,
+      "step": 426
+    },
+    {
+      "epoch": 213.5,
+      "grad_norm": 52.181732177734375,
+      "learning_rate": 0.0007565978045398663,
+      "loss": 1.1515,
+      "step": 427
+    },
+    {
+      "epoch": 214.0,
+      "grad_norm": 14.001968383789062,
+      "learning_rate": 0.0007555600024427027,
+      "loss": 0.7722,
+      "step": 428
+    },
+    {
+      "epoch": 214.5,
+      "grad_norm": 9.633649826049805,
+      "learning_rate": 0.0007545207078751857,
+      "loss": 1.152,
+      "step": 429
+    },
+    {
+      "epoch": 215.0,
+      "grad_norm": 15.116260528564453,
+      "learning_rate": 0.0007534799269067953,
+      "loss": 0.7705,
+      "step": 430
+    },
+    {
+      "epoch": 215.5,
+      "grad_norm": 13.952412605285645,
+      "learning_rate": 0.0007524376656156919,
+      "loss": 1.1232,
+      "step": 431
+    },
+    {
+      "epoch": 216.0,
+      "grad_norm": 21.036226272583008,
+      "learning_rate": 0.0007513939300886816,
+      "loss": 0.9027,
+      "step": 432
+    },
+    {
+      "epoch": 216.5,
+      "grad_norm": 9.916934967041016,
+      "learning_rate": 0.0007503487264211793,
+      "loss": 1.2218,
+      "step": 433
+    },
+    {
+      "epoch": 217.0,
+      "grad_norm": 12.98391056060791,
+      "learning_rate": 0.0007493020607171743,
+      "loss": 0.7828,
+      "step": 434
+    },
+    {
+      "epoch": 217.5,
+      "grad_norm": 13.72927474975586,
+      "learning_rate": 0.000748253939089194,
+      "loss": 1.0047,
+      "step": 435
+    },
+    {
+      "epoch": 218.0,
+      "grad_norm": 17.90583038330078,
+      "learning_rate": 0.0007472043676582685,
+      "loss": 1.0011,
+      "step": 436
+    },
+    {
+      "epoch": 218.5,
+      "grad_norm": 24.214080810546875,
+      "learning_rate": 0.0007461533525538947,
+      "loss": 1.0075,
+      "step": 437
+    },
+    {
+      "epoch": 219.0,
+      "grad_norm": 24.46935272216797,
+      "learning_rate": 0.0007451008999140005,
+      "loss": 0.9755,
+      "step": 438
+    },
+    {
+      "epoch": 219.5,
+      "grad_norm": 7.724076271057129,
+      "learning_rate": 0.0007440470158849091,
+      "loss": 1.114,
+      "step": 439
+    },
+    {
+      "epoch": 220.0,
+      "grad_norm": 22.03670310974121,
+      "learning_rate": 0.000742991706621303,
+      "loss": 0.8711,
+      "step": 440
+    },
+    {
+      "epoch": 220.5,
+      "grad_norm": 23.18583869934082,
+      "learning_rate": 0.0007419349782861881,
+      "loss": 0.8106,
+      "step": 441
+    },
+    {
+      "epoch": 221.0,
+      "grad_norm": 12.35851001739502,
+      "learning_rate": 0.0007408768370508576,
+      "loss": 1.0885,
+      "step": 442
+    },
+    {
+      "epoch": 221.5,
+      "grad_norm": 20.564716339111328,
+      "learning_rate": 0.000739817289094856,
+      "loss": 0.8728,
+      "step": 443
+    },
+    {
+      "epoch": 222.0,
+      "grad_norm": 21.132793426513672,
+      "learning_rate": 0.0007387563406059432,
+      "loss": 1.1393,
+      "step": 444
+    },
+    {
+      "epoch": 222.5,
+      "grad_norm": 10.72192096710205,
+      "learning_rate": 0.0007376939977800581,
+      "loss": 1.1019,
+      "step": 445
+    },
+    {
+      "epoch": 223.0,
+      "grad_norm": 25.528621673583984,
+      "learning_rate": 0.0007366302668212826,
+      "loss": 0.8652,
+      "step": 446
+    },
+    {
+      "epoch": 223.5,
+      "grad_norm": 17.804893493652344,
+      "learning_rate": 0.0007355651539418049,
+      "loss": 1.007,
+      "step": 447
+    },
+    {
+      "epoch": 224.0,
+      "grad_norm": 17.15072250366211,
+      "learning_rate": 0.0007344986653618844,
+      "loss": 1.1148,
+      "step": 448
+    },
+    {
+      "epoch": 224.5,
+      "grad_norm": 12.136202812194824,
+      "learning_rate": 0.0007334308073098139,
+      "loss": 0.974,
+      "step": 449
+    },
+    {
+      "epoch": 225.0,
+      "grad_norm": 10.720677375793457,
+      "learning_rate": 0.0007323615860218843,
+      "loss": 0.9262,
+      "step": 450
+    },
+    {
+      "epoch": 225.5,
+      "grad_norm": 17.299392700195312,
+      "learning_rate": 0.0007312910077423476,
+      "loss": 1.0953,
+      "step": 451
+    },
+    {
+      "epoch": 226.0,
+      "grad_norm": 16.680747985839844,
+      "learning_rate": 0.0007302190787233807,
+      "loss": 0.8696,
+      "step": 452
+    },
+    {
+      "epoch": 226.5,
+      "grad_norm": 18.39723014831543,
+      "learning_rate": 0.0007291458052250488,
+      "loss": 1.2192,
+      "step": 453
+    },
+    {
+      "epoch": 227.0,
+      "grad_norm": 78.7840576171875,
+      "learning_rate": 0.0007280711935152691,
+      "loss": 1.0245,
+      "step": 454
+    },
+    {
+      "epoch": 227.5,
+      "grad_norm": 16.80086326599121,
+      "learning_rate": 0.0007269952498697733,
+      "loss": 1.0091,
+      "step": 455
+    },
+    {
+      "epoch": 228.0,
+      "grad_norm": 22.319021224975586,
+      "learning_rate": 0.0007259179805720726,
+      "loss": 0.9518,
+      "step": 456
+    },
+    {
+      "epoch": 228.5,
+      "grad_norm": 20.36347007751465,
+      "learning_rate": 0.0007248393919134188,
+      "loss": 1.0327,
+      "step": 457
+    },
+    {
+      "epoch": 229.0,
+      "grad_norm": 18.762556076049805,
+      "learning_rate": 0.0007237594901927699,
+      "loss": 0.9982,
+      "step": 458
+    },
+    {
+      "epoch": 229.5,
+      "grad_norm": 30.659494400024414,
+      "learning_rate": 0.0007226782817167514,
+      "loss": 1.0825,
+      "step": 459
+    },
+    {
+      "epoch": 230.0,
+      "grad_norm": 23.812955856323242,
+      "learning_rate": 0.0007215957727996207,
+      "loss": 0.9434,
+      "step": 460
+    },
+    {
+      "epoch": 230.5,
+      "grad_norm": 14.057976722717285,
+      "learning_rate": 0.0007205119697632296,
+      "loss": 1.1124,
+      "step": 461
+    },
+    {
+      "epoch": 231.0,
+      "grad_norm": 20.08623695373535,
+      "learning_rate": 0.0007194268789369874,
+      "loss": 0.8576,
+      "step": 462
+    },
+    {
+      "epoch": 231.5,
+      "grad_norm": 18.500167846679688,
+      "learning_rate": 0.0007183405066578247,
+      "loss": 1.0948,
+      "step": 463
+    },
+    {
+      "epoch": 232.0,
+      "grad_norm": 16.56690788269043,
+      "learning_rate": 0.000717252859270155,
+      "loss": 0.8664,
+      "step": 464
+    },
+    {
+      "epoch": 232.5,
+      "grad_norm": 19.24199867248535,
+      "learning_rate": 0.0007161639431258386,
+      "loss": 0.8906,
+      "step": 465
+    },
+    {
+      "epoch": 233.0,
+      "grad_norm": 18.461132049560547,
+      "learning_rate": 0.000715073764584146,
+      "loss": 1.5437,
+      "step": 466
+    },
+    {
+      "epoch": 233.5,
+      "grad_norm": 8.657334327697754,
+      "learning_rate": 0.0007139823300117193,
+      "loss": 0.938,
+      "step": 467
+    },
+    {
+      "epoch": 234.0,
+      "grad_norm": 25.8443660736084,
+      "learning_rate": 0.0007128896457825364,
+      "loss": 0.9837,
+      "step": 468
+    },
+    {
+      "epoch": 234.5,
+      "grad_norm": 23.247228622436523,
+      "learning_rate": 0.0007117957182778727,
+      "loss": 1.0011,
+      "step": 469
+    },
+    {
+      "epoch": 235.0,
+      "grad_norm": 12.796981811523438,
+      "learning_rate": 0.0007107005538862646,
+      "loss": 0.9159,
+      "step": 470
+    },
+    {
+      "epoch": 235.5,
+      "grad_norm": 12.516453742980957,
+      "learning_rate": 0.0007096041590034719,
+      "loss": 0.9105,
+      "step": 471
+    },
+    {
+      "epoch": 236.0,
+      "grad_norm": 9.54037857055664,
+      "learning_rate": 0.0007085065400324407,
+      "loss": 1.0987,
+      "step": 472
+    },
+    {
+      "epoch": 236.5,
+      "grad_norm": 11.162910461425781,
+      "learning_rate": 0.0007074077033832652,
+      "loss": 0.6908,
+      "step": 473
+    },
+    {
+      "epoch": 237.0,
+      "grad_norm": 9.226560592651367,
+      "learning_rate": 0.0007063076554731512,
+      "loss": 1.126,
+      "step": 474
+    },
+    {
+      "epoch": 237.5,
+      "grad_norm": 8.407577514648438,
+      "learning_rate": 0.0007052064027263785,
+      "loss": 0.8868,
+      "step": 475
+    },
+    {
+      "epoch": 238.0,
+      "grad_norm": 13.981208801269531,
+      "learning_rate": 0.0007041039515742625,
+      "loss": 0.8783,
+      "step": 476
+    },
+    {
+      "epoch": 238.5,
+      "grad_norm": 9.400607109069824,
+      "learning_rate": 0.0007030003084551181,
+      "loss": 0.8821,
+      "step": 477
+    },
+    {
+      "epoch": 239.0,
+      "grad_norm": 13.614973068237305,
+      "learning_rate": 0.0007018954798142204,
+      "loss": 0.9008,
+      "step": 478
+    },
+    {
+      "epoch": 239.5,
+      "grad_norm": 7.8317551612854,
+      "learning_rate": 0.0007007894721037688,
+      "loss": 0.9235,
+      "step": 479
+    },
+    {
+      "epoch": 240.0,
+      "grad_norm": 14.0414457321167,
+      "learning_rate": 0.0006996822917828477,
+      "loss": 0.8601,
+      "step": 480
+    },
+    {
+      "epoch": 240.5,
+      "grad_norm": 9.88857364654541,
+      "learning_rate": 0.0006985739453173903,
+      "loss": 0.8393,
+      "step": 481
+    },
+    {
+      "epoch": 241.0,
+      "grad_norm": 18.54656982421875,
+      "learning_rate": 0.0006974644391801395,
+      "loss": 0.9919,
+      "step": 482
+    },
+    {
+      "epoch": 241.5,
+      "grad_norm": 14.761321067810059,
+      "learning_rate": 0.0006963537798506108,
+      "loss": 0.8549,
+      "step": 483
+    },
+    {
+      "epoch": 242.0,
+      "grad_norm": 12.16065788269043,
+      "learning_rate": 0.0006952419738150545,
+      "loss": 0.9851,
+      "step": 484
+    },
+    {
+      "epoch": 242.5,
+      "grad_norm": 14.52094554901123,
+      "learning_rate": 0.0006941290275664175,
+      "loss": 1.0646,
+      "step": 485
+    },
+    {
+      "epoch": 243.0,
+      "grad_norm": 11.686137199401855,
+      "learning_rate": 0.0006930149476043058,
+      "loss": 0.6886,
+      "step": 486
+    },
+    {
+      "epoch": 243.5,
+      "grad_norm": 10.54874324798584,
+      "learning_rate": 0.0006918997404349459,
+      "loss": 0.8949,
+      "step": 487
+    },
+    {
+      "epoch": 244.0,
+      "grad_norm": 10.815678596496582,
+      "learning_rate": 0.0006907834125711476,
+      "loss": 0.7689,
+      "step": 488
+    },
+    {
+      "epoch": 244.5,
+      "grad_norm": 12.013069152832031,
+      "learning_rate": 0.0006896659705322653,
+      "loss": 0.9847,
+      "step": 489
+    },
+    {
+      "epoch": 245.0,
+      "grad_norm": 25.461071014404297,
+      "learning_rate": 0.0006885474208441603,
+      "loss": 1.2144,
+      "step": 490
+    },
+    {
+      "epoch": 245.5,
+      "grad_norm": 15.028848648071289,
+      "learning_rate": 0.0006874277700391622,
+      "loss": 0.9946,
+      "step": 491
+    },
+    {
+      "epoch": 246.0,
+      "grad_norm": 18.335084915161133,
+      "learning_rate": 0.0006863070246560319,
+      "loss": 0.9055,
+      "step": 492
+    },
+    {
+      "epoch": 246.5,
+      "grad_norm": 20.676258087158203,
+      "learning_rate": 0.0006851851912399218,
+      "loss": 1.0797,
+      "step": 493
+    },
+    {
+      "epoch": 247.0,
+      "grad_norm": 49.90549087524414,
+      "learning_rate": 0.0006840622763423391,
+      "loss": 0.8035,
+      "step": 494
+    },
+    {
+      "epoch": 247.5,
+      "grad_norm": 15.933221817016602,
+      "learning_rate": 0.0006829382865211062,
+      "loss": 0.9064,
+      "step": 495
+    },
+    {
+      "epoch": 248.0,
+      "grad_norm": 17.790117263793945,
+      "learning_rate": 0.0006818132283403235,
+      "loss": 1.1309,
+      "step": 496
+    },
+    {
+      "epoch": 248.5,
+      "grad_norm": 13.757402420043945,
+      "learning_rate": 0.0006806871083703306,
+      "loss": 0.8081,
+      "step": 497
+    },
+    {
+      "epoch": 249.0,
+      "grad_norm": 18.95448875427246,
+      "learning_rate": 0.0006795599331876678,
+      "loss": 1.5002,
+      "step": 498
+    },
+    {
+      "epoch": 249.5,
+      "grad_norm": 15.33133602142334,
+      "learning_rate": 0.000678431709375038,
+      "loss": 1.0809,
+      "step": 499
+    },
+    {
+      "epoch": 250.0,
+      "grad_norm": 20.36852264404297,
+      "learning_rate": 0.0006773024435212678,
+      "loss": 0.7638,
+      "step": 500
+    },
+    {
+      "epoch": 250.5,
+      "grad_norm": 12.749167442321777,
+      "learning_rate": 0.0006761721422212696,
+      "loss": 0.682,
+      "step": 501
+    },
+    {
+      "epoch": 251.0,
+      "grad_norm": 8.280853271484375,
+      "learning_rate": 0.0006750408120760029,
+      "loss": 1.2348,
+      "step": 502
+    },
+    {
+      "epoch": 251.5,
+      "grad_norm": 9.779509544372559,
+      "learning_rate": 0.0006739084596924352,
+      "loss": 0.9073,
+      "step": 503
+    },
+    {
+      "epoch": 252.0,
+      "grad_norm": 11.670598030090332,
+      "learning_rate": 0.0006727750916835043,
+      "loss": 0.9868,
+      "step": 504
+    },
+    {
+      "epoch": 252.5,
+      "grad_norm": 10.319906234741211,
+      "learning_rate": 0.0006716407146680792,
+      "loss": 0.9016,
+      "step": 505
+    },
+    {
+      "epoch": 253.0,
+      "grad_norm": 8.60800552368164,
+      "learning_rate": 0.0006705053352709212,
+      "loss": 0.7661,
+      "step": 506
+    },
+    {
+      "epoch": 253.5,
+      "grad_norm": 11.923989295959473,
+      "learning_rate": 0.0006693689601226458,
+      "loss": 0.8448,
+      "step": 507
+    },
+    {
+      "epoch": 254.0,
+      "grad_norm": 12.463971138000488,
+      "learning_rate": 0.0006682315958596836,
+      "loss": 0.9236,
+      "step": 508
+    },
+    {
+      "epoch": 254.5,
+      "grad_norm": 11.198707580566406,
+      "learning_rate": 0.0006670932491242414,
+      "loss": 0.909,
+      "step": 509
+    },
+    {
+      "epoch": 255.0,
+      "grad_norm": 16.650724411010742,
+      "learning_rate": 0.0006659539265642643,
+      "loss": 0.9137,
+      "step": 510
+    },
+    {
+      "epoch": 255.5,
+      "grad_norm": 12.244513511657715,
+      "learning_rate": 0.0006648136348333954,
+      "loss": 0.7723,
+      "step": 511
+    },
+    {
+      "epoch": 256.0,
+      "grad_norm": 10.562124252319336,
+      "learning_rate": 0.0006636723805909383,
+      "loss": 1.0124,
+      "step": 512
+    },
+    {
+      "epoch": 256.5,
+      "grad_norm": 10.841675758361816,
+      "learning_rate": 0.0006625301705018178,
+      "loss": 1.0099,
+      "step": 513
+    },
+    {
+      "epoch": 257.0,
+      "grad_norm": 15.727170944213867,
+      "learning_rate": 0.0006613870112365398,
+      "loss": 0.9278,
+      "step": 514
+    },
+    {
+      "epoch": 257.5,
+      "grad_norm": 18.722354888916016,
+      "learning_rate": 0.0006602429094711549,
+      "loss": 0.859,
+      "step": 515
+    },
+    {
+      "epoch": 258.0,
+      "grad_norm": 10.05528736114502,
+      "learning_rate": 0.0006590978718872166,
+      "loss": 0.8563,
+      "step": 516
+    },
+    {
+      "epoch": 258.5,
+      "grad_norm": 13.065600395202637,
+      "learning_rate": 0.0006579519051717445,
+      "loss": 0.8833,
+      "step": 517
+    },
+    {
+      "epoch": 259.0,
+      "grad_norm": 16.173564910888672,
+      "learning_rate": 0.0006568050160171837,
+      "loss": 0.7301,
+      "step": 518
+    },
+    {
+      "epoch": 259.5,
+      "grad_norm": 15.837112426757812,
+      "learning_rate": 0.0006556572111213666,
+      "loss": 0.988,
+      "step": 519
+    },
+    {
+      "epoch": 260.0,
+      "grad_norm": 14.83794116973877,
+      "learning_rate": 0.0006545084971874737,
+      "loss": 0.6521,
+      "step": 520
+    },
+    {
+      "epoch": 260.5,
+      "grad_norm": 13.650358200073242,
+      "learning_rate": 0.0006533588809239941,
+      "loss": 0.7257,
+      "step": 521
+    },
+    {
+      "epoch": 261.0,
+      "grad_norm": 11.35278606414795,
+      "learning_rate": 0.0006522083690446862,
+      "loss": 1.0977,
+      "step": 522
+    },
+    {
+      "epoch": 261.5,
+      "grad_norm": 15.753032684326172,
+      "learning_rate": 0.0006510569682685394,
+      "loss": 0.8892,
+      "step": 523
+    },
+    {
+      "epoch": 262.0,
+      "grad_norm": 15.767334938049316,
+      "learning_rate": 0.0006499046853197338,
+      "loss": 0.8733,
+      "step": 524
+    },
+    {
+      "epoch": 262.5,
+      "grad_norm": 10.346399307250977,
+      "learning_rate": 0.0006487515269276015,
+      "loss": 0.9843,
+      "step": 525
+    },
+    {
+      "epoch": 263.0,
+      "grad_norm": 14.375774383544922,
+      "learning_rate": 0.0006475974998265874,
+      "loss": 0.7134,
+      "step": 526
+    },
+    {
+      "epoch": 263.5,
+      "grad_norm": 8.106711387634277,
+      "learning_rate": 0.0006464426107562095,
+      "loss": 0.8566,
+      "step": 527
+    },
+    {
+      "epoch": 264.0,
+      "grad_norm": 15.848772048950195,
+      "learning_rate": 0.0006452868664610196,
+      "loss": 0.7647,
+      "step": 528
+    },
+    {
+      "epoch": 264.5,
+      "grad_norm": 8.460439682006836,
+      "learning_rate": 0.0006441302736905641,
+      "loss": 0.8326,
+      "step": 529
+    },
+    {
+      "epoch": 265.0,
+      "grad_norm": 10.65987491607666,
+      "learning_rate": 0.0006429728391993446,
+      "loss": 0.6895,
+      "step": 530
+    },
+    {
+      "epoch": 265.5,
+      "grad_norm": 8.564724922180176,
+      "learning_rate": 0.0006418145697467783,
+      "loss": 0.814,
+      "step": 531
+    },
+    {
+      "epoch": 266.0,
+      "grad_norm": 18.43366241455078,
+      "learning_rate": 0.0006406554720971582,
+      "loss": 0.7862,
+      "step": 532
+    },
+    {
+      "epoch": 266.5,
+      "grad_norm": 11.917065620422363,
+      "learning_rate": 0.0006394955530196147,
+      "loss": 0.8101,
+      "step": 533
+    },
+    {
+      "epoch": 267.0,
+      "grad_norm": 12.14805793762207,
+      "learning_rate": 0.0006383348192880747,
+      "loss": 0.894,
+      "step": 534
+    },
+    {
+      "epoch": 267.5,
+      "grad_norm": 9.563876152038574,
+      "learning_rate": 0.000637173277681223,
+      "loss": 0.7676,
+      "step": 535
+    },
+    {
+      "epoch": 268.0,
+      "grad_norm": 13.440650939941406,
+      "learning_rate": 0.0006360109349824621,
+      "loss": 0.8504,
+      "step": 536
+    },
+    {
+      "epoch": 268.5,
+      "grad_norm": 15.660651206970215,
+      "learning_rate": 0.0006348477979798735,
+      "loss": 0.6358,
+      "step": 537
+    },
+    {
+      "epoch": 269.0,
+      "grad_norm": 18.38522720336914,
+      "learning_rate": 0.0006336838734661765,
+      "loss": 1.1614,
+      "step": 538
+    },
+    {
+      "epoch": 269.5,
+      "grad_norm": 15.530710220336914,
+      "learning_rate": 0.0006325191682386906,
+      "loss": 0.7249,
+      "step": 539
+    },
+    {
+      "epoch": 270.0,
+      "grad_norm": 15.080126762390137,
+      "learning_rate": 0.0006313536890992934,
+      "loss": 0.9466,
+      "step": 540
+    },
+    {
+      "epoch": 270.5,
+      "grad_norm": 11.354280471801758,
+      "learning_rate": 0.0006301874428543833,
+      "loss": 0.8383,
+      "step": 541
+    },
+    {
+      "epoch": 271.0,
+      "grad_norm": 10.717304229736328,
+      "learning_rate": 0.000629020436314838,
+      "loss": 0.9311,
+      "step": 542
+    },
+    {
+      "epoch": 271.5,
+      "grad_norm": 7.795917510986328,
+      "learning_rate": 0.0006278526762959751,
+      "loss": 0.9535,
+      "step": 543
+    },
+    {
+      "epoch": 272.0,
+      "grad_norm": 10.336143493652344,
+      "learning_rate": 0.0006266841696175131,
+      "loss": 0.6794,
+      "step": 544
+    },
+    {
+      "epoch": 272.5,
+      "grad_norm": 12.824676513671875,
+      "learning_rate": 0.0006255149231035309,
+      "loss": 0.7382,
+      "step": 545
+    },
+    {
+      "epoch": 273.0,
+      "grad_norm": 12.608689308166504,
+      "learning_rate": 0.0006243449435824276,
+      "loss": 0.7734,
+      "step": 546
+    },
+    {
+      "epoch": 273.5,
+      "grad_norm": 9.374356269836426,
+      "learning_rate": 0.0006231742378868831,
+      "loss": 0.9028,
+      "step": 547
+    },
+    {
+      "epoch": 274.0,
+      "grad_norm": 9.544020652770996,
+      "learning_rate": 0.0006220028128538187,
+      "loss": 0.6182,
+      "step": 548
+    },
+    {
+      "epoch": 274.5,
+      "grad_norm": 11.945455551147461,
+      "learning_rate": 0.0006208306753243564,
+      "loss": 0.753,
+      "step": 549
+    },
+    {
+      "epoch": 275.0,
+      "grad_norm": 12.08008861541748,
+      "learning_rate": 0.0006196578321437789,
+      "loss": 0.8524,
+      "step": 550
+    },
+    {
+      "epoch": 275.5,
+      "grad_norm": 8.308807373046875,
+      "learning_rate": 0.0006184842901614901,
+      "loss": 0.8412,
+      "step": 551
+    },
+    {
+      "epoch": 276.0,
+      "grad_norm": 11.580907821655273,
+      "learning_rate": 0.000617310056230975,
+      "loss": 0.6904,
+      "step": 552
+    },
+    {
+      "epoch": 276.5,
+      "grad_norm": 11.722846031188965,
+      "learning_rate": 0.0006161351372097596,
+      "loss": 0.6396,
+      "step": 553
+    },
+    {
+      "epoch": 277.0,
+      "grad_norm": 8.685089111328125,
+      "learning_rate": 0.0006149595399593703,
+      "loss": 1.016,
+      "step": 554
+    },
+    {
+      "epoch": 277.5,
+      "grad_norm": 9.909539222717285,
+      "learning_rate": 0.000613783271345295,
+      "loss": 0.9183,
+      "step": 555
+    },
+    {
+      "epoch": 278.0,
+      "grad_norm": 13.674391746520996,
+      "learning_rate": 0.0006126063382369423,
+      "loss": 0.7197,
+      "step": 556
+    },
+    {
+      "epoch": 278.5,
+      "grad_norm": 12.364531517028809,
+      "learning_rate": 0.0006114287475076009,
+      "loss": 0.8089,
+      "step": 557
+    },
+    {
+      "epoch": 279.0,
+      "grad_norm": 14.289377212524414,
+      "learning_rate": 0.0006102505060344006,
+      "loss": 0.7363,
+      "step": 558
+    },
+    {
+      "epoch": 279.5,
+      "grad_norm": 8.402495384216309,
+      "learning_rate": 0.0006090716206982714,
+      "loss": 0.6943,
+      "step": 559
+    },
+    {
+      "epoch": 280.0,
+      "grad_norm": 11.390583038330078,
+      "learning_rate": 0.0006078920983839031,
+      "loss": 0.9519,
+      "step": 560
+    },
+    {
+      "epoch": 280.5,
+      "grad_norm": 7.343291759490967,
+      "learning_rate": 0.000606711945979706,
+      "loss": 0.8611,
+      "step": 561
+    },
+    {
+      "epoch": 281.0,
+      "grad_norm": 10.02965259552002,
+      "learning_rate": 0.0006055311703777698,
+      "loss": 0.623,
+      "step": 562
+    },
+    {
+      "epoch": 281.5,
+      "grad_norm": 17.226282119750977,
+      "learning_rate": 0.0006043497784738237,
+      "loss": 0.7347,
+      "step": 563
+    },
+    {
+      "epoch": 282.0,
+      "grad_norm": 9.059028625488281,
+      "learning_rate": 0.0006031677771671962,
+      "loss": 0.671,
+      "step": 564
+    },
+    {
+      "epoch": 282.5,
+      "grad_norm": 7.1017913818359375,
+      "learning_rate": 0.0006019851733607744,
+      "loss": 0.932,
+      "step": 565
+    },
+    {
+      "epoch": 283.0,
+      "grad_norm": 10.799386024475098,
+      "learning_rate": 0.0006008019739609646,
+      "loss": 0.5403,
+      "step": 566
+    },
+    {
+      "epoch": 283.5,
+      "grad_norm": 8.003046035766602,
+      "learning_rate": 0.0005996181858776507,
+      "loss": 0.8054,
+      "step": 567
+    },
+    {
+      "epoch": 284.0,
+      "grad_norm": 8.339454650878906,
+      "learning_rate": 0.0005984338160241551,
+      "loss": 0.674,
+      "step": 568
+    },
+    {
+      "epoch": 284.5,
+      "grad_norm": 9.353204727172852,
+      "learning_rate": 0.0005972488713171974,
+      "loss": 0.6148,
+      "step": 569
+    },
+    {
+      "epoch": 285.0,
+      "grad_norm": 8.813953399658203,
+      "learning_rate": 0.0005960633586768543,
+      "loss": 0.768,
+      "step": 570
+    },
+    {
+      "epoch": 285.5,
+      "grad_norm": 21.869308471679688,
+      "learning_rate": 0.0005948772850265193,
+      "loss": 0.9291,
+      "step": 571
+    },
+    {
+      "epoch": 286.0,
+      "grad_norm": 13.392123222351074,
+      "learning_rate": 0.0005936906572928624,
+      "loss": 0.6294,
+      "step": 572
+    },
+    {
+      "epoch": 286.5,
+      "grad_norm": 15.045798301696777,
+      "learning_rate": 0.0005925034824057892,
+      "loss": 0.8838,
+      "step": 573
+    },
+    {
+      "epoch": 287.0,
+      "grad_norm": 12.13925552368164,
+      "learning_rate": 0.0005913157672984006,
+      "loss": 0.5734,
+      "step": 574
+    },
+    {
+      "epoch": 287.5,
+      "grad_norm": 13.351587295532227,
+      "learning_rate": 0.000590127518906953,
+      "loss": 0.7608,
+      "step": 575
+    },
+    {
+      "epoch": 288.0,
+      "grad_norm": 15.696539878845215,
+      "learning_rate": 0.0005889387441708161,
+      "loss": 0.6037,
+      "step": 576
+    },
+    {
+      "epoch": 288.5,
+      "grad_norm": 9.399062156677246,
+      "learning_rate": 0.0005877494500324346,
+      "loss": 0.8791,
+      "step": 577
+    },
+    {
+      "epoch": 289.0,
+      "grad_norm": 13.130828857421875,
+      "learning_rate": 0.0005865596434372857,
+      "loss": 0.6324,
+      "step": 578
+    },
+    {
+      "epoch": 289.5,
+      "grad_norm": 11.723708152770996,
+      "learning_rate": 0.0005853693313338399,
+      "loss": 0.824,
+      "step": 579
+    },
+    {
+      "epoch": 290.0,
+      "grad_norm": 17.031309127807617,
+      "learning_rate": 0.0005841785206735191,
+      "loss": 0.695,
+      "step": 580
+    },
+    {
+      "epoch": 290.5,
+      "grad_norm": 16.940290451049805,
+      "learning_rate": 0.0005829872184106579,
+      "loss": 0.7475,
+      "step": 581
+    },
+    {
+      "epoch": 291.0,
+      "grad_norm": 10.51325511932373,
+      "learning_rate": 0.0005817954315024609,
+      "loss": 0.7064,
+      "step": 582
+    },
+    {
+      "epoch": 291.5,
+      "grad_norm": 15.263915061950684,
+      "learning_rate": 0.0005806031669089634,
+      "loss": 0.5725,
+      "step": 583
+    },
+    {
+      "epoch": 292.0,
+      "grad_norm": 18.890596389770508,
+      "learning_rate": 0.0005794104315929903,
+      "loss": 1.2188,
+      "step": 584
+    },
+    {
+      "epoch": 292.5,
+      "grad_norm": 10.702483177185059,
+      "learning_rate": 0.0005782172325201155,
+      "loss": 0.6278,
+      "step": 585
+    },
+    {
+      "epoch": 293.0,
+      "grad_norm": 12.502796173095703,
+      "learning_rate": 0.0005770235766586215,
+      "loss": 0.8329,
+      "step": 586
+    },
+    {
+      "epoch": 293.5,
+      "grad_norm": 13.095947265625,
+      "learning_rate": 0.0005758294709794579,
+      "loss": 0.8683,
+      "step": 587
+    },
+    {
+      "epoch": 294.0,
+      "grad_norm": 18.511831283569336,
+      "learning_rate": 0.0005746349224562021,
+      "loss": 0.9004,
+      "step": 588
+    },
+    {
+      "epoch": 294.5,
+      "grad_norm": 16.944067001342773,
+      "learning_rate": 0.0005734399380650167,
+      "loss": 0.9079,
+      "step": 589
+    },
+    {
+      "epoch": 295.0,
+      "grad_norm": 14.659197807312012,
+      "learning_rate": 0.0005722445247846107,
+      "loss": 0.7,
+      "step": 590
+    },
+    {
+      "epoch": 295.5,
+      "grad_norm": 11.728489875793457,
+      "learning_rate": 0.0005710486895961971,
+      "loss": 0.9632,
+      "step": 591
+    },
+    {
+      "epoch": 296.0,
+      "grad_norm": 16.602502822875977,
+      "learning_rate": 0.000569852439483453,
+      "loss": 0.7472,
+      "step": 592
+    },
+    {
+      "epoch": 296.5,
+      "grad_norm": 10.83639907836914,
+      "learning_rate": 0.0005686557814324791,
+      "loss": 0.891,
+      "step": 593
+    },
+    {
+      "epoch": 297.0,
+      "grad_norm": 12.513603210449219,
+      "learning_rate": 0.0005674587224317579,
+      "loss": 0.6303,
+      "step": 594
+    },
+    {
+      "epoch": 297.5,
+      "grad_norm": 9.863957405090332,
+      "learning_rate": 0.0005662612694721139,
+      "loss": 0.7015,
+      "step": 595
+    },
+    {
+      "epoch": 298.0,
+      "grad_norm": 11.755051612854004,
+      "learning_rate": 0.0005650634295466716,
+      "loss": 0.7662,
+      "step": 596
+    },
+    {
+      "epoch": 298.5,
+      "grad_norm": 8.038734436035156,
+      "learning_rate": 0.0005638652096508165,
+      "loss": 0.8678,
+      "step": 597
+    },
+    {
+      "epoch": 299.0,
+      "grad_norm": 12.914886474609375,
+      "learning_rate": 0.0005626666167821521,
+      "loss": 0.6479,
+      "step": 598
+    },
+    {
+      "epoch": 299.5,
+      "grad_norm": 13.228520393371582,
+      "learning_rate": 0.0005614676579404607,
+      "loss": 0.7166,
+      "step": 599
+    },
+    {
+      "epoch": 300.0,
+      "grad_norm": 12.135201454162598,
+      "learning_rate": 0.0005602683401276614,
+      "loss": 0.8311,
+      "step": 600
+    },
+    {
+      "epoch": 300.5,
+      "grad_norm": 13.10306167602539,
+      "learning_rate": 0.0005590686703477705,
+      "loss": 0.7768,
+      "step": 601
+    },
+    {
+      "epoch": 301.0,
+      "grad_norm": 12.659260749816895,
+      "learning_rate": 0.0005578686556068585,
+      "loss": 0.7353,
+      "step": 602
+    },
+    {
+      "epoch": 301.5,
+      "grad_norm": 13.3078031539917,
+      "learning_rate": 0.0005566683029130115,
+      "loss": 0.692,
+      "step": 603
+    },
+    {
+      "epoch": 302.0,
+      "grad_norm": 12.916741371154785,
+      "learning_rate": 0.0005554676192762891,
+      "loss": 0.7299,
+      "step": 604
+    },
+    {
+      "epoch": 302.5,
+      "grad_norm": 8.21851634979248,
+      "learning_rate": 0.0005542666117086831,
+      "loss": 0.6872,
+      "step": 605
+    },
+    {
+      "epoch": 303.0,
+      "grad_norm": 24.944929122924805,
+      "learning_rate": 0.0005530652872240779,
+      "loss": 1.1807,
+      "step": 606
+    },
+    {
+      "epoch": 303.5,
+      "grad_norm": 24.231395721435547,
+      "learning_rate": 0.0005518636528382078,
+      "loss": 0.8926,
+      "step": 607
+    },
+    {
+      "epoch": 304.0,
+      "grad_norm": 14.253700256347656,
+      "learning_rate": 0.0005506617155686176,
+      "loss": 0.7283,
+      "step": 608
+    },
+    {
+      "epoch": 304.5,
+      "grad_norm": 14.889595031738281,
+      "learning_rate": 0.0005494594824346204,
+      "loss": 0.8387,
+      "step": 609
+    },
+    {
+      "epoch": 305.0,
+      "grad_norm": 19.897613525390625,
+      "learning_rate": 0.0005482569604572576,
+      "loss": 0.6447,
+      "step": 610
+    },
+    {
+      "epoch": 305.5,
+      "grad_norm": 9.73555850982666,
+      "learning_rate": 0.0005470541566592572,
+      "loss": 0.5864,
+      "step": 611
+    },
+    {
+      "epoch": 306.0,
+      "grad_norm": 11.731720924377441,
+      "learning_rate": 0.0005458510780649931,
+      "loss": 0.8982,
+      "step": 612
+    },
+    {
+      "epoch": 306.5,
+      "grad_norm": 9.136198043823242,
+      "learning_rate": 0.000544647731700444,
+      "loss": 0.6414,
+      "step": 613
+    },
+    {
+      "epoch": 307.0,
+      "grad_norm": 12.03184986114502,
+      "learning_rate": 0.0005434441245931524,
+      "loss": 0.9304,
+      "step": 614
+    },
+    {
+      "epoch": 307.5,
+      "grad_norm": 10.764842987060547,
+      "learning_rate": 0.0005422402637721836,
+      "loss": 0.6463,
+      "step": 615
+    },
+    {
+      "epoch": 308.0,
+      "grad_norm": 6.806069850921631,
+      "learning_rate": 0.0005410361562680842,
+      "loss": 0.7381,
+      "step": 616
+    },
+    {
+      "epoch": 308.5,
+      "grad_norm": 8.290765762329102,
+      "learning_rate": 0.0005398318091128422,
+      "loss": 0.5271,
+      "step": 617
+    },
+    {
+      "epoch": 309.0,
+      "grad_norm": 7.3509321212768555,
+      "learning_rate": 0.0005386272293398444,
+      "loss": 0.9144,
+      "step": 618
+    },
+    {
+      "epoch": 309.5,
+      "grad_norm": 8.734230041503906,
+      "learning_rate": 0.0005374224239838365,
+      "loss": 0.5191,
+      "step": 619
+    },
+    {
+      "epoch": 310.0,
+      "grad_norm": 9.371519088745117,
+      "learning_rate": 0.0005362174000808813,
+      "loss": 0.7544,
+      "step": 620
+    },
+    {
+      "epoch": 310.5,
+      "grad_norm": 9.852248191833496,
+      "learning_rate": 0.0005350121646683182,
+      "loss": 0.8389,
+      "step": 621
+    },
+    {
+      "epoch": 311.0,
+      "grad_norm": 10.86923885345459,
+      "learning_rate": 0.0005338067247847219,
+      "loss": 0.6607,
+      "step": 622
+    },
+    {
+      "epoch": 311.5,
+      "grad_norm": 6.235162258148193,
+      "learning_rate": 0.0005326010874698608,
+      "loss": 0.7564,
+      "step": 623
+    },
+    {
+      "epoch": 312.0,
+      "grad_norm": 22.363107681274414,
+      "learning_rate": 0.0005313952597646568,
+      "loss": 0.6608,
+      "step": 624
+    },
+    {
+      "epoch": 312.5,
+      "grad_norm": 12.136018753051758,
+      "learning_rate": 0.0005301892487111431,
+      "loss": 0.655,
+      "step": 625
+    },
+    {
+      "epoch": 313.0,
+      "grad_norm": 9.552675247192383,
+      "learning_rate": 0.0005289830613524241,
+      "loss": 0.6772,
+      "step": 626
+    },
+    {
+      "epoch": 313.5,
+      "grad_norm": 8.428141593933105,
+      "learning_rate": 0.0005277767047326338,
+      "loss": 0.781,
+      "step": 627
+    },
+    {
+      "epoch": 314.0,
+      "grad_norm": 14.407150268554688,
+      "learning_rate": 0.0005265701858968943,
+      "loss": 0.6146,
+      "step": 628
+    },
+    {
+      "epoch": 314.5,
+      "grad_norm": 9.657368659973145,
+      "learning_rate": 0.0005253635118912756,
+      "loss": 0.6905,
+      "step": 629
+    },
+    {
+      "epoch": 315.0,
+      "grad_norm": 10.933073997497559,
+      "learning_rate": 0.0005241566897627535,
+      "loss": 0.6248,
+      "step": 630
+    },
+    {
+      "epoch": 315.5,
+      "grad_norm": 10.306805610656738,
+      "learning_rate": 0.0005229497265591689,
+      "loss": 0.49,
+      "step": 631
+    },
+    {
+      "epoch": 316.0,
+      "grad_norm": 13.158028602600098,
+      "learning_rate": 0.0005217426293291868,
+      "loss": 0.924,
+      "step": 632
+    },
+    {
+      "epoch": 316.5,
+      "grad_norm": 13.890748023986816,
+      "learning_rate": 0.0005205354051222545,
+      "loss": 0.843,
+      "step": 633
+    },
+    {
+      "epoch": 317.0,
+      "grad_norm": 15.626507759094238,
+      "learning_rate": 0.0005193280609885611,
+      "loss": 0.5412,
+      "step": 634
+    },
+    {
+      "epoch": 317.5,
+      "grad_norm": 14.684675216674805,
+      "learning_rate": 0.0005181206039789962,
+      "loss": 0.588,
+      "step": 635
+    },
+    {
+      "epoch": 318.0,
+      "grad_norm": 12.251501083374023,
+      "learning_rate": 0.0005169130411451083,
+      "loss": 0.7938,
+      "step": 636
+    },
+    {
+      "epoch": 318.5,
+      "grad_norm": 8.811335563659668,
+      "learning_rate": 0.0005157053795390641,
+      "loss": 0.6782,
+      "step": 637
+    },
+    {
+      "epoch": 319.0,
+      "grad_norm": 8.832444190979004,
+      "learning_rate": 0.0005144976262136073,
+      "loss": 0.7023,
+      "step": 638
+    },
+    {
+      "epoch": 319.5,
+      "grad_norm": 8.185532569885254,
+      "learning_rate": 0.0005132897882220167,
+      "loss": 0.5533,
+      "step": 639
+    },
+    {
+      "epoch": 320.0,
+      "grad_norm": 8.193849563598633,
+      "learning_rate": 0.0005120818726180662,
+      "loss": 0.8821,
+      "step": 640
+    },
+    {
+      "epoch": 320.5,
+      "grad_norm": 8.693097114562988,
+      "learning_rate": 0.0005108738864559823,
+      "loss": 0.4721,
+      "step": 641
+    },
+    {
+      "epoch": 321.0,
+      "grad_norm": 7.98622465133667,
+      "learning_rate": 0.0005096658367904042,
+      "loss": 0.9157,
+      "step": 642
+    },
+    {
+      "epoch": 321.5,
+      "grad_norm": 7.520875453948975,
+      "learning_rate": 0.0005084577306763415,
+      "loss": 0.6165,
+      "step": 643
+    },
+    {
+      "epoch": 322.0,
+      "grad_norm": 11.647687911987305,
+      "learning_rate": 0.0005072495751691338,
+      "loss": 0.9636,
+      "step": 644
+    },
+    {
+      "epoch": 322.5,
+      "grad_norm": 10.56940746307373,
+      "learning_rate": 0.0005060413773244087,
+      "loss": 0.5884,
+      "step": 645
+    },
+    {
+      "epoch": 323.0,
+      "grad_norm": 9.552960395812988,
+      "learning_rate": 0.0005048331441980416,
+      "loss": 0.7567,
+      "step": 646
+    },
+    {
+      "epoch": 323.5,
+      "grad_norm": 10.163215637207031,
+      "learning_rate": 0.0005036248828461133,
+      "loss": 0.7326,
+      "step": 647
+    },
+    {
+      "epoch": 324.0,
+      "grad_norm": 17.775524139404297,
+      "learning_rate": 0.0005024166003248702,
+      "loss": 0.657,
+      "step": 648
+    },
+    {
+      "epoch": 324.5,
+      "grad_norm": 8.287296295166016,
+      "learning_rate": 0.0005012083036906817,
+      "loss": 0.6922,
+      "step": 649
+    },
+    {
+      "epoch": 325.0,
+      "grad_norm": 11.926411628723145,
+      "learning_rate": 0.0005,
+      "loss": 0.6564,
+      "step": 650
+    },
+    {
+      "epoch": 325.5,
+      "grad_norm": 10.875996589660645,
+      "learning_rate": 0.0004987916963093183,
+      "loss": 0.7068,
+      "step": 651
+    },
+    {
+      "epoch": 326.0,
+      "grad_norm": 9.844751358032227,
+      "learning_rate": 0.0004975833996751299,
+      "loss": 0.9535,
+      "step": 652
+    },
+    {
+      "epoch": 326.5,
+      "grad_norm": 15.329977989196777,
+      "learning_rate": 0.0004963751171538866,
+      "loss": 0.6984,
+      "step": 653
+    },
+    {
+      "epoch": 327.0,
+      "grad_norm": 14.797718048095703,
+      "learning_rate": 0.0004951668558019585,
+      "loss": 0.7783,
+      "step": 654
+    },
+    {
+      "epoch": 327.5,
+      "grad_norm": 15.537858009338379,
+      "learning_rate": 0.0004939586226755913,
+      "loss": 0.6451,
+      "step": 655
+    },
+    {
+      "epoch": 328.0,
+      "grad_norm": 8.54361629486084,
+      "learning_rate": 0.0004927504248308663,
+      "loss": 0.6667,
+      "step": 656
+    },
+    {
+      "epoch": 328.5,
+      "grad_norm": 14.354753494262695,
+      "learning_rate": 0.0004915422693236584,
+      "loss": 0.7056,
+      "step": 657
+    },
+    {
+      "epoch": 329.0,
+      "grad_norm": 16.983583450317383,
+      "learning_rate": 0.0004903341632095958,
+      "loss": 0.6078,
+      "step": 658
+    },
+    {
+      "epoch": 329.5,
+      "grad_norm": 10.012537002563477,
+      "learning_rate": 0.0004891261135440177,
+      "loss": 0.5635,
+      "step": 659
+    },
+    {
+      "epoch": 330.0,
+      "grad_norm": 10.801152229309082,
+      "learning_rate": 0.000487918127381934,
+      "loss": 0.7142,
+      "step": 660
+    },
+    {
+      "epoch": 330.5,
+      "grad_norm": 8.539676666259766,
+      "learning_rate": 0.0004867102117779834,
+      "loss": 0.694,
+      "step": 661
+    },
+    {
+      "epoch": 331.0,
+      "grad_norm": 7.145092964172363,
+      "learning_rate": 0.00048550237378639275,
+      "loss": 0.4908,
+      "step": 662
+    },
+    {
+      "epoch": 331.5,
+      "grad_norm": 8.803322792053223,
+      "learning_rate": 0.00048429462046093585,
+      "loss": 0.4381,
+      "step": 663
+    },
+    {
+      "epoch": 332.0,
+      "grad_norm": 9.998886108398438,
+      "learning_rate": 0.0004830869588548918,
+      "loss": 0.8832,
+      "step": 664
+    },
+    {
+      "epoch": 332.5,
+      "grad_norm": 8.418477058410645,
+      "learning_rate": 0.000481879396021004,
+      "loss": 0.7606,
+      "step": 665
+    },
+    {
+      "epoch": 333.0,
+      "grad_norm": 7.3144211769104,
+      "learning_rate": 0.00048067193901143887,
+      "loss": 0.4357,
+      "step": 666
+    },
+    {
+      "epoch": 333.5,
+      "grad_norm": 16.10854721069336,
+      "learning_rate": 0.00047946459487774556,
+      "loss": 0.5005,
+      "step": 667
+    },
+    {
+      "epoch": 334.0,
+      "grad_norm": 9.584819793701172,
+      "learning_rate": 0.00047825737067081327,
+      "loss": 0.7829,
+      "step": 668
+    },
+    {
+      "epoch": 334.5,
+      "grad_norm": 6.8481011390686035,
+      "learning_rate": 0.0004770502734408311,
+      "loss": 0.7825,
+      "step": 669
+    },
+    {
+      "epoch": 335.0,
+      "grad_norm": 15.377241134643555,
+      "learning_rate": 0.00047584331023724653,
+      "loss": 0.5575,
+      "step": 670
+    },
+    {
+      "epoch": 335.5,
+      "grad_norm": 12.38183307647705,
+      "learning_rate": 0.0004746364881087244,
+      "loss": 0.6552,
+      "step": 671
+    },
+    {
+      "epoch": 336.0,
+      "grad_norm": 21.3419132232666,
+      "learning_rate": 0.0004734298141031057,
+      "loss": 0.6596,
+      "step": 672
+    },
+    {
+      "epoch": 336.5,
+      "grad_norm": 14.396490097045898,
+      "learning_rate": 0.00047222329526736635,
+      "loss": 0.7737,
+      "step": 673
+    },
+    {
+      "epoch": 337.0,
+      "grad_norm": 16.65850257873535,
+      "learning_rate": 0.00047101693864757605,
+      "loss": 0.5682,
+      "step": 674
+    },
+    {
+      "epoch": 337.5,
+      "grad_norm": 6.612177848815918,
+      "learning_rate": 0.0004698107512888569,
+      "loss": 0.7978,
+      "step": 675
+    },
+    {
+      "epoch": 338.0,
+      "grad_norm": 13.071910858154297,
+      "learning_rate": 0.0004686047402353433,
+      "loss": 0.4835,
+      "step": 676
+    },
+    {
+      "epoch": 338.5,
+      "grad_norm": 7.844511032104492,
+      "learning_rate": 0.00046739891253013923,
+      "loss": 0.7081,
+      "step": 677
+    },
+    {
+      "epoch": 339.0,
+      "grad_norm": 8.101701736450195,
+      "learning_rate": 0.00046619327521527825,
+      "loss": 0.5562,
+      "step": 678
+    },
+    {
+      "epoch": 339.5,
+      "grad_norm": 7.12611722946167,
+      "learning_rate": 0.0004649878353316818,
+      "loss": 0.6736,
+      "step": 679
+    },
+    {
+      "epoch": 340.0,
+      "grad_norm": 14.323423385620117,
+      "learning_rate": 0.00046378259991911887,
+      "loss": 0.5943,
+      "step": 680
+    },
+    {
+      "epoch": 340.5,
+      "grad_norm": 12.41629409790039,
+      "learning_rate": 0.0004625775760161636,
+      "loss": 0.6548,
+      "step": 681
+    },
+    {
+      "epoch": 341.0,
+      "grad_norm": 13.218873023986816,
+      "learning_rate": 0.0004613727706601557,
+      "loss": 0.7024,
+      "step": 682
+    },
+    {
+      "epoch": 341.5,
+      "grad_norm": 10.877093315124512,
+      "learning_rate": 0.00046016819088715786,
+      "loss": 0.6225,
+      "step": 683
+    },
+    {
+      "epoch": 342.0,
+      "grad_norm": 9.745308876037598,
+      "learning_rate": 0.0004589638437319157,
+      "loss": 0.5913,
+      "step": 684
+    },
+    {
+      "epoch": 342.5,
+      "grad_norm": 8.511798858642578,
+      "learning_rate": 0.0004577597362278165,
+      "loss": 0.5908,
+      "step": 685
+    },
+    {
+      "epoch": 343.0,
+      "grad_norm": 9.225003242492676,
+      "learning_rate": 0.0004565558754068477,
+      "loss": 0.6472,
+      "step": 686
+    },
+    {
+      "epoch": 343.5,
+      "grad_norm": 6.896814823150635,
+      "learning_rate": 0.00045535226829955614,
+      "loss": 0.74,
+      "step": 687
+    },
+    {
+      "epoch": 344.0,
+      "grad_norm": 11.832365989685059,
+      "learning_rate": 0.0004541489219350069,
+      "loss": 0.5249,
+      "step": 688
+    },
+    {
+      "epoch": 344.5,
+      "grad_norm": 10.762080192565918,
+      "learning_rate": 0.00045294584334074284,
+      "loss": 0.6108,
+      "step": 689
+    },
+    {
+      "epoch": 345.0,
+      "grad_norm": 8.609845161437988,
+      "learning_rate": 0.00045174303954274245,
+      "loss": 0.6481,
+      "step": 690
+    },
+    {
+      "epoch": 345.5,
+      "grad_norm": 11.866127014160156,
+      "learning_rate": 0.0004505405175653797,
+      "loss": 0.6482,
+      "step": 691
+    },
+    {
+      "epoch": 346.0,
+      "grad_norm": 14.82781982421875,
+      "learning_rate": 0.0004493382844313826,
+      "loss": 0.6438,
+      "step": 692
+    },
+    {
+      "epoch": 346.5,
+      "grad_norm": 7.827607154846191,
+      "learning_rate": 0.00044813634716179217,
+      "loss": 0.7762,
+      "step": 693
+    },
+    {
+      "epoch": 347.0,
+      "grad_norm": 17.15668296813965,
+      "learning_rate": 0.0004469347127759222,
+      "loss": 0.4686,
+      "step": 694
+    },
+    {
+      "epoch": 347.5,
+      "grad_norm": 11.237194061279297,
+      "learning_rate": 0.00044573338829131693,
+      "loss": 0.6765,
+      "step": 695
+    },
+    {
+      "epoch": 348.0,
+      "grad_norm": 10.1303129196167,
+      "learning_rate": 0.00044453238072371116,
+      "loss": 0.5824,
+      "step": 696
+    },
+    {
+      "epoch": 348.5,
+      "grad_norm": 12.26428508758545,
+      "learning_rate": 0.0004433316970869885,
+      "loss": 0.4984,
+      "step": 697
+    },
+    {
+      "epoch": 349.0,
+      "grad_norm": 8.808486938476562,
+      "learning_rate": 0.0004421313443931416,
+      "loss": 0.8034,
+      "step": 698
+    },
+    {
+      "epoch": 349.5,
+      "grad_norm": 10.697589874267578,
+      "learning_rate": 0.0004409313296522297,
+      "loss": 0.6991,
+      "step": 699
+    },
+    {
+      "epoch": 350.0,
+      "grad_norm": 25.127532958984375,
+      "learning_rate": 0.00043973165987233853,
+      "loss": 0.4956,
+      "step": 700
+    },
+    {
+      "epoch": 350.5,
+      "grad_norm": 10.437840461730957,
+      "learning_rate": 0.00043853234205953943,
+      "loss": 0.684,
+      "step": 701
+    },
+    {
+      "epoch": 351.0,
+      "grad_norm": 14.081304550170898,
+      "learning_rate": 0.00043733338321784784,
+      "loss": 0.6126,
+      "step": 702
+    },
+    {
+      "epoch": 351.5,
+      "grad_norm": 16.336280822753906,
+      "learning_rate": 0.0004361347903491836,
+      "loss": 0.5764,
+      "step": 703
+    },
+    {
+      "epoch": 352.0,
+      "grad_norm": 9.603869438171387,
+      "learning_rate": 0.0004349365704533284,
+      "loss": 0.6636,
+      "step": 704
+    },
+    {
+      "epoch": 352.5,
+      "grad_norm": 13.625994682312012,
+      "learning_rate": 0.00043373873052788637,
+      "loss": 0.6112,
+      "step": 705
+    },
+    {
+      "epoch": 353.0,
+      "grad_norm": 14.83207893371582,
+      "learning_rate": 0.00043254127756824214,
+      "loss": 0.6036,
+      "step": 706
+    },
+    {
+      "epoch": 353.5,
+      "grad_norm": 6.32863712310791,
+      "learning_rate": 0.000431344218567521,
+      "loss": 0.6038,
+      "step": 707
+    },
+    {
+      "epoch": 354.0,
+      "grad_norm": 14.014519691467285,
+      "learning_rate": 0.00043014756051654705,
+      "loss": 0.6005,
+      "step": 708
+    },
+    {
+      "epoch": 354.5,
+      "grad_norm": 9.407151222229004,
+      "learning_rate": 0.00042895131040380315,
+      "loss": 0.6698,
+      "step": 709
+    },
+    {
+      "epoch": 355.0,
+      "grad_norm": 14.189062118530273,
+      "learning_rate": 0.0004277554752153895,
+      "loss": 0.6185,
+      "step": 710
+    },
+    {
+      "epoch": 355.5,
+      "grad_norm": 8.251945495605469,
+      "learning_rate": 0.0004265600619349832,
+      "loss": 0.5905,
+      "step": 711
+    },
+    {
+      "epoch": 356.0,
+      "grad_norm": 13.41107177734375,
+      "learning_rate": 0.000425365077543798,
+      "loss": 0.6068,
+      "step": 712
+    },
+    {
+      "epoch": 356.5,
+      "grad_norm": 9.325090408325195,
+      "learning_rate": 0.0004241705290205421,
+      "loss": 0.7014,
+      "step": 713
+    },
+    {
+      "epoch": 357.0,
+      "grad_norm": 16.509735107421875,
+      "learning_rate": 0.00042297642334137875,
+      "loss": 0.4876,
+      "step": 714
+    },
+    {
+      "epoch": 357.5,
+      "grad_norm": 8.346978187561035,
+      "learning_rate": 0.0004217827674798845,
+      "loss": 0.5311,
+      "step": 715
+    },
+    {
+      "epoch": 358.0,
+      "grad_norm": 13.908074378967285,
+      "learning_rate": 0.0004205895684070098,
+      "loss": 0.7135,
+      "step": 716
+    },
+    {
+      "epoch": 358.5,
+      "grad_norm": 7.795354843139648,
+      "learning_rate": 0.00041939683309103663,
+      "loss": 0.7331,
+      "step": 717
+    },
+    {
+      "epoch": 359.0,
+      "grad_norm": 11.836822509765625,
+      "learning_rate": 0.0004182045684975391,
+      "loss": 0.456,
+      "step": 718
+    },
+    {
+      "epoch": 359.5,
+      "grad_norm": 9.60282039642334,
+      "learning_rate": 0.0004170127815893421,
+      "loss": 0.7019,
+      "step": 719
+    },
+    {
+      "epoch": 360.0,
+      "grad_norm": 16.935016632080078,
+      "learning_rate": 0.0004158214793264807,
+      "loss": 0.5251,
+      "step": 720
+    },
+    {
+      "epoch": 360.5,
+      "grad_norm": 16.354005813598633,
+      "learning_rate": 0.0004146306686661602,
+      "loss": 0.5833,
+      "step": 721
+    },
+    {
+      "epoch": 361.0,
+      "grad_norm": 18.009193420410156,
+      "learning_rate": 0.00041344035656271436,
+      "loss": 0.7866,
+      "step": 722
+    },
+    {
+      "epoch": 361.5,
+      "grad_norm": 12.471964836120605,
+      "learning_rate": 0.0004122505499675655,
+      "loss": 0.7458,
+      "step": 723
+    },
+    {
+      "epoch": 362.0,
+      "grad_norm": 13.796614646911621,
+      "learning_rate": 0.00041106125582918385,
+      "loss": 0.4468,
+      "step": 724
+    },
+    {
+      "epoch": 362.5,
+      "grad_norm": 7.775569438934326,
+      "learning_rate": 0.00040987248109304716,
+      "loss": 0.7095,
+      "step": 725
+    },
+    {
+      "epoch": 363.0,
+      "grad_norm": 8.567879676818848,
+      "learning_rate": 0.00040868423270159945,
+      "loss": 0.6177,
+      "step": 726
+    },
+    {
+      "epoch": 363.5,
+      "grad_norm": 8.855268478393555,
+      "learning_rate": 0.000407496517594211,
+      "loss": 0.5348,
+      "step": 727
+    },
+    {
+      "epoch": 364.0,
+      "grad_norm": 7.385838031768799,
+      "learning_rate": 0.0004063093427071376,
+      "loss": 0.6498,
+      "step": 728
+    },
+    {
+      "epoch": 364.5,
+      "grad_norm": 6.559077262878418,
+      "learning_rate": 0.00040512271497348076,
+      "loss": 0.7253,
+      "step": 729
+    },
+    {
+      "epoch": 365.0,
+      "grad_norm": 12.2962007522583,
+      "learning_rate": 0.00040393664132314577,
+      "loss": 0.4726,
+      "step": 730
+    },
+    {
+      "epoch": 365.5,
+      "grad_norm": 9.821822166442871,
+      "learning_rate": 0.0004027511286828027,
+      "loss": 0.602,
+      "step": 731
+    },
+    {
+      "epoch": 366.0,
+      "grad_norm": 10.89181137084961,
+      "learning_rate": 0.000401566183975845,
+      "loss": 0.8707,
+      "step": 732
+    },
+    {
+      "epoch": 366.5,
+      "grad_norm": 10.412627220153809,
+      "learning_rate": 0.0004003818141223493,
+      "loss": 0.4606,
+      "step": 733
+    },
+    {
+      "epoch": 367.0,
+      "grad_norm": 10.142531394958496,
+      "learning_rate": 0.00039919802603903553,
+      "loss": 0.6974,
+      "step": 734
+    },
+    {
+      "epoch": 367.5,
+      "grad_norm": 8.779993057250977,
+      "learning_rate": 0.00039801482663922574,
+      "loss": 0.7246,
+      "step": 735
+    },
+    {
+      "epoch": 368.0,
+      "grad_norm": 9.068611145019531,
+      "learning_rate": 0.0003968322228328041,
+      "loss": 0.4536,
+      "step": 736
+    },
+    {
+      "epoch": 368.5,
+      "grad_norm": 11.373882293701172,
+      "learning_rate": 0.0003956502215261763,
+      "loss": 0.4862,
+      "step": 737
+    },
+    {
+      "epoch": 369.0,
+      "grad_norm": 14.528646469116211,
+      "learning_rate": 0.00039446882962223027,
+      "loss": 0.9259,
+      "step": 738
+    },
+    {
+      "epoch": 369.5,
+      "grad_norm": 9.39692497253418,
+      "learning_rate": 0.00039328805402029404,
+      "loss": 0.754,
+      "step": 739
+    },
+    {
+      "epoch": 370.0,
+      "grad_norm": 10.453376770019531,
+      "learning_rate": 0.000392107901616097,
+      "loss": 0.4258,
+      "step": 740
+    },
+    {
+      "epoch": 370.5,
+      "grad_norm": 9.095719337463379,
+      "learning_rate": 0.0003909283793017289,
+      "loss": 0.6962,
+      "step": 741
+    },
+    {
+      "epoch": 371.0,
+      "grad_norm": 8.22550106048584,
+      "learning_rate": 0.0003897494939655995,
+      "loss": 0.5201,
+      "step": 742
+    },
+    {
+      "epoch": 371.5,
+      "grad_norm": 15.37446117401123,
+      "learning_rate": 0.00038857125249239916,
+      "loss": 0.4339,
+      "step": 743
+    },
+    {
+      "epoch": 372.0,
+      "grad_norm": 8.430370330810547,
+      "learning_rate": 0.00038739366176305785,
+      "loss": 0.7832,
+      "step": 744
+    },
+    {
+      "epoch": 372.5,
+      "grad_norm": 8.98311996459961,
+      "learning_rate": 0.00038621672865470503,
+      "loss": 0.5513,
+      "step": 745
+    },
+    {
+      "epoch": 373.0,
+      "grad_norm": 11.335391998291016,
+      "learning_rate": 0.00038504046004062974,
+      "loss": 0.5625,
+      "step": 746
+    },
+    {
+      "epoch": 373.5,
+      "grad_norm": 10.32812786102295,
+      "learning_rate": 0.00038386486279024055,
+      "loss": 0.5655,
+      "step": 747
+    },
+    {
+      "epoch": 374.0,
+      "grad_norm": 7.775624752044678,
+      "learning_rate": 0.000382689943769025,
+      "loss": 0.5305,
+      "step": 748
+    },
+    {
+      "epoch": 374.5,
+      "grad_norm": 5.3197174072265625,
+      "learning_rate": 0.00038151570983851,
+      "loss": 0.4829,
+      "step": 749
+    },
+    {
+      "epoch": 375.0,
+      "grad_norm": 7.23258638381958,
+      "learning_rate": 0.00038034216785622126,
+      "loss": 0.6045,
+      "step": 750
+    },
+    {
+      "epoch": 375.5,
+      "grad_norm": 8.32296085357666,
+      "learning_rate": 0.0003791693246756436,
+      "loss": 0.6292,
+      "step": 751
+    },
+    {
+      "epoch": 376.0,
+      "grad_norm": 5.875463962554932,
+      "learning_rate": 0.00037799718714618124,
+      "loss": 0.5375,
+      "step": 752
+    },
+    {
+      "epoch": 376.5,
+      "grad_norm": 6.123970985412598,
+      "learning_rate": 0.00037682576211311704,
+      "loss": 0.3879,
+      "step": 753
+    },
+    {
+      "epoch": 377.0,
+      "grad_norm": 11.018393516540527,
+      "learning_rate": 0.0003756550564175727,
+      "loss": 0.6549,
+      "step": 754
+    },
+    {
+      "epoch": 377.5,
+      "grad_norm": 10.09932804107666,
+      "learning_rate": 0.00037448507689646915,
+      "loss": 0.6997,
+      "step": 755
+    },
+    {
+      "epoch": 378.0,
+      "grad_norm": 6.168050289154053,
+      "learning_rate": 0.0003733158303824868,
+      "loss": 0.3872,
+      "step": 756
+    },
+    {
+      "epoch": 378.5,
+      "grad_norm": 6.59968376159668,
+      "learning_rate": 0.00037214732370402493,
+      "loss": 0.5201,
+      "step": 757
+    },
+    {
+      "epoch": 379.0,
+      "grad_norm": 10.000802993774414,
+      "learning_rate": 0.0003709795636851622,
+      "loss": 0.5811,
+      "step": 758
+    },
+    {
+      "epoch": 379.5,
+      "grad_norm": 5.797168254852295,
+      "learning_rate": 0.0003698125571456168,
+      "loss": 0.6158,
+      "step": 759
+    },
+    {
+      "epoch": 380.0,
+      "grad_norm": 11.78331184387207,
+      "learning_rate": 0.0003686463109007065,
+      "loss": 0.4712,
+      "step": 760
+    },
+    {
+      "epoch": 380.5,
+      "grad_norm": 7.390726089477539,
+      "learning_rate": 0.0003674808317613095,
+      "loss": 0.6263,
+      "step": 761
+    },
+    {
+      "epoch": 381.0,
+      "grad_norm": 9.715678215026855,
+      "learning_rate": 0.0003663161265338235,
+      "loss": 0.5421,
+      "step": 762
+    },
+    {
+      "epoch": 381.5,
+      "grad_norm": 14.303598403930664,
+      "learning_rate": 0.0003651522020201268,
+      "loss": 0.6106,
+      "step": 763
+    },
+    {
+      "epoch": 382.0,
+      "grad_norm": 8.646794319152832,
+      "learning_rate": 0.00036398906501753785,
+      "loss": 0.4009,
+      "step": 764
+    },
+    {
+      "epoch": 382.5,
+      "grad_norm": 8.449315071105957,
+      "learning_rate": 0.0003628267223187771,
+      "loss": 0.6261,
+      "step": 765
+    },
+    {
+      "epoch": 383.0,
+      "grad_norm": 9.600123405456543,
+      "learning_rate": 0.00036166518071192546,
+      "loss": 0.5244,
+      "step": 766
+    },
+    {
+      "epoch": 383.5,
+      "grad_norm": 6.5802321434021,
+      "learning_rate": 0.0003605044469803854,
+      "loss": 0.5414,
+      "step": 767
+    },
+    {
+      "epoch": 384.0,
+      "grad_norm": 7.23771333694458,
+      "learning_rate": 0.00035934452790284177,
+      "loss": 0.5198,
+      "step": 768
+    },
+    {
+      "epoch": 384.5,
+      "grad_norm": 5.851401329040527,
+      "learning_rate": 0.0003581854302532218,
+      "loss": 0.5394,
+      "step": 769
+    },
+    {
+      "epoch": 385.0,
+      "grad_norm": 10.882811546325684,
+      "learning_rate": 0.00035702716080065545,
+      "loss": 0.5862,
+      "step": 770
+    },
+    {
+      "epoch": 385.5,
+      "grad_norm": 8.181309700012207,
+      "learning_rate": 0.0003558697263094359,
+      "loss": 0.575,
+      "step": 771
+    },
+    {
+      "epoch": 386.0,
+      "grad_norm": 12.346161842346191,
+      "learning_rate": 0.00035471313353898054,
+      "loss": 0.5971,
+      "step": 772
+    },
+    {
+      "epoch": 386.5,
+      "grad_norm": 8.454315185546875,
+      "learning_rate": 0.0003535573892437904,
+      "loss": 0.5886,
+      "step": 773
+    },
+    {
+      "epoch": 387.0,
+      "grad_norm": 6.908176898956299,
+      "learning_rate": 0.0003524025001734126,
+      "loss": 0.4844,
+      "step": 774
+    },
+    {
+      "epoch": 387.5,
+      "grad_norm": 5.884853839874268,
+      "learning_rate": 0.0003512484730723986,
+      "loss": 0.5931,
+      "step": 775
+    },
+    {
+      "epoch": 388.0,
+      "grad_norm": 6.892649173736572,
+      "learning_rate": 0.00035009531468026647,
+      "loss": 0.4987,
+      "step": 776
+    },
+    {
+      "epoch": 388.5,
+      "grad_norm": 6.073370456695557,
+      "learning_rate": 0.00034894303173146066,
+      "loss": 0.4831,
+      "step": 777
+    },
+    {
+      "epoch": 389.0,
+      "grad_norm": 10.121975898742676,
+      "learning_rate": 0.00034779163095531384,
+      "loss": 0.566,
+      "step": 778
+    },
+    {
+      "epoch": 389.5,
+      "grad_norm": 6.3472371101379395,
+      "learning_rate": 0.000346641119076006,
+      "loss": 0.627,
+      "step": 779
+    },
+    {
+      "epoch": 390.0,
+      "grad_norm": 5.79213285446167,
+      "learning_rate": 0.00034549150281252633,
+      "loss": 0.45,
+      "step": 780
+    },
+    {
+      "epoch": 390.5,
+      "grad_norm": 5.635915756225586,
+      "learning_rate": 0.00034434278887863346,
+      "loss": 0.5057,
+      "step": 781
+    },
+    {
+      "epoch": 391.0,
+      "grad_norm": 10.593151092529297,
+      "learning_rate": 0.00034319498398281635,
+      "loss": 0.6281,
+      "step": 782
+    },
+    {
+      "epoch": 391.5,
+      "grad_norm": 8.581198692321777,
+      "learning_rate": 0.0003420480948282556,
+      "loss": 0.5131,
+      "step": 783
+    },
+    {
+      "epoch": 392.0,
+      "grad_norm": 7.918695449829102,
+      "learning_rate": 0.0003409021281127835,
+      "loss": 0.5756,
+      "step": 784
+    },
+    {
+      "epoch": 392.5,
+      "grad_norm": 6.1155829429626465,
+      "learning_rate": 0.0003397570905288453,
+      "loss": 0.5428,
+      "step": 785
+    },
+    {
+      "epoch": 393.0,
+      "grad_norm": 13.031826972961426,
+      "learning_rate": 0.0003386129887634601,
+      "loss": 0.4786,
+      "step": 786
+    },
+    {
+      "epoch": 393.5,
+      "grad_norm": 6.871598243713379,
+      "learning_rate": 0.0003374698294981824,
+      "loss": 0.6786,
+      "step": 787
+    },
+    {
+      "epoch": 394.0,
+      "grad_norm": 11.693024635314941,
+      "learning_rate": 0.0003363276194090617,
+      "loss": 0.56,
+      "step": 788
+    },
+    {
+      "epoch": 394.5,
+      "grad_norm": 10.845073699951172,
+      "learning_rate": 0.00033518636516660465,
+      "loss": 0.6446,
+      "step": 789
+    },
+    {
+      "epoch": 395.0,
+      "grad_norm": 10.124814987182617,
+      "learning_rate": 0.0003340460734357359,
+      "loss": 0.4028,
+      "step": 790
+    },
+    {
+      "epoch": 395.5,
+      "grad_norm": 8.16336441040039,
+      "learning_rate": 0.00033290675087575854,
+      "loss": 0.4574,
+      "step": 791
+    },
+    {
+      "epoch": 396.0,
+      "grad_norm": 8.569724082946777,
+      "learning_rate": 0.0003317684041403165,
+      "loss": 0.6872,
+      "step": 792
+    },
+    {
+      "epoch": 396.5,
+      "grad_norm": 7.219727516174316,
+      "learning_rate": 0.0003306310398773543,
+      "loss": 0.3933,
+      "step": 793
+    },
+    {
+      "epoch": 397.0,
+      "grad_norm": 24.1346492767334,
+      "learning_rate": 0.00032949466472907896,
+      "loss": 0.9008,
+      "step": 794
+    },
+    {
+      "epoch": 397.5,
+      "grad_norm": 12.681283950805664,
+      "learning_rate": 0.0003283592853319208,
+      "loss": 0.5249,
+      "step": 795
+    },
+    {
+      "epoch": 398.0,
+      "grad_norm": 8.65915298461914,
+      "learning_rate": 0.0003272249083164957,
+      "loss": 0.5967,
+      "step": 796
+    },
+    {
+      "epoch": 398.5,
+      "grad_norm": 12.540446281433105,
+      "learning_rate": 0.00032609154030756484,
+      "loss": 0.497,
+      "step": 797
+    },
+    {
+      "epoch": 399.0,
+      "grad_norm": 20.23981285095215,
+      "learning_rate": 0.0003249591879239972,
+      "loss": 0.5853,
+      "step": 798
+    },
+    {
+      "epoch": 399.5,
+      "grad_norm": 8.258729934692383,
+      "learning_rate": 0.0003238278577787305,
+      "loss": 0.6033,
+      "step": 799
+    },
+    {
+      "epoch": 400.0,
+      "grad_norm": 7.564504623413086,
+      "learning_rate": 0.00032269755647873217,
+      "loss": 0.5812,
+      "step": 800
+    },
+    {
+      "epoch": 400.5,
+      "grad_norm": 7.34077262878418,
+      "learning_rate": 0.0003215682906249621,
+      "loss": 0.6082,
+      "step": 801
+    },
+    {
+      "epoch": 401.0,
+      "grad_norm": 7.603788375854492,
+      "learning_rate": 0.0003204400668123322,
+      "loss": 0.5583,
+      "step": 802
+    },
+    {
+      "epoch": 401.5,
+      "grad_norm": 7.669255256652832,
+      "learning_rate": 0.0003193128916296695,
+      "loss": 0.4974,
+      "step": 803
+    },
+    {
+      "epoch": 402.0,
+      "grad_norm": 7.351687431335449,
+      "learning_rate": 0.00031818677165967646,
+      "loss": 0.6591,
+      "step": 804
+    },
+    {
+      "epoch": 402.5,
+      "grad_norm": 4.807971954345703,
+      "learning_rate": 0.0003170617134788939,
+      "loss": 0.5355,
+      "step": 805
+    },
+    {
+      "epoch": 403.0,
+      "grad_norm": 6.753246307373047,
+      "learning_rate": 0.00031593772365766105,
+      "loss": 0.4731,
+      "step": 806
+    },
+    {
+      "epoch": 403.5,
+      "grad_norm": 5.101422309875488,
+      "learning_rate": 0.00031481480876007823,
+      "loss": 0.5292,
+      "step": 807
+    },
+    {
+      "epoch": 404.0,
+      "grad_norm": 8.22523307800293,
+      "learning_rate": 0.00031369297534396826,
+      "loss": 0.6046,
+      "step": 808
+    },
+    {
+      "epoch": 404.5,
+      "grad_norm": 10.836455345153809,
+      "learning_rate": 0.0003125722299608378,
+      "loss": 0.648,
+      "step": 809
+    },
+    {
+      "epoch": 405.0,
+      "grad_norm": 8.931352615356445,
+      "learning_rate": 0.0003114525791558398,
+      "loss": 0.3681,
+      "step": 810
+    },
+    {
+      "epoch": 405.5,
+      "grad_norm": 6.121690273284912,
+      "learning_rate": 0.00031033402946773473,
+      "loss": 0.4946,
+      "step": 811
+    },
+    {
+      "epoch": 406.0,
+      "grad_norm": 8.935819625854492,
+      "learning_rate": 0.0003092165874288525,
+      "loss": 0.6024,
+      "step": 812
+    },
+    {
+      "epoch": 406.5,
+      "grad_norm": 8.369319915771484,
+      "learning_rate": 0.00030810025956505403,
+      "loss": 0.5341,
+      "step": 813
+    },
+    {
+      "epoch": 407.0,
+      "grad_norm": 9.406549453735352,
+      "learning_rate": 0.00030698505239569424,
+      "loss": 0.4473,
+      "step": 814
+    },
+    {
+      "epoch": 407.5,
+      "grad_norm": 11.265097618103027,
+      "learning_rate": 0.0003058709724335825,
+      "loss": 0.5364,
+      "step": 815
+    },
+    {
+      "epoch": 408.0,
+      "grad_norm": 12.912116050720215,
+      "learning_rate": 0.0003047580261849456,
+      "loss": 0.5154,
+      "step": 816
+    },
+    {
+      "epoch": 408.5,
+      "grad_norm": 4.532247543334961,
+      "learning_rate": 0.00030364622014938925,
+      "loss": 0.4316,
+      "step": 817
+    },
+    {
+      "epoch": 409.0,
+      "grad_norm": 10.54069995880127,
+      "learning_rate": 0.0003025355608198606,
+      "loss": 0.6316,
+      "step": 818
+    },
+    {
+      "epoch": 409.5,
+      "grad_norm": 19.557851791381836,
+      "learning_rate": 0.00030142605468260977,
+      "loss": 0.4494,
+      "step": 819
+    },
+    {
+      "epoch": 410.0,
+      "grad_norm": 9.801682472229004,
+      "learning_rate": 0.0003003177082171523,
+      "loss": 0.6384,
+      "step": 820
+    },
+    {
+      "epoch": 410.5,
+      "grad_norm": 5.457016944885254,
+      "learning_rate": 0.0002992105278962314,
+      "loss": 0.6434,
+      "step": 821
+    },
+    {
+      "epoch": 411.0,
+      "grad_norm": 10.695679664611816,
+      "learning_rate": 0.0002981045201857796,
+      "loss": 0.3761,
+      "step": 822
+    },
+    {
+      "epoch": 411.5,
+      "grad_norm": 6.783427715301514,
+      "learning_rate": 0.00029699969154488206,
+      "loss": 0.6605,
+      "step": 823
+    },
+    {
+      "epoch": 412.0,
+      "grad_norm": 6.060352802276611,
+      "learning_rate": 0.00029589604842573757,
+      "loss": 0.348,
+      "step": 824
+    },
+    {
+      "epoch": 412.5,
+      "grad_norm": 3.516425371170044,
+      "learning_rate": 0.0002947935972736217,
+      "loss": 0.3123,
+      "step": 825
+    },
+    {
+      "epoch": 413.0,
+      "grad_norm": 31.699419021606445,
+      "learning_rate": 0.0002936923445268488,
+      "loss": 1.2623,
+      "step": 826
+    },
+    {
+      "epoch": 413.5,
+      "grad_norm": 10.0693359375,
+      "learning_rate": 0.00029259229661673486,
+      "loss": 0.6587,
+      "step": 827
+    },
+    {
+      "epoch": 414.0,
+      "grad_norm": 9.27851390838623,
+      "learning_rate": 0.00029149345996755936,
+      "loss": 0.3855,
+      "step": 828
+    },
+    {
+      "epoch": 414.5,
+      "grad_norm": 6.326018810272217,
+      "learning_rate": 0.00029039584099652816,
+      "loss": 0.4678,
+      "step": 829
+    },
+    {
+      "epoch": 415.0,
+      "grad_norm": 7.433679103851318,
+      "learning_rate": 0.00028929944611373555,
+      "loss": 0.5877,
+      "step": 830
+    },
+    {
+      "epoch": 415.5,
+      "grad_norm": 5.403927326202393,
+      "learning_rate": 0.0002882042817221273,
+      "loss": 0.4175,
+      "step": 831
+    },
+    {
+      "epoch": 416.0,
+      "grad_norm": 7.088924407958984,
+      "learning_rate": 0.00028711035421746366,
+      "loss": 0.5292,
+      "step": 832
+    },
+    {
+      "epoch": 416.5,
+      "grad_norm": 6.283558368682861,
+      "learning_rate": 0.00028601766998828074,
+      "loss": 0.3777,
+      "step": 833
+    },
+    {
+      "epoch": 417.0,
+      "grad_norm": 7.314220905303955,
+      "learning_rate": 0.00028492623541585404,
+      "loss": 0.6551,
+      "step": 834
+    },
+    {
+      "epoch": 417.5,
+      "grad_norm": 4.246515274047852,
+      "learning_rate": 0.00028383605687416123,
+      "loss": 0.3712,
+      "step": 835
+    },
+    {
+      "epoch": 418.0,
+      "grad_norm": 8.414041519165039,
+      "learning_rate": 0.00028274714072984506,
+      "loss": 0.5997,
+      "step": 836
+    },
+    {
+      "epoch": 418.5,
+      "grad_norm": 5.198728084564209,
+      "learning_rate": 0.00028165949334217535,
+      "loss": 0.43,
+      "step": 837
+    },
+    {
+      "epoch": 419.0,
+      "grad_norm": 9.946592330932617,
+      "learning_rate": 0.00028057312106301253,
+      "loss": 0.6369,
+      "step": 838
+    },
+    {
+      "epoch": 419.5,
+      "grad_norm": 8.879456520080566,
+      "learning_rate": 0.0002794880302367705,
+      "loss": 0.3973,
+      "step": 839
+    },
+    {
+      "epoch": 420.0,
+      "grad_norm": 6.711618900299072,
+      "learning_rate": 0.0002784042272003794,
+      "loss": 0.618,
+      "step": 840
+    },
+    {
+      "epoch": 420.5,
+      "grad_norm": 5.179981708526611,
+      "learning_rate": 0.0002773217182832487,
+      "loss": 0.4933,
+      "step": 841
+    },
+    {
+      "epoch": 421.0,
+      "grad_norm": 8.509923934936523,
+      "learning_rate": 0.0002762405098072303,
+      "loss": 0.5179,
+      "step": 842
+    },
+    {
+      "epoch": 421.5,
+      "grad_norm": 7.031705856323242,
+      "learning_rate": 0.0002751606080865814,
+      "loss": 0.5701,
+      "step": 843
+    },
+    {
+      "epoch": 422.0,
+      "grad_norm": 8.863751411437988,
+      "learning_rate": 0.00027408201942792756,
+      "loss": 0.426,
+      "step": 844
+    },
+    {
+      "epoch": 422.5,
+      "grad_norm": 6.162368297576904,
+      "learning_rate": 0.00027300475013022663,
+      "loss": 0.4994,
+      "step": 845
+    },
+    {
+      "epoch": 423.0,
+      "grad_norm": 14.237730026245117,
+      "learning_rate": 0.000271928806484731,
+      "loss": 0.7582,
+      "step": 846
+    },
+    {
+      "epoch": 423.5,
+      "grad_norm": 7.896010875701904,
+      "learning_rate": 0.0002708541947749512,
+      "loss": 0.5239,
+      "step": 847
+    },
+    {
+      "epoch": 424.0,
+      "grad_norm": 7.066755294799805,
+      "learning_rate": 0.00026978092127661945,
+      "loss": 0.4122,
+      "step": 848
+    },
+    {
+      "epoch": 424.5,
+      "grad_norm": 10.19424057006836,
+      "learning_rate": 0.0002687089922576524,
+      "loss": 0.5099,
+      "step": 849
+    },
+    {
+      "epoch": 425.0,
+      "grad_norm": 8.622855186462402,
+      "learning_rate": 0.00026763841397811573,
+      "loss": 0.4488,
+      "step": 850
+    },
+    {
+      "epoch": 425.5,
+      "grad_norm": 5.944708824157715,
+      "learning_rate": 0.0002665691926901862,
+      "loss": 0.3916,
+      "step": 851
+    },
+    {
+      "epoch": 426.0,
+      "grad_norm": 9.606409072875977,
+      "learning_rate": 0.0002655013346381158,
+      "loss": 0.6267,
+      "step": 852
+    },
+    {
+      "epoch": 426.5,
+      "grad_norm": 6.145974636077881,
+      "learning_rate": 0.00026443484605819504,
+      "loss": 0.4377,
+      "step": 853
+    },
+    {
+      "epoch": 427.0,
+      "grad_norm": 8.908138275146484,
+      "learning_rate": 0.00026336973317871756,
+      "loss": 0.4933,
+      "step": 854
+    },
+    {
+      "epoch": 427.5,
+      "grad_norm": 6.5088043212890625,
+      "learning_rate": 0.00026230600221994197,
+      "loss": 0.4638,
+      "step": 855
+    },
+    {
+      "epoch": 428.0,
+      "grad_norm": 6.629590034484863,
+      "learning_rate": 0.0002612436593940568,
+      "loss": 0.5333,
+      "step": 856
+    },
+    {
+      "epoch": 428.5,
+      "grad_norm": 7.923740863800049,
+      "learning_rate": 0.000260182710905144,
+      "loss": 0.4008,
+      "step": 857
+    },
+    {
+      "epoch": 429.0,
+      "grad_norm": 7.7602972984313965,
+      "learning_rate": 0.0002591231629491423,
+      "loss": 0.5405,
+      "step": 858
+    },
+    {
+      "epoch": 429.5,
+      "grad_norm": 10.004776000976562,
+      "learning_rate": 0.0002580650217138118,
+      "loss": 0.49,
+      "step": 859
+    },
+    {
+      "epoch": 430.0,
+      "grad_norm": 7.723813056945801,
+      "learning_rate": 0.00025700829337869696,
+      "loss": 0.4407,
+      "step": 860
+    },
+    {
+      "epoch": 430.5,
+      "grad_norm": 8.191097259521484,
+      "learning_rate": 0.00025595298411509095,
+      "loss": 0.4322,
+      "step": 861
+    },
+    {
+      "epoch": 431.0,
+      "grad_norm": 7.365472793579102,
+      "learning_rate": 0.00025489910008599966,
+      "loss": 0.4839,
+      "step": 862
+    },
+    {
+      "epoch": 431.5,
+      "grad_norm": 7.801701545715332,
+      "learning_rate": 0.0002538466474461055,
+      "loss": 0.3698,
+      "step": 863
+    },
+    {
+      "epoch": 432.0,
+      "grad_norm": 12.315910339355469,
+      "learning_rate": 0.00025279563234173176,
+      "loss": 0.7568,
+      "step": 864
+    },
+    {
+      "epoch": 432.5,
+      "grad_norm": 5.561842441558838,
+      "learning_rate": 0.00025174606091080625,
+      "loss": 0.5825,
+      "step": 865
+    },
+    {
+      "epoch": 433.0,
+      "grad_norm": 6.639138221740723,
+      "learning_rate": 0.0002506979392828258,
+      "loss": 0.3523,
+      "step": 866
+    },
+    {
+      "epoch": 433.5,
+      "grad_norm": 6.864770412445068,
+      "learning_rate": 0.0002496512735788208,
+      "loss": 0.5108,
+      "step": 867
+    },
+    {
+      "epoch": 434.0,
+      "grad_norm": 6.027656555175781,
+      "learning_rate": 0.00024860606991131855,
+      "loss": 0.3838,
+      "step": 868
+    },
+    {
+      "epoch": 434.5,
+      "grad_norm": 5.55900239944458,
+      "learning_rate": 0.0002475623343843082,
+      "loss": 0.5629,
+      "step": 869
+    },
+    {
+      "epoch": 435.0,
+      "grad_norm": 6.646783828735352,
+      "learning_rate": 0.00024652007309320496,
+      "loss": 0.3884,
+      "step": 870
+    },
+    {
+      "epoch": 435.5,
+      "grad_norm": 6.156187534332275,
+      "learning_rate": 0.00024547929212481435,
+      "loss": 0.574,
+      "step": 871
+    },
+    {
+      "epoch": 436.0,
+      "grad_norm": 8.191523551940918,
+      "learning_rate": 0.0002444399975572974,
+      "loss": 0.486,
+      "step": 872
+    },
+    {
+      "epoch": 436.5,
+      "grad_norm": 5.7925801277160645,
+      "learning_rate": 0.00024340219546013388,
+      "loss": 0.4231,
+      "step": 873
+    },
+    {
+      "epoch": 437.0,
+      "grad_norm": 6.512327671051025,
+      "learning_rate": 0.0002423658918940878,
+      "loss": 0.4904,
+      "step": 874
+    },
+    {
+      "epoch": 437.5,
+      "grad_norm": 6.390252590179443,
+      "learning_rate": 0.00024133109291117155,
+      "loss": 0.5785,
+      "step": 875
+    },
+    {
+      "epoch": 438.0,
+      "grad_norm": 8.99042797088623,
+      "learning_rate": 0.00024029780455461138,
+      "loss": 0.5899,
+      "step": 876
+    },
+    {
+      "epoch": 438.5,
+      "grad_norm": 4.705801963806152,
+      "learning_rate": 0.00023926603285881083,
+      "loss": 0.3328,
+      "step": 877
+    },
+    {
+      "epoch": 439.0,
+      "grad_norm": 9.56912899017334,
+      "learning_rate": 0.00023823578384931632,
+      "loss": 0.564,
+      "step": 878
+    },
+    {
+      "epoch": 439.5,
+      "grad_norm": 14.541672706604004,
+      "learning_rate": 0.00023720706354278187,
+      "loss": 0.4518,
+      "step": 879
+    },
+    {
+      "epoch": 440.0,
+      "grad_norm": 5.941550254821777,
+      "learning_rate": 0.00023617987794693357,
+      "loss": 0.4722,
+      "step": 880
+    },
+    {
+      "epoch": 440.5,
+      "grad_norm": 5.095841407775879,
+      "learning_rate": 0.00023515423306053553,
+      "loss": 0.5772,
+      "step": 881
+    },
+    {
+      "epoch": 441.0,
+      "grad_norm": 5.768101215362549,
+      "learning_rate": 0.00023413013487335333,
+      "loss": 0.317,
+      "step": 882
+    },
+    {
+      "epoch": 441.5,
+      "grad_norm": 5.711394309997559,
+      "learning_rate": 0.00023310758936612025,
+      "loss": 0.3962,
+      "step": 883
+    },
+    {
+      "epoch": 442.0,
+      "grad_norm": 7.340517997741699,
+      "learning_rate": 0.00023208660251050156,
+      "loss": 0.7695,
+      "step": 884
+    },
+    {
+      "epoch": 442.5,
+      "grad_norm": 6.200962543487549,
+      "learning_rate": 0.0002310671802690607,
+      "loss": 0.4834,
+      "step": 885
+    },
+    {
+      "epoch": 443.0,
+      "grad_norm": 9.566047668457031,
+      "learning_rate": 0.00023004932859522305,
+      "loss": 0.512,
+      "step": 886
+    },
+    {
+      "epoch": 443.5,
+      "grad_norm": 7.321904182434082,
+      "learning_rate": 0.00022903305343324217,
+      "loss": 0.5045,
+      "step": 887
+    },
+    {
+      "epoch": 444.0,
+      "grad_norm": 5.773847579956055,
+      "learning_rate": 0.00022801836071816473,
+      "loss": 0.4253,
+      "step": 888
+    },
+    {
+      "epoch": 444.5,
+      "grad_norm": 5.374971866607666,
+      "learning_rate": 0.00022700525637579599,
+      "loss": 0.4642,
+      "step": 889
+    },
+    {
+      "epoch": 445.0,
+      "grad_norm": 8.404210090637207,
+      "learning_rate": 0.00022599374632266512,
+      "loss": 0.4162,
+      "step": 890
+    },
+    {
+      "epoch": 445.5,
+      "grad_norm": 4.909769535064697,
+      "learning_rate": 0.00022498383646599046,
+      "loss": 0.4437,
+      "step": 891
+    },
+    {
+      "epoch": 446.0,
+      "grad_norm": 6.044703960418701,
+      "learning_rate": 0.00022397553270364545,
+      "loss": 0.5524,
+      "step": 892
+    },
+    {
+      "epoch": 446.5,
+      "grad_norm": 6.300393104553223,
+      "learning_rate": 0.00022296884092412356,
+      "loss": 0.4111,
+      "step": 893
+    },
+    {
+      "epoch": 447.0,
+      "grad_norm": 5.974217414855957,
+      "learning_rate": 0.00022196376700650496,
+      "loss": 0.5096,
+      "step": 894
+    },
+    {
+      "epoch": 447.5,
+      "grad_norm": 7.008927345275879,
+      "learning_rate": 0.0002209603168204209,
+      "loss": 0.5515,
+      "step": 895
+    },
+    {
+      "epoch": 448.0,
+      "grad_norm": 12.311294555664062,
+      "learning_rate": 0.00021995849622602015,
+      "loss": 0.4454,
+      "step": 896
+    },
+    {
+      "epoch": 448.5,
+      "grad_norm": 5.96365213394165,
+      "learning_rate": 0.0002189583110739348,
+      "loss": 0.3488,
+      "step": 897
+    },
+    {
+      "epoch": 449.0,
+      "grad_norm": 7.669370651245117,
+      "learning_rate": 0.0002179597672052458,
+      "loss": 0.6877,
+      "step": 898
+    },
+    {
+      "epoch": 449.5,
+      "grad_norm": 5.630612850189209,
+      "learning_rate": 0.00021696287045144935,
+      "loss": 0.4475,
+      "step": 899
+    },
+    {
+      "epoch": 450.0,
+      "grad_norm": 6.436372756958008,
+      "learning_rate": 0.00021596762663442215,
+      "loss": 0.4939,
+      "step": 900
+    },
+    {
+      "epoch": 450.5,
+      "grad_norm": 6.084460735321045,
+      "learning_rate": 0.00021497404156638783,
+      "loss": 0.4488,
+      "step": 901
+    },
+    {
+      "epoch": 451.0,
+      "grad_norm": 6.1659016609191895,
+      "learning_rate": 0.00021398212104988275,
+      "loss": 0.4145,
+      "step": 902
+    },
+    {
+      "epoch": 451.5,
+      "grad_norm": 8.662652015686035,
+      "learning_rate": 0.00021299187087772282,
+      "loss": 0.4035,
+      "step": 903
+    },
+    {
+      "epoch": 452.0,
+      "grad_norm": 10.43588638305664,
+      "learning_rate": 0.0002120032968329687,
+      "loss": 0.6303,
+      "step": 904
+    },
+    {
+      "epoch": 452.5,
+      "grad_norm": 6.475193023681641,
+      "learning_rate": 0.00021101640468889255,
+      "loss": 0.5216,
+      "step": 905
+    },
+    {
+      "epoch": 453.0,
+      "grad_norm": 9.8986234664917,
+      "learning_rate": 0.0002100312002089441,
+      "loss": 0.4792,
+      "step": 906
+    },
+    {
+      "epoch": 453.5,
+      "grad_norm": 8.121114730834961,
+      "learning_rate": 0.00020904768914671758,
+      "loss": 0.4651,
+      "step": 907
+    },
+    {
+      "epoch": 454.0,
+      "grad_norm": 9.362919807434082,
+      "learning_rate": 0.00020806587724591725,
+      "loss": 0.5024,
+      "step": 908
+    },
+    {
+      "epoch": 454.5,
+      "grad_norm": 8.308518409729004,
+      "learning_rate": 0.00020708577024032432,
+      "loss": 0.4471,
+      "step": 909
+    },
+    {
+      "epoch": 455.0,
+      "grad_norm": 8.217002868652344,
+      "learning_rate": 0.00020610737385376348,
+      "loss": 0.4705,
+      "step": 910
+    },
+    {
+      "epoch": 455.5,
+      "grad_norm": 5.221889972686768,
+      "learning_rate": 0.00020513069380006943,
+      "loss": 0.4823,
+      "step": 911
+    },
+    {
+      "epoch": 456.0,
+      "grad_norm": 6.983361721038818,
+      "learning_rate": 0.0002041557357830534,
+      "loss": 0.4204,
+      "step": 912
+    },
+    {
+      "epoch": 456.5,
+      "grad_norm": 8.31548023223877,
+      "learning_rate": 0.00020318250549647,
+      "loss": 0.4054,
+      "step": 913
+    },
+    {
+      "epoch": 457.0,
+      "grad_norm": 7.561813831329346,
+      "learning_rate": 0.00020221100862398374,
+      "loss": 0.5175,
+      "step": 914
+    },
+    {
+      "epoch": 457.5,
+      "grad_norm": 8.128759384155273,
+      "learning_rate": 0.00020124125083913636,
+      "loss": 0.4015,
+      "step": 915
+    },
+    {
+      "epoch": 458.0,
+      "grad_norm": 8.550854682922363,
+      "learning_rate": 0.0002002732378053131,
+      "loss": 0.5006,
+      "step": 916
+    },
+    {
+      "epoch": 458.5,
+      "grad_norm": 5.205315113067627,
+      "learning_rate": 0.00019930697517570967,
+      "loss": 0.4451,
+      "step": 917
+    },
+    {
+      "epoch": 459.0,
+      "grad_norm": 6.101719379425049,
+      "learning_rate": 0.00019834246859329964,
+      "loss": 0.4797,
+      "step": 918
+    },
+    {
+      "epoch": 459.5,
+      "grad_norm": 6.14967679977417,
+      "learning_rate": 0.00019737972369080132,
+      "loss": 0.4055,
+      "step": 919
+    },
+    {
+      "epoch": 460.0,
+      "grad_norm": 6.501387119293213,
+      "learning_rate": 0.00019641874609064441,
+      "loss": 0.4801,
+      "step": 920
+    },
+    {
+      "epoch": 460.5,
+      "grad_norm": 5.227033615112305,
+      "learning_rate": 0.00019545954140493826,
+      "loss": 0.4993,
+      "step": 921
+    },
+    {
+      "epoch": 461.0,
+      "grad_norm": 5.5304131507873535,
+      "learning_rate": 0.00019450211523543793,
+      "loss": 0.3284,
+      "step": 922
+    },
+    {
+      "epoch": 461.5,
+      "grad_norm": 5.791502475738525,
+      "learning_rate": 0.00019354647317351188,
+      "loss": 0.3947,
+      "step": 923
+    },
+    {
+      "epoch": 462.0,
+      "grad_norm": 6.1968865394592285,
+      "learning_rate": 0.00019259262080010937,
+      "loss": 0.4366,
+      "step": 924
+    },
+    {
+      "epoch": 462.5,
+      "grad_norm": 4.806795120239258,
+      "learning_rate": 0.00019164056368572847,
+      "loss": 0.3604,
+      "step": 925
+    },
+    {
+      "epoch": 463.0,
+      "grad_norm": 15.549742698669434,
+      "learning_rate": 0.00019069030739038222,
+      "loss": 0.7783,
+      "step": 926
+    },
+    {
+      "epoch": 463.5,
+      "grad_norm": 4.787899971008301,
+      "learning_rate": 0.00018974185746356727,
+      "loss": 0.4362,
+      "step": 927
+    },
+    {
+      "epoch": 464.0,
+      "grad_norm": 6.144554615020752,
+      "learning_rate": 0.0001887952194442309,
+      "loss": 0.4773,
+      "step": 928
+    },
+    {
+      "epoch": 464.5,
+      "grad_norm": 8.66558837890625,
+      "learning_rate": 0.00018785039886073875,
+      "loss": 0.5853,
+      "step": 929
+    },
+    {
+      "epoch": 465.0,
+      "grad_norm": 7.957023620605469,
+      "learning_rate": 0.00018690740123084316,
+      "loss": 0.3319,
+      "step": 930
+    },
+    {
+      "epoch": 465.5,
+      "grad_norm": 10.916908264160156,
+      "learning_rate": 0.00018596623206164987,
+      "loss": 0.3959,
+      "step": 931
+    },
+    {
+      "epoch": 466.0,
+      "grad_norm": 11.649430274963379,
+      "learning_rate": 0.00018502689684958662,
+      "loss": 0.5191,
+      "step": 932
+    },
+    {
+      "epoch": 466.5,
+      "grad_norm": 4.922072410583496,
+      "learning_rate": 0.00018408940108037065,
+      "loss": 0.4837,
+      "step": 933
+    },
+    {
+      "epoch": 467.0,
+      "grad_norm": 7.171119689941406,
+      "learning_rate": 0.00018315375022897736,
+      "loss": 0.3768,
+      "step": 934
+    },
+    {
+      "epoch": 467.5,
+      "grad_norm": 7.603716850280762,
+      "learning_rate": 0.00018221994975960737,
+      "loss": 0.4797,
+      "step": 935
+    },
+    {
+      "epoch": 468.0,
+      "grad_norm": 6.712286949157715,
+      "learning_rate": 0.00018128800512565513,
+      "loss": 0.409,
+      "step": 936
+    },
+    {
+      "epoch": 468.5,
+      "grad_norm": 4.100458145141602,
+      "learning_rate": 0.00018035792176967718,
+      "loss": 0.4202,
+      "step": 937
+    },
+    {
+      "epoch": 469.0,
+      "grad_norm": 5.814347743988037,
+      "learning_rate": 0.00017942970512335998,
+      "loss": 0.4199,
+      "step": 938
+    },
+    {
+      "epoch": 469.5,
+      "grad_norm": 6.216732978820801,
+      "learning_rate": 0.0001785033606074886,
+      "loss": 0.4251,
+      "step": 939
+    },
+    {
+      "epoch": 470.0,
+      "grad_norm": 17.838729858398438,
+      "learning_rate": 0.00017757889363191482,
+      "loss": 0.6967,
+      "step": 940
+    },
+    {
+      "epoch": 470.5,
+      "grad_norm": 6.291789531707764,
+      "learning_rate": 0.00017665630959552548,
+      "loss": 0.4926,
+      "step": 941
+    },
+    {
+      "epoch": 471.0,
+      "grad_norm": 6.933482646942139,
+      "learning_rate": 0.00017573561388621101,
+      "loss": 0.4059,
+      "step": 942
+    },
+    {
+      "epoch": 471.5,
+      "grad_norm": 4.735661029815674,
+      "learning_rate": 0.00017481681188083443,
+      "loss": 0.4105,
+      "step": 943
+    },
+    {
+      "epoch": 472.0,
+      "grad_norm": 9.255976676940918,
+      "learning_rate": 0.0001738999089451991,
+      "loss": 0.4637,
+      "step": 944
+    },
+    {
+      "epoch": 472.5,
+      "grad_norm": 5.270608901977539,
+      "learning_rate": 0.00017298491043401794,
+      "loss": 0.4829,
+      "step": 945
+    },
+    {
+      "epoch": 473.0,
+      "grad_norm": 5.520230770111084,
+      "learning_rate": 0.00017207182169088204,
+      "loss": 0.351,
+      "step": 946
+    },
+    {
+      "epoch": 473.5,
+      "grad_norm": 3.8876707553863525,
+      "learning_rate": 0.0001711606480482293,
+      "loss": 0.418,
+      "step": 947
+    },
+    {
+      "epoch": 474.0,
+      "grad_norm": 7.245790481567383,
+      "learning_rate": 0.00017025139482731384,
+      "loss": 0.4415,
+      "step": 948
+    },
+    {
+      "epoch": 474.5,
+      "grad_norm": 5.0944437980651855,
+      "learning_rate": 0.00016934406733817414,
+      "loss": 0.3564,
+      "step": 949
+    },
+    {
+      "epoch": 475.0,
+      "grad_norm": 9.916528701782227,
+      "learning_rate": 0.00016843867087960252,
+      "loss": 0.5148,
+      "step": 950
+    },
+    {
+      "epoch": 475.5,
+      "grad_norm": 5.615230560302734,
+      "learning_rate": 0.00016753521073911386,
+      "loss": 0.4289,
+      "step": 951
+    },
+    {
+      "epoch": 476.0,
+      "grad_norm": 5.100493431091309,
+      "learning_rate": 0.00016663369219291558,
+      "loss": 0.4166,
+      "step": 952
+    },
+    {
+      "epoch": 476.5,
+      "grad_norm": 5.147010803222656,
+      "learning_rate": 0.00016573412050587555,
+      "loss": 0.5036,
+      "step": 953
+    },
+    {
+      "epoch": 477.0,
+      "grad_norm": 4.724948883056641,
+      "learning_rate": 0.00016483650093149227,
+      "loss": 0.3994,
+      "step": 954
+    },
+    {
+      "epoch": 477.5,
+      "grad_norm": 3.7687876224517822,
+      "learning_rate": 0.0001639408387118636,
+      "loss": 0.321,
+      "step": 955
+    },
+    {
+      "epoch": 478.0,
+      "grad_norm": 6.119095802307129,
+      "learning_rate": 0.00016304713907765712,
+      "loss": 0.4817,
+      "step": 956
+    },
+    {
+      "epoch": 478.5,
+      "grad_norm": 5.543955326080322,
+      "learning_rate": 0.00016215540724807815,
+      "loss": 0.4323,
+      "step": 957
+    },
+    {
+      "epoch": 479.0,
+      "grad_norm": 5.692309856414795,
+      "learning_rate": 0.00016126564843084052,
+      "loss": 0.3544,
+      "step": 958
+    },
+    {
+      "epoch": 479.5,
+      "grad_norm": 4.498508930206299,
+      "learning_rate": 0.00016037786782213548,
+      "loss": 0.4085,
+      "step": 959
+    },
+    {
+      "epoch": 480.0,
+      "grad_norm": 17.858257293701172,
+      "learning_rate": 0.00015949207060660136,
+      "loss": 0.9732,
+      "step": 960
+    },
+    {
+      "epoch": 480.5,
+      "grad_norm": 8.283931732177734,
+      "learning_rate": 0.000158608261957294,
+      "loss": 0.4838,
+      "step": 961
+    },
+    {
+      "epoch": 481.0,
+      "grad_norm": 10.169431686401367,
+      "learning_rate": 0.00015772644703565563,
+      "loss": 0.4338,
+      "step": 962
+    },
+    {
+      "epoch": 481.5,
+      "grad_norm": 5.234196662902832,
+      "learning_rate": 0.00015684663099148512,
+      "loss": 0.3952,
+      "step": 963
+    },
+    {
+      "epoch": 482.0,
+      "grad_norm": 7.689177989959717,
+      "learning_rate": 0.000155968818962908,
+      "loss": 0.368,
+      "step": 964
+    },
+    {
+      "epoch": 482.5,
+      "grad_norm": 4.742186546325684,
+      "learning_rate": 0.00015509301607634617,
+      "loss": 0.393,
+      "step": 965
+    },
+    {
+      "epoch": 483.0,
+      "grad_norm": 7.027538776397705,
+      "learning_rate": 0.00015421922744648846,
+      "loss": 0.3858,
+      "step": 966
+    },
+    {
+      "epoch": 483.5,
+      "grad_norm": 3.57450270652771,
+      "learning_rate": 0.00015334745817626022,
+      "loss": 0.3668,
+      "step": 967
+    },
+    {
+      "epoch": 484.0,
+      "grad_norm": 7.151853084564209,
+      "learning_rate": 0.00015247771335679373,
+      "loss": 0.5291,
+      "step": 968
+    },
+    {
+      "epoch": 484.5,
+      "grad_norm": 8.65969467163086,
+      "learning_rate": 0.00015160999806739835,
+      "loss": 0.4775,
+      "step": 969
+    },
+    {
+      "epoch": 485.0,
+      "grad_norm": 4.9497857093811035,
+      "learning_rate": 0.00015074431737553158,
+      "loss": 0.3075,
+      "step": 970
+    },
+    {
+      "epoch": 485.5,
+      "grad_norm": 5.261447429656982,
+      "learning_rate": 0.00014988067633676816,
+      "loss": 0.3473,
+      "step": 971
+    },
+    {
+      "epoch": 486.0,
+      "grad_norm": 5.390052795410156,
+      "learning_rate": 0.00014901907999477165,
+      "loss": 0.4247,
+      "step": 972
+    },
+    {
+      "epoch": 486.5,
+      "grad_norm": 5.303692817687988,
+      "learning_rate": 0.00014815953338126432,
+      "loss": 0.3939,
+      "step": 973
+    },
+    {
+      "epoch": 487.0,
+      "grad_norm": 6.357856750488281,
+      "learning_rate": 0.00014730204151599846,
+      "loss": 0.4388,
+      "step": 974
+    },
+    {
+      "epoch": 487.5,
+      "grad_norm": 5.3975629806518555,
+      "learning_rate": 0.00014644660940672628,
+      "loss": 0.5074,
+      "step": 975
+    },
+    {
+      "epoch": 488.0,
+      "grad_norm": 7.748287200927734,
+      "learning_rate": 0.000145593242049171,
+      "loss": 0.3908,
+      "step": 976
+    },
+    {
+      "epoch": 488.5,
+      "grad_norm": 6.594232559204102,
+      "learning_rate": 0.00014474194442699783,
+      "loss": 0.4309,
+      "step": 977
+    },
+    {
+      "epoch": 489.0,
+      "grad_norm": 6.133785247802734,
+      "learning_rate": 0.00014389272151178452,
+      "loss": 0.365,
+      "step": 978
+    },
+    {
+      "epoch": 489.5,
+      "grad_norm": 5.716797828674316,
+      "learning_rate": 0.00014304557826299297,
+      "loss": 0.3816,
+      "step": 979
+    },
+    {
+      "epoch": 490.0,
+      "grad_norm": 10.08061695098877,
+      "learning_rate": 0.00014220051962793951,
+      "loss": 0.5574,
+      "step": 980
+    },
+    {
+      "epoch": 490.5,
+      "grad_norm": 7.06821870803833,
+      "learning_rate": 0.0001413575505417662,
+      "loss": 0.3802,
+      "step": 981
+    },
+    {
+      "epoch": 491.0,
+      "grad_norm": 9.366924285888672,
+      "learning_rate": 0.0001405166759274123,
+      "loss": 0.4535,
+      "step": 982
+    },
+    {
+      "epoch": 491.5,
+      "grad_norm": 6.311436176300049,
+      "learning_rate": 0.00013967790069558555,
+      "loss": 0.4467,
+      "step": 983
+    },
+    {
+      "epoch": 492.0,
+      "grad_norm": 5.222879409790039,
+      "learning_rate": 0.00013884122974473307,
+      "loss": 0.3241,
+      "step": 984
+    },
+    {
+      "epoch": 492.5,
+      "grad_norm": 3.3343677520751953,
+      "learning_rate": 0.00013800666796101292,
+      "loss": 0.348,
+      "step": 985
+    },
+    {
+      "epoch": 493.0,
+      "grad_norm": 5.018557071685791,
+      "learning_rate": 0.00013717422021826569,
+      "loss": 0.4471,
+      "step": 986
+    },
+    {
+      "epoch": 493.5,
+      "grad_norm": 4.397199630737305,
+      "learning_rate": 0.00013634389137798608,
+      "loss": 0.4251,
+      "step": 987
+    },
+    {
+      "epoch": 494.0,
+      "grad_norm": 7.692255020141602,
+      "learning_rate": 0.00013551568628929433,
+      "loss": 0.3511,
+      "step": 988
+    },
+    {
+      "epoch": 494.5,
+      "grad_norm": 4.789150714874268,
+      "learning_rate": 0.00013468960978890794,
+      "loss": 0.4473,
+      "step": 989
+    },
+    {
+      "epoch": 495.0,
+      "grad_norm": 13.000507354736328,
+      "learning_rate": 0.0001338656667011134,
+      "loss": 0.465,
+      "step": 990
+    },
+    {
+      "epoch": 495.5,
+      "grad_norm": 5.734309673309326,
+      "learning_rate": 0.0001330438618377381,
+      "loss": 0.3618,
+      "step": 991
+    },
+    {
+      "epoch": 496.0,
+      "grad_norm": 5.070725917816162,
+      "learning_rate": 0.00013222419999812246,
+      "loss": 0.3947,
+      "step": 992
+    },
+    {
+      "epoch": 496.5,
+      "grad_norm": 4.779550075531006,
+      "learning_rate": 0.00013140668596909133,
+      "loss": 0.4438,
+      "step": 993
+    },
+    {
+      "epoch": 497.0,
+      "grad_norm": 8.357994079589844,
+      "learning_rate": 0.00013059132452492651,
+      "loss": 0.4584,
+      "step": 994
+    },
+    {
+      "epoch": 497.5,
+      "grad_norm": 4.80888032913208,
+      "learning_rate": 0.0001297781204273385,
+      "loss": 0.3433,
+      "step": 995
+    },
+    {
+      "epoch": 498.0,
+      "grad_norm": 5.5599164962768555,
+      "learning_rate": 0.00012896707842543898,
+      "loss": 0.459,
+      "step": 996
+    },
+    {
+      "epoch": 498.5,
+      "grad_norm": 6.4827752113342285,
+      "learning_rate": 0.00012815820325571325,
+      "loss": 0.3157,
+      "step": 997
+    },
+    {
+      "epoch": 499.0,
+      "grad_norm": 8.241578102111816,
+      "learning_rate": 0.0001273514996419921,
+      "loss": 0.5344,
+      "step": 998
+    },
+    {
+      "epoch": 499.5,
+      "grad_norm": 4.791350841522217,
+      "learning_rate": 0.00012654697229542438,
+      "loss": 0.3443,
+      "step": 999
+    },
+    {
+      "epoch": 500.0,
+      "grad_norm": 8.930863380432129,
+      "learning_rate": 0.0001257446259144494,
+      "loss": 0.4661,
+      "step": 1000
+    },
+    {
+      "epoch": 500.5,
+      "grad_norm": 5.738401412963867,
+      "learning_rate": 0.0001249444651847702,
+      "loss": 0.3884,
+      "step": 1001
+    },
+    {
+      "epoch": 501.0,
+      "grad_norm": 4.781877040863037,
+      "learning_rate": 0.0001241464947793251,
+      "loss": 0.4991,
+      "step": 1002
+    },
+    {
+      "epoch": 501.5,
+      "grad_norm": 5.195380687713623,
+      "learning_rate": 0.0001233507193582611,
+      "loss": 0.4672,
+      "step": 1003
+    },
+    {
+      "epoch": 502.0,
+      "grad_norm": 4.799306392669678,
+      "learning_rate": 0.0001225571435689062,
+      "loss": 0.2987,
+      "step": 1004
+    },
+    {
+      "epoch": 502.5,
+      "grad_norm": 4.244041442871094,
+      "learning_rate": 0.00012176577204574318,
+      "loss": 0.3846,
+      "step": 1005
+    },
+    {
+      "epoch": 503.0,
+      "grad_norm": 4.8642120361328125,
+      "learning_rate": 0.00012097660941038147,
+      "loss": 0.3848,
+      "step": 1006
+    },
+    {
+      "epoch": 503.5,
+      "grad_norm": 4.1718430519104,
+      "learning_rate": 0.0001201896602715306,
+      "loss": 0.4666,
+      "step": 1007
+    },
+    {
+      "epoch": 504.0,
+      "grad_norm": 24.06111717224121,
+      "learning_rate": 0.00011940492922497337,
+      "loss": 0.6959,
+      "step": 1008
+    },
+    {
+      "epoch": 504.5,
+      "grad_norm": 6.994117259979248,
+      "learning_rate": 0.0001186224208535388,
+      "loss": 0.4226,
+      "step": 1009
+    },
+    {
+      "epoch": 505.0,
+      "grad_norm": 37.89023971557617,
+      "learning_rate": 0.0001178421397270758,
+      "loss": 0.4157,
+      "step": 1010
+    },
+    {
+      "epoch": 505.5,
+      "grad_norm": 6.107475757598877,
+      "learning_rate": 0.00011706409040242588,
+      "loss": 0.4958,
+      "step": 1011
+    },
+    {
+      "epoch": 506.0,
+      "grad_norm": 6.104827880859375,
+      "learning_rate": 0.00011628827742339687,
+      "loss": 0.3404,
+      "step": 1012
+    },
+    {
+      "epoch": 506.5,
+      "grad_norm": 6.03773832321167,
+      "learning_rate": 0.00011551470532073643,
+      "loss": 0.3242,
+      "step": 1013
+    },
+    {
+      "epoch": 507.0,
+      "grad_norm": 5.51300573348999,
+      "learning_rate": 0.00011474337861210544,
+      "loss": 0.5134,
+      "step": 1014
+    },
+    {
+      "epoch": 507.5,
+      "grad_norm": 4.085872650146484,
+      "learning_rate": 0.00011397430180205171,
+      "loss": 0.3406,
+      "step": 1015
+    },
+    {
+      "epoch": 508.0,
+      "grad_norm": 7.762851238250732,
+      "learning_rate": 0.00011320747938198356,
+      "loss": 0.4881,
+      "step": 1016
+    },
+    {
+      "epoch": 508.5,
+      "grad_norm": 5.115671157836914,
+      "learning_rate": 0.00011244291583014377,
+      "loss": 0.3987,
+      "step": 1017
+    },
+    {
+      "epoch": 509.0,
+      "grad_norm": 6.653552532196045,
+      "learning_rate": 0.00011168061561158321,
+      "loss": 0.3197,
+      "step": 1018
+    },
+    {
+      "epoch": 509.5,
+      "grad_norm": 5.976140975952148,
+      "learning_rate": 0.00011092058317813519,
+      "loss": 0.4605,
+      "step": 1019
+    },
+    {
+      "epoch": 510.0,
+      "grad_norm": 6.265192031860352,
+      "learning_rate": 0.00011016282296838886,
+      "loss": 0.3461,
+      "step": 1020
+    },
+    {
+      "epoch": 510.5,
+      "grad_norm": 7.543193340301514,
+      "learning_rate": 0.00010940733940766368,
+      "loss": 0.4457,
+      "step": 1021
+    },
+    {
+      "epoch": 511.0,
+      "grad_norm": 5.622101306915283,
+      "learning_rate": 0.00010865413690798321,
+      "loss": 0.2977,
+      "step": 1022
+    },
+    {
+      "epoch": 511.5,
+      "grad_norm": 3.9157474040985107,
+      "learning_rate": 0.00010790321986805024,
+      "loss": 0.3434,
+      "step": 1023
+    },
+    {
+      "epoch": 512.0,
+      "grad_norm": 5.5897111892700195,
+      "learning_rate": 0.00010715459267321997,
+      "loss": 0.4537,
+      "step": 1024
+    },
+    {
+      "epoch": 512.5,
+      "grad_norm": 4.161907196044922,
+      "learning_rate": 0.00010640825969547497,
+      "loss": 0.4286,
+      "step": 1025
+    },
+    {
+      "epoch": 513.0,
+      "grad_norm": 6.199113368988037,
+      "learning_rate": 0.0001056642252933997,
+      "loss": 0.3142,
+      "step": 1026
+    },
+    {
+      "epoch": 513.5,
+      "grad_norm": 4.078858852386475,
+      "learning_rate": 0.00010492249381215479,
+      "loss": 0.3703,
+      "step": 1027
+    },
+    {
+      "epoch": 514.0,
+      "grad_norm": 9.558767318725586,
+      "learning_rate": 0.00010418306958345213,
+      "loss": 0.4079,
+      "step": 1028
+    },
+    {
+      "epoch": 514.5,
+      "grad_norm": 4.761571407318115,
+      "learning_rate": 0.00010344595692552894,
+      "loss": 0.394,
+      "step": 1029
+    },
+    {
+      "epoch": 515.0,
+      "grad_norm": 7.909276008605957,
+      "learning_rate": 0.00010271116014312292,
+      "loss": 0.3326,
+      "step": 1030
+    },
+    {
+      "epoch": 515.5,
+      "grad_norm": 6.386188507080078,
+      "learning_rate": 0.000101978683527447,
+      "loss": 0.3796,
+      "step": 1031
+    },
+    {
+      "epoch": 516.0,
+      "grad_norm": 5.595881938934326,
+      "learning_rate": 0.00010124853135616475,
+      "loss": 0.3728,
+      "step": 1032
+    },
+    {
+      "epoch": 516.5,
+      "grad_norm": 4.94740629196167,
+      "learning_rate": 0.00010052070789336443,
+      "loss": 0.3789,
+      "step": 1033
+    },
+    {
+      "epoch": 517.0,
+      "grad_norm": 7.455836772918701,
+      "learning_rate": 9.9795217389535e-05,
+      "loss": 0.3287,
+      "step": 1034
+    },
+    {
+      "epoch": 517.5,
+      "grad_norm": 7.743784427642822,
+      "learning_rate": 9.907206408154079e-05,
+      "loss": 0.3538,
+      "step": 1035
+    },
+    {
+      "epoch": 518.0,
+      "grad_norm": 7.105027198791504,
+      "learning_rate": 9.835125219259695e-05,
+      "loss": 0.621,
+      "step": 1036
+    },
+    {
+      "epoch": 518.5,
+      "grad_norm": 5.381617546081543,
+      "learning_rate": 9.763278593224462e-05,
+      "loss": 0.4051,
+      "step": 1037
+    },
+    {
+      "epoch": 519.0,
+      "grad_norm": 5.529709339141846,
+      "learning_rate": 9.691666949632683e-05,
+      "loss": 0.4044,
+      "step": 1038
+    },
+    {
+      "epoch": 519.5,
+      "grad_norm": 4.6511993408203125,
+      "learning_rate": 9.620290706696322e-05,
+      "loss": 0.336,
+      "step": 1039
+    },
+    {
+      "epoch": 520.0,
+      "grad_norm": 5.753106117248535,
+      "learning_rate": 9.549150281252633e-05,
+      "loss": 0.4059,
+      "step": 1040
+    },
+    {
+      "epoch": 520.5,
+      "grad_norm": 3.1649749279022217,
+      "learning_rate": 9.478246088761671e-05,
+      "loss": 0.3182,
+      "step": 1041
+    },
+    {
+      "epoch": 521.0,
+      "grad_norm": 6.778075218200684,
+      "learning_rate": 9.407578543303913e-05,
+      "loss": 0.5503,
+      "step": 1042
+    },
+    {
+      "epoch": 521.5,
+      "grad_norm": 5.444674968719482,
+      "learning_rate": 9.3371480575778e-05,
+      "loss": 0.3873,
+      "step": 1043
+    },
+    {
+      "epoch": 522.0,
+      "grad_norm": 7.651487350463867,
+      "learning_rate": 9.266955042897358e-05,
+      "loss": 0.631,
+      "step": 1044
+    },
+    {
+      "epoch": 522.5,
+      "grad_norm": 4.1143317222595215,
+      "learning_rate": 9.196999909189763e-05,
+      "loss": 0.3124,
+      "step": 1045
+    },
+    {
+      "epoch": 523.0,
+      "grad_norm": 5.250977039337158,
+      "learning_rate": 9.127283064992997e-05,
+      "loss": 0.5015,
+      "step": 1046
+    },
+    {
+      "epoch": 523.5,
+      "grad_norm": 5.0915632247924805,
+      "learning_rate": 9.05780491745341e-05,
+      "loss": 0.3233,
+      "step": 1047
+    },
+    {
+      "epoch": 524.0,
+      "grad_norm": 5.412391185760498,
+      "learning_rate": 8.988565872323362e-05,
+      "loss": 0.4118,
+      "step": 1048
+    },
+    {
+      "epoch": 524.5,
+      "grad_norm": 4.853921890258789,
+      "learning_rate": 8.919566333958845e-05,
+      "loss": 0.3408,
+      "step": 1049
+    },
+    {
+      "epoch": 525.0,
+      "grad_norm": 5.188722133636475,
+      "learning_rate": 8.850806705317183e-05,
+      "loss": 0.3897,
+      "step": 1050
+    },
+    {
+      "epoch": 525.5,
+      "grad_norm": 4.502488136291504,
+      "learning_rate": 8.782287387954563e-05,
+      "loss": 0.4189,
+      "step": 1051
+    },
+    {
+      "epoch": 526.0,
+      "grad_norm": 3.9045989513397217,
+      "learning_rate": 8.714008782023796e-05,
+      "loss": 0.3246,
+      "step": 1052
+    },
+    {
+      "epoch": 526.5,
+      "grad_norm": 5.067163467407227,
+      "learning_rate": 8.645971286271903e-05,
+      "loss": 0.4849,
+      "step": 1053
+    },
+    {
+      "epoch": 527.0,
+      "grad_norm": 3.5333404541015625,
+      "learning_rate": 8.578175298037872e-05,
+      "loss": 0.3031,
+      "step": 1054
+    },
+    {
+      "epoch": 527.5,
+      "grad_norm": 3.9064810276031494,
+      "learning_rate": 8.510621213250246e-05,
+      "loss": 0.3609,
+      "step": 1055
+    },
+    {
+      "epoch": 528.0,
+      "grad_norm": 4.286402225494385,
+      "learning_rate": 8.443309426424861e-05,
+      "loss": 0.3398,
+      "step": 1056
+    },
+    {
+      "epoch": 528.5,
+      "grad_norm": 3.6806812286376953,
+      "learning_rate": 8.376240330662538e-05,
+      "loss": 0.379,
+      "step": 1057
+    },
+    {
+      "epoch": 529.0,
+      "grad_norm": 5.866394519805908,
+      "learning_rate": 8.309414317646769e-05,
+      "loss": 0.3236,
+      "step": 1058
+    },
+    {
+      "epoch": 529.5,
+      "grad_norm": 3.4704389572143555,
+      "learning_rate": 8.242831777641474e-05,
+      "loss": 0.2912,
+      "step": 1059
+    },
+    {
+      "epoch": 530.0,
+      "grad_norm": 5.50031042098999,
+      "learning_rate": 8.176493099488664e-05,
+      "loss": 0.487,
+      "step": 1060
+    },
+    {
+      "epoch": 530.5,
+      "grad_norm": 4.342918872833252,
+      "learning_rate": 8.1103986706062e-05,
+      "loss": 0.3729,
+      "step": 1061
+    },
+    {
+      "epoch": 531.0,
+      "grad_norm": 4.739935398101807,
+      "learning_rate": 8.04454887698553e-05,
+      "loss": 0.3319,
+      "step": 1062
+    },
+    {
+      "epoch": 531.5,
+      "grad_norm": 4.642815589904785,
+      "learning_rate": 7.97894410318944e-05,
+      "loss": 0.4028,
+      "step": 1063
+    },
+    {
+      "epoch": 532.0,
+      "grad_norm": 4.825720310211182,
+      "learning_rate": 7.913584732349787e-05,
+      "loss": 0.3468,
+      "step": 1064
+    },
+    {
+      "epoch": 532.5,
+      "grad_norm": 4.816979885101318,
+      "learning_rate": 7.848471146165287e-05,
+      "loss": 0.4116,
+      "step": 1065
+    },
+    {
+      "epoch": 533.0,
+      "grad_norm": 6.612005233764648,
+      "learning_rate": 7.783603724899258e-05,
+      "loss": 0.2982,
+      "step": 1066
+    },
+    {
+      "epoch": 533.5,
+      "grad_norm": 4.627198696136475,
+      "learning_rate": 7.71898284737742e-05,
+      "loss": 0.3676,
+      "step": 1067
+    },
+    {
+      "epoch": 534.0,
+      "grad_norm": 6.4206719398498535,
+      "learning_rate": 7.654608890985709e-05,
+      "loss": 0.4145,
+      "step": 1068
+    },
+    {
+      "epoch": 534.5,
+      "grad_norm": 5.362915515899658,
+      "learning_rate": 7.590482231667994e-05,
+      "loss": 0.3825,
+      "step": 1069
+    },
+    {
+      "epoch": 535.0,
+      "grad_norm": 6.360662460327148,
+      "learning_rate": 7.526603243923958e-05,
+      "loss": 0.421,
+      "step": 1070
+    },
+    {
+      "epoch": 535.5,
+      "grad_norm": 5.304653644561768,
+      "learning_rate": 7.462972300806858e-05,
+      "loss": 0.4078,
+      "step": 1071
+    },
+    {
+      "epoch": 536.0,
+      "grad_norm": 6.355183124542236,
+      "learning_rate": 7.399589773921412e-05,
+      "loss": 0.339,
+      "step": 1072
+    },
+    {
+      "epoch": 536.5,
+      "grad_norm": 4.909663200378418,
+      "learning_rate": 7.336456033421551e-05,
+      "loss": 0.3856,
+      "step": 1073
+    },
+    {
+      "epoch": 537.0,
+      "grad_norm": 9.781787872314453,
+      "learning_rate": 7.273571448008304e-05,
+      "loss": 0.4359,
+      "step": 1074
+    },
+    {
+      "epoch": 537.5,
+      "grad_norm": 4.054769992828369,
+      "learning_rate": 7.21093638492763e-05,
+      "loss": 0.4207,
+      "step": 1075
+    },
+    {
+      "epoch": 538.0,
+      "grad_norm": 5.431164741516113,
+      "learning_rate": 7.148551209968279e-05,
+      "loss": 0.3386,
+      "step": 1076
+    },
+    {
+      "epoch": 538.5,
+      "grad_norm": 5.20838737487793,
+      "learning_rate": 7.086416287459679e-05,
+      "loss": 0.3063,
+      "step": 1077
+    },
+    {
+      "epoch": 539.0,
+      "grad_norm": 7.026962757110596,
+      "learning_rate": 7.024531980269744e-05,
+      "loss": 0.4899,
+      "step": 1078
+    },
+    {
+      "epoch": 539.5,
+      "grad_norm": 5.351546287536621,
+      "learning_rate": 6.962898649802824e-05,
+      "loss": 0.5198,
+      "step": 1079
+    },
+    {
+      "epoch": 540.0,
+      "grad_norm": 4.879822731018066,
+      "learning_rate": 6.901516655997537e-05,
+      "loss": 0.2741,
+      "step": 1080
+    },
+    {
+      "epoch": 540.5,
+      "grad_norm": 3.786128520965576,
+      "learning_rate": 6.84038635732473e-05,
+      "loss": 0.3968,
+      "step": 1081
+    },
+    {
+      "epoch": 541.0,
+      "grad_norm": 7.543567180633545,
+      "learning_rate": 6.779508110785331e-05,
+      "loss": 0.4642,
+      "step": 1082
+    },
+    {
+      "epoch": 541.5,
+      "grad_norm": 4.167280673980713,
+      "learning_rate": 6.718882271908278e-05,
+      "loss": 0.3149,
+      "step": 1083
+    },
+    {
+      "epoch": 542.0,
+      "grad_norm": 17.56368064880371,
+      "learning_rate": 6.658509194748463e-05,
+      "loss": 0.6839,
+      "step": 1084
+    },
+    {
+      "epoch": 542.5,
+      "grad_norm": 3.6284666061401367,
+      "learning_rate": 6.598389231884628e-05,
+      "loss": 0.3534,
+      "step": 1085
+    },
+    {
+      "epoch": 543.0,
+      "grad_norm": 5.000655174255371,
+      "learning_rate": 6.538522734417357e-05,
+      "loss": 0.3762,
+      "step": 1086
+    },
+    {
+      "epoch": 543.5,
+      "grad_norm": 4.177554130554199,
+      "learning_rate": 6.47891005196698e-05,
+      "loss": 0.3389,
+      "step": 1087
+    },
+    {
+      "epoch": 544.0,
+      "grad_norm": 3.4086899757385254,
+      "learning_rate": 6.419551532671541e-05,
+      "loss": 0.3546,
+      "step": 1088
+    },
+    {
+      "epoch": 544.5,
+      "grad_norm": 4.838875770568848,
+      "learning_rate": 6.360447523184781e-05,
+      "loss": 0.4727,
+      "step": 1089
+    },
+    {
+      "epoch": 545.0,
+      "grad_norm": 3.9170193672180176,
+      "learning_rate": 6.301598368674105e-05,
+      "loss": 0.3687,
+      "step": 1090
+    },
+    {
+      "epoch": 545.5,
+      "grad_norm": 4.605557918548584,
+      "learning_rate": 6.243004412818559e-05,
+      "loss": 0.4083,
+      "step": 1091
+    },
+    {
+      "epoch": 546.0,
+      "grad_norm": 5.223203659057617,
+      "learning_rate": 6.184665997806832e-05,
+      "loss": 0.3547,
+      "step": 1092
+    },
+    {
+      "epoch": 546.5,
+      "grad_norm": 4.328655242919922,
+      "learning_rate": 6.126583464335239e-05,
+      "loss": 0.4399,
+      "step": 1093
+    },
+    {
+      "epoch": 547.0,
+      "grad_norm": 3.5133752822875977,
+      "learning_rate": 6.0687571516057803e-05,
+      "loss": 0.2432,
+      "step": 1094
+    },
+    {
+      "epoch": 547.5,
+      "grad_norm": 3.508127212524414,
+      "learning_rate": 6.011187397324114e-05,
+      "loss": 0.3199,
+      "step": 1095
+    },
+    {
+      "epoch": 548.0,
+      "grad_norm": 4.919780731201172,
+      "learning_rate": 5.9538745376975736e-05,
+      "loss": 0.4319,
+      "step": 1096
+    },
+    {
+      "epoch": 548.5,
+      "grad_norm": 4.67732048034668,
+      "learning_rate": 5.896818907433255e-05,
+      "loss": 0.5024,
+      "step": 1097
+    },
+    {
+      "epoch": 549.0,
+      "grad_norm": 5.879189968109131,
+      "learning_rate": 5.84002083973601e-05,
+      "loss": 0.3589,
+      "step": 1098
+    },
+    {
+      "epoch": 549.5,
+      "grad_norm": 4.247766971588135,
+      "learning_rate": 5.783480666306568e-05,
+      "loss": 0.3647,
+      "step": 1099
+    },
+    {
+      "epoch": 550.0,
+      "grad_norm": 21.226205825805664,
+      "learning_rate": 5.72719871733951e-05,
+      "loss": 0.5855,
+      "step": 1100
+    },
+    {
+      "epoch": 550.5,
+      "grad_norm": 3.5221686363220215,
+      "learning_rate": 5.6711753215214104e-05,
+      "loss": 0.3381,
+      "step": 1101
+    },
+    {
+      "epoch": 551.0,
+      "grad_norm": 11.048192977905273,
+      "learning_rate": 5.6154108060288756e-05,
+      "loss": 0.4596,
+      "step": 1102
+    },
+    {
+      "epoch": 551.5,
+      "grad_norm": 4.459977149963379,
+      "learning_rate": 5.5599054965266795e-05,
+      "loss": 0.319,
+      "step": 1103
+    },
+    {
+      "epoch": 552.0,
+      "grad_norm": 6.926024913787842,
+      "learning_rate": 5.5046597171658106e-05,
+      "loss": 0.4324,
+      "step": 1104
+    },
+    {
+      "epoch": 552.5,
+      "grad_norm": 4.505620002746582,
+      "learning_rate": 5.449673790581611e-05,
+      "loss": 0.3909,
+      "step": 1105
+    },
+    {
+      "epoch": 553.0,
+      "grad_norm": 7.049350261688232,
+      "learning_rate": 5.394948037891867e-05,
+      "loss": 0.4019,
+      "step": 1106
+    },
+    {
+      "epoch": 553.5,
+      "grad_norm": 4.105534076690674,
+      "learning_rate": 5.340482778694966e-05,
+      "loss": 0.3473,
+      "step": 1107
+    },
+    {
+      "epoch": 554.0,
+      "grad_norm": 5.129210472106934,
+      "learning_rate": 5.286278331068017e-05,
+      "loss": 0.4651,
+      "step": 1108
+    },
+    {
+      "epoch": 554.5,
+      "grad_norm": 3.4428176879882812,
+      "learning_rate": 5.2323350115649756e-05,
+      "loss": 0.2715,
+      "step": 1109
+    },
+    {
+      "epoch": 555.0,
+      "grad_norm": 7.101128101348877,
+      "learning_rate": 5.1786531352148115e-05,
+      "loss": 0.5381,
+      "step": 1110
+    },
+    {
+      "epoch": 555.5,
+      "grad_norm": 3.685354471206665,
+      "learning_rate": 5.125233015519676e-05,
+      "loss": 0.3892,
+      "step": 1111
+    },
+    {
+      "epoch": 556.0,
+      "grad_norm": 3.8293871879577637,
+      "learning_rate": 5.072074964453055e-05,
+      "loss": 0.2907,
+      "step": 1112
+    },
+    {
+      "epoch": 556.5,
+      "grad_norm": 4.309526443481445,
+      "learning_rate": 5.0191792924579485e-05,
+      "loss": 0.3554,
+      "step": 1113
+    },
+    {
+      "epoch": 557.0,
+      "grad_norm": 5.190085411071777,
+      "learning_rate": 4.966546308445074e-05,
+      "loss": 0.3533,
+      "step": 1114
+    },
+    {
+      "epoch": 557.5,
+      "grad_norm": 3.6892218589782715,
+      "learning_rate": 4.914176319791036e-05,
+      "loss": 0.3288,
+      "step": 1115
+    },
+    {
+      "epoch": 558.0,
+      "grad_norm": 5.344886779785156,
+      "learning_rate": 4.862069632336558e-05,
+      "loss": 0.4124,
+      "step": 1116
+    },
+    {
+      "epoch": 558.5,
+      "grad_norm": 3.764634132385254,
+      "learning_rate": 4.8102265503847e-05,
+      "loss": 0.3526,
+      "step": 1117
+    },
+    {
+      "epoch": 559.0,
+      "grad_norm": 17.656734466552734,
+      "learning_rate": 4.758647376699032e-05,
+      "loss": 0.6084,
+      "step": 1118
+    },
+    {
+      "epoch": 559.5,
+      "grad_norm": 4.092284202575684,
+      "learning_rate": 4.707332412501919e-05,
+      "loss": 0.3951,
+      "step": 1119
+    },
+    {
+      "epoch": 560.0,
+      "grad_norm": 3.377016305923462,
+      "learning_rate": 4.65628195747273e-05,
+      "loss": 0.2959,
+      "step": 1120
+    },
+    {
+      "epoch": 560.5,
+      "grad_norm": 3.9937169551849365,
+      "learning_rate": 4.605496309746127e-05,
+      "loss": 0.3967,
+      "step": 1121
+    },
+    {
+      "epoch": 561.0,
+      "grad_norm": 3.858347177505493,
+      "learning_rate": 4.5549757659102795e-05,
+      "loss": 0.296,
+      "step": 1122
+    },
+    {
+      "epoch": 561.5,
+      "grad_norm": 4.4640092849731445,
+      "learning_rate": 4.5047206210051426e-05,
+      "loss": 0.3005,
+      "step": 1123
+    },
+    {
+      "epoch": 562.0,
+      "grad_norm": 7.387597560882568,
+      "learning_rate": 4.4547311685207536e-05,
+      "loss": 0.4325,
+      "step": 1124
+    },
+    {
+      "epoch": 562.5,
+      "grad_norm": 3.441588878631592,
+      "learning_rate": 4.405007700395497e-05,
+      "loss": 0.343,
+      "step": 1125
+    },
+    {
+      "epoch": 563.0,
+      "grad_norm": 4.963720321655273,
+      "learning_rate": 4.3555505070144276e-05,
+      "loss": 0.3611,
+      "step": 1126
+    },
+    {
+      "epoch": 563.5,
+      "grad_norm": 3.0999765396118164,
+      "learning_rate": 4.306359877207538e-05,
+      "loss": 0.3693,
+      "step": 1127
+    },
+    {
+      "epoch": 564.0,
+      "grad_norm": 4.521252155303955,
+      "learning_rate": 4.257436098248091e-05,
+      "loss": 0.2997,
+      "step": 1128
+    },
+    {
+      "epoch": 564.5,
+      "grad_norm": 4.467068672180176,
+      "learning_rate": 4.208779455850942e-05,
+      "loss": 0.3511,
+      "step": 1129
+    },
+    {
+      "epoch": 565.0,
+      "grad_norm": 9.471026420593262,
+      "learning_rate": 4.16039023417088e-05,
+      "loss": 0.5284,
+      "step": 1130
+    },
+    {
+      "epoch": 565.5,
+      "grad_norm": 3.433380126953125,
+      "learning_rate": 4.112268715800943e-05,
+      "loss": 0.3337,
+      "step": 1131
+    },
+    {
+      "epoch": 566.0,
+      "grad_norm": 3.401965856552124,
+      "learning_rate": 4.0644151817707866e-05,
+      "loss": 0.3856,
+      "step": 1132
+    },
+    {
+      "epoch": 566.5,
+      "grad_norm": 3.8238401412963867,
+      "learning_rate": 4.0168299115450445e-05,
+      "loss": 0.4004,
+      "step": 1133
+    },
+    {
+      "epoch": 567.0,
+      "grad_norm": 3.430161476135254,
+      "learning_rate": 3.9695131830216616e-05,
+      "loss": 0.3297,
+      "step": 1134
+    },
+    {
+      "epoch": 567.5,
+      "grad_norm": 5.321807861328125,
+      "learning_rate": 3.922465272530351e-05,
+      "loss": 0.378,
+      "step": 1135
+    },
+    {
+      "epoch": 568.0,
+      "grad_norm": 5.820250988006592,
+      "learning_rate": 3.8756864548308845e-05,
+      "loss": 0.3594,
+      "step": 1136
+    },
+    {
+      "epoch": 568.5,
+      "grad_norm": 3.652477741241455,
+      "learning_rate": 3.8291770031115446e-05,
+      "loss": 0.3344,
+      "step": 1137
+    },
+    {
+      "epoch": 569.0,
+      "grad_norm": 7.754717826843262,
+      "learning_rate": 3.782937188987523e-05,
+      "loss": 0.5072,
+      "step": 1138
+    },
+    {
+      "epoch": 569.5,
+      "grad_norm": 3.714512586593628,
+      "learning_rate": 3.73696728249932e-05,
+      "loss": 0.3569,
+      "step": 1139
+    },
+    {
+      "epoch": 570.0,
+      "grad_norm": 8.195136070251465,
+      "learning_rate": 3.691267552111183e-05,
+      "loss": 0.4394,
+      "step": 1140
+    },
+    {
+      "epoch": 570.5,
+      "grad_norm": 3.334235191345215,
+      "learning_rate": 3.6458382647095164e-05,
+      "loss": 0.3002,
+      "step": 1141
+    },
+    {
+      "epoch": 571.0,
+      "grad_norm": 4.835788726806641,
+      "learning_rate": 3.6006796856013493e-05,
+      "loss": 0.433,
+      "step": 1142
+    },
+    {
+      "epoch": 571.5,
+      "grad_norm": 3.9299495220184326,
+      "learning_rate": 3.555792078512787e-05,
+      "loss": 0.3968,
+      "step": 1143
+    },
+    {
+      "epoch": 572.0,
+      "grad_norm": 4.243776798248291,
+      "learning_rate": 3.5111757055874326e-05,
+      "loss": 0.3452,
+      "step": 1144
+    },
+    {
+      "epoch": 572.5,
+      "grad_norm": 3.818819046020508,
+      "learning_rate": 3.466830827384898e-05,
+      "loss": 0.3706,
+      "step": 1145
+    },
+    {
+      "epoch": 573.0,
+      "grad_norm": 6.275313377380371,
+      "learning_rate": 3.422757702879259e-05,
+      "loss": 0.31,
+      "step": 1146
+    },
+    {
+      "epoch": 573.5,
+      "grad_norm": 4.341431140899658,
+      "learning_rate": 3.378956589457543e-05,
+      "loss": 0.3179,
+      "step": 1147
+    },
+    {
+      "epoch": 574.0,
+      "grad_norm": 6.777219295501709,
+      "learning_rate": 3.335427742918262e-05,
+      "loss": 0.6169,
+      "step": 1148
+    },
+    {
+      "epoch": 574.5,
+      "grad_norm": 4.447834491729736,
+      "learning_rate": 3.292171417469858e-05,
+      "loss": 0.3131,
+      "step": 1149
+    },
+    {
+      "epoch": 575.0,
+      "grad_norm": 5.12947940826416,
+      "learning_rate": 3.249187865729264e-05,
+      "loss": 0.4102,
+      "step": 1150
+    },
+    {
+      "epoch": 575.5,
+      "grad_norm": 2.9352805614471436,
+      "learning_rate": 3.206477338720398e-05,
+      "loss": 0.3037,
+      "step": 1151
+    },
+    {
+      "epoch": 576.0,
+      "grad_norm": 10.469058990478516,
+      "learning_rate": 3.164040085872755e-05,
+      "loss": 0.5653,
+      "step": 1152
+    },
+    {
+      "epoch": 576.5,
+      "grad_norm": 3.5477986335754395,
+      "learning_rate": 3.121876355019859e-05,
+      "loss": 0.2955,
+      "step": 1153
+    },
+    {
+      "epoch": 577.0,
+      "grad_norm": 4.3201069831848145,
+      "learning_rate": 3.079986392397899e-05,
+      "loss": 0.3803,
+      "step": 1154
+    },
+    {
+      "epoch": 577.5,
+      "grad_norm": 2.8223788738250732,
+      "learning_rate": 3.0383704426442393e-05,
+      "loss": 0.3088,
+      "step": 1155
+    },
+    {
+      "epoch": 578.0,
+      "grad_norm": 5.064321041107178,
+      "learning_rate": 2.9970287487960158e-05,
+      "loss": 0.4505,
+      "step": 1156
+    },
+    {
+      "epoch": 578.5,
+      "grad_norm": 3.5879404544830322,
+      "learning_rate": 2.9559615522887274e-05,
+      "loss": 0.3696,
+      "step": 1157
+    },
+    {
+      "epoch": 579.0,
+      "grad_norm": 6.456601619720459,
+      "learning_rate": 2.9151690929547726e-05,
+      "loss": 0.3103,
+      "step": 1158
+    },
+    {
+      "epoch": 579.5,
+      "grad_norm": 4.017468452453613,
+      "learning_rate": 2.8746516090221163e-05,
+      "loss": 0.3426,
+      "step": 1159
+    },
+    {
+      "epoch": 580.0,
+      "grad_norm": 5.80248498916626,
+      "learning_rate": 2.834409337112842e-05,
+      "loss": 0.4175,
+      "step": 1160
+    },
+    {
+      "epoch": 580.5,
+      "grad_norm": 3.2507662773132324,
+      "learning_rate": 2.7944425122418238e-05,
+      "loss": 0.3346,
+      "step": 1161
+    },
+    {
+      "epoch": 581.0,
+      "grad_norm": 4.0719475746154785,
+      "learning_rate": 2.7547513678153e-05,
+      "loss": 0.348,
+      "step": 1162
+    },
+    {
+      "epoch": 581.5,
+      "grad_norm": 2.771308422088623,
+      "learning_rate": 2.715336135629548e-05,
+      "loss": 0.3083,
+      "step": 1163
+    },
+    {
+      "epoch": 582.0,
+      "grad_norm": 7.4327073097229,
+      "learning_rate": 2.6761970458695107e-05,
+      "loss": 0.4607,
+      "step": 1164
+    },
+    {
+      "epoch": 582.5,
+      "grad_norm": 3.269991397857666,
+      "learning_rate": 2.6373343271074658e-05,
+      "loss": 0.3263,
+      "step": 1165
+    },
+    {
+      "epoch": 583.0,
+      "grad_norm": 3.781911849975586,
+      "learning_rate": 2.598748206301682e-05,
+      "loss": 0.3588,
+      "step": 1166
+    },
+    {
+      "epoch": 583.5,
+      "grad_norm": 4.1138529777526855,
+      "learning_rate": 2.5604389087950987e-05,
+      "loss": 0.4963,
+      "step": 1167
+    },
+    {
+      "epoch": 584.0,
+      "grad_norm": 3.5733642578125,
+      "learning_rate": 2.522406658313997e-05,
+      "loss": 0.2926,
+      "step": 1168
+    },
+    {
+      "epoch": 584.5,
+      "grad_norm": 3.2220914363861084,
+      "learning_rate": 2.484651676966704e-05,
+      "loss": 0.2948,
+      "step": 1169
+    },
+    {
+      "epoch": 585.0,
+      "grad_norm": 3.9002745151519775,
+      "learning_rate": 2.4471741852423235e-05,
+      "loss": 0.3924,
+      "step": 1170
+    },
+    {
+      "epoch": 585.5,
+      "grad_norm": 3.894941806793213,
+      "learning_rate": 2.409974402009385e-05,
+      "loss": 0.3104,
+      "step": 1171
+    },
+    {
+      "epoch": 586.0,
+      "grad_norm": 3.5766544342041016,
+      "learning_rate": 2.3730525445146145e-05,
+      "loss": 0.3071,
+      "step": 1172
+    },
+    {
+      "epoch": 586.5,
+      "grad_norm": 3.3516080379486084,
+      "learning_rate": 2.3364088283816532e-05,
+      "loss": 0.3251,
+      "step": 1173
+    },
+    {
+      "epoch": 587.0,
+      "grad_norm": 4.621022701263428,
+      "learning_rate": 2.3000434676097803e-05,
+      "loss": 0.3922,
+      "step": 1174
+    },
+    {
+      "epoch": 587.5,
+      "grad_norm": 3.8262898921966553,
+      "learning_rate": 2.2639566745727203e-05,
+      "loss": 0.396,
+      "step": 1175
+    },
+    {
+      "epoch": 588.0,
+      "grad_norm": 4.147953033447266,
+      "learning_rate": 2.2281486600173206e-05,
+      "loss": 0.3159,
+      "step": 1176
+    },
+    {
+      "epoch": 588.5,
+      "grad_norm": 3.6508471965789795,
+      "learning_rate": 2.192619633062376e-05,
+      "loss": 0.4781,
+      "step": 1177
+    },
+    {
+      "epoch": 589.0,
+      "grad_norm": 4.170773029327393,
+      "learning_rate": 2.1573698011973954e-05,
+      "loss": 0.2756,
+      "step": 1178
+    },
+    {
+      "epoch": 589.5,
+      "grad_norm": 4.0242919921875,
+      "learning_rate": 2.1223993702814027e-05,
+      "loss": 0.3965,
+      "step": 1179
+    },
+    {
+      "epoch": 590.0,
+      "grad_norm": 3.4660654067993164,
+      "learning_rate": 2.087708544541689e-05,
+      "loss": 0.2492,
+      "step": 1180
+    },
+    {
+      "epoch": 590.5,
+      "grad_norm": 3.067244529724121,
+      "learning_rate": 2.0532975265726785e-05,
+      "loss": 0.316,
+      "step": 1181
+    },
+    {
+      "epoch": 591.0,
+      "grad_norm": 5.82780122756958,
+      "learning_rate": 2.019166517334703e-05,
+      "loss": 0.3609,
+      "step": 1182
+    },
+    {
+      "epoch": 591.5,
+      "grad_norm": 2.495479106903076,
+      "learning_rate": 1.985315716152847e-05,
+      "loss": 0.2863,
+      "step": 1183
+    },
+    {
+      "epoch": 592.0,
+      "grad_norm": 18.98761558532715,
+      "learning_rate": 1.9517453207157864e-05,
+      "loss": 0.7458,
+      "step": 1184
+    },
+    {
+      "epoch": 592.5,
+      "grad_norm": 3.3956587314605713,
+      "learning_rate": 1.9184555270746197e-05,
+      "loss": 0.3393,
+      "step": 1185
+    },
+    {
+      "epoch": 593.0,
+      "grad_norm": 4.012298107147217,
+      "learning_rate": 1.885446529641732e-05,
+      "loss": 0.3109,
+      "step": 1186
+    },
+    {
+      "epoch": 593.5,
+      "grad_norm": 3.0160374641418457,
+      "learning_rate": 1.852718521189667e-05,
+      "loss": 0.3104,
+      "step": 1187
+    },
+    {
+      "epoch": 594.0,
+      "grad_norm": 4.000430583953857,
+      "learning_rate": 1.820271692849984e-05,
+      "loss": 0.3141,
+      "step": 1188
+    },
+    {
+      "epoch": 594.5,
+      "grad_norm": 3.232729196548462,
+      "learning_rate": 1.788106234112147e-05,
+      "loss": 0.2672,
+      "step": 1189
+    },
+    {
+      "epoch": 595.0,
+      "grad_norm": 4.076137065887451,
+      "learning_rate": 1.7562223328224324e-05,
+      "loss": 0.3995,
+      "step": 1190
+    },
+    {
+      "epoch": 595.5,
+      "grad_norm": 3.353762626647949,
+      "learning_rate": 1.7246201751828118e-05,
+      "loss": 0.358,
+      "step": 1191
+    },
+    {
+      "epoch": 596.0,
+      "grad_norm": 4.43498420715332,
+      "learning_rate": 1.693299945749882e-05,
+      "loss": 0.3444,
+      "step": 1192
+    },
+    {
+      "epoch": 596.5,
+      "grad_norm": 3.093390703201294,
+      "learning_rate": 1.6622618274337754e-05,
+      "loss": 0.3209,
+      "step": 1193
+    },
+    {
+      "epoch": 597.0,
+      "grad_norm": 3.8518967628479004,
+      "learning_rate": 1.6315060014971016e-05,
+      "loss": 0.3166,
+      "step": 1194
+    },
+    {
+      "epoch": 597.5,
+      "grad_norm": 2.752342700958252,
+      "learning_rate": 1.6010326475538628e-05,
+      "loss": 0.3471,
+      "step": 1195
+    },
+    {
+      "epoch": 598.0,
+      "grad_norm": 7.323949337005615,
+      "learning_rate": 1.5708419435684463e-05,
+      "loss": 0.3868,
+      "step": 1196
+    },
+    {
+      "epoch": 598.5,
+      "grad_norm": 3.212301015853882,
+      "learning_rate": 1.5409340658545546e-05,
+      "loss": 0.4111,
+      "step": 1197
+    },
+    {
+      "epoch": 599.0,
+      "grad_norm": 4.531469345092773,
+      "learning_rate": 1.5113091890741948e-05,
+      "loss": 0.2855,
+      "step": 1198
+    },
+    {
+      "epoch": 599.5,
+      "grad_norm": 3.206294059753418,
+      "learning_rate": 1.4819674862366295e-05,
+      "loss": 0.3358,
+      "step": 1199
+    },
+    {
+      "epoch": 600.0,
+      "grad_norm": 3.3697855472564697,
+      "learning_rate": 1.4529091286973995e-05,
+      "loss": 0.3771,
+      "step": 1200
+    },
+    {
+      "epoch": 600.5,
+      "grad_norm": 2.459050178527832,
+      "learning_rate": 1.4241342861573081e-05,
+      "loss": 0.3104,
+      "step": 1201
+    },
+    {
+      "epoch": 601.0,
+      "grad_norm": 4.774039268493652,
+      "learning_rate": 1.3956431266614278e-05,
+      "loss": 0.3315,
+      "step": 1202
+    },
+    {
+      "epoch": 601.5,
+      "grad_norm": 2.6110453605651855,
+      "learning_rate": 1.3674358165981226e-05,
+      "loss": 0.2895,
+      "step": 1203
+    },
+    {
+      "epoch": 602.0,
+      "grad_norm": 4.223430156707764,
+      "learning_rate": 1.3395125206980774e-05,
+      "loss": 0.3828,
+      "step": 1204
+    },
+    {
+      "epoch": 602.5,
+      "grad_norm": 3.436066150665283,
+      "learning_rate": 1.3118734020333256e-05,
+      "loss": 0.329,
+      "step": 1205
+    },
+    {
+      "epoch": 603.0,
+      "grad_norm": 6.114805698394775,
+      "learning_rate": 1.2845186220163286e-05,
+      "loss": 0.3458,
+      "step": 1206
+    },
+    {
+      "epoch": 603.5,
+      "grad_norm": 4.581478595733643,
+      "learning_rate": 1.2574483403989867e-05,
+      "loss": 0.3613,
+      "step": 1207
+    },
+    {
+      "epoch": 604.0,
+      "grad_norm": 3.1037673950195312,
+      "learning_rate": 1.2306627152717408e-05,
+      "loss": 0.3479,
+      "step": 1208
+    },
+    {
+      "epoch": 604.5,
+      "grad_norm": 4.485935211181641,
+      "learning_rate": 1.2041619030626282e-05,
+      "loss": 0.4547,
+      "step": 1209
+    },
+    {
+      "epoch": 605.0,
+      "grad_norm": 4.357248783111572,
+      "learning_rate": 1.1779460585363943e-05,
+      "loss": 0.3236,
+      "step": 1210
+    },
+    {
+      "epoch": 605.5,
+      "grad_norm": 3.081249952316284,
+      "learning_rate": 1.1520153347935658e-05,
+      "loss": 0.3684,
+      "step": 1211
+    },
+    {
+      "epoch": 606.0,
+      "grad_norm": 4.546634674072266,
+      "learning_rate": 1.1263698832695512e-05,
+      "loss": 0.3412,
+      "step": 1212
+    },
+    {
+      "epoch": 606.5,
+      "grad_norm": 3.5010457038879395,
+      "learning_rate": 1.1010098537337975e-05,
+      "loss": 0.329,
+      "step": 1213
+    },
+    {
+      "epoch": 607.0,
+      "grad_norm": 11.072281837463379,
+      "learning_rate": 1.0759353942888573e-05,
+      "loss": 0.5143,
+      "step": 1214
+    },
+    {
+      "epoch": 607.5,
+      "grad_norm": 3.3091318607330322,
+      "learning_rate": 1.0511466513695778e-05,
+      "loss": 0.3354,
+      "step": 1215
+    },
+    {
+      "epoch": 608.0,
+      "grad_norm": 5.469602108001709,
+      "learning_rate": 1.0266437697422026e-05,
+      "loss": 0.3451,
+      "step": 1216
+    },
+    {
+      "epoch": 608.5,
+      "grad_norm": 4.189046859741211,
+      "learning_rate": 1.0024268925035661e-05,
+      "loss": 0.3675,
+      "step": 1217
+    },
+    {
+      "epoch": 609.0,
+      "grad_norm": 3.173112630844116,
+      "learning_rate": 9.784961610802113e-06,
+      "loss": 0.3381,
+      "step": 1218
+    },
+    {
+      "epoch": 609.5,
+      "grad_norm": 4.656552314758301,
+      "learning_rate": 9.548517152276115e-06,
+      "loss": 0.4603,
+      "step": 1219
+    },
+    {
+      "epoch": 610.0,
+      "grad_norm": 3.8624956607818604,
+      "learning_rate": 9.314936930293284e-06,
+      "loss": 0.2615,
+      "step": 1220
+    },
+    {
+      "epoch": 610.5,
+      "grad_norm": 4.915114879608154,
+      "learning_rate": 9.084222308962054e-06,
+      "loss": 0.427,
+      "step": 1221
+    },
+    {
+      "epoch": 611.0,
+      "grad_norm": 5.323273658752441,
+      "learning_rate": 8.856374635655695e-06,
+      "loss": 0.3407,
+      "step": 1222
+    },
+    {
+      "epoch": 611.5,
+      "grad_norm": 3.060882091522217,
+      "learning_rate": 8.631395241004592e-06,
+      "loss": 0.3435,
+      "step": 1223
+    },
+    {
+      "epoch": 612.0,
+      "grad_norm": 8.344826698303223,
+      "learning_rate": 8.409285438888358e-06,
+      "loss": 0.4618,
+      "step": 1224
+    },
+    {
+      "epoch": 612.5,
+      "grad_norm": 3.8486576080322266,
+      "learning_rate": 8.190046526428241e-06,
+      "loss": 0.311,
+      "step": 1225
+    },
+    {
+      "epoch": 613.0,
+      "grad_norm": 4.950206756591797,
+      "learning_rate": 7.973679783979337e-06,
+      "loss": 0.3936,
+      "step": 1226
+    },
+    {
+      "epoch": 613.5,
+      "grad_norm": 3.3266513347625732,
+      "learning_rate": 7.760186475123442e-06,
+      "loss": 0.3228,
+      "step": 1227
+    },
+    {
+      "epoch": 614.0,
+      "grad_norm": 4.4216203689575195,
+      "learning_rate": 7.549567846661387e-06,
+      "loss": 0.3479,
+      "step": 1228
+    },
+    {
+      "epoch": 614.5,
+      "grad_norm": 5.311647415161133,
+      "learning_rate": 7.341825128605984e-06,
+      "loss": 0.3756,
+      "step": 1229
+    },
+    {
+      "epoch": 615.0,
+      "grad_norm": 3.6215133666992188,
+      "learning_rate": 7.136959534174592e-06,
+      "loss": 0.3585,
+      "step": 1230
+    },
+    {
+      "epoch": 615.5,
+      "grad_norm": 3.0667953491210938,
+      "learning_rate": 6.93497225978218e-06,
+      "loss": 0.337,
+      "step": 1231
+    },
+    {
+      "epoch": 616.0,
+      "grad_norm": 8.4304780960083,
+      "learning_rate": 6.735864485034493e-06,
+      "loss": 0.7068,
+      "step": 1232
+    },
+    {
+      "epoch": 616.5,
+      "grad_norm": 4.000027656555176,
+      "learning_rate": 6.539637372720897e-06,
+      "loss": 0.3664,
+      "step": 1233
+    },
+    {
+      "epoch": 617.0,
+      "grad_norm": 3.0169484615325928,
+      "learning_rate": 6.346292068807602e-06,
+      "loss": 0.3141,
+      "step": 1234
+    },
+    {
+      "epoch": 617.5,
+      "grad_norm": 3.082249402999878,
+      "learning_rate": 6.15582970243117e-06,
+      "loss": 0.3469,
+      "step": 1235
+    },
+    {
+      "epoch": 618.0,
+      "grad_norm": 4.726069927215576,
+      "learning_rate": 5.968251385891743e-06,
+      "loss": 0.3623,
+      "step": 1236
+    },
+    {
+      "epoch": 618.5,
+      "grad_norm": 3.9897854328155518,
+      "learning_rate": 5.783558214646712e-06,
+      "loss": 0.4163,
+      "step": 1237
+    },
+    {
+      "epoch": 619.0,
+      "grad_norm": 3.5751760005950928,
+      "learning_rate": 5.601751267304056e-06,
+      "loss": 0.2423,
+      "step": 1238
+    },
+    {
+      "epoch": 619.5,
+      "grad_norm": 3.0000417232513428,
+      "learning_rate": 5.422831605616352e-06,
+      "loss": 0.3621,
+      "step": 1239
+    },
+    {
+      "epoch": 620.0,
+      "grad_norm": 3.9752204418182373,
+      "learning_rate": 5.2468002744744395e-06,
+      "loss": 0.3447,
+      "step": 1240
+    },
+    {
+      "epoch": 620.5,
+      "grad_norm": 2.8096187114715576,
+      "learning_rate": 5.073658301901207e-06,
+      "loss": 0.2919,
+      "step": 1241
+    },
+    {
+      "epoch": 621.0,
+      "grad_norm": 4.048423767089844,
+      "learning_rate": 4.9034066990457095e-06,
+      "loss": 0.3403,
+      "step": 1242
+    },
+    {
+      "epoch": 621.5,
+      "grad_norm": 4.086081504821777,
+      "learning_rate": 4.7360464601773344e-06,
+      "loss": 0.4186,
+      "step": 1243
+    },
+    {
+      "epoch": 622.0,
+      "grad_norm": 3.309329032897949,
+      "learning_rate": 4.571578562679757e-06,
+      "loss": 0.2563,
+      "step": 1244
+    },
+    {
+      "epoch": 622.5,
+      "grad_norm": 2.5095250606536865,
+      "learning_rate": 4.41000396704544e-06,
+      "loss": 0.3657,
+      "step": 1245
+    },
+    {
+      "epoch": 623.0,
+      "grad_norm": 8.402181625366211,
+      "learning_rate": 4.2513236168700845e-06,
+      "loss": 0.3659,
+      "step": 1246
+    },
+    {
+      "epoch": 623.5,
+      "grad_norm": 2.3093628883361816,
+      "learning_rate": 4.095538438846802e-06,
+      "loss": 0.2832,
+      "step": 1247
+    },
+    {
+      "epoch": 624.0,
+      "grad_norm": 2.666494607925415,
+      "learning_rate": 3.942649342761117e-06,
+      "loss": 0.3632,
+      "step": 1248
+    },
+    {
+      "epoch": 624.5,
+      "grad_norm": 3.1523962020874023,
+      "learning_rate": 3.7926572214851405e-06,
+      "loss": 0.3168,
+      "step": 1249
+    },
+    {
+      "epoch": 625.0,
+      "grad_norm": 2.8384346961975098,
+      "learning_rate": 3.645562950973014e-06,
+      "loss": 0.3778,
+      "step": 1250
+    },
+    {
+      "epoch": 625.5,
+      "grad_norm": 2.507253885269165,
+      "learning_rate": 3.501367390255139e-06,
+      "loss": 0.3493,
+      "step": 1251
+    },
+    {
+      "epoch": 626.0,
+      "grad_norm": 13.728121757507324,
+      "learning_rate": 3.3600713814335158e-06,
+      "loss": 0.4467,
+      "step": 1252
+    },
+    {
+      "epoch": 626.5,
+      "grad_norm": 3.521282196044922,
+      "learning_rate": 3.221675749676911e-06,
+      "loss": 0.2923,
+      "step": 1253
+    },
+    {
+      "epoch": 627.0,
+      "grad_norm": 7.8102874755859375,
+      "learning_rate": 3.0861813032156404e-06,
+      "loss": 0.5423,
+      "step": 1254
+    },
+    {
+      "epoch": 627.5,
+      "grad_norm": 4.267757415771484,
+      "learning_rate": 2.9535888333374063e-06,
+      "loss": 0.3951,
+      "step": 1255
+    },
+    {
+      "epoch": 628.0,
+      "grad_norm": 2.7788209915161133,
+      "learning_rate": 2.823899114382078e-06,
+      "loss": 0.2925,
+      "step": 1256
+    },
+    {
+      "epoch": 628.5,
+      "grad_norm": 5.810858249664307,
+      "learning_rate": 2.697112903737642e-06,
+      "loss": 0.458,
+      "step": 1257
+    },
+    {
+      "epoch": 629.0,
+      "grad_norm": 3.2245874404907227,
+      "learning_rate": 2.573230941835536e-06,
+      "loss": 0.3175,
+      "step": 1258
+    },
+    {
+      "epoch": 629.5,
+      "grad_norm": 4.0286478996276855,
+      "learning_rate": 2.4522539521463216e-06,
+      "loss": 0.4371,
+      "step": 1259
+    },
+    {
+      "epoch": 630.0,
+      "grad_norm": 2.790431499481201,
+      "learning_rate": 2.334182641175686e-06,
+      "loss": 0.2823,
+      "step": 1260
+    },
+    {
+      "epoch": 630.5,
+      "grad_norm": 3.060202121734619,
+      "learning_rate": 2.219017698460002e-06,
+      "loss": 0.3139,
+      "step": 1261
+    },
+    {
+      "epoch": 631.0,
+      "grad_norm": 5.02834939956665,
+      "learning_rate": 2.106759796562496e-06,
+      "loss": 0.3908,
+      "step": 1262
+    },
+    {
+      "epoch": 631.5,
+      "grad_norm": 3.284142017364502,
+      "learning_rate": 1.99740959106931e-06,
+      "loss": 0.3755,
+      "step": 1263
+    },
+    {
+      "epoch": 632.0,
+      "grad_norm": 7.2658514976501465,
+      "learning_rate": 1.890967720585668e-06,
+      "loss": 0.3729,
+      "step": 1264
+    },
+    {
+      "epoch": 632.5,
+      "grad_norm": 2.9172821044921875,
+      "learning_rate": 1.7874348067319911e-06,
+      "loss": 0.3316,
+      "step": 1265
+    },
+    {
+      "epoch": 633.0,
+      "grad_norm": 2.3214914798736572,
+      "learning_rate": 1.6868114541404577e-06,
+      "loss": 0.3191,
+      "step": 1266
+    },
+    {
+      "epoch": 633.5,
+      "grad_norm": 2.2073729038238525,
+      "learning_rate": 1.5890982504515039e-06,
+      "loss": 0.3157,
+      "step": 1267
+    },
+    {
+      "epoch": 634.0,
+      "grad_norm": 5.47219181060791,
+      "learning_rate": 1.494295766310161e-06,
+      "loss": 0.3484,
+      "step": 1268
+    },
+    {
+      "epoch": 634.5,
+      "grad_norm": 3.0883967876434326,
+      "learning_rate": 1.402404555362946e-06,
+      "loss": 0.3987,
+      "step": 1269
+    },
+    {
+      "epoch": 635.0,
+      "grad_norm": 5.0094122886657715,
+      "learning_rate": 1.3134251542544773e-06,
+      "loss": 0.3422,
+      "step": 1270
+    },
+    {
+      "epoch": 635.5,
+      "grad_norm": 3.4778897762298584,
+      "learning_rate": 1.227358082624419e-06,
+      "loss": 0.3724,
+      "step": 1271
+    },
+    {
+      "epoch": 636.0,
+      "grad_norm": 3.8139805793762207,
+      "learning_rate": 1.1442038431044854e-06,
+      "loss": 0.2991,
+      "step": 1272
+    },
+    {
+      "epoch": 636.5,
+      "grad_norm": 2.7982492446899414,
+      "learning_rate": 1.0639629213154422e-06,
+      "loss": 0.34,
+      "step": 1273
+    },
+    {
+      "epoch": 637.0,
+      "grad_norm": 10.94028091430664,
+      "learning_rate": 9.866357858642206e-07,
+      "loss": 0.3995,
+      "step": 1274
+    },
+    {
+      "epoch": 637.5,
+      "grad_norm": 2.3486061096191406,
+      "learning_rate": 9.12222888341252e-07,
+      "loss": 0.2749,
+      "step": 1275
+    },
+    {
+      "epoch": 638.0,
+      "grad_norm": 3.3858132362365723,
+      "learning_rate": 8.4072466331786e-07,
+      "loss": 0.3499,
+      "step": 1276
+    },
+    {
+      "epoch": 638.5,
+      "grad_norm": 3.1323537826538086,
+      "learning_rate": 7.721415283437061e-07,
+      "loss": 0.3684,
+      "step": 1277
+    },
+    {
+      "epoch": 639.0,
+      "grad_norm": 3.36407732963562,
+      "learning_rate": 7.064738839442364e-07,
+      "loss": 0.2521,
+      "step": 1278
+    },
+    {
+      "epoch": 639.5,
+      "grad_norm": 2.5673577785491943,
+      "learning_rate": 6.437221136184613e-07,
+      "loss": 0.3119,
+      "step": 1279
+    },
+    {
+      "epoch": 640.0,
+      "grad_norm": 12.754891395568848,
+      "learning_rate": 5.838865838366791e-07,
+      "loss": 0.5441,
+      "step": 1280
+    },
+    {
+      "epoch": 640.5,
+      "grad_norm": 2.3102803230285645,
+      "learning_rate": 5.269676440384785e-07,
+      "loss": 0.2964,
+      "step": 1281
+    },
+    {
+      "epoch": 641.0,
+      "grad_norm": 3.5695204734802246,
+      "learning_rate": 4.729656266304061e-07,
+      "loss": 0.3217,
+      "step": 1282
+    },
+    {
+      "epoch": 641.5,
+      "grad_norm": 3.4399874210357666,
+      "learning_rate": 4.218808469841906e-07,
+      "loss": 0.3274,
+      "step": 1283
+    },
+    {
+      "epoch": 642.0,
+      "grad_norm": 8.23744010925293,
+      "learning_rate": 3.737136034349109e-07,
+      "loss": 0.4427,
+      "step": 1284
+    },
+    {
+      "epoch": 642.5,
+      "grad_norm": 2.2069194316864014,
+      "learning_rate": 3.2846417727938617e-07,
+      "loss": 0.3227,
+      "step": 1285
+    },
+    {
+      "epoch": 643.0,
+      "grad_norm": 8.936519622802734,
+      "learning_rate": 2.861328327741219e-07,
+      "loss": 0.4613,
+      "step": 1286
+    },
+    {
+      "epoch": 643.5,
+      "grad_norm": 2.81213116645813,
+      "learning_rate": 2.467198171342e-07,
+      "loss": 0.3443,
+      "step": 1287
+    },
+    {
+      "epoch": 644.0,
+      "grad_norm": 2.815112590789795,
+      "learning_rate": 2.1022536053166842e-07,
+      "loss": 0.3199,
+      "step": 1288
+    },
+    {
+      "epoch": 644.5,
+      "grad_norm": 2.6604883670806885,
+      "learning_rate": 1.7664967609409832e-07,
+      "loss": 0.3314,
+      "step": 1289
+    },
+    {
+      "epoch": 645.0,
+      "grad_norm": 6.471987247467041,
+      "learning_rate": 1.4599295990352924e-07,
+      "loss": 0.3571,
+      "step": 1290
+    },
+    {
+      "epoch": 645.5,
+      "grad_norm": 3.8412744998931885,
+      "learning_rate": 1.182553909950812e-07,
+      "loss": 0.4457,
+      "step": 1291
+    },
+    {
+      "epoch": 646.0,
+      "grad_norm": 2.6368608474731445,
+      "learning_rate": 9.343713135623322e-08,
+      "loss": 0.2413,
+      "step": 1292
+    },
+    {
+      "epoch": 646.5,
+      "grad_norm": 2.7168586254119873,
+      "learning_rate": 7.153832592560194e-08,
+      "loss": 0.3067,
+      "step": 1293
+    },
+    {
+      "epoch": 647.0,
+      "grad_norm": 5.021940231323242,
+      "learning_rate": 5.2559102592164565e-08,
+      "loss": 0.3971,
+      "step": 1294
+    },
+    {
+      "epoch": 647.5,
+      "grad_norm": 3.8227570056915283,
+      "learning_rate": 3.649957219464817e-08,
+      "loss": 0.4192,
+      "step": 1295
+    },
+    {
+      "epoch": 648.0,
+      "grad_norm": 5.346871376037598,
+      "learning_rate": 2.3359828520641556e-08,
+      "loss": 0.3641,
+      "step": 1296
+    },
+    {
+      "epoch": 648.5,
+      "grad_norm": 2.344628095626831,
+      "learning_rate": 1.3139948306262195e-08,
+      "loss": 0.2752,
+      "step": 1297
+    },
+    {
+      "epoch": 649.0,
+      "grad_norm": 4.132582664489746,
+      "learning_rate": 5.839991235656594e-09,
+      "loss": 0.3926,
+      "step": 1298
+    },
+    {
+      "epoch": 649.5,
+      "grad_norm": 3.1090128421783447,
+      "learning_rate": 1.4599999405007225e-09,
+      "loss": 0.32,
+      "step": 1299
+    },
+    {
+      "epoch": 650.0,
+      "grad_norm": 4.562283515930176,
+      "learning_rate": 0.0,
+      "loss": 0.4513,
+      "step": 1300
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 1300,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 650,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.340747841798144e+18,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}