diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,108652 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 15517,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 6.444544692917445e-05,
+      "grad_norm": 0.15129432154940006,
+      "learning_rate": 0.0,
+      "loss": 0.0012,
+      "step": 1
+    },
+    {
+      "epoch": 0.0001288908938583489,
+      "grad_norm": 0.26991460323889616,
+      "learning_rate": 9.43403983493194e-07,
+      "loss": 0.0003,
+      "step": 2
+    },
+    {
+      "epoch": 0.00019333634078752337,
+      "grad_norm": 1.9954004446079063,
+      "learning_rate": 1.4952599368676733e-06,
+      "loss": 0.0091,
+      "step": 3
+    },
+    {
+      "epoch": 0.0002577817877166978,
+      "grad_norm": 0.22931631126106178,
+      "learning_rate": 1.886807966986388e-06,
+      "loss": 0.0003,
+      "step": 4
+    },
+    {
+      "epoch": 0.00032222723464587225,
+      "grad_norm": 0.25371039245875826,
+      "learning_rate": 2.1905162141015006e-06,
+      "loss": 0.0005,
+      "step": 5
+    },
+    {
+      "epoch": 0.00038667268157504674,
+      "grad_norm": 1.1948746521302072,
+      "learning_rate": 2.4386639203608674e-06,
+      "loss": 0.006,
+      "step": 6
+    },
+    {
+      "epoch": 0.0004511181285042212,
+      "grad_norm": 0.30320543126617766,
+      "learning_rate": 2.648469816548369e-06,
+      "loss": 0.0027,
+      "step": 7
+    },
+    {
+      "epoch": 0.0005155635754333956,
+      "grad_norm": 0.10152956997270147,
+      "learning_rate": 2.8302119504795815e-06,
+      "loss": 0.0009,
+      "step": 8
+    },
+    {
+      "epoch": 0.0005800090223625701,
+      "grad_norm": 1.4987747433411953,
+      "learning_rate": 2.9905198737353466e-06,
+      "loss": 0.0129,
+      "step": 9
+    },
+    {
+      "epoch": 0.0006444544692917445,
+      "grad_norm": 0.02707402421982954,
+      "learning_rate": 3.1339201975946947e-06,
+      "loss": 0.0001,
+      "step": 10
+    },
+    {
+      "epoch": 0.000708899916220919,
+      "grad_norm": 0.018766550431254077,
+      "learning_rate": 3.2636415696447343e-06,
+      "loss": 0.0001,
+      "step": 11
+    },
+    {
+      "epoch": 0.0007733453631500935,
+      "grad_norm": 0.20907100771370063,
+      "learning_rate": 3.382067903854061e-06,
+      "loss": 0.0004,
+      "step": 12
+    },
+    {
+      "epoch": 0.0008377908100792679,
+      "grad_norm": 0.5817576027518055,
+      "learning_rate": 3.4910095707707385e-06,
+      "loss": 0.0026,
+      "step": 13
+    },
+    {
+      "epoch": 0.0009022362570084424,
+      "grad_norm": 0.01711225147567763,
+      "learning_rate": 3.5918738000415628e-06,
+      "loss": 0.0,
+      "step": 14
+    },
+    {
+      "epoch": 0.0009666817039376168,
+      "grad_norm": 0.07462783118349445,
+      "learning_rate": 3.6857761509691734e-06,
+      "loss": 0.0003,
+      "step": 15
+    },
+    {
+      "epoch": 0.0010311271508667912,
+      "grad_norm": 0.10816803494037058,
+      "learning_rate": 3.773615933972776e-06,
+      "loss": 0.0002,
+      "step": 16
+    },
+    {
+      "epoch": 0.0010955725977959657,
+      "grad_norm": 1.9217337579697105,
+      "learning_rate": 3.8561287268159795e-06,
+      "loss": 0.0167,
+      "step": 17
+    },
+    {
+      "epoch": 0.0011600180447251401,
+      "grad_norm": 0.1417044693166633,
+      "learning_rate": 3.93392385722854e-06,
+      "loss": 0.0002,
+      "step": 18
+    },
+    {
+      "epoch": 0.0012244634916543146,
+      "grad_norm": 0.023784749719864124,
+      "learning_rate": 4.0075117377730165e-06,
+      "loss": 0.0,
+      "step": 19
+    },
+    {
+      "epoch": 0.001288908938583489,
+      "grad_norm": 0.07828858355950565,
+      "learning_rate": 4.077324181087888e-06,
+      "loss": 0.0002,
+      "step": 20
+    },
+    {
+      "epoch": 0.0013533543855126634,
+      "grad_norm": 0.04722417233368723,
+      "learning_rate": 4.143729753416042e-06,
+      "loss": 0.0002,
+      "step": 21
+    },
+    {
+      "epoch": 0.001417799832441838,
+      "grad_norm": 0.08640805563755993,
+      "learning_rate": 4.207045553137928e-06,
+      "loss": 0.0002,
+      "step": 22
+    },
+    {
+      "epoch": 0.0014822452793710125,
+      "grad_norm": 1.7851848307852938,
+      "learning_rate": 4.267546368922451e-06,
+      "loss": 0.012,
+      "step": 23
+    },
+    {
+      "epoch": 0.001546690726300187,
+      "grad_norm": 0.18091976525222841,
+      "learning_rate": 4.325471887347255e-06,
+      "loss": 0.0004,
+      "step": 24
+    },
+    {
+      "epoch": 0.0016111361732293614,
+      "grad_norm": 0.03394466558933047,
+      "learning_rate": 4.381032428203001e-06,
+      "loss": 0.0004,
+      "step": 25
+    },
+    {
+      "epoch": 0.0016755816201585358,
+      "grad_norm": 0.041340723123166515,
+      "learning_rate": 4.434413554263933e-06,
+      "loss": 0.0002,
+      "step": 26
+    },
+    {
+      "epoch": 0.0017400270670877103,
+      "grad_norm": 0.20461789214303566,
+      "learning_rate": 4.485779810603019e-06,
+      "loss": 0.0022,
+      "step": 27
+    },
+    {
+      "epoch": 0.0018044725140168847,
+      "grad_norm": 0.12106531716915613,
+      "learning_rate": 4.5352777835347565e-06,
+      "loss": 0.0002,
+      "step": 28
+    },
+    {
+      "epoch": 0.0018689179609460592,
+      "grad_norm": 0.03492363492746817,
+      "learning_rate": 4.583038622537583e-06,
+      "loss": 0.0001,
+      "step": 29
+    },
+    {
+      "epoch": 0.0019333634078752336,
+      "grad_norm": 0.03259851462272291,
+      "learning_rate": 4.629180134462368e-06,
+      "loss": 0.0001,
+      "step": 30
+    },
+    {
+      "epoch": 0.001997808854804408,
+      "grad_norm": 0.03013889637859078,
+      "learning_rate": 4.673808534226263e-06,
+      "loss": 0.0002,
+      "step": 31
+    },
+    {
+      "epoch": 0.0020622543017335825,
+      "grad_norm": 0.027037052171979114,
+      "learning_rate": 4.717019917465971e-06,
+      "loss": 0.0003,
+      "step": 32
+    },
+    {
+      "epoch": 0.002126699748662757,
+      "grad_norm": 0.005106793730490385,
+      "learning_rate": 4.758901506512407e-06,
+      "loss": 0.0,
+      "step": 33
+    },
+    {
+      "epoch": 0.0021911451955919314,
+      "grad_norm": 0.3133135457884165,
+      "learning_rate": 4.799532710309174e-06,
+      "loss": 0.0008,
+      "step": 34
+    },
+    {
+      "epoch": 0.002255590642521106,
+      "grad_norm": 0.06576723596208979,
+      "learning_rate": 4.838986030649869e-06,
+      "loss": 0.0004,
+      "step": 35
+    },
+    {
+      "epoch": 0.0023200360894502802,
+      "grad_norm": 0.44200181870496286,
+      "learning_rate": 4.877327840721735e-06,
+      "loss": 0.0026,
+      "step": 36
+    },
+    {
+      "epoch": 0.0023844815363794547,
+      "grad_norm": 0.005113691502045184,
+      "learning_rate": 4.914619056956378e-06,
+      "loss": 0.0001,
+      "step": 37
+    },
+    {
+      "epoch": 0.002448926983308629,
+      "grad_norm": 0.011710030818769188,
+      "learning_rate": 4.9509157212662115e-06,
+      "loss": 0.0001,
+      "step": 38
+    },
+    {
+      "epoch": 0.0025133724302378035,
+      "grad_norm": 0.10730184004136714,
+      "learning_rate": 4.986269507638412e-06,
+      "loss": 0.0013,
+      "step": 39
+    },
+    {
+      "epoch": 0.002577817877166978,
+      "grad_norm": 0.06514135570120874,
+      "learning_rate": 5.0207281645810825e-06,
+      "loss": 0.0002,
+      "step": 40
+    },
+    {
+      "epoch": 0.0026422633240961524,
+      "grad_norm": 0.00949610941862856,
+      "learning_rate": 5.054335902928648e-06,
+      "loss": 0.0001,
+      "step": 41
+    },
+    {
+      "epoch": 0.002706708771025327,
+      "grad_norm": 0.011550266470039141,
+      "learning_rate": 5.087133736909237e-06,
+      "loss": 0.0001,
+      "step": 42
+    },
+    {
+      "epoch": 0.0027711542179545013,
+      "grad_norm": 0.007752723479024223,
+      "learning_rate": 5.119159785074679e-06,
+      "loss": 0.0,
+      "step": 43
+    },
+    {
+      "epoch": 0.002835599664883676,
+      "grad_norm": 0.5378045317232993,
+      "learning_rate": 5.1504495366311225e-06,
+      "loss": 0.0032,
+      "step": 44
+    },
+    {
+      "epoch": 0.0029000451118128506,
+      "grad_norm": 0.2128986742090359,
+      "learning_rate": 5.181036087836847e-06,
+      "loss": 0.0006,
+      "step": 45
+    },
+    {
+      "epoch": 0.002964490558742025,
+      "grad_norm": 0.000640446091847838,
+      "learning_rate": 5.210950352415644e-06,
+      "loss": 0.0,
+      "step": 46
+    },
+    {
+      "epoch": 0.0030289360056711995,
+      "grad_norm": 0.030881680917973543,
+      "learning_rate": 5.2402212493395684e-06,
+      "loss": 0.0001,
+      "step": 47
+    },
+    {
+      "epoch": 0.003093381452600374,
+      "grad_norm": 0.05658321101545766,
+      "learning_rate": 5.268875870840449e-06,
+      "loss": 0.0001,
+      "step": 48
+    },
+    {
+      "epoch": 0.0031578268995295484,
+      "grad_norm": 0.5703326443811637,
+      "learning_rate": 5.296939633096738e-06,
+      "loss": 0.0052,
+      "step": 49
+    },
+    {
+      "epoch": 0.003222272346458723,
+      "grad_norm": 2.8203730272893908,
+      "learning_rate": 5.324436411696195e-06,
+      "loss": 0.0161,
+      "step": 50
+    },
+    {
+      "epoch": 0.0032867177933878972,
+      "grad_norm": 0.001724963280434341,
+      "learning_rate": 5.351388663683652e-06,
+      "loss": 0.0,
+      "step": 51
+    },
+    {
+      "epoch": 0.0033511632403170717,
+      "grad_norm": 0.01416510359891742,
+      "learning_rate": 5.377817537757127e-06,
+      "loss": 0.0001,
+      "step": 52
+    },
+    {
+      "epoch": 0.003415608687246246,
+      "grad_norm": 0.0049279435077748575,
+      "learning_rate": 5.403742973967069e-06,
+      "loss": 0.0,
+      "step": 53
+    },
+    {
+      "epoch": 0.0034800541341754206,
+      "grad_norm": 0.7568150671265557,
+      "learning_rate": 5.4291837940962135e-06,
+      "loss": 0.0023,
+      "step": 54
+    },
+    {
+      "epoch": 0.003544499581104595,
+      "grad_norm": 0.18716012429739934,
+      "learning_rate": 5.454157783746235e-06,
+      "loss": 0.0006,
+      "step": 55
+    },
+    {
+      "epoch": 0.0036089450280337694,
+      "grad_norm": 0.03473747532236853,
+      "learning_rate": 5.4786817670279514e-06,
+      "loss": 0.0001,
+      "step": 56
+    },
+    {
+      "epoch": 0.003673390474962944,
+      "grad_norm": 0.08810403927552686,
+      "learning_rate": 5.50277167464069e-06,
+      "loss": 0.0003,
+      "step": 57
+    },
+    {
+      "epoch": 0.0037378359218921183,
+      "grad_norm": 0.017372241080586315,
+      "learning_rate": 5.526442606030776e-06,
+      "loss": 0.0001,
+      "step": 58
+    },
+    {
+      "epoch": 0.0038022813688212928,
+      "grad_norm": 0.21547829813614516,
+      "learning_rate": 5.549708886236511e-06,
+      "loss": 0.0017,
+      "step": 59
+    },
+    {
+      "epoch": 0.003866726815750467,
+      "grad_norm": 0.5537295329959373,
+      "learning_rate": 5.572584117955562e-06,
+      "loss": 0.002,
+      "step": 60
+    },
+    {
+      "epoch": 0.003931172262679642,
+      "grad_norm": 0.14335554933770492,
+      "learning_rate": 5.5950812293086465e-06,
+      "loss": 0.0004,
+      "step": 61
+    },
+    {
+      "epoch": 0.003995617709608816,
+      "grad_norm": 0.011859054511193099,
+      "learning_rate": 5.617212517719457e-06,
+      "loss": 0.0001,
+      "step": 62
+    },
+    {
+      "epoch": 0.004060063156537991,
+      "grad_norm": 0.004499399040304683,
+      "learning_rate": 5.638989690283715e-06,
+      "loss": 0.0,
+      "step": 63
+    },
+    {
+      "epoch": 0.004124508603467165,
+      "grad_norm": 0.9716421455530285,
+      "learning_rate": 5.660423900959163e-06,
+      "loss": 0.0033,
+      "step": 64
+    },
+    {
+      "epoch": 0.00418895405039634,
+      "grad_norm": 0.2602800996635666,
+      "learning_rate": 5.681525784872238e-06,
+      "loss": 0.004,
+      "step": 65
+    },
+    {
+      "epoch": 0.004253399497325514,
+      "grad_norm": 0.505029623796257,
+      "learning_rate": 5.7023054900056e-06,
+      "loss": 0.0048,
+      "step": 66
+    },
+    {
+      "epoch": 0.004317844944254689,
+      "grad_norm": 0.06494347848669335,
+      "learning_rate": 5.722772706502867e-06,
+      "loss": 0.0002,
+      "step": 67
+    },
+    {
+      "epoch": 0.004382290391183863,
+      "grad_norm": 0.05793100298650254,
+      "learning_rate": 5.742936693802368e-06,
+      "loss": 0.0004,
+      "step": 68
+    },
+    {
+      "epoch": 0.004446735838113038,
+      "grad_norm": 0.3872734676672814,
+      "learning_rate": 5.762806305790124e-06,
+      "loss": 0.0031,
+      "step": 69
+    },
+    {
+      "epoch": 0.004511181285042212,
+      "grad_norm": 0.032758724262618666,
+      "learning_rate": 5.782390014143064e-06,
+      "loss": 0.0001,
+      "step": 70
+    },
+    {
+      "epoch": 0.0045756267319713865,
+      "grad_norm": 0.510062996956511,
+      "learning_rate": 5.801695930016513e-06,
+      "loss": 0.0053,
+      "step": 71
+    },
+    {
+      "epoch": 0.0046400721789005605,
+      "grad_norm": 0.0020728367399568642,
+      "learning_rate": 5.820731824214928e-06,
+      "loss": 0.0,
+      "step": 72
+    },
+    {
+      "epoch": 0.004704517625829735,
+      "grad_norm": 0.022118583202706633,
+      "learning_rate": 5.8395051459714105e-06,
+      "loss": 0.0001,
+      "step": 73
+    },
+    {
+      "epoch": 0.004768963072758909,
+      "grad_norm": 0.023627041277547944,
+      "learning_rate": 5.858023040449572e-06,
+      "loss": 0.0,
+      "step": 74
+    },
+    {
+      "epoch": 0.004833408519688084,
+      "grad_norm": 0.046421888470687886,
+      "learning_rate": 5.876292365070674e-06,
+      "loss": 0.0004,
+      "step": 75
+    },
+    {
+      "epoch": 0.004897853966617258,
+      "grad_norm": 0.02261795227128185,
+      "learning_rate": 5.894319704759405e-06,
+      "loss": 0.0,
+      "step": 76
+    },
+    {
+      "epoch": 0.004962299413546433,
+      "grad_norm": 0.035546317724952904,
+      "learning_rate": 5.912111386193103e-06,
+      "loss": 0.0001,
+      "step": 77
+    },
+    {
+      "epoch": 0.005026744860475607,
+      "grad_norm": 0.01905267113714707,
+      "learning_rate": 5.929673491131606e-06,
+      "loss": 0.0,
+      "step": 78
+    },
+    {
+      "epoch": 0.005091190307404782,
+      "grad_norm": 0.015782799957426378,
+      "learning_rate": 5.9470118688979865e-06,
+      "loss": 0.0001,
+      "step": 79
+    },
+    {
+      "epoch": 0.005155635754333956,
+      "grad_norm": 0.10050158065115125,
+      "learning_rate": 5.964132148074277e-06,
+      "loss": 0.0003,
+      "step": 80
+    },
+    {
+      "epoch": 0.005220081201263131,
+      "grad_norm": 0.02889224712849086,
+      "learning_rate": 5.981039747470693e-06,
+      "loss": 0.0002,
+      "step": 81
+    },
+    {
+      "epoch": 0.005284526648192305,
+      "grad_norm": 0.04525759150992565,
+      "learning_rate": 5.997739886421841e-06,
+      "loss": 0.0003,
+      "step": 82
+    },
+    {
+      "epoch": 0.00534897209512148,
+      "grad_norm": 0.10955070109304078,
+      "learning_rate": 6.014237594458876e-06,
+      "loss": 0.0002,
+      "step": 83
+    },
+    {
+      "epoch": 0.005413417542050654,
+      "grad_norm": 0.005650621324755179,
+      "learning_rate": 6.030537720402429e-06,
+      "loss": 0.0,
+      "step": 84
+    },
+    {
+      "epoch": 0.005477862988979829,
+      "grad_norm": 0.22230525543403104,
+      "learning_rate": 6.04664494091748e-06,
+      "loss": 0.001,
+      "step": 85
+    },
+    {
+      "epoch": 0.005542308435909003,
+      "grad_norm": 0.1757163404784244,
+      "learning_rate": 6.062563768567872e-06,
+      "loss": 0.0021,
+      "step": 86
+    },
+    {
+      "epoch": 0.0056067538828381775,
+      "grad_norm": 0.10250988499729953,
+      "learning_rate": 6.0782985594052556e-06,
+      "loss": 0.0003,
+      "step": 87
+    },
+    {
+      "epoch": 0.005671199329767352,
+      "grad_norm": 0.07404739274091283,
+      "learning_rate": 6.093853520124317e-06,
+      "loss": 0.0006,
+      "step": 88
+    },
+    {
+      "epoch": 0.005735644776696526,
+      "grad_norm": 0.01653300445057603,
+      "learning_rate": 6.109232714813748e-06,
+      "loss": 0.0,
+      "step": 89
+    },
+    {
+      "epoch": 0.005800090223625701,
+      "grad_norm": 0.009526052586484979,
+      "learning_rate": 6.124440071330041e-06,
+      "loss": 0.0001,
+      "step": 90
+    },
+    {
+      "epoch": 0.005864535670554875,
+      "grad_norm": 0.010664951094544989,
+      "learning_rate": 6.139479387319107e-06,
+      "loss": 0.0,
+      "step": 91
+    },
+    {
+      "epoch": 0.00592898111748405,
+      "grad_norm": 0.020226888276105217,
+      "learning_rate": 6.154354335908838e-06,
+      "loss": 0.0001,
+      "step": 92
+    },
+    {
+      "epoch": 0.005993426564413224,
+      "grad_norm": 0.4540252129143341,
+      "learning_rate": 6.169068471093936e-06,
+      "loss": 0.0036,
+      "step": 93
+    },
+    {
+      "epoch": 0.006057872011342399,
+      "grad_norm": 0.0843852312896068,
+      "learning_rate": 6.183625232832763e-06,
+      "loss": 0.0008,
+      "step": 94
+    },
+    {
+      "epoch": 0.006122317458271573,
+      "grad_norm": 0.7314261216216805,
+      "learning_rate": 6.1980279518745175e-06,
+      "loss": 0.0058,
+      "step": 95
+    },
+    {
+      "epoch": 0.006186762905200748,
+      "grad_norm": 0.0012243406202469007,
+      "learning_rate": 6.212279854333643e-06,
+      "loss": 0.0,
+      "step": 96
+    },
+    {
+      "epoch": 0.006251208352129922,
+      "grad_norm": 0.028267818956160366,
+      "learning_rate": 6.226384066027224e-06,
+      "loss": 0.0001,
+      "step": 97
+    },
+    {
+      "epoch": 0.006315653799059097,
+      "grad_norm": 0.018740492262862375,
+      "learning_rate": 6.240343616589932e-06,
+      "loss": 0.0001,
+      "step": 98
+    },
+    {
+      "epoch": 0.006380099245988271,
+      "grad_norm": 0.051675413777484226,
+      "learning_rate": 6.25416144338008e-06,
+      "loss": 0.0001,
+      "step": 99
+    },
+    {
+      "epoch": 0.006444544692917446,
+      "grad_norm": 0.3241670981018575,
+      "learning_rate": 6.267840395189389e-06,
+      "loss": 0.0019,
+      "step": 100
+    },
+    {
+      "epoch": 0.00650899013984662,
+      "grad_norm": 4.658525973154353,
+      "learning_rate": 6.281383235768169e-06,
+      "loss": 0.0324,
+      "step": 101
+    },
+    {
+      "epoch": 0.0065734355867757945,
+      "grad_norm": 0.13965773384789484,
+      "learning_rate": 6.2947926471768465e-06,
+      "loss": 0.0004,
+      "step": 102
+    },
+    {
+      "epoch": 0.0066378810337049685,
+      "grad_norm": 0.027297791319953163,
+      "learning_rate": 6.30807123297399e-06,
+      "loss": 0.0001,
+      "step": 103
+    },
+    {
+      "epoch": 0.006702326480634143,
+      "grad_norm": 0.014509134729985261,
+      "learning_rate": 6.3212215212503204e-06,
+      "loss": 0.0,
+      "step": 104
+    },
+    {
+      "epoch": 0.006766771927563317,
+      "grad_norm": 0.011064436533537543,
+      "learning_rate": 6.334245967517543e-06,
+      "loss": 0.0,
+      "step": 105
+    },
+    {
+      "epoch": 0.006831217374492492,
+      "grad_norm": 0.33541193592494806,
+      "learning_rate": 6.347146957460263e-06,
+      "loss": 0.0002,
+      "step": 106
+    },
+    {
+      "epoch": 0.006895662821421666,
+      "grad_norm": 0.13366590525564065,
+      "learning_rate": 6.3599268095586996e-06,
+      "loss": 0.0002,
+      "step": 107
+    },
+    {
+      "epoch": 0.006960108268350841,
+      "grad_norm": 0.3804038820837792,
+      "learning_rate": 6.372587777589408e-06,
+      "loss": 0.0033,
+      "step": 108
+    },
+    {
+      "epoch": 0.007024553715280015,
+      "grad_norm": 0.004748762629093711,
+      "learning_rate": 6.385132053010746e-06,
+      "loss": 0.0,
+      "step": 109
+    },
+    {
+      "epoch": 0.00708899916220919,
+      "grad_norm": 0.18445047393599892,
+      "learning_rate": 6.397561767239429e-06,
+      "loss": 0.002,
+      "step": 110
+    },
+    {
+      "epoch": 0.007153444609138364,
+      "grad_norm": 0.10146856090062799,
+      "learning_rate": 6.40987899382405e-06,
+      "loss": 0.0003,
+      "step": 111
+    },
+    {
+      "epoch": 0.007217890056067539,
+      "grad_norm": 0.011157311312939988,
+      "learning_rate": 6.422085750521144e-06,
+      "loss": 0.0,
+      "step": 112
+    },
+    {
+      "epoch": 0.007282335502996713,
+      "grad_norm": 0.607664348129492,
+      "learning_rate": 6.434184001278967e-06,
+      "loss": 0.0028,
+      "step": 113
+    },
+    {
+      "epoch": 0.007346780949925888,
+      "grad_norm": 1.1590725239402295,
+      "learning_rate": 6.446175658133884e-06,
+      "loss": 0.0069,
+      "step": 114
+    },
+    {
+      "epoch": 0.007411226396855062,
+      "grad_norm": 0.0064697996621230795,
+      "learning_rate": 6.458062583023952e-06,
+      "loss": 0.0,
+      "step": 115
+    },
+    {
+      "epoch": 0.007475671843784237,
+      "grad_norm": 0.09498369574237263,
+      "learning_rate": 6.46984658952397e-06,
+      "loss": 0.0002,
+      "step": 116
+    },
+    {
+      "epoch": 0.0075401172907134115,
+      "grad_norm": 0.0215383638363165,
+      "learning_rate": 6.481529444506085e-06,
+      "loss": 0.0001,
+      "step": 117
+    },
+    {
+      "epoch": 0.0076045627376425855,
+      "grad_norm": 0.1171485120767354,
+      "learning_rate": 6.493112869729705e-06,
+      "loss": 0.0002,
+      "step": 118
+    },
+    {
+      "epoch": 0.00766900818457176,
+      "grad_norm": 0.06883933984294645,
+      "learning_rate": 6.5045985433643485e-06,
+      "loss": 0.0003,
+      "step": 119
+    },
+    {
+      "epoch": 0.007733453631500934,
+      "grad_norm": 0.015494437689087105,
+      "learning_rate": 6.515988101448755e-06,
+      "loss": 0.0,
+      "step": 120
+    },
+    {
+      "epoch": 0.007797899078430109,
+      "grad_norm": 1.084890231104155,
+      "learning_rate": 6.5272831392894685e-06,
+      "loss": 0.0045,
+      "step": 121
+    },
+    {
+      "epoch": 0.007862344525359284,
+      "grad_norm": 0.0041539375579682835,
+      "learning_rate": 6.538485212801841e-06,
+      "loss": 0.0,
+      "step": 122
+    },
+    {
+      "epoch": 0.007926789972288458,
+      "grad_norm": 0.021719405589740653,
+      "learning_rate": 6.54959583979632e-06,
+      "loss": 0.0001,
+      "step": 123
+    },
+    {
+      "epoch": 0.007991235419217632,
+      "grad_norm": 0.207106756440556,
+      "learning_rate": 6.560616501212651e-06,
+      "loss": 0.0015,
+      "step": 124
+    },
+    {
+      "epoch": 0.008055680866146806,
+      "grad_norm": 0.12389089981738528,
+      "learning_rate": 6.571548642304503e-06,
+      "loss": 0.0015,
+      "step": 125
+    },
+    {
+      "epoch": 0.008120126313075982,
+      "grad_norm": 0.08421375587621023,
+      "learning_rate": 6.582393673776909e-06,
+      "loss": 0.0017,
+      "step": 126
+    },
+    {
+      "epoch": 0.008184571760005156,
+      "grad_norm": 0.7895623314177207,
+      "learning_rate": 6.593152972878746e-06,
+      "loss": 0.0034,
+      "step": 127
+    },
+    {
+      "epoch": 0.00824901720693433,
+      "grad_norm": 0.009141865289784029,
+      "learning_rate": 6.603827884452357e-06,
+      "loss": 0.0001,
+      "step": 128
+    },
+    {
+      "epoch": 0.008313462653863504,
+      "grad_norm": 0.06799812572737098,
+      "learning_rate": 6.614419721942351e-06,
+      "loss": 0.0002,
+      "step": 129
+    },
+    {
+      "epoch": 0.00837790810079268,
+      "grad_norm": 0.010198638620555195,
+      "learning_rate": 6.624929768365432e-06,
+      "loss": 0.0001,
+      "step": 130
+    },
+    {
+      "epoch": 0.008442353547721854,
+      "grad_norm": 0.14960549750420293,
+      "learning_rate": 6.635359277243088e-06,
+      "loss": 0.0003,
+      "step": 131
+    },
+    {
+      "epoch": 0.008506798994651028,
+      "grad_norm": 0.03669934654162836,
+      "learning_rate": 6.6457094734987945e-06,
+      "loss": 0.0001,
+      "step": 132
+    },
+    {
+      "epoch": 0.008571244441580202,
+      "grad_norm": 0.0583445395178466,
+      "learning_rate": 6.6559815543213856e-06,
+      "loss": 0.0002,
+      "step": 133
+    },
+    {
+      "epoch": 0.008635689888509377,
+      "grad_norm": 0.21443872420801796,
+      "learning_rate": 6.666176689996061e-06,
+      "loss": 0.002,
+      "step": 134
+    },
+    {
+      "epoch": 0.008700135335438551,
+      "grad_norm": 0.5406641455353228,
+      "learning_rate": 6.67629602470452e-06,
+      "loss": 0.0055,
+      "step": 135
+    },
+    {
+      "epoch": 0.008764580782367725,
+      "grad_norm": 0.0021834450013692394,
+      "learning_rate": 6.686340677295562e-06,
+      "loss": 0.0,
+      "step": 136
+    },
+    {
+      "epoch": 0.0088290262292969,
+      "grad_norm": 0.008434527707541637,
+      "learning_rate": 6.696311742027455e-06,
+      "loss": 0.0,
+      "step": 137
+    },
+    {
+      "epoch": 0.008893471676226075,
+      "grad_norm": 0.04693062531604418,
+      "learning_rate": 6.706210289283318e-06,
+      "loss": 0.0002,
+      "step": 138
+    },
+    {
+      "epoch": 0.00895791712315525,
+      "grad_norm": 0.05186746336439399,
+      "learning_rate": 6.7160373662606684e-06,
+      "loss": 0.0002,
+      "step": 139
+    },
+    {
+      "epoch": 0.009022362570084423,
+      "grad_norm": 0.015129704926972309,
+      "learning_rate": 6.7257939976362575e-06,
+      "loss": 0.0,
+      "step": 140
+    },
+    {
+      "epoch": 0.009086808017013597,
+      "grad_norm": 0.021042850048326282,
+      "learning_rate": 6.735481186207243e-06,
+      "loss": 0.0001,
+      "step": 141
+    },
+    {
+      "epoch": 0.009151253463942773,
+      "grad_norm": 0.02537888704924219,
+      "learning_rate": 6.745099913509707e-06,
+      "loss": 0.0001,
+      "step": 142
+    },
+    {
+      "epoch": 0.009215698910871947,
+      "grad_norm": 0.02671439549274601,
+      "learning_rate": 6.754651140415472e-06,
+      "loss": 0.0,
+      "step": 143
+    },
+    {
+      "epoch": 0.009280144357801121,
+      "grad_norm": 0.008137524957251669,
+      "learning_rate": 6.764135807708122e-06,
+      "loss": 0.0,
+      "step": 144
+    },
+    {
+      "epoch": 0.009344589804730297,
+      "grad_norm": 0.003042221966208203,
+      "learning_rate": 6.773554836639082e-06,
+      "loss": 0.0,
+      "step": 145
+    },
+    {
+      "epoch": 0.00940903525165947,
+      "grad_norm": 0.06033152692120294,
+      "learning_rate": 6.782909129464605e-06,
+      "loss": 0.0002,
+      "step": 146
+    },
+    {
+      "epoch": 0.009473480698588645,
+      "grad_norm": 0.027788121319448158,
+      "learning_rate": 6.792199569964411e-06,
+      "loss": 0.0,
+      "step": 147
+    },
+    {
+      "epoch": 0.009537926145517819,
+      "grad_norm": 0.09830052518820728,
+      "learning_rate": 6.801427023942765e-06,
+      "loss": 0.0009,
+      "step": 148
+    },
+    {
+      "epoch": 0.009602371592446994,
+      "grad_norm": 0.013099850923853006,
+      "learning_rate": 6.810592339712671e-06,
+      "loss": 0.0001,
+      "step": 149
+    },
+    {
+      "epoch": 0.009666817039376168,
+      "grad_norm": 0.3610734402843211,
+      "learning_rate": 6.819696348563868e-06,
+      "loss": 0.005,
+      "step": 150
+    },
+    {
+      "epoch": 0.009731262486305342,
+      "grad_norm": 0.22420221826039138,
+      "learning_rate": 6.828739865215293e-06,
+      "loss": 0.0025,
+      "step": 151
+    },
+    {
+      "epoch": 0.009795707933234516,
+      "grad_norm": 0.026359216933519877,
+      "learning_rate": 6.837723688252599e-06,
+      "loss": 0.0001,
+      "step": 152
+    },
+    {
+      "epoch": 0.009860153380163692,
+      "grad_norm": 0.37353960421039445,
+      "learning_rate": 6.846648600551325e-06,
+      "loss": 0.0013,
+      "step": 153
+    },
+    {
+      "epoch": 0.009924598827092866,
+      "grad_norm": 0.37925933825010816,
+      "learning_rate": 6.8555153696862974e-06,
+      "loss": 0.0014,
+      "step": 154
+    },
+    {
+      "epoch": 0.00998904427402204,
+      "grad_norm": 0.015689512713385756,
+      "learning_rate": 6.864324748327763e-06,
+      "loss": 0.0001,
+      "step": 155
+    },
+    {
+      "epoch": 0.010053489720951214,
+      "grad_norm": 0.42026716718154256,
+      "learning_rate": 6.8730774746248e-06,
+      "loss": 0.0027,
+      "step": 156
+    },
+    {
+      "epoch": 0.01011793516788039,
+      "grad_norm": 0.03761871169419944,
+      "learning_rate": 6.881774272576467e-06,
+      "loss": 0.0001,
+      "step": 157
+    },
+    {
+      "epoch": 0.010182380614809564,
+      "grad_norm": 0.030528250735410357,
+      "learning_rate": 6.89041585239118e-06,
+      "loss": 0.0002,
+      "step": 158
+    },
+    {
+      "epoch": 0.010246826061738738,
+      "grad_norm": 0.20077781600093578,
+      "learning_rate": 6.899002910834743e-06,
+      "loss": 0.0009,
+      "step": 159
+    },
+    {
+      "epoch": 0.010311271508667912,
+      "grad_norm": 1.613127029104335,
+      "learning_rate": 6.907536131567471e-06,
+      "loss": 0.0053,
+      "step": 160
+    },
+    {
+      "epoch": 0.010375716955597088,
+      "grad_norm": 0.39481368200606176,
+      "learning_rate": 6.91601618547082e-06,
+      "loss": 0.0103,
+      "step": 161
+    },
+    {
+      "epoch": 0.010440162402526262,
+      "grad_norm": 0.03645218519897308,
+      "learning_rate": 6.9244437309638855e-06,
+      "loss": 0.0001,
+      "step": 162
+    },
+    {
+      "epoch": 0.010504607849455436,
+      "grad_norm": 0.026445971997361595,
+      "learning_rate": 6.932819414310185e-06,
+      "loss": 0.0001,
+      "step": 163
+    },
+    {
+      "epoch": 0.01056905329638461,
+      "grad_norm": 0.012293269592344393,
+      "learning_rate": 6.941143869915035e-06,
+      "loss": 0.0,
+      "step": 164
+    },
+    {
+      "epoch": 0.010633498743313785,
+      "grad_norm": 0.006382313439032722,
+      "learning_rate": 6.949417720613908e-06,
+      "loss": 0.0,
+      "step": 165
+    },
+    {
+      "epoch": 0.01069794419024296,
+      "grad_norm": 0.18303938626882663,
+      "learning_rate": 6.95764157795207e-06,
+      "loss": 0.0007,
+      "step": 166
+    },
+    {
+      "epoch": 0.010762389637172133,
+      "grad_norm": 0.08027807718869757,
+      "learning_rate": 6.965816042455817e-06,
+      "loss": 0.0002,
+      "step": 167
+    },
+    {
+      "epoch": 0.010826835084101307,
+      "grad_norm": 0.009543528622426802,
+      "learning_rate": 6.9739417038956234e-06,
+      "loss": 0.0,
+      "step": 168
+    },
+    {
+      "epoch": 0.010891280531030483,
+      "grad_norm": 0.005663156424081436,
+      "learning_rate": 6.982019141541477e-06,
+      "loss": 0.0,
+      "step": 169
+    },
+    {
+      "epoch": 0.010955725977959657,
+      "grad_norm": 0.06478321556126382,
+      "learning_rate": 6.990048924410674e-06,
+      "loss": 0.0001,
+      "step": 170
+    },
+    {
+      "epoch": 0.011020171424888831,
+      "grad_norm": 0.11265850722264495,
+      "learning_rate": 6.998031611508364e-06,
+      "loss": 0.0002,
+      "step": 171
+    },
+    {
+      "epoch": 0.011084616871818005,
+      "grad_norm": 0.13747799410071634,
+      "learning_rate": 7.005967752061066e-06,
+      "loss": 0.0017,
+      "step": 172
+    },
+    {
+      "epoch": 0.011149062318747181,
+      "grad_norm": 0.8610413781168746,
+      "learning_rate": 7.013857885743432e-06,
+      "loss": 0.0035,
+      "step": 173
+    },
+    {
+      "epoch": 0.011213507765676355,
+      "grad_norm": 2.1479235225041444,
+      "learning_rate": 7.02170254289845e-06,
+      "loss": 0.0081,
+      "step": 174
+    },
+    {
+      "epoch": 0.011277953212605529,
+      "grad_norm": 0.19717496733882242,
+      "learning_rate": 7.02950224475137e-06,
+      "loss": 0.0009,
+      "step": 175
+    },
+    {
+      "epoch": 0.011342398659534705,
+      "grad_norm": 0.12631252498003578,
+      "learning_rate": 7.037257503617509e-06,
+      "loss": 0.0007,
+      "step": 176
+    },
+    {
+      "epoch": 0.011406844106463879,
+      "grad_norm": 0.002589981572960044,
+      "learning_rate": 7.044968823104184e-06,
+      "loss": 0.0,
+      "step": 177
+    },
+    {
+      "epoch": 0.011471289553393053,
+      "grad_norm": 0.048464294586520046,
+      "learning_rate": 7.052636698306942e-06,
+      "loss": 0.0001,
+      "step": 178
+    },
+    {
+      "epoch": 0.011535735000322227,
+      "grad_norm": 0.008706107825396808,
+      "learning_rate": 7.060261616000313e-06,
+      "loss": 0.0,
+      "step": 179
+    },
+    {
+      "epoch": 0.011600180447251402,
+      "grad_norm": 0.016572784451954176,
+      "learning_rate": 7.067844054823235e-06,
+      "loss": 0.0002,
+      "step": 180
+    },
+    {
+      "epoch": 0.011664625894180576,
+      "grad_norm": 0.011129301311151215,
+      "learning_rate": 7.075384485459343e-06,
+      "loss": 0.0001,
+      "step": 181
+    },
+    {
+      "epoch": 0.01172907134110975,
+      "grad_norm": 0.039764761465697955,
+      "learning_rate": 7.082883370812301e-06,
+      "loss": 0.0001,
+      "step": 182
+    },
+    {
+      "epoch": 0.011793516788038924,
+      "grad_norm": 0.024318506875899433,
+      "learning_rate": 7.09034116617632e-06,
+      "loss": 0.0001,
+      "step": 183
+    },
+    {
+      "epoch": 0.0118579622349681,
+      "grad_norm": 2.2273904967445186,
+      "learning_rate": 7.0977583194020325e-06,
+      "loss": 0.012,
+      "step": 184
+    },
+    {
+      "epoch": 0.011922407681897274,
+      "grad_norm": 0.006233002881990179,
+      "learning_rate": 7.105135271057878e-06,
+      "loss": 0.0,
+      "step": 185
+    },
+    {
+      "epoch": 0.011986853128826448,
+      "grad_norm": 0.09233918702362297,
+      "learning_rate": 7.1124724545871305e-06,
+      "loss": 0.0004,
+      "step": 186
+    },
+    {
+      "epoch": 0.012051298575755622,
+      "grad_norm": 0.006889842787298746,
+      "learning_rate": 7.119770296460714e-06,
+      "loss": 0.0,
+      "step": 187
+    },
+    {
+      "epoch": 0.012115744022684798,
+      "grad_norm": 0.03291308421119183,
+      "learning_rate": 7.127029216325958e-06,
+      "loss": 0.0002,
+      "step": 188
+    },
+    {
+      "epoch": 0.012180189469613972,
+      "grad_norm": 0.4183982353790482,
+      "learning_rate": 7.1342496271513885e-06,
+      "loss": 0.0006,
+      "step": 189
+    },
+    {
+      "epoch": 0.012244634916543146,
+      "grad_norm": 0.0010894861958931816,
+      "learning_rate": 7.141431935367712e-06,
+      "loss": 0.0,
+      "step": 190
+    },
+    {
+      "epoch": 0.01230908036347232,
+      "grad_norm": 0.1424472482504031,
+      "learning_rate": 7.14857654100509e-06,
+      "loss": 0.0005,
+      "step": 191
+    },
+    {
+      "epoch": 0.012373525810401496,
+      "grad_norm": 0.0006208977366605882,
+      "learning_rate": 7.155683837826838e-06,
+      "loss": 0.0,
+      "step": 192
+    },
+    {
+      "epoch": 0.01243797125733067,
+      "grad_norm": 0.006579198044893609,
+      "learning_rate": 7.162754213459641e-06,
+      "loss": 0.0001,
+      "step": 193
+    },
+    {
+      "epoch": 0.012502416704259844,
+      "grad_norm": 0.0037932328500453966,
+      "learning_rate": 7.1697880495204184e-06,
+      "loss": 0.0,
+      "step": 194
+    },
+    {
+      "epoch": 0.012566862151189018,
+      "grad_norm": 3.7558033860150153,
+      "learning_rate": 7.176785721739913e-06,
+      "loss": 0.0326,
+      "step": 195
+    },
+    {
+      "epoch": 0.012631307598118193,
+      "grad_norm": 0.048965416061562524,
+      "learning_rate": 7.1837476000831255e-06,
+      "loss": 0.0001,
+      "step": 196
+    },
+    {
+      "epoch": 0.012695753045047367,
+      "grad_norm": 0.004427047393266682,
+      "learning_rate": 7.190674048866693e-06,
+      "loss": 0.0,
+      "step": 197
+    },
+    {
+      "epoch": 0.012760198491976541,
+      "grad_norm": 0.8024053711522192,
+      "learning_rate": 7.197565426873274e-06,
+      "loss": 0.0031,
+      "step": 198
+    },
+    {
+      "epoch": 0.012824643938905715,
+      "grad_norm": 0.16278141967045448,
+      "learning_rate": 7.20442208746308e-06,
+      "loss": 0.0005,
+      "step": 199
+    },
+    {
+      "epoch": 0.012889089385834891,
+      "grad_norm": 0.009971334443131889,
+      "learning_rate": 7.211244378682583e-06,
+      "loss": 0.0001,
+      "step": 200
+    },
+    {
+      "epoch": 0.012953534832764065,
+      "grad_norm": 0.0020690471456571028,
+      "learning_rate": 7.218032643370541e-06,
+      "loss": 0.0,
+      "step": 201
+    },
+    {
+      "epoch": 0.01301798027969324,
+      "grad_norm": 0.03277078691989995,
+      "learning_rate": 7.224787219261363e-06,
+      "loss": 0.0001,
+      "step": 202
+    },
+    {
+      "epoch": 0.013082425726622415,
+      "grad_norm": 0.01937230899628366,
+      "learning_rate": 7.231508439085952e-06,
+      "loss": 0.0001,
+      "step": 203
+    },
+    {
+      "epoch": 0.013146871173551589,
+      "grad_norm": 0.05567845986542249,
+      "learning_rate": 7.238196630670041e-06,
+      "loss": 0.0005,
+      "step": 204
+    },
+    {
+      "epoch": 0.013211316620480763,
+      "grad_norm": 0.004371745422922564,
+      "learning_rate": 7.244852117030148e-06,
+      "loss": 0.0,
+      "step": 205
+    },
+    {
+      "epoch": 0.013275762067409937,
+      "grad_norm": 0.0070734951743021,
+      "learning_rate": 7.251475216467184e-06,
+      "loss": 0.0001,
+      "step": 206
+    },
+    {
+      "epoch": 0.013340207514339113,
+      "grad_norm": 0.02153720307664906,
+      "learning_rate": 7.258066242657797e-06,
+      "loss": 0.0001,
+      "step": 207
+    },
+    {
+      "epoch": 0.013404652961268287,
+      "grad_norm": 0.12040073376471393,
+      "learning_rate": 7.2646255047435146e-06,
+      "loss": 0.0004,
+      "step": 208
+    },
+    {
+      "epoch": 0.01346909840819746,
+      "grad_norm": 0.0004398452218530347,
+      "learning_rate": 7.271153307417751e-06,
+      "loss": 0.0,
+      "step": 209
+    },
+    {
+      "epoch": 0.013533543855126635,
+      "grad_norm": 0.04042410483723827,
+      "learning_rate": 7.277649951010737e-06,
+      "loss": 0.0003,
+      "step": 210
+    },
+    {
+      "epoch": 0.01359798930205581,
+      "grad_norm": 0.06771473747313296,
+      "learning_rate": 7.284115731572427e-06,
+      "loss": 0.0001,
+      "step": 211
+    },
+    {
+      "epoch": 0.013662434748984984,
+      "grad_norm": 0.01151591523695472,
+      "learning_rate": 7.290550940953457e-06,
+      "loss": 0.0,
+      "step": 212
+    },
+    {
+      "epoch": 0.013726880195914158,
+      "grad_norm": 0.4189345173901024,
+      "learning_rate": 7.296955866884187e-06,
+      "loss": 0.003,
+      "step": 213
+    },
+    {
+      "epoch": 0.013791325642843332,
+      "grad_norm": 0.04164191329700527,
+      "learning_rate": 7.303330793051894e-06,
+      "loss": 0.0001,
+      "step": 214
+    },
+    {
+      "epoch": 0.013855771089772508,
+      "grad_norm": 0.45272719251647475,
+      "learning_rate": 7.309675999176179e-06,
+      "loss": 0.0042,
+      "step": 215
+    },
+    {
+      "epoch": 0.013920216536701682,
+      "grad_norm": 0.10562175672393527,
+      "learning_rate": 7.315991761082603e-06,
+      "loss": 0.0001,
+      "step": 216
+    },
+    {
+      "epoch": 0.013984661983630856,
+      "grad_norm": 0.004886530786615905,
+      "learning_rate": 7.322278350774632e-06,
+      "loss": 0.0,
+      "step": 217
+    },
+    {
+      "epoch": 0.01404910743056003,
+      "grad_norm": 0.0305665453017975,
+      "learning_rate": 7.328536036503941e-06,
+      "loss": 0.0001,
+      "step": 218
+    },
+    {
+      "epoch": 0.014113552877489206,
+      "grad_norm": 0.03056080236649342,
+      "learning_rate": 7.334765082839083e-06,
+      "loss": 0.0002,
+      "step": 219
+    },
+    {
+      "epoch": 0.01417799832441838,
+      "grad_norm": 0.01914917310382882,
+      "learning_rate": 7.3409657507326235e-06,
+      "loss": 0.0001,
+      "step": 220
+    },
+    {
+      "epoch": 0.014242443771347554,
+      "grad_norm": 0.03466315135259134,
+      "learning_rate": 7.347138297586717e-06,
+      "loss": 0.0,
+      "step": 221
+    },
+    {
+      "epoch": 0.014306889218276728,
+      "grad_norm": 0.005841761742250744,
+      "learning_rate": 7.353282977317245e-06,
+      "loss": 0.0001,
+      "step": 222
+    },
+    {
+      "epoch": 0.014371334665205904,
+      "grad_norm": 0.014125468048465728,
+      "learning_rate": 7.359400040416475e-06,
+      "loss": 0.0001,
+      "step": 223
+    },
+    {
+      "epoch": 0.014435780112135078,
+      "grad_norm": 0.1470687716683251,
+      "learning_rate": 7.365489734014338e-06,
+      "loss": 0.0013,
+      "step": 224
+    },
+    {
+      "epoch": 0.014500225559064252,
+      "grad_norm": 0.1316313746000964,
+      "learning_rate": 7.371552301938347e-06,
+      "loss": 0.0011,
+      "step": 225
+    },
+    {
+      "epoch": 0.014564671005993426,
+      "grad_norm": 1.216626395686913,
+      "learning_rate": 7.377587984772161e-06,
+      "loss": 0.003,
+      "step": 226
+    },
+    {
+      "epoch": 0.014629116452922602,
+      "grad_norm": 0.030215975039620327,
+      "learning_rate": 7.3835970199128815e-06,
+      "loss": 0.0001,
+      "step": 227
+    },
+    {
+      "epoch": 0.014693561899851776,
+      "grad_norm": 0.3925818447866758,
+      "learning_rate": 7.3895796416270785e-06,
+      "loss": 0.0014,
+      "step": 228
+    },
+    {
+      "epoch": 0.01475800734678095,
+      "grad_norm": 0.011962584082899389,
+      "learning_rate": 7.395536081105594e-06,
+      "loss": 0.0,
+      "step": 229
+    },
+    {
+      "epoch": 0.014822452793710124,
+      "grad_norm": 0.31894698647727304,
+      "learning_rate": 7.401466566517146e-06,
+      "loss": 0.0012,
+      "step": 230
+    },
+    {
+      "epoch": 0.0148868982406393,
+      "grad_norm": 0.003858933499673059,
+      "learning_rate": 7.407371323060776e-06,
+      "loss": 0.0,
+      "step": 231
+    },
+    {
+      "epoch": 0.014951343687568473,
+      "grad_norm": 0.002184925615570465,
+      "learning_rate": 7.413250573017164e-06,
+      "loss": 0.0,
+      "step": 232
+    },
+    {
+      "epoch": 0.015015789134497647,
+      "grad_norm": 0.0575590729604499,
+      "learning_rate": 7.4191045357988316e-06,
+      "loss": 0.0002,
+      "step": 233
+    },
+    {
+      "epoch": 0.015080234581426823,
+      "grad_norm": 0.0851483114943205,
+      "learning_rate": 7.424933427999279e-06,
+      "loss": 0.0001,
+      "step": 234
+    },
+    {
+      "epoch": 0.015144680028355997,
+      "grad_norm": 0.023247510036130875,
+      "learning_rate": 7.43073746344107e-06,
+      "loss": 0.0002,
+      "step": 235
+    },
+    {
+      "epoch": 0.015209125475285171,
+      "grad_norm": 0.005134519637574446,
+      "learning_rate": 7.436516853222899e-06,
+      "loss": 0.0,
+      "step": 236
+    },
+    {
+      "epoch": 0.015273570922214345,
+      "grad_norm": 0.012576069907377694,
+      "learning_rate": 7.442271805765659e-06,
+      "loss": 0.0,
+      "step": 237
+    },
+    {
+      "epoch": 0.01533801636914352,
+      "grad_norm": 0.0041381520945973395,
+      "learning_rate": 7.448002526857543e-06,
+      "loss": 0.0,
+      "step": 238
+    },
+    {
+      "epoch": 0.015402461816072695,
+      "grad_norm": 0.1408810734446625,
+      "learning_rate": 7.453709219698196e-06,
+      "loss": 0.0014,
+      "step": 239
+    },
+    {
+      "epoch": 0.015466907263001869,
+      "grad_norm": 0.02099080559759998,
+      "learning_rate": 7.4593920849419495e-06,
+      "loss": 0.0001,
+      "step": 240
+    },
+    {
+      "epoch": 0.015531352709931043,
+      "grad_norm": 0.34798627897458273,
+      "learning_rate": 7.465051320740162e-06,
+      "loss": 0.0014,
+      "step": 241
+    },
+    {
+      "epoch": 0.015595798156860219,
+      "grad_norm": 0.03753130495216673,
+      "learning_rate": 7.470687122782663e-06,
+      "loss": 0.0003,
+      "step": 242
+    },
+    {
+      "epoch": 0.01566024360378939,
+      "grad_norm": 0.26285752177303706,
+      "learning_rate": 7.476299684338365e-06,
+      "loss": 0.0014,
+      "step": 243
+    },
+    {
+      "epoch": 0.015724689050718568,
+      "grad_norm": 0.011810539302378758,
+      "learning_rate": 7.481889196295035e-06,
+      "loss": 0.0,
+      "step": 244
+    },
+    {
+      "epoch": 0.015789134497647742,
+      "grad_norm": 0.2277299056097155,
+      "learning_rate": 7.487455847198239e-06,
+      "loss": 0.0012,
+      "step": 245
+    },
+    {
+      "epoch": 0.015853579944576916,
+      "grad_norm": 0.004420655223345774,
+      "learning_rate": 7.492999823289514e-06,
+      "loss": 0.0,
+      "step": 246
+    },
+    {
+      "epoch": 0.01591802539150609,
+      "grad_norm": 0.09830975739963607,
+      "learning_rate": 7.498521308543755e-06,
+      "loss": 0.0005,
+      "step": 247
+    },
+    {
+      "epoch": 0.015982470838435264,
+      "grad_norm": 0.045239574227459146,
+      "learning_rate": 7.504020484705845e-06,
+      "loss": 0.0001,
+      "step": 248
+    },
+    {
+      "epoch": 0.01604691628536444,
+      "grad_norm": 0.004480344683799468,
+      "learning_rate": 7.5094975313265485e-06,
+      "loss": 0.0,
+      "step": 249
+    },
+    {
+      "epoch": 0.016111361732293612,
+      "grad_norm": 0.04672057822806287,
+      "learning_rate": 7.5149526257976954e-06,
+      "loss": 0.0001,
+      "step": 250
+    },
+    {
+      "epoch": 0.016175807179222786,
+      "grad_norm": 0.27883213565129883,
+      "learning_rate": 7.5203859433866514e-06,
+      "loss": 0.001,
+      "step": 251
+    },
+    {
+      "epoch": 0.016240252626151964,
+      "grad_norm": 0.06406857182641057,
+      "learning_rate": 7.525797657270103e-06,
+      "loss": 0.0001,
+      "step": 252
+    },
+    {
+      "epoch": 0.016304698073081138,
+      "grad_norm": 0.1080837493277941,
+      "learning_rate": 7.531187938567185e-06,
+      "loss": 0.0004,
+      "step": 253
+    },
+    {
+      "epoch": 0.016369143520010312,
+      "grad_norm": 0.03588783821123804,
+      "learning_rate": 7.53655695637194e-06,
+      "loss": 0.0003,
+      "step": 254
+    },
+    {
+      "epoch": 0.016433588966939486,
+      "grad_norm": 0.16026049803218403,
+      "learning_rate": 7.5419048777851525e-06,
+      "loss": 0.0018,
+      "step": 255
+    },
+    {
+      "epoch": 0.01649803441386866,
+      "grad_norm": 1.9165503701573037,
+      "learning_rate": 7.547231867945552e-06,
+      "loss": 0.0132,
+      "step": 256
+    },
+    {
+      "epoch": 0.016562479860797834,
+      "grad_norm": 0.015958733921597468,
+      "learning_rate": 7.55253809006041e-06,
+      "loss": 0.0,
+      "step": 257
+    },
+    {
+      "epoch": 0.016626925307727008,
+      "grad_norm": 0.0019534432535608308,
+      "learning_rate": 7.557823705435545e-06,
+      "loss": 0.0,
+      "step": 258
+    },
+    {
+      "epoch": 0.016691370754656182,
+      "grad_norm": 0.02074439107330492,
+      "learning_rate": 7.5630888735047455e-06,
+      "loss": 0.0,
+      "step": 259
+    },
+    {
+      "epoch": 0.01675581620158536,
+      "grad_norm": 1.1553521149673944,
+      "learning_rate": 7.568333751858627e-06,
+      "loss": 0.007,
+      "step": 260
+    },
+    {
+      "epoch": 0.016820261648514533,
+      "grad_norm": 0.3606061372170696,
+      "learning_rate": 7.573558496272929e-06,
+      "loss": 0.0028,
+      "step": 261
+    },
+    {
+      "epoch": 0.016884707095443707,
+      "grad_norm": 0.00275996815502147,
+      "learning_rate": 7.578763260736281e-06,
+      "loss": 0.0,
+      "step": 262
+    },
+    {
+      "epoch": 0.01694915254237288,
+      "grad_norm": 0.004268297014106991,
+      "learning_rate": 7.58394819747744e-06,
+      "loss": 0.0,
+      "step": 263
+    },
+    {
+      "epoch": 0.017013597989302055,
+      "grad_norm": 0.0367751156855776,
+      "learning_rate": 7.589113456991989e-06,
+      "loss": 0.0,
+      "step": 264
+    },
+    {
+      "epoch": 0.01707804343623123,
+      "grad_norm": 0.03272693147392111,
+      "learning_rate": 7.594259188068569e-06,
+      "loss": 0.0004,
+      "step": 265
+    },
+    {
+      "epoch": 0.017142488883160403,
+      "grad_norm": 0.2226594205262504,
+      "learning_rate": 7.59938553781458e-06,
+      "loss": 0.0008,
+      "step": 266
+    },
+    {
+      "epoch": 0.01720693433008958,
+      "grad_norm": 4.17899426567084,
+      "learning_rate": 7.604492651681422e-06,
+      "loss": 0.0028,
+      "step": 267
+    },
+    {
+      "epoch": 0.017271379777018755,
+      "grad_norm": 0.314194863453364,
+      "learning_rate": 7.609580673489255e-06,
+      "loss": 0.0017,
+      "step": 268
+    },
+    {
+      "epoch": 0.01733582522394793,
+      "grad_norm": 0.009341551827583042,
+      "learning_rate": 7.614649745451305e-06,
+      "loss": 0.0,
+      "step": 269
+    },
+    {
+      "epoch": 0.017400270670877103,
+      "grad_norm": 0.006294262625942665,
+      "learning_rate": 7.6197000081977145e-06,
+      "loss": 0.0,
+      "step": 270
+    },
+    {
+      "epoch": 0.017464716117806277,
+      "grad_norm": 0.042152326765031375,
+      "learning_rate": 7.624731600798942e-06,
+      "loss": 0.0003,
+      "step": 271
+    },
+    {
+      "epoch": 0.01752916156473545,
+      "grad_norm": 0.005840402112676206,
+      "learning_rate": 7.629744660788755e-06,
+      "loss": 0.0,
+      "step": 272
+    },
+    {
+      "epoch": 0.017593607011664625,
+      "grad_norm": 0.003961071319983515,
+      "learning_rate": 7.63473932418678e-06,
+      "loss": 0.0,
+      "step": 273
+    },
+    {
+      "epoch": 0.0176580524585938,
+      "grad_norm": 0.03416186617415247,
+      "learning_rate": 7.639715725520648e-06,
+      "loss": 0.0003,
+      "step": 274
+    },
+    {
+      "epoch": 0.017722497905522976,
+      "grad_norm": 0.011693800006046062,
+      "learning_rate": 7.644673997847735e-06,
+      "loss": 0.0,
+      "step": 275
+    },
+    {
+      "epoch": 0.01778694335245215,
+      "grad_norm": 0.013642164927569226,
+      "learning_rate": 7.64961427277651e-06,
+      "loss": 0.0001,
+      "step": 276
+    },
+    {
+      "epoch": 0.017851388799381324,
+      "grad_norm": 0.005290876124771607,
+      "learning_rate": 7.6545366804875e-06,
+      "loss": 0.0,
+      "step": 277
+    },
+    {
+      "epoch": 0.0179158342463105,
+      "grad_norm": 0.3685656717237363,
+      "learning_rate": 7.659441349753862e-06,
+      "loss": 0.0047,
+      "step": 278
+    },
+    {
+      "epoch": 0.017980279693239672,
+      "grad_norm": 0.162786871985292,
+      "learning_rate": 7.664328407961608e-06,
+      "loss": 0.0003,
+      "step": 279
+    },
+    {
+      "epoch": 0.018044725140168846,
+      "grad_norm": 0.0038340519806686416,
+      "learning_rate": 7.669197981129452e-06,
+      "loss": 0.0,
+      "step": 280
+    },
+    {
+      "epoch": 0.01810917058709802,
+      "grad_norm": 1.1800076890145348,
+      "learning_rate": 7.674050193928305e-06,
+      "loss": 0.0107,
+      "step": 281
+    },
+    {
+      "epoch": 0.018173616034027194,
+      "grad_norm": 0.4022661152723877,
+      "learning_rate": 7.678885169700438e-06,
+      "loss": 0.0011,
+      "step": 282
+    },
+    {
+      "epoch": 0.018238061480956372,
+      "grad_norm": 0.0010258398165785125,
+      "learning_rate": 7.683703030478275e-06,
+      "loss": 0.0,
+      "step": 283
+    },
+    {
+      "epoch": 0.018302506927885546,
+      "grad_norm": 0.03959304117747763,
+      "learning_rate": 7.688503897002901e-06,
+      "loss": 0.0002,
+      "step": 284
+    },
+    {
+      "epoch": 0.01836695237481472,
+      "grad_norm": 0.003900586596470166,
+      "learning_rate": 7.693287888742192e-06,
+      "loss": 0.0,
+      "step": 285
+    },
+    {
+      "epoch": 0.018431397821743894,
+      "grad_norm": 0.0016122389663264988,
+      "learning_rate": 7.698055123908667e-06,
+      "loss": 0.0,
+      "step": 286
+    },
+    {
+      "epoch": 0.018495843268673068,
+      "grad_norm": 0.04600308794351545,
+      "learning_rate": 7.702805719477016e-06,
+      "loss": 0.0001,
+      "step": 287
+    },
+    {
+      "epoch": 0.018560288715602242,
+      "grad_norm": 0.7386376867749295,
+      "learning_rate": 7.707539791201318e-06,
+      "loss": 0.0042,
+      "step": 288
+    },
+    {
+      "epoch": 0.018624734162531416,
+      "grad_norm": 0.07433124536949527,
+      "learning_rate": 7.712257453631959e-06,
+      "loss": 0.0001,
+      "step": 289
+    },
+    {
+      "epoch": 0.018689179609460593,
+      "grad_norm": 0.013048752330046877,
+      "learning_rate": 7.716958820132276e-06,
+      "loss": 0.0001,
+      "step": 290
+    },
+    {
+      "epoch": 0.018753625056389767,
+      "grad_norm": 0.007703258584517692,
+      "learning_rate": 7.721644002894899e-06,
+      "loss": 0.0001,
+      "step": 291
+    },
+    {
+      "epoch": 0.01881807050331894,
+      "grad_norm": 0.09900497813725398,
+      "learning_rate": 7.726313112957798e-06,
+      "loss": 0.0004,
+      "step": 292
+    },
+    {
+      "epoch": 0.018882515950248115,
+      "grad_norm": 0.7346640890359516,
+      "learning_rate": 7.730966260220105e-06,
+      "loss": 0.0083,
+      "step": 293
+    },
+    {
+      "epoch": 0.01894696139717729,
+      "grad_norm": 0.09893905196656488,
+      "learning_rate": 7.735603553457605e-06,
+      "loss": 0.0004,
+      "step": 294
+    },
+    {
+      "epoch": 0.019011406844106463,
+      "grad_norm": 0.24004125974315293,
+      "learning_rate": 7.740225100338013e-06,
+      "loss": 0.0021,
+      "step": 295
+    },
+    {
+      "epoch": 0.019075852291035637,
+      "grad_norm": 0.2778235435865419,
+      "learning_rate": 7.74483100743596e-06,
+      "loss": 0.0005,
+      "step": 296
+    },
+    {
+      "epoch": 0.01914029773796481,
+      "grad_norm": 0.011061506543972114,
+      "learning_rate": 7.749421380247754e-06,
+      "loss": 0.0,
+      "step": 297
+    },
+    {
+      "epoch": 0.01920474318489399,
+      "grad_norm": 0.0099751102256593,
+      "learning_rate": 7.753996323205866e-06,
+      "loss": 0.0,
+      "step": 298
+    },
+    {
+      "epoch": 0.019269188631823163,
+      "grad_norm": 0.04293211545664707,
+      "learning_rate": 7.758555939693189e-06,
+      "loss": 0.0001,
+      "step": 299
+    },
+    {
+      "epoch": 0.019333634078752337,
+      "grad_norm": 0.14890663148925204,
+      "learning_rate": 7.763100332057061e-06,
+      "loss": 0.0017,
+      "step": 300
+    },
+    {
+      "epoch": 0.01939807952568151,
+      "grad_norm": 0.051856462016533725,
+      "learning_rate": 7.767629601623046e-06,
+      "loss": 0.0003,
+      "step": 301
+    },
+    {
+      "epoch": 0.019462524972610685,
+      "grad_norm": 0.20118795719491775,
+      "learning_rate": 7.77214384870849e-06,
+      "loss": 0.0008,
+      "step": 302
+    },
+    {
+      "epoch": 0.01952697041953986,
+      "grad_norm": 0.01699588268895743,
+      "learning_rate": 7.776643172635842e-06,
+      "loss": 0.0001,
+      "step": 303
+    },
+    {
+      "epoch": 0.019591415866469033,
+      "grad_norm": 0.28016719067002793,
+      "learning_rate": 7.781127671745794e-06,
+      "loss": 0.0005,
+      "step": 304
+    },
+    {
+      "epoch": 0.019655861313398207,
+      "grad_norm": 0.13740708235192936,
+      "learning_rate": 7.785597443410148e-06,
+      "loss": 0.0002,
+      "step": 305
+    },
+    {
+      "epoch": 0.019720306760327384,
+      "grad_norm": 0.13476762217271904,
+      "learning_rate": 7.79005258404452e-06,
+      "loss": 0.0009,
+      "step": 306
+    },
+    {
+      "epoch": 0.01978475220725656,
+      "grad_norm": 0.014737783093005689,
+      "learning_rate": 7.794493189120813e-06,
+      "loss": 0.0001,
+      "step": 307
+    },
+    {
+      "epoch": 0.019849197654185732,
+      "grad_norm": 1.1847140609432278,
+      "learning_rate": 7.79891935317949e-06,
+      "loss": 0.0063,
+      "step": 308
+    },
+    {
+      "epoch": 0.019913643101114906,
+      "grad_norm": 0.004352597997975502,
+      "learning_rate": 7.803331169841663e-06,
+      "loss": 0.0,
+      "step": 309
+    },
+    {
+      "epoch": 0.01997808854804408,
+      "grad_norm": 0.025632327962959205,
+      "learning_rate": 7.807728731820957e-06,
+      "loss": 0.0002,
+      "step": 310
+    },
+    {
+      "epoch": 0.020042533994973254,
+      "grad_norm": 0.054951094255651874,
+      "learning_rate": 7.812112130935214e-06,
+      "loss": 0.0003,
+      "step": 311
+    },
+    {
+      "epoch": 0.02010697944190243,
+      "grad_norm": 0.10254521578536714,
+      "learning_rate": 7.816481458117993e-06,
+      "loss": 0.0002,
+      "step": 312
+    },
+    {
+      "epoch": 0.020171424888831602,
+      "grad_norm": 0.14173059175345795,
+      "learning_rate": 7.820836803429887e-06,
+      "loss": 0.0004,
+      "step": 313
+    },
+    {
+      "epoch": 0.02023587033576078,
+      "grad_norm": 0.010335749948364018,
+      "learning_rate": 7.825178256069662e-06,
+      "loss": 0.0,
+      "step": 314
+    },
+    {
+      "epoch": 0.020300315782689954,
+      "grad_norm": 0.009398006668248197,
+      "learning_rate": 7.829505904385217e-06,
+      "loss": 0.0,
+      "step": 315
+    },
+    {
+      "epoch": 0.020364761229619128,
+      "grad_norm": 0.0008854282801088228,
+      "learning_rate": 7.833819835884374e-06,
+      "loss": 0.0,
+      "step": 316
+    },
+    {
+      "epoch": 0.020429206676548302,
+      "grad_norm": 0.01692642260677151,
+      "learning_rate": 7.838120137245496e-06,
+      "loss": 0.0001,
+      "step": 317
+    },
+    {
+      "epoch": 0.020493652123477476,
+      "grad_norm": 0.012734900473579882,
+      "learning_rate": 7.842406894327936e-06,
+      "loss": 0.0001,
+      "step": 318
+    },
+    {
+      "epoch": 0.02055809757040665,
+      "grad_norm": 0.04965355559724808,
+      "learning_rate": 7.846680192182317e-06,
+      "loss": 0.0002,
+      "step": 319
+    },
+    {
+      "epoch": 0.020622543017335824,
+      "grad_norm": 0.027329077999220016,
+      "learning_rate": 7.850940115060665e-06,
+      "loss": 0.0002,
+      "step": 320
+    },
+    {
+      "epoch": 0.020686988464265,
+      "grad_norm": 0.012715309583353273,
+      "learning_rate": 7.855186746426372e-06,
+      "loss": 0.0,
+      "step": 321
+    },
+    {
+      "epoch": 0.020751433911194175,
+      "grad_norm": 0.022217729569863346,
+      "learning_rate": 7.859420168964014e-06,
+      "loss": 0.0001,
+      "step": 322
+    },
+    {
+      "epoch": 0.02081587935812335,
+      "grad_norm": 0.009745268168007587,
+      "learning_rate": 7.863640464588996e-06,
+      "loss": 0.0,
+      "step": 323
+    },
+    {
+      "epoch": 0.020880324805052523,
+      "grad_norm": 0.3033755741514754,
+      "learning_rate": 7.86784771445708e-06,
+      "loss": 0.001,
+      "step": 324
+    },
+    {
+      "epoch": 0.020944770251981697,
+      "grad_norm": 0.012576282814093155,
+      "learning_rate": 7.87204199897374e-06,
+      "loss": 0.0,
+      "step": 325
+    },
+    {
+      "epoch": 0.02100921569891087,
+      "grad_norm": 0.0025267722044404915,
+      "learning_rate": 7.87622339780338e-06,
+      "loss": 0.0,
+      "step": 326
+    },
+    {
+      "epoch": 0.021073661145840045,
+      "grad_norm": 0.06525056075694663,
+      "learning_rate": 7.88039198987842e-06,
+      "loss": 0.0001,
+      "step": 327
+    },
+    {
+      "epoch": 0.02113810659276922,
+      "grad_norm": 0.010063583865905463,
+      "learning_rate": 7.88454785340823e-06,
+      "loss": 0.0,
+      "step": 328
+    },
+    {
+      "epoch": 0.021202552039698397,
+      "grad_norm": 1.3302956737899603,
+      "learning_rate": 7.888691065887938e-06,
+      "loss": 0.0081,
+      "step": 329
+    },
+    {
+      "epoch": 0.02126699748662757,
+      "grad_norm": 0.17322090259007986,
+      "learning_rate": 7.892821704107102e-06,
+      "loss": 0.0023,
+      "step": 330
+    },
+    {
+      "epoch": 0.021331442933556745,
+      "grad_norm": 0.011463568326494302,
+      "learning_rate": 7.896939844158243e-06,
+      "loss": 0.0,
+      "step": 331
+    },
+    {
+      "epoch": 0.02139588838048592,
+      "grad_norm": 0.04771365515172803,
+      "learning_rate": 7.901045561445263e-06,
+      "loss": 0.0001,
+      "step": 332
+    },
+    {
+      "epoch": 0.021460333827415093,
+      "grad_norm": 0.007986473007665173,
+      "learning_rate": 7.905138930691726e-06,
+      "loss": 0.0,
+      "step": 333
+    },
+    {
+      "epoch": 0.021524779274344267,
+      "grad_norm": 0.13398699493738292,
+      "learning_rate": 7.909220025949012e-06,
+      "loss": 0.0001,
+      "step": 334
+    },
+    {
+      "epoch": 0.02158922472127344,
+      "grad_norm": 0.17768507053513713,
+      "learning_rate": 7.913288920604367e-06,
+      "loss": 0.0024,
+      "step": 335
+    },
+    {
+      "epoch": 0.021653670168202615,
+      "grad_norm": 1.1020287142645597,
+      "learning_rate": 7.917345687388818e-06,
+      "loss": 0.0027,
+      "step": 336
+    },
+    {
+      "epoch": 0.021718115615131792,
+      "grad_norm": 0.005485560200914321,
+      "learning_rate": 7.921390398384963e-06,
+      "loss": 0.0,
+      "step": 337
+    },
+    {
+      "epoch": 0.021782561062060966,
+      "grad_norm": 0.39644821828960675,
+      "learning_rate": 7.925423125034672e-06,
+      "loss": 0.0025,
+      "step": 338
+    },
+    {
+      "epoch": 0.02184700650899014,
+      "grad_norm": 0.1476993275933789,
+      "learning_rate": 7.92944393814664e-06,
+      "loss": 0.0006,
+      "step": 339
+    },
+    {
+      "epoch": 0.021911451955919314,
+      "grad_norm": 0.003948748132867952,
+      "learning_rate": 7.933452907903869e-06,
+      "loss": 0.0,
+      "step": 340
+    },
+    {
+      "epoch": 0.02197589740284849,
+      "grad_norm": 0.1684808299627317,
+      "learning_rate": 7.937450103870998e-06,
+      "loss": 0.0007,
+      "step": 341
+    },
+    {
+      "epoch": 0.022040342849777662,
+      "grad_norm": 0.03526063990809892,
+      "learning_rate": 7.941435595001557e-06,
+      "loss": 0.0001,
+      "step": 342
+    },
+    {
+      "epoch": 0.022104788296706836,
+      "grad_norm": 0.00300194528628094,
+      "learning_rate": 7.945409449645105e-06,
+      "loss": 0.0,
+      "step": 343
+    },
+    {
+      "epoch": 0.02216923374363601,
+      "grad_norm": 0.0015934897607704204,
+      "learning_rate": 7.94937173555426e-06,
+      "loss": 0.0,
+      "step": 344
+    },
+    {
+      "epoch": 0.022233679190565188,
+      "grad_norm": 0.0058229976179194655,
+      "learning_rate": 7.953322519891626e-06,
+      "loss": 0.0,
+      "step": 345
+    },
+    {
+      "epoch": 0.022298124637494362,
+      "grad_norm": 0.04014894770212523,
+      "learning_rate": 7.957261869236626e-06,
+      "loss": 0.0,
+      "step": 346
+    },
+    {
+      "epoch": 0.022362570084423536,
+      "grad_norm": 0.3194126538626638,
+      "learning_rate": 7.961189849592243e-06,
+      "loss": 0.0056,
+      "step": 347
+    },
+    {
+      "epoch": 0.02242701553135271,
+      "grad_norm": 0.0064705525401818125,
+      "learning_rate": 7.965106526391645e-06,
+      "loss": 0.0001,
+      "step": 348
+    },
+    {
+      "epoch": 0.022491460978281884,
+      "grad_norm": 3.5872106349028567,
+      "learning_rate": 7.969011964504728e-06,
+      "loss": 0.0476,
+      "step": 349
+    },
+    {
+      "epoch": 0.022555906425211058,
+      "grad_norm": 0.21691632631488653,
+      "learning_rate": 7.972906228244563e-06,
+      "loss": 0.0006,
+      "step": 350
+    },
+    {
+      "epoch": 0.022620351872140232,
+      "grad_norm": 0.013143918484912772,
+      "learning_rate": 7.976789381373757e-06,
+      "loss": 0.0,
+      "step": 351
+    },
+    {
+      "epoch": 0.02268479731906941,
+      "grad_norm": 0.02501359449813824,
+      "learning_rate": 7.980661487110704e-06,
+      "loss": 0.0001,
+      "step": 352
+    },
+    {
+      "epoch": 0.022749242765998583,
+      "grad_norm": 0.0017974908439855096,
+      "learning_rate": 7.98452260813577e-06,
+      "loss": 0.0,
+      "step": 353
+    },
+    {
+      "epoch": 0.022813688212927757,
+      "grad_norm": 0.023813737078225865,
+      "learning_rate": 7.988372806597378e-06,
+      "loss": 0.0,
+      "step": 354
+    },
+    {
+      "epoch": 0.02287813365985693,
+      "grad_norm": 0.36353022749573466,
+      "learning_rate": 7.992212144118013e-06,
+      "loss": 0.0034,
+      "step": 355
+    },
+    {
+      "epoch": 0.022942579106786105,
+      "grad_norm": 0.03440102538916525,
+      "learning_rate": 7.996040681800137e-06,
+      "loss": 0.0001,
+      "step": 356
+    },
+    {
+      "epoch": 0.02300702455371528,
+      "grad_norm": 0.39249523673234826,
+      "learning_rate": 7.99985848023202e-06,
+      "loss": 0.0033,
+      "step": 357
+    },
+    {
+      "epoch": 0.023071470000644453,
+      "grad_norm": 0.015484064838049798,
+      "learning_rate": 8.003665599493508e-06,
+      "loss": 0.0002,
+      "step": 358
+    },
+    {
+      "epoch": 0.023135915447573627,
+      "grad_norm": 0.05831849892882333,
+      "learning_rate": 8.007462099161673e-06,
+      "loss": 0.0002,
+      "step": 359
+    },
+    {
+      "epoch": 0.023200360894502805,
+      "grad_norm": 0.030189192169840014,
+      "learning_rate": 8.01124803831643e-06,
+      "loss": 0.0,
+      "step": 360
+    },
+    {
+      "epoch": 0.02326480634143198,
+      "grad_norm": 0.2871898356727702,
+      "learning_rate": 8.015023475546033e-06,
+      "loss": 0.0009,
+      "step": 361
+    },
+    {
+      "epoch": 0.023329251788361153,
+      "grad_norm": 0.0028194829163485577,
+      "learning_rate": 8.018788468952538e-06,
+      "loss": 0.0,
+      "step": 362
+    },
+    {
+      "epoch": 0.023393697235290327,
+      "grad_norm": 0.13778555057274436,
+      "learning_rate": 8.022543076157141e-06,
+      "loss": 0.0039,
+      "step": 363
+    },
+    {
+      "epoch": 0.0234581426822195,
+      "grad_norm": 0.021112274440135134,
+      "learning_rate": 8.026287354305495e-06,
+      "loss": 0.0001,
+      "step": 364
+    },
+    {
+      "epoch": 0.023522588129148675,
+      "grad_norm": 0.004636861815034096,
+      "learning_rate": 8.030021360072912e-06,
+      "loss": 0.0,
+      "step": 365
+    },
+    {
+      "epoch": 0.02358703357607785,
+      "grad_norm": 0.0028864734672826285,
+      "learning_rate": 8.033745149669513e-06,
+      "loss": 0.0,
+      "step": 366
+    },
+    {
+      "epoch": 0.023651479023007023,
+      "grad_norm": 0.22539105605376755,
+      "learning_rate": 8.037458778845317e-06,
+      "loss": 0.0007,
+      "step": 367
+    },
+    {
+      "epoch": 0.0237159244699362,
+      "grad_norm": 0.7308905296678736,
+      "learning_rate": 8.041162302895227e-06,
+      "loss": 0.0024,
+      "step": 368
+    },
+    {
+      "epoch": 0.023780369916865374,
+      "grad_norm": 0.0055914223650806285,
+      "learning_rate": 8.044855776663993e-06,
+      "loss": 0.0,
+      "step": 369
+    },
+    {
+      "epoch": 0.02384481536379455,
+      "grad_norm": 0.04858513212525571,
+      "learning_rate": 8.048539254551072e-06,
+      "loss": 0.0,
+      "step": 370
+    },
+    {
+      "epoch": 0.023909260810723722,
+      "grad_norm": 0.0032946095109740066,
+      "learning_rate": 8.052212790515438e-06,
+      "loss": 0.0,
+      "step": 371
+    },
+    {
+      "epoch": 0.023973706257652896,
+      "grad_norm": 0.00796063799442098,
+      "learning_rate": 8.055876438080324e-06,
+      "loss": 0.0,
+      "step": 372
+    },
+    {
+      "epoch": 0.02403815170458207,
+      "grad_norm": 0.07456883882132498,
+      "learning_rate": 8.059530250337904e-06,
+      "loss": 0.0015,
+      "step": 373
+    },
+    {
+      "epoch": 0.024102597151511244,
+      "grad_norm": 0.024193773638429922,
+      "learning_rate": 8.063174279953908e-06,
+      "loss": 0.0,
+      "step": 374
+    },
+    {
+      "epoch": 0.024167042598440422,
+      "grad_norm": 0.08240639532182745,
+      "learning_rate": 8.066808579172175e-06,
+      "loss": 0.0014,
+      "step": 375
+    },
+    {
+      "epoch": 0.024231488045369596,
+      "grad_norm": 0.06964492787404344,
+      "learning_rate": 8.070433199819152e-06,
+      "loss": 0.0001,
+      "step": 376
+    },
+    {
+      "epoch": 0.02429593349229877,
+      "grad_norm": 0.13324233929377452,
+      "learning_rate": 8.074048193308322e-06,
+      "loss": 0.0014,
+      "step": 377
+    },
+    {
+      "epoch": 0.024360378939227944,
+      "grad_norm": 0.31884756278114057,
+      "learning_rate": 8.077653610644582e-06,
+      "loss": 0.0025,
+      "step": 378
+    },
+    {
+      "epoch": 0.024424824386157118,
+      "grad_norm": 0.005066331055183986,
+      "learning_rate": 8.081249502428569e-06,
+      "loss": 0.0001,
+      "step": 379
+    },
+    {
+      "epoch": 0.024489269833086292,
+      "grad_norm": 0.11909933234504282,
+      "learning_rate": 8.084835918860906e-06,
+      "loss": 0.0006,
+      "step": 380
+    },
+    {
+      "epoch": 0.024553715280015466,
+      "grad_norm": 0.21011827473200473,
+      "learning_rate": 8.08841290974642e-06,
+      "loss": 0.0004,
+      "step": 381
+    },
+    {
+      "epoch": 0.02461816072694464,
+      "grad_norm": 0.027595556932529592,
+      "learning_rate": 8.091980524498285e-06,
+      "loss": 0.0,
+      "step": 382
+    },
+    {
+      "epoch": 0.024682606173873817,
+      "grad_norm": 0.006354662520049507,
+      "learning_rate": 8.095538812142118e-06,
+      "loss": 0.0,
+      "step": 383
+    },
+    {
+      "epoch": 0.02474705162080299,
+      "grad_norm": 0.013398883993908624,
+      "learning_rate": 8.099087821320032e-06,
+      "loss": 0.0001,
+      "step": 384
+    },
+    {
+      "epoch": 0.024811497067732165,
+      "grad_norm": 0.018757425423641675,
+      "learning_rate": 8.102627600294604e-06,
+      "loss": 0.0,
+      "step": 385
+    },
+    {
+      "epoch": 0.02487594251466134,
+      "grad_norm": 0.003749142883297565,
+      "learning_rate": 8.106158196952835e-06,
+      "loss": 0.0,
+      "step": 386
+    },
+    {
+      "epoch": 0.024940387961590513,
+      "grad_norm": 0.37095629505467437,
+      "learning_rate": 8.109679658810025e-06,
+      "loss": 0.0019,
+      "step": 387
+    },
+    {
+      "epoch": 0.025004833408519687,
+      "grad_norm": 0.049536866135149134,
+      "learning_rate": 8.113192033013613e-06,
+      "loss": 0.0001,
+      "step": 388
+    },
+    {
+      "epoch": 0.02506927885544886,
+      "grad_norm": 0.05720039117501582,
+      "learning_rate": 8.116695366346962e-06,
+      "loss": 0.0002,
+      "step": 389
+    },
+    {
+      "epoch": 0.025133724302378035,
+      "grad_norm": 0.013197769274937824,
+      "learning_rate": 8.120189705233107e-06,
+      "loss": 0.0,
+      "step": 390
+    },
+    {
+      "epoch": 0.025198169749307213,
+      "grad_norm": 1.5365355994845276,
+      "learning_rate": 8.12367509573843e-06,
+      "loss": 0.0082,
+      "step": 391
+    },
+    {
+      "epoch": 0.025262615196236387,
+      "grad_norm": 0.005337843706833008,
+      "learning_rate": 8.127151583576319e-06,
+      "loss": 0.0,
+      "step": 392
+    },
+    {
+      "epoch": 0.02532706064316556,
+      "grad_norm": 0.009108647589636271,
+      "learning_rate": 8.13061921411076e-06,
+      "loss": 0.0001,
+      "step": 393
+    },
+    {
+      "epoch": 0.025391506090094735,
+      "grad_norm": 0.012351875469581566,
+      "learning_rate": 8.134078032359886e-06,
+      "loss": 0.0001,
+      "step": 394
+    },
+    {
+      "epoch": 0.02545595153702391,
+      "grad_norm": 0.0014107577675921742,
+      "learning_rate": 8.137528082999486e-06,
+      "loss": 0.0,
+      "step": 395
+    },
+    {
+      "epoch": 0.025520396983953083,
+      "grad_norm": 0.05295340430722678,
+      "learning_rate": 8.140969410366469e-06,
+      "loss": 0.0001,
+      "step": 396
+    },
+    {
+      "epoch": 0.025584842430882257,
+      "grad_norm": 0.5690354196975049,
+      "learning_rate": 8.144402058462278e-06,
+      "loss": 0.0023,
+      "step": 397
+    },
+    {
+      "epoch": 0.02564928787781143,
+      "grad_norm": 0.03940801697463157,
+      "learning_rate": 8.147826070956273e-06,
+      "loss": 0.0001,
+      "step": 398
+    },
+    {
+      "epoch": 0.02571373332474061,
+      "grad_norm": 0.011530897889634606,
+      "learning_rate": 8.15124149118906e-06,
+      "loss": 0.0001,
+      "step": 399
+    },
+    {
+      "epoch": 0.025778178771669782,
+      "grad_norm": 0.02830996714923142,
+      "learning_rate": 8.154648362175777e-06,
+      "loss": 0.0002,
+      "step": 400
+    },
+    {
+      "epoch": 0.025842624218598956,
+      "grad_norm": 0.00046499532858108194,
+      "learning_rate": 8.15804672660936e-06,
+      "loss": 0.0,
+      "step": 401
+    },
+    {
+      "epoch": 0.02590706966552813,
+      "grad_norm": 0.26561429899763406,
+      "learning_rate": 8.161436626863734e-06,
+      "loss": 0.0004,
+      "step": 402
+    },
+    {
+      "epoch": 0.025971515112457304,
+      "grad_norm": 0.014311267292635437,
+      "learning_rate": 8.164818104996999e-06,
+      "loss": 0.0001,
+      "step": 403
+    },
+    {
+      "epoch": 0.02603596055938648,
+      "grad_norm": 0.033020361512505675,
+      "learning_rate": 8.168191202754557e-06,
+      "loss": 0.0001,
+      "step": 404
+    },
+    {
+      "epoch": 0.026100406006315652,
+      "grad_norm": 0.01233642367780627,
+      "learning_rate": 8.171555961572193e-06,
+      "loss": 0.0001,
+      "step": 405
+    },
+    {
+      "epoch": 0.02616485145324483,
+      "grad_norm": 0.030507303415610476,
+      "learning_rate": 8.174912422579145e-06,
+      "loss": 0.0,
+      "step": 406
+    },
+    {
+      "epoch": 0.026229296900174004,
+      "grad_norm": 0.000771348551102021,
+      "learning_rate": 8.178260626601112e-06,
+      "loss": 0.0,
+      "step": 407
+    },
+    {
+      "epoch": 0.026293742347103178,
+      "grad_norm": 0.04695325270252983,
+      "learning_rate": 8.181600614163234e-06,
+      "loss": 0.0001,
+      "step": 408
+    },
+    {
+      "epoch": 0.026358187794032352,
+      "grad_norm": 0.004360910000568842,
+      "learning_rate": 8.184932425493038e-06,
+      "loss": 0.0,
+      "step": 409
+    },
+    {
+      "epoch": 0.026422633240961526,
+      "grad_norm": 0.0022237242463113796,
+      "learning_rate": 8.188256100523343e-06,
+      "loss": 0.0,
+      "step": 410
+    },
+    {
+      "epoch": 0.0264870786878907,
+      "grad_norm": 0.11268887113016282,
+      "learning_rate": 8.191571678895127e-06,
+      "loss": 0.0018,
+      "step": 411
+    },
+    {
+      "epoch": 0.026551524134819874,
+      "grad_norm": 0.0014461989918299746,
+      "learning_rate": 8.194879199960378e-06,
+      "loss": 0.0,
+      "step": 412
+    },
+    {
+      "epoch": 0.026615969581749048,
+      "grad_norm": 0.03791107941859059,
+      "learning_rate": 8.19817870278488e-06,
+      "loss": 0.0001,
+      "step": 413
+    },
+    {
+      "epoch": 0.026680415028678225,
+      "grad_norm": 0.23998875971638906,
+      "learning_rate": 8.20147022615099e-06,
+      "loss": 0.002,
+      "step": 414
+    },
+    {
+      "epoch": 0.0267448604756074,
+      "grad_norm": 0.012383002073283897,
+      "learning_rate": 8.204753808560375e-06,
+      "loss": 0.0,
+      "step": 415
+    },
+    {
+      "epoch": 0.026809305922536573,
+      "grad_norm": 0.29471501301861663,
+      "learning_rate": 8.208029488236709e-06,
+      "loss": 0.0017,
+      "step": 416
+    },
+    {
+      "epoch": 0.026873751369465747,
+      "grad_norm": 0.20318785025191824,
+      "learning_rate": 8.211297303128342e-06,
+      "loss": 0.0015,
+      "step": 417
+    },
+    {
+      "epoch": 0.02693819681639492,
+      "grad_norm": 0.017450833706922735,
+      "learning_rate": 8.214557290910945e-06,
+      "loss": 0.0,
+      "step": 418
+    },
+    {
+      "epoch": 0.027002642263324095,
+      "grad_norm": 0.012397289560402282,
+      "learning_rate": 8.21780948899011e-06,
+      "loss": 0.0,
+      "step": 419
+    },
+    {
+      "epoch": 0.02706708771025327,
+      "grad_norm": 0.009342866096887397,
+      "learning_rate": 8.22105393450393e-06,
+      "loss": 0.0,
+      "step": 420
+    },
+    {
+      "epoch": 0.027131533157182443,
+      "grad_norm": 0.7392758325675876,
+      "learning_rate": 8.224290664325538e-06,
+      "loss": 0.0034,
+      "step": 421
+    },
+    {
+      "epoch": 0.02719597860411162,
+      "grad_norm": 0.19301211838897894,
+      "learning_rate": 8.227519715065621e-06,
+      "loss": 0.0007,
+      "step": 422
+    },
+    {
+      "epoch": 0.027260424051040795,
+      "grad_norm": 0.04688608763614122,
+      "learning_rate": 8.230741123074915e-06,
+      "loss": 0.0001,
+      "step": 423
+    },
+    {
+      "epoch": 0.02732486949796997,
+      "grad_norm": 0.04107700591487386,
+      "learning_rate": 8.233954924446651e-06,
+      "loss": 0.0005,
+      "step": 424
+    },
+    {
+      "epoch": 0.027389314944899143,
+      "grad_norm": 0.0016944504864118827,
+      "learning_rate": 8.23716115501898e-06,
+      "loss": 0.0,
+      "step": 425
+    },
+    {
+      "epoch": 0.027453760391828317,
+      "grad_norm": 0.14313535363424715,
+      "learning_rate": 8.24035985037738e-06,
+      "loss": 0.002,
+      "step": 426
+    },
+    {
+      "epoch": 0.02751820583875749,
+      "grad_norm": 0.08606880084376932,
+      "learning_rate": 8.243551045857016e-06,
+      "loss": 0.0003,
+      "step": 427
+    },
+    {
+      "epoch": 0.027582651285686665,
+      "grad_norm": 0.17907311266594605,
+      "learning_rate": 8.246734776545088e-06,
+      "loss": 0.0006,
+      "step": 428
+    },
+    {
+      "epoch": 0.02764709673261584,
+      "grad_norm": 0.11154235338706936,
+      "learning_rate": 8.249911077283146e-06,
+      "loss": 0.0002,
+      "step": 429
+    },
+    {
+      "epoch": 0.027711542179545016,
+      "grad_norm": 1.4785572282992487,
+      "learning_rate": 8.253079982669373e-06,
+      "loss": 0.0137,
+      "step": 430
+    },
+    {
+      "epoch": 0.02777598762647419,
+      "grad_norm": 0.012928009033203125,
+      "learning_rate": 8.25624152706085e-06,
+      "loss": 0.0,
+      "step": 431
+    },
+    {
+      "epoch": 0.027840433073403364,
+      "grad_norm": 0.003619215014048827,
+      "learning_rate": 8.259395744575797e-06,
+      "loss": 0.0,
+      "step": 432
+    },
+    {
+      "epoch": 0.02790487852033254,
+      "grad_norm": 0.0014624501580332932,
+      "learning_rate": 8.26254266909576e-06,
+      "loss": 0.0,
+      "step": 433
+    },
+    {
+      "epoch": 0.027969323967261712,
+      "grad_norm": 0.004199093101875544,
+      "learning_rate": 8.265682334267826e-06,
+      "loss": 0.0,
+      "step": 434
+    },
+    {
+      "epoch": 0.028033769414190886,
+      "grad_norm": 0.021314337555709988,
+      "learning_rate": 8.268814773506757e-06,
+      "loss": 0.0001,
+      "step": 435
+    },
+    {
+      "epoch": 0.02809821486112006,
+      "grad_norm": 0.03760360231074014,
+      "learning_rate": 8.271940019997134e-06,
+      "loss": 0.0001,
+      "step": 436
+    },
+    {
+      "epoch": 0.028162660308049238,
+      "grad_norm": 0.005209692266127515,
+      "learning_rate": 8.275058106695467e-06,
+      "loss": 0.0,
+      "step": 437
+    },
+    {
+      "epoch": 0.028227105754978412,
+      "grad_norm": 0.09817250645422441,
+      "learning_rate": 8.278169066332278e-06,
+      "loss": 0.0001,
+      "step": 438
+    },
+    {
+      "epoch": 0.028291551201907586,
+      "grad_norm": 0.28345436264734336,
+      "learning_rate": 8.28127293141416e-06,
+      "loss": 0.0005,
+      "step": 439
+    },
+    {
+      "epoch": 0.02835599664883676,
+      "grad_norm": 0.02705612753805196,
+      "learning_rate": 8.284369734225816e-06,
+      "loss": 0.0,
+      "step": 440
+    },
+    {
+      "epoch": 0.028420442095765934,
+      "grad_norm": 0.07754971769711609,
+      "learning_rate": 8.287459506832084e-06,
+      "loss": 0.0001,
+      "step": 441
+    },
+    {
+      "epoch": 0.028484887542695108,
+      "grad_norm": 0.03146103849257445,
+      "learning_rate": 8.290542281079913e-06,
+      "loss": 0.0,
+      "step": 442
+    },
+    {
+      "epoch": 0.028549332989624282,
+      "grad_norm": 0.1758081044770802,
+      "learning_rate": 8.293618088600338e-06,
+      "loss": 0.0003,
+      "step": 443
+    },
+    {
+      "epoch": 0.028613778436553456,
+      "grad_norm": 0.0029838889004879295,
+      "learning_rate": 8.29668696081044e-06,
+      "loss": 0.0,
+      "step": 444
+    },
+    {
+      "epoch": 0.028678223883482633,
+      "grad_norm": 0.45164235672802683,
+      "learning_rate": 8.299748928915249e-06,
+      "loss": 0.001,
+      "step": 445
+    },
+    {
+      "epoch": 0.028742669330411807,
+      "grad_norm": 0.0033308645237363874,
+      "learning_rate": 8.302804023909669e-06,
+      "loss": 0.0,
+      "step": 446
+    },
+    {
+      "epoch": 0.02880711477734098,
+      "grad_norm": 0.025905336972041268,
+      "learning_rate": 8.305852276580345e-06,
+      "loss": 0.0001,
+      "step": 447
+    },
+    {
+      "epoch": 0.028871560224270156,
+      "grad_norm": 0.0041218454884513304,
+      "learning_rate": 8.308893717507532e-06,
+      "loss": 0.0,
+      "step": 448
+    },
+    {
+      "epoch": 0.02893600567119933,
+      "grad_norm": 0.12197986303560157,
+      "learning_rate": 8.311928377066941e-06,
+      "loss": 0.0004,
+      "step": 449
+    },
+    {
+      "epoch": 0.029000451118128504,
+      "grad_norm": 0.08898476274019052,
+      "learning_rate": 8.314956285431542e-06,
+      "loss": 0.0018,
+      "step": 450
+    },
+    {
+      "epoch": 0.029064896565057678,
+      "grad_norm": 0.09684278236993904,
+      "learning_rate": 8.317977472573382e-06,
+      "loss": 0.0006,
+      "step": 451
+    },
+    {
+      "epoch": 0.02912934201198685,
+      "grad_norm": 0.05075467771608681,
+      "learning_rate": 8.320991968265357e-06,
+      "loss": 0.0005,
+      "step": 452
+    },
+    {
+      "epoch": 0.02919378745891603,
+      "grad_norm": 0.02043870866338334,
+      "learning_rate": 8.323999802082968e-06,
+      "loss": 0.0,
+      "step": 453
+    },
+    {
+      "epoch": 0.029258232905845203,
+      "grad_norm": 0.02380190989209756,
+      "learning_rate": 8.327001003406075e-06,
+      "loss": 0.0,
+      "step": 454
+    },
+    {
+      "epoch": 0.029322678352774377,
+      "grad_norm": 0.19246463681493037,
+      "learning_rate": 8.329995601420607e-06,
+      "loss": 0.0007,
+      "step": 455
+    },
+    {
+      "epoch": 0.02938712379970355,
+      "grad_norm": 0.0018403446570685027,
+      "learning_rate": 8.332983625120273e-06,
+      "loss": 0.0015,
+      "step": 456
+    },
+    {
+      "epoch": 0.029451569246632725,
+      "grad_norm": 0.1614785790568277,
+      "learning_rate": 8.335965103308235e-06,
+      "loss": 0.0003,
+      "step": 457
+    },
+    {
+      "epoch": 0.0295160146935619,
+      "grad_norm": 0.04677324625517432,
+      "learning_rate": 8.338940064598788e-06,
+      "loss": 0.0001,
+      "step": 458
+    },
+    {
+      "epoch": 0.029580460140491073,
+      "grad_norm": 0.023722940740272376,
+      "learning_rate": 8.341908537418999e-06,
+      "loss": 0.0002,
+      "step": 459
+    },
+    {
+      "epoch": 0.029644905587420247,
+      "grad_norm": 0.0071161812529181,
+      "learning_rate": 8.34487055001034e-06,
+      "loss": 0.0,
+      "step": 460
+    },
+    {
+      "epoch": 0.029709351034349425,
+      "grad_norm": 0.01802171434180495,
+      "learning_rate": 8.347826130430298e-06,
+      "loss": 0.0001,
+      "step": 461
+    },
+    {
+      "epoch": 0.0297737964812786,
+      "grad_norm": 0.35340842839961994,
+      "learning_rate": 8.35077530655397e-06,
+      "loss": 0.0026,
+      "step": 462
+    },
+    {
+      "epoch": 0.029838241928207773,
+      "grad_norm": 0.061611551431616104,
+      "learning_rate": 8.353718106075646e-06,
+      "loss": 0.0002,
+      "step": 463
+    },
+    {
+      "epoch": 0.029902687375136947,
+      "grad_norm": 0.04584732913277349,
+      "learning_rate": 8.356654556510358e-06,
+      "loss": 0.0002,
+      "step": 464
+    },
+    {
+      "epoch": 0.02996713282206612,
+      "grad_norm": 0.1855844961474537,
+      "learning_rate": 8.359584685195436e-06,
+      "loss": 0.0005,
+      "step": 465
+    },
+    {
+      "epoch": 0.030031578268995295,
+      "grad_norm": 0.02861397477113445,
+      "learning_rate": 8.362508519292026e-06,
+      "loss": 0.0003,
+      "step": 466
+    },
+    {
+      "epoch": 0.03009602371592447,
+      "grad_norm": 0.04034164553115,
+      "learning_rate": 8.365426085786605e-06,
+      "loss": 0.0001,
+      "step": 467
+    },
+    {
+      "epoch": 0.030160469162853646,
+      "grad_norm": 0.007763013674992482,
+      "learning_rate": 8.368337411492474e-06,
+      "loss": 0.0,
+      "step": 468
+    },
+    {
+      "epoch": 0.03022491460978282,
+      "grad_norm": 0.0920300413920213,
+      "learning_rate": 8.371242523051236e-06,
+      "loss": 0.0001,
+      "step": 469
+    },
+    {
+      "epoch": 0.030289360056711994,
+      "grad_norm": 0.13798576860239228,
+      "learning_rate": 8.374141446934264e-06,
+      "loss": 0.0006,
+      "step": 470
+    },
+    {
+      "epoch": 0.030353805503641168,
+      "grad_norm": 0.013517930261967707,
+      "learning_rate": 8.37703420944414e-06,
+      "loss": 0.0,
+      "step": 471
+    },
+    {
+      "epoch": 0.030418250950570342,
+      "grad_norm": 0.008036030340445328,
+      "learning_rate": 8.379920836716092e-06,
+      "loss": 0.0,
+      "step": 472
+    },
+    {
+      "epoch": 0.030482696397499516,
+      "grad_norm": 0.005974472065890517,
+      "learning_rate": 8.382801354719412e-06,
+      "loss": 0.0,
+      "step": 473
+    },
+    {
+      "epoch": 0.03054714184442869,
+      "grad_norm": 0.0008534880974867222,
+      "learning_rate": 8.385675789258854e-06,
+      "loss": 0.0,
+      "step": 474
+    },
+    {
+      "epoch": 0.030611587291357864,
+      "grad_norm": 0.03526061844791487,
+      "learning_rate": 8.388544165976018e-06,
+      "loss": 0.0001,
+      "step": 475
+    },
+    {
+      "epoch": 0.03067603273828704,
+      "grad_norm": 0.09276391030632433,
+      "learning_rate": 8.391406510350738e-06,
+      "loss": 0.0003,
+      "step": 476
+    },
+    {
+      "epoch": 0.030740478185216216,
+      "grad_norm": 0.006227281517178593,
+      "learning_rate": 8.394262847702416e-06,
+      "loss": 0.0,
+      "step": 477
+    },
+    {
+      "epoch": 0.03080492363214539,
+      "grad_norm": 0.0019055967033663934,
+      "learning_rate": 8.39711320319139e-06,
+      "loss": 0.0,
+      "step": 478
+    },
+    {
+      "epoch": 0.030869369079074564,
+      "grad_norm": 0.008035130758411991,
+      "learning_rate": 8.399957601820245e-06,
+      "loss": 0.0,
+      "step": 479
+    },
+    {
+      "epoch": 0.030933814526003738,
+      "grad_norm": 0.030565513157518043,
+      "learning_rate": 8.402796068435144e-06,
+      "loss": 0.0002,
+      "step": 480
+    },
+    {
+      "epoch": 0.03099825997293291,
+      "grad_norm": 0.0010238980188911407,
+      "learning_rate": 8.405628627727116e-06,
+      "loss": 0.0,
+      "step": 481
+    },
+    {
+      "epoch": 0.031062705419862086,
+      "grad_norm": 0.000852778331990603,
+      "learning_rate": 8.408455304233356e-06,
+      "loss": 0.0,
+      "step": 482
+    },
+    {
+      "epoch": 0.03112715086679126,
+      "grad_norm": 0.009300486191113138,
+      "learning_rate": 8.411276122338493e-06,
+      "loss": 0.0001,
+      "step": 483
+    },
+    {
+      "epoch": 0.031191596313720437,
+      "grad_norm": 0.006314212471481981,
+      "learning_rate": 8.414091106275857e-06,
+      "loss": 0.0001,
+      "step": 484
+    },
+    {
+      "epoch": 0.03125604176064961,
+      "grad_norm": 0.13175345772435,
+      "learning_rate": 8.416900280128724e-06,
+      "loss": 0.0012,
+      "step": 485
+    },
+    {
+      "epoch": 0.03132048720757878,
+      "grad_norm": 0.0034394697556541735,
+      "learning_rate": 8.41970366783156e-06,
+      "loss": 0.0,
+      "step": 486
+    },
+    {
+      "epoch": 0.03138493265450796,
+      "grad_norm": 0.3510331194805854,
+      "learning_rate": 8.422501293171231e-06,
+      "loss": 0.0008,
+      "step": 487
+    },
+    {
+      "epoch": 0.031449378101437137,
+      "grad_norm": 0.0006937932722181743,
+      "learning_rate": 8.42529317978823e-06,
+      "loss": 0.0,
+      "step": 488
+    },
+    {
+      "epoch": 0.03151382354836631,
+      "grad_norm": 0.0395256834106489,
+      "learning_rate": 8.42807935117786e-06,
+      "loss": 0.0002,
+      "step": 489
+    },
+    {
+      "epoch": 0.031578268995295485,
+      "grad_norm": 0.0195828954531926,
+      "learning_rate": 8.430859830691432e-06,
+      "loss": 0.0001,
+      "step": 490
+    },
+    {
+      "epoch": 0.031642714442224655,
+      "grad_norm": 0.2022134980256737,
+      "learning_rate": 8.43363464153744e-06,
+      "loss": 0.0012,
+      "step": 491
+    },
+    {
+      "epoch": 0.03170715988915383,
+      "grad_norm": 0.16900205486186845,
+      "learning_rate": 8.436403806782708e-06,
+      "loss": 0.0004,
+      "step": 492
+    },
+    {
+      "epoch": 0.031771605336083,
+      "grad_norm": 0.0006769181918546838,
+      "learning_rate": 8.43916734935356e-06,
+      "loss": 0.0,
+      "step": 493
+    },
+    {
+      "epoch": 0.03183605078301218,
+      "grad_norm": 0.013904361173820282,
+      "learning_rate": 8.44192529203695e-06,
+      "loss": 0.0001,
+      "step": 494
+    },
+    {
+      "epoch": 0.03190049622994136,
+      "grad_norm": 0.004376416113878596,
+      "learning_rate": 8.444677657481581e-06,
+      "loss": 0.0,
+      "step": 495
+    },
+    {
+      "epoch": 0.03196494167687053,
+      "grad_norm": 0.02817264340354307,
+      "learning_rate": 8.44742446819904e-06,
+      "loss": 0.0003,
+      "step": 496
+    },
+    {
+      "epoch": 0.032029387123799706,
+      "grad_norm": 2.2377603134379584,
+      "learning_rate": 8.450165746564882e-06,
+      "loss": 0.0117,
+      "step": 497
+    },
+    {
+      "epoch": 0.03209383257072888,
+      "grad_norm": 0.002553285998667623,
+      "learning_rate": 8.452901514819742e-06,
+      "loss": 0.0,
+      "step": 498
+    },
+    {
+      "epoch": 0.032158278017658054,
+      "grad_norm": 0.00034067570616675476,
+      "learning_rate": 8.455631795070407e-06,
+      "loss": 0.0,
+      "step": 499
+    },
+    {
+      "epoch": 0.032222723464587225,
+      "grad_norm": 0.35536890409234767,
+      "learning_rate": 8.45835660929089e-06,
+      "loss": 0.0018,
+      "step": 500
+    },
+    {
+      "epoch": 0.0322871689115164,
+      "grad_norm": 0.003508252971943265,
+      "learning_rate": 8.461075979323489e-06,
+      "loss": 0.0,
+      "step": 501
+    },
+    {
+      "epoch": 0.03235161435844557,
+      "grad_norm": 0.7101809909652864,
+      "learning_rate": 8.463789926879845e-06,
+      "loss": 0.0045,
+      "step": 502
+    },
+    {
+      "epoch": 0.03241605980537475,
+      "grad_norm": 0.0013628442685876426,
+      "learning_rate": 8.466498473541972e-06,
+      "loss": 0.0,
+      "step": 503
+    },
+    {
+      "epoch": 0.03248050525230393,
+      "grad_norm": 0.34192186522539375,
+      "learning_rate": 8.469201640763297e-06,
+      "loss": 0.0016,
+      "step": 504
+    },
+    {
+      "epoch": 0.0325449506992331,
+      "grad_norm": 0.0034762216357610365,
+      "learning_rate": 8.471899449869671e-06,
+      "loss": 0.0,
+      "step": 505
+    },
+    {
+      "epoch": 0.032609396146162276,
+      "grad_norm": 0.12617818828412725,
+      "learning_rate": 8.474591922060378e-06,
+      "loss": 0.0004,
+      "step": 506
+    },
+    {
+      "epoch": 0.032673841593091446,
+      "grad_norm": 0.004196338433318039,
+      "learning_rate": 8.47727907840915e-06,
+      "loss": 0.0,
+      "step": 507
+    },
+    {
+      "epoch": 0.032738287040020624,
+      "grad_norm": 0.00876693332198306,
+      "learning_rate": 8.479960939865135e-06,
+      "loss": 0.0001,
+      "step": 508
+    },
+    {
+      "epoch": 0.032802732486949794,
+      "grad_norm": 0.0004212675822051207,
+      "learning_rate": 8.482637527253888e-06,
+      "loss": 0.0,
+      "step": 509
+    },
+    {
+      "epoch": 0.03286717793387897,
+      "grad_norm": 0.004896906167491933,
+      "learning_rate": 8.485308861278346e-06,
+      "loss": 0.0,
+      "step": 510
+    },
+    {
+      "epoch": 0.03293162338080815,
+      "grad_norm": 0.0006167679203622598,
+      "learning_rate": 8.487974962519779e-06,
+      "loss": 0.0,
+      "step": 511
+    },
+    {
+      "epoch": 0.03299606882773732,
+      "grad_norm": 0.25578066656271004,
+      "learning_rate": 8.490635851438747e-06,
+      "loss": 0.0011,
+      "step": 512
+    },
+    {
+      "epoch": 0.0330605142746665,
+      "grad_norm": 0.016222739160781905,
+      "learning_rate": 8.493291548376036e-06,
+      "loss": 0.0001,
+      "step": 513
+    },
+    {
+      "epoch": 0.03312495972159567,
+      "grad_norm": 0.00399666735491616,
+      "learning_rate": 8.495942073553605e-06,
+      "loss": 0.0,
+      "step": 514
+    },
+    {
+      "epoch": 0.033189405168524845,
+      "grad_norm": 0.9266883816050113,
+      "learning_rate": 8.498587447075492e-06,
+      "loss": 0.0033,
+      "step": 515
+    },
+    {
+      "epoch": 0.033253850615454016,
+      "grad_norm": 0.0028993043429181614,
+      "learning_rate": 8.50122768892874e-06,
+      "loss": 0.0,
+      "step": 516
+    },
+    {
+      "epoch": 0.03331829606238319,
+      "grad_norm": 0.27063704065226896,
+      "learning_rate": 8.503862818984304e-06,
+      "loss": 0.0021,
+      "step": 517
+    },
+    {
+      "epoch": 0.033382741509312364,
+      "grad_norm": 0.010379239929541071,
+      "learning_rate": 8.50649285699794e-06,
+      "loss": 0.0,
+      "step": 518
+    },
+    {
+      "epoch": 0.03344718695624154,
+      "grad_norm": 0.35729035839734724,
+      "learning_rate": 8.509117822611103e-06,
+      "loss": 0.0023,
+      "step": 519
+    },
+    {
+      "epoch": 0.03351163240317072,
+      "grad_norm": 0.30007881033376443,
+      "learning_rate": 8.511737735351822e-06,
+      "loss": 0.0011,
+      "step": 520
+    },
+    {
+      "epoch": 0.03357607785009989,
+      "grad_norm": 0.12294530839587339,
+      "learning_rate": 8.514352614635567e-06,
+      "loss": 0.0001,
+      "step": 521
+    },
+    {
+      "epoch": 0.03364052329702907,
+      "grad_norm": 0.11914525257566484,
+      "learning_rate": 8.516962479766123e-06,
+      "loss": 0.0002,
+      "step": 522
+    },
+    {
+      "epoch": 0.03370496874395824,
+      "grad_norm": 0.03495502834888825,
+      "learning_rate": 8.51956734993644e-06,
+      "loss": 0.0001,
+      "step": 523
+    },
+    {
+      "epoch": 0.033769414190887415,
+      "grad_norm": 0.34509190983321814,
+      "learning_rate": 8.522167244229476e-06,
+      "loss": 0.0104,
+      "step": 524
+    },
+    {
+      "epoch": 0.033833859637816585,
+      "grad_norm": 0.2735239028658671,
+      "learning_rate": 8.524762181619042e-06,
+      "loss": 0.0005,
+      "step": 525
+    },
+    {
+      "epoch": 0.03389830508474576,
+      "grad_norm": 0.4540046998770102,
+      "learning_rate": 8.527352180970633e-06,
+      "loss": 0.0051,
+      "step": 526
+    },
+    {
+      "epoch": 0.03396275053167494,
+      "grad_norm": 0.016232761018629926,
+      "learning_rate": 8.529937261042241e-06,
+      "loss": 0.0,
+      "step": 527
+    },
+    {
+      "epoch": 0.03402719597860411,
+      "grad_norm": 0.08212446411675045,
+      "learning_rate": 8.532517440485183e-06,
+      "loss": 0.0001,
+      "step": 528
+    },
+    {
+      "epoch": 0.03409164142553329,
+      "grad_norm": 0.077195666137957,
+      "learning_rate": 8.535092737844902e-06,
+      "loss": 0.0001,
+      "step": 529
+    },
+    {
+      "epoch": 0.03415608687246246,
+      "grad_norm": 0.3339083512543114,
+      "learning_rate": 8.537663171561763e-06,
+      "loss": 0.0034,
+      "step": 530
+    },
+    {
+      "epoch": 0.034220532319391636,
+      "grad_norm": 0.06010152089013261,
+      "learning_rate": 8.540228759971857e-06,
+      "loss": 0.0001,
+      "step": 531
+    },
+    {
+      "epoch": 0.03428497776632081,
+      "grad_norm": 0.10862170243882367,
+      "learning_rate": 8.542789521307773e-06,
+      "loss": 0.0001,
+      "step": 532
+    },
+    {
+      "epoch": 0.034349423213249984,
+      "grad_norm": 0.0018621187038332806,
+      "learning_rate": 8.545345473699385e-06,
+      "loss": 0.0,
+      "step": 533
+    },
+    {
+      "epoch": 0.03441386866017916,
+      "grad_norm": 0.5433106478048796,
+      "learning_rate": 8.547896635174616e-06,
+      "loss": 0.0025,
+      "step": 534
+    },
+    {
+      "epoch": 0.03447831410710833,
+      "grad_norm": 0.02162303799301777,
+      "learning_rate": 8.550443023660201e-06,
+      "loss": 0.0,
+      "step": 535
+    },
+    {
+      "epoch": 0.03454275955403751,
+      "grad_norm": 0.16319136153064215,
+      "learning_rate": 8.55298465698245e-06,
+      "loss": 0.0017,
+      "step": 536
+    },
+    {
+      "epoch": 0.03460720500096668,
+      "grad_norm": 0.03575839023267142,
+      "learning_rate": 8.555521552867987e-06,
+      "loss": 0.0,
+      "step": 537
+    },
+    {
+      "epoch": 0.03467165044789586,
+      "grad_norm": 0.21404037256804487,
+      "learning_rate": 8.558053728944501e-06,
+      "loss": 0.0002,
+      "step": 538
+    },
+    {
+      "epoch": 0.03473609589482503,
+      "grad_norm": 0.007927826619897955,
+      "learning_rate": 8.560581202741473e-06,
+      "loss": 0.0,
+      "step": 539
+    },
+    {
+      "epoch": 0.034800541341754206,
+      "grad_norm": 0.0006026644939430921,
+      "learning_rate": 8.563103991690909e-06,
+      "loss": 0.0,
+      "step": 540
+    },
+    {
+      "epoch": 0.034864986788683376,
+      "grad_norm": 0.001941445168647103,
+      "learning_rate": 8.565622113128059e-06,
+      "loss": 0.0,
+      "step": 541
+    },
+    {
+      "epoch": 0.034929432235612554,
+      "grad_norm": 0.048530455674446166,
+      "learning_rate": 8.568135584292137e-06,
+      "loss": 0.0007,
+      "step": 542
+    },
+    {
+      "epoch": 0.03499387768254173,
+      "grad_norm": 0.0007215121741700993,
+      "learning_rate": 8.570644422327015e-06,
+      "loss": 0.0,
+      "step": 543
+    },
+    {
+      "epoch": 0.0350583231294709,
+      "grad_norm": 0.0026185945721582315,
+      "learning_rate": 8.57314864428195e-06,
+      "loss": 0.0,
+      "step": 544
+    },
+    {
+      "epoch": 0.03512276857640008,
+      "grad_norm": 0.3393699807556369,
+      "learning_rate": 8.575648267112246e-06,
+      "loss": 0.0049,
+      "step": 545
+    },
+    {
+      "epoch": 0.03518721402332925,
+      "grad_norm": 0.19151880886172956,
+      "learning_rate": 8.578143307679974e-06,
+      "loss": 0.0008,
+      "step": 546
+    },
+    {
+      "epoch": 0.03525165947025843,
+      "grad_norm": 0.04460153439407813,
+      "learning_rate": 8.580633782754635e-06,
+      "loss": 0.0005,
+      "step": 547
+    },
+    {
+      "epoch": 0.0353161049171876,
+      "grad_norm": 0.271443576628783,
+      "learning_rate": 8.583119709013842e-06,
+      "loss": 0.0013,
+      "step": 548
+    },
+    {
+      "epoch": 0.035380550364116775,
+      "grad_norm": 0.0041845092527643,
+      "learning_rate": 8.585601103043993e-06,
+      "loss": 0.0,
+      "step": 549
+    },
+    {
+      "epoch": 0.03544499581104595,
+      "grad_norm": 0.007688605143187043,
+      "learning_rate": 8.58807798134093e-06,
+      "loss": 0.0,
+      "step": 550
+    },
+    {
+      "epoch": 0.03550944125797512,
+      "grad_norm": 0.0011524979515473482,
+      "learning_rate": 8.5905503603106e-06,
+      "loss": 0.0,
+      "step": 551
+    },
+    {
+      "epoch": 0.0355738867049043,
+      "grad_norm": 0.004469706196756287,
+      "learning_rate": 8.593018256269704e-06,
+      "loss": 0.0,
+      "step": 552
+    },
+    {
+      "epoch": 0.03563833215183347,
+      "grad_norm": 0.0017258531883411576,
+      "learning_rate": 8.595481685446355e-06,
+      "loss": 0.0,
+      "step": 553
+    },
+    {
+      "epoch": 0.03570277759876265,
+      "grad_norm": 0.011111525603690426,
+      "learning_rate": 8.597940663980695e-06,
+      "loss": 0.0,
+      "step": 554
+    },
+    {
+      "epoch": 0.03576722304569182,
+      "grad_norm": 0.0024214964647003984,
+      "learning_rate": 8.600395207925551e-06,
+      "loss": 0.0,
+      "step": 555
+    },
+    {
+      "epoch": 0.035831668492621,
+      "grad_norm": 0.002982207597299224,
+      "learning_rate": 8.602845333247056e-06,
+      "loss": 0.0,
+      "step": 556
+    },
+    {
+      "epoch": 0.035896113939550174,
+      "grad_norm": 0.0003296500930890877,
+      "learning_rate": 8.605291055825273e-06,
+      "loss": 0.0,
+      "step": 557
+    },
+    {
+      "epoch": 0.035960559386479345,
+      "grad_norm": 0.02805379203644968,
+      "learning_rate": 8.607732391454803e-06,
+      "loss": 0.0,
+      "step": 558
+    },
+    {
+      "epoch": 0.03602500483340852,
+      "grad_norm": 0.007168128548352624,
+      "learning_rate": 8.610169355845417e-06,
+      "loss": 0.0,
+      "step": 559
+    },
+    {
+      "epoch": 0.03608945028033769,
+      "grad_norm": 0.008010737554365606,
+      "learning_rate": 8.612601964622646e-06,
+      "loss": 0.0,
+      "step": 560
+    },
+    {
+      "epoch": 0.03615389572726687,
+      "grad_norm": 0.006762966873986022,
+      "learning_rate": 8.615030233328387e-06,
+      "loss": 0.0001,
+      "step": 561
+    },
+    {
+      "epoch": 0.03621834117419604,
+      "grad_norm": 0.00104232572711473,
+      "learning_rate": 8.617454177421499e-06,
+      "loss": 0.0,
+      "step": 562
+    },
+    {
+      "epoch": 0.03628278662112522,
+      "grad_norm": 2.3966805549541053,
+      "learning_rate": 8.619873812278397e-06,
+      "loss": 0.0212,
+      "step": 563
+    },
+    {
+      "epoch": 0.03634723206805439,
+      "grad_norm": 0.002023202811533298,
+      "learning_rate": 8.622289153193632e-06,
+      "loss": 0.0,
+      "step": 564
+    },
+    {
+      "epoch": 0.036411677514983566,
+      "grad_norm": 0.08835435758202559,
+      "learning_rate": 8.624700215380469e-06,
+      "loss": 0.0,
+      "step": 565
+    },
+    {
+      "epoch": 0.036476122961912744,
+      "grad_norm": 0.008966135734965502,
+      "learning_rate": 8.62710701397147e-06,
+      "loss": 0.0,
+      "step": 566
+    },
+    {
+      "epoch": 0.036540568408841914,
+      "grad_norm": 0.3744501413902579,
+      "learning_rate": 8.62950956401906e-06,
+      "loss": 0.0006,
+      "step": 567
+    },
+    {
+      "epoch": 0.03660501385577109,
+      "grad_norm": 0.074906811338477,
+      "learning_rate": 8.631907880496095e-06,
+      "loss": 0.0002,
+      "step": 568
+    },
+    {
+      "epoch": 0.03666945930270026,
+      "grad_norm": 0.0002814712078328361,
+      "learning_rate": 8.634301978296409e-06,
+      "loss": 0.0,
+      "step": 569
+    },
+    {
+      "epoch": 0.03673390474962944,
+      "grad_norm": 0.0007091569765689754,
+      "learning_rate": 8.636691872235386e-06,
+      "loss": 0.0,
+      "step": 570
+    },
+    {
+      "epoch": 0.03679835019655861,
+      "grad_norm": 0.15370993098858876,
+      "learning_rate": 8.639077577050498e-06,
+      "loss": 0.0012,
+      "step": 571
+    },
+    {
+      "epoch": 0.03686279564348779,
+      "grad_norm": 0.09234985727520918,
+      "learning_rate": 8.64145910740186e-06,
+      "loss": 0.002,
+      "step": 572
+    },
+    {
+      "epoch": 0.036927241090416965,
+      "grad_norm": 0.19519613614308395,
+      "learning_rate": 8.643836477872763e-06,
+      "loss": 0.0006,
+      "step": 573
+    },
+    {
+      "epoch": 0.036991686537346136,
+      "grad_norm": 0.13928936096695185,
+      "learning_rate": 8.64620970297021e-06,
+      "loss": 0.0021,
+      "step": 574
+    },
+    {
+      "epoch": 0.03705613198427531,
+      "grad_norm": 0.013526277478610292,
+      "learning_rate": 8.648578797125451e-06,
+      "loss": 0.0,
+      "step": 575
+    },
+    {
+      "epoch": 0.037120577431204484,
+      "grad_norm": 0.036371678731640995,
+      "learning_rate": 8.65094377469451e-06,
+      "loss": 0.0001,
+      "step": 576
+    },
+    {
+      "epoch": 0.03718502287813366,
+      "grad_norm": 0.011606647574347847,
+      "learning_rate": 8.6533046499587e-06,
+      "loss": 0.0,
+      "step": 577
+    },
+    {
+      "epoch": 0.03724946832506283,
+      "grad_norm": 0.00018151496333934717,
+      "learning_rate": 8.655661437125153e-06,
+      "loss": 0.0,
+      "step": 578
+    },
+    {
+      "epoch": 0.03731391377199201,
+      "grad_norm": 0.017874406288697416,
+      "learning_rate": 8.658014150327315e-06,
+      "loss": 0.0002,
+      "step": 579
+    },
+    {
+      "epoch": 0.03737835921892119,
+      "grad_norm": 0.0006531733163911647,
+      "learning_rate": 8.66036280362547e-06,
+      "loss": 0.0,
+      "step": 580
+    },
+    {
+      "epoch": 0.03744280466585036,
+      "grad_norm": 0.0015501879208044541,
+      "learning_rate": 8.662707411007244e-06,
+      "loss": 0.0,
+      "step": 581
+    },
+    {
+      "epoch": 0.037507250112779535,
+      "grad_norm": 0.0014950575720748577,
+      "learning_rate": 8.665047986388091e-06,
+      "loss": 0.0,
+      "step": 582
+    },
+    {
+      "epoch": 0.037571695559708705,
+      "grad_norm": 0.00018370767475683753,
+      "learning_rate": 8.667384543611802e-06,
+      "loss": 0.0,
+      "step": 583
+    },
+    {
+      "epoch": 0.03763614100663788,
+      "grad_norm": 0.5265942085748113,
+      "learning_rate": 8.669717096450994e-06,
+      "loss": 0.0023,
+      "step": 584
+    },
+    {
+      "epoch": 0.03770058645356705,
+      "grad_norm": 0.0009574933177824663,
+      "learning_rate": 8.672045658607586e-06,
+      "loss": 0.0,
+      "step": 585
+    },
+    {
+      "epoch": 0.03776503190049623,
+      "grad_norm": 0.6674461997778207,
+      "learning_rate": 8.674370243713298e-06,
+      "loss": 0.0003,
+      "step": 586
+    },
+    {
+      "epoch": 0.0378294773474254,
+      "grad_norm": 0.7010925375594455,
+      "learning_rate": 8.676690865330125e-06,
+      "loss": 0.0029,
+      "step": 587
+    },
+    {
+      "epoch": 0.03789392279435458,
+      "grad_norm": 0.00532852021202004,
+      "learning_rate": 8.6790075369508e-06,
+      "loss": 0.0,
+      "step": 588
+    },
+    {
+      "epoch": 0.037958368241283756,
+      "grad_norm": 0.005449903198530545,
+      "learning_rate": 8.68132027199928e-06,
+      "loss": 0.0,
+      "step": 589
+    },
+    {
+      "epoch": 0.03802281368821293,
+      "grad_norm": 0.10746727764211438,
+      "learning_rate": 8.683629083831205e-06,
+      "loss": 0.0002,
+      "step": 590
+    },
+    {
+      "epoch": 0.038087259135142104,
+      "grad_norm": 0.016064785820550875,
+      "learning_rate": 8.685933985734367e-06,
+      "loss": 0.0,
+      "step": 591
+    },
+    {
+      "epoch": 0.038151704582071275,
+      "grad_norm": 0.0029501290312424723,
+      "learning_rate": 8.688234990929155e-06,
+      "loss": 0.0,
+      "step": 592
+    },
+    {
+      "epoch": 0.03821615002900045,
+      "grad_norm": 0.0013237907721366717,
+      "learning_rate": 8.690532112569025e-06,
+      "loss": 0.0,
+      "step": 593
+    },
+    {
+      "epoch": 0.03828059547592962,
+      "grad_norm": 0.2628805365475962,
+      "learning_rate": 8.692825363740948e-06,
+      "loss": 0.0008,
+      "step": 594
+    },
+    {
+      "epoch": 0.0383450409228588,
+      "grad_norm": 0.002112624002434478,
+      "learning_rate": 8.69511475746585e-06,
+      "loss": 0.0,
+      "step": 595
+    },
+    {
+      "epoch": 0.03840948636978798,
+      "grad_norm": 0.14168079395771904,
+      "learning_rate": 8.69740030669906e-06,
+      "loss": 0.0013,
+      "step": 596
+    },
+    {
+      "epoch": 0.03847393181671715,
+      "grad_norm": 0.21190947426606038,
+      "learning_rate": 8.699682024330754e-06,
+      "loss": 0.003,
+      "step": 597
+    },
+    {
+      "epoch": 0.038538377263646326,
+      "grad_norm": 0.003101969641557718,
+      "learning_rate": 8.701959923186383e-06,
+      "loss": 0.0,
+      "step": 598
+    },
+    {
+      "epoch": 0.038602822710575496,
+      "grad_norm": 0.009290021857474596,
+      "learning_rate": 8.704234016027115e-06,
+      "loss": 0.0,
+      "step": 599
+    },
+    {
+      "epoch": 0.038667268157504674,
+      "grad_norm": 0.004114591117807323,
+      "learning_rate": 8.706504315550256e-06,
+      "loss": 0.0,
+      "step": 600
+    },
+    {
+      "epoch": 0.038731713604433844,
+      "grad_norm": 0.16130475782513984,
+      "learning_rate": 8.708770834389678e-06,
+      "loss": 0.002,
+      "step": 601
+    },
+    {
+      "epoch": 0.03879615905136302,
+      "grad_norm": 0.003502346821683403,
+      "learning_rate": 8.71103358511624e-06,
+      "loss": 0.0,
+      "step": 602
+    },
+    {
+      "epoch": 0.03886060449829219,
+      "grad_norm": 0.002125771925744141,
+      "learning_rate": 8.713292580238214e-06,
+      "loss": 0.0,
+      "step": 603
+    },
+    {
+      "epoch": 0.03892504994522137,
+      "grad_norm": 0.00033336581023886793,
+      "learning_rate": 8.715547832201683e-06,
+      "loss": 0.0,
+      "step": 604
+    },
+    {
+      "epoch": 0.03898949539215055,
+      "grad_norm": 0.032290644222526255,
+      "learning_rate": 8.71779935339097e-06,
+      "loss": 0.0001,
+      "step": 605
+    },
+    {
+      "epoch": 0.03905394083907972,
+      "grad_norm": 14.158548400095667,
+      "learning_rate": 8.720047156129036e-06,
+      "loss": 0.0439,
+      "step": 606
+    },
+    {
+      "epoch": 0.039118386286008895,
+      "grad_norm": 0.12495020157562878,
+      "learning_rate": 8.722291252677892e-06,
+      "loss": 0.0006,
+      "step": 607
+    },
+    {
+      "epoch": 0.039182831732938066,
+      "grad_norm": 0.0017411839254716896,
+      "learning_rate": 8.724531655238988e-06,
+      "loss": 0.0,
+      "step": 608
+    },
+    {
+      "epoch": 0.03924727717986724,
+      "grad_norm": 0.0005025559677200699,
+      "learning_rate": 8.726768375953625e-06,
+      "loss": 0.0,
+      "step": 609
+    },
+    {
+      "epoch": 0.039311722626796414,
+      "grad_norm": 0.00043016709712481065,
+      "learning_rate": 8.729001426903342e-06,
+      "loss": 0.0,
+      "step": 610
+    },
+    {
+      "epoch": 0.03937616807372559,
+      "grad_norm": 0.011355258759439064,
+      "learning_rate": 8.731230820110307e-06,
+      "loss": 0.0,
+      "step": 611
+    },
+    {
+      "epoch": 0.03944061352065477,
+      "grad_norm": 0.018387622737482987,
+      "learning_rate": 8.733456567537714e-06,
+      "loss": 0.0002,
+      "step": 612
+    },
+    {
+      "epoch": 0.03950505896758394,
+      "grad_norm": 0.012411244035460034,
+      "learning_rate": 8.735678681090153e-06,
+      "loss": 0.0,
+      "step": 613
+    },
+    {
+      "epoch": 0.03956950441451312,
+      "grad_norm": 0.01917335761000915,
+      "learning_rate": 8.737897172614007e-06,
+      "loss": 0.0001,
+      "step": 614
+    },
+    {
+      "epoch": 0.03963394986144229,
+      "grad_norm": 0.0012099117817259695,
+      "learning_rate": 8.74011205389782e-06,
+      "loss": 0.0,
+      "step": 615
+    },
+    {
+      "epoch": 0.039698395308371465,
+      "grad_norm": 0.011013451182717668,
+      "learning_rate": 8.742323336672685e-06,
+      "loss": 0.0,
+      "step": 616
+    },
+    {
+      "epoch": 0.039762840755300635,
+      "grad_norm": 0.0009947618007444038,
+      "learning_rate": 8.744531032612602e-06,
+      "loss": 0.0,
+      "step": 617
+    },
+    {
+      "epoch": 0.03982728620222981,
+      "grad_norm": 0.007576749125775888,
+      "learning_rate": 8.746735153334857e-06,
+      "loss": 0.0001,
+      "step": 618
+    },
+    {
+      "epoch": 0.03989173164915899,
+      "grad_norm": 0.117186272137885,
+      "learning_rate": 8.74893571040039e-06,
+      "loss": 0.0017,
+      "step": 619
+    },
+    {
+      "epoch": 0.03995617709608816,
+      "grad_norm": 0.11402360435857845,
+      "learning_rate": 8.751132715314151e-06,
+      "loss": 0.0002,
+      "step": 620
+    },
+    {
+      "epoch": 0.04002062254301734,
+      "grad_norm": 2.0487442217612113,
+      "learning_rate": 8.75332617952547e-06,
+      "loss": 0.0063,
+      "step": 621
+    },
+    {
+      "epoch": 0.04008506798994651,
+      "grad_norm": 0.2021698686969095,
+      "learning_rate": 8.755516114428408e-06,
+      "loss": 0.0005,
+      "step": 622
+    },
+    {
+      "epoch": 0.040149513436875686,
+      "grad_norm": 0.0006979889077255704,
+      "learning_rate": 8.757702531362116e-06,
+      "loss": 0.0,
+      "step": 623
+    },
+    {
+      "epoch": 0.04021395888380486,
+      "grad_norm": 0.0057224222695230215,
+      "learning_rate": 8.759885441611187e-06,
+      "loss": 0.0,
+      "step": 624
+    },
+    {
+      "epoch": 0.040278404330734034,
+      "grad_norm": 0.0025807465767095474,
+      "learning_rate": 8.762064856406002e-06,
+      "loss": 0.0,
+      "step": 625
+    },
+    {
+      "epoch": 0.040342849777663205,
+      "grad_norm": 0.3693147077263644,
+      "learning_rate": 8.76424078692308e-06,
+      "loss": 0.0005,
+      "step": 626
+    },
+    {
+      "epoch": 0.04040729522459238,
+      "grad_norm": 0.0006090807770591003,
+      "learning_rate": 8.766413244285424e-06,
+      "loss": 0.0,
+      "step": 627
+    },
+    {
+      "epoch": 0.04047174067152156,
+      "grad_norm": 0.03648462138970883,
+      "learning_rate": 8.768582239562856e-06,
+      "loss": 0.0001,
+      "step": 628
+    },
+    {
+      "epoch": 0.04053618611845073,
+      "grad_norm": 0.0005767880342411257,
+      "learning_rate": 8.770747783772357e-06,
+      "loss": 0.0,
+      "step": 629
+    },
+    {
+      "epoch": 0.04060063156537991,
+      "grad_norm": 0.0006805918616400947,
+      "learning_rate": 8.77290988787841e-06,
+      "loss": 0.0,
+      "step": 630
+    },
+    {
+      "epoch": 0.04066507701230908,
+      "grad_norm": 0.004117036655468099,
+      "learning_rate": 8.775068562793323e-06,
+      "loss": 0.0,
+      "step": 631
+    },
+    {
+      "epoch": 0.040729522459238256,
+      "grad_norm": 0.0007426168119818877,
+      "learning_rate": 8.777223819377568e-06,
+      "loss": 0.0,
+      "step": 632
+    },
+    {
+      "epoch": 0.040793967906167426,
+      "grad_norm": 0.011304413014642186,
+      "learning_rate": 8.7793756684401e-06,
+      "loss": 0.0,
+      "step": 633
+    },
+    {
+      "epoch": 0.040858413353096604,
+      "grad_norm": 0.0004935244143254786,
+      "learning_rate": 8.78152412073869e-06,
+      "loss": 0.0,
+      "step": 634
+    },
+    {
+      "epoch": 0.04092285880002578,
+      "grad_norm": 0.007054770802510865,
+      "learning_rate": 8.783669186980247e-06,
+      "loss": 0.0,
+      "step": 635
+    },
+    {
+      "epoch": 0.04098730424695495,
+      "grad_norm": 0.01369389714554964,
+      "learning_rate": 8.78581087782113e-06,
+      "loss": 0.0,
+      "step": 636
+    },
+    {
+      "epoch": 0.04105174969388413,
+      "grad_norm": 0.011906267839883657,
+      "learning_rate": 8.787949203867476e-06,
+      "loss": 0.0,
+      "step": 637
+    },
+    {
+      "epoch": 0.0411161951408133,
+      "grad_norm": 0.006885284197584247,
+      "learning_rate": 8.790084175675511e-06,
+      "loss": 0.0,
+      "step": 638
+    },
+    {
+      "epoch": 0.04118064058774248,
+      "grad_norm": 0.4487910847199546,
+      "learning_rate": 8.792215803751859e-06,
+      "loss": 0.0024,
+      "step": 639
+    },
+    {
+      "epoch": 0.04124508603467165,
+      "grad_norm": 0.9246561480973362,
+      "learning_rate": 8.794344098553859e-06,
+      "loss": 0.0008,
+      "step": 640
+    },
+    {
+      "epoch": 0.041309531481600825,
+      "grad_norm": 0.04488668151610257,
+      "learning_rate": 8.796469070489866e-06,
+      "loss": 0.0005,
+      "step": 641
+    },
+    {
+      "epoch": 0.04137397692853,
+      "grad_norm": 0.027408085647675932,
+      "learning_rate": 8.798590729919567e-06,
+      "loss": 0.0,
+      "step": 642
+    },
+    {
+      "epoch": 0.04143842237545917,
+      "grad_norm": 0.0017001050288757774,
+      "learning_rate": 8.800709087154267e-06,
+      "loss": 0.0,
+      "step": 643
+    },
+    {
+      "epoch": 0.04150286782238835,
+      "grad_norm": 0.16256402501267625,
+      "learning_rate": 8.802824152457208e-06,
+      "loss": 0.0003,
+      "step": 644
+    },
+    {
+      "epoch": 0.04156731326931752,
+      "grad_norm": 0.011062200472460433,
+      "learning_rate": 8.804935936043852e-06,
+      "loss": 0.0,
+      "step": 645
+    },
+    {
+      "epoch": 0.0416317587162467,
+      "grad_norm": 0.021831654093761303,
+      "learning_rate": 8.80704444808219e-06,
+      "loss": 0.0001,
+      "step": 646
+    },
+    {
+      "epoch": 0.04169620416317587,
+      "grad_norm": 0.06078771214529044,
+      "learning_rate": 8.809149698693027e-06,
+      "loss": 0.0001,
+      "step": 647
+    },
+    {
+      "epoch": 0.04176064961010505,
+      "grad_norm": 0.005953470628247643,
+      "learning_rate": 8.811251697950276e-06,
+      "loss": 0.0,
+      "step": 648
+    },
+    {
+      "epoch": 0.04182509505703422,
+      "grad_norm": 0.2050906677043541,
+      "learning_rate": 8.813350455881246e-06,
+      "loss": 0.0007,
+      "step": 649
+    },
+    {
+      "epoch": 0.041889540503963395,
+      "grad_norm": 0.1971951160380812,
+      "learning_rate": 8.815445982466934e-06,
+      "loss": 0.0003,
+      "step": 650
+    },
+    {
+      "epoch": 0.04195398595089257,
+      "grad_norm": 0.00236395117113748,
+      "learning_rate": 8.817538287642305e-06,
+      "loss": 0.0,
+      "step": 651
+    },
+    {
+      "epoch": 0.04201843139782174,
+      "grad_norm": 0.1256566977010292,
+      "learning_rate": 8.819627381296574e-06,
+      "loss": 0.0007,
+      "step": 652
+    },
+    {
+      "epoch": 0.04208287684475092,
+      "grad_norm": 0.07903941680245861,
+      "learning_rate": 8.82171327327349e-06,
+      "loss": 0.0003,
+      "step": 653
+    },
+    {
+      "epoch": 0.04214732229168009,
+      "grad_norm": 0.39863904829441993,
+      "learning_rate": 8.823795973371614e-06,
+      "loss": 0.0012,
+      "step": 654
+    },
+    {
+      "epoch": 0.04221176773860927,
+      "grad_norm": 0.035371382107604585,
+      "learning_rate": 8.825875491344588e-06,
+      "loss": 0.0001,
+      "step": 655
+    },
+    {
+      "epoch": 0.04227621318553844,
+      "grad_norm": 0.02279703556094779,
+      "learning_rate": 8.827951836901422e-06,
+      "loss": 0.0,
+      "step": 656
+    },
+    {
+      "epoch": 0.042340658632467616,
+      "grad_norm": 0.26524165484813206,
+      "learning_rate": 8.830025019706755e-06,
+      "loss": 0.0005,
+      "step": 657
+    },
+    {
+      "epoch": 0.042405104079396794,
+      "grad_norm": 0.01918668875669408,
+      "learning_rate": 8.832095049381132e-06,
+      "loss": 0.0001,
+      "step": 658
+    },
+    {
+      "epoch": 0.042469549526325964,
+      "grad_norm": 0.02418418424942438,
+      "learning_rate": 8.834161935501262e-06,
+      "loss": 0.0002,
+      "step": 659
+    },
+    {
+      "epoch": 0.04253399497325514,
+      "grad_norm": 0.0029525133059810048,
+      "learning_rate": 8.836225687600296e-06,
+      "loss": 0.0,
+      "step": 660
+    },
+    {
+      "epoch": 0.04259844042018431,
+      "grad_norm": 0.06152243840450862,
+      "learning_rate": 8.838286315168083e-06,
+      "loss": 0.0001,
+      "step": 661
+    },
+    {
+      "epoch": 0.04266288586711349,
+      "grad_norm": 0.05055057189679331,
+      "learning_rate": 8.840343827651438e-06,
+      "loss": 0.0004,
+      "step": 662
+    },
+    {
+      "epoch": 0.04272733131404266,
+      "grad_norm": 0.00030230320615462757,
+      "learning_rate": 8.842398234454391e-06,
+      "loss": 0.0,
+      "step": 663
+    },
+    {
+      "epoch": 0.04279177676097184,
+      "grad_norm": 0.022563544469127325,
+      "learning_rate": 8.844449544938457e-06,
+      "loss": 0.0,
+      "step": 664
+    },
+    {
+      "epoch": 0.042856222207901015,
+      "grad_norm": 0.0010129811466353747,
+      "learning_rate": 8.846497768422887e-06,
+      "loss": 0.0,
+      "step": 665
+    },
+    {
+      "epoch": 0.042920667654830186,
+      "grad_norm": 0.10487420670137693,
+      "learning_rate": 8.84854291418492e-06,
+      "loss": 0.0001,
+      "step": 666
+    },
+    {
+      "epoch": 0.04298511310175936,
+      "grad_norm": 0.13032998283350847,
+      "learning_rate": 8.850584991460033e-06,
+      "loss": 0.0024,
+      "step": 667
+    },
+    {
+      "epoch": 0.043049558548688534,
+      "grad_norm": 0.12460349571325573,
+      "learning_rate": 8.852624009442204e-06,
+      "loss": 0.0001,
+      "step": 668
+    },
+    {
+      "epoch": 0.04311400399561771,
+      "grad_norm": 0.0003381800828124894,
+      "learning_rate": 8.854659977284147e-06,
+      "loss": 0.0,
+      "step": 669
+    },
+    {
+      "epoch": 0.04317844944254688,
+      "grad_norm": 0.007937901512867782,
+      "learning_rate": 8.856692904097561e-06,
+      "loss": 0.0001,
+      "step": 670
+    },
+    {
+      "epoch": 0.04324289488947606,
+      "grad_norm": 0.0008805466292661837,
+      "learning_rate": 8.85872279895338e-06,
+      "loss": 0.0,
+      "step": 671
+    },
+    {
+      "epoch": 0.04330734033640523,
+      "grad_norm": 0.17617806573889097,
+      "learning_rate": 8.860749670882013e-06,
+      "loss": 0.0005,
+      "step": 672
+    },
+    {
+      "epoch": 0.04337178578333441,
+      "grad_norm": 0.00395744069753078,
+      "learning_rate": 8.862773528873578e-06,
+      "loss": 0.0,
+      "step": 673
+    },
+    {
+      "epoch": 0.043436231230263585,
+      "grad_norm": 0.007247134466748965,
+      "learning_rate": 8.864794381878157e-06,
+      "loss": 0.0,
+      "step": 674
+    },
+    {
+      "epoch": 0.043500676677192755,
+      "grad_norm": 0.0030663660709385806,
+      "learning_rate": 8.86681223880602e-06,
+      "loss": 0.0,
+      "step": 675
+    },
+    {
+      "epoch": 0.04356512212412193,
+      "grad_norm": 0.017393769671059423,
+      "learning_rate": 8.868827108527866e-06,
+      "loss": 0.0,
+      "step": 676
+    },
+    {
+      "epoch": 0.0436295675710511,
+      "grad_norm": 0.0017702782273335328,
+      "learning_rate": 8.87083899987505e-06,
+      "loss": 0.0,
+      "step": 677
+    },
+    {
+      "epoch": 0.04369401301798028,
+      "grad_norm": 0.00022959416712921537,
+      "learning_rate": 8.872847921639834e-06,
+      "loss": 0.0,
+      "step": 678
+    },
+    {
+      "epoch": 0.04375845846490945,
+      "grad_norm": 0.003605868948068563,
+      "learning_rate": 8.874853882575593e-06,
+      "loss": 0.0,
+      "step": 679
+    },
+    {
+      "epoch": 0.04382290391183863,
+      "grad_norm": 0.01744342059980127,
+      "learning_rate": 8.876856891397061e-06,
+      "loss": 0.0001,
+      "step": 680
+    },
+    {
+      "epoch": 0.043887349358767806,
+      "grad_norm": 7.388464259112656e-05,
+      "learning_rate": 8.878856956780554e-06,
+      "loss": 0.0,
+      "step": 681
+    },
+    {
+      "epoch": 0.04395179480569698,
+      "grad_norm": 0.0013601676504224963,
+      "learning_rate": 8.880854087364192e-06,
+      "loss": 0.0,
+      "step": 682
+    },
+    {
+      "epoch": 0.044016240252626154,
+      "grad_norm": 0.2798160049980701,
+      "learning_rate": 8.882848291748122e-06,
+      "loss": 0.0011,
+      "step": 683
+    },
+    {
+      "epoch": 0.044080685699555325,
+      "grad_norm": 0.14433741459071334,
+      "learning_rate": 8.884839578494751e-06,
+      "loss": 0.0004,
+      "step": 684
+    },
+    {
+      "epoch": 0.0441451311464845,
+      "grad_norm": 0.24179156302192892,
+      "learning_rate": 8.886827956128954e-06,
+      "loss": 0.002,
+      "step": 685
+    },
+    {
+      "epoch": 0.04420957659341367,
+      "grad_norm": 0.0021927636562543086,
+      "learning_rate": 8.8888134331383e-06,
+      "loss": 0.0,
+      "step": 686
+    },
+    {
+      "epoch": 0.04427402204034285,
+      "grad_norm": 0.0022068358536655036,
+      "learning_rate": 8.890796017973267e-06,
+      "loss": 0.0,
+      "step": 687
+    },
+    {
+      "epoch": 0.04433846748727202,
+      "grad_norm": 0.005372660175485759,
+      "learning_rate": 8.892775719047455e-06,
+      "loss": 0.0,
+      "step": 688
+    },
+    {
+      "epoch": 0.0444029129342012,
+      "grad_norm": 0.03758966528205878,
+      "learning_rate": 8.894752544737809e-06,
+      "loss": 0.0001,
+      "step": 689
+    },
+    {
+      "epoch": 0.044467358381130376,
+      "grad_norm": 0.012306964234103844,
+      "learning_rate": 8.896726503384818e-06,
+      "loss": 0.0,
+      "step": 690
+    },
+    {
+      "epoch": 0.044531803828059546,
+      "grad_norm": 0.00020675266164698914,
+      "learning_rate": 8.898697603292746e-06,
+      "loss": 0.0,
+      "step": 691
+    },
+    {
+      "epoch": 0.044596249274988724,
+      "grad_norm": 0.0004554410369827967,
+      "learning_rate": 8.900665852729818e-06,
+      "loss": 0.0,
+      "step": 692
+    },
+    {
+      "epoch": 0.044660694721917894,
+      "grad_norm": 0.007791950816193934,
+      "learning_rate": 8.90263125992845e-06,
+      "loss": 0.0,
+      "step": 693
+    },
+    {
+      "epoch": 0.04472514016884707,
+      "grad_norm": 0.18602995977878675,
+      "learning_rate": 8.904593833085437e-06,
+      "loss": 0.0005,
+      "step": 694
+    },
+    {
+      "epoch": 0.04478958561577624,
+      "grad_norm": 0.013594795936494878,
+      "learning_rate": 8.90655358036217e-06,
+      "loss": 0.0001,
+      "step": 695
+    },
+    {
+      "epoch": 0.04485403106270542,
+      "grad_norm": 0.00023714633586855382,
+      "learning_rate": 8.908510509884837e-06,
+      "loss": 0.0,
+      "step": 696
+    },
+    {
+      "epoch": 0.0449184765096346,
+      "grad_norm": 6.832725945019009e-05,
+      "learning_rate": 8.910464629744626e-06,
+      "loss": 0.0,
+      "step": 697
+    },
+    {
+      "epoch": 0.04498292195656377,
+      "grad_norm": 0.016889844296641223,
+      "learning_rate": 8.912415947997922e-06,
+      "loss": 0.0003,
+      "step": 698
+    },
+    {
+      "epoch": 0.045047367403492945,
+      "grad_norm": 0.2184353592157856,
+      "learning_rate": 8.914364472666504e-06,
+      "loss": 0.0026,
+      "step": 699
+    },
+    {
+      "epoch": 0.045111812850422116,
+      "grad_norm": 0.25465223325641045,
+      "learning_rate": 8.916310211737758e-06,
+      "loss": 0.0008,
+      "step": 700
+    },
+    {
+      "epoch": 0.04517625829735129,
+      "grad_norm": 0.003655815925584125,
+      "learning_rate": 8.918253173164853e-06,
+      "loss": 0.0,
+      "step": 701
+    },
+    {
+      "epoch": 0.045240703744280464,
+      "grad_norm": 0.0006114981310893153,
+      "learning_rate": 8.92019336486695e-06,
+      "loss": 0.0,
+      "step": 702
+    },
+    {
+      "epoch": 0.04530514919120964,
+      "grad_norm": 0.00010137673386137189,
+      "learning_rate": 8.922130794729396e-06,
+      "loss": 0.0,
+      "step": 703
+    },
+    {
+      "epoch": 0.04536959463813882,
+      "grad_norm": 0.07228969160342366,
+      "learning_rate": 8.924065470603898e-06,
+      "loss": 0.0001,
+      "step": 704
+    },
+    {
+      "epoch": 0.04543404008506799,
+      "grad_norm": 0.1913084734088233,
+      "learning_rate": 8.925997400308744e-06,
+      "loss": 0.0004,
+      "step": 705
+    },
+    {
+      "epoch": 0.04549848553199717,
+      "grad_norm": 0.07917869010187947,
+      "learning_rate": 8.927926591628964e-06,
+      "loss": 0.0003,
+      "step": 706
+    },
+    {
+      "epoch": 0.04556293097892634,
+      "grad_norm": 0.06977438315324237,
+      "learning_rate": 8.929853052316538e-06,
+      "loss": 0.0004,
+      "step": 707
+    },
+    {
+      "epoch": 0.045627376425855515,
+      "grad_norm": 0.034029532033426,
+      "learning_rate": 8.931776790090572e-06,
+      "loss": 0.0001,
+      "step": 708
+    },
+    {
+      "epoch": 0.045691821872784685,
+      "grad_norm": 0.08348560267343069,
+      "learning_rate": 8.933697812637488e-06,
+      "loss": 0.0002,
+      "step": 709
+    },
+    {
+      "epoch": 0.04575626731971386,
+      "grad_norm": 0.051544153316252506,
+      "learning_rate": 8.935616127611207e-06,
+      "loss": 0.0017,
+      "step": 710
+    },
+    {
+      "epoch": 0.04582071276664303,
+      "grad_norm": 0.0763091812179522,
+      "learning_rate": 8.937531742633331e-06,
+      "loss": 0.0001,
+      "step": 711
+    },
+    {
+      "epoch": 0.04588515821357221,
+      "grad_norm": 0.0007173089122636117,
+      "learning_rate": 8.939444665293331e-06,
+      "loss": 0.0,
+      "step": 712
+    },
+    {
+      "epoch": 0.04594960366050139,
+      "grad_norm": 0.004598107264150894,
+      "learning_rate": 8.941354903148714e-06,
+      "loss": 0.0,
+      "step": 713
+    },
+    {
+      "epoch": 0.04601404910743056,
+      "grad_norm": 0.000392021647295444,
+      "learning_rate": 8.943262463725215e-06,
+      "loss": 0.0,
+      "step": 714
+    },
+    {
+      "epoch": 0.046078494554359736,
+      "grad_norm": 0.005239661838366242,
+      "learning_rate": 8.945167354516973e-06,
+      "loss": 0.0,
+      "step": 715
+    },
+    {
+      "epoch": 0.04614294000128891,
+      "grad_norm": 0.013039220056873272,
+      "learning_rate": 8.947069582986702e-06,
+      "loss": 0.0,
+      "step": 716
+    },
+    {
+      "epoch": 0.046207385448218084,
+      "grad_norm": 0.7156779940111023,
+      "learning_rate": 8.948969156565867e-06,
+      "loss": 0.0028,
+      "step": 717
+    },
+    {
+      "epoch": 0.046271830895147255,
+      "grad_norm": 0.0014232633069768458,
+      "learning_rate": 8.950866082654867e-06,
+      "loss": 0.0,
+      "step": 718
+    },
+    {
+      "epoch": 0.04633627634207643,
+      "grad_norm": 0.010360478298695039,
+      "learning_rate": 8.952760368623196e-06,
+      "loss": 0.0,
+      "step": 719
+    },
+    {
+      "epoch": 0.04640072178900561,
+      "grad_norm": 0.04740542080782063,
+      "learning_rate": 8.954652021809624e-06,
+      "loss": 0.0002,
+      "step": 720
+    },
+    {
+      "epoch": 0.04646516723593478,
+      "grad_norm": 0.0013033683657919647,
+      "learning_rate": 8.956541049522359e-06,
+      "loss": 0.0,
+      "step": 721
+    },
+    {
+      "epoch": 0.04652961268286396,
+      "grad_norm": 0.00024011095477306243,
+      "learning_rate": 8.958427459039227e-06,
+      "loss": 0.0,
+      "step": 722
+    },
+    {
+      "epoch": 0.04659405812979313,
+      "grad_norm": 0.03165036091424995,
+      "learning_rate": 8.960311257607835e-06,
+      "loss": 0.0016,
+      "step": 723
+    },
+    {
+      "epoch": 0.046658503576722306,
+      "grad_norm": 0.0037302307404365937,
+      "learning_rate": 8.96219245244573e-06,
+      "loss": 0.0,
+      "step": 724
+    },
+    {
+      "epoch": 0.046722949023651476,
+      "grad_norm": 0.11715760644466812,
+      "learning_rate": 8.964071050740584e-06,
+      "loss": 0.0004,
+      "step": 725
+    },
+    {
+      "epoch": 0.046787394470580654,
+      "grad_norm": 0.011645889650838596,
+      "learning_rate": 8.965947059650336e-06,
+      "loss": 0.0001,
+      "step": 726
+    },
+    {
+      "epoch": 0.04685183991750983,
+      "grad_norm": 0.06984104940667903,
+      "learning_rate": 8.967820486303374e-06,
+      "loss": 0.0003,
+      "step": 727
+    },
+    {
+      "epoch": 0.046916285364439,
+      "grad_norm": 0.32142027752446717,
+      "learning_rate": 8.96969133779869e-06,
+      "loss": 0.0005,
+      "step": 728
+    },
+    {
+      "epoch": 0.04698073081136818,
+      "grad_norm": 2.551444979388958,
+      "learning_rate": 8.971559621206039e-06,
+      "loss": 0.0197,
+      "step": 729
+    },
+    {
+      "epoch": 0.04704517625829735,
+      "grad_norm": 0.002081532217039823,
+      "learning_rate": 8.973425343566106e-06,
+      "loss": 0.0,
+      "step": 730
+    },
+    {
+      "epoch": 0.04710962170522653,
+      "grad_norm": 0.25030308471193197,
+      "learning_rate": 8.975288511890657e-06,
+      "loss": 0.0052,
+      "step": 731
+    },
+    {
+      "epoch": 0.0471740671521557,
+      "grad_norm": 0.005732779290674802,
+      "learning_rate": 8.977149133162707e-06,
+      "loss": 0.0,
+      "step": 732
+    },
+    {
+      "epoch": 0.047238512599084875,
+      "grad_norm": 0.10988615433689451,
+      "learning_rate": 8.979007214336669e-06,
+      "loss": 0.0003,
+      "step": 733
+    },
+    {
+      "epoch": 0.047302958046014046,
+      "grad_norm": 0.12829035230675495,
+      "learning_rate": 8.98086276233851e-06,
+      "loss": 0.0003,
+      "step": 734
+    },
+    {
+      "epoch": 0.04736740349294322,
+      "grad_norm": 0.007249099234747826,
+      "learning_rate": 8.982715784065911e-06,
+      "loss": 0.0,
+      "step": 735
+    },
+    {
+      "epoch": 0.0474318489398724,
+      "grad_norm": 0.019692113896025192,
+      "learning_rate": 8.984566286388422e-06,
+      "loss": 0.0001,
+      "step": 736
+    },
+    {
+      "epoch": 0.04749629438680157,
+      "grad_norm": 0.05206044806916652,
+      "learning_rate": 8.986414276147602e-06,
+      "loss": 0.0002,
+      "step": 737
+    },
+    {
+      "epoch": 0.04756073983373075,
+      "grad_norm": 0.028860954812302157,
+      "learning_rate": 8.988259760157187e-06,
+      "loss": 0.0001,
+      "step": 738
+    },
+    {
+      "epoch": 0.04762518528065992,
+      "grad_norm": 0.001301629719628138,
+      "learning_rate": 8.990102745203234e-06,
+      "loss": 0.0,
+      "step": 739
+    },
+    {
+      "epoch": 0.0476896307275891,
+      "grad_norm": 0.002345407602635431,
+      "learning_rate": 8.991943238044267e-06,
+      "loss": 0.0,
+      "step": 740
+    },
+    {
+      "epoch": 0.04775407617451827,
+      "grad_norm": 0.12231212697510714,
+      "learning_rate": 8.993781245411428e-06,
+      "loss": 0.0004,
+      "step": 741
+    },
+    {
+      "epoch": 0.047818521621447445,
+      "grad_norm": 0.049014023716665675,
+      "learning_rate": 8.995616774008632e-06,
+      "loss": 0.0001,
+      "step": 742
+    },
+    {
+      "epoch": 0.04788296706837662,
+      "grad_norm": 0.009068422831759652,
+      "learning_rate": 8.9974498305127e-06,
+      "loss": 0.0,
+      "step": 743
+    },
+    {
+      "epoch": 0.04794741251530579,
+      "grad_norm": 0.004244489514567139,
+      "learning_rate": 8.999280421573518e-06,
+      "loss": 0.0,
+      "step": 744
+    },
+    {
+      "epoch": 0.04801185796223497,
+      "grad_norm": 0.10947748900064816,
+      "learning_rate": 9.001108553814172e-06,
+      "loss": 0.0003,
+      "step": 745
+    },
+    {
+      "epoch": 0.04807630340916414,
+      "grad_norm": 0.008424072505390963,
+      "learning_rate": 9.002934233831098e-06,
+      "loss": 0.0,
+      "step": 746
+    },
+    {
+      "epoch": 0.04814074885609332,
+      "grad_norm": 0.006270599434854108,
+      "learning_rate": 9.004757468194222e-06,
+      "loss": 0.0,
+      "step": 747
+    },
+    {
+      "epoch": 0.04820519430302249,
+      "grad_norm": 0.01742854568377238,
+      "learning_rate": 9.006578263447102e-06,
+      "loss": 0.0001,
+      "step": 748
+    },
+    {
+      "epoch": 0.048269639749951666,
+      "grad_norm": 0.024492512070331957,
+      "learning_rate": 9.008396626107069e-06,
+      "loss": 0.0,
+      "step": 749
+    },
+    {
+      "epoch": 0.048334085196880844,
+      "grad_norm": 0.12119974988300376,
+      "learning_rate": 9.010212562665369e-06,
+      "loss": 0.0001,
+      "step": 750
+    },
+    {
+      "epoch": 0.048398530643810014,
+      "grad_norm": 0.06679123539568732,
+      "learning_rate": 9.012026079587298e-06,
+      "loss": 0.0001,
+      "step": 751
+    },
+    {
+      "epoch": 0.04846297609073919,
+      "grad_norm": 0.07117144798083709,
+      "learning_rate": 9.013837183312346e-06,
+      "loss": 0.0001,
+      "step": 752
+    },
+    {
+      "epoch": 0.04852742153766836,
+      "grad_norm": 0.00487018289232718,
+      "learning_rate": 9.015645880254325e-06,
+      "loss": 0.0,
+      "step": 753
+    },
+    {
+      "epoch": 0.04859186698459754,
+      "grad_norm": 0.612146846985091,
+      "learning_rate": 9.017452176801516e-06,
+      "loss": 0.0025,
+      "step": 754
+    },
+    {
+      "epoch": 0.04865631243152671,
+      "grad_norm": 0.0034136926349857006,
+      "learning_rate": 9.019256079316795e-06,
+      "loss": 0.0,
+      "step": 755
+    },
+    {
+      "epoch": 0.04872075787845589,
+      "grad_norm": 0.12108867773077127,
+      "learning_rate": 9.021057594137776e-06,
+      "loss": 0.0002,
+      "step": 756
+    },
+    {
+      "epoch": 0.04878520332538506,
+      "grad_norm": 0.004504167955212749,
+      "learning_rate": 9.022856727576939e-06,
+      "loss": 0.0,
+      "step": 757
+    },
+    {
+      "epoch": 0.048849648772314236,
+      "grad_norm": 0.07505442996143205,
+      "learning_rate": 9.024653485921763e-06,
+      "loss": 0.0015,
+      "step": 758
+    },
+    {
+      "epoch": 0.04891409421924341,
+      "grad_norm": 0.0005453544489624827,
+      "learning_rate": 9.026447875434859e-06,
+      "loss": 0.0,
+      "step": 759
+    },
+    {
+      "epoch": 0.048978539666172584,
+      "grad_norm": 0.34459969236495225,
+      "learning_rate": 9.0282399023541e-06,
+      "loss": 0.0002,
+      "step": 760
+    },
+    {
+      "epoch": 0.04904298511310176,
+      "grad_norm": 0.18131343537161349,
+      "learning_rate": 9.030029572892756e-06,
+      "loss": 0.0007,
+      "step": 761
+    },
+    {
+      "epoch": 0.04910743056003093,
+      "grad_norm": 0.02098021280926811,
+      "learning_rate": 9.031816893239614e-06,
+      "loss": 0.0,
+      "step": 762
+    },
+    {
+      "epoch": 0.04917187600696011,
+      "grad_norm": 0.08627074285048919,
+      "learning_rate": 9.033601869559115e-06,
+      "loss": 0.0035,
+      "step": 763
+    },
+    {
+      "epoch": 0.04923632145388928,
+      "grad_norm": 0.002511861290816773,
+      "learning_rate": 9.035384507991479e-06,
+      "loss": 0.0,
+      "step": 764
+    },
+    {
+      "epoch": 0.04930076690081846,
+      "grad_norm": 0.40511111030437946,
+      "learning_rate": 9.037164814652826e-06,
+      "loss": 0.0008,
+      "step": 765
+    },
+    {
+      "epoch": 0.049365212347747635,
+      "grad_norm": 0.14034011913080038,
+      "learning_rate": 9.038942795635312e-06,
+      "loss": 0.0001,
+      "step": 766
+    },
+    {
+      "epoch": 0.049429657794676805,
+      "grad_norm": 0.06861259699001675,
+      "learning_rate": 9.040718457007249e-06,
+      "loss": 0.0001,
+      "step": 767
+    },
+    {
+      "epoch": 0.04949410324160598,
+      "grad_norm": 0.001581944552875584,
+      "learning_rate": 9.042491804813226e-06,
+      "loss": 0.0,
+      "step": 768
+    },
+    {
+      "epoch": 0.04955854868853515,
+      "grad_norm": 0.005308040387458921,
+      "learning_rate": 9.044262845074235e-06,
+      "loss": 0.0,
+      "step": 769
+    },
+    {
+      "epoch": 0.04962299413546433,
+      "grad_norm": 0.0019728464218824234,
+      "learning_rate": 9.046031583787798e-06,
+      "loss": 0.0,
+      "step": 770
+    },
+    {
+      "epoch": 0.0496874395823935,
+      "grad_norm": 0.003223276699460636,
+      "learning_rate": 9.047798026928083e-06,
+      "loss": 0.0,
+      "step": 771
+    },
+    {
+      "epoch": 0.04975188502932268,
+      "grad_norm": 0.021868800717484028,
+      "learning_rate": 9.049562180446029e-06,
+      "loss": 0.0001,
+      "step": 772
+    },
+    {
+      "epoch": 0.04981633047625185,
+      "grad_norm": 0.1826315175518166,
+      "learning_rate": 9.051324050269462e-06,
+      "loss": 0.0003,
+      "step": 773
+    },
+    {
+      "epoch": 0.04988077592318103,
+      "grad_norm": 0.29998356456240793,
+      "learning_rate": 9.05308364230322e-06,
+      "loss": 0.0011,
+      "step": 774
+    },
+    {
+      "epoch": 0.049945221370110204,
+      "grad_norm": 0.5431729488475763,
+      "learning_rate": 9.054840962429265e-06,
+      "loss": 0.0004,
+      "step": 775
+    },
+    {
+      "epoch": 0.050009666817039375,
+      "grad_norm": 0.009489328773548649,
+      "learning_rate": 9.056596016506807e-06,
+      "loss": 0.0,
+      "step": 776
+    },
+    {
+      "epoch": 0.05007411226396855,
+      "grad_norm": 0.00046785632344008304,
+      "learning_rate": 9.05834881037242e-06,
+      "loss": 0.0,
+      "step": 777
+    },
+    {
+      "epoch": 0.05013855771089772,
+      "grad_norm": 0.01111245605224189,
+      "learning_rate": 9.060099349840158e-06,
+      "loss": 0.0001,
+      "step": 778
+    },
+    {
+      "epoch": 0.0502030031578269,
+      "grad_norm": 0.07606922165075893,
+      "learning_rate": 9.061847640701665e-06,
+      "loss": 0.0005,
+      "step": 779
+    },
+    {
+      "epoch": 0.05026744860475607,
+      "grad_norm": 0.000699417674606527,
+      "learning_rate": 9.063593688726301e-06,
+      "loss": 0.0,
+      "step": 780
+    },
+    {
+      "epoch": 0.05033189405168525,
+      "grad_norm": 0.010236254823261563,
+      "learning_rate": 9.065337499661248e-06,
+      "loss": 0.0,
+      "step": 781
+    },
+    {
+      "epoch": 0.050396339498614426,
+      "grad_norm": 0.0017980719947457786,
+      "learning_rate": 9.067079079231624e-06,
+      "loss": 0.0,
+      "step": 782
+    },
+    {
+      "epoch": 0.050460784945543596,
+      "grad_norm": 0.000939549117843024,
+      "learning_rate": 9.068818433140602e-06,
+      "loss": 0.0,
+      "step": 783
+    },
+    {
+      "epoch": 0.050525230392472774,
+      "grad_norm": 0.002781551161266302,
+      "learning_rate": 9.070555567069513e-06,
+      "loss": 0.0,
+      "step": 784
+    },
+    {
+      "epoch": 0.050589675839401944,
+      "grad_norm": 0.0035230904250238056,
+      "learning_rate": 9.072290486677968e-06,
+      "loss": 0.0,
+      "step": 785
+    },
+    {
+      "epoch": 0.05065412128633112,
+      "grad_norm": 0.01560564350924316,
+      "learning_rate": 9.074023197603955e-06,
+      "loss": 0.0001,
+      "step": 786
+    },
+    {
+      "epoch": 0.05071856673326029,
+      "grad_norm": 0.010400567996276689,
+      "learning_rate": 9.075753705463962e-06,
+      "loss": 0.0,
+      "step": 787
+    },
+    {
+      "epoch": 0.05078301218018947,
+      "grad_norm": 0.00015725166531810745,
+      "learning_rate": 9.07748201585308e-06,
+      "loss": 0.0,
+      "step": 788
+    },
+    {
+      "epoch": 0.05084745762711865,
+      "grad_norm": 0.0016245494880248916,
+      "learning_rate": 9.079208134345112e-06,
+      "loss": 0.0,
+      "step": 789
+    },
+    {
+      "epoch": 0.05091190307404782,
+      "grad_norm": 0.001113785553319258,
+      "learning_rate": 9.080932066492682e-06,
+      "loss": 0.0,
+      "step": 790
+    },
+    {
+      "epoch": 0.050976348520976995,
+      "grad_norm": 0.46446961485109706,
+      "learning_rate": 9.082653817827336e-06,
+      "loss": 0.0007,
+      "step": 791
+    },
+    {
+      "epoch": 0.051040793967906166,
+      "grad_norm": 0.0013411261450823374,
+      "learning_rate": 9.084373393859663e-06,
+      "loss": 0.0,
+      "step": 792
+    },
+    {
+      "epoch": 0.05110523941483534,
+      "grad_norm": 0.004991003306657957,
+      "learning_rate": 9.086090800079385e-06,
+      "loss": 0.0,
+      "step": 793
+    },
+    {
+      "epoch": 0.051169684861764514,
+      "grad_norm": 0.0013036785024449445,
+      "learning_rate": 9.087806041955472e-06,
+      "loss": 0.0,
+      "step": 794
+    },
+    {
+      "epoch": 0.05123413030869369,
+      "grad_norm": 0.00048091056047600136,
+      "learning_rate": 9.089519124936242e-06,
+      "loss": 0.0,
+      "step": 795
+    },
+    {
+      "epoch": 0.05129857575562286,
+      "grad_norm": 0.0016136776036838984,
+      "learning_rate": 9.091230054449467e-06,
+      "loss": 0.0,
+      "step": 796
+    },
+    {
+      "epoch": 0.05136302120255204,
+      "grad_norm": 0.4099738669208059,
+      "learning_rate": 9.092938835902477e-06,
+      "loss": 0.0029,
+      "step": 797
+    },
+    {
+      "epoch": 0.05142746664948122,
+      "grad_norm": 0.003942835183172,
+      "learning_rate": 9.094645474682253e-06,
+      "loss": 0.0,
+      "step": 798
+    },
+    {
+      "epoch": 0.05149191209641039,
+      "grad_norm": 0.0030768112635003782,
+      "learning_rate": 9.096349976155548e-06,
+      "loss": 0.0,
+      "step": 799
+    },
+    {
+      "epoch": 0.051556357543339565,
+      "grad_norm": 0.006397829046396678,
+      "learning_rate": 9.098052345668971e-06,
+      "loss": 0.0,
+      "step": 800
+    },
+    {
+      "epoch": 0.051620802990268735,
+      "grad_norm": 0.06002623182371448,
+      "learning_rate": 9.099752588549096e-06,
+      "loss": 0.0007,
+      "step": 801
+    },
+    {
+      "epoch": 0.05168524843719791,
+      "grad_norm": 3.3841404940161315,
+      "learning_rate": 9.101450710102554e-06,
+      "loss": 0.034,
+      "step": 802
+    },
+    {
+      "epoch": 0.05174969388412708,
+      "grad_norm": 0.0013937209491712162,
+      "learning_rate": 9.103146715616145e-06,
+      "loss": 0.0,
+      "step": 803
+    },
+    {
+      "epoch": 0.05181413933105626,
+      "grad_norm": 6.458562080363265e-05,
+      "learning_rate": 9.104840610356928e-06,
+      "loss": 0.0,
+      "step": 804
+    },
+    {
+      "epoch": 0.05187858477798544,
+      "grad_norm": 0.3579401105183063,
+      "learning_rate": 9.10653239957232e-06,
+      "loss": 0.0025,
+      "step": 805
+    },
+    {
+      "epoch": 0.05194303022491461,
+      "grad_norm": 0.008360636412067086,
+      "learning_rate": 9.108222088490193e-06,
+      "loss": 0.0,
+      "step": 806
+    },
+    {
+      "epoch": 0.052007475671843786,
+      "grad_norm": 0.005285542514094032,
+      "learning_rate": 9.10990968231898e-06,
+      "loss": 0.0,
+      "step": 807
+    },
+    {
+      "epoch": 0.05207192111877296,
+      "grad_norm": 0.0002770779211620817,
+      "learning_rate": 9.11159518624775e-06,
+      "loss": 0.0,
+      "step": 808
+    },
+    {
+      "epoch": 0.052136366565702134,
+      "grad_norm": 0.0011585447889919547,
+      "learning_rate": 9.113278605446331e-06,
+      "loss": 0.0,
+      "step": 809
+    },
+    {
+      "epoch": 0.052200812012631305,
+      "grad_norm": 0.049277994096405396,
+      "learning_rate": 9.114959945065387e-06,
+      "loss": 0.0004,
+      "step": 810
+    },
+    {
+      "epoch": 0.05226525745956048,
+      "grad_norm": 0.00015442997540374582,
+      "learning_rate": 9.116639210236513e-06,
+      "loss": 0.0,
+      "step": 811
+    },
+    {
+      "epoch": 0.05232970290648966,
+      "grad_norm": 0.0013635697356662283,
+      "learning_rate": 9.11831640607234e-06,
+      "loss": 0.0,
+      "step": 812
+    },
+    {
+      "epoch": 0.05239414835341883,
+      "grad_norm": 0.016716470890307505,
+      "learning_rate": 9.119991537666614e-06,
+      "loss": 0.0002,
+      "step": 813
+    },
+    {
+      "epoch": 0.05245859380034801,
+      "grad_norm": 0.0005278616955096009,
+      "learning_rate": 9.121664610094306e-06,
+      "loss": 0.0,
+      "step": 814
+    },
+    {
+      "epoch": 0.05252303924727718,
+      "grad_norm": 0.0025750142539637693,
+      "learning_rate": 9.123335628411687e-06,
+      "loss": 0.0,
+      "step": 815
+    },
+    {
+      "epoch": 0.052587484694206356,
+      "grad_norm": 0.00020521592683312736,
+      "learning_rate": 9.125004597656428e-06,
+      "loss": 0.0,
+      "step": 816
+    },
+    {
+      "epoch": 0.052651930141135526,
+      "grad_norm": 0.19445270116812716,
+      "learning_rate": 9.126671522847696e-06,
+      "loss": 0.0004,
+      "step": 817
+    },
+    {
+      "epoch": 0.052716375588064704,
+      "grad_norm": 0.3771810287388212,
+      "learning_rate": 9.128336408986232e-06,
+      "loss": 0.0025,
+      "step": 818
+    },
+    {
+      "epoch": 0.052780821034993874,
+      "grad_norm": 0.05887705540169662,
+      "learning_rate": 9.129999261054454e-06,
+      "loss": 0.0001,
+      "step": 819
+    },
+    {
+      "epoch": 0.05284526648192305,
+      "grad_norm": 0.1534011955489274,
+      "learning_rate": 9.131660084016536e-06,
+      "loss": 0.0003,
+      "step": 820
+    },
+    {
+      "epoch": 0.05290971192885223,
+      "grad_norm": 0.013809870076849355,
+      "learning_rate": 9.133318882818504e-06,
+      "loss": 0.0001,
+      "step": 821
+    },
+    {
+      "epoch": 0.0529741573757814,
+      "grad_norm": 0.0032549074867175556,
+      "learning_rate": 9.134975662388321e-06,
+      "loss": 0.0,
+      "step": 822
+    },
+    {
+      "epoch": 0.05303860282271058,
+      "grad_norm": 0.009309798570618937,
+      "learning_rate": 9.136630427635978e-06,
+      "loss": 0.0001,
+      "step": 823
+    },
+    {
+      "epoch": 0.05310304826963975,
+      "grad_norm": 0.0033493742577040365,
+      "learning_rate": 9.138283183453572e-06,
+      "loss": 0.0,
+      "step": 824
+    },
+    {
+      "epoch": 0.053167493716568925,
+      "grad_norm": 0.018859710507561127,
+      "learning_rate": 9.139933934715408e-06,
+      "loss": 0.0001,
+      "step": 825
+    },
+    {
+      "epoch": 0.053231939163498096,
+      "grad_norm": 0.01560803326672018,
+      "learning_rate": 9.141582686278072e-06,
+      "loss": 0.0,
+      "step": 826
+    },
+    {
+      "epoch": 0.05329638461042727,
+      "grad_norm": 0.00016964226944774627,
+      "learning_rate": 9.143229442980527e-06,
+      "loss": 0.0,
+      "step": 827
+    },
+    {
+      "epoch": 0.05336083005735645,
+      "grad_norm": 0.022061608980455484,
+      "learning_rate": 9.144874209644185e-06,
+      "loss": 0.0016,
+      "step": 828
+    },
+    {
+      "epoch": 0.05342527550428562,
+      "grad_norm": 0.006649718737306569,
+      "learning_rate": 9.146516991073004e-06,
+      "loss": 0.0,
+      "step": 829
+    },
+    {
+      "epoch": 0.0534897209512148,
+      "grad_norm": 0.00042874270227666826,
+      "learning_rate": 9.148157792053569e-06,
+      "loss": 0.0,
+      "step": 830
+    },
+    {
+      "epoch": 0.05355416639814397,
+      "grad_norm": 0.2741124949858014,
+      "learning_rate": 9.149796617355175e-06,
+      "loss": 0.0042,
+      "step": 831
+    },
+    {
+      "epoch": 0.05361861184507315,
+      "grad_norm": 0.03657944854462583,
+      "learning_rate": 9.151433471729903e-06,
+      "loss": 0.0001,
+      "step": 832
+    },
+    {
+      "epoch": 0.05368305729200232,
+      "grad_norm": 0.001000460494932859,
+      "learning_rate": 9.153068359912718e-06,
+      "loss": 0.0,
+      "step": 833
+    },
+    {
+      "epoch": 0.053747502738931495,
+      "grad_norm": 0.4291703666910219,
+      "learning_rate": 9.154701286621536e-06,
+      "loss": 0.0005,
+      "step": 834
+    },
+    {
+      "epoch": 0.05381194818586067,
+      "grad_norm": 0.011767546566647835,
+      "learning_rate": 9.156332256557316e-06,
+      "loss": 0.0,
+      "step": 835
+    },
+    {
+      "epoch": 0.05387639363278984,
+      "grad_norm": 0.02017886370790413,
+      "learning_rate": 9.157961274404139e-06,
+      "loss": 0.0001,
+      "step": 836
+    },
+    {
+      "epoch": 0.05394083907971902,
+      "grad_norm": 0.1635306058218916,
+      "learning_rate": 9.159588344829283e-06,
+      "loss": 0.0003,
+      "step": 837
+    },
+    {
+      "epoch": 0.05400528452664819,
+      "grad_norm": 0.03241790156834949,
+      "learning_rate": 9.161213472483306e-06,
+      "loss": 0.0001,
+      "step": 838
+    },
+    {
+      "epoch": 0.05406972997357737,
+      "grad_norm": 0.03645090452502005,
+      "learning_rate": 9.162836662000131e-06,
+      "loss": 0.0,
+      "step": 839
+    },
+    {
+      "epoch": 0.05413417542050654,
+      "grad_norm": 0.001018123489854683,
+      "learning_rate": 9.164457917997124e-06,
+      "loss": 0.0,
+      "step": 840
+    },
+    {
+      "epoch": 0.054198620867435716,
+      "grad_norm": 0.0006147523546883594,
+      "learning_rate": 9.166077245075165e-06,
+      "loss": 0.0,
+      "step": 841
+    },
+    {
+      "epoch": 0.05426306631436489,
+      "grad_norm": 0.03234380830020073,
+      "learning_rate": 9.167694647818732e-06,
+      "loss": 0.0002,
+      "step": 842
+    },
+    {
+      "epoch": 0.054327511761294064,
+      "grad_norm": 0.30908209151377397,
+      "learning_rate": 9.16931013079598e-06,
+      "loss": 0.0003,
+      "step": 843
+    },
+    {
+      "epoch": 0.05439195720822324,
+      "grad_norm": 0.03522377846818942,
+      "learning_rate": 9.170923698558816e-06,
+      "loss": 0.0005,
+      "step": 844
+    },
+    {
+      "epoch": 0.05445640265515241,
+      "grad_norm": 0.000277197422768193,
+      "learning_rate": 9.172535355642978e-06,
+      "loss": 0.0,
+      "step": 845
+    },
+    {
+      "epoch": 0.05452084810208159,
+      "grad_norm": 0.0016432842093718164,
+      "learning_rate": 9.174145106568109e-06,
+      "loss": 0.0,
+      "step": 846
+    },
+    {
+      "epoch": 0.05458529354901076,
+      "grad_norm": 0.013591225137140847,
+      "learning_rate": 9.175752955837838e-06,
+      "loss": 0.0,
+      "step": 847
+    },
+    {
+      "epoch": 0.05464973899593994,
+      "grad_norm": 0.8462395497498096,
+      "learning_rate": 9.177358907939845e-06,
+      "loss": 0.0031,
+      "step": 848
+    },
+    {
+      "epoch": 0.05471418444286911,
+      "grad_norm": 0.04412306889026068,
+      "learning_rate": 9.178962967345948e-06,
+      "loss": 0.0001,
+      "step": 849
+    },
+    {
+      "epoch": 0.054778629889798286,
+      "grad_norm": 0.005405304038502618,
+      "learning_rate": 9.180565138512175e-06,
+      "loss": 0.0,
+      "step": 850
+    },
+    {
+      "epoch": 0.05484307533672746,
+      "grad_norm": 0.3794935924368555,
+      "learning_rate": 9.182165425878829e-06,
+      "loss": 0.0019,
+      "step": 851
+    },
+    {
+      "epoch": 0.054907520783656634,
+      "grad_norm": 0.0020777573880359597,
+      "learning_rate": 9.183763833870574e-06,
+      "loss": 0.0,
+      "step": 852
+    },
+    {
+      "epoch": 0.05497196623058581,
+      "grad_norm": 0.004563794466339452,
+      "learning_rate": 9.185360366896503e-06,
+      "loss": 0.0,
+      "step": 853
+    },
+    {
+      "epoch": 0.05503641167751498,
+      "grad_norm": 0.007595697852477659,
+      "learning_rate": 9.186955029350209e-06,
+      "loss": 0.0001,
+      "step": 854
+    },
+    {
+      "epoch": 0.05510085712444416,
+      "grad_norm": 0.021470438571486155,
+      "learning_rate": 9.188547825609863e-06,
+      "loss": 0.0003,
+      "step": 855
+    },
+    {
+      "epoch": 0.05516530257137333,
+      "grad_norm": 0.03807929940940631,
+      "learning_rate": 9.190138760038282e-06,
+      "loss": 0.0001,
+      "step": 856
+    },
+    {
+      "epoch": 0.05522974801830251,
+      "grad_norm": 0.052610970411618635,
+      "learning_rate": 9.191727836983e-06,
+      "loss": 0.0002,
+      "step": 857
+    },
+    {
+      "epoch": 0.05529419346523168,
+      "grad_norm": 0.011456006523472022,
+      "learning_rate": 9.19331506077634e-06,
+      "loss": 0.0001,
+      "step": 858
+    },
+    {
+      "epoch": 0.055358638912160855,
+      "grad_norm": 0.010611982148422123,
+      "learning_rate": 9.19490043573549e-06,
+      "loss": 0.0,
+      "step": 859
+    },
+    {
+      "epoch": 0.05542308435909003,
+      "grad_norm": 0.00036207342749774215,
+      "learning_rate": 9.196483966162567e-06,
+      "loss": 0.0,
+      "step": 860
+    },
+    {
+      "epoch": 0.055487529806019203,
+      "grad_norm": 0.006096769717651004,
+      "learning_rate": 9.198065656344689e-06,
+      "loss": 0.0,
+      "step": 861
+    },
+    {
+      "epoch": 0.05555197525294838,
+      "grad_norm": 0.0033176289929112016,
+      "learning_rate": 9.199645510554044e-06,
+      "loss": 0.0,
+      "step": 862
+    },
+    {
+      "epoch": 0.05561642069987755,
+      "grad_norm": 0.07427952499156017,
+      "learning_rate": 9.201223533047966e-06,
+      "loss": 0.0007,
+      "step": 863
+    },
+    {
+      "epoch": 0.05568086614680673,
+      "grad_norm": 0.007857354582267988,
+      "learning_rate": 9.20279972806899e-06,
+      "loss": 0.0,
+      "step": 864
+    },
+    {
+      "epoch": 0.0557453115937359,
+      "grad_norm": 0.43490041132666496,
+      "learning_rate": 9.204374099844932e-06,
+      "loss": 0.003,
+      "step": 865
+    },
+    {
+      "epoch": 0.05580975704066508,
+      "grad_norm": 0.004341780147231225,
+      "learning_rate": 9.205946652588954e-06,
+      "loss": 0.0,
+      "step": 866
+    },
+    {
+      "epoch": 0.055874202487594254,
+      "grad_norm": 0.02007426507979371,
+      "learning_rate": 9.20751739049963e-06,
+      "loss": 0.0001,
+      "step": 867
+    },
+    {
+      "epoch": 0.055938647934523425,
+      "grad_norm": 6.445465882152348e-05,
+      "learning_rate": 9.20908631776102e-06,
+      "loss": 0.0,
+      "step": 868
+    },
+    {
+      "epoch": 0.0560030933814526,
+      "grad_norm": 0.01036632348269938,
+      "learning_rate": 9.21065343854272e-06,
+      "loss": 0.0001,
+      "step": 869
+    },
+    {
+      "epoch": 0.05606753882838177,
+      "grad_norm": 0.03182250548398591,
+      "learning_rate": 9.21221875699995e-06,
+      "loss": 0.0001,
+      "step": 870
+    },
+    {
+      "epoch": 0.05613198427531095,
+      "grad_norm": 0.0027731863618260023,
+      "learning_rate": 9.213782277273607e-06,
+      "loss": 0.0,
+      "step": 871
+    },
+    {
+      "epoch": 0.05619642972224012,
+      "grad_norm": 0.02141632130848708,
+      "learning_rate": 9.215344003490328e-06,
+      "loss": 0.0,
+      "step": 872
+    },
+    {
+      "epoch": 0.0562608751691693,
+      "grad_norm": 0.00841137676276659,
+      "learning_rate": 9.21690393976257e-06,
+      "loss": 0.0,
+      "step": 873
+    },
+    {
+      "epoch": 0.056325320616098476,
+      "grad_norm": 0.0026180638976685206,
+      "learning_rate": 9.21846209018866e-06,
+      "loss": 0.0,
+      "step": 874
+    },
+    {
+      "epoch": 0.056389766063027646,
+      "grad_norm": 0.1417267476215166,
+      "learning_rate": 9.220018458852871e-06,
+      "loss": 0.0002,
+      "step": 875
+    },
+    {
+      "epoch": 0.056454211509956824,
+      "grad_norm": 0.0016003214834285813,
+      "learning_rate": 9.221573049825472e-06,
+      "loss": 0.0,
+      "step": 876
+    },
+    {
+      "epoch": 0.056518656956885995,
+      "grad_norm": 0.0008617063924082198,
+      "learning_rate": 9.223125867162807e-06,
+      "loss": 0.0,
+      "step": 877
+    },
+    {
+      "epoch": 0.05658310240381517,
+      "grad_norm": 0.00019133509686306216,
+      "learning_rate": 9.224676914907353e-06,
+      "loss": 0.0,
+      "step": 878
+    },
+    {
+      "epoch": 0.05664754785074434,
+      "grad_norm": 0.09435838250675625,
+      "learning_rate": 9.226226197087778e-06,
+      "loss": 0.0009,
+      "step": 879
+    },
+    {
+      "epoch": 0.05671199329767352,
+      "grad_norm": 0.002137334572342066,
+      "learning_rate": 9.22777371771901e-06,
+      "loss": 0.0,
+      "step": 880
+    },
+    {
+      "epoch": 0.05677643874460269,
+      "grad_norm": 0.00500561291614537,
+      "learning_rate": 9.229319480802301e-06,
+      "loss": 0.0,
+      "step": 881
+    },
+    {
+      "epoch": 0.05684088419153187,
+      "grad_norm": 0.0018617593146569935,
+      "learning_rate": 9.230863490325278e-06,
+      "loss": 0.0,
+      "step": 882
+    },
+    {
+      "epoch": 0.056905329638461045,
+      "grad_norm": 0.003249829210823486,
+      "learning_rate": 9.232405750262018e-06,
+      "loss": 0.0,
+      "step": 883
+    },
+    {
+      "epoch": 0.056969775085390216,
+      "grad_norm": 4.155492727945956,
+      "learning_rate": 9.233946264573107e-06,
+      "loss": 0.0309,
+      "step": 884
+    },
+    {
+      "epoch": 0.057034220532319393,
+      "grad_norm": 0.0004600202922205456,
+      "learning_rate": 9.235485037205686e-06,
+      "loss": 0.0,
+      "step": 885
+    },
+    {
+      "epoch": 0.057098665979248564,
+      "grad_norm": 0.0009887524614545788,
+      "learning_rate": 9.237022072093532e-06,
+      "loss": 0.0,
+      "step": 886
+    },
+    {
+      "epoch": 0.05716311142617774,
+      "grad_norm": 0.12449591667280237,
+      "learning_rate": 9.238557373157111e-06,
+      "loss": 0.0002,
+      "step": 887
+    },
+    {
+      "epoch": 0.05722755687310691,
+      "grad_norm": 0.015552755601544609,
+      "learning_rate": 9.240090944303633e-06,
+      "loss": 0.0001,
+      "step": 888
+    },
+    {
+      "epoch": 0.05729200232003609,
+      "grad_norm": 0.23141060741991862,
+      "learning_rate": 9.241622789427116e-06,
+      "loss": 0.0004,
+      "step": 889
+    },
+    {
+      "epoch": 0.05735644776696527,
+      "grad_norm": 0.8822941904283358,
+      "learning_rate": 9.243152912408443e-06,
+      "loss": 0.0065,
+      "step": 890
+    },
+    {
+      "epoch": 0.05742089321389444,
+      "grad_norm": 0.00029090730639586807,
+      "learning_rate": 9.244681317115427e-06,
+      "loss": 0.0,
+      "step": 891
+    },
+    {
+      "epoch": 0.057485338660823615,
+      "grad_norm": 0.001579554953343648,
+      "learning_rate": 9.246208007402863e-06,
+      "loss": 0.0,
+      "step": 892
+    },
+    {
+      "epoch": 0.057549784107752786,
+      "grad_norm": 0.023213539630016298,
+      "learning_rate": 9.247732987112587e-06,
+      "loss": 0.0001,
+      "step": 893
+    },
+    {
+      "epoch": 0.05761422955468196,
+      "grad_norm": 0.0024890347909475733,
+      "learning_rate": 9.249256260073538e-06,
+      "loss": 0.0,
+      "step": 894
+    },
+    {
+      "epoch": 0.057678675001611134,
+      "grad_norm": 0.000585152467749838,
+      "learning_rate": 9.250777830101814e-06,
+      "loss": 0.0,
+      "step": 895
+    },
+    {
+      "epoch": 0.05774312044854031,
+      "grad_norm": 0.028285490200250774,
+      "learning_rate": 9.252297701000728e-06,
+      "loss": 0.0,
+      "step": 896
+    },
+    {
+      "epoch": 0.05780756589546949,
+      "grad_norm": 0.10915553690810925,
+      "learning_rate": 9.253815876560862e-06,
+      "loss": 0.0008,
+      "step": 897
+    },
+    {
+      "epoch": 0.05787201134239866,
+      "grad_norm": 0.0002786805844268407,
+      "learning_rate": 9.255332360560135e-06,
+      "loss": 0.0,
+      "step": 898
+    },
+    {
+      "epoch": 0.057936456789327836,
+      "grad_norm": 0.005925699088651387,
+      "learning_rate": 9.256847156763845e-06,
+      "loss": 0.0001,
+      "step": 899
+    },
+    {
+      "epoch": 0.05800090223625701,
+      "grad_norm": 0.0033125726030387532,
+      "learning_rate": 9.258360268924736e-06,
+      "loss": 0.0,
+      "step": 900
+    },
+    {
+      "epoch": 0.058065347683186185,
+      "grad_norm": 0.010476595702934775,
+      "learning_rate": 9.259871700783048e-06,
+      "loss": 0.0,
+      "step": 901
+    },
+    {
+      "epoch": 0.058129793130115355,
+      "grad_norm": 0.02884480152810691,
+      "learning_rate": 9.261381456066576e-06,
+      "loss": 0.0001,
+      "step": 902
+    },
+    {
+      "epoch": 0.05819423857704453,
+      "grad_norm": 0.02782736705714902,
+      "learning_rate": 9.26288953849072e-06,
+      "loss": 0.0003,
+      "step": 903
+    },
+    {
+      "epoch": 0.0582586840239737,
+      "grad_norm": 0.10276730883429447,
+      "learning_rate": 9.264395951758548e-06,
+      "loss": 0.0014,
+      "step": 904
+    },
+    {
+      "epoch": 0.05832312947090288,
+      "grad_norm": 0.021272532663555836,
+      "learning_rate": 9.265900699560844e-06,
+      "loss": 0.0,
+      "step": 905
+    },
+    {
+      "epoch": 0.05838757491783206,
+      "grad_norm": 0.32919439257374566,
+      "learning_rate": 9.267403785576162e-06,
+      "loss": 0.0004,
+      "step": 906
+    },
+    {
+      "epoch": 0.05845202036476123,
+      "grad_norm": 0.03212588290219352,
+      "learning_rate": 9.268905213470882e-06,
+      "loss": 0.0001,
+      "step": 907
+    },
+    {
+      "epoch": 0.058516465811690406,
+      "grad_norm": 0.005368526072576281,
+      "learning_rate": 9.270404986899269e-06,
+      "loss": 0.0,
+      "step": 908
+    },
+    {
+      "epoch": 0.05858091125861958,
+      "grad_norm": 0.004630751419976039,
+      "learning_rate": 9.271903109503515e-06,
+      "loss": 0.0,
+      "step": 909
+    },
+    {
+      "epoch": 0.058645356705548754,
+      "grad_norm": 0.04599784677651563,
+      "learning_rate": 9.273399584913803e-06,
+      "loss": 0.0001,
+      "step": 910
+    },
+    {
+      "epoch": 0.058709802152477925,
+      "grad_norm": 0.14592824909655777,
+      "learning_rate": 9.274894416748349e-06,
+      "loss": 0.0019,
+      "step": 911
+    },
+    {
+      "epoch": 0.0587742475994071,
+      "grad_norm": 0.165071235996826,
+      "learning_rate": 9.276387608613467e-06,
+      "loss": 0.0003,
+      "step": 912
+    },
+    {
+      "epoch": 0.05883869304633628,
+      "grad_norm": 0.0045161075821298306,
+      "learning_rate": 9.27787916410361e-06,
+      "loss": 0.0,
+      "step": 913
+    },
+    {
+      "epoch": 0.05890313849326545,
+      "grad_norm": 0.0609009449695222,
+      "learning_rate": 9.279369086801427e-06,
+      "loss": 0.0001,
+      "step": 914
+    },
+    {
+      "epoch": 0.05896758394019463,
+      "grad_norm": 0.01850613402613245,
+      "learning_rate": 9.28085738027782e-06,
+      "loss": 0.0001,
+      "step": 915
+    },
+    {
+      "epoch": 0.0590320293871238,
+      "grad_norm": 0.15463124032407352,
+      "learning_rate": 9.28234404809198e-06,
+      "loss": 0.0007,
+      "step": 916
+    },
+    {
+      "epoch": 0.059096474834052976,
+      "grad_norm": 0.003664506187491668,
+      "learning_rate": 9.283829093791457e-06,
+      "loss": 0.0,
+      "step": 917
+    },
+    {
+      "epoch": 0.059160920280982146,
+      "grad_norm": 0.00880322416467816,
+      "learning_rate": 9.285312520912193e-06,
+      "loss": 0.0001,
+      "step": 918
+    },
+    {
+      "epoch": 0.059225365727911324,
+      "grad_norm": 3.7186074109349265,
+      "learning_rate": 9.286794332978585e-06,
+      "loss": 0.0237,
+      "step": 919
+    },
+    {
+      "epoch": 0.059289811174840494,
+      "grad_norm": 8.817468231810405e-05,
+      "learning_rate": 9.288274533503533e-06,
+      "loss": 0.0,
+      "step": 920
+    },
+    {
+      "epoch": 0.05935425662176967,
+      "grad_norm": 0.004592331110218876,
+      "learning_rate": 9.289753125988485e-06,
+      "loss": 0.0,
+      "step": 921
+    },
+    {
+      "epoch": 0.05941870206869885,
+      "grad_norm": 1.0849440167306779,
+      "learning_rate": 9.291230113923492e-06,
+      "loss": 0.0055,
+      "step": 922
+    },
+    {
+      "epoch": 0.05948314751562802,
+      "grad_norm": 0.0024849397582142455,
+      "learning_rate": 9.29270550078725e-06,
+      "loss": 0.0,
+      "step": 923
+    },
+    {
+      "epoch": 0.0595475929625572,
+      "grad_norm": 0.9088584914373267,
+      "learning_rate": 9.294179290047164e-06,
+      "loss": 0.0059,
+      "step": 924
+    },
+    {
+      "epoch": 0.05961203840948637,
+      "grad_norm": 0.00017863757655568402,
+      "learning_rate": 9.29565148515938e-06,
+      "loss": 0.0,
+      "step": 925
+    },
+    {
+      "epoch": 0.059676483856415545,
+      "grad_norm": 0.16478213790474924,
+      "learning_rate": 9.29712208956884e-06,
+      "loss": 0.0028,
+      "step": 926
+    },
+    {
+      "epoch": 0.059740929303344716,
+      "grad_norm": 0.009325136506692883,
+      "learning_rate": 9.298591106709336e-06,
+      "loss": 0.0,
+      "step": 927
+    },
+    {
+      "epoch": 0.05980537475027389,
+      "grad_norm": 0.0018313403949757983,
+      "learning_rate": 9.300058540003553e-06,
+      "loss": 0.0,
+      "step": 928
+    },
+    {
+      "epoch": 0.05986982019720307,
+      "grad_norm": 0.036417811764517145,
+      "learning_rate": 9.301524392863113e-06,
+      "loss": 0.0001,
+      "step": 929
+    },
+    {
+      "epoch": 0.05993426564413224,
+      "grad_norm": 1.021395560902065,
+      "learning_rate": 9.30298866868863e-06,
+      "loss": 0.0049,
+      "step": 930
+    },
+    {
+      "epoch": 0.05999871109106142,
+      "grad_norm": 0.0024827359936627643,
+      "learning_rate": 9.304451370869755e-06,
+      "loss": 0.0,
+      "step": 931
+    },
+    {
+      "epoch": 0.06006315653799059,
+      "grad_norm": 0.7179352247112939,
+      "learning_rate": 9.30591250278522e-06,
+      "loss": 0.0021,
+      "step": 932
+    },
+    {
+      "epoch": 0.06012760198491977,
+      "grad_norm": 0.0631933717664277,
+      "learning_rate": 9.307372067802889e-06,
+      "loss": 0.0003,
+      "step": 933
+    },
+    {
+      "epoch": 0.06019204743184894,
+      "grad_norm": 0.060907736335031146,
+      "learning_rate": 9.308830069279799e-06,
+      "loss": 0.0004,
+      "step": 934
+    },
+    {
+      "epoch": 0.060256492878778115,
+      "grad_norm": 0.0008149091579210628,
+      "learning_rate": 9.310286510562214e-06,
+      "loss": 0.0,
+      "step": 935
+    },
+    {
+      "epoch": 0.06032093832570729,
+      "grad_norm": 0.006663428570939788,
+      "learning_rate": 9.311741394985666e-06,
+      "loss": 0.0,
+      "step": 936
+    },
+    {
+      "epoch": 0.06038538377263646,
+      "grad_norm": 0.12337773360999044,
+      "learning_rate": 9.313194725875004e-06,
+      "loss": 0.001,
+      "step": 937
+    },
+    {
+      "epoch": 0.06044982921956564,
+      "grad_norm": 0.19178069720437993,
+      "learning_rate": 9.31464650654443e-06,
+      "loss": 0.0005,
+      "step": 938
+    },
+    {
+      "epoch": 0.06051427466649481,
+      "grad_norm": 0.0015436894352676542,
+      "learning_rate": 9.316096740297561e-06,
+      "loss": 0.0,
+      "step": 939
+    },
+    {
+      "epoch": 0.06057872011342399,
+      "grad_norm": 0.15529906463269208,
+      "learning_rate": 9.317545430427458e-06,
+      "loss": 0.0009,
+      "step": 940
+    },
+    {
+      "epoch": 0.06064316556035316,
+      "grad_norm": 0.15753065047540682,
+      "learning_rate": 9.318992580216683e-06,
+      "loss": 0.0025,
+      "step": 941
+    },
+    {
+      "epoch": 0.060707611007282336,
+      "grad_norm": 0.0037847305096929324,
+      "learning_rate": 9.320438192937334e-06,
+      "loss": 0.0,
+      "step": 942
+    },
+    {
+      "epoch": 0.06077205645421151,
+      "grad_norm": 0.560417957966202,
+      "learning_rate": 9.321882271851098e-06,
+      "loss": 0.0015,
+      "step": 943
+    },
+    {
+      "epoch": 0.060836501901140684,
+      "grad_norm": 0.0015783399253200654,
+      "learning_rate": 9.323324820209286e-06,
+      "loss": 0.0,
+      "step": 944
+    },
+    {
+      "epoch": 0.06090094734806986,
+      "grad_norm": 0.08633490002918111,
+      "learning_rate": 9.32476584125289e-06,
+      "loss": 0.0003,
+      "step": 945
+    },
+    {
+      "epoch": 0.06096539279499903,
+      "grad_norm": 0.17250897245119487,
+      "learning_rate": 9.326205338212606e-06,
+      "loss": 0.0006,
+      "step": 946
+    },
+    {
+      "epoch": 0.06102983824192821,
+      "grad_norm": 0.11453406446909774,
+      "learning_rate": 9.327643314308903e-06,
+      "loss": 0.0003,
+      "step": 947
+    },
+    {
+      "epoch": 0.06109428368885738,
+      "grad_norm": 0.3036814978949101,
+      "learning_rate": 9.329079772752047e-06,
+      "loss": 0.0015,
+      "step": 948
+    },
+    {
+      "epoch": 0.06115872913578656,
+      "grad_norm": 0.011376020798526345,
+      "learning_rate": 9.33051471674215e-06,
+      "loss": 0.0,
+      "step": 949
+    },
+    {
+      "epoch": 0.06122317458271573,
+      "grad_norm": 0.012309696487391423,
+      "learning_rate": 9.331948149469212e-06,
+      "loss": 0.0001,
+      "step": 950
+    },
+    {
+      "epoch": 0.061287620029644906,
+      "grad_norm": 0.006174902085906057,
+      "learning_rate": 9.33338007411317e-06,
+      "loss": 0.0,
+      "step": 951
+    },
+    {
+      "epoch": 0.06135206547657408,
+      "grad_norm": 0.19661972321282462,
+      "learning_rate": 9.33481049384393e-06,
+      "loss": 0.0008,
+      "step": 952
+    },
+    {
+      "epoch": 0.061416510923503254,
+      "grad_norm": 0.005731411898032642,
+      "learning_rate": 9.336239411821416e-06,
+      "loss": 0.0,
+      "step": 953
+    },
+    {
+      "epoch": 0.06148095637043243,
+      "grad_norm": 0.6383001896675805,
+      "learning_rate": 9.33766683119561e-06,
+      "loss": 0.004,
+      "step": 954
+    },
+    {
+      "epoch": 0.0615454018173616,
+      "grad_norm": 0.0018123395465073323,
+      "learning_rate": 9.33909275510659e-06,
+      "loss": 0.0,
+      "step": 955
+    },
+    {
+      "epoch": 0.06160984726429078,
+      "grad_norm": 0.5576655806137351,
+      "learning_rate": 9.340517186684582e-06,
+      "loss": 0.002,
+      "step": 956
+    },
+    {
+      "epoch": 0.06167429271121995,
+      "grad_norm": 0.11143409554222802,
+      "learning_rate": 9.34194012904999e-06,
+      "loss": 0.0011,
+      "step": 957
+    },
+    {
+      "epoch": 0.06173873815814913,
+      "grad_norm": 0.163548453280282,
+      "learning_rate": 9.34336158531344e-06,
+      "loss": 0.0002,
+      "step": 958
+    },
+    {
+      "epoch": 0.061803183605078305,
+      "grad_norm": 0.0179151870815518,
+      "learning_rate": 9.344781558575823e-06,
+      "loss": 0.0,
+      "step": 959
+    },
+    {
+      "epoch": 0.061867629052007475,
+      "grad_norm": 0.00805664830135211,
+      "learning_rate": 9.346200051928338e-06,
+      "loss": 0.0001,
+      "step": 960
+    },
+    {
+      "epoch": 0.06193207449893665,
+      "grad_norm": 0.035669773452513494,
+      "learning_rate": 9.347617068452525e-06,
+      "loss": 0.0001,
+      "step": 961
+    },
+    {
+      "epoch": 0.06199651994586582,
+      "grad_norm": 0.016276996306397545,
+      "learning_rate": 9.34903261122031e-06,
+      "loss": 0.0001,
+      "step": 962
+    },
+    {
+      "epoch": 0.062060965392795,
+      "grad_norm": 0.006962232185746814,
+      "learning_rate": 9.350446683294047e-06,
+      "loss": 0.0,
+      "step": 963
+    },
+    {
+      "epoch": 0.06212541083972417,
+      "grad_norm": 0.017856100377398786,
+      "learning_rate": 9.35185928772655e-06,
+      "loss": 0.0,
+      "step": 964
+    },
+    {
+      "epoch": 0.06218985628665335,
+      "grad_norm": 0.25958305769630685,
+      "learning_rate": 9.353270427561143e-06,
+      "loss": 0.0011,
+      "step": 965
+    },
+    {
+      "epoch": 0.06225430173358252,
+      "grad_norm": 0.06486360650927218,
+      "learning_rate": 9.354680105831685e-06,
+      "loss": 0.0001,
+      "step": 966
+    },
+    {
+      "epoch": 0.0623187471805117,
+      "grad_norm": 3.134478879749426,
+      "learning_rate": 9.356088325562633e-06,
+      "loss": 0.0352,
+      "step": 967
+    },
+    {
+      "epoch": 0.062383192627440874,
+      "grad_norm": 0.19477207757671616,
+      "learning_rate": 9.35749508976905e-06,
+      "loss": 0.0002,
+      "step": 968
+    },
+    {
+      "epoch": 0.062447638074370045,
+      "grad_norm": 0.0303179640337168,
+      "learning_rate": 9.35890040145667e-06,
+      "loss": 0.0,
+      "step": 969
+    },
+    {
+      "epoch": 0.06251208352129922,
+      "grad_norm": 0.22131384688029207,
+      "learning_rate": 9.360304263621919e-06,
+      "loss": 0.0006,
+      "step": 970
+    },
+    {
+      "epoch": 0.06257652896822839,
+      "grad_norm": 0.021000072329624038,
+      "learning_rate": 9.361706679251967e-06,
+      "loss": 0.0,
+      "step": 971
+    },
+    {
+      "epoch": 0.06264097441515756,
+      "grad_norm": 0.017301594668236765,
+      "learning_rate": 9.363107651324754e-06,
+      "loss": 0.0,
+      "step": 972
+    },
+    {
+      "epoch": 0.06270541986208675,
+      "grad_norm": 0.005622783541268775,
+      "learning_rate": 9.364507182809037e-06,
+      "loss": 0.0,
+      "step": 973
+    },
+    {
+      "epoch": 0.06276986530901592,
+      "grad_norm": 0.23653247762425234,
+      "learning_rate": 9.365905276664425e-06,
+      "loss": 0.0012,
+      "step": 974
+    },
+    {
+      "epoch": 0.06283431075594509,
+      "grad_norm": 0.0019284558066157138,
+      "learning_rate": 9.367301935841413e-06,
+      "loss": 0.0,
+      "step": 975
+    },
+    {
+      "epoch": 0.06289875620287427,
+      "grad_norm": 0.013986864143152953,
+      "learning_rate": 9.368697163281422e-06,
+      "loss": 0.0,
+      "step": 976
+    },
+    {
+      "epoch": 0.06296320164980344,
+      "grad_norm": 0.0022328048482544325,
+      "learning_rate": 9.37009096191684e-06,
+      "loss": 0.0,
+      "step": 977
+    },
+    {
+      "epoch": 0.06302764709673261,
+      "grad_norm": 0.05739005222315148,
+      "learning_rate": 9.371483334671052e-06,
+      "loss": 0.0001,
+      "step": 978
+    },
+    {
+      "epoch": 0.06309209254366178,
+      "grad_norm": 0.0010009926174638238,
+      "learning_rate": 9.372874284458484e-06,
+      "loss": 0.0,
+      "step": 979
+    },
+    {
+      "epoch": 0.06315653799059097,
+      "grad_norm": 0.002213090174786511,
+      "learning_rate": 9.374263814184626e-06,
+      "loss": 0.0,
+      "step": 980
+    },
+    {
+      "epoch": 0.06322098343752014,
+      "grad_norm": 0.037727668382915094,
+      "learning_rate": 9.375651926746093e-06,
+      "loss": 0.0,
+      "step": 981
+    },
+    {
+      "epoch": 0.06328542888444931,
+      "grad_norm": 0.5743290801677684,
+      "learning_rate": 9.377038625030634e-06,
+      "loss": 0.0035,
+      "step": 982
+    },
+    {
+      "epoch": 0.0633498743313785,
+      "grad_norm": 0.2047254163261985,
+      "learning_rate": 9.378423911917185e-06,
+      "loss": 0.0007,
+      "step": 983
+    },
+    {
+      "epoch": 0.06341431977830767,
+      "grad_norm": 0.001514966733781814,
+      "learning_rate": 9.379807790275902e-06,
+      "loss": 0.0,
+      "step": 984
+    },
+    {
+      "epoch": 0.06347876522523684,
+      "grad_norm": 0.025498008379182133,
+      "learning_rate": 9.381190262968194e-06,
+      "loss": 0.0001,
+      "step": 985
+    },
+    {
+      "epoch": 0.063543210672166,
+      "grad_norm": 0.009340591807369656,
+      "learning_rate": 9.382571332846755e-06,
+      "loss": 0.0,
+      "step": 986
+    },
+    {
+      "epoch": 0.06360765611909519,
+      "grad_norm": 0.001099268463229261,
+      "learning_rate": 9.383951002755613e-06,
+      "loss": 0.0,
+      "step": 987
+    },
+    {
+      "epoch": 0.06367210156602436,
+      "grad_norm": 0.21566121032299157,
+      "learning_rate": 9.385329275530144e-06,
+      "loss": 0.0006,
+      "step": 988
+    },
+    {
+      "epoch": 0.06373654701295353,
+      "grad_norm": 0.0011949991086739977,
+      "learning_rate": 9.38670615399713e-06,
+      "loss": 0.0,
+      "step": 989
+    },
+    {
+      "epoch": 0.06380099245988272,
+      "grad_norm": 0.001856079925364757,
+      "learning_rate": 9.388081640974775e-06,
+      "loss": 0.0,
+      "step": 990
+    },
+    {
+      "epoch": 0.06386543790681189,
+      "grad_norm": 0.618543390978522,
+      "learning_rate": 9.389455739272754e-06,
+      "loss": 0.0067,
+      "step": 991
+    },
+    {
+      "epoch": 0.06392988335374106,
+      "grad_norm": 0.00102082338563046,
+      "learning_rate": 9.390828451692232e-06,
+      "loss": 0.0,
+      "step": 992
+    },
+    {
+      "epoch": 0.06399432880067023,
+      "grad_norm": 0.0004849908016925305,
+      "learning_rate": 9.392199781025916e-06,
+      "loss": 0.0,
+      "step": 993
+    },
+    {
+      "epoch": 0.06405877424759941,
+      "grad_norm": 0.006964614517327541,
+      "learning_rate": 9.393569730058076e-06,
+      "loss": 0.0,
+      "step": 994
+    },
+    {
+      "epoch": 0.06412321969452858,
+      "grad_norm": 0.0680559337719936,
+      "learning_rate": 9.394938301564581e-06,
+      "loss": 0.0001,
+      "step": 995
+    },
+    {
+      "epoch": 0.06418766514145775,
+      "grad_norm": 0.02046442622393038,
+      "learning_rate": 9.396305498312936e-06,
+      "loss": 0.0001,
+      "step": 996
+    },
+    {
+      "epoch": 0.06425211058838692,
+      "grad_norm": 0.022657457270191042,
+      "learning_rate": 9.397671323062319e-06,
+      "loss": 0.0002,
+      "step": 997
+    },
+    {
+      "epoch": 0.06431655603531611,
+      "grad_norm": 0.0010187417482833032,
+      "learning_rate": 9.399035778563601e-06,
+      "loss": 0.0,
+      "step": 998
+    },
+    {
+      "epoch": 0.06438100148224528,
+      "grad_norm": 0.011742133910265604,
+      "learning_rate": 9.400398867559398e-06,
+      "loss": 0.0001,
+      "step": 999
+    },
+    {
+      "epoch": 0.06444544692917445,
+      "grad_norm": 0.013586182883457138,
+      "learning_rate": 9.401760592784085e-06,
+      "loss": 0.0,
+      "step": 1000
+    },
+    {
+      "epoch": 0.06450989237610363,
+      "grad_norm": 0.2551176516648852,
+      "learning_rate": 9.403120956963842e-06,
+      "loss": 0.0034,
+      "step": 1001
+    },
+    {
+      "epoch": 0.0645743378230328,
+      "grad_norm": 0.03086066982715626,
+      "learning_rate": 9.404479962816683e-06,
+      "loss": 0.0001,
+      "step": 1002
+    },
+    {
+      "epoch": 0.06463878326996197,
+      "grad_norm": 0.2412145436747094,
+      "learning_rate": 9.405837613052491e-06,
+      "loss": 0.0006,
+      "step": 1003
+    },
+    {
+      "epoch": 0.06470322871689115,
+      "grad_norm": 0.037208169947243144,
+      "learning_rate": 9.407193910373039e-06,
+      "loss": 0.0004,
+      "step": 1004
+    },
+    {
+      "epoch": 0.06476767416382033,
+      "grad_norm": 0.0024333425810042523,
+      "learning_rate": 9.40854885747204e-06,
+      "loss": 0.0,
+      "step": 1005
+    },
+    {
+      "epoch": 0.0648321196107495,
+      "grad_norm": 0.08621493714841584,
+      "learning_rate": 9.409902457035166e-06,
+      "loss": 0.0002,
+      "step": 1006
+    },
+    {
+      "epoch": 0.06489656505767867,
+      "grad_norm": 0.02889134757488118,
+      "learning_rate": 9.411254711740086e-06,
+      "loss": 0.0004,
+      "step": 1007
+    },
+    {
+      "epoch": 0.06496101050460786,
+      "grad_norm": 0.228104563999611,
+      "learning_rate": 9.412605624256493e-06,
+      "loss": 0.0019,
+      "step": 1008
+    },
+    {
+      "epoch": 0.06502545595153703,
+      "grad_norm": 0.015025817973177263,
+      "learning_rate": 9.413955197246137e-06,
+      "loss": 0.0001,
+      "step": 1009
+    },
+    {
+      "epoch": 0.0650899013984662,
+      "grad_norm": 0.8140691497003968,
+      "learning_rate": 9.415303433362863e-06,
+      "loss": 0.0038,
+      "step": 1010
+    },
+    {
+      "epoch": 0.06515434684539537,
+      "grad_norm": 0.004443188188472599,
+      "learning_rate": 9.416650335252638e-06,
+      "loss": 0.0,
+      "step": 1011
+    },
+    {
+      "epoch": 0.06521879229232455,
+      "grad_norm": 0.16307514977324722,
+      "learning_rate": 9.417995905553573e-06,
+      "loss": 0.0004,
+      "step": 1012
+    },
+    {
+      "epoch": 0.06528323773925372,
+      "grad_norm": 0.08674513548595156,
+      "learning_rate": 9.419340146895971e-06,
+      "loss": 0.0003,
+      "step": 1013
+    },
+    {
+      "epoch": 0.06534768318618289,
+      "grad_norm": 0.0863520286012186,
+      "learning_rate": 9.420683061902345e-06,
+      "loss": 0.0004,
+      "step": 1014
+    },
+    {
+      "epoch": 0.06541212863311208,
+      "grad_norm": 0.0015631516269431248,
+      "learning_rate": 9.422024653187451e-06,
+      "loss": 0.0,
+      "step": 1015
+    },
+    {
+      "epoch": 0.06547657408004125,
+      "grad_norm": 0.0009583769723558712,
+      "learning_rate": 9.423364923358329e-06,
+      "loss": 0.0,
+      "step": 1016
+    },
+    {
+      "epoch": 0.06554101952697042,
+      "grad_norm": 0.3525857656721515,
+      "learning_rate": 9.424703875014313e-06,
+      "loss": 0.0033,
+      "step": 1017
+    },
+    {
+      "epoch": 0.06560546497389959,
+      "grad_norm": 0.01980719319717165,
+      "learning_rate": 9.426041510747082e-06,
+      "loss": 0.0001,
+      "step": 1018
+    },
+    {
+      "epoch": 0.06566991042082877,
+      "grad_norm": 0.05776367316483845,
+      "learning_rate": 9.42737783314068e-06,
+      "loss": 0.0019,
+      "step": 1019
+    },
+    {
+      "epoch": 0.06573435586775794,
+      "grad_norm": 0.029714163614339636,
+      "learning_rate": 9.42871284477154e-06,
+      "loss": 0.0001,
+      "step": 1020
+    },
+    {
+      "epoch": 0.06579880131468711,
+      "grad_norm": 0.1090029194822556,
+      "learning_rate": 9.430046548208533e-06,
+      "loss": 0.0002,
+      "step": 1021
+    },
+    {
+      "epoch": 0.0658632467616163,
+      "grad_norm": 0.79815705332845,
+      "learning_rate": 9.431378946012973e-06,
+      "loss": 0.0021,
+      "step": 1022
+    },
+    {
+      "epoch": 0.06592769220854547,
+      "grad_norm": 0.12051999387583368,
+      "learning_rate": 9.432710040738669e-06,
+      "loss": 0.0002,
+      "step": 1023
+    },
+    {
+      "epoch": 0.06599213765547464,
+      "grad_norm": 0.009057564683849412,
+      "learning_rate": 9.434039834931941e-06,
+      "loss": 0.0001,
+      "step": 1024
+    },
+    {
+      "epoch": 0.06605658310240381,
+      "grad_norm": 0.0362359697607633,
+      "learning_rate": 9.43536833113165e-06,
+      "loss": 0.0001,
+      "step": 1025
+    },
+    {
+      "epoch": 0.066121028549333,
+      "grad_norm": 0.029041901339865995,
+      "learning_rate": 9.436695531869232e-06,
+      "loss": 0.0001,
+      "step": 1026
+    },
+    {
+      "epoch": 0.06618547399626216,
+      "grad_norm": 0.005182049662164661,
+      "learning_rate": 9.438021439668724e-06,
+      "loss": 0.0,
+      "step": 1027
+    },
+    {
+      "epoch": 0.06624991944319134,
+      "grad_norm": 0.005158991102843603,
+      "learning_rate": 9.439346057046797e-06,
+      "loss": 0.0001,
+      "step": 1028
+    },
+    {
+      "epoch": 0.06631436489012052,
+      "grad_norm": 0.4060666330698772,
+      "learning_rate": 9.44066938651278e-06,
+      "loss": 0.0057,
+      "step": 1029
+    },
+    {
+      "epoch": 0.06637881033704969,
+      "grad_norm": 0.0013174037296815595,
+      "learning_rate": 9.441991430568686e-06,
+      "loss": 0.0,
+      "step": 1030
+    },
+    {
+      "epoch": 0.06644325578397886,
+      "grad_norm": 0.0033868625919964246,
+      "learning_rate": 9.443312191709244e-06,
+      "loss": 0.0,
+      "step": 1031
+    },
+    {
+      "epoch": 0.06650770123090803,
+      "grad_norm": 0.006117924134115885,
+      "learning_rate": 9.444631672421934e-06,
+      "loss": 0.0001,
+      "step": 1032
+    },
+    {
+      "epoch": 0.06657214667783722,
+      "grad_norm": 0.13950942529862392,
+      "learning_rate": 9.445949875187002e-06,
+      "loss": 0.0019,
+      "step": 1033
+    },
+    {
+      "epoch": 0.06663659212476639,
+      "grad_norm": 0.01851590001806397,
+      "learning_rate": 9.447266802477499e-06,
+      "loss": 0.0004,
+      "step": 1034
+    },
+    {
+      "epoch": 0.06670103757169556,
+      "grad_norm": 0.0016811966371559576,
+      "learning_rate": 9.448582456759298e-06,
+      "loss": 0.0,
+      "step": 1035
+    },
+    {
+      "epoch": 0.06676548301862473,
+      "grad_norm": 0.1653637147615305,
+      "learning_rate": 9.449896840491135e-06,
+      "loss": 0.0022,
+      "step": 1036
+    },
+    {
+      "epoch": 0.06682992846555391,
+      "grad_norm": 0.022010529514772826,
+      "learning_rate": 9.451209956124626e-06,
+      "loss": 0.0001,
+      "step": 1037
+    },
+    {
+      "epoch": 0.06689437391248308,
+      "grad_norm": 0.0071250187018150595,
+      "learning_rate": 9.452521806104297e-06,
+      "loss": 0.0,
+      "step": 1038
+    },
+    {
+      "epoch": 0.06695881935941225,
+      "grad_norm": 0.02253860670889963,
+      "learning_rate": 9.453832392867618e-06,
+      "loss": 0.0002,
+      "step": 1039
+    },
+    {
+      "epoch": 0.06702326480634144,
+      "grad_norm": 0.024827336967876906,
+      "learning_rate": 9.455141718845016e-06,
+      "loss": 0.0001,
+      "step": 1040
+    },
+    {
+      "epoch": 0.06708771025327061,
+      "grad_norm": 0.028985878209440076,
+      "learning_rate": 9.456449786459916e-06,
+      "loss": 0.0001,
+      "step": 1041
+    },
+    {
+      "epoch": 0.06715215570019978,
+      "grad_norm": 0.06560310114761658,
+      "learning_rate": 9.45775659812876e-06,
+      "loss": 0.0002,
+      "step": 1042
+    },
+    {
+      "epoch": 0.06721660114712895,
+      "grad_norm": 0.007292826751741267,
+      "learning_rate": 9.459062156261041e-06,
+      "loss": 0.0,
+      "step": 1043
+    },
+    {
+      "epoch": 0.06728104659405813,
+      "grad_norm": 0.001488974553572595,
+      "learning_rate": 9.460366463259316e-06,
+      "loss": 0.0,
+      "step": 1044
+    },
+    {
+      "epoch": 0.0673454920409873,
+      "grad_norm": 0.058802188234950835,
+      "learning_rate": 9.461669521519253e-06,
+      "loss": 0.0001,
+      "step": 1045
+    },
+    {
+      "epoch": 0.06740993748791647,
+      "grad_norm": 0.015528356733507939,
+      "learning_rate": 9.462971333429634e-06,
+      "loss": 0.0,
+      "step": 1046
+    },
+    {
+      "epoch": 0.06747438293484566,
+      "grad_norm": 0.010762498561983332,
+      "learning_rate": 9.4642719013724e-06,
+      "loss": 0.0001,
+      "step": 1047
+    },
+    {
+      "epoch": 0.06753882838177483,
+      "grad_norm": 0.027761143204603284,
+      "learning_rate": 9.46557122772267e-06,
+      "loss": 0.0001,
+      "step": 1048
+    },
+    {
+      "epoch": 0.067603273828704,
+      "grad_norm": 0.13193905647539747,
+      "learning_rate": 9.466869314848766e-06,
+      "loss": 0.0001,
+      "step": 1049
+    },
+    {
+      "epoch": 0.06766771927563317,
+      "grad_norm": 0.06437447835886814,
+      "learning_rate": 9.468166165112236e-06,
+      "loss": 0.0,
+      "step": 1050
+    },
+    {
+      "epoch": 0.06773216472256235,
+      "grad_norm": 0.39598322422926224,
+      "learning_rate": 9.469461780867895e-06,
+      "loss": 0.0015,
+      "step": 1051
+    },
+    {
+      "epoch": 0.06779661016949153,
+      "grad_norm": 0.020346673344546363,
+      "learning_rate": 9.470756164463827e-06,
+      "loss": 0.0001,
+      "step": 1052
+    },
+    {
+      "epoch": 0.0678610556164207,
+      "grad_norm": 0.0005355693822581008,
+      "learning_rate": 9.472049318241431e-06,
+      "loss": 0.0,
+      "step": 1053
+    },
+    {
+      "epoch": 0.06792550106334988,
+      "grad_norm": 0.025392949474745102,
+      "learning_rate": 9.473341244535435e-06,
+      "loss": 0.0,
+      "step": 1054
+    },
+    {
+      "epoch": 0.06798994651027905,
+      "grad_norm": 0.040790192372236425,
+      "learning_rate": 9.474631945673927e-06,
+      "loss": 0.0003,
+      "step": 1055
+    },
+    {
+      "epoch": 0.06805439195720822,
+      "grad_norm": 0.033763923209698024,
+      "learning_rate": 9.475921423978379e-06,
+      "loss": 0.0001,
+      "step": 1056
+    },
+    {
+      "epoch": 0.06811883740413739,
+      "grad_norm": 0.020406697261917767,
+      "learning_rate": 9.477209681763664e-06,
+      "loss": 0.0,
+      "step": 1057
+    },
+    {
+      "epoch": 0.06818328285106658,
+      "grad_norm": 0.0010645021910689245,
+      "learning_rate": 9.478496721338096e-06,
+      "loss": 0.0,
+      "step": 1058
+    },
+    {
+      "epoch": 0.06824772829799575,
+      "grad_norm": 0.05344905297559554,
+      "learning_rate": 9.479782545003442e-06,
+      "loss": 0.0006,
+      "step": 1059
+    },
+    {
+      "epoch": 0.06831217374492492,
+      "grad_norm": 0.22296419932981829,
+      "learning_rate": 9.481067155054957e-06,
+      "loss": 0.0005,
+      "step": 1060
+    },
+    {
+      "epoch": 0.0683766191918541,
+      "grad_norm": 0.0006634595339300624,
+      "learning_rate": 9.482350553781397e-06,
+      "loss": 0.0,
+      "step": 1061
+    },
+    {
+      "epoch": 0.06844106463878327,
+      "grad_norm": 0.15285856729547703,
+      "learning_rate": 9.48363274346505e-06,
+      "loss": 0.0003,
+      "step": 1062
+    },
+    {
+      "epoch": 0.06850551008571244,
+      "grad_norm": 0.030757339862208403,
+      "learning_rate": 9.484913726381766e-06,
+      "loss": 0.0002,
+      "step": 1063
+    },
+    {
+      "epoch": 0.06856995553264161,
+      "grad_norm": 0.20174093222005063,
+      "learning_rate": 9.486193504800969e-06,
+      "loss": 0.0006,
+      "step": 1064
+    },
+    {
+      "epoch": 0.0686344009795708,
+      "grad_norm": 0.030571583646065718,
+      "learning_rate": 9.487472080985686e-06,
+      "loss": 0.0001,
+      "step": 1065
+    },
+    {
+      "epoch": 0.06869884642649997,
+      "grad_norm": 0.0009779026849065075,
+      "learning_rate": 9.48874945719258e-06,
+      "loss": 0.0,
+      "step": 1066
+    },
+    {
+      "epoch": 0.06876329187342914,
+      "grad_norm": 0.00596599383421079,
+      "learning_rate": 9.49002563567196e-06,
+      "loss": 0.0001,
+      "step": 1067
+    },
+    {
+      "epoch": 0.06882773732035832,
+      "grad_norm": 0.12309505639517282,
+      "learning_rate": 9.49130061866781e-06,
+      "loss": 0.002,
+      "step": 1068
+    },
+    {
+      "epoch": 0.0688921827672875,
+      "grad_norm": 0.00540315072705124,
+      "learning_rate": 9.492574408417818e-06,
+      "loss": 0.0,
+      "step": 1069
+    },
+    {
+      "epoch": 0.06895662821421666,
+      "grad_norm": 0.0008161722316358159,
+      "learning_rate": 9.493847007153396e-06,
+      "loss": 0.0,
+      "step": 1070
+    },
+    {
+      "epoch": 0.06902107366114583,
+      "grad_norm": 0.0047131316080134674,
+      "learning_rate": 9.495118417099695e-06,
+      "loss": 0.0,
+      "step": 1071
+    },
+    {
+      "epoch": 0.06908551910807502,
+      "grad_norm": 0.051150246774193775,
+      "learning_rate": 9.496388640475643e-06,
+      "loss": 0.0002,
+      "step": 1072
+    },
+    {
+      "epoch": 0.06914996455500419,
+      "grad_norm": 0.0019308822291618391,
+      "learning_rate": 9.49765767949396e-06,
+      "loss": 0.0,
+      "step": 1073
+    },
+    {
+      "epoch": 0.06921441000193336,
+      "grad_norm": 0.006845832426320013,
+      "learning_rate": 9.498925536361181e-06,
+      "loss": 0.0001,
+      "step": 1074
+    },
+    {
+      "epoch": 0.06927885544886254,
+      "grad_norm": 0.07802347799089825,
+      "learning_rate": 9.500192213277679e-06,
+      "loss": 0.0001,
+      "step": 1075
+    },
+    {
+      "epoch": 0.06934330089579172,
+      "grad_norm": 0.06181839032778164,
+      "learning_rate": 9.501457712437695e-06,
+      "loss": 0.0,
+      "step": 1076
+    },
+    {
+      "epoch": 0.06940774634272089,
+      "grad_norm": 0.005407418144409315,
+      "learning_rate": 9.502722036029348e-06,
+      "loss": 0.0,
+      "step": 1077
+    },
+    {
+      "epoch": 0.06947219178965006,
+      "grad_norm": 0.013561958584136294,
+      "learning_rate": 9.503985186234666e-06,
+      "loss": 0.0016,
+      "step": 1078
+    },
+    {
+      "epoch": 0.06953663723657924,
+      "grad_norm": 0.041943084762526915,
+      "learning_rate": 9.505247165229614e-06,
+      "loss": 0.0001,
+      "step": 1079
+    },
+    {
+      "epoch": 0.06960108268350841,
+      "grad_norm": 0.4291865810230572,
+      "learning_rate": 9.506507975184103e-06,
+      "loss": 0.0015,
+      "step": 1080
+    },
+    {
+      "epoch": 0.06966552813043758,
+      "grad_norm": 0.18459478347266867,
+      "learning_rate": 9.507767618262019e-06,
+      "loss": 0.0009,
+      "step": 1081
+    },
+    {
+      "epoch": 0.06972997357736675,
+      "grad_norm": 0.55468446435232,
+      "learning_rate": 9.509026096621253e-06,
+      "loss": 0.0038,
+      "step": 1082
+    },
+    {
+      "epoch": 0.06979441902429594,
+      "grad_norm": 0.007788673964636509,
+      "learning_rate": 9.510283412413708e-06,
+      "loss": 0.0001,
+      "step": 1083
+    },
+    {
+      "epoch": 0.06985886447122511,
+      "grad_norm": 0.005881688910034065,
+      "learning_rate": 9.511539567785331e-06,
+      "loss": 0.0,
+      "step": 1084
+    },
+    {
+      "epoch": 0.06992330991815428,
+      "grad_norm": 0.010063309122665664,
+      "learning_rate": 9.512794564876132e-06,
+      "loss": 0.0,
+      "step": 1085
+    },
+    {
+      "epoch": 0.06998775536508346,
+      "grad_norm": 0.16376256774401285,
+      "learning_rate": 9.51404840582021e-06,
+      "loss": 0.0003,
+      "step": 1086
+    },
+    {
+      "epoch": 0.07005220081201263,
+      "grad_norm": 0.012820581196048885,
+      "learning_rate": 9.51530109274577e-06,
+      "loss": 0.0001,
+      "step": 1087
+    },
+    {
+      "epoch": 0.0701166462589418,
+      "grad_norm": 0.3957807941035052,
+      "learning_rate": 9.516552627775143e-06,
+      "loss": 0.0029,
+      "step": 1088
+    },
+    {
+      "epoch": 0.07018109170587097,
+      "grad_norm": 0.021148173145825185,
+      "learning_rate": 9.517803013024814e-06,
+      "loss": 0.0001,
+      "step": 1089
+    },
+    {
+      "epoch": 0.07024553715280016,
+      "grad_norm": 0.005577600092877411,
+      "learning_rate": 9.51905225060544e-06,
+      "loss": 0.0001,
+      "step": 1090
+    },
+    {
+      "epoch": 0.07030998259972933,
+      "grad_norm": 0.00015751397743556044,
+      "learning_rate": 9.52030034262187e-06,
+      "loss": 0.0,
+      "step": 1091
+    },
+    {
+      "epoch": 0.0703744280466585,
+      "grad_norm": 0.002895350355714667,
+      "learning_rate": 9.521547291173168e-06,
+      "loss": 0.0,
+      "step": 1092
+    },
+    {
+      "epoch": 0.07043887349358768,
+      "grad_norm": 0.0023124064473402568,
+      "learning_rate": 9.522793098352637e-06,
+      "loss": 0.0,
+      "step": 1093
+    },
+    {
+      "epoch": 0.07050331894051685,
+      "grad_norm": 0.20981965745450876,
+      "learning_rate": 9.524037766247829e-06,
+      "loss": 0.0017,
+      "step": 1094
+    },
+    {
+      "epoch": 0.07056776438744602,
+      "grad_norm": 0.016446824720652117,
+      "learning_rate": 9.525281296940584e-06,
+      "loss": 0.0,
+      "step": 1095
+    },
+    {
+      "epoch": 0.0706322098343752,
+      "grad_norm": 0.13535662307196408,
+      "learning_rate": 9.526523692507037e-06,
+      "loss": 0.0005,
+      "step": 1096
+    },
+    {
+      "epoch": 0.07069665528130438,
+      "grad_norm": 0.030537059287996025,
+      "learning_rate": 9.527764955017639e-06,
+      "loss": 0.0,
+      "step": 1097
+    },
+    {
+      "epoch": 0.07076110072823355,
+      "grad_norm": 0.014719741351277675,
+      "learning_rate": 9.529005086537187e-06,
+      "loss": 0.0001,
+      "step": 1098
+    },
+    {
+      "epoch": 0.07082554617516272,
+      "grad_norm": 0.22880329055924983,
+      "learning_rate": 9.530244089124837e-06,
+      "loss": 0.0012,
+      "step": 1099
+    },
+    {
+      "epoch": 0.0708899916220919,
+      "grad_norm": 0.00448704140852509,
+      "learning_rate": 9.531481964834122e-06,
+      "loss": 0.0,
+      "step": 1100
+    },
+    {
+      "epoch": 0.07095443706902108,
+      "grad_norm": 0.009209617813428274,
+      "learning_rate": 9.532718715712988e-06,
+      "loss": 0.0,
+      "step": 1101
+    },
+    {
+      "epoch": 0.07101888251595025,
+      "grad_norm": 0.09858833858855938,
+      "learning_rate": 9.533954343803793e-06,
+      "loss": 0.001,
+      "step": 1102
+    },
+    {
+      "epoch": 0.07108332796287942,
+      "grad_norm": 0.0003334525690795657,
+      "learning_rate": 9.535188851143343e-06,
+      "loss": 0.0,
+      "step": 1103
+    },
+    {
+      "epoch": 0.0711477734098086,
+      "grad_norm": 0.001099738770532882,
+      "learning_rate": 9.536422239762899e-06,
+      "loss": 0.0,
+      "step": 1104
+    },
+    {
+      "epoch": 0.07121221885673777,
+      "grad_norm": 0.012566891242134792,
+      "learning_rate": 9.537654511688219e-06,
+      "loss": 0.0,
+      "step": 1105
+    },
+    {
+      "epoch": 0.07127666430366694,
+      "grad_norm": 0.029662996653041767,
+      "learning_rate": 9.538885668939549e-06,
+      "loss": 0.0001,
+      "step": 1106
+    },
+    {
+      "epoch": 0.07134110975059613,
+      "grad_norm": 0.0039183110992792244,
+      "learning_rate": 9.540115713531667e-06,
+      "loss": 0.0,
+      "step": 1107
+    },
+    {
+      "epoch": 0.0714055551975253,
+      "grad_norm": 0.06460220983321396,
+      "learning_rate": 9.541344647473889e-06,
+      "loss": 0.0006,
+      "step": 1108
+    },
+    {
+      "epoch": 0.07147000064445447,
+      "grad_norm": 0.01392801660340169,
+      "learning_rate": 9.542572472770095e-06,
+      "loss": 0.0001,
+      "step": 1109
+    },
+    {
+      "epoch": 0.07153444609138364,
+      "grad_norm": 0.0075267292844853,
+      "learning_rate": 9.543799191418745e-06,
+      "loss": 0.0001,
+      "step": 1110
+    },
+    {
+      "epoch": 0.07159889153831282,
+      "grad_norm": 0.4668644333477323,
+      "learning_rate": 9.545024805412904e-06,
+      "loss": 0.0024,
+      "step": 1111
+    },
+    {
+      "epoch": 0.071663336985242,
+      "grad_norm": 0.04064895583802354,
+      "learning_rate": 9.546249316740252e-06,
+      "loss": 0.0001,
+      "step": 1112
+    },
+    {
+      "epoch": 0.07172778243217116,
+      "grad_norm": 0.01834313036665756,
+      "learning_rate": 9.54747272738311e-06,
+      "loss": 0.0,
+      "step": 1113
+    },
+    {
+      "epoch": 0.07179222787910035,
+      "grad_norm": 0.15668289299222576,
+      "learning_rate": 9.548695039318467e-06,
+      "loss": 0.0001,
+      "step": 1114
+    },
+    {
+      "epoch": 0.07185667332602952,
+      "grad_norm": 0.04903259619234823,
+      "learning_rate": 9.549916254517975e-06,
+      "loss": 0.0001,
+      "step": 1115
+    },
+    {
+      "epoch": 0.07192111877295869,
+      "grad_norm": 0.023982815736462482,
+      "learning_rate": 9.551136374947997e-06,
+      "loss": 0.0,
+      "step": 1116
+    },
+    {
+      "epoch": 0.07198556421988786,
+      "grad_norm": 0.0032611786740441946,
+      "learning_rate": 9.552355402569607e-06,
+      "loss": 0.0,
+      "step": 1117
+    },
+    {
+      "epoch": 0.07205000966681704,
+      "grad_norm": 0.03591521896676965,
+      "learning_rate": 9.553573339338611e-06,
+      "loss": 0.0001,
+      "step": 1118
+    },
+    {
+      "epoch": 0.07211445511374621,
+      "grad_norm": 0.002452457086472786,
+      "learning_rate": 9.554790187205577e-06,
+      "loss": 0.0,
+      "step": 1119
+    },
+    {
+      "epoch": 0.07217890056067539,
+      "grad_norm": 0.005269897617250751,
+      "learning_rate": 9.55600594811584e-06,
+      "loss": 0.0,
+      "step": 1120
+    },
+    {
+      "epoch": 0.07224334600760456,
+      "grad_norm": 0.4053084576654436,
+      "learning_rate": 9.557220624009528e-06,
+      "loss": 0.0009,
+      "step": 1121
+    },
+    {
+      "epoch": 0.07230779145453374,
+      "grad_norm": 0.04937713962457996,
+      "learning_rate": 9.55843421682158e-06,
+      "loss": 0.0004,
+      "step": 1122
+    },
+    {
+      "epoch": 0.07237223690146291,
+      "grad_norm": 0.31367174055920827,
+      "learning_rate": 9.559646728481764e-06,
+      "loss": 0.0005,
+      "step": 1123
+    },
+    {
+      "epoch": 0.07243668234839208,
+      "grad_norm": 0.0021878403973614306,
+      "learning_rate": 9.560858160914693e-06,
+      "loss": 0.0,
+      "step": 1124
+    },
+    {
+      "epoch": 0.07250112779532127,
+      "grad_norm": 0.04644155774287131,
+      "learning_rate": 9.562068516039848e-06,
+      "loss": 0.0001,
+      "step": 1125
+    },
+    {
+      "epoch": 0.07256557324225044,
+      "grad_norm": 0.005334428075335696,
+      "learning_rate": 9.563277795771592e-06,
+      "loss": 0.0,
+      "step": 1126
+    },
+    {
+      "epoch": 0.0726300186891796,
+      "grad_norm": 0.0009123395890019158,
+      "learning_rate": 9.564486002019189e-06,
+      "loss": 0.0,
+      "step": 1127
+    },
+    {
+      "epoch": 0.07269446413610878,
+      "grad_norm": 0.0032421208617262635,
+      "learning_rate": 9.565693136686826e-06,
+      "loss": 0.0,
+      "step": 1128
+    },
+    {
+      "epoch": 0.07275890958303796,
+      "grad_norm": 0.20209974356677035,
+      "learning_rate": 9.566899201673622e-06,
+      "loss": 0.0004,
+      "step": 1129
+    },
+    {
+      "epoch": 0.07282335502996713,
+      "grad_norm": 0.0017558729881317762,
+      "learning_rate": 9.568104198873661e-06,
+      "loss": 0.0,
+      "step": 1130
+    },
+    {
+      "epoch": 0.0728878004768963,
+      "grad_norm": 0.0007344998898286816,
+      "learning_rate": 9.569308130175994e-06,
+      "loss": 0.0,
+      "step": 1131
+    },
+    {
+      "epoch": 0.07295224592382549,
+      "grad_norm": 0.07438051013009608,
+      "learning_rate": 9.570510997464664e-06,
+      "loss": 0.0004,
+      "step": 1132
+    },
+    {
+      "epoch": 0.07301669137075466,
+      "grad_norm": 0.05245505196256475,
+      "learning_rate": 9.571712802618725e-06,
+      "loss": 0.0002,
+      "step": 1133
+    },
+    {
+      "epoch": 0.07308113681768383,
+      "grad_norm": 0.0009284186006608998,
+      "learning_rate": 9.572913547512255e-06,
+      "loss": 0.0,
+      "step": 1134
+    },
+    {
+      "epoch": 0.073145582264613,
+      "grad_norm": 0.0034999118506301983,
+      "learning_rate": 9.574113234014383e-06,
+      "loss": 0.0,
+      "step": 1135
+    },
+    {
+      "epoch": 0.07321002771154218,
+      "grad_norm": 0.07515344754801417,
+      "learning_rate": 9.57531186398929e-06,
+      "loss": 0.0002,
+      "step": 1136
+    },
+    {
+      "epoch": 0.07327447315847135,
+      "grad_norm": 0.008738924023498379,
+      "learning_rate": 9.576509439296242e-06,
+      "loss": 0.0,
+      "step": 1137
+    },
+    {
+      "epoch": 0.07333891860540052,
+      "grad_norm": 0.0008081805016731883,
+      "learning_rate": 9.577705961789603e-06,
+      "loss": 0.0,
+      "step": 1138
+    },
+    {
+      "epoch": 0.07340336405232971,
+      "grad_norm": 0.03078131770414565,
+      "learning_rate": 9.578901433318846e-06,
+      "loss": 0.0001,
+      "step": 1139
+    },
+    {
+      "epoch": 0.07346780949925888,
+      "grad_norm": 0.03254781257906103,
+      "learning_rate": 9.580095855728579e-06,
+      "loss": 0.0003,
+      "step": 1140
+    },
+    {
+      "epoch": 0.07353225494618805,
+      "grad_norm": 0.10584629442542785,
+      "learning_rate": 9.581289230858554e-06,
+      "loss": 0.0018,
+      "step": 1141
+    },
+    {
+      "epoch": 0.07359670039311722,
+      "grad_norm": 0.013347870485689773,
+      "learning_rate": 9.582481560543692e-06,
+      "loss": 0.0,
+      "step": 1142
+    },
+    {
+      "epoch": 0.0736611458400464,
+      "grad_norm": 0.0043998961185891345,
+      "learning_rate": 9.583672846614092e-06,
+      "loss": 0.0,
+      "step": 1143
+    },
+    {
+      "epoch": 0.07372559128697558,
+      "grad_norm": 0.04458813463929904,
+      "learning_rate": 9.584863090895054e-06,
+      "loss": 0.0001,
+      "step": 1144
+    },
+    {
+      "epoch": 0.07379003673390475,
+      "grad_norm": 0.026909427188295067,
+      "learning_rate": 9.586052295207094e-06,
+      "loss": 0.0,
+      "step": 1145
+    },
+    {
+      "epoch": 0.07385448218083393,
+      "grad_norm": 0.008740718491162603,
+      "learning_rate": 9.587240461365958e-06,
+      "loss": 0.0,
+      "step": 1146
+    },
+    {
+      "epoch": 0.0739189276277631,
+      "grad_norm": 0.2052019199750452,
+      "learning_rate": 9.58842759118264e-06,
+      "loss": 0.002,
+      "step": 1147
+    },
+    {
+      "epoch": 0.07398337307469227,
+      "grad_norm": 0.0039053163539573186,
+      "learning_rate": 9.589613686463404e-06,
+      "loss": 0.0001,
+      "step": 1148
+    },
+    {
+      "epoch": 0.07404781852162144,
+      "grad_norm": 0.007556923226473166,
+      "learning_rate": 9.590798749009793e-06,
+      "loss": 0.0,
+      "step": 1149
+    },
+    {
+      "epoch": 0.07411226396855063,
+      "grad_norm": 0.02777061566941155,
+      "learning_rate": 9.591982780618645e-06,
+      "loss": 0.0001,
+      "step": 1150
+    },
+    {
+      "epoch": 0.0741767094154798,
+      "grad_norm": 0.3086832496171256,
+      "learning_rate": 9.593165783082122e-06,
+      "loss": 0.0012,
+      "step": 1151
+    },
+    {
+      "epoch": 0.07424115486240897,
+      "grad_norm": 0.0005099813533373637,
+      "learning_rate": 9.594347758187705e-06,
+      "loss": 0.0,
+      "step": 1152
+    },
+    {
+      "epoch": 0.07430560030933815,
+      "grad_norm": 0.0010335636522138616,
+      "learning_rate": 9.59552870771823e-06,
+      "loss": 0.0,
+      "step": 1153
+    },
+    {
+      "epoch": 0.07437004575626732,
+      "grad_norm": 0.0005503441681810711,
+      "learning_rate": 9.596708633451897e-06,
+      "loss": 0.0,
+      "step": 1154
+    },
+    {
+      "epoch": 0.07443449120319649,
+      "grad_norm": 0.29321193774320214,
+      "learning_rate": 9.597887537162277e-06,
+      "loss": 0.0007,
+      "step": 1155
+    },
+    {
+      "epoch": 0.07449893665012566,
+      "grad_norm": 0.0003677550358705728,
+      "learning_rate": 9.599065420618347e-06,
+      "loss": 0.0,
+      "step": 1156
+    },
+    {
+      "epoch": 0.07456338209705485,
+      "grad_norm": 0.09427091300374382,
+      "learning_rate": 9.600242285584487e-06,
+      "loss": 0.0001,
+      "step": 1157
+    },
+    {
+      "epoch": 0.07462782754398402,
+      "grad_norm": 0.1102941329864884,
+      "learning_rate": 9.601418133820508e-06,
+      "loss": 0.0002,
+      "step": 1158
+    },
+    {
+      "epoch": 0.07469227299091319,
+      "grad_norm": 0.0016408962246154898,
+      "learning_rate": 9.602592967081665e-06,
+      "loss": 0.0,
+      "step": 1159
+    },
+    {
+      "epoch": 0.07475671843784237,
+      "grad_norm": 0.015343294270169556,
+      "learning_rate": 9.603766787118666e-06,
+      "loss": 0.0001,
+      "step": 1160
+    },
+    {
+      "epoch": 0.07482116388477154,
+      "grad_norm": 0.007299553047901523,
+      "learning_rate": 9.604939595677697e-06,
+      "loss": 0.0001,
+      "step": 1161
+    },
+    {
+      "epoch": 0.07488560933170071,
+      "grad_norm": 0.06251656417098318,
+      "learning_rate": 9.606111394500438e-06,
+      "loss": 0.0002,
+      "step": 1162
+    },
+    {
+      "epoch": 0.07495005477862988,
+      "grad_norm": 0.005736968608914812,
+      "learning_rate": 9.607282185324068e-06,
+      "loss": 0.0,
+      "step": 1163
+    },
+    {
+      "epoch": 0.07501450022555907,
+      "grad_norm": 0.0003761041372193296,
+      "learning_rate": 9.608451969881285e-06,
+      "loss": 0.0,
+      "step": 1164
+    },
+    {
+      "epoch": 0.07507894567248824,
+      "grad_norm": 0.004082930690161441,
+      "learning_rate": 9.609620749900332e-06,
+      "loss": 0.0,
+      "step": 1165
+    },
+    {
+      "epoch": 0.07514339111941741,
+      "grad_norm": 0.001868907810488472,
+      "learning_rate": 9.610788527104996e-06,
+      "loss": 0.0,
+      "step": 1166
+    },
+    {
+      "epoch": 0.07520783656634658,
+      "grad_norm": 0.01672035118957357,
+      "learning_rate": 9.611955303214636e-06,
+      "loss": 0.0,
+      "step": 1167
+    },
+    {
+      "epoch": 0.07527228201327577,
+      "grad_norm": 0.0003677767328121933,
+      "learning_rate": 9.613121079944188e-06,
+      "loss": 0.0,
+      "step": 1168
+    },
+    {
+      "epoch": 0.07533672746020494,
+      "grad_norm": 0.20056341800735186,
+      "learning_rate": 9.614285859004185e-06,
+      "loss": 0.0018,
+      "step": 1169
+    },
+    {
+      "epoch": 0.0754011729071341,
+      "grad_norm": 0.672537078225248,
+      "learning_rate": 9.61544964210078e-06,
+      "loss": 0.0034,
+      "step": 1170
+    },
+    {
+      "epoch": 0.07546561835406329,
+      "grad_norm": 0.0004565966029069957,
+      "learning_rate": 9.616612430935744e-06,
+      "loss": 0.0,
+      "step": 1171
+    },
+    {
+      "epoch": 0.07553006380099246,
+      "grad_norm": 0.00035533163882761317,
+      "learning_rate": 9.617774227206492e-06,
+      "loss": 0.0,
+      "step": 1172
+    },
+    {
+      "epoch": 0.07559450924792163,
+      "grad_norm": 0.192594228075119,
+      "learning_rate": 9.618935032606104e-06,
+      "loss": 0.0015,
+      "step": 1173
+    },
+    {
+      "epoch": 0.0756589546948508,
+      "grad_norm": 0.10144572697475107,
+      "learning_rate": 9.620094848823319e-06,
+      "loss": 0.0002,
+      "step": 1174
+    },
+    {
+      "epoch": 0.07572340014177999,
+      "grad_norm": 0.007117030604680507,
+      "learning_rate": 9.621253677542571e-06,
+      "loss": 0.0,
+      "step": 1175
+    },
+    {
+      "epoch": 0.07578784558870916,
+      "grad_norm": 0.0061437197610423756,
+      "learning_rate": 9.622411520443993e-06,
+      "loss": 0.0,
+      "step": 1176
+    },
+    {
+      "epoch": 0.07585229103563833,
+      "grad_norm": 0.04613563697317982,
+      "learning_rate": 9.623568379203435e-06,
+      "loss": 0.0001,
+      "step": 1177
+    },
+    {
+      "epoch": 0.07591673648256751,
+      "grad_norm": 0.058720324813408255,
+      "learning_rate": 9.624724255492474e-06,
+      "loss": 0.0001,
+      "step": 1178
+    },
+    {
+      "epoch": 0.07598118192949668,
+      "grad_norm": 0.006014910631233619,
+      "learning_rate": 9.625879150978434e-06,
+      "loss": 0.0,
+      "step": 1179
+    },
+    {
+      "epoch": 0.07604562737642585,
+      "grad_norm": 0.0006228941474103343,
+      "learning_rate": 9.6270330673244e-06,
+      "loss": 0.0,
+      "step": 1180
+    },
+    {
+      "epoch": 0.07611007282335502,
+      "grad_norm": 0.006748228304873041,
+      "learning_rate": 9.628186006189226e-06,
+      "loss": 0.0,
+      "step": 1181
+    },
+    {
+      "epoch": 0.07617451827028421,
+      "grad_norm": 0.5808579312109688,
+      "learning_rate": 9.62933796922756e-06,
+      "loss": 0.0007,
+      "step": 1182
+    },
+    {
+      "epoch": 0.07623896371721338,
+      "grad_norm": 0.043795489180758776,
+      "learning_rate": 9.630488958089847e-06,
+      "loss": 0.0002,
+      "step": 1183
+    },
+    {
+      "epoch": 0.07630340916414255,
+      "grad_norm": 0.0515382355616,
+      "learning_rate": 9.631638974422349e-06,
+      "loss": 0.0001,
+      "step": 1184
+    },
+    {
+      "epoch": 0.07636785461107173,
+      "grad_norm": 0.031008705248851732,
+      "learning_rate": 9.632788019867159e-06,
+      "loss": 0.0001,
+      "step": 1185
+    },
+    {
+      "epoch": 0.0764323000580009,
+      "grad_norm": 0.003564128836485725,
+      "learning_rate": 9.63393609606222e-06,
+      "loss": 0.0,
+      "step": 1186
+    },
+    {
+      "epoch": 0.07649674550493007,
+      "grad_norm": 0.0021392358347696247,
+      "learning_rate": 9.635083204641324e-06,
+      "loss": 0.0,
+      "step": 1187
+    },
+    {
+      "epoch": 0.07656119095185925,
+      "grad_norm": 0.0004078006923083002,
+      "learning_rate": 9.636229347234144e-06,
+      "loss": 0.0,
+      "step": 1188
+    },
+    {
+      "epoch": 0.07662563639878843,
+      "grad_norm": 0.20509931364044576,
+      "learning_rate": 9.637374525466229e-06,
+      "loss": 0.0004,
+      "step": 1189
+    },
+    {
+      "epoch": 0.0766900818457176,
+      "grad_norm": 0.06311748250374623,
+      "learning_rate": 9.638518740959042e-06,
+      "loss": 0.0004,
+      "step": 1190
+    },
+    {
+      "epoch": 0.07675452729264677,
+      "grad_norm": 0.022872006103306926,
+      "learning_rate": 9.639661995329951e-06,
+      "loss": 0.0,
+      "step": 1191
+    },
+    {
+      "epoch": 0.07681897273957596,
+      "grad_norm": 0.0007420031287808976,
+      "learning_rate": 9.640804290192255e-06,
+      "loss": 0.0,
+      "step": 1192
+    },
+    {
+      "epoch": 0.07688341818650513,
+      "grad_norm": 0.005457022771712689,
+      "learning_rate": 9.641945627155186e-06,
+      "loss": 0.0,
+      "step": 1193
+    },
+    {
+      "epoch": 0.0769478636334343,
+      "grad_norm": 0.008290051615120557,
+      "learning_rate": 9.643086007823948e-06,
+      "loss": 0.0001,
+      "step": 1194
+    },
+    {
+      "epoch": 0.07701230908036347,
+      "grad_norm": 0.023992458922295828,
+      "learning_rate": 9.644225433799694e-06,
+      "loss": 0.0001,
+      "step": 1195
+    },
+    {
+      "epoch": 0.07707675452729265,
+      "grad_norm": 0.003102753117205217,
+      "learning_rate": 9.645363906679577e-06,
+      "loss": 0.0,
+      "step": 1196
+    },
+    {
+      "epoch": 0.07714119997422182,
+      "grad_norm": 0.058512229636302565,
+      "learning_rate": 9.646501428056732e-06,
+      "loss": 0.0001,
+      "step": 1197
+    },
+    {
+      "epoch": 0.07720564542115099,
+      "grad_norm": 0.00014176378035999393,
+      "learning_rate": 9.64763799952031e-06,
+      "loss": 0.0,
+      "step": 1198
+    },
+    {
+      "epoch": 0.07727009086808018,
+      "grad_norm": 0.0003108339992887321,
+      "learning_rate": 9.64877362265548e-06,
+      "loss": 0.0,
+      "step": 1199
+    },
+    {
+      "epoch": 0.07733453631500935,
+      "grad_norm": 0.0032003990062005197,
+      "learning_rate": 9.649908299043451e-06,
+      "loss": 0.0,
+      "step": 1200
+    },
+    {
+      "epoch": 0.07739898176193852,
+      "grad_norm": 0.5838421549541228,
+      "learning_rate": 9.651042030261476e-06,
+      "loss": 0.0034,
+      "step": 1201
+    },
+    {
+      "epoch": 0.07746342720886769,
+      "grad_norm": 0.0008412995426426008,
+      "learning_rate": 9.652174817882872e-06,
+      "loss": 0.0,
+      "step": 1202
+    },
+    {
+      "epoch": 0.07752787265579687,
+      "grad_norm": 9.185586311117465e-05,
+      "learning_rate": 9.653306663477033e-06,
+      "loss": 0.0,
+      "step": 1203
+    },
+    {
+      "epoch": 0.07759231810272604,
+      "grad_norm": 0.23148305400119992,
+      "learning_rate": 9.654437568609434e-06,
+      "loss": 0.0019,
+      "step": 1204
+    },
+    {
+      "epoch": 0.07765676354965521,
+      "grad_norm": 0.00339606142341171,
+      "learning_rate": 9.655567534841662e-06,
+      "loss": 0.0,
+      "step": 1205
+    },
+    {
+      "epoch": 0.07772120899658438,
+      "grad_norm": 0.07896940352611387,
+      "learning_rate": 9.656696563731408e-06,
+      "loss": 0.0002,
+      "step": 1206
+    },
+    {
+      "epoch": 0.07778565444351357,
+      "grad_norm": 0.08238037779996558,
+      "learning_rate": 9.657824656832492e-06,
+      "loss": 0.0002,
+      "step": 1207
+    },
+    {
+      "epoch": 0.07785009989044274,
+      "grad_norm": 0.002891572609876939,
+      "learning_rate": 9.658951815694877e-06,
+      "loss": 0.0,
+      "step": 1208
+    },
+    {
+      "epoch": 0.07791454533737191,
+      "grad_norm": 0.0007835520232595968,
+      "learning_rate": 9.660078041864675e-06,
+      "loss": 0.0,
+      "step": 1209
+    },
+    {
+      "epoch": 0.0779789907843011,
+      "grad_norm": 0.0002144706980715807,
+      "learning_rate": 9.661203336884163e-06,
+      "loss": 0.0,
+      "step": 1210
+    },
+    {
+      "epoch": 0.07804343623123026,
+      "grad_norm": 0.141667232412957,
+      "learning_rate": 9.6623277022918e-06,
+      "loss": 0.0008,
+      "step": 1211
+    },
+    {
+      "epoch": 0.07810788167815944,
+      "grad_norm": 0.013447958714361321,
+      "learning_rate": 9.66345113962223e-06,
+      "loss": 0.0001,
+      "step": 1212
+    },
+    {
+      "epoch": 0.0781723271250886,
+      "grad_norm": 0.015480042443949904,
+      "learning_rate": 9.664573650406304e-06,
+      "loss": 0.0,
+      "step": 1213
+    },
+    {
+      "epoch": 0.07823677257201779,
+      "grad_norm": 0.023478707705058638,
+      "learning_rate": 9.665695236171086e-06,
+      "loss": 0.0001,
+      "step": 1214
+    },
+    {
+      "epoch": 0.07830121801894696,
+      "grad_norm": 0.19705356870810878,
+      "learning_rate": 9.666815898439866e-06,
+      "loss": 0.0026,
+      "step": 1215
+    },
+    {
+      "epoch": 0.07836566346587613,
+      "grad_norm": 0.03781223988640594,
+      "learning_rate": 9.667935638732182e-06,
+      "loss": 0.0,
+      "step": 1216
+    },
+    {
+      "epoch": 0.07843010891280532,
+      "grad_norm": 0.08106810676990649,
+      "learning_rate": 9.669054458563815e-06,
+      "loss": 0.0016,
+      "step": 1217
+    },
+    {
+      "epoch": 0.07849455435973449,
+      "grad_norm": 0.0032037268903943607,
+      "learning_rate": 9.67017235944682e-06,
+      "loss": 0.0,
+      "step": 1218
+    },
+    {
+      "epoch": 0.07855899980666366,
+      "grad_norm": 0.0014909723492173954,
+      "learning_rate": 9.67128934288952e-06,
+      "loss": 0.0,
+      "step": 1219
+    },
+    {
+      "epoch": 0.07862344525359283,
+      "grad_norm": 0.05132170474820264,
+      "learning_rate": 9.672405410396536e-06,
+      "loss": 0.0,
+      "step": 1220
+    },
+    {
+      "epoch": 0.07868789070052201,
+      "grad_norm": 0.05448380494674192,
+      "learning_rate": 9.673520563468786e-06,
+      "loss": 0.0002,
+      "step": 1221
+    },
+    {
+      "epoch": 0.07875233614745118,
+      "grad_norm": 0.00041079875761307306,
+      "learning_rate": 9.674634803603501e-06,
+      "loss": 0.0,
+      "step": 1222
+    },
+    {
+      "epoch": 0.07881678159438035,
+      "grad_norm": 1.6507721048111714,
+      "learning_rate": 9.675748132294243e-06,
+      "loss": 0.0026,
+      "step": 1223
+    },
+    {
+      "epoch": 0.07888122704130954,
+      "grad_norm": 0.0002573480081688759,
+      "learning_rate": 9.676860551030908e-06,
+      "loss": 0.0,
+      "step": 1224
+    },
+    {
+      "epoch": 0.07894567248823871,
+      "grad_norm": 0.05452053945885145,
+      "learning_rate": 9.677972061299738e-06,
+      "loss": 0.0001,
+      "step": 1225
+    },
+    {
+      "epoch": 0.07901011793516788,
+      "grad_norm": 0.0009042614824524367,
+      "learning_rate": 9.679082664583347e-06,
+      "loss": 0.0,
+      "step": 1226
+    },
+    {
+      "epoch": 0.07907456338209705,
+      "grad_norm": 0.12650184384658905,
+      "learning_rate": 9.68019236236071e-06,
+      "loss": 0.0005,
+      "step": 1227
+    },
+    {
+      "epoch": 0.07913900882902623,
+      "grad_norm": 0.0003593917699330583,
+      "learning_rate": 9.6813011561072e-06,
+      "loss": 0.0,
+      "step": 1228
+    },
+    {
+      "epoch": 0.0792034542759554,
+      "grad_norm": 0.00784809323102259,
+      "learning_rate": 9.682409047294576e-06,
+      "loss": 0.0,
+      "step": 1229
+    },
+    {
+      "epoch": 0.07926789972288457,
+      "grad_norm": 0.0034080110459679502,
+      "learning_rate": 9.683516037391016e-06,
+      "loss": 0.0,
+      "step": 1230
+    },
+    {
+      "epoch": 0.07933234516981376,
+      "grad_norm": 0.0008000263937962497,
+      "learning_rate": 9.684622127861108e-06,
+      "loss": 0.0,
+      "step": 1231
+    },
+    {
+      "epoch": 0.07939679061674293,
+      "grad_norm": 0.4572995042527525,
+      "learning_rate": 9.685727320165879e-06,
+      "loss": 0.0051,
+      "step": 1232
+    },
+    {
+      "epoch": 0.0794612360636721,
+      "grad_norm": 0.14509120287519361,
+      "learning_rate": 9.6868316157628e-06,
+      "loss": 0.0004,
+      "step": 1233
+    },
+    {
+      "epoch": 0.07952568151060127,
+      "grad_norm": 0.0019696790708333524,
+      "learning_rate": 9.687935016105796e-06,
+      "loss": 0.0,
+      "step": 1234
+    },
+    {
+      "epoch": 0.07959012695753045,
+      "grad_norm": 0.2617435443577594,
+      "learning_rate": 9.689037522645257e-06,
+      "loss": 0.0007,
+      "step": 1235
+    },
+    {
+      "epoch": 0.07965457240445963,
+      "grad_norm": 0.1585587871704741,
+      "learning_rate": 9.690139136828051e-06,
+      "loss": 0.0002,
+      "step": 1236
+    },
+    {
+      "epoch": 0.0797190178513888,
+      "grad_norm": 0.0007139002751616745,
+      "learning_rate": 9.69123986009754e-06,
+      "loss": 0.0,
+      "step": 1237
+    },
+    {
+      "epoch": 0.07978346329831798,
+      "grad_norm": 0.0005685128388628978,
+      "learning_rate": 9.692339693893584e-06,
+      "loss": 0.0,
+      "step": 1238
+    },
+    {
+      "epoch": 0.07984790874524715,
+      "grad_norm": 0.00020046062008239465,
+      "learning_rate": 9.693438639652553e-06,
+      "loss": 0.0,
+      "step": 1239
+    },
+    {
+      "epoch": 0.07991235419217632,
+      "grad_norm": 0.018666319797578546,
+      "learning_rate": 9.694536698807345e-06,
+      "loss": 0.0002,
+      "step": 1240
+    },
+    {
+      "epoch": 0.07997679963910549,
+      "grad_norm": 0.002108454970512216,
+      "learning_rate": 9.69563387278739e-06,
+      "loss": 0.0,
+      "step": 1241
+    },
+    {
+      "epoch": 0.08004124508603468,
+      "grad_norm": 0.005928479550208142,
+      "learning_rate": 9.696730163018664e-06,
+      "loss": 0.0001,
+      "step": 1242
+    },
+    {
+      "epoch": 0.08010569053296385,
+      "grad_norm": 0.007256955744477934,
+      "learning_rate": 9.697825570923702e-06,
+      "loss": 0.0,
+      "step": 1243
+    },
+    {
+      "epoch": 0.08017013597989302,
+      "grad_norm": 0.012520196155166474,
+      "learning_rate": 9.698920097921602e-06,
+      "loss": 0.0,
+      "step": 1244
+    },
+    {
+      "epoch": 0.0802345814268222,
+      "grad_norm": 0.0049347868056927174,
+      "learning_rate": 9.700013745428048e-06,
+      "loss": 0.0,
+      "step": 1245
+    },
+    {
+      "epoch": 0.08029902687375137,
+      "grad_norm": 0.14433231710557265,
+      "learning_rate": 9.701106514855312e-06,
+      "loss": 0.0002,
+      "step": 1246
+    },
+    {
+      "epoch": 0.08036347232068054,
+      "grad_norm": 0.16086143413594908,
+      "learning_rate": 9.70219840761226e-06,
+      "loss": 0.0004,
+      "step": 1247
+    },
+    {
+      "epoch": 0.08042791776760971,
+      "grad_norm": 0.022894449012242864,
+      "learning_rate": 9.703289425104382e-06,
+      "loss": 0.0001,
+      "step": 1248
+    },
+    {
+      "epoch": 0.0804923632145389,
+      "grad_norm": 0.0004916579296406989,
+      "learning_rate": 9.70437956873378e-06,
+      "loss": 0.0,
+      "step": 1249
+    },
+    {
+      "epoch": 0.08055680866146807,
+      "grad_norm": 0.006341910971692203,
+      "learning_rate": 9.705468839899196e-06,
+      "loss": 0.0,
+      "step": 1250
+    },
+    {
+      "epoch": 0.08062125410839724,
+      "grad_norm": 0.000566196866322031,
+      "learning_rate": 9.706557239996017e-06,
+      "loss": 0.0,
+      "step": 1251
+    },
+    {
+      "epoch": 0.08068569955532641,
+      "grad_norm": 4.660119618530064,
+      "learning_rate": 9.707644770416276e-06,
+      "loss": 0.0008,
+      "step": 1252
+    },
+    {
+      "epoch": 0.0807501450022556,
+      "grad_norm": 0.0011596501754506218,
+      "learning_rate": 9.708731432548683e-06,
+      "loss": 0.0,
+      "step": 1253
+    },
+    {
+      "epoch": 0.08081459044918476,
+      "grad_norm": 0.0002413284823318843,
+      "learning_rate": 9.70981722777862e-06,
+      "loss": 0.0,
+      "step": 1254
+    },
+    {
+      "epoch": 0.08087903589611393,
+      "grad_norm": 0.00018585046524750463,
+      "learning_rate": 9.71090215748815e-06,
+      "loss": 0.0,
+      "step": 1255
+    },
+    {
+      "epoch": 0.08094348134304312,
+      "grad_norm": 0.004867257547187822,
+      "learning_rate": 9.71198622305605e-06,
+      "loss": 0.0,
+      "step": 1256
+    },
+    {
+      "epoch": 0.08100792678997229,
+      "grad_norm": 0.006004401052693547,
+      "learning_rate": 9.713069425857785e-06,
+      "loss": 0.0,
+      "step": 1257
+    },
+    {
+      "epoch": 0.08107237223690146,
+      "grad_norm": 0.0062596726703548775,
+      "learning_rate": 9.714151767265551e-06,
+      "loss": 0.0,
+      "step": 1258
+    },
+    {
+      "epoch": 0.08113681768383063,
+      "grad_norm": 0.0001596291398008103,
+      "learning_rate": 9.71523324864827e-06,
+      "loss": 0.0,
+      "step": 1259
+    },
+    {
+      "epoch": 0.08120126313075982,
+      "grad_norm": 0.003137489508600633,
+      "learning_rate": 9.716313871371605e-06,
+      "loss": 0.0,
+      "step": 1260
+    },
+    {
+      "epoch": 0.08126570857768899,
+      "grad_norm": 0.03389845337090403,
+      "learning_rate": 9.717393636797964e-06,
+      "loss": 0.0002,
+      "step": 1261
+    },
+    {
+      "epoch": 0.08133015402461816,
+      "grad_norm": 0.0015749193613697652,
+      "learning_rate": 9.718472546286517e-06,
+      "loss": 0.0,
+      "step": 1262
+    },
+    {
+      "epoch": 0.08139459947154734,
+      "grad_norm": 0.0016629746994158803,
+      "learning_rate": 9.71955060119321e-06,
+      "loss": 0.0,
+      "step": 1263
+    },
+    {
+      "epoch": 0.08145904491847651,
+      "grad_norm": 0.9630747342468597,
+      "learning_rate": 9.720627802870762e-06,
+      "loss": 0.0016,
+      "step": 1264
+    },
+    {
+      "epoch": 0.08152349036540568,
+      "grad_norm": 0.005799075443889431,
+      "learning_rate": 9.721704152668686e-06,
+      "loss": 0.0,
+      "step": 1265
+    },
+    {
+      "epoch": 0.08158793581233485,
+      "grad_norm": 0.9226179200559602,
+      "learning_rate": 9.722779651933294e-06,
+      "loss": 0.0038,
+      "step": 1266
+    },
+    {
+      "epoch": 0.08165238125926404,
+      "grad_norm": 0.009570234618307968,
+      "learning_rate": 9.723854302007711e-06,
+      "loss": 0.0,
+      "step": 1267
+    },
+    {
+      "epoch": 0.08171682670619321,
+      "grad_norm": 0.007609505717263756,
+      "learning_rate": 9.724928104231886e-06,
+      "loss": 0.0,
+      "step": 1268
+    },
+    {
+      "epoch": 0.08178127215312238,
+      "grad_norm": 0.0037044836505944105,
+      "learning_rate": 9.72600105994259e-06,
+      "loss": 0.0,
+      "step": 1269
+    },
+    {
+      "epoch": 0.08184571760005156,
+      "grad_norm": 0.3324494028852565,
+      "learning_rate": 9.727073170473441e-06,
+      "loss": 0.0003,
+      "step": 1270
+    },
+    {
+      "epoch": 0.08191016304698073,
+      "grad_norm": 0.22125266870947377,
+      "learning_rate": 9.72814443715491e-06,
+      "loss": 0.0015,
+      "step": 1271
+    },
+    {
+      "epoch": 0.0819746084939099,
+      "grad_norm": 0.8368448698447373,
+      "learning_rate": 9.729214861314324e-06,
+      "loss": 0.0008,
+      "step": 1272
+    },
+    {
+      "epoch": 0.08203905394083907,
+      "grad_norm": 0.03696941188445865,
+      "learning_rate": 9.730284444275884e-06,
+      "loss": 0.0,
+      "step": 1273
+    },
+    {
+      "epoch": 0.08210349938776826,
+      "grad_norm": 0.09513810317509,
+      "learning_rate": 9.73135318736067e-06,
+      "loss": 0.0001,
+      "step": 1274
+    },
+    {
+      "epoch": 0.08216794483469743,
+      "grad_norm": 0.1407959614094262,
+      "learning_rate": 9.732421091886653e-06,
+      "loss": 0.0001,
+      "step": 1275
+    },
+    {
+      "epoch": 0.0822323902816266,
+      "grad_norm": 0.11572725150082791,
+      "learning_rate": 9.733488159168705e-06,
+      "loss": 0.0002,
+      "step": 1276
+    },
+    {
+      "epoch": 0.08229683572855578,
+      "grad_norm": 0.20268866551076187,
+      "learning_rate": 9.734554390518605e-06,
+      "loss": 0.0006,
+      "step": 1277
+    },
+    {
+      "epoch": 0.08236128117548495,
+      "grad_norm": 0.034702215271124964,
+      "learning_rate": 9.735619787245053e-06,
+      "loss": 0.0,
+      "step": 1278
+    },
+    {
+      "epoch": 0.08242572662241412,
+      "grad_norm": 0.618313305814074,
+      "learning_rate": 9.73668435065368e-06,
+      "loss": 0.0058,
+      "step": 1279
+    },
+    {
+      "epoch": 0.0824901720693433,
+      "grad_norm": 0.12879200409898806,
+      "learning_rate": 9.737748082047053e-06,
+      "loss": 0.002,
+      "step": 1280
+    },
+    {
+      "epoch": 0.08255461751627248,
+      "grad_norm": 0.15517708649995626,
+      "learning_rate": 9.738810982724688e-06,
+      "loss": 0.0014,
+      "step": 1281
+    },
+    {
+      "epoch": 0.08261906296320165,
+      "grad_norm": 0.022942057466187164,
+      "learning_rate": 9.739873053983062e-06,
+      "loss": 0.0,
+      "step": 1282
+    },
+    {
+      "epoch": 0.08268350841013082,
+      "grad_norm": 0.32524165834202085,
+      "learning_rate": 9.740934297115614e-06,
+      "loss": 0.0001,
+      "step": 1283
+    },
+    {
+      "epoch": 0.08274795385706,
+      "grad_norm": 0.017645141056179724,
+      "learning_rate": 9.74199471341276e-06,
+      "loss": 0.0001,
+      "step": 1284
+    },
+    {
+      "epoch": 0.08281239930398918,
+      "grad_norm": 0.001701689789313395,
+      "learning_rate": 9.743054304161912e-06,
+      "loss": 0.0,
+      "step": 1285
+    },
+    {
+      "epoch": 0.08287684475091835,
+      "grad_norm": 0.11248027509383528,
+      "learning_rate": 9.744113070647462e-06,
+      "loss": 0.002,
+      "step": 1286
+    },
+    {
+      "epoch": 0.08294129019784752,
+      "grad_norm": 0.009134959493058783,
+      "learning_rate": 9.745171014150819e-06,
+      "loss": 0.0,
+      "step": 1287
+    },
+    {
+      "epoch": 0.0830057356447767,
+      "grad_norm": 0.001851569614579538,
+      "learning_rate": 9.746228135950402e-06,
+      "loss": 0.0,
+      "step": 1288
+    },
+    {
+      "epoch": 0.08307018109170587,
+      "grad_norm": 0.004553438642060542,
+      "learning_rate": 9.747284437321652e-06,
+      "loss": 0.0,
+      "step": 1289
+    },
+    {
+      "epoch": 0.08313462653863504,
+      "grad_norm": 0.010993416473078717,
+      "learning_rate": 9.748339919537046e-06,
+      "loss": 0.0001,
+      "step": 1290
+    },
+    {
+      "epoch": 0.08319907198556421,
+      "grad_norm": 0.046090326040599205,
+      "learning_rate": 9.749394583866101e-06,
+      "loss": 0.0002,
+      "step": 1291
+    },
+    {
+      "epoch": 0.0832635174324934,
+      "grad_norm": 0.016645441025789916,
+      "learning_rate": 9.750448431575384e-06,
+      "loss": 0.0002,
+      "step": 1292
+    },
+    {
+      "epoch": 0.08332796287942257,
+      "grad_norm": 0.07861484502338792,
+      "learning_rate": 9.751501463928525e-06,
+      "loss": 0.0002,
+      "step": 1293
+    },
+    {
+      "epoch": 0.08339240832635174,
+      "grad_norm": 0.14598362209163068,
+      "learning_rate": 9.75255368218622e-06,
+      "loss": 0.0012,
+      "step": 1294
+    },
+    {
+      "epoch": 0.08345685377328092,
+      "grad_norm": 0.005253576712949334,
+      "learning_rate": 9.753605087606247e-06,
+      "loss": 0.0,
+      "step": 1295
+    },
+    {
+      "epoch": 0.0835212992202101,
+      "grad_norm": 0.06470684124601767,
+      "learning_rate": 9.75465568144347e-06,
+      "loss": 0.0001,
+      "step": 1296
+    },
+    {
+      "epoch": 0.08358574466713926,
+      "grad_norm": 0.0001312365358307035,
+      "learning_rate": 9.755705464949845e-06,
+      "loss": 0.0,
+      "step": 1297
+    },
+    {
+      "epoch": 0.08365019011406843,
+      "grad_norm": 0.009228063738194224,
+      "learning_rate": 9.756754439374438e-06,
+      "loss": 0.0,
+      "step": 1298
+    },
+    {
+      "epoch": 0.08371463556099762,
+      "grad_norm": 0.022493899301872504,
+      "learning_rate": 9.757802605963434e-06,
+      "loss": 0.0001,
+      "step": 1299
+    },
+    {
+      "epoch": 0.08377908100792679,
+      "grad_norm": 0.025017351223788283,
+      "learning_rate": 9.758849965960128e-06,
+      "loss": 0.0,
+      "step": 1300
+    },
+    {
+      "epoch": 0.08384352645485596,
+      "grad_norm": 0.0033973402368583138,
+      "learning_rate": 9.759896520604957e-06,
+      "loss": 0.0,
+      "step": 1301
+    },
+    {
+      "epoch": 0.08390797190178514,
+      "grad_norm": 0.016653590335366984,
+      "learning_rate": 9.760942271135499e-06,
+      "loss": 0.0001,
+      "step": 1302
+    },
+    {
+      "epoch": 0.08397241734871431,
+      "grad_norm": 0.05323761764710082,
+      "learning_rate": 9.761987218786474e-06,
+      "loss": 0.0001,
+      "step": 1303
+    },
+    {
+      "epoch": 0.08403686279564349,
+      "grad_norm": 0.00047998523145980336,
+      "learning_rate": 9.763031364789768e-06,
+      "loss": 0.0,
+      "step": 1304
+    },
+    {
+      "epoch": 0.08410130824257266,
+      "grad_norm": 0.0012154446343361834,
+      "learning_rate": 9.764074710374431e-06,
+      "loss": 0.0,
+      "step": 1305
+    },
+    {
+      "epoch": 0.08416575368950184,
+      "grad_norm": 0.1703871353470703,
+      "learning_rate": 9.765117256766685e-06,
+      "loss": 0.0003,
+      "step": 1306
+    },
+    {
+      "epoch": 0.08423019913643101,
+      "grad_norm": 0.3756810798293176,
+      "learning_rate": 9.766159005189943e-06,
+      "loss": 0.0032,
+      "step": 1307
+    },
+    {
+      "epoch": 0.08429464458336018,
+      "grad_norm": 0.0012046472375180962,
+      "learning_rate": 9.767199956864808e-06,
+      "loss": 0.0,
+      "step": 1308
+    },
+    {
+      "epoch": 0.08435909003028937,
+      "grad_norm": 0.0008695978340420592,
+      "learning_rate": 9.768240113009083e-06,
+      "loss": 0.0,
+      "step": 1309
+    },
+    {
+      "epoch": 0.08442353547721854,
+      "grad_norm": 0.035405811355180825,
+      "learning_rate": 9.769279474837782e-06,
+      "loss": 0.0003,
+      "step": 1310
+    },
+    {
+      "epoch": 0.08448798092414771,
+      "grad_norm": 0.05460181932338454,
+      "learning_rate": 9.770318043563141e-06,
+      "loss": 0.0003,
+      "step": 1311
+    },
+    {
+      "epoch": 0.08455242637107688,
+      "grad_norm": 0.0020692583491097017,
+      "learning_rate": 9.771355820394616e-06,
+      "loss": 0.0,
+      "step": 1312
+    },
+    {
+      "epoch": 0.08461687181800606,
+      "grad_norm": 0.0029885875308376477,
+      "learning_rate": 9.772392806538907e-06,
+      "loss": 0.0,
+      "step": 1313
+    },
+    {
+      "epoch": 0.08468131726493523,
+      "grad_norm": 0.0010176173067989608,
+      "learning_rate": 9.773429003199951e-06,
+      "loss": 0.0,
+      "step": 1314
+    },
+    {
+      "epoch": 0.0847457627118644,
+      "grad_norm": 0.00025002265076070745,
+      "learning_rate": 9.77446441157894e-06,
+      "loss": 0.0,
+      "step": 1315
+    },
+    {
+      "epoch": 0.08481020815879359,
+      "grad_norm": 0.011615362301719032,
+      "learning_rate": 9.775499032874327e-06,
+      "loss": 0.0,
+      "step": 1316
+    },
+    {
+      "epoch": 0.08487465360572276,
+      "grad_norm": 0.8283102080329923,
+      "learning_rate": 9.776532868281832e-06,
+      "loss": 0.0065,
+      "step": 1317
+    },
+    {
+      "epoch": 0.08493909905265193,
+      "grad_norm": 0.009607753739860307,
+      "learning_rate": 9.777565918994456e-06,
+      "loss": 0.0,
+      "step": 1318
+    },
+    {
+      "epoch": 0.0850035444995811,
+      "grad_norm": 0.0008744673144066277,
+      "learning_rate": 9.778598186202483e-06,
+      "loss": 0.0,
+      "step": 1319
+    },
+    {
+      "epoch": 0.08506798994651028,
+      "grad_norm": 0.35578931250669904,
+      "learning_rate": 9.77962967109349e-06,
+      "loss": 0.0015,
+      "step": 1320
+    },
+    {
+      "epoch": 0.08513243539343945,
+      "grad_norm": 0.0002869815156587147,
+      "learning_rate": 9.780660374852359e-06,
+      "loss": 0.0,
+      "step": 1321
+    },
+    {
+      "epoch": 0.08519688084036862,
+      "grad_norm": 9.750659056478101e-05,
+      "learning_rate": 9.781690298661277e-06,
+      "loss": 0.0,
+      "step": 1322
+    },
+    {
+      "epoch": 0.08526132628729781,
+      "grad_norm": 0.0008800151837540015,
+      "learning_rate": 9.782719443699757e-06,
+      "loss": 0.0,
+      "step": 1323
+    },
+    {
+      "epoch": 0.08532577173422698,
+      "grad_norm": 0.007283107294294607,
+      "learning_rate": 9.783747811144632e-06,
+      "loss": 0.0,
+      "step": 1324
+    },
+    {
+      "epoch": 0.08539021718115615,
+      "grad_norm": 0.00046669696933537125,
+      "learning_rate": 9.784775402170071e-06,
+      "loss": 0.0,
+      "step": 1325
+    },
+    {
+      "epoch": 0.08545466262808532,
+      "grad_norm": 0.011137442948644324,
+      "learning_rate": 9.785802217947585e-06,
+      "loss": 0.0,
+      "step": 1326
+    },
+    {
+      "epoch": 0.0855191080750145,
+      "grad_norm": 0.00046858581858597265,
+      "learning_rate": 9.786828259646038e-06,
+      "loss": 0.0,
+      "step": 1327
+    },
+    {
+      "epoch": 0.08558355352194368,
+      "grad_norm": 0.0011227359990139579,
+      "learning_rate": 9.787853528431651e-06,
+      "loss": 0.0,
+      "step": 1328
+    },
+    {
+      "epoch": 0.08564799896887285,
+      "grad_norm": 0.005764189022396666,
+      "learning_rate": 9.788878025468011e-06,
+      "loss": 0.0,
+      "step": 1329
+    },
+    {
+      "epoch": 0.08571244441580203,
+      "grad_norm": 0.00036195693870573616,
+      "learning_rate": 9.78990175191608e-06,
+      "loss": 0.0,
+      "step": 1330
+    },
+    {
+      "epoch": 0.0857768898627312,
+      "grad_norm": 0.0004031809736862309,
+      "learning_rate": 9.790924708934204e-06,
+      "loss": 0.0,
+      "step": 1331
+    },
+    {
+      "epoch": 0.08584133530966037,
+      "grad_norm": 0.12820426942731789,
+      "learning_rate": 9.791946897678112e-06,
+      "loss": 0.0015,
+      "step": 1332
+    },
+    {
+      "epoch": 0.08590578075658954,
+      "grad_norm": 0.007201268710137027,
+      "learning_rate": 9.792968319300942e-06,
+      "loss": 0.0,
+      "step": 1333
+    },
+    {
+      "epoch": 0.08597022620351873,
+      "grad_norm": 0.29705053711328777,
+      "learning_rate": 9.793988974953227e-06,
+      "loss": 0.0012,
+      "step": 1334
+    },
+    {
+      "epoch": 0.0860346716504479,
+      "grad_norm": 0.048011958608229134,
+      "learning_rate": 9.795008865782922e-06,
+      "loss": 0.0003,
+      "step": 1335
+    },
+    {
+      "epoch": 0.08609911709737707,
+      "grad_norm": 0.000588550762065899,
+      "learning_rate": 9.796027992935399e-06,
+      "loss": 0.0,
+      "step": 1336
+    },
+    {
+      "epoch": 0.08616356254430624,
+      "grad_norm": 0.00403323898251864,
+      "learning_rate": 9.79704635755346e-06,
+      "loss": 0.0,
+      "step": 1337
+    },
+    {
+      "epoch": 0.08622800799123542,
+      "grad_norm": 0.0770604017607048,
+      "learning_rate": 9.798063960777342e-06,
+      "loss": 0.0003,
+      "step": 1338
+    },
+    {
+      "epoch": 0.08629245343816459,
+      "grad_norm": 0.002188687958222415,
+      "learning_rate": 9.799080803744728e-06,
+      "loss": 0.0,
+      "step": 1339
+    },
+    {
+      "epoch": 0.08635689888509376,
+      "grad_norm": 0.0005241857617764945,
+      "learning_rate": 9.800096887590755e-06,
+      "loss": 0.0,
+      "step": 1340
+    },
+    {
+      "epoch": 0.08642134433202295,
+      "grad_norm": 0.0007191739447359967,
+      "learning_rate": 9.801112213448018e-06,
+      "loss": 0.0,
+      "step": 1341
+    },
+    {
+      "epoch": 0.08648578977895212,
+      "grad_norm": 0.0020778973471756966,
+      "learning_rate": 9.802126782446577e-06,
+      "loss": 0.0,
+      "step": 1342
+    },
+    {
+      "epoch": 0.08655023522588129,
+      "grad_norm": 0.023544600776087193,
+      "learning_rate": 9.803140595713964e-06,
+      "loss": 0.0001,
+      "step": 1343
+    },
+    {
+      "epoch": 0.08661468067281046,
+      "grad_norm": 0.005317380634932252,
+      "learning_rate": 9.804153654375207e-06,
+      "loss": 0.0,
+      "step": 1344
+    },
+    {
+      "epoch": 0.08667912611973964,
+      "grad_norm": 0.001298663308470156,
+      "learning_rate": 9.805165959552807e-06,
+      "loss": 0.0,
+      "step": 1345
+    },
+    {
+      "epoch": 0.08674357156666881,
+      "grad_norm": 0.001955103407708176,
+      "learning_rate": 9.806177512366772e-06,
+      "loss": 0.0,
+      "step": 1346
+    },
+    {
+      "epoch": 0.08680801701359799,
+      "grad_norm": 0.012739390414210567,
+      "learning_rate": 9.807188313934614e-06,
+      "loss": 0.0,
+      "step": 1347
+    },
+    {
+      "epoch": 0.08687246246052717,
+      "grad_norm": 0.0017430904771763144,
+      "learning_rate": 9.808198365371351e-06,
+      "loss": 0.0,
+      "step": 1348
+    },
+    {
+      "epoch": 0.08693690790745634,
+      "grad_norm": 0.007015913286311483,
+      "learning_rate": 9.80920766778953e-06,
+      "loss": 0.0,
+      "step": 1349
+    },
+    {
+      "epoch": 0.08700135335438551,
+      "grad_norm": 0.0018873731549218875,
+      "learning_rate": 9.810216222299216e-06,
+      "loss": 0.0,
+      "step": 1350
+    },
+    {
+      "epoch": 0.08706579880131468,
+      "grad_norm": 0.0003947696754434573,
+      "learning_rate": 9.81122403000801e-06,
+      "loss": 0.0,
+      "step": 1351
+    },
+    {
+      "epoch": 0.08713024424824387,
+      "grad_norm": 0.038484140741465445,
+      "learning_rate": 9.81223109202106e-06,
+      "loss": 0.0001,
+      "step": 1352
+    },
+    {
+      "epoch": 0.08719468969517304,
+      "grad_norm": 0.11864357590167482,
+      "learning_rate": 9.813237409441055e-06,
+      "loss": 0.0003,
+      "step": 1353
+    },
+    {
+      "epoch": 0.0872591351421022,
+      "grad_norm": 0.5198617615258171,
+      "learning_rate": 9.814242983368245e-06,
+      "loss": 0.0043,
+      "step": 1354
+    },
+    {
+      "epoch": 0.08732358058903139,
+      "grad_norm": 0.06844873037905837,
+      "learning_rate": 9.815247814900443e-06,
+      "loss": 0.0001,
+      "step": 1355
+    },
+    {
+      "epoch": 0.08738802603596056,
+      "grad_norm": 0.00017070857672494202,
+      "learning_rate": 9.816251905133028e-06,
+      "loss": 0.0,
+      "step": 1356
+    },
+    {
+      "epoch": 0.08745247148288973,
+      "grad_norm": 0.014964520116337086,
+      "learning_rate": 9.817255255158962e-06,
+      "loss": 0.0001,
+      "step": 1357
+    },
+    {
+      "epoch": 0.0875169169298189,
+      "grad_norm": 0.003770806539701692,
+      "learning_rate": 9.818257866068787e-06,
+      "loss": 0.0,
+      "step": 1358
+    },
+    {
+      "epoch": 0.08758136237674809,
+      "grad_norm": 0.09473215165314121,
+      "learning_rate": 9.81925973895064e-06,
+      "loss": 0.0005,
+      "step": 1359
+    },
+    {
+      "epoch": 0.08764580782367726,
+      "grad_norm": 0.005408603276892646,
+      "learning_rate": 9.820260874890255e-06,
+      "loss": 0.0,
+      "step": 1360
+    },
+    {
+      "epoch": 0.08771025327060643,
+      "grad_norm": 0.0015547945886781377,
+      "learning_rate": 9.821261274970973e-06,
+      "loss": 0.0,
+      "step": 1361
+    },
+    {
+      "epoch": 0.08777469871753561,
+      "grad_norm": 0.0015138486387163961,
+      "learning_rate": 9.822260940273748e-06,
+      "loss": 0.0,
+      "step": 1362
+    },
+    {
+      "epoch": 0.08783914416446478,
+      "grad_norm": 0.00069763105509275,
+      "learning_rate": 9.823259871877153e-06,
+      "loss": 0.0015,
+      "step": 1363
+    },
+    {
+      "epoch": 0.08790358961139395,
+      "grad_norm": 0.0025643315514372773,
+      "learning_rate": 9.824258070857386e-06,
+      "loss": 0.0,
+      "step": 1364
+    },
+    {
+      "epoch": 0.08796803505832312,
+      "grad_norm": 0.009340093648465875,
+      "learning_rate": 9.825255538288282e-06,
+      "loss": 0.0,
+      "step": 1365
+    },
+    {
+      "epoch": 0.08803248050525231,
+      "grad_norm": 0.015881523472943818,
+      "learning_rate": 9.826252275241316e-06,
+      "loss": 0.0001,
+      "step": 1366
+    },
+    {
+      "epoch": 0.08809692595218148,
+      "grad_norm": 0.017519239632679023,
+      "learning_rate": 9.827248282785611e-06,
+      "loss": 0.0,
+      "step": 1367
+    },
+    {
+      "epoch": 0.08816137139911065,
+      "grad_norm": 0.0016914266291122215,
+      "learning_rate": 9.828243561987945e-06,
+      "loss": 0.0,
+      "step": 1368
+    },
+    {
+      "epoch": 0.08822581684603983,
+      "grad_norm": 0.05371591804994142,
+      "learning_rate": 9.829238113912756e-06,
+      "loss": 0.0002,
+      "step": 1369
+    },
+    {
+      "epoch": 0.088290262292969,
+      "grad_norm": 0.002753948558404598,
+      "learning_rate": 9.83023193962215e-06,
+      "loss": 0.0,
+      "step": 1370
+    },
+    {
+      "epoch": 0.08835470773989818,
+      "grad_norm": 0.002328304292487111,
+      "learning_rate": 9.831225040175908e-06,
+      "loss": 0.0,
+      "step": 1371
+    },
+    {
+      "epoch": 0.08841915318682735,
+      "grad_norm": 0.001300685910040194,
+      "learning_rate": 9.832217416631494e-06,
+      "loss": 0.0,
+      "step": 1372
+    },
+    {
+      "epoch": 0.08848359863375653,
+      "grad_norm": 0.02016763377567185,
+      "learning_rate": 9.833209070044064e-06,
+      "loss": 0.0,
+      "step": 1373
+    },
+    {
+      "epoch": 0.0885480440806857,
+      "grad_norm": 0.0006239190779079104,
+      "learning_rate": 9.834200001466461e-06,
+      "loss": 0.0,
+      "step": 1374
+    },
+    {
+      "epoch": 0.08861248952761487,
+      "grad_norm": 0.17247998444561874,
+      "learning_rate": 9.835190211949236e-06,
+      "loss": 0.0006,
+      "step": 1375
+    },
+    {
+      "epoch": 0.08867693497454404,
+      "grad_norm": 0.02695488949887911,
+      "learning_rate": 9.83617970254065e-06,
+      "loss": 0.0,
+      "step": 1376
+    },
+    {
+      "epoch": 0.08874138042147323,
+      "grad_norm": 0.004205663295906084,
+      "learning_rate": 9.837168474286672e-06,
+      "loss": 0.0,
+      "step": 1377
+    },
+    {
+      "epoch": 0.0888058258684024,
+      "grad_norm": 0.23413007177194578,
+      "learning_rate": 9.838156528231003e-06,
+      "loss": 0.002,
+      "step": 1378
+    },
+    {
+      "epoch": 0.08887027131533157,
+      "grad_norm": 0.010107797521254137,
+      "learning_rate": 9.839143865415061e-06,
+      "loss": 0.0,
+      "step": 1379
+    },
+    {
+      "epoch": 0.08893471676226075,
+      "grad_norm": 0.22197280527991795,
+      "learning_rate": 9.840130486878012e-06,
+      "loss": 0.0034,
+      "step": 1380
+    },
+    {
+      "epoch": 0.08899916220918992,
+      "grad_norm": 0.011300355431752012,
+      "learning_rate": 9.841116393656755e-06,
+      "loss": 0.0,
+      "step": 1381
+    },
+    {
+      "epoch": 0.08906360765611909,
+      "grad_norm": 0.0006419008143487855,
+      "learning_rate": 9.84210158678594e-06,
+      "loss": 0.0,
+      "step": 1382
+    },
+    {
+      "epoch": 0.08912805310304826,
+      "grad_norm": 0.06378269833658995,
+      "learning_rate": 9.843086067297971e-06,
+      "loss": 0.0001,
+      "step": 1383
+    },
+    {
+      "epoch": 0.08919249854997745,
+      "grad_norm": 0.01747026315388834,
+      "learning_rate": 9.844069836223013e-06,
+      "loss": 0.0,
+      "step": 1384
+    },
+    {
+      "epoch": 0.08925694399690662,
+      "grad_norm": 0.03605956953767996,
+      "learning_rate": 9.845052894589003e-06,
+      "loss": 0.0001,
+      "step": 1385
+    },
+    {
+      "epoch": 0.08932138944383579,
+      "grad_norm": 0.008695684207934068,
+      "learning_rate": 9.846035243421644e-06,
+      "loss": 0.0,
+      "step": 1386
+    },
+    {
+      "epoch": 0.08938583489076497,
+      "grad_norm": 1.3870611063691909,
+      "learning_rate": 9.847016883744427e-06,
+      "loss": 0.009,
+      "step": 1387
+    },
+    {
+      "epoch": 0.08945028033769414,
+      "grad_norm": 0.0004034604019097972,
+      "learning_rate": 9.847997816578631e-06,
+      "loss": 0.0,
+      "step": 1388
+    },
+    {
+      "epoch": 0.08951472578462331,
+      "grad_norm": 0.24464011404589933,
+      "learning_rate": 9.848978042943318e-06,
+      "loss": 0.0054,
+      "step": 1389
+    },
+    {
+      "epoch": 0.08957917123155248,
+      "grad_norm": 0.006027357184420001,
+      "learning_rate": 9.849957563855364e-06,
+      "loss": 0.0,
+      "step": 1390
+    },
+    {
+      "epoch": 0.08964361667848167,
+      "grad_norm": 0.0032428463574663727,
+      "learning_rate": 9.85093638032944e-06,
+      "loss": 0.0,
+      "step": 1391
+    },
+    {
+      "epoch": 0.08970806212541084,
+      "grad_norm": 0.18261966601643068,
+      "learning_rate": 9.851914493378031e-06,
+      "loss": 0.0002,
+      "step": 1392
+    },
+    {
+      "epoch": 0.08977250757234001,
+      "grad_norm": 0.019904185252961712,
+      "learning_rate": 9.85289190401145e-06,
+      "loss": 0.0,
+      "step": 1393
+    },
+    {
+      "epoch": 0.0898369530192692,
+      "grad_norm": 0.006466688261427656,
+      "learning_rate": 9.853868613237822e-06,
+      "loss": 0.0,
+      "step": 1394
+    },
+    {
+      "epoch": 0.08990139846619837,
+      "grad_norm": 0.2036738682902856,
+      "learning_rate": 9.854844622063109e-06,
+      "loss": 0.0005,
+      "step": 1395
+    },
+    {
+      "epoch": 0.08996584391312754,
+      "grad_norm": 0.12498444221953549,
+      "learning_rate": 9.855819931491116e-06,
+      "loss": 0.0005,
+      "step": 1396
+    },
+    {
+      "epoch": 0.0900302893600567,
+      "grad_norm": 0.25550481384225326,
+      "learning_rate": 9.85679454252348e-06,
+      "loss": 0.0002,
+      "step": 1397
+    },
+    {
+      "epoch": 0.09009473480698589,
+      "grad_norm": 0.005078154637463445,
+      "learning_rate": 9.857768456159699e-06,
+      "loss": 0.0,
+      "step": 1398
+    },
+    {
+      "epoch": 0.09015918025391506,
+      "grad_norm": 0.006486574987916103,
+      "learning_rate": 9.85874167339712e-06,
+      "loss": 0.0,
+      "step": 1399
+    },
+    {
+      "epoch": 0.09022362570084423,
+      "grad_norm": 0.1849051672410749,
+      "learning_rate": 9.859714195230952e-06,
+      "loss": 0.0003,
+      "step": 1400
+    },
+    {
+      "epoch": 0.09028807114777342,
+      "grad_norm": 1.157685394105611,
+      "learning_rate": 9.860686022654276e-06,
+      "loss": 0.0045,
+      "step": 1401
+    },
+    {
+      "epoch": 0.09035251659470259,
+      "grad_norm": 0.05039861841158964,
+      "learning_rate": 9.861657156658047e-06,
+      "loss": 0.0002,
+      "step": 1402
+    },
+    {
+      "epoch": 0.09041696204163176,
+      "grad_norm": 0.003640545123375758,
+      "learning_rate": 9.862627598231097e-06,
+      "loss": 0.0,
+      "step": 1403
+    },
+    {
+      "epoch": 0.09048140748856093,
+      "grad_norm": 0.18785870532732124,
+      "learning_rate": 9.863597348360145e-06,
+      "loss": 0.0005,
+      "step": 1404
+    },
+    {
+      "epoch": 0.09054585293549011,
+      "grad_norm": 0.02162690258325965,
+      "learning_rate": 9.864566408029807e-06,
+      "loss": 0.0,
+      "step": 1405
+    },
+    {
+      "epoch": 0.09061029838241928,
+      "grad_norm": 0.3182465991267757,
+      "learning_rate": 9.86553477822259e-06,
+      "loss": 0.0013,
+      "step": 1406
+    },
+    {
+      "epoch": 0.09067474382934845,
+      "grad_norm": 0.00506957712645276,
+      "learning_rate": 9.866502459918909e-06,
+      "loss": 0.0,
+      "step": 1407
+    },
+    {
+      "epoch": 0.09073918927627764,
+      "grad_norm": 0.0008336997610315851,
+      "learning_rate": 9.867469454097092e-06,
+      "loss": 0.0,
+      "step": 1408
+    },
+    {
+      "epoch": 0.09080363472320681,
+      "grad_norm": 0.044061586416188465,
+      "learning_rate": 9.86843576173338e-06,
+      "loss": 0.0004,
+      "step": 1409
+    },
+    {
+      "epoch": 0.09086808017013598,
+      "grad_norm": 0.007409890122830501,
+      "learning_rate": 9.869401383801938e-06,
+      "loss": 0.0,
+      "step": 1410
+    },
+    {
+      "epoch": 0.09093252561706515,
+      "grad_norm": 0.018156252708728265,
+      "learning_rate": 9.870366321274855e-06,
+      "loss": 0.0,
+      "step": 1411
+    },
+    {
+      "epoch": 0.09099697106399433,
+      "grad_norm": 0.0006506659029600226,
+      "learning_rate": 9.871330575122158e-06,
+      "loss": 0.0,
+      "step": 1412
+    },
+    {
+      "epoch": 0.0910614165109235,
+      "grad_norm": 0.0008328508705257589,
+      "learning_rate": 9.872294146311813e-06,
+      "loss": 0.0,
+      "step": 1413
+    },
+    {
+      "epoch": 0.09112586195785267,
+      "grad_norm": 0.0007773650111156604,
+      "learning_rate": 9.87325703580973e-06,
+      "loss": 0.0,
+      "step": 1414
+    },
+    {
+      "epoch": 0.09119030740478186,
+      "grad_norm": 0.0012217894950347168,
+      "learning_rate": 9.874219244579777e-06,
+      "loss": 0.0,
+      "step": 1415
+    },
+    {
+      "epoch": 0.09125475285171103,
+      "grad_norm": 0.0480236903713464,
+      "learning_rate": 9.875180773583766e-06,
+      "loss": 0.0001,
+      "step": 1416
+    },
+    {
+      "epoch": 0.0913191982986402,
+      "grad_norm": 0.001227121457690193,
+      "learning_rate": 9.876141623781484e-06,
+      "loss": 0.0,
+      "step": 1417
+    },
+    {
+      "epoch": 0.09138364374556937,
+      "grad_norm": 0.004099157581286233,
+      "learning_rate": 9.877101796130682e-06,
+      "loss": 0.0,
+      "step": 1418
+    },
+    {
+      "epoch": 0.09144808919249856,
+      "grad_norm": 0.001988160165573771,
+      "learning_rate": 9.878061291587085e-06,
+      "loss": 0.0,
+      "step": 1419
+    },
+    {
+      "epoch": 0.09151253463942773,
+      "grad_norm": 0.00018577178633780048,
+      "learning_rate": 9.879020111104401e-06,
+      "loss": 0.0,
+      "step": 1420
+    },
+    {
+      "epoch": 0.0915769800863569,
+      "grad_norm": 0.22095755874515524,
+      "learning_rate": 9.87997825563432e-06,
+      "loss": 0.0003,
+      "step": 1421
+    },
+    {
+      "epoch": 0.09164142553328607,
+      "grad_norm": 2.3769321613123924,
+      "learning_rate": 9.880935726126525e-06,
+      "loss": 0.015,
+      "step": 1422
+    },
+    {
+      "epoch": 0.09170587098021525,
+      "grad_norm": 0.018600959082779225,
+      "learning_rate": 9.8818925235287e-06,
+      "loss": 0.0001,
+      "step": 1423
+    },
+    {
+      "epoch": 0.09177031642714442,
+      "grad_norm": 0.08022021561964109,
+      "learning_rate": 9.882848648786525e-06,
+      "loss": 0.0001,
+      "step": 1424
+    },
+    {
+      "epoch": 0.09183476187407359,
+      "grad_norm": 0.20565971377159456,
+      "learning_rate": 9.883804102843692e-06,
+      "loss": 0.0002,
+      "step": 1425
+    },
+    {
+      "epoch": 0.09189920732100278,
+      "grad_norm": 0.9405621745271036,
+      "learning_rate": 9.884758886641908e-06,
+      "loss": 0.0044,
+      "step": 1426
+    },
+    {
+      "epoch": 0.09196365276793195,
+      "grad_norm": 0.0004957289890303268,
+      "learning_rate": 9.885713001120897e-06,
+      "loss": 0.0,
+      "step": 1427
+    },
+    {
+      "epoch": 0.09202809821486112,
+      "grad_norm": 0.0017034345264256923,
+      "learning_rate": 9.886666447218409e-06,
+      "loss": 0.0,
+      "step": 1428
+    },
+    {
+      "epoch": 0.09209254366179029,
+      "grad_norm": 0.0339874361429872,
+      "learning_rate": 9.887619225870228e-06,
+      "loss": 0.0001,
+      "step": 1429
+    },
+    {
+      "epoch": 0.09215698910871947,
+      "grad_norm": 0.3589929540229609,
+      "learning_rate": 9.888571338010167e-06,
+      "loss": 0.0032,
+      "step": 1430
+    },
+    {
+      "epoch": 0.09222143455564864,
+      "grad_norm": 0.005050188803306223,
+      "learning_rate": 9.889522784570087e-06,
+      "loss": 0.0001,
+      "step": 1431
+    },
+    {
+      "epoch": 0.09228588000257781,
+      "grad_norm": 0.15730377931145592,
+      "learning_rate": 9.890473566479896e-06,
+      "loss": 0.0005,
+      "step": 1432
+    },
+    {
+      "epoch": 0.092350325449507,
+      "grad_norm": 0.0005229967634722769,
+      "learning_rate": 9.891423684667549e-06,
+      "loss": 0.0,
+      "step": 1433
+    },
+    {
+      "epoch": 0.09241477089643617,
+      "grad_norm": 0.005672397433659443,
+      "learning_rate": 9.892373140059061e-06,
+      "loss": 0.0001,
+      "step": 1434
+    },
+    {
+      "epoch": 0.09247921634336534,
+      "grad_norm": 0.0002519493904518232,
+      "learning_rate": 9.893321933578516e-06,
+      "loss": 0.0,
+      "step": 1435
+    },
+    {
+      "epoch": 0.09254366179029451,
+      "grad_norm": 0.027063703592058647,
+      "learning_rate": 9.894270066148061e-06,
+      "loss": 0.0001,
+      "step": 1436
+    },
+    {
+      "epoch": 0.0926081072372237,
+      "grad_norm": 0.026686603489626357,
+      "learning_rate": 9.895217538687919e-06,
+      "loss": 0.0003,
+      "step": 1437
+    },
+    {
+      "epoch": 0.09267255268415286,
+      "grad_norm": 0.000908574260126288,
+      "learning_rate": 9.89616435211639e-06,
+      "loss": 0.0,
+      "step": 1438
+    },
+    {
+      "epoch": 0.09273699813108204,
+      "grad_norm": 0.02027503235190547,
+      "learning_rate": 9.897110507349863e-06,
+      "loss": 0.0,
+      "step": 1439
+    },
+    {
+      "epoch": 0.09280144357801122,
+      "grad_norm": 0.019177680948524713,
+      "learning_rate": 9.898056005302818e-06,
+      "loss": 0.0001,
+      "step": 1440
+    },
+    {
+      "epoch": 0.09286588902494039,
+      "grad_norm": 0.010158676693733085,
+      "learning_rate": 9.899000846887821e-06,
+      "loss": 0.0,
+      "step": 1441
+    },
+    {
+      "epoch": 0.09293033447186956,
+      "grad_norm": 0.004541012965222062,
+      "learning_rate": 9.899945033015553e-06,
+      "loss": 0.0,
+      "step": 1442
+    },
+    {
+      "epoch": 0.09299477991879873,
+      "grad_norm": 0.002712991317341899,
+      "learning_rate": 9.900888564594789e-06,
+      "loss": 0.0,
+      "step": 1443
+    },
+    {
+      "epoch": 0.09305922536572792,
+      "grad_norm": 0.020957469761172257,
+      "learning_rate": 9.901831442532423e-06,
+      "loss": 0.0001,
+      "step": 1444
+    },
+    {
+      "epoch": 0.09312367081265709,
+      "grad_norm": 0.00033094373180639466,
+      "learning_rate": 9.902773667733459e-06,
+      "loss": 0.0,
+      "step": 1445
+    },
+    {
+      "epoch": 0.09318811625958626,
+      "grad_norm": 0.3864754190635272,
+      "learning_rate": 9.90371524110103e-06,
+      "loss": 0.0007,
+      "step": 1446
+    },
+    {
+      "epoch": 0.09325256170651544,
+      "grad_norm": 0.01617892127793432,
+      "learning_rate": 9.904656163536388e-06,
+      "loss": 0.0001,
+      "step": 1447
+    },
+    {
+      "epoch": 0.09331700715344461,
+      "grad_norm": 0.012048904136151128,
+      "learning_rate": 9.905596435938926e-06,
+      "loss": 0.0,
+      "step": 1448
+    },
+    {
+      "epoch": 0.09338145260037378,
+      "grad_norm": 0.0011809616144378202,
+      "learning_rate": 9.906536059206166e-06,
+      "loss": 0.0,
+      "step": 1449
+    },
+    {
+      "epoch": 0.09344589804730295,
+      "grad_norm": 0.0017681243992767163,
+      "learning_rate": 9.907475034233778e-06,
+      "loss": 0.0,
+      "step": 1450
+    },
+    {
+      "epoch": 0.09351034349423214,
+      "grad_norm": 0.0017163327437415454,
+      "learning_rate": 9.908413361915576e-06,
+      "loss": 0.0,
+      "step": 1451
+    },
+    {
+      "epoch": 0.09357478894116131,
+      "grad_norm": 0.01384325150685238,
+      "learning_rate": 9.90935104314353e-06,
+      "loss": 0.0,
+      "step": 1452
+    },
+    {
+      "epoch": 0.09363923438809048,
+      "grad_norm": 0.0867655132908843,
+      "learning_rate": 9.910288078807764e-06,
+      "loss": 0.0001,
+      "step": 1453
+    },
+    {
+      "epoch": 0.09370367983501966,
+      "grad_norm": 0.7641827369741804,
+      "learning_rate": 9.911224469796568e-06,
+      "loss": 0.0075,
+      "step": 1454
+    },
+    {
+      "epoch": 0.09376812528194883,
+      "grad_norm": 0.060422641060364934,
+      "learning_rate": 9.912160216996399e-06,
+      "loss": 0.0002,
+      "step": 1455
+    },
+    {
+      "epoch": 0.093832570728878,
+      "grad_norm": 2.010386913125508,
+      "learning_rate": 9.913095321291884e-06,
+      "loss": 0.0106,
+      "step": 1456
+    },
+    {
+      "epoch": 0.09389701617580717,
+      "grad_norm": 0.006303394784412231,
+      "learning_rate": 9.914029783565832e-06,
+      "loss": 0.0,
+      "step": 1457
+    },
+    {
+      "epoch": 0.09396146162273636,
+      "grad_norm": 0.009684100579074902,
+      "learning_rate": 9.914963604699233e-06,
+      "loss": 0.0,
+      "step": 1458
+    },
+    {
+      "epoch": 0.09402590706966553,
+      "grad_norm": 0.008250804826870545,
+      "learning_rate": 9.915896785571264e-06,
+      "loss": 0.0,
+      "step": 1459
+    },
+    {
+      "epoch": 0.0940903525165947,
+      "grad_norm": 0.0009466256485816172,
+      "learning_rate": 9.9168293270593e-06,
+      "loss": 0.0,
+      "step": 1460
+    },
+    {
+      "epoch": 0.09415479796352387,
+      "grad_norm": 0.048614013439532935,
+      "learning_rate": 9.917761230038905e-06,
+      "loss": 0.0017,
+      "step": 1461
+    },
+    {
+      "epoch": 0.09421924341045305,
+      "grad_norm": 0.01888382417348389,
+      "learning_rate": 9.918692495383852e-06,
+      "loss": 0.0,
+      "step": 1462
+    },
+    {
+      "epoch": 0.09428368885738223,
+      "grad_norm": 0.05641691448848606,
+      "learning_rate": 9.91962312396612e-06,
+      "loss": 0.0002,
+      "step": 1463
+    },
+    {
+      "epoch": 0.0943481343043114,
+      "grad_norm": 0.42407625379866715,
+      "learning_rate": 9.920553116655903e-06,
+      "loss": 0.0027,
+      "step": 1464
+    },
+    {
+      "epoch": 0.09441257975124058,
+      "grad_norm": 0.010436752034464663,
+      "learning_rate": 9.921482474321605e-06,
+      "loss": 0.0,
+      "step": 1465
+    },
+    {
+      "epoch": 0.09447702519816975,
+      "grad_norm": 0.8174559493256072,
+      "learning_rate": 9.922411197829863e-06,
+      "loss": 0.0033,
+      "step": 1466
+    },
+    {
+      "epoch": 0.09454147064509892,
+      "grad_norm": 0.0583152390449536,
+      "learning_rate": 9.923339288045531e-06,
+      "loss": 0.0,
+      "step": 1467
+    },
+    {
+      "epoch": 0.09460591609202809,
+      "grad_norm": 0.306076587996211,
+      "learning_rate": 9.924266745831703e-06,
+      "loss": 0.0071,
+      "step": 1468
+    },
+    {
+      "epoch": 0.09467036153895728,
+      "grad_norm": 0.23739539088773876,
+      "learning_rate": 9.925193572049705e-06,
+      "loss": 0.0005,
+      "step": 1469
+    },
+    {
+      "epoch": 0.09473480698588645,
+      "grad_norm": 0.250875088850879,
+      "learning_rate": 9.926119767559105e-06,
+      "loss": 0.0006,
+      "step": 1470
+    },
+    {
+      "epoch": 0.09479925243281562,
+      "grad_norm": 0.042556311330388334,
+      "learning_rate": 9.927045333217721e-06,
+      "loss": 0.0,
+      "step": 1471
+    },
+    {
+      "epoch": 0.0948636978797448,
+      "grad_norm": 0.03793214656428974,
+      "learning_rate": 9.927970269881614e-06,
+      "loss": 0.0,
+      "step": 1472
+    },
+    {
+      "epoch": 0.09492814332667397,
+      "grad_norm": 0.28249577284916433,
+      "learning_rate": 9.928894578405113e-06,
+      "loss": 0.0012,
+      "step": 1473
+    },
+    {
+      "epoch": 0.09499258877360314,
+      "grad_norm": 0.17475368245792194,
+      "learning_rate": 9.929818259640796e-06,
+      "loss": 0.0003,
+      "step": 1474
+    },
+    {
+      "epoch": 0.09505703422053231,
+      "grad_norm": 0.1445167057348758,
+      "learning_rate": 9.930741314439511e-06,
+      "loss": 0.0002,
+      "step": 1475
+    },
+    {
+      "epoch": 0.0951214796674615,
+      "grad_norm": 0.0236630083799695,
+      "learning_rate": 9.931663743650381e-06,
+      "loss": 0.0,
+      "step": 1476
+    },
+    {
+      "epoch": 0.09518592511439067,
+      "grad_norm": 0.029098778789947974,
+      "learning_rate": 9.932585548120796e-06,
+      "loss": 0.0001,
+      "step": 1477
+    },
+    {
+      "epoch": 0.09525037056131984,
+      "grad_norm": 0.009179967320519636,
+      "learning_rate": 9.933506728696428e-06,
+      "loss": 0.0001,
+      "step": 1478
+    },
+    {
+      "epoch": 0.09531481600824902,
+      "grad_norm": 0.001304011320141148,
+      "learning_rate": 9.934427286221235e-06,
+      "loss": 0.0,
+      "step": 1479
+    },
+    {
+      "epoch": 0.0953792614551782,
+      "grad_norm": 0.037167467586508234,
+      "learning_rate": 9.93534722153746e-06,
+      "loss": 0.0001,
+      "step": 1480
+    },
+    {
+      "epoch": 0.09544370690210736,
+      "grad_norm": 0.004253996176845752,
+      "learning_rate": 9.936266535485644e-06,
+      "loss": 0.0,
+      "step": 1481
+    },
+    {
+      "epoch": 0.09550815234903653,
+      "grad_norm": 0.012284348723612233,
+      "learning_rate": 9.937185228904622e-06,
+      "loss": 0.0,
+      "step": 1482
+    },
+    {
+      "epoch": 0.09557259779596572,
+      "grad_norm": 0.23995914820594377,
+      "learning_rate": 9.938103302631535e-06,
+      "loss": 0.0005,
+      "step": 1483
+    },
+    {
+      "epoch": 0.09563704324289489,
+      "grad_norm": 0.4843605408510047,
+      "learning_rate": 9.939020757501826e-06,
+      "loss": 0.0029,
+      "step": 1484
+    },
+    {
+      "epoch": 0.09570148868982406,
+      "grad_norm": 0.0006049933104365288,
+      "learning_rate": 9.939937594349256e-06,
+      "loss": 0.0,
+      "step": 1485
+    },
+    {
+      "epoch": 0.09576593413675324,
+      "grad_norm": 0.007013362420659339,
+      "learning_rate": 9.940853814005894e-06,
+      "loss": 0.0001,
+      "step": 1486
+    },
+    {
+      "epoch": 0.09583037958368242,
+      "grad_norm": 0.057614378941506235,
+      "learning_rate": 9.941769417302139e-06,
+      "loss": 0.0001,
+      "step": 1487
+    },
+    {
+      "epoch": 0.09589482503061159,
+      "grad_norm": 0.016872804892730364,
+      "learning_rate": 9.942684405066712e-06,
+      "loss": 0.0002,
+      "step": 1488
+    },
+    {
+      "epoch": 0.09595927047754076,
+      "grad_norm": 0.02935959737771146,
+      "learning_rate": 9.94359877812666e-06,
+      "loss": 0.0002,
+      "step": 1489
+    },
+    {
+      "epoch": 0.09602371592446994,
+      "grad_norm": 0.002290192727872697,
+      "learning_rate": 9.944512537307367e-06,
+      "loss": 0.0,
+      "step": 1490
+    },
+    {
+      "epoch": 0.09608816137139911,
+      "grad_norm": 0.022548370604501426,
+      "learning_rate": 9.945425683432555e-06,
+      "loss": 0.0001,
+      "step": 1491
+    },
+    {
+      "epoch": 0.09615260681832828,
+      "grad_norm": 0.007626670351049349,
+      "learning_rate": 9.946338217324292e-06,
+      "loss": 0.0,
+      "step": 1492
+    },
+    {
+      "epoch": 0.09621705226525747,
+      "grad_norm": 0.006175822573090903,
+      "learning_rate": 9.94725013980299e-06,
+      "loss": 0.0001,
+      "step": 1493
+    },
+    {
+      "epoch": 0.09628149771218664,
+      "grad_norm": 0.038742508647196895,
+      "learning_rate": 9.948161451687416e-06,
+      "loss": 0.0004,
+      "step": 1494
+    },
+    {
+      "epoch": 0.09634594315911581,
+      "grad_norm": 0.0005905164246859292,
+      "learning_rate": 9.949072153794689e-06,
+      "loss": 0.0,
+      "step": 1495
+    },
+    {
+      "epoch": 0.09641038860604498,
+      "grad_norm": 0.131122069226978,
+      "learning_rate": 9.949982246940294e-06,
+      "loss": 0.0005,
+      "step": 1496
+    },
+    {
+      "epoch": 0.09647483405297416,
+      "grad_norm": 0.0005022347779479968,
+      "learning_rate": 9.950891731938082e-06,
+      "loss": 0.0,
+      "step": 1497
+    },
+    {
+      "epoch": 0.09653927949990333,
+      "grad_norm": 0.0001437653529467265,
+      "learning_rate": 9.951800609600263e-06,
+      "loss": 0.0,
+      "step": 1498
+    },
+    {
+      "epoch": 0.0966037249468325,
+      "grad_norm": 0.0010122622831039668,
+      "learning_rate": 9.952708880737434e-06,
+      "loss": 0.0,
+      "step": 1499
+    },
+    {
+      "epoch": 0.09666817039376169,
+      "grad_norm": 0.33948919282434475,
+      "learning_rate": 9.953616546158563e-06,
+      "loss": 0.0024,
+      "step": 1500
+    },
+    {
+      "epoch": 0.09673261584069086,
+      "grad_norm": 0.003581175820871328,
+      "learning_rate": 9.954523606671003e-06,
+      "loss": 0.0,
+      "step": 1501
+    },
+    {
+      "epoch": 0.09679706128762003,
+      "grad_norm": 0.00042302368364188466,
+      "learning_rate": 9.955430063080492e-06,
+      "loss": 0.0,
+      "step": 1502
+    },
+    {
+      "epoch": 0.0968615067345492,
+      "grad_norm": 0.0017871230098158303,
+      "learning_rate": 9.956335916191164e-06,
+      "loss": 0.0,
+      "step": 1503
+    },
+    {
+      "epoch": 0.09692595218147838,
+      "grad_norm": 0.0777799107536665,
+      "learning_rate": 9.95724116680554e-06,
+      "loss": 0.0005,
+      "step": 1504
+    },
+    {
+      "epoch": 0.09699039762840755,
+      "grad_norm": 0.0012527611164778536,
+      "learning_rate": 9.958145815724548e-06,
+      "loss": 0.0,
+      "step": 1505
+    },
+    {
+      "epoch": 0.09705484307533672,
+      "grad_norm": 0.24797657097956194,
+      "learning_rate": 9.95904986374752e-06,
+      "loss": 0.0009,
+      "step": 1506
+    },
+    {
+      "epoch": 0.0971192885222659,
+      "grad_norm": 0.0014533547584727054,
+      "learning_rate": 9.95995331167219e-06,
+      "loss": 0.0,
+      "step": 1507
+    },
+    {
+      "epoch": 0.09718373396919508,
+      "grad_norm": 0.0015967057251301884,
+      "learning_rate": 9.96085616029471e-06,
+      "loss": 0.0,
+      "step": 1508
+    },
+    {
+      "epoch": 0.09724817941612425,
+      "grad_norm": 0.004208869780398821,
+      "learning_rate": 9.961758410409647e-06,
+      "loss": 0.0,
+      "step": 1509
+    },
+    {
+      "epoch": 0.09731262486305342,
+      "grad_norm": 0.01696486413837203,
+      "learning_rate": 9.96266006280999e-06,
+      "loss": 0.0,
+      "step": 1510
+    },
+    {
+      "epoch": 0.0973770703099826,
+      "grad_norm": 0.0018957704558056011,
+      "learning_rate": 9.963561118287149e-06,
+      "loss": 0.0,
+      "step": 1511
+    },
+    {
+      "epoch": 0.09744151575691178,
+      "grad_norm": 0.0040607474789808124,
+      "learning_rate": 9.96446157763097e-06,
+      "loss": 0.0,
+      "step": 1512
+    },
+    {
+      "epoch": 0.09750596120384095,
+      "grad_norm": 0.0011904989553570007,
+      "learning_rate": 9.965361441629727e-06,
+      "loss": 0.0,
+      "step": 1513
+    },
+    {
+      "epoch": 0.09757040665077012,
+      "grad_norm": 0.01142907492530569,
+      "learning_rate": 9.966260711070133e-06,
+      "loss": 0.0001,
+      "step": 1514
+    },
+    {
+      "epoch": 0.0976348520976993,
+      "grad_norm": 0.0046347557260846255,
+      "learning_rate": 9.967159386737342e-06,
+      "loss": 0.0,
+      "step": 1515
+    },
+    {
+      "epoch": 0.09769929754462847,
+      "grad_norm": 0.0010111396870745295,
+      "learning_rate": 9.968057469414957e-06,
+      "loss": 0.0,
+      "step": 1516
+    },
+    {
+      "epoch": 0.09776374299155764,
+      "grad_norm": 0.0036916821478951735,
+      "learning_rate": 9.968954959885024e-06,
+      "loss": 0.0,
+      "step": 1517
+    },
+    {
+      "epoch": 0.09782818843848683,
+      "grad_norm": 0.0547688530482111,
+      "learning_rate": 9.969851858928051e-06,
+      "loss": 0.0003,
+      "step": 1518
+    },
+    {
+      "epoch": 0.097892633885416,
+      "grad_norm": 0.00326234239409605,
+      "learning_rate": 9.970748167323e-06,
+      "loss": 0.0,
+      "step": 1519
+    },
+    {
+      "epoch": 0.09795707933234517,
+      "grad_norm": 0.004320389485947813,
+      "learning_rate": 9.971643885847294e-06,
+      "loss": 0.0,
+      "step": 1520
+    },
+    {
+      "epoch": 0.09802152477927434,
+      "grad_norm": 0.002041762479109093,
+      "learning_rate": 9.972539015276823e-06,
+      "loss": 0.0,
+      "step": 1521
+    },
+    {
+      "epoch": 0.09808597022620352,
+      "grad_norm": 0.06881354887199352,
+      "learning_rate": 9.97343355638595e-06,
+      "loss": 0.0001,
+      "step": 1522
+    },
+    {
+      "epoch": 0.0981504156731327,
+      "grad_norm": 0.06540829750124166,
+      "learning_rate": 9.974327509947506e-06,
+      "loss": 0.0001,
+      "step": 1523
+    },
+    {
+      "epoch": 0.09821486112006186,
+      "grad_norm": 0.00757338516581833,
+      "learning_rate": 9.975220876732808e-06,
+      "loss": 0.0,
+      "step": 1524
+    },
+    {
+      "epoch": 0.09827930656699105,
+      "grad_norm": 0.008557299181676596,
+      "learning_rate": 9.97611365751165e-06,
+      "loss": 0.0001,
+      "step": 1525
+    },
+    {
+      "epoch": 0.09834375201392022,
+      "grad_norm": 0.0004520848970033896,
+      "learning_rate": 9.977005853052309e-06,
+      "loss": 0.0,
+      "step": 1526
+    },
+    {
+      "epoch": 0.09840819746084939,
+      "grad_norm": 0.0021838000738325022,
+      "learning_rate": 9.977897464121563e-06,
+      "loss": 0.0,
+      "step": 1527
+    },
+    {
+      "epoch": 0.09847264290777856,
+      "grad_norm": 0.003258614152280576,
+      "learning_rate": 9.978788491484673e-06,
+      "loss": 0.0,
+      "step": 1528
+    },
+    {
+      "epoch": 0.09853708835470774,
+      "grad_norm": 0.0015698688676021998,
+      "learning_rate": 9.979678935905403e-06,
+      "loss": 0.0,
+      "step": 1529
+    },
+    {
+      "epoch": 0.09860153380163691,
+      "grad_norm": 0.01185302928350965,
+      "learning_rate": 9.98056879814602e-06,
+      "loss": 0.0001,
+      "step": 1530
+    },
+    {
+      "epoch": 0.09866597924856609,
+      "grad_norm": 0.001710585233948929,
+      "learning_rate": 9.981458078967295e-06,
+      "loss": 0.0,
+      "step": 1531
+    },
+    {
+      "epoch": 0.09873042469549527,
+      "grad_norm": 0.0593000578214577,
+      "learning_rate": 9.982346779128506e-06,
+      "loss": 0.0001,
+      "step": 1532
+    },
+    {
+      "epoch": 0.09879487014242444,
+      "grad_norm": 0.0037362333247285434,
+      "learning_rate": 9.983234899387453e-06,
+      "loss": 0.0,
+      "step": 1533
+    },
+    {
+      "epoch": 0.09885931558935361,
+      "grad_norm": 0.0004543681892485397,
+      "learning_rate": 9.984122440500443e-06,
+      "loss": 0.0,
+      "step": 1534
+    },
+    {
+      "epoch": 0.09892376103628278,
+      "grad_norm": 9.713128271191101e-05,
+      "learning_rate": 9.985009403222313e-06,
+      "loss": 0.0,
+      "step": 1535
+    },
+    {
+      "epoch": 0.09898820648321197,
+      "grad_norm": 0.038945052790278024,
+      "learning_rate": 9.98589578830642e-06,
+      "loss": 0.0002,
+      "step": 1536
+    },
+    {
+      "epoch": 0.09905265193014114,
+      "grad_norm": 0.00021423168057826362,
+      "learning_rate": 9.98678159650465e-06,
+      "loss": 0.0,
+      "step": 1537
+    },
+    {
+      "epoch": 0.0991170973770703,
+      "grad_norm": 0.017880657638670156,
+      "learning_rate": 9.987666828567429e-06,
+      "loss": 0.0001,
+      "step": 1538
+    },
+    {
+      "epoch": 0.09918154282399949,
+      "grad_norm": 0.01319963089075672,
+      "learning_rate": 9.98855148524371e-06,
+      "loss": 0.0,
+      "step": 1539
+    },
+    {
+      "epoch": 0.09924598827092866,
+      "grad_norm": 0.0003586439757768937,
+      "learning_rate": 9.989435567280993e-06,
+      "loss": 0.0,
+      "step": 1540
+    },
+    {
+      "epoch": 0.09931043371785783,
+      "grad_norm": 0.36412274922213167,
+      "learning_rate": 9.990319075425318e-06,
+      "loss": 0.0026,
+      "step": 1541
+    },
+    {
+      "epoch": 0.099374879164787,
+      "grad_norm": 0.24167075712069122,
+      "learning_rate": 9.991202010421277e-06,
+      "loss": 0.0021,
+      "step": 1542
+    },
+    {
+      "epoch": 0.09943932461171619,
+      "grad_norm": 0.008199802468034723,
+      "learning_rate": 9.992084373012014e-06,
+      "loss": 0.0001,
+      "step": 1543
+    },
+    {
+      "epoch": 0.09950377005864536,
+      "grad_norm": 0.08797115768979515,
+      "learning_rate": 9.992966163939223e-06,
+      "loss": 0.0006,
+      "step": 1544
+    },
+    {
+      "epoch": 0.09956821550557453,
+      "grad_norm": 0.00023174358708018033,
+      "learning_rate": 9.993847383943163e-06,
+      "loss": 0.0,
+      "step": 1545
+    },
+    {
+      "epoch": 0.0996326609525037,
+      "grad_norm": 0.11313614101504088,
+      "learning_rate": 9.994728033762655e-06,
+      "loss": 0.0002,
+      "step": 1546
+    },
+    {
+      "epoch": 0.09969710639943288,
+      "grad_norm": 0.07457094638349225,
+      "learning_rate": 9.995608114135087e-06,
+      "loss": 0.0002,
+      "step": 1547
+    },
+    {
+      "epoch": 0.09976155184636205,
+      "grad_norm": 0.012721014483883464,
+      "learning_rate": 9.996487625796414e-06,
+      "loss": 0.0001,
+      "step": 1548
+    },
+    {
+      "epoch": 0.09982599729329122,
+      "grad_norm": 0.0012972554044241161,
+      "learning_rate": 9.997366569481168e-06,
+      "loss": 0.0,
+      "step": 1549
+    },
+    {
+      "epoch": 0.09989044274022041,
+      "grad_norm": 0.0043032076280890485,
+      "learning_rate": 9.998244945922459e-06,
+      "loss": 0.0,
+      "step": 1550
+    },
+    {
+      "epoch": 0.09995488818714958,
+      "grad_norm": 0.0019208322998490745,
+      "learning_rate": 9.999122755851977e-06,
+      "loss": 0.0,
+      "step": 1551
+    },
+    {
+      "epoch": 0.10001933363407875,
+      "grad_norm": 0.0025537528524414454,
+      "learning_rate": 1e-05,
+      "loss": 0.0,
+      "step": 1552
+    },
+    {
+      "epoch": 0.10008377908100792,
+      "grad_norm": 0.01819473357393409,
+      "learning_rate": 1e-05,
+      "loss": 0.0001,
+      "step": 1553
+    },
+    {
+      "epoch": 0.1001482245279371,
+      "grad_norm": 0.023582341253408183,
+      "learning_rate": 9.999283924095955e-06,
+      "loss": 0.0001,
+      "step": 1554
+    },
+    {
+      "epoch": 0.10021266997486628,
+      "grad_norm": 0.0006501829399390779,
+      "learning_rate": 9.99856784819191e-06,
+      "loss": 0.0,
+      "step": 1555
+    },
+    {
+      "epoch": 0.10027711542179545,
+      "grad_norm": 0.0045095002079804025,
+      "learning_rate": 9.997851772287864e-06,
+      "loss": 0.0,
+      "step": 1556
+    },
+    {
+      "epoch": 0.10034156086872463,
+      "grad_norm": 0.0048714429017706305,
+      "learning_rate": 9.997135696383818e-06,
+      "loss": 0.0,
+      "step": 1557
+    },
+    {
+      "epoch": 0.1004060063156538,
+      "grad_norm": 0.06844209026375658,
+      "learning_rate": 9.99641962047977e-06,
+      "loss": 0.0002,
+      "step": 1558
+    },
+    {
+      "epoch": 0.10047045176258297,
+      "grad_norm": 0.011684224684139077,
+      "learning_rate": 9.995703544575725e-06,
+      "loss": 0.0,
+      "step": 1559
+    },
+    {
+      "epoch": 0.10053489720951214,
+      "grad_norm": 0.02074485054202999,
+      "learning_rate": 9.99498746867168e-06,
+      "loss": 0.0,
+      "step": 1560
+    },
+    {
+      "epoch": 0.10059934265644133,
+      "grad_norm": 0.0688557315716873,
+      "learning_rate": 9.994271392767634e-06,
+      "loss": 0.0002,
+      "step": 1561
+    },
+    {
+      "epoch": 0.1006637881033705,
+      "grad_norm": 0.04624153556322279,
+      "learning_rate": 9.993555316863588e-06,
+      "loss": 0.0001,
+      "step": 1562
+    },
+    {
+      "epoch": 0.10072823355029967,
+      "grad_norm": 0.008817848692303166,
+      "learning_rate": 9.992839240959542e-06,
+      "loss": 0.0,
+      "step": 1563
+    },
+    {
+      "epoch": 0.10079267899722885,
+      "grad_norm": 0.41477144476881117,
+      "learning_rate": 9.992123165055496e-06,
+      "loss": 0.0025,
+      "step": 1564
+    },
+    {
+      "epoch": 0.10085712444415802,
+      "grad_norm": 0.0028627288818668436,
+      "learning_rate": 9.99140708915145e-06,
+      "loss": 0.0,
+      "step": 1565
+    },
+    {
+      "epoch": 0.10092156989108719,
+      "grad_norm": 0.8068578764666766,
+      "learning_rate": 9.990691013247405e-06,
+      "loss": 0.0037,
+      "step": 1566
+    },
+    {
+      "epoch": 0.10098601533801636,
+      "grad_norm": 0.00030946291876683197,
+      "learning_rate": 9.98997493734336e-06,
+      "loss": 0.0,
+      "step": 1567
+    },
+    {
+      "epoch": 0.10105046078494555,
+      "grad_norm": 0.510590744176873,
+      "learning_rate": 9.989258861439314e-06,
+      "loss": 0.0015,
+      "step": 1568
+    },
+    {
+      "epoch": 0.10111490623187472,
+      "grad_norm": 0.005179627251681472,
+      "learning_rate": 9.988542785535268e-06,
+      "loss": 0.0,
+      "step": 1569
+    },
+    {
+      "epoch": 0.10117935167880389,
+      "grad_norm": 0.5898086335295064,
+      "learning_rate": 9.987826709631222e-06,
+      "loss": 0.0034,
+      "step": 1570
+    },
+    {
+      "epoch": 0.10124379712573307,
+      "grad_norm": 0.06419531237883218,
+      "learning_rate": 9.987110633727177e-06,
+      "loss": 0.0002,
+      "step": 1571
+    },
+    {
+      "epoch": 0.10130824257266224,
+      "grad_norm": 0.015524767153565688,
+      "learning_rate": 9.98639455782313e-06,
+      "loss": 0.0001,
+      "step": 1572
+    },
+    {
+      "epoch": 0.10137268801959141,
+      "grad_norm": 0.02811805669194153,
+      "learning_rate": 9.985678481919085e-06,
+      "loss": 0.0002,
+      "step": 1573
+    },
+    {
+      "epoch": 0.10143713346652058,
+      "grad_norm": 0.06079880538116993,
+      "learning_rate": 9.984962406015038e-06,
+      "loss": 0.0016,
+      "step": 1574
+    },
+    {
+      "epoch": 0.10150157891344977,
+      "grad_norm": 0.021396310851056387,
+      "learning_rate": 9.984246330110992e-06,
+      "loss": 0.0001,
+      "step": 1575
+    },
+    {
+      "epoch": 0.10156602436037894,
+      "grad_norm": 0.00487446275950087,
+      "learning_rate": 9.983530254206946e-06,
+      "loss": 0.0,
+      "step": 1576
+    },
+    {
+      "epoch": 0.10163046980730811,
+      "grad_norm": 0.001861853412206755,
+      "learning_rate": 9.9828141783029e-06,
+      "loss": 0.0,
+      "step": 1577
+    },
+    {
+      "epoch": 0.1016949152542373,
+      "grad_norm": 0.13226993615728455,
+      "learning_rate": 9.982098102398855e-06,
+      "loss": 0.0002,
+      "step": 1578
+    },
+    {
+      "epoch": 0.10175936070116647,
+      "grad_norm": 0.6347721275583404,
+      "learning_rate": 9.98138202649481e-06,
+      "loss": 0.0029,
+      "step": 1579
+    },
+    {
+      "epoch": 0.10182380614809564,
+      "grad_norm": 0.12557575131126444,
+      "learning_rate": 9.980665950590763e-06,
+      "loss": 0.0003,
+      "step": 1580
+    },
+    {
+      "epoch": 0.1018882515950248,
+      "grad_norm": 0.15937583048922008,
+      "learning_rate": 9.979949874686718e-06,
+      "loss": 0.0014,
+      "step": 1581
+    },
+    {
+      "epoch": 0.10195269704195399,
+      "grad_norm": 0.1975559703975177,
+      "learning_rate": 9.979233798782672e-06,
+      "loss": 0.0008,
+      "step": 1582
+    },
+    {
+      "epoch": 0.10201714248888316,
+      "grad_norm": 0.18503193927540512,
+      "learning_rate": 9.978517722878626e-06,
+      "loss": 0.0001,
+      "step": 1583
+    },
+    {
+      "epoch": 0.10208158793581233,
+      "grad_norm": 0.018288625469601905,
+      "learning_rate": 9.977801646974579e-06,
+      "loss": 0.0,
+      "step": 1584
+    },
+    {
+      "epoch": 0.10214603338274152,
+      "grad_norm": 0.06556653898587625,
+      "learning_rate": 9.977085571070533e-06,
+      "loss": 0.0016,
+      "step": 1585
+    },
+    {
+      "epoch": 0.10221047882967069,
+      "grad_norm": 0.12746553448387646,
+      "learning_rate": 9.976369495166488e-06,
+      "loss": 0.0,
+      "step": 1586
+    },
+    {
+      "epoch": 0.10227492427659986,
+      "grad_norm": 1.126843214046914,
+      "learning_rate": 9.975653419262442e-06,
+      "loss": 0.0058,
+      "step": 1587
+    },
+    {
+      "epoch": 0.10233936972352903,
+      "grad_norm": 0.0009175100005660954,
+      "learning_rate": 9.974937343358396e-06,
+      "loss": 0.0,
+      "step": 1588
+    },
+    {
+      "epoch": 0.10240381517045821,
+      "grad_norm": 0.025905761639659113,
+      "learning_rate": 9.974221267454352e-06,
+      "loss": 0.0,
+      "step": 1589
+    },
+    {
+      "epoch": 0.10246826061738738,
+      "grad_norm": 0.001112717209841313,
+      "learning_rate": 9.973505191550305e-06,
+      "loss": 0.0,
+      "step": 1590
+    },
+    {
+      "epoch": 0.10253270606431655,
+      "grad_norm": 0.007526676600103704,
+      "learning_rate": 9.972789115646259e-06,
+      "loss": 0.0,
+      "step": 1591
+    },
+    {
+      "epoch": 0.10259715151124572,
+      "grad_norm": 0.07737540406489828,
+      "learning_rate": 9.972073039742213e-06,
+      "loss": 0.0001,
+      "step": 1592
+    },
+    {
+      "epoch": 0.10266159695817491,
+      "grad_norm": 0.15650133541903946,
+      "learning_rate": 9.971356963838168e-06,
+      "loss": 0.0002,
+      "step": 1593
+    },
+    {
+      "epoch": 0.10272604240510408,
+      "grad_norm": 0.25322032122670035,
+      "learning_rate": 9.970640887934122e-06,
+      "loss": 0.0025,
+      "step": 1594
+    },
+    {
+      "epoch": 0.10279048785203325,
+      "grad_norm": 0.3750155962243933,
+      "learning_rate": 9.969924812030076e-06,
+      "loss": 0.0039,
+      "step": 1595
+    },
+    {
+      "epoch": 0.10285493329896243,
+      "grad_norm": 0.01156925484668394,
+      "learning_rate": 9.96920873612603e-06,
+      "loss": 0.0015,
+      "step": 1596
+    },
+    {
+      "epoch": 0.1029193787458916,
+      "grad_norm": 0.0010545124652149038,
+      "learning_rate": 9.968492660221985e-06,
+      "loss": 0.0,
+      "step": 1597
+    },
+    {
+      "epoch": 0.10298382419282077,
+      "grad_norm": 0.029305482633034747,
+      "learning_rate": 9.96777658431794e-06,
+      "loss": 0.0,
+      "step": 1598
+    },
+    {
+      "epoch": 0.10304826963974995,
+      "grad_norm": 0.0017090864848175315,
+      "learning_rate": 9.967060508413893e-06,
+      "loss": 0.0,
+      "step": 1599
+    },
+    {
+      "epoch": 0.10311271508667913,
+      "grad_norm": 0.05317997095655382,
+      "learning_rate": 9.966344432509846e-06,
+      "loss": 0.0001,
+      "step": 1600
+    },
+    {
+      "epoch": 0.1031771605336083,
+      "grad_norm": 0.22688678027798553,
+      "learning_rate": 9.9656283566058e-06,
+      "loss": 0.0026,
+      "step": 1601
+    },
+    {
+      "epoch": 0.10324160598053747,
+      "grad_norm": 0.30919653532738356,
+      "learning_rate": 9.964912280701755e-06,
+      "loss": 0.0004,
+      "step": 1602
+    },
+    {
+      "epoch": 0.10330605142746666,
+      "grad_norm": 0.014351292654744227,
+      "learning_rate": 9.964196204797709e-06,
+      "loss": 0.0,
+      "step": 1603
+    },
+    {
+      "epoch": 0.10337049687439583,
+      "grad_norm": 0.010907906211386922,
+      "learning_rate": 9.963480128893663e-06,
+      "loss": 0.0,
+      "step": 1604
+    },
+    {
+      "epoch": 0.103434942321325,
+      "grad_norm": 0.002965178042445306,
+      "learning_rate": 9.962764052989618e-06,
+      "loss": 0.0,
+      "step": 1605
+    },
+    {
+      "epoch": 0.10349938776825417,
+      "grad_norm": 0.003496357209876648,
+      "learning_rate": 9.962047977085572e-06,
+      "loss": 0.0,
+      "step": 1606
+    },
+    {
+      "epoch": 0.10356383321518335,
+      "grad_norm": 0.13440704324049624,
+      "learning_rate": 9.961331901181526e-06,
+      "loss": 0.0008,
+      "step": 1607
+    },
+    {
+      "epoch": 0.10362827866211252,
+      "grad_norm": 0.013390774687813115,
+      "learning_rate": 9.96061582527748e-06,
+      "loss": 0.0016,
+      "step": 1608
+    },
+    {
+      "epoch": 0.10369272410904169,
+      "grad_norm": 0.0007995069776809722,
+      "learning_rate": 9.959899749373435e-06,
+      "loss": 0.0,
+      "step": 1609
+    },
+    {
+      "epoch": 0.10375716955597088,
+      "grad_norm": 0.0002824253997624104,
+      "learning_rate": 9.959183673469387e-06,
+      "loss": 0.0,
+      "step": 1610
+    },
+    {
+      "epoch": 0.10382161500290005,
+      "grad_norm": 0.019146714592035718,
+      "learning_rate": 9.958467597565342e-06,
+      "loss": 0.0,
+      "step": 1611
+    },
+    {
+      "epoch": 0.10388606044982922,
+      "grad_norm": 0.3344483081280959,
+      "learning_rate": 9.957751521661298e-06,
+      "loss": 0.0014,
+      "step": 1612
+    },
+    {
+      "epoch": 0.10395050589675839,
+      "grad_norm": 0.007686443658249208,
+      "learning_rate": 9.957035445757252e-06,
+      "loss": 0.0016,
+      "step": 1613
+    },
+    {
+      "epoch": 0.10401495134368757,
+      "grad_norm": 0.004736729924681449,
+      "learning_rate": 9.956319369853206e-06,
+      "loss": 0.0001,
+      "step": 1614
+    },
+    {
+      "epoch": 0.10407939679061674,
+      "grad_norm": 0.2593175709583272,
+      "learning_rate": 9.95560329394916e-06,
+      "loss": 0.0045,
+      "step": 1615
+    },
+    {
+      "epoch": 0.10414384223754591,
+      "grad_norm": 1.196787952016555,
+      "learning_rate": 9.954887218045113e-06,
+      "loss": 0.0024,
+      "step": 1616
+    },
+    {
+      "epoch": 0.1042082876844751,
+      "grad_norm": 0.00788326272426621,
+      "learning_rate": 9.954171142141067e-06,
+      "loss": 0.0001,
+      "step": 1617
+    },
+    {
+      "epoch": 0.10427273313140427,
+      "grad_norm": 0.004238651176470014,
+      "learning_rate": 9.953455066237022e-06,
+      "loss": 0.0,
+      "step": 1618
+    },
+    {
+      "epoch": 0.10433717857833344,
+      "grad_norm": 0.006960126069193266,
+      "learning_rate": 9.952738990332976e-06,
+      "loss": 0.0001,
+      "step": 1619
+    },
+    {
+      "epoch": 0.10440162402526261,
+      "grad_norm": 0.3046891750387463,
+      "learning_rate": 9.95202291442893e-06,
+      "loss": 0.0005,
+      "step": 1620
+    },
+    {
+      "epoch": 0.1044660694721918,
+      "grad_norm": 0.19090559741277238,
+      "learning_rate": 9.951306838524885e-06,
+      "loss": 0.0003,
+      "step": 1621
+    },
+    {
+      "epoch": 0.10453051491912096,
+      "grad_norm": 0.12862948849226613,
+      "learning_rate": 9.950590762620839e-06,
+      "loss": 0.0009,
+      "step": 1622
+    },
+    {
+      "epoch": 0.10459496036605014,
+      "grad_norm": 0.000399543694424654,
+      "learning_rate": 9.949874686716793e-06,
+      "loss": 0.0,
+      "step": 1623
+    },
+    {
+      "epoch": 0.10465940581297932,
+      "grad_norm": 0.02006041245888554,
+      "learning_rate": 9.949158610812748e-06,
+      "loss": 0.0001,
+      "step": 1624
+    },
+    {
+      "epoch": 0.10472385125990849,
+      "grad_norm": 0.031908140114484264,
+      "learning_rate": 9.948442534908702e-06,
+      "loss": 0.0001,
+      "step": 1625
+    },
+    {
+      "epoch": 0.10478829670683766,
+      "grad_norm": 0.0009540502284184393,
+      "learning_rate": 9.947726459004654e-06,
+      "loss": 0.0,
+      "step": 1626
+    },
+    {
+      "epoch": 0.10485274215376683,
+      "grad_norm": 0.0008386468327602505,
+      "learning_rate": 9.947010383100609e-06,
+      "loss": 0.0,
+      "step": 1627
+    },
+    {
+      "epoch": 0.10491718760069602,
+      "grad_norm": 0.00263780282117747,
+      "learning_rate": 9.946294307196563e-06,
+      "loss": 0.0,
+      "step": 1628
+    },
+    {
+      "epoch": 0.10498163304762519,
+      "grad_norm": 0.006001984598497469,
+      "learning_rate": 9.945578231292517e-06,
+      "loss": 0.0001,
+      "step": 1629
+    },
+    {
+      "epoch": 0.10504607849455436,
+      "grad_norm": 0.00013517572587163083,
+      "learning_rate": 9.944862155388472e-06,
+      "loss": 0.0,
+      "step": 1630
+    },
+    {
+      "epoch": 0.10511052394148353,
+      "grad_norm": 0.03848265845278273,
+      "learning_rate": 9.944146079484426e-06,
+      "loss": 0.0001,
+      "step": 1631
+    },
+    {
+      "epoch": 0.10517496938841271,
+      "grad_norm": 0.022167288196300997,
+      "learning_rate": 9.94343000358038e-06,
+      "loss": 0.0001,
+      "step": 1632
+    },
+    {
+      "epoch": 0.10523941483534188,
+      "grad_norm": 0.002821403357723632,
+      "learning_rate": 9.942713927676335e-06,
+      "loss": 0.0,
+      "step": 1633
+    },
+    {
+      "epoch": 0.10530386028227105,
+      "grad_norm": 1.4247583351674593,
+      "learning_rate": 9.941997851772289e-06,
+      "loss": 0.0085,
+      "step": 1634
+    },
+    {
+      "epoch": 0.10536830572920024,
+      "grad_norm": 0.21941276183174943,
+      "learning_rate": 9.941281775868243e-06,
+      "loss": 0.0018,
+      "step": 1635
+    },
+    {
+      "epoch": 0.10543275117612941,
+      "grad_norm": 0.12458575422455666,
+      "learning_rate": 9.940565699964197e-06,
+      "loss": 0.0025,
+      "step": 1636
+    },
+    {
+      "epoch": 0.10549719662305858,
+      "grad_norm": 0.0027435063651743657,
+      "learning_rate": 9.939849624060152e-06,
+      "loss": 0.0,
+      "step": 1637
+    },
+    {
+      "epoch": 0.10556164206998775,
+      "grad_norm": 0.6987531166484844,
+      "learning_rate": 9.939133548156106e-06,
+      "loss": 0.0022,
+      "step": 1638
+    },
+    {
+      "epoch": 0.10562608751691693,
+      "grad_norm": 0.02080380749067857,
+      "learning_rate": 9.93841747225206e-06,
+      "loss": 0.0001,
+      "step": 1639
+    },
+    {
+      "epoch": 0.1056905329638461,
+      "grad_norm": 0.04035038836258241,
+      "learning_rate": 9.937701396348015e-06,
+      "loss": 0.0001,
+      "step": 1640
+    },
+    {
+      "epoch": 0.10575497841077527,
+      "grad_norm": 0.003756669122280707,
+      "learning_rate": 9.936985320443969e-06,
+      "loss": 0.0,
+      "step": 1641
+    },
+    {
+      "epoch": 0.10581942385770446,
+      "grad_norm": 0.009743641159397206,
+      "learning_rate": 9.936269244539922e-06,
+      "loss": 0.0001,
+      "step": 1642
+    },
+    {
+      "epoch": 0.10588386930463363,
+      "grad_norm": 0.06777027930780727,
+      "learning_rate": 9.935553168635876e-06,
+      "loss": 0.0004,
+      "step": 1643
+    },
+    {
+      "epoch": 0.1059483147515628,
+      "grad_norm": 0.005226804001337679,
+      "learning_rate": 9.93483709273183e-06,
+      "loss": 0.0,
+      "step": 1644
+    },
+    {
+      "epoch": 0.10601276019849197,
+      "grad_norm": 0.015324877669177634,
+      "learning_rate": 9.934121016827784e-06,
+      "loss": 0.0001,
+      "step": 1645
+    },
+    {
+      "epoch": 0.10607720564542115,
+      "grad_norm": 0.0344477850831631,
+      "learning_rate": 9.933404940923739e-06,
+      "loss": 0.0003,
+      "step": 1646
+    },
+    {
+      "epoch": 0.10614165109235033,
+      "grad_norm": 0.07882127139899352,
+      "learning_rate": 9.932688865019693e-06,
+      "loss": 0.0001,
+      "step": 1647
+    },
+    {
+      "epoch": 0.1062060965392795,
+      "grad_norm": 0.053166576465501644,
+      "learning_rate": 9.931972789115647e-06,
+      "loss": 0.0004,
+      "step": 1648
+    },
+    {
+      "epoch": 0.10627054198620868,
+      "grad_norm": 0.0001717063849772211,
+      "learning_rate": 9.931256713211602e-06,
+      "loss": 0.0,
+      "step": 1649
+    },
+    {
+      "epoch": 0.10633498743313785,
+      "grad_norm": 0.009019181123145955,
+      "learning_rate": 9.930540637307556e-06,
+      "loss": 0.0002,
+      "step": 1650
+    },
+    {
+      "epoch": 0.10639943288006702,
+      "grad_norm": 0.04910897784280347,
+      "learning_rate": 9.929824561403509e-06,
+      "loss": 0.0002,
+      "step": 1651
+    },
+    {
+      "epoch": 0.10646387832699619,
+      "grad_norm": 0.011014606916717328,
+      "learning_rate": 9.929108485499463e-06,
+      "loss": 0.0001,
+      "step": 1652
+    },
+    {
+      "epoch": 0.10652832377392538,
+      "grad_norm": 0.14360992510726237,
+      "learning_rate": 9.928392409595417e-06,
+      "loss": 0.0011,
+      "step": 1653
+    },
+    {
+      "epoch": 0.10659276922085455,
+      "grad_norm": 0.49932092507531767,
+      "learning_rate": 9.927676333691371e-06,
+      "loss": 0.0015,
+      "step": 1654
+    },
+    {
+      "epoch": 0.10665721466778372,
+      "grad_norm": 0.006795758715444581,
+      "learning_rate": 9.926960257787326e-06,
+      "loss": 0.0001,
+      "step": 1655
+    },
+    {
+      "epoch": 0.1067216601147129,
+      "grad_norm": 0.001934329060949019,
+      "learning_rate": 9.92624418188328e-06,
+      "loss": 0.0,
+      "step": 1656
+    },
+    {
+      "epoch": 0.10678610556164207,
+      "grad_norm": 0.09470259963083237,
+      "learning_rate": 9.925528105979234e-06,
+      "loss": 0.0003,
+      "step": 1657
+    },
+    {
+      "epoch": 0.10685055100857124,
+      "grad_norm": 0.02276080609148665,
+      "learning_rate": 9.924812030075189e-06,
+      "loss": 0.0002,
+      "step": 1658
+    },
+    {
+      "epoch": 0.10691499645550041,
+      "grad_norm": 0.004617006557617145,
+      "learning_rate": 9.924095954171143e-06,
+      "loss": 0.0,
+      "step": 1659
+    },
+    {
+      "epoch": 0.1069794419024296,
+      "grad_norm": 0.008264148745958573,
+      "learning_rate": 9.923379878267097e-06,
+      "loss": 0.0,
+      "step": 1660
+    },
+    {
+      "epoch": 0.10704388734935877,
+      "grad_norm": 0.42388602709453416,
+      "learning_rate": 9.922663802363052e-06,
+      "loss": 0.003,
+      "step": 1661
+    },
+    {
+      "epoch": 0.10710833279628794,
+      "grad_norm": 0.022281854083548366,
+      "learning_rate": 9.921947726459006e-06,
+      "loss": 0.0003,
+      "step": 1662
+    },
+    {
+      "epoch": 0.10717277824321712,
+      "grad_norm": 0.024515886161469536,
+      "learning_rate": 9.92123165055496e-06,
+      "loss": 0.0001,
+      "step": 1663
+    },
+    {
+      "epoch": 0.1072372236901463,
+      "grad_norm": 0.004597568827487063,
+      "learning_rate": 9.920515574650914e-06,
+      "loss": 0.0001,
+      "step": 1664
+    },
+    {
+      "epoch": 0.10730166913707546,
+      "grad_norm": 0.020596066669992474,
+      "learning_rate": 9.919799498746869e-06,
+      "loss": 0.0001,
+      "step": 1665
+    },
+    {
+      "epoch": 0.10736611458400463,
+      "grad_norm": 0.17759965674602354,
+      "learning_rate": 9.919083422842823e-06,
+      "loss": 0.0013,
+      "step": 1666
+    },
+    {
+      "epoch": 0.10743056003093382,
+      "grad_norm": 0.018403341874534827,
+      "learning_rate": 9.918367346938776e-06,
+      "loss": 0.0002,
+      "step": 1667
+    },
+    {
+      "epoch": 0.10749500547786299,
+      "grad_norm": 2.1243720529282752,
+      "learning_rate": 9.91765127103473e-06,
+      "loss": 0.011,
+      "step": 1668
+    },
+    {
+      "epoch": 0.10755945092479216,
+      "grad_norm": 0.009084960426039212,
+      "learning_rate": 9.916935195130684e-06,
+      "loss": 0.0001,
+      "step": 1669
+    },
+    {
+      "epoch": 0.10762389637172134,
+      "grad_norm": 0.0387725830078125,
+      "learning_rate": 9.916219119226639e-06,
+      "loss": 0.0003,
+      "step": 1670
+    },
+    {
+      "epoch": 0.10768834181865052,
+      "grad_norm": 0.22734220117519863,
+      "learning_rate": 9.915503043322593e-06,
+      "loss": 0.0006,
+      "step": 1671
+    },
+    {
+      "epoch": 0.10775278726557969,
+      "grad_norm": 0.839883847721016,
+      "learning_rate": 9.914786967418547e-06,
+      "loss": 0.0105,
+      "step": 1672
+    },
+    {
+      "epoch": 0.10781723271250886,
+      "grad_norm": 0.0015275560043602424,
+      "learning_rate": 9.914070891514501e-06,
+      "loss": 0.0,
+      "step": 1673
+    },
+    {
+      "epoch": 0.10788167815943804,
+      "grad_norm": 0.2546175456952681,
+      "learning_rate": 9.913354815610456e-06,
+      "loss": 0.0021,
+      "step": 1674
+    },
+    {
+      "epoch": 0.10794612360636721,
+      "grad_norm": 0.27017822568966937,
+      "learning_rate": 9.91263873970641e-06,
+      "loss": 0.0007,
+      "step": 1675
+    },
+    {
+      "epoch": 0.10801056905329638,
+      "grad_norm": 0.020131666449685264,
+      "learning_rate": 9.911922663802364e-06,
+      "loss": 0.0001,
+      "step": 1676
+    },
+    {
+      "epoch": 0.10807501450022555,
+      "grad_norm": 0.007008979154739355,
+      "learning_rate": 9.911206587898317e-06,
+      "loss": 0.0,
+      "step": 1677
+    },
+    {
+      "epoch": 0.10813945994715474,
+      "grad_norm": 0.31403968120823095,
+      "learning_rate": 9.910490511994271e-06,
+      "loss": 0.0008,
+      "step": 1678
+    },
+    {
+      "epoch": 0.10820390539408391,
+      "grad_norm": 0.08434008796606712,
+      "learning_rate": 9.909774436090226e-06,
+      "loss": 0.0001,
+      "step": 1679
+    },
+    {
+      "epoch": 0.10826835084101308,
+      "grad_norm": 0.21085719946650242,
+      "learning_rate": 9.90905836018618e-06,
+      "loss": 0.0007,
+      "step": 1680
+    },
+    {
+      "epoch": 0.10833279628794226,
+      "grad_norm": 0.06126331219652856,
+      "learning_rate": 9.908342284282134e-06,
+      "loss": 0.0001,
+      "step": 1681
+    },
+    {
+      "epoch": 0.10839724173487143,
+      "grad_norm": 0.02127553894163218,
+      "learning_rate": 9.90762620837809e-06,
+      "loss": 0.0,
+      "step": 1682
+    },
+    {
+      "epoch": 0.1084616871818006,
+      "grad_norm": 0.0020103520188776516,
+      "learning_rate": 9.906910132474043e-06,
+      "loss": 0.0,
+      "step": 1683
+    },
+    {
+      "epoch": 0.10852613262872977,
+      "grad_norm": 0.1730377089352632,
+      "learning_rate": 9.906194056569997e-06,
+      "loss": 0.0032,
+      "step": 1684
+    },
+    {
+      "epoch": 0.10859057807565896,
+      "grad_norm": 0.40038176736105546,
+      "learning_rate": 9.905477980665951e-06,
+      "loss": 0.0048,
+      "step": 1685
+    },
+    {
+      "epoch": 0.10865502352258813,
+      "grad_norm": 0.0349529717470364,
+      "learning_rate": 9.904761904761906e-06,
+      "loss": 0.0001,
+      "step": 1686
+    },
+    {
+      "epoch": 0.1087194689695173,
+      "grad_norm": 0.12564037620330987,
+      "learning_rate": 9.90404582885786e-06,
+      "loss": 0.0002,
+      "step": 1687
+    },
+    {
+      "epoch": 0.10878391441644648,
+      "grad_norm": 0.02573585559729339,
+      "learning_rate": 9.903329752953814e-06,
+      "loss": 0.0001,
+      "step": 1688
+    },
+    {
+      "epoch": 0.10884835986337565,
+      "grad_norm": 0.0182613506637033,
+      "learning_rate": 9.902613677049769e-06,
+      "loss": 0.0001,
+      "step": 1689
+    },
+    {
+      "epoch": 0.10891280531030482,
+      "grad_norm": 0.0033500290164589796,
+      "learning_rate": 9.901897601145723e-06,
+      "loss": 0.0,
+      "step": 1690
+    },
+    {
+      "epoch": 0.108977250757234,
+      "grad_norm": 0.22283454081114984,
+      "learning_rate": 9.901181525241677e-06,
+      "loss": 0.0009,
+      "step": 1691
+    },
+    {
+      "epoch": 0.10904169620416318,
+      "grad_norm": 0.0013649688327389462,
+      "learning_rate": 9.900465449337631e-06,
+      "loss": 0.0,
+      "step": 1692
+    },
+    {
+      "epoch": 0.10910614165109235,
+      "grad_norm": 0.0007646449219787592,
+      "learning_rate": 9.899749373433584e-06,
+      "loss": 0.0,
+      "step": 1693
+    },
+    {
+      "epoch": 0.10917058709802152,
+      "grad_norm": 0.009881301935125679,
+      "learning_rate": 9.899033297529538e-06,
+      "loss": 0.0001,
+      "step": 1694
+    },
+    {
+      "epoch": 0.1092350325449507,
+      "grad_norm": 0.22115118406281953,
+      "learning_rate": 9.898317221625493e-06,
+      "loss": 0.001,
+      "step": 1695
+    },
+    {
+      "epoch": 0.10929947799187988,
+      "grad_norm": 0.015784157099274273,
+      "learning_rate": 9.897601145721447e-06,
+      "loss": 0.0001,
+      "step": 1696
+    },
+    {
+      "epoch": 0.10936392343880905,
+      "grad_norm": 0.04523709863820392,
+      "learning_rate": 9.896885069817401e-06,
+      "loss": 0.0001,
+      "step": 1697
+    },
+    {
+      "epoch": 0.10942836888573822,
+      "grad_norm": 0.028951350795599847,
+      "learning_rate": 9.896168993913355e-06,
+      "loss": 0.0,
+      "step": 1698
+    },
+    {
+      "epoch": 0.1094928143326674,
+      "grad_norm": 0.00021010130849105123,
+      "learning_rate": 9.89545291800931e-06,
+      "loss": 0.0,
+      "step": 1699
+    },
+    {
+      "epoch": 0.10955725977959657,
+      "grad_norm": 0.01118543396035685,
+      "learning_rate": 9.894736842105264e-06,
+      "loss": 0.0001,
+      "step": 1700
+    },
+    {
+      "epoch": 0.10962170522652574,
+      "grad_norm": 0.00016900792184204182,
+      "learning_rate": 9.894020766201218e-06,
+      "loss": 0.0,
+      "step": 1701
+    },
+    {
+      "epoch": 0.10968615067345493,
+      "grad_norm": 0.00286496145638672,
+      "learning_rate": 9.893304690297173e-06,
+      "loss": 0.0,
+      "step": 1702
+    },
+    {
+      "epoch": 0.1097505961203841,
+      "grad_norm": 0.00041776218172744057,
+      "learning_rate": 9.892588614393125e-06,
+      "loss": 0.0,
+      "step": 1703
+    },
+    {
+      "epoch": 0.10981504156731327,
+      "grad_norm": 0.008368331222943564,
+      "learning_rate": 9.89187253848908e-06,
+      "loss": 0.0001,
+      "step": 1704
+    },
+    {
+      "epoch": 0.10987948701424244,
+      "grad_norm": 0.041839912630408586,
+      "learning_rate": 9.891156462585036e-06,
+      "loss": 0.0002,
+      "step": 1705
+    },
+    {
+      "epoch": 0.10994393246117162,
+      "grad_norm": 0.024497527264033654,
+      "learning_rate": 9.89044038668099e-06,
+      "loss": 0.0001,
+      "step": 1706
+    },
+    {
+      "epoch": 0.1100083779081008,
+      "grad_norm": 0.023590454305983673,
+      "learning_rate": 9.889724310776944e-06,
+      "loss": 0.0001,
+      "step": 1707
+    },
+    {
+      "epoch": 0.11007282335502996,
+      "grad_norm": 0.012773182132556847,
+      "learning_rate": 9.889008234872898e-06,
+      "loss": 0.0,
+      "step": 1708
+    },
+    {
+      "epoch": 0.11013726880195915,
+      "grad_norm": 0.004021249149665355,
+      "learning_rate": 9.888292158968851e-06,
+      "loss": 0.0,
+      "step": 1709
+    },
+    {
+      "epoch": 0.11020171424888832,
+      "grad_norm": 0.02710948476336297,
+      "learning_rate": 9.887576083064805e-06,
+      "loss": 0.0003,
+      "step": 1710
+    },
+    {
+      "epoch": 0.11026615969581749,
+      "grad_norm": 0.006102303146094259,
+      "learning_rate": 9.88686000716076e-06,
+      "loss": 0.0,
+      "step": 1711
+    },
+    {
+      "epoch": 0.11033060514274666,
+      "grad_norm": 0.07333378468360141,
+      "learning_rate": 9.886143931256714e-06,
+      "loss": 0.0004,
+      "step": 1712
+    },
+    {
+      "epoch": 0.11039505058967584,
+      "grad_norm": 0.03245848052794509,
+      "learning_rate": 9.885427855352668e-06,
+      "loss": 0.0,
+      "step": 1713
+    },
+    {
+      "epoch": 0.11045949603660501,
+      "grad_norm": 0.06893218377444407,
+      "learning_rate": 9.884711779448623e-06,
+      "loss": 0.0001,
+      "step": 1714
+    },
+    {
+      "epoch": 0.11052394148353419,
+      "grad_norm": 0.0598030877506716,
+      "learning_rate": 9.883995703544577e-06,
+      "loss": 0.0002,
+      "step": 1715
+    },
+    {
+      "epoch": 0.11058838693046336,
+      "grad_norm": 0.0013980899631203523,
+      "learning_rate": 9.883279627640531e-06,
+      "loss": 0.0,
+      "step": 1716
+    },
+    {
+      "epoch": 0.11065283237739254,
+      "grad_norm": 0.3934271253107235,
+      "learning_rate": 9.882563551736485e-06,
+      "loss": 0.0009,
+      "step": 1717
+    },
+    {
+      "epoch": 0.11071727782432171,
+      "grad_norm": 0.13356755708162912,
+      "learning_rate": 9.88184747583244e-06,
+      "loss": 0.0004,
+      "step": 1718
+    },
+    {
+      "epoch": 0.11078172327125088,
+      "grad_norm": 0.009251442707350574,
+      "learning_rate": 9.881131399928392e-06,
+      "loss": 0.0001,
+      "step": 1719
+    },
+    {
+      "epoch": 0.11084616871818007,
+      "grad_norm": 0.09078265506655032,
+      "learning_rate": 9.880415324024347e-06,
+      "loss": 0.0002,
+      "step": 1720
+    },
+    {
+      "epoch": 0.11091061416510924,
+      "grad_norm": 0.45446654387990904,
+      "learning_rate": 9.879699248120301e-06,
+      "loss": 0.0018,
+      "step": 1721
+    },
+    {
+      "epoch": 0.11097505961203841,
+      "grad_norm": 0.03796955507606215,
+      "learning_rate": 9.878983172216255e-06,
+      "loss": 0.0002,
+      "step": 1722
+    },
+    {
+      "epoch": 0.11103950505896758,
+      "grad_norm": 0.33390644346596626,
+      "learning_rate": 9.87826709631221e-06,
+      "loss": 0.0004,
+      "step": 1723
+    },
+    {
+      "epoch": 0.11110395050589676,
+      "grad_norm": 0.0017985687644987878,
+      "learning_rate": 9.877551020408164e-06,
+      "loss": 0.0,
+      "step": 1724
+    },
+    {
+      "epoch": 0.11116839595282593,
+      "grad_norm": 0.00012855284184480202,
+      "learning_rate": 9.876834944504118e-06,
+      "loss": 0.0,
+      "step": 1725
+    },
+    {
+      "epoch": 0.1112328413997551,
+      "grad_norm": 0.002766077239669994,
+      "learning_rate": 9.876118868600072e-06,
+      "loss": 0.0,
+      "step": 1726
+    },
+    {
+      "epoch": 0.11129728684668429,
+      "grad_norm": 0.007371952801886638,
+      "learning_rate": 9.875402792696027e-06,
+      "loss": 0.0,
+      "step": 1727
+    },
+    {
+      "epoch": 0.11136173229361346,
+      "grad_norm": 0.004698404644119558,
+      "learning_rate": 9.87468671679198e-06,
+      "loss": 0.0,
+      "step": 1728
+    },
+    {
+      "epoch": 0.11142617774054263,
+      "grad_norm": 0.003165305540191389,
+      "learning_rate": 9.873970640887935e-06,
+      "loss": 0.0,
+      "step": 1729
+    },
+    {
+      "epoch": 0.1114906231874718,
+      "grad_norm": 0.002938261901882307,
+      "learning_rate": 9.87325456498389e-06,
+      "loss": 0.0,
+      "step": 1730
+    },
+    {
+      "epoch": 0.11155506863440098,
+      "grad_norm": 0.004821551934840438,
+      "learning_rate": 9.872538489079844e-06,
+      "loss": 0.0001,
+      "step": 1731
+    },
+    {
+      "epoch": 0.11161951408133015,
+      "grad_norm": 4.844431380060839,
+      "learning_rate": 9.871822413175798e-06,
+      "loss": 0.0238,
+      "step": 1732
+    },
+    {
+      "epoch": 0.11168395952825932,
+      "grad_norm": 0.023224661434122617,
+      "learning_rate": 9.871106337271753e-06,
+      "loss": 0.0001,
+      "step": 1733
+    },
+    {
+      "epoch": 0.11174840497518851,
+      "grad_norm": 0.31371399393588917,
+      "learning_rate": 9.870390261367707e-06,
+      "loss": 0.0055,
+      "step": 1734
+    },
+    {
+      "epoch": 0.11181285042211768,
+      "grad_norm": 0.06859223690514762,
+      "learning_rate": 9.86967418546366e-06,
+      "loss": 0.0004,
+      "step": 1735
+    },
+    {
+      "epoch": 0.11187729586904685,
+      "grad_norm": 0.00032449914922864464,
+      "learning_rate": 9.868958109559614e-06,
+      "loss": 0.0,
+      "step": 1736
+    },
+    {
+      "epoch": 0.11194174131597602,
+      "grad_norm": 0.000419428479045938,
+      "learning_rate": 9.868242033655568e-06,
+      "loss": 0.0,
+      "step": 1737
+    },
+    {
+      "epoch": 0.1120061867629052,
+      "grad_norm": 0.00641962832500581,
+      "learning_rate": 9.867525957751522e-06,
+      "loss": 0.0,
+      "step": 1738
+    },
+    {
+      "epoch": 0.11207063220983438,
+      "grad_norm": 0.007188855193993879,
+      "learning_rate": 9.866809881847477e-06,
+      "loss": 0.0,
+      "step": 1739
+    },
+    {
+      "epoch": 0.11213507765676355,
+      "grad_norm": 0.5678866540904066,
+      "learning_rate": 9.866093805943431e-06,
+      "loss": 0.0022,
+      "step": 1740
+    },
+    {
+      "epoch": 0.11219952310369273,
+      "grad_norm": 0.05270199532192997,
+      "learning_rate": 9.865377730039385e-06,
+      "loss": 0.0002,
+      "step": 1741
+    },
+    {
+      "epoch": 0.1122639685506219,
+      "grad_norm": 0.0034274067808784683,
+      "learning_rate": 9.86466165413534e-06,
+      "loss": 0.0,
+      "step": 1742
+    },
+    {
+      "epoch": 0.11232841399755107,
+      "grad_norm": 0.0012332182509001428,
+      "learning_rate": 9.863945578231294e-06,
+      "loss": 0.0,
+      "step": 1743
+    },
+    {
+      "epoch": 0.11239285944448024,
+      "grad_norm": 0.019587984236910893,
+      "learning_rate": 9.863229502327246e-06,
+      "loss": 0.0001,
+      "step": 1744
+    },
+    {
+      "epoch": 0.11245730489140943,
+      "grad_norm": 0.0001058282062590549,
+      "learning_rate": 9.8625134264232e-06,
+      "loss": 0.0,
+      "step": 1745
+    },
+    {
+      "epoch": 0.1125217503383386,
+      "grad_norm": 0.0019646040441125955,
+      "learning_rate": 9.861797350519155e-06,
+      "loss": 0.0,
+      "step": 1746
+    },
+    {
+      "epoch": 0.11258619578526777,
+      "grad_norm": 0.0024159480603714364,
+      "learning_rate": 9.86108127461511e-06,
+      "loss": 0.0,
+      "step": 1747
+    },
+    {
+      "epoch": 0.11265064123219695,
+      "grad_norm": 0.03507614114998864,
+      "learning_rate": 9.860365198711064e-06,
+      "loss": 0.0002,
+      "step": 1748
+    },
+    {
+      "epoch": 0.11271508667912612,
+      "grad_norm": 0.02787627501469119,
+      "learning_rate": 9.859649122807018e-06,
+      "loss": 0.0001,
+      "step": 1749
+    },
+    {
+      "epoch": 0.11277953212605529,
+      "grad_norm": 0.008317143690736094,
+      "learning_rate": 9.858933046902972e-06,
+      "loss": 0.0,
+      "step": 1750
+    },
+    {
+      "epoch": 0.11284397757298446,
+      "grad_norm": 0.8252554064657068,
+      "learning_rate": 9.858216970998927e-06,
+      "loss": 0.0044,
+      "step": 1751
+    },
+    {
+      "epoch": 0.11290842301991365,
+      "grad_norm": 1.160233260658864,
+      "learning_rate": 9.857500895094881e-06,
+      "loss": 0.0055,
+      "step": 1752
+    },
+    {
+      "epoch": 0.11297286846684282,
+      "grad_norm": 0.007301981019052537,
+      "learning_rate": 9.856784819190835e-06,
+      "loss": 0.0,
+      "step": 1753
+    },
+    {
+      "epoch": 0.11303731391377199,
+      "grad_norm": 0.0021348737129215627,
+      "learning_rate": 9.85606874328679e-06,
+      "loss": 0.0,
+      "step": 1754
+    },
+    {
+      "epoch": 0.11310175936070116,
+      "grad_norm": 0.1850555238494288,
+      "learning_rate": 9.855352667382744e-06,
+      "loss": 0.0002,
+      "step": 1755
+    },
+    {
+      "epoch": 0.11316620480763034,
+      "grad_norm": 0.003926293491443531,
+      "learning_rate": 9.854636591478698e-06,
+      "loss": 0.0,
+      "step": 1756
+    },
+    {
+      "epoch": 0.11323065025455951,
+      "grad_norm": 0.03075867687050704,
+      "learning_rate": 9.853920515574652e-06,
+      "loss": 0.0,
+      "step": 1757
+    },
+    {
+      "epoch": 0.11329509570148869,
+      "grad_norm": 0.0006125017611817344,
+      "learning_rate": 9.853204439670607e-06,
+      "loss": 0.0,
+      "step": 1758
+    },
+    {
+      "epoch": 0.11335954114841787,
+      "grad_norm": 0.0019315106515052796,
+      "learning_rate": 9.852488363766561e-06,
+      "loss": 0.0,
+      "step": 1759
+    },
+    {
+      "epoch": 0.11342398659534704,
+      "grad_norm": 0.14089677959341357,
+      "learning_rate": 9.851772287862514e-06,
+      "loss": 0.0002,
+      "step": 1760
+    },
+    {
+      "epoch": 0.11348843204227621,
+      "grad_norm": 0.0015930161051359712,
+      "learning_rate": 9.851056211958468e-06,
+      "loss": 0.0,
+      "step": 1761
+    },
+    {
+      "epoch": 0.11355287748920538,
+      "grad_norm": 0.2209100850050719,
+      "learning_rate": 9.850340136054422e-06,
+      "loss": 0.0021,
+      "step": 1762
+    },
+    {
+      "epoch": 0.11361732293613457,
+      "grad_norm": 0.11348926830889935,
+      "learning_rate": 9.849624060150376e-06,
+      "loss": 0.0004,
+      "step": 1763
+    },
+    {
+      "epoch": 0.11368176838306374,
+      "grad_norm": 0.05025630142028179,
+      "learning_rate": 9.84890798424633e-06,
+      "loss": 0.0002,
+      "step": 1764
+    },
+    {
+      "epoch": 0.1137462138299929,
+      "grad_norm": 0.004523011762179798,
+      "learning_rate": 9.848191908342285e-06,
+      "loss": 0.0,
+      "step": 1765
+    },
+    {
+      "epoch": 0.11381065927692209,
+      "grad_norm": 0.03315617945049665,
+      "learning_rate": 9.84747583243824e-06,
+      "loss": 0.0004,
+      "step": 1766
+    },
+    {
+      "epoch": 0.11387510472385126,
+      "grad_norm": 0.004434986096414539,
+      "learning_rate": 9.846759756534194e-06,
+      "loss": 0.0,
+      "step": 1767
+    },
+    {
+      "epoch": 0.11393955017078043,
+      "grad_norm": 0.23389844924721334,
+      "learning_rate": 9.846043680630148e-06,
+      "loss": 0.0008,
+      "step": 1768
+    },
+    {
+      "epoch": 0.1140039956177096,
+      "grad_norm": 0.12071385320497696,
+      "learning_rate": 9.845327604726102e-06,
+      "loss": 0.0004,
+      "step": 1769
+    },
+    {
+      "epoch": 0.11406844106463879,
+      "grad_norm": 0.027456958188735676,
+      "learning_rate": 9.844611528822055e-06,
+      "loss": 0.0,
+      "step": 1770
+    },
+    {
+      "epoch": 0.11413288651156796,
+      "grad_norm": 0.2016661137400553,
+      "learning_rate": 9.843895452918009e-06,
+      "loss": 0.0026,
+      "step": 1771
+    },
+    {
+      "epoch": 0.11419733195849713,
+      "grad_norm": 0.0006652434216453054,
+      "learning_rate": 9.843179377013963e-06,
+      "loss": 0.0,
+      "step": 1772
+    },
+    {
+      "epoch": 0.11426177740542631,
+      "grad_norm": 0.06822878762253942,
+      "learning_rate": 9.842463301109918e-06,
+      "loss": 0.001,
+      "step": 1773
+    },
+    {
+      "epoch": 0.11432622285235548,
+      "grad_norm": 0.19988441517336292,
+      "learning_rate": 9.841747225205872e-06,
+      "loss": 0.0003,
+      "step": 1774
+    },
+    {
+      "epoch": 0.11439066829928465,
+      "grad_norm": 0.05110567110947775,
+      "learning_rate": 9.841031149301828e-06,
+      "loss": 0.0001,
+      "step": 1775
+    },
+    {
+      "epoch": 0.11445511374621382,
+      "grad_norm": 0.007954576835518657,
+      "learning_rate": 9.84031507339778e-06,
+      "loss": 0.0,
+      "step": 1776
+    },
+    {
+      "epoch": 0.11451955919314301,
+      "grad_norm": 0.0010392938517551102,
+      "learning_rate": 9.839598997493735e-06,
+      "loss": 0.0,
+      "step": 1777
+    },
+    {
+      "epoch": 0.11458400464007218,
+      "grad_norm": 0.0056564445665283,
+      "learning_rate": 9.83888292158969e-06,
+      "loss": 0.0,
+      "step": 1778
+    },
+    {
+      "epoch": 0.11464845008700135,
+      "grad_norm": 0.0014838747156959813,
+      "learning_rate": 9.838166845685644e-06,
+      "loss": 0.0,
+      "step": 1779
+    },
+    {
+      "epoch": 0.11471289553393053,
+      "grad_norm": 0.0030324228042027165,
+      "learning_rate": 9.837450769781598e-06,
+      "loss": 0.0,
+      "step": 1780
+    },
+    {
+      "epoch": 0.1147773409808597,
+      "grad_norm": 0.007153180921713235,
+      "learning_rate": 9.836734693877552e-06,
+      "loss": 0.0,
+      "step": 1781
+    },
+    {
+      "epoch": 0.11484178642778888,
+      "grad_norm": 0.0014775236203942192,
+      "learning_rate": 9.836018617973506e-06,
+      "loss": 0.0,
+      "step": 1782
+    },
+    {
+      "epoch": 0.11490623187471805,
+      "grad_norm": 0.042548194151211846,
+      "learning_rate": 9.83530254206946e-06,
+      "loss": 0.0001,
+      "step": 1783
+    },
+    {
+      "epoch": 0.11497067732164723,
+      "grad_norm": 0.012568469437944967,
+      "learning_rate": 9.834586466165415e-06,
+      "loss": 0.0,
+      "step": 1784
+    },
+    {
+      "epoch": 0.1150351227685764,
+      "grad_norm": 0.03218187318560303,
+      "learning_rate": 9.83387039026137e-06,
+      "loss": 0.0,
+      "step": 1785
+    },
+    {
+      "epoch": 0.11509956821550557,
+      "grad_norm": 0.007692717030373828,
+      "learning_rate": 9.833154314357322e-06,
+      "loss": 0.0001,
+      "step": 1786
+    },
+    {
+      "epoch": 0.11516401366243476,
+      "grad_norm": 0.012189991502221508,
+      "learning_rate": 9.832438238453276e-06,
+      "loss": 0.0,
+      "step": 1787
+    },
+    {
+      "epoch": 0.11522845910936393,
+      "grad_norm": 0.04904071543282355,
+      "learning_rate": 9.83172216254923e-06,
+      "loss": 0.0022,
+      "step": 1788
+    },
+    {
+      "epoch": 0.1152929045562931,
+      "grad_norm": 0.7873672933629802,
+      "learning_rate": 9.831006086645185e-06,
+      "loss": 0.0021,
+      "step": 1789
+    },
+    {
+      "epoch": 0.11535735000322227,
+      "grad_norm": 0.03984673321766176,
+      "learning_rate": 9.830290010741139e-06,
+      "loss": 0.0002,
+      "step": 1790
+    },
+    {
+      "epoch": 0.11542179545015145,
+      "grad_norm": 0.04043361873909708,
+      "learning_rate": 9.829573934837093e-06,
+      "loss": 0.0001,
+      "step": 1791
+    },
+    {
+      "epoch": 0.11548624089708062,
+      "grad_norm": 0.00460612867531268,
+      "learning_rate": 9.828857858933048e-06,
+      "loss": 0.0,
+      "step": 1792
+    },
+    {
+      "epoch": 0.11555068634400979,
+      "grad_norm": 0.006081790106934461,
+      "learning_rate": 9.828141783029002e-06,
+      "loss": 0.0,
+      "step": 1793
+    },
+    {
+      "epoch": 0.11561513179093898,
+      "grad_norm": 0.06486523973034726,
+      "learning_rate": 9.827425707124956e-06,
+      "loss": 0.0001,
+      "step": 1794
+    },
+    {
+      "epoch": 0.11567957723786815,
+      "grad_norm": 3.2129624564538615,
+      "learning_rate": 9.82670963122091e-06,
+      "loss": 0.03,
+      "step": 1795
+    },
+    {
+      "epoch": 0.11574402268479732,
+      "grad_norm": 0.0007317990986965962,
+      "learning_rate": 9.825993555316863e-06,
+      "loss": 0.0,
+      "step": 1796
+    },
+    {
+      "epoch": 0.11580846813172649,
+      "grad_norm": 0.007586954091167557,
+      "learning_rate": 9.825277479412818e-06,
+      "loss": 0.0,
+      "step": 1797
+    },
+    {
+      "epoch": 0.11587291357865567,
+      "grad_norm": 0.004224269569500741,
+      "learning_rate": 9.824561403508772e-06,
+      "loss": 0.0,
+      "step": 1798
+    },
+    {
+      "epoch": 0.11593735902558484,
+      "grad_norm": 0.0027508155825239165,
+      "learning_rate": 9.823845327604728e-06,
+      "loss": 0.0,
+      "step": 1799
+    },
+    {
+      "epoch": 0.11600180447251401,
+      "grad_norm": 0.0005093112787406999,
+      "learning_rate": 9.823129251700682e-06,
+      "loss": 0.0,
+      "step": 1800
+    },
+    {
+      "epoch": 0.11606624991944318,
+      "grad_norm": 0.2830816863572351,
+      "learning_rate": 9.822413175796636e-06,
+      "loss": 0.0006,
+      "step": 1801
+    },
+    {
+      "epoch": 0.11613069536637237,
+      "grad_norm": 0.008831027053601843,
+      "learning_rate": 9.821697099892589e-06,
+      "loss": 0.0001,
+      "step": 1802
+    },
+    {
+      "epoch": 0.11619514081330154,
+      "grad_norm": 0.041514996841974086,
+      "learning_rate": 9.820981023988543e-06,
+      "loss": 0.0003,
+      "step": 1803
+    },
+    {
+      "epoch": 0.11625958626023071,
+      "grad_norm": 0.0021810375241418326,
+      "learning_rate": 9.820264948084498e-06,
+      "loss": 0.0,
+      "step": 1804
+    },
+    {
+      "epoch": 0.1163240317071599,
+      "grad_norm": 0.004574236003985102,
+      "learning_rate": 9.819548872180452e-06,
+      "loss": 0.0,
+      "step": 1805
+    },
+    {
+      "epoch": 0.11638847715408907,
+      "grad_norm": 0.001797843287860481,
+      "learning_rate": 9.818832796276406e-06,
+      "loss": 0.0,
+      "step": 1806
+    },
+    {
+      "epoch": 0.11645292260101824,
+      "grad_norm": 0.03992228954057036,
+      "learning_rate": 9.81811672037236e-06,
+      "loss": 0.0003,
+      "step": 1807
+    },
+    {
+      "epoch": 0.1165173680479474,
+      "grad_norm": 2.652800530496413,
+      "learning_rate": 9.817400644468315e-06,
+      "loss": 0.0095,
+      "step": 1808
+    },
+    {
+      "epoch": 0.11658181349487659,
+      "grad_norm": 0.005735454108690199,
+      "learning_rate": 9.816684568564269e-06,
+      "loss": 0.0,
+      "step": 1809
+    },
+    {
+      "epoch": 0.11664625894180576,
+      "grad_norm": 0.46595162514166677,
+      "learning_rate": 9.815968492660223e-06,
+      "loss": 0.0022,
+      "step": 1810
+    },
+    {
+      "epoch": 0.11671070438873493,
+      "grad_norm": 0.008367182582977108,
+      "learning_rate": 9.815252416756178e-06,
+      "loss": 0.0001,
+      "step": 1811
+    },
+    {
+      "epoch": 0.11677514983566412,
+      "grad_norm": 0.010296350222608969,
+      "learning_rate": 9.81453634085213e-06,
+      "loss": 0.0001,
+      "step": 1812
+    },
+    {
+      "epoch": 0.11683959528259329,
+      "grad_norm": 0.002394567203813334,
+      "learning_rate": 9.813820264948085e-06,
+      "loss": 0.0,
+      "step": 1813
+    },
+    {
+      "epoch": 0.11690404072952246,
+      "grad_norm": 0.013784040528215559,
+      "learning_rate": 9.813104189044039e-06,
+      "loss": 0.0,
+      "step": 1814
+    },
+    {
+      "epoch": 0.11696848617645163,
+      "grad_norm": 0.003288869045197393,
+      "learning_rate": 9.812388113139993e-06,
+      "loss": 0.0,
+      "step": 1815
+    },
+    {
+      "epoch": 0.11703293162338081,
+      "grad_norm": 0.003724212986621817,
+      "learning_rate": 9.811672037235947e-06,
+      "loss": 0.0,
+      "step": 1816
+    },
+    {
+      "epoch": 0.11709737707030998,
+      "grad_norm": 0.0013027106074673877,
+      "learning_rate": 9.810955961331902e-06,
+      "loss": 0.0,
+      "step": 1817
+    },
+    {
+      "epoch": 0.11716182251723915,
+      "grad_norm": 0.5599757351897601,
+      "learning_rate": 9.810239885427856e-06,
+      "loss": 0.0036,
+      "step": 1818
+    },
+    {
+      "epoch": 0.11722626796416834,
+      "grad_norm": 0.012470146755273252,
+      "learning_rate": 9.80952380952381e-06,
+      "loss": 0.0,
+      "step": 1819
+    },
+    {
+      "epoch": 0.11729071341109751,
+      "grad_norm": 0.31943589725439364,
+      "learning_rate": 9.808807733619765e-06,
+      "loss": 0.0017,
+      "step": 1820
+    },
+    {
+      "epoch": 0.11735515885802668,
+      "grad_norm": 0.1360801427960747,
+      "learning_rate": 9.808091657715717e-06,
+      "loss": 0.0003,
+      "step": 1821
+    },
+    {
+      "epoch": 0.11741960430495585,
+      "grad_norm": 1.11044941842528,
+      "learning_rate": 9.807375581811673e-06,
+      "loss": 0.0037,
+      "step": 1822
+    },
+    {
+      "epoch": 0.11748404975188503,
+      "grad_norm": 0.046221214479398,
+      "learning_rate": 9.806659505907628e-06,
+      "loss": 0.0001,
+      "step": 1823
+    },
+    {
+      "epoch": 0.1175484951988142,
+      "grad_norm": 0.014703095802911305,
+      "learning_rate": 9.805943430003582e-06,
+      "loss": 0.0001,
+      "step": 1824
+    },
+    {
+      "epoch": 0.11761294064574337,
+      "grad_norm": 0.09168547607566063,
+      "learning_rate": 9.805227354099536e-06,
+      "loss": 0.0002,
+      "step": 1825
+    },
+    {
+      "epoch": 0.11767738609267256,
+      "grad_norm": 0.04476822615850487,
+      "learning_rate": 9.80451127819549e-06,
+      "loss": 0.0002,
+      "step": 1826
+    },
+    {
+      "epoch": 0.11774183153960173,
+      "grad_norm": 0.23735031819500035,
+      "learning_rate": 9.803795202291445e-06,
+      "loss": 0.0001,
+      "step": 1827
+    },
+    {
+      "epoch": 0.1178062769865309,
+      "grad_norm": 0.019883241428665027,
+      "learning_rate": 9.803079126387397e-06,
+      "loss": 0.0,
+      "step": 1828
+    },
+    {
+      "epoch": 0.11787072243346007,
+      "grad_norm": 0.021027037639753155,
+      "learning_rate": 9.802363050483352e-06,
+      "loss": 0.0002,
+      "step": 1829
+    },
+    {
+      "epoch": 0.11793516788038926,
+      "grad_norm": 0.10222835241112932,
+      "learning_rate": 9.801646974579306e-06,
+      "loss": 0.0009,
+      "step": 1830
+    },
+    {
+      "epoch": 0.11799961332731843,
+      "grad_norm": 0.019721556166337365,
+      "learning_rate": 9.80093089867526e-06,
+      "loss": 0.0001,
+      "step": 1831
+    },
+    {
+      "epoch": 0.1180640587742476,
+      "grad_norm": 0.011796718616704312,
+      "learning_rate": 9.800214822771215e-06,
+      "loss": 0.0001,
+      "step": 1832
+    },
+    {
+      "epoch": 0.11812850422117678,
+      "grad_norm": 1.1332119467653918,
+      "learning_rate": 9.799498746867169e-06,
+      "loss": 0.0074,
+      "step": 1833
+    },
+    {
+      "epoch": 0.11819294966810595,
+      "grad_norm": 0.04183516698015525,
+      "learning_rate": 9.798782670963123e-06,
+      "loss": 0.0001,
+      "step": 1834
+    },
+    {
+      "epoch": 0.11825739511503512,
+      "grad_norm": 0.020117152431605615,
+      "learning_rate": 9.798066595059077e-06,
+      "loss": 0.0,
+      "step": 1835
+    },
+    {
+      "epoch": 0.11832184056196429,
+      "grad_norm": 0.005855164840714329,
+      "learning_rate": 9.797350519155032e-06,
+      "loss": 0.0,
+      "step": 1836
+    },
+    {
+      "epoch": 0.11838628600889348,
+      "grad_norm": 1.0156371482709448,
+      "learning_rate": 9.796634443250986e-06,
+      "loss": 0.0036,
+      "step": 1837
+    },
+    {
+      "epoch": 0.11845073145582265,
+      "grad_norm": 0.14426022971488647,
+      "learning_rate": 9.795918367346939e-06,
+      "loss": 0.0004,
+      "step": 1838
+    },
+    {
+      "epoch": 0.11851517690275182,
+      "grad_norm": 0.059642671585571926,
+      "learning_rate": 9.795202291442893e-06,
+      "loss": 0.0002,
+      "step": 1839
+    },
+    {
+      "epoch": 0.11857962234968099,
+      "grad_norm": 0.01622077639010314,
+      "learning_rate": 9.794486215538847e-06,
+      "loss": 0.0,
+      "step": 1840
+    },
+    {
+      "epoch": 0.11864406779661017,
+      "grad_norm": 0.00355795310393443,
+      "learning_rate": 9.793770139634802e-06,
+      "loss": 0.0,
+      "step": 1841
+    },
+    {
+      "epoch": 0.11870851324353934,
+      "grad_norm": 0.22162360267499878,
+      "learning_rate": 9.793054063730756e-06,
+      "loss": 0.0016,
+      "step": 1842
+    },
+    {
+      "epoch": 0.11877295869046851,
+      "grad_norm": 0.03565730231915416,
+      "learning_rate": 9.79233798782671e-06,
+      "loss": 0.0,
+      "step": 1843
+    },
+    {
+      "epoch": 0.1188374041373977,
+      "grad_norm": 0.04782365939084201,
+      "learning_rate": 9.791621911922664e-06,
+      "loss": 0.0,
+      "step": 1844
+    },
+    {
+      "epoch": 0.11890184958432687,
+      "grad_norm": 0.007612927308443903,
+      "learning_rate": 9.790905836018619e-06,
+      "loss": 0.0,
+      "step": 1845
+    },
+    {
+      "epoch": 0.11896629503125604,
+      "grad_norm": 0.010512277541804959,
+      "learning_rate": 9.790189760114573e-06,
+      "loss": 0.0001,
+      "step": 1846
+    },
+    {
+      "epoch": 0.11903074047818521,
+      "grad_norm": 0.01802613225231296,
+      "learning_rate": 9.789473684210527e-06,
+      "loss": 0.0001,
+      "step": 1847
+    },
+    {
+      "epoch": 0.1190951859251144,
+      "grad_norm": 0.002060216204765636,
+      "learning_rate": 9.788757608306482e-06,
+      "loss": 0.0,
+      "step": 1848
+    },
+    {
+      "epoch": 0.11915963137204356,
+      "grad_norm": 0.01830879044055786,
+      "learning_rate": 9.788041532402436e-06,
+      "loss": 0.0001,
+      "step": 1849
+    },
+    {
+      "epoch": 0.11922407681897274,
+      "grad_norm": 0.1223366844699642,
+      "learning_rate": 9.78732545649839e-06,
+      "loss": 0.002,
+      "step": 1850
+    },
+    {
+      "epoch": 0.11928852226590192,
+      "grad_norm": 0.2997939887607467,
+      "learning_rate": 9.786609380594345e-06,
+      "loss": 0.0004,
+      "step": 1851
+    },
+    {
+      "epoch": 0.11935296771283109,
+      "grad_norm": 0.10824277371729986,
+      "learning_rate": 9.785893304690299e-06,
+      "loss": 0.0003,
+      "step": 1852
+    },
+    {
+      "epoch": 0.11941741315976026,
+      "grad_norm": 0.0021208647960036446,
+      "learning_rate": 9.785177228786253e-06,
+      "loss": 0.0,
+      "step": 1853
+    },
+    {
+      "epoch": 0.11948185860668943,
+      "grad_norm": 0.29415234850131244,
+      "learning_rate": 9.784461152882206e-06,
+      "loss": 0.0007,
+      "step": 1854
+    },
+    {
+      "epoch": 0.11954630405361862,
+      "grad_norm": 0.9598414365134162,
+      "learning_rate": 9.78374507697816e-06,
+      "loss": 0.0098,
+      "step": 1855
+    },
+    {
+      "epoch": 0.11961074950054779,
+      "grad_norm": 0.0016752527904032157,
+      "learning_rate": 9.783029001074114e-06,
+      "loss": 0.0,
+      "step": 1856
+    },
+    {
+      "epoch": 0.11967519494747696,
+      "grad_norm": 0.0483249174057824,
+      "learning_rate": 9.782312925170069e-06,
+      "loss": 0.0001,
+      "step": 1857
+    },
+    {
+      "epoch": 0.11973964039440614,
+      "grad_norm": 0.03142188899828553,
+      "learning_rate": 9.781596849266023e-06,
+      "loss": 0.0003,
+      "step": 1858
+    },
+    {
+      "epoch": 0.11980408584133531,
+      "grad_norm": 0.2874638451772468,
+      "learning_rate": 9.780880773361977e-06,
+      "loss": 0.0006,
+      "step": 1859
+    },
+    {
+      "epoch": 0.11986853128826448,
+      "grad_norm": 0.005053723657460722,
+      "learning_rate": 9.780164697457932e-06,
+      "loss": 0.0,
+      "step": 1860
+    },
+    {
+      "epoch": 0.11993297673519365,
+      "grad_norm": 0.02870584304045045,
+      "learning_rate": 9.779448621553886e-06,
+      "loss": 0.0001,
+      "step": 1861
+    },
+    {
+      "epoch": 0.11999742218212284,
+      "grad_norm": 0.2748382656294078,
+      "learning_rate": 9.77873254564984e-06,
+      "loss": 0.0004,
+      "step": 1862
+    },
+    {
+      "epoch": 0.12006186762905201,
+      "grad_norm": 0.0008884232880952846,
+      "learning_rate": 9.778016469745793e-06,
+      "loss": 0.0,
+      "step": 1863
+    },
+    {
+      "epoch": 0.12012631307598118,
+      "grad_norm": 0.17966616545364283,
+      "learning_rate": 9.777300393841747e-06,
+      "loss": 0.0022,
+      "step": 1864
+    },
+    {
+      "epoch": 0.12019075852291036,
+      "grad_norm": 0.004776275188142645,
+      "learning_rate": 9.776584317937701e-06,
+      "loss": 0.0,
+      "step": 1865
+    },
+    {
+      "epoch": 0.12025520396983953,
+      "grad_norm": 0.006079849315427602,
+      "learning_rate": 9.775868242033656e-06,
+      "loss": 0.0,
+      "step": 1866
+    },
+    {
+      "epoch": 0.1203196494167687,
+      "grad_norm": 0.2494943765783425,
+      "learning_rate": 9.77515216612961e-06,
+      "loss": 0.0005,
+      "step": 1867
+    },
+    {
+      "epoch": 0.12038409486369787,
+      "grad_norm": 0.011233876711569475,
+      "learning_rate": 9.774436090225564e-06,
+      "loss": 0.0001,
+      "step": 1868
+    },
+    {
+      "epoch": 0.12044854031062706,
+      "grad_norm": 0.22918849086157178,
+      "learning_rate": 9.77372001432152e-06,
+      "loss": 0.0007,
+      "step": 1869
+    },
+    {
+      "epoch": 0.12051298575755623,
+      "grad_norm": 0.003232146333357443,
+      "learning_rate": 9.773003938417473e-06,
+      "loss": 0.0,
+      "step": 1870
+    },
+    {
+      "epoch": 0.1205774312044854,
+      "grad_norm": 0.09946201213585422,
+      "learning_rate": 9.772287862513427e-06,
+      "loss": 0.0011,
+      "step": 1871
+    },
+    {
+      "epoch": 0.12064187665141458,
+      "grad_norm": 0.02818442187642415,
+      "learning_rate": 9.771571786609381e-06,
+      "loss": 0.0002,
+      "step": 1872
+    },
+    {
+      "epoch": 0.12070632209834375,
+      "grad_norm": 0.010201301486194175,
+      "learning_rate": 9.770855710705336e-06,
+      "loss": 0.0,
+      "step": 1873
+    },
+    {
+      "epoch": 0.12077076754527293,
+      "grad_norm": 0.6826322714867648,
+      "learning_rate": 9.77013963480129e-06,
+      "loss": 0.0058,
+      "step": 1874
+    },
+    {
+      "epoch": 0.1208352129922021,
+      "grad_norm": 7.93762296335806e-05,
+      "learning_rate": 9.769423558897244e-06,
+      "loss": 0.0,
+      "step": 1875
+    },
+    {
+      "epoch": 0.12089965843913128,
+      "grad_norm": 0.03652186301096464,
+      "learning_rate": 9.768707482993199e-06,
+      "loss": 0.0004,
+      "step": 1876
+    },
+    {
+      "epoch": 0.12096410388606045,
+      "grad_norm": 0.06363982698805158,
+      "learning_rate": 9.767991407089153e-06,
+      "loss": 0.0002,
+      "step": 1877
+    },
+    {
+      "epoch": 0.12102854933298962,
+      "grad_norm": 0.0020954344021100178,
+      "learning_rate": 9.767275331185107e-06,
+      "loss": 0.0,
+      "step": 1878
+    },
+    {
+      "epoch": 0.1210929947799188,
+      "grad_norm": 0.17501861119930448,
+      "learning_rate": 9.76655925528106e-06,
+      "loss": 0.0009,
+      "step": 1879
+    },
+    {
+      "epoch": 0.12115744022684798,
+      "grad_norm": 0.004144092077334751,
+      "learning_rate": 9.765843179377014e-06,
+      "loss": 0.0,
+      "step": 1880
+    },
+    {
+      "epoch": 0.12122188567377715,
+      "grad_norm": 0.0020211306433941265,
+      "learning_rate": 9.765127103472968e-06,
+      "loss": 0.0,
+      "step": 1881
+    },
+    {
+      "epoch": 0.12128633112070632,
+      "grad_norm": 0.02949696401101104,
+      "learning_rate": 9.764411027568923e-06,
+      "loss": 0.0002,
+      "step": 1882
+    },
+    {
+      "epoch": 0.1213507765676355,
+      "grad_norm": 0.016619784508678393,
+      "learning_rate": 9.763694951664877e-06,
+      "loss": 0.0001,
+      "step": 1883
+    },
+    {
+      "epoch": 0.12141522201456467,
+      "grad_norm": 0.3003303690624513,
+      "learning_rate": 9.762978875760831e-06,
+      "loss": 0.0015,
+      "step": 1884
+    },
+    {
+      "epoch": 0.12147966746149384,
+      "grad_norm": 0.0013416382681100284,
+      "learning_rate": 9.762262799856786e-06,
+      "loss": 0.0,
+      "step": 1885
+    },
+    {
+      "epoch": 0.12154411290842301,
+      "grad_norm": 0.00362328786595532,
+      "learning_rate": 9.76154672395274e-06,
+      "loss": 0.0,
+      "step": 1886
+    },
+    {
+      "epoch": 0.1216085583553522,
+      "grad_norm": 0.20061469461587267,
+      "learning_rate": 9.760830648048694e-06,
+      "loss": 0.0008,
+      "step": 1887
+    },
+    {
+      "epoch": 0.12167300380228137,
+      "grad_norm": 0.1765857512431687,
+      "learning_rate": 9.760114572144649e-06,
+      "loss": 0.0007,
+      "step": 1888
+    },
+    {
+      "epoch": 0.12173744924921054,
+      "grad_norm": 0.009202730838186858,
+      "learning_rate": 9.759398496240601e-06,
+      "loss": 0.0001,
+      "step": 1889
+    },
+    {
+      "epoch": 0.12180189469613972,
+      "grad_norm": 0.024221015879938333,
+      "learning_rate": 9.758682420336555e-06,
+      "loss": 0.0001,
+      "step": 1890
+    },
+    {
+      "epoch": 0.1218663401430689,
+      "grad_norm": 0.0032962728768922566,
+      "learning_rate": 9.75796634443251e-06,
+      "loss": 0.0,
+      "step": 1891
+    },
+    {
+      "epoch": 0.12193078558999806,
+      "grad_norm": 0.0017656631596894676,
+      "learning_rate": 9.757250268528466e-06,
+      "loss": 0.0,
+      "step": 1892
+    },
+    {
+      "epoch": 0.12199523103692723,
+      "grad_norm": 0.29931956751586863,
+      "learning_rate": 9.75653419262442e-06,
+      "loss": 0.001,
+      "step": 1893
+    },
+    {
+      "epoch": 0.12205967648385642,
+      "grad_norm": 0.005788445500950225,
+      "learning_rate": 9.755818116720374e-06,
+      "loss": 0.0,
+      "step": 1894
+    },
+    {
+      "epoch": 0.12212412193078559,
+      "grad_norm": 0.0004621153606314104,
+      "learning_rate": 9.755102040816327e-06,
+      "loss": 0.0,
+      "step": 1895
+    },
+    {
+      "epoch": 0.12218856737771476,
+      "grad_norm": 0.0032908027163354404,
+      "learning_rate": 9.754385964912281e-06,
+      "loss": 0.0,
+      "step": 1896
+    },
+    {
+      "epoch": 0.12225301282464394,
+      "grad_norm": 0.011861358115373295,
+      "learning_rate": 9.753669889008236e-06,
+      "loss": 0.0,
+      "step": 1897
+    },
+    {
+      "epoch": 0.12231745827157312,
+      "grad_norm": 0.28728746872163846,
+      "learning_rate": 9.75295381310419e-06,
+      "loss": 0.001,
+      "step": 1898
+    },
+    {
+      "epoch": 0.12238190371850229,
+      "grad_norm": 0.05119925405332734,
+      "learning_rate": 9.752237737200144e-06,
+      "loss": 0.0001,
+      "step": 1899
+    },
+    {
+      "epoch": 0.12244634916543146,
+      "grad_norm": 0.027689745751122395,
+      "learning_rate": 9.751521661296098e-06,
+      "loss": 0.0001,
+      "step": 1900
+    },
+    {
+      "epoch": 0.12251079461236064,
+      "grad_norm": 0.013967656639211944,
+      "learning_rate": 9.750805585392053e-06,
+      "loss": 0.0001,
+      "step": 1901
+    },
+    {
+      "epoch": 0.12257524005928981,
+      "grad_norm": 0.006609898455584625,
+      "learning_rate": 9.750089509488007e-06,
+      "loss": 0.0,
+      "step": 1902
+    },
+    {
+      "epoch": 0.12263968550621898,
+      "grad_norm": 0.0014454099959909776,
+      "learning_rate": 9.749373433583961e-06,
+      "loss": 0.0,
+      "step": 1903
+    },
+    {
+      "epoch": 0.12270413095314817,
+      "grad_norm": 0.00026203836987910003,
+      "learning_rate": 9.748657357679916e-06,
+      "loss": 0.0,
+      "step": 1904
+    },
+    {
+      "epoch": 0.12276857640007734,
+      "grad_norm": 0.00271610028264705,
+      "learning_rate": 9.747941281775868e-06,
+      "loss": 0.0,
+      "step": 1905
+    },
+    {
+      "epoch": 0.12283302184700651,
+      "grad_norm": 0.0016812886013362202,
+      "learning_rate": 9.747225205871823e-06,
+      "loss": 0.0,
+      "step": 1906
+    },
+    {
+      "epoch": 0.12289746729393568,
+      "grad_norm": 0.061560284842779436,
+      "learning_rate": 9.746509129967777e-06,
+      "loss": 0.0002,
+      "step": 1907
+    },
+    {
+      "epoch": 0.12296191274086486,
+      "grad_norm": 0.003250485974073988,
+      "learning_rate": 9.745793054063731e-06,
+      "loss": 0.0,
+      "step": 1908
+    },
+    {
+      "epoch": 0.12302635818779403,
+      "grad_norm": 0.0006700169758031093,
+      "learning_rate": 9.745076978159685e-06,
+      "loss": 0.0,
+      "step": 1909
+    },
+    {
+      "epoch": 0.1230908036347232,
+      "grad_norm": 0.005113968592353677,
+      "learning_rate": 9.74436090225564e-06,
+      "loss": 0.0001,
+      "step": 1910
+    },
+    {
+      "epoch": 0.12315524908165239,
+      "grad_norm": 0.41192248477347315,
+      "learning_rate": 9.743644826351594e-06,
+      "loss": 0.0015,
+      "step": 1911
+    },
+    {
+      "epoch": 0.12321969452858156,
+      "grad_norm": 0.0045803669893567924,
+      "learning_rate": 9.742928750447548e-06,
+      "loss": 0.0,
+      "step": 1912
+    },
+    {
+      "epoch": 0.12328413997551073,
+      "grad_norm": 0.17900576010418873,
+      "learning_rate": 9.742212674543503e-06,
+      "loss": 0.0022,
+      "step": 1913
+    },
+    {
+      "epoch": 0.1233485854224399,
+      "grad_norm": 0.04272674418974187,
+      "learning_rate": 9.741496598639457e-06,
+      "loss": 0.0001,
+      "step": 1914
+    },
+    {
+      "epoch": 0.12341303086936908,
+      "grad_norm": 0.021601893328035547,
+      "learning_rate": 9.740780522735411e-06,
+      "loss": 0.0,
+      "step": 1915
+    },
+    {
+      "epoch": 0.12347747631629825,
+      "grad_norm": 0.13924543879510617,
+      "learning_rate": 9.740064446831366e-06,
+      "loss": 0.001,
+      "step": 1916
+    },
+    {
+      "epoch": 0.12354192176322742,
+      "grad_norm": 0.028301274749120395,
+      "learning_rate": 9.73934837092732e-06,
+      "loss": 0.0001,
+      "step": 1917
+    },
+    {
+      "epoch": 0.12360636721015661,
+      "grad_norm": 0.009873854772588045,
+      "learning_rate": 9.738632295023274e-06,
+      "loss": 0.0001,
+      "step": 1918
+    },
+    {
+      "epoch": 0.12367081265708578,
+      "grad_norm": 0.16220766557710695,
+      "learning_rate": 9.737916219119228e-06,
+      "loss": 0.0007,
+      "step": 1919
+    },
+    {
+      "epoch": 0.12373525810401495,
+      "grad_norm": 0.022465447563904575,
+      "learning_rate": 9.737200143215183e-06,
+      "loss": 0.0001,
+      "step": 1920
+    },
+    {
+      "epoch": 0.12379970355094412,
+      "grad_norm": 0.009806271616587365,
+      "learning_rate": 9.736484067311135e-06,
+      "loss": 0.0,
+      "step": 1921
+    },
+    {
+      "epoch": 0.1238641489978733,
+      "grad_norm": 0.05567401244931437,
+      "learning_rate": 9.73576799140709e-06,
+      "loss": 0.0002,
+      "step": 1922
+    },
+    {
+      "epoch": 0.12392859444480248,
+      "grad_norm": 0.052939701804537706,
+      "learning_rate": 9.735051915503044e-06,
+      "loss": 0.0002,
+      "step": 1923
+    },
+    {
+      "epoch": 0.12399303989173165,
+      "grad_norm": 0.001525405124781685,
+      "learning_rate": 9.734335839598998e-06,
+      "loss": 0.0,
+      "step": 1924
+    },
+    {
+      "epoch": 0.12405748533866082,
+      "grad_norm": 0.37714290288877417,
+      "learning_rate": 9.733619763694953e-06,
+      "loss": 0.0011,
+      "step": 1925
+    },
+    {
+      "epoch": 0.12412193078559,
+      "grad_norm": 0.001764385053678129,
+      "learning_rate": 9.732903687790907e-06,
+      "loss": 0.0,
+      "step": 1926
+    },
+    {
+      "epoch": 0.12418637623251917,
+      "grad_norm": 0.019501183135983383,
+      "learning_rate": 9.732187611886861e-06,
+      "loss": 0.0002,
+      "step": 1927
+    },
+    {
+      "epoch": 0.12425082167944834,
+      "grad_norm": 0.018751990192806996,
+      "learning_rate": 9.731471535982815e-06,
+      "loss": 0.0001,
+      "step": 1928
+    },
+    {
+      "epoch": 0.12431526712637753,
+      "grad_norm": 0.004955295051767424,
+      "learning_rate": 9.73075546007877e-06,
+      "loss": 0.0,
+      "step": 1929
+    },
+    {
+      "epoch": 0.1243797125733067,
+      "grad_norm": 0.061112311234522064,
+      "learning_rate": 9.730039384174724e-06,
+      "loss": 0.0002,
+      "step": 1930
+    },
+    {
+      "epoch": 0.12444415802023587,
+      "grad_norm": 0.004004600983625894,
+      "learning_rate": 9.729323308270677e-06,
+      "loss": 0.0,
+      "step": 1931
+    },
+    {
+      "epoch": 0.12450860346716504,
+      "grad_norm": 0.0003309675091962049,
+      "learning_rate": 9.728607232366631e-06,
+      "loss": 0.0,
+      "step": 1932
+    },
+    {
+      "epoch": 0.12457304891409422,
+      "grad_norm": 0.15170736542210217,
+      "learning_rate": 9.727891156462585e-06,
+      "loss": 0.002,
+      "step": 1933
+    },
+    {
+      "epoch": 0.1246374943610234,
+      "grad_norm": 1.4362513882968126,
+      "learning_rate": 9.72717508055854e-06,
+      "loss": 0.0162,
+      "step": 1934
+    },
+    {
+      "epoch": 0.12470193980795256,
+      "grad_norm": 0.014300368475316579,
+      "learning_rate": 9.726459004654494e-06,
+      "loss": 0.0001,
+      "step": 1935
+    },
+    {
+      "epoch": 0.12476638525488175,
+      "grad_norm": 0.004234241794280436,
+      "learning_rate": 9.725742928750448e-06,
+      "loss": 0.0001,
+      "step": 1936
+    },
+    {
+      "epoch": 0.12483083070181092,
+      "grad_norm": 0.015785880095632743,
+      "learning_rate": 9.725026852846402e-06,
+      "loss": 0.0,
+      "step": 1937
+    },
+    {
+      "epoch": 0.12489527614874009,
+      "grad_norm": 0.16290879479847703,
+      "learning_rate": 9.724310776942357e-06,
+      "loss": 0.0019,
+      "step": 1938
+    },
+    {
+      "epoch": 0.12495972159566926,
+      "grad_norm": 0.0871295187201113,
+      "learning_rate": 9.723594701038311e-06,
+      "loss": 0.0017,
+      "step": 1939
+    },
+    {
+      "epoch": 0.12502416704259844,
+      "grad_norm": 0.04819015484545837,
+      "learning_rate": 9.722878625134265e-06,
+      "loss": 0.0001,
+      "step": 1940
+    },
+    {
+      "epoch": 0.12508861248952763,
+      "grad_norm": 0.043515775119977985,
+      "learning_rate": 9.72216254923022e-06,
+      "loss": 0.0006,
+      "step": 1941
+    },
+    {
+      "epoch": 0.12515305793645679,
+      "grad_norm": 0.019120648631456776,
+      "learning_rate": 9.721446473326174e-06,
+      "loss": 0.0002,
+      "step": 1942
+    },
+    {
+      "epoch": 0.12521750338338597,
+      "grad_norm": 0.06639534145403339,
+      "learning_rate": 9.720730397422128e-06,
+      "loss": 0.0002,
+      "step": 1943
+    },
+    {
+      "epoch": 0.12528194883031513,
+      "grad_norm": 0.32793971915972786,
+      "learning_rate": 9.720014321518082e-06,
+      "loss": 0.0003,
+      "step": 1944
+    },
+    {
+      "epoch": 0.1253463942772443,
+      "grad_norm": 0.23585167454332714,
+      "learning_rate": 9.719298245614037e-06,
+      "loss": 0.0004,
+      "step": 1945
+    },
+    {
+      "epoch": 0.1254108397241735,
+      "grad_norm": 0.08747161219538542,
+      "learning_rate": 9.718582169709991e-06,
+      "loss": 0.0003,
+      "step": 1946
+    },
+    {
+      "epoch": 0.12547528517110265,
+      "grad_norm": 0.08837184298199861,
+      "learning_rate": 9.717866093805944e-06,
+      "loss": 0.0001,
+      "step": 1947
+    },
+    {
+      "epoch": 0.12553973061803184,
+      "grad_norm": 0.048392791842107145,
+      "learning_rate": 9.717150017901898e-06,
+      "loss": 0.0001,
+      "step": 1948
+    },
+    {
+      "epoch": 0.12560417606496102,
+      "grad_norm": 0.9592803684743428,
+      "learning_rate": 9.716433941997852e-06,
+      "loss": 0.0082,
+      "step": 1949
+    },
+    {
+      "epoch": 0.12566862151189018,
+      "grad_norm": 0.2817725519397007,
+      "learning_rate": 9.715717866093807e-06,
+      "loss": 0.0003,
+      "step": 1950
+    },
+    {
+      "epoch": 0.12573306695881936,
+      "grad_norm": 0.012663876833443062,
+      "learning_rate": 9.715001790189761e-06,
+      "loss": 0.0,
+      "step": 1951
+    },
+    {
+      "epoch": 0.12579751240574855,
+      "grad_norm": 0.05103851906930751,
+      "learning_rate": 9.714285714285715e-06,
+      "loss": 0.0016,
+      "step": 1952
+    },
+    {
+      "epoch": 0.1258619578526777,
+      "grad_norm": 0.07201781829296526,
+      "learning_rate": 9.71356963838167e-06,
+      "loss": 0.0003,
+      "step": 1953
+    },
+    {
+      "epoch": 0.1259264032996069,
+      "grad_norm": 0.0009689525530489202,
+      "learning_rate": 9.712853562477624e-06,
+      "loss": 0.0,
+      "step": 1954
+    },
+    {
+      "epoch": 0.12599084874653604,
+      "grad_norm": 0.0145900601196395,
+      "learning_rate": 9.712137486573578e-06,
+      "loss": 0.0,
+      "step": 1955
+    },
+    {
+      "epoch": 0.12605529419346523,
+      "grad_norm": 0.14103773266063874,
+      "learning_rate": 9.71142141066953e-06,
+      "loss": 0.0005,
+      "step": 1956
+    },
+    {
+      "epoch": 0.1261197396403944,
+      "grad_norm": 0.007242582761787885,
+      "learning_rate": 9.710705334765485e-06,
+      "loss": 0.0,
+      "step": 1957
+    },
+    {
+      "epoch": 0.12618418508732357,
+      "grad_norm": 2.3089804801462774,
+      "learning_rate": 9.70998925886144e-06,
+      "loss": 0.018,
+      "step": 1958
+    },
+    {
+      "epoch": 0.12624863053425275,
+      "grad_norm": 0.001684282938530845,
+      "learning_rate": 9.709273182957394e-06,
+      "loss": 0.0,
+      "step": 1959
+    },
+    {
+      "epoch": 0.12631307598118194,
+      "grad_norm": 0.008110944387793166,
+      "learning_rate": 9.708557107053348e-06,
+      "loss": 0.0,
+      "step": 1960
+    },
+    {
+      "epoch": 0.1263775214281111,
+      "grad_norm": 0.0012180645839539397,
+      "learning_rate": 9.707841031149302e-06,
+      "loss": 0.0,
+      "step": 1961
+    },
+    {
+      "epoch": 0.12644196687504028,
+      "grad_norm": 0.0032079103488066195,
+      "learning_rate": 9.707124955245258e-06,
+      "loss": 0.0,
+      "step": 1962
+    },
+    {
+      "epoch": 0.12650641232196946,
+      "grad_norm": 0.0015223787814444203,
+      "learning_rate": 9.70640887934121e-06,
+      "loss": 0.0,
+      "step": 1963
+    },
+    {
+      "epoch": 0.12657085776889862,
+      "grad_norm": 0.002616986152407543,
+      "learning_rate": 9.705692803437165e-06,
+      "loss": 0.0,
+      "step": 1964
+    },
+    {
+      "epoch": 0.1266353032158278,
+      "grad_norm": 0.45123696786549056,
+      "learning_rate": 9.70497672753312e-06,
+      "loss": 0.0022,
+      "step": 1965
+    },
+    {
+      "epoch": 0.126699748662757,
+      "grad_norm": 0.5187885166714238,
+      "learning_rate": 9.704260651629074e-06,
+      "loss": 0.0025,
+      "step": 1966
+    },
+    {
+      "epoch": 0.12676419410968615,
+      "grad_norm": 0.002532257006504168,
+      "learning_rate": 9.703544575725028e-06,
+      "loss": 0.0,
+      "step": 1967
+    },
+    {
+      "epoch": 0.12682863955661533,
+      "grad_norm": 0.0024266808451328583,
+      "learning_rate": 9.702828499820982e-06,
+      "loss": 0.0,
+      "step": 1968
+    },
+    {
+      "epoch": 0.1268930850035445,
+      "grad_norm": 0.000678778908986063,
+      "learning_rate": 9.702112423916937e-06,
+      "loss": 0.0,
+      "step": 1969
+    },
+    {
+      "epoch": 0.12695753045047367,
+      "grad_norm": 0.2541039650735298,
+      "learning_rate": 9.701396348012891e-06,
+      "loss": 0.002,
+      "step": 1970
+    },
+    {
+      "epoch": 0.12702197589740286,
+      "grad_norm": 0.05886083764835366,
+      "learning_rate": 9.700680272108845e-06,
+      "loss": 0.0005,
+      "step": 1971
+    },
+    {
+      "epoch": 0.127086421344332,
+      "grad_norm": 0.01571323952426794,
+      "learning_rate": 9.699964196204798e-06,
+      "loss": 0.0001,
+      "step": 1972
+    },
+    {
+      "epoch": 0.1271508667912612,
+      "grad_norm": 0.003961847090334322,
+      "learning_rate": 9.699248120300752e-06,
+      "loss": 0.0,
+      "step": 1973
+    },
+    {
+      "epoch": 0.12721531223819038,
+      "grad_norm": 0.027488595208986197,
+      "learning_rate": 9.698532044396706e-06,
+      "loss": 0.0,
+      "step": 1974
+    },
+    {
+      "epoch": 0.12727975768511954,
+      "grad_norm": 0.011537951536441094,
+      "learning_rate": 9.69781596849266e-06,
+      "loss": 0.0,
+      "step": 1975
+    },
+    {
+      "epoch": 0.12734420313204872,
+      "grad_norm": 0.06099060148991771,
+      "learning_rate": 9.697099892588615e-06,
+      "loss": 0.0001,
+      "step": 1976
+    },
+    {
+      "epoch": 0.1274086485789779,
+      "grad_norm": 0.0044298324954258495,
+      "learning_rate": 9.69638381668457e-06,
+      "loss": 0.0,
+      "step": 1977
+    },
+    {
+      "epoch": 0.12747309402590706,
+      "grad_norm": 0.03492540826460728,
+      "learning_rate": 9.695667740780524e-06,
+      "loss": 0.0001,
+      "step": 1978
+    },
+    {
+      "epoch": 0.12753753947283625,
+      "grad_norm": 0.01319119997833318,
+      "learning_rate": 9.694951664876478e-06,
+      "loss": 0.0,
+      "step": 1979
+    },
+    {
+      "epoch": 0.12760198491976543,
+      "grad_norm": 0.007561792083493985,
+      "learning_rate": 9.694235588972432e-06,
+      "loss": 0.0,
+      "step": 1980
+    },
+    {
+      "epoch": 0.1276664303666946,
+      "grad_norm": 0.04029629379038578,
+      "learning_rate": 9.693519513068386e-06,
+      "loss": 0.0001,
+      "step": 1981
+    },
+    {
+      "epoch": 0.12773087581362377,
+      "grad_norm": 0.0435667968924196,
+      "learning_rate": 9.692803437164339e-06,
+      "loss": 0.0002,
+      "step": 1982
+    },
+    {
+      "epoch": 0.12779532126055293,
+      "grad_norm": 0.010267197277295997,
+      "learning_rate": 9.692087361260293e-06,
+      "loss": 0.0,
+      "step": 1983
+    },
+    {
+      "epoch": 0.12785976670748211,
+      "grad_norm": 0.11801684485489015,
+      "learning_rate": 9.691371285356248e-06,
+      "loss": 0.0023,
+      "step": 1984
+    },
+    {
+      "epoch": 0.1279242121544113,
+      "grad_norm": 0.02908363021812742,
+      "learning_rate": 9.690655209452202e-06,
+      "loss": 0.0001,
+      "step": 1985
+    },
+    {
+      "epoch": 0.12798865760134046,
+      "grad_norm": 0.0020590699766240743,
+      "learning_rate": 9.689939133548158e-06,
+      "loss": 0.0,
+      "step": 1986
+    },
+    {
+      "epoch": 0.12805310304826964,
+      "grad_norm": 0.18389281811274666,
+      "learning_rate": 9.689223057644112e-06,
+      "loss": 0.0002,
+      "step": 1987
+    },
+    {
+      "epoch": 0.12811754849519882,
+      "grad_norm": 0.09317822238790244,
+      "learning_rate": 9.688506981740065e-06,
+      "loss": 0.002,
+      "step": 1988
+    },
+    {
+      "epoch": 0.12818199394212798,
+      "grad_norm": 0.2188158447077549,
+      "learning_rate": 9.687790905836019e-06,
+      "loss": 0.001,
+      "step": 1989
+    },
+    {
+      "epoch": 0.12824643938905717,
+      "grad_norm": 0.0652653566331668,
+      "learning_rate": 9.687074829931973e-06,
+      "loss": 0.0001,
+      "step": 1990
+    },
+    {
+      "epoch": 0.12831088483598635,
+      "grad_norm": 0.08241202614232288,
+      "learning_rate": 9.686358754027928e-06,
+      "loss": 0.0001,
+      "step": 1991
+    },
+    {
+      "epoch": 0.1283753302829155,
+      "grad_norm": 0.006286295574519581,
+      "learning_rate": 9.685642678123882e-06,
+      "loss": 0.0,
+      "step": 1992
+    },
+    {
+      "epoch": 0.1284397757298447,
+      "grad_norm": 0.003714464869485905,
+      "learning_rate": 9.684926602219836e-06,
+      "loss": 0.0,
+      "step": 1993
+    },
+    {
+      "epoch": 0.12850422117677385,
+      "grad_norm": 0.005759371136794642,
+      "learning_rate": 9.68421052631579e-06,
+      "loss": 0.0,
+      "step": 1994
+    },
+    {
+      "epoch": 0.12856866662370303,
+      "grad_norm": 0.005675051030252726,
+      "learning_rate": 9.683494450411745e-06,
+      "loss": 0.0001,
+      "step": 1995
+    },
+    {
+      "epoch": 0.12863311207063222,
+      "grad_norm": 0.0024021179085006966,
+      "learning_rate": 9.6827783745077e-06,
+      "loss": 0.0,
+      "step": 1996
+    },
+    {
+      "epoch": 0.12869755751756137,
+      "grad_norm": 0.0003785387330054233,
+      "learning_rate": 9.682062298603654e-06,
+      "loss": 0.0,
+      "step": 1997
+    },
+    {
+      "epoch": 0.12876200296449056,
+      "grad_norm": 0.029128929405783818,
+      "learning_rate": 9.681346222699606e-06,
+      "loss": 0.0003,
+      "step": 1998
+    },
+    {
+      "epoch": 0.12882644841141974,
+      "grad_norm": 0.00965094471811793,
+      "learning_rate": 9.68063014679556e-06,
+      "loss": 0.0001,
+      "step": 1999
+    },
+    {
+      "epoch": 0.1288908938583489,
+      "grad_norm": 0.1631243083836886,
+      "learning_rate": 9.679914070891515e-06,
+      "loss": 0.0004,
+      "step": 2000
+    },
+    {
+      "epoch": 0.12895533930527808,
+      "grad_norm": 0.0008463747064084646,
+      "learning_rate": 9.679197994987469e-06,
+      "loss": 0.0,
+      "step": 2001
+    },
+    {
+      "epoch": 0.12901978475220727,
+      "grad_norm": 0.16633632762422967,
+      "learning_rate": 9.678481919083423e-06,
+      "loss": 0.0017,
+      "step": 2002
+    },
+    {
+      "epoch": 0.12908423019913642,
+      "grad_norm": 0.010527829476286266,
+      "learning_rate": 9.677765843179378e-06,
+      "loss": 0.0,
+      "step": 2003
+    },
+    {
+      "epoch": 0.1291486756460656,
+      "grad_norm": 0.0005172988190096706,
+      "learning_rate": 9.677049767275332e-06,
+      "loss": 0.0,
+      "step": 2004
+    },
+    {
+      "epoch": 0.1292131210929948,
+      "grad_norm": 0.5475268974403349,
+      "learning_rate": 9.676333691371286e-06,
+      "loss": 0.0049,
+      "step": 2005
+    },
+    {
+      "epoch": 0.12927756653992395,
+      "grad_norm": 0.3089875231942968,
+      "learning_rate": 9.67561761546724e-06,
+      "loss": 0.0015,
+      "step": 2006
+    },
+    {
+      "epoch": 0.12934201198685313,
+      "grad_norm": 0.4836295297972049,
+      "learning_rate": 9.674901539563195e-06,
+      "loss": 0.0014,
+      "step": 2007
+    },
+    {
+      "epoch": 0.1294064574337823,
+      "grad_norm": 0.027169375141539498,
+      "learning_rate": 9.674185463659147e-06,
+      "loss": 0.0001,
+      "step": 2008
+    },
+    {
+      "epoch": 0.12947090288071147,
+      "grad_norm": 0.021970903803559733,
+      "learning_rate": 9.673469387755103e-06,
+      "loss": 0.0001,
+      "step": 2009
+    },
+    {
+      "epoch": 0.12953534832764066,
+      "grad_norm": 0.009162414864943546,
+      "learning_rate": 9.672753311851058e-06,
+      "loss": 0.0,
+      "step": 2010
+    },
+    {
+      "epoch": 0.12959979377456982,
+      "grad_norm": 0.11454378921360125,
+      "learning_rate": 9.672037235947012e-06,
+      "loss": 0.0001,
+      "step": 2011
+    },
+    {
+      "epoch": 0.129664239221499,
+      "grad_norm": 0.029784219172073287,
+      "learning_rate": 9.671321160042966e-06,
+      "loss": 0.0,
+      "step": 2012
+    },
+    {
+      "epoch": 0.12972868466842818,
+      "grad_norm": 0.023949376703751586,
+      "learning_rate": 9.67060508413892e-06,
+      "loss": 0.0001,
+      "step": 2013
+    },
+    {
+      "epoch": 0.12979313011535734,
+      "grad_norm": 0.002974615832058479,
+      "learning_rate": 9.669889008234873e-06,
+      "loss": 0.0,
+      "step": 2014
+    },
+    {
+      "epoch": 0.12985757556228653,
+      "grad_norm": 0.0378054906125581,
+      "learning_rate": 9.669172932330828e-06,
+      "loss": 0.0002,
+      "step": 2015
+    },
+    {
+      "epoch": 0.1299220210092157,
+      "grad_norm": 0.9699590275654907,
+      "learning_rate": 9.668456856426782e-06,
+      "loss": 0.0052,
+      "step": 2016
+    },
+    {
+      "epoch": 0.12998646645614487,
+      "grad_norm": 0.04729061845179302,
+      "learning_rate": 9.667740780522736e-06,
+      "loss": 0.0002,
+      "step": 2017
+    },
+    {
+      "epoch": 0.13005091190307405,
+      "grad_norm": 0.0011672871369493613,
+      "learning_rate": 9.66702470461869e-06,
+      "loss": 0.0,
+      "step": 2018
+    },
+    {
+      "epoch": 0.13011535735000324,
+      "grad_norm": 0.16779780560629792,
+      "learning_rate": 9.666308628714645e-06,
+      "loss": 0.0013,
+      "step": 2019
+    },
+    {
+      "epoch": 0.1301798027969324,
+      "grad_norm": 0.009769002881575378,
+      "learning_rate": 9.665592552810599e-06,
+      "loss": 0.0,
+      "step": 2020
+    },
+    {
+      "epoch": 0.13024424824386158,
+      "grad_norm": 0.06925341571206166,
+      "learning_rate": 9.664876476906553e-06,
+      "loss": 0.0016,
+      "step": 2021
+    },
+    {
+      "epoch": 0.13030869369079073,
+      "grad_norm": 0.028873011933198297,
+      "learning_rate": 9.664160401002508e-06,
+      "loss": 0.0002,
+      "step": 2022
+    },
+    {
+      "epoch": 0.13037313913771992,
+      "grad_norm": 0.030437929164553016,
+      "learning_rate": 9.663444325098462e-06,
+      "loss": 0.0003,
+      "step": 2023
+    },
+    {
+      "epoch": 0.1304375845846491,
+      "grad_norm": 0.04157236987937166,
+      "learning_rate": 9.662728249194415e-06,
+      "loss": 0.0,
+      "step": 2024
+    },
+    {
+      "epoch": 0.13050203003157826,
+      "grad_norm": 0.002376335667268474,
+      "learning_rate": 9.662012173290369e-06,
+      "loss": 0.0,
+      "step": 2025
+    },
+    {
+      "epoch": 0.13056647547850744,
+      "grad_norm": 0.4035950618520248,
+      "learning_rate": 9.661296097386323e-06,
+      "loss": 0.0029,
+      "step": 2026
+    },
+    {
+      "epoch": 0.13063092092543663,
+      "grad_norm": 0.027696990872225418,
+      "learning_rate": 9.660580021482277e-06,
+      "loss": 0.0,
+      "step": 2027
+    },
+    {
+      "epoch": 0.13069536637236578,
+      "grad_norm": 0.04223727765421684,
+      "learning_rate": 9.659863945578232e-06,
+      "loss": 0.0001,
+      "step": 2028
+    },
+    {
+      "epoch": 0.13075981181929497,
+      "grad_norm": 0.07407476864774445,
+      "learning_rate": 9.659147869674186e-06,
+      "loss": 0.0001,
+      "step": 2029
+    },
+    {
+      "epoch": 0.13082425726622415,
+      "grad_norm": 0.06935889911574292,
+      "learning_rate": 9.65843179377014e-06,
+      "loss": 0.0001,
+      "step": 2030
+    },
+    {
+      "epoch": 0.1308887027131533,
+      "grad_norm": 0.048697351332985216,
+      "learning_rate": 9.657715717866095e-06,
+      "loss": 0.0003,
+      "step": 2031
+    },
+    {
+      "epoch": 0.1309531481600825,
+      "grad_norm": 0.005297919283381317,
+      "learning_rate": 9.656999641962049e-06,
+      "loss": 0.0,
+      "step": 2032
+    },
+    {
+      "epoch": 0.13101759360701165,
+      "grad_norm": 0.026038987175857987,
+      "learning_rate": 9.656283566058003e-06,
+      "loss": 0.0,
+      "step": 2033
+    },
+    {
+      "epoch": 0.13108203905394084,
+      "grad_norm": 0.031187994357722387,
+      "learning_rate": 9.655567490153958e-06,
+      "loss": 0.0,
+      "step": 2034
+    },
+    {
+      "epoch": 0.13114648450087002,
+      "grad_norm": 0.0026731337132794813,
+      "learning_rate": 9.654851414249912e-06,
+      "loss": 0.0,
+      "step": 2035
+    },
+    {
+      "epoch": 0.13121092994779918,
+      "grad_norm": 0.031899373503944894,
+      "learning_rate": 9.654135338345866e-06,
+      "loss": 0.0,
+      "step": 2036
+    },
+    {
+      "epoch": 0.13127537539472836,
+      "grad_norm": 0.19949602017699414,
+      "learning_rate": 9.65341926244182e-06,
+      "loss": 0.0018,
+      "step": 2037
+    },
+    {
+      "epoch": 0.13133982084165755,
+      "grad_norm": 0.08790331325073492,
+      "learning_rate": 9.652703186537775e-06,
+      "loss": 0.0001,
+      "step": 2038
+    },
+    {
+      "epoch": 0.1314042662885867,
+      "grad_norm": 0.017596495982531747,
+      "learning_rate": 9.651987110633729e-06,
+      "loss": 0.0001,
+      "step": 2039
+    },
+    {
+      "epoch": 0.1314687117355159,
+      "grad_norm": 0.019189778384638312,
+      "learning_rate": 9.651271034729682e-06,
+      "loss": 0.0002,
+      "step": 2040
+    },
+    {
+      "epoch": 0.13153315718244507,
+      "grad_norm": 0.06314370843579555,
+      "learning_rate": 9.650554958825636e-06,
+      "loss": 0.0003,
+      "step": 2041
+    },
+    {
+      "epoch": 0.13159760262937423,
+      "grad_norm": 0.0006894959370470676,
+      "learning_rate": 9.64983888292159e-06,
+      "loss": 0.0,
+      "step": 2042
+    },
+    {
+      "epoch": 0.1316620480763034,
+      "grad_norm": 0.37931082451387943,
+      "learning_rate": 9.649122807017545e-06,
+      "loss": 0.0007,
+      "step": 2043
+    },
+    {
+      "epoch": 0.1317264935232326,
+      "grad_norm": 0.0030844319112572124,
+      "learning_rate": 9.648406731113499e-06,
+      "loss": 0.0,
+      "step": 2044
+    },
+    {
+      "epoch": 0.13179093897016175,
+      "grad_norm": 0.08177165554426978,
+      "learning_rate": 9.647690655209453e-06,
+      "loss": 0.0004,
+      "step": 2045
+    },
+    {
+      "epoch": 0.13185538441709094,
+      "grad_norm": 0.43350847797617365,
+      "learning_rate": 9.646974579305407e-06,
+      "loss": 0.0023,
+      "step": 2046
+    },
+    {
+      "epoch": 0.1319198298640201,
+      "grad_norm": 0.017477741474904874,
+      "learning_rate": 9.646258503401362e-06,
+      "loss": 0.0,
+      "step": 2047
+    },
+    {
+      "epoch": 0.13198427531094928,
+      "grad_norm": 0.002203850556521162,
+      "learning_rate": 9.645542427497316e-06,
+      "loss": 0.0,
+      "step": 2048
+    },
+    {
+      "epoch": 0.13204872075787846,
+      "grad_norm": 0.31950717965940695,
+      "learning_rate": 9.644826351593269e-06,
+      "loss": 0.0006,
+      "step": 2049
+    },
+    {
+      "epoch": 0.13211316620480762,
+      "grad_norm": 0.2171599295152496,
+      "learning_rate": 9.644110275689223e-06,
+      "loss": 0.0012,
+      "step": 2050
+    },
+    {
+      "epoch": 0.1321776116517368,
+      "grad_norm": 0.006761549558222412,
+      "learning_rate": 9.643394199785177e-06,
+      "loss": 0.0001,
+      "step": 2051
+    },
+    {
+      "epoch": 0.132242057098666,
+      "grad_norm": 0.21121347000074073,
+      "learning_rate": 9.642678123881131e-06,
+      "loss": 0.0016,
+      "step": 2052
+    },
+    {
+      "epoch": 0.13230650254559514,
+      "grad_norm": 0.00621073689854682,
+      "learning_rate": 9.641962047977086e-06,
+      "loss": 0.0001,
+      "step": 2053
+    },
+    {
+      "epoch": 0.13237094799252433,
+      "grad_norm": 0.32037854095344764,
+      "learning_rate": 9.64124597207304e-06,
+      "loss": 0.0011,
+      "step": 2054
+    },
+    {
+      "epoch": 0.1324353934394535,
+      "grad_norm": 0.009164569823797965,
+      "learning_rate": 9.640529896168994e-06,
+      "loss": 0.0,
+      "step": 2055
+    },
+    {
+      "epoch": 0.13249983888638267,
+      "grad_norm": 0.02221471650999578,
+      "learning_rate": 9.639813820264949e-06,
+      "loss": 0.0002,
+      "step": 2056
+    },
+    {
+      "epoch": 0.13256428433331185,
+      "grad_norm": 0.000575011763455413,
+      "learning_rate": 9.639097744360903e-06,
+      "loss": 0.0,
+      "step": 2057
+    },
+    {
+      "epoch": 0.13262872978024104,
+      "grad_norm": 0.010129757102013255,
+      "learning_rate": 9.638381668456857e-06,
+      "loss": 0.0,
+      "step": 2058
+    },
+    {
+      "epoch": 0.1326931752271702,
+      "grad_norm": 0.1409973473472684,
+      "learning_rate": 9.637665592552812e-06,
+      "loss": 0.0013,
+      "step": 2059
+    },
+    {
+      "epoch": 0.13275762067409938,
+      "grad_norm": 0.005983310707828044,
+      "learning_rate": 9.636949516648766e-06,
+      "loss": 0.0,
+      "step": 2060
+    },
+    {
+      "epoch": 0.13282206612102854,
+      "grad_norm": 0.0009344883651601945,
+      "learning_rate": 9.63623344074472e-06,
+      "loss": 0.0,
+      "step": 2061
+    },
+    {
+      "epoch": 0.13288651156795772,
+      "grad_norm": 0.003960576942543866,
+      "learning_rate": 9.635517364840674e-06,
+      "loss": 0.0,
+      "step": 2062
+    },
+    {
+      "epoch": 0.1329509570148869,
+      "grad_norm": 0.5569627866898234,
+      "learning_rate": 9.634801288936629e-06,
+      "loss": 0.0028,
+      "step": 2063
+    },
+    {
+      "epoch": 0.13301540246181606,
+      "grad_norm": 0.0047079077069493705,
+      "learning_rate": 9.634085213032583e-06,
+      "loss": 0.0,
+      "step": 2064
+    },
+    {
+      "epoch": 0.13307984790874525,
+      "grad_norm": 0.003906507967835101,
+      "learning_rate": 9.633369137128536e-06,
+      "loss": 0.0,
+      "step": 2065
+    },
+    {
+      "epoch": 0.13314429335567443,
+      "grad_norm": 0.11023906442304471,
+      "learning_rate": 9.63265306122449e-06,
+      "loss": 0.0003,
+      "step": 2066
+    },
+    {
+      "epoch": 0.1332087388026036,
+      "grad_norm": 0.0004294632581472271,
+      "learning_rate": 9.631936985320444e-06,
+      "loss": 0.0,
+      "step": 2067
+    },
+    {
+      "epoch": 0.13327318424953277,
+      "grad_norm": 0.11307068278807467,
+      "learning_rate": 9.631220909416399e-06,
+      "loss": 0.0021,
+      "step": 2068
+    },
+    {
+      "epoch": 0.13333762969646196,
+      "grad_norm": 0.0006063610247870296,
+      "learning_rate": 9.630504833512353e-06,
+      "loss": 0.0,
+      "step": 2069
+    },
+    {
+      "epoch": 0.1334020751433911,
+      "grad_norm": 0.0014840911368628803,
+      "learning_rate": 9.629788757608307e-06,
+      "loss": 0.0,
+      "step": 2070
+    },
+    {
+      "epoch": 0.1334665205903203,
+      "grad_norm": 0.6738283019134732,
+      "learning_rate": 9.629072681704261e-06,
+      "loss": 0.0026,
+      "step": 2071
+    },
+    {
+      "epoch": 0.13353096603724945,
+      "grad_norm": 0.0011470253664121449,
+      "learning_rate": 9.628356605800216e-06,
+      "loss": 0.0,
+      "step": 2072
+    },
+    {
+      "epoch": 0.13359541148417864,
+      "grad_norm": 0.0026127623637299136,
+      "learning_rate": 9.62764052989617e-06,
+      "loss": 0.0,
+      "step": 2073
+    },
+    {
+      "epoch": 0.13365985693110782,
+      "grad_norm": 0.0012740458006002587,
+      "learning_rate": 9.626924453992124e-06,
+      "loss": 0.0,
+      "step": 2074
+    },
+    {
+      "epoch": 0.13372430237803698,
+      "grad_norm": 0.01763503065907178,
+      "learning_rate": 9.626208378088077e-06,
+      "loss": 0.0001,
+      "step": 2075
+    },
+    {
+      "epoch": 0.13378874782496616,
+      "grad_norm": 0.023128139248136018,
+      "learning_rate": 9.625492302184031e-06,
+      "loss": 0.0003,
+      "step": 2076
+    },
+    {
+      "epoch": 0.13385319327189535,
+      "grad_norm": 0.027094712760084454,
+      "learning_rate": 9.624776226279986e-06,
+      "loss": 0.0,
+      "step": 2077
+    },
+    {
+      "epoch": 0.1339176387188245,
+      "grad_norm": 0.006780221983481411,
+      "learning_rate": 9.62406015037594e-06,
+      "loss": 0.0,
+      "step": 2078
+    },
+    {
+      "epoch": 0.1339820841657537,
+      "grad_norm": 0.03709572145604888,
+      "learning_rate": 9.623344074471896e-06,
+      "loss": 0.0,
+      "step": 2079
+    },
+    {
+      "epoch": 0.13404652961268287,
+      "grad_norm": 0.013222114372806895,
+      "learning_rate": 9.62262799856785e-06,
+      "loss": 0.0,
+      "step": 2080
+    },
+    {
+      "epoch": 0.13411097505961203,
+      "grad_norm": 2.5401217554602096,
+      "learning_rate": 9.621911922663803e-06,
+      "loss": 0.0209,
+      "step": 2081
+    },
+    {
+      "epoch": 0.13417542050654122,
+      "grad_norm": 1.0853759081919199,
+      "learning_rate": 9.621195846759757e-06,
+      "loss": 0.004,
+      "step": 2082
+    },
+    {
+      "epoch": 0.1342398659534704,
+      "grad_norm": 0.0014960894831067125,
+      "learning_rate": 9.620479770855711e-06,
+      "loss": 0.0,
+      "step": 2083
+    },
+    {
+      "epoch": 0.13430431140039956,
+      "grad_norm": 0.0010822836961904576,
+      "learning_rate": 9.619763694951666e-06,
+      "loss": 0.0,
+      "step": 2084
+    },
+    {
+      "epoch": 0.13436875684732874,
+      "grad_norm": 0.0020887246339200602,
+      "learning_rate": 9.61904761904762e-06,
+      "loss": 0.0,
+      "step": 2085
+    },
+    {
+      "epoch": 0.1344332022942579,
+      "grad_norm": 0.0006119661795616889,
+      "learning_rate": 9.618331543143574e-06,
+      "loss": 0.0,
+      "step": 2086
+    },
+    {
+      "epoch": 0.13449764774118708,
+      "grad_norm": 0.0343665176866361,
+      "learning_rate": 9.617615467239529e-06,
+      "loss": 0.0001,
+      "step": 2087
+    },
+    {
+      "epoch": 0.13456209318811627,
+      "grad_norm": 0.013401859452242893,
+      "learning_rate": 9.616899391335483e-06,
+      "loss": 0.0,
+      "step": 2088
+    },
+    {
+      "epoch": 0.13462653863504542,
+      "grad_norm": 0.012735135863470763,
+      "learning_rate": 9.616183315431437e-06,
+      "loss": 0.0,
+      "step": 2089
+    },
+    {
+      "epoch": 0.1346909840819746,
+      "grad_norm": 0.00041242381199152895,
+      "learning_rate": 9.615467239527391e-06,
+      "loss": 0.0,
+      "step": 2090
+    },
+    {
+      "epoch": 0.1347554295289038,
+      "grad_norm": 0.0021930814514787705,
+      "learning_rate": 9.614751163623344e-06,
+      "loss": 0.0,
+      "step": 2091
+    },
+    {
+      "epoch": 0.13481987497583295,
+      "grad_norm": 0.010113932262379114,
+      "learning_rate": 9.614035087719298e-06,
+      "loss": 0.0,
+      "step": 2092
+    },
+    {
+      "epoch": 0.13488432042276213,
+      "grad_norm": 0.002889841846447938,
+      "learning_rate": 9.613319011815253e-06,
+      "loss": 0.0,
+      "step": 2093
+    },
+    {
+      "epoch": 0.13494876586969132,
+      "grad_norm": 0.10021186453941479,
+      "learning_rate": 9.612602935911207e-06,
+      "loss": 0.0017,
+      "step": 2094
+    },
+    {
+      "epoch": 0.13501321131662047,
+      "grad_norm": 0.1701542427445515,
+      "learning_rate": 9.611886860007161e-06,
+      "loss": 0.0005,
+      "step": 2095
+    },
+    {
+      "epoch": 0.13507765676354966,
+      "grad_norm": 0.2863434546765848,
+      "learning_rate": 9.611170784103116e-06,
+      "loss": 0.0003,
+      "step": 2096
+    },
+    {
+      "epoch": 0.13514210221047884,
+      "grad_norm": 0.030204838470596046,
+      "learning_rate": 9.61045470819907e-06,
+      "loss": 0.0001,
+      "step": 2097
+    },
+    {
+      "epoch": 0.135206547657408,
+      "grad_norm": 0.0006739822506401262,
+      "learning_rate": 9.609738632295024e-06,
+      "loss": 0.0,
+      "step": 2098
+    },
+    {
+      "epoch": 0.13527099310433718,
+      "grad_norm": 0.011925152450066019,
+      "learning_rate": 9.609022556390978e-06,
+      "loss": 0.0,
+      "step": 2099
+    },
+    {
+      "epoch": 0.13533543855126634,
+      "grad_norm": 0.000670276235356618,
+      "learning_rate": 9.608306480486933e-06,
+      "loss": 0.0,
+      "step": 2100
+    },
+    {
+      "epoch": 0.13539988399819552,
+      "grad_norm": 0.0018891649124682022,
+      "learning_rate": 9.607590404582885e-06,
+      "loss": 0.0,
+      "step": 2101
+    },
+    {
+      "epoch": 0.1354643294451247,
+      "grad_norm": 0.0008358051574337161,
+      "learning_rate": 9.606874328678841e-06,
+      "loss": 0.0,
+      "step": 2102
+    },
+    {
+      "epoch": 0.13552877489205387,
+      "grad_norm": 0.17796909637015565,
+      "learning_rate": 9.606158252774796e-06,
+      "loss": 0.0005,
+      "step": 2103
+    },
+    {
+      "epoch": 0.13559322033898305,
+      "grad_norm": 0.0029781535719557023,
+      "learning_rate": 9.60544217687075e-06,
+      "loss": 0.0,
+      "step": 2104
+    },
+    {
+      "epoch": 0.13565766578591223,
+      "grad_norm": 0.0007821579397496209,
+      "learning_rate": 9.604726100966704e-06,
+      "loss": 0.0,
+      "step": 2105
+    },
+    {
+      "epoch": 0.1357221112328414,
+      "grad_norm": 0.013895512731706612,
+      "learning_rate": 9.604010025062659e-06,
+      "loss": 0.0001,
+      "step": 2106
+    },
+    {
+      "epoch": 0.13578655667977058,
+      "grad_norm": 0.0006325714668676574,
+      "learning_rate": 9.603293949158611e-06,
+      "loss": 0.0,
+      "step": 2107
+    },
+    {
+      "epoch": 0.13585100212669976,
+      "grad_norm": 0.0036075466188459538,
+      "learning_rate": 9.602577873254565e-06,
+      "loss": 0.0,
+      "step": 2108
+    },
+    {
+      "epoch": 0.13591544757362892,
+      "grad_norm": 0.00022546715718579325,
+      "learning_rate": 9.60186179735052e-06,
+      "loss": 0.0,
+      "step": 2109
+    },
+    {
+      "epoch": 0.1359798930205581,
+      "grad_norm": 0.0019036321791445633,
+      "learning_rate": 9.601145721446474e-06,
+      "loss": 0.0,
+      "step": 2110
+    },
+    {
+      "epoch": 0.13604433846748729,
+      "grad_norm": 0.0013767631086754703,
+      "learning_rate": 9.600429645542428e-06,
+      "loss": 0.0,
+      "step": 2111
+    },
+    {
+      "epoch": 0.13610878391441644,
+      "grad_norm": 0.00033106270899061087,
+      "learning_rate": 9.599713569638383e-06,
+      "loss": 0.0,
+      "step": 2112
+    },
+    {
+      "epoch": 0.13617322936134563,
+      "grad_norm": 0.0004757141807601016,
+      "learning_rate": 9.598997493734337e-06,
+      "loss": 0.0,
+      "step": 2113
+    },
+    {
+      "epoch": 0.13623767480827478,
+      "grad_norm": 0.0012730227358612794,
+      "learning_rate": 9.598281417830291e-06,
+      "loss": 0.0,
+      "step": 2114
+    },
+    {
+      "epoch": 0.13630212025520397,
+      "grad_norm": 0.006321125741840267,
+      "learning_rate": 9.597565341926246e-06,
+      "loss": 0.0001,
+      "step": 2115
+    },
+    {
+      "epoch": 0.13636656570213315,
+      "grad_norm": 0.007981766124111247,
+      "learning_rate": 9.5968492660222e-06,
+      "loss": 0.0001,
+      "step": 2116
+    },
+    {
+      "epoch": 0.1364310111490623,
+      "grad_norm": 0.0013743153367953734,
+      "learning_rate": 9.596133190118152e-06,
+      "loss": 0.0,
+      "step": 2117
+    },
+    {
+      "epoch": 0.1364954565959915,
+      "grad_norm": 0.009479398745795132,
+      "learning_rate": 9.595417114214107e-06,
+      "loss": 0.0,
+      "step": 2118
+    },
+    {
+      "epoch": 0.13655990204292068,
+      "grad_norm": 0.0025998528844112315,
+      "learning_rate": 9.594701038310061e-06,
+      "loss": 0.0,
+      "step": 2119
+    },
+    {
+      "epoch": 0.13662434748984983,
+      "grad_norm": 0.0014551499795753477,
+      "learning_rate": 9.593984962406015e-06,
+      "loss": 0.0,
+      "step": 2120
+    },
+    {
+      "epoch": 0.13668879293677902,
+      "grad_norm": 0.00018595876394929454,
+      "learning_rate": 9.59326888650197e-06,
+      "loss": 0.0,
+      "step": 2121
+    },
+    {
+      "epoch": 0.1367532383837082,
+      "grad_norm": 0.16831879403294095,
+      "learning_rate": 9.592552810597924e-06,
+      "loss": 0.0001,
+      "step": 2122
+    },
+    {
+      "epoch": 0.13681768383063736,
+      "grad_norm": 0.006065252590868405,
+      "learning_rate": 9.591836734693878e-06,
+      "loss": 0.0001,
+      "step": 2123
+    },
+    {
+      "epoch": 0.13688212927756654,
+      "grad_norm": 0.0008141178296848007,
+      "learning_rate": 9.591120658789833e-06,
+      "loss": 0.0,
+      "step": 2124
+    },
+    {
+      "epoch": 0.1369465747244957,
+      "grad_norm": 0.010431545831266202,
+      "learning_rate": 9.590404582885787e-06,
+      "loss": 0.0,
+      "step": 2125
+    },
+    {
+      "epoch": 0.13701102017142489,
+      "grad_norm": 0.00022965953498677982,
+      "learning_rate": 9.589688506981741e-06,
+      "loss": 0.0,
+      "step": 2126
+    },
+    {
+      "epoch": 0.13707546561835407,
+      "grad_norm": 0.03739440462698081,
+      "learning_rate": 9.588972431077695e-06,
+      "loss": 0.0006,
+      "step": 2127
+    },
+    {
+      "epoch": 0.13713991106528323,
+      "grad_norm": 0.0012059992764552644,
+      "learning_rate": 9.58825635517365e-06,
+      "loss": 0.0,
+      "step": 2128
+    },
+    {
+      "epoch": 0.1372043565122124,
+      "grad_norm": 0.003077806228310975,
+      "learning_rate": 9.587540279269604e-06,
+      "loss": 0.0,
+      "step": 2129
+    },
+    {
+      "epoch": 0.1372688019591416,
+      "grad_norm": 0.0038935344462661513,
+      "learning_rate": 9.586824203365558e-06,
+      "loss": 0.0001,
+      "step": 2130
+    },
+    {
+      "epoch": 0.13733324740607075,
+      "grad_norm": 0.3498434423366078,
+      "learning_rate": 9.586108127461513e-06,
+      "loss": 0.0024,
+      "step": 2131
+    },
+    {
+      "epoch": 0.13739769285299994,
+      "grad_norm": 0.0030082327199185107,
+      "learning_rate": 9.585392051557467e-06,
+      "loss": 0.0,
+      "step": 2132
+    },
+    {
+      "epoch": 0.13746213829992912,
+      "grad_norm": 0.001989814893872896,
+      "learning_rate": 9.58467597565342e-06,
+      "loss": 0.0,
+      "step": 2133
+    },
+    {
+      "epoch": 0.13752658374685828,
+      "grad_norm": 0.2112891392170814,
+      "learning_rate": 9.583959899749374e-06,
+      "loss": 0.0004,
+      "step": 2134
+    },
+    {
+      "epoch": 0.13759102919378746,
+      "grad_norm": 0.007960537454691344,
+      "learning_rate": 9.583243823845328e-06,
+      "loss": 0.0001,
+      "step": 2135
+    },
+    {
+      "epoch": 0.13765547464071665,
+      "grad_norm": 0.14093893824348863,
+      "learning_rate": 9.582527747941282e-06,
+      "loss": 0.0002,
+      "step": 2136
+    },
+    {
+      "epoch": 0.1377199200876458,
+      "grad_norm": 0.926997838025005,
+      "learning_rate": 9.581811672037237e-06,
+      "loss": 0.0055,
+      "step": 2137
+    },
+    {
+      "epoch": 0.137784365534575,
+      "grad_norm": 0.0007762803125597636,
+      "learning_rate": 9.581095596133191e-06,
+      "loss": 0.0,
+      "step": 2138
+    },
+    {
+      "epoch": 0.13784881098150414,
+      "grad_norm": 0.011034256644735584,
+      "learning_rate": 9.580379520229145e-06,
+      "loss": 0.0001,
+      "step": 2139
+    },
+    {
+      "epoch": 0.13791325642843333,
+      "grad_norm": 0.005050776046657039,
+      "learning_rate": 9.5796634443251e-06,
+      "loss": 0.0,
+      "step": 2140
+    },
+    {
+      "epoch": 0.1379777018753625,
+      "grad_norm": 0.0058777191061129405,
+      "learning_rate": 9.578947368421054e-06,
+      "loss": 0.0,
+      "step": 2141
+    },
+    {
+      "epoch": 0.13804214732229167,
+      "grad_norm": 0.00010401014356011279,
+      "learning_rate": 9.578231292517007e-06,
+      "loss": 0.0,
+      "step": 2142
+    },
+    {
+      "epoch": 0.13810659276922085,
+      "grad_norm": 0.010355138189235565,
+      "learning_rate": 9.57751521661296e-06,
+      "loss": 0.0,
+      "step": 2143
+    },
+    {
+      "epoch": 0.13817103821615004,
+      "grad_norm": 0.0004001475512964918,
+      "learning_rate": 9.576799140708915e-06,
+      "loss": 0.0,
+      "step": 2144
+    },
+    {
+      "epoch": 0.1382354836630792,
+      "grad_norm": 0.004463203340507725,
+      "learning_rate": 9.57608306480487e-06,
+      "loss": 0.0,
+      "step": 2145
+    },
+    {
+      "epoch": 0.13829992911000838,
+      "grad_norm": 0.08375990333559151,
+      "learning_rate": 9.575366988900824e-06,
+      "loss": 0.0013,
+      "step": 2146
+    },
+    {
+      "epoch": 0.13836437455693756,
+      "grad_norm": 0.001991547368177983,
+      "learning_rate": 9.574650912996778e-06,
+      "loss": 0.0,
+      "step": 2147
+    },
+    {
+      "epoch": 0.13842882000386672,
+      "grad_norm": 0.00025540746774049394,
+      "learning_rate": 9.573934837092732e-06,
+      "loss": 0.0,
+      "step": 2148
+    },
+    {
+      "epoch": 0.1384932654507959,
+      "grad_norm": 0.00028334922210477735,
+      "learning_rate": 9.573218761188687e-06,
+      "loss": 0.0,
+      "step": 2149
+    },
+    {
+      "epoch": 0.1385577108977251,
+      "grad_norm": 0.0002980895207097618,
+      "learning_rate": 9.572502685284641e-06,
+      "loss": 0.0,
+      "step": 2150
+    },
+    {
+      "epoch": 0.13862215634465425,
+      "grad_norm": 0.06858295254675982,
+      "learning_rate": 9.571786609380595e-06,
+      "loss": 0.0007,
+      "step": 2151
+    },
+    {
+      "epoch": 0.13868660179158343,
+      "grad_norm": 0.0037622125428383424,
+      "learning_rate": 9.57107053347655e-06,
+      "loss": 0.0,
+      "step": 2152
+    },
+    {
+      "epoch": 0.1387510472385126,
+      "grad_norm": 0.002992232614199599,
+      "learning_rate": 9.570354457572504e-06,
+      "loss": 0.0,
+      "step": 2153
+    },
+    {
+      "epoch": 0.13881549268544177,
+      "grad_norm": 0.0015270018171054526,
+      "learning_rate": 9.569638381668458e-06,
+      "loss": 0.0,
+      "step": 2154
+    },
+    {
+      "epoch": 0.13887993813237096,
+      "grad_norm": 0.00048099787247835995,
+      "learning_rate": 9.568922305764412e-06,
+      "loss": 0.0,
+      "step": 2155
+    },
+    {
+      "epoch": 0.1389443835793001,
+      "grad_norm": 0.06242042743630073,
+      "learning_rate": 9.568206229860367e-06,
+      "loss": 0.0008,
+      "step": 2156
+    },
+    {
+      "epoch": 0.1390088290262293,
+      "grad_norm": 0.010198910788331253,
+      "learning_rate": 9.567490153956321e-06,
+      "loss": 0.0,
+      "step": 2157
+    },
+    {
+      "epoch": 0.13907327447315848,
+      "grad_norm": 0.02498183349411272,
+      "learning_rate": 9.566774078052274e-06,
+      "loss": 0.0002,
+      "step": 2158
+    },
+    {
+      "epoch": 0.13913771992008764,
+      "grad_norm": 0.02871299698962203,
+      "learning_rate": 9.566058002148228e-06,
+      "loss": 0.0002,
+      "step": 2159
+    },
+    {
+      "epoch": 0.13920216536701682,
+      "grad_norm": 0.0027904745167431897,
+      "learning_rate": 9.565341926244182e-06,
+      "loss": 0.0,
+      "step": 2160
+    },
+    {
+      "epoch": 0.139266610813946,
+      "grad_norm": 0.0017900588858772886,
+      "learning_rate": 9.564625850340137e-06,
+      "loss": 0.0,
+      "step": 2161
+    },
+    {
+      "epoch": 0.13933105626087516,
+      "grad_norm": 0.34400681092622376,
+      "learning_rate": 9.56390977443609e-06,
+      "loss": 0.0009,
+      "step": 2162
+    },
+    {
+      "epoch": 0.13939550170780435,
+      "grad_norm": 0.0004665528583749067,
+      "learning_rate": 9.563193698532045e-06,
+      "loss": 0.0,
+      "step": 2163
+    },
+    {
+      "epoch": 0.1394599471547335,
+      "grad_norm": 0.0022528881939270448,
+      "learning_rate": 9.562477622628e-06,
+      "loss": 0.0,
+      "step": 2164
+    },
+    {
+      "epoch": 0.1395243926016627,
+      "grad_norm": 0.0747928905148953,
+      "learning_rate": 9.561761546723954e-06,
+      "loss": 0.0008,
+      "step": 2165
+    },
+    {
+      "epoch": 0.13958883804859187,
+      "grad_norm": 0.15587924242670487,
+      "learning_rate": 9.561045470819908e-06,
+      "loss": 0.0015,
+      "step": 2166
+    },
+    {
+      "epoch": 0.13965328349552103,
+      "grad_norm": 0.010952113106743318,
+      "learning_rate": 9.560329394915862e-06,
+      "loss": 0.0001,
+      "step": 2167
+    },
+    {
+      "epoch": 0.13971772894245021,
+      "grad_norm": 0.0052290386293551245,
+      "learning_rate": 9.559613319011815e-06,
+      "loss": 0.0,
+      "step": 2168
+    },
+    {
+      "epoch": 0.1397821743893794,
+      "grad_norm": 0.0009271938531524405,
+      "learning_rate": 9.55889724310777e-06,
+      "loss": 0.0,
+      "step": 2169
+    },
+    {
+      "epoch": 0.13984661983630856,
+      "grad_norm": 0.0012394748834099222,
+      "learning_rate": 9.558181167203723e-06,
+      "loss": 0.0,
+      "step": 2170
+    },
+    {
+      "epoch": 0.13991106528323774,
+      "grad_norm": 0.058789488562754094,
+      "learning_rate": 9.557465091299678e-06,
+      "loss": 0.0001,
+      "step": 2171
+    },
+    {
+      "epoch": 0.13997551073016692,
+      "grad_norm": 0.0008058793089597991,
+      "learning_rate": 9.556749015395634e-06,
+      "loss": 0.0,
+      "step": 2172
+    },
+    {
+      "epoch": 0.14003995617709608,
+      "grad_norm": 0.010173480068610822,
+      "learning_rate": 9.556032939491588e-06,
+      "loss": 0.0001,
+      "step": 2173
+    },
+    {
+      "epoch": 0.14010440162402527,
+      "grad_norm": 0.015670921228439428,
+      "learning_rate": 9.55531686358754e-06,
+      "loss": 0.0001,
+      "step": 2174
+    },
+    {
+      "epoch": 0.14016884707095445,
+      "grad_norm": 0.03238827058306233,
+      "learning_rate": 9.554600787683495e-06,
+      "loss": 0.0001,
+      "step": 2175
+    },
+    {
+      "epoch": 0.1402332925178836,
+      "grad_norm": 0.001430104595176344,
+      "learning_rate": 9.55388471177945e-06,
+      "loss": 0.0,
+      "step": 2176
+    },
+    {
+      "epoch": 0.1402977379648128,
+      "grad_norm": 0.22314276162412025,
+      "learning_rate": 9.553168635875404e-06,
+      "loss": 0.0009,
+      "step": 2177
+    },
+    {
+      "epoch": 0.14036218341174195,
+      "grad_norm": 0.003981575309131401,
+      "learning_rate": 9.552452559971358e-06,
+      "loss": 0.0,
+      "step": 2178
+    },
+    {
+      "epoch": 0.14042662885867113,
+      "grad_norm": 0.022705913856466316,
+      "learning_rate": 9.551736484067312e-06,
+      "loss": 0.0,
+      "step": 2179
+    },
+    {
+      "epoch": 0.14049107430560032,
+      "grad_norm": 0.02675406863132256,
+      "learning_rate": 9.551020408163266e-06,
+      "loss": 0.0004,
+      "step": 2180
+    },
+    {
+      "epoch": 0.14055551975252947,
+      "grad_norm": 0.0002129580119729682,
+      "learning_rate": 9.55030433225922e-06,
+      "loss": 0.0,
+      "step": 2181
+    },
+    {
+      "epoch": 0.14061996519945866,
+      "grad_norm": 0.4309991929595454,
+      "learning_rate": 9.549588256355175e-06,
+      "loss": 0.001,
+      "step": 2182
+    },
+    {
+      "epoch": 0.14068441064638784,
+      "grad_norm": 0.027090977677346518,
+      "learning_rate": 9.54887218045113e-06,
+      "loss": 0.0002,
+      "step": 2183
+    },
+    {
+      "epoch": 0.140748856093317,
+      "grad_norm": 0.015662161214944636,
+      "learning_rate": 9.548156104547082e-06,
+      "loss": 0.0002,
+      "step": 2184
+    },
+    {
+      "epoch": 0.14081330154024618,
+      "grad_norm": 0.2672868382237579,
+      "learning_rate": 9.547440028643036e-06,
+      "loss": 0.0017,
+      "step": 2185
+    },
+    {
+      "epoch": 0.14087774698717537,
+      "grad_norm": 0.42117580116917186,
+      "learning_rate": 9.54672395273899e-06,
+      "loss": 0.0019,
+      "step": 2186
+    },
+    {
+      "epoch": 0.14094219243410452,
+      "grad_norm": 0.0010824486543928827,
+      "learning_rate": 9.546007876834945e-06,
+      "loss": 0.0,
+      "step": 2187
+    },
+    {
+      "epoch": 0.1410066378810337,
+      "grad_norm": 0.002333835033309739,
+      "learning_rate": 9.5452918009309e-06,
+      "loss": 0.0,
+      "step": 2188
+    },
+    {
+      "epoch": 0.1410710833279629,
+      "grad_norm": 0.01005892812360604,
+      "learning_rate": 9.544575725026853e-06,
+      "loss": 0.0001,
+      "step": 2189
+    },
+    {
+      "epoch": 0.14113552877489205,
+      "grad_norm": 0.04455364954326522,
+      "learning_rate": 9.543859649122808e-06,
+      "loss": 0.0001,
+      "step": 2190
+    },
+    {
+      "epoch": 0.14119997422182123,
+      "grad_norm": 0.011728766129579472,
+      "learning_rate": 9.543143573218762e-06,
+      "loss": 0.0,
+      "step": 2191
+    },
+    {
+      "epoch": 0.1412644196687504,
+      "grad_norm": 0.04750135049264757,
+      "learning_rate": 9.542427497314716e-06,
+      "loss": 0.0001,
+      "step": 2192
+    },
+    {
+      "epoch": 0.14132886511567957,
+      "grad_norm": 0.10700909026667592,
+      "learning_rate": 9.54171142141067e-06,
+      "loss": 0.0001,
+      "step": 2193
+    },
+    {
+      "epoch": 0.14139331056260876,
+      "grad_norm": 0.0018928932543482945,
+      "learning_rate": 9.540995345506623e-06,
+      "loss": 0.0,
+      "step": 2194
+    },
+    {
+      "epoch": 0.14145775600953792,
+      "grad_norm": 0.00035078335624548354,
+      "learning_rate": 9.540279269602578e-06,
+      "loss": 0.0,
+      "step": 2195
+    },
+    {
+      "epoch": 0.1415222014564671,
+      "grad_norm": 0.003913708593869081,
+      "learning_rate": 9.539563193698534e-06,
+      "loss": 0.0,
+      "step": 2196
+    },
+    {
+      "epoch": 0.14158664690339628,
+      "grad_norm": 0.32054424044004987,
+      "learning_rate": 9.538847117794488e-06,
+      "loss": 0.0025,
+      "step": 2197
+    },
+    {
+      "epoch": 0.14165109235032544,
+      "grad_norm": 0.00016171468966313804,
+      "learning_rate": 9.538131041890442e-06,
+      "loss": 0.0,
+      "step": 2198
+    },
+    {
+      "epoch": 0.14171553779725463,
+      "grad_norm": 0.03721486801308035,
+      "learning_rate": 9.537414965986396e-06,
+      "loss": 0.0003,
+      "step": 2199
+    },
+    {
+      "epoch": 0.1417799832441838,
+      "grad_norm": 0.009083953664856316,
+      "learning_rate": 9.536698890082349e-06,
+      "loss": 0.0,
+      "step": 2200
+    },
+    {
+      "epoch": 0.14184442869111297,
+      "grad_norm": 0.005521489335318659,
+      "learning_rate": 9.535982814178303e-06,
+      "loss": 0.0,
+      "step": 2201
+    },
+    {
+      "epoch": 0.14190887413804215,
+      "grad_norm": 0.0010786474428082949,
+      "learning_rate": 9.535266738274258e-06,
+      "loss": 0.0,
+      "step": 2202
+    },
+    {
+      "epoch": 0.1419733195849713,
+      "grad_norm": 0.0008330341741979345,
+      "learning_rate": 9.534550662370212e-06,
+      "loss": 0.0,
+      "step": 2203
+    },
+    {
+      "epoch": 0.1420377650319005,
+      "grad_norm": 0.004503439156467866,
+      "learning_rate": 9.533834586466166e-06,
+      "loss": 0.0,
+      "step": 2204
+    },
+    {
+      "epoch": 0.14210221047882968,
+      "grad_norm": 0.1447155653550225,
+      "learning_rate": 9.53311851056212e-06,
+      "loss": 0.0003,
+      "step": 2205
+    },
+    {
+      "epoch": 0.14216665592575883,
+      "grad_norm": 0.0006731134985055816,
+      "learning_rate": 9.532402434658075e-06,
+      "loss": 0.0,
+      "step": 2206
+    },
+    {
+      "epoch": 0.14223110137268802,
+      "grad_norm": 0.000276659143474887,
+      "learning_rate": 9.53168635875403e-06,
+      "loss": 0.0,
+      "step": 2207
+    },
+    {
+      "epoch": 0.1422955468196172,
+      "grad_norm": 0.0007131897573369064,
+      "learning_rate": 9.530970282849983e-06,
+      "loss": 0.0,
+      "step": 2208
+    },
+    {
+      "epoch": 0.14235999226654636,
+      "grad_norm": 0.007994417837624694,
+      "learning_rate": 9.530254206945938e-06,
+      "loss": 0.0,
+      "step": 2209
+    },
+    {
+      "epoch": 0.14242443771347554,
+      "grad_norm": 0.181136117707174,
+      "learning_rate": 9.52953813104189e-06,
+      "loss": 0.0005,
+      "step": 2210
+    },
+    {
+      "epoch": 0.14248888316040473,
+      "grad_norm": 0.15367599099740947,
+      "learning_rate": 9.528822055137845e-06,
+      "loss": 0.0005,
+      "step": 2211
+    },
+    {
+      "epoch": 0.14255332860733388,
+      "grad_norm": 0.15235054783448743,
+      "learning_rate": 9.528105979233799e-06,
+      "loss": 0.0007,
+      "step": 2212
+    },
+    {
+      "epoch": 0.14261777405426307,
+      "grad_norm": 0.004490466231885834,
+      "learning_rate": 9.527389903329753e-06,
+      "loss": 0.0,
+      "step": 2213
+    },
+    {
+      "epoch": 0.14268221950119225,
+      "grad_norm": 0.1511576771537659,
+      "learning_rate": 9.526673827425708e-06,
+      "loss": 0.0017,
+      "step": 2214
+    },
+    {
+      "epoch": 0.1427466649481214,
+      "grad_norm": 0.013676781391319589,
+      "learning_rate": 9.525957751521662e-06,
+      "loss": 0.0001,
+      "step": 2215
+    },
+    {
+      "epoch": 0.1428111103950506,
+      "grad_norm": 0.06545610037074183,
+      "learning_rate": 9.525241675617616e-06,
+      "loss": 0.0005,
+      "step": 2216
+    },
+    {
+      "epoch": 0.14287555584197975,
+      "grad_norm": 0.9775896697611458,
+      "learning_rate": 9.52452559971357e-06,
+      "loss": 0.0047,
+      "step": 2217
+    },
+    {
+      "epoch": 0.14294000128890894,
+      "grad_norm": 0.00031478760827874825,
+      "learning_rate": 9.523809523809525e-06,
+      "loss": 0.0,
+      "step": 2218
+    },
+    {
+      "epoch": 0.14300444673583812,
+      "grad_norm": 0.030597664652693826,
+      "learning_rate": 9.523093447905479e-06,
+      "loss": 0.0002,
+      "step": 2219
+    },
+    {
+      "epoch": 0.14306889218276728,
+      "grad_norm": 1.7293522593468222,
+      "learning_rate": 9.522377372001433e-06,
+      "loss": 0.0008,
+      "step": 2220
+    },
+    {
+      "epoch": 0.14313333762969646,
+      "grad_norm": 0.01601536033865269,
+      "learning_rate": 9.521661296097388e-06,
+      "loss": 0.0,
+      "step": 2221
+    },
+    {
+      "epoch": 0.14319778307662565,
+      "grad_norm": 0.0006310703572563955,
+      "learning_rate": 9.520945220193342e-06,
+      "loss": 0.0,
+      "step": 2222
+    },
+    {
+      "epoch": 0.1432622285235548,
+      "grad_norm": 0.0005210939538044766,
+      "learning_rate": 9.520229144289296e-06,
+      "loss": 0.0,
+      "step": 2223
+    },
+    {
+      "epoch": 0.143326673970484,
+      "grad_norm": 0.018941119371520383,
+      "learning_rate": 9.51951306838525e-06,
+      "loss": 0.0002,
+      "step": 2224
+    },
+    {
+      "epoch": 0.14339111941741317,
+      "grad_norm": 0.05937819495828138,
+      "learning_rate": 9.518796992481205e-06,
+      "loss": 0.0001,
+      "step": 2225
+    },
+    {
+      "epoch": 0.14345556486434233,
+      "grad_norm": 0.018362872540066737,
+      "learning_rate": 9.518080916577157e-06,
+      "loss": 0.0001,
+      "step": 2226
+    },
+    {
+      "epoch": 0.1435200103112715,
+      "grad_norm": 0.01831749307485675,
+      "learning_rate": 9.517364840673112e-06,
+      "loss": 0.0001,
+      "step": 2227
+    },
+    {
+      "epoch": 0.1435844557582007,
+      "grad_norm": 0.006862687030695175,
+      "learning_rate": 9.516648764769066e-06,
+      "loss": 0.0,
+      "step": 2228
+    },
+    {
+      "epoch": 0.14364890120512985,
+      "grad_norm": 7.471194955517663e-05,
+      "learning_rate": 9.51593268886502e-06,
+      "loss": 0.0,
+      "step": 2229
+    },
+    {
+      "epoch": 0.14371334665205904,
+      "grad_norm": 0.003645056892950574,
+      "learning_rate": 9.515216612960975e-06,
+      "loss": 0.0,
+      "step": 2230
+    },
+    {
+      "epoch": 0.1437777920989882,
+      "grad_norm": 0.016415402507622183,
+      "learning_rate": 9.514500537056929e-06,
+      "loss": 0.0001,
+      "step": 2231
+    },
+    {
+      "epoch": 0.14384223754591738,
+      "grad_norm": 0.0016457805006108836,
+      "learning_rate": 9.513784461152883e-06,
+      "loss": 0.0,
+      "step": 2232
+    },
+    {
+      "epoch": 0.14390668299284656,
+      "grad_norm": 0.026616648264700046,
+      "learning_rate": 9.513068385248838e-06,
+      "loss": 0.0001,
+      "step": 2233
+    },
+    {
+      "epoch": 0.14397112843977572,
+      "grad_norm": 0.5394570109498802,
+      "learning_rate": 9.512352309344792e-06,
+      "loss": 0.0036,
+      "step": 2234
+    },
+    {
+      "epoch": 0.1440355738867049,
+      "grad_norm": 0.006396931689943378,
+      "learning_rate": 9.511636233440744e-06,
+      "loss": 0.0,
+      "step": 2235
+    },
+    {
+      "epoch": 0.1441000193336341,
+      "grad_norm": 0.012580438401703474,
+      "learning_rate": 9.510920157536699e-06,
+      "loss": 0.0,
+      "step": 2236
+    },
+    {
+      "epoch": 0.14416446478056325,
+      "grad_norm": 0.01920853147496425,
+      "learning_rate": 9.510204081632653e-06,
+      "loss": 0.0001,
+      "step": 2237
+    },
+    {
+      "epoch": 0.14422891022749243,
+      "grad_norm": 0.002245388255484627,
+      "learning_rate": 9.509488005728607e-06,
+      "loss": 0.0,
+      "step": 2238
+    },
+    {
+      "epoch": 0.14429335567442161,
+      "grad_norm": 0.579263982784339,
+      "learning_rate": 9.508771929824562e-06,
+      "loss": 0.0168,
+      "step": 2239
+    },
+    {
+      "epoch": 0.14435780112135077,
+      "grad_norm": 0.00028066203840411235,
+      "learning_rate": 9.508055853920516e-06,
+      "loss": 0.0,
+      "step": 2240
+    },
+    {
+      "epoch": 0.14442224656827995,
+      "grad_norm": 0.021214743917220127,
+      "learning_rate": 9.50733977801647e-06,
+      "loss": 0.0001,
+      "step": 2241
+    },
+    {
+      "epoch": 0.1444866920152091,
+      "grad_norm": 0.00032764501369088723,
+      "learning_rate": 9.506623702112425e-06,
+      "loss": 0.0,
+      "step": 2242
+    },
+    {
+      "epoch": 0.1445511374621383,
+      "grad_norm": 0.0058459261688533605,
+      "learning_rate": 9.505907626208379e-06,
+      "loss": 0.0,
+      "step": 2243
+    },
+    {
+      "epoch": 0.14461558290906748,
+      "grad_norm": 0.051741941735831465,
+      "learning_rate": 9.505191550304333e-06,
+      "loss": 0.0,
+      "step": 2244
+    },
+    {
+      "epoch": 0.14468002835599664,
+      "grad_norm": 0.32622149655365945,
+      "learning_rate": 9.504475474400287e-06,
+      "loss": 0.0003,
+      "step": 2245
+    },
+    {
+      "epoch": 0.14474447380292582,
+      "grad_norm": 0.055378686377618995,
+      "learning_rate": 9.503759398496242e-06,
+      "loss": 0.0001,
+      "step": 2246
+    },
+    {
+      "epoch": 0.144808919249855,
+      "grad_norm": 0.02396991245736209,
+      "learning_rate": 9.503043322592196e-06,
+      "loss": 0.0001,
+      "step": 2247
+    },
+    {
+      "epoch": 0.14487336469678416,
+      "grad_norm": 0.03345312463997792,
+      "learning_rate": 9.50232724668815e-06,
+      "loss": 0.0015,
+      "step": 2248
+    },
+    {
+      "epoch": 0.14493781014371335,
+      "grad_norm": 0.15419614497691245,
+      "learning_rate": 9.501611170784105e-06,
+      "loss": 0.002,
+      "step": 2249
+    },
+    {
+      "epoch": 0.14500225559064253,
+      "grad_norm": 0.25495189154503606,
+      "learning_rate": 9.500895094880059e-06,
+      "loss": 0.001,
+      "step": 2250
+    },
+    {
+      "epoch": 0.1450667010375717,
+      "grad_norm": 0.06784105322185756,
+      "learning_rate": 9.500179018976012e-06,
+      "loss": 0.0004,
+      "step": 2251
+    },
+    {
+      "epoch": 0.14513114648450087,
+      "grad_norm": 0.0008401214103608888,
+      "learning_rate": 9.499462943071966e-06,
+      "loss": 0.0,
+      "step": 2252
+    },
+    {
+      "epoch": 0.14519559193143006,
+      "grad_norm": 0.7134616185045286,
+      "learning_rate": 9.49874686716792e-06,
+      "loss": 0.0016,
+      "step": 2253
+    },
+    {
+      "epoch": 0.1452600373783592,
+      "grad_norm": 0.005298159616167539,
+      "learning_rate": 9.498030791263874e-06,
+      "loss": 0.0,
+      "step": 2254
+    },
+    {
+      "epoch": 0.1453244828252884,
+      "grad_norm": 0.24674281553149796,
+      "learning_rate": 9.497314715359829e-06,
+      "loss": 0.0005,
+      "step": 2255
+    },
+    {
+      "epoch": 0.14538892827221755,
+      "grad_norm": 0.015423563114479719,
+      "learning_rate": 9.496598639455783e-06,
+      "loss": 0.0001,
+      "step": 2256
+    },
+    {
+      "epoch": 0.14545337371914674,
+      "grad_norm": 0.014220395291028429,
+      "learning_rate": 9.495882563551737e-06,
+      "loss": 0.0001,
+      "step": 2257
+    },
+    {
+      "epoch": 0.14551781916607592,
+      "grad_norm": 0.0020908002801696174,
+      "learning_rate": 9.495166487647692e-06,
+      "loss": 0.0,
+      "step": 2258
+    },
+    {
+      "epoch": 0.14558226461300508,
+      "grad_norm": 0.004630563261824797,
+      "learning_rate": 9.494450411743646e-06,
+      "loss": 0.0,
+      "step": 2259
+    },
+    {
+      "epoch": 0.14564671005993426,
+      "grad_norm": 0.0005906543258460047,
+      "learning_rate": 9.4937343358396e-06,
+      "loss": 0.0,
+      "step": 2260
+    },
+    {
+      "epoch": 0.14571115550686345,
+      "grad_norm": 0.00957402495164633,
+      "learning_rate": 9.493018259935553e-06,
+      "loss": 0.0,
+      "step": 2261
+    },
+    {
+      "epoch": 0.1457756009537926,
+      "grad_norm": 0.014087719869458042,
+      "learning_rate": 9.492302184031507e-06,
+      "loss": 0.0,
+      "step": 2262
+    },
+    {
+      "epoch": 0.1458400464007218,
+      "grad_norm": 0.0029143890782602393,
+      "learning_rate": 9.491586108127461e-06,
+      "loss": 0.0,
+      "step": 2263
+    },
+    {
+      "epoch": 0.14590449184765097,
+      "grad_norm": 0.03503983481883921,
+      "learning_rate": 9.490870032223416e-06,
+      "loss": 0.0004,
+      "step": 2264
+    },
+    {
+      "epoch": 0.14596893729458013,
+      "grad_norm": 0.013767286296889813,
+      "learning_rate": 9.49015395631937e-06,
+      "loss": 0.0,
+      "step": 2265
+    },
+    {
+      "epoch": 0.14603338274150932,
+      "grad_norm": 0.25470119158679877,
+      "learning_rate": 9.489437880415326e-06,
+      "loss": 0.0005,
+      "step": 2266
+    },
+    {
+      "epoch": 0.1460978281884385,
+      "grad_norm": 0.007543495843438581,
+      "learning_rate": 9.488721804511279e-06,
+      "loss": 0.0,
+      "step": 2267
+    },
+    {
+      "epoch": 0.14616227363536766,
+      "grad_norm": 0.0020839204456079203,
+      "learning_rate": 9.488005728607233e-06,
+      "loss": 0.0,
+      "step": 2268
+    },
+    {
+      "epoch": 0.14622671908229684,
+      "grad_norm": 0.06732914003313965,
+      "learning_rate": 9.487289652703187e-06,
+      "loss": 0.0003,
+      "step": 2269
+    },
+    {
+      "epoch": 0.146291164529226,
+      "grad_norm": 0.0016212466885061447,
+      "learning_rate": 9.486573576799142e-06,
+      "loss": 0.0,
+      "step": 2270
+    },
+    {
+      "epoch": 0.14635560997615518,
+      "grad_norm": 0.3236607034805371,
+      "learning_rate": 9.485857500895096e-06,
+      "loss": 0.0045,
+      "step": 2271
+    },
+    {
+      "epoch": 0.14642005542308437,
+      "grad_norm": 0.004995574742021267,
+      "learning_rate": 9.48514142499105e-06,
+      "loss": 0.0,
+      "step": 2272
+    },
+    {
+      "epoch": 0.14648450087001352,
+      "grad_norm": 0.00117536297122479,
+      "learning_rate": 9.484425349087004e-06,
+      "loss": 0.0,
+      "step": 2273
+    },
+    {
+      "epoch": 0.1465489463169427,
+      "grad_norm": 0.0005086225519439412,
+      "learning_rate": 9.483709273182959e-06,
+      "loss": 0.0,
+      "step": 2274
+    },
+    {
+      "epoch": 0.1466133917638719,
+      "grad_norm": 0.0030946496454177484,
+      "learning_rate": 9.482993197278913e-06,
+      "loss": 0.0,
+      "step": 2275
+    },
+    {
+      "epoch": 0.14667783721080105,
+      "grad_norm": 0.0010062472442629099,
+      "learning_rate": 9.482277121374867e-06,
+      "loss": 0.0,
+      "step": 2276
+    },
+    {
+      "epoch": 0.14674228265773023,
+      "grad_norm": 0.03180384283377841,
+      "learning_rate": 9.48156104547082e-06,
+      "loss": 0.0002,
+      "step": 2277
+    },
+    {
+      "epoch": 0.14680672810465942,
+      "grad_norm": 0.21418460822901342,
+      "learning_rate": 9.480844969566774e-06,
+      "loss": 0.0015,
+      "step": 2278
+    },
+    {
+      "epoch": 0.14687117355158857,
+      "grad_norm": 0.010110387519656235,
+      "learning_rate": 9.480128893662729e-06,
+      "loss": 0.0,
+      "step": 2279
+    },
+    {
+      "epoch": 0.14693561899851776,
+      "grad_norm": 0.0013718162505616696,
+      "learning_rate": 9.479412817758683e-06,
+      "loss": 0.0,
+      "step": 2280
+    },
+    {
+      "epoch": 0.14700006444544694,
+      "grad_norm": 0.001147183938640018,
+      "learning_rate": 9.478696741854637e-06,
+      "loss": 0.0,
+      "step": 2281
+    },
+    {
+      "epoch": 0.1470645098923761,
+      "grad_norm": 0.0038179217062423933,
+      "learning_rate": 9.477980665950591e-06,
+      "loss": 0.0,
+      "step": 2282
+    },
+    {
+      "epoch": 0.14712895533930528,
+      "grad_norm": 0.16823613707940346,
+      "learning_rate": 9.477264590046546e-06,
+      "loss": 0.0001,
+      "step": 2283
+    },
+    {
+      "epoch": 0.14719340078623444,
+      "grad_norm": 0.0007041555804207547,
+      "learning_rate": 9.4765485141425e-06,
+      "loss": 0.0,
+      "step": 2284
+    },
+    {
+      "epoch": 0.14725784623316363,
+      "grad_norm": 0.17469604142040665,
+      "learning_rate": 9.475832438238454e-06,
+      "loss": 0.0006,
+      "step": 2285
+    },
+    {
+      "epoch": 0.1473222916800928,
+      "grad_norm": 0.07642142365060478,
+      "learning_rate": 9.475116362334409e-06,
+      "loss": 0.0002,
+      "step": 2286
+    },
+    {
+      "epoch": 0.14738673712702197,
+      "grad_norm": 0.026083890142809724,
+      "learning_rate": 9.474400286430361e-06,
+      "loss": 0.0003,
+      "step": 2287
+    },
+    {
+      "epoch": 0.14745118257395115,
+      "grad_norm": 0.0013347215920775026,
+      "learning_rate": 9.473684210526315e-06,
+      "loss": 0.0,
+      "step": 2288
+    },
+    {
+      "epoch": 0.14751562802088033,
+      "grad_norm": 0.0656069517862333,
+      "learning_rate": 9.472968134622271e-06,
+      "loss": 0.0001,
+      "step": 2289
+    },
+    {
+      "epoch": 0.1475800734678095,
+      "grad_norm": 0.0008470576602853886,
+      "learning_rate": 9.472252058718226e-06,
+      "loss": 0.0,
+      "step": 2290
+    },
+    {
+      "epoch": 0.14764451891473868,
+      "grad_norm": 0.16749845691760115,
+      "learning_rate": 9.47153598281418e-06,
+      "loss": 0.003,
+      "step": 2291
+    },
+    {
+      "epoch": 0.14770896436166786,
+      "grad_norm": 0.0016121071686281076,
+      "learning_rate": 9.470819906910134e-06,
+      "loss": 0.0,
+      "step": 2292
+    },
+    {
+      "epoch": 0.14777340980859702,
+      "grad_norm": 0.006441744083881955,
+      "learning_rate": 9.470103831006087e-06,
+      "loss": 0.0,
+      "step": 2293
+    },
+    {
+      "epoch": 0.1478378552555262,
+      "grad_norm": 0.016728643202810173,
+      "learning_rate": 9.469387755102041e-06,
+      "loss": 0.0001,
+      "step": 2294
+    },
+    {
+      "epoch": 0.14790230070245536,
+      "grad_norm": 0.005027691402007882,
+      "learning_rate": 9.468671679197996e-06,
+      "loss": 0.0,
+      "step": 2295
+    },
+    {
+      "epoch": 0.14796674614938454,
+      "grad_norm": 0.001770350858388834,
+      "learning_rate": 9.46795560329395e-06,
+      "loss": 0.0,
+      "step": 2296
+    },
+    {
+      "epoch": 0.14803119159631373,
+      "grad_norm": 0.00011038801131939273,
+      "learning_rate": 9.467239527389904e-06,
+      "loss": 0.0,
+      "step": 2297
+    },
+    {
+      "epoch": 0.14809563704324288,
+      "grad_norm": 0.30113663671343743,
+      "learning_rate": 9.466523451485858e-06,
+      "loss": 0.0011,
+      "step": 2298
+    },
+    {
+      "epoch": 0.14816008249017207,
+      "grad_norm": 0.0003501895384979152,
+      "learning_rate": 9.465807375581813e-06,
+      "loss": 0.0,
+      "step": 2299
+    },
+    {
+      "epoch": 0.14822452793710125,
+      "grad_norm": 0.001076580935332914,
+      "learning_rate": 9.465091299677767e-06,
+      "loss": 0.0,
+      "step": 2300
+    },
+    {
+      "epoch": 0.1482889733840304,
+      "grad_norm": 0.880960361418177,
+      "learning_rate": 9.464375223773721e-06,
+      "loss": 0.0051,
+      "step": 2301
+    },
+    {
+      "epoch": 0.1483534188309596,
+      "grad_norm": 0.03526416745467504,
+      "learning_rate": 9.463659147869676e-06,
+      "loss": 0.0002,
+      "step": 2302
+    },
+    {
+      "epoch": 0.14841786427788878,
+      "grad_norm": 0.0025750307660047872,
+      "learning_rate": 9.462943071965628e-06,
+      "loss": 0.0,
+      "step": 2303
+    },
+    {
+      "epoch": 0.14848230972481793,
+      "grad_norm": 0.004023243719837641,
+      "learning_rate": 9.462226996061583e-06,
+      "loss": 0.0,
+      "step": 2304
+    },
+    {
+      "epoch": 0.14854675517174712,
+      "grad_norm": 0.9786404284533876,
+      "learning_rate": 9.461510920157537e-06,
+      "loss": 0.0027,
+      "step": 2305
+    },
+    {
+      "epoch": 0.1486112006186763,
+      "grad_norm": 0.13937884764855718,
+      "learning_rate": 9.460794844253491e-06,
+      "loss": 0.0018,
+      "step": 2306
+    },
+    {
+      "epoch": 0.14867564606560546,
+      "grad_norm": 0.004105793329801603,
+      "learning_rate": 9.460078768349445e-06,
+      "loss": 0.0,
+      "step": 2307
+    },
+    {
+      "epoch": 0.14874009151253464,
+      "grad_norm": 0.0005642601189968466,
+      "learning_rate": 9.4593626924454e-06,
+      "loss": 0.0,
+      "step": 2308
+    },
+    {
+      "epoch": 0.1488045369594638,
+      "grad_norm": 0.041690372482371,
+      "learning_rate": 9.458646616541354e-06,
+      "loss": 0.0001,
+      "step": 2309
+    },
+    {
+      "epoch": 0.14886898240639299,
+      "grad_norm": 0.0004829795518174653,
+      "learning_rate": 9.457930540637308e-06,
+      "loss": 0.0,
+      "step": 2310
+    },
+    {
+      "epoch": 0.14893342785332217,
+      "grad_norm": 0.0004682333152799085,
+      "learning_rate": 9.457214464733263e-06,
+      "loss": 0.0,
+      "step": 2311
+    },
+    {
+      "epoch": 0.14899787330025133,
+      "grad_norm": 0.016004459520340634,
+      "learning_rate": 9.456498388829217e-06,
+      "loss": 0.0002,
+      "step": 2312
+    },
+    {
+      "epoch": 0.1490623187471805,
+      "grad_norm": 0.6353641212082908,
+      "learning_rate": 9.455782312925171e-06,
+      "loss": 0.0041,
+      "step": 2313
+    },
+    {
+      "epoch": 0.1491267641941097,
+      "grad_norm": 0.0006008079600321058,
+      "learning_rate": 9.455066237021126e-06,
+      "loss": 0.0,
+      "step": 2314
+    },
+    {
+      "epoch": 0.14919120964103885,
+      "grad_norm": 0.002256655415846708,
+      "learning_rate": 9.45435016111708e-06,
+      "loss": 0.0,
+      "step": 2315
+    },
+    {
+      "epoch": 0.14925565508796804,
+      "grad_norm": 0.00014758511457507473,
+      "learning_rate": 9.453634085213034e-06,
+      "loss": 0.0,
+      "step": 2316
+    },
+    {
+      "epoch": 0.14932010053489722,
+      "grad_norm": 0.006368467140729355,
+      "learning_rate": 9.452918009308988e-06,
+      "loss": 0.0,
+      "step": 2317
+    },
+    {
+      "epoch": 0.14938454598182638,
+      "grad_norm": 0.0032855975036902166,
+      "learning_rate": 9.452201933404943e-06,
+      "loss": 0.0,
+      "step": 2318
+    },
+    {
+      "epoch": 0.14944899142875556,
+      "grad_norm": 0.6131575149796049,
+      "learning_rate": 9.451485857500895e-06,
+      "loss": 0.003,
+      "step": 2319
+    },
+    {
+      "epoch": 0.14951343687568475,
+      "grad_norm": 0.275169933440258,
+      "learning_rate": 9.45076978159685e-06,
+      "loss": 0.0007,
+      "step": 2320
+    },
+    {
+      "epoch": 0.1495778823226139,
+      "grad_norm": 0.0023692457357848964,
+      "learning_rate": 9.450053705692804e-06,
+      "loss": 0.0,
+      "step": 2321
+    },
+    {
+      "epoch": 0.1496423277695431,
+      "grad_norm": 0.0025468429265798303,
+      "learning_rate": 9.449337629788758e-06,
+      "loss": 0.0,
+      "step": 2322
+    },
+    {
+      "epoch": 0.14970677321647224,
+      "grad_norm": 0.6888988525450286,
+      "learning_rate": 9.448621553884713e-06,
+      "loss": 0.0026,
+      "step": 2323
+    },
+    {
+      "epoch": 0.14977121866340143,
+      "grad_norm": 0.000541195102515391,
+      "learning_rate": 9.447905477980667e-06,
+      "loss": 0.0,
+      "step": 2324
+    },
+    {
+      "epoch": 0.1498356641103306,
+      "grad_norm": 0.04337376750890723,
+      "learning_rate": 9.447189402076621e-06,
+      "loss": 0.0001,
+      "step": 2325
+    },
+    {
+      "epoch": 0.14990010955725977,
+      "grad_norm": 0.05995612202794425,
+      "learning_rate": 9.446473326172575e-06,
+      "loss": 0.0002,
+      "step": 2326
+    },
+    {
+      "epoch": 0.14996455500418895,
+      "grad_norm": 0.014729528683343611,
+      "learning_rate": 9.44575725026853e-06,
+      "loss": 0.0001,
+      "step": 2327
+    },
+    {
+      "epoch": 0.15002900045111814,
+      "grad_norm": 0.015224141839539764,
+      "learning_rate": 9.445041174364482e-06,
+      "loss": 0.0,
+      "step": 2328
+    },
+    {
+      "epoch": 0.1500934458980473,
+      "grad_norm": 0.0020922033316262286,
+      "learning_rate": 9.444325098460437e-06,
+      "loss": 0.0,
+      "step": 2329
+    },
+    {
+      "epoch": 0.15015789134497648,
+      "grad_norm": 0.00976562052965062,
+      "learning_rate": 9.443609022556391e-06,
+      "loss": 0.0,
+      "step": 2330
+    },
+    {
+      "epoch": 0.15022233679190566,
+      "grad_norm": 0.003141049531346141,
+      "learning_rate": 9.442892946652345e-06,
+      "loss": 0.0,
+      "step": 2331
+    },
+    {
+      "epoch": 0.15028678223883482,
+      "grad_norm": 0.2375710487754487,
+      "learning_rate": 9.4421768707483e-06,
+      "loss": 0.0007,
+      "step": 2332
+    },
+    {
+      "epoch": 0.150351227685764,
+      "grad_norm": 0.10324509375902188,
+      "learning_rate": 9.441460794844254e-06,
+      "loss": 0.0003,
+      "step": 2333
+    },
+    {
+      "epoch": 0.15041567313269316,
+      "grad_norm": 0.011364228373043003,
+      "learning_rate": 9.440744718940208e-06,
+      "loss": 0.0,
+      "step": 2334
+    },
+    {
+      "epoch": 0.15048011857962235,
+      "grad_norm": 0.033959039268354946,
+      "learning_rate": 9.440028643036162e-06,
+      "loss": 0.0001,
+      "step": 2335
+    },
+    {
+      "epoch": 0.15054456402655153,
+      "grad_norm": 0.0010414789426565346,
+      "learning_rate": 9.439312567132117e-06,
+      "loss": 0.0,
+      "step": 2336
+    },
+    {
+      "epoch": 0.1506090094734807,
+      "grad_norm": 0.011555880377593744,
+      "learning_rate": 9.438596491228071e-06,
+      "loss": 0.0001,
+      "step": 2337
+    },
+    {
+      "epoch": 0.15067345492040987,
+      "grad_norm": 0.13693876274950872,
+      "learning_rate": 9.437880415324025e-06,
+      "loss": 0.0005,
+      "step": 2338
+    },
+    {
+      "epoch": 0.15073790036733906,
+      "grad_norm": 0.046467830614872876,
+      "learning_rate": 9.43716433941998e-06,
+      "loss": 0.0002,
+      "step": 2339
+    },
+    {
+      "epoch": 0.1508023458142682,
+      "grad_norm": 0.28248522299966256,
+      "learning_rate": 9.436448263515934e-06,
+      "loss": 0.0005,
+      "step": 2340
+    },
+    {
+      "epoch": 0.1508667912611974,
+      "grad_norm": 0.010665106642022313,
+      "learning_rate": 9.435732187611888e-06,
+      "loss": 0.0001,
+      "step": 2341
+    },
+    {
+      "epoch": 0.15093123670812658,
+      "grad_norm": 0.01410729256099583,
+      "learning_rate": 9.435016111707843e-06,
+      "loss": 0.0001,
+      "step": 2342
+    },
+    {
+      "epoch": 0.15099568215505574,
+      "grad_norm": 0.006377867812248754,
+      "learning_rate": 9.434300035803797e-06,
+      "loss": 0.0,
+      "step": 2343
+    },
+    {
+      "epoch": 0.15106012760198492,
+      "grad_norm": 0.3001965872068028,
+      "learning_rate": 9.43358395989975e-06,
+      "loss": 0.0005,
+      "step": 2344
+    },
+    {
+      "epoch": 0.1511245730489141,
+      "grad_norm": 0.0054143970512442385,
+      "learning_rate": 9.432867883995704e-06,
+      "loss": 0.0,
+      "step": 2345
+    },
+    {
+      "epoch": 0.15118901849584326,
+      "grad_norm": 0.003879576657613195,
+      "learning_rate": 9.432151808091658e-06,
+      "loss": 0.0,
+      "step": 2346
+    },
+    {
+      "epoch": 0.15125346394277245,
+      "grad_norm": 0.017826363177415604,
+      "learning_rate": 9.431435732187612e-06,
+      "loss": 0.0002,
+      "step": 2347
+    },
+    {
+      "epoch": 0.1513179093897016,
+      "grad_norm": 0.14413647606933994,
+      "learning_rate": 9.430719656283567e-06,
+      "loss": 0.0004,
+      "step": 2348
+    },
+    {
+      "epoch": 0.1513823548366308,
+      "grad_norm": 0.0006606705735611096,
+      "learning_rate": 9.430003580379521e-06,
+      "loss": 0.0,
+      "step": 2349
+    },
+    {
+      "epoch": 0.15144680028355997,
+      "grad_norm": 0.001281065290532309,
+      "learning_rate": 9.429287504475475e-06,
+      "loss": 0.0,
+      "step": 2350
+    },
+    {
+      "epoch": 0.15151124573048913,
+      "grad_norm": 0.002033793603108425,
+      "learning_rate": 9.42857142857143e-06,
+      "loss": 0.0,
+      "step": 2351
+    },
+    {
+      "epoch": 0.15157569117741831,
+      "grad_norm": 0.006509579136491605,
+      "learning_rate": 9.427855352667384e-06,
+      "loss": 0.0,
+      "step": 2352
+    },
+    {
+      "epoch": 0.1516401366243475,
+      "grad_norm": 0.015385489864271826,
+      "learning_rate": 9.427139276763338e-06,
+      "loss": 0.0001,
+      "step": 2353
+    },
+    {
+      "epoch": 0.15170458207127666,
+      "grad_norm": 0.00034424994771698977,
+      "learning_rate": 9.42642320085929e-06,
+      "loss": 0.0,
+      "step": 2354
+    },
+    {
+      "epoch": 0.15176902751820584,
+      "grad_norm": 0.0008804023377011956,
+      "learning_rate": 9.425707124955245e-06,
+      "loss": 0.0,
+      "step": 2355
+    },
+    {
+      "epoch": 0.15183347296513502,
+      "grad_norm": 0.547700177660738,
+      "learning_rate": 9.4249910490512e-06,
+      "loss": 0.0022,
+      "step": 2356
+    },
+    {
+      "epoch": 0.15189791841206418,
+      "grad_norm": 0.0032957801919053283,
+      "learning_rate": 9.424274973147154e-06,
+      "loss": 0.0,
+      "step": 2357
+    },
+    {
+      "epoch": 0.15196236385899337,
+      "grad_norm": 0.0027416642189764164,
+      "learning_rate": 9.423558897243108e-06,
+      "loss": 0.0,
+      "step": 2358
+    },
+    {
+      "epoch": 0.15202680930592255,
+      "grad_norm": 0.001389136393539497,
+      "learning_rate": 9.422842821339064e-06,
+      "loss": 0.0,
+      "step": 2359
+    },
+    {
+      "epoch": 0.1520912547528517,
+      "grad_norm": 0.0016453047623652915,
+      "learning_rate": 9.422126745435017e-06,
+      "loss": 0.0,
+      "step": 2360
+    },
+    {
+      "epoch": 0.1521557001997809,
+      "grad_norm": 0.003564651534562986,
+      "learning_rate": 9.42141066953097e-06,
+      "loss": 0.0,
+      "step": 2361
+    },
+    {
+      "epoch": 0.15222014564671005,
+      "grad_norm": 0.001581220919916646,
+      "learning_rate": 9.420694593626925e-06,
+      "loss": 0.0,
+      "step": 2362
+    },
+    {
+      "epoch": 0.15228459109363923,
+      "grad_norm": 1.166109780962107,
+      "learning_rate": 9.41997851772288e-06,
+      "loss": 0.0068,
+      "step": 2363
+    },
+    {
+      "epoch": 0.15234903654056842,
+      "grad_norm": 0.001095153365991885,
+      "learning_rate": 9.419262441818834e-06,
+      "loss": 0.0,
+      "step": 2364
+    },
+    {
+      "epoch": 0.15241348198749757,
+      "grad_norm": 0.00018614029556956357,
+      "learning_rate": 9.418546365914788e-06,
+      "loss": 0.0,
+      "step": 2365
+    },
+    {
+      "epoch": 0.15247792743442676,
+      "grad_norm": 0.1916720432757476,
+      "learning_rate": 9.417830290010742e-06,
+      "loss": 0.0003,
+      "step": 2366
+    },
+    {
+      "epoch": 0.15254237288135594,
+      "grad_norm": 0.17782862752708178,
+      "learning_rate": 9.417114214106697e-06,
+      "loss": 0.0004,
+      "step": 2367
+    },
+    {
+      "epoch": 0.1526068183282851,
+      "grad_norm": 0.01575983901171653,
+      "learning_rate": 9.416398138202651e-06,
+      "loss": 0.0,
+      "step": 2368
+    },
+    {
+      "epoch": 0.15267126377521428,
+      "grad_norm": 0.0023644075189573855,
+      "learning_rate": 9.415682062298605e-06,
+      "loss": 0.0,
+      "step": 2369
+    },
+    {
+      "epoch": 0.15273570922214347,
+      "grad_norm": 0.03693351516886743,
+      "learning_rate": 9.414965986394558e-06,
+      "loss": 0.0001,
+      "step": 2370
+    },
+    {
+      "epoch": 0.15280015466907262,
+      "grad_norm": 0.0012187265154810075,
+      "learning_rate": 9.414249910490512e-06,
+      "loss": 0.0,
+      "step": 2371
+    },
+    {
+      "epoch": 0.1528646001160018,
+      "grad_norm": 0.024398932537146707,
+      "learning_rate": 9.413533834586466e-06,
+      "loss": 0.0001,
+      "step": 2372
+    },
+    {
+      "epoch": 0.15292904556293097,
+      "grad_norm": 0.038167990425758816,
+      "learning_rate": 9.41281775868242e-06,
+      "loss": 0.0,
+      "step": 2373
+    },
+    {
+      "epoch": 0.15299349100986015,
+      "grad_norm": 0.0026435490808041713,
+      "learning_rate": 9.412101682778375e-06,
+      "loss": 0.0,
+      "step": 2374
+    },
+    {
+      "epoch": 0.15305793645678933,
+      "grad_norm": 0.06960011244006364,
+      "learning_rate": 9.41138560687433e-06,
+      "loss": 0.0001,
+      "step": 2375
+    },
+    {
+      "epoch": 0.1531223819037185,
+      "grad_norm": 0.0007201561743522177,
+      "learning_rate": 9.410669530970284e-06,
+      "loss": 0.0,
+      "step": 2376
+    },
+    {
+      "epoch": 0.15318682735064768,
+      "grad_norm": 0.003125982065041039,
+      "learning_rate": 9.409953455066238e-06,
+      "loss": 0.0,
+      "step": 2377
+    },
+    {
+      "epoch": 0.15325127279757686,
+      "grad_norm": 0.0042100056123061,
+      "learning_rate": 9.409237379162192e-06,
+      "loss": 0.0,
+      "step": 2378
+    },
+    {
+      "epoch": 0.15331571824450602,
+      "grad_norm": 0.004587416136859462,
+      "learning_rate": 9.408521303258147e-06,
+      "loss": 0.0,
+      "step": 2379
+    },
+    {
+      "epoch": 0.1533801636914352,
+      "grad_norm": 0.04717204972693149,
+      "learning_rate": 9.407805227354099e-06,
+      "loss": 0.0,
+      "step": 2380
+    },
+    {
+      "epoch": 0.15344460913836439,
+      "grad_norm": 0.04171196733747788,
+      "learning_rate": 9.407089151450053e-06,
+      "loss": 0.0001,
+      "step": 2381
+    },
+    {
+      "epoch": 0.15350905458529354,
+      "grad_norm": 0.17253016101069935,
+      "learning_rate": 9.406373075546008e-06,
+      "loss": 0.0005,
+      "step": 2382
+    },
+    {
+      "epoch": 0.15357350003222273,
+      "grad_norm": 0.03761734075604174,
+      "learning_rate": 9.405656999641964e-06,
+      "loss": 0.0,
+      "step": 2383
+    },
+    {
+      "epoch": 0.1536379454791519,
+      "grad_norm": 0.011965065388645795,
+      "learning_rate": 9.404940923737918e-06,
+      "loss": 0.0,
+      "step": 2384
+    },
+    {
+      "epoch": 0.15370239092608107,
+      "grad_norm": 0.04018939250144867,
+      "learning_rate": 9.404224847833872e-06,
+      "loss": 0.0,
+      "step": 2385
+    },
+    {
+      "epoch": 0.15376683637301025,
+      "grad_norm": 0.3395482345981454,
+      "learning_rate": 9.403508771929825e-06,
+      "loss": 0.0008,
+      "step": 2386
+    },
+    {
+      "epoch": 0.1538312818199394,
+      "grad_norm": 0.03146444998541529,
+      "learning_rate": 9.40279269602578e-06,
+      "loss": 0.0,
+      "step": 2387
+    },
+    {
+      "epoch": 0.1538957272668686,
+      "grad_norm": 0.01010570402105599,
+      "learning_rate": 9.402076620121734e-06,
+      "loss": 0.0,
+      "step": 2388
+    },
+    {
+      "epoch": 0.15396017271379778,
+      "grad_norm": 0.006316466860131675,
+      "learning_rate": 9.401360544217688e-06,
+      "loss": 0.0,
+      "step": 2389
+    },
+    {
+      "epoch": 0.15402461816072693,
+      "grad_norm": 0.004210654737544568,
+      "learning_rate": 9.400644468313642e-06,
+      "loss": 0.0,
+      "step": 2390
+    },
+    {
+      "epoch": 0.15408906360765612,
+      "grad_norm": 0.1352283370985708,
+      "learning_rate": 9.399928392409596e-06,
+      "loss": 0.0004,
+      "step": 2391
+    },
+    {
+      "epoch": 0.1541535090545853,
+      "grad_norm": 0.02761470179688728,
+      "learning_rate": 9.39921231650555e-06,
+      "loss": 0.0001,
+      "step": 2392
+    },
+    {
+      "epoch": 0.15421795450151446,
+      "grad_norm": 0.0005356983755468877,
+      "learning_rate": 9.398496240601505e-06,
+      "loss": 0.0,
+      "step": 2393
+    },
+    {
+      "epoch": 0.15428239994844364,
+      "grad_norm": 0.11308521952425876,
+      "learning_rate": 9.39778016469746e-06,
+      "loss": 0.0004,
+      "step": 2394
+    },
+    {
+      "epoch": 0.15434684539537283,
+      "grad_norm": 0.0003210339035201644,
+      "learning_rate": 9.397064088793414e-06,
+      "loss": 0.0,
+      "step": 2395
+    },
+    {
+      "epoch": 0.15441129084230198,
+      "grad_norm": 0.36937508378132805,
+      "learning_rate": 9.396348012889366e-06,
+      "loss": 0.0027,
+      "step": 2396
+    },
+    {
+      "epoch": 0.15447573628923117,
+      "grad_norm": 0.03054608444162845,
+      "learning_rate": 9.39563193698532e-06,
+      "loss": 0.0001,
+      "step": 2397
+    },
+    {
+      "epoch": 0.15454018173616035,
+      "grad_norm": 0.49067405346123016,
+      "learning_rate": 9.394915861081275e-06,
+      "loss": 0.0011,
+      "step": 2398
+    },
+    {
+      "epoch": 0.1546046271830895,
+      "grad_norm": 0.0164183152203303,
+      "learning_rate": 9.394199785177229e-06,
+      "loss": 0.0001,
+      "step": 2399
+    },
+    {
+      "epoch": 0.1546690726300187,
+      "grad_norm": 0.0048164873590324534,
+      "learning_rate": 9.393483709273183e-06,
+      "loss": 0.0,
+      "step": 2400
+    },
+    {
+      "epoch": 0.15473351807694785,
+      "grad_norm": 0.003437130484469497,
+      "learning_rate": 9.392767633369138e-06,
+      "loss": 0.0,
+      "step": 2401
+    },
+    {
+      "epoch": 0.15479796352387704,
+      "grad_norm": 1.1498831648148473,
+      "learning_rate": 9.392051557465092e-06,
+      "loss": 0.0075,
+      "step": 2402
+    },
+    {
+      "epoch": 0.15486240897080622,
+      "grad_norm": 0.5590278432647267,
+      "learning_rate": 9.391335481561046e-06,
+      "loss": 0.0039,
+      "step": 2403
+    },
+    {
+      "epoch": 0.15492685441773538,
+      "grad_norm": 0.006440467338636328,
+      "learning_rate": 9.390619405657e-06,
+      "loss": 0.0,
+      "step": 2404
+    },
+    {
+      "epoch": 0.15499129986466456,
+      "grad_norm": 0.005089887857211715,
+      "learning_rate": 9.389903329752955e-06,
+      "loss": 0.0,
+      "step": 2405
+    },
+    {
+      "epoch": 0.15505574531159375,
+      "grad_norm": 0.03226839406679151,
+      "learning_rate": 9.38918725384891e-06,
+      "loss": 0.0002,
+      "step": 2406
+    },
+    {
+      "epoch": 0.1551201907585229,
+      "grad_norm": 0.005760741049848393,
+      "learning_rate": 9.388471177944863e-06,
+      "loss": 0.0,
+      "step": 2407
+    },
+    {
+      "epoch": 0.1551846362054521,
+      "grad_norm": 0.002409811586602318,
+      "learning_rate": 9.387755102040818e-06,
+      "loss": 0.0,
+      "step": 2408
+    },
+    {
+      "epoch": 0.15524908165238127,
+      "grad_norm": 0.0002274626718670526,
+      "learning_rate": 9.387039026136772e-06,
+      "loss": 0.0,
+      "step": 2409
+    },
+    {
+      "epoch": 0.15531352709931043,
+      "grad_norm": 0.009568083954336773,
+      "learning_rate": 9.386322950232726e-06,
+      "loss": 0.0,
+      "step": 2410
+    },
+    {
+      "epoch": 0.1553779725462396,
+      "grad_norm": 0.0010595213282637791,
+      "learning_rate": 9.38560687432868e-06,
+      "loss": 0.0,
+      "step": 2411
+    },
+    {
+      "epoch": 0.15544241799316877,
+      "grad_norm": 0.0015494971480794625,
+      "learning_rate": 9.384890798424633e-06,
+      "loss": 0.0,
+      "step": 2412
+    },
+    {
+      "epoch": 0.15550686344009795,
+      "grad_norm": 0.027000724745569585,
+      "learning_rate": 9.384174722520588e-06,
+      "loss": 0.0001,
+      "step": 2413
+    },
+    {
+      "epoch": 0.15557130888702714,
+      "grad_norm": 0.00021975748346756332,
+      "learning_rate": 9.383458646616542e-06,
+      "loss": 0.0,
+      "step": 2414
+    },
+    {
+      "epoch": 0.1556357543339563,
+      "grad_norm": 0.22442388753739564,
+      "learning_rate": 9.382742570712496e-06,
+      "loss": 0.001,
+      "step": 2415
+    },
+    {
+      "epoch": 0.15570019978088548,
+      "grad_norm": 0.058095847062474275,
+      "learning_rate": 9.38202649480845e-06,
+      "loss": 0.0002,
+      "step": 2416
+    },
+    {
+      "epoch": 0.15576464522781466,
+      "grad_norm": 0.004947700956219011,
+      "learning_rate": 9.381310418904405e-06,
+      "loss": 0.0,
+      "step": 2417
+    },
+    {
+      "epoch": 0.15582909067474382,
+      "grad_norm": 0.013964975233892866,
+      "learning_rate": 9.380594343000359e-06,
+      "loss": 0.0,
+      "step": 2418
+    },
+    {
+      "epoch": 0.155893536121673,
+      "grad_norm": 0.00045088995588918313,
+      "learning_rate": 9.379878267096313e-06,
+      "loss": 0.0,
+      "step": 2419
+    },
+    {
+      "epoch": 0.1559579815686022,
+      "grad_norm": 0.009997687306932127,
+      "learning_rate": 9.379162191192268e-06,
+      "loss": 0.0,
+      "step": 2420
+    },
+    {
+      "epoch": 0.15602242701553135,
+      "grad_norm": 0.013743444217281438,
+      "learning_rate": 9.378446115288222e-06,
+      "loss": 0.0001,
+      "step": 2421
+    },
+    {
+      "epoch": 0.15608687246246053,
+      "grad_norm": 0.0011079386020701972,
+      "learning_rate": 9.377730039384175e-06,
+      "loss": 0.0,
+      "step": 2422
+    },
+    {
+      "epoch": 0.15615131790938971,
+      "grad_norm": 0.0005776427554553396,
+      "learning_rate": 9.377013963480129e-06,
+      "loss": 0.0,
+      "step": 2423
+    },
+    {
+      "epoch": 0.15621576335631887,
+      "grad_norm": 0.02744104327440318,
+      "learning_rate": 9.376297887576083e-06,
+      "loss": 0.0001,
+      "step": 2424
+    },
+    {
+      "epoch": 0.15628020880324806,
+      "grad_norm": 0.0002999203618423623,
+      "learning_rate": 9.375581811672037e-06,
+      "loss": 0.0,
+      "step": 2425
+    },
+    {
+      "epoch": 0.1563446542501772,
+      "grad_norm": 0.0001923817727009459,
+      "learning_rate": 9.374865735767992e-06,
+      "loss": 0.0,
+      "step": 2426
+    },
+    {
+      "epoch": 0.1564090996971064,
+      "grad_norm": 0.15707550729938025,
+      "learning_rate": 9.374149659863946e-06,
+      "loss": 0.0003,
+      "step": 2427
+    },
+    {
+      "epoch": 0.15647354514403558,
+      "grad_norm": 0.0005202558905786767,
+      "learning_rate": 9.3734335839599e-06,
+      "loss": 0.0,
+      "step": 2428
+    },
+    {
+      "epoch": 0.15653799059096474,
+      "grad_norm": 0.1341350078461987,
+      "learning_rate": 9.372717508055855e-06,
+      "loss": 0.0005,
+      "step": 2429
+    },
+    {
+      "epoch": 0.15660243603789392,
+      "grad_norm": 0.03295119511432381,
+      "learning_rate": 9.372001432151809e-06,
+      "loss": 0.0001,
+      "step": 2430
+    },
+    {
+      "epoch": 0.1566668814848231,
+      "grad_norm": 0.008991741543612479,
+      "learning_rate": 9.371285356247763e-06,
+      "loss": 0.0001,
+      "step": 2431
+    },
+    {
+      "epoch": 0.15673132693175226,
+      "grad_norm": 0.005882962626867871,
+      "learning_rate": 9.370569280343718e-06,
+      "loss": 0.0,
+      "step": 2432
+    },
+    {
+      "epoch": 0.15679577237868145,
+      "grad_norm": 0.14532828296553948,
+      "learning_rate": 9.369853204439672e-06,
+      "loss": 0.0001,
+      "step": 2433
+    },
+    {
+      "epoch": 0.15686021782561063,
+      "grad_norm": 0.001661131622482854,
+      "learning_rate": 9.369137128535626e-06,
+      "loss": 0.0,
+      "step": 2434
+    },
+    {
+      "epoch": 0.1569246632725398,
+      "grad_norm": 0.0013453870636462811,
+      "learning_rate": 9.36842105263158e-06,
+      "loss": 0.0,
+      "step": 2435
+    },
+    {
+      "epoch": 0.15698910871946897,
+      "grad_norm": 0.0007773121412426143,
+      "learning_rate": 9.367704976727535e-06,
+      "loss": 0.0,
+      "step": 2436
+    },
+    {
+      "epoch": 0.15705355416639816,
+      "grad_norm": 0.002945040172602348,
+      "learning_rate": 9.366988900823489e-06,
+      "loss": 0.0,
+      "step": 2437
+    },
+    {
+      "epoch": 0.1571179996133273,
+      "grad_norm": 0.007691798343803482,
+      "learning_rate": 9.366272824919442e-06,
+      "loss": 0.0,
+      "step": 2438
+    },
+    {
+      "epoch": 0.1571824450602565,
+      "grad_norm": 0.001754172241999642,
+      "learning_rate": 9.365556749015396e-06,
+      "loss": 0.0,
+      "step": 2439
+    },
+    {
+      "epoch": 0.15724689050718565,
+      "grad_norm": 0.039875313526018905,
+      "learning_rate": 9.36484067311135e-06,
+      "loss": 0.0004,
+      "step": 2440
+    },
+    {
+      "epoch": 0.15731133595411484,
+      "grad_norm": 0.15183961423677111,
+      "learning_rate": 9.364124597207305e-06,
+      "loss": 0.0005,
+      "step": 2441
+    },
+    {
+      "epoch": 0.15737578140104402,
+      "grad_norm": 0.32147288064350765,
+      "learning_rate": 9.363408521303259e-06,
+      "loss": 0.0019,
+      "step": 2442
+    },
+    {
+      "epoch": 0.15744022684797318,
+      "grad_norm": 0.015395113417899842,
+      "learning_rate": 9.362692445399213e-06,
+      "loss": 0.0,
+      "step": 2443
+    },
+    {
+      "epoch": 0.15750467229490236,
+      "grad_norm": 0.007149663212477922,
+      "learning_rate": 9.361976369495167e-06,
+      "loss": 0.0,
+      "step": 2444
+    },
+    {
+      "epoch": 0.15756911774183155,
+      "grad_norm": 0.006734754031812976,
+      "learning_rate": 9.361260293591122e-06,
+      "loss": 0.0,
+      "step": 2445
+    },
+    {
+      "epoch": 0.1576335631887607,
+      "grad_norm": 0.0014427696968621231,
+      "learning_rate": 9.360544217687076e-06,
+      "loss": 0.0,
+      "step": 2446
+    },
+    {
+      "epoch": 0.1576980086356899,
+      "grad_norm": 0.0067118994039908,
+      "learning_rate": 9.359828141783029e-06,
+      "loss": 0.0,
+      "step": 2447
+    },
+    {
+      "epoch": 0.15776245408261907,
+      "grad_norm": 0.003678528787548895,
+      "learning_rate": 9.359112065878983e-06,
+      "loss": 0.0,
+      "step": 2448
+    },
+    {
+      "epoch": 0.15782689952954823,
+      "grad_norm": 0.0029926002214066875,
+      "learning_rate": 9.358395989974937e-06,
+      "loss": 0.0,
+      "step": 2449
+    },
+    {
+      "epoch": 0.15789134497647742,
+      "grad_norm": 0.0005757762038000473,
+      "learning_rate": 9.357679914070892e-06,
+      "loss": 0.0,
+      "step": 2450
+    },
+    {
+      "epoch": 0.1579557904234066,
+      "grad_norm": 0.16343268918931092,
+      "learning_rate": 9.356963838166846e-06,
+      "loss": 0.0018,
+      "step": 2451
+    },
+    {
+      "epoch": 0.15802023587033576,
+      "grad_norm": 0.011417497251374545,
+      "learning_rate": 9.3562477622628e-06,
+      "loss": 0.0001,
+      "step": 2452
+    },
+    {
+      "epoch": 0.15808468131726494,
+      "grad_norm": 0.00034124654959428265,
+      "learning_rate": 9.355531686358756e-06,
+      "loss": 0.0,
+      "step": 2453
+    },
+    {
+      "epoch": 0.1581491267641941,
+      "grad_norm": 0.009350126488007813,
+      "learning_rate": 9.354815610454709e-06,
+      "loss": 0.0,
+      "step": 2454
+    },
+    {
+      "epoch": 0.15821357221112328,
+      "grad_norm": 0.012536311772587556,
+      "learning_rate": 9.354099534550663e-06,
+      "loss": 0.0,
+      "step": 2455
+    },
+    {
+      "epoch": 0.15827801765805247,
+      "grad_norm": 0.0006414442846759456,
+      "learning_rate": 9.353383458646617e-06,
+      "loss": 0.0,
+      "step": 2456
+    },
+    {
+      "epoch": 0.15834246310498162,
+      "grad_norm": 0.0014955118526876147,
+      "learning_rate": 9.352667382742572e-06,
+      "loss": 0.0,
+      "step": 2457
+    },
+    {
+      "epoch": 0.1584069085519108,
+      "grad_norm": 0.012644152882045408,
+      "learning_rate": 9.351951306838526e-06,
+      "loss": 0.0,
+      "step": 2458
+    },
+    {
+      "epoch": 0.15847135399884,
+      "grad_norm": 0.0022283322330553974,
+      "learning_rate": 9.35123523093448e-06,
+      "loss": 0.0,
+      "step": 2459
+    },
+    {
+      "epoch": 0.15853579944576915,
+      "grad_norm": 0.00216890313118297,
+      "learning_rate": 9.350519155030435e-06,
+      "loss": 0.0,
+      "step": 2460
+    },
+    {
+      "epoch": 0.15860024489269833,
+      "grad_norm": 4.773645306590904e-05,
+      "learning_rate": 9.349803079126389e-06,
+      "loss": 0.0,
+      "step": 2461
+    },
+    {
+      "epoch": 0.15866469033962752,
+      "grad_norm": 0.0026673034609326894,
+      "learning_rate": 9.349087003222343e-06,
+      "loss": 0.0,
+      "step": 2462
+    },
+    {
+      "epoch": 0.15872913578655667,
+      "grad_norm": 0.00012853098665182515,
+      "learning_rate": 9.348370927318296e-06,
+      "loss": 0.0,
+      "step": 2463
+    },
+    {
+      "epoch": 0.15879358123348586,
+      "grad_norm": 0.023990334793908007,
+      "learning_rate": 9.34765485141425e-06,
+      "loss": 0.0002,
+      "step": 2464
+    },
+    {
+      "epoch": 0.15885802668041502,
+      "grad_norm": 0.0015653462913017993,
+      "learning_rate": 9.346938775510204e-06,
+      "loss": 0.0,
+      "step": 2465
+    },
+    {
+      "epoch": 0.1589224721273442,
+      "grad_norm": 0.0017677698246296816,
+      "learning_rate": 9.346222699606159e-06,
+      "loss": 0.0,
+      "step": 2466
+    },
+    {
+      "epoch": 0.15898691757427338,
+      "grad_norm": 0.19936779468096666,
+      "learning_rate": 9.345506623702113e-06,
+      "loss": 0.0053,
+      "step": 2467
+    },
+    {
+      "epoch": 0.15905136302120254,
+      "grad_norm": 0.010805132866747108,
+      "learning_rate": 9.344790547798067e-06,
+      "loss": 0.0001,
+      "step": 2468
+    },
+    {
+      "epoch": 0.15911580846813173,
+      "grad_norm": 0.014871396083823452,
+      "learning_rate": 9.344074471894022e-06,
+      "loss": 0.0002,
+      "step": 2469
+    },
+    {
+      "epoch": 0.1591802539150609,
+      "grad_norm": 0.003370385419454828,
+      "learning_rate": 9.343358395989976e-06,
+      "loss": 0.0,
+      "step": 2470
+    },
+    {
+      "epoch": 0.15924469936199007,
+      "grad_norm": 0.0023136418019346424,
+      "learning_rate": 9.34264232008593e-06,
+      "loss": 0.0,
+      "step": 2471
+    },
+    {
+      "epoch": 0.15930914480891925,
+      "grad_norm": 0.006850854848157116,
+      "learning_rate": 9.341926244181884e-06,
+      "loss": 0.0001,
+      "step": 2472
+    },
+    {
+      "epoch": 0.15937359025584844,
+      "grad_norm": 0.011098353460658784,
+      "learning_rate": 9.341210168277837e-06,
+      "loss": 0.0,
+      "step": 2473
+    },
+    {
+      "epoch": 0.1594380357027776,
+      "grad_norm": 0.0040408311809586265,
+      "learning_rate": 9.340494092373791e-06,
+      "loss": 0.0,
+      "step": 2474
+    },
+    {
+      "epoch": 0.15950248114970678,
+      "grad_norm": 0.0017714230238855028,
+      "learning_rate": 9.339778016469746e-06,
+      "loss": 0.0,
+      "step": 2475
+    },
+    {
+      "epoch": 0.15956692659663596,
+      "grad_norm": 0.0035697117594498335,
+      "learning_rate": 9.339061940565702e-06,
+      "loss": 0.0,
+      "step": 2476
+    },
+    {
+      "epoch": 0.15963137204356512,
+      "grad_norm": 0.0014668561974660174,
+      "learning_rate": 9.338345864661656e-06,
+      "loss": 0.0,
+      "step": 2477
+    },
+    {
+      "epoch": 0.1596958174904943,
+      "grad_norm": 1.4915300288922788,
+      "learning_rate": 9.33762978875761e-06,
+      "loss": 0.008,
+      "step": 2478
+    },
+    {
+      "epoch": 0.15976026293742346,
+      "grad_norm": 0.0040878630786619,
+      "learning_rate": 9.336913712853563e-06,
+      "loss": 0.0,
+      "step": 2479
+    },
+    {
+      "epoch": 0.15982470838435264,
+      "grad_norm": 0.0013943551804445802,
+      "learning_rate": 9.336197636949517e-06,
+      "loss": 0.0,
+      "step": 2480
+    },
+    {
+      "epoch": 0.15988915383128183,
+      "grad_norm": 0.006055781747204741,
+      "learning_rate": 9.335481561045471e-06,
+      "loss": 0.0,
+      "step": 2481
+    },
+    {
+      "epoch": 0.15995359927821098,
+      "grad_norm": 0.02593562146286775,
+      "learning_rate": 9.334765485141426e-06,
+      "loss": 0.0001,
+      "step": 2482
+    },
+    {
+      "epoch": 0.16001804472514017,
+      "grad_norm": 0.00026559044513447465,
+      "learning_rate": 9.33404940923738e-06,
+      "loss": 0.0,
+      "step": 2483
+    },
+    {
+      "epoch": 0.16008249017206935,
+      "grad_norm": 0.0022454719980279157,
+      "learning_rate": 9.333333333333334e-06,
+      "loss": 0.0,
+      "step": 2484
+    },
+    {
+      "epoch": 0.1601469356189985,
+      "grad_norm": 0.019448823897386905,
+      "learning_rate": 9.332617257429289e-06,
+      "loss": 0.0001,
+      "step": 2485
+    },
+    {
+      "epoch": 0.1602113810659277,
+      "grad_norm": 0.002218050878402869,
+      "learning_rate": 9.331901181525243e-06,
+      "loss": 0.0,
+      "step": 2486
+    },
+    {
+      "epoch": 0.16027582651285688,
+      "grad_norm": 0.3381958283192187,
+      "learning_rate": 9.331185105621197e-06,
+      "loss": 0.0011,
+      "step": 2487
+    },
+    {
+      "epoch": 0.16034027195978603,
+      "grad_norm": 0.00022144595144902289,
+      "learning_rate": 9.330469029717152e-06,
+      "loss": 0.0,
+      "step": 2488
+    },
+    {
+      "epoch": 0.16040471740671522,
+      "grad_norm": 0.016463564310275253,
+      "learning_rate": 9.329752953813104e-06,
+      "loss": 0.0001,
+      "step": 2489
+    },
+    {
+      "epoch": 0.1604691628536444,
+      "grad_norm": 0.0029835897225247788,
+      "learning_rate": 9.329036877909058e-06,
+      "loss": 0.0,
+      "step": 2490
+    },
+    {
+      "epoch": 0.16053360830057356,
+      "grad_norm": 0.18282084405229299,
+      "learning_rate": 9.328320802005013e-06,
+      "loss": 0.0017,
+      "step": 2491
+    },
+    {
+      "epoch": 0.16059805374750274,
+      "grad_norm": 0.011583676702772308,
+      "learning_rate": 9.327604726100967e-06,
+      "loss": 0.0001,
+      "step": 2492
+    },
+    {
+      "epoch": 0.1606624991944319,
+      "grad_norm": 0.0034710967822639046,
+      "learning_rate": 9.326888650196921e-06,
+      "loss": 0.0,
+      "step": 2493
+    },
+    {
+      "epoch": 0.16072694464136109,
+      "grad_norm": 0.0011947749959884256,
+      "learning_rate": 9.326172574292876e-06,
+      "loss": 0.0,
+      "step": 2494
+    },
+    {
+      "epoch": 0.16079139008829027,
+      "grad_norm": 0.3944731943453859,
+      "learning_rate": 9.32545649838883e-06,
+      "loss": 0.0024,
+      "step": 2495
+    },
+    {
+      "epoch": 0.16085583553521943,
+      "grad_norm": 0.0018753792654615559,
+      "learning_rate": 9.324740422484784e-06,
+      "loss": 0.0,
+      "step": 2496
+    },
+    {
+      "epoch": 0.1609202809821486,
+      "grad_norm": 0.00033919158385829816,
+      "learning_rate": 9.324024346580739e-06,
+      "loss": 0.0,
+      "step": 2497
+    },
+    {
+      "epoch": 0.1609847264290778,
+      "grad_norm": 0.008896867719477038,
+      "learning_rate": 9.323308270676693e-06,
+      "loss": 0.0001,
+      "step": 2498
+    },
+    {
+      "epoch": 0.16104917187600695,
+      "grad_norm": 0.0010966606224628967,
+      "learning_rate": 9.322592194772647e-06,
+      "loss": 0.0,
+      "step": 2499
+    },
+    {
+      "epoch": 0.16111361732293614,
+      "grad_norm": 6.286197510243029e-05,
+      "learning_rate": 9.321876118868601e-06,
+      "loss": 0.0,
+      "step": 2500
+    },
+    {
+      "epoch": 0.16117806276986532,
+      "grad_norm": 0.012560768212650635,
+      "learning_rate": 9.321160042964556e-06,
+      "loss": 0.0001,
+      "step": 2501
+    },
+    {
+      "epoch": 0.16124250821679448,
+      "grad_norm": 0.01912257629322411,
+      "learning_rate": 9.32044396706051e-06,
+      "loss": 0.0002,
+      "step": 2502
+    },
+    {
+      "epoch": 0.16130695366372366,
+      "grad_norm": 0.016693878853747644,
+      "learning_rate": 9.319727891156464e-06,
+      "loss": 0.0001,
+      "step": 2503
+    },
+    {
+      "epoch": 0.16137139911065282,
+      "grad_norm": 0.07219035829834394,
+      "learning_rate": 9.319011815252419e-06,
+      "loss": 0.0002,
+      "step": 2504
+    },
+    {
+      "epoch": 0.161435844557582,
+      "grad_norm": 0.0012386043214729372,
+      "learning_rate": 9.318295739348371e-06,
+      "loss": 0.0,
+      "step": 2505
+    },
+    {
+      "epoch": 0.1615002900045112,
+      "grad_norm": 0.016528571301615474,
+      "learning_rate": 9.317579663444326e-06,
+      "loss": 0.0001,
+      "step": 2506
+    },
+    {
+      "epoch": 0.16156473545144034,
+      "grad_norm": 0.001469249498037701,
+      "learning_rate": 9.31686358754028e-06,
+      "loss": 0.0,
+      "step": 2507
+    },
+    {
+      "epoch": 0.16162918089836953,
+      "grad_norm": 0.034161891732316735,
+      "learning_rate": 9.316147511636234e-06,
+      "loss": 0.0001,
+      "step": 2508
+    },
+    {
+      "epoch": 0.1616936263452987,
+      "grad_norm": 0.4474646446177699,
+      "learning_rate": 9.315431435732188e-06,
+      "loss": 0.0014,
+      "step": 2509
+    },
+    {
+      "epoch": 0.16175807179222787,
+      "grad_norm": 0.00017682347512700295,
+      "learning_rate": 9.314715359828143e-06,
+      "loss": 0.0,
+      "step": 2510
+    },
+    {
+      "epoch": 0.16182251723915705,
+      "grad_norm": 0.0003216794260017332,
+      "learning_rate": 9.313999283924097e-06,
+      "loss": 0.0,
+      "step": 2511
+    },
+    {
+      "epoch": 0.16188696268608624,
+      "grad_norm": 0.14043387697255008,
+      "learning_rate": 9.313283208020051e-06,
+      "loss": 0.0005,
+      "step": 2512
+    },
+    {
+      "epoch": 0.1619514081330154,
+      "grad_norm": 0.005593976827827833,
+      "learning_rate": 9.312567132116006e-06,
+      "loss": 0.0,
+      "step": 2513
+    },
+    {
+      "epoch": 0.16201585357994458,
+      "grad_norm": 0.00858530863042198,
+      "learning_rate": 9.31185105621196e-06,
+      "loss": 0.0,
+      "step": 2514
+    },
+    {
+      "epoch": 0.16208029902687376,
+      "grad_norm": 0.21605553042851372,
+      "learning_rate": 9.311134980307913e-06,
+      "loss": 0.0015,
+      "step": 2515
+    },
+    {
+      "epoch": 0.16214474447380292,
+      "grad_norm": 0.5249323210916667,
+      "learning_rate": 9.310418904403867e-06,
+      "loss": 0.0019,
+      "step": 2516
+    },
+    {
+      "epoch": 0.1622091899207321,
+      "grad_norm": 0.0031300865219468605,
+      "learning_rate": 9.309702828499821e-06,
+      "loss": 0.0,
+      "step": 2517
+    },
+    {
+      "epoch": 0.16227363536766126,
+      "grad_norm": 0.016066443398272942,
+      "learning_rate": 9.308986752595775e-06,
+      "loss": 0.0002,
+      "step": 2518
+    },
+    {
+      "epoch": 0.16233808081459045,
+      "grad_norm": 0.011800571617004192,
+      "learning_rate": 9.30827067669173e-06,
+      "loss": 0.0001,
+      "step": 2519
+    },
+    {
+      "epoch": 0.16240252626151963,
+      "grad_norm": 0.006443131808993591,
+      "learning_rate": 9.307554600787684e-06,
+      "loss": 0.0,
+      "step": 2520
+    },
+    {
+      "epoch": 0.1624669717084488,
+      "grad_norm": 0.0002683456713499837,
+      "learning_rate": 9.306838524883638e-06,
+      "loss": 0.0,
+      "step": 2521
+    },
+    {
+      "epoch": 0.16253141715537797,
+      "grad_norm": 0.012876411028502148,
+      "learning_rate": 9.306122448979593e-06,
+      "loss": 0.0,
+      "step": 2522
+    },
+    {
+      "epoch": 0.16259586260230716,
+      "grad_norm": 0.052757710574246656,
+      "learning_rate": 9.305406373075547e-06,
+      "loss": 0.0,
+      "step": 2523
+    },
+    {
+      "epoch": 0.1626603080492363,
+      "grad_norm": 0.001491306592329016,
+      "learning_rate": 9.304690297171501e-06,
+      "loss": 0.0,
+      "step": 2524
+    },
+    {
+      "epoch": 0.1627247534961655,
+      "grad_norm": 0.03234166304160187,
+      "learning_rate": 9.303974221267455e-06,
+      "loss": 0.0002,
+      "step": 2525
+    },
+    {
+      "epoch": 0.16278919894309468,
+      "grad_norm": 0.003989245770081959,
+      "learning_rate": 9.30325814536341e-06,
+      "loss": 0.0,
+      "step": 2526
+    },
+    {
+      "epoch": 0.16285364439002384,
+      "grad_norm": 0.03451099249804787,
+      "learning_rate": 9.302542069459364e-06,
+      "loss": 0.0004,
+      "step": 2527
+    },
+    {
+      "epoch": 0.16291808983695302,
+      "grad_norm": 0.0022142978663875178,
+      "learning_rate": 9.301825993555318e-06,
+      "loss": 0.0,
+      "step": 2528
+    },
+    {
+      "epoch": 0.1629825352838822,
+      "grad_norm": 0.0008906830224453514,
+      "learning_rate": 9.301109917651273e-06,
+      "loss": 0.0,
+      "step": 2529
+    },
+    {
+      "epoch": 0.16304698073081136,
+      "grad_norm": 0.40542301956458016,
+      "learning_rate": 9.300393841747227e-06,
+      "loss": 0.0016,
+      "step": 2530
+    },
+    {
+      "epoch": 0.16311142617774055,
+      "grad_norm": 0.0012230303771541364,
+      "learning_rate": 9.29967776584318e-06,
+      "loss": 0.0,
+      "step": 2531
+    },
+    {
+      "epoch": 0.1631758716246697,
+      "grad_norm": 0.056274097549309655,
+      "learning_rate": 9.298961689939134e-06,
+      "loss": 0.0,
+      "step": 2532
+    },
+    {
+      "epoch": 0.1632403170715989,
+      "grad_norm": 0.001701698073510463,
+      "learning_rate": 9.298245614035088e-06,
+      "loss": 0.0,
+      "step": 2533
+    },
+    {
+      "epoch": 0.16330476251852807,
+      "grad_norm": 0.004243688258801596,
+      "learning_rate": 9.297529538131042e-06,
+      "loss": 0.0,
+      "step": 2534
+    },
+    {
+      "epoch": 0.16336920796545723,
+      "grad_norm": 0.0004493868624061053,
+      "learning_rate": 9.296813462226997e-06,
+      "loss": 0.0,
+      "step": 2535
+    },
+    {
+      "epoch": 0.16343365341238641,
+      "grad_norm": 0.15120959538283224,
+      "learning_rate": 9.296097386322951e-06,
+      "loss": 0.0004,
+      "step": 2536
+    },
+    {
+      "epoch": 0.1634980988593156,
+      "grad_norm": 0.012329401352514077,
+      "learning_rate": 9.295381310418905e-06,
+      "loss": 0.0001,
+      "step": 2537
+    },
+    {
+      "epoch": 0.16356254430624476,
+      "grad_norm": 0.006242194356227512,
+      "learning_rate": 9.29466523451486e-06,
+      "loss": 0.0,
+      "step": 2538
+    },
+    {
+      "epoch": 0.16362698975317394,
+      "grad_norm": 0.0004984472123917352,
+      "learning_rate": 9.293949158610814e-06,
+      "loss": 0.0,
+      "step": 2539
+    },
+    {
+      "epoch": 0.16369143520010312,
+      "grad_norm": 0.005776456899099466,
+      "learning_rate": 9.293233082706767e-06,
+      "loss": 0.0,
+      "step": 2540
+    },
+    {
+      "epoch": 0.16375588064703228,
+      "grad_norm": 0.0032904222645318234,
+      "learning_rate": 9.292517006802721e-06,
+      "loss": 0.0,
+      "step": 2541
+    },
+    {
+      "epoch": 0.16382032609396147,
+      "grad_norm": 0.005594085108149277,
+      "learning_rate": 9.291800930898675e-06,
+      "loss": 0.0,
+      "step": 2542
+    },
+    {
+      "epoch": 0.16388477154089062,
+      "grad_norm": 0.7958945806678982,
+      "learning_rate": 9.29108485499463e-06,
+      "loss": 0.0036,
+      "step": 2543
+    },
+    {
+      "epoch": 0.1639492169878198,
+      "grad_norm": 0.007064345315062998,
+      "learning_rate": 9.290368779090584e-06,
+      "loss": 0.0,
+      "step": 2544
+    },
+    {
+      "epoch": 0.164013662434749,
+      "grad_norm": 0.020122688256661223,
+      "learning_rate": 9.289652703186538e-06,
+      "loss": 0.0002,
+      "step": 2545
+    },
+    {
+      "epoch": 0.16407810788167815,
+      "grad_norm": 0.024931361317891115,
+      "learning_rate": 9.288936627282494e-06,
+      "loss": 0.0,
+      "step": 2546
+    },
+    {
+      "epoch": 0.16414255332860733,
+      "grad_norm": 0.2923833720903259,
+      "learning_rate": 9.288220551378447e-06,
+      "loss": 0.0008,
+      "step": 2547
+    },
+    {
+      "epoch": 0.16420699877553652,
+      "grad_norm": 0.002371580849569522,
+      "learning_rate": 9.287504475474401e-06,
+      "loss": 0.0,
+      "step": 2548
+    },
+    {
+      "epoch": 0.16427144422246567,
+      "grad_norm": 0.043302525215061886,
+      "learning_rate": 9.286788399570355e-06,
+      "loss": 0.0001,
+      "step": 2549
+    },
+    {
+      "epoch": 0.16433588966939486,
+      "grad_norm": 0.002335608467180129,
+      "learning_rate": 9.28607232366631e-06,
+      "loss": 0.0,
+      "step": 2550
+    },
+    {
+      "epoch": 0.16440033511632404,
+      "grad_norm": 0.06442228576493117,
+      "learning_rate": 9.285356247762264e-06,
+      "loss": 0.0006,
+      "step": 2551
+    },
+    {
+      "epoch": 0.1644647805632532,
+      "grad_norm": 0.0255342682065745,
+      "learning_rate": 9.284640171858218e-06,
+      "loss": 0.0002,
+      "step": 2552
+    },
+    {
+      "epoch": 0.16452922601018238,
+      "grad_norm": 0.023540083579128342,
+      "learning_rate": 9.283924095954172e-06,
+      "loss": 0.0001,
+      "step": 2553
+    },
+    {
+      "epoch": 0.16459367145711157,
+      "grad_norm": 0.0017502331113364701,
+      "learning_rate": 9.283208020050127e-06,
+      "loss": 0.0,
+      "step": 2554
+    },
+    {
+      "epoch": 0.16465811690404072,
+      "grad_norm": 0.0006908666154918732,
+      "learning_rate": 9.282491944146081e-06,
+      "loss": 0.0,
+      "step": 2555
+    },
+    {
+      "epoch": 0.1647225623509699,
+      "grad_norm": 0.00048313697026926197,
+      "learning_rate": 9.281775868242034e-06,
+      "loss": 0.0,
+      "step": 2556
+    },
+    {
+      "epoch": 0.16478700779789907,
+      "grad_norm": 0.0008612412057671483,
+      "learning_rate": 9.281059792337988e-06,
+      "loss": 0.0,
+      "step": 2557
+    },
+    {
+      "epoch": 0.16485145324482825,
+      "grad_norm": 0.6262135406738908,
+      "learning_rate": 9.280343716433942e-06,
+      "loss": 0.0034,
+      "step": 2558
+    },
+    {
+      "epoch": 0.16491589869175743,
+      "grad_norm": 0.0006356426223387895,
+      "learning_rate": 9.279627640529897e-06,
+      "loss": 0.0,
+      "step": 2559
+    },
+    {
+      "epoch": 0.1649803441386866,
+      "grad_norm": 0.001305826023262251,
+      "learning_rate": 9.278911564625851e-06,
+      "loss": 0.0,
+      "step": 2560
+    },
+    {
+      "epoch": 0.16504478958561578,
+      "grad_norm": 0.2166476208464855,
+      "learning_rate": 9.278195488721805e-06,
+      "loss": 0.0007,
+      "step": 2561
+    },
+    {
+      "epoch": 0.16510923503254496,
+      "grad_norm": 0.0017773730090488518,
+      "learning_rate": 9.27747941281776e-06,
+      "loss": 0.0,
+      "step": 2562
+    },
+    {
+      "epoch": 0.16517368047947412,
+      "grad_norm": 0.01864715096330182,
+      "learning_rate": 9.276763336913714e-06,
+      "loss": 0.0,
+      "step": 2563
+    },
+    {
+      "epoch": 0.1652381259264033,
+      "grad_norm": 0.0009938659133838495,
+      "learning_rate": 9.276047261009668e-06,
+      "loss": 0.0,
+      "step": 2564
+    },
+    {
+      "epoch": 0.16530257137333249,
+      "grad_norm": 0.1604910932287349,
+      "learning_rate": 9.275331185105622e-06,
+      "loss": 0.0001,
+      "step": 2565
+    },
+    {
+      "epoch": 0.16536701682026164,
+      "grad_norm": 0.013365406361419188,
+      "learning_rate": 9.274615109201575e-06,
+      "loss": 0.0,
+      "step": 2566
+    },
+    {
+      "epoch": 0.16543146226719083,
+      "grad_norm": 0.0015641780087289204,
+      "learning_rate": 9.27389903329753e-06,
+      "loss": 0.0,
+      "step": 2567
+    },
+    {
+      "epoch": 0.16549590771412,
+      "grad_norm": 0.0015147951019427674,
+      "learning_rate": 9.273182957393484e-06,
+      "loss": 0.0,
+      "step": 2568
+    },
+    {
+      "epoch": 0.16556035316104917,
+      "grad_norm": 0.003322727915939341,
+      "learning_rate": 9.27246688148944e-06,
+      "loss": 0.0,
+      "step": 2569
+    },
+    {
+      "epoch": 0.16562479860797835,
+      "grad_norm": 0.3481242640776877,
+      "learning_rate": 9.271750805585394e-06,
+      "loss": 0.0082,
+      "step": 2570
+    },
+    {
+      "epoch": 0.1656892440549075,
+      "grad_norm": 0.05511486226657862,
+      "learning_rate": 9.271034729681348e-06,
+      "loss": 0.0003,
+      "step": 2571
+    },
+    {
+      "epoch": 0.1657536895018367,
+      "grad_norm": 0.0013772392389835638,
+      "learning_rate": 9.2703186537773e-06,
+      "loss": 0.0,
+      "step": 2572
+    },
+    {
+      "epoch": 0.16581813494876588,
+      "grad_norm": 0.0028322900559294185,
+      "learning_rate": 9.269602577873255e-06,
+      "loss": 0.0,
+      "step": 2573
+    },
+    {
+      "epoch": 0.16588258039569503,
+      "grad_norm": 0.0008771691379030808,
+      "learning_rate": 9.26888650196921e-06,
+      "loss": 0.0,
+      "step": 2574
+    },
+    {
+      "epoch": 0.16594702584262422,
+      "grad_norm": 0.001837532065417171,
+      "learning_rate": 9.268170426065164e-06,
+      "loss": 0.0,
+      "step": 2575
+    },
+    {
+      "epoch": 0.1660114712895534,
+      "grad_norm": 0.011364499836188676,
+      "learning_rate": 9.267454350161118e-06,
+      "loss": 0.0,
+      "step": 2576
+    },
+    {
+      "epoch": 0.16607591673648256,
+      "grad_norm": 0.43601692392264785,
+      "learning_rate": 9.266738274257072e-06,
+      "loss": 0.0017,
+      "step": 2577
+    },
+    {
+      "epoch": 0.16614036218341174,
+      "grad_norm": 0.2836992981645201,
+      "learning_rate": 9.266022198353027e-06,
+      "loss": 0.0005,
+      "step": 2578
+    },
+    {
+      "epoch": 0.16620480763034093,
+      "grad_norm": 0.006381917270187372,
+      "learning_rate": 9.26530612244898e-06,
+      "loss": 0.0,
+      "step": 2579
+    },
+    {
+      "epoch": 0.16626925307727008,
+      "grad_norm": 0.0995379968224912,
+      "learning_rate": 9.264590046544935e-06,
+      "loss": 0.0001,
+      "step": 2580
+    },
+    {
+      "epoch": 0.16633369852419927,
+      "grad_norm": 0.005838220002381634,
+      "learning_rate": 9.26387397064089e-06,
+      "loss": 0.0,
+      "step": 2581
+    },
+    {
+      "epoch": 0.16639814397112843,
+      "grad_norm": 0.1734023075655848,
+      "learning_rate": 9.263157894736842e-06,
+      "loss": 0.0015,
+      "step": 2582
+    },
+    {
+      "epoch": 0.1664625894180576,
+      "grad_norm": 0.32543608770325944,
+      "learning_rate": 9.262441818832796e-06,
+      "loss": 0.001,
+      "step": 2583
+    },
+    {
+      "epoch": 0.1665270348649868,
+      "grad_norm": 0.035015456749274816,
+      "learning_rate": 9.26172574292875e-06,
+      "loss": 0.002,
+      "step": 2584
+    },
+    {
+      "epoch": 0.16659148031191595,
+      "grad_norm": 0.00900476722572773,
+      "learning_rate": 9.261009667024705e-06,
+      "loss": 0.0,
+      "step": 2585
+    },
+    {
+      "epoch": 0.16665592575884514,
+      "grad_norm": 0.0008566073306815689,
+      "learning_rate": 9.26029359112066e-06,
+      "loss": 0.0,
+      "step": 2586
+    },
+    {
+      "epoch": 0.16672037120577432,
+      "grad_norm": 0.4923978083746918,
+      "learning_rate": 9.259577515216614e-06,
+      "loss": 0.0022,
+      "step": 2587
+    },
+    {
+      "epoch": 0.16678481665270348,
+      "grad_norm": 0.0018108477937682785,
+      "learning_rate": 9.258861439312568e-06,
+      "loss": 0.0,
+      "step": 2588
+    },
+    {
+      "epoch": 0.16684926209963266,
+      "grad_norm": 0.004562858927769996,
+      "learning_rate": 9.258145363408522e-06,
+      "loss": 0.0,
+      "step": 2589
+    },
+    {
+      "epoch": 0.16691370754656185,
+      "grad_norm": 0.07446345657383704,
+      "learning_rate": 9.257429287504476e-06,
+      "loss": 0.0001,
+      "step": 2590
+    },
+    {
+      "epoch": 0.166978152993491,
+      "grad_norm": 0.0009006018433352618,
+      "learning_rate": 9.25671321160043e-06,
+      "loss": 0.0,
+      "step": 2591
+    },
+    {
+      "epoch": 0.1670425984404202,
+      "grad_norm": 0.007707075664561584,
+      "learning_rate": 9.255997135696383e-06,
+      "loss": 0.0,
+      "step": 2592
+    },
+    {
+      "epoch": 0.16710704388734937,
+      "grad_norm": 0.1405618313798484,
+      "learning_rate": 9.25528105979234e-06,
+      "loss": 0.0002,
+      "step": 2593
+    },
+    {
+      "epoch": 0.16717148933427853,
+      "grad_norm": 0.0008934174065720114,
+      "learning_rate": 9.254564983888294e-06,
+      "loss": 0.0,
+      "step": 2594
+    },
+    {
+      "epoch": 0.1672359347812077,
+      "grad_norm": 0.5161244256757117,
+      "learning_rate": 9.253848907984248e-06,
+      "loss": 0.0014,
+      "step": 2595
+    },
+    {
+      "epoch": 0.16730038022813687,
+      "grad_norm": 0.05120134815256295,
+      "learning_rate": 9.253132832080202e-06,
+      "loss": 0.0001,
+      "step": 2596
+    },
+    {
+      "epoch": 0.16736482567506605,
+      "grad_norm": 0.021043779453902272,
+      "learning_rate": 9.252416756176157e-06,
+      "loss": 0.0002,
+      "step": 2597
+    },
+    {
+      "epoch": 0.16742927112199524,
+      "grad_norm": 0.0009231523664426139,
+      "learning_rate": 9.251700680272109e-06,
+      "loss": 0.0,
+      "step": 2598
+    },
+    {
+      "epoch": 0.1674937165689244,
+      "grad_norm": 0.016473526227955518,
+      "learning_rate": 9.250984604368063e-06,
+      "loss": 0.0,
+      "step": 2599
+    },
+    {
+      "epoch": 0.16755816201585358,
+      "grad_norm": 0.0006848207539497617,
+      "learning_rate": 9.250268528464018e-06,
+      "loss": 0.0,
+      "step": 2600
+    },
+    {
+      "epoch": 0.16762260746278276,
+      "grad_norm": 0.001094208815293132,
+      "learning_rate": 9.249552452559972e-06,
+      "loss": 0.0,
+      "step": 2601
+    },
+    {
+      "epoch": 0.16768705290971192,
+      "grad_norm": 0.00046785349861243374,
+      "learning_rate": 9.248836376655926e-06,
+      "loss": 0.0,
+      "step": 2602
+    },
+    {
+      "epoch": 0.1677514983566411,
+      "grad_norm": 0.00014457733525141594,
+      "learning_rate": 9.24812030075188e-06,
+      "loss": 0.0,
+      "step": 2603
+    },
+    {
+      "epoch": 0.1678159438035703,
+      "grad_norm": 0.029550263242095656,
+      "learning_rate": 9.247404224847835e-06,
+      "loss": 0.0001,
+      "step": 2604
+    },
+    {
+      "epoch": 0.16788038925049945,
+      "grad_norm": 0.8862753948667788,
+      "learning_rate": 9.24668814894379e-06,
+      "loss": 0.003,
+      "step": 2605
+    },
+    {
+      "epoch": 0.16794483469742863,
+      "grad_norm": 0.011991181543029776,
+      "learning_rate": 9.245972073039744e-06,
+      "loss": 0.0001,
+      "step": 2606
+    },
+    {
+      "epoch": 0.16800928014435781,
+      "grad_norm": 0.05132832108348396,
+      "learning_rate": 9.245255997135698e-06,
+      "loss": 0.0016,
+      "step": 2607
+    },
+    {
+      "epoch": 0.16807372559128697,
+      "grad_norm": 0.01007424389617214,
+      "learning_rate": 9.24453992123165e-06,
+      "loss": 0.0001,
+      "step": 2608
+    },
+    {
+      "epoch": 0.16813817103821616,
+      "grad_norm": 0.3279116481888165,
+      "learning_rate": 9.243823845327605e-06,
+      "loss": 0.0008,
+      "step": 2609
+    },
+    {
+      "epoch": 0.1682026164851453,
+      "grad_norm": 0.0010222354990563067,
+      "learning_rate": 9.243107769423559e-06,
+      "loss": 0.0,
+      "step": 2610
+    },
+    {
+      "epoch": 0.1682670619320745,
+      "grad_norm": 0.0002879770515881767,
+      "learning_rate": 9.242391693519513e-06,
+      "loss": 0.0,
+      "step": 2611
+    },
+    {
+      "epoch": 0.16833150737900368,
+      "grad_norm": 0.04284265966657461,
+      "learning_rate": 9.241675617615468e-06,
+      "loss": 0.0004,
+      "step": 2612
+    },
+    {
+      "epoch": 0.16839595282593284,
+      "grad_norm": 0.02611223447159595,
+      "learning_rate": 9.240959541711422e-06,
+      "loss": 0.0004,
+      "step": 2613
+    },
+    {
+      "epoch": 0.16846039827286202,
+      "grad_norm": 0.000562746261733419,
+      "learning_rate": 9.240243465807376e-06,
+      "loss": 0.0,
+      "step": 2614
+    },
+    {
+      "epoch": 0.1685248437197912,
+      "grad_norm": 0.006530159238405348,
+      "learning_rate": 9.23952738990333e-06,
+      "loss": 0.0,
+      "step": 2615
+    },
+    {
+      "epoch": 0.16858928916672036,
+      "grad_norm": 0.004683470390970493,
+      "learning_rate": 9.238811313999285e-06,
+      "loss": 0.0,
+      "step": 2616
+    },
+    {
+      "epoch": 0.16865373461364955,
+      "grad_norm": 0.4705708583019044,
+      "learning_rate": 9.238095238095239e-06,
+      "loss": 0.0032,
+      "step": 2617
+    },
+    {
+      "epoch": 0.16871818006057873,
+      "grad_norm": 0.0005363429561250486,
+      "learning_rate": 9.237379162191193e-06,
+      "loss": 0.0,
+      "step": 2618
+    },
+    {
+      "epoch": 0.1687826255075079,
+      "grad_norm": 0.15183939342721997,
+      "learning_rate": 9.236663086287148e-06,
+      "loss": 0.0017,
+      "step": 2619
+    },
+    {
+      "epoch": 0.16884707095443707,
+      "grad_norm": 0.0001052157179584856,
+      "learning_rate": 9.235947010383102e-06,
+      "loss": 0.0,
+      "step": 2620
+    },
+    {
+      "epoch": 0.16891151640136626,
+      "grad_norm": 0.0007306970136187449,
+      "learning_rate": 9.235230934479056e-06,
+      "loss": 0.0,
+      "step": 2621
+    },
+    {
+      "epoch": 0.16897596184829541,
+      "grad_norm": 0.002508690093415649,
+      "learning_rate": 9.23451485857501e-06,
+      "loss": 0.0,
+      "step": 2622
+    },
+    {
+      "epoch": 0.1690404072952246,
+      "grad_norm": 0.0005834624117502399,
+      "learning_rate": 9.233798782670965e-06,
+      "loss": 0.0,
+      "step": 2623
+    },
+    {
+      "epoch": 0.16910485274215376,
+      "grad_norm": 0.002580928592817899,
+      "learning_rate": 9.233082706766918e-06,
+      "loss": 0.0,
+      "step": 2624
+    },
+    {
+      "epoch": 0.16916929818908294,
+      "grad_norm": 0.004280836334426163,
+      "learning_rate": 9.232366630862872e-06,
+      "loss": 0.0,
+      "step": 2625
+    },
+    {
+      "epoch": 0.16923374363601212,
+      "grad_norm": 0.00018424144724289227,
+      "learning_rate": 9.231650554958826e-06,
+      "loss": 0.0,
+      "step": 2626
+    },
+    {
+      "epoch": 0.16929818908294128,
+      "grad_norm": 0.0010581367877079988,
+      "learning_rate": 9.23093447905478e-06,
+      "loss": 0.0,
+      "step": 2627
+    },
+    {
+      "epoch": 0.16936263452987046,
+      "grad_norm": 0.0007131972095357654,
+      "learning_rate": 9.230218403150735e-06,
+      "loss": 0.0,
+      "step": 2628
+    },
+    {
+      "epoch": 0.16942707997679965,
+      "grad_norm": 0.00043519554939918065,
+      "learning_rate": 9.229502327246689e-06,
+      "loss": 0.0,
+      "step": 2629
+    },
+    {
+      "epoch": 0.1694915254237288,
+      "grad_norm": 0.001351474322390404,
+      "learning_rate": 9.228786251342643e-06,
+      "loss": 0.0,
+      "step": 2630
+    },
+    {
+      "epoch": 0.169555970870658,
+      "grad_norm": 0.020752774250390287,
+      "learning_rate": 9.228070175438598e-06,
+      "loss": 0.0002,
+      "step": 2631
+    },
+    {
+      "epoch": 0.16962041631758717,
+      "grad_norm": 0.018716232806307055,
+      "learning_rate": 9.227354099534552e-06,
+      "loss": 0.0001,
+      "step": 2632
+    },
+    {
+      "epoch": 0.16968486176451633,
+      "grad_norm": 0.00029643028839327425,
+      "learning_rate": 9.226638023630505e-06,
+      "loss": 0.0,
+      "step": 2633
+    },
+    {
+      "epoch": 0.16974930721144552,
+      "grad_norm": 0.0022362696824027137,
+      "learning_rate": 9.225921947726459e-06,
+      "loss": 0.0,
+      "step": 2634
+    },
+    {
+      "epoch": 0.16981375265837467,
+      "grad_norm": 0.00021475709016773186,
+      "learning_rate": 9.225205871822413e-06,
+      "loss": 0.0,
+      "step": 2635
+    },
+    {
+      "epoch": 0.16987819810530386,
+      "grad_norm": 0.16918964578636836,
+      "learning_rate": 9.224489795918367e-06,
+      "loss": 0.0019,
+      "step": 2636
+    },
+    {
+      "epoch": 0.16994264355223304,
+      "grad_norm": 0.12910555113432312,
+      "learning_rate": 9.223773720014322e-06,
+      "loss": 0.0002,
+      "step": 2637
+    },
+    {
+      "epoch": 0.1700070889991622,
+      "grad_norm": 0.024268443228424336,
+      "learning_rate": 9.223057644110276e-06,
+      "loss": 0.0001,
+      "step": 2638
+    },
+    {
+      "epoch": 0.17007153444609138,
+      "grad_norm": 0.0003196529914171969,
+      "learning_rate": 9.222341568206232e-06,
+      "loss": 0.0,
+      "step": 2639
+    },
+    {
+      "epoch": 0.17013597989302057,
+      "grad_norm": 0.09965735102900866,
+      "learning_rate": 9.221625492302185e-06,
+      "loss": 0.0038,
+      "step": 2640
+    },
+    {
+      "epoch": 0.17020042533994972,
+      "grad_norm": 0.009036427241455914,
+      "learning_rate": 9.220909416398139e-06,
+      "loss": 0.0001,
+      "step": 2641
+    },
+    {
+      "epoch": 0.1702648707868789,
+      "grad_norm": 0.00011327834905210328,
+      "learning_rate": 9.220193340494093e-06,
+      "loss": 0.0,
+      "step": 2642
+    },
+    {
+      "epoch": 0.1703293162338081,
+      "grad_norm": 0.00287235194554443,
+      "learning_rate": 9.219477264590047e-06,
+      "loss": 0.0,
+      "step": 2643
+    },
+    {
+      "epoch": 0.17039376168073725,
+      "grad_norm": 0.015226695642371184,
+      "learning_rate": 9.218761188686002e-06,
+      "loss": 0.0001,
+      "step": 2644
+    },
+    {
+      "epoch": 0.17045820712766643,
+      "grad_norm": 0.00020810014028015193,
+      "learning_rate": 9.218045112781956e-06,
+      "loss": 0.0,
+      "step": 2645
+    },
+    {
+      "epoch": 0.17052265257459562,
+      "grad_norm": 0.00020155726396008463,
+      "learning_rate": 9.21732903687791e-06,
+      "loss": 0.0,
+      "step": 2646
+    },
+    {
+      "epoch": 0.17058709802152477,
+      "grad_norm": 0.0010949986795731515,
+      "learning_rate": 9.216612960973865e-06,
+      "loss": 0.0,
+      "step": 2647
+    },
+    {
+      "epoch": 0.17065154346845396,
+      "grad_norm": 0.0076949646895825684,
+      "learning_rate": 9.215896885069819e-06,
+      "loss": 0.0001,
+      "step": 2648
+    },
+    {
+      "epoch": 0.17071598891538312,
+      "grad_norm": 0.001981525942460112,
+      "learning_rate": 9.215180809165772e-06,
+      "loss": 0.0,
+      "step": 2649
+    },
+    {
+      "epoch": 0.1707804343623123,
+      "grad_norm": 0.038440089378423435,
+      "learning_rate": 9.214464733261726e-06,
+      "loss": 0.0001,
+      "step": 2650
+    },
+    {
+      "epoch": 0.17084487980924148,
+      "grad_norm": 0.10907353425293582,
+      "learning_rate": 9.21374865735768e-06,
+      "loss": 0.0001,
+      "step": 2651
+    },
+    {
+      "epoch": 0.17090932525617064,
+      "grad_norm": 0.0063255364810587255,
+      "learning_rate": 9.213032581453634e-06,
+      "loss": 0.0,
+      "step": 2652
+    },
+    {
+      "epoch": 0.17097377070309983,
+      "grad_norm": 0.012754438453606867,
+      "learning_rate": 9.212316505549589e-06,
+      "loss": 0.0,
+      "step": 2653
+    },
+    {
+      "epoch": 0.171038216150029,
+      "grad_norm": 0.00026067457484949056,
+      "learning_rate": 9.211600429645543e-06,
+      "loss": 0.0,
+      "step": 2654
+    },
+    {
+      "epoch": 0.17110266159695817,
+      "grad_norm": 0.00021310395223525356,
+      "learning_rate": 9.210884353741497e-06,
+      "loss": 0.0,
+      "step": 2655
+    },
+    {
+      "epoch": 0.17116710704388735,
+      "grad_norm": 0.010862934374835192,
+      "learning_rate": 9.210168277837452e-06,
+      "loss": 0.0,
+      "step": 2656
+    },
+    {
+      "epoch": 0.17123155249081654,
+      "grad_norm": 0.06877897773526727,
+      "learning_rate": 9.209452201933406e-06,
+      "loss": 0.0002,
+      "step": 2657
+    },
+    {
+      "epoch": 0.1712959979377457,
+      "grad_norm": 0.0003992406670996045,
+      "learning_rate": 9.20873612602936e-06,
+      "loss": 0.0,
+      "step": 2658
+    },
+    {
+      "epoch": 0.17136044338467488,
+      "grad_norm": 0.000843099911044625,
+      "learning_rate": 9.208020050125313e-06,
+      "loss": 0.0,
+      "step": 2659
+    },
+    {
+      "epoch": 0.17142488883160406,
+      "grad_norm": 0.03777898066302125,
+      "learning_rate": 9.207303974221267e-06,
+      "loss": 0.0004,
+      "step": 2660
+    },
+    {
+      "epoch": 0.17148933427853322,
+      "grad_norm": 0.0013253862296127193,
+      "learning_rate": 9.206587898317221e-06,
+      "loss": 0.0,
+      "step": 2661
+    },
+    {
+      "epoch": 0.1715537797254624,
+      "grad_norm": 0.0019887129236439605,
+      "learning_rate": 9.205871822413176e-06,
+      "loss": 0.0,
+      "step": 2662
+    },
+    {
+      "epoch": 0.17161822517239156,
+      "grad_norm": 0.01141948088125074,
+      "learning_rate": 9.205155746509132e-06,
+      "loss": 0.0,
+      "step": 2663
+    },
+    {
+      "epoch": 0.17168267061932074,
+      "grad_norm": 0.007296598030753204,
+      "learning_rate": 9.204439670605086e-06,
+      "loss": 0.0,
+      "step": 2664
+    },
+    {
+      "epoch": 0.17174711606624993,
+      "grad_norm": 0.015167112561033686,
+      "learning_rate": 9.203723594701039e-06,
+      "loss": 0.0002,
+      "step": 2665
+    },
+    {
+      "epoch": 0.17181156151317908,
+      "grad_norm": 0.006666472705684092,
+      "learning_rate": 9.203007518796993e-06,
+      "loss": 0.0,
+      "step": 2666
+    },
+    {
+      "epoch": 0.17187600696010827,
+      "grad_norm": 0.005332642179530365,
+      "learning_rate": 9.202291442892947e-06,
+      "loss": 0.0,
+      "step": 2667
+    },
+    {
+      "epoch": 0.17194045240703745,
+      "grad_norm": 0.0003287436028234088,
+      "learning_rate": 9.201575366988902e-06,
+      "loss": 0.0,
+      "step": 2668
+    },
+    {
+      "epoch": 0.1720048978539666,
+      "grad_norm": 0.008185411046763031,
+      "learning_rate": 9.200859291084856e-06,
+      "loss": 0.0,
+      "step": 2669
+    },
+    {
+      "epoch": 0.1720693433008958,
+      "grad_norm": 0.0003072117989800363,
+      "learning_rate": 9.20014321518081e-06,
+      "loss": 0.0,
+      "step": 2670
+    },
+    {
+      "epoch": 0.17213378874782498,
+      "grad_norm": 0.14845676673531352,
+      "learning_rate": 9.199427139276764e-06,
+      "loss": 0.0001,
+      "step": 2671
+    },
+    {
+      "epoch": 0.17219823419475414,
+      "grad_norm": 0.0023122495109587056,
+      "learning_rate": 9.198711063372719e-06,
+      "loss": 0.0,
+      "step": 2672
+    },
+    {
+      "epoch": 0.17226267964168332,
+      "grad_norm": 2.9962439866178707,
+      "learning_rate": 9.197994987468673e-06,
+      "loss": 0.0269,
+      "step": 2673
+    },
+    {
+      "epoch": 0.17232712508861248,
+      "grad_norm": 0.040175158651624854,
+      "learning_rate": 9.197278911564627e-06,
+      "loss": 0.0002,
+      "step": 2674
+    },
+    {
+      "epoch": 0.17239157053554166,
+      "grad_norm": 0.0012490798280579728,
+      "learning_rate": 9.19656283566058e-06,
+      "loss": 0.0,
+      "step": 2675
+    },
+    {
+      "epoch": 0.17245601598247084,
+      "grad_norm": 0.003016418042880189,
+      "learning_rate": 9.195846759756534e-06,
+      "loss": 0.0,
+      "step": 2676
+    },
+    {
+      "epoch": 0.1725204614294,
+      "grad_norm": 0.003284682926061627,
+      "learning_rate": 9.195130683852489e-06,
+      "loss": 0.0,
+      "step": 2677
+    },
+    {
+      "epoch": 0.17258490687632919,
+      "grad_norm": 0.011853928536630667,
+      "learning_rate": 9.194414607948443e-06,
+      "loss": 0.0001,
+      "step": 2678
+    },
+    {
+      "epoch": 0.17264935232325837,
+      "grad_norm": 0.09941864921166386,
+      "learning_rate": 9.193698532044397e-06,
+      "loss": 0.0002,
+      "step": 2679
+    },
+    {
+      "epoch": 0.17271379777018753,
+      "grad_norm": 0.008067099418627759,
+      "learning_rate": 9.192982456140351e-06,
+      "loss": 0.0,
+      "step": 2680
+    },
+    {
+      "epoch": 0.1727782432171167,
+      "grad_norm": 0.5154138768623525,
+      "learning_rate": 9.192266380236306e-06,
+      "loss": 0.0166,
+      "step": 2681
+    },
+    {
+      "epoch": 0.1728426886640459,
+      "grad_norm": 0.01469303930545766,
+      "learning_rate": 9.19155030433226e-06,
+      "loss": 0.0001,
+      "step": 2682
+    },
+    {
+      "epoch": 0.17290713411097505,
+      "grad_norm": 0.220314544641192,
+      "learning_rate": 9.190834228428214e-06,
+      "loss": 0.006,
+      "step": 2683
+    },
+    {
+      "epoch": 0.17297157955790424,
+      "grad_norm": 0.09990222740725906,
+      "learning_rate": 9.190118152524169e-06,
+      "loss": 0.0004,
+      "step": 2684
+    },
+    {
+      "epoch": 0.17303602500483342,
+      "grad_norm": 0.00873880537834303,
+      "learning_rate": 9.189402076620121e-06,
+      "loss": 0.0,
+      "step": 2685
+    },
+    {
+      "epoch": 0.17310047045176258,
+      "grad_norm": 0.0012456913305398055,
+      "learning_rate": 9.188686000716077e-06,
+      "loss": 0.0,
+      "step": 2686
+    },
+    {
+      "epoch": 0.17316491589869176,
+      "grad_norm": 0.003980550458896421,
+      "learning_rate": 9.187969924812032e-06,
+      "loss": 0.0,
+      "step": 2687
+    },
+    {
+      "epoch": 0.17322936134562092,
+      "grad_norm": 0.24734072155041284,
+      "learning_rate": 9.187253848907986e-06,
+      "loss": 0.0062,
+      "step": 2688
+    },
+    {
+      "epoch": 0.1732938067925501,
+      "grad_norm": 0.0014917908984364223,
+      "learning_rate": 9.18653777300394e-06,
+      "loss": 0.0,
+      "step": 2689
+    },
+    {
+      "epoch": 0.1733582522394793,
+      "grad_norm": 0.018215838914686967,
+      "learning_rate": 9.185821697099894e-06,
+      "loss": 0.0,
+      "step": 2690
+    },
+    {
+      "epoch": 0.17342269768640844,
+      "grad_norm": 0.0007429930106887829,
+      "learning_rate": 9.185105621195847e-06,
+      "loss": 0.0,
+      "step": 2691
+    },
+    {
+      "epoch": 0.17348714313333763,
+      "grad_norm": 0.0001631497864257658,
+      "learning_rate": 9.184389545291801e-06,
+      "loss": 0.0,
+      "step": 2692
+    },
+    {
+      "epoch": 0.1735515885802668,
+      "grad_norm": 0.010403012017535249,
+      "learning_rate": 9.183673469387756e-06,
+      "loss": 0.0,
+      "step": 2693
+    },
+    {
+      "epoch": 0.17361603402719597,
+      "grad_norm": 0.008222731931976913,
+      "learning_rate": 9.18295739348371e-06,
+      "loss": 0.0001,
+      "step": 2694
+    },
+    {
+      "epoch": 0.17368047947412515,
+      "grad_norm": 0.03993690792998894,
+      "learning_rate": 9.182241317579664e-06,
+      "loss": 0.0001,
+      "step": 2695
+    },
+    {
+      "epoch": 0.17374492492105434,
+      "grad_norm": 0.00030802079541501006,
+      "learning_rate": 9.181525241675619e-06,
+      "loss": 0.0,
+      "step": 2696
+    },
+    {
+      "epoch": 0.1738093703679835,
+      "grad_norm": 0.005414459370388794,
+      "learning_rate": 9.180809165771573e-06,
+      "loss": 0.0,
+      "step": 2697
+    },
+    {
+      "epoch": 0.17387381581491268,
+      "grad_norm": 0.006613869105241104,
+      "learning_rate": 9.180093089867527e-06,
+      "loss": 0.0,
+      "step": 2698
+    },
+    {
+      "epoch": 0.17393826126184186,
+      "grad_norm": 0.3907094673384912,
+      "learning_rate": 9.179377013963481e-06,
+      "loss": 0.0019,
+      "step": 2699
+    },
+    {
+      "epoch": 0.17400270670877102,
+      "grad_norm": 0.009057081484954533,
+      "learning_rate": 9.178660938059436e-06,
+      "loss": 0.0001,
+      "step": 2700
+    },
+    {
+      "epoch": 0.1740671521557002,
+      "grad_norm": 0.7250289730333702,
+      "learning_rate": 9.177944862155388e-06,
+      "loss": 0.0027,
+      "step": 2701
+    },
+    {
+      "epoch": 0.17413159760262936,
+      "grad_norm": 0.005397337466009714,
+      "learning_rate": 9.177228786251343e-06,
+      "loss": 0.0001,
+      "step": 2702
+    },
+    {
+      "epoch": 0.17419604304955855,
+      "grad_norm": 0.0006209733526635416,
+      "learning_rate": 9.176512710347297e-06,
+      "loss": 0.0,
+      "step": 2703
+    },
+    {
+      "epoch": 0.17426048849648773,
+      "grad_norm": 0.04377721584989303,
+      "learning_rate": 9.175796634443251e-06,
+      "loss": 0.0002,
+      "step": 2704
+    },
+    {
+      "epoch": 0.1743249339434169,
+      "grad_norm": 0.0001704777087367316,
+      "learning_rate": 9.175080558539206e-06,
+      "loss": 0.0,
+      "step": 2705
+    },
+    {
+      "epoch": 0.17438937939034607,
+      "grad_norm": 0.002415898461971982,
+      "learning_rate": 9.17436448263516e-06,
+      "loss": 0.0,
+      "step": 2706
+    },
+    {
+      "epoch": 0.17445382483727526,
+      "grad_norm": 0.00010784267160302466,
+      "learning_rate": 9.173648406731114e-06,
+      "loss": 0.0,
+      "step": 2707
+    },
+    {
+      "epoch": 0.1745182702842044,
+      "grad_norm": 0.382636049048164,
+      "learning_rate": 9.172932330827068e-06,
+      "loss": 0.0106,
+      "step": 2708
+    },
+    {
+      "epoch": 0.1745827157311336,
+      "grad_norm": 0.0011183936260544516,
+      "learning_rate": 9.172216254923023e-06,
+      "loss": 0.0,
+      "step": 2709
+    },
+    {
+      "epoch": 0.17464716117806278,
+      "grad_norm": 0.0002463259427477652,
+      "learning_rate": 9.171500179018977e-06,
+      "loss": 0.0,
+      "step": 2710
+    },
+    {
+      "epoch": 0.17471160662499194,
+      "grad_norm": 1.1865234375,
+      "learning_rate": 9.170784103114931e-06,
+      "loss": 0.0072,
+      "step": 2711
+    },
+    {
+      "epoch": 0.17477605207192112,
+      "grad_norm": 0.002127015512866035,
+      "learning_rate": 9.170068027210886e-06,
+      "loss": 0.0,
+      "step": 2712
+    },
+    {
+      "epoch": 0.17484049751885028,
+      "grad_norm": 0.12030690802546253,
+      "learning_rate": 9.16935195130684e-06,
+      "loss": 0.0005,
+      "step": 2713
+    },
+    {
+      "epoch": 0.17490494296577946,
+      "grad_norm": 0.521378924391397,
+      "learning_rate": 9.168635875402794e-06,
+      "loss": 0.0034,
+      "step": 2714
+    },
+    {
+      "epoch": 0.17496938841270865,
+      "grad_norm": 0.018006203519038057,
+      "learning_rate": 9.167919799498749e-06,
+      "loss": 0.0001,
+      "step": 2715
+    },
+    {
+      "epoch": 0.1750338338596378,
+      "grad_norm": 0.004143970490399163,
+      "learning_rate": 9.167203723594703e-06,
+      "loss": 0.0,
+      "step": 2716
+    },
+    {
+      "epoch": 0.175098279306567,
+      "grad_norm": 0.00027283137921355467,
+      "learning_rate": 9.166487647690655e-06,
+      "loss": 0.0,
+      "step": 2717
+    },
+    {
+      "epoch": 0.17516272475349617,
+      "grad_norm": 0.0491949441081807,
+      "learning_rate": 9.16577157178661e-06,
+      "loss": 0.0001,
+      "step": 2718
+    },
+    {
+      "epoch": 0.17522717020042533,
+      "grad_norm": 0.0031770309749900405,
+      "learning_rate": 9.165055495882564e-06,
+      "loss": 0.0,
+      "step": 2719
+    },
+    {
+      "epoch": 0.17529161564735452,
+      "grad_norm": 0.10751967746263864,
+      "learning_rate": 9.164339419978518e-06,
+      "loss": 0.0005,
+      "step": 2720
+    },
+    {
+      "epoch": 0.1753560610942837,
+      "grad_norm": 0.14063707273371076,
+      "learning_rate": 9.163623344074473e-06,
+      "loss": 0.0003,
+      "step": 2721
+    },
+    {
+      "epoch": 0.17542050654121286,
+      "grad_norm": 0.0011253487254859079,
+      "learning_rate": 9.162907268170427e-06,
+      "loss": 0.0,
+      "step": 2722
+    },
+    {
+      "epoch": 0.17548495198814204,
+      "grad_norm": 0.030041388852838077,
+      "learning_rate": 9.162191192266381e-06,
+      "loss": 0.0001,
+      "step": 2723
+    },
+    {
+      "epoch": 0.17554939743507122,
+      "grad_norm": 0.0030749777231569345,
+      "learning_rate": 9.161475116362336e-06,
+      "loss": 0.0,
+      "step": 2724
+    },
+    {
+      "epoch": 0.17561384288200038,
+      "grad_norm": 0.01139113188354577,
+      "learning_rate": 9.16075904045829e-06,
+      "loss": 0.0001,
+      "step": 2725
+    },
+    {
+      "epoch": 0.17567828832892957,
+      "grad_norm": 0.005132686334298028,
+      "learning_rate": 9.160042964554242e-06,
+      "loss": 0.0,
+      "step": 2726
+    },
+    {
+      "epoch": 0.17574273377585872,
+      "grad_norm": 0.011252321098629682,
+      "learning_rate": 9.159326888650197e-06,
+      "loss": 0.0,
+      "step": 2727
+    },
+    {
+      "epoch": 0.1758071792227879,
+      "grad_norm": 0.0001379375049884014,
+      "learning_rate": 9.158610812746151e-06,
+      "loss": 0.0,
+      "step": 2728
+    },
+    {
+      "epoch": 0.1758716246697171,
+      "grad_norm": 0.009681077505664085,
+      "learning_rate": 9.157894736842105e-06,
+      "loss": 0.0001,
+      "step": 2729
+    },
+    {
+      "epoch": 0.17593607011664625,
+      "grad_norm": 0.005116961384670569,
+      "learning_rate": 9.15717866093806e-06,
+      "loss": 0.0,
+      "step": 2730
+    },
+    {
+      "epoch": 0.17600051556357543,
+      "grad_norm": 0.0266595739957785,
+      "learning_rate": 9.156462585034014e-06,
+      "loss": 0.0001,
+      "step": 2731
+    },
+    {
+      "epoch": 0.17606496101050462,
+      "grad_norm": 0.00023531374018092132,
+      "learning_rate": 9.155746509129968e-06,
+      "loss": 0.0,
+      "step": 2732
+    },
+    {
+      "epoch": 0.17612940645743377,
+      "grad_norm": 0.004908907718591901,
+      "learning_rate": 9.155030433225923e-06,
+      "loss": 0.0,
+      "step": 2733
+    },
+    {
+      "epoch": 0.17619385190436296,
+      "grad_norm": 0.0026669341550138548,
+      "learning_rate": 9.154314357321877e-06,
+      "loss": 0.0,
+      "step": 2734
+    },
+    {
+      "epoch": 0.17625829735129214,
+      "grad_norm": 0.024920427777757822,
+      "learning_rate": 9.153598281417831e-06,
+      "loss": 0.0001,
+      "step": 2735
+    },
+    {
+      "epoch": 0.1763227427982213,
+      "grad_norm": 0.0002157229860912765,
+      "learning_rate": 9.152882205513785e-06,
+      "loss": 0.0,
+      "step": 2736
+    },
+    {
+      "epoch": 0.17638718824515048,
+      "grad_norm": 0.003320399939081684,
+      "learning_rate": 9.15216612960974e-06,
+      "loss": 0.0,
+      "step": 2737
+    },
+    {
+      "epoch": 0.17645163369207967,
+      "grad_norm": 0.0021092644703280157,
+      "learning_rate": 9.151450053705694e-06,
+      "loss": 0.0,
+      "step": 2738
+    },
+    {
+      "epoch": 0.17651607913900882,
+      "grad_norm": 0.28572773821913267,
+      "learning_rate": 9.150733977801648e-06,
+      "loss": 0.0014,
+      "step": 2739
+    },
+    {
+      "epoch": 0.176580524585938,
+      "grad_norm": 0.007860353343515316,
+      "learning_rate": 9.150017901897603e-06,
+      "loss": 0.0,
+      "step": 2740
+    },
+    {
+      "epoch": 0.17664497003286717,
+      "grad_norm": 0.0074805734351772474,
+      "learning_rate": 9.149301825993557e-06,
+      "loss": 0.0001,
+      "step": 2741
+    },
+    {
+      "epoch": 0.17670941547979635,
+      "grad_norm": 0.00011039618974302341,
+      "learning_rate": 9.14858575008951e-06,
+      "loss": 0.0,
+      "step": 2742
+    },
+    {
+      "epoch": 0.17677386092672553,
+      "grad_norm": 0.0070548084467875975,
+      "learning_rate": 9.147869674185464e-06,
+      "loss": 0.0,
+      "step": 2743
+    },
+    {
+      "epoch": 0.1768383063736547,
+      "grad_norm": 0.010001704121558984,
+      "learning_rate": 9.147153598281418e-06,
+      "loss": 0.0,
+      "step": 2744
+    },
+    {
+      "epoch": 0.17690275182058388,
+      "grad_norm": 0.0019253162831213417,
+      "learning_rate": 9.146437522377372e-06,
+      "loss": 0.0,
+      "step": 2745
+    },
+    {
+      "epoch": 0.17696719726751306,
+      "grad_norm": 0.01767115100918315,
+      "learning_rate": 9.145721446473327e-06,
+      "loss": 0.0001,
+      "step": 2746
+    },
+    {
+      "epoch": 0.17703164271444222,
+      "grad_norm": 0.012779574033902407,
+      "learning_rate": 9.145005370569281e-06,
+      "loss": 0.0001,
+      "step": 2747
+    },
+    {
+      "epoch": 0.1770960881613714,
+      "grad_norm": 0.0015406080229237796,
+      "learning_rate": 9.144289294665235e-06,
+      "loss": 0.0,
+      "step": 2748
+    },
+    {
+      "epoch": 0.17716053360830059,
+      "grad_norm": 0.008029459760126534,
+      "learning_rate": 9.14357321876119e-06,
+      "loss": 0.0,
+      "step": 2749
+    },
+    {
+      "epoch": 0.17722497905522974,
+      "grad_norm": 0.007691325124199665,
+      "learning_rate": 9.142857142857144e-06,
+      "loss": 0.0,
+      "step": 2750
+    },
+    {
+      "epoch": 0.17728942450215893,
+      "grad_norm": 0.0007767056720782113,
+      "learning_rate": 9.142141066953098e-06,
+      "loss": 0.0,
+      "step": 2751
+    },
+    {
+      "epoch": 0.17735386994908808,
+      "grad_norm": 0.0006003394971469371,
+      "learning_rate": 9.14142499104905e-06,
+      "loss": 0.0,
+      "step": 2752
+    },
+    {
+      "epoch": 0.17741831539601727,
+      "grad_norm": 2.471258603101776,
+      "learning_rate": 9.140708915145005e-06,
+      "loss": 0.0177,
+      "step": 2753
+    },
+    {
+      "epoch": 0.17748276084294645,
+      "grad_norm": 0.022019753049725126,
+      "learning_rate": 9.13999283924096e-06,
+      "loss": 0.0001,
+      "step": 2754
+    },
+    {
+      "epoch": 0.1775472062898756,
+      "grad_norm": 0.42206641552426616,
+      "learning_rate": 9.139276763336914e-06,
+      "loss": 0.0028,
+      "step": 2755
+    },
+    {
+      "epoch": 0.1776116517368048,
+      "grad_norm": 0.0005502695640614431,
+      "learning_rate": 9.13856068743287e-06,
+      "loss": 0.0,
+      "step": 2756
+    },
+    {
+      "epoch": 0.17767609718373398,
+      "grad_norm": 0.01922958508482105,
+      "learning_rate": 9.137844611528824e-06,
+      "loss": 0.0,
+      "step": 2757
+    },
+    {
+      "epoch": 0.17774054263066313,
+      "grad_norm": 0.19847522624781685,
+      "learning_rate": 9.137128535624777e-06,
+      "loss": 0.0001,
+      "step": 2758
+    },
+    {
+      "epoch": 0.17780498807759232,
+      "grad_norm": 0.46543913466278036,
+      "learning_rate": 9.136412459720731e-06,
+      "loss": 0.0013,
+      "step": 2759
+    },
+    {
+      "epoch": 0.1778694335245215,
+      "grad_norm": 0.02188902143711718,
+      "learning_rate": 9.135696383816685e-06,
+      "loss": 0.0001,
+      "step": 2760
+    },
+    {
+      "epoch": 0.17793387897145066,
+      "grad_norm": 0.35200723492504954,
+      "learning_rate": 9.13498030791264e-06,
+      "loss": 0.0011,
+      "step": 2761
+    },
+    {
+      "epoch": 0.17799832441837984,
+      "grad_norm": 0.000621212590110667,
+      "learning_rate": 9.134264232008594e-06,
+      "loss": 0.0,
+      "step": 2762
+    },
+    {
+      "epoch": 0.17806276986530903,
+      "grad_norm": 0.0041229362491163645,
+      "learning_rate": 9.133548156104548e-06,
+      "loss": 0.0,
+      "step": 2763
+    },
+    {
+      "epoch": 0.17812721531223819,
+      "grad_norm": 0.1805057863321118,
+      "learning_rate": 9.132832080200502e-06,
+      "loss": 0.0005,
+      "step": 2764
+    },
+    {
+      "epoch": 0.17819166075916737,
+      "grad_norm": 0.04216859049336486,
+      "learning_rate": 9.132116004296457e-06,
+      "loss": 0.0005,
+      "step": 2765
+    },
+    {
+      "epoch": 0.17825610620609653,
+      "grad_norm": 0.007158825226122906,
+      "learning_rate": 9.131399928392411e-06,
+      "loss": 0.0,
+      "step": 2766
+    },
+    {
+      "epoch": 0.1783205516530257,
+      "grad_norm": 0.0007151610788651363,
+      "learning_rate": 9.130683852488365e-06,
+      "loss": 0.0,
+      "step": 2767
+    },
+    {
+      "epoch": 0.1783849970999549,
+      "grad_norm": 0.026471367081296403,
+      "learning_rate": 9.129967776584318e-06,
+      "loss": 0.0,
+      "step": 2768
+    },
+    {
+      "epoch": 0.17844944254688405,
+      "grad_norm": 0.028729129156446532,
+      "learning_rate": 9.129251700680272e-06,
+      "loss": 0.0001,
+      "step": 2769
+    },
+    {
+      "epoch": 0.17851388799381324,
+      "grad_norm": 0.05488665115941958,
+      "learning_rate": 9.128535624776226e-06,
+      "loss": 0.0001,
+      "step": 2770
+    },
+    {
+      "epoch": 0.17857833344074242,
+      "grad_norm": 0.06607381270845136,
+      "learning_rate": 9.12781954887218e-06,
+      "loss": 0.0002,
+      "step": 2771
+    },
+    {
+      "epoch": 0.17864277888767158,
+      "grad_norm": 0.04787404155176431,
+      "learning_rate": 9.127103472968135e-06,
+      "loss": 0.0001,
+      "step": 2772
+    },
+    {
+      "epoch": 0.17870722433460076,
+      "grad_norm": 0.10605978049032681,
+      "learning_rate": 9.12638739706409e-06,
+      "loss": 0.0003,
+      "step": 2773
+    },
+    {
+      "epoch": 0.17877166978152995,
+      "grad_norm": 0.0007944870144507999,
+      "learning_rate": 9.125671321160044e-06,
+      "loss": 0.0,
+      "step": 2774
+    },
+    {
+      "epoch": 0.1788361152284591,
+      "grad_norm": 0.00026383111149123895,
+      "learning_rate": 9.124955245255998e-06,
+      "loss": 0.0,
+      "step": 2775
+    },
+    {
+      "epoch": 0.1789005606753883,
+      "grad_norm": 0.00323695762220021,
+      "learning_rate": 9.124239169351952e-06,
+      "loss": 0.0,
+      "step": 2776
+    },
+    {
+      "epoch": 0.17896500612231747,
+      "grad_norm": 0.12140674287801527,
+      "learning_rate": 9.123523093447907e-06,
+      "loss": 0.0002,
+      "step": 2777
+    },
+    {
+      "epoch": 0.17902945156924663,
+      "grad_norm": 0.05398408619186501,
+      "learning_rate": 9.12280701754386e-06,
+      "loss": 0.0001,
+      "step": 2778
+    },
+    {
+      "epoch": 0.1790938970161758,
+      "grad_norm": 0.0087486923606527,
+      "learning_rate": 9.122090941639813e-06,
+      "loss": 0.0001,
+      "step": 2779
+    },
+    {
+      "epoch": 0.17915834246310497,
+      "grad_norm": 0.0002230895636351413,
+      "learning_rate": 9.12137486573577e-06,
+      "loss": 0.0,
+      "step": 2780
+    },
+    {
+      "epoch": 0.17922278791003415,
+      "grad_norm": 0.33086147690502815,
+      "learning_rate": 9.120658789831724e-06,
+      "loss": 0.0008,
+      "step": 2781
+    },
+    {
+      "epoch": 0.17928723335696334,
+      "grad_norm": 0.909170935370485,
+      "learning_rate": 9.119942713927678e-06,
+      "loss": 0.0027,
+      "step": 2782
+    },
+    {
+      "epoch": 0.1793516788038925,
+      "grad_norm": 0.0024790891799992814,
+      "learning_rate": 9.119226638023632e-06,
+      "loss": 0.0,
+      "step": 2783
+    },
+    {
+      "epoch": 0.17941612425082168,
+      "grad_norm": 0.10431613976366509,
+      "learning_rate": 9.118510562119585e-06,
+      "loss": 0.0004,
+      "step": 2784
+    },
+    {
+      "epoch": 0.17948056969775086,
+      "grad_norm": 0.012409698026434208,
+      "learning_rate": 9.11779448621554e-06,
+      "loss": 0.0,
+      "step": 2785
+    },
+    {
+      "epoch": 0.17954501514468002,
+      "grad_norm": 0.2690126294650971,
+      "learning_rate": 9.117078410311494e-06,
+      "loss": 0.0038,
+      "step": 2786
+    },
+    {
+      "epoch": 0.1796094605916092,
+      "grad_norm": 0.0033128166764828226,
+      "learning_rate": 9.116362334407448e-06,
+      "loss": 0.0,
+      "step": 2787
+    },
+    {
+      "epoch": 0.1796739060385384,
+      "grad_norm": 0.0009229370739567911,
+      "learning_rate": 9.115646258503402e-06,
+      "loss": 0.0,
+      "step": 2788
+    },
+    {
+      "epoch": 0.17973835148546755,
+      "grad_norm": 0.012281191385820282,
+      "learning_rate": 9.114930182599356e-06,
+      "loss": 0.0,
+      "step": 2789
+    },
+    {
+      "epoch": 0.17980279693239673,
+      "grad_norm": 0.0010946180994970097,
+      "learning_rate": 9.11421410669531e-06,
+      "loss": 0.0,
+      "step": 2790
+    },
+    {
+      "epoch": 0.1798672423793259,
+      "grad_norm": 0.0022886774893961163,
+      "learning_rate": 9.113498030791265e-06,
+      "loss": 0.0,
+      "step": 2791
+    },
+    {
+      "epoch": 0.17993168782625507,
+      "grad_norm": 0.18050407336452123,
+      "learning_rate": 9.11278195488722e-06,
+      "loss": 0.0004,
+      "step": 2792
+    },
+    {
+      "epoch": 0.17999613327318426,
+      "grad_norm": 0.00776095822730694,
+      "learning_rate": 9.112065878983174e-06,
+      "loss": 0.0,
+      "step": 2793
+    },
+    {
+      "epoch": 0.1800605787201134,
+      "grad_norm": 0.010017379040207797,
+      "learning_rate": 9.111349803079126e-06,
+      "loss": 0.0,
+      "step": 2794
+    },
+    {
+      "epoch": 0.1801250241670426,
+      "grad_norm": 0.06924958627733009,
+      "learning_rate": 9.11063372717508e-06,
+      "loss": 0.0003,
+      "step": 2795
+    },
+    {
+      "epoch": 0.18018946961397178,
+      "grad_norm": 0.2812996264226042,
+      "learning_rate": 9.109917651271035e-06,
+      "loss": 0.0004,
+      "step": 2796
+    },
+    {
+      "epoch": 0.18025391506090094,
+      "grad_norm": 0.0031109287427372747,
+      "learning_rate": 9.10920157536699e-06,
+      "loss": 0.0,
+      "step": 2797
+    },
+    {
+      "epoch": 0.18031836050783012,
+      "grad_norm": 0.005299891749785947,
+      "learning_rate": 9.108485499462943e-06,
+      "loss": 0.0,
+      "step": 2798
+    },
+    {
+      "epoch": 0.1803828059547593,
+      "grad_norm": 0.159811229772372,
+      "learning_rate": 9.107769423558898e-06,
+      "loss": 0.0094,
+      "step": 2799
+    },
+    {
+      "epoch": 0.18044725140168846,
+      "grad_norm": 0.3035861307779333,
+      "learning_rate": 9.107053347654852e-06,
+      "loss": 0.0011,
+      "step": 2800
+    },
+    {
+      "epoch": 0.18051169684861765,
+      "grad_norm": 0.10506831640956556,
+      "learning_rate": 9.106337271750806e-06,
+      "loss": 0.0003,
+      "step": 2801
+    },
+    {
+      "epoch": 0.18057614229554683,
+      "grad_norm": 0.06075173568196115,
+      "learning_rate": 9.10562119584676e-06,
+      "loss": 0.0001,
+      "step": 2802
+    },
+    {
+      "epoch": 0.180640587742476,
+      "grad_norm": 0.011675944604708841,
+      "learning_rate": 9.104905119942715e-06,
+      "loss": 0.0,
+      "step": 2803
+    },
+    {
+      "epoch": 0.18070503318940517,
+      "grad_norm": 0.017210122448086982,
+      "learning_rate": 9.10418904403867e-06,
+      "loss": 0.0001,
+      "step": 2804
+    },
+    {
+      "epoch": 0.18076947863633433,
+      "grad_norm": 0.0010844863697502473,
+      "learning_rate": 9.103472968134624e-06,
+      "loss": 0.0,
+      "step": 2805
+    },
+    {
+      "epoch": 0.18083392408326351,
+      "grad_norm": 0.03426975746377976,
+      "learning_rate": 9.102756892230578e-06,
+      "loss": 0.0,
+      "step": 2806
+    },
+    {
+      "epoch": 0.1808983695301927,
+      "grad_norm": 0.0514928133710475,
+      "learning_rate": 9.102040816326532e-06,
+      "loss": 0.0001,
+      "step": 2807
+    },
+    {
+      "epoch": 0.18096281497712186,
+      "grad_norm": 0.00034102055632519174,
+      "learning_rate": 9.101324740422486e-06,
+      "loss": 0.0,
+      "step": 2808
+    },
+    {
+      "epoch": 0.18102726042405104,
+      "grad_norm": 0.007900137649701421,
+      "learning_rate": 9.10060866451844e-06,
+      "loss": 0.0,
+      "step": 2809
+    },
+    {
+      "epoch": 0.18109170587098022,
+      "grad_norm": 0.0012440836350956035,
+      "learning_rate": 9.099892588614393e-06,
+      "loss": 0.0,
+      "step": 2810
+    },
+    {
+      "epoch": 0.18115615131790938,
+      "grad_norm": 0.0045855459787961725,
+      "learning_rate": 9.099176512710348e-06,
+      "loss": 0.0,
+      "step": 2811
+    },
+    {
+      "epoch": 0.18122059676483857,
+      "grad_norm": 0.1941159407275899,
+      "learning_rate": 9.098460436806302e-06,
+      "loss": 0.0007,
+      "step": 2812
+    },
+    {
+      "epoch": 0.18128504221176775,
+      "grad_norm": 0.025967662808676583,
+      "learning_rate": 9.097744360902256e-06,
+      "loss": 0.0001,
+      "step": 2813
+    },
+    {
+      "epoch": 0.1813494876586969,
+      "grad_norm": 0.0331286074506034,
+      "learning_rate": 9.09702828499821e-06,
+      "loss": 0.0001,
+      "step": 2814
+    },
+    {
+      "epoch": 0.1814139331056261,
+      "grad_norm": 0.08495830115250962,
+      "learning_rate": 9.096312209094165e-06,
+      "loss": 0.0022,
+      "step": 2815
+    },
+    {
+      "epoch": 0.18147837855255528,
+      "grad_norm": 0.00813836807265155,
+      "learning_rate": 9.095596133190119e-06,
+      "loss": 0.0,
+      "step": 2816
+    },
+    {
+      "epoch": 0.18154282399948443,
+      "grad_norm": 0.008071176914842531,
+      "learning_rate": 9.094880057286073e-06,
+      "loss": 0.0,
+      "step": 2817
+    },
+    {
+      "epoch": 0.18160726944641362,
+      "grad_norm": 0.03360537068790458,
+      "learning_rate": 9.094163981382028e-06,
+      "loss": 0.0,
+      "step": 2818
+    },
+    {
+      "epoch": 0.18167171489334277,
+      "grad_norm": 3.9930071502552744,
+      "learning_rate": 9.09344790547798e-06,
+      "loss": 0.0428,
+      "step": 2819
+    },
+    {
+      "epoch": 0.18173616034027196,
+      "grad_norm": 0.3067133750386272,
+      "learning_rate": 9.092731829573935e-06,
+      "loss": 0.0021,
+      "step": 2820
+    },
+    {
+      "epoch": 0.18180060578720114,
+      "grad_norm": 0.09692007100777007,
+      "learning_rate": 9.092015753669889e-06,
+      "loss": 0.0003,
+      "step": 2821
+    },
+    {
+      "epoch": 0.1818650512341303,
+      "grad_norm": 0.008326567539547143,
+      "learning_rate": 9.091299677765843e-06,
+      "loss": 0.0,
+      "step": 2822
+    },
+    {
+      "epoch": 0.18192949668105948,
+      "grad_norm": 0.09641887182616365,
+      "learning_rate": 9.090583601861798e-06,
+      "loss": 0.0011,
+      "step": 2823
+    },
+    {
+      "epoch": 0.18199394212798867,
+      "grad_norm": 0.007352684257062721,
+      "learning_rate": 9.089867525957752e-06,
+      "loss": 0.0,
+      "step": 2824
+    },
+    {
+      "epoch": 0.18205838757491782,
+      "grad_norm": 0.030274536320525708,
+      "learning_rate": 9.089151450053706e-06,
+      "loss": 0.0001,
+      "step": 2825
+    },
+    {
+      "epoch": 0.182122833021847,
+      "grad_norm": 0.0006278500122715883,
+      "learning_rate": 9.08843537414966e-06,
+      "loss": 0.0,
+      "step": 2826
+    },
+    {
+      "epoch": 0.1821872784687762,
+      "grad_norm": 0.0030603911056034895,
+      "learning_rate": 9.087719298245615e-06,
+      "loss": 0.0,
+      "step": 2827
+    },
+    {
+      "epoch": 0.18225172391570535,
+      "grad_norm": 0.0004505991303399384,
+      "learning_rate": 9.087003222341569e-06,
+      "loss": 0.0,
+      "step": 2828
+    },
+    {
+      "epoch": 0.18231616936263453,
+      "grad_norm": 0.017218593559380928,
+      "learning_rate": 9.086287146437523e-06,
+      "loss": 0.0001,
+      "step": 2829
+    },
+    {
+      "epoch": 0.18238061480956372,
+      "grad_norm": 0.0021886677003908977,
+      "learning_rate": 9.085571070533478e-06,
+      "loss": 0.0,
+      "step": 2830
+    },
+    {
+      "epoch": 0.18244506025649287,
+      "grad_norm": 0.0004787494491278347,
+      "learning_rate": 9.084854994629432e-06,
+      "loss": 0.0,
+      "step": 2831
+    },
+    {
+      "epoch": 0.18250950570342206,
+      "grad_norm": 0.0060502932594086204,
+      "learning_rate": 9.084138918725386e-06,
+      "loss": 0.0,
+      "step": 2832
+    },
+    {
+      "epoch": 0.18257395115035122,
+      "grad_norm": 0.00035090977070178326,
+      "learning_rate": 9.08342284282134e-06,
+      "loss": 0.0,
+      "step": 2833
+    },
+    {
+      "epoch": 0.1826383965972804,
+      "grad_norm": 0.12514816384262056,
+      "learning_rate": 9.082706766917295e-06,
+      "loss": 0.0004,
+      "step": 2834
+    },
+    {
+      "epoch": 0.18270284204420958,
+      "grad_norm": 0.3302171515942961,
+      "learning_rate": 9.081990691013247e-06,
+      "loss": 0.0015,
+      "step": 2835
+    },
+    {
+      "epoch": 0.18276728749113874,
+      "grad_norm": 0.012892377663801916,
+      "learning_rate": 9.081274615109202e-06,
+      "loss": 0.0001,
+      "step": 2836
+    },
+    {
+      "epoch": 0.18283173293806793,
+      "grad_norm": 0.1393849681809212,
+      "learning_rate": 9.080558539205156e-06,
+      "loss": 0.0001,
+      "step": 2837
+    },
+    {
+      "epoch": 0.1828961783849971,
+      "grad_norm": 0.004492926812105762,
+      "learning_rate": 9.07984246330111e-06,
+      "loss": 0.0,
+      "step": 2838
+    },
+    {
+      "epoch": 0.18296062383192627,
+      "grad_norm": 0.03560397375172943,
+      "learning_rate": 9.079126387397065e-06,
+      "loss": 0.0,
+      "step": 2839
+    },
+    {
+      "epoch": 0.18302506927885545,
+      "grad_norm": 0.32795772375842486,
+      "learning_rate": 9.078410311493019e-06,
+      "loss": 0.0007,
+      "step": 2840
+    },
+    {
+      "epoch": 0.18308951472578464,
+      "grad_norm": 0.01319255954466059,
+      "learning_rate": 9.077694235588973e-06,
+      "loss": 0.0,
+      "step": 2841
+    },
+    {
+      "epoch": 0.1831539601727138,
+      "grad_norm": 0.1890710931196515,
+      "learning_rate": 9.076978159684928e-06,
+      "loss": 0.0016,
+      "step": 2842
+    },
+    {
+      "epoch": 0.18321840561964298,
+      "grad_norm": 0.019405285126641924,
+      "learning_rate": 9.076262083780882e-06,
+      "loss": 0.0001,
+      "step": 2843
+    },
+    {
+      "epoch": 0.18328285106657213,
+      "grad_norm": 0.000751944682332039,
+      "learning_rate": 9.075546007876836e-06,
+      "loss": 0.0,
+      "step": 2844
+    },
+    {
+      "epoch": 0.18334729651350132,
+      "grad_norm": 0.3132701562643847,
+      "learning_rate": 9.074829931972789e-06,
+      "loss": 0.0006,
+      "step": 2845
+    },
+    {
+      "epoch": 0.1834117419604305,
+      "grad_norm": 0.000840455758111162,
+      "learning_rate": 9.074113856068743e-06,
+      "loss": 0.0,
+      "step": 2846
+    },
+    {
+      "epoch": 0.18347618740735966,
+      "grad_norm": 0.010120712128541575,
+      "learning_rate": 9.073397780164697e-06,
+      "loss": 0.0001,
+      "step": 2847
+    },
+    {
+      "epoch": 0.18354063285428884,
+      "grad_norm": 0.009966635973211107,
+      "learning_rate": 9.072681704260652e-06,
+      "loss": 0.0001,
+      "step": 2848
+    },
+    {
+      "epoch": 0.18360507830121803,
+      "grad_norm": 0.006565756711698041,
+      "learning_rate": 9.071965628356606e-06,
+      "loss": 0.0,
+      "step": 2849
+    },
+    {
+      "epoch": 0.18366952374814718,
+      "grad_norm": 0.0005553512116047497,
+      "learning_rate": 9.071249552452562e-06,
+      "loss": 0.0,
+      "step": 2850
+    },
+    {
+      "epoch": 0.18373396919507637,
+      "grad_norm": 0.00028353594343350657,
+      "learning_rate": 9.070533476548515e-06,
+      "loss": 0.0,
+      "step": 2851
+    },
+    {
+      "epoch": 0.18379841464200555,
+      "grad_norm": 0.20578208174425175,
+      "learning_rate": 9.069817400644469e-06,
+      "loss": 0.0014,
+      "step": 2852
+    },
+    {
+      "epoch": 0.1838628600889347,
+      "grad_norm": 0.04055908694973615,
+      "learning_rate": 9.069101324740423e-06,
+      "loss": 0.0002,
+      "step": 2853
+    },
+    {
+      "epoch": 0.1839273055358639,
+      "grad_norm": 0.06899265144614615,
+      "learning_rate": 9.068385248836377e-06,
+      "loss": 0.0007,
+      "step": 2854
+    },
+    {
+      "epoch": 0.18399175098279308,
+      "grad_norm": 0.0006986279102526963,
+      "learning_rate": 9.067669172932332e-06,
+      "loss": 0.0,
+      "step": 2855
+    },
+    {
+      "epoch": 0.18405619642972224,
+      "grad_norm": 0.00019407767512415275,
+      "learning_rate": 9.066953097028286e-06,
+      "loss": 0.0,
+      "step": 2856
+    },
+    {
+      "epoch": 0.18412064187665142,
+      "grad_norm": 0.15103341496607176,
+      "learning_rate": 9.06623702112424e-06,
+      "loss": 0.0008,
+      "step": 2857
+    },
+    {
+      "epoch": 0.18418508732358058,
+      "grad_norm": 0.12748822634696041,
+      "learning_rate": 9.065520945220195e-06,
+      "loss": 0.0002,
+      "step": 2858
+    },
+    {
+      "epoch": 0.18424953277050976,
+      "grad_norm": 0.010185844246403907,
+      "learning_rate": 9.064804869316149e-06,
+      "loss": 0.0,
+      "step": 2859
+    },
+    {
+      "epoch": 0.18431397821743895,
+      "grad_norm": 0.0005492418436434324,
+      "learning_rate": 9.064088793412103e-06,
+      "loss": 0.0,
+      "step": 2860
+    },
+    {
+      "epoch": 0.1843784236643681,
+      "grad_norm": 0.06054778444468528,
+      "learning_rate": 9.063372717508056e-06,
+      "loss": 0.0002,
+      "step": 2861
+    },
+    {
+      "epoch": 0.1844428691112973,
+      "grad_norm": 0.00015870494723684052,
+      "learning_rate": 9.06265664160401e-06,
+      "loss": 0.0,
+      "step": 2862
+    },
+    {
+      "epoch": 0.18450731455822647,
+      "grad_norm": 0.2531298430886071,
+      "learning_rate": 9.061940565699964e-06,
+      "loss": 0.0008,
+      "step": 2863
+    },
+    {
+      "epoch": 0.18457176000515563,
+      "grad_norm": 0.00035488257119721176,
+      "learning_rate": 9.061224489795919e-06,
+      "loss": 0.0,
+      "step": 2864
+    },
+    {
+      "epoch": 0.1846362054520848,
+      "grad_norm": 0.001592378038638373,
+      "learning_rate": 9.060508413891873e-06,
+      "loss": 0.0,
+      "step": 2865
+    },
+    {
+      "epoch": 0.184700650899014,
+      "grad_norm": 0.005239850862371177,
+      "learning_rate": 9.059792337987827e-06,
+      "loss": 0.0,
+      "step": 2866
+    },
+    {
+      "epoch": 0.18476509634594315,
+      "grad_norm": 0.0012565288853653319,
+      "learning_rate": 9.059076262083782e-06,
+      "loss": 0.0,
+      "step": 2867
+    },
+    {
+      "epoch": 0.18482954179287234,
+      "grad_norm": 0.003256885058237763,
+      "learning_rate": 9.058360186179736e-06,
+      "loss": 0.0,
+      "step": 2868
+    },
+    {
+      "epoch": 0.18489398723980152,
+      "grad_norm": 0.0013340724323056313,
+      "learning_rate": 9.05764411027569e-06,
+      "loss": 0.0,
+      "step": 2869
+    },
+    {
+      "epoch": 0.18495843268673068,
+      "grad_norm": 0.009623622364578288,
+      "learning_rate": 9.056928034371644e-06,
+      "loss": 0.0001,
+      "step": 2870
+    },
+    {
+      "epoch": 0.18502287813365986,
+      "grad_norm": 0.02283470135952233,
+      "learning_rate": 9.056211958467597e-06,
+      "loss": 0.0,
+      "step": 2871
+    },
+    {
+      "epoch": 0.18508732358058902,
+      "grad_norm": 3.1352344007944293,
+      "learning_rate": 9.055495882563551e-06,
+      "loss": 0.0312,
+      "step": 2872
+    },
+    {
+      "epoch": 0.1851517690275182,
+      "grad_norm": 0.1283916257810219,
+      "learning_rate": 9.054779806659507e-06,
+      "loss": 0.0022,
+      "step": 2873
+    },
+    {
+      "epoch": 0.1852162144744474,
+      "grad_norm": 0.0017243622373718673,
+      "learning_rate": 9.054063730755462e-06,
+      "loss": 0.0,
+      "step": 2874
+    },
+    {
+      "epoch": 0.18528065992137654,
+      "grad_norm": 0.02626256695329013,
+      "learning_rate": 9.053347654851416e-06,
+      "loss": 0.0001,
+      "step": 2875
+    },
+    {
+      "epoch": 0.18534510536830573,
+      "grad_norm": 1.8560834970519415,
+      "learning_rate": 9.05263157894737e-06,
+      "loss": 0.0055,
+      "step": 2876
+    },
+    {
+      "epoch": 0.1854095508152349,
+      "grad_norm": 0.22453528301728987,
+      "learning_rate": 9.051915503043323e-06,
+      "loss": 0.0003,
+      "step": 2877
+    },
+    {
+      "epoch": 0.18547399626216407,
+      "grad_norm": 0.01902833171278348,
+      "learning_rate": 9.051199427139277e-06,
+      "loss": 0.0001,
+      "step": 2878
+    },
+    {
+      "epoch": 0.18553844170909325,
+      "grad_norm": 0.0015275143263481297,
+      "learning_rate": 9.050483351235231e-06,
+      "loss": 0.0,
+      "step": 2879
+    },
+    {
+      "epoch": 0.18560288715602244,
+      "grad_norm": 0.0001940454087729942,
+      "learning_rate": 9.049767275331186e-06,
+      "loss": 0.0,
+      "step": 2880
+    },
+    {
+      "epoch": 0.1856673326029516,
+      "grad_norm": 0.00048513116384772117,
+      "learning_rate": 9.04905119942714e-06,
+      "loss": 0.0,
+      "step": 2881
+    },
+    {
+      "epoch": 0.18573177804988078,
+      "grad_norm": 0.0037608151201876908,
+      "learning_rate": 9.048335123523094e-06,
+      "loss": 0.0,
+      "step": 2882
+    },
+    {
+      "epoch": 0.18579622349680994,
+      "grad_norm": 1.1987838702145488,
+      "learning_rate": 9.047619047619049e-06,
+      "loss": 0.001,
+      "step": 2883
+    },
+    {
+      "epoch": 0.18586066894373912,
+      "grad_norm": 0.010770438174594095,
+      "learning_rate": 9.046902971715003e-06,
+      "loss": 0.0001,
+      "step": 2884
+    },
+    {
+      "epoch": 0.1859251143906683,
+      "grad_norm": 0.003791891232412315,
+      "learning_rate": 9.046186895810957e-06,
+      "loss": 0.0,
+      "step": 2885
+    },
+    {
+      "epoch": 0.18598955983759746,
+      "grad_norm": 1.4171702761376253,
+      "learning_rate": 9.045470819906912e-06,
+      "loss": 0.0036,
+      "step": 2886
+    },
+    {
+      "epoch": 0.18605400528452665,
+      "grad_norm": 0.013678521963010951,
+      "learning_rate": 9.044754744002864e-06,
+      "loss": 0.0,
+      "step": 2887
+    },
+    {
+      "epoch": 0.18611845073145583,
+      "grad_norm": 0.2775792210281082,
+      "learning_rate": 9.044038668098818e-06,
+      "loss": 0.0021,
+      "step": 2888
+    },
+    {
+      "epoch": 0.186182896178385,
+      "grad_norm": 0.021857381142348316,
+      "learning_rate": 9.043322592194773e-06,
+      "loss": 0.0001,
+      "step": 2889
+    },
+    {
+      "epoch": 0.18624734162531417,
+      "grad_norm": 0.10352626332013336,
+      "learning_rate": 9.042606516290727e-06,
+      "loss": 0.0002,
+      "step": 2890
+    },
+    {
+      "epoch": 0.18631178707224336,
+      "grad_norm": 0.043887566667352385,
+      "learning_rate": 9.041890440386681e-06,
+      "loss": 0.0001,
+      "step": 2891
+    },
+    {
+      "epoch": 0.1863762325191725,
+      "grad_norm": 0.00426374508594024,
+      "learning_rate": 9.041174364482636e-06,
+      "loss": 0.0,
+      "step": 2892
+    },
+    {
+      "epoch": 0.1864406779661017,
+      "grad_norm": 0.0017327150241980725,
+      "learning_rate": 9.04045828857859e-06,
+      "loss": 0.0,
+      "step": 2893
+    },
+    {
+      "epoch": 0.18650512341303088,
+      "grad_norm": 0.16207233258521805,
+      "learning_rate": 9.039742212674544e-06,
+      "loss": 0.0003,
+      "step": 2894
+    },
+    {
+      "epoch": 0.18656956885996004,
+      "grad_norm": 0.08029956264882238,
+      "learning_rate": 9.039026136770499e-06,
+      "loss": 0.0002,
+      "step": 2895
+    },
+    {
+      "epoch": 0.18663401430688922,
+      "grad_norm": 0.002952531018276622,
+      "learning_rate": 9.038310060866453e-06,
+      "loss": 0.0,
+      "step": 2896
+    },
+    {
+      "epoch": 0.18669845975381838,
+      "grad_norm": 0.001077306109601978,
+      "learning_rate": 9.037593984962407e-06,
+      "loss": 0.0,
+      "step": 2897
+    },
+    {
+      "epoch": 0.18676290520074756,
+      "grad_norm": 0.5327763508814157,
+      "learning_rate": 9.036877909058361e-06,
+      "loss": 0.0018,
+      "step": 2898
+    },
+    {
+      "epoch": 0.18682735064767675,
+      "grad_norm": 0.0011653264573898313,
+      "learning_rate": 9.036161833154316e-06,
+      "loss": 0.0,
+      "step": 2899
+    },
+    {
+      "epoch": 0.1868917960946059,
+      "grad_norm": 0.010833147865541404,
+      "learning_rate": 9.03544575725027e-06,
+      "loss": 0.0001,
+      "step": 2900
+    },
+    {
+      "epoch": 0.1869562415415351,
+      "grad_norm": 0.0035407077531820256,
+      "learning_rate": 9.034729681346224e-06,
+      "loss": 0.0,
+      "step": 2901
+    },
+    {
+      "epoch": 0.18702068698846427,
+      "grad_norm": 0.19864257068351776,
+      "learning_rate": 9.034013605442179e-06,
+      "loss": 0.0003,
+      "step": 2902
+    },
+    {
+      "epoch": 0.18708513243539343,
+      "grad_norm": 0.5965458646506104,
+      "learning_rate": 9.033297529538131e-06,
+      "loss": 0.0033,
+      "step": 2903
+    },
+    {
+      "epoch": 0.18714957788232262,
+      "grad_norm": 0.0016764956381874533,
+      "learning_rate": 9.032581453634086e-06,
+      "loss": 0.0,
+      "step": 2904
+    },
+    {
+      "epoch": 0.1872140233292518,
+      "grad_norm": 0.3784683095611159,
+      "learning_rate": 9.03186537773004e-06,
+      "loss": 0.0007,
+      "step": 2905
+    },
+    {
+      "epoch": 0.18727846877618096,
+      "grad_norm": 0.005355622180015852,
+      "learning_rate": 9.031149301825994e-06,
+      "loss": 0.0,
+      "step": 2906
+    },
+    {
+      "epoch": 0.18734291422311014,
+      "grad_norm": 0.1916487189541943,
+      "learning_rate": 9.030433225921948e-06,
+      "loss": 0.0003,
+      "step": 2907
+    },
+    {
+      "epoch": 0.18740735967003933,
+      "grad_norm": 0.08662085151423458,
+      "learning_rate": 9.029717150017903e-06,
+      "loss": 0.0006,
+      "step": 2908
+    },
+    {
+      "epoch": 0.18747180511696848,
+      "grad_norm": 0.1926151557278316,
+      "learning_rate": 9.029001074113857e-06,
+      "loss": 0.0004,
+      "step": 2909
+    },
+    {
+      "epoch": 0.18753625056389767,
+      "grad_norm": 0.3587466842551305,
+      "learning_rate": 9.028284998209811e-06,
+      "loss": 0.003,
+      "step": 2910
+    },
+    {
+      "epoch": 0.18760069601082682,
+      "grad_norm": 0.01522610071722062,
+      "learning_rate": 9.027568922305766e-06,
+      "loss": 0.0001,
+      "step": 2911
+    },
+    {
+      "epoch": 0.187665141457756,
+      "grad_norm": 0.09025906889595814,
+      "learning_rate": 9.026852846401718e-06,
+      "loss": 0.001,
+      "step": 2912
+    },
+    {
+      "epoch": 0.1877295869046852,
+      "grad_norm": 0.0005104981904117869,
+      "learning_rate": 9.026136770497673e-06,
+      "loss": 0.0,
+      "step": 2913
+    },
+    {
+      "epoch": 0.18779403235161435,
+      "grad_norm": 0.009975611318601715,
+      "learning_rate": 9.025420694593627e-06,
+      "loss": 0.0,
+      "step": 2914
+    },
+    {
+      "epoch": 0.18785847779854353,
+      "grad_norm": 0.04623598655275114,
+      "learning_rate": 9.024704618689581e-06,
+      "loss": 0.0002,
+      "step": 2915
+    },
+    {
+      "epoch": 0.18792292324547272,
+      "grad_norm": 0.003555617732486728,
+      "learning_rate": 9.023988542785535e-06,
+      "loss": 0.0,
+      "step": 2916
+    },
+    {
+      "epoch": 0.18798736869240187,
+      "grad_norm": 0.018560679936225576,
+      "learning_rate": 9.02327246688149e-06,
+      "loss": 0.0,
+      "step": 2917
+    },
+    {
+      "epoch": 0.18805181413933106,
+      "grad_norm": 0.43325214848147997,
+      "learning_rate": 9.022556390977444e-06,
+      "loss": 0.0021,
+      "step": 2918
+    },
+    {
+      "epoch": 0.18811625958626024,
+      "grad_norm": 0.07998140821266131,
+      "learning_rate": 9.021840315073398e-06,
+      "loss": 0.0017,
+      "step": 2919
+    },
+    {
+      "epoch": 0.1881807050331894,
+      "grad_norm": 0.07237012820470437,
+      "learning_rate": 9.021124239169353e-06,
+      "loss": 0.0002,
+      "step": 2920
+    },
+    {
+      "epoch": 0.18824515048011858,
+      "grad_norm": 0.000759668939913165,
+      "learning_rate": 9.020408163265307e-06,
+      "loss": 0.0,
+      "step": 2921
+    },
+    {
+      "epoch": 0.18830959592704774,
+      "grad_norm": 0.18338005331275695,
+      "learning_rate": 9.019692087361261e-06,
+      "loss": 0.0006,
+      "step": 2922
+    },
+    {
+      "epoch": 0.18837404137397692,
+      "grad_norm": 0.2098501933593441,
+      "learning_rate": 9.018976011457216e-06,
+      "loss": 0.0005,
+      "step": 2923
+    },
+    {
+      "epoch": 0.1884384868209061,
+      "grad_norm": 0.0018100893660102637,
+      "learning_rate": 9.01825993555317e-06,
+      "loss": 0.0,
+      "step": 2924
+    },
+    {
+      "epoch": 0.18850293226783527,
+      "grad_norm": 0.0450810406327063,
+      "learning_rate": 9.017543859649124e-06,
+      "loss": 0.0001,
+      "step": 2925
+    },
+    {
+      "epoch": 0.18856737771476445,
+      "grad_norm": 0.014848699673825443,
+      "learning_rate": 9.016827783745078e-06,
+      "loss": 0.0,
+      "step": 2926
+    },
+    {
+      "epoch": 0.18863182316169363,
+      "grad_norm": 0.015089833418442411,
+      "learning_rate": 9.016111707841033e-06,
+      "loss": 0.0,
+      "step": 2927
+    },
+    {
+      "epoch": 0.1886962686086228,
+      "grad_norm": 0.37162824586463794,
+      "learning_rate": 9.015395631936985e-06,
+      "loss": 0.003,
+      "step": 2928
+    },
+    {
+      "epoch": 0.18876071405555198,
+      "grad_norm": 0.00679900766430204,
+      "learning_rate": 9.01467955603294e-06,
+      "loss": 0.0,
+      "step": 2929
+    },
+    {
+      "epoch": 0.18882515950248116,
+      "grad_norm": 0.19478660354221924,
+      "learning_rate": 9.013963480128894e-06,
+      "loss": 0.0007,
+      "step": 2930
+    },
+    {
+      "epoch": 0.18888960494941032,
+      "grad_norm": 0.08228354880526026,
+      "learning_rate": 9.013247404224848e-06,
+      "loss": 0.0003,
+      "step": 2931
+    },
+    {
+      "epoch": 0.1889540503963395,
+      "grad_norm": 0.07801816609421663,
+      "learning_rate": 9.012531328320803e-06,
+      "loss": 0.0002,
+      "step": 2932
+    },
+    {
+      "epoch": 0.18901849584326869,
+      "grad_norm": 0.015425874953560122,
+      "learning_rate": 9.011815252416757e-06,
+      "loss": 0.0,
+      "step": 2933
+    },
+    {
+      "epoch": 0.18908294129019784,
+      "grad_norm": 0.12362967331985299,
+      "learning_rate": 9.011099176512711e-06,
+      "loss": 0.0001,
+      "step": 2934
+    },
+    {
+      "epoch": 0.18914738673712703,
+      "grad_norm": 0.002316840415064825,
+      "learning_rate": 9.010383100608665e-06,
+      "loss": 0.0,
+      "step": 2935
+    },
+    {
+      "epoch": 0.18921183218405618,
+      "grad_norm": 0.0053035649105632595,
+      "learning_rate": 9.00966702470462e-06,
+      "loss": 0.0,
+      "step": 2936
+    },
+    {
+      "epoch": 0.18927627763098537,
+      "grad_norm": 0.016294097918252204,
+      "learning_rate": 9.008950948800574e-06,
+      "loss": 0.0,
+      "step": 2937
+    },
+    {
+      "epoch": 0.18934072307791455,
+      "grad_norm": 0.008288165719355206,
+      "learning_rate": 9.008234872896527e-06,
+      "loss": 0.0,
+      "step": 2938
+    },
+    {
+      "epoch": 0.1894051685248437,
+      "grad_norm": 0.19698208553230082,
+      "learning_rate": 9.007518796992481e-06,
+      "loss": 0.0014,
+      "step": 2939
+    },
+    {
+      "epoch": 0.1894696139717729,
+      "grad_norm": 0.003667401671959796,
+      "learning_rate": 9.006802721088435e-06,
+      "loss": 0.0,
+      "step": 2940
+    },
+    {
+      "epoch": 0.18953405941870208,
+      "grad_norm": 0.002796827125301907,
+      "learning_rate": 9.00608664518439e-06,
+      "loss": 0.0,
+      "step": 2941
+    },
+    {
+      "epoch": 0.18959850486563123,
+      "grad_norm": 0.3992818972816783,
+      "learning_rate": 9.005370569280344e-06,
+      "loss": 0.0002,
+      "step": 2942
+    },
+    {
+      "epoch": 0.18966295031256042,
+      "grad_norm": 0.05893500611468803,
+      "learning_rate": 9.0046544933763e-06,
+      "loss": 0.0004,
+      "step": 2943
+    },
+    {
+      "epoch": 0.1897273957594896,
+      "grad_norm": 0.002527947056167966,
+      "learning_rate": 9.003938417472252e-06,
+      "loss": 0.0,
+      "step": 2944
+    },
+    {
+      "epoch": 0.18979184120641876,
+      "grad_norm": 0.03283716045698447,
+      "learning_rate": 9.003222341568207e-06,
+      "loss": 0.0001,
+      "step": 2945
+    },
+    {
+      "epoch": 0.18985628665334794,
+      "grad_norm": 0.003239810900464001,
+      "learning_rate": 9.002506265664161e-06,
+      "loss": 0.0,
+      "step": 2946
+    },
+    {
+      "epoch": 0.18992073210027713,
+      "grad_norm": 0.034203012205389266,
+      "learning_rate": 9.001790189760115e-06,
+      "loss": 0.0003,
+      "step": 2947
+    },
+    {
+      "epoch": 0.18998517754720629,
+      "grad_norm": 0.01083160668170668,
+      "learning_rate": 9.00107411385607e-06,
+      "loss": 0.0,
+      "step": 2948
+    },
+    {
+      "epoch": 0.19004962299413547,
+      "grad_norm": 0.007940915211813882,
+      "learning_rate": 9.000358037952024e-06,
+      "loss": 0.0,
+      "step": 2949
+    },
+    {
+      "epoch": 0.19011406844106463,
+      "grad_norm": 0.1084319956272911,
+      "learning_rate": 8.999641962047978e-06,
+      "loss": 0.0008,
+      "step": 2950
+    },
+    {
+      "epoch": 0.1901785138879938,
+      "grad_norm": 0.01166456268566552,
+      "learning_rate": 8.998925886143933e-06,
+      "loss": 0.0,
+      "step": 2951
+    },
+    {
+      "epoch": 0.190242959334923,
+      "grad_norm": 0.009255105740318771,
+      "learning_rate": 8.998209810239887e-06,
+      "loss": 0.0,
+      "step": 2952
+    },
+    {
+      "epoch": 0.19030740478185215,
+      "grad_norm": 0.046569376752111036,
+      "learning_rate": 8.997493734335841e-06,
+      "loss": 0.0001,
+      "step": 2953
+    },
+    {
+      "epoch": 0.19037185022878134,
+      "grad_norm": 0.01405700880200295,
+      "learning_rate": 8.996777658431794e-06,
+      "loss": 0.0001,
+      "step": 2954
+    },
+    {
+      "epoch": 0.19043629567571052,
+      "grad_norm": 0.01268672480723258,
+      "learning_rate": 8.996061582527748e-06,
+      "loss": 0.0,
+      "step": 2955
+    },
+    {
+      "epoch": 0.19050074112263968,
+      "grad_norm": 0.002450229581765563,
+      "learning_rate": 8.995345506623702e-06,
+      "loss": 0.0,
+      "step": 2956
+    },
+    {
+      "epoch": 0.19056518656956886,
+      "grad_norm": 0.0003489092380548279,
+      "learning_rate": 8.994629430719657e-06,
+      "loss": 0.0,
+      "step": 2957
+    },
+    {
+      "epoch": 0.19062963201649805,
+      "grad_norm": 0.0029454670887810084,
+      "learning_rate": 8.993913354815611e-06,
+      "loss": 0.0,
+      "step": 2958
+    },
+    {
+      "epoch": 0.1906940774634272,
+      "grad_norm": 0.1718268218574252,
+      "learning_rate": 8.993197278911565e-06,
+      "loss": 0.0018,
+      "step": 2959
+    },
+    {
+      "epoch": 0.1907585229103564,
+      "grad_norm": 0.11376702629245902,
+      "learning_rate": 8.99248120300752e-06,
+      "loss": 0.0003,
+      "step": 2960
+    },
+    {
+      "epoch": 0.19082296835728554,
+      "grad_norm": 0.0007795738133778679,
+      "learning_rate": 8.991765127103474e-06,
+      "loss": 0.0,
+      "step": 2961
+    },
+    {
+      "epoch": 0.19088741380421473,
+      "grad_norm": 1.4979985713009711,
+      "learning_rate": 8.991049051199428e-06,
+      "loss": 0.0109,
+      "step": 2962
+    },
+    {
+      "epoch": 0.1909518592511439,
+      "grad_norm": 0.8103406896663958,
+      "learning_rate": 8.990332975295382e-06,
+      "loss": 0.0002,
+      "step": 2963
+    },
+    {
+      "epoch": 0.19101630469807307,
+      "grad_norm": 0.02842256055845789,
+      "learning_rate": 8.989616899391335e-06,
+      "loss": 0.0001,
+      "step": 2964
+    },
+    {
+      "epoch": 0.19108075014500225,
+      "grad_norm": 0.020678844417695182,
+      "learning_rate": 8.98890082348729e-06,
+      "loss": 0.0002,
+      "step": 2965
+    },
+    {
+      "epoch": 0.19114519559193144,
+      "grad_norm": 0.0028593214061483157,
+      "learning_rate": 8.988184747583245e-06,
+      "loss": 0.0,
+      "step": 2966
+    },
+    {
+      "epoch": 0.1912096410388606,
+      "grad_norm": 0.009822979528012175,
+      "learning_rate": 8.9874686716792e-06,
+      "loss": 0.0,
+      "step": 2967
+    },
+    {
+      "epoch": 0.19127408648578978,
+      "grad_norm": 0.022687103734926797,
+      "learning_rate": 8.986752595775154e-06,
+      "loss": 0.0001,
+      "step": 2968
+    },
+    {
+      "epoch": 0.19133853193271896,
+      "grad_norm": 0.06149599999979963,
+      "learning_rate": 8.986036519871108e-06,
+      "loss": 0.0002,
+      "step": 2969
+    },
+    {
+      "epoch": 0.19140297737964812,
+      "grad_norm": 1.2065422771333003,
+      "learning_rate": 8.98532044396706e-06,
+      "loss": 0.0034,
+      "step": 2970
+    },
+    {
+      "epoch": 0.1914674228265773,
+      "grad_norm": 0.018639562538012196,
+      "learning_rate": 8.984604368063015e-06,
+      "loss": 0.0,
+      "step": 2971
+    },
+    {
+      "epoch": 0.1915318682735065,
+      "grad_norm": 0.01474286627896713,
+      "learning_rate": 8.98388829215897e-06,
+      "loss": 0.0001,
+      "step": 2972
+    },
+    {
+      "epoch": 0.19159631372043565,
+      "grad_norm": 1.5631704036642082,
+      "learning_rate": 8.983172216254924e-06,
+      "loss": 0.0107,
+      "step": 2973
+    },
+    {
+      "epoch": 0.19166075916736483,
+      "grad_norm": 0.00574327599457403,
+      "learning_rate": 8.982456140350878e-06,
+      "loss": 0.0001,
+      "step": 2974
+    },
+    {
+      "epoch": 0.191725204614294,
+      "grad_norm": 0.23094776449504884,
+      "learning_rate": 8.981740064446832e-06,
+      "loss": 0.0028,
+      "step": 2975
+    },
+    {
+      "epoch": 0.19178965006122317,
+      "grad_norm": 0.0028454445462574494,
+      "learning_rate": 8.981023988542787e-06,
+      "loss": 0.0,
+      "step": 2976
+    },
+    {
+      "epoch": 0.19185409550815236,
+      "grad_norm": 0.004666714035464535,
+      "learning_rate": 8.980307912638741e-06,
+      "loss": 0.0,
+      "step": 2977
+    },
+    {
+      "epoch": 0.1919185409550815,
+      "grad_norm": 0.004573597119030054,
+      "learning_rate": 8.979591836734695e-06,
+      "loss": 0.0,
+      "step": 2978
+    },
+    {
+      "epoch": 0.1919829864020107,
+      "grad_norm": 0.011798074217064835,
+      "learning_rate": 8.97887576083065e-06,
+      "loss": 0.0,
+      "step": 2979
+    },
+    {
+      "epoch": 0.19204743184893988,
+      "grad_norm": 0.0004367517808790769,
+      "learning_rate": 8.978159684926602e-06,
+      "loss": 0.0,
+      "step": 2980
+    },
+    {
+      "epoch": 0.19211187729586904,
+      "grad_norm": 0.0033874185502569326,
+      "learning_rate": 8.977443609022556e-06,
+      "loss": 0.0,
+      "step": 2981
+    },
+    {
+      "epoch": 0.19217632274279822,
+      "grad_norm": 0.003283052887576775,
+      "learning_rate": 8.97672753311851e-06,
+      "loss": 0.0,
+      "step": 2982
+    },
+    {
+      "epoch": 0.1922407681897274,
+      "grad_norm": 0.19926155789775196,
+      "learning_rate": 8.976011457214465e-06,
+      "loss": 0.0006,
+      "step": 2983
+    },
+    {
+      "epoch": 0.19230521363665656,
+      "grad_norm": 0.004192863160750905,
+      "learning_rate": 8.97529538131042e-06,
+      "loss": 0.0,
+      "step": 2984
+    },
+    {
+      "epoch": 0.19236965908358575,
+      "grad_norm": 0.00110007929077268,
+      "learning_rate": 8.974579305406374e-06,
+      "loss": 0.0,
+      "step": 2985
+    },
+    {
+      "epoch": 0.19243410453051493,
+      "grad_norm": 0.039425082084022116,
+      "learning_rate": 8.973863229502328e-06,
+      "loss": 0.0001,
+      "step": 2986
+    },
+    {
+      "epoch": 0.1924985499774441,
+      "grad_norm": 0.00043550038157316295,
+      "learning_rate": 8.973147153598282e-06,
+      "loss": 0.0,
+      "step": 2987
+    },
+    {
+      "epoch": 0.19256299542437327,
+      "grad_norm": 0.05183727983491678,
+      "learning_rate": 8.972431077694236e-06,
+      "loss": 0.0016,
+      "step": 2988
+    },
+    {
+      "epoch": 0.19262744087130243,
+      "grad_norm": 0.15039922020860932,
+      "learning_rate": 8.97171500179019e-06,
+      "loss": 0.0019,
+      "step": 2989
+    },
+    {
+      "epoch": 0.19269188631823161,
+      "grad_norm": 0.0006762037012585999,
+      "learning_rate": 8.970998925886145e-06,
+      "loss": 0.0,
+      "step": 2990
+    },
+    {
+      "epoch": 0.1927563317651608,
+      "grad_norm": 0.10825269799498864,
+      "learning_rate": 8.9702828499821e-06,
+      "loss": 0.0011,
+      "step": 2991
+    },
+    {
+      "epoch": 0.19282077721208996,
+      "grad_norm": 0.3270196347969351,
+      "learning_rate": 8.969566774078054e-06,
+      "loss": 0.0011,
+      "step": 2992
+    },
+    {
+      "epoch": 0.19288522265901914,
+      "grad_norm": 0.007802911038854596,
+      "learning_rate": 8.968850698174008e-06,
+      "loss": 0.0,
+      "step": 2993
+    },
+    {
+      "epoch": 0.19294966810594832,
+      "grad_norm": 0.0008345035135385656,
+      "learning_rate": 8.968134622269962e-06,
+      "loss": 0.0,
+      "step": 2994
+    },
+    {
+      "epoch": 0.19301411355287748,
+      "grad_norm": 0.08876152359166628,
+      "learning_rate": 8.967418546365917e-06,
+      "loss": 0.0001,
+      "step": 2995
+    },
+    {
+      "epoch": 0.19307855899980667,
+      "grad_norm": 0.1668909055741231,
+      "learning_rate": 8.96670247046187e-06,
+      "loss": 0.0005,
+      "step": 2996
+    },
+    {
+      "epoch": 0.19314300444673585,
+      "grad_norm": 0.060401569099000595,
+      "learning_rate": 8.965986394557823e-06,
+      "loss": 0.0003,
+      "step": 2997
+    },
+    {
+      "epoch": 0.193207449893665,
+      "grad_norm": 0.0018142782884662798,
+      "learning_rate": 8.965270318653778e-06,
+      "loss": 0.0,
+      "step": 2998
+    },
+    {
+      "epoch": 0.1932718953405942,
+      "grad_norm": 0.01917432223078058,
+      "learning_rate": 8.964554242749732e-06,
+      "loss": 0.0001,
+      "step": 2999
+    },
+    {
+      "epoch": 0.19333634078752338,
+      "grad_norm": 0.00013875258473411826,
+      "learning_rate": 8.963838166845686e-06,
+      "loss": 0.0,
+      "step": 3000
+    },
+    {
+      "epoch": 0.19340078623445253,
+      "grad_norm": 0.11576714445206707,
+      "learning_rate": 8.96312209094164e-06,
+      "loss": 0.0028,
+      "step": 3001
+    },
+    {
+      "epoch": 0.19346523168138172,
+      "grad_norm": 0.6482484840760784,
+      "learning_rate": 8.962406015037595e-06,
+      "loss": 0.0014,
+      "step": 3002
+    },
+    {
+      "epoch": 0.19352967712831087,
+      "grad_norm": 0.6482484840760784,
+      "learning_rate": 8.962406015037595e-06,
+      "loss": 0.001,
+      "step": 3003
+    },
+    {
+      "epoch": 0.19359412257524006,
+      "grad_norm": 0.0007944590031801373,
+      "learning_rate": 8.96168993913355e-06,
+      "loss": 0.0,
+      "step": 3004
+    },
+    {
+      "epoch": 0.19365856802216924,
+      "grad_norm": 0.003729115719289283,
+      "learning_rate": 8.960973863229504e-06,
+      "loss": 0.0,
+      "step": 3005
+    },
+    {
+      "epoch": 0.1937230134690984,
+      "grad_norm": 0.0008843086516153837,
+      "learning_rate": 8.960257787325458e-06,
+      "loss": 0.0,
+      "step": 3006
+    },
+    {
+      "epoch": 0.19378745891602758,
+      "grad_norm": 0.21990022366392237,
+      "learning_rate": 8.95954171142141e-06,
+      "loss": 0.0021,
+      "step": 3007
+    },
+    {
+      "epoch": 0.19385190436295677,
+      "grad_norm": 0.15541553592994967,
+      "learning_rate": 8.958825635517365e-06,
+      "loss": 0.0018,
+      "step": 3008
+    },
+    {
+      "epoch": 0.19391634980988592,
+      "grad_norm": 0.02511120376754912,
+      "learning_rate": 8.958109559613319e-06,
+      "loss": 0.0001,
+      "step": 3009
+    },
+    {
+      "epoch": 0.1939807952568151,
+      "grad_norm": 0.001173436761283878,
+      "learning_rate": 8.957393483709273e-06,
+      "loss": 0.0,
+      "step": 3010
+    },
+    {
+      "epoch": 0.1940452407037443,
+      "grad_norm": 0.014118849072544383,
+      "learning_rate": 8.956677407805228e-06,
+      "loss": 0.0001,
+      "step": 3011
+    },
+    {
+      "epoch": 0.19410968615067345,
+      "grad_norm": 0.02563359621196255,
+      "learning_rate": 8.955961331901182e-06,
+      "loss": 0.0015,
+      "step": 3012
+    },
+    {
+      "epoch": 0.19417413159760263,
+      "grad_norm": 0.0006445896627271484,
+      "learning_rate": 8.955245255997136e-06,
+      "loss": 0.0,
+      "step": 3013
+    },
+    {
+      "epoch": 0.1942385770445318,
+      "grad_norm": 0.008495704221291287,
+      "learning_rate": 8.95452918009309e-06,
+      "loss": 0.0001,
+      "step": 3014
+    },
+    {
+      "epoch": 0.19430302249146097,
+      "grad_norm": 0.9487614727206138,
+      "learning_rate": 8.953813104189045e-06,
+      "loss": 0.0081,
+      "step": 3015
+    },
+    {
+      "epoch": 0.19436746793839016,
+      "grad_norm": 0.0034573415451793093,
+      "learning_rate": 8.953097028285e-06,
+      "loss": 0.0001,
+      "step": 3016
+    },
+    {
+      "epoch": 0.19443191338531932,
+      "grad_norm": 0.024194100837835554,
+      "learning_rate": 8.952380952380953e-06,
+      "loss": 0.0002,
+      "step": 3017
+    },
+    {
+      "epoch": 0.1944963588322485,
+      "grad_norm": 0.08266217977115524,
+      "learning_rate": 8.951664876476908e-06,
+      "loss": 0.0002,
+      "step": 3018
+    },
+    {
+      "epoch": 0.19456080427917768,
+      "grad_norm": 0.05253149672988689,
+      "learning_rate": 8.950948800572862e-06,
+      "loss": 0.0001,
+      "step": 3019
+    },
+    {
+      "epoch": 0.19462524972610684,
+      "grad_norm": 0.26598516057325405,
+      "learning_rate": 8.950232724668816e-06,
+      "loss": 0.0012,
+      "step": 3020
+    },
+    {
+      "epoch": 0.19468969517303603,
+      "grad_norm": 0.001171713922511211,
+      "learning_rate": 8.94951664876477e-06,
+      "loss": 0.0,
+      "step": 3021
+    },
+    {
+      "epoch": 0.1947541406199652,
+      "grad_norm": 0.006140044099780373,
+      "learning_rate": 8.948800572860725e-06,
+      "loss": 0.0,
+      "step": 3022
+    },
+    {
+      "epoch": 0.19481858606689437,
+      "grad_norm": 0.17568814673045194,
+      "learning_rate": 8.948084496956678e-06,
+      "loss": 0.0008,
+      "step": 3023
+    },
+    {
+      "epoch": 0.19488303151382355,
+      "grad_norm": 0.0013122715834712388,
+      "learning_rate": 8.947368421052632e-06,
+      "loss": 0.0,
+      "step": 3024
+    },
+    {
+      "epoch": 0.19494747696075274,
+      "grad_norm": 0.004215928276184871,
+      "learning_rate": 8.946652345148586e-06,
+      "loss": 0.0,
+      "step": 3025
+    },
+    {
+      "epoch": 0.1950119224076819,
+      "grad_norm": 0.17655713718844582,
+      "learning_rate": 8.94593626924454e-06,
+      "loss": 0.0013,
+      "step": 3026
+    },
+    {
+      "epoch": 0.19507636785461108,
+      "grad_norm": 0.0023065439930651447,
+      "learning_rate": 8.945220193340495e-06,
+      "loss": 0.0,
+      "step": 3027
+    },
+    {
+      "epoch": 0.19514081330154023,
+      "grad_norm": 0.00540466461391632,
+      "learning_rate": 8.944504117436449e-06,
+      "loss": 0.0,
+      "step": 3028
+    },
+    {
+      "epoch": 0.19520525874846942,
+      "grad_norm": 0.0007297223900245063,
+      "learning_rate": 8.943788041532403e-06,
+      "loss": 0.0,
+      "step": 3029
+    },
+    {
+      "epoch": 0.1952697041953986,
+      "grad_norm": 0.05844247682053747,
+      "learning_rate": 8.943071965628358e-06,
+      "loss": 0.0003,
+      "step": 3030
+    },
+    {
+      "epoch": 0.19533414964232776,
+      "grad_norm": 0.31681117297173256,
+      "learning_rate": 8.942355889724312e-06,
+      "loss": 0.001,
+      "step": 3031
+    },
+    {
+      "epoch": 0.19539859508925694,
+      "grad_norm": 0.28899940884754083,
+      "learning_rate": 8.941639813820265e-06,
+      "loss": 0.0007,
+      "step": 3032
+    },
+    {
+      "epoch": 0.19546304053618613,
+      "grad_norm": 0.0033367080062562983,
+      "learning_rate": 8.940923737916219e-06,
+      "loss": 0.0,
+      "step": 3033
+    },
+    {
+      "epoch": 0.19552748598311528,
+      "grad_norm": 0.09525066093311885,
+      "learning_rate": 8.940207662012173e-06,
+      "loss": 0.0007,
+      "step": 3034
+    },
+    {
+      "epoch": 0.19559193143004447,
+      "grad_norm": 0.00012517123703986328,
+      "learning_rate": 8.939491586108127e-06,
+      "loss": 0.0,
+      "step": 3035
+    },
+    {
+      "epoch": 0.19565637687697365,
+      "grad_norm": 0.05815377551468865,
+      "learning_rate": 8.938775510204082e-06,
+      "loss": 0.0003,
+      "step": 3036
+    },
+    {
+      "epoch": 0.1957208223239028,
+      "grad_norm": 0.0007073388646762897,
+      "learning_rate": 8.938059434300038e-06,
+      "loss": 0.0,
+      "step": 3037
+    },
+    {
+      "epoch": 0.195785267770832,
+      "grad_norm": 0.015940901006851475,
+      "learning_rate": 8.937343358395992e-06,
+      "loss": 0.0001,
+      "step": 3038
+    },
+    {
+      "epoch": 0.19584971321776118,
+      "grad_norm": 0.0006422838104322162,
+      "learning_rate": 8.936627282491945e-06,
+      "loss": 0.0,
+      "step": 3039
+    },
+    {
+      "epoch": 0.19591415866469034,
+      "grad_norm": 0.010320363651081372,
+      "learning_rate": 8.935911206587899e-06,
+      "loss": 0.0,
+      "step": 3040
+    },
+    {
+      "epoch": 0.19597860411161952,
+      "grad_norm": 0.016090234798197458,
+      "learning_rate": 8.935195130683853e-06,
+      "loss": 0.0001,
+      "step": 3041
+    },
+    {
+      "epoch": 0.19604304955854868,
+      "grad_norm": 0.35970697411549507,
+      "learning_rate": 8.934479054779808e-06,
+      "loss": 0.0013,
+      "step": 3042
+    },
+    {
+      "epoch": 0.19610749500547786,
+      "grad_norm": 0.0001557431387520329,
+      "learning_rate": 8.933762978875762e-06,
+      "loss": 0.0,
+      "step": 3043
+    },
+    {
+      "epoch": 0.19617194045240705,
+      "grad_norm": 0.001276511107407491,
+      "learning_rate": 8.933046902971716e-06,
+      "loss": 0.0,
+      "step": 3044
+    },
+    {
+      "epoch": 0.1962363858993362,
+      "grad_norm": 0.0005064007813871364,
+      "learning_rate": 8.93233082706767e-06,
+      "loss": 0.0,
+      "step": 3045
+    },
+    {
+      "epoch": 0.1963008313462654,
+      "grad_norm": 0.006093401575577427,
+      "learning_rate": 8.931614751163625e-06,
+      "loss": 0.0001,
+      "step": 3046
+    },
+    {
+      "epoch": 0.19636527679319457,
+      "grad_norm": 0.11769669691171146,
+      "learning_rate": 8.930898675259579e-06,
+      "loss": 0.0003,
+      "step": 3047
+    },
+    {
+      "epoch": 0.19642972224012373,
+      "grad_norm": 0.0010719782400106074,
+      "learning_rate": 8.930182599355532e-06,
+      "loss": 0.0,
+      "step": 3048
+    },
+    {
+      "epoch": 0.1964941676870529,
+      "grad_norm": 0.0005714457417295161,
+      "learning_rate": 8.929466523451486e-06,
+      "loss": 0.0,
+      "step": 3049
+    },
+    {
+      "epoch": 0.1965586131339821,
+      "grad_norm": 0.4581438037557388,
+      "learning_rate": 8.92875044754744e-06,
+      "loss": 0.0014,
+      "step": 3050
+    },
+    {
+      "epoch": 0.19662305858091125,
+      "grad_norm": 0.0068193888884800335,
+      "learning_rate": 8.928034371643395e-06,
+      "loss": 0.0001,
+      "step": 3051
+    },
+    {
+      "epoch": 0.19668750402784044,
+      "grad_norm": 0.02994317793845591,
+      "learning_rate": 8.927318295739349e-06,
+      "loss": 0.0,
+      "step": 3052
+    },
+    {
+      "epoch": 0.1967519494747696,
+      "grad_norm": 0.014345473768226343,
+      "learning_rate": 8.926602219835303e-06,
+      "loss": 0.0001,
+      "step": 3053
+    },
+    {
+      "epoch": 0.19681639492169878,
+      "grad_norm": 0.0005049755597789164,
+      "learning_rate": 8.925886143931257e-06,
+      "loss": 0.0,
+      "step": 3054
+    },
+    {
+      "epoch": 0.19688084036862796,
+      "grad_norm": 0.2326998450872154,
+      "learning_rate": 8.925170068027212e-06,
+      "loss": 0.0002,
+      "step": 3055
+    },
+    {
+      "epoch": 0.19694528581555712,
+      "grad_norm": 0.2033664753568645,
+      "learning_rate": 8.924453992123166e-06,
+      "loss": 0.0006,
+      "step": 3056
+    },
+    {
+      "epoch": 0.1970097312624863,
+      "grad_norm": 0.06406544280271197,
+      "learning_rate": 8.92373791621912e-06,
+      "loss": 0.0002,
+      "step": 3057
+    },
+    {
+      "epoch": 0.1970741767094155,
+      "grad_norm": 0.01635489248533111,
+      "learning_rate": 8.923021840315073e-06,
+      "loss": 0.0001,
+      "step": 3058
+    },
+    {
+      "epoch": 0.19713862215634465,
+      "grad_norm": 0.0020399615266398115,
+      "learning_rate": 8.922305764411027e-06,
+      "loss": 0.0,
+      "step": 3059
+    },
+    {
+      "epoch": 0.19720306760327383,
+      "grad_norm": 0.007834388015339783,
+      "learning_rate": 8.921589688506982e-06,
+      "loss": 0.0,
+      "step": 3060
+    },
+    {
+      "epoch": 0.197267513050203,
+      "grad_norm": 0.0030011656472259296,
+      "learning_rate": 8.920873612602938e-06,
+      "loss": 0.0,
+      "step": 3061
+    },
+    {
+      "epoch": 0.19733195849713217,
+      "grad_norm": 0.010005728413372437,
+      "learning_rate": 8.920157536698892e-06,
+      "loss": 0.0001,
+      "step": 3062
+    },
+    {
+      "epoch": 0.19739640394406135,
+      "grad_norm": 0.0002145815971270866,
+      "learning_rate": 8.919441460794846e-06,
+      "loss": 0.0,
+      "step": 3063
+    },
+    {
+      "epoch": 0.19746084939099054,
+      "grad_norm": 0.11205250164730646,
+      "learning_rate": 8.918725384890799e-06,
+      "loss": 0.0019,
+      "step": 3064
+    },
+    {
+      "epoch": 0.1975252948379197,
+      "grad_norm": 0.02060544100981016,
+      "learning_rate": 8.918009308986753e-06,
+      "loss": 0.0,
+      "step": 3065
+    },
+    {
+      "epoch": 0.19758974028484888,
+      "grad_norm": 0.001616753284672874,
+      "learning_rate": 8.917293233082707e-06,
+      "loss": 0.0,
+      "step": 3066
+    },
+    {
+      "epoch": 0.19765418573177804,
+      "grad_norm": 0.005112549724463105,
+      "learning_rate": 8.916577157178662e-06,
+      "loss": 0.0,
+      "step": 3067
+    },
+    {
+      "epoch": 0.19771863117870722,
+      "grad_norm": 0.003087953257347784,
+      "learning_rate": 8.915861081274616e-06,
+      "loss": 0.0,
+      "step": 3068
+    },
+    {
+      "epoch": 0.1977830766256364,
+      "grad_norm": 0.012052220127453446,
+      "learning_rate": 8.91514500537057e-06,
+      "loss": 0.0001,
+      "step": 3069
+    },
+    {
+      "epoch": 0.19784752207256556,
+      "grad_norm": 0.0009411924538636635,
+      "learning_rate": 8.914428929466525e-06,
+      "loss": 0.0,
+      "step": 3070
+    },
+    {
+      "epoch": 0.19791196751949475,
+      "grad_norm": 0.018949288941459586,
+      "learning_rate": 8.913712853562479e-06,
+      "loss": 0.0002,
+      "step": 3071
+    },
+    {
+      "epoch": 0.19797641296642393,
+      "grad_norm": 0.011494391843931702,
+      "learning_rate": 8.912996777658433e-06,
+      "loss": 0.0,
+      "step": 3072
+    },
+    {
+      "epoch": 0.1980408584133531,
+      "grad_norm": 0.0031116002090642793,
+      "learning_rate": 8.912280701754387e-06,
+      "loss": 0.0,
+      "step": 3073
+    },
+    {
+      "epoch": 0.19810530386028227,
+      "grad_norm": 0.022021763285966188,
+      "learning_rate": 8.91156462585034e-06,
+      "loss": 0.0001,
+      "step": 3074
+    },
+    {
+      "epoch": 0.19816974930721146,
+      "grad_norm": 0.007751726489644879,
+      "learning_rate": 8.910848549946294e-06,
+      "loss": 0.0,
+      "step": 3075
+    },
+    {
+      "epoch": 0.1982341947541406,
+      "grad_norm": 0.0032938459913306323,
+      "learning_rate": 8.910132474042249e-06,
+      "loss": 0.0,
+      "step": 3076
+    },
+    {
+      "epoch": 0.1982986402010698,
+      "grad_norm": 0.026446706021640256,
+      "learning_rate": 8.909416398138203e-06,
+      "loss": 0.0003,
+      "step": 3077
+    },
+    {
+      "epoch": 0.19836308564799898,
+      "grad_norm": 0.0027147322304744165,
+      "learning_rate": 8.908700322234157e-06,
+      "loss": 0.0,
+      "step": 3078
+    },
+    {
+      "epoch": 0.19842753109492814,
+      "grad_norm": 0.03446864700507342,
+      "learning_rate": 8.907984246330112e-06,
+      "loss": 0.0,
+      "step": 3079
+    },
+    {
+      "epoch": 0.19849197654185732,
+      "grad_norm": 0.002107508255067804,
+      "learning_rate": 8.907268170426066e-06,
+      "loss": 0.0,
+      "step": 3080
+    },
+    {
+      "epoch": 0.19855642198878648,
+      "grad_norm": 0.0015772077216705193,
+      "learning_rate": 8.90655209452202e-06,
+      "loss": 0.0,
+      "step": 3081
+    },
+    {
+      "epoch": 0.19862086743571566,
+      "grad_norm": 0.0014874691914585742,
+      "learning_rate": 8.905836018617974e-06,
+      "loss": 0.0,
+      "step": 3082
+    },
+    {
+      "epoch": 0.19868531288264485,
+      "grad_norm": 0.0007431047392924327,
+      "learning_rate": 8.905119942713929e-06,
+      "loss": 0.0,
+      "step": 3083
+    },
+    {
+      "epoch": 0.198749758329574,
+      "grad_norm": 0.0018363159310733638,
+      "learning_rate": 8.904403866809883e-06,
+      "loss": 0.0,
+      "step": 3084
+    },
+    {
+      "epoch": 0.1988142037765032,
+      "grad_norm": 0.0013460365521944676,
+      "learning_rate": 8.903687790905837e-06,
+      "loss": 0.0,
+      "step": 3085
+    },
+    {
+      "epoch": 0.19887864922343237,
+      "grad_norm": 0.15260297166684111,
+      "learning_rate": 8.902971715001792e-06,
+      "loss": 0.0019,
+      "step": 3086
+    },
+    {
+      "epoch": 0.19894309467036153,
+      "grad_norm": 0.005164714457936164,
+      "learning_rate": 8.902255639097746e-06,
+      "loss": 0.0,
+      "step": 3087
+    },
+    {
+      "epoch": 0.19900754011729072,
+      "grad_norm": 0.005644038945726389,
+      "learning_rate": 8.9015395631937e-06,
+      "loss": 0.0,
+      "step": 3088
+    },
+    {
+      "epoch": 0.1990719855642199,
+      "grad_norm": 0.05408689312074812,
+      "learning_rate": 8.900823487289655e-06,
+      "loss": 0.0005,
+      "step": 3089
+    },
+    {
+      "epoch": 0.19913643101114906,
+      "grad_norm": 0.016265788623832402,
+      "learning_rate": 8.900107411385607e-06,
+      "loss": 0.0,
+      "step": 3090
+    },
+    {
+      "epoch": 0.19920087645807824,
+      "grad_norm": 0.2519519865025074,
+      "learning_rate": 8.899391335481561e-06,
+      "loss": 0.0026,
+      "step": 3091
+    },
+    {
+      "epoch": 0.1992653219050074,
+      "grad_norm": 0.019179351740819434,
+      "learning_rate": 8.898675259577516e-06,
+      "loss": 0.0003,
+      "step": 3092
+    },
+    {
+      "epoch": 0.19932976735193658,
+      "grad_norm": 0.00027986388418838086,
+      "learning_rate": 8.89795918367347e-06,
+      "loss": 0.0,
+      "step": 3093
+    },
+    {
+      "epoch": 0.19939421279886577,
+      "grad_norm": 0.015353708675322387,
+      "learning_rate": 8.897243107769424e-06,
+      "loss": 0.0,
+      "step": 3094
+    },
+    {
+      "epoch": 0.19945865824579492,
+      "grad_norm": 0.0026218758351446818,
+      "learning_rate": 8.896527031865379e-06,
+      "loss": 0.0,
+      "step": 3095
+    },
+    {
+      "epoch": 0.1995231036927241,
+      "grad_norm": 0.0033979739761200935,
+      "learning_rate": 8.895810955961333e-06,
+      "loss": 0.0,
+      "step": 3096
+    },
+    {
+      "epoch": 0.1995875491396533,
+      "grad_norm": 0.18854867217886892,
+      "learning_rate": 8.895094880057287e-06,
+      "loss": 0.0007,
+      "step": 3097
+    },
+    {
+      "epoch": 0.19965199458658245,
+      "grad_norm": 0.20573892851899744,
+      "learning_rate": 8.894378804153241e-06,
+      "loss": 0.0006,
+      "step": 3098
+    },
+    {
+      "epoch": 0.19971644003351163,
+      "grad_norm": 0.497828072749145,
+      "learning_rate": 8.893662728249196e-06,
+      "loss": 0.0014,
+      "step": 3099
+    },
+    {
+      "epoch": 0.19978088548044082,
+      "grad_norm": 0.00453579094805275,
+      "learning_rate": 8.892946652345148e-06,
+      "loss": 0.0,
+      "step": 3100
+    },
+    {
+      "epoch": 0.19984533092736997,
+      "grad_norm": 0.005429255738304247,
+      "learning_rate": 8.892230576441103e-06,
+      "loss": 0.0,
+      "step": 3101
+    },
+    {
+      "epoch": 0.19990977637429916,
+      "grad_norm": 0.0028107191266686154,
+      "learning_rate": 8.891514500537057e-06,
+      "loss": 0.0,
+      "step": 3102
+    },
+    {
+      "epoch": 0.19997422182122834,
+      "grad_norm": 0.001090017515992669,
+      "learning_rate": 8.890798424633011e-06,
+      "loss": 0.0,
+      "step": 3103
+    },
+    {
+      "epoch": 0.2000386672681575,
+      "grad_norm": 0.005462728398524046,
+      "learning_rate": 8.890082348728966e-06,
+      "loss": 0.0001,
+      "step": 3104
+    },
+    {
+      "epoch": 0.20010311271508668,
+      "grad_norm": 0.21584340658087367,
+      "learning_rate": 8.88936627282492e-06,
+      "loss": 0.0011,
+      "step": 3105
+    },
+    {
+      "epoch": 0.20016755816201584,
+      "grad_norm": 0.026409045302834108,
+      "learning_rate": 8.888650196920874e-06,
+      "loss": 0.0001,
+      "step": 3106
+    },
+    {
+      "epoch": 0.20023200360894503,
+      "grad_norm": 0.00022268280929360672,
+      "learning_rate": 8.887934121016828e-06,
+      "loss": 0.0,
+      "step": 3107
+    },
+    {
+      "epoch": 0.2002964490558742,
+      "grad_norm": 0.7561233808588724,
+      "learning_rate": 8.887218045112783e-06,
+      "loss": 0.0056,
+      "step": 3108
+    },
+    {
+      "epoch": 0.20036089450280337,
+      "grad_norm": 0.0008315596068303258,
+      "learning_rate": 8.886501969208737e-06,
+      "loss": 0.0,
+      "step": 3109
+    },
+    {
+      "epoch": 0.20042533994973255,
+      "grad_norm": 0.03906717868442174,
+      "learning_rate": 8.885785893304691e-06,
+      "loss": 0.0001,
+      "step": 3110
+    },
+    {
+      "epoch": 0.20048978539666173,
+      "grad_norm": 0.004649140480185962,
+      "learning_rate": 8.885069817400646e-06,
+      "loss": 0.0,
+      "step": 3111
+    },
+    {
+      "epoch": 0.2005542308435909,
+      "grad_norm": 7.197497173441826e-05,
+      "learning_rate": 8.8843537414966e-06,
+      "loss": 0.0,
+      "step": 3112
+    },
+    {
+      "epoch": 0.20061867629052008,
+      "grad_norm": 0.01971261560067173,
+      "learning_rate": 8.883637665592554e-06,
+      "loss": 0.0,
+      "step": 3113
+    },
+    {
+      "epoch": 0.20068312173744926,
+      "grad_norm": 0.008505374211518214,
+      "learning_rate": 8.882921589688509e-06,
+      "loss": 0.0,
+      "step": 3114
+    },
+    {
+      "epoch": 0.20074756718437842,
+      "grad_norm": 0.0005514684611620498,
+      "learning_rate": 8.882205513784463e-06,
+      "loss": 0.0,
+      "step": 3115
+    },
+    {
+      "epoch": 0.2008120126313076,
+      "grad_norm": 0.3043897714692487,
+      "learning_rate": 8.881489437880415e-06,
+      "loss": 0.0014,
+      "step": 3116
+    },
+    {
+      "epoch": 0.20087645807823679,
+      "grad_norm": 0.0006227491004178154,
+      "learning_rate": 8.88077336197637e-06,
+      "loss": 0.0,
+      "step": 3117
+    },
+    {
+      "epoch": 0.20094090352516594,
+      "grad_norm": 0.2678906919268471,
+      "learning_rate": 8.880057286072324e-06,
+      "loss": 0.0005,
+      "step": 3118
+    },
+    {
+      "epoch": 0.20100534897209513,
+      "grad_norm": 0.015165462236630976,
+      "learning_rate": 8.879341210168278e-06,
+      "loss": 0.0,
+      "step": 3119
+    },
+    {
+      "epoch": 0.20106979441902428,
+      "grad_norm": 2.0520360976719396,
+      "learning_rate": 8.878625134264233e-06,
+      "loss": 0.0165,
+      "step": 3120
+    },
+    {
+      "epoch": 0.20113423986595347,
+      "grad_norm": 0.35805703099009856,
+      "learning_rate": 8.877909058360187e-06,
+      "loss": 0.0012,
+      "step": 3121
+    },
+    {
+      "epoch": 0.20119868531288265,
+      "grad_norm": 0.17738543831285275,
+      "learning_rate": 8.877192982456141e-06,
+      "loss": 0.0019,
+      "step": 3122
+    },
+    {
+      "epoch": 0.2012631307598118,
+      "grad_norm": 0.00017845454458807985,
+      "learning_rate": 8.876476906552096e-06,
+      "loss": 0.0,
+      "step": 3123
+    },
+    {
+      "epoch": 0.201327576206741,
+      "grad_norm": 0.1073608615304861,
+      "learning_rate": 8.87576083064805e-06,
+      "loss": 0.0004,
+      "step": 3124
+    },
+    {
+      "epoch": 0.20139202165367018,
+      "grad_norm": 0.0004145412500641842,
+      "learning_rate": 8.875044754744002e-06,
+      "loss": 0.0,
+      "step": 3125
+    },
+    {
+      "epoch": 0.20145646710059933,
+      "grad_norm": 0.003809234403064291,
+      "learning_rate": 8.874328678839957e-06,
+      "loss": 0.0001,
+      "step": 3126
+    },
+    {
+      "epoch": 0.20152091254752852,
+      "grad_norm": 0.00033783976806183987,
+      "learning_rate": 8.873612602935911e-06,
+      "loss": 0.0,
+      "step": 3127
+    },
+    {
+      "epoch": 0.2015853579944577,
+      "grad_norm": 0.013849292965756455,
+      "learning_rate": 8.872896527031865e-06,
+      "loss": 0.0,
+      "step": 3128
+    },
+    {
+      "epoch": 0.20164980344138686,
+      "grad_norm": 0.045313850949965165,
+      "learning_rate": 8.87218045112782e-06,
+      "loss": 0.0002,
+      "step": 3129
+    },
+    {
+      "epoch": 0.20171424888831604,
+      "grad_norm": 0.236568006038209,
+      "learning_rate": 8.871464375223774e-06,
+      "loss": 0.0003,
+      "step": 3130
+    },
+    {
+      "epoch": 0.2017786943352452,
+      "grad_norm": 0.0038200420844738477,
+      "learning_rate": 8.87074829931973e-06,
+      "loss": 0.0,
+      "step": 3131
+    },
+    {
+      "epoch": 0.20184313978217439,
+      "grad_norm": 0.014478036074284993,
+      "learning_rate": 8.870032223415683e-06,
+      "loss": 0.0001,
+      "step": 3132
+    },
+    {
+      "epoch": 0.20190758522910357,
+      "grad_norm": 0.015151909350715245,
+      "learning_rate": 8.869316147511637e-06,
+      "loss": 0.0,
+      "step": 3133
+    },
+    {
+      "epoch": 0.20197203067603273,
+      "grad_norm": 0.009762932361485831,
+      "learning_rate": 8.868600071607591e-06,
+      "loss": 0.0001,
+      "step": 3134
+    },
+    {
+      "epoch": 0.2020364761229619,
+      "grad_norm": 0.00021053239735446382,
+      "learning_rate": 8.867883995703545e-06,
+      "loss": 0.0,
+      "step": 3135
+    },
+    {
+      "epoch": 0.2021009215698911,
+      "grad_norm": 0.44029179548069913,
+      "learning_rate": 8.8671679197995e-06,
+      "loss": 0.0016,
+      "step": 3136
+    },
+    {
+      "epoch": 0.20216536701682025,
+      "grad_norm": 0.0019125566904534606,
+      "learning_rate": 8.866451843895454e-06,
+      "loss": 0.0,
+      "step": 3137
+    },
+    {
+      "epoch": 0.20222981246374944,
+      "grad_norm": 0.0030594121784513393,
+      "learning_rate": 8.865735767991408e-06,
+      "loss": 0.0,
+      "step": 3138
+    },
+    {
+      "epoch": 0.20229425791067862,
+      "grad_norm": 0.002322376722079305,
+      "learning_rate": 8.865019692087363e-06,
+      "loss": 0.0,
+      "step": 3139
+    },
+    {
+      "epoch": 0.20235870335760778,
+      "grad_norm": 0.025830919381848335,
+      "learning_rate": 8.864303616183317e-06,
+      "loss": 0.0001,
+      "step": 3140
+    },
+    {
+      "epoch": 0.20242314880453696,
+      "grad_norm": 0.004267439275289735,
+      "learning_rate": 8.86358754027927e-06,
+      "loss": 0.0,
+      "step": 3141
+    },
+    {
+      "epoch": 0.20248759425146615,
+      "grad_norm": 0.0005659775357358226,
+      "learning_rate": 8.862871464375224e-06,
+      "loss": 0.0,
+      "step": 3142
+    },
+    {
+      "epoch": 0.2025520396983953,
+      "grad_norm": 0.03815054914874814,
+      "learning_rate": 8.862155388471178e-06,
+      "loss": 0.0003,
+      "step": 3143
+    },
+    {
+      "epoch": 0.2026164851453245,
+      "grad_norm": 0.3565567469487967,
+      "learning_rate": 8.861439312567132e-06,
+      "loss": 0.0031,
+      "step": 3144
+    },
+    {
+      "epoch": 0.20268093059225364,
+      "grad_norm": 0.040210932144915466,
+      "learning_rate": 8.860723236663087e-06,
+      "loss": 0.0002,
+      "step": 3145
+    },
+    {
+      "epoch": 0.20274537603918283,
+      "grad_norm": 0.41041854464235217,
+      "learning_rate": 8.860007160759041e-06,
+      "loss": 0.0008,
+      "step": 3146
+    },
+    {
+      "epoch": 0.202809821486112,
+      "grad_norm": 0.25541055057218365,
+      "learning_rate": 8.859291084854995e-06,
+      "loss": 0.001,
+      "step": 3147
+    },
+    {
+      "epoch": 0.20287426693304117,
+      "grad_norm": 0.010501430279528583,
+      "learning_rate": 8.85857500895095e-06,
+      "loss": 0.0001,
+      "step": 3148
+    },
+    {
+      "epoch": 0.20293871237997035,
+      "grad_norm": 0.003696393179009786,
+      "learning_rate": 8.857858933046904e-06,
+      "loss": 0.0,
+      "step": 3149
+    },
+    {
+      "epoch": 0.20300315782689954,
+      "grad_norm": 0.003631618092946609,
+      "learning_rate": 8.857142857142858e-06,
+      "loss": 0.0,
+      "step": 3150
+    },
+    {
+      "epoch": 0.2030676032738287,
+      "grad_norm": 0.00258595354685201,
+      "learning_rate": 8.856426781238811e-06,
+      "loss": 0.0,
+      "step": 3151
+    },
+    {
+      "epoch": 0.20313204872075788,
+      "grad_norm": 0.0008821024753706213,
+      "learning_rate": 8.855710705334765e-06,
+      "loss": 0.0,
+      "step": 3152
+    },
+    {
+      "epoch": 0.20319649416768706,
+      "grad_norm": 0.3704561075782924,
+      "learning_rate": 8.85499462943072e-06,
+      "loss": 0.0043,
+      "step": 3153
+    },
+    {
+      "epoch": 0.20326093961461622,
+      "grad_norm": 0.014684063583619683,
+      "learning_rate": 8.854278553526675e-06,
+      "loss": 0.0,
+      "step": 3154
+    },
+    {
+      "epoch": 0.2033253850615454,
+      "grad_norm": 0.006794537516212787,
+      "learning_rate": 8.85356247762263e-06,
+      "loss": 0.0,
+      "step": 3155
+    },
+    {
+      "epoch": 0.2033898305084746,
+      "grad_norm": 1.0445839023372976,
+      "learning_rate": 8.852846401718584e-06,
+      "loss": 0.0022,
+      "step": 3156
+    },
+    {
+      "epoch": 0.20345427595540375,
+      "grad_norm": 0.004375746475512265,
+      "learning_rate": 8.852130325814537e-06,
+      "loss": 0.0,
+      "step": 3157
+    },
+    {
+      "epoch": 0.20351872140233293,
+      "grad_norm": 0.15032337592505507,
+      "learning_rate": 8.851414249910491e-06,
+      "loss": 0.0002,
+      "step": 3158
+    },
+    {
+      "epoch": 0.2035831668492621,
+      "grad_norm": 0.0040612406359037175,
+      "learning_rate": 8.850698174006445e-06,
+      "loss": 0.0,
+      "step": 3159
+    },
+    {
+      "epoch": 0.20364761229619127,
+      "grad_norm": 0.0018185464603422453,
+      "learning_rate": 8.8499820981024e-06,
+      "loss": 0.0,
+      "step": 3160
+    },
+    {
+      "epoch": 0.20371205774312046,
+      "grad_norm": 0.11170526669906726,
+      "learning_rate": 8.849266022198354e-06,
+      "loss": 0.0005,
+      "step": 3161
+    },
+    {
+      "epoch": 0.2037765031900496,
+      "grad_norm": 0.1659282005347964,
+      "learning_rate": 8.848549946294308e-06,
+      "loss": 0.0016,
+      "step": 3162
+    },
+    {
+      "epoch": 0.2038409486369788,
+      "grad_norm": 0.016492687346802256,
+      "learning_rate": 8.847833870390262e-06,
+      "loss": 0.0001,
+      "step": 3163
+    },
+    {
+      "epoch": 0.20390539408390798,
+      "grad_norm": 0.9390481563748461,
+      "learning_rate": 8.847117794486217e-06,
+      "loss": 0.0031,
+      "step": 3164
+    },
+    {
+      "epoch": 0.20396983953083714,
+      "grad_norm": 0.0022533603744609204,
+      "learning_rate": 8.846401718582171e-06,
+      "loss": 0.0,
+      "step": 3165
+    },
+    {
+      "epoch": 0.20403428497776632,
+      "grad_norm": 0.07041904205568279,
+      "learning_rate": 8.845685642678125e-06,
+      "loss": 0.0001,
+      "step": 3166
+    },
+    {
+      "epoch": 0.2040987304246955,
+      "grad_norm": 0.2668981613989854,
+      "learning_rate": 8.844969566774078e-06,
+      "loss": 0.0052,
+      "step": 3167
+    },
+    {
+      "epoch": 0.20416317587162466,
+      "grad_norm": 0.027716524548335512,
+      "learning_rate": 8.844253490870032e-06,
+      "loss": 0.0,
+      "step": 3168
+    },
+    {
+      "epoch": 0.20422762131855385,
+      "grad_norm": 0.003601240274471943,
+      "learning_rate": 8.843537414965987e-06,
+      "loss": 0.0,
+      "step": 3169
+    },
+    {
+      "epoch": 0.20429206676548303,
+      "grad_norm": 0.012017656279619291,
+      "learning_rate": 8.84282133906194e-06,
+      "loss": 0.0,
+      "step": 3170
+    },
+    {
+      "epoch": 0.2043565122124122,
+      "grad_norm": 0.02095705244991158,
+      "learning_rate": 8.842105263157895e-06,
+      "loss": 0.0,
+      "step": 3171
+    },
+    {
+      "epoch": 0.20442095765934137,
+      "grad_norm": 0.040138181546046925,
+      "learning_rate": 8.84138918725385e-06,
+      "loss": 0.0,
+      "step": 3172
+    },
+    {
+      "epoch": 0.20448540310627053,
+      "grad_norm": 0.009901321405853671,
+      "learning_rate": 8.840673111349804e-06,
+      "loss": 0.0,
+      "step": 3173
+    },
+    {
+      "epoch": 0.20454984855319971,
+      "grad_norm": 0.007040968629763532,
+      "learning_rate": 8.839957035445758e-06,
+      "loss": 0.0,
+      "step": 3174
+    },
+    {
+      "epoch": 0.2046142940001289,
+      "grad_norm": 0.014848915765067815,
+      "learning_rate": 8.839240959541712e-06,
+      "loss": 0.0,
+      "step": 3175
+    },
+    {
+      "epoch": 0.20467873944705806,
+      "grad_norm": 0.23642315018145912,
+      "learning_rate": 8.838524883637667e-06,
+      "loss": 0.0008,
+      "step": 3176
+    },
+    {
+      "epoch": 0.20474318489398724,
+      "grad_norm": 0.0029172454636086998,
+      "learning_rate": 8.83780880773362e-06,
+      "loss": 0.0,
+      "step": 3177
+    },
+    {
+      "epoch": 0.20480763034091642,
+      "grad_norm": 0.10179572652957916,
+      "learning_rate": 8.837092731829575e-06,
+      "loss": 0.0002,
+      "step": 3178
+    },
+    {
+      "epoch": 0.20487207578784558,
+      "grad_norm": 0.24513314487750718,
+      "learning_rate": 8.83637665592553e-06,
+      "loss": 0.0041,
+      "step": 3179
+    },
+    {
+      "epoch": 0.20493652123477477,
+      "grad_norm": 0.012036311721826174,
+      "learning_rate": 8.835660580021484e-06,
+      "loss": 0.0,
+      "step": 3180
+    },
+    {
+      "epoch": 0.20500096668170395,
+      "grad_norm": 0.2696918479020835,
+      "learning_rate": 8.834944504117438e-06,
+      "loss": 0.0005,
+      "step": 3181
+    },
+    {
+      "epoch": 0.2050654121286331,
+      "grad_norm": 0.0003936673405125266,
+      "learning_rate": 8.834228428213392e-06,
+      "loss": 0.0,
+      "step": 3182
+    },
+    {
+      "epoch": 0.2051298575755623,
+      "grad_norm": 0.058986318316246796,
+      "learning_rate": 8.833512352309345e-06,
+      "loss": 0.0,
+      "step": 3183
+    },
+    {
+      "epoch": 0.20519430302249145,
+      "grad_norm": 0.13323453158189,
+      "learning_rate": 8.8327962764053e-06,
+      "loss": 0.0013,
+      "step": 3184
+    },
+    {
+      "epoch": 0.20525874846942063,
+      "grad_norm": 1.0706027674551744,
+      "learning_rate": 8.832080200501254e-06,
+      "loss": 0.0052,
+      "step": 3185
+    },
+    {
+      "epoch": 0.20532319391634982,
+      "grad_norm": 0.04365196867916086,
+      "learning_rate": 8.831364124597208e-06,
+      "loss": 0.0001,
+      "step": 3186
+    },
+    {
+      "epoch": 0.20538763936327897,
+      "grad_norm": 0.00400724211170088,
+      "learning_rate": 8.830648048693162e-06,
+      "loss": 0.0,
+      "step": 3187
+    },
+    {
+      "epoch": 0.20545208481020816,
+      "grad_norm": 0.25079876728614287,
+      "learning_rate": 8.829931972789117e-06,
+      "loss": 0.0051,
+      "step": 3188
+    },
+    {
+      "epoch": 0.20551653025713734,
+      "grad_norm": 0.6116119554626651,
+      "learning_rate": 8.82921589688507e-06,
+      "loss": 0.002,
+      "step": 3189
+    },
+    {
+      "epoch": 0.2055809757040665,
+      "grad_norm": 0.9003768555752997,
+      "learning_rate": 8.828499820981025e-06,
+      "loss": 0.0081,
+      "step": 3190
+    },
+    {
+      "epoch": 0.20564542115099568,
+      "grad_norm": 0.14866784067739025,
+      "learning_rate": 8.82778374507698e-06,
+      "loss": 0.0005,
+      "step": 3191
+    },
+    {
+      "epoch": 0.20570986659792487,
+      "grad_norm": 0.004681531754656857,
+      "learning_rate": 8.827067669172934e-06,
+      "loss": 0.0001,
+      "step": 3192
+    },
+    {
+      "epoch": 0.20577431204485402,
+      "grad_norm": 0.0016474154499777846,
+      "learning_rate": 8.826351593268886e-06,
+      "loss": 0.0,
+      "step": 3193
+    },
+    {
+      "epoch": 0.2058387574917832,
+      "grad_norm": 0.7216092334678865,
+      "learning_rate": 8.82563551736484e-06,
+      "loss": 0.0047,
+      "step": 3194
+    },
+    {
+      "epoch": 0.2059032029387124,
+      "grad_norm": 0.046842534027833414,
+      "learning_rate": 8.824919441460795e-06,
+      "loss": 0.0,
+      "step": 3195
+    },
+    {
+      "epoch": 0.20596764838564155,
+      "grad_norm": 0.008319201874511975,
+      "learning_rate": 8.82420336555675e-06,
+      "loss": 0.0,
+      "step": 3196
+    },
+    {
+      "epoch": 0.20603209383257073,
+      "grad_norm": 0.07627724121336131,
+      "learning_rate": 8.823487289652704e-06,
+      "loss": 0.0001,
+      "step": 3197
+    },
+    {
+      "epoch": 0.2060965392794999,
+      "grad_norm": 0.04090758410395465,
+      "learning_rate": 8.822771213748658e-06,
+      "loss": 0.0001,
+      "step": 3198
+    },
+    {
+      "epoch": 0.20616098472642908,
+      "grad_norm": 0.004577226024324245,
+      "learning_rate": 8.822055137844612e-06,
+      "loss": 0.0,
+      "step": 3199
+    },
+    {
+      "epoch": 0.20622543017335826,
+      "grad_norm": 0.5695110071506763,
+      "learning_rate": 8.821339061940566e-06,
+      "loss": 0.0015,
+      "step": 3200
+    },
+    {
+      "epoch": 0.20628987562028742,
+      "grad_norm": 0.20783832970335034,
+      "learning_rate": 8.82062298603652e-06,
+      "loss": 0.0007,
+      "step": 3201
+    },
+    {
+      "epoch": 0.2063543210672166,
+      "grad_norm": 0.07589570440082365,
+      "learning_rate": 8.819906910132475e-06,
+      "loss": 0.0003,
+      "step": 3202
+    },
+    {
+      "epoch": 0.20641876651414579,
+      "grad_norm": 0.025595976228796117,
+      "learning_rate": 8.81919083422843e-06,
+      "loss": 0.0001,
+      "step": 3203
+    },
+    {
+      "epoch": 0.20648321196107494,
+      "grad_norm": 0.008649726581106711,
+      "learning_rate": 8.818474758324384e-06,
+      "loss": 0.0001,
+      "step": 3204
+    },
+    {
+      "epoch": 0.20654765740800413,
+      "grad_norm": 0.019401712300670126,
+      "learning_rate": 8.817758682420338e-06,
+      "loss": 0.0001,
+      "step": 3205
+    },
+    {
+      "epoch": 0.2066121028549333,
+      "grad_norm": 0.018649523956299447,
+      "learning_rate": 8.817042606516292e-06,
+      "loss": 0.0,
+      "step": 3206
+    },
+    {
+      "epoch": 0.20667654830186247,
+      "grad_norm": 0.004175074407233036,
+      "learning_rate": 8.816326530612247e-06,
+      "loss": 0.0,
+      "step": 3207
+    },
+    {
+      "epoch": 0.20674099374879165,
+      "grad_norm": 0.001537868439082998,
+      "learning_rate": 8.8156104547082e-06,
+      "loss": 0.0,
+      "step": 3208
+    },
+    {
+      "epoch": 0.20680543919572084,
+      "grad_norm": 0.004367193650683198,
+      "learning_rate": 8.814894378804153e-06,
+      "loss": 0.0,
+      "step": 3209
+    },
+    {
+      "epoch": 0.20686988464265,
+      "grad_norm": 0.0008591282411366313,
+      "learning_rate": 8.814178302900108e-06,
+      "loss": 0.0,
+      "step": 3210
+    },
+    {
+      "epoch": 0.20693433008957918,
+      "grad_norm": 0.029196901412854644,
+      "learning_rate": 8.813462226996062e-06,
+      "loss": 0.0001,
+      "step": 3211
+    },
+    {
+      "epoch": 0.20699877553650833,
+      "grad_norm": 0.014151467748943703,
+      "learning_rate": 8.812746151092016e-06,
+      "loss": 0.0,
+      "step": 3212
+    },
+    {
+      "epoch": 0.20706322098343752,
+      "grad_norm": 0.0005395895476555126,
+      "learning_rate": 8.81203007518797e-06,
+      "loss": 0.0,
+      "step": 3213
+    },
+    {
+      "epoch": 0.2071276664303667,
+      "grad_norm": 0.1063896173481837,
+      "learning_rate": 8.811313999283925e-06,
+      "loss": 0.0008,
+      "step": 3214
+    },
+    {
+      "epoch": 0.20719211187729586,
+      "grad_norm": 0.04907908321108309,
+      "learning_rate": 8.81059792337988e-06,
+      "loss": 0.0001,
+      "step": 3215
+    },
+    {
+      "epoch": 0.20725655732422504,
+      "grad_norm": 0.0009487456876486122,
+      "learning_rate": 8.809881847475833e-06,
+      "loss": 0.0,
+      "step": 3216
+    },
+    {
+      "epoch": 0.20732100277115423,
+      "grad_norm": 0.012295272892583854,
+      "learning_rate": 8.809165771571788e-06,
+      "loss": 0.0,
+      "step": 3217
+    },
+    {
+      "epoch": 0.20738544821808338,
+      "grad_norm": 0.003048664850237199,
+      "learning_rate": 8.80844969566774e-06,
+      "loss": 0.0,
+      "step": 3218
+    },
+    {
+      "epoch": 0.20744989366501257,
+      "grad_norm": 0.007495688176306054,
+      "learning_rate": 8.807733619763695e-06,
+      "loss": 0.0001,
+      "step": 3219
+    },
+    {
+      "epoch": 0.20751433911194175,
+      "grad_norm": 0.0035060227618026463,
+      "learning_rate": 8.807017543859649e-06,
+      "loss": 0.0,
+      "step": 3220
+    },
+    {
+      "epoch": 0.2075787845588709,
+      "grad_norm": 0.03656436499960002,
+      "learning_rate": 8.806301467955603e-06,
+      "loss": 0.0002,
+      "step": 3221
+    },
+    {
+      "epoch": 0.2076432300058001,
+      "grad_norm": 0.29249599460157977,
+      "learning_rate": 8.805585392051558e-06,
+      "loss": 0.0006,
+      "step": 3222
+    },
+    {
+      "epoch": 0.20770767545272925,
+      "grad_norm": 0.011994526837623365,
+      "learning_rate": 8.804869316147512e-06,
+      "loss": 0.0,
+      "step": 3223
+    },
+    {
+      "epoch": 0.20777212089965844,
+      "grad_norm": 0.004782641801488116,
+      "learning_rate": 8.804153240243468e-06,
+      "loss": 0.0,
+      "step": 3224
+    },
+    {
+      "epoch": 0.20783656634658762,
+      "grad_norm": 0.04384486055700967,
+      "learning_rate": 8.80343716433942e-06,
+      "loss": 0.0001,
+      "step": 3225
+    },
+    {
+      "epoch": 0.20790101179351678,
+      "grad_norm": 0.0033315971745620817,
+      "learning_rate": 8.802721088435375e-06,
+      "loss": 0.0,
+      "step": 3226
+    },
+    {
+      "epoch": 0.20796545724044596,
+      "grad_norm": 0.016471005833125225,
+      "learning_rate": 8.802005012531329e-06,
+      "loss": 0.0,
+      "step": 3227
+    },
+    {
+      "epoch": 0.20802990268737515,
+      "grad_norm": 0.008197529248738967,
+      "learning_rate": 8.801288936627283e-06,
+      "loss": 0.0001,
+      "step": 3228
+    },
+    {
+      "epoch": 0.2080943481343043,
+      "grad_norm": 0.012575679376179708,
+      "learning_rate": 8.800572860723238e-06,
+      "loss": 0.0,
+      "step": 3229
+    },
+    {
+      "epoch": 0.2081587935812335,
+      "grad_norm": 0.0986802487508774,
+      "learning_rate": 8.799856784819192e-06,
+      "loss": 0.0021,
+      "step": 3230
+    },
+    {
+      "epoch": 0.20822323902816267,
+      "grad_norm": 0.0008527289712501518,
+      "learning_rate": 8.799140708915146e-06,
+      "loss": 0.0,
+      "step": 3231
+    },
+    {
+      "epoch": 0.20828768447509183,
+      "grad_norm": 0.9886804912648364,
+      "learning_rate": 8.7984246330111e-06,
+      "loss": 0.0021,
+      "step": 3232
+    },
+    {
+      "epoch": 0.208352129922021,
+      "grad_norm": 0.016484567955490047,
+      "learning_rate": 8.797708557107055e-06,
+      "loss": 0.0002,
+      "step": 3233
+    },
+    {
+      "epoch": 0.2084165753689502,
+      "grad_norm": 0.28071698703188924,
+      "learning_rate": 8.796992481203007e-06,
+      "loss": 0.0022,
+      "step": 3234
+    },
+    {
+      "epoch": 0.20848102081587935,
+      "grad_norm": 0.021459780780179235,
+      "learning_rate": 8.796276405298962e-06,
+      "loss": 0.0002,
+      "step": 3235
+    },
+    {
+      "epoch": 0.20854546626280854,
+      "grad_norm": 0.058684126375482114,
+      "learning_rate": 8.795560329394916e-06,
+      "loss": 0.0017,
+      "step": 3236
+    },
+    {
+      "epoch": 0.2086099117097377,
+      "grad_norm": 0.0011527392584699309,
+      "learning_rate": 8.79484425349087e-06,
+      "loss": 0.0,
+      "step": 3237
+    },
+    {
+      "epoch": 0.20867435715666688,
+      "grad_norm": 0.0020617141823999893,
+      "learning_rate": 8.794128177586825e-06,
+      "loss": 0.0,
+      "step": 3238
+    },
+    {
+      "epoch": 0.20873880260359606,
+      "grad_norm": 0.2818069506547426,
+      "learning_rate": 8.793412101682779e-06,
+      "loss": 0.0003,
+      "step": 3239
+    },
+    {
+      "epoch": 0.20880324805052522,
+      "grad_norm": 0.0039035938963142623,
+      "learning_rate": 8.792696025778733e-06,
+      "loss": 0.0,
+      "step": 3240
+    },
+    {
+      "epoch": 0.2088676934974544,
+      "grad_norm": 0.2977870433530856,
+      "learning_rate": 8.791979949874688e-06,
+      "loss": 0.0011,
+      "step": 3241
+    },
+    {
+      "epoch": 0.2089321389443836,
+      "grad_norm": 0.0770880788983326,
+      "learning_rate": 8.791263873970642e-06,
+      "loss": 0.0003,
+      "step": 3242
+    },
+    {
+      "epoch": 0.20899658439131275,
+      "grad_norm": 2.4195809605252174,
+      "learning_rate": 8.790547798066596e-06,
+      "loss": 0.0161,
+      "step": 3243
+    },
+    {
+      "epoch": 0.20906102983824193,
+      "grad_norm": 0.0061660864423312214,
+      "learning_rate": 8.789831722162549e-06,
+      "loss": 0.0,
+      "step": 3244
+    },
+    {
+      "epoch": 0.20912547528517111,
+      "grad_norm": 0.12889428516693582,
+      "learning_rate": 8.789115646258503e-06,
+      "loss": 0.0001,
+      "step": 3245
+    },
+    {
+      "epoch": 0.20918992073210027,
+      "grad_norm": 0.09151028984381641,
+      "learning_rate": 8.788399570354457e-06,
+      "loss": 0.0001,
+      "step": 3246
+    },
+    {
+      "epoch": 0.20925436617902946,
+      "grad_norm": 0.009779283910463333,
+      "learning_rate": 8.787683494450412e-06,
+      "loss": 0.0,
+      "step": 3247
+    },
+    {
+      "epoch": 0.20931881162595864,
+      "grad_norm": 0.006560546257411386,
+      "learning_rate": 8.786967418546368e-06,
+      "loss": 0.0,
+      "step": 3248
+    },
+    {
+      "epoch": 0.2093832570728878,
+      "grad_norm": 0.0038019288472966287,
+      "learning_rate": 8.786251342642322e-06,
+      "loss": 0.0,
+      "step": 3249
+    },
+    {
+      "epoch": 0.20944770251981698,
+      "grad_norm": 0.09019723577914596,
+      "learning_rate": 8.785535266738275e-06,
+      "loss": 0.0029,
+      "step": 3250
+    },
+    {
+      "epoch": 0.20951214796674614,
+      "grad_norm": 0.050902314027495524,
+      "learning_rate": 8.784819190834229e-06,
+      "loss": 0.0001,
+      "step": 3251
+    },
+    {
+      "epoch": 0.20957659341367532,
+      "grad_norm": 0.006432140117759434,
+      "learning_rate": 8.784103114930183e-06,
+      "loss": 0.0001,
+      "step": 3252
+    },
+    {
+      "epoch": 0.2096410388606045,
+      "grad_norm": 0.006941493151334226,
+      "learning_rate": 8.783387039026137e-06,
+      "loss": 0.0,
+      "step": 3253
+    },
+    {
+      "epoch": 0.20970548430753366,
+      "grad_norm": 0.13453681844557985,
+      "learning_rate": 8.782670963122092e-06,
+      "loss": 0.0007,
+      "step": 3254
+    },
+    {
+      "epoch": 0.20976992975446285,
+      "grad_norm": 0.16346047845287467,
+      "learning_rate": 8.781954887218046e-06,
+      "loss": 0.0008,
+      "step": 3255
+    },
+    {
+      "epoch": 0.20983437520139203,
+      "grad_norm": 0.0006596375563727548,
+      "learning_rate": 8.781238811314e-06,
+      "loss": 0.0,
+      "step": 3256
+    },
+    {
+      "epoch": 0.2098988206483212,
+      "grad_norm": 0.004409057604210094,
+      "learning_rate": 8.780522735409955e-06,
+      "loss": 0.0,
+      "step": 3257
+    },
+    {
+      "epoch": 0.20996326609525037,
+      "grad_norm": 0.003006464597436745,
+      "learning_rate": 8.779806659505909e-06,
+      "loss": 0.0,
+      "step": 3258
+    },
+    {
+      "epoch": 0.21002771154217956,
+      "grad_norm": 2.8408788922178685,
+      "learning_rate": 8.779090583601863e-06,
+      "loss": 0.0237,
+      "step": 3259
+    },
+    {
+      "epoch": 0.2100921569891087,
+      "grad_norm": 0.01016413326544834,
+      "learning_rate": 8.778374507697816e-06,
+      "loss": 0.0,
+      "step": 3260
+    },
+    {
+      "epoch": 0.2101566024360379,
+      "grad_norm": 0.01772877014468529,
+      "learning_rate": 8.77765843179377e-06,
+      "loss": 0.0,
+      "step": 3261
+    },
+    {
+      "epoch": 0.21022104788296705,
+      "grad_norm": 0.0024843355972861882,
+      "learning_rate": 8.776942355889724e-06,
+      "loss": 0.0,
+      "step": 3262
+    },
+    {
+      "epoch": 0.21028549332989624,
+      "grad_norm": 0.016083136451063594,
+      "learning_rate": 8.776226279985679e-06,
+      "loss": 0.0001,
+      "step": 3263
+    },
+    {
+      "epoch": 0.21034993877682542,
+      "grad_norm": 0.006143627681656301,
+      "learning_rate": 8.775510204081633e-06,
+      "loss": 0.0,
+      "step": 3264
+    },
+    {
+      "epoch": 0.21041438422375458,
+      "grad_norm": 0.6915285034718149,
+      "learning_rate": 8.774794128177587e-06,
+      "loss": 0.0009,
+      "step": 3265
+    },
+    {
+      "epoch": 0.21047882967068376,
+      "grad_norm": 0.003211719254704211,
+      "learning_rate": 8.774078052273542e-06,
+      "loss": 0.0,
+      "step": 3266
+    },
+    {
+      "epoch": 0.21054327511761295,
+      "grad_norm": 0.04889199609424895,
+      "learning_rate": 8.773361976369496e-06,
+      "loss": 0.0003,
+      "step": 3267
+    },
+    {
+      "epoch": 0.2106077205645421,
+      "grad_norm": 0.005836996667072234,
+      "learning_rate": 8.77264590046545e-06,
+      "loss": 0.0,
+      "step": 3268
+    },
+    {
+      "epoch": 0.2106721660114713,
+      "grad_norm": 0.04504886589983157,
+      "learning_rate": 8.771929824561405e-06,
+      "loss": 0.0002,
+      "step": 3269
+    },
+    {
+      "epoch": 0.21073661145840047,
+      "grad_norm": 0.007049011440276931,
+      "learning_rate": 8.771213748657357e-06,
+      "loss": 0.0,
+      "step": 3270
+    },
+    {
+      "epoch": 0.21080105690532963,
+      "grad_norm": 0.01300277264543531,
+      "learning_rate": 8.770497672753313e-06,
+      "loss": 0.0,
+      "step": 3271
+    },
+    {
+      "epoch": 0.21086550235225882,
+      "grad_norm": 0.09114491995853359,
+      "learning_rate": 8.769781596849267e-06,
+      "loss": 0.0004,
+      "step": 3272
+    },
+    {
+      "epoch": 0.210929947799188,
+      "grad_norm": 0.20353056693551833,
+      "learning_rate": 8.769065520945222e-06,
+      "loss": 0.0003,
+      "step": 3273
+    },
+    {
+      "epoch": 0.21099439324611716,
+      "grad_norm": 0.7785442417855192,
+      "learning_rate": 8.768349445041176e-06,
+      "loss": 0.0016,
+      "step": 3274
+    },
+    {
+      "epoch": 0.21105883869304634,
+      "grad_norm": 0.979300330893555,
+      "learning_rate": 8.76763336913713e-06,
+      "loss": 0.0006,
+      "step": 3275
+    },
+    {
+      "epoch": 0.2111232841399755,
+      "grad_norm": 0.004026414319159687,
+      "learning_rate": 8.766917293233083e-06,
+      "loss": 0.0,
+      "step": 3276
+    },
+    {
+      "epoch": 0.21118772958690468,
+      "grad_norm": 1.9066745097770412,
+      "learning_rate": 8.766201217329037e-06,
+      "loss": 0.0091,
+      "step": 3277
+    },
+    {
+      "epoch": 0.21125217503383387,
+      "grad_norm": 0.02170203435620058,
+      "learning_rate": 8.765485141424992e-06,
+      "loss": 0.0001,
+      "step": 3278
+    },
+    {
+      "epoch": 0.21131662048076302,
+      "grad_norm": 0.001396143696547249,
+      "learning_rate": 8.764769065520946e-06,
+      "loss": 0.0,
+      "step": 3279
+    },
+    {
+      "epoch": 0.2113810659276922,
+      "grad_norm": 0.07585580675491908,
+      "learning_rate": 8.7640529896169e-06,
+      "loss": 0.0002,
+      "step": 3280
+    },
+    {
+      "epoch": 0.2114455113746214,
+      "grad_norm": 0.05296037073165522,
+      "learning_rate": 8.763336913712854e-06,
+      "loss": 0.0001,
+      "step": 3281
+    },
+    {
+      "epoch": 0.21150995682155055,
+      "grad_norm": 0.00801547689700731,
+      "learning_rate": 8.762620837808809e-06,
+      "loss": 0.0,
+      "step": 3282
+    },
+    {
+      "epoch": 0.21157440226847973,
+      "grad_norm": 0.05558661669894557,
+      "learning_rate": 8.761904761904763e-06,
+      "loss": 0.0001,
+      "step": 3283
+    },
+    {
+      "epoch": 0.21163884771540892,
+      "grad_norm": 0.03368207035083663,
+      "learning_rate": 8.761188686000717e-06,
+      "loss": 0.0,
+      "step": 3284
+    },
+    {
+      "epoch": 0.21170329316233807,
+      "grad_norm": 0.16717958253313203,
+      "learning_rate": 8.760472610096672e-06,
+      "loss": 0.0005,
+      "step": 3285
+    },
+    {
+      "epoch": 0.21176773860926726,
+      "grad_norm": 0.20278401364321336,
+      "learning_rate": 8.759756534192624e-06,
+      "loss": 0.002,
+      "step": 3286
+    },
+    {
+      "epoch": 0.21183218405619644,
+      "grad_norm": 0.03134582958365876,
+      "learning_rate": 8.759040458288579e-06,
+      "loss": 0.0002,
+      "step": 3287
+    },
+    {
+      "epoch": 0.2118966295031256,
+      "grad_norm": 0.003064029577034968,
+      "learning_rate": 8.758324382384533e-06,
+      "loss": 0.0,
+      "step": 3288
+    },
+    {
+      "epoch": 0.21196107495005478,
+      "grad_norm": 0.013964800171850807,
+      "learning_rate": 8.757608306480487e-06,
+      "loss": 0.0001,
+      "step": 3289
+    },
+    {
+      "epoch": 0.21202552039698394,
+      "grad_norm": 0.3455218362415116,
+      "learning_rate": 8.756892230576441e-06,
+      "loss": 0.0008,
+      "step": 3290
+    },
+    {
+      "epoch": 0.21208996584391313,
+      "grad_norm": 0.1274534928593735,
+      "learning_rate": 8.756176154672396e-06,
+      "loss": 0.0003,
+      "step": 3291
+    },
+    {
+      "epoch": 0.2121544112908423,
+      "grad_norm": 0.02204250914795714,
+      "learning_rate": 8.75546007876835e-06,
+      "loss": 0.0001,
+      "step": 3292
+    },
+    {
+      "epoch": 0.21221885673777147,
+      "grad_norm": 0.0020332618231904947,
+      "learning_rate": 8.754744002864304e-06,
+      "loss": 0.0,
+      "step": 3293
+    },
+    {
+      "epoch": 0.21228330218470065,
+      "grad_norm": 0.010261200257673677,
+      "learning_rate": 8.754027926960259e-06,
+      "loss": 0.0,
+      "step": 3294
+    },
+    {
+      "epoch": 0.21234774763162984,
+      "grad_norm": 0.005997398083936662,
+      "learning_rate": 8.753311851056213e-06,
+      "loss": 0.0,
+      "step": 3295
+    },
+    {
+      "epoch": 0.212412193078559,
+      "grad_norm": 0.007660184481136306,
+      "learning_rate": 8.752595775152167e-06,
+      "loss": 0.0,
+      "step": 3296
+    },
+    {
+      "epoch": 0.21247663852548818,
+      "grad_norm": 0.038827415225720155,
+      "learning_rate": 8.751879699248122e-06,
+      "loss": 0.0,
+      "step": 3297
+    },
+    {
+      "epoch": 0.21254108397241736,
+      "grad_norm": 0.33644458291898005,
+      "learning_rate": 8.751163623344076e-06,
+      "loss": 0.0024,
+      "step": 3298
+    },
+    {
+      "epoch": 0.21260552941934652,
+      "grad_norm": 0.017198641204569848,
+      "learning_rate": 8.75044754744003e-06,
+      "loss": 0.0001,
+      "step": 3299
+    },
+    {
+      "epoch": 0.2126699748662757,
+      "grad_norm": 0.15987451125906116,
+      "learning_rate": 8.749731471535984e-06,
+      "loss": 0.0014,
+      "step": 3300
+    },
+    {
+      "epoch": 0.21273442031320486,
+      "grad_norm": 0.007898780452521827,
+      "learning_rate": 8.749015395631939e-06,
+      "loss": 0.0,
+      "step": 3301
+    },
+    {
+      "epoch": 0.21279886576013404,
+      "grad_norm": 0.2562249607552893,
+      "learning_rate": 8.748299319727891e-06,
+      "loss": 0.003,
+      "step": 3302
+    },
+    {
+      "epoch": 0.21286331120706323,
+      "grad_norm": 0.0024193017311194913,
+      "learning_rate": 8.747583243823846e-06,
+      "loss": 0.0,
+      "step": 3303
+    },
+    {
+      "epoch": 0.21292775665399238,
+      "grad_norm": 0.005256035814212943,
+      "learning_rate": 8.7468671679198e-06,
+      "loss": 0.0,
+      "step": 3304
+    },
+    {
+      "epoch": 0.21299220210092157,
+      "grad_norm": 0.012534387630057374,
+      "learning_rate": 8.746151092015754e-06,
+      "loss": 0.0,
+      "step": 3305
+    },
+    {
+      "epoch": 0.21305664754785075,
+      "grad_norm": 0.002335593572439098,
+      "learning_rate": 8.745435016111709e-06,
+      "loss": 0.0,
+      "step": 3306
+    },
+    {
+      "epoch": 0.2131210929947799,
+      "grad_norm": 3.6138689872499303,
+      "learning_rate": 8.744718940207663e-06,
+      "loss": 0.0088,
+      "step": 3307
+    },
+    {
+      "epoch": 0.2131855384417091,
+      "grad_norm": 0.021828677739768283,
+      "learning_rate": 8.744002864303617e-06,
+      "loss": 0.0001,
+      "step": 3308
+    },
+    {
+      "epoch": 0.21324998388863828,
+      "grad_norm": 0.008283227528941124,
+      "learning_rate": 8.743286788399571e-06,
+      "loss": 0.0,
+      "step": 3309
+    },
+    {
+      "epoch": 0.21331442933556743,
+      "grad_norm": 0.05812750326426742,
+      "learning_rate": 8.742570712495526e-06,
+      "loss": 0.0006,
+      "step": 3310
+    },
+    {
+      "epoch": 0.21337887478249662,
+      "grad_norm": 0.016621163486587685,
+      "learning_rate": 8.741854636591478e-06,
+      "loss": 0.0,
+      "step": 3311
+    },
+    {
+      "epoch": 0.2134433202294258,
+      "grad_norm": 0.00017105550687940947,
+      "learning_rate": 8.741138560687433e-06,
+      "loss": 0.0,
+      "step": 3312
+    },
+    {
+      "epoch": 0.21350776567635496,
+      "grad_norm": 0.6441706949390169,
+      "learning_rate": 8.740422484783387e-06,
+      "loss": 0.0009,
+      "step": 3313
+    },
+    {
+      "epoch": 0.21357221112328414,
+      "grad_norm": 0.04369000739798586,
+      "learning_rate": 8.739706408879341e-06,
+      "loss": 0.0,
+      "step": 3314
+    },
+    {
+      "epoch": 0.2136366565702133,
+      "grad_norm": 0.09402632719574942,
+      "learning_rate": 8.738990332975296e-06,
+      "loss": 0.0003,
+      "step": 3315
+    },
+    {
+      "epoch": 0.21370110201714249,
+      "grad_norm": 0.05016807446108145,
+      "learning_rate": 8.73827425707125e-06,
+      "loss": 0.0001,
+      "step": 3316
+    },
+    {
+      "epoch": 0.21376554746407167,
+      "grad_norm": 0.010129704308617241,
+      "learning_rate": 8.737558181167204e-06,
+      "loss": 0.0,
+      "step": 3317
+    },
+    {
+      "epoch": 0.21382999291100083,
+      "grad_norm": 0.06499619789764004,
+      "learning_rate": 8.736842105263158e-06,
+      "loss": 0.0001,
+      "step": 3318
+    },
+    {
+      "epoch": 0.21389443835793,
+      "grad_norm": 0.005310078119373596,
+      "learning_rate": 8.736126029359113e-06,
+      "loss": 0.0,
+      "step": 3319
+    },
+    {
+      "epoch": 0.2139588838048592,
+      "grad_norm": 0.0006348497178582359,
+      "learning_rate": 8.735409953455067e-06,
+      "loss": 0.0,
+      "step": 3320
+    },
+    {
+      "epoch": 0.21402332925178835,
+      "grad_norm": 0.00472366162278607,
+      "learning_rate": 8.734693877551021e-06,
+      "loss": 0.0,
+      "step": 3321
+    },
+    {
+      "epoch": 0.21408777469871754,
+      "grad_norm": 0.015531686896680753,
+      "learning_rate": 8.733977801646976e-06,
+      "loss": 0.0002,
+      "step": 3322
+    },
+    {
+      "epoch": 0.21415222014564672,
+      "grad_norm": 0.02151993293429727,
+      "learning_rate": 8.73326172574293e-06,
+      "loss": 0.0002,
+      "step": 3323
+    },
+    {
+      "epoch": 0.21421666559257588,
+      "grad_norm": 0.010721432144462448,
+      "learning_rate": 8.732545649838884e-06,
+      "loss": 0.0,
+      "step": 3324
+    },
+    {
+      "epoch": 0.21428111103950506,
+      "grad_norm": 0.07421909194164812,
+      "learning_rate": 8.731829573934839e-06,
+      "loss": 0.0034,
+      "step": 3325
+    },
+    {
+      "epoch": 0.21434555648643425,
+      "grad_norm": 0.003490125792385133,
+      "learning_rate": 8.731113498030793e-06,
+      "loss": 0.0,
+      "step": 3326
+    },
+    {
+      "epoch": 0.2144100019333634,
+      "grad_norm": 0.6003774389582377,
+      "learning_rate": 8.730397422126745e-06,
+      "loss": 0.003,
+      "step": 3327
+    },
+    {
+      "epoch": 0.2144744473802926,
+      "grad_norm": 0.0031710861298442842,
+      "learning_rate": 8.7296813462227e-06,
+      "loss": 0.0,
+      "step": 3328
+    },
+    {
+      "epoch": 0.21453889282722174,
+      "grad_norm": 0.006062084484470189,
+      "learning_rate": 8.728965270318654e-06,
+      "loss": 0.0,
+      "step": 3329
+    },
+    {
+      "epoch": 0.21460333827415093,
+      "grad_norm": 0.0046390183140608615,
+      "learning_rate": 8.728249194414608e-06,
+      "loss": 0.0,
+      "step": 3330
+    },
+    {
+      "epoch": 0.2146677837210801,
+      "grad_norm": 0.0028126984205885065,
+      "learning_rate": 8.727533118510563e-06,
+      "loss": 0.0,
+      "step": 3331
+    },
+    {
+      "epoch": 0.21473222916800927,
+      "grad_norm": 0.33433982271918034,
+      "learning_rate": 8.726817042606517e-06,
+      "loss": 0.0026,
+      "step": 3332
+    },
+    {
+      "epoch": 0.21479667461493845,
+      "grad_norm": 0.09134759936800937,
+      "learning_rate": 8.726100966702471e-06,
+      "loss": 0.0004,
+      "step": 3333
+    },
+    {
+      "epoch": 0.21486112006186764,
+      "grad_norm": 0.10058142364892066,
+      "learning_rate": 8.725384890798425e-06,
+      "loss": 0.0001,
+      "step": 3334
+    },
+    {
+      "epoch": 0.2149255655087968,
+      "grad_norm": 0.0016853936002382036,
+      "learning_rate": 8.72466881489438e-06,
+      "loss": 0.0,
+      "step": 3335
+    },
+    {
+      "epoch": 0.21499001095572598,
+      "grad_norm": 0.15968984634818126,
+      "learning_rate": 8.723952738990334e-06,
+      "loss": 0.0004,
+      "step": 3336
+    },
+    {
+      "epoch": 0.21505445640265516,
+      "grad_norm": 0.014139192268035285,
+      "learning_rate": 8.723236663086287e-06,
+      "loss": 0.0,
+      "step": 3337
+    },
+    {
+      "epoch": 0.21511890184958432,
+      "grad_norm": 0.0609662983623595,
+      "learning_rate": 8.722520587182241e-06,
+      "loss": 0.0001,
+      "step": 3338
+    },
+    {
+      "epoch": 0.2151833472965135,
+      "grad_norm": 0.00030584942007946705,
+      "learning_rate": 8.721804511278195e-06,
+      "loss": 0.0,
+      "step": 3339
+    },
+    {
+      "epoch": 0.2152477927434427,
+      "grad_norm": 0.010851115075375938,
+      "learning_rate": 8.72108843537415e-06,
+      "loss": 0.0,
+      "step": 3340
+    },
+    {
+      "epoch": 0.21531223819037185,
+      "grad_norm": 0.0025828937431415975,
+      "learning_rate": 8.720372359470106e-06,
+      "loss": 0.0,
+      "step": 3341
+    },
+    {
+      "epoch": 0.21537668363730103,
+      "grad_norm": 0.27412838166825926,
+      "learning_rate": 8.71965628356606e-06,
+      "loss": 0.0002,
+      "step": 3342
+    },
+    {
+      "epoch": 0.2154411290842302,
+      "grad_norm": 0.0014394433601773147,
+      "learning_rate": 8.718940207662012e-06,
+      "loss": 0.0,
+      "step": 3343
+    },
+    {
+      "epoch": 0.21550557453115937,
+      "grad_norm": 0.03302873013551384,
+      "learning_rate": 8.718224131757967e-06,
+      "loss": 0.0001,
+      "step": 3344
+    },
+    {
+      "epoch": 0.21557001997808856,
+      "grad_norm": 0.003994157454168405,
+      "learning_rate": 8.717508055853921e-06,
+      "loss": 0.0,
+      "step": 3345
+    },
+    {
+      "epoch": 0.2156344654250177,
+      "grad_norm": 0.03668763155929853,
+      "learning_rate": 8.716791979949875e-06,
+      "loss": 0.0001,
+      "step": 3346
+    },
+    {
+      "epoch": 0.2156989108719469,
+      "grad_norm": 0.0035984155267912584,
+      "learning_rate": 8.71607590404583e-06,
+      "loss": 0.0,
+      "step": 3347
+    },
+    {
+      "epoch": 0.21576335631887608,
+      "grad_norm": 0.000645077782702887,
+      "learning_rate": 8.715359828141784e-06,
+      "loss": 0.0,
+      "step": 3348
+    },
+    {
+      "epoch": 0.21582780176580524,
+      "grad_norm": 0.0007157999193808737,
+      "learning_rate": 8.714643752237738e-06,
+      "loss": 0.0,
+      "step": 3349
+    },
+    {
+      "epoch": 0.21589224721273442,
+      "grad_norm": 0.08779469924341486,
+      "learning_rate": 8.713927676333693e-06,
+      "loss": 0.0007,
+      "step": 3350
+    },
+    {
+      "epoch": 0.2159566926596636,
+      "grad_norm": 0.0009734634672225255,
+      "learning_rate": 8.713211600429647e-06,
+      "loss": 0.0,
+      "step": 3351
+    },
+    {
+      "epoch": 0.21602113810659276,
+      "grad_norm": 0.009129763667514236,
+      "learning_rate": 8.712495524525601e-06,
+      "loss": 0.0001,
+      "step": 3352
+    },
+    {
+      "epoch": 0.21608558355352195,
+      "grad_norm": 0.0015141456217384983,
+      "learning_rate": 8.711779448621554e-06,
+      "loss": 0.0,
+      "step": 3353
+    },
+    {
+      "epoch": 0.2161500290004511,
+      "grad_norm": 0.009750178906905728,
+      "learning_rate": 8.711063372717508e-06,
+      "loss": 0.0,
+      "step": 3354
+    },
+    {
+      "epoch": 0.2162144744473803,
+      "grad_norm": 0.7822016070286729,
+      "learning_rate": 8.710347296813462e-06,
+      "loss": 0.0092,
+      "step": 3355
+    },
+    {
+      "epoch": 0.21627891989430947,
+      "grad_norm": 0.0008353497719524981,
+      "learning_rate": 8.709631220909417e-06,
+      "loss": 0.0,
+      "step": 3356
+    },
+    {
+      "epoch": 0.21634336534123863,
+      "grad_norm": 0.0837409934184648,
+      "learning_rate": 8.708915145005371e-06,
+      "loss": 0.0001,
+      "step": 3357
+    },
+    {
+      "epoch": 0.21640781078816781,
+      "grad_norm": 0.022414822248022617,
+      "learning_rate": 8.708199069101325e-06,
+      "loss": 0.0001,
+      "step": 3358
+    },
+    {
+      "epoch": 0.216472256235097,
+      "grad_norm": 0.11345686946837005,
+      "learning_rate": 8.70748299319728e-06,
+      "loss": 0.0001,
+      "step": 3359
+    },
+    {
+      "epoch": 0.21653670168202616,
+      "grad_norm": 0.08448661016414556,
+      "learning_rate": 8.706766917293234e-06,
+      "loss": 0.0009,
+      "step": 3360
+    },
+    {
+      "epoch": 0.21660114712895534,
+      "grad_norm": 0.0220006539202053,
+      "learning_rate": 8.706050841389188e-06,
+      "loss": 0.0,
+      "step": 3361
+    },
+    {
+      "epoch": 0.21666559257588452,
+      "grad_norm": 0.08631784358443949,
+      "learning_rate": 8.705334765485142e-06,
+      "loss": 0.0001,
+      "step": 3362
+    },
+    {
+      "epoch": 0.21673003802281368,
+      "grad_norm": 0.029033055171980918,
+      "learning_rate": 8.704618689581095e-06,
+      "loss": 0.0001,
+      "step": 3363
+    },
+    {
+      "epoch": 0.21679448346974287,
+      "grad_norm": 0.4350533533262354,
+      "learning_rate": 8.703902613677051e-06,
+      "loss": 0.0039,
+      "step": 3364
+    },
+    {
+      "epoch": 0.21685892891667205,
+      "grad_norm": 0.001498889728926459,
+      "learning_rate": 8.703186537773005e-06,
+      "loss": 0.0,
+      "step": 3365
+    },
+    {
+      "epoch": 0.2169233743636012,
+      "grad_norm": 0.010148400387620866,
+      "learning_rate": 8.70247046186896e-06,
+      "loss": 0.0,
+      "step": 3366
+    },
+    {
+      "epoch": 0.2169878198105304,
+      "grad_norm": 0.007033440864080341,
+      "learning_rate": 8.701754385964914e-06,
+      "loss": 0.0,
+      "step": 3367
+    },
+    {
+      "epoch": 0.21705226525745955,
+      "grad_norm": 0.05006730273660623,
+      "learning_rate": 8.701038310060868e-06,
+      "loss": 0.0001,
+      "step": 3368
+    },
+    {
+      "epoch": 0.21711671070438873,
+      "grad_norm": 0.007402992324705263,
+      "learning_rate": 8.700322234156821e-06,
+      "loss": 0.0,
+      "step": 3369
+    },
+    {
+      "epoch": 0.21718115615131792,
+      "grad_norm": 0.0057464255085265354,
+      "learning_rate": 8.699606158252775e-06,
+      "loss": 0.0,
+      "step": 3370
+    },
+    {
+      "epoch": 0.21724560159824707,
+      "grad_norm": 0.05918935218428574,
+      "learning_rate": 8.69889008234873e-06,
+      "loss": 0.0003,
+      "step": 3371
+    },
+    {
+      "epoch": 0.21731004704517626,
+      "grad_norm": 0.025653915972399525,
+      "learning_rate": 8.698174006444684e-06,
+      "loss": 0.0002,
+      "step": 3372
+    },
+    {
+      "epoch": 0.21737449249210544,
+      "grad_norm": 0.012820261677368707,
+      "learning_rate": 8.697457930540638e-06,
+      "loss": 0.0002,
+      "step": 3373
+    },
+    {
+      "epoch": 0.2174389379390346,
+      "grad_norm": 0.003222630195440105,
+      "learning_rate": 8.696741854636592e-06,
+      "loss": 0.0,
+      "step": 3374
+    },
+    {
+      "epoch": 0.21750338338596378,
+      "grad_norm": 0.005536295847970089,
+      "learning_rate": 8.696025778732547e-06,
+      "loss": 0.0,
+      "step": 3375
+    },
+    {
+      "epoch": 0.21756782883289297,
+      "grad_norm": 0.0475632780772114,
+      "learning_rate": 8.695309702828501e-06,
+      "loss": 0.0001,
+      "step": 3376
+    },
+    {
+      "epoch": 0.21763227427982212,
+      "grad_norm": 0.013851927429286565,
+      "learning_rate": 8.694593626924455e-06,
+      "loss": 0.0,
+      "step": 3377
+    },
+    {
+      "epoch": 0.2176967197267513,
+      "grad_norm": 0.006665526372697656,
+      "learning_rate": 8.69387755102041e-06,
+      "loss": 0.0,
+      "step": 3378
+    },
+    {
+      "epoch": 0.2177611651736805,
+      "grad_norm": 0.0003302123249551213,
+      "learning_rate": 8.693161475116362e-06,
+      "loss": 0.0,
+      "step": 3379
+    },
+    {
+      "epoch": 0.21782561062060965,
+      "grad_norm": 0.04042210473255954,
+      "learning_rate": 8.692445399212316e-06,
+      "loss": 0.0,
+      "step": 3380
+    },
+    {
+      "epoch": 0.21789005606753883,
+      "grad_norm": 0.032240357119762836,
+      "learning_rate": 8.69172932330827e-06,
+      "loss": 0.0001,
+      "step": 3381
+    },
+    {
+      "epoch": 0.217954501514468,
+      "grad_norm": 0.028356624417666038,
+      "learning_rate": 8.691013247404225e-06,
+      "loss": 0.0001,
+      "step": 3382
+    },
+    {
+      "epoch": 0.21801894696139718,
+      "grad_norm": 0.24312603511442854,
+      "learning_rate": 8.69029717150018e-06,
+      "loss": 0.0007,
+      "step": 3383
+    },
+    {
+      "epoch": 0.21808339240832636,
+      "grad_norm": 0.06490065794875387,
+      "learning_rate": 8.689581095596134e-06,
+      "loss": 0.0001,
+      "step": 3384
+    },
+    {
+      "epoch": 0.21814783785525552,
+      "grad_norm": 0.0003474384346755857,
+      "learning_rate": 8.688865019692088e-06,
+      "loss": 0.0,
+      "step": 3385
+    },
+    {
+      "epoch": 0.2182122833021847,
+      "grad_norm": 0.010435808402401832,
+      "learning_rate": 8.688148943788042e-06,
+      "loss": 0.0001,
+      "step": 3386
+    },
+    {
+      "epoch": 0.21827672874911389,
+      "grad_norm": 0.10267596869027987,
+      "learning_rate": 8.687432867883997e-06,
+      "loss": 0.0004,
+      "step": 3387
+    },
+    {
+      "epoch": 0.21834117419604304,
+      "grad_norm": 0.00680259145453217,
+      "learning_rate": 8.686716791979951e-06,
+      "loss": 0.0,
+      "step": 3388
+    },
+    {
+      "epoch": 0.21840561964297223,
+      "grad_norm": 0.0007357830666917632,
+      "learning_rate": 8.686000716075905e-06,
+      "loss": 0.0,
+      "step": 3389
+    },
+    {
+      "epoch": 0.2184700650899014,
+      "grad_norm": 0.017961030447330713,
+      "learning_rate": 8.68528464017186e-06,
+      "loss": 0.0,
+      "step": 3390
+    },
+    {
+      "epoch": 0.21853451053683057,
+      "grad_norm": 0.015245958178808318,
+      "learning_rate": 8.684568564267814e-06,
+      "loss": 0.0,
+      "step": 3391
+    },
+    {
+      "epoch": 0.21859895598375975,
+      "grad_norm": 0.2522366576990323,
+      "learning_rate": 8.683852488363768e-06,
+      "loss": 0.0013,
+      "step": 3392
+    },
+    {
+      "epoch": 0.2186634014306889,
+      "grad_norm": 0.004980891866569009,
+      "learning_rate": 8.683136412459722e-06,
+      "loss": 0.0,
+      "step": 3393
+    },
+    {
+      "epoch": 0.2187278468776181,
+      "grad_norm": 0.005101799462129407,
+      "learning_rate": 8.682420336555677e-06,
+      "loss": 0.0,
+      "step": 3394
+    },
+    {
+      "epoch": 0.21879229232454728,
+      "grad_norm": 0.0034149603184703468,
+      "learning_rate": 8.68170426065163e-06,
+      "loss": 0.0,
+      "step": 3395
+    },
+    {
+      "epoch": 0.21885673777147643,
+      "grad_norm": 0.03652058159275285,
+      "learning_rate": 8.680988184747584e-06,
+      "loss": 0.0002,
+      "step": 3396
+    },
+    {
+      "epoch": 0.21892118321840562,
+      "grad_norm": 0.007019698589223098,
+      "learning_rate": 8.680272108843538e-06,
+      "loss": 0.0,
+      "step": 3397
+    },
+    {
+      "epoch": 0.2189856286653348,
+      "grad_norm": 0.0005208419070086863,
+      "learning_rate": 8.679556032939492e-06,
+      "loss": 0.0,
+      "step": 3398
+    },
+    {
+      "epoch": 0.21905007411226396,
+      "grad_norm": 0.022129151887111145,
+      "learning_rate": 8.678839957035446e-06,
+      "loss": 0.0,
+      "step": 3399
+    },
+    {
+      "epoch": 0.21911451955919314,
+      "grad_norm": 0.008741674890565838,
+      "learning_rate": 8.6781238811314e-06,
+      "loss": 0.0,
+      "step": 3400
+    },
+    {
+      "epoch": 0.21917896500612233,
+      "grad_norm": 0.0028274312509968943,
+      "learning_rate": 8.677407805227355e-06,
+      "loss": 0.0,
+      "step": 3401
+    },
+    {
+      "epoch": 0.21924341045305148,
+      "grad_norm": 0.027357096054692588,
+      "learning_rate": 8.67669172932331e-06,
+      "loss": 0.0001,
+      "step": 3402
+    },
+    {
+      "epoch": 0.21930785589998067,
+      "grad_norm": 9.549562120148108e-05,
+      "learning_rate": 8.675975653419264e-06,
+      "loss": 0.0,
+      "step": 3403
+    },
+    {
+      "epoch": 0.21937230134690985,
+      "grad_norm": 0.00035971321640775695,
+      "learning_rate": 8.675259577515216e-06,
+      "loss": 0.0,
+      "step": 3404
+    },
+    {
+      "epoch": 0.219436746793839,
+      "grad_norm": 0.001148672309388114,
+      "learning_rate": 8.67454350161117e-06,
+      "loss": 0.0,
+      "step": 3405
+    },
+    {
+      "epoch": 0.2195011922407682,
+      "grad_norm": 0.0022274004349375665,
+      "learning_rate": 8.673827425707125e-06,
+      "loss": 0.0,
+      "step": 3406
+    },
+    {
+      "epoch": 0.21956563768769735,
+      "grad_norm": 0.001988663302904296,
+      "learning_rate": 8.673111349803079e-06,
+      "loss": 0.0,
+      "step": 3407
+    },
+    {
+      "epoch": 0.21963008313462654,
+      "grad_norm": 0.004393998501194764,
+      "learning_rate": 8.672395273899033e-06,
+      "loss": 0.0,
+      "step": 3408
+    },
+    {
+      "epoch": 0.21969452858155572,
+      "grad_norm": 0.026292318215849746,
+      "learning_rate": 8.671679197994988e-06,
+      "loss": 0.0,
+      "step": 3409
+    },
+    {
+      "epoch": 0.21975897402848488,
+      "grad_norm": 0.037110821168728526,
+      "learning_rate": 8.670963122090942e-06,
+      "loss": 0.0004,
+      "step": 3410
+    },
+    {
+      "epoch": 0.21982341947541406,
+      "grad_norm": 0.0004869931471118294,
+      "learning_rate": 8.670247046186896e-06,
+      "loss": 0.0,
+      "step": 3411
+    },
+    {
+      "epoch": 0.21988786492234325,
+      "grad_norm": 0.035249077579844044,
+      "learning_rate": 8.66953097028285e-06,
+      "loss": 0.0003,
+      "step": 3412
+    },
+    {
+      "epoch": 0.2199523103692724,
+      "grad_norm": 0.001970382648606595,
+      "learning_rate": 8.668814894378805e-06,
+      "loss": 0.0,
+      "step": 3413
+    },
+    {
+      "epoch": 0.2200167558162016,
+      "grad_norm": 0.011623401973596743,
+      "learning_rate": 8.66809881847476e-06,
+      "loss": 0.0001,
+      "step": 3414
+    },
+    {
+      "epoch": 0.22008120126313077,
+      "grad_norm": 0.00034764643982887803,
+      "learning_rate": 8.667382742570714e-06,
+      "loss": 0.0,
+      "step": 3415
+    },
+    {
+      "epoch": 0.22014564671005993,
+      "grad_norm": 8.975251794623639e-05,
+      "learning_rate": 8.666666666666668e-06,
+      "loss": 0.0,
+      "step": 3416
+    },
+    {
+      "epoch": 0.2202100921569891,
+      "grad_norm": 0.07822764627206331,
+      "learning_rate": 8.665950590762622e-06,
+      "loss": 0.0007,
+      "step": 3417
+    },
+    {
+      "epoch": 0.2202745376039183,
+      "grad_norm": 0.020361744152594136,
+      "learning_rate": 8.665234514858576e-06,
+      "loss": 0.0001,
+      "step": 3418
+    },
+    {
+      "epoch": 0.22033898305084745,
+      "grad_norm": 0.031516272508776165,
+      "learning_rate": 8.66451843895453e-06,
+      "loss": 0.0001,
+      "step": 3419
+    },
+    {
+      "epoch": 0.22040342849777664,
+      "grad_norm": 0.0008974531050145935,
+      "learning_rate": 8.663802363050483e-06,
+      "loss": 0.0,
+      "step": 3420
+    },
+    {
+      "epoch": 0.2204678739447058,
+      "grad_norm": 0.16912508093612774,
+      "learning_rate": 8.663086287146438e-06,
+      "loss": 0.0003,
+      "step": 3421
+    },
+    {
+      "epoch": 0.22053231939163498,
+      "grad_norm": 0.00018829353512949963,
+      "learning_rate": 8.662370211242392e-06,
+      "loss": 0.0,
+      "step": 3422
+    },
+    {
+      "epoch": 0.22059676483856416,
+      "grad_norm": 0.00521659484946839,
+      "learning_rate": 8.661654135338346e-06,
+      "loss": 0.0,
+      "step": 3423
+    },
+    {
+      "epoch": 0.22066121028549332,
+      "grad_norm": 0.35939284985206355,
+      "learning_rate": 8.6609380594343e-06,
+      "loss": 0.0027,
+      "step": 3424
+    },
+    {
+      "epoch": 0.2207256557324225,
+      "grad_norm": 0.07403550301307933,
+      "learning_rate": 8.660221983530255e-06,
+      "loss": 0.0009,
+      "step": 3425
+    },
+    {
+      "epoch": 0.2207901011793517,
+      "grad_norm": 0.004889930625470678,
+      "learning_rate": 8.659505907626209e-06,
+      "loss": 0.0,
+      "step": 3426
+    },
+    {
+      "epoch": 0.22085454662628085,
+      "grad_norm": 0.003421520565400845,
+      "learning_rate": 8.658789831722163e-06,
+      "loss": 0.0,
+      "step": 3427
+    },
+    {
+      "epoch": 0.22091899207321003,
+      "grad_norm": 0.0053065978274310565,
+      "learning_rate": 8.658073755818118e-06,
+      "loss": 0.0,
+      "step": 3428
+    },
+    {
+      "epoch": 0.22098343752013921,
+      "grad_norm": 0.013211383273678805,
+      "learning_rate": 8.657357679914072e-06,
+      "loss": 0.0001,
+      "step": 3429
+    },
+    {
+      "epoch": 0.22104788296706837,
+      "grad_norm": 0.0009982449528573667,
+      "learning_rate": 8.656641604010025e-06,
+      "loss": 0.0,
+      "step": 3430
+    },
+    {
+      "epoch": 0.22111232841399756,
+      "grad_norm": 0.0015086585130352167,
+      "learning_rate": 8.655925528105979e-06,
+      "loss": 0.0,
+      "step": 3431
+    },
+    {
+      "epoch": 0.2211767738609267,
+      "grad_norm": 0.02614516921378581,
+      "learning_rate": 8.655209452201933e-06,
+      "loss": 0.0001,
+      "step": 3432
+    },
+    {
+      "epoch": 0.2212412193078559,
+      "grad_norm": 0.018234096510467625,
+      "learning_rate": 8.654493376297888e-06,
+      "loss": 0.0001,
+      "step": 3433
+    },
+    {
+      "epoch": 0.22130566475478508,
+      "grad_norm": 0.0011630213225405184,
+      "learning_rate": 8.653777300393844e-06,
+      "loss": 0.0,
+      "step": 3434
+    },
+    {
+      "epoch": 0.22137011020171424,
+      "grad_norm": 0.7110074187051323,
+      "learning_rate": 8.653061224489798e-06,
+      "loss": 0.0061,
+      "step": 3435
+    },
+    {
+      "epoch": 0.22143455564864342,
+      "grad_norm": 0.00033229507672846427,
+      "learning_rate": 8.65234514858575e-06,
+      "loss": 0.0,
+      "step": 3436
+    },
+    {
+      "epoch": 0.2214990010955726,
+      "grad_norm": 0.002534487956548706,
+      "learning_rate": 8.651629072681705e-06,
+      "loss": 0.0,
+      "step": 3437
+    },
+    {
+      "epoch": 0.22156344654250176,
+      "grad_norm": 0.17261034087608992,
+      "learning_rate": 8.650912996777659e-06,
+      "loss": 0.0003,
+      "step": 3438
+    },
+    {
+      "epoch": 0.22162789198943095,
+      "grad_norm": 0.002774698501918162,
+      "learning_rate": 8.650196920873613e-06,
+      "loss": 0.0,
+      "step": 3439
+    },
+    {
+      "epoch": 0.22169233743636013,
+      "grad_norm": 0.008339194776038943,
+      "learning_rate": 8.649480844969568e-06,
+      "loss": 0.0,
+      "step": 3440
+    },
+    {
+      "epoch": 0.2217567828832893,
+      "grad_norm": 0.004585672715958108,
+      "learning_rate": 8.648764769065522e-06,
+      "loss": 0.0,
+      "step": 3441
+    },
+    {
+      "epoch": 0.22182122833021847,
+      "grad_norm": 0.0030527512602284202,
+      "learning_rate": 8.648048693161476e-06,
+      "loss": 0.0,
+      "step": 3442
+    },
+    {
+      "epoch": 0.22188567377714766,
+      "grad_norm": 0.0939224861790215,
+      "learning_rate": 8.64733261725743e-06,
+      "loss": 0.0001,
+      "step": 3443
+    },
+    {
+      "epoch": 0.22195011922407681,
+      "grad_norm": 0.003773801458241794,
+      "learning_rate": 8.646616541353385e-06,
+      "loss": 0.0,
+      "step": 3444
+    },
+    {
+      "epoch": 0.222014564671006,
+      "grad_norm": 0.019486883556046962,
+      "learning_rate": 8.645900465449339e-06,
+      "loss": 0.0,
+      "step": 3445
+    },
+    {
+      "epoch": 0.22207901011793515,
+      "grad_norm": 0.07323750163029638,
+      "learning_rate": 8.645184389545292e-06,
+      "loss": 0.0005,
+      "step": 3446
+    },
+    {
+      "epoch": 0.22214345556486434,
+      "grad_norm": 0.00814745306731119,
+      "learning_rate": 8.644468313641246e-06,
+      "loss": 0.0,
+      "step": 3447
+    },
+    {
+      "epoch": 0.22220790101179352,
+      "grad_norm": 0.29745805104008016,
+      "learning_rate": 8.6437522377372e-06,
+      "loss": 0.0002,
+      "step": 3448
+    },
+    {
+      "epoch": 0.22227234645872268,
+      "grad_norm": 0.010189280976991384,
+      "learning_rate": 8.643036161833155e-06,
+      "loss": 0.0,
+      "step": 3449
+    },
+    {
+      "epoch": 0.22233679190565186,
+      "grad_norm": 0.350625219243027,
+      "learning_rate": 8.642320085929109e-06,
+      "loss": 0.0011,
+      "step": 3450
+    },
+    {
+      "epoch": 0.22240123735258105,
+      "grad_norm": 0.09586779868305924,
+      "learning_rate": 8.641604010025063e-06,
+      "loss": 0.0014,
+      "step": 3451
+    },
+    {
+      "epoch": 0.2224656827995102,
+      "grad_norm": 0.002273792676528464,
+      "learning_rate": 8.640887934121017e-06,
+      "loss": 0.0,
+      "step": 3452
+    },
+    {
+      "epoch": 0.2225301282464394,
+      "grad_norm": 0.0036304762924196463,
+      "learning_rate": 8.640171858216972e-06,
+      "loss": 0.0,
+      "step": 3453
+    },
+    {
+      "epoch": 0.22259457369336857,
+      "grad_norm": 0.24947871309025466,
+      "learning_rate": 8.639455782312926e-06,
+      "loss": 0.0037,
+      "step": 3454
+    },
+    {
+      "epoch": 0.22265901914029773,
+      "grad_norm": 0.00028613456653341616,
+      "learning_rate": 8.63873970640888e-06,
+      "loss": 0.0,
+      "step": 3455
+    },
+    {
+      "epoch": 0.22272346458722692,
+      "grad_norm": 7.548515570305155e-05,
+      "learning_rate": 8.638023630504833e-06,
+      "loss": 0.0,
+      "step": 3456
+    },
+    {
+      "epoch": 0.2227879100341561,
+      "grad_norm": 0.07466611177857914,
+      "learning_rate": 8.637307554600787e-06,
+      "loss": 0.0002,
+      "step": 3457
+    },
+    {
+      "epoch": 0.22285235548108526,
+      "grad_norm": 0.0011913602007814256,
+      "learning_rate": 8.636591478696743e-06,
+      "loss": 0.0,
+      "step": 3458
+    },
+    {
+      "epoch": 0.22291680092801444,
+      "grad_norm": 0.0005066447257724092,
+      "learning_rate": 8.635875402792698e-06,
+      "loss": 0.0,
+      "step": 3459
+    },
+    {
+      "epoch": 0.2229812463749436,
+      "grad_norm": 0.00195445982760958,
+      "learning_rate": 8.635159326888652e-06,
+      "loss": 0.0,
+      "step": 3460
+    },
+    {
+      "epoch": 0.22304569182187278,
+      "grad_norm": 0.00460945396853242,
+      "learning_rate": 8.634443250984606e-06,
+      "loss": 0.0,
+      "step": 3461
+    },
+    {
+      "epoch": 0.22311013726880197,
+      "grad_norm": 0.22631486151336772,
+      "learning_rate": 8.633727175080559e-06,
+      "loss": 0.0005,
+      "step": 3462
+    },
+    {
+      "epoch": 0.22317458271573112,
+      "grad_norm": 1.5542394097752636,
+      "learning_rate": 8.633011099176513e-06,
+      "loss": 0.0056,
+      "step": 3463
+    },
+    {
+      "epoch": 0.2232390281626603,
+      "grad_norm": 0.0003889234814934518,
+      "learning_rate": 8.632295023272467e-06,
+      "loss": 0.0,
+      "step": 3464
+    },
+    {
+      "epoch": 0.2233034736095895,
+      "grad_norm": 0.014474523316253513,
+      "learning_rate": 8.631578947368422e-06,
+      "loss": 0.0001,
+      "step": 3465
+    },
+    {
+      "epoch": 0.22336791905651865,
+      "grad_norm": 0.000770039970009564,
+      "learning_rate": 8.630862871464376e-06,
+      "loss": 0.0,
+      "step": 3466
+    },
+    {
+      "epoch": 0.22343236450344783,
+      "grad_norm": 0.02406613506883519,
+      "learning_rate": 8.63014679556033e-06,
+      "loss": 0.0002,
+      "step": 3467
+    },
+    {
+      "epoch": 0.22349680995037702,
+      "grad_norm": 0.8807060632696435,
+      "learning_rate": 8.629430719656285e-06,
+      "loss": 0.0038,
+      "step": 3468
+    },
+    {
+      "epoch": 0.22356125539730617,
+      "grad_norm": 0.010200052529701138,
+      "learning_rate": 8.628714643752239e-06,
+      "loss": 0.0001,
+      "step": 3469
+    },
+    {
+      "epoch": 0.22362570084423536,
+      "grad_norm": 0.004726596178967311,
+      "learning_rate": 8.627998567848193e-06,
+      "loss": 0.0001,
+      "step": 3470
+    },
+    {
+      "epoch": 0.22369014629116452,
+      "grad_norm": 0.0035062183509143477,
+      "learning_rate": 8.627282491944147e-06,
+      "loss": 0.0,
+      "step": 3471
+    },
+    {
+      "epoch": 0.2237545917380937,
+      "grad_norm": 0.006413737365501235,
+      "learning_rate": 8.6265664160401e-06,
+      "loss": 0.0,
+      "step": 3472
+    },
+    {
+      "epoch": 0.22381903718502288,
+      "grad_norm": 0.17002202513660772,
+      "learning_rate": 8.625850340136054e-06,
+      "loss": 0.002,
+      "step": 3473
+    },
+    {
+      "epoch": 0.22388348263195204,
+      "grad_norm": 0.000330794695075691,
+      "learning_rate": 8.625134264232009e-06,
+      "loss": 0.0,
+      "step": 3474
+    },
+    {
+      "epoch": 0.22394792807888123,
+      "grad_norm": 0.2625006482706694,
+      "learning_rate": 8.624418188327963e-06,
+      "loss": 0.0024,
+      "step": 3475
+    },
+    {
+      "epoch": 0.2240123735258104,
+      "grad_norm": 0.00048218474812608737,
+      "learning_rate": 8.623702112423917e-06,
+      "loss": 0.0,
+      "step": 3476
+    },
+    {
+      "epoch": 0.22407681897273957,
+      "grad_norm": 1.2268953297000005,
+      "learning_rate": 8.622986036519872e-06,
+      "loss": 0.0004,
+      "step": 3477
+    },
+    {
+      "epoch": 0.22414126441966875,
+      "grad_norm": 0.03838107792094849,
+      "learning_rate": 8.622269960615826e-06,
+      "loss": 0.0,
+      "step": 3478
+    },
+    {
+      "epoch": 0.22420570986659794,
+      "grad_norm": 0.0007060704934618854,
+      "learning_rate": 8.62155388471178e-06,
+      "loss": 0.0,
+      "step": 3479
+    },
+    {
+      "epoch": 0.2242701553135271,
+      "grad_norm": 0.01650770562641032,
+      "learning_rate": 8.620837808807734e-06,
+      "loss": 0.0,
+      "step": 3480
+    },
+    {
+      "epoch": 0.22433460076045628,
+      "grad_norm": 0.18594924064241675,
+      "learning_rate": 8.620121732903689e-06,
+      "loss": 0.0019,
+      "step": 3481
+    },
+    {
+      "epoch": 0.22439904620738546,
+      "grad_norm": 0.0001218993788027957,
+      "learning_rate": 8.619405656999643e-06,
+      "loss": 0.0,
+      "step": 3482
+    },
+    {
+      "epoch": 0.22446349165431462,
+      "grad_norm": 0.012246611324123326,
+      "learning_rate": 8.618689581095597e-06,
+      "loss": 0.0,
+      "step": 3483
+    },
+    {
+      "epoch": 0.2245279371012438,
+      "grad_norm": 0.0007658951102295896,
+      "learning_rate": 8.617973505191552e-06,
+      "loss": 0.0,
+      "step": 3484
+    },
+    {
+      "epoch": 0.22459238254817296,
+      "grad_norm": 0.024163193440480502,
+      "learning_rate": 8.617257429287506e-06,
+      "loss": 0.0002,
+      "step": 3485
+    },
+    {
+      "epoch": 0.22465682799510214,
+      "grad_norm": 0.00452679815357398,
+      "learning_rate": 8.61654135338346e-06,
+      "loss": 0.0,
+      "step": 3486
+    },
+    {
+      "epoch": 0.22472127344203133,
+      "grad_norm": 0.11612035923873511,
+      "learning_rate": 8.615825277479415e-06,
+      "loss": 0.0006,
+      "step": 3487
+    },
+    {
+      "epoch": 0.22478571888896048,
+      "grad_norm": 0.040972251040244025,
+      "learning_rate": 8.615109201575367e-06,
+      "loss": 0.0,
+      "step": 3488
+    },
+    {
+      "epoch": 0.22485016433588967,
+      "grad_norm": 0.001675342434363501,
+      "learning_rate": 8.614393125671321e-06,
+      "loss": 0.0,
+      "step": 3489
+    },
+    {
+      "epoch": 0.22491460978281885,
+      "grad_norm": 0.002689598627667504,
+      "learning_rate": 8.613677049767276e-06,
+      "loss": 0.0,
+      "step": 3490
+    },
+    {
+      "epoch": 0.224979055229748,
+      "grad_norm": 0.0001554998765619099,
+      "learning_rate": 8.61296097386323e-06,
+      "loss": 0.0,
+      "step": 3491
+    },
+    {
+      "epoch": 0.2250435006766772,
+      "grad_norm": 0.0007973969882462369,
+      "learning_rate": 8.612244897959184e-06,
+      "loss": 0.0,
+      "step": 3492
+    },
+    {
+      "epoch": 0.22510794612360638,
+      "grad_norm": 0.00010330225671837098,
+      "learning_rate": 8.611528822055139e-06,
+      "loss": 0.0,
+      "step": 3493
+    },
+    {
+      "epoch": 0.22517239157053553,
+      "grad_norm": 0.00965204197384073,
+      "learning_rate": 8.610812746151093e-06,
+      "loss": 0.0,
+      "step": 3494
+    },
+    {
+      "epoch": 0.22523683701746472,
+      "grad_norm": 0.28994619764491436,
+      "learning_rate": 8.610096670247047e-06,
+      "loss": 0.0006,
+      "step": 3495
+    },
+    {
+      "epoch": 0.2253012824643939,
+      "grad_norm": 0.45118794268115947,
+      "learning_rate": 8.609380594343002e-06,
+      "loss": 0.0011,
+      "step": 3496
+    },
+    {
+      "epoch": 0.22536572791132306,
+      "grad_norm": 0.056652606321714157,
+      "learning_rate": 8.608664518438954e-06,
+      "loss": 0.0001,
+      "step": 3497
+    },
+    {
+      "epoch": 0.22543017335825224,
+      "grad_norm": 0.002017892148967793,
+      "learning_rate": 8.607948442534908e-06,
+      "loss": 0.0,
+      "step": 3498
+    },
+    {
+      "epoch": 0.2254946188051814,
+      "grad_norm": 0.005166001572256319,
+      "learning_rate": 8.607232366630863e-06,
+      "loss": 0.0,
+      "step": 3499
+    },
+    {
+      "epoch": 0.22555906425211059,
+      "grad_norm": 0.002040933558769715,
+      "learning_rate": 8.606516290726817e-06,
+      "loss": 0.0,
+      "step": 3500
+    },
+    {
+      "epoch": 0.22562350969903977,
+      "grad_norm": 0.0007145762792561211,
+      "learning_rate": 8.605800214822771e-06,
+      "loss": 0.0,
+      "step": 3501
+    },
+    {
+      "epoch": 0.22568795514596893,
+      "grad_norm": 0.0016717906006389666,
+      "learning_rate": 8.605084138918726e-06,
+      "loss": 0.0,
+      "step": 3502
+    },
+    {
+      "epoch": 0.2257524005928981,
+      "grad_norm": 0.05478180477274337,
+      "learning_rate": 8.60436806301468e-06,
+      "loss": 0.0001,
+      "step": 3503
+    },
+    {
+      "epoch": 0.2258168460398273,
+      "grad_norm": 0.009749687123417545,
+      "learning_rate": 8.603651987110634e-06,
+      "loss": 0.0,
+      "step": 3504
+    },
+    {
+      "epoch": 0.22588129148675645,
+      "grad_norm": 0.023990730276947616,
+      "learning_rate": 8.602935911206589e-06,
+      "loss": 0.0,
+      "step": 3505
+    },
+    {
+      "epoch": 0.22594573693368564,
+      "grad_norm": 0.07996477268825748,
+      "learning_rate": 8.602219835302543e-06,
+      "loss": 0.0016,
+      "step": 3506
+    },
+    {
+      "epoch": 0.22601018238061482,
+      "grad_norm": 0.0019798777230370587,
+      "learning_rate": 8.601503759398497e-06,
+      "loss": 0.0,
+      "step": 3507
+    },
+    {
+      "epoch": 0.22607462782754398,
+      "grad_norm": 0.0003005708285917906,
+      "learning_rate": 8.600787683494451e-06,
+      "loss": 0.0,
+      "step": 3508
+    },
+    {
+      "epoch": 0.22613907327447316,
+      "grad_norm": 0.05497608294365255,
+      "learning_rate": 8.600071607590406e-06,
+      "loss": 0.0001,
+      "step": 3509
+    },
+    {
+      "epoch": 0.22620351872140232,
+      "grad_norm": 0.00847252292242908,
+      "learning_rate": 8.59935553168636e-06,
+      "loss": 0.0,
+      "step": 3510
+    },
+    {
+      "epoch": 0.2262679641683315,
+      "grad_norm": 0.0018614056583017003,
+      "learning_rate": 8.598639455782314e-06,
+      "loss": 0.0,
+      "step": 3511
+    },
+    {
+      "epoch": 0.2263324096152607,
+      "grad_norm": 0.005697144685217808,
+      "learning_rate": 8.597923379878269e-06,
+      "loss": 0.0,
+      "step": 3512
+    },
+    {
+      "epoch": 0.22639685506218984,
+      "grad_norm": 0.01307997836978058,
+      "learning_rate": 8.597207303974223e-06,
+      "loss": 0.0001,
+      "step": 3513
+    },
+    {
+      "epoch": 0.22646130050911903,
+      "grad_norm": 0.1307518360599829,
+      "learning_rate": 8.596491228070176e-06,
+      "loss": 0.0004,
+      "step": 3514
+    },
+    {
+      "epoch": 0.2265257459560482,
+      "grad_norm": 0.016693761174956682,
+      "learning_rate": 8.59577515216613e-06,
+      "loss": 0.0,
+      "step": 3515
+    },
+    {
+      "epoch": 0.22659019140297737,
+      "grad_norm": 0.033633965786489854,
+      "learning_rate": 8.595059076262084e-06,
+      "loss": 0.0001,
+      "step": 3516
+    },
+    {
+      "epoch": 0.22665463684990655,
+      "grad_norm": 0.034370924494014785,
+      "learning_rate": 8.594343000358038e-06,
+      "loss": 0.0,
+      "step": 3517
+    },
+    {
+      "epoch": 0.22671908229683574,
+      "grad_norm": 0.4692173535367172,
+      "learning_rate": 8.593626924453993e-06,
+      "loss": 0.002,
+      "step": 3518
+    },
+    {
+      "epoch": 0.2267835277437649,
+      "grad_norm": 0.16348664515859737,
+      "learning_rate": 8.592910848549947e-06,
+      "loss": 0.0003,
+      "step": 3519
+    },
+    {
+      "epoch": 0.22684797319069408,
+      "grad_norm": 0.02663825689310431,
+      "learning_rate": 8.592194772645901e-06,
+      "loss": 0.0002,
+      "step": 3520
+    },
+    {
+      "epoch": 0.22691241863762326,
+      "grad_norm": 0.23243357524753472,
+      "learning_rate": 8.591478696741856e-06,
+      "loss": 0.0002,
+      "step": 3521
+    },
+    {
+      "epoch": 0.22697686408455242,
+      "grad_norm": 0.00017325221716987894,
+      "learning_rate": 8.59076262083781e-06,
+      "loss": 0.0,
+      "step": 3522
+    },
+    {
+      "epoch": 0.2270413095314816,
+      "grad_norm": 0.0010475463938276882,
+      "learning_rate": 8.590046544933763e-06,
+      "loss": 0.0,
+      "step": 3523
+    },
+    {
+      "epoch": 0.22710575497841076,
+      "grad_norm": 0.0006956761215428136,
+      "learning_rate": 8.589330469029717e-06,
+      "loss": 0.0,
+      "step": 3524
+    },
+    {
+      "epoch": 0.22717020042533995,
+      "grad_norm": 0.026205720830062656,
+      "learning_rate": 8.588614393125671e-06,
+      "loss": 0.0003,
+      "step": 3525
+    },
+    {
+      "epoch": 0.22723464587226913,
+      "grad_norm": 0.007105951163493245,
+      "learning_rate": 8.587898317221625e-06,
+      "loss": 0.0,
+      "step": 3526
+    },
+    {
+      "epoch": 0.2272990913191983,
+      "grad_norm": 0.0004057145569787648,
+      "learning_rate": 8.58718224131758e-06,
+      "loss": 0.0,
+      "step": 3527
+    },
+    {
+      "epoch": 0.22736353676612747,
+      "grad_norm": 0.02197380725177132,
+      "learning_rate": 8.586466165413536e-06,
+      "loss": 0.0001,
+      "step": 3528
+    },
+    {
+      "epoch": 0.22742798221305666,
+      "grad_norm": 0.001348719214099292,
+      "learning_rate": 8.58575008950949e-06,
+      "loss": 0.0,
+      "step": 3529
+    },
+    {
+      "epoch": 0.2274924276599858,
+      "grad_norm": 0.010772461928954678,
+      "learning_rate": 8.585034013605443e-06,
+      "loss": 0.0001,
+      "step": 3530
+    },
+    {
+      "epoch": 0.227556873106915,
+      "grad_norm": 0.08330517966318159,
+      "learning_rate": 8.584317937701397e-06,
+      "loss": 0.0019,
+      "step": 3531
+    },
+    {
+      "epoch": 0.22762131855384418,
+      "grad_norm": 0.19010628847131272,
+      "learning_rate": 8.583601861797351e-06,
+      "loss": 0.0027,
+      "step": 3532
+    },
+    {
+      "epoch": 0.22768576400077334,
+      "grad_norm": 0.0011382736201224818,
+      "learning_rate": 8.582885785893306e-06,
+      "loss": 0.0,
+      "step": 3533
+    },
+    {
+      "epoch": 0.22775020944770252,
+      "grad_norm": 0.01510820616315032,
+      "learning_rate": 8.58216970998926e-06,
+      "loss": 0.0001,
+      "step": 3534
+    },
+    {
+      "epoch": 0.2278146548946317,
+      "grad_norm": 0.0022582360408176425,
+      "learning_rate": 8.581453634085214e-06,
+      "loss": 0.0,
+      "step": 3535
+    },
+    {
+      "epoch": 0.22787910034156086,
+      "grad_norm": 0.0024001167201978904,
+      "learning_rate": 8.580737558181168e-06,
+      "loss": 0.0,
+      "step": 3536
+    },
+    {
+      "epoch": 0.22794354578849005,
+      "grad_norm": 0.00014460807293517424,
+      "learning_rate": 8.580021482277123e-06,
+      "loss": 0.0,
+      "step": 3537
+    },
+    {
+      "epoch": 0.2280079912354192,
+      "grad_norm": 0.004371144356167792,
+      "learning_rate": 8.579305406373077e-06,
+      "loss": 0.0,
+      "step": 3538
+    },
+    {
+      "epoch": 0.2280724366823484,
+      "grad_norm": 0.12752316946926834,
+      "learning_rate": 8.57858933046903e-06,
+      "loss": 0.0012,
+      "step": 3539
+    },
+    {
+      "epoch": 0.22813688212927757,
+      "grad_norm": 6.635877081708909e-05,
+      "learning_rate": 8.577873254564984e-06,
+      "loss": 0.0,
+      "step": 3540
+    },
+    {
+      "epoch": 0.22820132757620673,
+      "grad_norm": 0.035709400460215875,
+      "learning_rate": 8.577157178660938e-06,
+      "loss": 0.0,
+      "step": 3541
+    },
+    {
+      "epoch": 0.22826577302313591,
+      "grad_norm": 0.043633021594940764,
+      "learning_rate": 8.576441102756893e-06,
+      "loss": 0.0,
+      "step": 3542
+    },
+    {
+      "epoch": 0.2283302184700651,
+      "grad_norm": 0.198884992954439,
+      "learning_rate": 8.575725026852847e-06,
+      "loss": 0.0002,
+      "step": 3543
+    },
+    {
+      "epoch": 0.22839466391699426,
+      "grad_norm": 0.03614280375678468,
+      "learning_rate": 8.575008950948801e-06,
+      "loss": 0.0001,
+      "step": 3544
+    },
+    {
+      "epoch": 0.22845910936392344,
+      "grad_norm": 0.0005592311449027554,
+      "learning_rate": 8.574292875044755e-06,
+      "loss": 0.0,
+      "step": 3545
+    },
+    {
+      "epoch": 0.22852355481085262,
+      "grad_norm": 0.0407157248262875,
+      "learning_rate": 8.57357679914071e-06,
+      "loss": 0.0001,
+      "step": 3546
+    },
+    {
+      "epoch": 0.22858800025778178,
+      "grad_norm": 0.04273274208048958,
+      "learning_rate": 8.572860723236664e-06,
+      "loss": 0.0001,
+      "step": 3547
+    },
+    {
+      "epoch": 0.22865244570471097,
+      "grad_norm": 0.00011219605860635525,
+      "learning_rate": 8.572144647332618e-06,
+      "loss": 0.0,
+      "step": 3548
+    },
+    {
+      "epoch": 0.22871689115164015,
+      "grad_norm": 0.0011935116431631536,
+      "learning_rate": 8.571428571428571e-06,
+      "loss": 0.0,
+      "step": 3549
+    },
+    {
+      "epoch": 0.2287813365985693,
+      "grad_norm": 0.041032109467459874,
+      "learning_rate": 8.570712495524525e-06,
+      "loss": 0.0005,
+      "step": 3550
+    },
+    {
+      "epoch": 0.2288457820454985,
+      "grad_norm": 0.0033587200385650448,
+      "learning_rate": 8.569996419620481e-06,
+      "loss": 0.0,
+      "step": 3551
+    },
+    {
+      "epoch": 0.22891022749242765,
+      "grad_norm": 0.25359755604274065,
+      "learning_rate": 8.569280343716436e-06,
+      "loss": 0.0005,
+      "step": 3552
+    },
+    {
+      "epoch": 0.22897467293935683,
+      "grad_norm": 0.00030970704197332206,
+      "learning_rate": 8.56856426781239e-06,
+      "loss": 0.0,
+      "step": 3553
+    },
+    {
+      "epoch": 0.22903911838628602,
+      "grad_norm": 0.003074699240476695,
+      "learning_rate": 8.567848191908344e-06,
+      "loss": 0.0,
+      "step": 3554
+    },
+    {
+      "epoch": 0.22910356383321517,
+      "grad_norm": 0.04174109869575229,
+      "learning_rate": 8.567132116004297e-06,
+      "loss": 0.0001,
+      "step": 3555
+    },
+    {
+      "epoch": 0.22916800928014436,
+      "grad_norm": 0.00035973275162782593,
+      "learning_rate": 8.566416040100251e-06,
+      "loss": 0.0,
+      "step": 3556
+    },
+    {
+      "epoch": 0.22923245472707354,
+      "grad_norm": 0.17487531971851106,
+      "learning_rate": 8.565699964196205e-06,
+      "loss": 0.0078,
+      "step": 3557
+    },
+    {
+      "epoch": 0.2292969001740027,
+      "grad_norm": 0.0021126300294926044,
+      "learning_rate": 8.56498388829216e-06,
+      "loss": 0.0,
+      "step": 3558
+    },
+    {
+      "epoch": 0.22936134562093188,
+      "grad_norm": 0.0005282738112423214,
+      "learning_rate": 8.564267812388114e-06,
+      "loss": 0.0,
+      "step": 3559
+    },
+    {
+      "epoch": 0.22942579106786107,
+      "grad_norm": 0.09196037584755035,
+      "learning_rate": 8.563551736484068e-06,
+      "loss": 0.0008,
+      "step": 3560
+    },
+    {
+      "epoch": 0.22949023651479022,
+      "grad_norm": 0.0020641696227278188,
+      "learning_rate": 8.562835660580023e-06,
+      "loss": 0.0,
+      "step": 3561
+    },
+    {
+      "epoch": 0.2295546819617194,
+      "grad_norm": 0.0005468917117046053,
+      "learning_rate": 8.562119584675977e-06,
+      "loss": 0.0,
+      "step": 3562
+    },
+    {
+      "epoch": 0.22961912740864857,
+      "grad_norm": 0.1214281701890994,
+      "learning_rate": 8.561403508771931e-06,
+      "loss": 0.0002,
+      "step": 3563
+    },
+    {
+      "epoch": 0.22968357285557775,
+      "grad_norm": 0.027091342936798053,
+      "learning_rate": 8.560687432867885e-06,
+      "loss": 0.0001,
+      "step": 3564
+    },
+    {
+      "epoch": 0.22974801830250693,
+      "grad_norm": 0.003703475196088253,
+      "learning_rate": 8.559971356963838e-06,
+      "loss": 0.0,
+      "step": 3565
+    },
+    {
+      "epoch": 0.2298124637494361,
+      "grad_norm": 0.0004199587784974058,
+      "learning_rate": 8.559255281059792e-06,
+      "loss": 0.0,
+      "step": 3566
+    },
+    {
+      "epoch": 0.22987690919636528,
+      "grad_norm": 2.4043078821272714,
+      "learning_rate": 8.558539205155747e-06,
+      "loss": 0.0121,
+      "step": 3567
+    },
+    {
+      "epoch": 0.22994135464329446,
+      "grad_norm": 0.00202259349133905,
+      "learning_rate": 8.557823129251701e-06,
+      "loss": 0.0,
+      "step": 3568
+    },
+    {
+      "epoch": 0.23000580009022362,
+      "grad_norm": 0.3955692008143024,
+      "learning_rate": 8.557107053347655e-06,
+      "loss": 0.0015,
+      "step": 3569
+    },
+    {
+      "epoch": 0.2300702455371528,
+      "grad_norm": 0.10374625174088346,
+      "learning_rate": 8.55639097744361e-06,
+      "loss": 0.0004,
+      "step": 3570
+    },
+    {
+      "epoch": 0.23013469098408199,
+      "grad_norm": 0.0004160911240873582,
+      "learning_rate": 8.555674901539564e-06,
+      "loss": 0.0,
+      "step": 3571
+    },
+    {
+      "epoch": 0.23019913643101114,
+      "grad_norm": 0.38638667095974427,
+      "learning_rate": 8.554958825635518e-06,
+      "loss": 0.0014,
+      "step": 3572
+    },
+    {
+      "epoch": 0.23026358187794033,
+      "grad_norm": 0.0017868331589009291,
+      "learning_rate": 8.554242749731472e-06,
+      "loss": 0.0,
+      "step": 3573
+    },
+    {
+      "epoch": 0.2303280273248695,
+      "grad_norm": 0.0008067980443222023,
+      "learning_rate": 8.553526673827427e-06,
+      "loss": 0.0,
+      "step": 3574
+    },
+    {
+      "epoch": 0.23039247277179867,
+      "grad_norm": 0.14120487685539893,
+      "learning_rate": 8.552810597923381e-06,
+      "loss": 0.0002,
+      "step": 3575
+    },
+    {
+      "epoch": 0.23045691821872785,
+      "grad_norm": 0.01688292958076883,
+      "learning_rate": 8.552094522019335e-06,
+      "loss": 0.0001,
+      "step": 3576
+    },
+    {
+      "epoch": 0.230521363665657,
+      "grad_norm": 0.016395282165946354,
+      "learning_rate": 8.55137844611529e-06,
+      "loss": 0.0,
+      "step": 3577
+    },
+    {
+      "epoch": 0.2305858091125862,
+      "grad_norm": 0.0002731364490621126,
+      "learning_rate": 8.550662370211244e-06,
+      "loss": 0.0,
+      "step": 3578
+    },
+    {
+      "epoch": 0.23065025455951538,
+      "grad_norm": 0.0024728902830418137,
+      "learning_rate": 8.549946294307198e-06,
+      "loss": 0.0,
+      "step": 3579
+    },
+    {
+      "epoch": 0.23071470000644453,
+      "grad_norm": 0.0009555167639653805,
+      "learning_rate": 8.549230218403152e-06,
+      "loss": 0.0,
+      "step": 3580
+    },
+    {
+      "epoch": 0.23077914545337372,
+      "grad_norm": 0.10064114749816049,
+      "learning_rate": 8.548514142499105e-06,
+      "loss": 0.0001,
+      "step": 3581
+    },
+    {
+      "epoch": 0.2308435909003029,
+      "grad_norm": 0.10534812510748648,
+      "learning_rate": 8.54779806659506e-06,
+      "loss": 0.0002,
+      "step": 3582
+    },
+    {
+      "epoch": 0.23090803634723206,
+      "grad_norm": 0.003143158927519917,
+      "learning_rate": 8.547081990691014e-06,
+      "loss": 0.0,
+      "step": 3583
+    },
+    {
+      "epoch": 0.23097248179416124,
+      "grad_norm": 0.0013283154727609836,
+      "learning_rate": 8.546365914786968e-06,
+      "loss": 0.0,
+      "step": 3584
+    },
+    {
+      "epoch": 0.23103692724109043,
+      "grad_norm": 0.0009492733557983951,
+      "learning_rate": 8.545649838882922e-06,
+      "loss": 0.0,
+      "step": 3585
+    },
+    {
+      "epoch": 0.23110137268801959,
+      "grad_norm": 0.000137084553934003,
+      "learning_rate": 8.544933762978877e-06,
+      "loss": 0.0,
+      "step": 3586
+    },
+    {
+      "epoch": 0.23116581813494877,
+      "grad_norm": 0.013543951428487857,
+      "learning_rate": 8.544217687074831e-06,
+      "loss": 0.0001,
+      "step": 3587
+    },
+    {
+      "epoch": 0.23123026358187795,
+      "grad_norm": 0.004368855218247446,
+      "learning_rate": 8.543501611170785e-06,
+      "loss": 0.0,
+      "step": 3588
+    },
+    {
+      "epoch": 0.2312947090288071,
+      "grad_norm": 0.08383096778560901,
+      "learning_rate": 8.54278553526674e-06,
+      "loss": 0.0001,
+      "step": 3589
+    },
+    {
+      "epoch": 0.2313591544757363,
+      "grad_norm": 0.014596576080533902,
+      "learning_rate": 8.542069459362694e-06,
+      "loss": 0.0001,
+      "step": 3590
+    },
+    {
+      "epoch": 0.23142359992266545,
+      "grad_norm": 0.003054285641641687,
+      "learning_rate": 8.541353383458646e-06,
+      "loss": 0.0,
+      "step": 3591
+    },
+    {
+      "epoch": 0.23148804536959464,
+      "grad_norm": 0.010529457621076216,
+      "learning_rate": 8.5406373075546e-06,
+      "loss": 0.0001,
+      "step": 3592
+    },
+    {
+      "epoch": 0.23155249081652382,
+      "grad_norm": 0.03386443048252686,
+      "learning_rate": 8.539921231650555e-06,
+      "loss": 0.0001,
+      "step": 3593
+    },
+    {
+      "epoch": 0.23161693626345298,
+      "grad_norm": 0.000382320509569171,
+      "learning_rate": 8.53920515574651e-06,
+      "loss": 0.0,
+      "step": 3594
+    },
+    {
+      "epoch": 0.23168138171038216,
+      "grad_norm": 0.03725659818404199,
+      "learning_rate": 8.538489079842464e-06,
+      "loss": 0.0,
+      "step": 3595
+    },
+    {
+      "epoch": 0.23174582715731135,
+      "grad_norm": 0.002715356890662232,
+      "learning_rate": 8.537773003938418e-06,
+      "loss": 0.0,
+      "step": 3596
+    },
+    {
+      "epoch": 0.2318102726042405,
+      "grad_norm": 0.0005005922412607236,
+      "learning_rate": 8.537056928034372e-06,
+      "loss": 0.0,
+      "step": 3597
+    },
+    {
+      "epoch": 0.2318747180511697,
+      "grad_norm": 0.23895372802173373,
+      "learning_rate": 8.536340852130326e-06,
+      "loss": 0.0005,
+      "step": 3598
+    },
+    {
+      "epoch": 0.23193916349809887,
+      "grad_norm": 0.0035535297376758264,
+      "learning_rate": 8.53562477622628e-06,
+      "loss": 0.0,
+      "step": 3599
+    },
+    {
+      "epoch": 0.23200360894502803,
+      "grad_norm": 1.192982066141991,
+      "learning_rate": 8.534908700322235e-06,
+      "loss": 0.0016,
+      "step": 3600
+    },
+    {
+      "epoch": 0.2320680543919572,
+      "grad_norm": 0.0006909154871493346,
+      "learning_rate": 8.53419262441819e-06,
+      "loss": 0.0,
+      "step": 3601
+    },
+    {
+      "epoch": 0.23213249983888637,
+      "grad_norm": 0.003394634640801154,
+      "learning_rate": 8.533476548514144e-06,
+      "loss": 0.0,
+      "step": 3602
+    },
+    {
+      "epoch": 0.23219694528581555,
+      "grad_norm": 0.16565397364526332,
+      "learning_rate": 8.532760472610098e-06,
+      "loss": 0.0029,
+      "step": 3603
+    },
+    {
+      "epoch": 0.23226139073274474,
+      "grad_norm": 0.0007924288655583169,
+      "learning_rate": 8.532044396706052e-06,
+      "loss": 0.0,
+      "step": 3604
+    },
+    {
+      "epoch": 0.2323258361796739,
+      "grad_norm": 0.0008002846613981609,
+      "learning_rate": 8.531328320802007e-06,
+      "loss": 0.0,
+      "step": 3605
+    },
+    {
+      "epoch": 0.23239028162660308,
+      "grad_norm": 0.028553978107853775,
+      "learning_rate": 8.530612244897961e-06,
+      "loss": 0.0001,
+      "step": 3606
+    },
+    {
+      "epoch": 0.23245472707353226,
+      "grad_norm": 0.018178990149575903,
+      "learning_rate": 8.529896168993913e-06,
+      "loss": 0.0001,
+      "step": 3607
+    },
+    {
+      "epoch": 0.23251917252046142,
+      "grad_norm": 0.026175255165739573,
+      "learning_rate": 8.529180093089868e-06,
+      "loss": 0.0,
+      "step": 3608
+    },
+    {
+      "epoch": 0.2325836179673906,
+      "grad_norm": 0.00863108323449065,
+      "learning_rate": 8.528464017185822e-06,
+      "loss": 0.0,
+      "step": 3609
+    },
+    {
+      "epoch": 0.2326480634143198,
+      "grad_norm": 0.00367795649645988,
+      "learning_rate": 8.527747941281776e-06,
+      "loss": 0.0,
+      "step": 3610
+    },
+    {
+      "epoch": 0.23271250886124895,
+      "grad_norm": 0.023703811777085085,
+      "learning_rate": 8.52703186537773e-06,
+      "loss": 0.0001,
+      "step": 3611
+    },
+    {
+      "epoch": 0.23277695430817813,
+      "grad_norm": 0.41825952843291414,
+      "learning_rate": 8.526315789473685e-06,
+      "loss": 0.0023,
+      "step": 3612
+    },
+    {
+      "epoch": 0.23284139975510731,
+      "grad_norm": 0.0002994700953356274,
+      "learning_rate": 8.52559971356964e-06,
+      "loss": 0.0,
+      "step": 3613
+    },
+    {
+      "epoch": 0.23290584520203647,
+      "grad_norm": 0.03925458351539499,
+      "learning_rate": 8.524883637665594e-06,
+      "loss": 0.0004,
+      "step": 3614
+    },
+    {
+      "epoch": 0.23297029064896566,
+      "grad_norm": 0.010272126599741443,
+      "learning_rate": 8.524167561761548e-06,
+      "loss": 0.0001,
+      "step": 3615
+    },
+    {
+      "epoch": 0.2330347360958948,
+      "grad_norm": 0.0008049833577190571,
+      "learning_rate": 8.5234514858575e-06,
+      "loss": 0.0,
+      "step": 3616
+    },
+    {
+      "epoch": 0.233099181542824,
+      "grad_norm": 0.00012036471779414499,
+      "learning_rate": 8.522735409953455e-06,
+      "loss": 0.0,
+      "step": 3617
+    },
+    {
+      "epoch": 0.23316362698975318,
+      "grad_norm": 0.0011984086577387977,
+      "learning_rate": 8.522019334049409e-06,
+      "loss": 0.0,
+      "step": 3618
+    },
+    {
+      "epoch": 0.23322807243668234,
+      "grad_norm": 0.0835934790491009,
+      "learning_rate": 8.521303258145363e-06,
+      "loss": 0.0006,
+      "step": 3619
+    },
+    {
+      "epoch": 0.23329251788361152,
+      "grad_norm": 0.08836004941212419,
+      "learning_rate": 8.520587182241318e-06,
+      "loss": 0.0002,
+      "step": 3620
+    },
+    {
+      "epoch": 0.2333569633305407,
+      "grad_norm": 0.18029412205686446,
+      "learning_rate": 8.519871106337274e-06,
+      "loss": 0.0038,
+      "step": 3621
+    },
+    {
+      "epoch": 0.23342140877746986,
+      "grad_norm": 0.00026738501476959226,
+      "learning_rate": 8.519155030433228e-06,
+      "loss": 0.0,
+      "step": 3622
+    },
+    {
+      "epoch": 0.23348585422439905,
+      "grad_norm": 0.005370139991002869,
+      "learning_rate": 8.51843895452918e-06,
+      "loss": 0.0001,
+      "step": 3623
+    },
+    {
+      "epoch": 0.23355029967132823,
+      "grad_norm": 0.2452207171892442,
+      "learning_rate": 8.517722878625135e-06,
+      "loss": 0.0005,
+      "step": 3624
+    },
+    {
+      "epoch": 0.2336147451182574,
+      "grad_norm": 0.0010621327473918718,
+      "learning_rate": 8.517006802721089e-06,
+      "loss": 0.0,
+      "step": 3625
+    },
+    {
+      "epoch": 0.23367919056518657,
+      "grad_norm": 0.00016653019881507862,
+      "learning_rate": 8.516290726817043e-06,
+      "loss": 0.0,
+      "step": 3626
+    },
+    {
+      "epoch": 0.23374363601211576,
+      "grad_norm": 0.06785625871955342,
+      "learning_rate": 8.515574650912998e-06,
+      "loss": 0.0005,
+      "step": 3627
+    },
+    {
+      "epoch": 0.23380808145904491,
+      "grad_norm": 1.7781309962381047,
+      "learning_rate": 8.514858575008952e-06,
+      "loss": 0.0043,
+      "step": 3628
+    },
+    {
+      "epoch": 0.2338725269059741,
+      "grad_norm": 0.00018354352976572647,
+      "learning_rate": 8.514142499104906e-06,
+      "loss": 0.0,
+      "step": 3629
+    },
+    {
+      "epoch": 0.23393697235290326,
+      "grad_norm": 0.0003587856822891483,
+      "learning_rate": 8.51342642320086e-06,
+      "loss": 0.0,
+      "step": 3630
+    },
+    {
+      "epoch": 0.23400141779983244,
+      "grad_norm": 0.003481401644091111,
+      "learning_rate": 8.512710347296815e-06,
+      "loss": 0.0,
+      "step": 3631
+    },
+    {
+      "epoch": 0.23406586324676162,
+      "grad_norm": 0.0014548914336011787,
+      "learning_rate": 8.511994271392768e-06,
+      "loss": 0.0,
+      "step": 3632
+    },
+    {
+      "epoch": 0.23413030869369078,
+      "grad_norm": 0.02106593887574402,
+      "learning_rate": 8.511278195488722e-06,
+      "loss": 0.0001,
+      "step": 3633
+    },
+    {
+      "epoch": 0.23419475414061997,
+      "grad_norm": 0.1377033136626864,
+      "learning_rate": 8.510562119584676e-06,
+      "loss": 0.002,
+      "step": 3634
+    },
+    {
+      "epoch": 0.23425919958754915,
+      "grad_norm": 0.04813451217798247,
+      "learning_rate": 8.50984604368063e-06,
+      "loss": 0.0002,
+      "step": 3635
+    },
+    {
+      "epoch": 0.2343236450344783,
+      "grad_norm": 0.012981266824628336,
+      "learning_rate": 8.509129967776585e-06,
+      "loss": 0.0,
+      "step": 3636
+    },
+    {
+      "epoch": 0.2343880904814075,
+      "grad_norm": 0.12003216797220177,
+      "learning_rate": 8.508413891872539e-06,
+      "loss": 0.0001,
+      "step": 3637
+    },
+    {
+      "epoch": 0.23445253592833667,
+      "grad_norm": 0.003240766907488695,
+      "learning_rate": 8.507697815968493e-06,
+      "loss": 0.0,
+      "step": 3638
+    },
+    {
+      "epoch": 0.23451698137526583,
+      "grad_norm": 0.07065803673474678,
+      "learning_rate": 8.506981740064448e-06,
+      "loss": 0.0003,
+      "step": 3639
+    },
+    {
+      "epoch": 0.23458142682219502,
+      "grad_norm": 0.8096147071804877,
+      "learning_rate": 8.506265664160402e-06,
+      "loss": 0.0021,
+      "step": 3640
+    },
+    {
+      "epoch": 0.23464587226912417,
+      "grad_norm": 0.00021237418484407522,
+      "learning_rate": 8.505549588256356e-06,
+      "loss": 0.0,
+      "step": 3641
+    },
+    {
+      "epoch": 0.23471031771605336,
+      "grad_norm": 0.002760565859295997,
+      "learning_rate": 8.504833512352309e-06,
+      "loss": 0.0,
+      "step": 3642
+    },
+    {
+      "epoch": 0.23477476316298254,
+      "grad_norm": 0.004335590145388964,
+      "learning_rate": 8.504117436448263e-06,
+      "loss": 0.0001,
+      "step": 3643
+    },
+    {
+      "epoch": 0.2348392086099117,
+      "grad_norm": 0.3139593026011861,
+      "learning_rate": 8.503401360544217e-06,
+      "loss": 0.0005,
+      "step": 3644
+    },
+    {
+      "epoch": 0.23490365405684088,
+      "grad_norm": 0.15569056856024935,
+      "learning_rate": 8.502685284640173e-06,
+      "loss": 0.0021,
+      "step": 3645
+    },
+    {
+      "epoch": 0.23496809950377007,
+      "grad_norm": 0.0906748991919936,
+      "learning_rate": 8.501969208736128e-06,
+      "loss": 0.0001,
+      "step": 3646
+    },
+    {
+      "epoch": 0.23503254495069922,
+      "grad_norm": 0.014439834483945311,
+      "learning_rate": 8.501253132832082e-06,
+      "loss": 0.0001,
+      "step": 3647
+    },
+    {
+      "epoch": 0.2350969903976284,
+      "grad_norm": 0.019942436114608407,
+      "learning_rate": 8.500537056928035e-06,
+      "loss": 0.0001,
+      "step": 3648
+    },
+    {
+      "epoch": 0.2351614358445576,
+      "grad_norm": 0.0053793905060603225,
+      "learning_rate": 8.499820981023989e-06,
+      "loss": 0.0,
+      "step": 3649
+    },
+    {
+      "epoch": 0.23522588129148675,
+      "grad_norm": 0.00658721510723763,
+      "learning_rate": 8.499104905119943e-06,
+      "loss": 0.0001,
+      "step": 3650
+    },
+    {
+      "epoch": 0.23529032673841593,
+      "grad_norm": 0.0024141843857270616,
+      "learning_rate": 8.498388829215898e-06,
+      "loss": 0.0,
+      "step": 3651
+    },
+    {
+      "epoch": 0.23535477218534512,
+      "grad_norm": 0.00412149132628881,
+      "learning_rate": 8.497672753311852e-06,
+      "loss": 0.0,
+      "step": 3652
+    },
+    {
+      "epoch": 0.23541921763227427,
+      "grad_norm": 0.0017195526767637631,
+      "learning_rate": 8.496956677407806e-06,
+      "loss": 0.0,
+      "step": 3653
+    },
+    {
+      "epoch": 0.23548366307920346,
+      "grad_norm": 0.0001600318501965816,
+      "learning_rate": 8.49624060150376e-06,
+      "loss": 0.0,
+      "step": 3654
+    },
+    {
+      "epoch": 0.23554810852613262,
+      "grad_norm": 0.0015997134786524389,
+      "learning_rate": 8.495524525599715e-06,
+      "loss": 0.0,
+      "step": 3655
+    },
+    {
+      "epoch": 0.2356125539730618,
+      "grad_norm": 0.00019996158198574133,
+      "learning_rate": 8.494808449695669e-06,
+      "loss": 0.0,
+      "step": 3656
+    },
+    {
+      "epoch": 0.23567699941999098,
+      "grad_norm": 0.022086918620967465,
+      "learning_rate": 8.494092373791623e-06,
+      "loss": 0.0001,
+      "step": 3657
+    },
+    {
+      "epoch": 0.23574144486692014,
+      "grad_norm": 0.010196585178414033,
+      "learning_rate": 8.493376297887576e-06,
+      "loss": 0.0,
+      "step": 3658
+    },
+    {
+      "epoch": 0.23580589031384933,
+      "grad_norm": 0.006909177117970666,
+      "learning_rate": 8.49266022198353e-06,
+      "loss": 0.0,
+      "step": 3659
+    },
+    {
+      "epoch": 0.2358703357607785,
+      "grad_norm": 2.2550082099579014,
+      "learning_rate": 8.491944146079485e-06,
+      "loss": 0.015,
+      "step": 3660
+    },
+    {
+      "epoch": 0.23593478120770767,
+      "grad_norm": 0.023382369159834774,
+      "learning_rate": 8.491228070175439e-06,
+      "loss": 0.0002,
+      "step": 3661
+    },
+    {
+      "epoch": 0.23599922665463685,
+      "grad_norm": 0.6735246278908694,
+      "learning_rate": 8.490511994271393e-06,
+      "loss": 0.0021,
+      "step": 3662
+    },
+    {
+      "epoch": 0.23606367210156604,
+      "grad_norm": 0.0016992829250492205,
+      "learning_rate": 8.489795918367347e-06,
+      "loss": 0.0,
+      "step": 3663
+    },
+    {
+      "epoch": 0.2361281175484952,
+      "grad_norm": 0.14514165931232095,
+      "learning_rate": 8.489079842463302e-06,
+      "loss": 0.0003,
+      "step": 3664
+    },
+    {
+      "epoch": 0.23619256299542438,
+      "grad_norm": 0.056902135808290705,
+      "learning_rate": 8.488363766559256e-06,
+      "loss": 0.0006,
+      "step": 3665
+    },
+    {
+      "epoch": 0.23625700844235356,
+      "grad_norm": 0.00035760265499879115,
+      "learning_rate": 8.48764769065521e-06,
+      "loss": 0.0,
+      "step": 3666
+    },
+    {
+      "epoch": 0.23632145388928272,
+      "grad_norm": 0.015011496814008625,
+      "learning_rate": 8.486931614751165e-06,
+      "loss": 0.0001,
+      "step": 3667
+    },
+    {
+      "epoch": 0.2363858993362119,
+      "grad_norm": 0.00026337167364298936,
+      "learning_rate": 8.486215538847119e-06,
+      "loss": 0.0,
+      "step": 3668
+    },
+    {
+      "epoch": 0.23645034478314106,
+      "grad_norm": 0.0029167551721233755,
+      "learning_rate": 8.485499462943073e-06,
+      "loss": 0.0,
+      "step": 3669
+    },
+    {
+      "epoch": 0.23651479023007024,
+      "grad_norm": 0.01064145374267631,
+      "learning_rate": 8.484783387039028e-06,
+      "loss": 0.0,
+      "step": 3670
+    },
+    {
+      "epoch": 0.23657923567699943,
+      "grad_norm": 0.002452641484903878,
+      "learning_rate": 8.484067311134982e-06,
+      "loss": 0.0,
+      "step": 3671
+    },
+    {
+      "epoch": 0.23664368112392858,
+      "grad_norm": 0.009032862405537249,
+      "learning_rate": 8.483351235230936e-06,
+      "loss": 0.0001,
+      "step": 3672
+    },
+    {
+      "epoch": 0.23670812657085777,
+      "grad_norm": 0.000362237704258534,
+      "learning_rate": 8.48263515932689e-06,
+      "loss": 0.0,
+      "step": 3673
+    },
+    {
+      "epoch": 0.23677257201778695,
+      "grad_norm": 0.0007830415697379507,
+      "learning_rate": 8.481919083422843e-06,
+      "loss": 0.0,
+      "step": 3674
+    },
+    {
+      "epoch": 0.2368370174647161,
+      "grad_norm": 0.0008006848452688169,
+      "learning_rate": 8.481203007518797e-06,
+      "loss": 0.0,
+      "step": 3675
+    },
+    {
+      "epoch": 0.2369014629116453,
+      "grad_norm": 1.025715803628338,
+      "learning_rate": 8.480486931614752e-06,
+      "loss": 0.005,
+      "step": 3676
+    },
+    {
+      "epoch": 0.23696590835857448,
+      "grad_norm": 0.06137980239797591,
+      "learning_rate": 8.479770855710706e-06,
+      "loss": 0.0002,
+      "step": 3677
+    },
+    {
+      "epoch": 0.23703035380550364,
+      "grad_norm": 0.37055178769912805,
+      "learning_rate": 8.47905477980666e-06,
+      "loss": 0.0035,
+      "step": 3678
+    },
+    {
+      "epoch": 0.23709479925243282,
+      "grad_norm": 0.00016133238319453695,
+      "learning_rate": 8.478338703902615e-06,
+      "loss": 0.0,
+      "step": 3679
+    },
+    {
+      "epoch": 0.23715924469936198,
+      "grad_norm": 0.45889901524037224,
+      "learning_rate": 8.477622627998569e-06,
+      "loss": 0.0006,
+      "step": 3680
+    },
+    {
+      "epoch": 0.23722369014629116,
+      "grad_norm": 0.04829375159957358,
+      "learning_rate": 8.476906552094523e-06,
+      "loss": 0.0001,
+      "step": 3681
+    },
+    {
+      "epoch": 0.23728813559322035,
+      "grad_norm": 0.04483302025827112,
+      "learning_rate": 8.476190476190477e-06,
+      "loss": 0.0,
+      "step": 3682
+    },
+    {
+      "epoch": 0.2373525810401495,
+      "grad_norm": 0.006915118970124431,
+      "learning_rate": 8.475474400286432e-06,
+      "loss": 0.0015,
+      "step": 3683
+    },
+    {
+      "epoch": 0.2374170264870787,
+      "grad_norm": 0.029743070065202375,
+      "learning_rate": 8.474758324382384e-06,
+      "loss": 0.0001,
+      "step": 3684
+    },
+    {
+      "epoch": 0.23748147193400787,
+      "grad_norm": 0.010346678082478153,
+      "learning_rate": 8.474042248478339e-06,
+      "loss": 0.0001,
+      "step": 3685
+    },
+    {
+      "epoch": 0.23754591738093703,
+      "grad_norm": 0.003808709571335096,
+      "learning_rate": 8.473326172574293e-06,
+      "loss": 0.0,
+      "step": 3686
+    },
+    {
+      "epoch": 0.2376103628278662,
+      "grad_norm": 0.03439312438102102,
+      "learning_rate": 8.472610096670247e-06,
+      "loss": 0.0001,
+      "step": 3687
+    },
+    {
+      "epoch": 0.2376748082747954,
+      "grad_norm": 0.0006371055023281539,
+      "learning_rate": 8.471894020766201e-06,
+      "loss": 0.0,
+      "step": 3688
+    },
+    {
+      "epoch": 0.23773925372172455,
+      "grad_norm": 0.008961421343413054,
+      "learning_rate": 8.471177944862156e-06,
+      "loss": 0.0,
+      "step": 3689
+    },
+    {
+      "epoch": 0.23780369916865374,
+      "grad_norm": 0.01069256581449394,
+      "learning_rate": 8.47046186895811e-06,
+      "loss": 0.0,
+      "step": 3690
+    },
+    {
+      "epoch": 0.23786814461558292,
+      "grad_norm": 0.07071569525956135,
+      "learning_rate": 8.469745793054064e-06,
+      "loss": 0.0005,
+      "step": 3691
+    },
+    {
+      "epoch": 0.23793259006251208,
+      "grad_norm": 0.001481707000711012,
+      "learning_rate": 8.469029717150019e-06,
+      "loss": 0.0,
+      "step": 3692
+    },
+    {
+      "epoch": 0.23799703550944126,
+      "grad_norm": 0.02562734064927472,
+      "learning_rate": 8.468313641245973e-06,
+      "loss": 0.0004,
+      "step": 3693
+    },
+    {
+      "epoch": 0.23806148095637042,
+      "grad_norm": 0.009852533383060125,
+      "learning_rate": 8.467597565341927e-06,
+      "loss": 0.0,
+      "step": 3694
+    },
+    {
+      "epoch": 0.2381259264032996,
+      "grad_norm": 0.024841071931428666,
+      "learning_rate": 8.466881489437882e-06,
+      "loss": 0.0,
+      "step": 3695
+    },
+    {
+      "epoch": 0.2381903718502288,
+      "grad_norm": 0.0019675953712899397,
+      "learning_rate": 8.466165413533836e-06,
+      "loss": 0.0,
+      "step": 3696
+    },
+    {
+      "epoch": 0.23825481729715794,
+      "grad_norm": 0.005357394982959551,
+      "learning_rate": 8.46544933762979e-06,
+      "loss": 0.0001,
+      "step": 3697
+    },
+    {
+      "epoch": 0.23831926274408713,
+      "grad_norm": 0.018464397355722775,
+      "learning_rate": 8.464733261725744e-06,
+      "loss": 0.0002,
+      "step": 3698
+    },
+    {
+      "epoch": 0.2383837081910163,
+      "grad_norm": 0.030809908730295932,
+      "learning_rate": 8.464017185821699e-06,
+      "loss": 0.0004,
+      "step": 3699
+    },
+    {
+      "epoch": 0.23844815363794547,
+      "grad_norm": 0.0016479319719454062,
+      "learning_rate": 8.463301109917651e-06,
+      "loss": 0.0,
+      "step": 3700
+    },
+    {
+      "epoch": 0.23851259908487465,
+      "grad_norm": 0.0053921943411062425,
+      "learning_rate": 8.462585034013606e-06,
+      "loss": 0.0,
+      "step": 3701
+    },
+    {
+      "epoch": 0.23857704453180384,
+      "grad_norm": 0.010372835272828807,
+      "learning_rate": 8.46186895810956e-06,
+      "loss": 0.0,
+      "step": 3702
+    },
+    {
+      "epoch": 0.238641489978733,
+      "grad_norm": 0.009654674728061838,
+      "learning_rate": 8.461152882205514e-06,
+      "loss": 0.0,
+      "step": 3703
+    },
+    {
+      "epoch": 0.23870593542566218,
+      "grad_norm": 0.0007935941029292266,
+      "learning_rate": 8.460436806301469e-06,
+      "loss": 0.0,
+      "step": 3704
+    },
+    {
+      "epoch": 0.23877038087259136,
+      "grad_norm": 0.0005360464652635607,
+      "learning_rate": 8.459720730397423e-06,
+      "loss": 0.0,
+      "step": 3705
+    },
+    {
+      "epoch": 0.23883482631952052,
+      "grad_norm": 0.0045528072264818704,
+      "learning_rate": 8.459004654493377e-06,
+      "loss": 0.0,
+      "step": 3706
+    },
+    {
+      "epoch": 0.2388992717664497,
+      "grad_norm": 0.0004918566041346934,
+      "learning_rate": 8.458288578589331e-06,
+      "loss": 0.0,
+      "step": 3707
+    },
+    {
+      "epoch": 0.23896371721337886,
+      "grad_norm": 0.01280604837715533,
+      "learning_rate": 8.457572502685286e-06,
+      "loss": 0.0,
+      "step": 3708
+    },
+    {
+      "epoch": 0.23902816266030805,
+      "grad_norm": 0.014364710962126858,
+      "learning_rate": 8.456856426781238e-06,
+      "loss": 0.0001,
+      "step": 3709
+    },
+    {
+      "epoch": 0.23909260810723723,
+      "grad_norm": 0.002629268415838148,
+      "learning_rate": 8.456140350877193e-06,
+      "loss": 0.0,
+      "step": 3710
+    },
+    {
+      "epoch": 0.2391570535541664,
+      "grad_norm": 0.07447493721494977,
+      "learning_rate": 8.455424274973147e-06,
+      "loss": 0.0008,
+      "step": 3711
+    },
+    {
+      "epoch": 0.23922149900109557,
+      "grad_norm": 0.5730290447183728,
+      "learning_rate": 8.454708199069101e-06,
+      "loss": 0.0059,
+      "step": 3712
+    },
+    {
+      "epoch": 0.23928594444802476,
+      "grad_norm": 0.0012635572436632849,
+      "learning_rate": 8.453992123165056e-06,
+      "loss": 0.0,
+      "step": 3713
+    },
+    {
+      "epoch": 0.2393503898949539,
+      "grad_norm": 0.024047417080047616,
+      "learning_rate": 8.45327604726101e-06,
+      "loss": 0.0,
+      "step": 3714
+    },
+    {
+      "epoch": 0.2394148353418831,
+      "grad_norm": 0.2183628402735261,
+      "learning_rate": 8.452559971356966e-06,
+      "loss": 0.0009,
+      "step": 3715
+    },
+    {
+      "epoch": 0.23947928078881228,
+      "grad_norm": 0.0006938746252782697,
+      "learning_rate": 8.451843895452918e-06,
+      "loss": 0.0,
+      "step": 3716
+    },
+    {
+      "epoch": 0.23954372623574144,
+      "grad_norm": 0.0038200120856466476,
+      "learning_rate": 8.451127819548873e-06,
+      "loss": 0.0,
+      "step": 3717
+    },
+    {
+      "epoch": 0.23960817168267062,
+      "grad_norm": 0.21408100848688394,
+      "learning_rate": 8.450411743644827e-06,
+      "loss": 0.0116,
+      "step": 3718
+    },
+    {
+      "epoch": 0.2396726171295998,
+      "grad_norm": 0.0025670123515598884,
+      "learning_rate": 8.449695667740781e-06,
+      "loss": 0.0,
+      "step": 3719
+    },
+    {
+      "epoch": 0.23973706257652896,
+      "grad_norm": 0.017562171034058886,
+      "learning_rate": 8.448979591836736e-06,
+      "loss": 0.0,
+      "step": 3720
+    },
+    {
+      "epoch": 0.23980150802345815,
+      "grad_norm": 0.0006715029546785951,
+      "learning_rate": 8.44826351593269e-06,
+      "loss": 0.0,
+      "step": 3721
+    },
+    {
+      "epoch": 0.2398659534703873,
+      "grad_norm": 0.03602618355161491,
+      "learning_rate": 8.447547440028644e-06,
+      "loss": 0.0001,
+      "step": 3722
+    },
+    {
+      "epoch": 0.2399303989173165,
+      "grad_norm": 0.018207445749766946,
+      "learning_rate": 8.446831364124599e-06,
+      "loss": 0.0001,
+      "step": 3723
+    },
+    {
+      "epoch": 0.23999484436424567,
+      "grad_norm": 0.12231537824269463,
+      "learning_rate": 8.446115288220553e-06,
+      "loss": 0.0002,
+      "step": 3724
+    },
+    {
+      "epoch": 0.24005928981117483,
+      "grad_norm": 0.11532271765071904,
+      "learning_rate": 8.445399212316505e-06,
+      "loss": 0.0019,
+      "step": 3725
+    },
+    {
+      "epoch": 0.24012373525810402,
+      "grad_norm": 2.264262664560703,
+      "learning_rate": 8.44468313641246e-06,
+      "loss": 0.0274,
+      "step": 3726
+    },
+    {
+      "epoch": 0.2401881807050332,
+      "grad_norm": 0.0012899088975698734,
+      "learning_rate": 8.443967060508414e-06,
+      "loss": 0.0,
+      "step": 3727
+    },
+    {
+      "epoch": 0.24025262615196236,
+      "grad_norm": 0.01366843452761101,
+      "learning_rate": 8.443250984604368e-06,
+      "loss": 0.0,
+      "step": 3728
+    },
+    {
+      "epoch": 0.24031707159889154,
+      "grad_norm": 0.15249607222806652,
+      "learning_rate": 8.442534908700323e-06,
+      "loss": 0.0006,
+      "step": 3729
+    },
+    {
+      "epoch": 0.24038151704582073,
+      "grad_norm": 0.021124642403800617,
+      "learning_rate": 8.441818832796277e-06,
+      "loss": 0.0,
+      "step": 3730
+    },
+    {
+      "epoch": 0.24044596249274988,
+      "grad_norm": 0.12230660647504508,
+      "learning_rate": 8.441102756892231e-06,
+      "loss": 0.0011,
+      "step": 3731
+    },
+    {
+      "epoch": 0.24051040793967907,
+      "grad_norm": 0.9040274496763755,
+      "learning_rate": 8.440386680988186e-06,
+      "loss": 0.0071,
+      "step": 3732
+    },
+    {
+      "epoch": 0.24057485338660822,
+      "grad_norm": 0.0016440136482255797,
+      "learning_rate": 8.43967060508414e-06,
+      "loss": 0.0,
+      "step": 3733
+    },
+    {
+      "epoch": 0.2406392988335374,
+      "grad_norm": 0.003047586958538098,
+      "learning_rate": 8.438954529180094e-06,
+      "loss": 0.0,
+      "step": 3734
+    },
+    {
+      "epoch": 0.2407037442804666,
+      "grad_norm": 0.0009673710376870563,
+      "learning_rate": 8.438238453276047e-06,
+      "loss": 0.0,
+      "step": 3735
+    },
+    {
+      "epoch": 0.24076818972739575,
+      "grad_norm": 0.016850687581852576,
+      "learning_rate": 8.437522377372001e-06,
+      "loss": 0.0001,
+      "step": 3736
+    },
+    {
+      "epoch": 0.24083263517432493,
+      "grad_norm": 0.007526774476757326,
+      "learning_rate": 8.436806301467955e-06,
+      "loss": 0.0001,
+      "step": 3737
+    },
+    {
+      "epoch": 0.24089708062125412,
+      "grad_norm": 0.13067564235859247,
+      "learning_rate": 8.436090225563911e-06,
+      "loss": 0.0003,
+      "step": 3738
+    },
+    {
+      "epoch": 0.24096152606818327,
+      "grad_norm": 0.10502113619162073,
+      "learning_rate": 8.435374149659866e-06,
+      "loss": 0.0004,
+      "step": 3739
+    },
+    {
+      "epoch": 0.24102597151511246,
+      "grad_norm": 0.05960794534318693,
+      "learning_rate": 8.43465807375582e-06,
+      "loss": 0.0001,
+      "step": 3740
+    },
+    {
+      "epoch": 0.24109041696204164,
+      "grad_norm": 0.04238541013913969,
+      "learning_rate": 8.433941997851773e-06,
+      "loss": 0.0003,
+      "step": 3741
+    },
+    {
+      "epoch": 0.2411548624089708,
+      "grad_norm": 0.018773961960864012,
+      "learning_rate": 8.433225921947727e-06,
+      "loss": 0.0016,
+      "step": 3742
+    },
+    {
+      "epoch": 0.24121930785589998,
+      "grad_norm": 0.002479640335606827,
+      "learning_rate": 8.432509846043681e-06,
+      "loss": 0.0,
+      "step": 3743
+    },
+    {
+      "epoch": 0.24128375330282917,
+      "grad_norm": 0.06248413383738307,
+      "learning_rate": 8.431793770139635e-06,
+      "loss": 0.0001,
+      "step": 3744
+    },
+    {
+      "epoch": 0.24134819874975832,
+      "grad_norm": 0.0005678648591910123,
+      "learning_rate": 8.43107769423559e-06,
+      "loss": 0.0,
+      "step": 3745
+    },
+    {
+      "epoch": 0.2414126441966875,
+      "grad_norm": 0.0013594156974299353,
+      "learning_rate": 8.430361618331544e-06,
+      "loss": 0.0,
+      "step": 3746
+    },
+    {
+      "epoch": 0.24147708964361667,
+      "grad_norm": 0.0005139733092100992,
+      "learning_rate": 8.429645542427498e-06,
+      "loss": 0.0,
+      "step": 3747
+    },
+    {
+      "epoch": 0.24154153509054585,
+      "grad_norm": 0.08670038037296418,
+      "learning_rate": 8.428929466523453e-06,
+      "loss": 0.0001,
+      "step": 3748
+    },
+    {
+      "epoch": 0.24160598053747503,
+      "grad_norm": 0.8153278218675962,
+      "learning_rate": 8.428213390619407e-06,
+      "loss": 0.0049,
+      "step": 3749
+    },
+    {
+      "epoch": 0.2416704259844042,
+      "grad_norm": 0.0035986476689472626,
+      "learning_rate": 8.427497314715361e-06,
+      "loss": 0.0,
+      "step": 3750
+    },
+    {
+      "epoch": 0.24173487143133338,
+      "grad_norm": 0.00044391004078189354,
+      "learning_rate": 8.426781238811314e-06,
+      "loss": 0.0,
+      "step": 3751
+    },
+    {
+      "epoch": 0.24179931687826256,
+      "grad_norm": 0.01262247356549068,
+      "learning_rate": 8.426065162907268e-06,
+      "loss": 0.0001,
+      "step": 3752
+    },
+    {
+      "epoch": 0.24186376232519172,
+      "grad_norm": 0.02262652431529532,
+      "learning_rate": 8.425349087003222e-06,
+      "loss": 0.0,
+      "step": 3753
+    },
+    {
+      "epoch": 0.2419282077721209,
+      "grad_norm": 0.0008362116585841875,
+      "learning_rate": 8.424633011099177e-06,
+      "loss": 0.0,
+      "step": 3754
+    },
+    {
+      "epoch": 0.24199265321905009,
+      "grad_norm": 0.005311226408071963,
+      "learning_rate": 8.423916935195131e-06,
+      "loss": 0.0,
+      "step": 3755
+    },
+    {
+      "epoch": 0.24205709866597924,
+      "grad_norm": 7.897391349834373e-05,
+      "learning_rate": 8.423200859291085e-06,
+      "loss": 0.0,
+      "step": 3756
+    },
+    {
+      "epoch": 0.24212154411290843,
+      "grad_norm": 0.007048575324203542,
+      "learning_rate": 8.42248478338704e-06,
+      "loss": 0.0001,
+      "step": 3757
+    },
+    {
+      "epoch": 0.2421859895598376,
+      "grad_norm": 0.002582242938985846,
+      "learning_rate": 8.421768707482994e-06,
+      "loss": 0.0,
+      "step": 3758
+    },
+    {
+      "epoch": 0.24225043500676677,
+      "grad_norm": 0.0048435534680228,
+      "learning_rate": 8.421052631578948e-06,
+      "loss": 0.0,
+      "step": 3759
+    },
+    {
+      "epoch": 0.24231488045369595,
+      "grad_norm": 0.4563282507589461,
+      "learning_rate": 8.420336555674903e-06,
+      "loss": 0.0009,
+      "step": 3760
+    },
+    {
+      "epoch": 0.2423793259006251,
+      "grad_norm": 0.002779418637291888,
+      "learning_rate": 8.419620479770857e-06,
+      "loss": 0.0,
+      "step": 3761
+    },
+    {
+      "epoch": 0.2424437713475543,
+      "grad_norm": 0.0027071330977519806,
+      "learning_rate": 8.418904403866811e-06,
+      "loss": 0.0,
+      "step": 3762
+    },
+    {
+      "epoch": 0.24250821679448348,
+      "grad_norm": 0.010813946842943904,
+      "learning_rate": 8.418188327962765e-06,
+      "loss": 0.0001,
+      "step": 3763
+    },
+    {
+      "epoch": 0.24257266224141263,
+      "grad_norm": 0.4170509314868395,
+      "learning_rate": 8.41747225205872e-06,
+      "loss": 0.0019,
+      "step": 3764
+    },
+    {
+      "epoch": 0.24263710768834182,
+      "grad_norm": 0.005691707930402684,
+      "learning_rate": 8.416756176154674e-06,
+      "loss": 0.0,
+      "step": 3765
+    },
+    {
+      "epoch": 0.242701553135271,
+      "grad_norm": 0.03813354713701013,
+      "learning_rate": 8.416040100250628e-06,
+      "loss": 0.0004,
+      "step": 3766
+    },
+    {
+      "epoch": 0.24276599858220016,
+      "grad_norm": 0.032524933022695664,
+      "learning_rate": 8.415324024346581e-06,
+      "loss": 0.0003,
+      "step": 3767
+    },
+    {
+      "epoch": 0.24283044402912934,
+      "grad_norm": 0.004065155547747669,
+      "learning_rate": 8.414607948442535e-06,
+      "loss": 0.0,
+      "step": 3768
+    },
+    {
+      "epoch": 0.24289488947605853,
+      "grad_norm": 0.25707170231896603,
+      "learning_rate": 8.41389187253849e-06,
+      "loss": 0.0003,
+      "step": 3769
+    },
+    {
+      "epoch": 0.24295933492298769,
+      "grad_norm": 0.9013399427931535,
+      "learning_rate": 8.413175796634444e-06,
+      "loss": 0.0075,
+      "step": 3770
+    },
+    {
+      "epoch": 0.24302378036991687,
+      "grad_norm": 0.0021442310344541708,
+      "learning_rate": 8.412459720730398e-06,
+      "loss": 0.0,
+      "step": 3771
+    },
+    {
+      "epoch": 0.24308822581684603,
+      "grad_norm": 0.10145635743766157,
+      "learning_rate": 8.411743644826352e-06,
+      "loss": 0.0005,
+      "step": 3772
+    },
+    {
+      "epoch": 0.2431526712637752,
+      "grad_norm": 0.0016105101427803788,
+      "learning_rate": 8.411027568922307e-06,
+      "loss": 0.0,
+      "step": 3773
+    },
+    {
+      "epoch": 0.2432171167107044,
+      "grad_norm": 0.00012493315542511565,
+      "learning_rate": 8.410311493018261e-06,
+      "loss": 0.0,
+      "step": 3774
+    },
+    {
+      "epoch": 0.24328156215763355,
+      "grad_norm": 0.019223677016477023,
+      "learning_rate": 8.409595417114215e-06,
+      "loss": 0.0001,
+      "step": 3775
+    },
+    {
+      "epoch": 0.24334600760456274,
+      "grad_norm": 0.005873850051817299,
+      "learning_rate": 8.40887934121017e-06,
+      "loss": 0.0001,
+      "step": 3776
+    },
+    {
+      "epoch": 0.24341045305149192,
+      "grad_norm": 0.011179190813958233,
+      "learning_rate": 8.408163265306122e-06,
+      "loss": 0.0001,
+      "step": 3777
+    },
+    {
+      "epoch": 0.24347489849842108,
+      "grad_norm": 0.0015477369522274402,
+      "learning_rate": 8.407447189402077e-06,
+      "loss": 0.0,
+      "step": 3778
+    },
+    {
+      "epoch": 0.24353934394535026,
+      "grad_norm": 0.0017133007265412168,
+      "learning_rate": 8.40673111349803e-06,
+      "loss": 0.0,
+      "step": 3779
+    },
+    {
+      "epoch": 0.24360378939227945,
+      "grad_norm": 0.0008757344192784564,
+      "learning_rate": 8.406015037593985e-06,
+      "loss": 0.0,
+      "step": 3780
+    },
+    {
+      "epoch": 0.2436682348392086,
+      "grad_norm": 0.007403871914327963,
+      "learning_rate": 8.40529896168994e-06,
+      "loss": 0.0,
+      "step": 3781
+    },
+    {
+      "epoch": 0.2437326802861378,
+      "grad_norm": 0.010201067541391566,
+      "learning_rate": 8.404582885785894e-06,
+      "loss": 0.0,
+      "step": 3782
+    },
+    {
+      "epoch": 0.24379712573306697,
+      "grad_norm": 0.015937982305306662,
+      "learning_rate": 8.403866809881848e-06,
+      "loss": 0.0,
+      "step": 3783
+    },
+    {
+      "epoch": 0.24386157117999613,
+      "grad_norm": 0.028991708253557905,
+      "learning_rate": 8.403150733977802e-06,
+      "loss": 0.0002,
+      "step": 3784
+    },
+    {
+      "epoch": 0.2439260166269253,
+      "grad_norm": 0.04578634280546854,
+      "learning_rate": 8.402434658073757e-06,
+      "loss": 0.0003,
+      "step": 3785
+    },
+    {
+      "epoch": 0.24399046207385447,
+      "grad_norm": 0.003052409221744512,
+      "learning_rate": 8.401718582169711e-06,
+      "loss": 0.0,
+      "step": 3786
+    },
+    {
+      "epoch": 0.24405490752078365,
+      "grad_norm": 0.010366611251185617,
+      "learning_rate": 8.401002506265665e-06,
+      "loss": 0.0,
+      "step": 3787
+    },
+    {
+      "epoch": 0.24411935296771284,
+      "grad_norm": 0.012989847995797284,
+      "learning_rate": 8.40028643036162e-06,
+      "loss": 0.0001,
+      "step": 3788
+    },
+    {
+      "epoch": 0.244183798414642,
+      "grad_norm": 1.509528493086036,
+      "learning_rate": 8.399570354457574e-06,
+      "loss": 0.0121,
+      "step": 3789
+    },
+    {
+      "epoch": 0.24424824386157118,
+      "grad_norm": 0.13255768762843395,
+      "learning_rate": 8.398854278553528e-06,
+      "loss": 0.0015,
+      "step": 3790
+    },
+    {
+      "epoch": 0.24431268930850036,
+      "grad_norm": 0.07067210243967802,
+      "learning_rate": 8.398138202649482e-06,
+      "loss": 0.0003,
+      "step": 3791
+    },
+    {
+      "epoch": 0.24437713475542952,
+      "grad_norm": 0.014960046766105765,
+      "learning_rate": 8.397422126745437e-06,
+      "loss": 0.0,
+      "step": 3792
+    },
+    {
+      "epoch": 0.2444415802023587,
+      "grad_norm": 0.008710688093756856,
+      "learning_rate": 8.39670605084139e-06,
+      "loss": 0.0,
+      "step": 3793
+    },
+    {
+      "epoch": 0.2445060256492879,
+      "grad_norm": 0.255980528473747,
+      "learning_rate": 8.395989974937344e-06,
+      "loss": 0.0017,
+      "step": 3794
+    },
+    {
+      "epoch": 0.24457047109621705,
+      "grad_norm": 0.00026442442359102594,
+      "learning_rate": 8.395273899033298e-06,
+      "loss": 0.0,
+      "step": 3795
+    },
+    {
+      "epoch": 0.24463491654314623,
+      "grad_norm": 0.3572974900567217,
+      "learning_rate": 8.394557823129252e-06,
+      "loss": 0.0004,
+      "step": 3796
+    },
+    {
+      "epoch": 0.24469936199007541,
+      "grad_norm": 0.04774931035227209,
+      "learning_rate": 8.393841747225207e-06,
+      "loss": 0.0007,
+      "step": 3797
+    },
+    {
+      "epoch": 0.24476380743700457,
+      "grad_norm": 0.10715923279977019,
+      "learning_rate": 8.39312567132116e-06,
+      "loss": 0.0005,
+      "step": 3798
+    },
+    {
+      "epoch": 0.24482825288393376,
+      "grad_norm": 0.0012820152532673366,
+      "learning_rate": 8.392409595417115e-06,
+      "loss": 0.0,
+      "step": 3799
+    },
+    {
+      "epoch": 0.2448926983308629,
+      "grad_norm": 0.3422857221960594,
+      "learning_rate": 8.39169351951307e-06,
+      "loss": 0.0016,
+      "step": 3800
+    },
+    {
+      "epoch": 0.2449571437777921,
+      "grad_norm": 0.004484513715100537,
+      "learning_rate": 8.390977443609024e-06,
+      "loss": 0.0,
+      "step": 3801
+    },
+    {
+      "epoch": 0.24502158922472128,
+      "grad_norm": 1.0571975791421366,
+      "learning_rate": 8.390261367704976e-06,
+      "loss": 0.0039,
+      "step": 3802
+    },
+    {
+      "epoch": 0.24508603467165044,
+      "grad_norm": 0.0014573274750767575,
+      "learning_rate": 8.38954529180093e-06,
+      "loss": 0.0,
+      "step": 3803
+    },
+    {
+      "epoch": 0.24515048011857962,
+      "grad_norm": 0.018570197831693735,
+      "learning_rate": 8.388829215896885e-06,
+      "loss": 0.0,
+      "step": 3804
+    },
+    {
+      "epoch": 0.2452149255655088,
+      "grad_norm": 0.0013457453855705087,
+      "learning_rate": 8.38811313999284e-06,
+      "loss": 0.0,
+      "step": 3805
+    },
+    {
+      "epoch": 0.24527937101243796,
+      "grad_norm": 0.0009432171359150775,
+      "learning_rate": 8.387397064088793e-06,
+      "loss": 0.0,
+      "step": 3806
+    },
+    {
+      "epoch": 0.24534381645936715,
+      "grad_norm": 0.04670797332971043,
+      "learning_rate": 8.386680988184748e-06,
+      "loss": 0.0003,
+      "step": 3807
+    },
+    {
+      "epoch": 0.24540826190629633,
+      "grad_norm": 0.0033869906812700606,
+      "learning_rate": 8.385964912280704e-06,
+      "loss": 0.0,
+      "step": 3808
+    },
+    {
+      "epoch": 0.2454727073532255,
+      "grad_norm": 0.001629549809419056,
+      "learning_rate": 8.385248836376656e-06,
+      "loss": 0.0,
+      "step": 3809
+    },
+    {
+      "epoch": 0.24553715280015467,
+      "grad_norm": 0.0008565285923683144,
+      "learning_rate": 8.38453276047261e-06,
+      "loss": 0.0,
+      "step": 3810
+    },
+    {
+      "epoch": 0.24560159824708383,
+      "grad_norm": 0.07435944046824011,
+      "learning_rate": 8.383816684568565e-06,
+      "loss": 0.0002,
+      "step": 3811
+    },
+    {
+      "epoch": 0.24566604369401301,
+      "grad_norm": 1.455835009863066,
+      "learning_rate": 8.38310060866452e-06,
+      "loss": 0.0065,
+      "step": 3812
+    },
+    {
+      "epoch": 0.2457304891409422,
+      "grad_norm": 0.02158769033318039,
+      "learning_rate": 8.382384532760474e-06,
+      "loss": 0.0,
+      "step": 3813
+    },
+    {
+      "epoch": 0.24579493458787136,
+      "grad_norm": 0.009037092296314405,
+      "learning_rate": 8.381668456856428e-06,
+      "loss": 0.0,
+      "step": 3814
+    },
+    {
+      "epoch": 0.24585938003480054,
+      "grad_norm": 0.01045907662154638,
+      "learning_rate": 8.380952380952382e-06,
+      "loss": 0.0,
+      "step": 3815
+    },
+    {
+      "epoch": 0.24592382548172972,
+      "grad_norm": 0.2750826120917977,
+      "learning_rate": 8.380236305048336e-06,
+      "loss": 0.001,
+      "step": 3816
+    },
+    {
+      "epoch": 0.24598827092865888,
+      "grad_norm": 0.00485543121189977,
+      "learning_rate": 8.37952022914429e-06,
+      "loss": 0.0,
+      "step": 3817
+    },
+    {
+      "epoch": 0.24605271637558807,
+      "grad_norm": 0.0069404249703027745,
+      "learning_rate": 8.378804153240243e-06,
+      "loss": 0.0,
+      "step": 3818
+    },
+    {
+      "epoch": 0.24611716182251725,
+      "grad_norm": 0.0008645888421194356,
+      "learning_rate": 8.378088077336198e-06,
+      "loss": 0.0,
+      "step": 3819
+    },
+    {
+      "epoch": 0.2461816072694464,
+      "grad_norm": 0.0006523125352911542,
+      "learning_rate": 8.377372001432152e-06,
+      "loss": 0.0,
+      "step": 3820
+    },
+    {
+      "epoch": 0.2462460527163756,
+      "grad_norm": 0.022340756216589642,
+      "learning_rate": 8.376655925528106e-06,
+      "loss": 0.0003,
+      "step": 3821
+    },
+    {
+      "epoch": 0.24631049816330478,
+      "grad_norm": 0.02572042481418643,
+      "learning_rate": 8.37593984962406e-06,
+      "loss": 0.0001,
+      "step": 3822
+    },
+    {
+      "epoch": 0.24637494361023393,
+      "grad_norm": 0.05793174450618309,
+      "learning_rate": 8.375223773720015e-06,
+      "loss": 0.0001,
+      "step": 3823
+    },
+    {
+      "epoch": 0.24643938905716312,
+      "grad_norm": 0.0013580941737972847,
+      "learning_rate": 8.37450769781597e-06,
+      "loss": 0.0,
+      "step": 3824
+    },
+    {
+      "epoch": 0.24650383450409227,
+      "grad_norm": 0.023467460744772406,
+      "learning_rate": 8.373791621911923e-06,
+      "loss": 0.0001,
+      "step": 3825
+    },
+    {
+      "epoch": 0.24656827995102146,
+      "grad_norm": 0.002280545682215033,
+      "learning_rate": 8.373075546007878e-06,
+      "loss": 0.0,
+      "step": 3826
+    },
+    {
+      "epoch": 0.24663272539795064,
+      "grad_norm": 0.20302869274283905,
+      "learning_rate": 8.372359470103832e-06,
+      "loss": 0.0019,
+      "step": 3827
+    },
+    {
+      "epoch": 0.2466971708448798,
+      "grad_norm": 2.142446703431541,
+      "learning_rate": 8.371643394199785e-06,
+      "loss": 0.0132,
+      "step": 3828
+    },
+    {
+      "epoch": 0.24676161629180898,
+      "grad_norm": 0.11213382494207665,
+      "learning_rate": 8.370927318295739e-06,
+      "loss": 0.0018,
+      "step": 3829
+    },
+    {
+      "epoch": 0.24682606173873817,
+      "grad_norm": 0.5988444067924492,
+      "learning_rate": 8.370211242391693e-06,
+      "loss": 0.0016,
+      "step": 3830
+    },
+    {
+      "epoch": 0.24689050718566732,
+      "grad_norm": 0.002282277636076149,
+      "learning_rate": 8.36949516648765e-06,
+      "loss": 0.0,
+      "step": 3831
+    },
+    {
+      "epoch": 0.2469549526325965,
+      "grad_norm": 0.017599929260292843,
+      "learning_rate": 8.368779090583604e-06,
+      "loss": 0.0002,
+      "step": 3832
+    },
+    {
+      "epoch": 0.2470193980795257,
+      "grad_norm": 8.669601616320829e-05,
+      "learning_rate": 8.368063014679558e-06,
+      "loss": 0.0,
+      "step": 3833
+    },
+    {
+      "epoch": 0.24708384352645485,
+      "grad_norm": 1.5981672071508441,
+      "learning_rate": 8.36734693877551e-06,
+      "loss": 0.0037,
+      "step": 3834
+    },
+    {
+      "epoch": 0.24714828897338403,
+      "grad_norm": 0.008489611550036908,
+      "learning_rate": 8.366630862871465e-06,
+      "loss": 0.0001,
+      "step": 3835
+    },
+    {
+      "epoch": 0.24721273442031322,
+      "grad_norm": 0.003644974177941349,
+      "learning_rate": 8.365914786967419e-06,
+      "loss": 0.0,
+      "step": 3836
+    },
+    {
+      "epoch": 0.24727717986724237,
+      "grad_norm": 0.003976308666016474,
+      "learning_rate": 8.365198711063373e-06,
+      "loss": 0.0,
+      "step": 3837
+    },
+    {
+      "epoch": 0.24734162531417156,
+      "grad_norm": 0.27547633413259537,
+      "learning_rate": 8.364482635159328e-06,
+      "loss": 0.0092,
+      "step": 3838
+    },
+    {
+      "epoch": 0.24740607076110072,
+      "grad_norm": 0.0015722282850001267,
+      "learning_rate": 8.363766559255282e-06,
+      "loss": 0.0,
+      "step": 3839
+    },
+    {
+      "epoch": 0.2474705162080299,
+      "grad_norm": 0.005728081698721112,
+      "learning_rate": 8.363050483351236e-06,
+      "loss": 0.0,
+      "step": 3840
+    },
+    {
+      "epoch": 0.24753496165495908,
+      "grad_norm": 0.019088249596522798,
+      "learning_rate": 8.36233440744719e-06,
+      "loss": 0.0002,
+      "step": 3841
+    },
+    {
+      "epoch": 0.24759940710188824,
+      "grad_norm": 0.13053990981847335,
+      "learning_rate": 8.361618331543145e-06,
+      "loss": 0.0004,
+      "step": 3842
+    },
+    {
+      "epoch": 0.24766385254881743,
+      "grad_norm": 0.022335887407941846,
+      "learning_rate": 8.3609022556391e-06,
+      "loss": 0.0002,
+      "step": 3843
+    },
+    {
+      "epoch": 0.2477282979957466,
+      "grad_norm": 0.029563943186087686,
+      "learning_rate": 8.360186179735052e-06,
+      "loss": 0.0001,
+      "step": 3844
+    },
+    {
+      "epoch": 0.24779274344267577,
+      "grad_norm": 0.009087314706417401,
+      "learning_rate": 8.359470103831006e-06,
+      "loss": 0.0,
+      "step": 3845
+    },
+    {
+      "epoch": 0.24785718888960495,
+      "grad_norm": 0.00653539308431923,
+      "learning_rate": 8.35875402792696e-06,
+      "loss": 0.0,
+      "step": 3846
+    },
+    {
+      "epoch": 0.24792163433653414,
+      "grad_norm": 0.4713887646308475,
+      "learning_rate": 8.358037952022915e-06,
+      "loss": 0.0011,
+      "step": 3847
+    },
+    {
+      "epoch": 0.2479860797834633,
+      "grad_norm": 0.024816317722805743,
+      "learning_rate": 8.357321876118869e-06,
+      "loss": 0.0001,
+      "step": 3848
+    },
+    {
+      "epoch": 0.24805052523039248,
+      "grad_norm": 0.038245842028925114,
+      "learning_rate": 8.356605800214823e-06,
+      "loss": 0.0005,
+      "step": 3849
+    },
+    {
+      "epoch": 0.24811497067732163,
+      "grad_norm": 0.1631073794859977,
+      "learning_rate": 8.355889724310778e-06,
+      "loss": 0.0007,
+      "step": 3850
+    },
+    {
+      "epoch": 0.24817941612425082,
+      "grad_norm": 0.12254443599798615,
+      "learning_rate": 8.355173648406732e-06,
+      "loss": 0.0006,
+      "step": 3851
+    },
+    {
+      "epoch": 0.24824386157118,
+      "grad_norm": 0.000737590175483497,
+      "learning_rate": 8.354457572502686e-06,
+      "loss": 0.0,
+      "step": 3852
+    },
+    {
+      "epoch": 0.24830830701810916,
+      "grad_norm": 0.0006945422121030539,
+      "learning_rate": 8.35374149659864e-06,
+      "loss": 0.0,
+      "step": 3853
+    },
+    {
+      "epoch": 0.24837275246503834,
+      "grad_norm": 0.019502762791774902,
+      "learning_rate": 8.353025420694593e-06,
+      "loss": 0.0003,
+      "step": 3854
+    },
+    {
+      "epoch": 0.24843719791196753,
+      "grad_norm": 0.08053178332273953,
+      "learning_rate": 8.352309344790549e-06,
+      "loss": 0.0009,
+      "step": 3855
+    },
+    {
+      "epoch": 0.24850164335889668,
+      "grad_norm": 0.14025133875090617,
+      "learning_rate": 8.351593268886503e-06,
+      "loss": 0.0012,
+      "step": 3856
+    },
+    {
+      "epoch": 0.24856608880582587,
+      "grad_norm": 0.006439356158143003,
+      "learning_rate": 8.350877192982458e-06,
+      "loss": 0.0,
+      "step": 3857
+    },
+    {
+      "epoch": 0.24863053425275505,
+      "grad_norm": 0.00029574026453269123,
+      "learning_rate": 8.350161117078412e-06,
+      "loss": 0.0,
+      "step": 3858
+    },
+    {
+      "epoch": 0.2486949796996842,
+      "grad_norm": 0.004987156108916005,
+      "learning_rate": 8.349445041174366e-06,
+      "loss": 0.0,
+      "step": 3859
+    },
+    {
+      "epoch": 0.2487594251466134,
+      "grad_norm": 0.015322867790270124,
+      "learning_rate": 8.348728965270319e-06,
+      "loss": 0.0,
+      "step": 3860
+    },
+    {
+      "epoch": 0.24882387059354258,
+      "grad_norm": 0.0010423756801406121,
+      "learning_rate": 8.348012889366273e-06,
+      "loss": 0.0,
+      "step": 3861
+    },
+    {
+      "epoch": 0.24888831604047174,
+      "grad_norm": 0.38380098203053353,
+      "learning_rate": 8.347296813462227e-06,
+      "loss": 0.0018,
+      "step": 3862
+    },
+    {
+      "epoch": 0.24895276148740092,
+      "grad_norm": 0.0022487493770144092,
+      "learning_rate": 8.346580737558182e-06,
+      "loss": 0.0,
+      "step": 3863
+    },
+    {
+      "epoch": 0.24901720693433008,
+      "grad_norm": 0.021213837093024088,
+      "learning_rate": 8.345864661654136e-06,
+      "loss": 0.0001,
+      "step": 3864
+    },
+    {
+      "epoch": 0.24908165238125926,
+      "grad_norm": 0.0027843664534250444,
+      "learning_rate": 8.34514858575009e-06,
+      "loss": 0.0,
+      "step": 3865
+    },
+    {
+      "epoch": 0.24914609782818845,
+      "grad_norm": 0.13952617361567596,
+      "learning_rate": 8.344432509846045e-06,
+      "loss": 0.0002,
+      "step": 3866
+    },
+    {
+      "epoch": 0.2492105432751176,
+      "grad_norm": 0.0010491756332098327,
+      "learning_rate": 8.343716433941999e-06,
+      "loss": 0.0,
+      "step": 3867
+    },
+    {
+      "epoch": 0.2492749887220468,
+      "grad_norm": 0.05011367783332657,
+      "learning_rate": 8.343000358037953e-06,
+      "loss": 0.0002,
+      "step": 3868
+    },
+    {
+      "epoch": 0.24933943416897597,
+      "grad_norm": 0.4646252671662538,
+      "learning_rate": 8.342284282133908e-06,
+      "loss": 0.0106,
+      "step": 3869
+    },
+    {
+      "epoch": 0.24940387961590513,
+      "grad_norm": 0.00023720217135674875,
+      "learning_rate": 8.34156820622986e-06,
+      "loss": 0.0,
+      "step": 3870
+    },
+    {
+      "epoch": 0.2494683250628343,
+      "grad_norm": 0.03269248525753471,
+      "learning_rate": 8.340852130325814e-06,
+      "loss": 0.0001,
+      "step": 3871
+    },
+    {
+      "epoch": 0.2495327705097635,
+      "grad_norm": 0.0042010391733460955,
+      "learning_rate": 8.340136054421769e-06,
+      "loss": 0.0,
+      "step": 3872
+    },
+    {
+      "epoch": 0.24959721595669265,
+      "grad_norm": 0.0717919685100648,
+      "learning_rate": 8.339419978517723e-06,
+      "loss": 0.0002,
+      "step": 3873
+    },
+    {
+      "epoch": 0.24966166140362184,
+      "grad_norm": 0.00015835011294514,
+      "learning_rate": 8.338703902613677e-06,
+      "loss": 0.0,
+      "step": 3874
+    },
+    {
+      "epoch": 0.24972610685055102,
+      "grad_norm": 0.013157551152446972,
+      "learning_rate": 8.337987826709632e-06,
+      "loss": 0.0,
+      "step": 3875
+    },
+    {
+      "epoch": 0.24979055229748018,
+      "grad_norm": 0.0015622204880051153,
+      "learning_rate": 8.337271750805586e-06,
+      "loss": 0.0,
+      "step": 3876
+    },
+    {
+      "epoch": 0.24985499774440936,
+      "grad_norm": 0.03520397086657459,
+      "learning_rate": 8.33655567490154e-06,
+      "loss": 0.0004,
+      "step": 3877
+    },
+    {
+      "epoch": 0.24991944319133852,
+      "grad_norm": 0.02710640237915254,
+      "learning_rate": 8.335839598997495e-06,
+      "loss": 0.0001,
+      "step": 3878
+    },
+    {
+      "epoch": 0.2499838886382677,
+      "grad_norm": 0.0010845842766640747,
+      "learning_rate": 8.335123523093449e-06,
+      "loss": 0.0,
+      "step": 3879
+    },
+    {
+      "epoch": 0.2500483340851969,
+      "grad_norm": 0.18885238021353518,
+      "learning_rate": 8.334407447189403e-06,
+      "loss": 0.0005,
+      "step": 3880
+    },
+    {
+      "epoch": 0.2501127795321261,
+      "grad_norm": 0.14326937212325316,
+      "learning_rate": 8.333691371285357e-06,
+      "loss": 0.0007,
+      "step": 3881
+    },
+    {
+      "epoch": 0.25017722497905526,
+      "grad_norm": 0.17603865426681994,
+      "learning_rate": 8.332975295381312e-06,
+      "loss": 0.0005,
+      "step": 3882
+    },
+    {
+      "epoch": 0.2502416704259844,
+      "grad_norm": 0.08609897712323253,
+      "learning_rate": 8.332259219477266e-06,
+      "loss": 0.0003,
+      "step": 3883
+    },
+    {
+      "epoch": 0.25030611587291357,
+      "grad_norm": 0.007551340900007415,
+      "learning_rate": 8.33154314357322e-06,
+      "loss": 0.0,
+      "step": 3884
+    },
+    {
+      "epoch": 0.25037056131984275,
+      "grad_norm": 0.057125706868368396,
+      "learning_rate": 8.330827067669175e-06,
+      "loss": 0.0001,
+      "step": 3885
+    },
+    {
+      "epoch": 0.25043500676677194,
+      "grad_norm": 0.0034849557503283534,
+      "learning_rate": 8.330110991765127e-06,
+      "loss": 0.0,
+      "step": 3886
+    },
+    {
+      "epoch": 0.2504994522137011,
+      "grad_norm": 0.0025515581732295125,
+      "learning_rate": 8.329394915861082e-06,
+      "loss": 0.0,
+      "step": 3887
+    },
+    {
+      "epoch": 0.25056389766063025,
+      "grad_norm": 0.0006230279157326317,
+      "learning_rate": 8.328678839957036e-06,
+      "loss": 0.0,
+      "step": 3888
+    },
+    {
+      "epoch": 0.25062834310755944,
+      "grad_norm": 0.01758640478425065,
+      "learning_rate": 8.32796276405299e-06,
+      "loss": 0.0002,
+      "step": 3889
+    },
+    {
+      "epoch": 0.2506927885544886,
+      "grad_norm": 0.46716078933857036,
+      "learning_rate": 8.327246688148944e-06,
+      "loss": 0.0034,
+      "step": 3890
+    },
+    {
+      "epoch": 0.2507572340014178,
+      "grad_norm": 0.067406871204759,
+      "learning_rate": 8.326530612244899e-06,
+      "loss": 0.0001,
+      "step": 3891
+    },
+    {
+      "epoch": 0.250821679448347,
+      "grad_norm": 0.0006241835739047378,
+      "learning_rate": 8.325814536340853e-06,
+      "loss": 0.0,
+      "step": 3892
+    },
+    {
+      "epoch": 0.2508861248952762,
+      "grad_norm": 0.31971812344868983,
+      "learning_rate": 8.325098460436807e-06,
+      "loss": 0.0009,
+      "step": 3893
+    },
+    {
+      "epoch": 0.2509505703422053,
+      "grad_norm": 0.028947342519578857,
+      "learning_rate": 8.324382384532762e-06,
+      "loss": 0.0002,
+      "step": 3894
+    },
+    {
+      "epoch": 0.2510150157891345,
+      "grad_norm": 0.004368482565717253,
+      "learning_rate": 8.323666308628714e-06,
+      "loss": 0.0,
+      "step": 3895
+    },
+    {
+      "epoch": 0.2510794612360637,
+      "grad_norm": 0.010188551518575773,
+      "learning_rate": 8.322950232724669e-06,
+      "loss": 0.0,
+      "step": 3896
+    },
+    {
+      "epoch": 0.25114390668299286,
+      "grad_norm": 0.16556811322935083,
+      "learning_rate": 8.322234156820623e-06,
+      "loss": 0.0008,
+      "step": 3897
+    },
+    {
+      "epoch": 0.25120835212992204,
+      "grad_norm": 0.04164728894373619,
+      "learning_rate": 8.321518080916577e-06,
+      "loss": 0.0001,
+      "step": 3898
+    },
+    {
+      "epoch": 0.25127279757685117,
+      "grad_norm": 0.002195380129568669,
+      "learning_rate": 8.320802005012531e-06,
+      "loss": 0.0,
+      "step": 3899
+    },
+    {
+      "epoch": 0.25133724302378035,
+      "grad_norm": 0.001985210248564425,
+      "learning_rate": 8.320085929108486e-06,
+      "loss": 0.0,
+      "step": 3900
+    },
+    {
+      "epoch": 0.25140168847070954,
+      "grad_norm": 0.001143081809185654,
+      "learning_rate": 8.319369853204442e-06,
+      "loss": 0.0,
+      "step": 3901
+    },
+    {
+      "epoch": 0.2514661339176387,
+      "grad_norm": 0.001505042883211427,
+      "learning_rate": 8.318653777300394e-06,
+      "loss": 0.0,
+      "step": 3902
+    },
+    {
+      "epoch": 0.2515305793645679,
+      "grad_norm": 0.07406394901749401,
+      "learning_rate": 8.317937701396349e-06,
+      "loss": 0.0006,
+      "step": 3903
+    },
+    {
+      "epoch": 0.2515950248114971,
+      "grad_norm": 0.0056360193079743005,
+      "learning_rate": 8.317221625492303e-06,
+      "loss": 0.0,
+      "step": 3904
+    },
+    {
+      "epoch": 0.2516594702584262,
+      "grad_norm": 0.061877072578334864,
+      "learning_rate": 8.316505549588257e-06,
+      "loss": 0.0002,
+      "step": 3905
+    },
+    {
+      "epoch": 0.2517239157053554,
+      "grad_norm": 0.1699128915889434,
+      "learning_rate": 8.315789473684212e-06,
+      "loss": 0.0005,
+      "step": 3906
+    },
+    {
+      "epoch": 0.2517883611522846,
+      "grad_norm": 0.00045409563031198523,
+      "learning_rate": 8.315073397780166e-06,
+      "loss": 0.0,
+      "step": 3907
+    },
+    {
+      "epoch": 0.2518528065992138,
+      "grad_norm": 0.025290433653521596,
+      "learning_rate": 8.31435732187612e-06,
+      "loss": 0.0,
+      "step": 3908
+    },
+    {
+      "epoch": 0.25191725204614296,
+      "grad_norm": 0.0014953556261391822,
+      "learning_rate": 8.313641245972074e-06,
+      "loss": 0.0,
+      "step": 3909
+    },
+    {
+      "epoch": 0.2519816974930721,
+      "grad_norm": 0.8381777895944582,
+      "learning_rate": 8.312925170068029e-06,
+      "loss": 0.0056,
+      "step": 3910
+    },
+    {
+      "epoch": 0.25204614294000127,
+      "grad_norm": 0.018405931304765668,
+      "learning_rate": 8.312209094163981e-06,
+      "loss": 0.0,
+      "step": 3911
+    },
+    {
+      "epoch": 0.25211058838693046,
+      "grad_norm": 0.11354576666309285,
+      "learning_rate": 8.311493018259936e-06,
+      "loss": 0.0002,
+      "step": 3912
+    },
+    {
+      "epoch": 0.25217503383385964,
+      "grad_norm": 0.053129490844862515,
+      "learning_rate": 8.31077694235589e-06,
+      "loss": 0.0001,
+      "step": 3913
+    },
+    {
+      "epoch": 0.2522394792807888,
+      "grad_norm": 0.0058663139011346475,
+      "learning_rate": 8.310060866451844e-06,
+      "loss": 0.0,
+      "step": 3914
+    },
+    {
+      "epoch": 0.252303924727718,
+      "grad_norm": 0.08415443369084287,
+      "learning_rate": 8.309344790547799e-06,
+      "loss": 0.0001,
+      "step": 3915
+    },
+    {
+      "epoch": 0.25236837017464714,
+      "grad_norm": 0.47324517501410757,
+      "learning_rate": 8.308628714643753e-06,
+      "loss": 0.0034,
+      "step": 3916
+    },
+    {
+      "epoch": 0.2524328156215763,
+      "grad_norm": 0.00036435772833055455,
+      "learning_rate": 8.307912638739707e-06,
+      "loss": 0.0,
+      "step": 3917
+    },
+    {
+      "epoch": 0.2524972610685055,
+      "grad_norm": 0.0016095858469143312,
+      "learning_rate": 8.307196562835661e-06,
+      "loss": 0.0,
+      "step": 3918
+    },
+    {
+      "epoch": 0.2525617065154347,
+      "grad_norm": 0.680525427646642,
+      "learning_rate": 8.306480486931616e-06,
+      "loss": 0.0049,
+      "step": 3919
+    },
+    {
+      "epoch": 0.2526261519623639,
+      "grad_norm": 0.0160279352790546,
+      "learning_rate": 8.30576441102757e-06,
+      "loss": 0.0,
+      "step": 3920
+    },
+    {
+      "epoch": 0.25269059740929306,
+      "grad_norm": 0.04248723436608822,
+      "learning_rate": 8.305048335123523e-06,
+      "loss": 0.0001,
+      "step": 3921
+    },
+    {
+      "epoch": 0.2527550428562222,
+      "grad_norm": 0.009271066669047413,
+      "learning_rate": 8.304332259219477e-06,
+      "loss": 0.0,
+      "step": 3922
+    },
+    {
+      "epoch": 0.2528194883031514,
+      "grad_norm": 0.0045926226252857045,
+      "learning_rate": 8.303616183315431e-06,
+      "loss": 0.0,
+      "step": 3923
+    },
+    {
+      "epoch": 0.25288393375008056,
+      "grad_norm": 0.0683818303419785,
+      "learning_rate": 8.302900107411385e-06,
+      "loss": 0.0001,
+      "step": 3924
+    },
+    {
+      "epoch": 0.25294837919700974,
+      "grad_norm": 0.4241263926590266,
+      "learning_rate": 8.302184031507341e-06,
+      "loss": 0.0016,
+      "step": 3925
+    },
+    {
+      "epoch": 0.2530128246439389,
+      "grad_norm": 0.002698765370629597,
+      "learning_rate": 8.301467955603296e-06,
+      "loss": 0.0,
+      "step": 3926
+    },
+    {
+      "epoch": 0.25307727009086806,
+      "grad_norm": 0.021524961969480675,
+      "learning_rate": 8.300751879699248e-06,
+      "loss": 0.0,
+      "step": 3927
+    },
+    {
+      "epoch": 0.25314171553779724,
+      "grad_norm": 0.0322740450649284,
+      "learning_rate": 8.300035803795203e-06,
+      "loss": 0.0,
+      "step": 3928
+    },
+    {
+      "epoch": 0.2532061609847264,
+      "grad_norm": 0.018648881771948716,
+      "learning_rate": 8.299319727891157e-06,
+      "loss": 0.0,
+      "step": 3929
+    },
+    {
+      "epoch": 0.2532706064316556,
+      "grad_norm": 0.11561774475944939,
+      "learning_rate": 8.298603651987111e-06,
+      "loss": 0.0001,
+      "step": 3930
+    },
+    {
+      "epoch": 0.2533350518785848,
+      "grad_norm": 0.0038557643494287752,
+      "learning_rate": 8.297887576083066e-06,
+      "loss": 0.0,
+      "step": 3931
+    },
+    {
+      "epoch": 0.253399497325514,
+      "grad_norm": 0.010070178706289873,
+      "learning_rate": 8.29717150017902e-06,
+      "loss": 0.0,
+      "step": 3932
+    },
+    {
+      "epoch": 0.2534639427724431,
+      "grad_norm": 1.0039694205866754,
+      "learning_rate": 8.296455424274974e-06,
+      "loss": 0.0032,
+      "step": 3933
+    },
+    {
+      "epoch": 0.2535283882193723,
+      "grad_norm": 0.06947501096791438,
+      "learning_rate": 8.295739348370928e-06,
+      "loss": 0.0,
+      "step": 3934
+    },
+    {
+      "epoch": 0.2535928336663015,
+      "grad_norm": 0.0042569530407342215,
+      "learning_rate": 8.295023272466883e-06,
+      "loss": 0.0,
+      "step": 3935
+    },
+    {
+      "epoch": 0.25365727911323066,
+      "grad_norm": 0.0014290376859672905,
+      "learning_rate": 8.294307196562837e-06,
+      "loss": 0.0,
+      "step": 3936
+    },
+    {
+      "epoch": 0.25372172456015984,
+      "grad_norm": 0.03599591192270969,
+      "learning_rate": 8.29359112065879e-06,
+      "loss": 0.0001,
+      "step": 3937
+    },
+    {
+      "epoch": 0.253786170007089,
+      "grad_norm": 0.03868331751936182,
+      "learning_rate": 8.292875044754744e-06,
+      "loss": 0.0,
+      "step": 3938
+    },
+    {
+      "epoch": 0.25385061545401816,
+      "grad_norm": 0.019385192551703966,
+      "learning_rate": 8.292158968850698e-06,
+      "loss": 0.0,
+      "step": 3939
+    },
+    {
+      "epoch": 0.25391506090094734,
+      "grad_norm": 0.002924765983032503,
+      "learning_rate": 8.291442892946653e-06,
+      "loss": 0.0,
+      "step": 3940
+    },
+    {
+      "epoch": 0.2539795063478765,
+      "grad_norm": 0.0067314597492258285,
+      "learning_rate": 8.290726817042607e-06,
+      "loss": 0.0,
+      "step": 3941
+    },
+    {
+      "epoch": 0.2540439517948057,
+      "grad_norm": 0.004637551407635732,
+      "learning_rate": 8.290010741138561e-06,
+      "loss": 0.0,
+      "step": 3942
+    },
+    {
+      "epoch": 0.2541083972417349,
+      "grad_norm": 0.0018488212029575274,
+      "learning_rate": 8.289294665234515e-06,
+      "loss": 0.0,
+      "step": 3943
+    },
+    {
+      "epoch": 0.254172842688664,
+      "grad_norm": 0.060456514226417284,
+      "learning_rate": 8.28857858933047e-06,
+      "loss": 0.0002,
+      "step": 3944
+    },
+    {
+      "epoch": 0.2542372881355932,
+      "grad_norm": 0.020665116807158542,
+      "learning_rate": 8.287862513426424e-06,
+      "loss": 0.0001,
+      "step": 3945
+    },
+    {
+      "epoch": 0.2543017335825224,
+      "grad_norm": 0.1538017089756088,
+      "learning_rate": 8.287146437522378e-06,
+      "loss": 0.0014,
+      "step": 3946
+    },
+    {
+      "epoch": 0.2543661790294516,
+      "grad_norm": 4.183620420739566,
+      "learning_rate": 8.286430361618331e-06,
+      "loss": 0.0425,
+      "step": 3947
+    },
+    {
+      "epoch": 0.25443062447638076,
+      "grad_norm": 0.2073996343380915,
+      "learning_rate": 8.285714285714287e-06,
+      "loss": 0.0004,
+      "step": 3948
+    },
+    {
+      "epoch": 0.2544950699233099,
+      "grad_norm": 0.020734861291480815,
+      "learning_rate": 8.284998209810241e-06,
+      "loss": 0.0016,
+      "step": 3949
+    },
+    {
+      "epoch": 0.2545595153702391,
+      "grad_norm": 0.0010731652208121033,
+      "learning_rate": 8.284282133906196e-06,
+      "loss": 0.0,
+      "step": 3950
+    },
+    {
+      "epoch": 0.25462396081716826,
+      "grad_norm": 0.0006424096037444866,
+      "learning_rate": 8.28356605800215e-06,
+      "loss": 0.0,
+      "step": 3951
+    },
+    {
+      "epoch": 0.25468840626409744,
+      "grad_norm": 0.003079481410634368,
+      "learning_rate": 8.282849982098104e-06,
+      "loss": 0.0,
+      "step": 3952
+    },
+    {
+      "epoch": 0.25475285171102663,
+      "grad_norm": 0.23498132479301437,
+      "learning_rate": 8.282133906194057e-06,
+      "loss": 0.0006,
+      "step": 3953
+    },
+    {
+      "epoch": 0.2548172971579558,
+      "grad_norm": 0.013251046524471814,
+      "learning_rate": 8.281417830290011e-06,
+      "loss": 0.0,
+      "step": 3954
+    },
+    {
+      "epoch": 0.25488174260488494,
+      "grad_norm": 0.005255771405272541,
+      "learning_rate": 8.280701754385965e-06,
+      "loss": 0.0,
+      "step": 3955
+    },
+    {
+      "epoch": 0.2549461880518141,
+      "grad_norm": 0.005244493955979896,
+      "learning_rate": 8.27998567848192e-06,
+      "loss": 0.0,
+      "step": 3956
+    },
+    {
+      "epoch": 0.2550106334987433,
+      "grad_norm": 0.0006158653952264748,
+      "learning_rate": 8.279269602577874e-06,
+      "loss": 0.0,
+      "step": 3957
+    },
+    {
+      "epoch": 0.2550750789456725,
+      "grad_norm": 0.00035900625196237423,
+      "learning_rate": 8.278553526673828e-06,
+      "loss": 0.0,
+      "step": 3958
+    },
+    {
+      "epoch": 0.2551395243926017,
+      "grad_norm": 0.008736003219196615,
+      "learning_rate": 8.277837450769783e-06,
+      "loss": 0.0001,
+      "step": 3959
+    },
+    {
+      "epoch": 0.25520396983953086,
+      "grad_norm": 0.0017198369442972868,
+      "learning_rate": 8.277121374865737e-06,
+      "loss": 0.0,
+      "step": 3960
+    },
+    {
+      "epoch": 0.25526841528646,
+      "grad_norm": 0.000357485762814059,
+      "learning_rate": 8.276405298961691e-06,
+      "loss": 0.0,
+      "step": 3961
+    },
+    {
+      "epoch": 0.2553328607333892,
+      "grad_norm": 0.004416114733046556,
+      "learning_rate": 8.275689223057645e-06,
+      "loss": 0.0,
+      "step": 3962
+    },
+    {
+      "epoch": 0.25539730618031836,
+      "grad_norm": 0.05105998267534113,
+      "learning_rate": 8.274973147153598e-06,
+      "loss": 0.0001,
+      "step": 3963
+    },
+    {
+      "epoch": 0.25546175162724755,
+      "grad_norm": 2.68921207630397,
+      "learning_rate": 8.274257071249552e-06,
+      "loss": 0.0131,
+      "step": 3964
+    },
+    {
+      "epoch": 0.25552619707417673,
+      "grad_norm": 0.0021671550621980688,
+      "learning_rate": 8.273540995345507e-06,
+      "loss": 0.0,
+      "step": 3965
+    },
+    {
+      "epoch": 0.25559064252110586,
+      "grad_norm": 0.047147107595446595,
+      "learning_rate": 8.272824919441461e-06,
+      "loss": 0.0004,
+      "step": 3966
+    },
+    {
+      "epoch": 0.25565508796803504,
+      "grad_norm": 0.0007389136439259691,
+      "learning_rate": 8.272108843537415e-06,
+      "loss": 0.0,
+      "step": 3967
+    },
+    {
+      "epoch": 0.25571953341496423,
+      "grad_norm": 0.0011897303535312082,
+      "learning_rate": 8.27139276763337e-06,
+      "loss": 0.0,
+      "step": 3968
+    },
+    {
+      "epoch": 0.2557839788618934,
+      "grad_norm": 0.0008199297088428887,
+      "learning_rate": 8.270676691729324e-06,
+      "loss": 0.0,
+      "step": 3969
+    },
+    {
+      "epoch": 0.2558484243088226,
+      "grad_norm": 0.526262460456898,
+      "learning_rate": 8.269960615825278e-06,
+      "loss": 0.0039,
+      "step": 3970
+    },
+    {
+      "epoch": 0.2559128697557518,
+      "grad_norm": 0.005483051269792115,
+      "learning_rate": 8.269244539921232e-06,
+      "loss": 0.0,
+      "step": 3971
+    },
+    {
+      "epoch": 0.2559773152026809,
+      "grad_norm": 0.020236496196260945,
+      "learning_rate": 8.268528464017187e-06,
+      "loss": 0.0,
+      "step": 3972
+    },
+    {
+      "epoch": 0.2560417606496101,
+      "grad_norm": 0.22438319054480052,
+      "learning_rate": 8.267812388113141e-06,
+      "loss": 0.0021,
+      "step": 3973
+    },
+    {
+      "epoch": 0.2561062060965393,
+      "grad_norm": 0.1771382826278181,
+      "learning_rate": 8.267096312209095e-06,
+      "loss": 0.0002,
+      "step": 3974
+    },
+    {
+      "epoch": 0.25617065154346846,
+      "grad_norm": 0.047601454923286886,
+      "learning_rate": 8.26638023630505e-06,
+      "loss": 0.0001,
+      "step": 3975
+    },
+    {
+      "epoch": 0.25623509699039765,
+      "grad_norm": 0.06528823065675737,
+      "learning_rate": 8.265664160401004e-06,
+      "loss": 0.0001,
+      "step": 3976
+    },
+    {
+      "epoch": 0.2562995424373268,
+      "grad_norm": 0.021962709957251546,
+      "learning_rate": 8.264948084496958e-06,
+      "loss": 0.0,
+      "step": 3977
+    },
+    {
+      "epoch": 0.25636398788425596,
+      "grad_norm": 0.011551023631307096,
+      "learning_rate": 8.264232008592913e-06,
+      "loss": 0.0,
+      "step": 3978
+    },
+    {
+      "epoch": 0.25642843333118515,
+      "grad_norm": 0.0026298561406031358,
+      "learning_rate": 8.263515932688865e-06,
+      "loss": 0.0,
+      "step": 3979
+    },
+    {
+      "epoch": 0.25649287877811433,
+      "grad_norm": 0.002949573749046438,
+      "learning_rate": 8.26279985678482e-06,
+      "loss": 0.0,
+      "step": 3980
+    },
+    {
+      "epoch": 0.2565573242250435,
+      "grad_norm": 0.0030800614041981695,
+      "learning_rate": 8.262083780880774e-06,
+      "loss": 0.0,
+      "step": 3981
+    },
+    {
+      "epoch": 0.2566217696719727,
+      "grad_norm": 0.02935547035373371,
+      "learning_rate": 8.261367704976728e-06,
+      "loss": 0.0003,
+      "step": 3982
+    },
+    {
+      "epoch": 0.25668621511890183,
+      "grad_norm": 0.018741440340363177,
+      "learning_rate": 8.260651629072682e-06,
+      "loss": 0.0001,
+      "step": 3983
+    },
+    {
+      "epoch": 0.256750660565831,
+      "grad_norm": 0.002872186655947733,
+      "learning_rate": 8.259935553168637e-06,
+      "loss": 0.0,
+      "step": 3984
+    },
+    {
+      "epoch": 0.2568151060127602,
+      "grad_norm": 0.054660745871351675,
+      "learning_rate": 8.259219477264591e-06,
+      "loss": 0.0002,
+      "step": 3985
+    },
+    {
+      "epoch": 0.2568795514596894,
+      "grad_norm": 0.0025098992277484064,
+      "learning_rate": 8.258503401360545e-06,
+      "loss": 0.0,
+      "step": 3986
+    },
+    {
+      "epoch": 0.25694399690661857,
+      "grad_norm": 0.11741832015214243,
+      "learning_rate": 8.2577873254565e-06,
+      "loss": 0.0006,
+      "step": 3987
+    },
+    {
+      "epoch": 0.2570084423535477,
+      "grad_norm": 0.008282736051549501,
+      "learning_rate": 8.257071249552452e-06,
+      "loss": 0.0,
+      "step": 3988
+    },
+    {
+      "epoch": 0.2570728878004769,
+      "grad_norm": 0.05705696246490212,
+      "learning_rate": 8.256355173648406e-06,
+      "loss": 0.0001,
+      "step": 3989
+    },
+    {
+      "epoch": 0.25713733324740606,
+      "grad_norm": 0.0049394262891318905,
+      "learning_rate": 8.25563909774436e-06,
+      "loss": 0.0,
+      "step": 3990
+    },
+    {
+      "epoch": 0.25720177869433525,
+      "grad_norm": 0.04868299851585092,
+      "learning_rate": 8.254923021840315e-06,
+      "loss": 0.0001,
+      "step": 3991
+    },
+    {
+      "epoch": 0.25726622414126443,
+      "grad_norm": 0.0073026446153531,
+      "learning_rate": 8.25420694593627e-06,
+      "loss": 0.0,
+      "step": 3992
+    },
+    {
+      "epoch": 0.2573306695881936,
+      "grad_norm": 0.006003148253976413,
+      "learning_rate": 8.253490870032224e-06,
+      "loss": 0.0,
+      "step": 3993
+    },
+    {
+      "epoch": 0.25739511503512275,
+      "grad_norm": 0.0030958094222771513,
+      "learning_rate": 8.252774794128178e-06,
+      "loss": 0.0,
+      "step": 3994
+    },
+    {
+      "epoch": 0.25745956048205193,
+      "grad_norm": 0.005458472731249199,
+      "learning_rate": 8.252058718224132e-06,
+      "loss": 0.0,
+      "step": 3995
+    },
+    {
+      "epoch": 0.2575240059289811,
+      "grad_norm": 0.00035485698227203874,
+      "learning_rate": 8.251342642320087e-06,
+      "loss": 0.0,
+      "step": 3996
+    },
+    {
+      "epoch": 0.2575884513759103,
+      "grad_norm": 0.00017725530470848552,
+      "learning_rate": 8.25062656641604e-06,
+      "loss": 0.0,
+      "step": 3997
+    },
+    {
+      "epoch": 0.2576528968228395,
+      "grad_norm": 0.00317466520196752,
+      "learning_rate": 8.249910490511995e-06,
+      "loss": 0.0,
+      "step": 3998
+    },
+    {
+      "epoch": 0.25771734226976867,
+      "grad_norm": 0.012355914561483782,
+      "learning_rate": 8.24919441460795e-06,
+      "loss": 0.0001,
+      "step": 3999
+    },
+    {
+      "epoch": 0.2577817877166978,
+      "grad_norm": 0.3189945787922834,
+      "learning_rate": 8.248478338703904e-06,
+      "loss": 0.0054,
+      "step": 4000
+    },
+    {
+      "epoch": 0.257846233163627,
+      "grad_norm": 0.0002133263309856516,
+      "learning_rate": 8.247762262799858e-06,
+      "loss": 0.0,
+      "step": 4001
+    },
+    {
+      "epoch": 0.25791067861055617,
+      "grad_norm": 0.005477701385231475,
+      "learning_rate": 8.247046186895812e-06,
+      "loss": 0.0,
+      "step": 4002
+    },
+    {
+      "epoch": 0.25797512405748535,
+      "grad_norm": 0.0050049522296231825,
+      "learning_rate": 8.246330110991767e-06,
+      "loss": 0.0,
+      "step": 4003
+    },
+    {
+      "epoch": 0.25803956950441453,
+      "grad_norm": 0.12513390461850646,
+      "learning_rate": 8.24561403508772e-06,
+      "loss": 0.0008,
+      "step": 4004
+    },
+    {
+      "epoch": 0.25810401495134366,
+      "grad_norm": 0.387589101391816,
+      "learning_rate": 8.244897959183674e-06,
+      "loss": 0.0008,
+      "step": 4005
+    },
+    {
+      "epoch": 0.25816846039827285,
+      "grad_norm": 0.01585431965727312,
+      "learning_rate": 8.244181883279628e-06,
+      "loss": 0.0,
+      "step": 4006
+    },
+    {
+      "epoch": 0.25823290584520203,
+      "grad_norm": 0.00618341276123621,
+      "learning_rate": 8.243465807375582e-06,
+      "loss": 0.0,
+      "step": 4007
+    },
+    {
+      "epoch": 0.2582973512921312,
+      "grad_norm": 0.0026786699895131457,
+      "learning_rate": 8.242749731471536e-06,
+      "loss": 0.0,
+      "step": 4008
+    },
+    {
+      "epoch": 0.2583617967390604,
+      "grad_norm": 0.0036054754536920027,
+      "learning_rate": 8.24203365556749e-06,
+      "loss": 0.0,
+      "step": 4009
+    },
+    {
+      "epoch": 0.2584262421859896,
+      "grad_norm": 0.0007546240302236562,
+      "learning_rate": 8.241317579663445e-06,
+      "loss": 0.0,
+      "step": 4010
+    },
+    {
+      "epoch": 0.2584906876329187,
+      "grad_norm": 0.012423956570293417,
+      "learning_rate": 8.2406015037594e-06,
+      "loss": 0.0001,
+      "step": 4011
+    },
+    {
+      "epoch": 0.2585551330798479,
+      "grad_norm": 0.003081257517508017,
+      "learning_rate": 8.239885427855354e-06,
+      "loss": 0.0,
+      "step": 4012
+    },
+    {
+      "epoch": 0.2586195785267771,
+      "grad_norm": 0.3443340411812879,
+      "learning_rate": 8.239169351951308e-06,
+      "loss": 0.0029,
+      "step": 4013
+    },
+    {
+      "epoch": 0.25868402397370627,
+      "grad_norm": 0.025767518490395797,
+      "learning_rate": 8.23845327604726e-06,
+      "loss": 0.0002,
+      "step": 4014
+    },
+    {
+      "epoch": 0.25874846942063545,
+      "grad_norm": 0.048692089363689406,
+      "learning_rate": 8.237737200143215e-06,
+      "loss": 0.0001,
+      "step": 4015
+    },
+    {
+      "epoch": 0.2588129148675646,
+      "grad_norm": 0.9627859124434754,
+      "learning_rate": 8.237021124239169e-06,
+      "loss": 0.0037,
+      "step": 4016
+    },
+    {
+      "epoch": 0.25887736031449377,
+      "grad_norm": 0.05837690042372548,
+      "learning_rate": 8.236305048335123e-06,
+      "loss": 0.0002,
+      "step": 4017
+    },
+    {
+      "epoch": 0.25894180576142295,
+      "grad_norm": 0.004356229156364595,
+      "learning_rate": 8.23558897243108e-06,
+      "loss": 0.0,
+      "step": 4018
+    },
+    {
+      "epoch": 0.25900625120835213,
+      "grad_norm": 0.03680381490022786,
+      "learning_rate": 8.234872896527034e-06,
+      "loss": 0.0,
+      "step": 4019
+    },
+    {
+      "epoch": 0.2590706966552813,
+      "grad_norm": 0.05317580278091146,
+      "learning_rate": 8.234156820622986e-06,
+      "loss": 0.0017,
+      "step": 4020
+    },
+    {
+      "epoch": 0.2591351421022105,
+      "grad_norm": 0.0026499589169380762,
+      "learning_rate": 8.23344074471894e-06,
+      "loss": 0.0,
+      "step": 4021
+    },
+    {
+      "epoch": 0.25919958754913963,
+      "grad_norm": 0.06186301694838707,
+      "learning_rate": 8.232724668814895e-06,
+      "loss": 0.0007,
+      "step": 4022
+    },
+    {
+      "epoch": 0.2592640329960688,
+      "grad_norm": 0.008444060526545723,
+      "learning_rate": 8.23200859291085e-06,
+      "loss": 0.0,
+      "step": 4023
+    },
+    {
+      "epoch": 0.259328478442998,
+      "grad_norm": 0.0017911749821692064,
+      "learning_rate": 8.231292517006804e-06,
+      "loss": 0.0,
+      "step": 4024
+    },
+    {
+      "epoch": 0.2593929238899272,
+      "grad_norm": 2.548784534768542,
+      "learning_rate": 8.230576441102758e-06,
+      "loss": 0.0384,
+      "step": 4025
+    },
+    {
+      "epoch": 0.25945736933685637,
+      "grad_norm": 2.548784534768542,
+      "learning_rate": 8.230576441102758e-06,
+      "loss": 0.0226,
+      "step": 4026
+    },
+    {
+      "epoch": 0.2595218147837855,
+      "grad_norm": 0.005779961270392248,
+      "learning_rate": 8.229860365198712e-06,
+      "loss": 0.0,
+      "step": 4027
+    },
+    {
+      "epoch": 0.2595862602307147,
+      "grad_norm": 0.0009377977132422742,
+      "learning_rate": 8.229144289294666e-06,
+      "loss": 0.0,
+      "step": 4028
+    },
+    {
+      "epoch": 0.25965070567764387,
+      "grad_norm": 0.03774370585557605,
+      "learning_rate": 8.22842821339062e-06,
+      "loss": 0.0,
+      "step": 4029
+    },
+    {
+      "epoch": 0.25971515112457305,
+      "grad_norm": 0.005797201910791572,
+      "learning_rate": 8.227712137486575e-06,
+      "loss": 0.0,
+      "step": 4030
+    },
+    {
+      "epoch": 0.25977959657150224,
+      "grad_norm": 0.06964904980649746,
+      "learning_rate": 8.226996061582528e-06,
+      "loss": 0.0015,
+      "step": 4031
+    },
+    {
+      "epoch": 0.2598440420184314,
+      "grad_norm": 0.060964751645412936,
+      "learning_rate": 8.226279985678482e-06,
+      "loss": 0.0002,
+      "step": 4032
+    },
+    {
+      "epoch": 0.25990848746536055,
+      "grad_norm": 0.060964751645412936,
+      "learning_rate": 8.226279985678482e-06,
+      "loss": 0.0187,
+      "step": 4033
+    },
+    {
+      "epoch": 0.25997293291228973,
+      "grad_norm": 0.060525086543292334,
+      "learning_rate": 8.225563909774436e-06,
+      "loss": 0.0002,
+      "step": 4034
+    },
+    {
+      "epoch": 0.2600373783592189,
+      "grad_norm": 0.19072903311402004,
+      "learning_rate": 8.22484783387039e-06,
+      "loss": 0.0002,
+      "step": 4035
+    },
+    {
+      "epoch": 0.2601018238061481,
+      "grad_norm": 0.031776800989206204,
+      "learning_rate": 8.224131757966345e-06,
+      "loss": 0.0,
+      "step": 4036
+    },
+    {
+      "epoch": 0.2601662692530773,
+      "grad_norm": 0.08727196607202162,
+      "learning_rate": 8.223415682062299e-06,
+      "loss": 0.0002,
+      "step": 4037
+    },
+    {
+      "epoch": 0.26023071470000647,
+      "grad_norm": 0.08446679894310626,
+      "learning_rate": 8.222699606158253e-06,
+      "loss": 0.0002,
+      "step": 4038
+    },
+    {
+      "epoch": 0.2602951601469356,
+      "grad_norm": 0.002980232543945297,
+      "learning_rate": 8.221983530254208e-06,
+      "loss": 0.0,
+      "step": 4039
+    },
+    {
+      "epoch": 0.2603596055938648,
+      "grad_norm": 0.0011543615042102016,
+      "learning_rate": 8.221267454350162e-06,
+      "loss": 0.0,
+      "step": 4040
+    },
+    {
+      "epoch": 0.26042405104079397,
+      "grad_norm": 0.41192820033215677,
+      "learning_rate": 8.220551378446116e-06,
+      "loss": 0.0012,
+      "step": 4041
+    },
+    {
+      "epoch": 0.26048849648772315,
+      "grad_norm": 0.0010378573531816071,
+      "learning_rate": 8.219835302542069e-06,
+      "loss": 0.0,
+      "step": 4042
+    },
+    {
+      "epoch": 0.26055294193465234,
+      "grad_norm": 0.0006749139181435023,
+      "learning_rate": 8.219119226638023e-06,
+      "loss": 0.0,
+      "step": 4043
+    },
+    {
+      "epoch": 0.26061738738158147,
+      "grad_norm": 0.07496953795127487,
+      "learning_rate": 8.21840315073398e-06,
+      "loss": 0.0002,
+      "step": 4044
+    },
+    {
+      "epoch": 0.26068183282851065,
+      "grad_norm": 0.004533902303890554,
+      "learning_rate": 8.217687074829933e-06,
+      "loss": 0.0,
+      "step": 4045
+    },
+    {
+      "epoch": 0.26074627827543984,
+      "grad_norm": 0.0896993077211288,
+      "learning_rate": 8.216970998925888e-06,
+      "loss": 0.0001,
+      "step": 4046
+    },
+    {
+      "epoch": 0.260810723722369,
+      "grad_norm": 0.0005955184982546869,
+      "learning_rate": 8.216254923021842e-06,
+      "loss": 0.0,
+      "step": 4047
+    },
+    {
+      "epoch": 0.2608751691692982,
+      "grad_norm": 0.004027601831837156,
+      "learning_rate": 8.215538847117795e-06,
+      "loss": 0.0,
+      "step": 4048
+    },
+    {
+      "epoch": 0.2609396146162274,
+      "grad_norm": 0.0069240323019456535,
+      "learning_rate": 8.214822771213749e-06,
+      "loss": 0.0001,
+      "step": 4049
+    },
+    {
+      "epoch": 0.2610040600631565,
+      "grad_norm": 0.0008218871824397413,
+      "learning_rate": 8.214106695309703e-06,
+      "loss": 0.0,
+      "step": 4050
+    },
+    {
+      "epoch": 0.2610685055100857,
+      "grad_norm": 0.0008479160595231805,
+      "learning_rate": 8.213390619405658e-06,
+      "loss": 0.0,
+      "step": 4051
+    },
+    {
+      "epoch": 0.2611329509570149,
+      "grad_norm": 0.1893188846496353,
+      "learning_rate": 8.212674543501612e-06,
+      "loss": 0.0006,
+      "step": 4052
+    },
+    {
+      "epoch": 0.26119739640394407,
+      "grad_norm": 0.015341010774691261,
+      "learning_rate": 8.211958467597566e-06,
+      "loss": 0.0001,
+      "step": 4053
+    },
+    {
+      "epoch": 0.26126184185087326,
+      "grad_norm": 0.0028778119216130286,
+      "learning_rate": 8.21124239169352e-06,
+      "loss": 0.0,
+      "step": 4054
+    },
+    {
+      "epoch": 0.2613262872978024,
+      "grad_norm": 0.07925056622743644,
+      "learning_rate": 8.210526315789475e-06,
+      "loss": 0.0009,
+      "step": 4055
+    },
+    {
+      "epoch": 0.26139073274473157,
+      "grad_norm": 0.20684729806969399,
+      "learning_rate": 8.209810239885429e-06,
+      "loss": 0.0043,
+      "step": 4056
+    },
+    {
+      "epoch": 0.26145517819166075,
+      "grad_norm": 0.0006323088838524224,
+      "learning_rate": 8.209094163981383e-06,
+      "loss": 0.0,
+      "step": 4057
+    },
+    {
+      "epoch": 0.26151962363858994,
+      "grad_norm": 0.0011074759344691717,
+      "learning_rate": 8.208378088077336e-06,
+      "loss": 0.0,
+      "step": 4058
+    },
+    {
+      "epoch": 0.2615840690855191,
+      "grad_norm": 0.00026941518339601263,
+      "learning_rate": 8.20766201217329e-06,
+      "loss": 0.0,
+      "step": 4059
+    },
+    {
+      "epoch": 0.2616485145324483,
+      "grad_norm": 0.04375307634300663,
+      "learning_rate": 8.206945936269245e-06,
+      "loss": 0.0001,
+      "step": 4060
+    },
+    {
+      "epoch": 0.26171295997937744,
+      "grad_norm": 0.05213391882349801,
+      "learning_rate": 8.206229860365199e-06,
+      "loss": 0.0001,
+      "step": 4061
+    },
+    {
+      "epoch": 0.2617774054263066,
+      "grad_norm": 1.516026237311075,
+      "learning_rate": 8.205513784461153e-06,
+      "loss": 0.0053,
+      "step": 4062
+    },
+    {
+      "epoch": 0.2618418508732358,
+      "grad_norm": 0.27357712314436866,
+      "learning_rate": 8.204797708557107e-06,
+      "loss": 0.0021,
+      "step": 4063
+    },
+    {
+      "epoch": 0.261906296320165,
+      "grad_norm": 0.005267339834329356,
+      "learning_rate": 8.204081632653062e-06,
+      "loss": 0.0,
+      "step": 4064
+    },
+    {
+      "epoch": 0.2619707417670942,
+      "grad_norm": 0.006003939047317733,
+      "learning_rate": 8.203365556749016e-06,
+      "loss": 0.0,
+      "step": 4065
+    },
+    {
+      "epoch": 0.2620351872140233,
+      "grad_norm": 0.006873744433309013,
+      "learning_rate": 8.20264948084497e-06,
+      "loss": 0.0,
+      "step": 4066
+    },
+    {
+      "epoch": 0.2620996326609525,
+      "grad_norm": 0.0010702078088623004,
+      "learning_rate": 8.201933404940925e-06,
+      "loss": 0.0,
+      "step": 4067
+    },
+    {
+      "epoch": 0.26216407810788167,
+      "grad_norm": 0.00016045440845831096,
+      "learning_rate": 8.201217329036879e-06,
+      "loss": 0.0,
+      "step": 4068
+    },
+    {
+      "epoch": 0.26222852355481086,
+      "grad_norm": 0.33688526859804885,
+      "learning_rate": 8.200501253132833e-06,
+      "loss": 0.0033,
+      "step": 4069
+    },
+    {
+      "epoch": 0.26229296900174004,
+      "grad_norm": 0.005965846264202467,
+      "learning_rate": 8.199785177228788e-06,
+      "loss": 0.0,
+      "step": 4070
+    },
+    {
+      "epoch": 0.2623574144486692,
+      "grad_norm": 1.2753300557497391,
+      "learning_rate": 8.199069101324742e-06,
+      "loss": 0.007,
+      "step": 4071
+    },
+    {
+      "epoch": 0.26242185989559835,
+      "grad_norm": 0.42094141481840447,
+      "learning_rate": 8.198353025420696e-06,
+      "loss": 0.0034,
+      "step": 4072
+    },
+    {
+      "epoch": 0.26248630534252754,
+      "grad_norm": 0.0007287004405519097,
+      "learning_rate": 8.19763694951665e-06,
+      "loss": 0.0,
+      "step": 4073
+    },
+    {
+      "epoch": 0.2625507507894567,
+      "grad_norm": 0.00040297765390191563,
+      "learning_rate": 8.196920873612603e-06,
+      "loss": 0.0,
+      "step": 4074
+    },
+    {
+      "epoch": 0.2626151962363859,
+      "grad_norm": 0.013456809470448914,
+      "learning_rate": 8.196204797708557e-06,
+      "loss": 0.0,
+      "step": 4075
+    },
+    {
+      "epoch": 0.2626796416833151,
+      "grad_norm": 0.37379560179625704,
+      "learning_rate": 8.195488721804512e-06,
+      "loss": 0.0044,
+      "step": 4076
+    },
+    {
+      "epoch": 0.2627440871302443,
+      "grad_norm": 0.19677700298270354,
+      "learning_rate": 8.194772645900466e-06,
+      "loss": 0.0001,
+      "step": 4077
+    },
+    {
+      "epoch": 0.2628085325771734,
+      "grad_norm": 0.4003689829862226,
+      "learning_rate": 8.19405656999642e-06,
+      "loss": 0.0004,
+      "step": 4078
+    },
+    {
+      "epoch": 0.2628729780241026,
+      "grad_norm": 0.004796831837730866,
+      "learning_rate": 8.193340494092375e-06,
+      "loss": 0.0001,
+      "step": 4079
+    },
+    {
+      "epoch": 0.2629374234710318,
+      "grad_norm": 0.0019953814802931044,
+      "learning_rate": 8.192624418188329e-06,
+      "loss": 0.0,
+      "step": 4080
+    },
+    {
+      "epoch": 0.26300186891796096,
+      "grad_norm": 0.08354990881299791,
+      "learning_rate": 8.191908342284283e-06,
+      "loss": 0.0017,
+      "step": 4081
+    },
+    {
+      "epoch": 0.26306631436489014,
+      "grad_norm": 0.022514688404505278,
+      "learning_rate": 8.191192266380237e-06,
+      "loss": 0.0002,
+      "step": 4082
+    },
+    {
+      "epoch": 0.26313075981181927,
+      "grad_norm": 0.029840040645768735,
+      "learning_rate": 8.190476190476192e-06,
+      "loss": 0.0001,
+      "step": 4083
+    },
+    {
+      "epoch": 0.26319520525874845,
+      "grad_norm": 0.0012259527385874386,
+      "learning_rate": 8.189760114572144e-06,
+      "loss": 0.0,
+      "step": 4084
+    },
+    {
+      "epoch": 0.26325965070567764,
+      "grad_norm": 0.0016036067659748598,
+      "learning_rate": 8.189044038668099e-06,
+      "loss": 0.0,
+      "step": 4085
+    },
+    {
+      "epoch": 0.2633240961526068,
+      "grad_norm": 0.00031105862440330756,
+      "learning_rate": 8.188327962764053e-06,
+      "loss": 0.0,
+      "step": 4086
+    },
+    {
+      "epoch": 0.263388541599536,
+      "grad_norm": 0.01397653280989205,
+      "learning_rate": 8.187611886860007e-06,
+      "loss": 0.0001,
+      "step": 4087
+    },
+    {
+      "epoch": 0.2634529870464652,
+      "grad_norm": 0.001967928846776797,
+      "learning_rate": 8.186895810955962e-06,
+      "loss": 0.0,
+      "step": 4088
+    },
+    {
+      "epoch": 0.2635174324933943,
+      "grad_norm": 0.00274836594875593,
+      "learning_rate": 8.186179735051916e-06,
+      "loss": 0.0,
+      "step": 4089
+    },
+    {
+      "epoch": 0.2635818779403235,
+      "grad_norm": 0.0032691538261679083,
+      "learning_rate": 8.18546365914787e-06,
+      "loss": 0.0,
+      "step": 4090
+    },
+    {
+      "epoch": 0.2636463233872527,
+      "grad_norm": 0.14504496675654013,
+      "learning_rate": 8.184747583243824e-06,
+      "loss": 0.0013,
+      "step": 4091
+    },
+    {
+      "epoch": 0.2637107688341819,
+      "grad_norm": 0.5235929400677679,
+      "learning_rate": 8.184031507339779e-06,
+      "loss": 0.0018,
+      "step": 4092
+    },
+    {
+      "epoch": 0.26377521428111106,
+      "grad_norm": 0.0067012557335894415,
+      "learning_rate": 8.183315431435733e-06,
+      "loss": 0.0,
+      "step": 4093
+    },
+    {
+      "epoch": 0.2638396597280402,
+      "grad_norm": 0.6293055528436677,
+      "learning_rate": 8.182599355531687e-06,
+      "loss": 0.0015,
+      "step": 4094
+    },
+    {
+      "epoch": 0.2639041051749694,
+      "grad_norm": 0.01867268633251904,
+      "learning_rate": 8.181883279627642e-06,
+      "loss": 0.0001,
+      "step": 4095
+    },
+    {
+      "epoch": 0.26396855062189856,
+      "grad_norm": 0.24328595672003867,
+      "learning_rate": 8.181167203723596e-06,
+      "loss": 0.0001,
+      "step": 4096
+    },
+    {
+      "epoch": 0.26403299606882774,
+      "grad_norm": 0.0032203777060472484,
+      "learning_rate": 8.18045112781955e-06,
+      "loss": 0.0,
+      "step": 4097
+    },
+    {
+      "epoch": 0.2640974415157569,
+      "grad_norm": 0.06334367560512613,
+      "learning_rate": 8.179735051915505e-06,
+      "loss": 0.0001,
+      "step": 4098
+    },
+    {
+      "epoch": 0.2641618869626861,
+      "grad_norm": 0.0012013743522975488,
+      "learning_rate": 8.179018976011459e-06,
+      "loss": 0.0,
+      "step": 4099
+    },
+    {
+      "epoch": 0.26422633240961524,
+      "grad_norm": 0.0037318730977894373,
+      "learning_rate": 8.178302900107411e-06,
+      "loss": 0.0,
+      "step": 4100
+    },
+    {
+      "epoch": 0.2642907778565444,
+      "grad_norm": 0.00883380236499844,
+      "learning_rate": 8.177586824203366e-06,
+      "loss": 0.0001,
+      "step": 4101
+    },
+    {
+      "epoch": 0.2643552233034736,
+      "grad_norm": 0.0044728075765404236,
+      "learning_rate": 8.17687074829932e-06,
+      "loss": 0.0,
+      "step": 4102
+    },
+    {
+      "epoch": 0.2644196687504028,
+      "grad_norm": 0.0070594660207348395,
+      "learning_rate": 8.176154672395274e-06,
+      "loss": 0.0001,
+      "step": 4103
+    },
+    {
+      "epoch": 0.264484114197332,
+      "grad_norm": 0.10401460655221516,
+      "learning_rate": 8.175438596491229e-06,
+      "loss": 0.0001,
+      "step": 4104
+    },
+    {
+      "epoch": 0.2645485596442611,
+      "grad_norm": 0.03844934332167701,
+      "learning_rate": 8.174722520587183e-06,
+      "loss": 0.0002,
+      "step": 4105
+    },
+    {
+      "epoch": 0.2646130050911903,
+      "grad_norm": 1.0716477578622337,
+      "learning_rate": 8.174006444683137e-06,
+      "loss": 0.0032,
+      "step": 4106
+    },
+    {
+      "epoch": 0.2646774505381195,
+      "grad_norm": 1.3784877152724,
+      "learning_rate": 8.173290368779092e-06,
+      "loss": 0.0164,
+      "step": 4107
+    },
+    {
+      "epoch": 0.26474189598504866,
+      "grad_norm": 0.029587838605416625,
+      "learning_rate": 8.172574292875046e-06,
+      "loss": 0.0001,
+      "step": 4108
+    },
+    {
+      "epoch": 0.26480634143197784,
+      "grad_norm": 0.18620288726645495,
+      "learning_rate": 8.171858216970998e-06,
+      "loss": 0.0019,
+      "step": 4109
+    },
+    {
+      "epoch": 0.264870786878907,
+      "grad_norm": 0.0058924889771335845,
+      "learning_rate": 8.171142141066953e-06,
+      "loss": 0.0,
+      "step": 4110
+    },
+    {
+      "epoch": 0.26493523232583616,
+      "grad_norm": 0.025689142320420872,
+      "learning_rate": 8.170426065162907e-06,
+      "loss": 0.0001,
+      "step": 4111
+    },
+    {
+      "epoch": 0.26499967777276534,
+      "grad_norm": 0.007680422605328349,
+      "learning_rate": 8.169709989258861e-06,
+      "loss": 0.0,
+      "step": 4112
+    },
+    {
+      "epoch": 0.2650641232196945,
+      "grad_norm": 0.025682199960681367,
+      "learning_rate": 8.168993913354816e-06,
+      "loss": 0.0001,
+      "step": 4113
+    },
+    {
+      "epoch": 0.2651285686666237,
+      "grad_norm": 0.6048429639215468,
+      "learning_rate": 8.168277837450772e-06,
+      "loss": 0.0029,
+      "step": 4114
+    },
+    {
+      "epoch": 0.2651930141135529,
+      "grad_norm": 0.21238982942042506,
+      "learning_rate": 8.167561761546726e-06,
+      "loss": 0.0002,
+      "step": 4115
+    },
+    {
+      "epoch": 0.2652574595604821,
+      "grad_norm": 0.1002031654121083,
+      "learning_rate": 8.166845685642679e-06,
+      "loss": 0.0001,
+      "step": 4116
+    },
+    {
+      "epoch": 0.2653219050074112,
+      "grad_norm": 0.07676032076806907,
+      "learning_rate": 8.166129609738633e-06,
+      "loss": 0.0,
+      "step": 4117
+    },
+    {
+      "epoch": 0.2653863504543404,
+      "grad_norm": 0.0556255801510828,
+      "learning_rate": 8.165413533834587e-06,
+      "loss": 0.0001,
+      "step": 4118
+    },
+    {
+      "epoch": 0.2654507959012696,
+      "grad_norm": 0.38215803137244037,
+      "learning_rate": 8.164697457930541e-06,
+      "loss": 0.0012,
+      "step": 4119
+    },
+    {
+      "epoch": 0.26551524134819876,
+      "grad_norm": 0.367995408179826,
+      "learning_rate": 8.163981382026496e-06,
+      "loss": 0.0004,
+      "step": 4120
+    },
+    {
+      "epoch": 0.26557968679512794,
+      "grad_norm": 0.12432520116844026,
+      "learning_rate": 8.16326530612245e-06,
+      "loss": 0.0002,
+      "step": 4121
+    },
+    {
+      "epoch": 0.2656441322420571,
+      "grad_norm": 0.40048772983471204,
+      "learning_rate": 8.162549230218404e-06,
+      "loss": 0.0012,
+      "step": 4122
+    },
+    {
+      "epoch": 0.26570857768898626,
+      "grad_norm": 0.028347294384729757,
+      "learning_rate": 8.161833154314359e-06,
+      "loss": 0.0001,
+      "step": 4123
+    },
+    {
+      "epoch": 0.26577302313591544,
+      "grad_norm": 0.001553801958047865,
+      "learning_rate": 8.161117078410313e-06,
+      "loss": 0.0,
+      "step": 4124
+    },
+    {
+      "epoch": 0.2658374685828446,
+      "grad_norm": 0.0048637148912617,
+      "learning_rate": 8.160401002506266e-06,
+      "loss": 0.0,
+      "step": 4125
+    },
+    {
+      "epoch": 0.2659019140297738,
+      "grad_norm": 0.06001730600695284,
+      "learning_rate": 8.15968492660222e-06,
+      "loss": 0.0004,
+      "step": 4126
+    },
+    {
+      "epoch": 0.265966359476703,
+      "grad_norm": 0.002554106564452214,
+      "learning_rate": 8.158968850698174e-06,
+      "loss": 0.0,
+      "step": 4127
+    },
+    {
+      "epoch": 0.2660308049236321,
+      "grad_norm": 0.01876722425713162,
+      "learning_rate": 8.158252774794128e-06,
+      "loss": 0.0002,
+      "step": 4128
+    },
+    {
+      "epoch": 0.2660952503705613,
+      "grad_norm": 0.0034013080274845862,
+      "learning_rate": 8.157536698890083e-06,
+      "loss": 0.0,
+      "step": 4129
+    },
+    {
+      "epoch": 0.2661596958174905,
+      "grad_norm": 0.0013784421649856453,
+      "learning_rate": 8.156820622986037e-06,
+      "loss": 0.0,
+      "step": 4130
+    },
+    {
+      "epoch": 0.2662241412644197,
+      "grad_norm": 0.002031356068066092,
+      "learning_rate": 8.156104547081991e-06,
+      "loss": 0.0,
+      "step": 4131
+    },
+    {
+      "epoch": 0.26628858671134886,
+      "grad_norm": 0.008120560940019509,
+      "learning_rate": 8.155388471177946e-06,
+      "loss": 0.0,
+      "step": 4132
+    },
+    {
+      "epoch": 0.266353032158278,
+      "grad_norm": 0.09823169443657727,
+      "learning_rate": 8.1546723952739e-06,
+      "loss": 0.0001,
+      "step": 4133
+    },
+    {
+      "epoch": 0.2664174776052072,
+      "grad_norm": 0.016965823095737853,
+      "learning_rate": 8.153956319369854e-06,
+      "loss": 0.0001,
+      "step": 4134
+    },
+    {
+      "epoch": 0.26648192305213636,
+      "grad_norm": 0.0015814950437502196,
+      "learning_rate": 8.153240243465807e-06,
+      "loss": 0.0,
+      "step": 4135
+    },
+    {
+      "epoch": 0.26654636849906554,
+      "grad_norm": 0.05691408457785069,
+      "learning_rate": 8.152524167561761e-06,
+      "loss": 0.0001,
+      "step": 4136
+    },
+    {
+      "epoch": 0.26661081394599473,
+      "grad_norm": 0.03462659186650775,
+      "learning_rate": 8.151808091657717e-06,
+      "loss": 0.0001,
+      "step": 4137
+    },
+    {
+      "epoch": 0.2666752593929239,
+      "grad_norm": 0.014375004746504964,
+      "learning_rate": 8.151092015753671e-06,
+      "loss": 0.0001,
+      "step": 4138
+    },
+    {
+      "epoch": 0.26673970483985304,
+      "grad_norm": 0.0024106756135800498,
+      "learning_rate": 8.150375939849626e-06,
+      "loss": 0.0,
+      "step": 4139
+    },
+    {
+      "epoch": 0.2668041502867822,
+      "grad_norm": 0.007668585018300137,
+      "learning_rate": 8.14965986394558e-06,
+      "loss": 0.0,
+      "step": 4140
+    },
+    {
+      "epoch": 0.2668685957337114,
+      "grad_norm": 0.006620185098797049,
+      "learning_rate": 8.148943788041533e-06,
+      "loss": 0.0,
+      "step": 4141
+    },
+    {
+      "epoch": 0.2669330411806406,
+      "grad_norm": 0.001317973851116255,
+      "learning_rate": 8.148227712137487e-06,
+      "loss": 0.0,
+      "step": 4142
+    },
+    {
+      "epoch": 0.2669974866275698,
+      "grad_norm": 0.035794016515809664,
+      "learning_rate": 8.147511636233441e-06,
+      "loss": 0.0004,
+      "step": 4143
+    },
+    {
+      "epoch": 0.2670619320744989,
+      "grad_norm": 0.00035408094955500456,
+      "learning_rate": 8.146795560329396e-06,
+      "loss": 0.0,
+      "step": 4144
+    },
+    {
+      "epoch": 0.2671263775214281,
+      "grad_norm": 0.008474903101516811,
+      "learning_rate": 8.14607948442535e-06,
+      "loss": 0.0001,
+      "step": 4145
+    },
+    {
+      "epoch": 0.2671908229683573,
+      "grad_norm": 0.0012700962973577897,
+      "learning_rate": 8.145363408521304e-06,
+      "loss": 0.0,
+      "step": 4146
+    },
+    {
+      "epoch": 0.26725526841528646,
+      "grad_norm": 0.0002888536530388627,
+      "learning_rate": 8.144647332617258e-06,
+      "loss": 0.0,
+      "step": 4147
+    },
+    {
+      "epoch": 0.26731971386221565,
+      "grad_norm": 0.0006534653339654185,
+      "learning_rate": 8.143931256713213e-06,
+      "loss": 0.0,
+      "step": 4148
+    },
+    {
+      "epoch": 0.26738415930914483,
+      "grad_norm": 0.07163475431684561,
+      "learning_rate": 8.143215180809167e-06,
+      "loss": 0.0001,
+      "step": 4149
+    },
+    {
+      "epoch": 0.26744860475607396,
+      "grad_norm": 0.0025577957639277035,
+      "learning_rate": 8.142499104905121e-06,
+      "loss": 0.0,
+      "step": 4150
+    },
+    {
+      "epoch": 0.26751305020300314,
+      "grad_norm": 0.00018299976222950143,
+      "learning_rate": 8.141783029001074e-06,
+      "loss": 0.0,
+      "step": 4151
+    },
+    {
+      "epoch": 0.26757749564993233,
+      "grad_norm": 0.2443402427946589,
+      "learning_rate": 8.141066953097028e-06,
+      "loss": 0.0007,
+      "step": 4152
+    },
+    {
+      "epoch": 0.2676419410968615,
+      "grad_norm": 0.4529980777695485,
+      "learning_rate": 8.140350877192983e-06,
+      "loss": 0.002,
+      "step": 4153
+    },
+    {
+      "epoch": 0.2677063865437907,
+      "grad_norm": 0.023886939799844756,
+      "learning_rate": 8.139634801288937e-06,
+      "loss": 0.0003,
+      "step": 4154
+    },
+    {
+      "epoch": 0.2677708319907199,
+      "grad_norm": 0.0007809454694696253,
+      "learning_rate": 8.138918725384891e-06,
+      "loss": 0.0,
+      "step": 4155
+    },
+    {
+      "epoch": 0.267835277437649,
+      "grad_norm": 0.18457986917933467,
+      "learning_rate": 8.138202649480845e-06,
+      "loss": 0.0043,
+      "step": 4156
+    },
+    {
+      "epoch": 0.2678997228845782,
+      "grad_norm": 0.005621835272940629,
+      "learning_rate": 8.1374865735768e-06,
+      "loss": 0.0001,
+      "step": 4157
+    },
+    {
+      "epoch": 0.2679641683315074,
+      "grad_norm": 0.0006343240890120217,
+      "learning_rate": 8.136770497672754e-06,
+      "loss": 0.0,
+      "step": 4158
+    },
+    {
+      "epoch": 0.26802861377843656,
+      "grad_norm": 0.07103669554810184,
+      "learning_rate": 8.136054421768708e-06,
+      "loss": 0.0002,
+      "step": 4159
+    },
+    {
+      "epoch": 0.26809305922536575,
+      "grad_norm": 0.00017467493782411902,
+      "learning_rate": 8.135338345864663e-06,
+      "loss": 0.0,
+      "step": 4160
+    },
+    {
+      "epoch": 0.2681575046722949,
+      "grad_norm": 0.00046530142134858303,
+      "learning_rate": 8.134622269960617e-06,
+      "loss": 0.0,
+      "step": 4161
+    },
+    {
+      "epoch": 0.26822195011922406,
+      "grad_norm": 0.00028367250039852975,
+      "learning_rate": 8.133906194056571e-06,
+      "loss": 0.0,
+      "step": 4162
+    },
+    {
+      "epoch": 0.26828639556615325,
+      "grad_norm": 0.07753849113407583,
+      "learning_rate": 8.133190118152525e-06,
+      "loss": 0.0002,
+      "step": 4163
+    },
+    {
+      "epoch": 0.26835084101308243,
+      "grad_norm": 0.00028427710331693025,
+      "learning_rate": 8.13247404224848e-06,
+      "loss": 0.0,
+      "step": 4164
+    },
+    {
+      "epoch": 0.2684152864600116,
+      "grad_norm": 0.0039621051110842725,
+      "learning_rate": 8.131757966344434e-06,
+      "loss": 0.0,
+      "step": 4165
+    },
+    {
+      "epoch": 0.2684797319069408,
+      "grad_norm": 0.021191026836945282,
+      "learning_rate": 8.131041890440388e-06,
+      "loss": 0.0001,
+      "step": 4166
+    },
+    {
+      "epoch": 0.26854417735386993,
+      "grad_norm": 0.0008846425877893397,
+      "learning_rate": 8.130325814536341e-06,
+      "loss": 0.0,
+      "step": 4167
+    },
+    {
+      "epoch": 0.2686086228007991,
+      "grad_norm": 0.00047771099607565615,
+      "learning_rate": 8.129609738632295e-06,
+      "loss": 0.0,
+      "step": 4168
+    },
+    {
+      "epoch": 0.2686730682477283,
+      "grad_norm": 0.011498115816560874,
+      "learning_rate": 8.12889366272825e-06,
+      "loss": 0.0,
+      "step": 4169
+    },
+    {
+      "epoch": 0.2687375136946575,
+      "grad_norm": 0.00035771112664960457,
+      "learning_rate": 8.128177586824204e-06,
+      "loss": 0.0,
+      "step": 4170
+    },
+    {
+      "epoch": 0.26880195914158667,
+      "grad_norm": 0.0011935188824502712,
+      "learning_rate": 8.127461510920158e-06,
+      "loss": 0.0,
+      "step": 4171
+    },
+    {
+      "epoch": 0.2688664045885158,
+      "grad_norm": 0.0036977720277806395,
+      "learning_rate": 8.126745435016112e-06,
+      "loss": 0.0,
+      "step": 4172
+    },
+    {
+      "epoch": 0.268930850035445,
+      "grad_norm": 0.0014385408918984439,
+      "learning_rate": 8.126029359112067e-06,
+      "loss": 0.0,
+      "step": 4173
+    },
+    {
+      "epoch": 0.26899529548237416,
+      "grad_norm": 0.0024186235474048293,
+      "learning_rate": 8.125313283208021e-06,
+      "loss": 0.0,
+      "step": 4174
+    },
+    {
+      "epoch": 0.26905974092930335,
+      "grad_norm": 0.034542698594028746,
+      "learning_rate": 8.124597207303975e-06,
+      "loss": 0.0001,
+      "step": 4175
+    },
+    {
+      "epoch": 0.26912418637623253,
+      "grad_norm": 0.009862969678027288,
+      "learning_rate": 8.12388113139993e-06,
+      "loss": 0.0,
+      "step": 4176
+    },
+    {
+      "epoch": 0.2691886318231617,
+      "grad_norm": 0.22107264717946118,
+      "learning_rate": 8.123165055495882e-06,
+      "loss": 0.002,
+      "step": 4177
+    },
+    {
+      "epoch": 0.26925307727009085,
+      "grad_norm": 0.0001773784457137841,
+      "learning_rate": 8.122448979591837e-06,
+      "loss": 0.0,
+      "step": 4178
+    },
+    {
+      "epoch": 0.26931752271702003,
+      "grad_norm": 0.7825109224501692,
+      "learning_rate": 8.121732903687791e-06,
+      "loss": 0.0013,
+      "step": 4179
+    },
+    {
+      "epoch": 0.2693819681639492,
+      "grad_norm": 1.3505293144001949,
+      "learning_rate": 8.121016827783745e-06,
+      "loss": 0.0069,
+      "step": 4180
+    },
+    {
+      "epoch": 0.2694464136108784,
+      "grad_norm": 0.15621311825343512,
+      "learning_rate": 8.1203007518797e-06,
+      "loss": 0.0,
+      "step": 4181
+    },
+    {
+      "epoch": 0.2695108590578076,
+      "grad_norm": 0.007509552035810147,
+      "learning_rate": 8.119584675975654e-06,
+      "loss": 0.0,
+      "step": 4182
+    },
+    {
+      "epoch": 0.26957530450473677,
+      "grad_norm": 0.0005021826837052551,
+      "learning_rate": 8.118868600071608e-06,
+      "loss": 0.0,
+      "step": 4183
+    },
+    {
+      "epoch": 0.2696397499516659,
+      "grad_norm": 0.0031436756764531048,
+      "learning_rate": 8.118152524167562e-06,
+      "loss": 0.0,
+      "step": 4184
+    },
+    {
+      "epoch": 0.2697041953985951,
+      "grad_norm": 0.9572962491351598,
+      "learning_rate": 8.117436448263517e-06,
+      "loss": 0.0021,
+      "step": 4185
+    },
+    {
+      "epoch": 0.26976864084552427,
+      "grad_norm": 1.3736213361417184,
+      "learning_rate": 8.116720372359471e-06,
+      "loss": 0.0054,
+      "step": 4186
+    },
+    {
+      "epoch": 0.26983308629245345,
+      "grad_norm": 0.3207216905616851,
+      "learning_rate": 8.116004296455425e-06,
+      "loss": 0.0012,
+      "step": 4187
+    },
+    {
+      "epoch": 0.26989753173938263,
+      "grad_norm": 0.029056574864155083,
+      "learning_rate": 8.11528822055138e-06,
+      "loss": 0.0001,
+      "step": 4188
+    },
+    {
+      "epoch": 0.26996197718631176,
+      "grad_norm": 0.0008124575126002671,
+      "learning_rate": 8.114572144647334e-06,
+      "loss": 0.0,
+      "step": 4189
+    },
+    {
+      "epoch": 0.27002642263324095,
+      "grad_norm": 0.007161701968818947,
+      "learning_rate": 8.113856068743288e-06,
+      "loss": 0.0,
+      "step": 4190
+    },
+    {
+      "epoch": 0.27009086808017013,
+      "grad_norm": 0.002907445352707109,
+      "learning_rate": 8.113139992839242e-06,
+      "loss": 0.0,
+      "step": 4191
+    },
+    {
+      "epoch": 0.2701553135270993,
+      "grad_norm": 0.0014930615116478428,
+      "learning_rate": 8.112423916935197e-06,
+      "loss": 0.0,
+      "step": 4192
+    },
+    {
+      "epoch": 0.2702197589740285,
+      "grad_norm": 0.0064780529217881066,
+      "learning_rate": 8.11170784103115e-06,
+      "loss": 0.0,
+      "step": 4193
+    },
+    {
+      "epoch": 0.2702842044209577,
+      "grad_norm": 0.0006701570723918696,
+      "learning_rate": 8.110991765127104e-06,
+      "loss": 0.0,
+      "step": 4194
+    },
+    {
+      "epoch": 0.2703486498678868,
+      "grad_norm": 0.17730787477319992,
+      "learning_rate": 8.110275689223058e-06,
+      "loss": 0.0007,
+      "step": 4195
+    },
+    {
+      "epoch": 0.270413095314816,
+      "grad_norm": 0.4032809813305308,
+      "learning_rate": 8.109559613319012e-06,
+      "loss": 0.0016,
+      "step": 4196
+    },
+    {
+      "epoch": 0.2704775407617452,
+      "grad_norm": 0.00046503311607447816,
+      "learning_rate": 8.108843537414967e-06,
+      "loss": 0.0,
+      "step": 4197
+    },
+    {
+      "epoch": 0.27054198620867437,
+      "grad_norm": 0.0011172217789278758,
+      "learning_rate": 8.108127461510921e-06,
+      "loss": 0.0,
+      "step": 4198
+    },
+    {
+      "epoch": 0.27060643165560355,
+      "grad_norm": 0.0008143952554526554,
+      "learning_rate": 8.107411385606875e-06,
+      "loss": 0.0,
+      "step": 4199
+    },
+    {
+      "epoch": 0.2706708771025327,
+      "grad_norm": 0.008522984751307939,
+      "learning_rate": 8.10669530970283e-06,
+      "loss": 0.0001,
+      "step": 4200
+    },
+    {
+      "epoch": 0.27073532254946187,
+      "grad_norm": 0.0024840219283999553,
+      "learning_rate": 8.105979233798784e-06,
+      "loss": 0.0,
+      "step": 4201
+    },
+    {
+      "epoch": 0.27079976799639105,
+      "grad_norm": 0.00039293305919349785,
+      "learning_rate": 8.105263157894736e-06,
+      "loss": 0.0,
+      "step": 4202
+    },
+    {
+      "epoch": 0.27086421344332023,
+      "grad_norm": 0.005540428539705254,
+      "learning_rate": 8.10454708199069e-06,
+      "loss": 0.0001,
+      "step": 4203
+    },
+    {
+      "epoch": 0.2709286588902494,
+      "grad_norm": 0.011686594502197983,
+      "learning_rate": 8.103831006086645e-06,
+      "loss": 0.0,
+      "step": 4204
+    },
+    {
+      "epoch": 0.2709931043371786,
+      "grad_norm": 0.16621984375021687,
+      "learning_rate": 8.1031149301826e-06,
+      "loss": 0.0001,
+      "step": 4205
+    },
+    {
+      "epoch": 0.27105754978410773,
+      "grad_norm": 0.0004452170447672888,
+      "learning_rate": 8.102398854278554e-06,
+      "loss": 0.0,
+      "step": 4206
+    },
+    {
+      "epoch": 0.2711219952310369,
+      "grad_norm": 0.00228342254717163,
+      "learning_rate": 8.10168277837451e-06,
+      "loss": 0.0,
+      "step": 4207
+    },
+    {
+      "epoch": 0.2711864406779661,
+      "grad_norm": 0.00729092715793383,
+      "learning_rate": 8.100966702470464e-06,
+      "loss": 0.0,
+      "step": 4208
+    },
+    {
+      "epoch": 0.2712508861248953,
+      "grad_norm": 0.01637368052505697,
+      "learning_rate": 8.100250626566416e-06,
+      "loss": 0.0001,
+      "step": 4209
+    },
+    {
+      "epoch": 0.27131533157182447,
+      "grad_norm": 0.47464788147246495,
+      "learning_rate": 8.09953455066237e-06,
+      "loss": 0.0007,
+      "step": 4210
+    },
+    {
+      "epoch": 0.2713797770187536,
+      "grad_norm": 0.017767601083250248,
+      "learning_rate": 8.098818474758325e-06,
+      "loss": 0.0001,
+      "step": 4211
+    },
+    {
+      "epoch": 0.2714442224656828,
+      "grad_norm": 0.0002587533399039369,
+      "learning_rate": 8.09810239885428e-06,
+      "loss": 0.0,
+      "step": 4212
+    },
+    {
+      "epoch": 0.27150866791261197,
+      "grad_norm": 1.0716687262551394,
+      "learning_rate": 8.097386322950234e-06,
+      "loss": 0.0001,
+      "step": 4213
+    },
+    {
+      "epoch": 0.27157311335954115,
+      "grad_norm": 0.0024058953844227756,
+      "learning_rate": 8.096670247046188e-06,
+      "loss": 0.0,
+      "step": 4214
+    },
+    {
+      "epoch": 0.27163755880647034,
+      "grad_norm": 0.00714180532387043,
+      "learning_rate": 8.095954171142142e-06,
+      "loss": 0.0,
+      "step": 4215
+    },
+    {
+      "epoch": 0.2717020042533995,
+      "grad_norm": 0.2157260896229057,
+      "learning_rate": 8.095238095238097e-06,
+      "loss": 0.0004,
+      "step": 4216
+    },
+    {
+      "epoch": 0.27176644970032865,
+      "grad_norm": 0.015134509852193008,
+      "learning_rate": 8.09452201933405e-06,
+      "loss": 0.0,
+      "step": 4217
+    },
+    {
+      "epoch": 0.27183089514725783,
+      "grad_norm": 0.013603620153944285,
+      "learning_rate": 8.093805943430003e-06,
+      "loss": 0.0001,
+      "step": 4218
+    },
+    {
+      "epoch": 0.271895340594187,
+      "grad_norm": 0.014178400987857479,
+      "learning_rate": 8.093089867525958e-06,
+      "loss": 0.0001,
+      "step": 4219
+    },
+    {
+      "epoch": 0.2719597860411162,
+      "grad_norm": 0.00024165479162664222,
+      "learning_rate": 8.092373791621912e-06,
+      "loss": 0.0,
+      "step": 4220
+    },
+    {
+      "epoch": 0.2720242314880454,
+      "grad_norm": 0.0020571868036936888,
+      "learning_rate": 8.091657715717866e-06,
+      "loss": 0.0,
+      "step": 4221
+    },
+    {
+      "epoch": 0.27208867693497457,
+      "grad_norm": 0.005309349799260391,
+      "learning_rate": 8.09094163981382e-06,
+      "loss": 0.0,
+      "step": 4222
+    },
+    {
+      "epoch": 0.2721531223819037,
+      "grad_norm": 0.30123737973572057,
+      "learning_rate": 8.090225563909775e-06,
+      "loss": 0.005,
+      "step": 4223
+    },
+    {
+      "epoch": 0.2722175678288329,
+      "grad_norm": 0.0012302217094057092,
+      "learning_rate": 8.08950948800573e-06,
+      "loss": 0.0,
+      "step": 4224
+    },
+    {
+      "epoch": 0.27228201327576207,
+      "grad_norm": 0.0002683422688236845,
+      "learning_rate": 8.088793412101684e-06,
+      "loss": 0.0,
+      "step": 4225
+    },
+    {
+      "epoch": 0.27234645872269125,
+      "grad_norm": 0.0006782941113959453,
+      "learning_rate": 8.088077336197638e-06,
+      "loss": 0.0,
+      "step": 4226
+    },
+    {
+      "epoch": 0.27241090416962044,
+      "grad_norm": 0.010038923617897768,
+      "learning_rate": 8.087361260293592e-06,
+      "loss": 0.0001,
+      "step": 4227
+    },
+    {
+      "epoch": 0.27247534961654957,
+      "grad_norm": 0.003118605860282275,
+      "learning_rate": 8.086645184389545e-06,
+      "loss": 0.0,
+      "step": 4228
+    },
+    {
+      "epoch": 0.27253979506347875,
+      "grad_norm": 0.01232286451285321,
+      "learning_rate": 8.085929108485499e-06,
+      "loss": 0.0001,
+      "step": 4229
+    },
+    {
+      "epoch": 0.27260424051040794,
+      "grad_norm": 0.8026916418279838,
+      "learning_rate": 8.085213032581455e-06,
+      "loss": 0.005,
+      "step": 4230
+    },
+    {
+      "epoch": 0.2726686859573371,
+      "grad_norm": 0.25647062712724145,
+      "learning_rate": 8.08449695667741e-06,
+      "loss": 0.0021,
+      "step": 4231
+    },
+    {
+      "epoch": 0.2727331314042663,
+      "grad_norm": 0.0792547996356328,
+      "learning_rate": 8.083780880773364e-06,
+      "loss": 0.0001,
+      "step": 4232
+    },
+    {
+      "epoch": 0.2727975768511955,
+      "grad_norm": 0.07577762607384415,
+      "learning_rate": 8.083064804869318e-06,
+      "loss": 0.0007,
+      "step": 4233
+    },
+    {
+      "epoch": 0.2728620222981246,
+      "grad_norm": 0.2807311066234324,
+      "learning_rate": 8.08234872896527e-06,
+      "loss": 0.0008,
+      "step": 4234
+    },
+    {
+      "epoch": 0.2729264677450538,
+      "grad_norm": 0.0006046875437334746,
+      "learning_rate": 8.081632653061225e-06,
+      "loss": 0.0,
+      "step": 4235
+    },
+    {
+      "epoch": 0.272990913191983,
+      "grad_norm": 0.030921507784656604,
+      "learning_rate": 8.080916577157179e-06,
+      "loss": 0.0002,
+      "step": 4236
+    },
+    {
+      "epoch": 0.27305535863891217,
+      "grad_norm": 0.17561644139596558,
+      "learning_rate": 8.080200501253133e-06,
+      "loss": 0.0018,
+      "step": 4237
+    },
+    {
+      "epoch": 0.27311980408584136,
+      "grad_norm": 0.0016910747966463998,
+      "learning_rate": 8.079484425349088e-06,
+      "loss": 0.0,
+      "step": 4238
+    },
+    {
+      "epoch": 0.2731842495327705,
+      "grad_norm": 0.017976164972360383,
+      "learning_rate": 8.078768349445042e-06,
+      "loss": 0.0,
+      "step": 4239
+    },
+    {
+      "epoch": 0.27324869497969967,
+      "grad_norm": 0.0002662013235983584,
+      "learning_rate": 8.078052273540996e-06,
+      "loss": 0.0,
+      "step": 4240
+    },
+    {
+      "epoch": 0.27331314042662885,
+      "grad_norm": 0.004663524159279211,
+      "learning_rate": 8.07733619763695e-06,
+      "loss": 0.0,
+      "step": 4241
+    },
+    {
+      "epoch": 0.27337758587355804,
+      "grad_norm": 0.21042938315959228,
+      "learning_rate": 8.076620121732905e-06,
+      "loss": 0.0005,
+      "step": 4242
+    },
+    {
+      "epoch": 0.2734420313204872,
+      "grad_norm": 0.03295338194526715,
+      "learning_rate": 8.07590404582886e-06,
+      "loss": 0.0001,
+      "step": 4243
+    },
+    {
+      "epoch": 0.2735064767674164,
+      "grad_norm": 0.0071647418191086885,
+      "learning_rate": 8.075187969924812e-06,
+      "loss": 0.0,
+      "step": 4244
+    },
+    {
+      "epoch": 0.27357092221434554,
+      "grad_norm": 0.09229751786692166,
+      "learning_rate": 8.074471894020766e-06,
+      "loss": 0.0009,
+      "step": 4245
+    },
+    {
+      "epoch": 0.2736353676612747,
+      "grad_norm": 0.005337680304002737,
+      "learning_rate": 8.07375581811672e-06,
+      "loss": 0.0,
+      "step": 4246
+    },
+    {
+      "epoch": 0.2736998131082039,
+      "grad_norm": 0.0066058648404035405,
+      "learning_rate": 8.073039742212675e-06,
+      "loss": 0.0,
+      "step": 4247
+    },
+    {
+      "epoch": 0.2737642585551331,
+      "grad_norm": 0.018948461852127226,
+      "learning_rate": 8.072323666308629e-06,
+      "loss": 0.0,
+      "step": 4248
+    },
+    {
+      "epoch": 0.2738287040020623,
+      "grad_norm": 0.28756954658064515,
+      "learning_rate": 8.071607590404583e-06,
+      "loss": 0.0011,
+      "step": 4249
+    },
+    {
+      "epoch": 0.2738931494489914,
+      "grad_norm": 0.1112823266641712,
+      "learning_rate": 8.070891514500538e-06,
+      "loss": 0.0002,
+      "step": 4250
+    },
+    {
+      "epoch": 0.2739575948959206,
+      "grad_norm": 0.0033629401611851456,
+      "learning_rate": 8.070175438596492e-06,
+      "loss": 0.0,
+      "step": 4251
+    },
+    {
+      "epoch": 0.27402204034284977,
+      "grad_norm": 0.01336159347321719,
+      "learning_rate": 8.069459362692446e-06,
+      "loss": 0.0001,
+      "step": 4252
+    },
+    {
+      "epoch": 0.27408648578977896,
+      "grad_norm": 0.08284823562061203,
+      "learning_rate": 8.0687432867884e-06,
+      "loss": 0.0005,
+      "step": 4253
+    },
+    {
+      "epoch": 0.27415093123670814,
+      "grad_norm": 0.02694515727281307,
+      "learning_rate": 8.068027210884355e-06,
+      "loss": 0.0,
+      "step": 4254
+    },
+    {
+      "epoch": 0.2742153766836373,
+      "grad_norm": 0.012616425954789283,
+      "learning_rate": 8.067311134980309e-06,
+      "loss": 0.0001,
+      "step": 4255
+    },
+    {
+      "epoch": 0.27427982213056645,
+      "grad_norm": 0.0007318360328273589,
+      "learning_rate": 8.066595059076263e-06,
+      "loss": 0.0,
+      "step": 4256
+    },
+    {
+      "epoch": 0.27434426757749564,
+      "grad_norm": 1.0259591406667061,
+      "learning_rate": 8.065878983172218e-06,
+      "loss": 0.0031,
+      "step": 4257
+    },
+    {
+      "epoch": 0.2744087130244248,
+      "grad_norm": 0.006475353661629028,
+      "learning_rate": 8.065162907268172e-06,
+      "loss": 0.0,
+      "step": 4258
+    },
+    {
+      "epoch": 0.274473158471354,
+      "grad_norm": 0.44145669902026735,
+      "learning_rate": 8.064446831364126e-06,
+      "loss": 0.0013,
+      "step": 4259
+    },
+    {
+      "epoch": 0.2745376039182832,
+      "grad_norm": 0.0496036830400151,
+      "learning_rate": 8.063730755460079e-06,
+      "loss": 0.0004,
+      "step": 4260
+    },
+    {
+      "epoch": 0.2746020493652124,
+      "grad_norm": 0.08654972122820045,
+      "learning_rate": 8.063014679556033e-06,
+      "loss": 0.0005,
+      "step": 4261
+    },
+    {
+      "epoch": 0.2746664948121415,
+      "grad_norm": 0.0032045456562780393,
+      "learning_rate": 8.062298603651988e-06,
+      "loss": 0.0,
+      "step": 4262
+    },
+    {
+      "epoch": 0.2747309402590707,
+      "grad_norm": 0.027642436666024946,
+      "learning_rate": 8.061582527747942e-06,
+      "loss": 0.0,
+      "step": 4263
+    },
+    {
+      "epoch": 0.2747953857059999,
+      "grad_norm": 0.026773698904416023,
+      "learning_rate": 8.060866451843896e-06,
+      "loss": 0.0001,
+      "step": 4264
+    },
+    {
+      "epoch": 0.27485983115292906,
+      "grad_norm": 0.009627766904490808,
+      "learning_rate": 8.06015037593985e-06,
+      "loss": 0.0,
+      "step": 4265
+    },
+    {
+      "epoch": 0.27492427659985824,
+      "grad_norm": 0.0016107713773131466,
+      "learning_rate": 8.059434300035805e-06,
+      "loss": 0.0,
+      "step": 4266
+    },
+    {
+      "epoch": 0.27498872204678737,
+      "grad_norm": 0.025908643128907324,
+      "learning_rate": 8.058718224131759e-06,
+      "loss": 0.0001,
+      "step": 4267
+    },
+    {
+      "epoch": 0.27505316749371655,
+      "grad_norm": 0.006806098258419837,
+      "learning_rate": 8.058002148227713e-06,
+      "loss": 0.0,
+      "step": 4268
+    },
+    {
+      "epoch": 0.27511761294064574,
+      "grad_norm": 0.007748768085792652,
+      "learning_rate": 8.057286072323668e-06,
+      "loss": 0.0,
+      "step": 4269
+    },
+    {
+      "epoch": 0.2751820583875749,
+      "grad_norm": 0.25058081332503046,
+      "learning_rate": 8.05656999641962e-06,
+      "loss": 0.0007,
+      "step": 4270
+    },
+    {
+      "epoch": 0.2752465038345041,
+      "grad_norm": 0.024260379358677264,
+      "learning_rate": 8.055853920515575e-06,
+      "loss": 0.0003,
+      "step": 4271
+    },
+    {
+      "epoch": 0.2753109492814333,
+      "grad_norm": 0.007065816195131135,
+      "learning_rate": 8.055137844611529e-06,
+      "loss": 0.0,
+      "step": 4272
+    },
+    {
+      "epoch": 0.2753753947283624,
+      "grad_norm": 0.0032001312958800823,
+      "learning_rate": 8.054421768707483e-06,
+      "loss": 0.0,
+      "step": 4273
+    },
+    {
+      "epoch": 0.2754398401752916,
+      "grad_norm": 0.0041486393690888055,
+      "learning_rate": 8.053705692803437e-06,
+      "loss": 0.0,
+      "step": 4274
+    },
+    {
+      "epoch": 0.2755042856222208,
+      "grad_norm": 0.002187301162407734,
+      "learning_rate": 8.052989616899392e-06,
+      "loss": 0.0,
+      "step": 4275
+    },
+    {
+      "epoch": 0.27556873106915,
+      "grad_norm": 0.019823731982545546,
+      "learning_rate": 8.052273540995346e-06,
+      "loss": 0.0001,
+      "step": 4276
+    },
+    {
+      "epoch": 0.27563317651607916,
+      "grad_norm": 0.03341653553794979,
+      "learning_rate": 8.0515574650913e-06,
+      "loss": 0.0002,
+      "step": 4277
+    },
+    {
+      "epoch": 0.2756976219630083,
+      "grad_norm": 0.0022856206212265636,
+      "learning_rate": 8.050841389187255e-06,
+      "loss": 0.0,
+      "step": 4278
+    },
+    {
+      "epoch": 0.2757620674099375,
+      "grad_norm": 0.09553208442519484,
+      "learning_rate": 8.050125313283209e-06,
+      "loss": 0.0002,
+      "step": 4279
+    },
+    {
+      "epoch": 0.27582651285686666,
+      "grad_norm": 0.008370866194229725,
+      "learning_rate": 8.049409237379163e-06,
+      "loss": 0.0001,
+      "step": 4280
+    },
+    {
+      "epoch": 0.27589095830379584,
+      "grad_norm": 0.0013321094427172684,
+      "learning_rate": 8.048693161475117e-06,
+      "loss": 0.0,
+      "step": 4281
+    },
+    {
+      "epoch": 0.275955403750725,
+      "grad_norm": 0.07517994067131512,
+      "learning_rate": 8.047977085571072e-06,
+      "loss": 0.0001,
+      "step": 4282
+    },
+    {
+      "epoch": 0.2760198491976542,
+      "grad_norm": 0.0026881408795613724,
+      "learning_rate": 8.047261009667026e-06,
+      "loss": 0.0,
+      "step": 4283
+    },
+    {
+      "epoch": 0.27608429464458334,
+      "grad_norm": 0.5881114557634267,
+      "learning_rate": 8.04654493376298e-06,
+      "loss": 0.0026,
+      "step": 4284
+    },
+    {
+      "epoch": 0.2761487400915125,
+      "grad_norm": 0.0552722037753931,
+      "learning_rate": 8.045828857858935e-06,
+      "loss": 0.0005,
+      "step": 4285
+    },
+    {
+      "epoch": 0.2762131855384417,
+      "grad_norm": 0.0019059926219596923,
+      "learning_rate": 8.045112781954887e-06,
+      "loss": 0.0,
+      "step": 4286
+    },
+    {
+      "epoch": 0.2762776309853709,
+      "grad_norm": 0.004945866811174837,
+      "learning_rate": 8.044396706050842e-06,
+      "loss": 0.0,
+      "step": 4287
+    },
+    {
+      "epoch": 0.2763420764323001,
+      "grad_norm": 0.020964249731356712,
+      "learning_rate": 8.043680630146796e-06,
+      "loss": 0.0001,
+      "step": 4288
+    },
+    {
+      "epoch": 0.2764065218792292,
+      "grad_norm": 2.334616909744486,
+      "learning_rate": 8.04296455424275e-06,
+      "loss": 0.0045,
+      "step": 4289
+    },
+    {
+      "epoch": 0.2764709673261584,
+      "grad_norm": 0.001980315912694081,
+      "learning_rate": 8.042248478338704e-06,
+      "loss": 0.0,
+      "step": 4290
+    },
+    {
+      "epoch": 0.2765354127730876,
+      "grad_norm": 0.07433033072093934,
+      "learning_rate": 8.041532402434659e-06,
+      "loss": 0.0003,
+      "step": 4291
+    },
+    {
+      "epoch": 0.27659985822001676,
+      "grad_norm": 0.015762347565720446,
+      "learning_rate": 8.040816326530613e-06,
+      "loss": 0.0,
+      "step": 4292
+    },
+    {
+      "epoch": 0.27666430366694594,
+      "grad_norm": 9.599061102562011e-05,
+      "learning_rate": 8.040100250626567e-06,
+      "loss": 0.0,
+      "step": 4293
+    },
+    {
+      "epoch": 0.2767287491138751,
+      "grad_norm": 0.011378677333695802,
+      "learning_rate": 8.039384174722522e-06,
+      "loss": 0.0001,
+      "step": 4294
+    },
+    {
+      "epoch": 0.27679319456080426,
+      "grad_norm": 0.0012122533737657336,
+      "learning_rate": 8.038668098818474e-06,
+      "loss": 0.0,
+      "step": 4295
+    },
+    {
+      "epoch": 0.27685764000773344,
+      "grad_norm": 0.00712413864184761,
+      "learning_rate": 8.037952022914429e-06,
+      "loss": 0.0,
+      "step": 4296
+    },
+    {
+      "epoch": 0.2769220854546626,
+      "grad_norm": 0.0021291666023090256,
+      "learning_rate": 8.037235947010383e-06,
+      "loss": 0.0,
+      "step": 4297
+    },
+    {
+      "epoch": 0.2769865309015918,
+      "grad_norm": 0.006989701767573375,
+      "learning_rate": 8.036519871106337e-06,
+      "loss": 0.0,
+      "step": 4298
+    },
+    {
+      "epoch": 0.277050976348521,
+      "grad_norm": 0.0030630408264887,
+      "learning_rate": 8.035803795202291e-06,
+      "loss": 0.0,
+      "step": 4299
+    },
+    {
+      "epoch": 0.2771154217954502,
+      "grad_norm": 0.42474275334818506,
+      "learning_rate": 8.035087719298247e-06,
+      "loss": 0.0018,
+      "step": 4300
+    },
+    {
+      "epoch": 0.2771798672423793,
+      "grad_norm": 0.0020389694047471857,
+      "learning_rate": 8.034371643394202e-06,
+      "loss": 0.0,
+      "step": 4301
+    },
+    {
+      "epoch": 0.2772443126893085,
+      "grad_norm": 0.02280851230690095,
+      "learning_rate": 8.033655567490154e-06,
+      "loss": 0.0001,
+      "step": 4302
+    },
+    {
+      "epoch": 0.2773087581362377,
+      "grad_norm": 0.8225653820081554,
+      "learning_rate": 8.032939491586109e-06,
+      "loss": 0.0045,
+      "step": 4303
+    },
+    {
+      "epoch": 0.27737320358316686,
+      "grad_norm": 0.007784121116634503,
+      "learning_rate": 8.032223415682063e-06,
+      "loss": 0.0,
+      "step": 4304
+    },
+    {
+      "epoch": 0.27743764903009605,
+      "grad_norm": 0.09095866422393512,
+      "learning_rate": 8.031507339778017e-06,
+      "loss": 0.0009,
+      "step": 4305
+    },
+    {
+      "epoch": 0.2775020944770252,
+      "grad_norm": 0.24910336875888767,
+      "learning_rate": 8.030791263873972e-06,
+      "loss": 0.0003,
+      "step": 4306
+    },
+    {
+      "epoch": 0.27756653992395436,
+      "grad_norm": 0.973263011252569,
+      "learning_rate": 8.030075187969926e-06,
+      "loss": 0.0138,
+      "step": 4307
+    },
+    {
+      "epoch": 0.27763098537088354,
+      "grad_norm": 0.49882107567641787,
+      "learning_rate": 8.02935911206588e-06,
+      "loss": 0.0013,
+      "step": 4308
+    },
+    {
+      "epoch": 0.2776954308178127,
+      "grad_norm": 0.001363859972825628,
+      "learning_rate": 8.028643036161834e-06,
+      "loss": 0.0,
+      "step": 4309
+    },
+    {
+      "epoch": 0.2777598762647419,
+      "grad_norm": 0.021673290387127987,
+      "learning_rate": 8.027926960257789e-06,
+      "loss": 0.0,
+      "step": 4310
+    },
+    {
+      "epoch": 0.2778243217116711,
+      "grad_norm": 0.02073911943213021,
+      "learning_rate": 8.027210884353741e-06,
+      "loss": 0.0001,
+      "step": 4311
+    },
+    {
+      "epoch": 0.2778887671586002,
+      "grad_norm": 0.35953358592667545,
+      "learning_rate": 8.026494808449696e-06,
+      "loss": 0.0029,
+      "step": 4312
+    },
+    {
+      "epoch": 0.2779532126055294,
+      "grad_norm": 0.004120392236843992,
+      "learning_rate": 8.02577873254565e-06,
+      "loss": 0.0,
+      "step": 4313
+    },
+    {
+      "epoch": 0.2780176580524586,
+      "grad_norm": 0.0016659283892527326,
+      "learning_rate": 8.025062656641604e-06,
+      "loss": 0.0,
+      "step": 4314
+    },
+    {
+      "epoch": 0.2780821034993878,
+      "grad_norm": 0.000579688410185921,
+      "learning_rate": 8.024346580737559e-06,
+      "loss": 0.0,
+      "step": 4315
+    },
+    {
+      "epoch": 0.27814654894631696,
+      "grad_norm": 0.1932598493658158,
+      "learning_rate": 8.023630504833513e-06,
+      "loss": 0.0009,
+      "step": 4316
+    },
+    {
+      "epoch": 0.2782109943932461,
+      "grad_norm": 0.009979227631807589,
+      "learning_rate": 8.022914428929467e-06,
+      "loss": 0.0,
+      "step": 4317
+    },
+    {
+      "epoch": 0.2782754398401753,
+      "grad_norm": 0.5246637414184626,
+      "learning_rate": 8.022198353025421e-06,
+      "loss": 0.0009,
+      "step": 4318
+    },
+    {
+      "epoch": 0.27833988528710446,
+      "grad_norm": 0.0019687456576445756,
+      "learning_rate": 8.021482277121376e-06,
+      "loss": 0.0,
+      "step": 4319
+    },
+    {
+      "epoch": 0.27840433073403364,
+      "grad_norm": 0.0011185073687473165,
+      "learning_rate": 8.02076620121733e-06,
+      "loss": 0.0,
+      "step": 4320
+    },
+    {
+      "epoch": 0.27846877618096283,
+      "grad_norm": 0.04931498752101843,
+      "learning_rate": 8.020050125313283e-06,
+      "loss": 0.0002,
+      "step": 4321
+    },
+    {
+      "epoch": 0.278533221627892,
+      "grad_norm": 0.00027083873114944093,
+      "learning_rate": 8.019334049409237e-06,
+      "loss": 0.0,
+      "step": 4322
+    },
+    {
+      "epoch": 0.27859766707482114,
+      "grad_norm": 0.05839166362732932,
+      "learning_rate": 8.018617973505191e-06,
+      "loss": 0.0008,
+      "step": 4323
+    },
+    {
+      "epoch": 0.2786621125217503,
+      "grad_norm": 0.0001520224323067307,
+      "learning_rate": 8.017901897601147e-06,
+      "loss": 0.0,
+      "step": 4324
+    },
+    {
+      "epoch": 0.2787265579686795,
+      "grad_norm": 0.0007621996242916202,
+      "learning_rate": 8.017185821697102e-06,
+      "loss": 0.0,
+      "step": 4325
+    },
+    {
+      "epoch": 0.2787910034156087,
+      "grad_norm": 0.00018054994650765078,
+      "learning_rate": 8.016469745793056e-06,
+      "loss": 0.0,
+      "step": 4326
+    },
+    {
+      "epoch": 0.2788554488625379,
+      "grad_norm": 0.004102618013888196,
+      "learning_rate": 8.015753669889008e-06,
+      "loss": 0.0,
+      "step": 4327
+    },
+    {
+      "epoch": 0.278919894309467,
+      "grad_norm": 0.0015162126017946738,
+      "learning_rate": 8.015037593984963e-06,
+      "loss": 0.0,
+      "step": 4328
+    },
+    {
+      "epoch": 0.2789843397563962,
+      "grad_norm": 0.002592029774412762,
+      "learning_rate": 8.014321518080917e-06,
+      "loss": 0.0,
+      "step": 4329
+    },
+    {
+      "epoch": 0.2790487852033254,
+      "grad_norm": 0.0014765765187576673,
+      "learning_rate": 8.013605442176871e-06,
+      "loss": 0.0,
+      "step": 4330
+    },
+    {
+      "epoch": 0.27911323065025456,
+      "grad_norm": 0.024509027084424147,
+      "learning_rate": 8.012889366272826e-06,
+      "loss": 0.0,
+      "step": 4331
+    },
+    {
+      "epoch": 0.27917767609718375,
+      "grad_norm": 0.0017077207017750734,
+      "learning_rate": 8.01217329036878e-06,
+      "loss": 0.0,
+      "step": 4332
+    },
+    {
+      "epoch": 0.27924212154411293,
+      "grad_norm": 0.0022571001131268175,
+      "learning_rate": 8.011457214464734e-06,
+      "loss": 0.0,
+      "step": 4333
+    },
+    {
+      "epoch": 0.27930656699104206,
+      "grad_norm": 0.0006027117464969319,
+      "learning_rate": 8.010741138560689e-06,
+      "loss": 0.0,
+      "step": 4334
+    },
+    {
+      "epoch": 0.27937101243797124,
+      "grad_norm": 0.00018672067099711776,
+      "learning_rate": 8.010025062656643e-06,
+      "loss": 0.0,
+      "step": 4335
+    },
+    {
+      "epoch": 0.27943545788490043,
+      "grad_norm": 0.0009781388881216495,
+      "learning_rate": 8.009308986752597e-06,
+      "loss": 0.0,
+      "step": 4336
+    },
+    {
+      "epoch": 0.2794999033318296,
+      "grad_norm": 0.004771590672136837,
+      "learning_rate": 8.00859291084855e-06,
+      "loss": 0.0,
+      "step": 4337
+    },
+    {
+      "epoch": 0.2795643487787588,
+      "grad_norm": 0.14926098480487418,
+      "learning_rate": 8.007876834944504e-06,
+      "loss": 0.0005,
+      "step": 4338
+    },
+    {
+      "epoch": 0.279628794225688,
+      "grad_norm": 0.486617649409908,
+      "learning_rate": 8.007160759040458e-06,
+      "loss": 0.0008,
+      "step": 4339
+    },
+    {
+      "epoch": 0.2796932396726171,
+      "grad_norm": 0.009931994546620311,
+      "learning_rate": 8.006444683136413e-06,
+      "loss": 0.0001,
+      "step": 4340
+    },
+    {
+      "epoch": 0.2797576851195463,
+      "grad_norm": 0.00038062467703721604,
+      "learning_rate": 8.005728607232367e-06,
+      "loss": 0.0,
+      "step": 4341
+    },
+    {
+      "epoch": 0.2798221305664755,
+      "grad_norm": 0.0003990964290874825,
+      "learning_rate": 8.005012531328321e-06,
+      "loss": 0.0,
+      "step": 4342
+    },
+    {
+      "epoch": 0.27988657601340466,
+      "grad_norm": 0.008947741416100167,
+      "learning_rate": 8.004296455424276e-06,
+      "loss": 0.0001,
+      "step": 4343
+    },
+    {
+      "epoch": 0.27995102146033385,
+      "grad_norm": 0.007823324713702041,
+      "learning_rate": 8.00358037952023e-06,
+      "loss": 0.0,
+      "step": 4344
+    },
+    {
+      "epoch": 0.280015466907263,
+      "grad_norm": 0.040479600314342654,
+      "learning_rate": 8.002864303616184e-06,
+      "loss": 0.0001,
+      "step": 4345
+    },
+    {
+      "epoch": 0.28007991235419216,
+      "grad_norm": 0.0055250204423881845,
+      "learning_rate": 8.002148227712138e-06,
+      "loss": 0.0,
+      "step": 4346
+    },
+    {
+      "epoch": 0.28014435780112135,
+      "grad_norm": 0.010728338781064616,
+      "learning_rate": 8.001432151808093e-06,
+      "loss": 0.0001,
+      "step": 4347
+    },
+    {
+      "epoch": 0.28020880324805053,
+      "grad_norm": 0.00040887627223743986,
+      "learning_rate": 8.000716075904047e-06,
+      "loss": 0.0,
+      "step": 4348
+    },
+    {
+      "epoch": 0.2802732486949797,
+      "grad_norm": 0.13571115180144624,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.0004,
+      "step": 4349
+    },
+    {
+      "epoch": 0.2803376941419089,
+      "grad_norm": 0.5591548156003548,
+      "learning_rate": 7.999283924095956e-06,
+      "loss": 0.004,
+      "step": 4350
+    },
+    {
+      "epoch": 0.28040213958883803,
+      "grad_norm": 2.2364727922981085,
+      "learning_rate": 7.99856784819191e-06,
+      "loss": 0.0049,
+      "step": 4351
+    },
+    {
+      "epoch": 0.2804665850357672,
+      "grad_norm": 0.015054656240965792,
+      "learning_rate": 7.997851772287864e-06,
+      "loss": 0.0,
+      "step": 4352
+    },
+    {
+      "epoch": 0.2805310304826964,
+      "grad_norm": 0.25799236381304563,
+      "learning_rate": 7.997135696383817e-06,
+      "loss": 0.0008,
+      "step": 4353
+    },
+    {
+      "epoch": 0.2805954759296256,
+      "grad_norm": 0.047828245320771644,
+      "learning_rate": 7.996419620479771e-06,
+      "loss": 0.0001,
+      "step": 4354
+    },
+    {
+      "epoch": 0.28065992137655477,
+      "grad_norm": 0.5221935978641291,
+      "learning_rate": 7.995703544575725e-06,
+      "loss": 0.0039,
+      "step": 4355
+    },
+    {
+      "epoch": 0.2807243668234839,
+      "grad_norm": 0.011907581028104474,
+      "learning_rate": 7.99498746867168e-06,
+      "loss": 0.0001,
+      "step": 4356
+    },
+    {
+      "epoch": 0.2807888122704131,
+      "grad_norm": 0.0031201953035860396,
+      "learning_rate": 7.994271392767634e-06,
+      "loss": 0.0,
+      "step": 4357
+    },
+    {
+      "epoch": 0.28085325771734226,
+      "grad_norm": 1.5818540874230766,
+      "learning_rate": 7.993555316863588e-06,
+      "loss": 0.0024,
+      "step": 4358
+    },
+    {
+      "epoch": 0.28091770316427145,
+      "grad_norm": 0.01792528071689751,
+      "learning_rate": 7.992839240959543e-06,
+      "loss": 0.0,
+      "step": 4359
+    },
+    {
+      "epoch": 0.28098214861120063,
+      "grad_norm": 0.03182447723050989,
+      "learning_rate": 7.992123165055497e-06,
+      "loss": 0.0003,
+      "step": 4360
+    },
+    {
+      "epoch": 0.2810465940581298,
+      "grad_norm": 0.0016257270435471858,
+      "learning_rate": 7.991407089151451e-06,
+      "loss": 0.0,
+      "step": 4361
+    },
+    {
+      "epoch": 0.28111103950505895,
+      "grad_norm": 0.4480030194664001,
+      "learning_rate": 7.990691013247406e-06,
+      "loss": 0.0019,
+      "step": 4362
+    },
+    {
+      "epoch": 0.28117548495198813,
+      "grad_norm": 0.051033991221149974,
+      "learning_rate": 7.989974937343358e-06,
+      "loss": 0.0001,
+      "step": 4363
+    },
+    {
+      "epoch": 0.2812399303989173,
+      "grad_norm": 0.2753491742477933,
+      "learning_rate": 7.989258861439312e-06,
+      "loss": 0.0011,
+      "step": 4364
+    },
+    {
+      "epoch": 0.2813043758458465,
+      "grad_norm": 0.009632521139026014,
+      "learning_rate": 7.988542785535267e-06,
+      "loss": 0.0001,
+      "step": 4365
+    },
+    {
+      "epoch": 0.2813688212927757,
+      "grad_norm": 0.39363273433598595,
+      "learning_rate": 7.987826709631221e-06,
+      "loss": 0.0048,
+      "step": 4366
+    },
+    {
+      "epoch": 0.2814332667397048,
+      "grad_norm": 0.004705003054986105,
+      "learning_rate": 7.987110633727175e-06,
+      "loss": 0.0,
+      "step": 4367
+    },
+    {
+      "epoch": 0.281497712186634,
+      "grad_norm": 0.002300404489696207,
+      "learning_rate": 7.98639455782313e-06,
+      "loss": 0.0,
+      "step": 4368
+    },
+    {
+      "epoch": 0.2815621576335632,
+      "grad_norm": 0.11177615728153535,
+      "learning_rate": 7.985678481919084e-06,
+      "loss": 0.0002,
+      "step": 4369
+    },
+    {
+      "epoch": 0.28162660308049237,
+      "grad_norm": 0.004602240365965075,
+      "learning_rate": 7.984962406015038e-06,
+      "loss": 0.0,
+      "step": 4370
+    },
+    {
+      "epoch": 0.28169104852742155,
+      "grad_norm": 0.05880574974903835,
+      "learning_rate": 7.984246330110993e-06,
+      "loss": 0.0017,
+      "step": 4371
+    },
+    {
+      "epoch": 0.28175549397435073,
+      "grad_norm": 0.008953983911220216,
+      "learning_rate": 7.983530254206947e-06,
+      "loss": 0.0001,
+      "step": 4372
+    },
+    {
+      "epoch": 0.28181993942127986,
+      "grad_norm": 0.007779673626552251,
+      "learning_rate": 7.982814178302901e-06,
+      "loss": 0.0001,
+      "step": 4373
+    },
+    {
+      "epoch": 0.28188438486820905,
+      "grad_norm": 0.0016397288000667067,
+      "learning_rate": 7.982098102398855e-06,
+      "loss": 0.0,
+      "step": 4374
+    },
+    {
+      "epoch": 0.28194883031513823,
+      "grad_norm": 0.07536983363889126,
+      "learning_rate": 7.98138202649481e-06,
+      "loss": 0.0005,
+      "step": 4375
+    },
+    {
+      "epoch": 0.2820132757620674,
+      "grad_norm": 0.0662605672125797,
+      "learning_rate": 7.980665950590764e-06,
+      "loss": 0.0002,
+      "step": 4376
+    },
+    {
+      "epoch": 0.2820777212089966,
+      "grad_norm": 0.001330320124680005,
+      "learning_rate": 7.979949874686718e-06,
+      "loss": 0.0,
+      "step": 4377
+    },
+    {
+      "epoch": 0.2821421666559258,
+      "grad_norm": 0.00196304783207142,
+      "learning_rate": 7.979233798782673e-06,
+      "loss": 0.0,
+      "step": 4378
+    },
+    {
+      "epoch": 0.2822066121028549,
+      "grad_norm": 0.037367839809333316,
+      "learning_rate": 7.978517722878625e-06,
+      "loss": 0.0001,
+      "step": 4379
+    },
+    {
+      "epoch": 0.2822710575497841,
+      "grad_norm": 0.006716713789545665,
+      "learning_rate": 7.97780164697458e-06,
+      "loss": 0.0,
+      "step": 4380
+    },
+    {
+      "epoch": 0.2823355029967133,
+      "grad_norm": 0.6094877187697938,
+      "learning_rate": 7.977085571070534e-06,
+      "loss": 0.0014,
+      "step": 4381
+    },
+    {
+      "epoch": 0.28239994844364247,
+      "grad_norm": 0.06727481563265361,
+      "learning_rate": 7.976369495166488e-06,
+      "loss": 0.0001,
+      "step": 4382
+    },
+    {
+      "epoch": 0.28246439389057165,
+      "grad_norm": 0.18548593022155066,
+      "learning_rate": 7.975653419262442e-06,
+      "loss": 0.0005,
+      "step": 4383
+    },
+    {
+      "epoch": 0.2825288393375008,
+      "grad_norm": 0.3365746156429773,
+      "learning_rate": 7.974937343358397e-06,
+      "loss": 0.0024,
+      "step": 4384
+    },
+    {
+      "epoch": 0.28259328478442997,
+      "grad_norm": 0.00029314361765416907,
+      "learning_rate": 7.974221267454351e-06,
+      "loss": 0.0,
+      "step": 4385
+    },
+    {
+      "epoch": 0.28265773023135915,
+      "grad_norm": 0.16602354030813277,
+      "learning_rate": 7.973505191550305e-06,
+      "loss": 0.0003,
+      "step": 4386
+    },
+    {
+      "epoch": 0.28272217567828833,
+      "grad_norm": 0.0002632260427845974,
+      "learning_rate": 7.97278911564626e-06,
+      "loss": 0.0,
+      "step": 4387
+    },
+    {
+      "epoch": 0.2827866211252175,
+      "grad_norm": 0.005457814706330409,
+      "learning_rate": 7.972073039742212e-06,
+      "loss": 0.0,
+      "step": 4388
+    },
+    {
+      "epoch": 0.2828510665721467,
+      "grad_norm": 0.3241865073108496,
+      "learning_rate": 7.971356963838167e-06,
+      "loss": 0.0074,
+      "step": 4389
+    },
+    {
+      "epoch": 0.28291551201907583,
+      "grad_norm": 0.03676520602108709,
+      "learning_rate": 7.97064088793412e-06,
+      "loss": 0.0001,
+      "step": 4390
+    },
+    {
+      "epoch": 0.282979957466005,
+      "grad_norm": 0.5804222704426779,
+      "learning_rate": 7.969924812030075e-06,
+      "loss": 0.0032,
+      "step": 4391
+    },
+    {
+      "epoch": 0.2830444029129342,
+      "grad_norm": 0.0040055974290435425,
+      "learning_rate": 7.96920873612603e-06,
+      "loss": 0.0,
+      "step": 4392
+    },
+    {
+      "epoch": 0.2831088483598634,
+      "grad_norm": 0.005350671320858573,
+      "learning_rate": 7.968492660221984e-06,
+      "loss": 0.0,
+      "step": 4393
+    },
+    {
+      "epoch": 0.28317329380679257,
+      "grad_norm": 0.004508154975757049,
+      "learning_rate": 7.96777658431794e-06,
+      "loss": 0.0,
+      "step": 4394
+    },
+    {
+      "epoch": 0.2832377392537217,
+      "grad_norm": 0.04889548900087098,
+      "learning_rate": 7.967060508413892e-06,
+      "loss": 0.0001,
+      "step": 4395
+    },
+    {
+      "epoch": 0.2833021847006509,
+      "grad_norm": 0.0443628869935588,
+      "learning_rate": 7.966344432509847e-06,
+      "loss": 0.0002,
+      "step": 4396
+    },
+    {
+      "epoch": 0.28336663014758007,
+      "grad_norm": 0.08155526279809674,
+      "learning_rate": 7.965628356605801e-06,
+      "loss": 0.0003,
+      "step": 4397
+    },
+    {
+      "epoch": 0.28343107559450925,
+      "grad_norm": 0.18657846288264684,
+      "learning_rate": 7.964912280701755e-06,
+      "loss": 0.0022,
+      "step": 4398
+    },
+    {
+      "epoch": 0.28349552104143844,
+      "grad_norm": 0.02187256732999685,
+      "learning_rate": 7.96419620479771e-06,
+      "loss": 0.0001,
+      "step": 4399
+    },
+    {
+      "epoch": 0.2835599664883676,
+      "grad_norm": 0.012205506467956272,
+      "learning_rate": 7.963480128893664e-06,
+      "loss": 0.0,
+      "step": 4400
+    },
+    {
+      "epoch": 0.28362441193529675,
+      "grad_norm": 0.037582196116246906,
+      "learning_rate": 7.962764052989618e-06,
+      "loss": 0.0,
+      "step": 4401
+    },
+    {
+      "epoch": 0.28368885738222593,
+      "grad_norm": 0.06266569361901952,
+      "learning_rate": 7.962047977085572e-06,
+      "loss": 0.0002,
+      "step": 4402
+    },
+    {
+      "epoch": 0.2837533028291551,
+      "grad_norm": 0.0008525386340146979,
+      "learning_rate": 7.961331901181527e-06,
+      "loss": 0.0,
+      "step": 4403
+    },
+    {
+      "epoch": 0.2838177482760843,
+      "grad_norm": 0.005498334319090381,
+      "learning_rate": 7.96061582527748e-06,
+      "loss": 0.0,
+      "step": 4404
+    },
+    {
+      "epoch": 0.2838821937230135,
+      "grad_norm": 0.010821442272882176,
+      "learning_rate": 7.959899749373434e-06,
+      "loss": 0.0,
+      "step": 4405
+    },
+    {
+      "epoch": 0.2839466391699426,
+      "grad_norm": 0.24158699435794354,
+      "learning_rate": 7.959183673469388e-06,
+      "loss": 0.002,
+      "step": 4406
+    },
+    {
+      "epoch": 0.2840110846168718,
+      "grad_norm": 0.3462137556730702,
+      "learning_rate": 7.958467597565342e-06,
+      "loss": 0.0012,
+      "step": 4407
+    },
+    {
+      "epoch": 0.284075530063801,
+      "grad_norm": 0.2733609092257243,
+      "learning_rate": 7.957751521661296e-06,
+      "loss": 0.0023,
+      "step": 4408
+    },
+    {
+      "epoch": 0.28413997551073017,
+      "grad_norm": 0.16384960734985596,
+      "learning_rate": 7.95703544575725e-06,
+      "loss": 0.0003,
+      "step": 4409
+    },
+    {
+      "epoch": 0.28420442095765935,
+      "grad_norm": 0.38844305999655787,
+      "learning_rate": 7.956319369853205e-06,
+      "loss": 0.0028,
+      "step": 4410
+    },
+    {
+      "epoch": 0.28426886640458854,
+      "grad_norm": 0.004082978805036018,
+      "learning_rate": 7.95560329394916e-06,
+      "loss": 0.0,
+      "step": 4411
+    },
+    {
+      "epoch": 0.28433331185151767,
+      "grad_norm": 0.004148427370944359,
+      "learning_rate": 7.954887218045114e-06,
+      "loss": 0.0,
+      "step": 4412
+    },
+    {
+      "epoch": 0.28439775729844685,
+      "grad_norm": 0.021951817193776223,
+      "learning_rate": 7.954171142141068e-06,
+      "loss": 0.0,
+      "step": 4413
+    },
+    {
+      "epoch": 0.28446220274537604,
+      "grad_norm": 0.006462803659022571,
+      "learning_rate": 7.95345506623702e-06,
+      "loss": 0.0,
+      "step": 4414
+    },
+    {
+      "epoch": 0.2845266481923052,
+      "grad_norm": 0.13909423337469742,
+      "learning_rate": 7.952738990332975e-06,
+      "loss": 0.0008,
+      "step": 4415
+    },
+    {
+      "epoch": 0.2845910936392344,
+      "grad_norm": 0.11019298248775587,
+      "learning_rate": 7.95202291442893e-06,
+      "loss": 0.002,
+      "step": 4416
+    },
+    {
+      "epoch": 0.2846555390861636,
+      "grad_norm": 0.015269646596399714,
+      "learning_rate": 7.951306838524885e-06,
+      "loss": 0.0,
+      "step": 4417
+    },
+    {
+      "epoch": 0.2847199845330927,
+      "grad_norm": 0.021452810796437523,
+      "learning_rate": 7.95059076262084e-06,
+      "loss": 0.0001,
+      "step": 4418
+    },
+    {
+      "epoch": 0.2847844299800219,
+      "grad_norm": 0.517738806776239,
+      "learning_rate": 7.949874686716794e-06,
+      "loss": 0.0012,
+      "step": 4419
+    },
+    {
+      "epoch": 0.2848488754269511,
+      "grad_norm": 0.03322008169773669,
+      "learning_rate": 7.949158610812746e-06,
+      "loss": 0.0003,
+      "step": 4420
+    },
+    {
+      "epoch": 0.28491332087388027,
+      "grad_norm": 0.009138953843116304,
+      "learning_rate": 7.9484425349087e-06,
+      "loss": 0.0,
+      "step": 4421
+    },
+    {
+      "epoch": 0.28497776632080946,
+      "grad_norm": 0.47363823956398626,
+      "learning_rate": 7.947726459004655e-06,
+      "loss": 0.0039,
+      "step": 4422
+    },
+    {
+      "epoch": 0.2850422117677386,
+      "grad_norm": 0.0020104430634621095,
+      "learning_rate": 7.94701038310061e-06,
+      "loss": 0.0,
+      "step": 4423
+    },
+    {
+      "epoch": 0.28510665721466777,
+      "grad_norm": 0.053079706230507256,
+      "learning_rate": 7.946294307196564e-06,
+      "loss": 0.0002,
+      "step": 4424
+    },
+    {
+      "epoch": 0.28517110266159695,
+      "grad_norm": 0.02456466714186916,
+      "learning_rate": 7.945578231292518e-06,
+      "loss": 0.0001,
+      "step": 4425
+    },
+    {
+      "epoch": 0.28523554810852614,
+      "grad_norm": 0.0008166402254194522,
+      "learning_rate": 7.944862155388472e-06,
+      "loss": 0.0,
+      "step": 4426
+    },
+    {
+      "epoch": 0.2852999935554553,
+      "grad_norm": 0.006199516264046085,
+      "learning_rate": 7.944146079484426e-06,
+      "loss": 0.0,
+      "step": 4427
+    },
+    {
+      "epoch": 0.2853644390023845,
+      "grad_norm": 0.0008081691424822122,
+      "learning_rate": 7.94343000358038e-06,
+      "loss": 0.0,
+      "step": 4428
+    },
+    {
+      "epoch": 0.28542888444931364,
+      "grad_norm": 0.0029845301345706203,
+      "learning_rate": 7.942713927676335e-06,
+      "loss": 0.0,
+      "step": 4429
+    },
+    {
+      "epoch": 0.2854933298962428,
+      "grad_norm": 0.011622539974228662,
+      "learning_rate": 7.941997851772288e-06,
+      "loss": 0.0,
+      "step": 4430
+    },
+    {
+      "epoch": 0.285557775343172,
+      "grad_norm": 0.00507248422770294,
+      "learning_rate": 7.941281775868242e-06,
+      "loss": 0.0,
+      "step": 4431
+    },
+    {
+      "epoch": 0.2856222207901012,
+      "grad_norm": 0.004660809038102969,
+      "learning_rate": 7.940565699964196e-06,
+      "loss": 0.0,
+      "step": 4432
+    },
+    {
+      "epoch": 0.2856866662370304,
+      "grad_norm": 0.00253746912322247,
+      "learning_rate": 7.93984962406015e-06,
+      "loss": 0.0,
+      "step": 4433
+    },
+    {
+      "epoch": 0.2857511116839595,
+      "grad_norm": 0.061854504871958836,
+      "learning_rate": 7.939133548156105e-06,
+      "loss": 0.002,
+      "step": 4434
+    },
+    {
+      "epoch": 0.2858155571308887,
+      "grad_norm": 0.008229746313761522,
+      "learning_rate": 7.93841747225206e-06,
+      "loss": 0.0,
+      "step": 4435
+    },
+    {
+      "epoch": 0.28588000257781787,
+      "grad_norm": 0.003418590816872071,
+      "learning_rate": 7.937701396348013e-06,
+      "loss": 0.0,
+      "step": 4436
+    },
+    {
+      "epoch": 0.28594444802474706,
+      "grad_norm": 0.4305533788055168,
+      "learning_rate": 7.936985320443968e-06,
+      "loss": 0.0026,
+      "step": 4437
+    },
+    {
+      "epoch": 0.28600889347167624,
+      "grad_norm": 0.006124099355040623,
+      "learning_rate": 7.936269244539922e-06,
+      "loss": 0.0,
+      "step": 4438
+    },
+    {
+      "epoch": 0.2860733389186054,
+      "grad_norm": 0.0031271294943862695,
+      "learning_rate": 7.935553168635876e-06,
+      "loss": 0.0,
+      "step": 4439
+    },
+    {
+      "epoch": 0.28613778436553455,
+      "grad_norm": 0.01045592306874938,
+      "learning_rate": 7.934837092731829e-06,
+      "loss": 0.0001,
+      "step": 4440
+    },
+    {
+      "epoch": 0.28620222981246374,
+      "grad_norm": 0.01239300328247604,
+      "learning_rate": 7.934121016827785e-06,
+      "loss": 0.0,
+      "step": 4441
+    },
+    {
+      "epoch": 0.2862666752593929,
+      "grad_norm": 0.1479763660766948,
+      "learning_rate": 7.93340494092374e-06,
+      "loss": 0.0003,
+      "step": 4442
+    },
+    {
+      "epoch": 0.2863311207063221,
+      "grad_norm": 0.002779436143763441,
+      "learning_rate": 7.932688865019694e-06,
+      "loss": 0.0,
+      "step": 4443
+    },
+    {
+      "epoch": 0.2863955661532513,
+      "grad_norm": 0.009939447193650896,
+      "learning_rate": 7.931972789115648e-06,
+      "loss": 0.0,
+      "step": 4444
+    },
+    {
+      "epoch": 0.2864600116001804,
+      "grad_norm": 0.03607640825679185,
+      "learning_rate": 7.931256713211602e-06,
+      "loss": 0.0001,
+      "step": 4445
+    },
+    {
+      "epoch": 0.2865244570471096,
+      "grad_norm": 0.030201390696854517,
+      "learning_rate": 7.930540637307555e-06,
+      "loss": 0.0,
+      "step": 4446
+    },
+    {
+      "epoch": 0.2865889024940388,
+      "grad_norm": 0.07210643736495957,
+      "learning_rate": 7.929824561403509e-06,
+      "loss": 0.0001,
+      "step": 4447
+    },
+    {
+      "epoch": 0.286653347940968,
+      "grad_norm": 0.27512171988414574,
+      "learning_rate": 7.929108485499463e-06,
+      "loss": 0.0023,
+      "step": 4448
+    },
+    {
+      "epoch": 0.28671779338789716,
+      "grad_norm": 0.007823387955755273,
+      "learning_rate": 7.928392409595418e-06,
+      "loss": 0.0,
+      "step": 4449
+    },
+    {
+      "epoch": 0.28678223883482634,
+      "grad_norm": 0.14412453491842261,
+      "learning_rate": 7.927676333691372e-06,
+      "loss": 0.0003,
+      "step": 4450
+    },
+    {
+      "epoch": 0.28684668428175547,
+      "grad_norm": 0.00010022605205979735,
+      "learning_rate": 7.926960257787326e-06,
+      "loss": 0.0,
+      "step": 4451
+    },
+    {
+      "epoch": 0.28691112972868466,
+      "grad_norm": 0.00852258052113132,
+      "learning_rate": 7.92624418188328e-06,
+      "loss": 0.0,
+      "step": 4452
+    },
+    {
+      "epoch": 0.28697557517561384,
+      "grad_norm": 0.005646042880733156,
+      "learning_rate": 7.925528105979235e-06,
+      "loss": 0.0,
+      "step": 4453
+    },
+    {
+      "epoch": 0.287040020622543,
+      "grad_norm": 0.0008066836868218597,
+      "learning_rate": 7.924812030075189e-06,
+      "loss": 0.0,
+      "step": 4454
+    },
+    {
+      "epoch": 0.2871044660694722,
+      "grad_norm": 0.001507509646450024,
+      "learning_rate": 7.924095954171143e-06,
+      "loss": 0.0,
+      "step": 4455
+    },
+    {
+      "epoch": 0.2871689115164014,
+      "grad_norm": 0.002629905680913306,
+      "learning_rate": 7.923379878267096e-06,
+      "loss": 0.0,
+      "step": 4456
+    },
+    {
+      "epoch": 0.2872333569633305,
+      "grad_norm": 0.0059772441425043265,
+      "learning_rate": 7.92266380236305e-06,
+      "loss": 0.0,
+      "step": 4457
+    },
+    {
+      "epoch": 0.2872978024102597,
+      "grad_norm": 0.0352865641772406,
+      "learning_rate": 7.921947726459005e-06,
+      "loss": 0.0001,
+      "step": 4458
+    },
+    {
+      "epoch": 0.2873622478571889,
+      "grad_norm": 0.0002449629222558241,
+      "learning_rate": 7.921231650554959e-06,
+      "loss": 0.0,
+      "step": 4459
+    },
+    {
+      "epoch": 0.2874266933041181,
+      "grad_norm": 0.00020933698575602438,
+      "learning_rate": 7.920515574650913e-06,
+      "loss": 0.0,
+      "step": 4460
+    },
+    {
+      "epoch": 0.28749113875104726,
+      "grad_norm": 0.0031949037566283696,
+      "learning_rate": 7.919799498746868e-06,
+      "loss": 0.0,
+      "step": 4461
+    },
+    {
+      "epoch": 0.2875555841979764,
+      "grad_norm": 0.0065847528646073294,
+      "learning_rate": 7.919083422842822e-06,
+      "loss": 0.0,
+      "step": 4462
+    },
+    {
+      "epoch": 0.2876200296449056,
+      "grad_norm": 0.007622588936978267,
+      "learning_rate": 7.918367346938776e-06,
+      "loss": 0.0001,
+      "step": 4463
+    },
+    {
+      "epoch": 0.28768447509183476,
+      "grad_norm": 0.0017648578089107282,
+      "learning_rate": 7.91765127103473e-06,
+      "loss": 0.0,
+      "step": 4464
+    },
+    {
+      "epoch": 0.28774892053876394,
+      "grad_norm": 0.02611680379581983,
+      "learning_rate": 7.916935195130685e-06,
+      "loss": 0.0001,
+      "step": 4465
+    },
+    {
+      "epoch": 0.2878133659856931,
+      "grad_norm": 0.0010693996340695561,
+      "learning_rate": 7.916219119226639e-06,
+      "loss": 0.0,
+      "step": 4466
+    },
+    {
+      "epoch": 0.2878778114326223,
+      "grad_norm": 0.05102518528801154,
+      "learning_rate": 7.915503043322593e-06,
+      "loss": 0.0004,
+      "step": 4467
+    },
+    {
+      "epoch": 0.28794225687955144,
+      "grad_norm": 0.03612863993394471,
+      "learning_rate": 7.914786967418548e-06,
+      "loss": 0.0001,
+      "step": 4468
+    },
+    {
+      "epoch": 0.2880067023264806,
+      "grad_norm": 0.017686609282882126,
+      "learning_rate": 7.914070891514502e-06,
+      "loss": 0.0001,
+      "step": 4469
+    },
+    {
+      "epoch": 0.2880711477734098,
+      "grad_norm": 0.0008417679838710003,
+      "learning_rate": 7.913354815610456e-06,
+      "loss": 0.0,
+      "step": 4470
+    },
+    {
+      "epoch": 0.288135593220339,
+      "grad_norm": 0.00041560685689716803,
+      "learning_rate": 7.91263873970641e-06,
+      "loss": 0.0,
+      "step": 4471
+    },
+    {
+      "epoch": 0.2882000386672682,
+      "grad_norm": 0.18941918295187352,
+      "learning_rate": 7.911922663802363e-06,
+      "loss": 0.0004,
+      "step": 4472
+    },
+    {
+      "epoch": 0.2882644841141973,
+      "grad_norm": 0.21404756925539772,
+      "learning_rate": 7.911206587898317e-06,
+      "loss": 0.0006,
+      "step": 4473
+    },
+    {
+      "epoch": 0.2883289295611265,
+      "grad_norm": 0.013452269074163891,
+      "learning_rate": 7.910490511994272e-06,
+      "loss": 0.0,
+      "step": 4474
+    },
+    {
+      "epoch": 0.2883933750080557,
+      "grad_norm": 0.09559103668822853,
+      "learning_rate": 7.909774436090226e-06,
+      "loss": 0.0004,
+      "step": 4475
+    },
+    {
+      "epoch": 0.28845782045498486,
+      "grad_norm": 0.025086608219520978,
+      "learning_rate": 7.90905836018618e-06,
+      "loss": 0.0001,
+      "step": 4476
+    },
+    {
+      "epoch": 0.28852226590191404,
+      "grad_norm": 0.006432836326093911,
+      "learning_rate": 7.908342284282135e-06,
+      "loss": 0.0,
+      "step": 4477
+    },
+    {
+      "epoch": 0.28858671134884323,
+      "grad_norm": 0.2498886709881648,
+      "learning_rate": 7.907626208378089e-06,
+      "loss": 0.0004,
+      "step": 4478
+    },
+    {
+      "epoch": 0.28865115679577236,
+      "grad_norm": 0.0026041772410369305,
+      "learning_rate": 7.906910132474043e-06,
+      "loss": 0.0,
+      "step": 4479
+    },
+    {
+      "epoch": 0.28871560224270154,
+      "grad_norm": 0.009591682875869855,
+      "learning_rate": 7.906194056569998e-06,
+      "loss": 0.0,
+      "step": 4480
+    },
+    {
+      "epoch": 0.2887800476896307,
+      "grad_norm": 0.00032052997232582793,
+      "learning_rate": 7.90547798066595e-06,
+      "loss": 0.0,
+      "step": 4481
+    },
+    {
+      "epoch": 0.2888444931365599,
+      "grad_norm": 0.00873159164063012,
+      "learning_rate": 7.904761904761904e-06,
+      "loss": 0.0,
+      "step": 4482
+    },
+    {
+      "epoch": 0.2889089385834891,
+      "grad_norm": 0.35534598777334087,
+      "learning_rate": 7.904045828857859e-06,
+      "loss": 0.0054,
+      "step": 4483
+    },
+    {
+      "epoch": 0.2889733840304182,
+      "grad_norm": 0.002808476140444069,
+      "learning_rate": 7.903329752953813e-06,
+      "loss": 0.0,
+      "step": 4484
+    },
+    {
+      "epoch": 0.2890378294773474,
+      "grad_norm": 0.007032176873919268,
+      "learning_rate": 7.902613677049767e-06,
+      "loss": 0.0,
+      "step": 4485
+    },
+    {
+      "epoch": 0.2891022749242766,
+      "grad_norm": 0.0008951899821003473,
+      "learning_rate": 7.901897601145722e-06,
+      "loss": 0.0,
+      "step": 4486
+    },
+    {
+      "epoch": 0.2891667203712058,
+      "grad_norm": 0.07280670081014678,
+      "learning_rate": 7.901181525241678e-06,
+      "loss": 0.0017,
+      "step": 4487
+    },
+    {
+      "epoch": 0.28923116581813496,
+      "grad_norm": 0.31948080480773533,
+      "learning_rate": 7.90046544933763e-06,
+      "loss": 0.0009,
+      "step": 4488
+    },
+    {
+      "epoch": 0.28929561126506415,
+      "grad_norm": 0.0005501602348737609,
+      "learning_rate": 7.899749373433585e-06,
+      "loss": 0.0,
+      "step": 4489
+    },
+    {
+      "epoch": 0.2893600567119933,
+      "grad_norm": 0.018795969206918124,
+      "learning_rate": 7.899033297529539e-06,
+      "loss": 0.0,
+      "step": 4490
+    },
+    {
+      "epoch": 0.28942450215892246,
+      "grad_norm": 0.0006946576011490777,
+      "learning_rate": 7.898317221625493e-06,
+      "loss": 0.0,
+      "step": 4491
+    },
+    {
+      "epoch": 0.28948894760585164,
+      "grad_norm": 0.00017523844631172394,
+      "learning_rate": 7.897601145721447e-06,
+      "loss": 0.0,
+      "step": 4492
+    },
+    {
+      "epoch": 0.2895533930527808,
+      "grad_norm": 0.00040692935676826147,
+      "learning_rate": 7.896885069817402e-06,
+      "loss": 0.0,
+      "step": 4493
+    },
+    {
+      "epoch": 0.28961783849971,
+      "grad_norm": 0.002279674125137819,
+      "learning_rate": 7.896168993913356e-06,
+      "loss": 0.0,
+      "step": 4494
+    },
+    {
+      "epoch": 0.2896822839466392,
+      "grad_norm": 0.2672872842209903,
+      "learning_rate": 7.89545291800931e-06,
+      "loss": 0.0027,
+      "step": 4495
+    },
+    {
+      "epoch": 0.2897467293935683,
+      "grad_norm": 0.006540003647810904,
+      "learning_rate": 7.894736842105265e-06,
+      "loss": 0.0,
+      "step": 4496
+    },
+    {
+      "epoch": 0.2898111748404975,
+      "grad_norm": 0.015930352995606616,
+      "learning_rate": 7.894020766201217e-06,
+      "loss": 0.0,
+      "step": 4497
+    },
+    {
+      "epoch": 0.2898756202874267,
+      "grad_norm": 0.004882923326128659,
+      "learning_rate": 7.893304690297172e-06,
+      "loss": 0.0,
+      "step": 4498
+    },
+    {
+      "epoch": 0.2899400657343559,
+      "grad_norm": 0.31889832402219837,
+      "learning_rate": 7.892588614393126e-06,
+      "loss": 0.0005,
+      "step": 4499
+    },
+    {
+      "epoch": 0.29000451118128506,
+      "grad_norm": 0.0013620853215409057,
+      "learning_rate": 7.89187253848908e-06,
+      "loss": 0.0,
+      "step": 4500
+    },
+    {
+      "epoch": 0.2900689566282142,
+      "grad_norm": 0.015338372594140916,
+      "learning_rate": 7.891156462585034e-06,
+      "loss": 0.0,
+      "step": 4501
+    },
+    {
+      "epoch": 0.2901334020751434,
+      "grad_norm": 0.05778266089969748,
+      "learning_rate": 7.890440386680989e-06,
+      "loss": 0.0001,
+      "step": 4502
+    },
+    {
+      "epoch": 0.29019784752207256,
+      "grad_norm": 0.00270977776113452,
+      "learning_rate": 7.889724310776943e-06,
+      "loss": 0.0,
+      "step": 4503
+    },
+    {
+      "epoch": 0.29026229296900175,
+      "grad_norm": 0.0037655202102685075,
+      "learning_rate": 7.889008234872897e-06,
+      "loss": 0.0,
+      "step": 4504
+    },
+    {
+      "epoch": 0.29032673841593093,
+      "grad_norm": 0.015389182378204158,
+      "learning_rate": 7.888292158968852e-06,
+      "loss": 0.0001,
+      "step": 4505
+    },
+    {
+      "epoch": 0.2903911838628601,
+      "grad_norm": 0.015059754222407136,
+      "learning_rate": 7.887576083064806e-06,
+      "loss": 0.0,
+      "step": 4506
+    },
+    {
+      "epoch": 0.29045562930978924,
+      "grad_norm": 0.03073885991750051,
+      "learning_rate": 7.886860007160759e-06,
+      "loss": 0.0,
+      "step": 4507
+    },
+    {
+      "epoch": 0.2905200747567184,
+      "grad_norm": 0.015867032314513176,
+      "learning_rate": 7.886143931256713e-06,
+      "loss": 0.0001,
+      "step": 4508
+    },
+    {
+      "epoch": 0.2905845202036476,
+      "grad_norm": 0.00045300063330874063,
+      "learning_rate": 7.885427855352667e-06,
+      "loss": 0.0,
+      "step": 4509
+    },
+    {
+      "epoch": 0.2906489656505768,
+      "grad_norm": 0.0004942003374035947,
+      "learning_rate": 7.884711779448621e-06,
+      "loss": 0.0,
+      "step": 4510
+    },
+    {
+      "epoch": 0.290713411097506,
+      "grad_norm": 0.0012337128738821176,
+      "learning_rate": 7.883995703544577e-06,
+      "loss": 0.0,
+      "step": 4511
+    },
+    {
+      "epoch": 0.2907778565444351,
+      "grad_norm": 0.0008876628842310143,
+      "learning_rate": 7.883279627640532e-06,
+      "loss": 0.0,
+      "step": 4512
+    },
+    {
+      "epoch": 0.2908423019913643,
+      "grad_norm": 0.007350333417431606,
+      "learning_rate": 7.882563551736484e-06,
+      "loss": 0.0,
+      "step": 4513
+    },
+    {
+      "epoch": 0.2909067474382935,
+      "grad_norm": 0.0002106284948672585,
+      "learning_rate": 7.881847475832439e-06,
+      "loss": 0.0,
+      "step": 4514
+    },
+    {
+      "epoch": 0.29097119288522266,
+      "grad_norm": 0.00035708859633882503,
+      "learning_rate": 7.881131399928393e-06,
+      "loss": 0.0,
+      "step": 4515
+    },
+    {
+      "epoch": 0.29103563833215185,
+      "grad_norm": 0.029065303728135556,
+      "learning_rate": 7.880415324024347e-06,
+      "loss": 0.0,
+      "step": 4516
+    },
+    {
+      "epoch": 0.29110008377908103,
+      "grad_norm": 0.2127117840682844,
+      "learning_rate": 7.879699248120301e-06,
+      "loss": 0.001,
+      "step": 4517
+    },
+    {
+      "epoch": 0.29116452922601016,
+      "grad_norm": 0.0006066203387944084,
+      "learning_rate": 7.878983172216256e-06,
+      "loss": 0.0,
+      "step": 4518
+    },
+    {
+      "epoch": 0.29122897467293934,
+      "grad_norm": 0.018112009074947217,
+      "learning_rate": 7.87826709631221e-06,
+      "loss": 0.0,
+      "step": 4519
+    },
+    {
+      "epoch": 0.29129342011986853,
+      "grad_norm": 0.002583304637280309,
+      "learning_rate": 7.877551020408164e-06,
+      "loss": 0.0,
+      "step": 4520
+    },
+    {
+      "epoch": 0.2913578655667977,
+      "grad_norm": 0.266955702883381,
+      "learning_rate": 7.876834944504119e-06,
+      "loss": 0.0019,
+      "step": 4521
+    },
+    {
+      "epoch": 0.2914223110137269,
+      "grad_norm": 0.018432121396502184,
+      "learning_rate": 7.876118868600073e-06,
+      "loss": 0.0,
+      "step": 4522
+    },
+    {
+      "epoch": 0.2914867564606561,
+      "grad_norm": 0.03903561799608369,
+      "learning_rate": 7.875402792696026e-06,
+      "loss": 0.0003,
+      "step": 4523
+    },
+    {
+      "epoch": 0.2915512019075852,
+      "grad_norm": 0.012237118359079405,
+      "learning_rate": 7.87468671679198e-06,
+      "loss": 0.0,
+      "step": 4524
+    },
+    {
+      "epoch": 0.2916156473545144,
+      "grad_norm": 0.0033078222190907323,
+      "learning_rate": 7.873970640887934e-06,
+      "loss": 0.0,
+      "step": 4525
+    },
+    {
+      "epoch": 0.2916800928014436,
+      "grad_norm": 0.003472865748525989,
+      "learning_rate": 7.873254564983888e-06,
+      "loss": 0.0,
+      "step": 4526
+    },
+    {
+      "epoch": 0.29174453824837276,
+      "grad_norm": 0.0013939818695185597,
+      "learning_rate": 7.872538489079843e-06,
+      "loss": 0.0,
+      "step": 4527
+    },
+    {
+      "epoch": 0.29180898369530195,
+      "grad_norm": 0.00018821632106576328,
+      "learning_rate": 7.871822413175797e-06,
+      "loss": 0.0,
+      "step": 4528
+    },
+    {
+      "epoch": 0.2918734291422311,
+      "grad_norm": 0.00306179393129678,
+      "learning_rate": 7.871106337271751e-06,
+      "loss": 0.0,
+      "step": 4529
+    },
+    {
+      "epoch": 0.29193787458916026,
+      "grad_norm": 0.0007617039287525343,
+      "learning_rate": 7.870390261367706e-06,
+      "loss": 0.0,
+      "step": 4530
+    },
+    {
+      "epoch": 0.29200232003608945,
+      "grad_norm": 1.2425142734099195,
+      "learning_rate": 7.86967418546366e-06,
+      "loss": 0.0022,
+      "step": 4531
+    },
+    {
+      "epoch": 0.29206676548301863,
+      "grad_norm": 0.0022768657242729392,
+      "learning_rate": 7.868958109559614e-06,
+      "loss": 0.0,
+      "step": 4532
+    },
+    {
+      "epoch": 0.2921312109299478,
+      "grad_norm": 0.00020864275355323165,
+      "learning_rate": 7.868242033655567e-06,
+      "loss": 0.0,
+      "step": 4533
+    },
+    {
+      "epoch": 0.292195656376877,
+      "grad_norm": 0.000213358795277892,
+      "learning_rate": 7.867525957751523e-06,
+      "loss": 0.0,
+      "step": 4534
+    },
+    {
+      "epoch": 0.29226010182380613,
+      "grad_norm": 0.00015255038864135203,
+      "learning_rate": 7.866809881847477e-06,
+      "loss": 0.0,
+      "step": 4535
+    },
+    {
+      "epoch": 0.2923245472707353,
+      "grad_norm": 0.00015455752183585395,
+      "learning_rate": 7.866093805943431e-06,
+      "loss": 0.0,
+      "step": 4536
+    },
+    {
+      "epoch": 0.2923889927176645,
+      "grad_norm": 0.12145031071190686,
+      "learning_rate": 7.865377730039386e-06,
+      "loss": 0.0001,
+      "step": 4537
+    },
+    {
+      "epoch": 0.2924534381645937,
+      "grad_norm": 0.031089749014432462,
+      "learning_rate": 7.86466165413534e-06,
+      "loss": 0.0002,
+      "step": 4538
+    },
+    {
+      "epoch": 0.29251788361152287,
+      "grad_norm": 0.3512590264114246,
+      "learning_rate": 7.863945578231293e-06,
+      "loss": 0.0025,
+      "step": 4539
+    },
+    {
+      "epoch": 0.292582329058452,
+      "grad_norm": 0.02734669707799407,
+      "learning_rate": 7.863229502327247e-06,
+      "loss": 0.0001,
+      "step": 4540
+    },
+    {
+      "epoch": 0.2926467745053812,
+      "grad_norm": 0.0010073198792487593,
+      "learning_rate": 7.862513426423201e-06,
+      "loss": 0.0,
+      "step": 4541
+    },
+    {
+      "epoch": 0.29271121995231036,
+      "grad_norm": 0.00019868880896533006,
+      "learning_rate": 7.861797350519156e-06,
+      "loss": 0.0,
+      "step": 4542
+    },
+    {
+      "epoch": 0.29277566539923955,
+      "grad_norm": 0.0006258420716790086,
+      "learning_rate": 7.86108127461511e-06,
+      "loss": 0.0,
+      "step": 4543
+    },
+    {
+      "epoch": 0.29284011084616873,
+      "grad_norm": 0.0005874389946818668,
+      "learning_rate": 7.860365198711064e-06,
+      "loss": 0.0,
+      "step": 4544
+    },
+    {
+      "epoch": 0.2929045562930979,
+      "grad_norm": 0.003848245562625219,
+      "learning_rate": 7.859649122807018e-06,
+      "loss": 0.0,
+      "step": 4545
+    },
+    {
+      "epoch": 0.29296900174002705,
+      "grad_norm": 0.0007591827858018054,
+      "learning_rate": 7.858933046902973e-06,
+      "loss": 0.0,
+      "step": 4546
+    },
+    {
+      "epoch": 0.29303344718695623,
+      "grad_norm": 0.0027661824546528937,
+      "learning_rate": 7.858216970998927e-06,
+      "loss": 0.0,
+      "step": 4547
+    },
+    {
+      "epoch": 0.2930978926338854,
+      "grad_norm": 0.004813073623290894,
+      "learning_rate": 7.857500895094881e-06,
+      "loss": 0.0,
+      "step": 4548
+    },
+    {
+      "epoch": 0.2931623380808146,
+      "grad_norm": 0.7728116364248644,
+      "learning_rate": 7.856784819190834e-06,
+      "loss": 0.0034,
+      "step": 4549
+    },
+    {
+      "epoch": 0.2932267835277438,
+      "grad_norm": 0.014514382210677952,
+      "learning_rate": 7.856068743286788e-06,
+      "loss": 0.0001,
+      "step": 4550
+    },
+    {
+      "epoch": 0.2932912289746729,
+      "grad_norm": 0.18422850996311158,
+      "learning_rate": 7.855352667382743e-06,
+      "loss": 0.0017,
+      "step": 4551
+    },
+    {
+      "epoch": 0.2933556744216021,
+      "grad_norm": 0.004560710887887605,
+      "learning_rate": 7.854636591478697e-06,
+      "loss": 0.0,
+      "step": 4552
+    },
+    {
+      "epoch": 0.2934201198685313,
+      "grad_norm": 0.03367021315660142,
+      "learning_rate": 7.853920515574651e-06,
+      "loss": 0.0001,
+      "step": 4553
+    },
+    {
+      "epoch": 0.29348456531546047,
+      "grad_norm": 0.0755510030149236,
+      "learning_rate": 7.853204439670605e-06,
+      "loss": 0.0008,
+      "step": 4554
+    },
+    {
+      "epoch": 0.29354901076238965,
+      "grad_norm": 0.27525725415499364,
+      "learning_rate": 7.85248836376656e-06,
+      "loss": 0.0008,
+      "step": 4555
+    },
+    {
+      "epoch": 0.29361345620931883,
+      "grad_norm": 0.019564531699342042,
+      "learning_rate": 7.851772287862514e-06,
+      "loss": 0.0,
+      "step": 4556
+    },
+    {
+      "epoch": 0.29367790165624796,
+      "grad_norm": 0.008114938435770352,
+      "learning_rate": 7.851056211958468e-06,
+      "loss": 0.0,
+      "step": 4557
+    },
+    {
+      "epoch": 0.29374234710317715,
+      "grad_norm": 0.3528800646587041,
+      "learning_rate": 7.850340136054423e-06,
+      "loss": 0.0002,
+      "step": 4558
+    },
+    {
+      "epoch": 0.29380679255010633,
+      "grad_norm": 0.0006609426373109618,
+      "learning_rate": 7.849624060150377e-06,
+      "loss": 0.0,
+      "step": 4559
+    },
+    {
+      "epoch": 0.2938712379970355,
+      "grad_norm": 0.18983152295207498,
+      "learning_rate": 7.848907984246331e-06,
+      "loss": 0.0019,
+      "step": 4560
+    },
+    {
+      "epoch": 0.2939356834439647,
+      "grad_norm": 0.051502671807416966,
+      "learning_rate": 7.848191908342286e-06,
+      "loss": 0.0001,
+      "step": 4561
+    },
+    {
+      "epoch": 0.2940001288908939,
+      "grad_norm": 0.004743945694445782,
+      "learning_rate": 7.84747583243824e-06,
+      "loss": 0.0,
+      "step": 4562
+    },
+    {
+      "epoch": 0.294064574337823,
+      "grad_norm": 0.3051454696860592,
+      "learning_rate": 7.846759756534194e-06,
+      "loss": 0.0007,
+      "step": 4563
+    },
+    {
+      "epoch": 0.2941290197847522,
+      "grad_norm": 0.0005357593062468779,
+      "learning_rate": 7.846043680630148e-06,
+      "loss": 0.0,
+      "step": 4564
+    },
+    {
+      "epoch": 0.2941934652316814,
+      "grad_norm": 0.0004975399317757935,
+      "learning_rate": 7.845327604726101e-06,
+      "loss": 0.0,
+      "step": 4565
+    },
+    {
+      "epoch": 0.29425791067861057,
+      "grad_norm": 0.002784029827135192,
+      "learning_rate": 7.844611528822055e-06,
+      "loss": 0.0,
+      "step": 4566
+    },
+    {
+      "epoch": 0.29432235612553975,
+      "grad_norm": 0.00492558169817448,
+      "learning_rate": 7.84389545291801e-06,
+      "loss": 0.0,
+      "step": 4567
+    },
+    {
+      "epoch": 0.2943868015724689,
+      "grad_norm": 0.0001998417877650629,
+      "learning_rate": 7.843179377013964e-06,
+      "loss": 0.0,
+      "step": 4568
+    },
+    {
+      "epoch": 0.29445124701939807,
+      "grad_norm": 0.01305385781332726,
+      "learning_rate": 7.842463301109918e-06,
+      "loss": 0.0001,
+      "step": 4569
+    },
+    {
+      "epoch": 0.29451569246632725,
+      "grad_norm": 0.010528942460724023,
+      "learning_rate": 7.841747225205873e-06,
+      "loss": 0.0001,
+      "step": 4570
+    },
+    {
+      "epoch": 0.29458013791325643,
+      "grad_norm": 0.00020059010498940666,
+      "learning_rate": 7.841031149301827e-06,
+      "loss": 0.0,
+      "step": 4571
+    },
+    {
+      "epoch": 0.2946445833601856,
+      "grad_norm": 0.0032149860570045243,
+      "learning_rate": 7.840315073397781e-06,
+      "loss": 0.0,
+      "step": 4572
+    },
+    {
+      "epoch": 0.2947090288071148,
+      "grad_norm": 0.71958861976571,
+      "learning_rate": 7.839598997493735e-06,
+      "loss": 0.0035,
+      "step": 4573
+    },
+    {
+      "epoch": 0.29477347425404393,
+      "grad_norm": 3.0767340446139673,
+      "learning_rate": 7.838882921589688e-06,
+      "loss": 0.025,
+      "step": 4574
+    },
+    {
+      "epoch": 0.2948379197009731,
+      "grad_norm": 0.001674350790710976,
+      "learning_rate": 7.838166845685642e-06,
+      "loss": 0.0,
+      "step": 4575
+    },
+    {
+      "epoch": 0.2949023651479023,
+      "grad_norm": 0.012657985235324564,
+      "learning_rate": 7.837450769781597e-06,
+      "loss": 0.0001,
+      "step": 4576
+    },
+    {
+      "epoch": 0.2949668105948315,
+      "grad_norm": 0.005560305992674056,
+      "learning_rate": 7.836734693877551e-06,
+      "loss": 0.0,
+      "step": 4577
+    },
+    {
+      "epoch": 0.29503125604176067,
+      "grad_norm": 0.039299774754964066,
+      "learning_rate": 7.836018617973505e-06,
+      "loss": 0.0001,
+      "step": 4578
+    },
+    {
+      "epoch": 0.2950957014886898,
+      "grad_norm": 0.00030549551142350253,
+      "learning_rate": 7.83530254206946e-06,
+      "loss": 0.0,
+      "step": 4579
+    },
+    {
+      "epoch": 0.295160146935619,
+      "grad_norm": 0.004564936822918925,
+      "learning_rate": 7.834586466165414e-06,
+      "loss": 0.0,
+      "step": 4580
+    },
+    {
+      "epoch": 0.29522459238254817,
+      "grad_norm": 0.001231723360405645,
+      "learning_rate": 7.833870390261368e-06,
+      "loss": 0.0,
+      "step": 4581
+    },
+    {
+      "epoch": 0.29528903782947735,
+      "grad_norm": 0.01125057962057763,
+      "learning_rate": 7.833154314357322e-06,
+      "loss": 0.0,
+      "step": 4582
+    },
+    {
+      "epoch": 0.29535348327640654,
+      "grad_norm": 0.025788512570669428,
+      "learning_rate": 7.832438238453277e-06,
+      "loss": 0.0,
+      "step": 4583
+    },
+    {
+      "epoch": 0.2954179287233357,
+      "grad_norm": 0.03530924012953212,
+      "learning_rate": 7.831722162549231e-06,
+      "loss": 0.0,
+      "step": 4584
+    },
+    {
+      "epoch": 0.29548237417026485,
+      "grad_norm": 0.03815530609659034,
+      "learning_rate": 7.831006086645185e-06,
+      "loss": 0.0005,
+      "step": 4585
+    },
+    {
+      "epoch": 0.29554681961719403,
+      "grad_norm": 0.7514677785081176,
+      "learning_rate": 7.83029001074114e-06,
+      "loss": 0.0039,
+      "step": 4586
+    },
+    {
+      "epoch": 0.2956112650641232,
+      "grad_norm": 0.009560576602998675,
+      "learning_rate": 7.829573934837094e-06,
+      "loss": 0.0001,
+      "step": 4587
+    },
+    {
+      "epoch": 0.2956757105110524,
+      "grad_norm": 0.032842624976011536,
+      "learning_rate": 7.828857858933048e-06,
+      "loss": 0.0,
+      "step": 4588
+    },
+    {
+      "epoch": 0.2957401559579816,
+      "grad_norm": 0.031006444152063366,
+      "learning_rate": 7.828141783029003e-06,
+      "loss": 0.0017,
+      "step": 4589
+    },
+    {
+      "epoch": 0.2958046014049107,
+      "grad_norm": 0.24915392579693496,
+      "learning_rate": 7.827425707124955e-06,
+      "loss": 0.0007,
+      "step": 4590
+    },
+    {
+      "epoch": 0.2958690468518399,
+      "grad_norm": 0.03304664469135916,
+      "learning_rate": 7.82670963122091e-06,
+      "loss": 0.0,
+      "step": 4591
+    },
+    {
+      "epoch": 0.2959334922987691,
+      "grad_norm": 0.05655894885454582,
+      "learning_rate": 7.825993555316864e-06,
+      "loss": 0.0001,
+      "step": 4592
+    },
+    {
+      "epoch": 0.29599793774569827,
+      "grad_norm": 0.38388087603718396,
+      "learning_rate": 7.825277479412818e-06,
+      "loss": 0.0011,
+      "step": 4593
+    },
+    {
+      "epoch": 0.29606238319262745,
+      "grad_norm": 0.566841116400412,
+      "learning_rate": 7.824561403508772e-06,
+      "loss": 0.0035,
+      "step": 4594
+    },
+    {
+      "epoch": 0.29612682863955664,
+      "grad_norm": 0.011698133493121187,
+      "learning_rate": 7.823845327604727e-06,
+      "loss": 0.0001,
+      "step": 4595
+    },
+    {
+      "epoch": 0.29619127408648577,
+      "grad_norm": 0.027770648157643973,
+      "learning_rate": 7.823129251700681e-06,
+      "loss": 0.0,
+      "step": 4596
+    },
+    {
+      "epoch": 0.29625571953341495,
+      "grad_norm": 2.278856981939525,
+      "learning_rate": 7.822413175796635e-06,
+      "loss": 0.0001,
+      "step": 4597
+    },
+    {
+      "epoch": 0.29632016498034414,
+      "grad_norm": 0.0020159815640930436,
+      "learning_rate": 7.82169709989259e-06,
+      "loss": 0.0,
+      "step": 4598
+    },
+    {
+      "epoch": 0.2963846104272733,
+      "grad_norm": 0.14718794658617437,
+      "learning_rate": 7.820981023988544e-06,
+      "loss": 0.0008,
+      "step": 4599
+    },
+    {
+      "epoch": 0.2964490558742025,
+      "grad_norm": 0.0021516636260371985,
+      "learning_rate": 7.820264948084496e-06,
+      "loss": 0.0,
+      "step": 4600
+    },
+    {
+      "epoch": 0.2965135013211317,
+      "grad_norm": 0.010642395207132137,
+      "learning_rate": 7.81954887218045e-06,
+      "loss": 0.0,
+      "step": 4601
+    },
+    {
+      "epoch": 0.2965779467680608,
+      "grad_norm": 0.04448229876604851,
+      "learning_rate": 7.818832796276405e-06,
+      "loss": 0.0,
+      "step": 4602
+    },
+    {
+      "epoch": 0.29664239221499,
+      "grad_norm": 0.0033751772275503853,
+      "learning_rate": 7.81811672037236e-06,
+      "loss": 0.0,
+      "step": 4603
+    },
+    {
+      "epoch": 0.2967068376619192,
+      "grad_norm": 0.0037728265986638255,
+      "learning_rate": 7.817400644468315e-06,
+      "loss": 0.0,
+      "step": 4604
+    },
+    {
+      "epoch": 0.29677128310884837,
+      "grad_norm": 0.0027119067769277696,
+      "learning_rate": 7.81668456856427e-06,
+      "loss": 0.0,
+      "step": 4605
+    },
+    {
+      "epoch": 0.29683572855577756,
+      "grad_norm": 0.06407639553398353,
+      "learning_rate": 7.815968492660222e-06,
+      "loss": 0.0002,
+      "step": 4606
+    },
+    {
+      "epoch": 0.2969001740027067,
+      "grad_norm": 0.03252517104288065,
+      "learning_rate": 7.815252416756177e-06,
+      "loss": 0.0001,
+      "step": 4607
+    },
+    {
+      "epoch": 0.29696461944963587,
+      "grad_norm": 0.04116721890821116,
+      "learning_rate": 7.81453634085213e-06,
+      "loss": 0.0003,
+      "step": 4608
+    },
+    {
+      "epoch": 0.29702906489656505,
+      "grad_norm": 0.0015415324517583455,
+      "learning_rate": 7.813820264948085e-06,
+      "loss": 0.0,
+      "step": 4609
+    },
+    {
+      "epoch": 0.29709351034349424,
+      "grad_norm": 0.08608630962120033,
+      "learning_rate": 7.81310418904404e-06,
+      "loss": 0.0002,
+      "step": 4610
+    },
+    {
+      "epoch": 0.2971579557904234,
+      "grad_norm": 0.00012577019306394426,
+      "learning_rate": 7.812388113139994e-06,
+      "loss": 0.0,
+      "step": 4611
+    },
+    {
+      "epoch": 0.2972224012373526,
+      "grad_norm": 0.006028051862400764,
+      "learning_rate": 7.811672037235948e-06,
+      "loss": 0.0001,
+      "step": 4612
+    },
+    {
+      "epoch": 0.29728684668428174,
+      "grad_norm": 0.10264788711136673,
+      "learning_rate": 7.810955961331902e-06,
+      "loss": 0.0004,
+      "step": 4613
+    },
+    {
+      "epoch": 0.2973512921312109,
+      "grad_norm": 0.037061054366988276,
+      "learning_rate": 7.810239885427857e-06,
+      "loss": 0.0,
+      "step": 4614
+    },
+    {
+      "epoch": 0.2974157375781401,
+      "grad_norm": 0.0002439607188696404,
+      "learning_rate": 7.809523809523811e-06,
+      "loss": 0.0,
+      "step": 4615
+    },
+    {
+      "epoch": 0.2974801830250693,
+      "grad_norm": 0.014972837410916137,
+      "learning_rate": 7.808807733619764e-06,
+      "loss": 0.0,
+      "step": 4616
+    },
+    {
+      "epoch": 0.2975446284719985,
+      "grad_norm": 0.0008584027435497128,
+      "learning_rate": 7.808091657715718e-06,
+      "loss": 0.0,
+      "step": 4617
+    },
+    {
+      "epoch": 0.2976090739189276,
+      "grad_norm": 0.0006271920321097305,
+      "learning_rate": 7.807375581811672e-06,
+      "loss": 0.0,
+      "step": 4618
+    },
+    {
+      "epoch": 0.2976735193658568,
+      "grad_norm": 0.005431255360080073,
+      "learning_rate": 7.806659505907626e-06,
+      "loss": 0.0,
+      "step": 4619
+    },
+    {
+      "epoch": 0.29773796481278597,
+      "grad_norm": 0.0010751240437544664,
+      "learning_rate": 7.80594343000358e-06,
+      "loss": 0.0,
+      "step": 4620
+    },
+    {
+      "epoch": 0.29780241025971516,
+      "grad_norm": 0.3879795137802696,
+      "learning_rate": 7.805227354099535e-06,
+      "loss": 0.0039,
+      "step": 4621
+    },
+    {
+      "epoch": 0.29786685570664434,
+      "grad_norm": 0.0020645484028067747,
+      "learning_rate": 7.80451127819549e-06,
+      "loss": 0.0,
+      "step": 4622
+    },
+    {
+      "epoch": 0.2979313011535735,
+      "grad_norm": 0.0010723496273858072,
+      "learning_rate": 7.803795202291444e-06,
+      "loss": 0.0,
+      "step": 4623
+    },
+    {
+      "epoch": 0.29799574660050265,
+      "grad_norm": 0.0027411407806028714,
+      "learning_rate": 7.803079126387398e-06,
+      "loss": 0.0,
+      "step": 4624
+    },
+    {
+      "epoch": 0.29806019204743184,
+      "grad_norm": 0.00014802318816350044,
+      "learning_rate": 7.802363050483352e-06,
+      "loss": 0.0,
+      "step": 4625
+    },
+    {
+      "epoch": 0.298124637494361,
+      "grad_norm": 0.031220465912154896,
+      "learning_rate": 7.801646974579305e-06,
+      "loss": 0.0002,
+      "step": 4626
+    },
+    {
+      "epoch": 0.2981890829412902,
+      "grad_norm": 0.002613055270924557,
+      "learning_rate": 7.80093089867526e-06,
+      "loss": 0.0,
+      "step": 4627
+    },
+    {
+      "epoch": 0.2982535283882194,
+      "grad_norm": 0.002614199435021259,
+      "learning_rate": 7.800214822771215e-06,
+      "loss": 0.0,
+      "step": 4628
+    },
+    {
+      "epoch": 0.2983179738351485,
+      "grad_norm": 0.0026466108391232086,
+      "learning_rate": 7.79949874686717e-06,
+      "loss": 0.0,
+      "step": 4629
+    },
+    {
+      "epoch": 0.2983824192820777,
+      "grad_norm": 0.0011872838771710653,
+      "learning_rate": 7.798782670963124e-06,
+      "loss": 0.0,
+      "step": 4630
+    },
+    {
+      "epoch": 0.2984468647290069,
+      "grad_norm": 0.46395645085834897,
+      "learning_rate": 7.798066595059078e-06,
+      "loss": 0.0014,
+      "step": 4631
+    },
+    {
+      "epoch": 0.2985113101759361,
+      "grad_norm": 0.0013353272721039187,
+      "learning_rate": 7.79735051915503e-06,
+      "loss": 0.0,
+      "step": 4632
+    },
+    {
+      "epoch": 0.29857575562286526,
+      "grad_norm": 0.0017705315561493894,
+      "learning_rate": 7.796634443250985e-06,
+      "loss": 0.0,
+      "step": 4633
+    },
+    {
+      "epoch": 0.29864020106979444,
+      "grad_norm": 0.007422942160128908,
+      "learning_rate": 7.79591836734694e-06,
+      "loss": 0.0001,
+      "step": 4634
+    },
+    {
+      "epoch": 0.29870464651672357,
+      "grad_norm": 0.0017911138589993113,
+      "learning_rate": 7.795202291442893e-06,
+      "loss": 0.0,
+      "step": 4635
+    },
+    {
+      "epoch": 0.29876909196365276,
+      "grad_norm": 0.023585737352795046,
+      "learning_rate": 7.794486215538848e-06,
+      "loss": 0.0002,
+      "step": 4636
+    },
+    {
+      "epoch": 0.29883353741058194,
+      "grad_norm": 0.0016310814255533401,
+      "learning_rate": 7.793770139634802e-06,
+      "loss": 0.0,
+      "step": 4637
+    },
+    {
+      "epoch": 0.2988979828575111,
+      "grad_norm": 0.10732582740168191,
+      "learning_rate": 7.793054063730756e-06,
+      "loss": 0.0011,
+      "step": 4638
+    },
+    {
+      "epoch": 0.2989624283044403,
+      "grad_norm": 0.15895521683864702,
+      "learning_rate": 7.79233798782671e-06,
+      "loss": 0.0006,
+      "step": 4639
+    },
+    {
+      "epoch": 0.2990268737513695,
+      "grad_norm": 0.00880201531000441,
+      "learning_rate": 7.791621911922665e-06,
+      "loss": 0.0,
+      "step": 4640
+    },
+    {
+      "epoch": 0.2990913191982986,
+      "grad_norm": 0.003968743597232801,
+      "learning_rate": 7.79090583601862e-06,
+      "loss": 0.0,
+      "step": 4641
+    },
+    {
+      "epoch": 0.2991557646452278,
+      "grad_norm": 0.001546892372756122,
+      "learning_rate": 7.790189760114572e-06,
+      "loss": 0.0,
+      "step": 4642
+    },
+    {
+      "epoch": 0.299220210092157,
+      "grad_norm": 0.003255836015276672,
+      "learning_rate": 7.789473684210526e-06,
+      "loss": 0.0,
+      "step": 4643
+    },
+    {
+      "epoch": 0.2992846555390862,
+      "grad_norm": 0.004197248189523486,
+      "learning_rate": 7.78875760830648e-06,
+      "loss": 0.0,
+      "step": 4644
+    },
+    {
+      "epoch": 0.29934910098601536,
+      "grad_norm": 0.0008416042110688216,
+      "learning_rate": 7.788041532402435e-06,
+      "loss": 0.0,
+      "step": 4645
+    },
+    {
+      "epoch": 0.2994135464329445,
+      "grad_norm": 0.0043030888462125645,
+      "learning_rate": 7.787325456498389e-06,
+      "loss": 0.0,
+      "step": 4646
+    },
+    {
+      "epoch": 0.2994779918798737,
+      "grad_norm": 0.033809603114627454,
+      "learning_rate": 7.786609380594343e-06,
+      "loss": 0.0002,
+      "step": 4647
+    },
+    {
+      "epoch": 0.29954243732680286,
+      "grad_norm": 0.0003606526883201009,
+      "learning_rate": 7.785893304690298e-06,
+      "loss": 0.0,
+      "step": 4648
+    },
+    {
+      "epoch": 0.29960688277373204,
+      "grad_norm": 0.014667166220949858,
+      "learning_rate": 7.785177228786252e-06,
+      "loss": 0.0,
+      "step": 4649
+    },
+    {
+      "epoch": 0.2996713282206612,
+      "grad_norm": 0.0005881952328653598,
+      "learning_rate": 7.784461152882206e-06,
+      "loss": 0.0,
+      "step": 4650
+    },
+    {
+      "epoch": 0.2997357736675904,
+      "grad_norm": 0.0008412704886720019,
+      "learning_rate": 7.78374507697816e-06,
+      "loss": 0.0,
+      "step": 4651
+    },
+    {
+      "epoch": 0.29980021911451954,
+      "grad_norm": 0.04959488604565864,
+      "learning_rate": 7.783029001074115e-06,
+      "loss": 0.0002,
+      "step": 4652
+    },
+    {
+      "epoch": 0.2998646645614487,
+      "grad_norm": 0.002363653848485576,
+      "learning_rate": 7.78231292517007e-06,
+      "loss": 0.0,
+      "step": 4653
+    },
+    {
+      "epoch": 0.2999291100083779,
+      "grad_norm": 0.00021418820248923426,
+      "learning_rate": 7.781596849266023e-06,
+      "loss": 0.0,
+      "step": 4654
+    },
+    {
+      "epoch": 0.2999935554553071,
+      "grad_norm": 0.0019327509427194745,
+      "learning_rate": 7.780880773361978e-06,
+      "loss": 0.0,
+      "step": 4655
+    },
+    {
+      "epoch": 0.3000580009022363,
+      "grad_norm": 0.004825567350965219,
+      "learning_rate": 7.780164697457932e-06,
+      "loss": 0.0,
+      "step": 4656
+    },
+    {
+      "epoch": 0.3001224463491654,
+      "grad_norm": 0.02280787046614858,
+      "learning_rate": 7.779448621553886e-06,
+      "loss": 0.0001,
+      "step": 4657
+    },
+    {
+      "epoch": 0.3001868917960946,
+      "grad_norm": 0.000839246267523498,
+      "learning_rate": 7.778732545649839e-06,
+      "loss": 0.0,
+      "step": 4658
+    },
+    {
+      "epoch": 0.3002513372430238,
+      "grad_norm": 7.830966965893415e-05,
+      "learning_rate": 7.778016469745793e-06,
+      "loss": 0.0,
+      "step": 4659
+    },
+    {
+      "epoch": 0.30031578268995296,
+      "grad_norm": 0.27347772847300145,
+      "learning_rate": 7.777300393841748e-06,
+      "loss": 0.002,
+      "step": 4660
+    },
+    {
+      "epoch": 0.30038022813688214,
+      "grad_norm": 0.004087371353731553,
+      "learning_rate": 7.776584317937702e-06,
+      "loss": 0.0,
+      "step": 4661
+    },
+    {
+      "epoch": 0.30044467358381133,
+      "grad_norm": 0.0006904881544805334,
+      "learning_rate": 7.775868242033656e-06,
+      "loss": 0.0,
+      "step": 4662
+    },
+    {
+      "epoch": 0.30050911903074046,
+      "grad_norm": 0.005781394579225731,
+      "learning_rate": 7.77515216612961e-06,
+      "loss": 0.0001,
+      "step": 4663
+    },
+    {
+      "epoch": 0.30057356447766964,
+      "grad_norm": 0.10011603023948211,
+      "learning_rate": 7.774436090225565e-06,
+      "loss": 0.0017,
+      "step": 4664
+    },
+    {
+      "epoch": 0.3006380099245988,
+      "grad_norm": 0.0003700707387533223,
+      "learning_rate": 7.773720014321519e-06,
+      "loss": 0.0,
+      "step": 4665
+    },
+    {
+      "epoch": 0.300702455371528,
+      "grad_norm": 0.0036266337136206684,
+      "learning_rate": 7.773003938417473e-06,
+      "loss": 0.0,
+      "step": 4666
+    },
+    {
+      "epoch": 0.3007669008184572,
+      "grad_norm": 0.006053707020341861,
+      "learning_rate": 7.772287862513428e-06,
+      "loss": 0.0,
+      "step": 4667
+    },
+    {
+      "epoch": 0.3008313462653863,
+      "grad_norm": 0.002798110998151991,
+      "learning_rate": 7.77157178660938e-06,
+      "loss": 0.0,
+      "step": 4668
+    },
+    {
+      "epoch": 0.3008957917123155,
+      "grad_norm": 0.002980137785368464,
+      "learning_rate": 7.770855710705335e-06,
+      "loss": 0.0,
+      "step": 4669
+    },
+    {
+      "epoch": 0.3009602371592447,
+      "grad_norm": 0.002200906356600856,
+      "learning_rate": 7.770139634801289e-06,
+      "loss": 0.0,
+      "step": 4670
+    },
+    {
+      "epoch": 0.3010246826061739,
+      "grad_norm": 0.3781393095729475,
+      "learning_rate": 7.769423558897243e-06,
+      "loss": 0.0006,
+      "step": 4671
+    },
+    {
+      "epoch": 0.30108912805310306,
+      "grad_norm": 0.012304835706529734,
+      "learning_rate": 7.768707482993197e-06,
+      "loss": 0.0,
+      "step": 4672
+    },
+    {
+      "epoch": 0.30115357350003225,
+      "grad_norm": 0.030847193216389755,
+      "learning_rate": 7.767991407089152e-06,
+      "loss": 0.0,
+      "step": 4673
+    },
+    {
+      "epoch": 0.3012180189469614,
+      "grad_norm": 0.0013446218988361794,
+      "learning_rate": 7.767275331185106e-06,
+      "loss": 0.0,
+      "step": 4674
+    },
+    {
+      "epoch": 0.30128246439389056,
+      "grad_norm": 0.3333385313145346,
+      "learning_rate": 7.76655925528106e-06,
+      "loss": 0.0004,
+      "step": 4675
+    },
+    {
+      "epoch": 0.30134690984081974,
+      "grad_norm": 0.05260483629009791,
+      "learning_rate": 7.765843179377015e-06,
+      "loss": 0.0001,
+      "step": 4676
+    },
+    {
+      "epoch": 0.3014113552877489,
+      "grad_norm": 0.012214662431553574,
+      "learning_rate": 7.765127103472969e-06,
+      "loss": 0.0,
+      "step": 4677
+    },
+    {
+      "epoch": 0.3014758007346781,
+      "grad_norm": 0.00023572483317842684,
+      "learning_rate": 7.764411027568923e-06,
+      "loss": 0.0,
+      "step": 4678
+    },
+    {
+      "epoch": 0.3015402461816073,
+      "grad_norm": 0.15849859996043447,
+      "learning_rate": 7.763694951664878e-06,
+      "loss": 0.0002,
+      "step": 4679
+    },
+    {
+      "epoch": 0.3016046916285364,
+      "grad_norm": 0.028688948102707078,
+      "learning_rate": 7.762978875760832e-06,
+      "loss": 0.0001,
+      "step": 4680
+    },
+    {
+      "epoch": 0.3016691370754656,
+      "grad_norm": 0.2802043258044797,
+      "learning_rate": 7.762262799856786e-06,
+      "loss": 0.0006,
+      "step": 4681
+    },
+    {
+      "epoch": 0.3017335825223948,
+      "grad_norm": 0.01589553270615149,
+      "learning_rate": 7.76154672395274e-06,
+      "loss": 0.0002,
+      "step": 4682
+    },
+    {
+      "epoch": 0.301798027969324,
+      "grad_norm": 0.005081411054152497,
+      "learning_rate": 7.760830648048695e-06,
+      "loss": 0.0,
+      "step": 4683
+    },
+    {
+      "epoch": 0.30186247341625316,
+      "grad_norm": 0.05606253532474472,
+      "learning_rate": 7.760114572144647e-06,
+      "loss": 0.0001,
+      "step": 4684
+    },
+    {
+      "epoch": 0.3019269188631823,
+      "grad_norm": 0.01129783697631848,
+      "learning_rate": 7.759398496240602e-06,
+      "loss": 0.0001,
+      "step": 4685
+    },
+    {
+      "epoch": 0.3019913643101115,
+      "grad_norm": 0.0004596890283011299,
+      "learning_rate": 7.758682420336556e-06,
+      "loss": 0.0,
+      "step": 4686
+    },
+    {
+      "epoch": 0.30205580975704066,
+      "grad_norm": 0.000857765967735999,
+      "learning_rate": 7.75796634443251e-06,
+      "loss": 0.0,
+      "step": 4687
+    },
+    {
+      "epoch": 0.30212025520396985,
+      "grad_norm": 0.0026360514695217855,
+      "learning_rate": 7.757250268528465e-06,
+      "loss": 0.0,
+      "step": 4688
+    },
+    {
+      "epoch": 0.30218470065089903,
+      "grad_norm": 0.019969502601820874,
+      "learning_rate": 7.756534192624419e-06,
+      "loss": 0.0,
+      "step": 4689
+    },
+    {
+      "epoch": 0.3022491460978282,
+      "grad_norm": 0.009782519107099928,
+      "learning_rate": 7.755818116720373e-06,
+      "loss": 0.0,
+      "step": 4690
+    },
+    {
+      "epoch": 0.30231359154475734,
+      "grad_norm": 0.001139921838035331,
+      "learning_rate": 7.755102040816327e-06,
+      "loss": 0.0,
+      "step": 4691
+    },
+    {
+      "epoch": 0.3023780369916865,
+      "grad_norm": 0.010498980062060898,
+      "learning_rate": 7.754385964912282e-06,
+      "loss": 0.0,
+      "step": 4692
+    },
+    {
+      "epoch": 0.3024424824386157,
+      "grad_norm": 0.209275470748402,
+      "learning_rate": 7.753669889008234e-06,
+      "loss": 0.0008,
+      "step": 4693
+    },
+    {
+      "epoch": 0.3025069278855449,
+      "grad_norm": 0.006151414616092774,
+      "learning_rate": 7.752953813104189e-06,
+      "loss": 0.0,
+      "step": 4694
+    },
+    {
+      "epoch": 0.3025713733324741,
+      "grad_norm": 0.2086179229931362,
+      "learning_rate": 7.752237737200143e-06,
+      "loss": 0.0015,
+      "step": 4695
+    },
+    {
+      "epoch": 0.3026358187794032,
+      "grad_norm": 0.00013166228230322507,
+      "learning_rate": 7.751521661296097e-06,
+      "loss": 0.0,
+      "step": 4696
+    },
+    {
+      "epoch": 0.3027002642263324,
+      "grad_norm": 9.777865384865902e-05,
+      "learning_rate": 7.750805585392053e-06,
+      "loss": 0.0,
+      "step": 4697
+    },
+    {
+      "epoch": 0.3027647096732616,
+      "grad_norm": 0.734374634762937,
+      "learning_rate": 7.750089509488008e-06,
+      "loss": 0.0025,
+      "step": 4698
+    },
+    {
+      "epoch": 0.30282915512019076,
+      "grad_norm": 0.006581345635300139,
+      "learning_rate": 7.749373433583962e-06,
+      "loss": 0.0,
+      "step": 4699
+    },
+    {
+      "epoch": 0.30289360056711995,
+      "grad_norm": 0.1299164592212838,
+      "learning_rate": 7.748657357679914e-06,
+      "loss": 0.0009,
+      "step": 4700
+    },
+    {
+      "epoch": 0.30295804601404913,
+      "grad_norm": 0.005917647420111509,
+      "learning_rate": 7.747941281775869e-06,
+      "loss": 0.0,
+      "step": 4701
+    },
+    {
+      "epoch": 0.30302249146097826,
+      "grad_norm": 0.46501049288836965,
+      "learning_rate": 7.747225205871823e-06,
+      "loss": 0.0088,
+      "step": 4702
+    },
+    {
+      "epoch": 0.30308693690790744,
+      "grad_norm": 0.0004601057773333485,
+      "learning_rate": 7.746509129967777e-06,
+      "loss": 0.0,
+      "step": 4703
+    },
+    {
+      "epoch": 0.30315138235483663,
+      "grad_norm": 0.16790902384417963,
+      "learning_rate": 7.745793054063732e-06,
+      "loss": 0.0002,
+      "step": 4704
+    },
+    {
+      "epoch": 0.3032158278017658,
+      "grad_norm": 0.009253831686704374,
+      "learning_rate": 7.745076978159686e-06,
+      "loss": 0.0,
+      "step": 4705
+    },
+    {
+      "epoch": 0.303280273248695,
+      "grad_norm": 0.05335917375409336,
+      "learning_rate": 7.74436090225564e-06,
+      "loss": 0.0002,
+      "step": 4706
+    },
+    {
+      "epoch": 0.3033447186956241,
+      "grad_norm": 0.0001441868512219672,
+      "learning_rate": 7.743644826351595e-06,
+      "loss": 0.0,
+      "step": 4707
+    },
+    {
+      "epoch": 0.3034091641425533,
+      "grad_norm": 0.0002741329134754317,
+      "learning_rate": 7.742928750447549e-06,
+      "loss": 0.0,
+      "step": 4708
+    },
+    {
+      "epoch": 0.3034736095894825,
+      "grad_norm": 0.01443467281263996,
+      "learning_rate": 7.742212674543501e-06,
+      "loss": 0.0001,
+      "step": 4709
+    },
+    {
+      "epoch": 0.3035380550364117,
+      "grad_norm": 0.015183455470397374,
+      "learning_rate": 7.741496598639456e-06,
+      "loss": 0.0001,
+      "step": 4710
+    },
+    {
+      "epoch": 0.30360250048334086,
+      "grad_norm": 0.023355433462827355,
+      "learning_rate": 7.74078052273541e-06,
+      "loss": 0.0002,
+      "step": 4711
+    },
+    {
+      "epoch": 0.30366694593027005,
+      "grad_norm": 0.004497382272455204,
+      "learning_rate": 7.740064446831364e-06,
+      "loss": 0.0,
+      "step": 4712
+    },
+    {
+      "epoch": 0.3037313913771992,
+      "grad_norm": 0.03580885399675683,
+      "learning_rate": 7.739348370927319e-06,
+      "loss": 0.0004,
+      "step": 4713
+    },
+    {
+      "epoch": 0.30379583682412836,
+      "grad_norm": 0.00021593148826366693,
+      "learning_rate": 7.738632295023273e-06,
+      "loss": 0.0,
+      "step": 4714
+    },
+    {
+      "epoch": 0.30386028227105755,
+      "grad_norm": 0.0006532279015124764,
+      "learning_rate": 7.737916219119227e-06,
+      "loss": 0.0,
+      "step": 4715
+    },
+    {
+      "epoch": 0.30392472771798673,
+      "grad_norm": 0.0016168136163807064,
+      "learning_rate": 7.737200143215182e-06,
+      "loss": 0.0,
+      "step": 4716
+    },
+    {
+      "epoch": 0.3039891731649159,
+      "grad_norm": 0.00016003900954974685,
+      "learning_rate": 7.736484067311136e-06,
+      "loss": 0.0,
+      "step": 4717
+    },
+    {
+      "epoch": 0.3040536186118451,
+      "grad_norm": 0.04574878624670092,
+      "learning_rate": 7.73576799140709e-06,
+      "loss": 0.0001,
+      "step": 4718
+    },
+    {
+      "epoch": 0.30411806405877423,
+      "grad_norm": 0.09921795946091631,
+      "learning_rate": 7.735051915503043e-06,
+      "loss": 0.0009,
+      "step": 4719
+    },
+    {
+      "epoch": 0.3041825095057034,
+      "grad_norm": 0.0009618883446051705,
+      "learning_rate": 7.734335839598997e-06,
+      "loss": 0.0,
+      "step": 4720
+    },
+    {
+      "epoch": 0.3042469549526326,
+      "grad_norm": 0.00015935827088833944,
+      "learning_rate": 7.733619763694953e-06,
+      "loss": 0.0,
+      "step": 4721
+    },
+    {
+      "epoch": 0.3043114003995618,
+      "grad_norm": 0.004121915434638065,
+      "learning_rate": 7.732903687790907e-06,
+      "loss": 0.0,
+      "step": 4722
+    },
+    {
+      "epoch": 0.30437584584649097,
+      "grad_norm": 1.251110299054438,
+      "learning_rate": 7.732187611886862e-06,
+      "loss": 0.0038,
+      "step": 4723
+    },
+    {
+      "epoch": 0.3044402912934201,
+      "grad_norm": 0.0019873520206396024,
+      "learning_rate": 7.731471535982816e-06,
+      "loss": 0.0,
+      "step": 4724
+    },
+    {
+      "epoch": 0.3045047367403493,
+      "grad_norm": 0.00038713677005660534,
+      "learning_rate": 7.730755460078769e-06,
+      "loss": 0.0,
+      "step": 4725
+    },
+    {
+      "epoch": 0.30456918218727846,
+      "grad_norm": 0.007009648691403379,
+      "learning_rate": 7.730039384174723e-06,
+      "loss": 0.0001,
+      "step": 4726
+    },
+    {
+      "epoch": 0.30463362763420765,
+      "grad_norm": 0.27969282933703976,
+      "learning_rate": 7.729323308270677e-06,
+      "loss": 0.0006,
+      "step": 4727
+    },
+    {
+      "epoch": 0.30469807308113683,
+      "grad_norm": 0.0012857714808325838,
+      "learning_rate": 7.728607232366631e-06,
+      "loss": 0.0,
+      "step": 4728
+    },
+    {
+      "epoch": 0.304762518528066,
+      "grad_norm": 0.5750549352359965,
+      "learning_rate": 7.727891156462586e-06,
+      "loss": 0.0035,
+      "step": 4729
+    },
+    {
+      "epoch": 0.30482696397499515,
+      "grad_norm": 0.2587270122489265,
+      "learning_rate": 7.72717508055854e-06,
+      "loss": 0.0004,
+      "step": 4730
+    },
+    {
+      "epoch": 0.30489140942192433,
+      "grad_norm": 0.002896226310871362,
+      "learning_rate": 7.726459004654494e-06,
+      "loss": 0.0,
+      "step": 4731
+    },
+    {
+      "epoch": 0.3049558548688535,
+      "grad_norm": 0.00044256071047343225,
+      "learning_rate": 7.725742928750449e-06,
+      "loss": 0.0,
+      "step": 4732
+    },
+    {
+      "epoch": 0.3050203003157827,
+      "grad_norm": 0.003308588148680433,
+      "learning_rate": 7.725026852846403e-06,
+      "loss": 0.0,
+      "step": 4733
+    },
+    {
+      "epoch": 0.3050847457627119,
+      "grad_norm": 0.001083691728375002,
+      "learning_rate": 7.724310776942357e-06,
+      "loss": 0.0,
+      "step": 4734
+    },
+    {
+      "epoch": 0.305149191209641,
+      "grad_norm": 0.0007827155588311144,
+      "learning_rate": 7.72359470103831e-06,
+      "loss": 0.0,
+      "step": 4735
+    },
+    {
+      "epoch": 0.3052136366565702,
+      "grad_norm": 0.036329813310400326,
+      "learning_rate": 7.722878625134264e-06,
+      "loss": 0.0004,
+      "step": 4736
+    },
+    {
+      "epoch": 0.3052780821034994,
+      "grad_norm": 0.0210308823587032,
+      "learning_rate": 7.722162549230218e-06,
+      "loss": 0.0,
+      "step": 4737
+    },
+    {
+      "epoch": 0.30534252755042857,
+      "grad_norm": 0.16520561188861155,
+      "learning_rate": 7.721446473326173e-06,
+      "loss": 0.0023,
+      "step": 4738
+    },
+    {
+      "epoch": 0.30540697299735775,
+      "grad_norm": 0.005275712354562943,
+      "learning_rate": 7.720730397422127e-06,
+      "loss": 0.0,
+      "step": 4739
+    },
+    {
+      "epoch": 0.30547141844428694,
+      "grad_norm": 0.900496966853056,
+      "learning_rate": 7.720014321518081e-06,
+      "loss": 0.0048,
+      "step": 4740
+    },
+    {
+      "epoch": 0.30553586389121606,
+      "grad_norm": 0.008231191698063514,
+      "learning_rate": 7.719298245614036e-06,
+      "loss": 0.0,
+      "step": 4741
+    },
+    {
+      "epoch": 0.30560030933814525,
+      "grad_norm": 0.3738767171786509,
+      "learning_rate": 7.71858216970999e-06,
+      "loss": 0.0023,
+      "step": 4742
+    },
+    {
+      "epoch": 0.30566475478507443,
+      "grad_norm": 0.3131810158747463,
+      "learning_rate": 7.717866093805944e-06,
+      "loss": 0.001,
+      "step": 4743
+    },
+    {
+      "epoch": 0.3057292002320036,
+      "grad_norm": 0.20129595328139735,
+      "learning_rate": 7.717150017901898e-06,
+      "loss": 0.0007,
+      "step": 4744
+    },
+    {
+      "epoch": 0.3057936456789328,
+      "grad_norm": 0.021132452629210614,
+      "learning_rate": 7.716433941997853e-06,
+      "loss": 0.0001,
+      "step": 4745
+    },
+    {
+      "epoch": 0.30585809112586193,
+      "grad_norm": 0.037783536362330825,
+      "learning_rate": 7.715717866093807e-06,
+      "loss": 0.0001,
+      "step": 4746
+    },
+    {
+      "epoch": 0.3059225365727911,
+      "grad_norm": 0.45276547013303264,
+      "learning_rate": 7.715001790189761e-06,
+      "loss": 0.0015,
+      "step": 4747
+    },
+    {
+      "epoch": 0.3059869820197203,
+      "grad_norm": 0.49549389414773587,
+      "learning_rate": 7.714285714285716e-06,
+      "loss": 0.0024,
+      "step": 4748
+    },
+    {
+      "epoch": 0.3060514274666495,
+      "grad_norm": 0.00019081945517868202,
+      "learning_rate": 7.71356963838167e-06,
+      "loss": 0.0,
+      "step": 4749
+    },
+    {
+      "epoch": 0.30611587291357867,
+      "grad_norm": 0.0014320023238470958,
+      "learning_rate": 7.712853562477624e-06,
+      "loss": 0.0,
+      "step": 4750
+    },
+    {
+      "epoch": 0.30618031836050785,
+      "grad_norm": 0.06193467670974739,
+      "learning_rate": 7.712137486573577e-06,
+      "loss": 0.0001,
+      "step": 4751
+    },
+    {
+      "epoch": 0.306244763807437,
+      "grad_norm": 0.024614945322434304,
+      "learning_rate": 7.711421410669531e-06,
+      "loss": 0.0,
+      "step": 4752
+    },
+    {
+      "epoch": 0.30630920925436617,
+      "grad_norm": 0.0035196920149812207,
+      "learning_rate": 7.710705334765485e-06,
+      "loss": 0.0,
+      "step": 4753
+    },
+    {
+      "epoch": 0.30637365470129535,
+      "grad_norm": 0.13125440414169431,
+      "learning_rate": 7.70998925886144e-06,
+      "loss": 0.0002,
+      "step": 4754
+    },
+    {
+      "epoch": 0.30643810014822453,
+      "grad_norm": 0.004681982640320887,
+      "learning_rate": 7.709273182957394e-06,
+      "loss": 0.0,
+      "step": 4755
+    },
+    {
+      "epoch": 0.3065025455951537,
+      "grad_norm": 0.006949093560674066,
+      "learning_rate": 7.708557107053348e-06,
+      "loss": 0.0,
+      "step": 4756
+    },
+    {
+      "epoch": 0.3065669910420829,
+      "grad_norm": 0.0036677102878780133,
+      "learning_rate": 7.707841031149303e-06,
+      "loss": 0.0,
+      "step": 4757
+    },
+    {
+      "epoch": 0.30663143648901203,
+      "grad_norm": 0.12488582731538928,
+      "learning_rate": 7.707124955245257e-06,
+      "loss": 0.0002,
+      "step": 4758
+    },
+    {
+      "epoch": 0.3066958819359412,
+      "grad_norm": 0.10194604235196442,
+      "learning_rate": 7.706408879341211e-06,
+      "loss": 0.002,
+      "step": 4759
+    },
+    {
+      "epoch": 0.3067603273828704,
+      "grad_norm": 0.029717985220950458,
+      "learning_rate": 7.705692803437166e-06,
+      "loss": 0.0003,
+      "step": 4760
+    },
+    {
+      "epoch": 0.3068247728297996,
+      "grad_norm": 0.29767039057667016,
+      "learning_rate": 7.704976727533118e-06,
+      "loss": 0.0043,
+      "step": 4761
+    },
+    {
+      "epoch": 0.30688921827672877,
+      "grad_norm": 0.009892853815882239,
+      "learning_rate": 7.704260651629072e-06,
+      "loss": 0.0001,
+      "step": 4762
+    },
+    {
+      "epoch": 0.3069536637236579,
+      "grad_norm": 0.9851313213488757,
+      "learning_rate": 7.703544575725027e-06,
+      "loss": 0.0037,
+      "step": 4763
+    },
+    {
+      "epoch": 0.3070181091705871,
+      "grad_norm": 0.17237959830448984,
+      "learning_rate": 7.702828499820981e-06,
+      "loss": 0.0004,
+      "step": 4764
+    },
+    {
+      "epoch": 0.30708255461751627,
+      "grad_norm": 0.20161969607969146,
+      "learning_rate": 7.702112423916935e-06,
+      "loss": 0.0007,
+      "step": 4765
+    },
+    {
+      "epoch": 0.30714700006444545,
+      "grad_norm": 0.009081867704277675,
+      "learning_rate": 7.70139634801289e-06,
+      "loss": 0.0,
+      "step": 4766
+    },
+    {
+      "epoch": 0.30721144551137464,
+      "grad_norm": 0.00021591472203924571,
+      "learning_rate": 7.700680272108844e-06,
+      "loss": 0.0,
+      "step": 4767
+    },
+    {
+      "epoch": 0.3072758909583038,
+      "grad_norm": 0.1620586212402397,
+      "learning_rate": 7.699964196204798e-06,
+      "loss": 0.0014,
+      "step": 4768
+    },
+    {
+      "epoch": 0.30734033640523295,
+      "grad_norm": 0.0013959799328598175,
+      "learning_rate": 7.699248120300753e-06,
+      "loss": 0.0,
+      "step": 4769
+    },
+    {
+      "epoch": 0.30740478185216213,
+      "grad_norm": 0.0035084092927934003,
+      "learning_rate": 7.698532044396707e-06,
+      "loss": 0.0,
+      "step": 4770
+    },
+    {
+      "epoch": 0.3074692272990913,
+      "grad_norm": 0.0012145909662963545,
+      "learning_rate": 7.697815968492661e-06,
+      "loss": 0.0,
+      "step": 4771
+    },
+    {
+      "epoch": 0.3075336727460205,
+      "grad_norm": 0.0020774691234807267,
+      "learning_rate": 7.697099892588615e-06,
+      "loss": 0.0,
+      "step": 4772
+    },
+    {
+      "epoch": 0.3075981181929497,
+      "grad_norm": 0.02511536306626666,
+      "learning_rate": 7.69638381668457e-06,
+      "loss": 0.0001,
+      "step": 4773
+    },
+    {
+      "epoch": 0.3076625636398788,
+      "grad_norm": 0.3414683043347056,
+      "learning_rate": 7.695667740780524e-06,
+      "loss": 0.001,
+      "step": 4774
+    },
+    {
+      "epoch": 0.307727009086808,
+      "grad_norm": 0.018721636445177253,
+      "learning_rate": 7.694951664876478e-06,
+      "loss": 0.0,
+      "step": 4775
+    },
+    {
+      "epoch": 0.3077914545337372,
+      "grad_norm": 0.004516405023989719,
+      "learning_rate": 7.694235588972433e-06,
+      "loss": 0.0,
+      "step": 4776
+    },
+    {
+      "epoch": 0.30785589998066637,
+      "grad_norm": 0.001999490388142513,
+      "learning_rate": 7.693519513068385e-06,
+      "loss": 0.0,
+      "step": 4777
+    },
+    {
+      "epoch": 0.30792034542759555,
+      "grad_norm": 0.009126433007855877,
+      "learning_rate": 7.69280343716434e-06,
+      "loss": 0.0,
+      "step": 4778
+    },
+    {
+      "epoch": 0.30798479087452474,
+      "grad_norm": 0.1669292889199868,
+      "learning_rate": 7.692087361260294e-06,
+      "loss": 0.0003,
+      "step": 4779
+    },
+    {
+      "epoch": 0.30804923632145387,
+      "grad_norm": 0.1951716678231349,
+      "learning_rate": 7.691371285356248e-06,
+      "loss": 0.0021,
+      "step": 4780
+    },
+    {
+      "epoch": 0.30811368176838305,
+      "grad_norm": 0.03909516756695523,
+      "learning_rate": 7.690655209452202e-06,
+      "loss": 0.0001,
+      "step": 4781
+    },
+    {
+      "epoch": 0.30817812721531224,
+      "grad_norm": 0.03378777956318579,
+      "learning_rate": 7.689939133548157e-06,
+      "loss": 0.0001,
+      "step": 4782
+    },
+    {
+      "epoch": 0.3082425726622414,
+      "grad_norm": 0.019446393540969095,
+      "learning_rate": 7.689223057644111e-06,
+      "loss": 0.0001,
+      "step": 4783
+    },
+    {
+      "epoch": 0.3083070181091706,
+      "grad_norm": 0.00664535455497235,
+      "learning_rate": 7.688506981740065e-06,
+      "loss": 0.0,
+      "step": 4784
+    },
+    {
+      "epoch": 0.30837146355609973,
+      "grad_norm": 0.03097241392532757,
+      "learning_rate": 7.68779090583602e-06,
+      "loss": 0.0,
+      "step": 4785
+    },
+    {
+      "epoch": 0.3084359090030289,
+      "grad_norm": 0.00806853471227027,
+      "learning_rate": 7.687074829931972e-06,
+      "loss": 0.0001,
+      "step": 4786
+    },
+    {
+      "epoch": 0.3085003544499581,
+      "grad_norm": 0.018839860233995302,
+      "learning_rate": 7.686358754027927e-06,
+      "loss": 0.0001,
+      "step": 4787
+    },
+    {
+      "epoch": 0.3085647998968873,
+      "grad_norm": 0.04220925687343892,
+      "learning_rate": 7.685642678123881e-06,
+      "loss": 0.0001,
+      "step": 4788
+    },
+    {
+      "epoch": 0.30862924534381647,
+      "grad_norm": 0.0034860404645811554,
+      "learning_rate": 7.684926602219835e-06,
+      "loss": 0.0,
+      "step": 4789
+    },
+    {
+      "epoch": 0.30869369079074566,
+      "grad_norm": 0.0022177384041537087,
+      "learning_rate": 7.68421052631579e-06,
+      "loss": 0.0,
+      "step": 4790
+    },
+    {
+      "epoch": 0.3087581362376748,
+      "grad_norm": 0.005044939837968819,
+      "learning_rate": 7.683494450411745e-06,
+      "loss": 0.0,
+      "step": 4791
+    },
+    {
+      "epoch": 0.30882258168460397,
+      "grad_norm": 0.0005836039764449223,
+      "learning_rate": 7.6827783745077e-06,
+      "loss": 0.0,
+      "step": 4792
+    },
+    {
+      "epoch": 0.30888702713153315,
+      "grad_norm": 0.0037036582704997596,
+      "learning_rate": 7.682062298603652e-06,
+      "loss": 0.0,
+      "step": 4793
+    },
+    {
+      "epoch": 0.30895147257846234,
+      "grad_norm": 0.006685573662889619,
+      "learning_rate": 7.681346222699607e-06,
+      "loss": 0.0,
+      "step": 4794
+    },
+    {
+      "epoch": 0.3090159180253915,
+      "grad_norm": 0.0004702842280755588,
+      "learning_rate": 7.680630146795561e-06,
+      "loss": 0.0,
+      "step": 4795
+    },
+    {
+      "epoch": 0.3090803634723207,
+      "grad_norm": 0.041626408650233614,
+      "learning_rate": 7.679914070891515e-06,
+      "loss": 0.0002,
+      "step": 4796
+    },
+    {
+      "epoch": 0.30914480891924984,
+      "grad_norm": 0.0007088733267549225,
+      "learning_rate": 7.67919799498747e-06,
+      "loss": 0.0,
+      "step": 4797
+    },
+    {
+      "epoch": 0.309209254366179,
+      "grad_norm": 0.01583183010603208,
+      "learning_rate": 7.678481919083424e-06,
+      "loss": 0.0,
+      "step": 4798
+    },
+    {
+      "epoch": 0.3092736998131082,
+      "grad_norm": 0.005335127912471382,
+      "learning_rate": 7.677765843179378e-06,
+      "loss": 0.0,
+      "step": 4799
+    },
+    {
+      "epoch": 0.3093381452600374,
+      "grad_norm": 0.0010392895308987736,
+      "learning_rate": 7.677049767275332e-06,
+      "loss": 0.0,
+      "step": 4800
+    },
+    {
+      "epoch": 0.3094025907069666,
+      "grad_norm": 0.04592088900990831,
+      "learning_rate": 7.676333691371287e-06,
+      "loss": 0.0001,
+      "step": 4801
+    },
+    {
+      "epoch": 0.3094670361538957,
+      "grad_norm": 0.05353814637908716,
+      "learning_rate": 7.67561761546724e-06,
+      "loss": 0.0001,
+      "step": 4802
+    },
+    {
+      "epoch": 0.3095314816008249,
+      "grad_norm": 0.004138582846488795,
+      "learning_rate": 7.674901539563194e-06,
+      "loss": 0.0,
+      "step": 4803
+    },
+    {
+      "epoch": 0.30959592704775407,
+      "grad_norm": 0.02532980585580661,
+      "learning_rate": 7.674185463659148e-06,
+      "loss": 0.0002,
+      "step": 4804
+    },
+    {
+      "epoch": 0.30966037249468326,
+      "grad_norm": 0.22175176851799505,
+      "learning_rate": 7.673469387755102e-06,
+      "loss": 0.0006,
+      "step": 4805
+    },
+    {
+      "epoch": 0.30972481794161244,
+      "grad_norm": 0.02077356363290854,
+      "learning_rate": 7.672753311851057e-06,
+      "loss": 0.0001,
+      "step": 4806
+    },
+    {
+      "epoch": 0.3097892633885416,
+      "grad_norm": 0.0009383453196531269,
+      "learning_rate": 7.67203723594701e-06,
+      "loss": 0.0,
+      "step": 4807
+    },
+    {
+      "epoch": 0.30985370883547075,
+      "grad_norm": 0.0021745621033941653,
+      "learning_rate": 7.671321160042965e-06,
+      "loss": 0.0,
+      "step": 4808
+    },
+    {
+      "epoch": 0.30991815428239994,
+      "grad_norm": 0.012943322912088268,
+      "learning_rate": 7.67060508413892e-06,
+      "loss": 0.0001,
+      "step": 4809
+    },
+    {
+      "epoch": 0.3099825997293291,
+      "grad_norm": 1.1511934908108268,
+      "learning_rate": 7.669889008234874e-06,
+      "loss": 0.0127,
+      "step": 4810
+    },
+    {
+      "epoch": 0.3100470451762583,
+      "grad_norm": 0.0009821788142993016,
+      "learning_rate": 7.669172932330828e-06,
+      "loss": 0.0,
+      "step": 4811
+    },
+    {
+      "epoch": 0.3101114906231875,
+      "grad_norm": 0.0007450914756910727,
+      "learning_rate": 7.66845685642678e-06,
+      "loss": 0.0,
+      "step": 4812
+    },
+    {
+      "epoch": 0.3101759360701166,
+      "grad_norm": 0.0015614974705150063,
+      "learning_rate": 7.667740780522735e-06,
+      "loss": 0.0,
+      "step": 4813
+    },
+    {
+      "epoch": 0.3102403815170458,
+      "grad_norm": 0.007181676464111988,
+      "learning_rate": 7.667024704618691e-06,
+      "loss": 0.0,
+      "step": 4814
+    },
+    {
+      "epoch": 0.310304826963975,
+      "grad_norm": 0.019878176986525983,
+      "learning_rate": 7.666308628714645e-06,
+      "loss": 0.0,
+      "step": 4815
+    },
+    {
+      "epoch": 0.3103692724109042,
+      "grad_norm": 0.0002867674669890647,
+      "learning_rate": 7.6655925528106e-06,
+      "loss": 0.0,
+      "step": 4816
+    },
+    {
+      "epoch": 0.31043371785783336,
+      "grad_norm": 1.71531500456341,
+      "learning_rate": 7.664876476906554e-06,
+      "loss": 0.0089,
+      "step": 4817
+    },
+    {
+      "epoch": 0.31049816330476254,
+      "grad_norm": 0.020545370898791864,
+      "learning_rate": 7.664160401002506e-06,
+      "loss": 0.0,
+      "step": 4818
+    },
+    {
+      "epoch": 0.31056260875169167,
+      "grad_norm": 0.0020818963706780835,
+      "learning_rate": 7.66344432509846e-06,
+      "loss": 0.0,
+      "step": 4819
+    },
+    {
+      "epoch": 0.31062705419862086,
+      "grad_norm": 0.0007152623732576892,
+      "learning_rate": 7.662728249194415e-06,
+      "loss": 0.0,
+      "step": 4820
+    },
+    {
+      "epoch": 0.31069149964555004,
+      "grad_norm": 0.0009722607342199007,
+      "learning_rate": 7.66201217329037e-06,
+      "loss": 0.0,
+      "step": 4821
+    },
+    {
+      "epoch": 0.3107559450924792,
+      "grad_norm": 0.023470547345033253,
+      "learning_rate": 7.661296097386324e-06,
+      "loss": 0.0,
+      "step": 4822
+    },
+    {
+      "epoch": 0.3108203905394084,
+      "grad_norm": 0.023717350679689185,
+      "learning_rate": 7.660580021482278e-06,
+      "loss": 0.0001,
+      "step": 4823
+    },
+    {
+      "epoch": 0.31088483598633754,
+      "grad_norm": 0.04864605815166363,
+      "learning_rate": 7.659863945578232e-06,
+      "loss": 0.0001,
+      "step": 4824
+    },
+    {
+      "epoch": 0.3109492814332667,
+      "grad_norm": 0.34392935235481353,
+      "learning_rate": 7.659147869674187e-06,
+      "loss": 0.0009,
+      "step": 4825
+    },
+    {
+      "epoch": 0.3110137268801959,
+      "grad_norm": 0.0010634566579679346,
+      "learning_rate": 7.65843179377014e-06,
+      "loss": 0.0,
+      "step": 4826
+    },
+    {
+      "epoch": 0.3110781723271251,
+      "grad_norm": 0.011746606897081312,
+      "learning_rate": 7.657715717866095e-06,
+      "loss": 0.0,
+      "step": 4827
+    },
+    {
+      "epoch": 0.3111426177740543,
+      "grad_norm": 0.03706126639581151,
+      "learning_rate": 7.656999641962048e-06,
+      "loss": 0.0015,
+      "step": 4828
+    },
+    {
+      "epoch": 0.31120706322098346,
+      "grad_norm": 0.2990590368965242,
+      "learning_rate": 7.656283566058002e-06,
+      "loss": 0.0015,
+      "step": 4829
+    },
+    {
+      "epoch": 0.3112715086679126,
+      "grad_norm": 0.0010528169780914909,
+      "learning_rate": 7.655567490153956e-06,
+      "loss": 0.0,
+      "step": 4830
+    },
+    {
+      "epoch": 0.3113359541148418,
+      "grad_norm": 0.042425665637751686,
+      "learning_rate": 7.65485141424991e-06,
+      "loss": 0.0,
+      "step": 4831
+    },
+    {
+      "epoch": 0.31140039956177096,
+      "grad_norm": 0.001670182439381502,
+      "learning_rate": 7.654135338345865e-06,
+      "loss": 0.0,
+      "step": 4832
+    },
+    {
+      "epoch": 0.31146484500870014,
+      "grad_norm": 0.0009654571293139372,
+      "learning_rate": 7.65341926244182e-06,
+      "loss": 0.0,
+      "step": 4833
+    },
+    {
+      "epoch": 0.3115292904556293,
+      "grad_norm": 0.011802421818662695,
+      "learning_rate": 7.652703186537774e-06,
+      "loss": 0.0,
+      "step": 4834
+    },
+    {
+      "epoch": 0.3115937359025585,
+      "grad_norm": 0.25027796074816694,
+      "learning_rate": 7.651987110633728e-06,
+      "loss": 0.0001,
+      "step": 4835
+    },
+    {
+      "epoch": 0.31165818134948764,
+      "grad_norm": 0.009832647673374325,
+      "learning_rate": 7.651271034729682e-06,
+      "loss": 0.0,
+      "step": 4836
+    },
+    {
+      "epoch": 0.3117226267964168,
+      "grad_norm": 0.29034152459702184,
+      "learning_rate": 7.650554958825636e-06,
+      "loss": 0.0042,
+      "step": 4837
+    },
+    {
+      "epoch": 0.311787072243346,
+      "grad_norm": 0.0031092830812778157,
+      "learning_rate": 7.64983888292159e-06,
+      "loss": 0.0,
+      "step": 4838
+    },
+    {
+      "epoch": 0.3118515176902752,
+      "grad_norm": 0.04667664571851142,
+      "learning_rate": 7.649122807017545e-06,
+      "loss": 0.0003,
+      "step": 4839
+    },
+    {
+      "epoch": 0.3119159631372044,
+      "grad_norm": 0.007606067281312005,
+      "learning_rate": 7.6484067311135e-06,
+      "loss": 0.0,
+      "step": 4840
+    },
+    {
+      "epoch": 0.3119804085841335,
+      "grad_norm": 0.02780274263883628,
+      "learning_rate": 7.647690655209454e-06,
+      "loss": 0.0,
+      "step": 4841
+    },
+    {
+      "epoch": 0.3120448540310627,
+      "grad_norm": 0.023759968008487852,
+      "learning_rate": 7.646974579305408e-06,
+      "loss": 0.0001,
+      "step": 4842
+    },
+    {
+      "epoch": 0.3121092994779919,
+      "grad_norm": 0.0008431761617935588,
+      "learning_rate": 7.646258503401362e-06,
+      "loss": 0.0,
+      "step": 4843
+    },
+    {
+      "epoch": 0.31217374492492106,
+      "grad_norm": 0.015393384023410159,
+      "learning_rate": 7.645542427497315e-06,
+      "loss": 0.0,
+      "step": 4844
+    },
+    {
+      "epoch": 0.31223819037185024,
+      "grad_norm": 0.004651234971706581,
+      "learning_rate": 7.644826351593269e-06,
+      "loss": 0.0,
+      "step": 4845
+    },
+    {
+      "epoch": 0.31230263581877943,
+      "grad_norm": 0.3541639505543996,
+      "learning_rate": 7.644110275689223e-06,
+      "loss": 0.0008,
+      "step": 4846
+    },
+    {
+      "epoch": 0.31236708126570856,
+      "grad_norm": 0.5029302800739834,
+      "learning_rate": 7.643394199785178e-06,
+      "loss": 0.0007,
+      "step": 4847
+    },
+    {
+      "epoch": 0.31243152671263774,
+      "grad_norm": 1.1434380840791063,
+      "learning_rate": 7.642678123881132e-06,
+      "loss": 0.0061,
+      "step": 4848
+    },
+    {
+      "epoch": 0.3124959721595669,
+      "grad_norm": 0.09346831740869416,
+      "learning_rate": 7.641962047977086e-06,
+      "loss": 0.0004,
+      "step": 4849
+    },
+    {
+      "epoch": 0.3125604176064961,
+      "grad_norm": 0.05341842719165527,
+      "learning_rate": 7.64124597207304e-06,
+      "loss": 0.0001,
+      "step": 4850
+    },
+    {
+      "epoch": 0.3126248630534253,
+      "grad_norm": 0.00365746558821454,
+      "learning_rate": 7.640529896168995e-06,
+      "loss": 0.0,
+      "step": 4851
+    },
+    {
+      "epoch": 0.3126893085003544,
+      "grad_norm": 0.004174417788371213,
+      "learning_rate": 7.63981382026495e-06,
+      "loss": 0.0,
+      "step": 4852
+    },
+    {
+      "epoch": 0.3127537539472836,
+      "grad_norm": 0.12209112752832342,
+      "learning_rate": 7.639097744360904e-06,
+      "loss": 0.0006,
+      "step": 4853
+    },
+    {
+      "epoch": 0.3128181993942128,
+      "grad_norm": 4.058324927935555,
+      "learning_rate": 7.638381668456856e-06,
+      "loss": 0.0265,
+      "step": 4854
+    },
+    {
+      "epoch": 0.312882644841142,
+      "grad_norm": 0.006323243901996307,
+      "learning_rate": 7.63766559255281e-06,
+      "loss": 0.0,
+      "step": 4855
+    },
+    {
+      "epoch": 0.31294709028807116,
+      "grad_norm": 0.08450654898943195,
+      "learning_rate": 7.636949516648765e-06,
+      "loss": 0.0003,
+      "step": 4856
+    },
+    {
+      "epoch": 0.31301153573500035,
+      "grad_norm": 0.036397043543195534,
+      "learning_rate": 7.636233440744719e-06,
+      "loss": 0.0,
+      "step": 4857
+    },
+    {
+      "epoch": 0.3130759811819295,
+      "grad_norm": 0.1896315550434969,
+      "learning_rate": 7.635517364840673e-06,
+      "loss": 0.0002,
+      "step": 4858
+    },
+    {
+      "epoch": 0.31314042662885866,
+      "grad_norm": 0.09269444465632841,
+      "learning_rate": 7.634801288936628e-06,
+      "loss": 0.0002,
+      "step": 4859
+    },
+    {
+      "epoch": 0.31320487207578784,
+      "grad_norm": 1.4154748765446705,
+      "learning_rate": 7.634085213032582e-06,
+      "loss": 0.0064,
+      "step": 4860
+    },
+    {
+      "epoch": 0.31326931752271703,
+      "grad_norm": 0.1601978399193165,
+      "learning_rate": 7.633369137128536e-06,
+      "loss": 0.0002,
+      "step": 4861
+    },
+    {
+      "epoch": 0.3133337629696462,
+      "grad_norm": 0.001664992046050716,
+      "learning_rate": 7.63265306122449e-06,
+      "loss": 0.0,
+      "step": 4862
+    },
+    {
+      "epoch": 0.31339820841657534,
+      "grad_norm": 0.005741379817492573,
+      "learning_rate": 7.631936985320445e-06,
+      "loss": 0.0,
+      "step": 4863
+    },
+    {
+      "epoch": 0.3134626538635045,
+      "grad_norm": 0.3956785422217856,
+      "learning_rate": 7.631220909416399e-06,
+      "loss": 0.0017,
+      "step": 4864
+    },
+    {
+      "epoch": 0.3135270993104337,
+      "grad_norm": 0.02680946006956498,
+      "learning_rate": 7.630504833512353e-06,
+      "loss": 0.0,
+      "step": 4865
+    },
+    {
+      "epoch": 0.3135915447573629,
+      "grad_norm": 0.4638488605292008,
+      "learning_rate": 7.629788757608308e-06,
+      "loss": 0.0007,
+      "step": 4866
+    },
+    {
+      "epoch": 0.3136559902042921,
+      "grad_norm": 0.14234355024120665,
+      "learning_rate": 7.629072681704261e-06,
+      "loss": 0.0013,
+      "step": 4867
+    },
+    {
+      "epoch": 0.31372043565122126,
+      "grad_norm": 0.07415628314107933,
+      "learning_rate": 7.6283566058002154e-06,
+      "loss": 0.0001,
+      "step": 4868
+    },
+    {
+      "epoch": 0.3137848810981504,
+      "grad_norm": 0.06906294810141103,
+      "learning_rate": 7.62764052989617e-06,
+      "loss": 0.0001,
+      "step": 4869
+    },
+    {
+      "epoch": 0.3138493265450796,
+      "grad_norm": 0.13154624617698507,
+      "learning_rate": 7.626924453992124e-06,
+      "loss": 0.0018,
+      "step": 4870
+    },
+    {
+      "epoch": 0.31391377199200876,
+      "grad_norm": 1.0724474365407197,
+      "learning_rate": 7.626208378088078e-06,
+      "loss": 0.0054,
+      "step": 4871
+    },
+    {
+      "epoch": 0.31397821743893795,
+      "grad_norm": 0.3422724004482794,
+      "learning_rate": 7.625492302184032e-06,
+      "loss": 0.0024,
+      "step": 4872
+    },
+    {
+      "epoch": 0.31404266288586713,
+      "grad_norm": 0.0016748577841006922,
+      "learning_rate": 7.624776226279986e-06,
+      "loss": 0.0,
+      "step": 4873
+    },
+    {
+      "epoch": 0.3141071083327963,
+      "grad_norm": 0.041638429798124134,
+      "learning_rate": 7.62406015037594e-06,
+      "loss": 0.0001,
+      "step": 4874
+    },
+    {
+      "epoch": 0.31417155377972544,
+      "grad_norm": 0.003857485529783318,
+      "learning_rate": 7.623344074471895e-06,
+      "loss": 0.0,
+      "step": 4875
+    },
+    {
+      "epoch": 0.3142359992266546,
+      "grad_norm": 0.03168840340461246,
+      "learning_rate": 7.622627998567849e-06,
+      "loss": 0.0,
+      "step": 4876
+    },
+    {
+      "epoch": 0.3143004446735838,
+      "grad_norm": 0.006814045902594915,
+      "learning_rate": 7.6219119226638024e-06,
+      "loss": 0.0,
+      "step": 4877
+    },
+    {
+      "epoch": 0.314364890120513,
+      "grad_norm": 0.18182873983099754,
+      "learning_rate": 7.621195846759757e-06,
+      "loss": 0.0003,
+      "step": 4878
+    },
+    {
+      "epoch": 0.3144293355674422,
+      "grad_norm": 0.00901951630925349,
+      "learning_rate": 7.620479770855711e-06,
+      "loss": 0.0,
+      "step": 4879
+    },
+    {
+      "epoch": 0.3144937810143713,
+      "grad_norm": 1.9066806994546444,
+      "learning_rate": 7.619763694951665e-06,
+      "loss": 0.0055,
+      "step": 4880
+    },
+    {
+      "epoch": 0.3145582264613005,
+      "grad_norm": 0.04556863316205216,
+      "learning_rate": 7.61904761904762e-06,
+      "loss": 0.0002,
+      "step": 4881
+    },
+    {
+      "epoch": 0.3146226719082297,
+      "grad_norm": 0.0012398452923164066,
+      "learning_rate": 7.618331543143573e-06,
+      "loss": 0.0,
+      "step": 4882
+    },
+    {
+      "epoch": 0.31468711735515886,
+      "grad_norm": 0.04519157202135707,
+      "learning_rate": 7.617615467239527e-06,
+      "loss": 0.0001,
+      "step": 4883
+    },
+    {
+      "epoch": 0.31475156280208805,
+      "grad_norm": 0.02342436591337108,
+      "learning_rate": 7.6168993913354825e-06,
+      "loss": 0.0017,
+      "step": 4884
+    },
+    {
+      "epoch": 0.31481600824901723,
+      "grad_norm": 0.018536176514005024,
+      "learning_rate": 7.616183315431437e-06,
+      "loss": 0.0002,
+      "step": 4885
+    },
+    {
+      "epoch": 0.31488045369594636,
+      "grad_norm": 0.172805904026286,
+      "learning_rate": 7.615467239527391e-06,
+      "loss": 0.0019,
+      "step": 4886
+    },
+    {
+      "epoch": 0.31494489914287555,
+      "grad_norm": 0.05709872304307744,
+      "learning_rate": 7.614751163623345e-06,
+      "loss": 0.0002,
+      "step": 4887
+    },
+    {
+      "epoch": 0.31500934458980473,
+      "grad_norm": 0.04901532527782089,
+      "learning_rate": 7.614035087719299e-06,
+      "loss": 0.0,
+      "step": 4888
+    },
+    {
+      "epoch": 0.3150737900367339,
+      "grad_norm": 0.045398359407119136,
+      "learning_rate": 7.613319011815253e-06,
+      "loss": 0.0001,
+      "step": 4889
+    },
+    {
+      "epoch": 0.3151382354836631,
+      "grad_norm": 0.01797741643383171,
+      "learning_rate": 7.6126029359112075e-06,
+      "loss": 0.0,
+      "step": 4890
+    },
+    {
+      "epoch": 0.3152026809305922,
+      "grad_norm": 0.003458750480485351,
+      "learning_rate": 7.611886860007162e-06,
+      "loss": 0.0,
+      "step": 4891
+    },
+    {
+      "epoch": 0.3152671263775214,
+      "grad_norm": 0.041268147514423606,
+      "learning_rate": 7.611170784103116e-06,
+      "loss": 0.0003,
+      "step": 4892
+    },
+    {
+      "epoch": 0.3153315718244506,
+      "grad_norm": 0.10172780046470717,
+      "learning_rate": 7.6104547081990695e-06,
+      "loss": 0.0001,
+      "step": 4893
+    },
+    {
+      "epoch": 0.3153960172713798,
+      "grad_norm": 0.0037263810735243897,
+      "learning_rate": 7.609738632295024e-06,
+      "loss": 0.0,
+      "step": 4894
+    },
+    {
+      "epoch": 0.31546046271830896,
+      "grad_norm": 0.0032725613126103575,
+      "learning_rate": 7.609022556390978e-06,
+      "loss": 0.0,
+      "step": 4895
+    },
+    {
+      "epoch": 0.31552490816523815,
+      "grad_norm": 0.09012161182014738,
+      "learning_rate": 7.608306480486932e-06,
+      "loss": 0.0008,
+      "step": 4896
+    },
+    {
+      "epoch": 0.3155893536121673,
+      "grad_norm": 0.09166959806201995,
+      "learning_rate": 7.607590404582887e-06,
+      "loss": 0.0017,
+      "step": 4897
+    },
+    {
+      "epoch": 0.31565379905909646,
+      "grad_norm": 0.011971259717290881,
+      "learning_rate": 7.60687432867884e-06,
+      "loss": 0.0,
+      "step": 4898
+    },
+    {
+      "epoch": 0.31571824450602565,
+      "grad_norm": 0.07688257728135989,
+      "learning_rate": 7.6061582527747945e-06,
+      "loss": 0.0001,
+      "step": 4899
+    },
+    {
+      "epoch": 0.31578268995295483,
+      "grad_norm": 0.000963422072308008,
+      "learning_rate": 7.605442176870749e-06,
+      "loss": 0.0,
+      "step": 4900
+    },
+    {
+      "epoch": 0.315847135399884,
+      "grad_norm": 0.04923102333675415,
+      "learning_rate": 7.604726100966703e-06,
+      "loss": 0.0001,
+      "step": 4901
+    },
+    {
+      "epoch": 0.3159115808468132,
+      "grad_norm": 0.017858915009544632,
+      "learning_rate": 7.604010025062657e-06,
+      "loss": 0.0,
+      "step": 4902
+    },
+    {
+      "epoch": 0.31597602629374233,
+      "grad_norm": 0.0074076590984602534,
+      "learning_rate": 7.603293949158611e-06,
+      "loss": 0.0,
+      "step": 4903
+    },
+    {
+      "epoch": 0.3160404717406715,
+      "grad_norm": 0.004138050993269019,
+      "learning_rate": 7.602577873254565e-06,
+      "loss": 0.0,
+      "step": 4904
+    },
+    {
+      "epoch": 0.3161049171876007,
+      "grad_norm": 0.007657744900865728,
+      "learning_rate": 7.601861797350519e-06,
+      "loss": 0.0001,
+      "step": 4905
+    },
+    {
+      "epoch": 0.3161693626345299,
+      "grad_norm": 0.015958886199231824,
+      "learning_rate": 7.601145721446474e-06,
+      "loss": 0.0001,
+      "step": 4906
+    },
+    {
+      "epoch": 0.31623380808145907,
+      "grad_norm": 0.11232840246170804,
+      "learning_rate": 7.600429645542427e-06,
+      "loss": 0.0003,
+      "step": 4907
+    },
+    {
+      "epoch": 0.3162982535283882,
+      "grad_norm": 0.0019847038300378613,
+      "learning_rate": 7.599713569638383e-06,
+      "loss": 0.0,
+      "step": 4908
+    },
+    {
+      "epoch": 0.3163626989753174,
+      "grad_norm": 0.015164377914120451,
+      "learning_rate": 7.598997493734337e-06,
+      "loss": 0.0,
+      "step": 4909
+    },
+    {
+      "epoch": 0.31642714442224656,
+      "grad_norm": 0.0005348704955189589,
+      "learning_rate": 7.598281417830291e-06,
+      "loss": 0.0,
+      "step": 4910
+    },
+    {
+      "epoch": 0.31649158986917575,
+      "grad_norm": 0.4584278384947784,
+      "learning_rate": 7.597565341926245e-06,
+      "loss": 0.0147,
+      "step": 4911
+    },
+    {
+      "epoch": 0.31655603531610493,
+      "grad_norm": 0.01919581211447712,
+      "learning_rate": 7.5968492660221995e-06,
+      "loss": 0.0,
+      "step": 4912
+    },
+    {
+      "epoch": 0.3166204807630341,
+      "grad_norm": 0.0255280418873319,
+      "learning_rate": 7.596133190118154e-06,
+      "loss": 0.0001,
+      "step": 4913
+    },
+    {
+      "epoch": 0.31668492620996325,
+      "grad_norm": 0.022365740209846784,
+      "learning_rate": 7.595417114214107e-06,
+      "loss": 0.0,
+      "step": 4914
+    },
+    {
+      "epoch": 0.31674937165689243,
+      "grad_norm": 0.0017737070032865822,
+      "learning_rate": 7.5947010383100616e-06,
+      "loss": 0.0,
+      "step": 4915
+    },
+    {
+      "epoch": 0.3168138171038216,
+      "grad_norm": 0.000988195283464662,
+      "learning_rate": 7.593984962406016e-06,
+      "loss": 0.0,
+      "step": 4916
+    },
+    {
+      "epoch": 0.3168782625507508,
+      "grad_norm": 0.11666120524551912,
+      "learning_rate": 7.59326888650197e-06,
+      "loss": 0.0002,
+      "step": 4917
+    },
+    {
+      "epoch": 0.31694270799768,
+      "grad_norm": 0.02640212358308717,
+      "learning_rate": 7.5925528105979244e-06,
+      "loss": 0.0001,
+      "step": 4918
+    },
+    {
+      "epoch": 0.3170071534446091,
+      "grad_norm": 0.0005211830684490821,
+      "learning_rate": 7.591836734693878e-06,
+      "loss": 0.0,
+      "step": 4919
+    },
+    {
+      "epoch": 0.3170715988915383,
+      "grad_norm": 0.007234303323305554,
+      "learning_rate": 7.591120658789832e-06,
+      "loss": 0.0,
+      "step": 4920
+    },
+    {
+      "epoch": 0.3171360443384675,
+      "grad_norm": 0.07723608651426664,
+      "learning_rate": 7.5904045828857865e-06,
+      "loss": 0.0002,
+      "step": 4921
+    },
+    {
+      "epoch": 0.31720048978539667,
+      "grad_norm": 0.0030668760381343594,
+      "learning_rate": 7.589688506981741e-06,
+      "loss": 0.0,
+      "step": 4922
+    },
+    {
+      "epoch": 0.31726493523232585,
+      "grad_norm": 0.08688448284729575,
+      "learning_rate": 7.588972431077694e-06,
+      "loss": 0.0002,
+      "step": 4923
+    },
+    {
+      "epoch": 0.31732938067925504,
+      "grad_norm": 0.06115100927977217,
+      "learning_rate": 7.5882563551736485e-06,
+      "loss": 0.0005,
+      "step": 4924
+    },
+    {
+      "epoch": 0.31739382612618416,
+      "grad_norm": 0.004299995558535509,
+      "learning_rate": 7.587540279269603e-06,
+      "loss": 0.0,
+      "step": 4925
+    },
+    {
+      "epoch": 0.31745827157311335,
+      "grad_norm": 0.010539844593121827,
+      "learning_rate": 7.586824203365557e-06,
+      "loss": 0.0,
+      "step": 4926
+    },
+    {
+      "epoch": 0.31752271702004253,
+      "grad_norm": 0.0030391662830343918,
+      "learning_rate": 7.5861081274615114e-06,
+      "loss": 0.0,
+      "step": 4927
+    },
+    {
+      "epoch": 0.3175871624669717,
+      "grad_norm": 0.03277884812642534,
+      "learning_rate": 7.585392051557465e-06,
+      "loss": 0.0001,
+      "step": 4928
+    },
+    {
+      "epoch": 0.3176516079139009,
+      "grad_norm": 0.03166498928316155,
+      "learning_rate": 7.584675975653419e-06,
+      "loss": 0.0001,
+      "step": 4929
+    },
+    {
+      "epoch": 0.31771605336083003,
+      "grad_norm": 0.005663952936868069,
+      "learning_rate": 7.5839598997493735e-06,
+      "loss": 0.0,
+      "step": 4930
+    },
+    {
+      "epoch": 0.3177804988077592,
+      "grad_norm": 0.016582148083116855,
+      "learning_rate": 7.583243823845329e-06,
+      "loss": 0.0,
+      "step": 4931
+    },
+    {
+      "epoch": 0.3178449442546884,
+      "grad_norm": 0.20244802858863004,
+      "learning_rate": 7.582527747941283e-06,
+      "loss": 0.0004,
+      "step": 4932
+    },
+    {
+      "epoch": 0.3179093897016176,
+      "grad_norm": 0.001806635185564558,
+      "learning_rate": 7.581811672037237e-06,
+      "loss": 0.0,
+      "step": 4933
+    },
+    {
+      "epoch": 0.31797383514854677,
+      "grad_norm": 1.1378611169584365,
+      "learning_rate": 7.5810955961331915e-06,
+      "loss": 0.0016,
+      "step": 4934
+    },
+    {
+      "epoch": 0.31803828059547595,
+      "grad_norm": 0.10410084035311608,
+      "learning_rate": 7.580379520229145e-06,
+      "loss": 0.0002,
+      "step": 4935
+    },
+    {
+      "epoch": 0.3181027260424051,
+      "grad_norm": 0.0009001930670387458,
+      "learning_rate": 7.579663444325099e-06,
+      "loss": 0.0,
+      "step": 4936
+    },
+    {
+      "epoch": 0.31816717148933427,
+      "grad_norm": 0.00022304137717573243,
+      "learning_rate": 7.578947368421054e-06,
+      "loss": 0.0,
+      "step": 4937
+    },
+    {
+      "epoch": 0.31823161693626345,
+      "grad_norm": 0.021512349972738248,
+      "learning_rate": 7.578231292517008e-06,
+      "loss": 0.0,
+      "step": 4938
+    },
+    {
+      "epoch": 0.31829606238319264,
+      "grad_norm": 0.10591603729833651,
+      "learning_rate": 7.577515216612961e-06,
+      "loss": 0.0002,
+      "step": 4939
+    },
+    {
+      "epoch": 0.3183605078301218,
+      "grad_norm": 0.013542071059994441,
+      "learning_rate": 7.576799140708916e-06,
+      "loss": 0.0001,
+      "step": 4940
+    },
+    {
+      "epoch": 0.318424953277051,
+      "grad_norm": 0.00536616140223203,
+      "learning_rate": 7.57608306480487e-06,
+      "loss": 0.0,
+      "step": 4941
+    },
+    {
+      "epoch": 0.31848939872398013,
+      "grad_norm": 0.01473934950136891,
+      "learning_rate": 7.575366988900824e-06,
+      "loss": 0.0001,
+      "step": 4942
+    },
+    {
+      "epoch": 0.3185538441709093,
+      "grad_norm": 0.02923624998055832,
+      "learning_rate": 7.5746509129967785e-06,
+      "loss": 0.0001,
+      "step": 4943
+    },
+    {
+      "epoch": 0.3186182896178385,
+      "grad_norm": 0.00951459181256414,
+      "learning_rate": 7.573934837092732e-06,
+      "loss": 0.0001,
+      "step": 4944
+    },
+    {
+      "epoch": 0.3186827350647677,
+      "grad_norm": 0.002520293197353339,
+      "learning_rate": 7.573218761188686e-06,
+      "loss": 0.0,
+      "step": 4945
+    },
+    {
+      "epoch": 0.31874718051169687,
+      "grad_norm": 0.0006642405582552634,
+      "learning_rate": 7.5725026852846406e-06,
+      "loss": 0.0,
+      "step": 4946
+    },
+    {
+      "epoch": 0.318811625958626,
+      "grad_norm": 0.0023286591485846588,
+      "learning_rate": 7.571786609380595e-06,
+      "loss": 0.0,
+      "step": 4947
+    },
+    {
+      "epoch": 0.3188760714055552,
+      "grad_norm": 0.010581924490633903,
+      "learning_rate": 7.571070533476549e-06,
+      "loss": 0.0001,
+      "step": 4948
+    },
+    {
+      "epoch": 0.31894051685248437,
+      "grad_norm": 0.5570829541365085,
+      "learning_rate": 7.570354457572503e-06,
+      "loss": 0.0034,
+      "step": 4949
+    },
+    {
+      "epoch": 0.31900496229941355,
+      "grad_norm": 0.01401707071364209,
+      "learning_rate": 7.569638381668457e-06,
+      "loss": 0.0,
+      "step": 4950
+    },
+    {
+      "epoch": 0.31906940774634274,
+      "grad_norm": 0.0011558668939177993,
+      "learning_rate": 7.568922305764411e-06,
+      "loss": 0.0,
+      "step": 4951
+    },
+    {
+      "epoch": 0.3191338531932719,
+      "grad_norm": 0.2016895448269546,
+      "learning_rate": 7.5682062298603655e-06,
+      "loss": 0.0011,
+      "step": 4952
+    },
+    {
+      "epoch": 0.31919829864020105,
+      "grad_norm": 1.4378902071553488,
+      "learning_rate": 7.56749015395632e-06,
+      "loss": 0.0022,
+      "step": 4953
+    },
+    {
+      "epoch": 0.31926274408713023,
+      "grad_norm": 0.002368113513227609,
+      "learning_rate": 7.566774078052275e-06,
+      "loss": 0.0,
+      "step": 4954
+    },
+    {
+      "epoch": 0.3193271895340594,
+      "grad_norm": 0.0015807385536410412,
+      "learning_rate": 7.566058002148228e-06,
+      "loss": 0.0,
+      "step": 4955
+    },
+    {
+      "epoch": 0.3193916349809886,
+      "grad_norm": 0.023490105392632896,
+      "learning_rate": 7.565341926244183e-06,
+      "loss": 0.0002,
+      "step": 4956
+    },
+    {
+      "epoch": 0.3194560804279178,
+      "grad_norm": 0.003028390952271756,
+      "learning_rate": 7.564625850340137e-06,
+      "loss": 0.0,
+      "step": 4957
+    },
+    {
+      "epoch": 0.3195205258748469,
+      "grad_norm": 0.02639448995249658,
+      "learning_rate": 7.563909774436091e-06,
+      "loss": 0.0001,
+      "step": 4958
+    },
+    {
+      "epoch": 0.3195849713217761,
+      "grad_norm": 0.0002196733853071634,
+      "learning_rate": 7.563193698532046e-06,
+      "loss": 0.0,
+      "step": 4959
+    },
+    {
+      "epoch": 0.3196494167687053,
+      "grad_norm": 0.00525486369614464,
+      "learning_rate": 7.562477622627999e-06,
+      "loss": 0.0,
+      "step": 4960
+    },
+    {
+      "epoch": 0.31971386221563447,
+      "grad_norm": 6.02780353327736,
+      "learning_rate": 7.561761546723953e-06,
+      "loss": 0.0112,
+      "step": 4961
+    },
+    {
+      "epoch": 0.31977830766256365,
+      "grad_norm": 0.027587832602718406,
+      "learning_rate": 7.561045470819908e-06,
+      "loss": 0.0002,
+      "step": 4962
+    },
+    {
+      "epoch": 0.31984275310949284,
+      "grad_norm": 0.0010843918612778584,
+      "learning_rate": 7.560329394915862e-06,
+      "loss": 0.0,
+      "step": 4963
+    },
+    {
+      "epoch": 0.31990719855642197,
+      "grad_norm": 0.019770665104542517,
+      "learning_rate": 7.559613319011816e-06,
+      "loss": 0.0001,
+      "step": 4964
+    },
+    {
+      "epoch": 0.31997164400335115,
+      "grad_norm": 0.0004921501486983964,
+      "learning_rate": 7.55889724310777e-06,
+      "loss": 0.0,
+      "step": 4965
+    },
+    {
+      "epoch": 0.32003608945028034,
+      "grad_norm": 0.06938545379726922,
+      "learning_rate": 7.558181167203724e-06,
+      "loss": 0.0001,
+      "step": 4966
+    },
+    {
+      "epoch": 0.3201005348972095,
+      "grad_norm": 0.004667437215269022,
+      "learning_rate": 7.557465091299678e-06,
+      "loss": 0.0,
+      "step": 4967
+    },
+    {
+      "epoch": 0.3201649803441387,
+      "grad_norm": 0.014288407805833092,
+      "learning_rate": 7.556749015395633e-06,
+      "loss": 0.0001,
+      "step": 4968
+    },
+    {
+      "epoch": 0.32022942579106783,
+      "grad_norm": 0.008945516329945936,
+      "learning_rate": 7.556032939491587e-06,
+      "loss": 0.0001,
+      "step": 4969
+    },
+    {
+      "epoch": 0.320293871237997,
+      "grad_norm": 0.00023568888504286596,
+      "learning_rate": 7.55531686358754e-06,
+      "loss": 0.0,
+      "step": 4970
+    },
+    {
+      "epoch": 0.3203583166849262,
+      "grad_norm": 0.01571903394353528,
+      "learning_rate": 7.554600787683495e-06,
+      "loss": 0.0001,
+      "step": 4971
+    },
+    {
+      "epoch": 0.3204227621318554,
+      "grad_norm": 0.00460776685039601,
+      "learning_rate": 7.553884711779449e-06,
+      "loss": 0.0,
+      "step": 4972
+    },
+    {
+      "epoch": 0.32048720757878457,
+      "grad_norm": 0.04010418068407689,
+      "learning_rate": 7.553168635875403e-06,
+      "loss": 0.0001,
+      "step": 4973
+    },
+    {
+      "epoch": 0.32055165302571376,
+      "grad_norm": 0.00035000815771177833,
+      "learning_rate": 7.5524525599713576e-06,
+      "loss": 0.0,
+      "step": 4974
+    },
+    {
+      "epoch": 0.3206160984726429,
+      "grad_norm": 0.0024140885061416342,
+      "learning_rate": 7.551736484067311e-06,
+      "loss": 0.0,
+      "step": 4975
+    },
+    {
+      "epoch": 0.32068054391957207,
+      "grad_norm": 0.007612368659569762,
+      "learning_rate": 7.551020408163265e-06,
+      "loss": 0.0,
+      "step": 4976
+    },
+    {
+      "epoch": 0.32074498936650125,
+      "grad_norm": 0.001607101033468048,
+      "learning_rate": 7.55030433225922e-06,
+      "loss": 0.0,
+      "step": 4977
+    },
+    {
+      "epoch": 0.32080943481343044,
+      "grad_norm": 0.0698845675905335,
+      "learning_rate": 7.549588256355175e-06,
+      "loss": 0.0001,
+      "step": 4978
+    },
+    {
+      "epoch": 0.3208738802603596,
+      "grad_norm": 0.00011869231645262834,
+      "learning_rate": 7.548872180451129e-06,
+      "loss": 0.0,
+      "step": 4979
+    },
+    {
+      "epoch": 0.3209383257072888,
+      "grad_norm": 0.002316536849549659,
+      "learning_rate": 7.548156104547083e-06,
+      "loss": 0.0,
+      "step": 4980
+    },
+    {
+      "epoch": 0.32100277115421794,
+      "grad_norm": 0.2341897152622086,
+      "learning_rate": 7.547440028643037e-06,
+      "loss": 0.0015,
+      "step": 4981
+    },
+    {
+      "epoch": 0.3210672166011471,
+      "grad_norm": 0.28748342580289754,
+      "learning_rate": 7.546723952738991e-06,
+      "loss": 0.0002,
+      "step": 4982
+    },
+    {
+      "epoch": 0.3211316620480763,
+      "grad_norm": 0.0005328684311772538,
+      "learning_rate": 7.546007876834945e-06,
+      "loss": 0.0,
+      "step": 4983
+    },
+    {
+      "epoch": 0.3211961074950055,
+      "grad_norm": 0.02484436391275856,
+      "learning_rate": 7.5452918009309e-06,
+      "loss": 0.0001,
+      "step": 4984
+    },
+    {
+      "epoch": 0.3212605529419347,
+      "grad_norm": 0.008556152524764708,
+      "learning_rate": 7.544575725026854e-06,
+      "loss": 0.0001,
+      "step": 4985
+    },
+    {
+      "epoch": 0.3213249983888638,
+      "grad_norm": 0.038374263358831415,
+      "learning_rate": 7.5438596491228074e-06,
+      "loss": 0.0003,
+      "step": 4986
+    },
+    {
+      "epoch": 0.321389443835793,
+      "grad_norm": 0.0023323979630988576,
+      "learning_rate": 7.543143573218762e-06,
+      "loss": 0.0,
+      "step": 4987
+    },
+    {
+      "epoch": 0.32145388928272217,
+      "grad_norm": 0.0018241033879360137,
+      "learning_rate": 7.542427497314716e-06,
+      "loss": 0.0,
+      "step": 4988
+    },
+    {
+      "epoch": 0.32151833472965136,
+      "grad_norm": 0.014841046251769881,
+      "learning_rate": 7.54171142141067e-06,
+      "loss": 0.0001,
+      "step": 4989
+    },
+    {
+      "epoch": 0.32158278017658054,
+      "grad_norm": 0.00044722802438275366,
+      "learning_rate": 7.540995345506625e-06,
+      "loss": 0.0,
+      "step": 4990
+    },
+    {
+      "epoch": 0.3216472256235097,
+      "grad_norm": 0.012483224214803168,
+      "learning_rate": 7.540279269602578e-06,
+      "loss": 0.0,
+      "step": 4991
+    },
+    {
+      "epoch": 0.32171167107043885,
+      "grad_norm": 0.0073138178050027175,
+      "learning_rate": 7.539563193698532e-06,
+      "loss": 0.0,
+      "step": 4992
+    },
+    {
+      "epoch": 0.32177611651736804,
+      "grad_norm": 0.1553142062233551,
+      "learning_rate": 7.538847117794487e-06,
+      "loss": 0.0002,
+      "step": 4993
+    },
+    {
+      "epoch": 0.3218405619642972,
+      "grad_norm": 0.0007977184599504469,
+      "learning_rate": 7.538131041890441e-06,
+      "loss": 0.0,
+      "step": 4994
+    },
+    {
+      "epoch": 0.3219050074112264,
+      "grad_norm": 0.0020614767277034192,
+      "learning_rate": 7.537414965986395e-06,
+      "loss": 0.0,
+      "step": 4995
+    },
+    {
+      "epoch": 0.3219694528581556,
+      "grad_norm": 0.00481232716002087,
+      "learning_rate": 7.536698890082349e-06,
+      "loss": 0.0,
+      "step": 4996
+    },
+    {
+      "epoch": 0.3220338983050847,
+      "grad_norm": 0.004769241063000902,
+      "learning_rate": 7.535982814178303e-06,
+      "loss": 0.0001,
+      "step": 4997
+    },
+    {
+      "epoch": 0.3220983437520139,
+      "grad_norm": 0.000924509781990204,
+      "learning_rate": 7.535266738274257e-06,
+      "loss": 0.0,
+      "step": 4998
+    },
+    {
+      "epoch": 0.3221627891989431,
+      "grad_norm": 0.002373089335570934,
+      "learning_rate": 7.534550662370212e-06,
+      "loss": 0.0,
+      "step": 4999
+    },
+    {
+      "epoch": 0.3222272346458723,
+      "grad_norm": 2.5394377915254367,
+      "learning_rate": 7.533834586466165e-06,
+      "loss": 0.0116,
+      "step": 5000
+    },
+    {
+      "epoch": 0.32229168009280146,
+      "grad_norm": 2.836771188312746,
+      "learning_rate": 7.533118510562121e-06,
+      "loss": 0.0571,
+      "step": 5001
+    },
+    {
+      "epoch": 0.32235612553973064,
+      "grad_norm": 0.0867760639624333,
+      "learning_rate": 7.5324024346580745e-06,
+      "loss": 0.0004,
+      "step": 5002
+    },
+    {
+      "epoch": 0.32242057098665977,
+      "grad_norm": 0.009055324801599299,
+      "learning_rate": 7.531686358754029e-06,
+      "loss": 0.0001,
+      "step": 5003
+    },
+    {
+      "epoch": 0.32248501643358896,
+      "grad_norm": 0.0007850045675465759,
+      "learning_rate": 7.530970282849983e-06,
+      "loss": 0.0,
+      "step": 5004
+    },
+    {
+      "epoch": 0.32254946188051814,
+      "grad_norm": 0.0054402043559637694,
+      "learning_rate": 7.530254206945937e-06,
+      "loss": 0.0001,
+      "step": 5005
+    },
+    {
+      "epoch": 0.3226139073274473,
+      "grad_norm": 0.0003870119441539209,
+      "learning_rate": 7.529538131041892e-06,
+      "loss": 0.0,
+      "step": 5006
+    },
+    {
+      "epoch": 0.3226783527743765,
+      "grad_norm": 0.002394564450146463,
+      "learning_rate": 7.528822055137845e-06,
+      "loss": 0.0,
+      "step": 5007
+    },
+    {
+      "epoch": 0.32274279822130564,
+      "grad_norm": 0.0009554164590431097,
+      "learning_rate": 7.5281059792337995e-06,
+      "loss": 0.0,
+      "step": 5008
+    },
+    {
+      "epoch": 0.3228072436682348,
+      "grad_norm": 0.018775061036906286,
+      "learning_rate": 7.527389903329754e-06,
+      "loss": 0.0001,
+      "step": 5009
+    },
+    {
+      "epoch": 0.322871689115164,
+      "grad_norm": 0.0004067308282074071,
+      "learning_rate": 7.526673827425708e-06,
+      "loss": 0.0,
+      "step": 5010
+    },
+    {
+      "epoch": 0.3229361345620932,
+      "grad_norm": 0.0022833677798131395,
+      "learning_rate": 7.525957751521662e-06,
+      "loss": 0.0,
+      "step": 5011
+    },
+    {
+      "epoch": 0.3230005800090224,
+      "grad_norm": 0.0034419373129303036,
+      "learning_rate": 7.525241675617616e-06,
+      "loss": 0.0,
+      "step": 5012
+    },
+    {
+      "epoch": 0.32306502545595156,
+      "grad_norm": 0.0013269891500375182,
+      "learning_rate": 7.52452559971357e-06,
+      "loss": 0.0,
+      "step": 5013
+    },
+    {
+      "epoch": 0.3231294709028807,
+      "grad_norm": 0.005102314456417867,
+      "learning_rate": 7.523809523809524e-06,
+      "loss": 0.0,
+      "step": 5014
+    },
+    {
+      "epoch": 0.3231939163498099,
+      "grad_norm": 0.0011584206982300578,
+      "learning_rate": 7.523093447905479e-06,
+      "loss": 0.0,
+      "step": 5015
+    },
+    {
+      "epoch": 0.32325836179673906,
+      "grad_norm": 0.004096714247036613,
+      "learning_rate": 7.522377372001432e-06,
+      "loss": 0.0,
+      "step": 5016
+    },
+    {
+      "epoch": 0.32332280724366824,
+      "grad_norm": 0.02179641642182101,
+      "learning_rate": 7.5216612960973865e-06,
+      "loss": 0.0001,
+      "step": 5017
+    },
+    {
+      "epoch": 0.3233872526905974,
+      "grad_norm": 0.0008365472971686693,
+      "learning_rate": 7.520945220193341e-06,
+      "loss": 0.0,
+      "step": 5018
+    },
+    {
+      "epoch": 0.3234516981375266,
+      "grad_norm": 0.06371238259548284,
+      "learning_rate": 7.520229144289295e-06,
+      "loss": 0.0001,
+      "step": 5019
+    },
+    {
+      "epoch": 0.32351614358445574,
+      "grad_norm": 0.009985623889348058,
+      "learning_rate": 7.519513068385249e-06,
+      "loss": 0.0,
+      "step": 5020
+    },
+    {
+      "epoch": 0.3235805890313849,
+      "grad_norm": 0.007412286860036669,
+      "learning_rate": 7.518796992481203e-06,
+      "loss": 0.0001,
+      "step": 5021
+    },
+    {
+      "epoch": 0.3236450344783141,
+      "grad_norm": 1.3540490710721447,
+      "learning_rate": 7.518080916577157e-06,
+      "loss": 0.0054,
+      "step": 5022
+    },
+    {
+      "epoch": 0.3237094799252433,
+      "grad_norm": 0.000858304434348953,
+      "learning_rate": 7.517364840673111e-06,
+      "loss": 0.0,
+      "step": 5023
+    },
+    {
+      "epoch": 0.3237739253721725,
+      "grad_norm": 0.0041181096965707915,
+      "learning_rate": 7.5166487647690666e-06,
+      "loss": 0.0,
+      "step": 5024
+    },
+    {
+      "epoch": 0.3238383708191016,
+      "grad_norm": 0.0003157757437743913,
+      "learning_rate": 7.515932688865021e-06,
+      "loss": 0.0,
+      "step": 5025
+    },
+    {
+      "epoch": 0.3239028162660308,
+      "grad_norm": 0.06434296094244776,
+      "learning_rate": 7.515216612960975e-06,
+      "loss": 0.0003,
+      "step": 5026
+    },
+    {
+      "epoch": 0.32396726171296,
+      "grad_norm": 0.3501482917633669,
+      "learning_rate": 7.5145005370569295e-06,
+      "loss": 0.0012,
+      "step": 5027
+    },
+    {
+      "epoch": 0.32403170715988916,
+      "grad_norm": 0.003286186092908012,
+      "learning_rate": 7.513784461152883e-06,
+      "loss": 0.0,
+      "step": 5028
+    },
+    {
+      "epoch": 0.32409615260681834,
+      "grad_norm": 0.006841491514237018,
+      "learning_rate": 7.513068385248837e-06,
+      "loss": 0.0,
+      "step": 5029
+    },
+    {
+      "epoch": 0.32416059805374753,
+      "grad_norm": 0.0026199554590787223,
+      "learning_rate": 7.5123523093447915e-06,
+      "loss": 0.0,
+      "step": 5030
+    },
+    {
+      "epoch": 0.32422504350067666,
+      "grad_norm": 0.002010553894782923,
+      "learning_rate": 7.511636233440746e-06,
+      "loss": 0.0,
+      "step": 5031
+    },
+    {
+      "epoch": 0.32428948894760584,
+      "grad_norm": 0.01111582230251598,
+      "learning_rate": 7.510920157536699e-06,
+      "loss": 0.0,
+      "step": 5032
+    },
+    {
+      "epoch": 0.324353934394535,
+      "grad_norm": 0.012234168882052655,
+      "learning_rate": 7.5102040816326536e-06,
+      "loss": 0.0,
+      "step": 5033
+    },
+    {
+      "epoch": 0.3244183798414642,
+      "grad_norm": 0.24919705042346788,
+      "learning_rate": 7.509488005728608e-06,
+      "loss": 0.0018,
+      "step": 5034
+    },
+    {
+      "epoch": 0.3244828252883934,
+      "grad_norm": 0.0015482715286001473,
+      "learning_rate": 7.508771929824562e-06,
+      "loss": 0.0,
+      "step": 5035
+    },
+    {
+      "epoch": 0.3245472707353225,
+      "grad_norm": 0.004696018029830818,
+      "learning_rate": 7.5080558539205164e-06,
+      "loss": 0.0,
+      "step": 5036
+    },
+    {
+      "epoch": 0.3246117161822517,
+      "grad_norm": 0.014119543213684357,
+      "learning_rate": 7.50733977801647e-06,
+      "loss": 0.0,
+      "step": 5037
+    },
+    {
+      "epoch": 0.3246761616291809,
+      "grad_norm": 0.3867381890545684,
+      "learning_rate": 7.506623702112424e-06,
+      "loss": 0.0025,
+      "step": 5038
+    },
+    {
+      "epoch": 0.3247406070761101,
+      "grad_norm": 0.00042987711144992877,
+      "learning_rate": 7.5059076262083785e-06,
+      "loss": 0.0,
+      "step": 5039
+    },
+    {
+      "epoch": 0.32480505252303926,
+      "grad_norm": 0.2663883013006666,
+      "learning_rate": 7.505191550304333e-06,
+      "loss": 0.0002,
+      "step": 5040
+    },
+    {
+      "epoch": 0.32486949796996845,
+      "grad_norm": 0.0004538031013607848,
+      "learning_rate": 7.504475474400287e-06,
+      "loss": 0.0,
+      "step": 5041
+    },
+    {
+      "epoch": 0.3249339434168976,
+      "grad_norm": 0.001688240594975126,
+      "learning_rate": 7.5037593984962405e-06,
+      "loss": 0.0,
+      "step": 5042
+    },
+    {
+      "epoch": 0.32499838886382676,
+      "grad_norm": 0.0006231698876898269,
+      "learning_rate": 7.503043322592195e-06,
+      "loss": 0.0,
+      "step": 5043
+    },
+    {
+      "epoch": 0.32506283431075594,
+      "grad_norm": 0.001326850866979831,
+      "learning_rate": 7.502327246688149e-06,
+      "loss": 0.0,
+      "step": 5044
+    },
+    {
+      "epoch": 0.32512727975768513,
+      "grad_norm": 0.09123642767439222,
+      "learning_rate": 7.5016111707841034e-06,
+      "loss": 0.0003,
+      "step": 5045
+    },
+    {
+      "epoch": 0.3251917252046143,
+      "grad_norm": 0.00031140489873245536,
+      "learning_rate": 7.500895094880058e-06,
+      "loss": 0.0,
+      "step": 5046
+    },
+    {
+      "epoch": 0.32525617065154344,
+      "grad_norm": 0.010423918369235648,
+      "learning_rate": 7.500179018976011e-06,
+      "loss": 0.0001,
+      "step": 5047
+    },
+    {
+      "epoch": 0.3253206160984726,
+      "grad_norm": 0.027732214205477675,
+      "learning_rate": 7.499462943071967e-06,
+      "loss": 0.0001,
+      "step": 5048
+    },
+    {
+      "epoch": 0.3253850615454018,
+      "grad_norm": 0.0008578557245992876,
+      "learning_rate": 7.498746867167921e-06,
+      "loss": 0.0,
+      "step": 5049
+    },
+    {
+      "epoch": 0.325449506992331,
+      "grad_norm": 0.015105272515006055,
+      "learning_rate": 7.498030791263875e-06,
+      "loss": 0.0001,
+      "step": 5050
+    },
+    {
+      "epoch": 0.3255139524392602,
+      "grad_norm": 0.3200695232774652,
+      "learning_rate": 7.497314715359829e-06,
+      "loss": 0.0018,
+      "step": 5051
+    },
+    {
+      "epoch": 0.32557839788618936,
+      "grad_norm": 0.06262675857097338,
+      "learning_rate": 7.4965986394557835e-06,
+      "loss": 0.0001,
+      "step": 5052
+    },
+    {
+      "epoch": 0.3256428433331185,
+      "grad_norm": 0.008279700894868208,
+      "learning_rate": 7.495882563551737e-06,
+      "loss": 0.0,
+      "step": 5053
+    },
+    {
+      "epoch": 0.3257072887800477,
+      "grad_norm": 0.01809988351456632,
+      "learning_rate": 7.495166487647691e-06,
+      "loss": 0.0001,
+      "step": 5054
+    },
+    {
+      "epoch": 0.32577173422697686,
+      "grad_norm": 0.00032587838609959163,
+      "learning_rate": 7.494450411743646e-06,
+      "loss": 0.0,
+      "step": 5055
+    },
+    {
+      "epoch": 0.32583617967390605,
+      "grad_norm": 0.16255369537207687,
+      "learning_rate": 7.4937343358396e-06,
+      "loss": 0.0021,
+      "step": 5056
+    },
+    {
+      "epoch": 0.32590062512083523,
+      "grad_norm": 0.008922854562315093,
+      "learning_rate": 7.493018259935554e-06,
+      "loss": 0.0,
+      "step": 5057
+    },
+    {
+      "epoch": 0.3259650705677644,
+      "grad_norm": 0.002154537714572031,
+      "learning_rate": 7.492302184031508e-06,
+      "loss": 0.0,
+      "step": 5058
+    },
+    {
+      "epoch": 0.32602951601469354,
+      "grad_norm": 0.0004839898917475037,
+      "learning_rate": 7.491586108127462e-06,
+      "loss": 0.0,
+      "step": 5059
+    },
+    {
+      "epoch": 0.32609396146162273,
+      "grad_norm": 0.0004548610049905596,
+      "learning_rate": 7.490870032223416e-06,
+      "loss": 0.0,
+      "step": 5060
+    },
+    {
+      "epoch": 0.3261584069085519,
+      "grad_norm": 0.44944039557347387,
+      "learning_rate": 7.4901539563193705e-06,
+      "loss": 0.0026,
+      "step": 5061
+    },
+    {
+      "epoch": 0.3262228523554811,
+      "grad_norm": 0.00016835646080685742,
+      "learning_rate": 7.489437880415325e-06,
+      "loss": 0.0,
+      "step": 5062
+    },
+    {
+      "epoch": 0.3262872978024103,
+      "grad_norm": 0.29209044726483824,
+      "learning_rate": 7.488721804511278e-06,
+      "loss": 0.0009,
+      "step": 5063
+    },
+    {
+      "epoch": 0.3263517432493394,
+      "grad_norm": 0.0001249842603296567,
+      "learning_rate": 7.4880057286072326e-06,
+      "loss": 0.0,
+      "step": 5064
+    },
+    {
+      "epoch": 0.3264161886962686,
+      "grad_norm": 0.011162206390915428,
+      "learning_rate": 7.487289652703187e-06,
+      "loss": 0.0,
+      "step": 5065
+    },
+    {
+      "epoch": 0.3264806341431978,
+      "grad_norm": 0.005577277546888436,
+      "learning_rate": 7.486573576799141e-06,
+      "loss": 0.0001,
+      "step": 5066
+    },
+    {
+      "epoch": 0.32654507959012696,
+      "grad_norm": 0.008560459027039776,
+      "learning_rate": 7.4858575008950955e-06,
+      "loss": 0.0,
+      "step": 5067
+    },
+    {
+      "epoch": 0.32660952503705615,
+      "grad_norm": 0.22203547121366796,
+      "learning_rate": 7.485141424991049e-06,
+      "loss": 0.0013,
+      "step": 5068
+    },
+    {
+      "epoch": 0.32667397048398533,
+      "grad_norm": 0.1355473815182943,
+      "learning_rate": 7.484425349087003e-06,
+      "loss": 0.0019,
+      "step": 5069
+    },
+    {
+      "epoch": 0.32673841593091446,
+      "grad_norm": 0.0003878992249302429,
+      "learning_rate": 7.4837092731829575e-06,
+      "loss": 0.0,
+      "step": 5070
+    },
+    {
+      "epoch": 0.32680286137784365,
+      "grad_norm": 0.0011205206516367939,
+      "learning_rate": 7.482993197278913e-06,
+      "loss": 0.0,
+      "step": 5071
+    },
+    {
+      "epoch": 0.32686730682477283,
+      "grad_norm": 0.00033044966655028444,
+      "learning_rate": 7.482277121374867e-06,
+      "loss": 0.0,
+      "step": 5072
+    },
+    {
+      "epoch": 0.326931752271702,
+      "grad_norm": 0.02445801950067175,
+      "learning_rate": 7.481561045470821e-06,
+      "loss": 0.0001,
+      "step": 5073
+    },
+    {
+      "epoch": 0.3269961977186312,
+      "grad_norm": 0.012846676525802831,
+      "learning_rate": 7.480844969566775e-06,
+      "loss": 0.0,
+      "step": 5074
+    },
+    {
+      "epoch": 0.3270606431655603,
+      "grad_norm": 0.006191962266799229,
+      "learning_rate": 7.480128893662729e-06,
+      "loss": 0.0,
+      "step": 5075
+    },
+    {
+      "epoch": 0.3271250886124895,
+      "grad_norm": 0.0037964312015699485,
+      "learning_rate": 7.479412817758683e-06,
+      "loss": 0.0,
+      "step": 5076
+    },
+    {
+      "epoch": 0.3271895340594187,
+      "grad_norm": 0.0003184726203052082,
+      "learning_rate": 7.478696741854638e-06,
+      "loss": 0.0,
+      "step": 5077
+    },
+    {
+      "epoch": 0.3272539795063479,
+      "grad_norm": 0.47123552026941196,
+      "learning_rate": 7.477980665950592e-06,
+      "loss": 0.004,
+      "step": 5078
+    },
+    {
+      "epoch": 0.32731842495327707,
+      "grad_norm": 0.35992211698106236,
+      "learning_rate": 7.477264590046545e-06,
+      "loss": 0.0007,
+      "step": 5079
+    },
+    {
+      "epoch": 0.32738287040020625,
+      "grad_norm": 0.18303458304005787,
+      "learning_rate": 7.4765485141425e-06,
+      "loss": 0.0002,
+      "step": 5080
+    },
+    {
+      "epoch": 0.3274473158471354,
+      "grad_norm": 0.005515852076379496,
+      "learning_rate": 7.475832438238454e-06,
+      "loss": 0.0,
+      "step": 5081
+    },
+    {
+      "epoch": 0.32751176129406456,
+      "grad_norm": 0.2568020526304833,
+      "learning_rate": 7.475116362334408e-06,
+      "loss": 0.0008,
+      "step": 5082
+    },
+    {
+      "epoch": 0.32757620674099375,
+      "grad_norm": 0.003086437234108956,
+      "learning_rate": 7.4744002864303626e-06,
+      "loss": 0.0,
+      "step": 5083
+    },
+    {
+      "epoch": 0.32764065218792293,
+      "grad_norm": 1.3171834943355232,
+      "learning_rate": 7.473684210526316e-06,
+      "loss": 0.0013,
+      "step": 5084
+    },
+    {
+      "epoch": 0.3277050976348521,
+      "grad_norm": 0.033313737024242854,
+      "learning_rate": 7.47296813462227e-06,
+      "loss": 0.0003,
+      "step": 5085
+    },
+    {
+      "epoch": 0.32776954308178124,
+      "grad_norm": 0.7567478842426889,
+      "learning_rate": 7.472252058718225e-06,
+      "loss": 0.0061,
+      "step": 5086
+    },
+    {
+      "epoch": 0.32783398852871043,
+      "grad_norm": 0.0049614953742675985,
+      "learning_rate": 7.471535982814179e-06,
+      "loss": 0.0,
+      "step": 5087
+    },
+    {
+      "epoch": 0.3278984339756396,
+      "grad_norm": 0.025865455721768113,
+      "learning_rate": 7.470819906910133e-06,
+      "loss": 0.0001,
+      "step": 5088
+    },
+    {
+      "epoch": 0.3279628794225688,
+      "grad_norm": 0.06708022955203878,
+      "learning_rate": 7.470103831006087e-06,
+      "loss": 0.0005,
+      "step": 5089
+    },
+    {
+      "epoch": 0.328027324869498,
+      "grad_norm": 0.11167039041291603,
+      "learning_rate": 7.469387755102041e-06,
+      "loss": 0.001,
+      "step": 5090
+    },
+    {
+      "epoch": 0.32809177031642717,
+      "grad_norm": 0.07170793525024977,
+      "learning_rate": 7.468671679197995e-06,
+      "loss": 0.0001,
+      "step": 5091
+    },
+    {
+      "epoch": 0.3281562157633563,
+      "grad_norm": 0.7082797245233898,
+      "learning_rate": 7.4679556032939496e-06,
+      "loss": 0.0016,
+      "step": 5092
+    },
+    {
+      "epoch": 0.3282206612102855,
+      "grad_norm": 0.006456634595156348,
+      "learning_rate": 7.467239527389904e-06,
+      "loss": 0.0,
+      "step": 5093
+    },
+    {
+      "epoch": 0.32828510665721466,
+      "grad_norm": 0.01645892325423116,
+      "learning_rate": 7.466523451485859e-06,
+      "loss": 0.0003,
+      "step": 5094
+    },
+    {
+      "epoch": 0.32834955210414385,
+      "grad_norm": 0.7233672896851489,
+      "learning_rate": 7.4658073755818124e-06,
+      "loss": 0.0031,
+      "step": 5095
+    },
+    {
+      "epoch": 0.32841399755107303,
+      "grad_norm": 0.1790257063440346,
+      "learning_rate": 7.465091299677767e-06,
+      "loss": 0.0004,
+      "step": 5096
+    },
+    {
+      "epoch": 0.3284784429980022,
+      "grad_norm": 0.004959311847834531,
+      "learning_rate": 7.464375223773721e-06,
+      "loss": 0.0,
+      "step": 5097
+    },
+    {
+      "epoch": 0.32854288844493135,
+      "grad_norm": 0.33592448098438615,
+      "learning_rate": 7.463659147869675e-06,
+      "loss": 0.0036,
+      "step": 5098
+    },
+    {
+      "epoch": 0.32860733389186053,
+      "grad_norm": 0.006044847333106452,
+      "learning_rate": 7.46294307196563e-06,
+      "loss": 0.0,
+      "step": 5099
+    },
+    {
+      "epoch": 0.3286717793387897,
+      "grad_norm": 0.015346268232809247,
+      "learning_rate": 7.462226996061583e-06,
+      "loss": 0.0,
+      "step": 5100
+    },
+    {
+      "epoch": 0.3287362247857189,
+      "grad_norm": 0.03213329618545794,
+      "learning_rate": 7.461510920157537e-06,
+      "loss": 0.0,
+      "step": 5101
+    },
+    {
+      "epoch": 0.3288006702326481,
+      "grad_norm": 0.0006729679357930884,
+      "learning_rate": 7.460794844253492e-06,
+      "loss": 0.0,
+      "step": 5102
+    },
+    {
+      "epoch": 0.3288651156795772,
+      "grad_norm": 0.3873201952901252,
+      "learning_rate": 7.460078768349446e-06,
+      "loss": 0.0025,
+      "step": 5103
+    },
+    {
+      "epoch": 0.3289295611265064,
+      "grad_norm": 0.0012658050799610712,
+      "learning_rate": 7.4593626924454e-06,
+      "loss": 0.0,
+      "step": 5104
+    },
+    {
+      "epoch": 0.3289940065734356,
+      "grad_norm": 0.1685233110152655,
+      "learning_rate": 7.458646616541354e-06,
+      "loss": 0.0004,
+      "step": 5105
+    },
+    {
+      "epoch": 0.32905845202036477,
+      "grad_norm": 0.003145091526784752,
+      "learning_rate": 7.457930540637308e-06,
+      "loss": 0.0,
+      "step": 5106
+    },
+    {
+      "epoch": 0.32912289746729395,
+      "grad_norm": 0.006677561097412213,
+      "learning_rate": 7.457214464733262e-06,
+      "loss": 0.0,
+      "step": 5107
+    },
+    {
+      "epoch": 0.32918734291422314,
+      "grad_norm": 0.008704954859180948,
+      "learning_rate": 7.456498388829217e-06,
+      "loss": 0.0,
+      "step": 5108
+    },
+    {
+      "epoch": 0.32925178836115226,
+      "grad_norm": 0.020684504992077225,
+      "learning_rate": 7.455782312925171e-06,
+      "loss": 0.0,
+      "step": 5109
+    },
+    {
+      "epoch": 0.32931623380808145,
+      "grad_norm": 0.017915273181646683,
+      "learning_rate": 7.455066237021124e-06,
+      "loss": 0.0001,
+      "step": 5110
+    },
+    {
+      "epoch": 0.32938067925501063,
+      "grad_norm": 0.0006413246172499555,
+      "learning_rate": 7.454350161117079e-06,
+      "loss": 0.0,
+      "step": 5111
+    },
+    {
+      "epoch": 0.3294451247019398,
+      "grad_norm": 0.016389241157989606,
+      "learning_rate": 7.453634085213033e-06,
+      "loss": 0.0,
+      "step": 5112
+    },
+    {
+      "epoch": 0.329509570148869,
+      "grad_norm": 0.04613150357756326,
+      "learning_rate": 7.452918009308987e-06,
+      "loss": 0.0002,
+      "step": 5113
+    },
+    {
+      "epoch": 0.32957401559579813,
+      "grad_norm": 0.006047588655278598,
+      "learning_rate": 7.452201933404941e-06,
+      "loss": 0.0,
+      "step": 5114
+    },
+    {
+      "epoch": 0.3296384610427273,
+      "grad_norm": 0.11379492971456501,
+      "learning_rate": 7.451485857500895e-06,
+      "loss": 0.0002,
+      "step": 5115
+    },
+    {
+      "epoch": 0.3297029064896565,
+      "grad_norm": 0.18467343189665436,
+      "learning_rate": 7.450769781596849e-06,
+      "loss": 0.0017,
+      "step": 5116
+    },
+    {
+      "epoch": 0.3297673519365857,
+      "grad_norm": 0.00666060152594704,
+      "learning_rate": 7.450053705692804e-06,
+      "loss": 0.0,
+      "step": 5117
+    },
+    {
+      "epoch": 0.32983179738351487,
+      "grad_norm": 0.15716715098903433,
+      "learning_rate": 7.449337629788759e-06,
+      "loss": 0.0002,
+      "step": 5118
+    },
+    {
+      "epoch": 0.32989624283044405,
+      "grad_norm": 0.03158307087649183,
+      "learning_rate": 7.448621553884713e-06,
+      "loss": 0.0001,
+      "step": 5119
+    },
+    {
+      "epoch": 0.3299606882773732,
+      "grad_norm": 6.076223041511504,
+      "learning_rate": 7.447905477980667e-06,
+      "loss": 0.0233,
+      "step": 5120
+    },
+    {
+      "epoch": 0.33002513372430237,
+      "grad_norm": 0.04275050059957507,
+      "learning_rate": 7.447189402076621e-06,
+      "loss": 0.0001,
+      "step": 5121
+    },
+    {
+      "epoch": 0.33008957917123155,
+      "grad_norm": 0.017601381917265038,
+      "learning_rate": 7.446473326172575e-06,
+      "loss": 0.0001,
+      "step": 5122
+    },
+    {
+      "epoch": 0.33015402461816074,
+      "grad_norm": 0.0008449752226208954,
+      "learning_rate": 7.445757250268529e-06,
+      "loss": 0.0,
+      "step": 5123
+    },
+    {
+      "epoch": 0.3302184700650899,
+      "grad_norm": 0.1010258416027035,
+      "learning_rate": 7.445041174364484e-06,
+      "loss": 0.0014,
+      "step": 5124
+    },
+    {
+      "epoch": 0.33028291551201905,
+      "grad_norm": 0.0027437871382046137,
+      "learning_rate": 7.444325098460438e-06,
+      "loss": 0.0,
+      "step": 5125
+    },
+    {
+      "epoch": 0.33034736095894823,
+      "grad_norm": 0.7007602556671023,
+      "learning_rate": 7.4436090225563915e-06,
+      "loss": 0.0044,
+      "step": 5126
+    },
+    {
+      "epoch": 0.3304118064058774,
+      "grad_norm": 0.06670512931405594,
+      "learning_rate": 7.442892946652346e-06,
+      "loss": 0.0002,
+      "step": 5127
+    },
+    {
+      "epoch": 0.3304762518528066,
+      "grad_norm": 0.00035463289021841213,
+      "learning_rate": 7.4421768707483e-06,
+      "loss": 0.0,
+      "step": 5128
+    },
+    {
+      "epoch": 0.3305406972997358,
+      "grad_norm": 0.011838379957474788,
+      "learning_rate": 7.441460794844254e-06,
+      "loss": 0.0,
+      "step": 5129
+    },
+    {
+      "epoch": 0.33060514274666497,
+      "grad_norm": 0.10076242075093453,
+      "learning_rate": 7.440744718940208e-06,
+      "loss": 0.0001,
+      "step": 5130
+    },
+    {
+      "epoch": 0.3306695881935941,
+      "grad_norm": 0.03547042955610553,
+      "learning_rate": 7.440028643036162e-06,
+      "loss": 0.0,
+      "step": 5131
+    },
+    {
+      "epoch": 0.3307340336405233,
+      "grad_norm": 0.02167577114795024,
+      "learning_rate": 7.439312567132116e-06,
+      "loss": 0.0,
+      "step": 5132
+    },
+    {
+      "epoch": 0.33079847908745247,
+      "grad_norm": 0.17646025275384114,
+      "learning_rate": 7.438596491228071e-06,
+      "loss": 0.0005,
+      "step": 5133
+    },
+    {
+      "epoch": 0.33086292453438165,
+      "grad_norm": 0.2529165492824762,
+      "learning_rate": 7.437880415324025e-06,
+      "loss": 0.0008,
+      "step": 5134
+    },
+    {
+      "epoch": 0.33092736998131084,
+      "grad_norm": 0.020512357560091968,
+      "learning_rate": 7.4371643394199785e-06,
+      "loss": 0.0001,
+      "step": 5135
+    },
+    {
+      "epoch": 0.33099181542824,
+      "grad_norm": 0.48623961290478873,
+      "learning_rate": 7.436448263515933e-06,
+      "loss": 0.0024,
+      "step": 5136
+    },
+    {
+      "epoch": 0.33105626087516915,
+      "grad_norm": 0.011717598828756112,
+      "learning_rate": 7.435732187611887e-06,
+      "loss": 0.0,
+      "step": 5137
+    },
+    {
+      "epoch": 0.33112070632209833,
+      "grad_norm": 0.10853024365035156,
+      "learning_rate": 7.435016111707841e-06,
+      "loss": 0.0016,
+      "step": 5138
+    },
+    {
+      "epoch": 0.3311851517690275,
+      "grad_norm": 0.011491480938313548,
+      "learning_rate": 7.434300035803796e-06,
+      "loss": 0.0001,
+      "step": 5139
+    },
+    {
+      "epoch": 0.3312495972159567,
+      "grad_norm": 0.018680412427338224,
+      "learning_rate": 7.433583959899749e-06,
+      "loss": 0.0001,
+      "step": 5140
+    },
+    {
+      "epoch": 0.3313140426628859,
+      "grad_norm": 0.0033161842348454814,
+      "learning_rate": 7.432867883995705e-06,
+      "loss": 0.0,
+      "step": 5141
+    },
+    {
+      "epoch": 0.331378488109815,
+      "grad_norm": 0.00034673562412390456,
+      "learning_rate": 7.4321518080916586e-06,
+      "loss": 0.0,
+      "step": 5142
+    },
+    {
+      "epoch": 0.3314429335567442,
+      "grad_norm": 0.0029976854056438,
+      "learning_rate": 7.431435732187613e-06,
+      "loss": 0.0,
+      "step": 5143
+    },
+    {
+      "epoch": 0.3315073790036734,
+      "grad_norm": 0.0018374341236920763,
+      "learning_rate": 7.430719656283567e-06,
+      "loss": 0.0,
+      "step": 5144
+    },
+    {
+      "epoch": 0.33157182445060257,
+      "grad_norm": 0.003731978135310744,
+      "learning_rate": 7.4300035803795215e-06,
+      "loss": 0.0,
+      "step": 5145
+    },
+    {
+      "epoch": 0.33163626989753175,
+      "grad_norm": 0.01021389306752017,
+      "learning_rate": 7.429287504475475e-06,
+      "loss": 0.0001,
+      "step": 5146
+    },
+    {
+      "epoch": 0.33170071534446094,
+      "grad_norm": 0.002822017216866906,
+      "learning_rate": 7.428571428571429e-06,
+      "loss": 0.0,
+      "step": 5147
+    },
+    {
+      "epoch": 0.33176516079139007,
+      "grad_norm": 0.0014409824839609782,
+      "learning_rate": 7.4278553526673835e-06,
+      "loss": 0.0,
+      "step": 5148
+    },
+    {
+      "epoch": 0.33182960623831925,
+      "grad_norm": 0.03521087826211135,
+      "learning_rate": 7.427139276763338e-06,
+      "loss": 0.0,
+      "step": 5149
+    },
+    {
+      "epoch": 0.33189405168524844,
+      "grad_norm": 0.0005318137644412104,
+      "learning_rate": 7.426423200859292e-06,
+      "loss": 0.0,
+      "step": 5150
+    },
+    {
+      "epoch": 0.3319584971321776,
+      "grad_norm": 0.001292276645488789,
+      "learning_rate": 7.4257071249552456e-06,
+      "loss": 0.0,
+      "step": 5151
+    },
+    {
+      "epoch": 0.3320229425791068,
+      "grad_norm": 0.1877081827917423,
+      "learning_rate": 7.4249910490512e-06,
+      "loss": 0.0018,
+      "step": 5152
+    },
+    {
+      "epoch": 0.33208738802603593,
+      "grad_norm": 0.0013139168940337887,
+      "learning_rate": 7.424274973147154e-06,
+      "loss": 0.0,
+      "step": 5153
+    },
+    {
+      "epoch": 0.3321518334729651,
+      "grad_norm": 0.0013585570143822931,
+      "learning_rate": 7.4235588972431084e-06,
+      "loss": 0.0,
+      "step": 5154
+    },
+    {
+      "epoch": 0.3322162789198943,
+      "grad_norm": 0.01602709453277668,
+      "learning_rate": 7.422842821339063e-06,
+      "loss": 0.0,
+      "step": 5155
+    },
+    {
+      "epoch": 0.3322807243668235,
+      "grad_norm": 0.17357639773672925,
+      "learning_rate": 7.422126745435016e-06,
+      "loss": 0.0008,
+      "step": 5156
+    },
+    {
+      "epoch": 0.33234516981375267,
+      "grad_norm": 0.527027119841736,
+      "learning_rate": 7.4214106695309705e-06,
+      "loss": 0.0079,
+      "step": 5157
+    },
+    {
+      "epoch": 0.33240961526068186,
+      "grad_norm": 0.1942045831769039,
+      "learning_rate": 7.420694593626925e-06,
+      "loss": 0.0004,
+      "step": 5158
+    },
+    {
+      "epoch": 0.332474060707611,
+      "grad_norm": 0.001252549471896302,
+      "learning_rate": 7.419978517722879e-06,
+      "loss": 0.0,
+      "step": 5159
+    },
+    {
+      "epoch": 0.33253850615454017,
+      "grad_norm": 0.004385940819667968,
+      "learning_rate": 7.419262441818833e-06,
+      "loss": 0.0,
+      "step": 5160
+    },
+    {
+      "epoch": 0.33260295160146935,
+      "grad_norm": 0.0018455546544822695,
+      "learning_rate": 7.418546365914787e-06,
+      "loss": 0.0,
+      "step": 5161
+    },
+    {
+      "epoch": 0.33266739704839854,
+      "grad_norm": 0.0013265551464794725,
+      "learning_rate": 7.417830290010741e-06,
+      "loss": 0.0,
+      "step": 5162
+    },
+    {
+      "epoch": 0.3327318424953277,
+      "grad_norm": 0.0005334555103447912,
+      "learning_rate": 7.4171142141066954e-06,
+      "loss": 0.0,
+      "step": 5163
+    },
+    {
+      "epoch": 0.33279628794225685,
+      "grad_norm": 0.0003283532295627354,
+      "learning_rate": 7.41639813820265e-06,
+      "loss": 0.0,
+      "step": 5164
+    },
+    {
+      "epoch": 0.33286073338918604,
+      "grad_norm": 0.05352444132349498,
+      "learning_rate": 7.415682062298605e-06,
+      "loss": 0.0001,
+      "step": 5165
+    },
+    {
+      "epoch": 0.3329251788361152,
+      "grad_norm": 0.008953477244246817,
+      "learning_rate": 7.414965986394559e-06,
+      "loss": 0.0,
+      "step": 5166
+    },
+    {
+      "epoch": 0.3329896242830444,
+      "grad_norm": 0.14708287994672403,
+      "learning_rate": 7.414249910490513e-06,
+      "loss": 0.0003,
+      "step": 5167
+    },
+    {
+      "epoch": 0.3330540697299736,
+      "grad_norm": 0.0017898316957946575,
+      "learning_rate": 7.413533834586467e-06,
+      "loss": 0.0,
+      "step": 5168
+    },
+    {
+      "epoch": 0.3331185151769028,
+      "grad_norm": 0.019036091577501586,
+      "learning_rate": 7.412817758682421e-06,
+      "loss": 0.0001,
+      "step": 5169
+    },
+    {
+      "epoch": 0.3331829606238319,
+      "grad_norm": 0.0005750557774482397,
+      "learning_rate": 7.4121016827783755e-06,
+      "loss": 0.0,
+      "step": 5170
+    },
+    {
+      "epoch": 0.3332474060707611,
+      "grad_norm": 0.00792877003601551,
+      "learning_rate": 7.41138560687433e-06,
+      "loss": 0.0,
+      "step": 5171
+    },
+    {
+      "epoch": 0.33331185151769027,
+      "grad_norm": 0.46352720952438115,
+      "learning_rate": 7.410669530970283e-06,
+      "loss": 0.0031,
+      "step": 5172
+    },
+    {
+      "epoch": 0.33337629696461946,
+      "grad_norm": 0.04526218280174097,
+      "learning_rate": 7.409953455066238e-06,
+      "loss": 0.0001,
+      "step": 5173
+    },
+    {
+      "epoch": 0.33344074241154864,
+      "grad_norm": 0.00010384908638211477,
+      "learning_rate": 7.409237379162192e-06,
+      "loss": 0.0,
+      "step": 5174
+    },
+    {
+      "epoch": 0.3335051878584778,
+      "grad_norm": 0.0001965139918647116,
+      "learning_rate": 7.408521303258146e-06,
+      "loss": 0.0,
+      "step": 5175
+    },
+    {
+      "epoch": 0.33356963330540695,
+      "grad_norm": 0.002679323084956987,
+      "learning_rate": 7.4078052273541005e-06,
+      "loss": 0.0,
+      "step": 5176
+    },
+    {
+      "epoch": 0.33363407875233614,
+      "grad_norm": 0.0040892885241793665,
+      "learning_rate": 7.407089151450054e-06,
+      "loss": 0.0,
+      "step": 5177
+    },
+    {
+      "epoch": 0.3336985241992653,
+      "grad_norm": 0.004283341959280961,
+      "learning_rate": 7.406373075546008e-06,
+      "loss": 0.0,
+      "step": 5178
+    },
+    {
+      "epoch": 0.3337629696461945,
+      "grad_norm": 0.010563679890308148,
+      "learning_rate": 7.4056569996419625e-06,
+      "loss": 0.0001,
+      "step": 5179
+    },
+    {
+      "epoch": 0.3338274150931237,
+      "grad_norm": 0.014411587506700527,
+      "learning_rate": 7.404940923737917e-06,
+      "loss": 0.0,
+      "step": 5180
+    },
+    {
+      "epoch": 0.3338918605400528,
+      "grad_norm": 0.002117332996749874,
+      "learning_rate": 7.404224847833871e-06,
+      "loss": 0.0,
+      "step": 5181
+    },
+    {
+      "epoch": 0.333956305986982,
+      "grad_norm": 0.0051531902656213316,
+      "learning_rate": 7.4035087719298246e-06,
+      "loss": 0.0,
+      "step": 5182
+    },
+    {
+      "epoch": 0.3340207514339112,
+      "grad_norm": 0.008657921238361855,
+      "learning_rate": 7.402792696025779e-06,
+      "loss": 0.0,
+      "step": 5183
+    },
+    {
+      "epoch": 0.3340851968808404,
+      "grad_norm": 0.08266300786287209,
+      "learning_rate": 7.402076620121733e-06,
+      "loss": 0.0001,
+      "step": 5184
+    },
+    {
+      "epoch": 0.33414964232776956,
+      "grad_norm": 0.053964652856025946,
+      "learning_rate": 7.4013605442176875e-06,
+      "loss": 0.0001,
+      "step": 5185
+    },
+    {
+      "epoch": 0.33421408777469874,
+      "grad_norm": 0.00027337132137712447,
+      "learning_rate": 7.400644468313642e-06,
+      "loss": 0.0,
+      "step": 5186
+    },
+    {
+      "epoch": 0.33427853322162787,
+      "grad_norm": 0.017739007588697363,
+      "learning_rate": 7.399928392409595e-06,
+      "loss": 0.0001,
+      "step": 5187
+    },
+    {
+      "epoch": 0.33434297866855706,
+      "grad_norm": 0.10516118308706995,
+      "learning_rate": 7.39921231650555e-06,
+      "loss": 0.0001,
+      "step": 5188
+    },
+    {
+      "epoch": 0.33440742411548624,
+      "grad_norm": 0.0012310758511546937,
+      "learning_rate": 7.398496240601505e-06,
+      "loss": 0.0,
+      "step": 5189
+    },
+    {
+      "epoch": 0.3344718695624154,
+      "grad_norm": 0.0024450255738462916,
+      "learning_rate": 7.397780164697459e-06,
+      "loss": 0.0,
+      "step": 5190
+    },
+    {
+      "epoch": 0.3345363150093446,
+      "grad_norm": 0.008509604235843085,
+      "learning_rate": 7.397064088793413e-06,
+      "loss": 0.0001,
+      "step": 5191
+    },
+    {
+      "epoch": 0.33460076045627374,
+      "grad_norm": 0.016462490353666547,
+      "learning_rate": 7.3963480128893676e-06,
+      "loss": 0.0002,
+      "step": 5192
+    },
+    {
+      "epoch": 0.3346652059032029,
+      "grad_norm": 0.001184509139402617,
+      "learning_rate": 7.395631936985321e-06,
+      "loss": 0.0,
+      "step": 5193
+    },
+    {
+      "epoch": 0.3347296513501321,
+      "grad_norm": 0.0011018125580738482,
+      "learning_rate": 7.394915861081275e-06,
+      "loss": 0.0,
+      "step": 5194
+    },
+    {
+      "epoch": 0.3347940967970613,
+      "grad_norm": 0.001489303290882407,
+      "learning_rate": 7.39419978517723e-06,
+      "loss": 0.0,
+      "step": 5195
+    },
+    {
+      "epoch": 0.3348585422439905,
+      "grad_norm": 0.0018044500319175656,
+      "learning_rate": 7.393483709273184e-06,
+      "loss": 0.0,
+      "step": 5196
+    },
+    {
+      "epoch": 0.33492298769091966,
+      "grad_norm": 0.17798532857531452,
+      "learning_rate": 7.392767633369138e-06,
+      "loss": 0.0001,
+      "step": 5197
+    },
+    {
+      "epoch": 0.3349874331378488,
+      "grad_norm": 0.023186446715734367,
+      "learning_rate": 7.392051557465092e-06,
+      "loss": 0.0,
+      "step": 5198
+    },
+    {
+      "epoch": 0.335051878584778,
+      "grad_norm": 0.14059826808740464,
+      "learning_rate": 7.391335481561046e-06,
+      "loss": 0.0005,
+      "step": 5199
+    },
+    {
+      "epoch": 0.33511632403170716,
+      "grad_norm": 0.003308997020900199,
+      "learning_rate": 7.390619405657e-06,
+      "loss": 0.0,
+      "step": 5200
+    },
+    {
+      "epoch": 0.33518076947863634,
+      "grad_norm": 0.031233637035030917,
+      "learning_rate": 7.3899033297529546e-06,
+      "loss": 0.0001,
+      "step": 5201
+    },
+    {
+      "epoch": 0.3352452149255655,
+      "grad_norm": 0.0012639814431313896,
+      "learning_rate": 7.389187253848909e-06,
+      "loss": 0.0,
+      "step": 5202
+    },
+    {
+      "epoch": 0.33530966037249466,
+      "grad_norm": 0.005084546256411885,
+      "learning_rate": 7.388471177944862e-06,
+      "loss": 0.0,
+      "step": 5203
+    },
+    {
+      "epoch": 0.33537410581942384,
+      "grad_norm": 0.39486029094773817,
+      "learning_rate": 7.387755102040817e-06,
+      "loss": 0.0007,
+      "step": 5204
+    },
+    {
+      "epoch": 0.335438551266353,
+      "grad_norm": 0.035950594309817285,
+      "learning_rate": 7.387039026136771e-06,
+      "loss": 0.0004,
+      "step": 5205
+    },
+    {
+      "epoch": 0.3355029967132822,
+      "grad_norm": 0.022437522697819116,
+      "learning_rate": 7.386322950232725e-06,
+      "loss": 0.0002,
+      "step": 5206
+    },
+    {
+      "epoch": 0.3355674421602114,
+      "grad_norm": 0.002273460860477751,
+      "learning_rate": 7.385606874328679e-06,
+      "loss": 0.0,
+      "step": 5207
+    },
+    {
+      "epoch": 0.3356318876071406,
+      "grad_norm": 0.019155860026953915,
+      "learning_rate": 7.384890798424633e-06,
+      "loss": 0.0,
+      "step": 5208
+    },
+    {
+      "epoch": 0.3356963330540697,
+      "grad_norm": 0.008933259082885026,
+      "learning_rate": 7.384174722520587e-06,
+      "loss": 0.0,
+      "step": 5209
+    },
+    {
+      "epoch": 0.3357607785009989,
+      "grad_norm": 0.0018883702706931679,
+      "learning_rate": 7.3834586466165416e-06,
+      "loss": 0.0,
+      "step": 5210
+    },
+    {
+      "epoch": 0.3358252239479281,
+      "grad_norm": 0.041224874042787886,
+      "learning_rate": 7.382742570712497e-06,
+      "loss": 0.0002,
+      "step": 5211
+    },
+    {
+      "epoch": 0.33588966939485726,
+      "grad_norm": 0.005298047347779996,
+      "learning_rate": 7.382026494808451e-06,
+      "loss": 0.0,
+      "step": 5212
+    },
+    {
+      "epoch": 0.33595411484178644,
+      "grad_norm": 0.012037538191049219,
+      "learning_rate": 7.381310418904405e-06,
+      "loss": 0.0,
+      "step": 5213
+    },
+    {
+      "epoch": 0.33601856028871563,
+      "grad_norm": 0.018119530926582156,
+      "learning_rate": 7.380594343000359e-06,
+      "loss": 0.0,
+      "step": 5214
+    },
+    {
+      "epoch": 0.33608300573564476,
+      "grad_norm": 0.0065361923963192506,
+      "learning_rate": 7.379878267096313e-06,
+      "loss": 0.0001,
+      "step": 5215
+    },
+    {
+      "epoch": 0.33614745118257394,
+      "grad_norm": 0.03795719088023908,
+      "learning_rate": 7.379162191192267e-06,
+      "loss": 0.0002,
+      "step": 5216
+    },
+    {
+      "epoch": 0.3362118966295031,
+      "grad_norm": 0.0003327354220714893,
+      "learning_rate": 7.378446115288222e-06,
+      "loss": 0.0,
+      "step": 5217
+    },
+    {
+      "epoch": 0.3362763420764323,
+      "grad_norm": 0.002622280967886348,
+      "learning_rate": 7.377730039384176e-06,
+      "loss": 0.0,
+      "step": 5218
+    },
+    {
+      "epoch": 0.3363407875233615,
+      "grad_norm": 0.003362164567001042,
+      "learning_rate": 7.377013963480129e-06,
+      "loss": 0.0,
+      "step": 5219
+    },
+    {
+      "epoch": 0.3364052329702906,
+      "grad_norm": 0.002488965637931309,
+      "learning_rate": 7.376297887576084e-06,
+      "loss": 0.0,
+      "step": 5220
+    },
+    {
+      "epoch": 0.3364696784172198,
+      "grad_norm": 0.021178266136436676,
+      "learning_rate": 7.375581811672038e-06,
+      "loss": 0.0017,
+      "step": 5221
+    },
+    {
+      "epoch": 0.336534123864149,
+      "grad_norm": 5.556497552128451e-05,
+      "learning_rate": 7.374865735767992e-06,
+      "loss": 0.0,
+      "step": 5222
+    },
+    {
+      "epoch": 0.3365985693110782,
+      "grad_norm": 1.27603100623005,
+      "learning_rate": 7.374149659863946e-06,
+      "loss": 0.0302,
+      "step": 5223
+    },
+    {
+      "epoch": 0.33666301475800736,
+      "grad_norm": 0.10044723039261633,
+      "learning_rate": 7.3734335839599e-06,
+      "loss": 0.0005,
+      "step": 5224
+    },
+    {
+      "epoch": 0.33672746020493655,
+      "grad_norm": 0.002654275139626637,
+      "learning_rate": 7.372717508055854e-06,
+      "loss": 0.0,
+      "step": 5225
+    },
+    {
+      "epoch": 0.3367919056518657,
+      "grad_norm": 0.15670975755353522,
+      "learning_rate": 7.372001432151809e-06,
+      "loss": 0.0001,
+      "step": 5226
+    },
+    {
+      "epoch": 0.33685635109879486,
+      "grad_norm": 0.006504992868248114,
+      "learning_rate": 7.371285356247763e-06,
+      "loss": 0.0,
+      "step": 5227
+    },
+    {
+      "epoch": 0.33692079654572404,
+      "grad_norm": 0.0029773298244788306,
+      "learning_rate": 7.370569280343716e-06,
+      "loss": 0.0,
+      "step": 5228
+    },
+    {
+      "epoch": 0.33698524199265323,
+      "grad_norm": 0.001564768362004572,
+      "learning_rate": 7.369853204439671e-06,
+      "loss": 0.0,
+      "step": 5229
+    },
+    {
+      "epoch": 0.3370496874395824,
+      "grad_norm": 0.46053742972565676,
+      "learning_rate": 7.369137128535625e-06,
+      "loss": 0.0004,
+      "step": 5230
+    },
+    {
+      "epoch": 0.33711413288651154,
+      "grad_norm": 0.03325454502955247,
+      "learning_rate": 7.368421052631579e-06,
+      "loss": 0.0001,
+      "step": 5231
+    },
+    {
+      "epoch": 0.3371785783334407,
+      "grad_norm": 0.017572879836545905,
+      "learning_rate": 7.367704976727534e-06,
+      "loss": 0.0,
+      "step": 5232
+    },
+    {
+      "epoch": 0.3372430237803699,
+      "grad_norm": 0.008048314701695543,
+      "learning_rate": 7.366988900823487e-06,
+      "loss": 0.0,
+      "step": 5233
+    },
+    {
+      "epoch": 0.3373074692272991,
+      "grad_norm": 0.0466043386887468,
+      "learning_rate": 7.366272824919441e-06,
+      "loss": 0.0001,
+      "step": 5234
+    },
+    {
+      "epoch": 0.3373719146742283,
+      "grad_norm": 0.17321825195418455,
+      "learning_rate": 7.3655567490153965e-06,
+      "loss": 0.0002,
+      "step": 5235
+    },
+    {
+      "epoch": 0.33743636012115746,
+      "grad_norm": 0.014061451726326977,
+      "learning_rate": 7.364840673111351e-06,
+      "loss": 0.0001,
+      "step": 5236
+    },
+    {
+      "epoch": 0.3375008055680866,
+      "grad_norm": 0.0063745520315426,
+      "learning_rate": 7.364124597207305e-06,
+      "loss": 0.0,
+      "step": 5237
+    },
+    {
+      "epoch": 0.3375652510150158,
+      "grad_norm": 0.09786029955600638,
+      "learning_rate": 7.363408521303259e-06,
+      "loss": 0.0007,
+      "step": 5238
+    },
+    {
+      "epoch": 0.33762969646194496,
+      "grad_norm": 0.0045003900085370694,
+      "learning_rate": 7.362692445399213e-06,
+      "loss": 0.0,
+      "step": 5239
+    },
+    {
+      "epoch": 0.33769414190887415,
+      "grad_norm": 0.07432003388342096,
+      "learning_rate": 7.361976369495167e-06,
+      "loss": 0.0001,
+      "step": 5240
+    },
+    {
+      "epoch": 0.33775858735580333,
+      "grad_norm": 0.013922346916827281,
+      "learning_rate": 7.361260293591121e-06,
+      "loss": 0.0,
+      "step": 5241
+    },
+    {
+      "epoch": 0.3378230328027325,
+      "grad_norm": 0.0038764664985424606,
+      "learning_rate": 7.360544217687076e-06,
+      "loss": 0.0,
+      "step": 5242
+    },
+    {
+      "epoch": 0.33788747824966164,
+      "grad_norm": 0.004857801852726184,
+      "learning_rate": 7.35982814178303e-06,
+      "loss": 0.0,
+      "step": 5243
+    },
+    {
+      "epoch": 0.33795192369659083,
+      "grad_norm": 0.2933163297644183,
+      "learning_rate": 7.3591120658789835e-06,
+      "loss": 0.0012,
+      "step": 5244
+    },
+    {
+      "epoch": 0.33801636914352,
+      "grad_norm": 0.013560541083518562,
+      "learning_rate": 7.358395989974938e-06,
+      "loss": 0.0,
+      "step": 5245
+    },
+    {
+      "epoch": 0.3380808145904492,
+      "grad_norm": 0.0003962089272216273,
+      "learning_rate": 7.357679914070892e-06,
+      "loss": 0.0,
+      "step": 5246
+    },
+    {
+      "epoch": 0.3381452600373784,
+      "grad_norm": 0.009339175938270656,
+      "learning_rate": 7.356963838166846e-06,
+      "loss": 0.0,
+      "step": 5247
+    },
+    {
+      "epoch": 0.3382097054843075,
+      "grad_norm": 0.0007190247019661537,
+      "learning_rate": 7.356247762262801e-06,
+      "loss": 0.0,
+      "step": 5248
+    },
+    {
+      "epoch": 0.3382741509312367,
+      "grad_norm": 0.0059187646550370775,
+      "learning_rate": 7.355531686358754e-06,
+      "loss": 0.0,
+      "step": 5249
+    },
+    {
+      "epoch": 0.3383385963781659,
+      "grad_norm": 0.4797958686513202,
+      "learning_rate": 7.354815610454708e-06,
+      "loss": 0.0016,
+      "step": 5250
+    },
+    {
+      "epoch": 0.33840304182509506,
+      "grad_norm": 0.20533320593971888,
+      "learning_rate": 7.354099534550663e-06,
+      "loss": 0.0026,
+      "step": 5251
+    },
+    {
+      "epoch": 0.33846748727202425,
+      "grad_norm": 0.0036076111581535755,
+      "learning_rate": 7.353383458646617e-06,
+      "loss": 0.0,
+      "step": 5252
+    },
+    {
+      "epoch": 0.33853193271895343,
+      "grad_norm": 0.004331472841271792,
+      "learning_rate": 7.352667382742571e-06,
+      "loss": 0.0,
+      "step": 5253
+    },
+    {
+      "epoch": 0.33859637816588256,
+      "grad_norm": 0.08988593500177013,
+      "learning_rate": 7.351951306838525e-06,
+      "loss": 0.0002,
+      "step": 5254
+    },
+    {
+      "epoch": 0.33866082361281175,
+      "grad_norm": 0.004191599223747066,
+      "learning_rate": 7.351235230934479e-06,
+      "loss": 0.0,
+      "step": 5255
+    },
+    {
+      "epoch": 0.33872526905974093,
+      "grad_norm": 0.36925171890969277,
+      "learning_rate": 7.350519155030433e-06,
+      "loss": 0.0003,
+      "step": 5256
+    },
+    {
+      "epoch": 0.3387897145066701,
+      "grad_norm": 0.04986970061517144,
+      "learning_rate": 7.349803079126388e-06,
+      "loss": 0.0001,
+      "step": 5257
+    },
+    {
+      "epoch": 0.3388541599535993,
+      "grad_norm": 0.010657370180991166,
+      "learning_rate": 7.349087003222343e-06,
+      "loss": 0.0,
+      "step": 5258
+    },
+    {
+      "epoch": 0.3389186054005284,
+      "grad_norm": 0.5126343899815423,
+      "learning_rate": 7.348370927318297e-06,
+      "loss": 0.0013,
+      "step": 5259
+    },
+    {
+      "epoch": 0.3389830508474576,
+      "grad_norm": 0.05698766120418015,
+      "learning_rate": 7.3476548514142506e-06,
+      "loss": 0.0001,
+      "step": 5260
+    },
+    {
+      "epoch": 0.3390474962943868,
+      "grad_norm": 0.004494436469781267,
+      "learning_rate": 7.346938775510205e-06,
+      "loss": 0.0,
+      "step": 5261
+    },
+    {
+      "epoch": 0.339111941741316,
+      "grad_norm": 0.0020461885320504514,
+      "learning_rate": 7.346222699606159e-06,
+      "loss": 0.0,
+      "step": 5262
+    },
+    {
+      "epoch": 0.33917638718824517,
+      "grad_norm": 0.02113158221440442,
+      "learning_rate": 7.3455066237021135e-06,
+      "loss": 0.0002,
+      "step": 5263
+    },
+    {
+      "epoch": 0.33924083263517435,
+      "grad_norm": 0.11814162932010543,
+      "learning_rate": 7.344790547798068e-06,
+      "loss": 0.0005,
+      "step": 5264
+    },
+    {
+      "epoch": 0.3393052780821035,
+      "grad_norm": 0.011550950562999067,
+      "learning_rate": 7.344074471894021e-06,
+      "loss": 0.0,
+      "step": 5265
+    },
+    {
+      "epoch": 0.33936972352903266,
+      "grad_norm": 0.02042146368376844,
+      "learning_rate": 7.3433583959899755e-06,
+      "loss": 0.0002,
+      "step": 5266
+    },
+    {
+      "epoch": 0.33943416897596185,
+      "grad_norm": 0.005358844574434535,
+      "learning_rate": 7.34264232008593e-06,
+      "loss": 0.0,
+      "step": 5267
+    },
+    {
+      "epoch": 0.33949861442289103,
+      "grad_norm": 0.003798950955890751,
+      "learning_rate": 7.341926244181884e-06,
+      "loss": 0.0,
+      "step": 5268
+    },
+    {
+      "epoch": 0.3395630598698202,
+      "grad_norm": 0.008024759503129957,
+      "learning_rate": 7.341210168277838e-06,
+      "loss": 0.0,
+      "step": 5269
+    },
+    {
+      "epoch": 0.33962750531674935,
+      "grad_norm": 0.008326903955132609,
+      "learning_rate": 7.340494092373792e-06,
+      "loss": 0.0,
+      "step": 5270
+    },
+    {
+      "epoch": 0.33969195076367853,
+      "grad_norm": 0.11832471047203198,
+      "learning_rate": 7.339778016469746e-06,
+      "loss": 0.0019,
+      "step": 5271
+    },
+    {
+      "epoch": 0.3397563962106077,
+      "grad_norm": 0.0015766559886541335,
+      "learning_rate": 7.3390619405657004e-06,
+      "loss": 0.0,
+      "step": 5272
+    },
+    {
+      "epoch": 0.3398208416575369,
+      "grad_norm": 0.07010359991070295,
+      "learning_rate": 7.338345864661655e-06,
+      "loss": 0.0001,
+      "step": 5273
+    },
+    {
+      "epoch": 0.3398852871044661,
+      "grad_norm": 0.0008208800151459135,
+      "learning_rate": 7.337629788757609e-06,
+      "loss": 0.0,
+      "step": 5274
+    },
+    {
+      "epoch": 0.33994973255139527,
+      "grad_norm": 0.0007744022262015843,
+      "learning_rate": 7.3369137128535625e-06,
+      "loss": 0.0,
+      "step": 5275
+    },
+    {
+      "epoch": 0.3400141779983244,
+      "grad_norm": 0.19669594039359198,
+      "learning_rate": 7.336197636949517e-06,
+      "loss": 0.0024,
+      "step": 5276
+    },
+    {
+      "epoch": 0.3400786234452536,
+      "grad_norm": 0.001586631846665345,
+      "learning_rate": 7.335481561045471e-06,
+      "loss": 0.0,
+      "step": 5277
+    },
+    {
+      "epoch": 0.34014306889218276,
+      "grad_norm": 0.0016974822848838252,
+      "learning_rate": 7.334765485141425e-06,
+      "loss": 0.0,
+      "step": 5278
+    },
+    {
+      "epoch": 0.34020751433911195,
+      "grad_norm": 0.010299373551429326,
+      "learning_rate": 7.33404940923738e-06,
+      "loss": 0.0,
+      "step": 5279
+    },
+    {
+      "epoch": 0.34027195978604113,
+      "grad_norm": 0.021553520264792618,
+      "learning_rate": 7.333333333333333e-06,
+      "loss": 0.0,
+      "step": 5280
+    },
+    {
+      "epoch": 0.3403364052329703,
+      "grad_norm": 0.0010162259051668598,
+      "learning_rate": 7.332617257429288e-06,
+      "loss": 0.0,
+      "step": 5281
+    },
+    {
+      "epoch": 0.34040085067989945,
+      "grad_norm": 0.19239838048229532,
+      "learning_rate": 7.331901181525243e-06,
+      "loss": 0.0016,
+      "step": 5282
+    },
+    {
+      "epoch": 0.34046529612682863,
+      "grad_norm": 0.0019388955843136493,
+      "learning_rate": 7.331185105621197e-06,
+      "loss": 0.0,
+      "step": 5283
+    },
+    {
+      "epoch": 0.3405297415737578,
+      "grad_norm": 0.10166173451835614,
+      "learning_rate": 7.330469029717151e-06,
+      "loss": 0.0002,
+      "step": 5284
+    },
+    {
+      "epoch": 0.340594187020687,
+      "grad_norm": 0.2888227447924931,
+      "learning_rate": 7.3297529538131055e-06,
+      "loss": 0.0002,
+      "step": 5285
+    },
+    {
+      "epoch": 0.3406586324676162,
+      "grad_norm": 0.10953043213619218,
+      "learning_rate": 7.329036877909059e-06,
+      "loss": 0.0018,
+      "step": 5286
+    },
+    {
+      "epoch": 0.3407230779145453,
+      "grad_norm": 0.08343678035229286,
+      "learning_rate": 7.328320802005013e-06,
+      "loss": 0.0003,
+      "step": 5287
+    },
+    {
+      "epoch": 0.3407875233614745,
+      "grad_norm": 0.008743033563584184,
+      "learning_rate": 7.3276047261009675e-06,
+      "loss": 0.0,
+      "step": 5288
+    },
+    {
+      "epoch": 0.3408519688084037,
+      "grad_norm": 0.015663007613879978,
+      "learning_rate": 7.326888650196922e-06,
+      "loss": 0.0016,
+      "step": 5289
+    },
+    {
+      "epoch": 0.34091641425533287,
+      "grad_norm": 0.16035313832000075,
+      "learning_rate": 7.326172574292876e-06,
+      "loss": 0.0004,
+      "step": 5290
+    },
+    {
+      "epoch": 0.34098085970226205,
+      "grad_norm": 0.02837167493002224,
+      "learning_rate": 7.32545649838883e-06,
+      "loss": 0.0,
+      "step": 5291
+    },
+    {
+      "epoch": 0.34104530514919124,
+      "grad_norm": 0.05193844351158227,
+      "learning_rate": 7.324740422484784e-06,
+      "loss": 0.0,
+      "step": 5292
+    },
+    {
+      "epoch": 0.34110975059612036,
+      "grad_norm": 0.002242715562046738,
+      "learning_rate": 7.324024346580738e-06,
+      "loss": 0.0,
+      "step": 5293
+    },
+    {
+      "epoch": 0.34117419604304955,
+      "grad_norm": 0.11049409118395545,
+      "learning_rate": 7.3233082706766925e-06,
+      "loss": 0.0005,
+      "step": 5294
+    },
+    {
+      "epoch": 0.34123864148997873,
+      "grad_norm": 0.0005390615075129139,
+      "learning_rate": 7.322592194772647e-06,
+      "loss": 0.0,
+      "step": 5295
+    },
+    {
+      "epoch": 0.3413030869369079,
+      "grad_norm": 0.0013760686804202237,
+      "learning_rate": 7.3218761188686e-06,
+      "loss": 0.0,
+      "step": 5296
+    },
+    {
+      "epoch": 0.3413675323838371,
+      "grad_norm": 0.09472930550290676,
+      "learning_rate": 7.3211600429645545e-06,
+      "loss": 0.0011,
+      "step": 5297
+    },
+    {
+      "epoch": 0.34143197783076623,
+      "grad_norm": 0.2669654151871651,
+      "learning_rate": 7.320443967060509e-06,
+      "loss": 0.0006,
+      "step": 5298
+    },
+    {
+      "epoch": 0.3414964232776954,
+      "grad_norm": 0.005662463238322093,
+      "learning_rate": 7.319727891156463e-06,
+      "loss": 0.0,
+      "step": 5299
+    },
+    {
+      "epoch": 0.3415608687246246,
+      "grad_norm": 1.0199606272634265,
+      "learning_rate": 7.3190118152524166e-06,
+      "loss": 0.001,
+      "step": 5300
+    },
+    {
+      "epoch": 0.3416253141715538,
+      "grad_norm": 0.0006392407831279938,
+      "learning_rate": 7.318295739348371e-06,
+      "loss": 0.0,
+      "step": 5301
+    },
+    {
+      "epoch": 0.34168975961848297,
+      "grad_norm": 0.0009464675577917425,
+      "learning_rate": 7.317579663444325e-06,
+      "loss": 0.0,
+      "step": 5302
+    },
+    {
+      "epoch": 0.34175420506541215,
+      "grad_norm": 0.33982247253106124,
+      "learning_rate": 7.3168635875402795e-06,
+      "loss": 0.0008,
+      "step": 5303
+    },
+    {
+      "epoch": 0.3418186505123413,
+      "grad_norm": 0.0005554538138186463,
+      "learning_rate": 7.316147511636234e-06,
+      "loss": 0.0,
+      "step": 5304
+    },
+    {
+      "epoch": 0.34188309595927047,
+      "grad_norm": 0.011539964270771048,
+      "learning_rate": 7.315431435732189e-06,
+      "loss": 0.0,
+      "step": 5305
+    },
+    {
+      "epoch": 0.34194754140619965,
+      "grad_norm": 0.004195442787445282,
+      "learning_rate": 7.314715359828143e-06,
+      "loss": 0.0,
+      "step": 5306
+    },
+    {
+      "epoch": 0.34201198685312884,
+      "grad_norm": 0.007462997480980597,
+      "learning_rate": 7.313999283924097e-06,
+      "loss": 0.0,
+      "step": 5307
+    },
+    {
+      "epoch": 0.342076432300058,
+      "grad_norm": 0.00024238609706196129,
+      "learning_rate": 7.313283208020051e-06,
+      "loss": 0.0,
+      "step": 5308
+    },
+    {
+      "epoch": 0.34214087774698715,
+      "grad_norm": 0.0022106063684528892,
+      "learning_rate": 7.312567132116005e-06,
+      "loss": 0.0,
+      "step": 5309
+    },
+    {
+      "epoch": 0.34220532319391633,
+      "grad_norm": 0.01913270911313999,
+      "learning_rate": 7.3118510562119596e-06,
+      "loss": 0.0001,
+      "step": 5310
+    },
+    {
+      "epoch": 0.3422697686408455,
+      "grad_norm": 0.002142273071297971,
+      "learning_rate": 7.311134980307914e-06,
+      "loss": 0.0,
+      "step": 5311
+    },
+    {
+      "epoch": 0.3423342140877747,
+      "grad_norm": 0.0065981141478347735,
+      "learning_rate": 7.310418904403867e-06,
+      "loss": 0.0,
+      "step": 5312
+    },
+    {
+      "epoch": 0.3423986595347039,
+      "grad_norm": 0.004802954010512389,
+      "learning_rate": 7.309702828499822e-06,
+      "loss": 0.0,
+      "step": 5313
+    },
+    {
+      "epoch": 0.34246310498163307,
+      "grad_norm": 0.00110266347810728,
+      "learning_rate": 7.308986752595776e-06,
+      "loss": 0.0,
+      "step": 5314
+    },
+    {
+      "epoch": 0.3425275504285622,
+      "grad_norm": 0.0005134383818776354,
+      "learning_rate": 7.30827067669173e-06,
+      "loss": 0.0,
+      "step": 5315
+    },
+    {
+      "epoch": 0.3425919958754914,
+      "grad_norm": 0.538310230953227,
+      "learning_rate": 7.307554600787684e-06,
+      "loss": 0.001,
+      "step": 5316
+    },
+    {
+      "epoch": 0.34265644132242057,
+      "grad_norm": 0.00018143450225810087,
+      "learning_rate": 7.306838524883638e-06,
+      "loss": 0.0,
+      "step": 5317
+    },
+    {
+      "epoch": 0.34272088676934975,
+      "grad_norm": 0.40119536868709277,
+      "learning_rate": 7.306122448979592e-06,
+      "loss": 0.0025,
+      "step": 5318
+    },
+    {
+      "epoch": 0.34278533221627894,
+      "grad_norm": 0.024010840365756462,
+      "learning_rate": 7.3054063730755466e-06,
+      "loss": 0.0,
+      "step": 5319
+    },
+    {
+      "epoch": 0.3428497776632081,
+      "grad_norm": 0.0008069614846977794,
+      "learning_rate": 7.304690297171501e-06,
+      "loss": 0.0,
+      "step": 5320
+    },
+    {
+      "epoch": 0.34291422311013725,
+      "grad_norm": 0.028020230829516272,
+      "learning_rate": 7.303974221267454e-06,
+      "loss": 0.0,
+      "step": 5321
+    },
+    {
+      "epoch": 0.34297866855706644,
+      "grad_norm": 0.00010121037613932627,
+      "learning_rate": 7.303258145363409e-06,
+      "loss": 0.0,
+      "step": 5322
+    },
+    {
+      "epoch": 0.3430431140039956,
+      "grad_norm": 0.10079432169841467,
+      "learning_rate": 7.302542069459363e-06,
+      "loss": 0.0003,
+      "step": 5323
+    },
+    {
+      "epoch": 0.3431075594509248,
+      "grad_norm": 0.03269509886827942,
+      "learning_rate": 7.301825993555317e-06,
+      "loss": 0.0002,
+      "step": 5324
+    },
+    {
+      "epoch": 0.343172004897854,
+      "grad_norm": 0.0022781756443171625,
+      "learning_rate": 7.3011099176512715e-06,
+      "loss": 0.0,
+      "step": 5325
+    },
+    {
+      "epoch": 0.3432364503447831,
+      "grad_norm": 0.0014035525781737842,
+      "learning_rate": 7.300393841747225e-06,
+      "loss": 0.0,
+      "step": 5326
+    },
+    {
+      "epoch": 0.3433008957917123,
+      "grad_norm": 0.0006713917775058356,
+      "learning_rate": 7.299677765843179e-06,
+      "loss": 0.0,
+      "step": 5327
+    },
+    {
+      "epoch": 0.3433653412386415,
+      "grad_norm": 0.01338775709200437,
+      "learning_rate": 7.298961689939134e-06,
+      "loss": 0.0,
+      "step": 5328
+    },
+    {
+      "epoch": 0.34342978668557067,
+      "grad_norm": 0.0006338464533540836,
+      "learning_rate": 7.298245614035089e-06,
+      "loss": 0.0,
+      "step": 5329
+    },
+    {
+      "epoch": 0.34349423213249985,
+      "grad_norm": 0.0021448701466866347,
+      "learning_rate": 7.297529538131043e-06,
+      "loss": 0.0,
+      "step": 5330
+    },
+    {
+      "epoch": 0.34355867757942904,
+      "grad_norm": 0.012620152073128372,
+      "learning_rate": 7.296813462226997e-06,
+      "loss": 0.0,
+      "step": 5331
+    },
+    {
+      "epoch": 0.34362312302635817,
+      "grad_norm": 0.000746423385854066,
+      "learning_rate": 7.296097386322952e-06,
+      "loss": 0.0,
+      "step": 5332
+    },
+    {
+      "epoch": 0.34368756847328735,
+      "grad_norm": 9.873649287161303e-05,
+      "learning_rate": 7.295381310418905e-06,
+      "loss": 0.0,
+      "step": 5333
+    },
+    {
+      "epoch": 0.34375201392021654,
+      "grad_norm": 0.00031365937247385795,
+      "learning_rate": 7.294665234514859e-06,
+      "loss": 0.0,
+      "step": 5334
+    },
+    {
+      "epoch": 0.3438164593671457,
+      "grad_norm": 0.00528864045369544,
+      "learning_rate": 7.293949158610814e-06,
+      "loss": 0.0,
+      "step": 5335
+    },
+    {
+      "epoch": 0.3438809048140749,
+      "grad_norm": 0.0005738690377691539,
+      "learning_rate": 7.293233082706768e-06,
+      "loss": 0.0,
+      "step": 5336
+    },
+    {
+      "epoch": 0.34394535026100403,
+      "grad_norm": 0.0009102301472844842,
+      "learning_rate": 7.292517006802721e-06,
+      "loss": 0.0,
+      "step": 5337
+    },
+    {
+      "epoch": 0.3440097957079332,
+      "grad_norm": 0.00034346797846901265,
+      "learning_rate": 7.291800930898676e-06,
+      "loss": 0.0,
+      "step": 5338
+    },
+    {
+      "epoch": 0.3440742411548624,
+      "grad_norm": 0.049793358991671866,
+      "learning_rate": 7.29108485499463e-06,
+      "loss": 0.0004,
+      "step": 5339
+    },
+    {
+      "epoch": 0.3441386866017916,
+      "grad_norm": 0.014751395360373839,
+      "learning_rate": 7.290368779090584e-06,
+      "loss": 0.0,
+      "step": 5340
+    },
+    {
+      "epoch": 0.3442031320487208,
+      "grad_norm": 0.005169372080883912,
+      "learning_rate": 7.289652703186539e-06,
+      "loss": 0.0,
+      "step": 5341
+    },
+    {
+      "epoch": 0.34426757749564996,
+      "grad_norm": 2.6960228307514163e-05,
+      "learning_rate": 7.288936627282492e-06,
+      "loss": 0.0,
+      "step": 5342
+    },
+    {
+      "epoch": 0.3443320229425791,
+      "grad_norm": 0.050153460105654975,
+      "learning_rate": 7.288220551378446e-06,
+      "loss": 0.0001,
+      "step": 5343
+    },
+    {
+      "epoch": 0.34439646838950827,
+      "grad_norm": 0.00024310397212061517,
+      "learning_rate": 7.287504475474401e-06,
+      "loss": 0.0,
+      "step": 5344
+    },
+    {
+      "epoch": 0.34446091383643745,
+      "grad_norm": 0.05808561250266076,
+      "learning_rate": 7.286788399570355e-06,
+      "loss": 0.0006,
+      "step": 5345
+    },
+    {
+      "epoch": 0.34452535928336664,
+      "grad_norm": 0.00013547144495678156,
+      "learning_rate": 7.286072323666309e-06,
+      "loss": 0.0,
+      "step": 5346
+    },
+    {
+      "epoch": 0.3445898047302958,
+      "grad_norm": 0.3373302818885634,
+      "learning_rate": 7.285356247762263e-06,
+      "loss": 0.0025,
+      "step": 5347
+    },
+    {
+      "epoch": 0.34465425017722495,
+      "grad_norm": 0.016717329194680847,
+      "learning_rate": 7.284640171858217e-06,
+      "loss": 0.0001,
+      "step": 5348
+    },
+    {
+      "epoch": 0.34471869562415414,
+      "grad_norm": 0.00030821020629320307,
+      "learning_rate": 7.283924095954171e-06,
+      "loss": 0.0,
+      "step": 5349
+    },
+    {
+      "epoch": 0.3447831410710833,
+      "grad_norm": 0.0033671255527705796,
+      "learning_rate": 7.283208020050126e-06,
+      "loss": 0.0,
+      "step": 5350
+    },
+    {
+      "epoch": 0.3448475865180125,
+      "grad_norm": 0.005871330298462043,
+      "learning_rate": 7.282491944146081e-06,
+      "loss": 0.0,
+      "step": 5351
+    },
+    {
+      "epoch": 0.3449120319649417,
+      "grad_norm": 0.0003919803199134911,
+      "learning_rate": 7.281775868242035e-06,
+      "loss": 0.0,
+      "step": 5352
+    },
+    {
+      "epoch": 0.3449764774118709,
+      "grad_norm": 0.0004434323271049176,
+      "learning_rate": 7.2810597923379885e-06,
+      "loss": 0.0,
+      "step": 5353
+    },
+    {
+      "epoch": 0.3450409228588,
+      "grad_norm": 0.17988686282623115,
+      "learning_rate": 7.280343716433943e-06,
+      "loss": 0.0011,
+      "step": 5354
+    },
+    {
+      "epoch": 0.3451053683057292,
+      "grad_norm": 5.375923874340952e-05,
+      "learning_rate": 7.279627640529897e-06,
+      "loss": 0.0,
+      "step": 5355
+    },
+    {
+      "epoch": 0.34516981375265837,
+      "grad_norm": 0.0009465090573032732,
+      "learning_rate": 7.278911564625851e-06,
+      "loss": 0.0,
+      "step": 5356
+    },
+    {
+      "epoch": 0.34523425919958756,
+      "grad_norm": 0.0020489958458895278,
+      "learning_rate": 7.278195488721806e-06,
+      "loss": 0.0,
+      "step": 5357
+    },
+    {
+      "epoch": 0.34529870464651674,
+      "grad_norm": 0.00010094518209825967,
+      "learning_rate": 7.277479412817759e-06,
+      "loss": 0.0,
+      "step": 5358
+    },
+    {
+      "epoch": 0.3453631500934459,
+      "grad_norm": 0.0006345648995169979,
+      "learning_rate": 7.276763336913713e-06,
+      "loss": 0.0,
+      "step": 5359
+    },
+    {
+      "epoch": 0.34542759554037505,
+      "grad_norm": 0.41101601724437337,
+      "learning_rate": 7.276047261009668e-06,
+      "loss": 0.0011,
+      "step": 5360
+    },
+    {
+      "epoch": 0.34549204098730424,
+      "grad_norm": 0.01636080831375019,
+      "learning_rate": 7.275331185105622e-06,
+      "loss": 0.0001,
+      "step": 5361
+    },
+    {
+      "epoch": 0.3455564864342334,
+      "grad_norm": 0.030371541605649418,
+      "learning_rate": 7.274615109201576e-06,
+      "loss": 0.0001,
+      "step": 5362
+    },
+    {
+      "epoch": 0.3456209318811626,
+      "grad_norm": 0.0009595921015281959,
+      "learning_rate": 7.27389903329753e-06,
+      "loss": 0.0,
+      "step": 5363
+    },
+    {
+      "epoch": 0.3456853773280918,
+      "grad_norm": 0.0028069042707871083,
+      "learning_rate": 7.273182957393484e-06,
+      "loss": 0.0,
+      "step": 5364
+    },
+    {
+      "epoch": 0.3457498227750209,
+      "grad_norm": 0.006725710750406669,
+      "learning_rate": 7.272466881489438e-06,
+      "loss": 0.0,
+      "step": 5365
+    },
+    {
+      "epoch": 0.3458142682219501,
+      "grad_norm": 0.02202792303496806,
+      "learning_rate": 7.271750805585393e-06,
+      "loss": 0.0002,
+      "step": 5366
+    },
+    {
+      "epoch": 0.3458787136688793,
+      "grad_norm": 0.0033364088445729323,
+      "learning_rate": 7.271034729681347e-06,
+      "loss": 0.0,
+      "step": 5367
+    },
+    {
+      "epoch": 0.3459431591158085,
+      "grad_norm": 0.4054470012322448,
+      "learning_rate": 7.2703186537773e-06,
+      "loss": 0.0021,
+      "step": 5368
+    },
+    {
+      "epoch": 0.34600760456273766,
+      "grad_norm": 0.006543067397167515,
+      "learning_rate": 7.269602577873255e-06,
+      "loss": 0.0,
+      "step": 5369
+    },
+    {
+      "epoch": 0.34607205000966684,
+      "grad_norm": 4.882931587784024e-05,
+      "learning_rate": 7.268886501969209e-06,
+      "loss": 0.0,
+      "step": 5370
+    },
+    {
+      "epoch": 0.34613649545659597,
+      "grad_norm": 0.2890103138391657,
+      "learning_rate": 7.268170426065163e-06,
+      "loss": 0.0013,
+      "step": 5371
+    },
+    {
+      "epoch": 0.34620094090352516,
+      "grad_norm": 0.02928876168441755,
+      "learning_rate": 7.267454350161118e-06,
+      "loss": 0.0002,
+      "step": 5372
+    },
+    {
+      "epoch": 0.34626538635045434,
+      "grad_norm": 0.03327625299827368,
+      "learning_rate": 7.266738274257071e-06,
+      "loss": 0.0,
+      "step": 5373
+    },
+    {
+      "epoch": 0.3463298317973835,
+      "grad_norm": 0.004407488781814231,
+      "learning_rate": 7.266022198353025e-06,
+      "loss": 0.0,
+      "step": 5374
+    },
+    {
+      "epoch": 0.3463942772443127,
+      "grad_norm": 0.00046352221804896843,
+      "learning_rate": 7.2653061224489805e-06,
+      "loss": 0.0,
+      "step": 5375
+    },
+    {
+      "epoch": 0.34645872269124184,
+      "grad_norm": 1.0442598045825846,
+      "learning_rate": 7.264590046544935e-06,
+      "loss": 0.0033,
+      "step": 5376
+    },
+    {
+      "epoch": 0.346523168138171,
+      "grad_norm": 0.0087028528833674,
+      "learning_rate": 7.263873970640889e-06,
+      "loss": 0.0001,
+      "step": 5377
+    },
+    {
+      "epoch": 0.3465876135851002,
+      "grad_norm": 0.04466082821105845,
+      "learning_rate": 7.263157894736843e-06,
+      "loss": 0.0,
+      "step": 5378
+    },
+    {
+      "epoch": 0.3466520590320294,
+      "grad_norm": 0.00022414270993919194,
+      "learning_rate": 7.262441818832797e-06,
+      "loss": 0.0,
+      "step": 5379
+    },
+    {
+      "epoch": 0.3467165044789586,
+      "grad_norm": 0.13606521537237062,
+      "learning_rate": 7.261725742928751e-06,
+      "loss": 0.0017,
+      "step": 5380
+    },
+    {
+      "epoch": 0.34678094992588776,
+      "grad_norm": 0.0032358758394306617,
+      "learning_rate": 7.2610096670247055e-06,
+      "loss": 0.0,
+      "step": 5381
+    },
+    {
+      "epoch": 0.3468453953728169,
+      "grad_norm": 0.10357858884203731,
+      "learning_rate": 7.26029359112066e-06,
+      "loss": 0.0012,
+      "step": 5382
+    },
+    {
+      "epoch": 0.3469098408197461,
+      "grad_norm": 9.994017630786523e-05,
+      "learning_rate": 7.259577515216614e-06,
+      "loss": 0.0,
+      "step": 5383
+    },
+    {
+      "epoch": 0.34697428626667526,
+      "grad_norm": 9.690410155464549e-05,
+      "learning_rate": 7.2588614393125675e-06,
+      "loss": 0.0,
+      "step": 5384
+    },
+    {
+      "epoch": 0.34703873171360444,
+      "grad_norm": 0.005215186637607239,
+      "learning_rate": 7.258145363408522e-06,
+      "loss": 0.0,
+      "step": 5385
+    },
+    {
+      "epoch": 0.3471031771605336,
+      "grad_norm": 0.0011850453431943883,
+      "learning_rate": 7.257429287504476e-06,
+      "loss": 0.0,
+      "step": 5386
+    },
+    {
+      "epoch": 0.34716762260746276,
+      "grad_norm": 0.0797241166546595,
+      "learning_rate": 7.25671321160043e-06,
+      "loss": 0.0005,
+      "step": 5387
+    },
+    {
+      "epoch": 0.34723206805439194,
+      "grad_norm": 0.14444908177693655,
+      "learning_rate": 7.255997135696385e-06,
+      "loss": 0.0022,
+      "step": 5388
+    },
+    {
+      "epoch": 0.3472965135013211,
+      "grad_norm": 0.0015494501170997498,
+      "learning_rate": 7.255281059792338e-06,
+      "loss": 0.0,
+      "step": 5389
+    },
+    {
+      "epoch": 0.3473609589482503,
+      "grad_norm": 0.0723502269609664,
+      "learning_rate": 7.2545649838882924e-06,
+      "loss": 0.0002,
+      "step": 5390
+    },
+    {
+      "epoch": 0.3474254043951795,
+      "grad_norm": 0.13437147135869884,
+      "learning_rate": 7.253848907984247e-06,
+      "loss": 0.002,
+      "step": 5391
+    },
+    {
+      "epoch": 0.3474898498421087,
+      "grad_norm": 0.07793027930166455,
+      "learning_rate": 7.253132832080201e-06,
+      "loss": 0.0005,
+      "step": 5392
+    },
+    {
+      "epoch": 0.3475542952890378,
+      "grad_norm": 0.0005459188861817741,
+      "learning_rate": 7.252416756176155e-06,
+      "loss": 0.0,
+      "step": 5393
+    },
+    {
+      "epoch": 0.347618740735967,
+      "grad_norm": 0.0005229998610681955,
+      "learning_rate": 7.251700680272109e-06,
+      "loss": 0.0,
+      "step": 5394
+    },
+    {
+      "epoch": 0.3476831861828962,
+      "grad_norm": 0.014832489254280722,
+      "learning_rate": 7.250984604368063e-06,
+      "loss": 0.0,
+      "step": 5395
+    },
+    {
+      "epoch": 0.34774763162982536,
+      "grad_norm": 0.001830075605712059,
+      "learning_rate": 7.250268528464017e-06,
+      "loss": 0.0,
+      "step": 5396
+    },
+    {
+      "epoch": 0.34781207707675454,
+      "grad_norm": 0.00035845492018812364,
+      "learning_rate": 7.249552452559972e-06,
+      "loss": 0.0015,
+      "step": 5397
+    },
+    {
+      "epoch": 0.34787652252368373,
+      "grad_norm": 0.22111401246031417,
+      "learning_rate": 7.248836376655927e-06,
+      "loss": 0.0018,
+      "step": 5398
+    },
+    {
+      "epoch": 0.34794096797061286,
+      "grad_norm": 0.34508336003421725,
+      "learning_rate": 7.248120300751881e-06,
+      "loss": 0.0013,
+      "step": 5399
+    },
+    {
+      "epoch": 0.34800541341754204,
+      "grad_norm": 0.11998791833124678,
+      "learning_rate": 7.247404224847835e-06,
+      "loss": 0.0003,
+      "step": 5400
+    },
+    {
+      "epoch": 0.3480698588644712,
+      "grad_norm": 0.0010438379812748842,
+      "learning_rate": 7.246688148943789e-06,
+      "loss": 0.0,
+      "step": 5401
+    },
+    {
+      "epoch": 0.3481343043114004,
+      "grad_norm": 0.0009734335987623431,
+      "learning_rate": 7.245972073039743e-06,
+      "loss": 0.0,
+      "step": 5402
+    },
+    {
+      "epoch": 0.3481987497583296,
+      "grad_norm": 0.0018563996895338923,
+      "learning_rate": 7.2452559971356975e-06,
+      "loss": 0.0,
+      "step": 5403
+    },
+    {
+      "epoch": 0.3482631952052587,
+      "grad_norm": 0.20984205384467244,
+      "learning_rate": 7.244539921231652e-06,
+      "loss": 0.0004,
+      "step": 5404
+    },
+    {
+      "epoch": 0.3483276406521879,
+      "grad_norm": 0.015448803826021713,
+      "learning_rate": 7.243823845327605e-06,
+      "loss": 0.0002,
+      "step": 5405
+    },
+    {
+      "epoch": 0.3483920860991171,
+      "grad_norm": 0.03937960007927294,
+      "learning_rate": 7.2431077694235595e-06,
+      "loss": 0.0,
+      "step": 5406
+    },
+    {
+      "epoch": 0.3484565315460463,
+      "grad_norm": 0.022501578310819947,
+      "learning_rate": 7.242391693519514e-06,
+      "loss": 0.0,
+      "step": 5407
+    },
+    {
+      "epoch": 0.34852097699297546,
+      "grad_norm": 0.03170367334764925,
+      "learning_rate": 7.241675617615468e-06,
+      "loss": 0.0016,
+      "step": 5408
+    },
+    {
+      "epoch": 0.34858542243990465,
+      "grad_norm": 0.0038783959984967516,
+      "learning_rate": 7.2409595417114224e-06,
+      "loss": 0.0,
+      "step": 5409
+    },
+    {
+      "epoch": 0.3486498678868338,
+      "grad_norm": 0.24674016584155095,
+      "learning_rate": 7.240243465807376e-06,
+      "loss": 0.0024,
+      "step": 5410
+    },
+    {
+      "epoch": 0.34871431333376296,
+      "grad_norm": 0.000528044999434226,
+      "learning_rate": 7.23952738990333e-06,
+      "loss": 0.0,
+      "step": 5411
+    },
+    {
+      "epoch": 0.34877875878069214,
+      "grad_norm": 0.04335431754627749,
+      "learning_rate": 7.2388113139992845e-06,
+      "loss": 0.0001,
+      "step": 5412
+    },
+    {
+      "epoch": 0.34884320422762133,
+      "grad_norm": 9.593325914361795e-05,
+      "learning_rate": 7.238095238095239e-06,
+      "loss": 0.0,
+      "step": 5413
+    },
+    {
+      "epoch": 0.3489076496745505,
+      "grad_norm": 0.0014304134812169755,
+      "learning_rate": 7.237379162191192e-06,
+      "loss": 0.0,
+      "step": 5414
+    },
+    {
+      "epoch": 0.34897209512147964,
+      "grad_norm": 0.004716281375558154,
+      "learning_rate": 7.2366630862871465e-06,
+      "loss": 0.0,
+      "step": 5415
+    },
+    {
+      "epoch": 0.3490365405684088,
+      "grad_norm": 0.0077143613239694445,
+      "learning_rate": 7.235947010383101e-06,
+      "loss": 0.0,
+      "step": 5416
+    },
+    {
+      "epoch": 0.349100986015338,
+      "grad_norm": 0.019198324219276017,
+      "learning_rate": 7.235230934479055e-06,
+      "loss": 0.0002,
+      "step": 5417
+    },
+    {
+      "epoch": 0.3491654314622672,
+      "grad_norm": 0.001355964329896379,
+      "learning_rate": 7.234514858575009e-06,
+      "loss": 0.0,
+      "step": 5418
+    },
+    {
+      "epoch": 0.3492298769091964,
+      "grad_norm": 0.022225915169465784,
+      "learning_rate": 7.233798782670963e-06,
+      "loss": 0.0,
+      "step": 5419
+    },
+    {
+      "epoch": 0.34929432235612556,
+      "grad_norm": 0.017610877868679407,
+      "learning_rate": 7.233082706766917e-06,
+      "loss": 0.0,
+      "step": 5420
+    },
+    {
+      "epoch": 0.3493587678030547,
+      "grad_norm": 0.007438692190637162,
+      "learning_rate": 7.232366630862872e-06,
+      "loss": 0.0,
+      "step": 5421
+    },
+    {
+      "epoch": 0.3494232132499839,
+      "grad_norm": 0.007550065798805919,
+      "learning_rate": 7.231650554958827e-06,
+      "loss": 0.0001,
+      "step": 5422
+    },
+    {
+      "epoch": 0.34948765869691306,
+      "grad_norm": 1.1858373848494488,
+      "learning_rate": 7.230934479054781e-06,
+      "loss": 0.0052,
+      "step": 5423
+    },
+    {
+      "epoch": 0.34955210414384225,
+      "grad_norm": 0.13834342205455954,
+      "learning_rate": 7.230218403150735e-06,
+      "loss": 0.0013,
+      "step": 5424
+    },
+    {
+      "epoch": 0.34961654959077143,
+      "grad_norm": 0.0019500085784708264,
+      "learning_rate": 7.2295023272466895e-06,
+      "loss": 0.0,
+      "step": 5425
+    },
+    {
+      "epoch": 0.34968099503770056,
+      "grad_norm": 0.274820237577197,
+      "learning_rate": 7.228786251342643e-06,
+      "loss": 0.0014,
+      "step": 5426
+    },
+    {
+      "epoch": 0.34974544048462974,
+      "grad_norm": 0.16902668597399362,
+      "learning_rate": 7.228070175438597e-06,
+      "loss": 0.0004,
+      "step": 5427
+    },
+    {
+      "epoch": 0.34980988593155893,
+      "grad_norm": 0.0031173492434176304,
+      "learning_rate": 7.2273540995345516e-06,
+      "loss": 0.0,
+      "step": 5428
+    },
+    {
+      "epoch": 0.3498743313784881,
+      "grad_norm": 0.0015306366566559873,
+      "learning_rate": 7.226638023630506e-06,
+      "loss": 0.0,
+      "step": 5429
+    },
+    {
+      "epoch": 0.3499387768254173,
+      "grad_norm": 0.0010650450143334291,
+      "learning_rate": 7.225921947726459e-06,
+      "loss": 0.0,
+      "step": 5430
+    },
+    {
+      "epoch": 0.3500032222723465,
+      "grad_norm": 0.00024179158076437042,
+      "learning_rate": 7.225205871822414e-06,
+      "loss": 0.0,
+      "step": 5431
+    },
+    {
+      "epoch": 0.3500676677192756,
+      "grad_norm": 0.007258163297915242,
+      "learning_rate": 7.224489795918368e-06,
+      "loss": 0.0,
+      "step": 5432
+    },
+    {
+      "epoch": 0.3501321131662048,
+      "grad_norm": 2.513392435115923,
+      "learning_rate": 7.223773720014322e-06,
+      "loss": 0.0187,
+      "step": 5433
+    },
+    {
+      "epoch": 0.350196558613134,
+      "grad_norm": 0.0008973330386759609,
+      "learning_rate": 7.2230576441102765e-06,
+      "loss": 0.0,
+      "step": 5434
+    },
+    {
+      "epoch": 0.35026100406006316,
+      "grad_norm": 0.011253913607002416,
+      "learning_rate": 7.22234156820623e-06,
+      "loss": 0.0,
+      "step": 5435
+    },
+    {
+      "epoch": 0.35032544950699235,
+      "grad_norm": 0.6856501140185183,
+      "learning_rate": 7.221625492302184e-06,
+      "loss": 0.0033,
+      "step": 5436
+    },
+    {
+      "epoch": 0.35038989495392153,
+      "grad_norm": 0.002744451414998837,
+      "learning_rate": 7.2209094163981386e-06,
+      "loss": 0.0,
+      "step": 5437
+    },
+    {
+      "epoch": 0.35045434040085066,
+      "grad_norm": 0.1043653877480623,
+      "learning_rate": 7.220193340494093e-06,
+      "loss": 0.0001,
+      "step": 5438
+    },
+    {
+      "epoch": 0.35051878584777985,
+      "grad_norm": 0.04470527560901967,
+      "learning_rate": 7.219477264590047e-06,
+      "loss": 0.0001,
+      "step": 5439
+    },
+    {
+      "epoch": 0.35058323129470903,
+      "grad_norm": 0.0028691901618445375,
+      "learning_rate": 7.218761188686001e-06,
+      "loss": 0.0,
+      "step": 5440
+    },
+    {
+      "epoch": 0.3506476767416382,
+      "grad_norm": 0.014048890434080821,
+      "learning_rate": 7.218045112781955e-06,
+      "loss": 0.0001,
+      "step": 5441
+    },
+    {
+      "epoch": 0.3507121221885674,
+      "grad_norm": 0.0001558104690521515,
+      "learning_rate": 7.217329036877909e-06,
+      "loss": 0.0,
+      "step": 5442
+    },
+    {
+      "epoch": 0.35077656763549653,
+      "grad_norm": 0.0006746233065462414,
+      "learning_rate": 7.2166129609738635e-06,
+      "loss": 0.0,
+      "step": 5443
+    },
+    {
+      "epoch": 0.3508410130824257,
+      "grad_norm": 0.0024750607397397654,
+      "learning_rate": 7.215896885069818e-06,
+      "loss": 0.0,
+      "step": 5444
+    },
+    {
+      "epoch": 0.3509054585293549,
+      "grad_norm": 0.006352597500439202,
+      "learning_rate": 7.215180809165773e-06,
+      "loss": 0.0,
+      "step": 5445
+    },
+    {
+      "epoch": 0.3509699039762841,
+      "grad_norm": 0.1022585984155468,
+      "learning_rate": 7.214464733261726e-06,
+      "loss": 0.0001,
+      "step": 5446
+    },
+    {
+      "epoch": 0.35103434942321327,
+      "grad_norm": 0.003986765644523545,
+      "learning_rate": 7.213748657357681e-06,
+      "loss": 0.0,
+      "step": 5447
+    },
+    {
+      "epoch": 0.35109879487014245,
+      "grad_norm": 0.35015860097590396,
+      "learning_rate": 7.213032581453635e-06,
+      "loss": 0.0015,
+      "step": 5448
+    },
+    {
+      "epoch": 0.3511632403170716,
+      "grad_norm": 0.004855895917974423,
+      "learning_rate": 7.212316505549589e-06,
+      "loss": 0.0,
+      "step": 5449
+    },
+    {
+      "epoch": 0.35122768576400076,
+      "grad_norm": 0.00476196278000269,
+      "learning_rate": 7.211600429645544e-06,
+      "loss": 0.0,
+      "step": 5450
+    },
+    {
+      "epoch": 0.35129213121092995,
+      "grad_norm": 4.2020598196443586e-05,
+      "learning_rate": 7.210884353741497e-06,
+      "loss": 0.0,
+      "step": 5451
+    },
+    {
+      "epoch": 0.35135657665785913,
+      "grad_norm": 0.000898003348724344,
+      "learning_rate": 7.210168277837451e-06,
+      "loss": 0.0,
+      "step": 5452
+    },
+    {
+      "epoch": 0.3514210221047883,
+      "grad_norm": 0.00018579366299213354,
+      "learning_rate": 7.209452201933406e-06,
+      "loss": 0.0,
+      "step": 5453
+    },
+    {
+      "epoch": 0.35148546755171745,
+      "grad_norm": 0.158084677871953,
+      "learning_rate": 7.20873612602936e-06,
+      "loss": 0.001,
+      "step": 5454
+    },
+    {
+      "epoch": 0.35154991299864663,
+      "grad_norm": 0.028924409174075277,
+      "learning_rate": 7.208020050125314e-06,
+      "loss": 0.0001,
+      "step": 5455
+    },
+    {
+      "epoch": 0.3516143584455758,
+      "grad_norm": 0.10685927277426384,
+      "learning_rate": 7.207303974221268e-06,
+      "loss": 0.0016,
+      "step": 5456
+    },
+    {
+      "epoch": 0.351678803892505,
+      "grad_norm": 0.003219382873925816,
+      "learning_rate": 7.206587898317222e-06,
+      "loss": 0.0,
+      "step": 5457
+    },
+    {
+      "epoch": 0.3517432493394342,
+      "grad_norm": 0.023248814491553707,
+      "learning_rate": 7.205871822413176e-06,
+      "loss": 0.0002,
+      "step": 5458
+    },
+    {
+      "epoch": 0.35180769478636337,
+      "grad_norm": 0.008260090754341405,
+      "learning_rate": 7.205155746509131e-06,
+      "loss": 0.0,
+      "step": 5459
+    },
+    {
+      "epoch": 0.3518721402332925,
+      "grad_norm": 0.02113740518555449,
+      "learning_rate": 7.204439670605085e-06,
+      "loss": 0.0002,
+      "step": 5460
+    },
+    {
+      "epoch": 0.3519365856802217,
+      "grad_norm": 0.00722376485948921,
+      "learning_rate": 7.203723594701038e-06,
+      "loss": 0.0001,
+      "step": 5461
+    },
+    {
+      "epoch": 0.35200103112715087,
+      "grad_norm": 0.0833738754293488,
+      "learning_rate": 7.203007518796993e-06,
+      "loss": 0.0001,
+      "step": 5462
+    },
+    {
+      "epoch": 0.35206547657408005,
+      "grad_norm": 0.00026359184926858333,
+      "learning_rate": 7.202291442892947e-06,
+      "loss": 0.0,
+      "step": 5463
+    },
+    {
+      "epoch": 0.35212992202100923,
+      "grad_norm": 0.14296674674876744,
+      "learning_rate": 7.201575366988901e-06,
+      "loss": 0.0005,
+      "step": 5464
+    },
+    {
+      "epoch": 0.35219436746793836,
+      "grad_norm": 0.0016080032851724442,
+      "learning_rate": 7.2008592910848555e-06,
+      "loss": 0.0,
+      "step": 5465
+    },
+    {
+      "epoch": 0.35225881291486755,
+      "grad_norm": 0.0009047653986284783,
+      "learning_rate": 7.200143215180809e-06,
+      "loss": 0.0,
+      "step": 5466
+    },
+    {
+      "epoch": 0.35232325836179673,
+      "grad_norm": 0.0308639411701821,
+      "learning_rate": 7.199427139276763e-06,
+      "loss": 0.0001,
+      "step": 5467
+    },
+    {
+      "epoch": 0.3523877038087259,
+      "grad_norm": 0.4904011446517464,
+      "learning_rate": 7.1987110633727184e-06,
+      "loss": 0.0024,
+      "step": 5468
+    },
+    {
+      "epoch": 0.3524521492556551,
+      "grad_norm": 0.04524271869956901,
+      "learning_rate": 7.197994987468673e-06,
+      "loss": 0.0001,
+      "step": 5469
+    },
+    {
+      "epoch": 0.3525165947025843,
+      "grad_norm": 0.04035843557537505,
+      "learning_rate": 7.197278911564627e-06,
+      "loss": 0.0002,
+      "step": 5470
+    },
+    {
+      "epoch": 0.3525810401495134,
+      "grad_norm": 0.006825177577924673,
+      "learning_rate": 7.196562835660581e-06,
+      "loss": 0.0,
+      "step": 5471
+    },
+    {
+      "epoch": 0.3526454855964426,
+      "grad_norm": 0.13986985453114945,
+      "learning_rate": 7.195846759756535e-06,
+      "loss": 0.0006,
+      "step": 5472
+    },
+    {
+      "epoch": 0.3527099310433718,
+      "grad_norm": 0.0017970094026375993,
+      "learning_rate": 7.195130683852489e-06,
+      "loss": 0.0,
+      "step": 5473
+    },
+    {
+      "epoch": 0.35277437649030097,
+      "grad_norm": 0.006094273782439035,
+      "learning_rate": 7.194414607948443e-06,
+      "loss": 0.0,
+      "step": 5474
+    },
+    {
+      "epoch": 0.35283882193723015,
+      "grad_norm": 0.008172755866807201,
+      "learning_rate": 7.193698532044398e-06,
+      "loss": 0.0001,
+      "step": 5475
+    },
+    {
+      "epoch": 0.35290326738415934,
+      "grad_norm": 5.158375250050825,
+      "learning_rate": 7.192982456140352e-06,
+      "loss": 0.0934,
+      "step": 5476
+    },
+    {
+      "epoch": 0.35296771283108846,
+      "grad_norm": 0.22788083518064708,
+      "learning_rate": 7.192266380236305e-06,
+      "loss": 0.0008,
+      "step": 5477
+    },
+    {
+      "epoch": 0.35303215827801765,
+      "grad_norm": 1.3529695452401322,
+      "learning_rate": 7.19155030433226e-06,
+      "loss": 0.0108,
+      "step": 5478
+    },
+    {
+      "epoch": 0.35309660372494683,
+      "grad_norm": 0.001636092079940293,
+      "learning_rate": 7.190834228428214e-06,
+      "loss": 0.0,
+      "step": 5479
+    },
+    {
+      "epoch": 0.353161049171876,
+      "grad_norm": 1.787735879849132e-05,
+      "learning_rate": 7.190118152524168e-06,
+      "loss": 0.0,
+      "step": 5480
+    },
+    {
+      "epoch": 0.3532254946188052,
+      "grad_norm": 0.004427899683739982,
+      "learning_rate": 7.189402076620123e-06,
+      "loss": 0.0,
+      "step": 5481
+    },
+    {
+      "epoch": 0.35328994006573433,
+      "grad_norm": 0.0019082844941634223,
+      "learning_rate": 7.188686000716076e-06,
+      "loss": 0.0,
+      "step": 5482
+    },
+    {
+      "epoch": 0.3533543855126635,
+      "grad_norm": 0.0046977578647996515,
+      "learning_rate": 7.18796992481203e-06,
+      "loss": 0.0,
+      "step": 5483
+    },
+    {
+      "epoch": 0.3534188309595927,
+      "grad_norm": 0.040097359922366724,
+      "learning_rate": 7.187253848907985e-06,
+      "loss": 0.0002,
+      "step": 5484
+    },
+    {
+      "epoch": 0.3534832764065219,
+      "grad_norm": 0.001663645825594052,
+      "learning_rate": 7.186537773003939e-06,
+      "loss": 0.0,
+      "step": 5485
+    },
+    {
+      "epoch": 0.35354772185345107,
+      "grad_norm": 0.07505223049971804,
+      "learning_rate": 7.185821697099893e-06,
+      "loss": 0.0007,
+      "step": 5486
+    },
+    {
+      "epoch": 0.35361216730038025,
+      "grad_norm": 3.3424625948722473,
+      "learning_rate": 7.185105621195847e-06,
+      "loss": 0.0171,
+      "step": 5487
+    },
+    {
+      "epoch": 0.3536766127473094,
+      "grad_norm": 0.002825290381998295,
+      "learning_rate": 7.184389545291801e-06,
+      "loss": 0.0,
+      "step": 5488
+    },
+    {
+      "epoch": 0.35374105819423857,
+      "grad_norm": 0.009157336320531529,
+      "learning_rate": 7.183673469387755e-06,
+      "loss": 0.0,
+      "step": 5489
+    },
+    {
+      "epoch": 0.35380550364116775,
+      "grad_norm": 0.002773670880667257,
+      "learning_rate": 7.18295739348371e-06,
+      "loss": 0.0,
+      "step": 5490
+    },
+    {
+      "epoch": 0.35386994908809694,
+      "grad_norm": 0.08855313046890825,
+      "learning_rate": 7.182241317579665e-06,
+      "loss": 0.0003,
+      "step": 5491
+    },
+    {
+      "epoch": 0.3539343945350261,
+      "grad_norm": 0.004166906161995427,
+      "learning_rate": 7.181525241675619e-06,
+      "loss": 0.0,
+      "step": 5492
+    },
+    {
+      "epoch": 0.35399883998195525,
+      "grad_norm": 0.005168623229808937,
+      "learning_rate": 7.1808091657715725e-06,
+      "loss": 0.0,
+      "step": 5493
+    },
+    {
+      "epoch": 0.35406328542888443,
+      "grad_norm": 0.003997300840861494,
+      "learning_rate": 7.180093089867527e-06,
+      "loss": 0.0,
+      "step": 5494
+    },
+    {
+      "epoch": 0.3541277308758136,
+      "grad_norm": 0.009798578823934631,
+      "learning_rate": 7.179377013963481e-06,
+      "loss": 0.0,
+      "step": 5495
+    },
+    {
+      "epoch": 0.3541921763227428,
+      "grad_norm": 0.003605997833364615,
+      "learning_rate": 7.178660938059435e-06,
+      "loss": 0.0,
+      "step": 5496
+    },
+    {
+      "epoch": 0.354256621769672,
+      "grad_norm": 0.39026054546592165,
+      "learning_rate": 7.17794486215539e-06,
+      "loss": 0.0028,
+      "step": 5497
+    },
+    {
+      "epoch": 0.35432106721660117,
+      "grad_norm": 0.22522868316952996,
+      "learning_rate": 7.177228786251343e-06,
+      "loss": 0.0006,
+      "step": 5498
+    },
+    {
+      "epoch": 0.3543855126635303,
+      "grad_norm": 0.0007982156110889913,
+      "learning_rate": 7.1765127103472975e-06,
+      "loss": 0.0,
+      "step": 5499
+    },
+    {
+      "epoch": 0.3544499581104595,
+      "grad_norm": 0.26714035756006893,
+      "learning_rate": 7.175796634443252e-06,
+      "loss": 0.002,
+      "step": 5500
+    },
+    {
+      "epoch": 0.35451440355738867,
+      "grad_norm": 0.0017835782053076952,
+      "learning_rate": 7.175080558539206e-06,
+      "loss": 0.0,
+      "step": 5501
+    },
+    {
+      "epoch": 0.35457884900431785,
+      "grad_norm": 0.001526153733418559,
+      "learning_rate": 7.17436448263516e-06,
+      "loss": 0.0,
+      "step": 5502
+    },
+    {
+      "epoch": 0.35464329445124704,
+      "grad_norm": 0.006026490988079468,
+      "learning_rate": 7.173648406731114e-06,
+      "loss": 0.0,
+      "step": 5503
+    },
+    {
+      "epoch": 0.35470773989817617,
+      "grad_norm": 0.017582708992891122,
+      "learning_rate": 7.172932330827068e-06,
+      "loss": 0.0015,
+      "step": 5504
+    },
+    {
+      "epoch": 0.35477218534510535,
+      "grad_norm": 0.30581867098200244,
+      "learning_rate": 7.172216254923022e-06,
+      "loss": 0.0019,
+      "step": 5505
+    },
+    {
+      "epoch": 0.35483663079203454,
+      "grad_norm": 0.0036598724000132385,
+      "learning_rate": 7.171500179018977e-06,
+      "loss": 0.0,
+      "step": 5506
+    },
+    {
+      "epoch": 0.3549010762389637,
+      "grad_norm": 0.0026242730294129815,
+      "learning_rate": 7.17078410311493e-06,
+      "loss": 0.0,
+      "step": 5507
+    },
+    {
+      "epoch": 0.3549655216858929,
+      "grad_norm": 0.0011075331112727902,
+      "learning_rate": 7.1700680272108844e-06,
+      "loss": 0.0,
+      "step": 5508
+    },
+    {
+      "epoch": 0.3550299671328221,
+      "grad_norm": 0.19952681974901088,
+      "learning_rate": 7.169351951306839e-06,
+      "loss": 0.0004,
+      "step": 5509
+    },
+    {
+      "epoch": 0.3550944125797512,
+      "grad_norm": 0.000878946801695103,
+      "learning_rate": 7.168635875402793e-06,
+      "loss": 0.0,
+      "step": 5510
+    },
+    {
+      "epoch": 0.3551588580266804,
+      "grad_norm": 0.24027537949617142,
+      "learning_rate": 7.167919799498747e-06,
+      "loss": 0.0018,
+      "step": 5511
+    },
+    {
+      "epoch": 0.3552233034736096,
+      "grad_norm": 0.015573056221064105,
+      "learning_rate": 7.167203723594701e-06,
+      "loss": 0.0,
+      "step": 5512
+    },
+    {
+      "epoch": 0.35528774892053877,
+      "grad_norm": 0.012980932203767375,
+      "learning_rate": 7.166487647690655e-06,
+      "loss": 0.0,
+      "step": 5513
+    },
+    {
+      "epoch": 0.35535219436746796,
+      "grad_norm": 0.02937732815710757,
+      "learning_rate": 7.165771571786609e-06,
+      "loss": 0.0001,
+      "step": 5514
+    },
+    {
+      "epoch": 0.35541663981439714,
+      "grad_norm": 0.003133471086012606,
+      "learning_rate": 7.1650554958825645e-06,
+      "loss": 0.0,
+      "step": 5515
+    },
+    {
+      "epoch": 0.35548108526132627,
+      "grad_norm": 0.0001353438010940419,
+      "learning_rate": 7.164339419978519e-06,
+      "loss": 0.0,
+      "step": 5516
+    },
+    {
+      "epoch": 0.35554553070825545,
+      "grad_norm": 0.0015784288790910163,
+      "learning_rate": 7.163623344074473e-06,
+      "loss": 0.0,
+      "step": 5517
+    },
+    {
+      "epoch": 0.35560997615518464,
+      "grad_norm": 0.25518015621623313,
+      "learning_rate": 7.1629072681704274e-06,
+      "loss": 0.0006,
+      "step": 5518
+    },
+    {
+      "epoch": 0.3556744216021138,
+      "grad_norm": 0.05870437907137682,
+      "learning_rate": 7.162191192266381e-06,
+      "loss": 0.0014,
+      "step": 5519
+    },
+    {
+      "epoch": 0.355738867049043,
+      "grad_norm": 0.008105418444959963,
+      "learning_rate": 7.161475116362335e-06,
+      "loss": 0.0,
+      "step": 5520
+    },
+    {
+      "epoch": 0.35580331249597213,
+      "grad_norm": 0.07082251511150572,
+      "learning_rate": 7.1607590404582895e-06,
+      "loss": 0.002,
+      "step": 5521
+    },
+    {
+      "epoch": 0.3558677579429013,
+      "grad_norm": 0.025880416565134647,
+      "learning_rate": 7.160042964554244e-06,
+      "loss": 0.0,
+      "step": 5522
+    },
+    {
+      "epoch": 0.3559322033898305,
+      "grad_norm": 0.0074415682950392205,
+      "learning_rate": 7.159326888650197e-06,
+      "loss": 0.0,
+      "step": 5523
+    },
+    {
+      "epoch": 0.3559966488367597,
+      "grad_norm": 0.010568843378593547,
+      "learning_rate": 7.1586108127461515e-06,
+      "loss": 0.0,
+      "step": 5524
+    },
+    {
+      "epoch": 0.3560610942836889,
+      "grad_norm": 0.02883981875404712,
+      "learning_rate": 7.157894736842106e-06,
+      "loss": 0.0001,
+      "step": 5525
+    },
+    {
+      "epoch": 0.35612553973061806,
+      "grad_norm": 0.1443228572447026,
+      "learning_rate": 7.15717866093806e-06,
+      "loss": 0.0004,
+      "step": 5526
+    },
+    {
+      "epoch": 0.3561899851775472,
+      "grad_norm": 0.1543692648358986,
+      "learning_rate": 7.1564625850340144e-06,
+      "loss": 0.0007,
+      "step": 5527
+    },
+    {
+      "epoch": 0.35625443062447637,
+      "grad_norm": 0.006400996936471855,
+      "learning_rate": 7.155746509129968e-06,
+      "loss": 0.0,
+      "step": 5528
+    },
+    {
+      "epoch": 0.35631887607140555,
+      "grad_norm": 0.023857215512726463,
+      "learning_rate": 7.155030433225922e-06,
+      "loss": 0.0,
+      "step": 5529
+    },
+    {
+      "epoch": 0.35638332151833474,
+      "grad_norm": 0.0010024296612209454,
+      "learning_rate": 7.1543143573218765e-06,
+      "loss": 0.0,
+      "step": 5530
+    },
+    {
+      "epoch": 0.3564477669652639,
+      "grad_norm": 0.46602568507627373,
+      "learning_rate": 7.153598281417831e-06,
+      "loss": 0.0007,
+      "step": 5531
+    },
+    {
+      "epoch": 0.35651221241219305,
+      "grad_norm": 0.007157312719751308,
+      "learning_rate": 7.152882205513785e-06,
+      "loss": 0.0,
+      "step": 5532
+    },
+    {
+      "epoch": 0.35657665785912224,
+      "grad_norm": 0.0022695185644662445,
+      "learning_rate": 7.1521661296097385e-06,
+      "loss": 0.0,
+      "step": 5533
+    },
+    {
+      "epoch": 0.3566411033060514,
+      "grad_norm": 0.0007219926736973189,
+      "learning_rate": 7.151450053705693e-06,
+      "loss": 0.0,
+      "step": 5534
+    },
+    {
+      "epoch": 0.3567055487529806,
+      "grad_norm": 0.007866555084452822,
+      "learning_rate": 7.150733977801647e-06,
+      "loss": 0.0,
+      "step": 5535
+    },
+    {
+      "epoch": 0.3567699941999098,
+      "grad_norm": 0.001497095775340077,
+      "learning_rate": 7.150017901897601e-06,
+      "loss": 0.0,
+      "step": 5536
+    },
+    {
+      "epoch": 0.356834439646839,
+      "grad_norm": 0.0021093867092792707,
+      "learning_rate": 7.149301825993556e-06,
+      "loss": 0.0,
+      "step": 5537
+    },
+    {
+      "epoch": 0.3568988850937681,
+      "grad_norm": 0.0030429109849377583,
+      "learning_rate": 7.148585750089511e-06,
+      "loss": 0.0,
+      "step": 5538
+    },
+    {
+      "epoch": 0.3569633305406973,
+      "grad_norm": 0.005652474576479686,
+      "learning_rate": 7.147869674185464e-06,
+      "loss": 0.0,
+      "step": 5539
+    },
+    {
+      "epoch": 0.3570277759876265,
+      "grad_norm": 0.0002939338741031065,
+      "learning_rate": 7.147153598281419e-06,
+      "loss": 0.0,
+      "step": 5540
+    },
+    {
+      "epoch": 0.35709222143455566,
+      "grad_norm": 0.0005272829418229353,
+      "learning_rate": 7.146437522377373e-06,
+      "loss": 0.0,
+      "step": 5541
+    },
+    {
+      "epoch": 0.35715666688148484,
+      "grad_norm": 0.0014691027056069703,
+      "learning_rate": 7.145721446473327e-06,
+      "loss": 0.0,
+      "step": 5542
+    },
+    {
+      "epoch": 0.35722111232841397,
+      "grad_norm": 0.0019714445798182743,
+      "learning_rate": 7.1450053705692815e-06,
+      "loss": 0.0,
+      "step": 5543
+    },
+    {
+      "epoch": 0.35728555777534315,
+      "grad_norm": 0.0017453958343376377,
+      "learning_rate": 7.144289294665235e-06,
+      "loss": 0.0,
+      "step": 5544
+    },
+    {
+      "epoch": 0.35735000322227234,
+      "grad_norm": 0.009047168410867279,
+      "learning_rate": 7.143573218761189e-06,
+      "loss": 0.0,
+      "step": 5545
+    },
+    {
+      "epoch": 0.3574144486692015,
+      "grad_norm": 0.2842775943064396,
+      "learning_rate": 7.1428571428571436e-06,
+      "loss": 0.0018,
+      "step": 5546
+    },
+    {
+      "epoch": 0.3574788941161307,
+      "grad_norm": 0.001421204330789419,
+      "learning_rate": 7.142141066953098e-06,
+      "loss": 0.0,
+      "step": 5547
+    },
+    {
+      "epoch": 0.3575433395630599,
+      "grad_norm": 0.003525207027181184,
+      "learning_rate": 7.141424991049052e-06,
+      "loss": 0.0,
+      "step": 5548
+    },
+    {
+      "epoch": 0.357607785009989,
+      "grad_norm": 0.05103926720907329,
+      "learning_rate": 7.140708915145006e-06,
+      "loss": 0.0001,
+      "step": 5549
+    },
+    {
+      "epoch": 0.3576722304569182,
+      "grad_norm": 0.5437392661251964,
+      "learning_rate": 7.13999283924096e-06,
+      "loss": 0.001,
+      "step": 5550
+    },
+    {
+      "epoch": 0.3577366759038474,
+      "grad_norm": 0.0900597256928726,
+      "learning_rate": 7.139276763336914e-06,
+      "loss": 0.0006,
+      "step": 5551
+    },
+    {
+      "epoch": 0.3578011213507766,
+      "grad_norm": 0.0013100415151973432,
+      "learning_rate": 7.1385606874328685e-06,
+      "loss": 0.0,
+      "step": 5552
+    },
+    {
+      "epoch": 0.35786556679770576,
+      "grad_norm": 0.287950904363612,
+      "learning_rate": 7.137844611528823e-06,
+      "loss": 0.0014,
+      "step": 5553
+    },
+    {
+      "epoch": 0.35793001224463494,
+      "grad_norm": 0.0008622037652708815,
+      "learning_rate": 7.137128535624776e-06,
+      "loss": 0.0,
+      "step": 5554
+    },
+    {
+      "epoch": 0.35799445769156407,
+      "grad_norm": 0.0015019054203790178,
+      "learning_rate": 7.1364124597207306e-06,
+      "loss": 0.0,
+      "step": 5555
+    },
+    {
+      "epoch": 0.35805890313849326,
+      "grad_norm": 0.020955137988755305,
+      "learning_rate": 7.135696383816685e-06,
+      "loss": 0.0001,
+      "step": 5556
+    },
+    {
+      "epoch": 0.35812334858542244,
+      "grad_norm": 0.3577950619108931,
+      "learning_rate": 7.134980307912639e-06,
+      "loss": 0.0015,
+      "step": 5557
+    },
+    {
+      "epoch": 0.3581877940323516,
+      "grad_norm": 0.0007089497421827163,
+      "learning_rate": 7.1342642320085935e-06,
+      "loss": 0.0,
+      "step": 5558
+    },
+    {
+      "epoch": 0.3582522394792808,
+      "grad_norm": 0.0246501357034617,
+      "learning_rate": 7.133548156104547e-06,
+      "loss": 0.0001,
+      "step": 5559
+    },
+    {
+      "epoch": 0.35831668492620994,
+      "grad_norm": 0.5051342395314098,
+      "learning_rate": 7.132832080200501e-06,
+      "loss": 0.0037,
+      "step": 5560
+    },
+    {
+      "epoch": 0.3583811303731391,
+      "grad_norm": 0.02137042368144486,
+      "learning_rate": 7.1321160042964555e-06,
+      "loss": 0.0002,
+      "step": 5561
+    },
+    {
+      "epoch": 0.3584455758200683,
+      "grad_norm": 0.0010559639345811118,
+      "learning_rate": 7.131399928392411e-06,
+      "loss": 0.0,
+      "step": 5562
+    },
+    {
+      "epoch": 0.3585100212669975,
+      "grad_norm": 0.559211362846977,
+      "learning_rate": 7.130683852488365e-06,
+      "loss": 0.0012,
+      "step": 5563
+    },
+    {
+      "epoch": 0.3585744667139267,
+      "grad_norm": 0.017202826393729526,
+      "learning_rate": 7.129967776584319e-06,
+      "loss": 0.0,
+      "step": 5564
+    },
+    {
+      "epoch": 0.35863891216085586,
+      "grad_norm": 0.005808692085628072,
+      "learning_rate": 7.129251700680273e-06,
+      "loss": 0.0,
+      "step": 5565
+    },
+    {
+      "epoch": 0.358703357607785,
+      "grad_norm": 0.06375497502894896,
+      "learning_rate": 7.128535624776227e-06,
+      "loss": 0.0002,
+      "step": 5566
+    },
+    {
+      "epoch": 0.3587678030547142,
+      "grad_norm": 0.0058093430870949725,
+      "learning_rate": 7.127819548872181e-06,
+      "loss": 0.0,
+      "step": 5567
+    },
+    {
+      "epoch": 0.35883224850164336,
+      "grad_norm": 0.032179049666154896,
+      "learning_rate": 7.127103472968136e-06,
+      "loss": 0.0,
+      "step": 5568
+    },
+    {
+      "epoch": 0.35889669394857254,
+      "grad_norm": 0.03255481688625018,
+      "learning_rate": 7.12638739706409e-06,
+      "loss": 0.0,
+      "step": 5569
+    },
+    {
+      "epoch": 0.3589611393955017,
+      "grad_norm": 0.03756726107666502,
+      "learning_rate": 7.125671321160043e-06,
+      "loss": 0.0002,
+      "step": 5570
+    },
+    {
+      "epoch": 0.35902558484243086,
+      "grad_norm": 0.0024971589116067573,
+      "learning_rate": 7.124955245255998e-06,
+      "loss": 0.0,
+      "step": 5571
+    },
+    {
+      "epoch": 0.35909003028936004,
+      "grad_norm": 0.00042623072682908026,
+      "learning_rate": 7.124239169351952e-06,
+      "loss": 0.0,
+      "step": 5572
+    },
+    {
+      "epoch": 0.3591544757362892,
+      "grad_norm": 0.0017245227030109347,
+      "learning_rate": 7.123523093447906e-06,
+      "loss": 0.0,
+      "step": 5573
+    },
+    {
+      "epoch": 0.3592189211832184,
+      "grad_norm": 3.358811867999132,
+      "learning_rate": 7.1228070175438605e-06,
+      "loss": 0.007,
+      "step": 5574
+    },
+    {
+      "epoch": 0.3592833666301476,
+      "grad_norm": 0.005075338771603977,
+      "learning_rate": 7.122090941639814e-06,
+      "loss": 0.0,
+      "step": 5575
+    },
+    {
+      "epoch": 0.3593478120770768,
+      "grad_norm": 0.0017143627481109437,
+      "learning_rate": 7.121374865735768e-06,
+      "loss": 0.0,
+      "step": 5576
+    },
+    {
+      "epoch": 0.3594122575240059,
+      "grad_norm": 0.0020707422092871275,
+      "learning_rate": 7.120658789831723e-06,
+      "loss": 0.0,
+      "step": 5577
+    },
+    {
+      "epoch": 0.3594767029709351,
+      "grad_norm": 0.0007588354010330688,
+      "learning_rate": 7.119942713927677e-06,
+      "loss": 0.0,
+      "step": 5578
+    },
+    {
+      "epoch": 0.3595411484178643,
+      "grad_norm": 0.08036000371090321,
+      "learning_rate": 7.119226638023631e-06,
+      "loss": 0.0004,
+      "step": 5579
+    },
+    {
+      "epoch": 0.35960559386479346,
+      "grad_norm": 0.00012982236447759736,
+      "learning_rate": 7.118510562119585e-06,
+      "loss": 0.0,
+      "step": 5580
+    },
+    {
+      "epoch": 0.35967003931172264,
+      "grad_norm": 0.029504824766621456,
+      "learning_rate": 7.117794486215539e-06,
+      "loss": 0.0002,
+      "step": 5581
+    },
+    {
+      "epoch": 0.3597344847586518,
+      "grad_norm": 0.0033183237279156897,
+      "learning_rate": 7.117078410311493e-06,
+      "loss": 0.0,
+      "step": 5582
+    },
+    {
+      "epoch": 0.35979893020558096,
+      "grad_norm": 0.004723393791790135,
+      "learning_rate": 7.1163623344074475e-06,
+      "loss": 0.0,
+      "step": 5583
+    },
+    {
+      "epoch": 0.35986337565251014,
+      "grad_norm": 0.08625253997566465,
+      "learning_rate": 7.115646258503401e-06,
+      "loss": 0.0003,
+      "step": 5584
+    },
+    {
+      "epoch": 0.3599278210994393,
+      "grad_norm": 0.009677660664644353,
+      "learning_rate": 7.114930182599357e-06,
+      "loss": 0.0,
+      "step": 5585
+    },
+    {
+      "epoch": 0.3599922665463685,
+      "grad_norm": 0.00018866588452496163,
+      "learning_rate": 7.1142141066953104e-06,
+      "loss": 0.0,
+      "step": 5586
+    },
+    {
+      "epoch": 0.3600567119932977,
+      "grad_norm": 0.0016751659241547368,
+      "learning_rate": 7.113498030791265e-06,
+      "loss": 0.0,
+      "step": 5587
+    },
+    {
+      "epoch": 0.3601211574402268,
+      "grad_norm": 0.0017755997623683422,
+      "learning_rate": 7.112781954887219e-06,
+      "loss": 0.0,
+      "step": 5588
+    },
+    {
+      "epoch": 0.360185602887156,
+      "grad_norm": 0.013514959361658361,
+      "learning_rate": 7.112065878983173e-06,
+      "loss": 0.0,
+      "step": 5589
+    },
+    {
+      "epoch": 0.3602500483340852,
+      "grad_norm": 0.00037803318271111,
+      "learning_rate": 7.111349803079128e-06,
+      "loss": 0.0,
+      "step": 5590
+    },
+    {
+      "epoch": 0.3603144937810144,
+      "grad_norm": 0.050001389420844165,
+      "learning_rate": 7.110633727175081e-06,
+      "loss": 0.0002,
+      "step": 5591
+    },
+    {
+      "epoch": 0.36037893922794356,
+      "grad_norm": 0.0014661494256198607,
+      "learning_rate": 7.109917651271035e-06,
+      "loss": 0.0,
+      "step": 5592
+    },
+    {
+      "epoch": 0.36044338467487275,
+      "grad_norm": 0.0172112343024897,
+      "learning_rate": 7.10920157536699e-06,
+      "loss": 0.0,
+      "step": 5593
+    },
+    {
+      "epoch": 0.3605078301218019,
+      "grad_norm": 0.003739827721603054,
+      "learning_rate": 7.108485499462944e-06,
+      "loss": 0.0,
+      "step": 5594
+    },
+    {
+      "epoch": 0.36057227556873106,
+      "grad_norm": 0.0035318203953639943,
+      "learning_rate": 7.107769423558898e-06,
+      "loss": 0.0,
+      "step": 5595
+    },
+    {
+      "epoch": 0.36063672101566024,
+      "grad_norm": 0.024730250178133278,
+      "learning_rate": 7.107053347654852e-06,
+      "loss": 0.0,
+      "step": 5596
+    },
+    {
+      "epoch": 0.36070116646258943,
+      "grad_norm": 0.00018136025456127962,
+      "learning_rate": 7.106337271750806e-06,
+      "loss": 0.0,
+      "step": 5597
+    },
+    {
+      "epoch": 0.3607656119095186,
+      "grad_norm": 0.006373948769455688,
+      "learning_rate": 7.10562119584676e-06,
+      "loss": 0.0015,
+      "step": 5598
+    },
+    {
+      "epoch": 0.36083005735644774,
+      "grad_norm": 0.0007691793894941653,
+      "learning_rate": 7.104905119942715e-06,
+      "loss": 0.0,
+      "step": 5599
+    },
+    {
+      "epoch": 0.3608945028033769,
+      "grad_norm": 0.009427548311023269,
+      "learning_rate": 7.104189044038668e-06,
+      "loss": 0.0,
+      "step": 5600
+    },
+    {
+      "epoch": 0.3609589482503061,
+      "grad_norm": 0.14387279738820716,
+      "learning_rate": 7.103472968134622e-06,
+      "loss": 0.0001,
+      "step": 5601
+    },
+    {
+      "epoch": 0.3610233936972353,
+      "grad_norm": 0.060341764475685185,
+      "learning_rate": 7.102756892230577e-06,
+      "loss": 0.0016,
+      "step": 5602
+    },
+    {
+      "epoch": 0.3610878391441645,
+      "grad_norm": 0.00019170525434062738,
+      "learning_rate": 7.102040816326531e-06,
+      "loss": 0.0,
+      "step": 5603
+    },
+    {
+      "epoch": 0.36115228459109366,
+      "grad_norm": 0.018135760065123022,
+      "learning_rate": 7.101324740422485e-06,
+      "loss": 0.0,
+      "step": 5604
+    },
+    {
+      "epoch": 0.3612167300380228,
+      "grad_norm": 0.07888118928994316,
+      "learning_rate": 7.100608664518439e-06,
+      "loss": 0.0001,
+      "step": 5605
+    },
+    {
+      "epoch": 0.361281175484952,
+      "grad_norm": 0.0009429972317331661,
+      "learning_rate": 7.099892588614393e-06,
+      "loss": 0.0,
+      "step": 5606
+    },
+    {
+      "epoch": 0.36134562093188116,
+      "grad_norm": 0.0005612899489036782,
+      "learning_rate": 7.099176512710347e-06,
+      "loss": 0.0,
+      "step": 5607
+    },
+    {
+      "epoch": 0.36141006637881035,
+      "grad_norm": 0.0023027160039246065,
+      "learning_rate": 7.0984604368063025e-06,
+      "loss": 0.0,
+      "step": 5608
+    },
+    {
+      "epoch": 0.36147451182573953,
+      "grad_norm": 0.0004911979683064441,
+      "learning_rate": 7.097744360902257e-06,
+      "loss": 0.0,
+      "step": 5609
+    },
+    {
+      "epoch": 0.36153895727266866,
+      "grad_norm": 0.15138024304600262,
+      "learning_rate": 7.097028284998211e-06,
+      "loss": 0.0017,
+      "step": 5610
+    },
+    {
+      "epoch": 0.36160340271959784,
+      "grad_norm": 0.007569795518407454,
+      "learning_rate": 7.096312209094165e-06,
+      "loss": 0.0,
+      "step": 5611
+    },
+    {
+      "epoch": 0.36166784816652703,
+      "grad_norm": 0.009712033820648556,
+      "learning_rate": 7.095596133190119e-06,
+      "loss": 0.0,
+      "step": 5612
+    },
+    {
+      "epoch": 0.3617322936134562,
+      "grad_norm": 0.12356386491916947,
+      "learning_rate": 7.094880057286073e-06,
+      "loss": 0.0003,
+      "step": 5613
+    },
+    {
+      "epoch": 0.3617967390603854,
+      "grad_norm": 0.0013265063818277874,
+      "learning_rate": 7.094163981382027e-06,
+      "loss": 0.0,
+      "step": 5614
+    },
+    {
+      "epoch": 0.3618611845073146,
+      "grad_norm": 0.0016747608507945689,
+      "learning_rate": 7.093447905477982e-06,
+      "loss": 0.0,
+      "step": 5615
+    },
+    {
+      "epoch": 0.3619256299542437,
+      "grad_norm": 0.0008821640464383144,
+      "learning_rate": 7.092731829573936e-06,
+      "loss": 0.0,
+      "step": 5616
+    },
+    {
+      "epoch": 0.3619900754011729,
+      "grad_norm": 0.00030392545722019647,
+      "learning_rate": 7.0920157536698895e-06,
+      "loss": 0.0,
+      "step": 5617
+    },
+    {
+      "epoch": 0.3620545208481021,
+      "grad_norm": 0.0054529868195787165,
+      "learning_rate": 7.091299677765844e-06,
+      "loss": 0.0,
+      "step": 5618
+    },
+    {
+      "epoch": 0.36211896629503126,
+      "grad_norm": 0.13914422730415102,
+      "learning_rate": 7.090583601861798e-06,
+      "loss": 0.0001,
+      "step": 5619
+    },
+    {
+      "epoch": 0.36218341174196045,
+      "grad_norm": 0.02705458926689199,
+      "learning_rate": 7.089867525957752e-06,
+      "loss": 0.0001,
+      "step": 5620
+    },
+    {
+      "epoch": 0.36224785718888963,
+      "grad_norm": 0.13120434966247765,
+      "learning_rate": 7.089151450053706e-06,
+      "loss": 0.0008,
+      "step": 5621
+    },
+    {
+      "epoch": 0.36231230263581876,
+      "grad_norm": 0.12314011810163554,
+      "learning_rate": 7.08843537414966e-06,
+      "loss": 0.0015,
+      "step": 5622
+    },
+    {
+      "epoch": 0.36237674808274795,
+      "grad_norm": 0.01434271131636332,
+      "learning_rate": 7.087719298245614e-06,
+      "loss": 0.0001,
+      "step": 5623
+    },
+    {
+      "epoch": 0.36244119352967713,
+      "grad_norm": 0.021528547431360342,
+      "learning_rate": 7.087003222341569e-06,
+      "loss": 0.0,
+      "step": 5624
+    },
+    {
+      "epoch": 0.3625056389766063,
+      "grad_norm": 0.01073329697163935,
+      "learning_rate": 7.086287146437523e-06,
+      "loss": 0.0,
+      "step": 5625
+    },
+    {
+      "epoch": 0.3625700844235355,
+      "grad_norm": 1.960513490832337,
+      "learning_rate": 7.0855710705334764e-06,
+      "loss": 0.0038,
+      "step": 5626
+    },
+    {
+      "epoch": 0.36263452987046463,
+      "grad_norm": 0.025460006296816034,
+      "learning_rate": 7.084854994629431e-06,
+      "loss": 0.0,
+      "step": 5627
+    },
+    {
+      "epoch": 0.3626989753173938,
+      "grad_norm": 0.20646837479210997,
+      "learning_rate": 7.084138918725385e-06,
+      "loss": 0.0004,
+      "step": 5628
+    },
+    {
+      "epoch": 0.362763420764323,
+      "grad_norm": 0.25142192289102916,
+      "learning_rate": 7.083422842821339e-06,
+      "loss": 0.001,
+      "step": 5629
+    },
+    {
+      "epoch": 0.3628278662112522,
+      "grad_norm": 0.006383531148001112,
+      "learning_rate": 7.082706766917294e-06,
+      "loss": 0.0,
+      "step": 5630
+    },
+    {
+      "epoch": 0.36289231165818137,
+      "grad_norm": 0.0029308486291541765,
+      "learning_rate": 7.081990691013247e-06,
+      "loss": 0.0,
+      "step": 5631
+    },
+    {
+      "epoch": 0.36295675710511055,
+      "grad_norm": 0.0007120470554510387,
+      "learning_rate": 7.081274615109203e-06,
+      "loss": 0.0,
+      "step": 5632
+    },
+    {
+      "epoch": 0.3630212025520397,
+      "grad_norm": 0.0012225325945689409,
+      "learning_rate": 7.0805585392051565e-06,
+      "loss": 0.0,
+      "step": 5633
+    },
+    {
+      "epoch": 0.36308564799896886,
+      "grad_norm": 0.019594588990991643,
+      "learning_rate": 7.079842463301111e-06,
+      "loss": 0.0001,
+      "step": 5634
+    },
+    {
+      "epoch": 0.36315009344589805,
+      "grad_norm": 0.003277557610881603,
+      "learning_rate": 7.079126387397065e-06,
+      "loss": 0.0,
+      "step": 5635
+    },
+    {
+      "epoch": 0.36321453889282723,
+      "grad_norm": 0.04047106517839488,
+      "learning_rate": 7.0784103114930194e-06,
+      "loss": 0.0001,
+      "step": 5636
+    },
+    {
+      "epoch": 0.3632789843397564,
+      "grad_norm": 0.006228676430910245,
+      "learning_rate": 7.077694235588973e-06,
+      "loss": 0.0,
+      "step": 5637
+    },
+    {
+      "epoch": 0.36334342978668555,
+      "grad_norm": 0.004968203124256135,
+      "learning_rate": 7.076978159684927e-06,
+      "loss": 0.0,
+      "step": 5638
+    },
+    {
+      "epoch": 0.36340787523361473,
+      "grad_norm": 0.004763822483671288,
+      "learning_rate": 7.0762620837808815e-06,
+      "loss": 0.0,
+      "step": 5639
+    },
+    {
+      "epoch": 0.3634723206805439,
+      "grad_norm": 0.47286776074125114,
+      "learning_rate": 7.075546007876836e-06,
+      "loss": 0.0032,
+      "step": 5640
+    },
+    {
+      "epoch": 0.3635367661274731,
+      "grad_norm": 0.21851157069067817,
+      "learning_rate": 7.07482993197279e-06,
+      "loss": 0.0024,
+      "step": 5641
+    },
+    {
+      "epoch": 0.3636012115744023,
+      "grad_norm": 0.000831134689962334,
+      "learning_rate": 7.0741138560687435e-06,
+      "loss": 0.0,
+      "step": 5642
+    },
+    {
+      "epoch": 0.36366565702133147,
+      "grad_norm": 0.0012023844883557063,
+      "learning_rate": 7.073397780164698e-06,
+      "loss": 0.0,
+      "step": 5643
+    },
+    {
+      "epoch": 0.3637301024682606,
+      "grad_norm": 0.00047426837176558134,
+      "learning_rate": 7.072681704260652e-06,
+      "loss": 0.0,
+      "step": 5644
+    },
+    {
+      "epoch": 0.3637945479151898,
+      "grad_norm": 0.0028780370890813933,
+      "learning_rate": 7.0719656283566064e-06,
+      "loss": 0.0,
+      "step": 5645
+    },
+    {
+      "epoch": 0.36385899336211897,
+      "grad_norm": 0.16501946869057837,
+      "learning_rate": 7.071249552452561e-06,
+      "loss": 0.0012,
+      "step": 5646
+    },
+    {
+      "epoch": 0.36392343880904815,
+      "grad_norm": 0.26383039160046534,
+      "learning_rate": 7.070533476548514e-06,
+      "loss": 0.0006,
+      "step": 5647
+    },
+    {
+      "epoch": 0.36398788425597733,
+      "grad_norm": 0.7559218269984767,
+      "learning_rate": 7.0698174006444685e-06,
+      "loss": 0.0082,
+      "step": 5648
+    },
+    {
+      "epoch": 0.36405232970290646,
+      "grad_norm": 0.11576493211318237,
+      "learning_rate": 7.069101324740423e-06,
+      "loss": 0.001,
+      "step": 5649
+    },
+    {
+      "epoch": 0.36411677514983565,
+      "grad_norm": 0.3949228552741045,
+      "learning_rate": 7.068385248836377e-06,
+      "loss": 0.0015,
+      "step": 5650
+    },
+    {
+      "epoch": 0.36418122059676483,
+      "grad_norm": 0.0030827061663709814,
+      "learning_rate": 7.067669172932331e-06,
+      "loss": 0.0,
+      "step": 5651
+    },
+    {
+      "epoch": 0.364245666043694,
+      "grad_norm": 0.05707504512960632,
+      "learning_rate": 7.066953097028285e-06,
+      "loss": 0.0044,
+      "step": 5652
+    },
+    {
+      "epoch": 0.3643101114906232,
+      "grad_norm": 0.0029535415213127816,
+      "learning_rate": 7.066237021124239e-06,
+      "loss": 0.0,
+      "step": 5653
+    },
+    {
+      "epoch": 0.3643745569375524,
+      "grad_norm": 0.6487911363102642,
+      "learning_rate": 7.065520945220193e-06,
+      "loss": 0.0012,
+      "step": 5654
+    },
+    {
+      "epoch": 0.3644390023844815,
+      "grad_norm": 8.800743879255332e-05,
+      "learning_rate": 7.0648048693161486e-06,
+      "loss": 0.0,
+      "step": 5655
+    },
+    {
+      "epoch": 0.3645034478314107,
+      "grad_norm": 0.003355957561265595,
+      "learning_rate": 7.064088793412103e-06,
+      "loss": 0.0,
+      "step": 5656
+    },
+    {
+      "epoch": 0.3645678932783399,
+      "grad_norm": 0.007201152011268038,
+      "learning_rate": 7.063372717508057e-06,
+      "loss": 0.0,
+      "step": 5657
+    },
+    {
+      "epoch": 0.36463233872526907,
+      "grad_norm": 0.021779140689082172,
+      "learning_rate": 7.062656641604011e-06,
+      "loss": 0.0001,
+      "step": 5658
+    },
+    {
+      "epoch": 0.36469678417219825,
+      "grad_norm": 0.1626429279984157,
+      "learning_rate": 7.061940565699965e-06,
+      "loss": 0.0009,
+      "step": 5659
+    },
+    {
+      "epoch": 0.36476122961912744,
+      "grad_norm": 0.8840851851432028,
+      "learning_rate": 7.061224489795919e-06,
+      "loss": 0.0044,
+      "step": 5660
+    },
+    {
+      "epoch": 0.36482567506605657,
+      "grad_norm": 0.3878680980398993,
+      "learning_rate": 7.0605084138918735e-06,
+      "loss": 0.0022,
+      "step": 5661
+    },
+    {
+      "epoch": 0.36489012051298575,
+      "grad_norm": 0.010117391982138556,
+      "learning_rate": 7.059792337987828e-06,
+      "loss": 0.0,
+      "step": 5662
+    },
+    {
+      "epoch": 0.36495456595991493,
+      "grad_norm": 0.05438971549517224,
+      "learning_rate": 7.059076262083781e-06,
+      "loss": 0.0002,
+      "step": 5663
+    },
+    {
+      "epoch": 0.3650190114068441,
+      "grad_norm": 0.0023272584511669445,
+      "learning_rate": 7.0583601861797356e-06,
+      "loss": 0.0,
+      "step": 5664
+    },
+    {
+      "epoch": 0.3650834568537733,
+      "grad_norm": 0.008117755104337068,
+      "learning_rate": 7.05764411027569e-06,
+      "loss": 0.0,
+      "step": 5665
+    },
+    {
+      "epoch": 0.36514790230070243,
+      "grad_norm": 0.1603883944921648,
+      "learning_rate": 7.056928034371644e-06,
+      "loss": 0.0011,
+      "step": 5666
+    },
+    {
+      "epoch": 0.3652123477476316,
+      "grad_norm": 0.1780197137734566,
+      "learning_rate": 7.0562119584675985e-06,
+      "loss": 0.0018,
+      "step": 5667
+    },
+    {
+      "epoch": 0.3652767931945608,
+      "grad_norm": 0.05630731138250431,
+      "learning_rate": 7.055495882563552e-06,
+      "loss": 0.0001,
+      "step": 5668
+    },
+    {
+      "epoch": 0.36534123864149,
+      "grad_norm": 0.004059539192963445,
+      "learning_rate": 7.054779806659506e-06,
+      "loss": 0.0,
+      "step": 5669
+    },
+    {
+      "epoch": 0.36540568408841917,
+      "grad_norm": 0.2651977187882061,
+      "learning_rate": 7.0540637307554605e-06,
+      "loss": 0.0011,
+      "step": 5670
+    },
+    {
+      "epoch": 0.36547012953534835,
+      "grad_norm": 0.18956324729432547,
+      "learning_rate": 7.053347654851415e-06,
+      "loss": 0.0001,
+      "step": 5671
+    },
+    {
+      "epoch": 0.3655345749822775,
+      "grad_norm": 0.06572701027378827,
+      "learning_rate": 7.052631578947369e-06,
+      "loss": 0.0001,
+      "step": 5672
+    },
+    {
+      "epoch": 0.36559902042920667,
+      "grad_norm": 0.0016957948449696752,
+      "learning_rate": 7.0519155030433226e-06,
+      "loss": 0.0,
+      "step": 5673
+    },
+    {
+      "epoch": 0.36566346587613585,
+      "grad_norm": 0.017716966280276046,
+      "learning_rate": 7.051199427139277e-06,
+      "loss": 0.0001,
+      "step": 5674
+    },
+    {
+      "epoch": 0.36572791132306504,
+      "grad_norm": 0.001855080590519422,
+      "learning_rate": 7.050483351235231e-06,
+      "loss": 0.0,
+      "step": 5675
+    },
+    {
+      "epoch": 0.3657923567699942,
+      "grad_norm": 0.0014049639301356737,
+      "learning_rate": 7.0497672753311855e-06,
+      "loss": 0.0,
+      "step": 5676
+    },
+    {
+      "epoch": 0.36585680221692335,
+      "grad_norm": 0.006962279213334821,
+      "learning_rate": 7.04905119942714e-06,
+      "loss": 0.0,
+      "step": 5677
+    },
+    {
+      "epoch": 0.36592124766385253,
+      "grad_norm": 0.020090188529929526,
+      "learning_rate": 7.048335123523095e-06,
+      "loss": 0.0,
+      "step": 5678
+    },
+    {
+      "epoch": 0.3659856931107817,
+      "grad_norm": 0.10503809921713848,
+      "learning_rate": 7.047619047619048e-06,
+      "loss": 0.0002,
+      "step": 5679
+    },
+    {
+      "epoch": 0.3660501385577109,
+      "grad_norm": 0.004166492745336915,
+      "learning_rate": 7.046902971715003e-06,
+      "loss": 0.0,
+      "step": 5680
+    },
+    {
+      "epoch": 0.3661145840046401,
+      "grad_norm": 0.0009582915293647691,
+      "learning_rate": 7.046186895810957e-06,
+      "loss": 0.0,
+      "step": 5681
+    },
+    {
+      "epoch": 0.36617902945156927,
+      "grad_norm": 0.037545327208160215,
+      "learning_rate": 7.045470819906911e-06,
+      "loss": 0.0001,
+      "step": 5682
+    },
+    {
+      "epoch": 0.3662434748984984,
+      "grad_norm": 0.10824613353346786,
+      "learning_rate": 7.0447547440028655e-06,
+      "loss": 0.0003,
+      "step": 5683
+    },
+    {
+      "epoch": 0.3663079203454276,
+      "grad_norm": 0.0014164057504567702,
+      "learning_rate": 7.044038668098819e-06,
+      "loss": 0.0,
+      "step": 5684
+    },
+    {
+      "epoch": 0.36637236579235677,
+      "grad_norm": 0.729636427012959,
+      "learning_rate": 7.043322592194773e-06,
+      "loss": 0.0045,
+      "step": 5685
+    },
+    {
+      "epoch": 0.36643681123928595,
+      "grad_norm": 0.05235399263431841,
+      "learning_rate": 7.042606516290728e-06,
+      "loss": 0.0001,
+      "step": 5686
+    },
+    {
+      "epoch": 0.36650125668621514,
+      "grad_norm": 0.0010319380225699353,
+      "learning_rate": 7.041890440386682e-06,
+      "loss": 0.0,
+      "step": 5687
+    },
+    {
+      "epoch": 0.36656570213314427,
+      "grad_norm": 0.0012176516510191088,
+      "learning_rate": 7.041174364482636e-06,
+      "loss": 0.0,
+      "step": 5688
+    },
+    {
+      "epoch": 0.36663014758007345,
+      "grad_norm": 0.0004752984702248449,
+      "learning_rate": 7.04045828857859e-06,
+      "loss": 0.0,
+      "step": 5689
+    },
+    {
+      "epoch": 0.36669459302700264,
+      "grad_norm": 0.019731819460332183,
+      "learning_rate": 7.039742212674544e-06,
+      "loss": 0.0001,
+      "step": 5690
+    },
+    {
+      "epoch": 0.3667590384739318,
+      "grad_norm": 0.0057392792271502915,
+      "learning_rate": 7.039026136770498e-06,
+      "loss": 0.0,
+      "step": 5691
+    },
+    {
+      "epoch": 0.366823483920861,
+      "grad_norm": 0.46144410757981064,
+      "learning_rate": 7.0383100608664525e-06,
+      "loss": 0.0028,
+      "step": 5692
+    },
+    {
+      "epoch": 0.3668879293677902,
+      "grad_norm": 0.0238733995461195,
+      "learning_rate": 7.037593984962407e-06,
+      "loss": 0.0001,
+      "step": 5693
+    },
+    {
+      "epoch": 0.3669523748147193,
+      "grad_norm": 0.30712166491100057,
+      "learning_rate": 7.03687790905836e-06,
+      "loss": 0.0024,
+      "step": 5694
+    },
+    {
+      "epoch": 0.3670168202616485,
+      "grad_norm": 0.00197018473504978,
+      "learning_rate": 7.036161833154315e-06,
+      "loss": 0.0,
+      "step": 5695
+    },
+    {
+      "epoch": 0.3670812657085777,
+      "grad_norm": 0.007235413846213685,
+      "learning_rate": 7.035445757250269e-06,
+      "loss": 0.0,
+      "step": 5696
+    },
+    {
+      "epoch": 0.36714571115550687,
+      "grad_norm": 0.0008705440038263109,
+      "learning_rate": 7.034729681346223e-06,
+      "loss": 0.0,
+      "step": 5697
+    },
+    {
+      "epoch": 0.36721015660243606,
+      "grad_norm": 0.016319481775685544,
+      "learning_rate": 7.034013605442177e-06,
+      "loss": 0.0,
+      "step": 5698
+    },
+    {
+      "epoch": 0.36727460204936524,
+      "grad_norm": 0.060190379923427566,
+      "learning_rate": 7.033297529538131e-06,
+      "loss": 0.0001,
+      "step": 5699
+    },
+    {
+      "epoch": 0.36733904749629437,
+      "grad_norm": 0.07271395616914396,
+      "learning_rate": 7.032581453634085e-06,
+      "loss": 0.0001,
+      "step": 5700
+    },
+    {
+      "epoch": 0.36740349294322355,
+      "grad_norm": 0.004665248624292797,
+      "learning_rate": 7.0318653777300395e-06,
+      "loss": 0.0,
+      "step": 5701
+    },
+    {
+      "epoch": 0.36746793839015274,
+      "grad_norm": 0.08727091492317983,
+      "learning_rate": 7.031149301825995e-06,
+      "loss": 0.0001,
+      "step": 5702
+    },
+    {
+      "epoch": 0.3675323838370819,
+      "grad_norm": 0.00788484775339447,
+      "learning_rate": 7.030433225921949e-06,
+      "loss": 0.0,
+      "step": 5703
+    },
+    {
+      "epoch": 0.3675968292840111,
+      "grad_norm": 0.035432436020510494,
+      "learning_rate": 7.029717150017903e-06,
+      "loss": 0.0001,
+      "step": 5704
+    },
+    {
+      "epoch": 0.36766127473094024,
+      "grad_norm": 0.031635243714684846,
+      "learning_rate": 7.029001074113857e-06,
+      "loss": 0.0001,
+      "step": 5705
+    },
+    {
+      "epoch": 0.3677257201778694,
+      "grad_norm": 0.4187037698648444,
+      "learning_rate": 7.028284998209811e-06,
+      "loss": 0.0015,
+      "step": 5706
+    },
+    {
+      "epoch": 0.3677901656247986,
+      "grad_norm": 0.001485610405980241,
+      "learning_rate": 7.027568922305765e-06,
+      "loss": 0.0,
+      "step": 5707
+    },
+    {
+      "epoch": 0.3678546110717278,
+      "grad_norm": 0.0032538214693292274,
+      "learning_rate": 7.02685284640172e-06,
+      "loss": 0.0,
+      "step": 5708
+    },
+    {
+      "epoch": 0.367919056518657,
+      "grad_norm": 0.4946226701409237,
+      "learning_rate": 7.026136770497674e-06,
+      "loss": 0.0016,
+      "step": 5709
+    },
+    {
+      "epoch": 0.36798350196558616,
+      "grad_norm": 0.02520561557442119,
+      "learning_rate": 7.025420694593627e-06,
+      "loss": 0.0001,
+      "step": 5710
+    },
+    {
+      "epoch": 0.3680479474125153,
+      "grad_norm": 0.010406505327414488,
+      "learning_rate": 7.024704618689582e-06,
+      "loss": 0.0,
+      "step": 5711
+    },
+    {
+      "epoch": 0.36811239285944447,
+      "grad_norm": 0.009454967464173939,
+      "learning_rate": 7.023988542785536e-06,
+      "loss": 0.0,
+      "step": 5712
+    },
+    {
+      "epoch": 0.36817683830637365,
+      "grad_norm": 0.5461984263936956,
+      "learning_rate": 7.02327246688149e-06,
+      "loss": 0.0025,
+      "step": 5713
+    },
+    {
+      "epoch": 0.36824128375330284,
+      "grad_norm": 0.011822903242047569,
+      "learning_rate": 7.022556390977444e-06,
+      "loss": 0.0001,
+      "step": 5714
+    },
+    {
+      "epoch": 0.368305729200232,
+      "grad_norm": 0.002773322625933541,
+      "learning_rate": 7.021840315073398e-06,
+      "loss": 0.0,
+      "step": 5715
+    },
+    {
+      "epoch": 0.36837017464716115,
+      "grad_norm": 0.14855857353534296,
+      "learning_rate": 7.021124239169352e-06,
+      "loss": 0.0002,
+      "step": 5716
+    },
+    {
+      "epoch": 0.36843462009409034,
+      "grad_norm": 0.027646845724586845,
+      "learning_rate": 7.020408163265307e-06,
+      "loss": 0.0,
+      "step": 5717
+    },
+    {
+      "epoch": 0.3684990655410195,
+      "grad_norm": 0.0027564969182885253,
+      "learning_rate": 7.019692087361261e-06,
+      "loss": 0.0,
+      "step": 5718
+    },
+    {
+      "epoch": 0.3685635109879487,
+      "grad_norm": 0.0005528286557136186,
+      "learning_rate": 7.018976011457214e-06,
+      "loss": 0.0,
+      "step": 5719
+    },
+    {
+      "epoch": 0.3686279564348779,
+      "grad_norm": 1.5152654319623102,
+      "learning_rate": 7.018259935553169e-06,
+      "loss": 0.0032,
+      "step": 5720
+    },
+    {
+      "epoch": 0.3686924018818071,
+      "grad_norm": 0.0014066507855464159,
+      "learning_rate": 7.017543859649123e-06,
+      "loss": 0.0,
+      "step": 5721
+    },
+    {
+      "epoch": 0.3687568473287362,
+      "grad_norm": 0.0002940719089644006,
+      "learning_rate": 7.016827783745077e-06,
+      "loss": 0.0,
+      "step": 5722
+    },
+    {
+      "epoch": 0.3688212927756654,
+      "grad_norm": 0.7755664462750681,
+      "learning_rate": 7.0161117078410316e-06,
+      "loss": 0.0042,
+      "step": 5723
+    },
+    {
+      "epoch": 0.3688857382225946,
+      "grad_norm": 0.011148354046125965,
+      "learning_rate": 7.015395631936985e-06,
+      "loss": 0.0,
+      "step": 5724
+    },
+    {
+      "epoch": 0.36895018366952376,
+      "grad_norm": 0.037286561287033865,
+      "learning_rate": 7.014679556032941e-06,
+      "loss": 0.0001,
+      "step": 5725
+    },
+    {
+      "epoch": 0.36901462911645294,
+      "grad_norm": 0.28917756239200254,
+      "learning_rate": 7.0139634801288945e-06,
+      "loss": 0.001,
+      "step": 5726
+    },
+    {
+      "epoch": 0.36907907456338207,
+      "grad_norm": 0.02074656065833119,
+      "learning_rate": 7.013247404224849e-06,
+      "loss": 0.0001,
+      "step": 5727
+    },
+    {
+      "epoch": 0.36914352001031125,
+      "grad_norm": 0.009874683740494761,
+      "learning_rate": 7.012531328320803e-06,
+      "loss": 0.0,
+      "step": 5728
+    },
+    {
+      "epoch": 0.36920796545724044,
+      "grad_norm": 0.0010247783570523034,
+      "learning_rate": 7.011815252416757e-06,
+      "loss": 0.0,
+      "step": 5729
+    },
+    {
+      "epoch": 0.3692724109041696,
+      "grad_norm": 0.12852172866055342,
+      "learning_rate": 7.011099176512711e-06,
+      "loss": 0.0017,
+      "step": 5730
+    },
+    {
+      "epoch": 0.3693368563510988,
+      "grad_norm": 0.022266211156909587,
+      "learning_rate": 7.010383100608665e-06,
+      "loss": 0.0,
+      "step": 5731
+    },
+    {
+      "epoch": 0.369401301798028,
+      "grad_norm": 0.03223963132931994,
+      "learning_rate": 7.009667024704619e-06,
+      "loss": 0.0,
+      "step": 5732
+    },
+    {
+      "epoch": 0.3694657472449571,
+      "grad_norm": 0.01174380559418009,
+      "learning_rate": 7.008950948800574e-06,
+      "loss": 0.0,
+      "step": 5733
+    },
+    {
+      "epoch": 0.3695301926918863,
+      "grad_norm": 0.0016373883989292677,
+      "learning_rate": 7.008234872896528e-06,
+      "loss": 0.0,
+      "step": 5734
+    },
+    {
+      "epoch": 0.3695946381388155,
+      "grad_norm": 0.015438489387722118,
+      "learning_rate": 7.0075187969924815e-06,
+      "loss": 0.0,
+      "step": 5735
+    },
+    {
+      "epoch": 0.3696590835857447,
+      "grad_norm": 0.004045239143780003,
+      "learning_rate": 7.006802721088436e-06,
+      "loss": 0.0,
+      "step": 5736
+    },
+    {
+      "epoch": 0.36972352903267386,
+      "grad_norm": 0.014195178455213005,
+      "learning_rate": 7.00608664518439e-06,
+      "loss": 0.0,
+      "step": 5737
+    },
+    {
+      "epoch": 0.36978797447960304,
+      "grad_norm": 0.024569513616097965,
+      "learning_rate": 7.005370569280344e-06,
+      "loss": 0.0,
+      "step": 5738
+    },
+    {
+      "epoch": 0.36985241992653217,
+      "grad_norm": 0.020241459484012674,
+      "learning_rate": 7.004654493376299e-06,
+      "loss": 0.0,
+      "step": 5739
+    },
+    {
+      "epoch": 0.36991686537346136,
+      "grad_norm": 0.004866278303404176,
+      "learning_rate": 7.003938417472252e-06,
+      "loss": 0.0,
+      "step": 5740
+    },
+    {
+      "epoch": 0.36998131082039054,
+      "grad_norm": 0.010689225892144915,
+      "learning_rate": 7.003222341568206e-06,
+      "loss": 0.0001,
+      "step": 5741
+    },
+    {
+      "epoch": 0.3700457562673197,
+      "grad_norm": 0.03240860651724199,
+      "learning_rate": 7.002506265664161e-06,
+      "loss": 0.0,
+      "step": 5742
+    },
+    {
+      "epoch": 0.3701102017142489,
+      "grad_norm": 0.6012695391805059,
+      "learning_rate": 7.001790189760115e-06,
+      "loss": 0.0037,
+      "step": 5743
+    },
+    {
+      "epoch": 0.37017464716117804,
+      "grad_norm": 0.04119383182365942,
+      "learning_rate": 7.001074113856069e-06,
+      "loss": 0.0001,
+      "step": 5744
+    },
+    {
+      "epoch": 0.3702390926081072,
+      "grad_norm": 0.008040074272905727,
+      "learning_rate": 7.000358037952023e-06,
+      "loss": 0.0,
+      "step": 5745
+    },
+    {
+      "epoch": 0.3703035380550364,
+      "grad_norm": 0.14133781336635506,
+      "learning_rate": 6.999641962047977e-06,
+      "loss": 0.0022,
+      "step": 5746
+    },
+    {
+      "epoch": 0.3703679835019656,
+      "grad_norm": 0.024683611120630507,
+      "learning_rate": 6.998925886143931e-06,
+      "loss": 0.0,
+      "step": 5747
+    },
+    {
+      "epoch": 0.3704324289488948,
+      "grad_norm": 0.08903047211636257,
+      "learning_rate": 6.9982098102398865e-06,
+      "loss": 0.0003,
+      "step": 5748
+    },
+    {
+      "epoch": 0.37049687439582396,
+      "grad_norm": 0.45973190145044607,
+      "learning_rate": 6.997493734335841e-06,
+      "loss": 0.0009,
+      "step": 5749
+    },
+    {
+      "epoch": 0.3705613198427531,
+      "grad_norm": 0.06435428612509528,
+      "learning_rate": 6.996777658431795e-06,
+      "loss": 0.0001,
+      "step": 5750
+    },
+    {
+      "epoch": 0.3706257652896823,
+      "grad_norm": 0.018558351255393086,
+      "learning_rate": 6.9960615825277485e-06,
+      "loss": 0.0,
+      "step": 5751
+    },
+    {
+      "epoch": 0.37069021073661146,
+      "grad_norm": 0.008864585347216576,
+      "learning_rate": 6.995345506623703e-06,
+      "loss": 0.0,
+      "step": 5752
+    },
+    {
+      "epoch": 0.37075465618354064,
+      "grad_norm": 0.009975094178598903,
+      "learning_rate": 6.994629430719657e-06,
+      "loss": 0.0,
+      "step": 5753
+    },
+    {
+      "epoch": 0.3708191016304698,
+      "grad_norm": 0.016556182540155888,
+      "learning_rate": 6.9939133548156114e-06,
+      "loss": 0.0001,
+      "step": 5754
+    },
+    {
+      "epoch": 0.37088354707739896,
+      "grad_norm": 0.13527741899174603,
+      "learning_rate": 6.993197278911566e-06,
+      "loss": 0.0019,
+      "step": 5755
+    },
+    {
+      "epoch": 0.37094799252432814,
+      "grad_norm": 0.06295406905782694,
+      "learning_rate": 6.992481203007519e-06,
+      "loss": 0.0001,
+      "step": 5756
+    },
+    {
+      "epoch": 0.3710124379712573,
+      "grad_norm": 0.0010064822738222764,
+      "learning_rate": 6.9917651271034735e-06,
+      "loss": 0.0,
+      "step": 5757
+    },
+    {
+      "epoch": 0.3710768834181865,
+      "grad_norm": 0.024290977389738094,
+      "learning_rate": 6.991049051199428e-06,
+      "loss": 0.0015,
+      "step": 5758
+    },
+    {
+      "epoch": 0.3711413288651157,
+      "grad_norm": 0.010974935412825312,
+      "learning_rate": 6.990332975295382e-06,
+      "loss": 0.0,
+      "step": 5759
+    },
+    {
+      "epoch": 0.3712057743120449,
+      "grad_norm": 0.10455954243655033,
+      "learning_rate": 6.989616899391336e-06,
+      "loss": 0.0003,
+      "step": 5760
+    },
+    {
+      "epoch": 0.371270219758974,
+      "grad_norm": 0.12336619134229843,
+      "learning_rate": 6.98890082348729e-06,
+      "loss": 0.0002,
+      "step": 5761
+    },
+    {
+      "epoch": 0.3713346652059032,
+      "grad_norm": 0.21171697777713713,
+      "learning_rate": 6.988184747583244e-06,
+      "loss": 0.0009,
+      "step": 5762
+    },
+    {
+      "epoch": 0.3713991106528324,
+      "grad_norm": 0.19279942206943126,
+      "learning_rate": 6.9874686716791984e-06,
+      "loss": 0.0026,
+      "step": 5763
+    },
+    {
+      "epoch": 0.37146355609976156,
+      "grad_norm": 0.01216910669966571,
+      "learning_rate": 6.986752595775153e-06,
+      "loss": 0.0001,
+      "step": 5764
+    },
+    {
+      "epoch": 0.37152800154669074,
+      "grad_norm": 0.00037731451971985587,
+      "learning_rate": 6.986036519871107e-06,
+      "loss": 0.0,
+      "step": 5765
+    },
+    {
+      "epoch": 0.3715924469936199,
+      "grad_norm": 0.00018807497398390133,
+      "learning_rate": 6.9853204439670605e-06,
+      "loss": 0.0,
+      "step": 5766
+    },
+    {
+      "epoch": 0.37165689244054906,
+      "grad_norm": 0.0026170447983087115,
+      "learning_rate": 6.984604368063015e-06,
+      "loss": 0.0,
+      "step": 5767
+    },
+    {
+      "epoch": 0.37172133788747824,
+      "grad_norm": 0.017747659184786897,
+      "learning_rate": 6.983888292158969e-06,
+      "loss": 0.0002,
+      "step": 5768
+    },
+    {
+      "epoch": 0.3717857833344074,
+      "grad_norm": 0.3520613627754859,
+      "learning_rate": 6.983172216254923e-06,
+      "loss": 0.0017,
+      "step": 5769
+    },
+    {
+      "epoch": 0.3718502287813366,
+      "grad_norm": 0.00016192167565096303,
+      "learning_rate": 6.982456140350878e-06,
+      "loss": 0.0,
+      "step": 5770
+    },
+    {
+      "epoch": 0.3719146742282658,
+      "grad_norm": 0.0008348356854251009,
+      "learning_rate": 6.981740064446831e-06,
+      "loss": 0.0,
+      "step": 5771
+    },
+    {
+      "epoch": 0.3719791196751949,
+      "grad_norm": 0.005497509674599073,
+      "learning_rate": 6.981023988542786e-06,
+      "loss": 0.0001,
+      "step": 5772
+    },
+    {
+      "epoch": 0.3720435651221241,
+      "grad_norm": 0.0037413784761350334,
+      "learning_rate": 6.9803079126387406e-06,
+      "loss": 0.0,
+      "step": 5773
+    },
+    {
+      "epoch": 0.3721080105690533,
+      "grad_norm": 0.0026230430236184187,
+      "learning_rate": 6.979591836734695e-06,
+      "loss": 0.0,
+      "step": 5774
+    },
+    {
+      "epoch": 0.3721724560159825,
+      "grad_norm": 0.41783501811893103,
+      "learning_rate": 6.978875760830649e-06,
+      "loss": 0.0034,
+      "step": 5775
+    },
+    {
+      "epoch": 0.37223690146291166,
+      "grad_norm": 0.024433689572735673,
+      "learning_rate": 6.9781596849266035e-06,
+      "loss": 0.0,
+      "step": 5776
+    },
+    {
+      "epoch": 0.37230134690984085,
+      "grad_norm": 0.20971572218454174,
+      "learning_rate": 6.977443609022557e-06,
+      "loss": 0.0025,
+      "step": 5777
+    },
+    {
+      "epoch": 0.37236579235677,
+      "grad_norm": 0.006778586627347576,
+      "learning_rate": 6.976727533118511e-06,
+      "loss": 0.0,
+      "step": 5778
+    },
+    {
+      "epoch": 0.37243023780369916,
+      "grad_norm": 0.004765972105939649,
+      "learning_rate": 6.9760114572144655e-06,
+      "loss": 0.0,
+      "step": 5779
+    },
+    {
+      "epoch": 0.37249468325062834,
+      "grad_norm": 0.00936975226703011,
+      "learning_rate": 6.97529538131042e-06,
+      "loss": 0.0,
+      "step": 5780
+    },
+    {
+      "epoch": 0.37255912869755753,
+      "grad_norm": 0.03314593066305474,
+      "learning_rate": 6.974579305406374e-06,
+      "loss": 0.0002,
+      "step": 5781
+    },
+    {
+      "epoch": 0.3726235741444867,
+      "grad_norm": 0.10569911541069012,
+      "learning_rate": 6.9738632295023276e-06,
+      "loss": 0.0001,
+      "step": 5782
+    },
+    {
+      "epoch": 0.37268801959141584,
+      "grad_norm": 0.005564481327632969,
+      "learning_rate": 6.973147153598282e-06,
+      "loss": 0.0,
+      "step": 5783
+    },
+    {
+      "epoch": 0.372752465038345,
+      "grad_norm": 0.0011321496959383888,
+      "learning_rate": 6.972431077694236e-06,
+      "loss": 0.0,
+      "step": 5784
+    },
+    {
+      "epoch": 0.3728169104852742,
+      "grad_norm": 0.004791011659612873,
+      "learning_rate": 6.9717150017901905e-06,
+      "loss": 0.0,
+      "step": 5785
+    },
+    {
+      "epoch": 0.3728813559322034,
+      "grad_norm": 0.005532310001008372,
+      "learning_rate": 6.970998925886145e-06,
+      "loss": 0.0001,
+      "step": 5786
+    },
+    {
+      "epoch": 0.3729458013791326,
+      "grad_norm": 0.7089383028892337,
+      "learning_rate": 6.970282849982098e-06,
+      "loss": 0.0035,
+      "step": 5787
+    },
+    {
+      "epoch": 0.37301024682606176,
+      "grad_norm": 0.10018688504803142,
+      "learning_rate": 6.9695667740780525e-06,
+      "loss": 0.0001,
+      "step": 5788
+    },
+    {
+      "epoch": 0.3730746922729909,
+      "grad_norm": 0.0001188095476045239,
+      "learning_rate": 6.968850698174007e-06,
+      "loss": 0.0,
+      "step": 5789
+    },
+    {
+      "epoch": 0.3731391377199201,
+      "grad_norm": 0.31996598923516045,
+      "learning_rate": 6.968134622269961e-06,
+      "loss": 0.0045,
+      "step": 5790
+    },
+    {
+      "epoch": 0.37320358316684926,
+      "grad_norm": 0.00032090989078132145,
+      "learning_rate": 6.9674185463659146e-06,
+      "loss": 0.0,
+      "step": 5791
+    },
+    {
+      "epoch": 0.37326802861377845,
+      "grad_norm": 0.0027813093535380045,
+      "learning_rate": 6.966702470461869e-06,
+      "loss": 0.0,
+      "step": 5792
+    },
+    {
+      "epoch": 0.37333247406070763,
+      "grad_norm": 0.0006011110428928335,
+      "learning_rate": 6.965986394557823e-06,
+      "loss": 0.0,
+      "step": 5793
+    },
+    {
+      "epoch": 0.37339691950763676,
+      "grad_norm": 0.006022162366228975,
+      "learning_rate": 6.9652703186537775e-06,
+      "loss": 0.0,
+      "step": 5794
+    },
+    {
+      "epoch": 0.37346136495456594,
+      "grad_norm": 0.0009564569230516909,
+      "learning_rate": 6.964554242749733e-06,
+      "loss": 0.0,
+      "step": 5795
+    },
+    {
+      "epoch": 0.37352581040149513,
+      "grad_norm": 0.03388242371408433,
+      "learning_rate": 6.963838166845687e-06,
+      "loss": 0.0,
+      "step": 5796
+    },
+    {
+      "epoch": 0.3735902558484243,
+      "grad_norm": 0.00852415166096573,
+      "learning_rate": 6.963122090941641e-06,
+      "loss": 0.0,
+      "step": 5797
+    },
+    {
+      "epoch": 0.3736547012953535,
+      "grad_norm": 0.014353758440833062,
+      "learning_rate": 6.962406015037595e-06,
+      "loss": 0.0001,
+      "step": 5798
+    },
+    {
+      "epoch": 0.3737191467422827,
+      "grad_norm": 0.001276369092390287,
+      "learning_rate": 6.961689939133549e-06,
+      "loss": 0.0,
+      "step": 5799
+    },
+    {
+      "epoch": 0.3737835921892118,
+      "grad_norm": 0.0008902047543594542,
+      "learning_rate": 6.960973863229503e-06,
+      "loss": 0.0,
+      "step": 5800
+    },
+    {
+      "epoch": 0.373848037636141,
+      "grad_norm": 0.004717272669452164,
+      "learning_rate": 6.9602577873254575e-06,
+      "loss": 0.0,
+      "step": 5801
+    },
+    {
+      "epoch": 0.3739124830830702,
+      "grad_norm": 0.014702761769368668,
+      "learning_rate": 6.959541711421412e-06,
+      "loss": 0.0001,
+      "step": 5802
+    },
+    {
+      "epoch": 0.37397692852999936,
+      "grad_norm": 0.03831489377639327,
+      "learning_rate": 6.958825635517365e-06,
+      "loss": 0.0001,
+      "step": 5803
+    },
+    {
+      "epoch": 0.37404137397692855,
+      "grad_norm": 0.5141810225466743,
+      "learning_rate": 6.95810955961332e-06,
+      "loss": 0.0018,
+      "step": 5804
+    },
+    {
+      "epoch": 0.3741058194238577,
+      "grad_norm": 0.0005838929170589029,
+      "learning_rate": 6.957393483709274e-06,
+      "loss": 0.0,
+      "step": 5805
+    },
+    {
+      "epoch": 0.37417026487078686,
+      "grad_norm": 0.0002700920403723447,
+      "learning_rate": 6.956677407805228e-06,
+      "loss": 0.0,
+      "step": 5806
+    },
+    {
+      "epoch": 0.37423471031771605,
+      "grad_norm": 0.06860687546291591,
+      "learning_rate": 6.955961331901182e-06,
+      "loss": 0.0017,
+      "step": 5807
+    },
+    {
+      "epoch": 0.37429915576464523,
+      "grad_norm": 0.004176513204807584,
+      "learning_rate": 6.955245255997136e-06,
+      "loss": 0.0,
+      "step": 5808
+    },
+    {
+      "epoch": 0.3743636012115744,
+      "grad_norm": 0.005652395412120658,
+      "learning_rate": 6.95452918009309e-06,
+      "loss": 0.0,
+      "step": 5809
+    },
+    {
+      "epoch": 0.3744280466585036,
+      "grad_norm": 0.019276447759645723,
+      "learning_rate": 6.9538131041890445e-06,
+      "loss": 0.0,
+      "step": 5810
+    },
+    {
+      "epoch": 0.37449249210543273,
+      "grad_norm": 0.004779306271954092,
+      "learning_rate": 6.953097028284999e-06,
+      "loss": 0.0,
+      "step": 5811
+    },
+    {
+      "epoch": 0.3745569375523619,
+      "grad_norm": 0.05639422399771823,
+      "learning_rate": 6.952380952380952e-06,
+      "loss": 0.0003,
+      "step": 5812
+    },
+    {
+      "epoch": 0.3746213829992911,
+      "grad_norm": 0.013047953267711436,
+      "learning_rate": 6.951664876476907e-06,
+      "loss": 0.0,
+      "step": 5813
+    },
+    {
+      "epoch": 0.3746858284462203,
+      "grad_norm": 0.0800606691129166,
+      "learning_rate": 6.950948800572861e-06,
+      "loss": 0.0,
+      "step": 5814
+    },
+    {
+      "epoch": 0.37475027389314947,
+      "grad_norm": 0.12582311229794294,
+      "learning_rate": 6.950232724668815e-06,
+      "loss": 0.0017,
+      "step": 5815
+    },
+    {
+      "epoch": 0.37481471934007865,
+      "grad_norm": 0.019311332882545055,
+      "learning_rate": 6.9495166487647695e-06,
+      "loss": 0.0001,
+      "step": 5816
+    },
+    {
+      "epoch": 0.3748791647870078,
+      "grad_norm": 0.2597594869935363,
+      "learning_rate": 6.948800572860723e-06,
+      "loss": 0.0006,
+      "step": 5817
+    },
+    {
+      "epoch": 0.37494361023393696,
+      "grad_norm": 0.013510250554619812,
+      "learning_rate": 6.948084496956679e-06,
+      "loss": 0.0,
+      "step": 5818
+    },
+    {
+      "epoch": 0.37500805568086615,
+      "grad_norm": 0.004594365982372799,
+      "learning_rate": 6.947368421052632e-06,
+      "loss": 0.0,
+      "step": 5819
+    },
+    {
+      "epoch": 0.37507250112779533,
+      "grad_norm": 0.001138969696631606,
+      "learning_rate": 6.946652345148587e-06,
+      "loss": 0.0,
+      "step": 5820
+    },
+    {
+      "epoch": 0.3751369465747245,
+      "grad_norm": 0.0006642511054819653,
+      "learning_rate": 6.945936269244541e-06,
+      "loss": 0.0,
+      "step": 5821
+    },
+    {
+      "epoch": 0.37520139202165365,
+      "grad_norm": 0.044176775604888044,
+      "learning_rate": 6.945220193340495e-06,
+      "loss": 0.0001,
+      "step": 5822
+    },
+    {
+      "epoch": 0.37526583746858283,
+      "grad_norm": 0.0004816241632529937,
+      "learning_rate": 6.944504117436449e-06,
+      "loss": 0.0,
+      "step": 5823
+    },
+    {
+      "epoch": 0.375330282915512,
+      "grad_norm": 0.0020537732340232817,
+      "learning_rate": 6.943788041532403e-06,
+      "loss": 0.0,
+      "step": 5824
+    },
+    {
+      "epoch": 0.3753947283624412,
+      "grad_norm": 0.3551568350092731,
+      "learning_rate": 6.943071965628357e-06,
+      "loss": 0.0015,
+      "step": 5825
+    },
+    {
+      "epoch": 0.3754591738093704,
+      "grad_norm": 0.0009804151741977017,
+      "learning_rate": 6.942355889724312e-06,
+      "loss": 0.0,
+      "step": 5826
+    },
+    {
+      "epoch": 0.37552361925629957,
+      "grad_norm": 1.3854608552621122,
+      "learning_rate": 6.941639813820266e-06,
+      "loss": 0.0075,
+      "step": 5827
+    },
+    {
+      "epoch": 0.3755880647032287,
+      "grad_norm": 0.000444632576506098,
+      "learning_rate": 6.940923737916219e-06,
+      "loss": 0.0,
+      "step": 5828
+    },
+    {
+      "epoch": 0.3756525101501579,
+      "grad_norm": 0.24999937415044649,
+      "learning_rate": 6.940207662012174e-06,
+      "loss": 0.001,
+      "step": 5829
+    },
+    {
+      "epoch": 0.37571695559708707,
+      "grad_norm": 0.001790503591942046,
+      "learning_rate": 6.939491586108128e-06,
+      "loss": 0.0,
+      "step": 5830
+    },
+    {
+      "epoch": 0.37578140104401625,
+      "grad_norm": 0.00021567154767956597,
+      "learning_rate": 6.938775510204082e-06,
+      "loss": 0.0,
+      "step": 5831
+    },
+    {
+      "epoch": 0.37584584649094543,
+      "grad_norm": 0.001552857672059101,
+      "learning_rate": 6.9380594343000366e-06,
+      "loss": 0.0,
+      "step": 5832
+    },
+    {
+      "epoch": 0.37591029193787456,
+      "grad_norm": 0.0009687577369576802,
+      "learning_rate": 6.93734335839599e-06,
+      "loss": 0.0,
+      "step": 5833
+    },
+    {
+      "epoch": 0.37597473738480375,
+      "grad_norm": 0.22006186439099512,
+      "learning_rate": 6.936627282491944e-06,
+      "loss": 0.0018,
+      "step": 5834
+    },
+    {
+      "epoch": 0.37603918283173293,
+      "grad_norm": 0.0022716589278178995,
+      "learning_rate": 6.935911206587899e-06,
+      "loss": 0.0,
+      "step": 5835
+    },
+    {
+      "epoch": 0.3761036282786621,
+      "grad_norm": 0.07283411149368349,
+      "learning_rate": 6.935195130683853e-06,
+      "loss": 0.0003,
+      "step": 5836
+    },
+    {
+      "epoch": 0.3761680737255913,
+      "grad_norm": 0.0513859471633266,
+      "learning_rate": 6.934479054779807e-06,
+      "loss": 0.0002,
+      "step": 5837
+    },
+    {
+      "epoch": 0.3762325191725205,
+      "grad_norm": 0.216646623524965,
+      "learning_rate": 6.933762978875761e-06,
+      "loss": 0.0014,
+      "step": 5838
+    },
+    {
+      "epoch": 0.3762969646194496,
+      "grad_norm": 0.003388630037906614,
+      "learning_rate": 6.933046902971715e-06,
+      "loss": 0.0,
+      "step": 5839
+    },
+    {
+      "epoch": 0.3763614100663788,
+      "grad_norm": 0.028671905161233677,
+      "learning_rate": 6.932330827067669e-06,
+      "loss": 0.0,
+      "step": 5840
+    },
+    {
+      "epoch": 0.376425855513308,
+      "grad_norm": 0.024152302581481176,
+      "learning_rate": 6.9316147511636236e-06,
+      "loss": 0.0,
+      "step": 5841
+    },
+    {
+      "epoch": 0.37649030096023717,
+      "grad_norm": 0.0009817171541373576,
+      "learning_rate": 6.930898675259579e-06,
+      "loss": 0.0,
+      "step": 5842
+    },
+    {
+      "epoch": 0.37655474640716635,
+      "grad_norm": 0.0013014101235195464,
+      "learning_rate": 6.930182599355533e-06,
+      "loss": 0.0,
+      "step": 5843
+    },
+    {
+      "epoch": 0.3766191918540955,
+      "grad_norm": 0.0028327639840403755,
+      "learning_rate": 6.9294665234514865e-06,
+      "loss": 0.0,
+      "step": 5844
+    },
+    {
+      "epoch": 0.37668363730102467,
+      "grad_norm": 0.0032929799671172245,
+      "learning_rate": 6.928750447547441e-06,
+      "loss": 0.0,
+      "step": 5845
+    },
+    {
+      "epoch": 0.37674808274795385,
+      "grad_norm": 0.030509106413843804,
+      "learning_rate": 6.928034371643395e-06,
+      "loss": 0.0002,
+      "step": 5846
+    },
+    {
+      "epoch": 0.37681252819488303,
+      "grad_norm": 0.004954742198539419,
+      "learning_rate": 6.927318295739349e-06,
+      "loss": 0.0,
+      "step": 5847
+    },
+    {
+      "epoch": 0.3768769736418122,
+      "grad_norm": 0.0002230131182336926,
+      "learning_rate": 6.926602219835304e-06,
+      "loss": 0.0,
+      "step": 5848
+    },
+    {
+      "epoch": 0.3769414190887414,
+      "grad_norm": 0.023321539143309727,
+      "learning_rate": 6.925886143931257e-06,
+      "loss": 0.0001,
+      "step": 5849
+    },
+    {
+      "epoch": 0.37700586453567053,
+      "grad_norm": 0.2192969636607764,
+      "learning_rate": 6.925170068027211e-06,
+      "loss": 0.0005,
+      "step": 5850
+    },
+    {
+      "epoch": 0.3770703099825997,
+      "grad_norm": 0.0029399686461097315,
+      "learning_rate": 6.924453992123166e-06,
+      "loss": 0.0,
+      "step": 5851
+    },
+    {
+      "epoch": 0.3771347554295289,
+      "grad_norm": 0.006376825308723674,
+      "learning_rate": 6.92373791621912e-06,
+      "loss": 0.0,
+      "step": 5852
+    },
+    {
+      "epoch": 0.3771992008764581,
+      "grad_norm": 0.15974989372908052,
+      "learning_rate": 6.923021840315074e-06,
+      "loss": 0.0006,
+      "step": 5853
+    },
+    {
+      "epoch": 0.37726364632338727,
+      "grad_norm": 0.003535994812387964,
+      "learning_rate": 6.922305764411028e-06,
+      "loss": 0.0,
+      "step": 5854
+    },
+    {
+      "epoch": 0.37732809177031645,
+      "grad_norm": 0.001929273478040189,
+      "learning_rate": 6.921589688506982e-06,
+      "loss": 0.0,
+      "step": 5855
+    },
+    {
+      "epoch": 0.3773925372172456,
+      "grad_norm": 0.001803497544205884,
+      "learning_rate": 6.920873612602936e-06,
+      "loss": 0.0,
+      "step": 5856
+    },
+    {
+      "epoch": 0.37745698266417477,
+      "grad_norm": 1.0116954786061811,
+      "learning_rate": 6.920157536698891e-06,
+      "loss": 0.0011,
+      "step": 5857
+    },
+    {
+      "epoch": 0.37752142811110395,
+      "grad_norm": 0.0044865772161063605,
+      "learning_rate": 6.919441460794845e-06,
+      "loss": 0.0,
+      "step": 5858
+    },
+    {
+      "epoch": 0.37758587355803314,
+      "grad_norm": 0.002553314405904031,
+      "learning_rate": 6.918725384890798e-06,
+      "loss": 0.0,
+      "step": 5859
+    },
+    {
+      "epoch": 0.3776503190049623,
+      "grad_norm": 0.0049119126899543186,
+      "learning_rate": 6.918009308986753e-06,
+      "loss": 0.0,
+      "step": 5860
+    },
+    {
+      "epoch": 0.37771476445189145,
+      "grad_norm": 0.0008667354202076583,
+      "learning_rate": 6.917293233082707e-06,
+      "loss": 0.0,
+      "step": 5861
+    },
+    {
+      "epoch": 0.37777920989882063,
+      "grad_norm": 0.0016562458722985242,
+      "learning_rate": 6.916577157178661e-06,
+      "loss": 0.0,
+      "step": 5862
+    },
+    {
+      "epoch": 0.3778436553457498,
+      "grad_norm": 0.0004253058844486389,
+      "learning_rate": 6.915861081274616e-06,
+      "loss": 0.0,
+      "step": 5863
+    },
+    {
+      "epoch": 0.377908100792679,
+      "grad_norm": 0.010075668762335689,
+      "learning_rate": 6.915145005370569e-06,
+      "loss": 0.0001,
+      "step": 5864
+    },
+    {
+      "epoch": 0.3779725462396082,
+      "grad_norm": 0.004239878134591785,
+      "learning_rate": 6.914428929466524e-06,
+      "loss": 0.0,
+      "step": 5865
+    },
+    {
+      "epoch": 0.37803699168653737,
+      "grad_norm": 0.30177530528108326,
+      "learning_rate": 6.9137128535624785e-06,
+      "loss": 0.0011,
+      "step": 5866
+    },
+    {
+      "epoch": 0.3781014371334665,
+      "grad_norm": 0.00040066692664154734,
+      "learning_rate": 6.912996777658433e-06,
+      "loss": 0.0,
+      "step": 5867
+    },
+    {
+      "epoch": 0.3781658825803957,
+      "grad_norm": 0.10780252724976731,
+      "learning_rate": 6.912280701754387e-06,
+      "loss": 0.0004,
+      "step": 5868
+    },
+    {
+      "epoch": 0.37823032802732487,
+      "grad_norm": 0.002191261152124998,
+      "learning_rate": 6.911564625850341e-06,
+      "loss": 0.0,
+      "step": 5869
+    },
+    {
+      "epoch": 0.37829477347425405,
+      "grad_norm": 0.2787256869160595,
+      "learning_rate": 6.910848549946295e-06,
+      "loss": 0.0014,
+      "step": 5870
+    },
+    {
+      "epoch": 0.37835921892118324,
+      "grad_norm": 0.0006742879125156188,
+      "learning_rate": 6.910132474042249e-06,
+      "loss": 0.0,
+      "step": 5871
+    },
+    {
+      "epoch": 0.37842366436811237,
+      "grad_norm": 0.012795925913342132,
+      "learning_rate": 6.9094163981382034e-06,
+      "loss": 0.0001,
+      "step": 5872
+    },
+    {
+      "epoch": 0.37848810981504155,
+      "grad_norm": 0.0007479872914136376,
+      "learning_rate": 6.908700322234158e-06,
+      "loss": 0.0,
+      "step": 5873
+    },
+    {
+      "epoch": 0.37855255526197074,
+      "grad_norm": 0.00018037472474745242,
+      "learning_rate": 6.907984246330112e-06,
+      "loss": 0.0,
+      "step": 5874
+    },
+    {
+      "epoch": 0.3786170007088999,
+      "grad_norm": 0.0015442612656284755,
+      "learning_rate": 6.9072681704260655e-06,
+      "loss": 0.0,
+      "step": 5875
+    },
+    {
+      "epoch": 0.3786814461558291,
+      "grad_norm": 0.0024565796104232176,
+      "learning_rate": 6.90655209452202e-06,
+      "loss": 0.0,
+      "step": 5876
+    },
+    {
+      "epoch": 0.3787458916027583,
+      "grad_norm": 0.500224241993818,
+      "learning_rate": 6.905836018617974e-06,
+      "loss": 0.0036,
+      "step": 5877
+    },
+    {
+      "epoch": 0.3788103370496874,
+      "grad_norm": 0.00017855047645974685,
+      "learning_rate": 6.905119942713928e-06,
+      "loss": 0.0,
+      "step": 5878
+    },
+    {
+      "epoch": 0.3788747824966166,
+      "grad_norm": 0.012190953634215032,
+      "learning_rate": 6.904403866809883e-06,
+      "loss": 0.0001,
+      "step": 5879
+    },
+    {
+      "epoch": 0.3789392279435458,
+      "grad_norm": 0.004909180990656321,
+      "learning_rate": 6.903687790905836e-06,
+      "loss": 0.0,
+      "step": 5880
+    },
+    {
+      "epoch": 0.37900367339047497,
+      "grad_norm": 0.5589098802746703,
+      "learning_rate": 6.9029717150017904e-06,
+      "loss": 0.0005,
+      "step": 5881
+    },
+    {
+      "epoch": 0.37906811883740416,
+      "grad_norm": 0.003212166081683288,
+      "learning_rate": 6.902255639097745e-06,
+      "loss": 0.0,
+      "step": 5882
+    },
+    {
+      "epoch": 0.3791325642843333,
+      "grad_norm": 0.08818240476463624,
+      "learning_rate": 6.901539563193699e-06,
+      "loss": 0.0016,
+      "step": 5883
+    },
+    {
+      "epoch": 0.37919700973126247,
+      "grad_norm": 0.003632769423772264,
+      "learning_rate": 6.9008234872896525e-06,
+      "loss": 0.0,
+      "step": 5884
+    },
+    {
+      "epoch": 0.37926145517819165,
+      "grad_norm": 0.009019636505328735,
+      "learning_rate": 6.900107411385607e-06,
+      "loss": 0.0001,
+      "step": 5885
+    },
+    {
+      "epoch": 0.37932590062512084,
+      "grad_norm": 0.0004343660014682446,
+      "learning_rate": 6.899391335481561e-06,
+      "loss": 0.0,
+      "step": 5886
+    },
+    {
+      "epoch": 0.37939034607205,
+      "grad_norm": 0.00027411815184107786,
+      "learning_rate": 6.898675259577515e-06,
+      "loss": 0.0,
+      "step": 5887
+    },
+    {
+      "epoch": 0.3794547915189792,
+      "grad_norm": 0.018313065196746546,
+      "learning_rate": 6.8979591836734705e-06,
+      "loss": 0.0,
+      "step": 5888
+    },
+    {
+      "epoch": 0.37951923696590834,
+      "grad_norm": 0.05844586306158265,
+      "learning_rate": 6.897243107769425e-06,
+      "loss": 0.0001,
+      "step": 5889
+    },
+    {
+      "epoch": 0.3795836824128375,
+      "grad_norm": 0.008817164212300388,
+      "learning_rate": 6.896527031865379e-06,
+      "loss": 0.0,
+      "step": 5890
+    },
+    {
+      "epoch": 0.3796481278597667,
+      "grad_norm": 0.440502219983676,
+      "learning_rate": 6.8958109559613326e-06,
+      "loss": 0.0005,
+      "step": 5891
+    },
+    {
+      "epoch": 0.3797125733066959,
+      "grad_norm": 0.03877985442103465,
+      "learning_rate": 6.895094880057287e-06,
+      "loss": 0.0001,
+      "step": 5892
+    },
+    {
+      "epoch": 0.3797770187536251,
+      "grad_norm": 0.0006761241313574944,
+      "learning_rate": 6.894378804153241e-06,
+      "loss": 0.0,
+      "step": 5893
+    },
+    {
+      "epoch": 0.37984146420055426,
+      "grad_norm": 0.0003246361163867012,
+      "learning_rate": 6.8936627282491955e-06,
+      "loss": 0.0,
+      "step": 5894
+    },
+    {
+      "epoch": 0.3799059096474834,
+      "grad_norm": 0.001269483495536825,
+      "learning_rate": 6.89294665234515e-06,
+      "loss": 0.0,
+      "step": 5895
+    },
+    {
+      "epoch": 0.37997035509441257,
+      "grad_norm": 0.6018471230283085,
+      "learning_rate": 6.892230576441103e-06,
+      "loss": 0.003,
+      "step": 5896
+    },
+    {
+      "epoch": 0.38003480054134176,
+      "grad_norm": 0.00046990029288776074,
+      "learning_rate": 6.8915145005370575e-06,
+      "loss": 0.0,
+      "step": 5897
+    },
+    {
+      "epoch": 0.38009924598827094,
+      "grad_norm": 0.2449174500328802,
+      "learning_rate": 6.890798424633012e-06,
+      "loss": 0.0001,
+      "step": 5898
+    },
+    {
+      "epoch": 0.3801636914352001,
+      "grad_norm": 0.00953486729292568,
+      "learning_rate": 6.890082348728966e-06,
+      "loss": 0.0,
+      "step": 5899
+    },
+    {
+      "epoch": 0.38022813688212925,
+      "grad_norm": 0.0004494877435115907,
+      "learning_rate": 6.88936627282492e-06,
+      "loss": 0.0,
+      "step": 5900
+    },
+    {
+      "epoch": 0.38029258232905844,
+      "grad_norm": 0.007622277277941377,
+      "learning_rate": 6.888650196920874e-06,
+      "loss": 0.0,
+      "step": 5901
+    },
+    {
+      "epoch": 0.3803570277759876,
+      "grad_norm": 0.00021103624434805576,
+      "learning_rate": 6.887934121016828e-06,
+      "loss": 0.0,
+      "step": 5902
+    },
+    {
+      "epoch": 0.3804214732229168,
+      "grad_norm": 0.00270139356453862,
+      "learning_rate": 6.8872180451127825e-06,
+      "loss": 0.0,
+      "step": 5903
+    },
+    {
+      "epoch": 0.380485918669846,
+      "grad_norm": 0.01119123283242716,
+      "learning_rate": 6.886501969208737e-06,
+      "loss": 0.0001,
+      "step": 5904
+    },
+    {
+      "epoch": 0.3805503641167752,
+      "grad_norm": 0.0005282343466651458,
+      "learning_rate": 6.88578589330469e-06,
+      "loss": 0.0,
+      "step": 5905
+    },
+    {
+      "epoch": 0.3806148095637043,
+      "grad_norm": 2.001432859226168,
+      "learning_rate": 6.8850698174006445e-06,
+      "loss": 0.0158,
+      "step": 5906
+    },
+    {
+      "epoch": 0.3806792550106335,
+      "grad_norm": 0.001497510037133082,
+      "learning_rate": 6.884353741496599e-06,
+      "loss": 0.0,
+      "step": 5907
+    },
+    {
+      "epoch": 0.3807437004575627,
+      "grad_norm": 0.0058557848933884354,
+      "learning_rate": 6.883637665592553e-06,
+      "loss": 0.0,
+      "step": 5908
+    },
+    {
+      "epoch": 0.38080814590449186,
+      "grad_norm": 0.15156723462407826,
+      "learning_rate": 6.882921589688507e-06,
+      "loss": 0.0003,
+      "step": 5909
+    },
+    {
+      "epoch": 0.38087259135142104,
+      "grad_norm": 0.0022827647541258317,
+      "learning_rate": 6.882205513784461e-06,
+      "loss": 0.0,
+      "step": 5910
+    },
+    {
+      "epoch": 0.38093703679835017,
+      "grad_norm": 0.001730530080398193,
+      "learning_rate": 6.881489437880415e-06,
+      "loss": 0.0,
+      "step": 5911
+    },
+    {
+      "epoch": 0.38100148224527935,
+      "grad_norm": 0.0014464979963454568,
+      "learning_rate": 6.88077336197637e-06,
+      "loss": 0.0,
+      "step": 5912
+    },
+    {
+      "epoch": 0.38106592769220854,
+      "grad_norm": 0.0013056269425528884,
+      "learning_rate": 6.880057286072325e-06,
+      "loss": 0.0,
+      "step": 5913
+    },
+    {
+      "epoch": 0.3811303731391377,
+      "grad_norm": 0.000246217387139311,
+      "learning_rate": 6.879341210168279e-06,
+      "loss": 0.0,
+      "step": 5914
+    },
+    {
+      "epoch": 0.3811948185860669,
+      "grad_norm": 0.0029653896752515046,
+      "learning_rate": 6.878625134264233e-06,
+      "loss": 0.0,
+      "step": 5915
+    },
+    {
+      "epoch": 0.3812592640329961,
+      "grad_norm": 0.004897105078602228,
+      "learning_rate": 6.8779090583601875e-06,
+      "loss": 0.0001,
+      "step": 5916
+    },
+    {
+      "epoch": 0.3813237094799252,
+      "grad_norm": 0.0006053366262376849,
+      "learning_rate": 6.877192982456141e-06,
+      "loss": 0.0,
+      "step": 5917
+    },
+    {
+      "epoch": 0.3813881549268544,
+      "grad_norm": 0.0014126298196204864,
+      "learning_rate": 6.876476906552095e-06,
+      "loss": 0.0,
+      "step": 5918
+    },
+    {
+      "epoch": 0.3814526003737836,
+      "grad_norm": 0.020675918183340834,
+      "learning_rate": 6.8757608306480495e-06,
+      "loss": 0.0002,
+      "step": 5919
+    },
+    {
+      "epoch": 0.3815170458207128,
+      "grad_norm": 0.0027330285181974686,
+      "learning_rate": 6.875044754744004e-06,
+      "loss": 0.0,
+      "step": 5920
+    },
+    {
+      "epoch": 0.38158149126764196,
+      "grad_norm": 0.020531958035086353,
+      "learning_rate": 6.874328678839957e-06,
+      "loss": 0.0,
+      "step": 5921
+    },
+    {
+      "epoch": 0.3816459367145711,
+      "grad_norm": 0.49557236434042445,
+      "learning_rate": 6.873612602935912e-06,
+      "loss": 0.002,
+      "step": 5922
+    },
+    {
+      "epoch": 0.3817103821615003,
+      "grad_norm": 0.0013880445470229727,
+      "learning_rate": 6.872896527031866e-06,
+      "loss": 0.0,
+      "step": 5923
+    },
+    {
+      "epoch": 0.38177482760842946,
+      "grad_norm": 0.001162689409539696,
+      "learning_rate": 6.87218045112782e-06,
+      "loss": 0.0,
+      "step": 5924
+    },
+    {
+      "epoch": 0.38183927305535864,
+      "grad_norm": 0.11035111520052028,
+      "learning_rate": 6.8714643752237745e-06,
+      "loss": 0.0017,
+      "step": 5925
+    },
+    {
+      "epoch": 0.3819037185022878,
+      "grad_norm": 0.0007650463249493047,
+      "learning_rate": 6.870748299319728e-06,
+      "loss": 0.0,
+      "step": 5926
+    },
+    {
+      "epoch": 0.381968163949217,
+      "grad_norm": 0.019925055939437754,
+      "learning_rate": 6.870032223415682e-06,
+      "loss": 0.0001,
+      "step": 5927
+    },
+    {
+      "epoch": 0.38203260939614614,
+      "grad_norm": 0.014557531030572302,
+      "learning_rate": 6.8693161475116365e-06,
+      "loss": 0.0,
+      "step": 5928
+    },
+    {
+      "epoch": 0.3820970548430753,
+      "grad_norm": 0.0002573063824412536,
+      "learning_rate": 6.868600071607591e-06,
+      "loss": 0.0,
+      "step": 5929
+    },
+    {
+      "epoch": 0.3821615002900045,
+      "grad_norm": 0.0003335346718924549,
+      "learning_rate": 6.867883995703545e-06,
+      "loss": 0.0,
+      "step": 5930
+    },
+    {
+      "epoch": 0.3822259457369337,
+      "grad_norm": 0.0008118441796438887,
+      "learning_rate": 6.867167919799499e-06,
+      "loss": 0.0,
+      "step": 5931
+    },
+    {
+      "epoch": 0.3822903911838629,
+      "grad_norm": 0.005326369326512274,
+      "learning_rate": 6.866451843895453e-06,
+      "loss": 0.0,
+      "step": 5932
+    },
+    {
+      "epoch": 0.38235483663079206,
+      "grad_norm": 0.005278252984737066,
+      "learning_rate": 6.865735767991407e-06,
+      "loss": 0.0,
+      "step": 5933
+    },
+    {
+      "epoch": 0.3824192820777212,
+      "grad_norm": 0.03136676899392069,
+      "learning_rate": 6.8650196920873615e-06,
+      "loss": 0.0001,
+      "step": 5934
+    },
+    {
+      "epoch": 0.3824837275246504,
+      "grad_norm": 0.0009959009747822732,
+      "learning_rate": 6.864303616183317e-06,
+      "loss": 0.0,
+      "step": 5935
+    },
+    {
+      "epoch": 0.38254817297157956,
+      "grad_norm": 0.00042236544398543236,
+      "learning_rate": 6.863587540279271e-06,
+      "loss": 0.0,
+      "step": 5936
+    },
+    {
+      "epoch": 0.38261261841850874,
+      "grad_norm": 0.0006273076220670741,
+      "learning_rate": 6.862871464375224e-06,
+      "loss": 0.0,
+      "step": 5937
+    },
+    {
+      "epoch": 0.3826770638654379,
+      "grad_norm": 0.013919804187017291,
+      "learning_rate": 6.862155388471179e-06,
+      "loss": 0.0,
+      "step": 5938
+    },
+    {
+      "epoch": 0.38274150931236706,
+      "grad_norm": 0.006399515087818912,
+      "learning_rate": 6.861439312567133e-06,
+      "loss": 0.0,
+      "step": 5939
+    },
+    {
+      "epoch": 0.38280595475929624,
+      "grad_norm": 0.0009560729498972583,
+      "learning_rate": 6.860723236663087e-06,
+      "loss": 0.0,
+      "step": 5940
+    },
+    {
+      "epoch": 0.3828704002062254,
+      "grad_norm": 0.0574728807346303,
+      "learning_rate": 6.8600071607590416e-06,
+      "loss": 0.0002,
+      "step": 5941
+    },
+    {
+      "epoch": 0.3829348456531546,
+      "grad_norm": 0.005010008731110634,
+      "learning_rate": 6.859291084854995e-06,
+      "loss": 0.0,
+      "step": 5942
+    },
+    {
+      "epoch": 0.3829992911000838,
+      "grad_norm": 0.010932127051166089,
+      "learning_rate": 6.858575008950949e-06,
+      "loss": 0.0001,
+      "step": 5943
+    },
+    {
+      "epoch": 0.383063736547013,
+      "grad_norm": 0.24358331776633155,
+      "learning_rate": 6.857858933046904e-06,
+      "loss": 0.0002,
+      "step": 5944
+    },
+    {
+      "epoch": 0.3831281819939421,
+      "grad_norm": 0.005314563855596481,
+      "learning_rate": 6.857142857142858e-06,
+      "loss": 0.0,
+      "step": 5945
+    },
+    {
+      "epoch": 0.3831926274408713,
+      "grad_norm": 0.005676835426711824,
+      "learning_rate": 6.856426781238812e-06,
+      "loss": 0.0,
+      "step": 5946
+    },
+    {
+      "epoch": 0.3832570728878005,
+      "grad_norm": 0.0003028851649869988,
+      "learning_rate": 6.855710705334766e-06,
+      "loss": 0.0,
+      "step": 5947
+    },
+    {
+      "epoch": 0.38332151833472966,
+      "grad_norm": 0.29716372253412965,
+      "learning_rate": 6.85499462943072e-06,
+      "loss": 0.0332,
+      "step": 5948
+    },
+    {
+      "epoch": 0.38338596378165885,
+      "grad_norm": 0.001941066789563445,
+      "learning_rate": 6.854278553526674e-06,
+      "loss": 0.0,
+      "step": 5949
+    },
+    {
+      "epoch": 0.383450409228588,
+      "grad_norm": 0.21121318779958853,
+      "learning_rate": 6.8535624776226286e-06,
+      "loss": 0.0013,
+      "step": 5950
+    },
+    {
+      "epoch": 0.38351485467551716,
+      "grad_norm": 0.023995077734359705,
+      "learning_rate": 6.852846401718583e-06,
+      "loss": 0.0001,
+      "step": 5951
+    },
+    {
+      "epoch": 0.38357930012244634,
+      "grad_norm": 0.05211629073163899,
+      "learning_rate": 6.852130325814536e-06,
+      "loss": 0.0004,
+      "step": 5952
+    },
+    {
+      "epoch": 0.3836437455693755,
+      "grad_norm": 0.000493550789505138,
+      "learning_rate": 6.851414249910491e-06,
+      "loss": 0.0,
+      "step": 5953
+    },
+    {
+      "epoch": 0.3837081910163047,
+      "grad_norm": 0.0003923175907974961,
+      "learning_rate": 6.850698174006445e-06,
+      "loss": 0.0,
+      "step": 5954
+    },
+    {
+      "epoch": 0.3837726364632339,
+      "grad_norm": 0.009923200087034062,
+      "learning_rate": 6.849982098102399e-06,
+      "loss": 0.0,
+      "step": 5955
+    },
+    {
+      "epoch": 0.383837081910163,
+      "grad_norm": 0.03499818907090998,
+      "learning_rate": 6.8492660221983535e-06,
+      "loss": 0.0001,
+      "step": 5956
+    },
+    {
+      "epoch": 0.3839015273570922,
+      "grad_norm": 0.0037882630463259115,
+      "learning_rate": 6.848549946294307e-06,
+      "loss": 0.0,
+      "step": 5957
+    },
+    {
+      "epoch": 0.3839659728040214,
+      "grad_norm": 0.00421029141255445,
+      "learning_rate": 6.847833870390261e-06,
+      "loss": 0.0,
+      "step": 5958
+    },
+    {
+      "epoch": 0.3840304182509506,
+      "grad_norm": 0.0005011436316182269,
+      "learning_rate": 6.847117794486216e-06,
+      "loss": 0.0,
+      "step": 5959
+    },
+    {
+      "epoch": 0.38409486369787976,
+      "grad_norm": 0.014270384782207152,
+      "learning_rate": 6.846401718582171e-06,
+      "loss": 0.0,
+      "step": 5960
+    },
+    {
+      "epoch": 0.38415930914480895,
+      "grad_norm": 0.32118606075803946,
+      "learning_rate": 6.845685642678125e-06,
+      "loss": 0.0023,
+      "step": 5961
+    },
+    {
+      "epoch": 0.3842237545917381,
+      "grad_norm": 0.0011372247327576077,
+      "learning_rate": 6.844969566774079e-06,
+      "loss": 0.0,
+      "step": 5962
+    },
+    {
+      "epoch": 0.38428820003866726,
+      "grad_norm": 0.00047110262412149026,
+      "learning_rate": 6.844253490870033e-06,
+      "loss": 0.0,
+      "step": 5963
+    },
+    {
+      "epoch": 0.38435264548559644,
+      "grad_norm": 0.004175332756483595,
+      "learning_rate": 6.843537414965987e-06,
+      "loss": 0.0,
+      "step": 5964
+    },
+    {
+      "epoch": 0.38441709093252563,
+      "grad_norm": 0.00842632735103172,
+      "learning_rate": 6.842821339061941e-06,
+      "loss": 0.0001,
+      "step": 5965
+    },
+    {
+      "epoch": 0.3844815363794548,
+      "grad_norm": 0.001385774924796425,
+      "learning_rate": 6.842105263157896e-06,
+      "loss": 0.0,
+      "step": 5966
+    },
+    {
+      "epoch": 0.38454598182638394,
+      "grad_norm": 0.00960274895131003,
+      "learning_rate": 6.84138918725385e-06,
+      "loss": 0.0001,
+      "step": 5967
+    },
+    {
+      "epoch": 0.3846104272733131,
+      "grad_norm": 0.0014732819087795253,
+      "learning_rate": 6.840673111349803e-06,
+      "loss": 0.0,
+      "step": 5968
+    },
+    {
+      "epoch": 0.3846748727202423,
+      "grad_norm": 0.0019206416515151602,
+      "learning_rate": 6.839957035445758e-06,
+      "loss": 0.0,
+      "step": 5969
+    },
+    {
+      "epoch": 0.3847393181671715,
+      "grad_norm": 0.010897749206447984,
+      "learning_rate": 6.839240959541712e-06,
+      "loss": 0.0001,
+      "step": 5970
+    },
+    {
+      "epoch": 0.3848037636141007,
+      "grad_norm": 0.003925340168642234,
+      "learning_rate": 6.838524883637666e-06,
+      "loss": 0.0,
+      "step": 5971
+    },
+    {
+      "epoch": 0.38486820906102986,
+      "grad_norm": 0.012405496974591106,
+      "learning_rate": 6.837808807733621e-06,
+      "loss": 0.0,
+      "step": 5972
+    },
+    {
+      "epoch": 0.384932654507959,
+      "grad_norm": 0.06288781766963841,
+      "learning_rate": 6.837092731829574e-06,
+      "loss": 0.0001,
+      "step": 5973
+    },
+    {
+      "epoch": 0.3849970999548882,
+      "grad_norm": 0.09090745498393638,
+      "learning_rate": 6.836376655925528e-06,
+      "loss": 0.0,
+      "step": 5974
+    },
+    {
+      "epoch": 0.38506154540181736,
+      "grad_norm": 0.0006160974128895556,
+      "learning_rate": 6.835660580021483e-06,
+      "loss": 0.0,
+      "step": 5975
+    },
+    {
+      "epoch": 0.38512599084874655,
+      "grad_norm": 0.0007932396092801955,
+      "learning_rate": 6.834944504117437e-06,
+      "loss": 0.0,
+      "step": 5976
+    },
+    {
+      "epoch": 0.38519043629567573,
+      "grad_norm": 0.05446525313130436,
+      "learning_rate": 6.834228428213391e-06,
+      "loss": 0.0001,
+      "step": 5977
+    },
+    {
+      "epoch": 0.38525488174260486,
+      "grad_norm": 0.006795695546119066,
+      "learning_rate": 6.833512352309345e-06,
+      "loss": 0.0,
+      "step": 5978
+    },
+    {
+      "epoch": 0.38531932718953404,
+      "grad_norm": 0.2266163268767816,
+      "learning_rate": 6.832796276405299e-06,
+      "loss": 0.0017,
+      "step": 5979
+    },
+    {
+      "epoch": 0.38538377263646323,
+      "grad_norm": 0.1509233551229267,
+      "learning_rate": 6.832080200501253e-06,
+      "loss": 0.0018,
+      "step": 5980
+    },
+    {
+      "epoch": 0.3854482180833924,
+      "grad_norm": 0.0038038159387376977,
+      "learning_rate": 6.831364124597208e-06,
+      "loss": 0.0,
+      "step": 5981
+    },
+    {
+      "epoch": 0.3855126635303216,
+      "grad_norm": 0.11071187248613129,
+      "learning_rate": 6.830648048693163e-06,
+      "loss": 0.0001,
+      "step": 5982
+    },
+    {
+      "epoch": 0.3855771089772508,
+      "grad_norm": 0.008326152462626822,
+      "learning_rate": 6.829931972789117e-06,
+      "loss": 0.0001,
+      "step": 5983
+    },
+    {
+      "epoch": 0.3856415544241799,
+      "grad_norm": 0.0029565684972451584,
+      "learning_rate": 6.8292158968850705e-06,
+      "loss": 0.0,
+      "step": 5984
+    },
+    {
+      "epoch": 0.3857059998711091,
+      "grad_norm": 0.5216882714265499,
+      "learning_rate": 6.828499820981025e-06,
+      "loss": 0.0014,
+      "step": 5985
+    },
+    {
+      "epoch": 0.3857704453180383,
+      "grad_norm": 0.002250543081538071,
+      "learning_rate": 6.827783745076979e-06,
+      "loss": 0.0,
+      "step": 5986
+    },
+    {
+      "epoch": 0.38583489076496746,
+      "grad_norm": 0.2361621589188941,
+      "learning_rate": 6.827067669172933e-06,
+      "loss": 0.0011,
+      "step": 5987
+    },
+    {
+      "epoch": 0.38589933621189665,
+      "grad_norm": 0.031054470860727275,
+      "learning_rate": 6.826351593268888e-06,
+      "loss": 0.0001,
+      "step": 5988
+    },
+    {
+      "epoch": 0.3859637816588258,
+      "grad_norm": 0.014908333265461695,
+      "learning_rate": 6.825635517364841e-06,
+      "loss": 0.0001,
+      "step": 5989
+    },
+    {
+      "epoch": 0.38602822710575496,
+      "grad_norm": 0.042477362343784304,
+      "learning_rate": 6.8249194414607954e-06,
+      "loss": 0.0001,
+      "step": 5990
+    },
+    {
+      "epoch": 0.38609267255268415,
+      "grad_norm": 0.011556256262070665,
+      "learning_rate": 6.82420336555675e-06,
+      "loss": 0.0,
+      "step": 5991
+    },
+    {
+      "epoch": 0.38615711799961333,
+      "grad_norm": 0.0008571054739306326,
+      "learning_rate": 6.823487289652704e-06,
+      "loss": 0.0,
+      "step": 5992
+    },
+    {
+      "epoch": 0.3862215634465425,
+      "grad_norm": 0.00012599860103708978,
+      "learning_rate": 6.822771213748658e-06,
+      "loss": 0.0,
+      "step": 5993
+    },
+    {
+      "epoch": 0.3862860088934717,
+      "grad_norm": 0.016258295239014172,
+      "learning_rate": 6.822055137844612e-06,
+      "loss": 0.0001,
+      "step": 5994
+    },
+    {
+      "epoch": 0.38635045434040083,
+      "grad_norm": 0.007365234092956695,
+      "learning_rate": 6.821339061940566e-06,
+      "loss": 0.0,
+      "step": 5995
+    },
+    {
+      "epoch": 0.38641489978733,
+      "grad_norm": 0.010457210348021977,
+      "learning_rate": 6.82062298603652e-06,
+      "loss": 0.0,
+      "step": 5996
+    },
+    {
+      "epoch": 0.3864793452342592,
+      "grad_norm": 0.0017911283307170947,
+      "learning_rate": 6.819906910132475e-06,
+      "loss": 0.0,
+      "step": 5997
+    },
+    {
+      "epoch": 0.3865437906811884,
+      "grad_norm": 0.021666974075202527,
+      "learning_rate": 6.819190834228428e-06,
+      "loss": 0.0001,
+      "step": 5998
+    },
+    {
+      "epoch": 0.38660823612811757,
+      "grad_norm": 0.09290394888785891,
+      "learning_rate": 6.8184747583243824e-06,
+      "loss": 0.001,
+      "step": 5999
+    },
+    {
+      "epoch": 0.38667268157504675,
+      "grad_norm": 0.04081432866768783,
+      "learning_rate": 6.817758682420337e-06,
+      "loss": 0.0,
+      "step": 6000
+    },
+    {
+      "epoch": 0.3867371270219759,
+      "grad_norm": 0.0030581764431921246,
+      "learning_rate": 6.817042606516291e-06,
+      "loss": 0.0,
+      "step": 6001
+    },
+    {
+      "epoch": 0.38680157246890506,
+      "grad_norm": 0.04727984482479228,
+      "learning_rate": 6.816326530612245e-06,
+      "loss": 0.0001,
+      "step": 6002
+    },
+    {
+      "epoch": 0.38686601791583425,
+      "grad_norm": 0.07452768727150452,
+      "learning_rate": 6.815610454708199e-06,
+      "loss": 0.0001,
+      "step": 6003
+    },
+    {
+      "epoch": 0.38693046336276343,
+      "grad_norm": 0.20755748626077528,
+      "learning_rate": 6.814894378804153e-06,
+      "loss": 0.0005,
+      "step": 6004
+    },
+    {
+      "epoch": 0.3869949088096926,
+      "grad_norm": 0.006755925785793144,
+      "learning_rate": 6.814178302900108e-06,
+      "loss": 0.0,
+      "step": 6005
+    },
+    {
+      "epoch": 0.38705935425662175,
+      "grad_norm": 0.17547190254891795,
+      "learning_rate": 6.8134622269960625e-06,
+      "loss": 0.0002,
+      "step": 6006
+    },
+    {
+      "epoch": 0.38712379970355093,
+      "grad_norm": 0.012661615225328758,
+      "learning_rate": 6.812746151092017e-06,
+      "loss": 0.0002,
+      "step": 6007
+    },
+    {
+      "epoch": 0.3871882451504801,
+      "grad_norm": 0.003090067111566923,
+      "learning_rate": 6.812030075187971e-06,
+      "loss": 0.0,
+      "step": 6008
+    },
+    {
+      "epoch": 0.3872526905974093,
+      "grad_norm": 0.19586265328413646,
+      "learning_rate": 6.811313999283925e-06,
+      "loss": 0.0012,
+      "step": 6009
+    },
+    {
+      "epoch": 0.3873171360443385,
+      "grad_norm": 4.473975694094967,
+      "learning_rate": 6.810597923379879e-06,
+      "loss": 0.0279,
+      "step": 6010
+    },
+    {
+      "epoch": 0.38738158149126767,
+      "grad_norm": 0.2445356530106942,
+      "learning_rate": 6.809881847475833e-06,
+      "loss": 0.0051,
+      "step": 6011
+    },
+    {
+      "epoch": 0.3874460269381968,
+      "grad_norm": 0.032982293469763065,
+      "learning_rate": 6.8091657715717875e-06,
+      "loss": 0.0003,
+      "step": 6012
+    },
+    {
+      "epoch": 0.387510472385126,
+      "grad_norm": 0.009209936588663899,
+      "learning_rate": 6.808449695667742e-06,
+      "loss": 0.0,
+      "step": 6013
+    },
+    {
+      "epoch": 0.38757491783205517,
+      "grad_norm": 0.00518529117122379,
+      "learning_rate": 6.807733619763695e-06,
+      "loss": 0.0001,
+      "step": 6014
+    },
+    {
+      "epoch": 0.38763936327898435,
+      "grad_norm": 0.07908431500910323,
+      "learning_rate": 6.8070175438596495e-06,
+      "loss": 0.0003,
+      "step": 6015
+    },
+    {
+      "epoch": 0.38770380872591353,
+      "grad_norm": 0.27541477016357735,
+      "learning_rate": 6.806301467955604e-06,
+      "loss": 0.0022,
+      "step": 6016
+    },
+    {
+      "epoch": 0.38776825417284266,
+      "grad_norm": 0.006634024515463209,
+      "learning_rate": 6.805585392051558e-06,
+      "loss": 0.0,
+      "step": 6017
+    },
+    {
+      "epoch": 0.38783269961977185,
+      "grad_norm": 0.011857648203153355,
+      "learning_rate": 6.804869316147512e-06,
+      "loss": 0.0,
+      "step": 6018
+    },
+    {
+      "epoch": 0.38789714506670103,
+      "grad_norm": 0.0028286459308458976,
+      "learning_rate": 6.804153240243466e-06,
+      "loss": 0.0,
+      "step": 6019
+    },
+    {
+      "epoch": 0.3879615905136302,
+      "grad_norm": 0.01959234457512042,
+      "learning_rate": 6.80343716433942e-06,
+      "loss": 0.0,
+      "step": 6020
+    },
+    {
+      "epoch": 0.3880260359605594,
+      "grad_norm": 0.5422197019045307,
+      "learning_rate": 6.8027210884353745e-06,
+      "loss": 0.0033,
+      "step": 6021
+    },
+    {
+      "epoch": 0.3880904814074886,
+      "grad_norm": 0.2587556781468276,
+      "learning_rate": 6.802005012531329e-06,
+      "loss": 0.0003,
+      "step": 6022
+    },
+    {
+      "epoch": 0.3881549268544177,
+      "grad_norm": 0.008435287658794083,
+      "learning_rate": 6.801288936627283e-06,
+      "loss": 0.0,
+      "step": 6023
+    },
+    {
+      "epoch": 0.3882193723013469,
+      "grad_norm": 0.01238562473581905,
+      "learning_rate": 6.8005728607232365e-06,
+      "loss": 0.0,
+      "step": 6024
+    },
+    {
+      "epoch": 0.3882838177482761,
+      "grad_norm": 1.987109665357907,
+      "learning_rate": 6.799856784819191e-06,
+      "loss": 0.0106,
+      "step": 6025
+    },
+    {
+      "epoch": 0.38834826319520527,
+      "grad_norm": 0.04223822440592762,
+      "learning_rate": 6.799140708915145e-06,
+      "loss": 0.0001,
+      "step": 6026
+    },
+    {
+      "epoch": 0.38841270864213445,
+      "grad_norm": 0.002983615328333293,
+      "learning_rate": 6.798424633011099e-06,
+      "loss": 0.0,
+      "step": 6027
+    },
+    {
+      "epoch": 0.3884771540890636,
+      "grad_norm": 0.14209061717448843,
+      "learning_rate": 6.797708557107054e-06,
+      "loss": 0.0002,
+      "step": 6028
+    },
+    {
+      "epoch": 0.38854159953599277,
+      "grad_norm": 0.07302858039418998,
+      "learning_rate": 6.796992481203009e-06,
+      "loss": 0.0001,
+      "step": 6029
+    },
+    {
+      "epoch": 0.38860604498292195,
+      "grad_norm": 0.03682687955726151,
+      "learning_rate": 6.796276405298962e-06,
+      "loss": 0.0,
+      "step": 6030
+    },
+    {
+      "epoch": 0.38867049042985113,
+      "grad_norm": 0.009217186042876992,
+      "learning_rate": 6.795560329394917e-06,
+      "loss": 0.0001,
+      "step": 6031
+    },
+    {
+      "epoch": 0.3887349358767803,
+      "grad_norm": 0.00352494553698736,
+      "learning_rate": 6.794844253490871e-06,
+      "loss": 0.0,
+      "step": 6032
+    },
+    {
+      "epoch": 0.3887993813237095,
+      "grad_norm": 0.029486892316410652,
+      "learning_rate": 6.794128177586825e-06,
+      "loss": 0.0,
+      "step": 6033
+    },
+    {
+      "epoch": 0.38886382677063863,
+      "grad_norm": 0.00390147836480369,
+      "learning_rate": 6.7934121016827795e-06,
+      "loss": 0.0,
+      "step": 6034
+    },
+    {
+      "epoch": 0.3889282722175678,
+      "grad_norm": 0.013006879238566681,
+      "learning_rate": 6.792696025778733e-06,
+      "loss": 0.0,
+      "step": 6035
+    },
+    {
+      "epoch": 0.388992717664497,
+      "grad_norm": 0.11093205133006082,
+      "learning_rate": 6.791979949874687e-06,
+      "loss": 0.0002,
+      "step": 6036
+    },
+    {
+      "epoch": 0.3890571631114262,
+      "grad_norm": 0.06791830821969097,
+      "learning_rate": 6.7912638739706415e-06,
+      "loss": 0.0001,
+      "step": 6037
+    },
+    {
+      "epoch": 0.38912160855835537,
+      "grad_norm": 0.003451845629040953,
+      "learning_rate": 6.790547798066596e-06,
+      "loss": 0.0,
+      "step": 6038
+    },
+    {
+      "epoch": 0.38918605400528455,
+      "grad_norm": 0.09567374634863354,
+      "learning_rate": 6.78983172216255e-06,
+      "loss": 0.0001,
+      "step": 6039
+    },
+    {
+      "epoch": 0.3892504994522137,
+      "grad_norm": 0.08102874455585664,
+      "learning_rate": 6.789115646258504e-06,
+      "loss": 0.0001,
+      "step": 6040
+    },
+    {
+      "epoch": 0.38931494489914287,
+      "grad_norm": 0.09263870092366862,
+      "learning_rate": 6.788399570354458e-06,
+      "loss": 0.0002,
+      "step": 6041
+    },
+    {
+      "epoch": 0.38937939034607205,
+      "grad_norm": 0.058662957804021545,
+      "learning_rate": 6.787683494450412e-06,
+      "loss": 0.0001,
+      "step": 6042
+    },
+    {
+      "epoch": 0.38944383579300124,
+      "grad_norm": 0.028517596008494257,
+      "learning_rate": 6.7869674185463665e-06,
+      "loss": 0.0,
+      "step": 6043
+    },
+    {
+      "epoch": 0.3895082812399304,
+      "grad_norm": 0.4768442821733463,
+      "learning_rate": 6.786251342642321e-06,
+      "loss": 0.0004,
+      "step": 6044
+    },
+    {
+      "epoch": 0.38957272668685955,
+      "grad_norm": 0.07491564763077205,
+      "learning_rate": 6.785535266738274e-06,
+      "loss": 0.0002,
+      "step": 6045
+    },
+    {
+      "epoch": 0.38963717213378873,
+      "grad_norm": 0.13349859351138746,
+      "learning_rate": 6.7848191908342285e-06,
+      "loss": 0.0002,
+      "step": 6046
+    },
+    {
+      "epoch": 0.3897016175807179,
+      "grad_norm": 0.01829255571236091,
+      "learning_rate": 6.784103114930183e-06,
+      "loss": 0.0001,
+      "step": 6047
+    },
+    {
+      "epoch": 0.3897660630276471,
+      "grad_norm": 0.005630099985821416,
+      "learning_rate": 6.783387039026137e-06,
+      "loss": 0.0,
+      "step": 6048
+    },
+    {
+      "epoch": 0.3898305084745763,
+      "grad_norm": 0.008533366939692776,
+      "learning_rate": 6.7826709631220914e-06,
+      "loss": 0.0,
+      "step": 6049
+    },
+    {
+      "epoch": 0.38989495392150547,
+      "grad_norm": 0.21604861616295312,
+      "learning_rate": 6.781954887218045e-06,
+      "loss": 0.0008,
+      "step": 6050
+    },
+    {
+      "epoch": 0.3899593993684346,
+      "grad_norm": 0.04892616443141344,
+      "learning_rate": 6.781238811313999e-06,
+      "loss": 0.0001,
+      "step": 6051
+    },
+    {
+      "epoch": 0.3900238448153638,
+      "grad_norm": 0.003492646098469076,
+      "learning_rate": 6.780522735409954e-06,
+      "loss": 0.0,
+      "step": 6052
+    },
+    {
+      "epoch": 0.39008829026229297,
+      "grad_norm": 0.17101291092330484,
+      "learning_rate": 6.779806659505909e-06,
+      "loss": 0.0007,
+      "step": 6053
+    },
+    {
+      "epoch": 0.39015273570922215,
+      "grad_norm": 0.0006625030070719406,
+      "learning_rate": 6.779090583601863e-06,
+      "loss": 0.0,
+      "step": 6054
+    },
+    {
+      "epoch": 0.39021718115615134,
+      "grad_norm": 0.0407639185230089,
+      "learning_rate": 6.778374507697817e-06,
+      "loss": 0.0001,
+      "step": 6055
+    },
+    {
+      "epoch": 0.39028162660308047,
+      "grad_norm": 0.017097571873581942,
+      "learning_rate": 6.777658431793771e-06,
+      "loss": 0.0,
+      "step": 6056
+    },
+    {
+      "epoch": 0.39034607205000965,
+      "grad_norm": 0.0012163750468424425,
+      "learning_rate": 6.776942355889725e-06,
+      "loss": 0.0,
+      "step": 6057
+    },
+    {
+      "epoch": 0.39041051749693884,
+      "grad_norm": 0.008539081604194343,
+      "learning_rate": 6.776226279985679e-06,
+      "loss": 0.0,
+      "step": 6058
+    },
+    {
+      "epoch": 0.390474962943868,
+      "grad_norm": 0.02127820832464942,
+      "learning_rate": 6.7755102040816336e-06,
+      "loss": 0.0016,
+      "step": 6059
+    },
+    {
+      "epoch": 0.3905394083907972,
+      "grad_norm": 0.05713715860478264,
+      "learning_rate": 6.774794128177588e-06,
+      "loss": 0.0002,
+      "step": 6060
+    },
+    {
+      "epoch": 0.3906038538377264,
+      "grad_norm": 0.040865544077726086,
+      "learning_rate": 6.774078052273541e-06,
+      "loss": 0.0003,
+      "step": 6061
+    },
+    {
+      "epoch": 0.3906682992846555,
+      "grad_norm": 0.0008292018234182559,
+      "learning_rate": 6.773361976369496e-06,
+      "loss": 0.0,
+      "step": 6062
+    },
+    {
+      "epoch": 0.3907327447315847,
+      "grad_norm": 0.10168170810440107,
+      "learning_rate": 6.77264590046545e-06,
+      "loss": 0.0004,
+      "step": 6063
+    },
+    {
+      "epoch": 0.3907971901785139,
+      "grad_norm": 0.003782680501955245,
+      "learning_rate": 6.771929824561404e-06,
+      "loss": 0.0,
+      "step": 6064
+    },
+    {
+      "epoch": 0.39086163562544307,
+      "grad_norm": 0.07099556273375927,
+      "learning_rate": 6.7712137486573585e-06,
+      "loss": 0.0001,
+      "step": 6065
+    },
+    {
+      "epoch": 0.39092608107237226,
+      "grad_norm": 0.0079087437697562,
+      "learning_rate": 6.770497672753312e-06,
+      "loss": 0.0,
+      "step": 6066
+    },
+    {
+      "epoch": 0.3909905265193014,
+      "grad_norm": 0.0015870679381851005,
+      "learning_rate": 6.769781596849266e-06,
+      "loss": 0.0,
+      "step": 6067
+    },
+    {
+      "epoch": 0.39105497196623057,
+      "grad_norm": 0.23510664546596716,
+      "learning_rate": 6.7690655209452206e-06,
+      "loss": 0.001,
+      "step": 6068
+    },
+    {
+      "epoch": 0.39111941741315975,
+      "grad_norm": 0.00031801601169482763,
+      "learning_rate": 6.768349445041175e-06,
+      "loss": 0.0,
+      "step": 6069
+    },
+    {
+      "epoch": 0.39118386286008894,
+      "grad_norm": 0.004575363429178179,
+      "learning_rate": 6.767633369137129e-06,
+      "loss": 0.0,
+      "step": 6070
+    },
+    {
+      "epoch": 0.3912483083070181,
+      "grad_norm": 0.060214690750501815,
+      "learning_rate": 6.766917293233083e-06,
+      "loss": 0.0002,
+      "step": 6071
+    },
+    {
+      "epoch": 0.3913127537539473,
+      "grad_norm": 0.0010421286728905228,
+      "learning_rate": 6.766201217329037e-06,
+      "loss": 0.0,
+      "step": 6072
+    },
+    {
+      "epoch": 0.39137719920087644,
+      "grad_norm": 0.4869503651905972,
+      "learning_rate": 6.765485141424991e-06,
+      "loss": 0.0004,
+      "step": 6073
+    },
+    {
+      "epoch": 0.3914416446478056,
+      "grad_norm": 0.21821732891056522,
+      "learning_rate": 6.7647690655209455e-06,
+      "loss": 0.0003,
+      "step": 6074
+    },
+    {
+      "epoch": 0.3915060900947348,
+      "grad_norm": 0.006221830004070171,
+      "learning_rate": 6.764052989616901e-06,
+      "loss": 0.0,
+      "step": 6075
+    },
+    {
+      "epoch": 0.391570535541664,
+      "grad_norm": 0.0037303944601452836,
+      "learning_rate": 6.763336913712855e-06,
+      "loss": 0.0,
+      "step": 6076
+    },
+    {
+      "epoch": 0.3916349809885932,
+      "grad_norm": 0.00043963797693598783,
+      "learning_rate": 6.762620837808808e-06,
+      "loss": 0.0,
+      "step": 6077
+    },
+    {
+      "epoch": 0.39169942643552236,
+      "grad_norm": 0.3138164567095846,
+      "learning_rate": 6.761904761904763e-06,
+      "loss": 0.0012,
+      "step": 6078
+    },
+    {
+      "epoch": 0.3917638718824515,
+      "grad_norm": 0.011132950687805643,
+      "learning_rate": 6.761188686000717e-06,
+      "loss": 0.0,
+      "step": 6079
+    },
+    {
+      "epoch": 0.39182831732938067,
+      "grad_norm": 0.46684292043449677,
+      "learning_rate": 6.760472610096671e-06,
+      "loss": 0.0011,
+      "step": 6080
+    },
+    {
+      "epoch": 0.39189276277630986,
+      "grad_norm": 0.0036284189679277653,
+      "learning_rate": 6.759756534192626e-06,
+      "loss": 0.0,
+      "step": 6081
+    },
+    {
+      "epoch": 0.39195720822323904,
+      "grad_norm": 0.00039873388616156857,
+      "learning_rate": 6.759040458288579e-06,
+      "loss": 0.0,
+      "step": 6082
+    },
+    {
+      "epoch": 0.3920216536701682,
+      "grad_norm": 0.009515007808803775,
+      "learning_rate": 6.758324382384533e-06,
+      "loss": 0.0,
+      "step": 6083
+    },
+    {
+      "epoch": 0.39208609911709735,
+      "grad_norm": 0.05785692576468786,
+      "learning_rate": 6.757608306480488e-06,
+      "loss": 0.0002,
+      "step": 6084
+    },
+    {
+      "epoch": 0.39215054456402654,
+      "grad_norm": 0.00735022972663383,
+      "learning_rate": 6.756892230576442e-06,
+      "loss": 0.0,
+      "step": 6085
+    },
+    {
+      "epoch": 0.3922149900109557,
+      "grad_norm": 0.0009415286435785793,
+      "learning_rate": 6.756176154672396e-06,
+      "loss": 0.0,
+      "step": 6086
+    },
+    {
+      "epoch": 0.3922794354578849,
+      "grad_norm": 0.02875433070604497,
+      "learning_rate": 6.75546007876835e-06,
+      "loss": 0.0,
+      "step": 6087
+    },
+    {
+      "epoch": 0.3923438809048141,
+      "grad_norm": 0.04135847135380983,
+      "learning_rate": 6.754744002864304e-06,
+      "loss": 0.0,
+      "step": 6088
+    },
+    {
+      "epoch": 0.3924083263517433,
+      "grad_norm": 0.004634106539941361,
+      "learning_rate": 6.754027926960258e-06,
+      "loss": 0.0,
+      "step": 6089
+    },
+    {
+      "epoch": 0.3924727717986724,
+      "grad_norm": 0.37803271956724577,
+      "learning_rate": 6.753311851056213e-06,
+      "loss": 0.003,
+      "step": 6090
+    },
+    {
+      "epoch": 0.3925372172456016,
+      "grad_norm": 0.0013209306774567687,
+      "learning_rate": 6.752595775152166e-06,
+      "loss": 0.0,
+      "step": 6091
+    },
+    {
+      "epoch": 0.3926016626925308,
+      "grad_norm": 0.008602022225160995,
+      "learning_rate": 6.75187969924812e-06,
+      "loss": 0.0,
+      "step": 6092
+    },
+    {
+      "epoch": 0.39266610813945996,
+      "grad_norm": 0.0026656787122006495,
+      "learning_rate": 6.751163623344075e-06,
+      "loss": 0.0,
+      "step": 6093
+    },
+    {
+      "epoch": 0.39273055358638914,
+      "grad_norm": 0.12914930173706923,
+      "learning_rate": 6.750447547440029e-06,
+      "loss": 0.0003,
+      "step": 6094
+    },
+    {
+      "epoch": 0.39279499903331827,
+      "grad_norm": 0.10282103505984658,
+      "learning_rate": 6.749731471535983e-06,
+      "loss": 0.0001,
+      "step": 6095
+    },
+    {
+      "epoch": 0.39285944448024746,
+      "grad_norm": 0.013621648850027368,
+      "learning_rate": 6.749015395631937e-06,
+      "loss": 0.0,
+      "step": 6096
+    },
+    {
+      "epoch": 0.39292388992717664,
+      "grad_norm": 0.038448006543109814,
+      "learning_rate": 6.748299319727891e-06,
+      "loss": 0.0001,
+      "step": 6097
+    },
+    {
+      "epoch": 0.3929883353741058,
+      "grad_norm": 0.044204079857528784,
+      "learning_rate": 6.747583243823845e-06,
+      "loss": 0.0006,
+      "step": 6098
+    },
+    {
+      "epoch": 0.393052780821035,
+      "grad_norm": 0.046062306311024155,
+      "learning_rate": 6.7468671679198004e-06,
+      "loss": 0.0007,
+      "step": 6099
+    },
+    {
+      "epoch": 0.3931172262679642,
+      "grad_norm": 0.007857325413024335,
+      "learning_rate": 6.746151092015755e-06,
+      "loss": 0.0001,
+      "step": 6100
+    },
+    {
+      "epoch": 0.3931816717148933,
+      "grad_norm": 0.01512873154302197,
+      "learning_rate": 6.745435016111709e-06,
+      "loss": 0.0001,
+      "step": 6101
+    },
+    {
+      "epoch": 0.3932461171618225,
+      "grad_norm": 0.0032276045978629686,
+      "learning_rate": 6.744718940207663e-06,
+      "loss": 0.0,
+      "step": 6102
+    },
+    {
+      "epoch": 0.3933105626087517,
+      "grad_norm": 0.0009057025388732807,
+      "learning_rate": 6.744002864303617e-06,
+      "loss": 0.0,
+      "step": 6103
+    },
+    {
+      "epoch": 0.3933750080556809,
+      "grad_norm": 0.03844354321300612,
+      "learning_rate": 6.743286788399571e-06,
+      "loss": 0.0001,
+      "step": 6104
+    },
+    {
+      "epoch": 0.39343945350261006,
+      "grad_norm": 0.0012272262348689547,
+      "learning_rate": 6.742570712495525e-06,
+      "loss": 0.0,
+      "step": 6105
+    },
+    {
+      "epoch": 0.3935038989495392,
+      "grad_norm": 0.001039082765129518,
+      "learning_rate": 6.74185463659148e-06,
+      "loss": 0.0,
+      "step": 6106
+    },
+    {
+      "epoch": 0.3935683443964684,
+      "grad_norm": 0.11303522260833805,
+      "learning_rate": 6.741138560687433e-06,
+      "loss": 0.0003,
+      "step": 6107
+    },
+    {
+      "epoch": 0.39363278984339756,
+      "grad_norm": 0.14047053251100844,
+      "learning_rate": 6.7404224847833874e-06,
+      "loss": 0.0009,
+      "step": 6108
+    },
+    {
+      "epoch": 0.39369723529032674,
+      "grad_norm": 0.00028781397467615895,
+      "learning_rate": 6.739706408879342e-06,
+      "loss": 0.0,
+      "step": 6109
+    },
+    {
+      "epoch": 0.3937616807372559,
+      "grad_norm": 0.0004087873048829891,
+      "learning_rate": 6.738990332975296e-06,
+      "loss": 0.0,
+      "step": 6110
+    },
+    {
+      "epoch": 0.3938261261841851,
+      "grad_norm": 0.00040317603909862624,
+      "learning_rate": 6.73827425707125e-06,
+      "loss": 0.0,
+      "step": 6111
+    },
+    {
+      "epoch": 0.39389057163111424,
+      "grad_norm": 0.00040317603909862624,
+      "learning_rate": 6.73827425707125e-06,
+      "loss": 0.0043,
+      "step": 6112
+    },
+    {
+      "epoch": 0.3939550170780434,
+      "grad_norm": 0.0003019461882047249,
+      "learning_rate": 6.737558181167204e-06,
+      "loss": 0.0,
+      "step": 6113
+    },
+    {
+      "epoch": 0.3940194625249726,
+      "grad_norm": 0.006519207135955577,
+      "learning_rate": 6.736842105263158e-06,
+      "loss": 0.0,
+      "step": 6114
+    },
+    {
+      "epoch": 0.3940839079719018,
+      "grad_norm": 0.0011064302177114479,
+      "learning_rate": 6.736126029359112e-06,
+      "loss": 0.0,
+      "step": 6115
+    },
+    {
+      "epoch": 0.394148353418831,
+      "grad_norm": 0.0011027793071489889,
+      "learning_rate": 6.735409953455067e-06,
+      "loss": 0.0,
+      "step": 6116
+    },
+    {
+      "epoch": 0.39421279886576016,
+      "grad_norm": 0.003786054726373903,
+      "learning_rate": 6.734693877551021e-06,
+      "loss": 0.0,
+      "step": 6117
+    },
+    {
+      "epoch": 0.3942772443126893,
+      "grad_norm": 0.28807663983357745,
+      "learning_rate": 6.7339778016469744e-06,
+      "loss": 0.0002,
+      "step": 6118
+    },
+    {
+      "epoch": 0.3943416897596185,
+      "grad_norm": 0.19951675602869184,
+      "learning_rate": 6.733261725742929e-06,
+      "loss": 0.0021,
+      "step": 6119
+    },
+    {
+      "epoch": 0.39440613520654766,
+      "grad_norm": 0.0014406398795617093,
+      "learning_rate": 6.732545649838883e-06,
+      "loss": 0.0,
+      "step": 6120
+    },
+    {
+      "epoch": 0.39447058065347684,
+      "grad_norm": 0.05896528389795827,
+      "learning_rate": 6.731829573934837e-06,
+      "loss": 0.0002,
+      "step": 6121
+    },
+    {
+      "epoch": 0.394535026100406,
+      "grad_norm": 0.0004485638853455591,
+      "learning_rate": 6.731113498030792e-06,
+      "loss": 0.0,
+      "step": 6122
+    },
+    {
+      "epoch": 0.39459947154733516,
+      "grad_norm": 0.000655233336787404,
+      "learning_rate": 6.730397422126747e-06,
+      "loss": 0.0,
+      "step": 6123
+    },
+    {
+      "epoch": 0.39466391699426434,
+      "grad_norm": 1.2452657214560217,
+      "learning_rate": 6.7296813462227e-06,
+      "loss": 0.0059,
+      "step": 6124
+    },
+    {
+      "epoch": 0.3947283624411935,
+      "grad_norm": 0.006466952664939254,
+      "learning_rate": 6.7289652703186545e-06,
+      "loss": 0.0,
+      "step": 6125
+    },
+    {
+      "epoch": 0.3947928078881227,
+      "grad_norm": 0.11695563902920848,
+      "learning_rate": 6.728249194414609e-06,
+      "loss": 0.0009,
+      "step": 6126
+    },
+    {
+      "epoch": 0.3948572533350519,
+      "grad_norm": 0.0054357978734342165,
+      "learning_rate": 6.727533118510563e-06,
+      "loss": 0.0,
+      "step": 6127
+    },
+    {
+      "epoch": 0.3949216987819811,
+      "grad_norm": 0.010678078004628366,
+      "learning_rate": 6.726817042606517e-06,
+      "loss": 0.0,
+      "step": 6128
+    },
+    {
+      "epoch": 0.3949861442289102,
+      "grad_norm": 0.004239491570611307,
+      "learning_rate": 6.726100966702471e-06,
+      "loss": 0.0,
+      "step": 6129
+    },
+    {
+      "epoch": 0.3950505896758394,
+      "grad_norm": 0.4474075958280737,
+      "learning_rate": 6.725384890798425e-06,
+      "loss": 0.0036,
+      "step": 6130
+    },
+    {
+      "epoch": 0.3951150351227686,
+      "grad_norm": 0.008688036460937971,
+      "learning_rate": 6.7246688148943795e-06,
+      "loss": 0.0,
+      "step": 6131
+    },
+    {
+      "epoch": 0.39517948056969776,
+      "grad_norm": 0.0029086975973019197,
+      "learning_rate": 6.723952738990334e-06,
+      "loss": 0.0,
+      "step": 6132
+    },
+    {
+      "epoch": 0.39524392601662695,
+      "grad_norm": 0.00804262993046448,
+      "learning_rate": 6.723236663086288e-06,
+      "loss": 0.0001,
+      "step": 6133
+    },
+    {
+      "epoch": 0.3953083714635561,
+      "grad_norm": 0.001383582353238081,
+      "learning_rate": 6.7225205871822415e-06,
+      "loss": 0.0,
+      "step": 6134
+    },
+    {
+      "epoch": 0.39537281691048526,
+      "grad_norm": 0.002657203034535678,
+      "learning_rate": 6.721804511278196e-06,
+      "loss": 0.0,
+      "step": 6135
+    },
+    {
+      "epoch": 0.39543726235741444,
+      "grad_norm": 0.22318623860445258,
+      "learning_rate": 6.72108843537415e-06,
+      "loss": 0.0007,
+      "step": 6136
+    },
+    {
+      "epoch": 0.3955017078043436,
+      "grad_norm": 0.060305021991659505,
+      "learning_rate": 6.720372359470104e-06,
+      "loss": 0.0001,
+      "step": 6137
+    },
+    {
+      "epoch": 0.3955661532512728,
+      "grad_norm": 0.005253792241862154,
+      "learning_rate": 6.719656283566059e-06,
+      "loss": 0.0,
+      "step": 6138
+    },
+    {
+      "epoch": 0.395630598698202,
+      "grad_norm": 0.0004886858584370133,
+      "learning_rate": 6.718940207662012e-06,
+      "loss": 0.0,
+      "step": 6139
+    },
+    {
+      "epoch": 0.3956950441451311,
+      "grad_norm": 0.005571864209276483,
+      "learning_rate": 6.7182241317579665e-06,
+      "loss": 0.0,
+      "step": 6140
+    },
+    {
+      "epoch": 0.3957594895920603,
+      "grad_norm": 0.0005897999561558652,
+      "learning_rate": 6.717508055853921e-06,
+      "loss": 0.0,
+      "step": 6141
+    },
+    {
+      "epoch": 0.3958239350389895,
+      "grad_norm": 0.018974760515579954,
+      "learning_rate": 6.716791979949875e-06,
+      "loss": 0.0,
+      "step": 6142
+    },
+    {
+      "epoch": 0.3958883804859187,
+      "grad_norm": 0.005510618869750594,
+      "learning_rate": 6.716075904045829e-06,
+      "loss": 0.0,
+      "step": 6143
+    },
+    {
+      "epoch": 0.39595282593284786,
+      "grad_norm": 0.005582761552538098,
+      "learning_rate": 6.715359828141783e-06,
+      "loss": 0.0,
+      "step": 6144
+    },
+    {
+      "epoch": 0.396017271379777,
+      "grad_norm": 0.0009583769130436984,
+      "learning_rate": 6.714643752237737e-06,
+      "loss": 0.0,
+      "step": 6145
+    },
+    {
+      "epoch": 0.3960817168267062,
+      "grad_norm": 0.0028119638281370062,
+      "learning_rate": 6.713927676333692e-06,
+      "loss": 0.0,
+      "step": 6146
+    },
+    {
+      "epoch": 0.39614616227363536,
+      "grad_norm": 0.0350255340781061,
+      "learning_rate": 6.7132116004296465e-06,
+      "loss": 0.0001,
+      "step": 6147
+    },
+    {
+      "epoch": 0.39621060772056454,
+      "grad_norm": 0.0018648797668935515,
+      "learning_rate": 6.712495524525601e-06,
+      "loss": 0.0,
+      "step": 6148
+    },
+    {
+      "epoch": 0.39627505316749373,
+      "grad_norm": 0.0003898422824717099,
+      "learning_rate": 6.711779448621555e-06,
+      "loss": 0.0,
+      "step": 6149
+    },
+    {
+      "epoch": 0.3963394986144229,
+      "grad_norm": 0.005369384247029605,
+      "learning_rate": 6.711063372717509e-06,
+      "loss": 0.0,
+      "step": 6150
+    },
+    {
+      "epoch": 0.39640394406135204,
+      "grad_norm": 0.0025898006320916687,
+      "learning_rate": 6.710347296813463e-06,
+      "loss": 0.0,
+      "step": 6151
+    },
+    {
+      "epoch": 0.3964683895082812,
+      "grad_norm": 0.05196281984407206,
+      "learning_rate": 6.709631220909417e-06,
+      "loss": 0.0005,
+      "step": 6152
+    },
+    {
+      "epoch": 0.3965328349552104,
+      "grad_norm": 0.07452371645686365,
+      "learning_rate": 6.7089151450053715e-06,
+      "loss": 0.0,
+      "step": 6153
+    },
+    {
+      "epoch": 0.3965972804021396,
+      "grad_norm": 0.19650088216253572,
+      "learning_rate": 6.708199069101326e-06,
+      "loss": 0.0024,
+      "step": 6154
+    },
+    {
+      "epoch": 0.3966617258490688,
+      "grad_norm": 0.12044238442729664,
+      "learning_rate": 6.707482993197279e-06,
+      "loss": 0.0003,
+      "step": 6155
+    },
+    {
+      "epoch": 0.39672617129599796,
+      "grad_norm": 0.3482266044446364,
+      "learning_rate": 6.7067669172932335e-06,
+      "loss": 0.0012,
+      "step": 6156
+    },
+    {
+      "epoch": 0.3967906167429271,
+      "grad_norm": 0.011913191458644948,
+      "learning_rate": 6.706050841389188e-06,
+      "loss": 0.0,
+      "step": 6157
+    },
+    {
+      "epoch": 0.3968550621898563,
+      "grad_norm": 0.006744656352361722,
+      "learning_rate": 6.705334765485142e-06,
+      "loss": 0.0,
+      "step": 6158
+    },
+    {
+      "epoch": 0.39691950763678546,
+      "grad_norm": 0.0047692479281992144,
+      "learning_rate": 6.7046186895810964e-06,
+      "loss": 0.0,
+      "step": 6159
+    },
+    {
+      "epoch": 0.39698395308371465,
+      "grad_norm": 0.39355205601269155,
+      "learning_rate": 6.70390261367705e-06,
+      "loss": 0.0013,
+      "step": 6160
+    },
+    {
+      "epoch": 0.39704839853064383,
+      "grad_norm": 0.20822374123710258,
+      "learning_rate": 6.703186537773004e-06,
+      "loss": 0.001,
+      "step": 6161
+    },
+    {
+      "epoch": 0.39711284397757296,
+      "grad_norm": 0.00042890634310612283,
+      "learning_rate": 6.7024704618689585e-06,
+      "loss": 0.0,
+      "step": 6162
+    },
+    {
+      "epoch": 0.39717728942450214,
+      "grad_norm": 0.32812195731069266,
+      "learning_rate": 6.701754385964913e-06,
+      "loss": 0.0003,
+      "step": 6163
+    },
+    {
+      "epoch": 0.39724173487143133,
+      "grad_norm": 0.19006787676608586,
+      "learning_rate": 6.701038310060867e-06,
+      "loss": 0.0004,
+      "step": 6164
+    },
+    {
+      "epoch": 0.3973061803183605,
+      "grad_norm": 0.17109802580415745,
+      "learning_rate": 6.7003222341568205e-06,
+      "loss": 0.0005,
+      "step": 6165
+    },
+    {
+      "epoch": 0.3973706257652897,
+      "grad_norm": 0.009018840277326589,
+      "learning_rate": 6.699606158252775e-06,
+      "loss": 0.0001,
+      "step": 6166
+    },
+    {
+      "epoch": 0.3974350712122189,
+      "grad_norm": 0.0005974862827873541,
+      "learning_rate": 6.698890082348729e-06,
+      "loss": 0.0,
+      "step": 6167
+    },
+    {
+      "epoch": 0.397499516659148,
+      "grad_norm": 0.0005127417844990209,
+      "learning_rate": 6.6981740064446834e-06,
+      "loss": 0.0,
+      "step": 6168
+    },
+    {
+      "epoch": 0.3975639621060772,
+      "grad_norm": 0.013986048446789382,
+      "learning_rate": 6.697457930540638e-06,
+      "loss": 0.0001,
+      "step": 6169
+    },
+    {
+      "epoch": 0.3976284075530064,
+      "grad_norm": 0.2875345027993739,
+      "learning_rate": 6.696741854636593e-06,
+      "loss": 0.0023,
+      "step": 6170
+    },
+    {
+      "epoch": 0.39769285299993556,
+      "grad_norm": 0.00013710756560736585,
+      "learning_rate": 6.696025778732546e-06,
+      "loss": 0.0,
+      "step": 6171
+    },
+    {
+      "epoch": 0.39775729844686475,
+      "grad_norm": 0.0073792311350031,
+      "learning_rate": 6.695309702828501e-06,
+      "loss": 0.0,
+      "step": 6172
+    },
+    {
+      "epoch": 0.3978217438937939,
+      "grad_norm": 0.003827701575672846,
+      "learning_rate": 6.694593626924455e-06,
+      "loss": 0.0,
+      "step": 6173
+    },
+    {
+      "epoch": 0.39788618934072306,
+      "grad_norm": 0.0034136686566082096,
+      "learning_rate": 6.693877551020409e-06,
+      "loss": 0.0,
+      "step": 6174
+    },
+    {
+      "epoch": 0.39795063478765225,
+      "grad_norm": 0.02060817741945624,
+      "learning_rate": 6.6931614751163635e-06,
+      "loss": 0.0002,
+      "step": 6175
+    },
+    {
+      "epoch": 0.39801508023458143,
+      "grad_norm": 0.0006820699695873835,
+      "learning_rate": 6.692445399212317e-06,
+      "loss": 0.0,
+      "step": 6176
+    },
+    {
+      "epoch": 0.3980795256815106,
+      "grad_norm": 0.007219121822066729,
+      "learning_rate": 6.691729323308271e-06,
+      "loss": 0.0,
+      "step": 6177
+    },
+    {
+      "epoch": 0.3981439711284398,
+      "grad_norm": 0.0012653471260716737,
+      "learning_rate": 6.6910132474042256e-06,
+      "loss": 0.0,
+      "step": 6178
+    },
+    {
+      "epoch": 0.39820841657536893,
+      "grad_norm": 0.017223988011398236,
+      "learning_rate": 6.69029717150018e-06,
+      "loss": 0.0001,
+      "step": 6179
+    },
+    {
+      "epoch": 0.3982728620222981,
+      "grad_norm": 0.6595386799147767,
+      "learning_rate": 6.689581095596134e-06,
+      "loss": 0.0037,
+      "step": 6180
+    },
+    {
+      "epoch": 0.3983373074692273,
+      "grad_norm": 0.016393956086872264,
+      "learning_rate": 6.688865019692088e-06,
+      "loss": 0.0,
+      "step": 6181
+    },
+    {
+      "epoch": 0.3984017529161565,
+      "grad_norm": 0.006429886962872136,
+      "learning_rate": 6.688148943788042e-06,
+      "loss": 0.0,
+      "step": 6182
+    },
+    {
+      "epoch": 0.39846619836308567,
+      "grad_norm": 0.03475362813153784,
+      "learning_rate": 6.687432867883996e-06,
+      "loss": 0.0,
+      "step": 6183
+    },
+    {
+      "epoch": 0.3985306438100148,
+      "grad_norm": 0.002141625646072484,
+      "learning_rate": 6.6867167919799505e-06,
+      "loss": 0.0,
+      "step": 6184
+    },
+    {
+      "epoch": 0.398595089256944,
+      "grad_norm": 0.009711719164443825,
+      "learning_rate": 6.686000716075905e-06,
+      "loss": 0.0001,
+      "step": 6185
+    },
+    {
+      "epoch": 0.39865953470387316,
+      "grad_norm": 0.007483122794986775,
+      "learning_rate": 6.685284640171858e-06,
+      "loss": 0.0001,
+      "step": 6186
+    },
+    {
+      "epoch": 0.39872398015080235,
+      "grad_norm": 0.0013225585110462972,
+      "learning_rate": 6.6845685642678126e-06,
+      "loss": 0.0,
+      "step": 6187
+    },
+    {
+      "epoch": 0.39878842559773153,
+      "grad_norm": 0.0007087233184781159,
+      "learning_rate": 6.683852488363767e-06,
+      "loss": 0.0,
+      "step": 6188
+    },
+    {
+      "epoch": 0.3988528710446607,
+      "grad_norm": 0.005808769746201201,
+      "learning_rate": 6.683136412459721e-06,
+      "loss": 0.0,
+      "step": 6189
+    },
+    {
+      "epoch": 0.39891731649158985,
+      "grad_norm": 0.03269624359247558,
+      "learning_rate": 6.682420336555675e-06,
+      "loss": 0.0001,
+      "step": 6190
+    },
+    {
+      "epoch": 0.39898176193851903,
+      "grad_norm": 0.06306389772063731,
+      "learning_rate": 6.681704260651629e-06,
+      "loss": 0.0003,
+      "step": 6191
+    },
+    {
+      "epoch": 0.3990462073854482,
+      "grad_norm": 0.0007749667094711631,
+      "learning_rate": 6.680988184747583e-06,
+      "loss": 0.0,
+      "step": 6192
+    },
+    {
+      "epoch": 0.3991106528323774,
+      "grad_norm": 1.1743724405160312,
+      "learning_rate": 6.680272108843538e-06,
+      "loss": 0.0051,
+      "step": 6193
+    },
+    {
+      "epoch": 0.3991750982793066,
+      "grad_norm": 0.0006233756383026622,
+      "learning_rate": 6.679556032939493e-06,
+      "loss": 0.0,
+      "step": 6194
+    },
+    {
+      "epoch": 0.39923954372623577,
+      "grad_norm": 0.1845996974583024,
+      "learning_rate": 6.678839957035447e-06,
+      "loss": 0.0012,
+      "step": 6195
+    },
+    {
+      "epoch": 0.3993039891731649,
+      "grad_norm": 0.04961964414596718,
+      "learning_rate": 6.678123881131401e-06,
+      "loss": 0.0001,
+      "step": 6196
+    },
+    {
+      "epoch": 0.3993684346200941,
+      "grad_norm": 0.002305596368655089,
+      "learning_rate": 6.677407805227355e-06,
+      "loss": 0.0,
+      "step": 6197
+    },
+    {
+      "epoch": 0.39943288006702327,
+      "grad_norm": 0.03989655149228882,
+      "learning_rate": 6.676691729323309e-06,
+      "loss": 0.0,
+      "step": 6198
+    },
+    {
+      "epoch": 0.39949732551395245,
+      "grad_norm": 0.11639607532249155,
+      "learning_rate": 6.675975653419263e-06,
+      "loss": 0.0017,
+      "step": 6199
+    },
+    {
+      "epoch": 0.39956177096088163,
+      "grad_norm": 0.08423818726228123,
+      "learning_rate": 6.675259577515218e-06,
+      "loss": 0.0017,
+      "step": 6200
+    },
+    {
+      "epoch": 0.39962621640781076,
+      "grad_norm": 0.0009415812576501265,
+      "learning_rate": 6.674543501611172e-06,
+      "loss": 0.0,
+      "step": 6201
+    },
+    {
+      "epoch": 0.39969066185473995,
+      "grad_norm": 0.03231412558563971,
+      "learning_rate": 6.673827425707125e-06,
+      "loss": 0.0,
+      "step": 6202
+    },
+    {
+      "epoch": 0.39975510730166913,
+      "grad_norm": 0.022533085776875975,
+      "learning_rate": 6.67311134980308e-06,
+      "loss": 0.0001,
+      "step": 6203
+    },
+    {
+      "epoch": 0.3998195527485983,
+      "grad_norm": 0.0004932371042244677,
+      "learning_rate": 6.672395273899034e-06,
+      "loss": 0.0,
+      "step": 6204
+    },
+    {
+      "epoch": 0.3998839981955275,
+      "grad_norm": 0.0016411312170160164,
+      "learning_rate": 6.671679197994988e-06,
+      "loss": 0.0,
+      "step": 6205
+    },
+    {
+      "epoch": 0.3999484436424567,
+      "grad_norm": 0.00045268622319586543,
+      "learning_rate": 6.670963122090942e-06,
+      "loss": 0.0,
+      "step": 6206
+    },
+    {
+      "epoch": 0.4000128890893858,
+      "grad_norm": 0.06922491009062504,
+      "learning_rate": 6.670247046186896e-06,
+      "loss": 0.0001,
+      "step": 6207
+    },
+    {
+      "epoch": 0.400077334536315,
+      "grad_norm": 0.012176174861812075,
+      "learning_rate": 6.66953097028285e-06,
+      "loss": 0.0,
+      "step": 6208
+    },
+    {
+      "epoch": 0.4001417799832442,
+      "grad_norm": 0.02192805518688391,
+      "learning_rate": 6.668814894378805e-06,
+      "loss": 0.0,
+      "step": 6209
+    },
+    {
+      "epoch": 0.40020622543017337,
+      "grad_norm": 0.01031527857257621,
+      "learning_rate": 6.668098818474759e-06,
+      "loss": 0.0,
+      "step": 6210
+    },
+    {
+      "epoch": 0.40027067087710255,
+      "grad_norm": 0.2088355398948857,
+      "learning_rate": 6.667382742570712e-06,
+      "loss": 0.0004,
+      "step": 6211
+    },
+    {
+      "epoch": 0.4003351163240317,
+      "grad_norm": 0.006589784635158972,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.0,
+      "step": 6212
+    },
+    {
+      "epoch": 0.40039956177096087,
+      "grad_norm": 0.590162759370787,
+      "learning_rate": 6.665950590762621e-06,
+      "loss": 0.0016,
+      "step": 6213
+    },
+    {
+      "epoch": 0.40046400721789005,
+      "grad_norm": 0.4982781116509973,
+      "learning_rate": 6.665234514858575e-06,
+      "loss": 0.0037,
+      "step": 6214
+    },
+    {
+      "epoch": 0.40052845266481923,
+      "grad_norm": 0.12883591900735675,
+      "learning_rate": 6.6645184389545295e-06,
+      "loss": 0.0,
+      "step": 6215
+    },
+    {
+      "epoch": 0.4005928981117484,
+      "grad_norm": 0.15875782090355953,
+      "learning_rate": 6.663802363050485e-06,
+      "loss": 0.003,
+      "step": 6216
+    },
+    {
+      "epoch": 0.4006573435586776,
+      "grad_norm": 0.021264245585666988,
+      "learning_rate": 6.663086287146439e-06,
+      "loss": 0.0,
+      "step": 6217
+    },
+    {
+      "epoch": 0.40072178900560673,
+      "grad_norm": 0.088981737988645,
+      "learning_rate": 6.6623702112423924e-06,
+      "loss": 0.0001,
+      "step": 6218
+    },
+    {
+      "epoch": 0.4007862344525359,
+      "grad_norm": 0.002934769665413855,
+      "learning_rate": 6.661654135338347e-06,
+      "loss": 0.0,
+      "step": 6219
+    },
+    {
+      "epoch": 0.4008506798994651,
+      "grad_norm": 0.15586118601461513,
+      "learning_rate": 6.660938059434301e-06,
+      "loss": 0.0006,
+      "step": 6220
+    },
+    {
+      "epoch": 0.4009151253463943,
+      "grad_norm": 0.28255903730610116,
+      "learning_rate": 6.660221983530255e-06,
+      "loss": 0.0004,
+      "step": 6221
+    },
+    {
+      "epoch": 0.40097957079332347,
+      "grad_norm": 0.028550108747655678,
+      "learning_rate": 6.659505907626209e-06,
+      "loss": 0.0001,
+      "step": 6222
+    },
+    {
+      "epoch": 0.4010440162402526,
+      "grad_norm": 0.27616440845200096,
+      "learning_rate": 6.658789831722163e-06,
+      "loss": 0.002,
+      "step": 6223
+    },
+    {
+      "epoch": 0.4011084616871818,
+      "grad_norm": 0.008006083560591043,
+      "learning_rate": 6.658073755818117e-06,
+      "loss": 0.0,
+      "step": 6224
+    },
+    {
+      "epoch": 0.40117290713411097,
+      "grad_norm": 0.07311599032007556,
+      "learning_rate": 6.657357679914072e-06,
+      "loss": 0.0002,
+      "step": 6225
+    },
+    {
+      "epoch": 0.40123735258104015,
+      "grad_norm": 0.0009471099356845897,
+      "learning_rate": 6.656641604010026e-06,
+      "loss": 0.0,
+      "step": 6226
+    },
+    {
+      "epoch": 0.40130179802796934,
+      "grad_norm": 0.00239955307892596,
+      "learning_rate": 6.6559255281059794e-06,
+      "loss": 0.0,
+      "step": 6227
+    },
+    {
+      "epoch": 0.4013662434748985,
+      "grad_norm": 0.03330112643076966,
+      "learning_rate": 6.655209452201934e-06,
+      "loss": 0.0,
+      "step": 6228
+    },
+    {
+      "epoch": 0.40143068892182765,
+      "grad_norm": 0.003579085964833083,
+      "learning_rate": 6.654493376297888e-06,
+      "loss": 0.0,
+      "step": 6229
+    },
+    {
+      "epoch": 0.40149513436875683,
+      "grad_norm": 0.0001288371734857651,
+      "learning_rate": 6.653777300393842e-06,
+      "loss": 0.0,
+      "step": 6230
+    },
+    {
+      "epoch": 0.401559579815686,
+      "grad_norm": 0.0001809409710459574,
+      "learning_rate": 6.653061224489797e-06,
+      "loss": 0.0,
+      "step": 6231
+    },
+    {
+      "epoch": 0.4016240252626152,
+      "grad_norm": 0.0018872319576630108,
+      "learning_rate": 6.65234514858575e-06,
+      "loss": 0.0,
+      "step": 6232
+    },
+    {
+      "epoch": 0.4016884707095444,
+      "grad_norm": 0.00037414685857723075,
+      "learning_rate": 6.651629072681704e-06,
+      "loss": 0.0,
+      "step": 6233
+    },
+    {
+      "epoch": 0.40175291615647357,
+      "grad_norm": 0.005488616829534191,
+      "learning_rate": 6.650912996777659e-06,
+      "loss": 0.0,
+      "step": 6234
+    },
+    {
+      "epoch": 0.4018173616034027,
+      "grad_norm": 0.0043051522679139476,
+      "learning_rate": 6.650196920873613e-06,
+      "loss": 0.0,
+      "step": 6235
+    },
+    {
+      "epoch": 0.4018818070503319,
+      "grad_norm": 0.18108599082511184,
+      "learning_rate": 6.649480844969567e-06,
+      "loss": 0.0005,
+      "step": 6236
+    },
+    {
+      "epoch": 0.40194625249726107,
+      "grad_norm": 0.0002171191700035768,
+      "learning_rate": 6.648764769065521e-06,
+      "loss": 0.0,
+      "step": 6237
+    },
+    {
+      "epoch": 0.40201069794419025,
+      "grad_norm": 0.0021956388300873758,
+      "learning_rate": 6.648048693161475e-06,
+      "loss": 0.0,
+      "step": 6238
+    },
+    {
+      "epoch": 0.40207514339111944,
+      "grad_norm": 0.009351226362029775,
+      "learning_rate": 6.647332617257429e-06,
+      "loss": 0.0001,
+      "step": 6239
+    },
+    {
+      "epoch": 0.40213958883804857,
+      "grad_norm": 0.00015364747358863305,
+      "learning_rate": 6.6466165413533845e-06,
+      "loss": 0.0,
+      "step": 6240
+    },
+    {
+      "epoch": 0.40220403428497775,
+      "grad_norm": 0.0046499385674727165,
+      "learning_rate": 6.645900465449339e-06,
+      "loss": 0.0,
+      "step": 6241
+    },
+    {
+      "epoch": 0.40226847973190694,
+      "grad_norm": 0.0010105762902176386,
+      "learning_rate": 6.645184389545293e-06,
+      "loss": 0.0,
+      "step": 6242
+    },
+    {
+      "epoch": 0.4023329251788361,
+      "grad_norm": 0.039143908131220026,
+      "learning_rate": 6.6444683136412465e-06,
+      "loss": 0.0001,
+      "step": 6243
+    },
+    {
+      "epoch": 0.4023973706257653,
+      "grad_norm": 0.0024909498493526964,
+      "learning_rate": 6.643752237737201e-06,
+      "loss": 0.0,
+      "step": 6244
+    },
+    {
+      "epoch": 0.4024618160726945,
+      "grad_norm": 0.030333604358856733,
+      "learning_rate": 6.643036161833155e-06,
+      "loss": 0.0,
+      "step": 6245
+    },
+    {
+      "epoch": 0.4025262615196236,
+      "grad_norm": 0.0003121790638401315,
+      "learning_rate": 6.642320085929109e-06,
+      "loss": 0.0,
+      "step": 6246
+    },
+    {
+      "epoch": 0.4025907069665528,
+      "grad_norm": 0.05674777091108059,
+      "learning_rate": 6.641604010025064e-06,
+      "loss": 0.0,
+      "step": 6247
+    },
+    {
+      "epoch": 0.402655152413482,
+      "grad_norm": 0.029250476772840036,
+      "learning_rate": 6.640887934121017e-06,
+      "loss": 0.0,
+      "step": 6248
+    },
+    {
+      "epoch": 0.40271959786041117,
+      "grad_norm": 9.698981643813166e-05,
+      "learning_rate": 6.6401718582169715e-06,
+      "loss": 0.0,
+      "step": 6249
+    },
+    {
+      "epoch": 0.40278404330734036,
+      "grad_norm": 0.27877300990720555,
+      "learning_rate": 6.639455782312926e-06,
+      "loss": 0.0026,
+      "step": 6250
+    },
+    {
+      "epoch": 0.4028484887542695,
+      "grad_norm": 0.0027097435262068716,
+      "learning_rate": 6.63873970640888e-06,
+      "loss": 0.0,
+      "step": 6251
+    },
+    {
+      "epoch": 0.40291293420119867,
+      "grad_norm": 0.05563554116756958,
+      "learning_rate": 6.638023630504834e-06,
+      "loss": 0.0001,
+      "step": 6252
+    },
+    {
+      "epoch": 0.40297737964812785,
+      "grad_norm": 0.03642531035428272,
+      "learning_rate": 6.637307554600788e-06,
+      "loss": 0.0001,
+      "step": 6253
+    },
+    {
+      "epoch": 0.40304182509505704,
+      "grad_norm": 0.029172964117816117,
+      "learning_rate": 6.636591478696742e-06,
+      "loss": 0.0001,
+      "step": 6254
+    },
+    {
+      "epoch": 0.4031062705419862,
+      "grad_norm": 0.002905543593009424,
+      "learning_rate": 6.635875402792696e-06,
+      "loss": 0.0,
+      "step": 6255
+    },
+    {
+      "epoch": 0.4031707159889154,
+      "grad_norm": 0.012390442078684512,
+      "learning_rate": 6.635159326888651e-06,
+      "loss": 0.0,
+      "step": 6256
+    },
+    {
+      "epoch": 0.40323516143584454,
+      "grad_norm": 0.021038528188060623,
+      "learning_rate": 6.634443250984605e-06,
+      "loss": 0.0002,
+      "step": 6257
+    },
+    {
+      "epoch": 0.4032996068827737,
+      "grad_norm": 0.3627454641874893,
+      "learning_rate": 6.6337271750805585e-06,
+      "loss": 0.0011,
+      "step": 6258
+    },
+    {
+      "epoch": 0.4033640523297029,
+      "grad_norm": 0.004103812286494867,
+      "learning_rate": 6.633011099176513e-06,
+      "loss": 0.0,
+      "step": 6259
+    },
+    {
+      "epoch": 0.4034284977766321,
+      "grad_norm": 0.46436891426010923,
+      "learning_rate": 6.632295023272467e-06,
+      "loss": 0.0023,
+      "step": 6260
+    },
+    {
+      "epoch": 0.4034929432235613,
+      "grad_norm": 0.05719079838357991,
+      "learning_rate": 6.631578947368421e-06,
+      "loss": 0.0001,
+      "step": 6261
+    },
+    {
+      "epoch": 0.4035573886704904,
+      "grad_norm": 0.001427135059155809,
+      "learning_rate": 6.630862871464376e-06,
+      "loss": 0.0,
+      "step": 6262
+    },
+    {
+      "epoch": 0.4036218341174196,
+      "grad_norm": 0.00043134871254642114,
+      "learning_rate": 6.630146795560331e-06,
+      "loss": 0.0,
+      "step": 6263
+    },
+    {
+      "epoch": 0.40368627956434877,
+      "grad_norm": 0.005838851823537722,
+      "learning_rate": 6.629430719656284e-06,
+      "loss": 0.0,
+      "step": 6264
+    },
+    {
+      "epoch": 0.40375072501127796,
+      "grad_norm": 0.0031162694259560825,
+      "learning_rate": 6.6287146437522385e-06,
+      "loss": 0.0,
+      "step": 6265
+    },
+    {
+      "epoch": 0.40381517045820714,
+      "grad_norm": 0.16612453815009942,
+      "learning_rate": 6.627998567848193e-06,
+      "loss": 0.0005,
+      "step": 6266
+    },
+    {
+      "epoch": 0.4038796159051363,
+      "grad_norm": 8.110659193095382e-05,
+      "learning_rate": 6.627282491944147e-06,
+      "loss": 0.0,
+      "step": 6267
+    },
+    {
+      "epoch": 0.40394406135206545,
+      "grad_norm": 0.09589340309400302,
+      "learning_rate": 6.6265664160401014e-06,
+      "loss": 0.0001,
+      "step": 6268
+    },
+    {
+      "epoch": 0.40400850679899464,
+      "grad_norm": 0.0001381144958984862,
+      "learning_rate": 6.625850340136055e-06,
+      "loss": 0.0,
+      "step": 6269
+    },
+    {
+      "epoch": 0.4040729522459238,
+      "grad_norm": 0.0010787557333930568,
+      "learning_rate": 6.625134264232009e-06,
+      "loss": 0.0,
+      "step": 6270
+    },
+    {
+      "epoch": 0.404137397692853,
+      "grad_norm": 0.0003107536158255425,
+      "learning_rate": 6.6244181883279635e-06,
+      "loss": 0.0,
+      "step": 6271
+    },
+    {
+      "epoch": 0.4042018431397822,
+      "grad_norm": 0.008608214525285639,
+      "learning_rate": 6.623702112423918e-06,
+      "loss": 0.0,
+      "step": 6272
+    },
+    {
+      "epoch": 0.4042662885867114,
+      "grad_norm": 0.00040518102927398483,
+      "learning_rate": 6.622986036519872e-06,
+      "loss": 0.0,
+      "step": 6273
+    },
+    {
+      "epoch": 0.4043307340336405,
+      "grad_norm": 0.36758800740593717,
+      "learning_rate": 6.6222699606158255e-06,
+      "loss": 0.0007,
+      "step": 6274
+    },
+    {
+      "epoch": 0.4043951794805697,
+      "grad_norm": 0.012724112433642078,
+      "learning_rate": 6.62155388471178e-06,
+      "loss": 0.0001,
+      "step": 6275
+    },
+    {
+      "epoch": 0.4044596249274989,
+      "grad_norm": 0.0034801321594823327,
+      "learning_rate": 6.620837808807734e-06,
+      "loss": 0.0015,
+      "step": 6276
+    },
+    {
+      "epoch": 0.40452407037442806,
+      "grad_norm": 0.0022402441000457106,
+      "learning_rate": 6.6201217329036884e-06,
+      "loss": 0.0,
+      "step": 6277
+    },
+    {
+      "epoch": 0.40458851582135724,
+      "grad_norm": 0.001198815179465232,
+      "learning_rate": 6.619405656999643e-06,
+      "loss": 0.0,
+      "step": 6278
+    },
+    {
+      "epoch": 0.40465296126828637,
+      "grad_norm": 0.02394095008108672,
+      "learning_rate": 6.618689581095596e-06,
+      "loss": 0.0001,
+      "step": 6279
+    },
+    {
+      "epoch": 0.40471740671521556,
+      "grad_norm": 0.008677542084958455,
+      "learning_rate": 6.6179735051915505e-06,
+      "loss": 0.0001,
+      "step": 6280
+    },
+    {
+      "epoch": 0.40478185216214474,
+      "grad_norm": 0.006015602815613465,
+      "learning_rate": 6.617257429287505e-06,
+      "loss": 0.0,
+      "step": 6281
+    },
+    {
+      "epoch": 0.4048462976090739,
+      "grad_norm": 0.00022256715957847932,
+      "learning_rate": 6.616541353383459e-06,
+      "loss": 0.0,
+      "step": 6282
+    },
+    {
+      "epoch": 0.4049107430560031,
+      "grad_norm": 0.0021269805569372334,
+      "learning_rate": 6.6158252774794125e-06,
+      "loss": 0.0,
+      "step": 6283
+    },
+    {
+      "epoch": 0.4049751885029323,
+      "grad_norm": 0.0019124917188764595,
+      "learning_rate": 6.615109201575367e-06,
+      "loss": 0.0,
+      "step": 6284
+    },
+    {
+      "epoch": 0.4050396339498614,
+      "grad_norm": 0.0006371406991821277,
+      "learning_rate": 6.614393125671321e-06,
+      "loss": 0.0,
+      "step": 6285
+    },
+    {
+      "epoch": 0.4051040793967906,
+      "grad_norm": 0.012221513144843947,
+      "learning_rate": 6.613677049767276e-06,
+      "loss": 0.0001,
+      "step": 6286
+    },
+    {
+      "epoch": 0.4051685248437198,
+      "grad_norm": 0.003629192668777277,
+      "learning_rate": 6.612960973863231e-06,
+      "loss": 0.0,
+      "step": 6287
+    },
+    {
+      "epoch": 0.405232970290649,
+      "grad_norm": 0.03566383629410653,
+      "learning_rate": 6.612244897959185e-06,
+      "loss": 0.0001,
+      "step": 6288
+    },
+    {
+      "epoch": 0.40529741573757816,
+      "grad_norm": 0.011285903547055491,
+      "learning_rate": 6.611528822055139e-06,
+      "loss": 0.0,
+      "step": 6289
+    },
+    {
+      "epoch": 0.4053618611845073,
+      "grad_norm": 0.0027852915169590963,
+      "learning_rate": 6.610812746151093e-06,
+      "loss": 0.0,
+      "step": 6290
+    },
+    {
+      "epoch": 0.4054263066314365,
+      "grad_norm": 0.10941760050982087,
+      "learning_rate": 6.610096670247047e-06,
+      "loss": 0.0009,
+      "step": 6291
+    },
+    {
+      "epoch": 0.40549075207836566,
+      "grad_norm": 0.002644033621041055,
+      "learning_rate": 6.609380594343001e-06,
+      "loss": 0.0,
+      "step": 6292
+    },
+    {
+      "epoch": 0.40555519752529484,
+      "grad_norm": 0.00019279917264443554,
+      "learning_rate": 6.6086645184389555e-06,
+      "loss": 0.0,
+      "step": 6293
+    },
+    {
+      "epoch": 0.405619642972224,
+      "grad_norm": 0.0005731582641601562,
+      "learning_rate": 6.60794844253491e-06,
+      "loss": 0.0,
+      "step": 6294
+    },
+    {
+      "epoch": 0.4056840884191532,
+      "grad_norm": 0.006801553876835982,
+      "learning_rate": 6.607232366630863e-06,
+      "loss": 0.0,
+      "step": 6295
+    },
+    {
+      "epoch": 0.40574853386608234,
+      "grad_norm": 0.003933302052609157,
+      "learning_rate": 6.6065162907268176e-06,
+      "loss": 0.0,
+      "step": 6296
+    },
+    {
+      "epoch": 0.4058129793130115,
+      "grad_norm": 0.0006329129133961246,
+      "learning_rate": 6.605800214822772e-06,
+      "loss": 0.0,
+      "step": 6297
+    },
+    {
+      "epoch": 0.4058774247599407,
+      "grad_norm": 0.0004284577227252947,
+      "learning_rate": 6.605084138918726e-06,
+      "loss": 0.0,
+      "step": 6298
+    },
+    {
+      "epoch": 0.4059418702068699,
+      "grad_norm": 0.01899693347187632,
+      "learning_rate": 6.60436806301468e-06,
+      "loss": 0.0001,
+      "step": 6299
+    },
+    {
+      "epoch": 0.4060063156537991,
+      "grad_norm": 0.008696248565383551,
+      "learning_rate": 6.603651987110634e-06,
+      "loss": 0.0001,
+      "step": 6300
+    },
+    {
+      "epoch": 0.4060707611007282,
+      "grad_norm": 0.00046881937012844134,
+      "learning_rate": 6.602935911206588e-06,
+      "loss": 0.0,
+      "step": 6301
+    },
+    {
+      "epoch": 0.4061352065476574,
+      "grad_norm": 0.0021209476661694926,
+      "learning_rate": 6.6022198353025425e-06,
+      "loss": 0.0,
+      "step": 6302
+    },
+    {
+      "epoch": 0.4061996519945866,
+      "grad_norm": 0.0002067565192842928,
+      "learning_rate": 6.601503759398497e-06,
+      "loss": 0.0,
+      "step": 6303
+    },
+    {
+      "epoch": 0.40626409744151576,
+      "grad_norm": 0.005868360340397581,
+      "learning_rate": 6.60078768349445e-06,
+      "loss": 0.0,
+      "step": 6304
+    },
+    {
+      "epoch": 0.40632854288844494,
+      "grad_norm": 0.0006698871174751843,
+      "learning_rate": 6.6000716075904046e-06,
+      "loss": 0.0,
+      "step": 6305
+    },
+    {
+      "epoch": 0.40639298833537413,
+      "grad_norm": 0.008249500911233668,
+      "learning_rate": 6.599355531686359e-06,
+      "loss": 0.0,
+      "step": 6306
+    },
+    {
+      "epoch": 0.40645743378230326,
+      "grad_norm": 0.0003297776276702966,
+      "learning_rate": 6.598639455782313e-06,
+      "loss": 0.0,
+      "step": 6307
+    },
+    {
+      "epoch": 0.40652187922923244,
+      "grad_norm": 0.763014566614935,
+      "learning_rate": 6.5979233798782675e-06,
+      "loss": 0.0032,
+      "step": 6308
+    },
+    {
+      "epoch": 0.4065863246761616,
+      "grad_norm": 0.0030389339003442026,
+      "learning_rate": 6.597207303974221e-06,
+      "loss": 0.0,
+      "step": 6309
+    },
+    {
+      "epoch": 0.4066507701230908,
+      "grad_norm": 0.002413791142186581,
+      "learning_rate": 6.596491228070177e-06,
+      "loss": 0.0,
+      "step": 6310
+    },
+    {
+      "epoch": 0.40671521557002,
+      "grad_norm": 0.037547443349476,
+      "learning_rate": 6.59577515216613e-06,
+      "loss": 0.0001,
+      "step": 6311
+    },
+    {
+      "epoch": 0.4067796610169492,
+      "grad_norm": 0.0012816616887669533,
+      "learning_rate": 6.595059076262085e-06,
+      "loss": 0.0,
+      "step": 6312
+    },
+    {
+      "epoch": 0.4068441064638783,
+      "grad_norm": 0.4274243489749732,
+      "learning_rate": 6.594343000358039e-06,
+      "loss": 0.0025,
+      "step": 6313
+    },
+    {
+      "epoch": 0.4069085519108075,
+      "grad_norm": 0.0009116007250358199,
+      "learning_rate": 6.593626924453993e-06,
+      "loss": 0.0,
+      "step": 6314
+    },
+    {
+      "epoch": 0.4069729973577367,
+      "grad_norm": 0.4122105153435198,
+      "learning_rate": 6.592910848549947e-06,
+      "loss": 0.0007,
+      "step": 6315
+    },
+    {
+      "epoch": 0.40703744280466586,
+      "grad_norm": 0.032060547766800215,
+      "learning_rate": 6.592194772645901e-06,
+      "loss": 0.0001,
+      "step": 6316
+    },
+    {
+      "epoch": 0.40710188825159505,
+      "grad_norm": 0.0014138136504818558,
+      "learning_rate": 6.591478696741855e-06,
+      "loss": 0.0,
+      "step": 6317
+    },
+    {
+      "epoch": 0.4071663336985242,
+      "grad_norm": 0.020313332069530073,
+      "learning_rate": 6.59076262083781e-06,
+      "loss": 0.0,
+      "step": 6318
+    },
+    {
+      "epoch": 0.40723077914545336,
+      "grad_norm": 0.0225892251709381,
+      "learning_rate": 6.590046544933764e-06,
+      "loss": 0.0,
+      "step": 6319
+    },
+    {
+      "epoch": 0.40729522459238254,
+      "grad_norm": 0.06057017977650917,
+      "learning_rate": 6.589330469029717e-06,
+      "loss": 0.0002,
+      "step": 6320
+    },
+    {
+      "epoch": 0.4073596700393117,
+      "grad_norm": 0.005657030130166821,
+      "learning_rate": 6.588614393125672e-06,
+      "loss": 0.0,
+      "step": 6321
+    },
+    {
+      "epoch": 0.4074241154862409,
+      "grad_norm": 0.27193709574173774,
+      "learning_rate": 6.587898317221626e-06,
+      "loss": 0.0003,
+      "step": 6322
+    },
+    {
+      "epoch": 0.4074885609331701,
+      "grad_norm": 0.005401159223127921,
+      "learning_rate": 6.58718224131758e-06,
+      "loss": 0.0,
+      "step": 6323
+    },
+    {
+      "epoch": 0.4075530063800992,
+      "grad_norm": 0.190735859372399,
+      "learning_rate": 6.5864661654135345e-06,
+      "loss": 0.0008,
+      "step": 6324
+    },
+    {
+      "epoch": 0.4076174518270284,
+      "grad_norm": 0.02942071707522714,
+      "learning_rate": 6.585750089509488e-06,
+      "loss": 0.0,
+      "step": 6325
+    },
+    {
+      "epoch": 0.4076818972739576,
+      "grad_norm": 0.04759147327995806,
+      "learning_rate": 6.585034013605442e-06,
+      "loss": 0.0002,
+      "step": 6326
+    },
+    {
+      "epoch": 0.4077463427208868,
+      "grad_norm": 0.0783865454767804,
+      "learning_rate": 6.584317937701397e-06,
+      "loss": 0.0001,
+      "step": 6327
+    },
+    {
+      "epoch": 0.40781078816781596,
+      "grad_norm": 0.009902174526617688,
+      "learning_rate": 6.583601861797351e-06,
+      "loss": 0.0015,
+      "step": 6328
+    },
+    {
+      "epoch": 0.4078752336147451,
+      "grad_norm": 0.3001118059233158,
+      "learning_rate": 6.582885785893305e-06,
+      "loss": 0.0004,
+      "step": 6329
+    },
+    {
+      "epoch": 0.4079396790616743,
+      "grad_norm": 0.023572379667072773,
+      "learning_rate": 6.582169709989259e-06,
+      "loss": 0.0001,
+      "step": 6330
+    },
+    {
+      "epoch": 0.40800412450860346,
+      "grad_norm": 0.005017211603975058,
+      "learning_rate": 6.581453634085213e-06,
+      "loss": 0.0,
+      "step": 6331
+    },
+    {
+      "epoch": 0.40806856995553265,
+      "grad_norm": 0.03636838824935784,
+      "learning_rate": 6.580737558181167e-06,
+      "loss": 0.0,
+      "step": 6332
+    },
+    {
+      "epoch": 0.40813301540246183,
+      "grad_norm": 0.1252912317573341,
+      "learning_rate": 6.580021482277122e-06,
+      "loss": 0.0001,
+      "step": 6333
+    },
+    {
+      "epoch": 0.408197460849391,
+      "grad_norm": 0.0002858440281521137,
+      "learning_rate": 6.579305406373077e-06,
+      "loss": 0.0,
+      "step": 6334
+    },
+    {
+      "epoch": 0.40826190629632014,
+      "grad_norm": 0.008682301258015307,
+      "learning_rate": 6.578589330469031e-06,
+      "loss": 0.0001,
+      "step": 6335
+    },
+    {
+      "epoch": 0.4083263517432493,
+      "grad_norm": 0.0008786334737352812,
+      "learning_rate": 6.5778732545649844e-06,
+      "loss": 0.0,
+      "step": 6336
+    },
+    {
+      "epoch": 0.4083907971901785,
+      "grad_norm": 0.054658848205107594,
+      "learning_rate": 6.577157178660939e-06,
+      "loss": 0.0001,
+      "step": 6337
+    },
+    {
+      "epoch": 0.4084552426371077,
+      "grad_norm": 0.009756242819836084,
+      "learning_rate": 6.576441102756893e-06,
+      "loss": 0.0,
+      "step": 6338
+    },
+    {
+      "epoch": 0.4085196880840369,
+      "grad_norm": 0.24148468360646674,
+      "learning_rate": 6.575725026852847e-06,
+      "loss": 0.0006,
+      "step": 6339
+    },
+    {
+      "epoch": 0.40858413353096606,
+      "grad_norm": 0.007297027798895586,
+      "learning_rate": 6.575008950948802e-06,
+      "loss": 0.0,
+      "step": 6340
+    },
+    {
+      "epoch": 0.4086485789778952,
+      "grad_norm": 0.003341223520785786,
+      "learning_rate": 6.574292875044755e-06,
+      "loss": 0.0,
+      "step": 6341
+    },
+    {
+      "epoch": 0.4087130244248244,
+      "grad_norm": 0.0011016815095849727,
+      "learning_rate": 6.573576799140709e-06,
+      "loss": 0.0,
+      "step": 6342
+    },
+    {
+      "epoch": 0.40877746987175356,
+      "grad_norm": 0.027504932317853434,
+      "learning_rate": 6.572860723236664e-06,
+      "loss": 0.0001,
+      "step": 6343
+    },
+    {
+      "epoch": 0.40884191531868275,
+      "grad_norm": 0.01037361559934303,
+      "learning_rate": 6.572144647332618e-06,
+      "loss": 0.0,
+      "step": 6344
+    },
+    {
+      "epoch": 0.40890636076561193,
+      "grad_norm": 0.2539894847889759,
+      "learning_rate": 6.571428571428572e-06,
+      "loss": 0.0018,
+      "step": 6345
+    },
+    {
+      "epoch": 0.40897080621254106,
+      "grad_norm": 0.01065426405305903,
+      "learning_rate": 6.570712495524526e-06,
+      "loss": 0.0001,
+      "step": 6346
+    },
+    {
+      "epoch": 0.40903525165947024,
+      "grad_norm": 0.0003099955461148386,
+      "learning_rate": 6.56999641962048e-06,
+      "loss": 0.0,
+      "step": 6347
+    },
+    {
+      "epoch": 0.40909969710639943,
+      "grad_norm": 0.0006434004654581663,
+      "learning_rate": 6.569280343716434e-06,
+      "loss": 0.0,
+      "step": 6348
+    },
+    {
+      "epoch": 0.4091641425533286,
+      "grad_norm": 0.004250379747714681,
+      "learning_rate": 6.568564267812389e-06,
+      "loss": 0.0,
+      "step": 6349
+    },
+    {
+      "epoch": 0.4092285880002578,
+      "grad_norm": 0.0016217436448476318,
+      "learning_rate": 6.567848191908343e-06,
+      "loss": 0.0,
+      "step": 6350
+    },
+    {
+      "epoch": 0.409293033447187,
+      "grad_norm": 0.10920754880728271,
+      "learning_rate": 6.567132116004296e-06,
+      "loss": 0.0014,
+      "step": 6351
+    },
+    {
+      "epoch": 0.4093574788941161,
+      "grad_norm": 0.0021528177987047534,
+      "learning_rate": 6.566416040100251e-06,
+      "loss": 0.0,
+      "step": 6352
+    },
+    {
+      "epoch": 0.4094219243410453,
+      "grad_norm": 0.002116073833938389,
+      "learning_rate": 6.565699964196205e-06,
+      "loss": 0.0,
+      "step": 6353
+    },
+    {
+      "epoch": 0.4094863697879745,
+      "grad_norm": 0.29838004832477744,
+      "learning_rate": 6.564983888292159e-06,
+      "loss": 0.0012,
+      "step": 6354
+    },
+    {
+      "epoch": 0.40955081523490366,
+      "grad_norm": 0.001972656939178585,
+      "learning_rate": 6.5642678123881136e-06,
+      "loss": 0.0,
+      "step": 6355
+    },
+    {
+      "epoch": 0.40961526068183285,
+      "grad_norm": 0.051000456672673784,
+      "learning_rate": 6.563551736484067e-06,
+      "loss": 0.0004,
+      "step": 6356
+    },
+    {
+      "epoch": 0.409679706128762,
+      "grad_norm": 0.002971262747549498,
+      "learning_rate": 6.562835660580022e-06,
+      "loss": 0.0,
+      "step": 6357
+    },
+    {
+      "epoch": 0.40974415157569116,
+      "grad_norm": 0.001269607342146986,
+      "learning_rate": 6.5621195846759765e-06,
+      "loss": 0.0,
+      "step": 6358
+    },
+    {
+      "epoch": 0.40980859702262035,
+      "grad_norm": 0.0013160284007964068,
+      "learning_rate": 6.561403508771931e-06,
+      "loss": 0.0,
+      "step": 6359
+    },
+    {
+      "epoch": 0.40987304246954953,
+      "grad_norm": 0.003960992563566679,
+      "learning_rate": 6.560687432867885e-06,
+      "loss": 0.0,
+      "step": 6360
+    },
+    {
+      "epoch": 0.4099374879164787,
+      "grad_norm": 0.000711062058098579,
+      "learning_rate": 6.559971356963839e-06,
+      "loss": 0.0,
+      "step": 6361
+    },
+    {
+      "epoch": 0.4100019333634079,
+      "grad_norm": 0.0006166932865461904,
+      "learning_rate": 6.559255281059793e-06,
+      "loss": 0.0,
+      "step": 6362
+    },
+    {
+      "epoch": 0.41006637881033703,
+      "grad_norm": 0.0008622193241519165,
+      "learning_rate": 6.558539205155747e-06,
+      "loss": 0.0,
+      "step": 6363
+    },
+    {
+      "epoch": 0.4101308242572662,
+      "grad_norm": 0.022749650694368222,
+      "learning_rate": 6.557823129251701e-06,
+      "loss": 0.0,
+      "step": 6364
+    },
+    {
+      "epoch": 0.4101952697041954,
+      "grad_norm": 0.0001210514390543481,
+      "learning_rate": 6.557107053347656e-06,
+      "loss": 0.0,
+      "step": 6365
+    },
+    {
+      "epoch": 0.4102597151511246,
+      "grad_norm": 0.30534664406291645,
+      "learning_rate": 6.55639097744361e-06,
+      "loss": 0.0007,
+      "step": 6366
+    },
+    {
+      "epoch": 0.41032416059805377,
+      "grad_norm": 0.0009480997115783098,
+      "learning_rate": 6.5556749015395635e-06,
+      "loss": 0.0,
+      "step": 6367
+    },
+    {
+      "epoch": 0.4103886060449829,
+      "grad_norm": 0.0030546703447422857,
+      "learning_rate": 6.554958825635518e-06,
+      "loss": 0.0,
+      "step": 6368
+    },
+    {
+      "epoch": 0.4104530514919121,
+      "grad_norm": 3.549411554795448e-05,
+      "learning_rate": 6.554242749731472e-06,
+      "loss": 0.0,
+      "step": 6369
+    },
+    {
+      "epoch": 0.41051749693884126,
+      "grad_norm": 0.06431834983958828,
+      "learning_rate": 6.553526673827426e-06,
+      "loss": 0.0006,
+      "step": 6370
+    },
+    {
+      "epoch": 0.41058194238577045,
+      "grad_norm": 0.0015877173040344272,
+      "learning_rate": 6.552810597923381e-06,
+      "loss": 0.0,
+      "step": 6371
+    },
+    {
+      "epoch": 0.41064638783269963,
+      "grad_norm": 0.3677186573657693,
+      "learning_rate": 6.552094522019334e-06,
+      "loss": 0.0003,
+      "step": 6372
+    },
+    {
+      "epoch": 0.4107108332796288,
+      "grad_norm": 0.015533271144025876,
+      "learning_rate": 6.551378446115288e-06,
+      "loss": 0.0001,
+      "step": 6373
+    },
+    {
+      "epoch": 0.41077527872655795,
+      "grad_norm": 0.0001311708652832042,
+      "learning_rate": 6.550662370211243e-06,
+      "loss": 0.0,
+      "step": 6374
+    },
+    {
+      "epoch": 0.41083972417348713,
+      "grad_norm": 0.03579685409460785,
+      "learning_rate": 6.549946294307197e-06,
+      "loss": 0.0001,
+      "step": 6375
+    },
+    {
+      "epoch": 0.4109041696204163,
+      "grad_norm": 0.010680600901401511,
+      "learning_rate": 6.5492302184031505e-06,
+      "loss": 0.0,
+      "step": 6376
+    },
+    {
+      "epoch": 0.4109686150673455,
+      "grad_norm": 0.0014740273723926735,
+      "learning_rate": 6.548514142499105e-06,
+      "loss": 0.0,
+      "step": 6377
+    },
+    {
+      "epoch": 0.4110330605142747,
+      "grad_norm": 0.17344039901681835,
+      "learning_rate": 6.547798066595059e-06,
+      "loss": 0.0016,
+      "step": 6378
+    },
+    {
+      "epoch": 0.41109750596120387,
+      "grad_norm": 2.5766377467299022e-05,
+      "learning_rate": 6.547081990691013e-06,
+      "loss": 0.0,
+      "step": 6379
+    },
+    {
+      "epoch": 0.411161951408133,
+      "grad_norm": 0.4452945638691878,
+      "learning_rate": 6.5463659147869685e-06,
+      "loss": 0.0052,
+      "step": 6380
+    },
+    {
+      "epoch": 0.4112263968550622,
+      "grad_norm": 0.0008473765258398587,
+      "learning_rate": 6.545649838882923e-06,
+      "loss": 0.0,
+      "step": 6381
+    },
+    {
+      "epoch": 0.41129084230199137,
+      "grad_norm": 0.000153973536380325,
+      "learning_rate": 6.544933762978877e-06,
+      "loss": 0.0,
+      "step": 6382
+    },
+    {
+      "epoch": 0.41135528774892055,
+      "grad_norm": 0.0008808541464115172,
+      "learning_rate": 6.5442176870748305e-06,
+      "loss": 0.0,
+      "step": 6383
+    },
+    {
+      "epoch": 0.41141973319584974,
+      "grad_norm": 0.04910274289866252,
+      "learning_rate": 6.543501611170785e-06,
+      "loss": 0.0001,
+      "step": 6384
+    },
+    {
+      "epoch": 0.41148417864277886,
+      "grad_norm": 0.6920763852250857,
+      "learning_rate": 6.542785535266739e-06,
+      "loss": 0.0039,
+      "step": 6385
+    },
+    {
+      "epoch": 0.41154862408970805,
+      "grad_norm": 0.10766323025445916,
+      "learning_rate": 6.5420694593626934e-06,
+      "loss": 0.0002,
+      "step": 6386
+    },
+    {
+      "epoch": 0.41161306953663723,
+      "grad_norm": 0.0026613364721375463,
+      "learning_rate": 6.541353383458648e-06,
+      "loss": 0.0,
+      "step": 6387
+    },
+    {
+      "epoch": 0.4116775149835664,
+      "grad_norm": 0.007791129047614624,
+      "learning_rate": 6.540637307554601e-06,
+      "loss": 0.0,
+      "step": 6388
+    },
+    {
+      "epoch": 0.4117419604304956,
+      "grad_norm": 0.0023654626963615804,
+      "learning_rate": 6.5399212316505555e-06,
+      "loss": 0.0,
+      "step": 6389
+    },
+    {
+      "epoch": 0.4118064058774248,
+      "grad_norm": 0.007302414206805107,
+      "learning_rate": 6.53920515574651e-06,
+      "loss": 0.0001,
+      "step": 6390
+    },
+    {
+      "epoch": 0.4118708513243539,
+      "grad_norm": 0.045215327175843444,
+      "learning_rate": 6.538489079842464e-06,
+      "loss": 0.0,
+      "step": 6391
+    },
+    {
+      "epoch": 0.4119352967712831,
+      "grad_norm": 0.0042823292236435235,
+      "learning_rate": 6.5377730039384175e-06,
+      "loss": 0.0,
+      "step": 6392
+    },
+    {
+      "epoch": 0.4119997422182123,
+      "grad_norm": 0.009907462456016054,
+      "learning_rate": 6.537056928034372e-06,
+      "loss": 0.0,
+      "step": 6393
+    },
+    {
+      "epoch": 0.41206418766514147,
+      "grad_norm": 1.2236248238341578,
+      "learning_rate": 6.536340852130326e-06,
+      "loss": 0.0013,
+      "step": 6394
+    },
+    {
+      "epoch": 0.41212863311207065,
+      "grad_norm": 0.04217290112438516,
+      "learning_rate": 6.5356247762262804e-06,
+      "loss": 0.0002,
+      "step": 6395
+    },
+    {
+      "epoch": 0.4121930785589998,
+      "grad_norm": 0.004078594013241326,
+      "learning_rate": 6.534908700322235e-06,
+      "loss": 0.0,
+      "step": 6396
+    },
+    {
+      "epoch": 0.41225752400592897,
+      "grad_norm": 0.06957932522015041,
+      "learning_rate": 6.534192624418188e-06,
+      "loss": 0.0001,
+      "step": 6397
+    },
+    {
+      "epoch": 0.41232196945285815,
+      "grad_norm": 0.3205191480641675,
+      "learning_rate": 6.5334765485141425e-06,
+      "loss": 0.0037,
+      "step": 6398
+    },
+    {
+      "epoch": 0.41238641489978733,
+      "grad_norm": 0.010984131143033277,
+      "learning_rate": 6.532760472610097e-06,
+      "loss": 0.0001,
+      "step": 6399
+    },
+    {
+      "epoch": 0.4124508603467165,
+      "grad_norm": 0.3163248357698386,
+      "learning_rate": 6.532044396706051e-06,
+      "loss": 0.0022,
+      "step": 6400
+    },
+    {
+      "epoch": 0.4125153057936457,
+      "grad_norm": 0.0038705281878985156,
+      "learning_rate": 6.531328320802005e-06,
+      "loss": 0.0,
+      "step": 6401
+    },
+    {
+      "epoch": 0.41257975124057483,
+      "grad_norm": 0.000454026447775615,
+      "learning_rate": 6.530612244897959e-06,
+      "loss": 0.0,
+      "step": 6402
+    },
+    {
+      "epoch": 0.412644196687504,
+      "grad_norm": 0.018020890708024186,
+      "learning_rate": 6.529896168993915e-06,
+      "loss": 0.0001,
+      "step": 6403
+    },
+    {
+      "epoch": 0.4127086421344332,
+      "grad_norm": 0.0011114438458940314,
+      "learning_rate": 6.529180093089868e-06,
+      "loss": 0.0,
+      "step": 6404
+    },
+    {
+      "epoch": 0.4127730875813624,
+      "grad_norm": 1.4026249680627463,
+      "learning_rate": 6.528464017185823e-06,
+      "loss": 0.0068,
+      "step": 6405
+    },
+    {
+      "epoch": 0.41283753302829157,
+      "grad_norm": 0.001006156064450549,
+      "learning_rate": 6.527747941281777e-06,
+      "loss": 0.0,
+      "step": 6406
+    },
+    {
+      "epoch": 0.4129019784752207,
+      "grad_norm": 0.7100981903682718,
+      "learning_rate": 6.527031865377731e-06,
+      "loss": 0.0024,
+      "step": 6407
+    },
+    {
+      "epoch": 0.4129664239221499,
+      "grad_norm": 0.00165587619774657,
+      "learning_rate": 6.526315789473685e-06,
+      "loss": 0.0,
+      "step": 6408
+    },
+    {
+      "epoch": 0.41303086936907907,
+      "grad_norm": 0.27788876991628714,
+      "learning_rate": 6.525599713569639e-06,
+      "loss": 0.0012,
+      "step": 6409
+    },
+    {
+      "epoch": 0.41309531481600825,
+      "grad_norm": 0.0005789179176985119,
+      "learning_rate": 6.524883637665593e-06,
+      "loss": 0.0,
+      "step": 6410
+    },
+    {
+      "epoch": 0.41315976026293744,
+      "grad_norm": 0.005029905999029246,
+      "learning_rate": 6.5241675617615475e-06,
+      "loss": 0.0001,
+      "step": 6411
+    },
+    {
+      "epoch": 0.4132242057098666,
+      "grad_norm": 0.0034671998073162475,
+      "learning_rate": 6.523451485857502e-06,
+      "loss": 0.0,
+      "step": 6412
+    },
+    {
+      "epoch": 0.41328865115679575,
+      "grad_norm": 0.005740657103420485,
+      "learning_rate": 6.522735409953455e-06,
+      "loss": 0.0,
+      "step": 6413
+    },
+    {
+      "epoch": 0.41335309660372493,
+      "grad_norm": 0.00789786293433695,
+      "learning_rate": 6.5220193340494096e-06,
+      "loss": 0.0,
+      "step": 6414
+    },
+    {
+      "epoch": 0.4134175420506541,
+      "grad_norm": 0.04233837478909208,
+      "learning_rate": 6.521303258145364e-06,
+      "loss": 0.0,
+      "step": 6415
+    },
+    {
+      "epoch": 0.4134819874975833,
+      "grad_norm": 0.031302123632725096,
+      "learning_rate": 6.520587182241318e-06,
+      "loss": 0.0001,
+      "step": 6416
+    },
+    {
+      "epoch": 0.4135464329445125,
+      "grad_norm": 0.011742804966906406,
+      "learning_rate": 6.5198711063372725e-06,
+      "loss": 0.0,
+      "step": 6417
+    },
+    {
+      "epoch": 0.41361087839144167,
+      "grad_norm": 0.000555710875784751,
+      "learning_rate": 6.519155030433226e-06,
+      "loss": 0.0,
+      "step": 6418
+    },
+    {
+      "epoch": 0.4136753238383708,
+      "grad_norm": 0.1739312573212879,
+      "learning_rate": 6.51843895452918e-06,
+      "loss": 0.0005,
+      "step": 6419
+    },
+    {
+      "epoch": 0.4137397692853,
+      "grad_norm": 0.004575221895012938,
+      "learning_rate": 6.5177228786251345e-06,
+      "loss": 0.0,
+      "step": 6420
+    },
+    {
+      "epoch": 0.41380421473222917,
+      "grad_norm": 0.00011079464582073852,
+      "learning_rate": 6.517006802721089e-06,
+      "loss": 0.0,
+      "step": 6421
+    },
+    {
+      "epoch": 0.41386866017915835,
+      "grad_norm": 0.005995697259705144,
+      "learning_rate": 6.516290726817043e-06,
+      "loss": 0.0,
+      "step": 6422
+    },
+    {
+      "epoch": 0.41393310562608754,
+      "grad_norm": 0.0034095905482489468,
+      "learning_rate": 6.5155746509129966e-06,
+      "loss": 0.0,
+      "step": 6423
+    },
+    {
+      "epoch": 0.41399755107301667,
+      "grad_norm": 0.00577780198281472,
+      "learning_rate": 6.514858575008951e-06,
+      "loss": 0.0,
+      "step": 6424
+    },
+    {
+      "epoch": 0.41406199651994585,
+      "grad_norm": 0.46146711537749807,
+      "learning_rate": 6.514142499104905e-06,
+      "loss": 0.0012,
+      "step": 6425
+    },
+    {
+      "epoch": 0.41412644196687504,
+      "grad_norm": 0.0009439974016115886,
+      "learning_rate": 6.5134264232008595e-06,
+      "loss": 0.0,
+      "step": 6426
+    },
+    {
+      "epoch": 0.4141908874138042,
+      "grad_norm": 0.03680339894602934,
+      "learning_rate": 6.512710347296815e-06,
+      "loss": 0.0,
+      "step": 6427
+    },
+    {
+      "epoch": 0.4142553328607334,
+      "grad_norm": 0.0009701746796509103,
+      "learning_rate": 6.511994271392769e-06,
+      "loss": 0.0,
+      "step": 6428
+    },
+    {
+      "epoch": 0.4143197783076626,
+      "grad_norm": 0.0010483196330762464,
+      "learning_rate": 6.511278195488722e-06,
+      "loss": 0.0,
+      "step": 6429
+    },
+    {
+      "epoch": 0.4143842237545917,
+      "grad_norm": 0.0006480988198016862,
+      "learning_rate": 6.510562119584677e-06,
+      "loss": 0.0,
+      "step": 6430
+    },
+    {
+      "epoch": 0.4144486692015209,
+      "grad_norm": 0.07760918026875319,
+      "learning_rate": 6.509846043680631e-06,
+      "loss": 0.0002,
+      "step": 6431
+    },
+    {
+      "epoch": 0.4145131146484501,
+      "grad_norm": 0.12619666898186474,
+      "learning_rate": 6.509129967776585e-06,
+      "loss": 0.0003,
+      "step": 6432
+    },
+    {
+      "epoch": 0.41457756009537927,
+      "grad_norm": 0.06872515229381396,
+      "learning_rate": 6.5084138918725396e-06,
+      "loss": 0.0001,
+      "step": 6433
+    },
+    {
+      "epoch": 0.41464200554230846,
+      "grad_norm": 0.005405009744319238,
+      "learning_rate": 6.507697815968493e-06,
+      "loss": 0.0,
+      "step": 6434
+    },
+    {
+      "epoch": 0.4147064509892376,
+      "grad_norm": 0.0006958994388959971,
+      "learning_rate": 6.506981740064447e-06,
+      "loss": 0.0,
+      "step": 6435
+    },
+    {
+      "epoch": 0.41477089643616677,
+      "grad_norm": 0.00012114286637400527,
+      "learning_rate": 6.506265664160402e-06,
+      "loss": 0.0,
+      "step": 6436
+    },
+    {
+      "epoch": 0.41483534188309595,
+      "grad_norm": 0.016547558757412303,
+      "learning_rate": 6.505549588256356e-06,
+      "loss": 0.0,
+      "step": 6437
+    },
+    {
+      "epoch": 0.41489978733002514,
+      "grad_norm": 0.013458047052151412,
+      "learning_rate": 6.50483351235231e-06,
+      "loss": 0.0,
+      "step": 6438
+    },
+    {
+      "epoch": 0.4149642327769543,
+      "grad_norm": 0.0021026141925494257,
+      "learning_rate": 6.504117436448264e-06,
+      "loss": 0.0,
+      "step": 6439
+    },
+    {
+      "epoch": 0.4150286782238835,
+      "grad_norm": 0.00038938549783562923,
+      "learning_rate": 6.503401360544218e-06,
+      "loss": 0.0,
+      "step": 6440
+    },
+    {
+      "epoch": 0.41509312367081264,
+      "grad_norm": 0.0020553606478648863,
+      "learning_rate": 6.502685284640172e-06,
+      "loss": 0.0,
+      "step": 6441
+    },
+    {
+      "epoch": 0.4151575691177418,
+      "grad_norm": 0.0011236673450610109,
+      "learning_rate": 6.5019692087361265e-06,
+      "loss": 0.0,
+      "step": 6442
+    },
+    {
+      "epoch": 0.415222014564671,
+      "grad_norm": 0.0008312099867464999,
+      "learning_rate": 6.501253132832081e-06,
+      "loss": 0.0,
+      "step": 6443
+    },
+    {
+      "epoch": 0.4152864600116002,
+      "grad_norm": 0.016113894443693048,
+      "learning_rate": 6.500537056928034e-06,
+      "loss": 0.0,
+      "step": 6444
+    },
+    {
+      "epoch": 0.4153509054585294,
+      "grad_norm": 0.001129702907296028,
+      "learning_rate": 6.499820981023989e-06,
+      "loss": 0.0,
+      "step": 6445
+    },
+    {
+      "epoch": 0.4154153509054585,
+      "grad_norm": 0.0036121507193785907,
+      "learning_rate": 6.499104905119943e-06,
+      "loss": 0.0,
+      "step": 6446
+    },
+    {
+      "epoch": 0.4154797963523877,
+      "grad_norm": 0.4637558492255657,
+      "learning_rate": 6.498388829215897e-06,
+      "loss": 0.0018,
+      "step": 6447
+    },
+    {
+      "epoch": 0.41554424179931687,
+      "grad_norm": 0.002091072027737738,
+      "learning_rate": 6.4976727533118515e-06,
+      "loss": 0.0,
+      "step": 6448
+    },
+    {
+      "epoch": 0.41560868724624606,
+      "grad_norm": 0.005425796910282488,
+      "learning_rate": 6.496956677407805e-06,
+      "loss": 0.0,
+      "step": 6449
+    },
+    {
+      "epoch": 0.41567313269317524,
+      "grad_norm": 0.0006827120071557798,
+      "learning_rate": 6.49624060150376e-06,
+      "loss": 0.0,
+      "step": 6450
+    },
+    {
+      "epoch": 0.4157375781401044,
+      "grad_norm": 0.0009370663355223362,
+      "learning_rate": 6.495524525599714e-06,
+      "loss": 0.0,
+      "step": 6451
+    },
+    {
+      "epoch": 0.41580202358703355,
+      "grad_norm": 0.005268021478741142,
+      "learning_rate": 6.494808449695669e-06,
+      "loss": 0.0,
+      "step": 6452
+    },
+    {
+      "epoch": 0.41586646903396274,
+      "grad_norm": 0.0018895030845587368,
+      "learning_rate": 6.494092373791623e-06,
+      "loss": 0.0,
+      "step": 6453
+    },
+    {
+      "epoch": 0.4159309144808919,
+      "grad_norm": 0.005175601673834675,
+      "learning_rate": 6.493376297887577e-06,
+      "loss": 0.0,
+      "step": 6454
+    },
+    {
+      "epoch": 0.4159953599278211,
+      "grad_norm": 0.4407284520497393,
+      "learning_rate": 6.492660221983531e-06,
+      "loss": 0.0032,
+      "step": 6455
+    },
+    {
+      "epoch": 0.4160598053747503,
+      "grad_norm": 0.0007340834875259358,
+      "learning_rate": 6.491944146079485e-06,
+      "loss": 0.0,
+      "step": 6456
+    },
+    {
+      "epoch": 0.4161242508216795,
+      "grad_norm": 0.010261719996274735,
+      "learning_rate": 6.491228070175439e-06,
+      "loss": 0.0,
+      "step": 6457
+    },
+    {
+      "epoch": 0.4161886962686086,
+      "grad_norm": 0.0008106852255236457,
+      "learning_rate": 6.490511994271394e-06,
+      "loss": 0.0,
+      "step": 6458
+    },
+    {
+      "epoch": 0.4162531417155378,
+      "grad_norm": 0.010090045530888448,
+      "learning_rate": 6.489795918367348e-06,
+      "loss": 0.0,
+      "step": 6459
+    },
+    {
+      "epoch": 0.416317587162467,
+      "grad_norm": 0.008465445410221185,
+      "learning_rate": 6.489079842463301e-06,
+      "loss": 0.0,
+      "step": 6460
+    },
+    {
+      "epoch": 0.41638203260939616,
+      "grad_norm": 0.02899654347956042,
+      "learning_rate": 6.488363766559256e-06,
+      "loss": 0.0001,
+      "step": 6461
+    },
+    {
+      "epoch": 0.41644647805632534,
+      "grad_norm": 0.001056778135030049,
+      "learning_rate": 6.48764769065521e-06,
+      "loss": 0.0,
+      "step": 6462
+    },
+    {
+      "epoch": 0.41651092350325447,
+      "grad_norm": 0.3688163746079339,
+      "learning_rate": 6.486931614751164e-06,
+      "loss": 0.0006,
+      "step": 6463
+    },
+    {
+      "epoch": 0.41657536895018366,
+      "grad_norm": 0.005144527660957336,
+      "learning_rate": 6.486215538847119e-06,
+      "loss": 0.0,
+      "step": 6464
+    },
+    {
+      "epoch": 0.41663981439711284,
+      "grad_norm": 0.017784187385494993,
+      "learning_rate": 6.485499462943072e-06,
+      "loss": 0.0002,
+      "step": 6465
+    },
+    {
+      "epoch": 0.416704259844042,
+      "grad_norm": 0.06417901822958386,
+      "learning_rate": 6.484783387039026e-06,
+      "loss": 0.0001,
+      "step": 6466
+    },
+    {
+      "epoch": 0.4167687052909712,
+      "grad_norm": 0.0006121339328757751,
+      "learning_rate": 6.484067311134981e-06,
+      "loss": 0.0,
+      "step": 6467
+    },
+    {
+      "epoch": 0.4168331507379004,
+      "grad_norm": 0.18229139305275863,
+      "learning_rate": 6.483351235230935e-06,
+      "loss": 0.0004,
+      "step": 6468
+    },
+    {
+      "epoch": 0.4168975961848295,
+      "grad_norm": 0.041040044281895095,
+      "learning_rate": 6.482635159326889e-06,
+      "loss": 0.0001,
+      "step": 6469
+    },
+    {
+      "epoch": 0.4169620416317587,
+      "grad_norm": 0.021999025416732904,
+      "learning_rate": 6.481919083422843e-06,
+      "loss": 0.0001,
+      "step": 6470
+    },
+    {
+      "epoch": 0.4170264870786879,
+      "grad_norm": 0.002544240789084966,
+      "learning_rate": 6.481203007518797e-06,
+      "loss": 0.0,
+      "step": 6471
+    },
+    {
+      "epoch": 0.4170909325256171,
+      "grad_norm": 0.0026852858379955535,
+      "learning_rate": 6.480486931614751e-06,
+      "loss": 0.0,
+      "step": 6472
+    },
+    {
+      "epoch": 0.41715537797254626,
+      "grad_norm": 0.43469944439074326,
+      "learning_rate": 6.479770855710706e-06,
+      "loss": 0.0012,
+      "step": 6473
+    },
+    {
+      "epoch": 0.4172198234194754,
+      "grad_norm": 0.347051651967674,
+      "learning_rate": 6.479054779806661e-06,
+      "loss": 0.0006,
+      "step": 6474
+    },
+    {
+      "epoch": 0.4172842688664046,
+      "grad_norm": 0.25698550840535966,
+      "learning_rate": 6.478338703902615e-06,
+      "loss": 0.0005,
+      "step": 6475
+    },
+    {
+      "epoch": 0.41734871431333376,
+      "grad_norm": 0.2126062926753528,
+      "learning_rate": 6.4776226279985685e-06,
+      "loss": 0.0019,
+      "step": 6476
+    },
+    {
+      "epoch": 0.41741315976026294,
+      "grad_norm": 0.28151246910276284,
+      "learning_rate": 6.476906552094523e-06,
+      "loss": 0.0012,
+      "step": 6477
+    },
+    {
+      "epoch": 0.4174776052071921,
+      "grad_norm": 0.0017983784302793271,
+      "learning_rate": 6.476190476190477e-06,
+      "loss": 0.0,
+      "step": 6478
+    },
+    {
+      "epoch": 0.4175420506541213,
+      "grad_norm": 0.029167512556955478,
+      "learning_rate": 6.475474400286431e-06,
+      "loss": 0.0002,
+      "step": 6479
+    },
+    {
+      "epoch": 0.41760649610105044,
+      "grad_norm": 6.0273887407271015e-05,
+      "learning_rate": 6.474758324382386e-06,
+      "loss": 0.0,
+      "step": 6480
+    },
+    {
+      "epoch": 0.4176709415479796,
+      "grad_norm": 0.003980392115880526,
+      "learning_rate": 6.474042248478339e-06,
+      "loss": 0.0,
+      "step": 6481
+    },
+    {
+      "epoch": 0.4177353869949088,
+      "grad_norm": 0.0011862001687265543,
+      "learning_rate": 6.473326172574293e-06,
+      "loss": 0.0,
+      "step": 6482
+    },
+    {
+      "epoch": 0.417799832441838,
+      "grad_norm": 0.28784543359479664,
+      "learning_rate": 6.472610096670248e-06,
+      "loss": 0.0004,
+      "step": 6483
+    },
+    {
+      "epoch": 0.4178642778887672,
+      "grad_norm": 0.002859249041893461,
+      "learning_rate": 6.471894020766202e-06,
+      "loss": 0.0,
+      "step": 6484
+    },
+    {
+      "epoch": 0.4179287233356963,
+      "grad_norm": 0.0022861943502021026,
+      "learning_rate": 6.471177944862156e-06,
+      "loss": 0.0,
+      "step": 6485
+    },
+    {
+      "epoch": 0.4179931687826255,
+      "grad_norm": 0.0020342299031710404,
+      "learning_rate": 6.47046186895811e-06,
+      "loss": 0.0,
+      "step": 6486
+    },
+    {
+      "epoch": 0.4180576142295547,
+      "grad_norm": 8.07128737944767e-05,
+      "learning_rate": 6.469745793054064e-06,
+      "loss": 0.0,
+      "step": 6487
+    },
+    {
+      "epoch": 0.41812205967648386,
+      "grad_norm": 0.1435686221463562,
+      "learning_rate": 6.469029717150018e-06,
+      "loss": 0.0017,
+      "step": 6488
+    },
+    {
+      "epoch": 0.41818650512341304,
+      "grad_norm": 9.83520528507581e-05,
+      "learning_rate": 6.468313641245973e-06,
+      "loss": 0.0,
+      "step": 6489
+    },
+    {
+      "epoch": 0.41825095057034223,
+      "grad_norm": 5.192385256500709e-05,
+      "learning_rate": 6.467597565341926e-06,
+      "loss": 0.0,
+      "step": 6490
+    },
+    {
+      "epoch": 0.41831539601727136,
+      "grad_norm": 0.005567069063858515,
+      "learning_rate": 6.46688148943788e-06,
+      "loss": 0.0,
+      "step": 6491
+    },
+    {
+      "epoch": 0.41837984146420054,
+      "grad_norm": 0.00021615406446108586,
+      "learning_rate": 6.466165413533835e-06,
+      "loss": 0.0,
+      "step": 6492
+    },
+    {
+      "epoch": 0.4184442869111297,
+      "grad_norm": 0.0021729118234671005,
+      "learning_rate": 6.465449337629789e-06,
+      "loss": 0.0,
+      "step": 6493
+    },
+    {
+      "epoch": 0.4185087323580589,
+      "grad_norm": 0.006202733531252822,
+      "learning_rate": 6.464733261725743e-06,
+      "loss": 0.0,
+      "step": 6494
+    },
+    {
+      "epoch": 0.4185731778049881,
+      "grad_norm": 0.0008673489649001512,
+      "learning_rate": 6.464017185821697e-06,
+      "loss": 0.0,
+      "step": 6495
+    },
+    {
+      "epoch": 0.4186376232519173,
+      "grad_norm": 0.01283363549355941,
+      "learning_rate": 6.463301109917651e-06,
+      "loss": 0.0,
+      "step": 6496
+    },
+    {
+      "epoch": 0.4187020686988464,
+      "grad_norm": 0.0971751521920101,
+      "learning_rate": 6.462585034013606e-06,
+      "loss": 0.0001,
+      "step": 6497
+    },
+    {
+      "epoch": 0.4187665141457756,
+      "grad_norm": 0.037483021995562936,
+      "learning_rate": 6.4618689581095605e-06,
+      "loss": 0.0002,
+      "step": 6498
+    },
+    {
+      "epoch": 0.4188309595927048,
+      "grad_norm": 0.002619854872960852,
+      "learning_rate": 6.461152882205515e-06,
+      "loss": 0.0,
+      "step": 6499
+    },
+    {
+      "epoch": 0.41889540503963396,
+      "grad_norm": 0.4253153570613154,
+      "learning_rate": 6.460436806301469e-06,
+      "loss": 0.0033,
+      "step": 6500
+    },
+    {
+      "epoch": 0.41895985048656315,
+      "grad_norm": 0.30405405417039943,
+      "learning_rate": 6.459720730397423e-06,
+      "loss": 0.0014,
+      "step": 6501
+    },
+    {
+      "epoch": 0.4190242959334923,
+      "grad_norm": 0.061062530101131295,
+      "learning_rate": 6.459004654493377e-06,
+      "loss": 0.0,
+      "step": 6502
+    },
+    {
+      "epoch": 0.41908874138042146,
+      "grad_norm": 0.0022728235938479116,
+      "learning_rate": 6.458288578589331e-06,
+      "loss": 0.0,
+      "step": 6503
+    },
+    {
+      "epoch": 0.41915318682735064,
+      "grad_norm": 0.018741473727967656,
+      "learning_rate": 6.4575725026852854e-06,
+      "loss": 0.0015,
+      "step": 6504
+    },
+    {
+      "epoch": 0.41921763227427983,
+      "grad_norm": 0.00014293611391536147,
+      "learning_rate": 6.45685642678124e-06,
+      "loss": 0.0,
+      "step": 6505
+    },
+    {
+      "epoch": 0.419282077721209,
+      "grad_norm": 0.1856076392009028,
+      "learning_rate": 6.456140350877193e-06,
+      "loss": 0.0007,
+      "step": 6506
+    },
+    {
+      "epoch": 0.4193465231681382,
+      "grad_norm": 0.0038470830664775165,
+      "learning_rate": 6.4554242749731475e-06,
+      "loss": 0.0,
+      "step": 6507
+    },
+    {
+      "epoch": 0.4194109686150673,
+      "grad_norm": 0.0002562884533311774,
+      "learning_rate": 6.454708199069102e-06,
+      "loss": 0.0,
+      "step": 6508
+    },
+    {
+      "epoch": 0.4194754140619965,
+      "grad_norm": 0.1724127247382639,
+      "learning_rate": 6.453992123165056e-06,
+      "loss": 0.0002,
+      "step": 6509
+    },
+    {
+      "epoch": 0.4195398595089257,
+      "grad_norm": 0.00018884989328603747,
+      "learning_rate": 6.45327604726101e-06,
+      "loss": 0.0,
+      "step": 6510
+    },
+    {
+      "epoch": 0.4196043049558549,
+      "grad_norm": 0.005897050359999245,
+      "learning_rate": 6.452559971356964e-06,
+      "loss": 0.0,
+      "step": 6511
+    },
+    {
+      "epoch": 0.41966875040278406,
+      "grad_norm": 0.08656624785462537,
+      "learning_rate": 6.451843895452918e-06,
+      "loss": 0.001,
+      "step": 6512
+    },
+    {
+      "epoch": 0.4197331958497132,
+      "grad_norm": 0.0008088807705909768,
+      "learning_rate": 6.4511278195488724e-06,
+      "loss": 0.0,
+      "step": 6513
+    },
+    {
+      "epoch": 0.4197976412966424,
+      "grad_norm": 0.04547051038808133,
+      "learning_rate": 6.450411743644827e-06,
+      "loss": 0.0,
+      "step": 6514
+    },
+    {
+      "epoch": 0.41986208674357156,
+      "grad_norm": 6.935269685480056e-05,
+      "learning_rate": 6.449695667740781e-06,
+      "loss": 0.0,
+      "step": 6515
+    },
+    {
+      "epoch": 0.41992653219050075,
+      "grad_norm": 0.013221494184816038,
+      "learning_rate": 6.4489795918367345e-06,
+      "loss": 0.0,
+      "step": 6516
+    },
+    {
+      "epoch": 0.41999097763742993,
+      "grad_norm": 0.004337338260439512,
+      "learning_rate": 6.448263515932689e-06,
+      "loss": 0.0,
+      "step": 6517
+    },
+    {
+      "epoch": 0.4200554230843591,
+      "grad_norm": 0.0007477333484548366,
+      "learning_rate": 6.447547440028643e-06,
+      "loss": 0.0,
+      "step": 6518
+    },
+    {
+      "epoch": 0.42011986853128824,
+      "grad_norm": 0.22266532645213608,
+      "learning_rate": 6.446831364124597e-06,
+      "loss": 0.0007,
+      "step": 6519
+    },
+    {
+      "epoch": 0.4201843139782174,
+      "grad_norm": 0.001061479785357458,
+      "learning_rate": 6.4461152882205525e-06,
+      "loss": 0.0,
+      "step": 6520
+    },
+    {
+      "epoch": 0.4202487594251466,
+      "grad_norm": 0.01430381819889332,
+      "learning_rate": 6.445399212316507e-06,
+      "loss": 0.0001,
+      "step": 6521
+    },
+    {
+      "epoch": 0.4203132048720758,
+      "grad_norm": 0.2277960656102833,
+      "learning_rate": 6.44468313641246e-06,
+      "loss": 0.0082,
+      "step": 6522
+    },
+    {
+      "epoch": 0.420377650319005,
+      "grad_norm": 0.009277152777889666,
+      "learning_rate": 6.443967060508415e-06,
+      "loss": 0.0001,
+      "step": 6523
+    },
+    {
+      "epoch": 0.4204420957659341,
+      "grad_norm": 0.014832602078491002,
+      "learning_rate": 6.443250984604369e-06,
+      "loss": 0.0,
+      "step": 6524
+    },
+    {
+      "epoch": 0.4205065412128633,
+      "grad_norm": 0.000828742706597286,
+      "learning_rate": 6.442534908700323e-06,
+      "loss": 0.0,
+      "step": 6525
+    },
+    {
+      "epoch": 0.4205709866597925,
+      "grad_norm": 0.00023983841067913176,
+      "learning_rate": 6.4418188327962775e-06,
+      "loss": 0.0,
+      "step": 6526
+    },
+    {
+      "epoch": 0.42063543210672166,
+      "grad_norm": 0.0060839587052440545,
+      "learning_rate": 6.441102756892231e-06,
+      "loss": 0.0001,
+      "step": 6527
+    },
+    {
+      "epoch": 0.42069987755365085,
+      "grad_norm": 0.12049660035438453,
+      "learning_rate": 6.440386680988185e-06,
+      "loss": 0.0004,
+      "step": 6528
+    },
+    {
+      "epoch": 0.42076432300058003,
+      "grad_norm": 0.0002864290770283571,
+      "learning_rate": 6.4396706050841395e-06,
+      "loss": 0.0,
+      "step": 6529
+    },
+    {
+      "epoch": 0.42082876844750916,
+      "grad_norm": 0.008323643209694061,
+      "learning_rate": 6.438954529180094e-06,
+      "loss": 0.0,
+      "step": 6530
+    },
+    {
+      "epoch": 0.42089321389443834,
+      "grad_norm": 0.0008086087638468122,
+      "learning_rate": 6.438238453276048e-06,
+      "loss": 0.0,
+      "step": 6531
+    },
+    {
+      "epoch": 0.42095765934136753,
+      "grad_norm": 0.005185411844307165,
+      "learning_rate": 6.4375223773720016e-06,
+      "loss": 0.0,
+      "step": 6532
+    },
+    {
+      "epoch": 0.4210221047882967,
+      "grad_norm": 0.0006719877031553268,
+      "learning_rate": 6.436806301467956e-06,
+      "loss": 0.0,
+      "step": 6533
+    },
+    {
+      "epoch": 0.4210865502352259,
+      "grad_norm": 2.6934940144879813e-05,
+      "learning_rate": 6.43609022556391e-06,
+      "loss": 0.0,
+      "step": 6534
+    },
+    {
+      "epoch": 0.4211509956821551,
+      "grad_norm": 0.00015571736547918388,
+      "learning_rate": 6.4353741496598645e-06,
+      "loss": 0.0,
+      "step": 6535
+    },
+    {
+      "epoch": 0.4212154411290842,
+      "grad_norm": 0.00315484098134091,
+      "learning_rate": 6.434658073755819e-06,
+      "loss": 0.0,
+      "step": 6536
+    },
+    {
+      "epoch": 0.4212798865760134,
+      "grad_norm": 0.005351614273942891,
+      "learning_rate": 6.433941997851772e-06,
+      "loss": 0.0001,
+      "step": 6537
+    },
+    {
+      "epoch": 0.4213443320229426,
+      "grad_norm": 0.010094460236600215,
+      "learning_rate": 6.4332259219477265e-06,
+      "loss": 0.0,
+      "step": 6538
+    },
+    {
+      "epoch": 0.42140877746987176,
+      "grad_norm": 0.010835629956553901,
+      "learning_rate": 6.432509846043681e-06,
+      "loss": 0.0002,
+      "step": 6539
+    },
+    {
+      "epoch": 0.42147322291680095,
+      "grad_norm": 0.00319520008482836,
+      "learning_rate": 6.431793770139635e-06,
+      "loss": 0.0,
+      "step": 6540
+    },
+    {
+      "epoch": 0.4215376683637301,
+      "grad_norm": 0.010853856846134343,
+      "learning_rate": 6.431077694235589e-06,
+      "loss": 0.0,
+      "step": 6541
+    },
+    {
+      "epoch": 0.42160211381065926,
+      "grad_norm": 0.0114451280760472,
+      "learning_rate": 6.430361618331543e-06,
+      "loss": 0.0,
+      "step": 6542
+    },
+    {
+      "epoch": 0.42166655925758845,
+      "grad_norm": 0.02084433664255218,
+      "learning_rate": 6.429645542427498e-06,
+      "loss": 0.0001,
+      "step": 6543
+    },
+    {
+      "epoch": 0.42173100470451763,
+      "grad_norm": 0.0003189129349114517,
+      "learning_rate": 6.428929466523452e-06,
+      "loss": 0.0,
+      "step": 6544
+    },
+    {
+      "epoch": 0.4217954501514468,
+      "grad_norm": 0.0035933766135850926,
+      "learning_rate": 6.428213390619407e-06,
+      "loss": 0.0,
+      "step": 6545
+    },
+    {
+      "epoch": 0.421859895598376,
+      "grad_norm": 0.024217145235977393,
+      "learning_rate": 6.427497314715361e-06,
+      "loss": 0.0001,
+      "step": 6546
+    },
+    {
+      "epoch": 0.42192434104530513,
+      "grad_norm": 0.007078477096816555,
+      "learning_rate": 6.426781238811315e-06,
+      "loss": 0.0,
+      "step": 6547
+    },
+    {
+      "epoch": 0.4219887864922343,
+      "grad_norm": 0.16918780173284304,
+      "learning_rate": 6.426065162907269e-06,
+      "loss": 0.0017,
+      "step": 6548
+    },
+    {
+      "epoch": 0.4220532319391635,
+      "grad_norm": 0.00042276683359260573,
+      "learning_rate": 6.425349087003223e-06,
+      "loss": 0.0,
+      "step": 6549
+    },
+    {
+      "epoch": 0.4221176773860927,
+      "grad_norm": 0.005739171150647594,
+      "learning_rate": 6.424633011099177e-06,
+      "loss": 0.0,
+      "step": 6550
+    },
+    {
+      "epoch": 0.42218212283302187,
+      "grad_norm": 0.00012378405994403383,
+      "learning_rate": 6.4239169351951316e-06,
+      "loss": 0.0,
+      "step": 6551
+    },
+    {
+      "epoch": 0.422246568279951,
+      "grad_norm": 0.00235690386382886,
+      "learning_rate": 6.423200859291086e-06,
+      "loss": 0.0,
+      "step": 6552
+    },
+    {
+      "epoch": 0.4223110137268802,
+      "grad_norm": 7.031785182137648e-05,
+      "learning_rate": 6.422484783387039e-06,
+      "loss": 0.0,
+      "step": 6553
+    },
+    {
+      "epoch": 0.42237545917380936,
+      "grad_norm": 0.0026047197359357352,
+      "learning_rate": 6.421768707482994e-06,
+      "loss": 0.0,
+      "step": 6554
+    },
+    {
+      "epoch": 0.42243990462073855,
+      "grad_norm": 0.12465103807190193,
+      "learning_rate": 6.421052631578948e-06,
+      "loss": 0.0001,
+      "step": 6555
+    },
+    {
+      "epoch": 0.42250435006766773,
+      "grad_norm": 0.0011322028151738962,
+      "learning_rate": 6.420336555674902e-06,
+      "loss": 0.0,
+      "step": 6556
+    },
+    {
+      "epoch": 0.4225687955145969,
+      "grad_norm": 0.005403863207130577,
+      "learning_rate": 6.4196204797708565e-06,
+      "loss": 0.0001,
+      "step": 6557
+    },
+    {
+      "epoch": 0.42263324096152605,
+      "grad_norm": 0.09145143117725939,
+      "learning_rate": 6.41890440386681e-06,
+      "loss": 0.0008,
+      "step": 6558
+    },
+    {
+      "epoch": 0.42269768640845523,
+      "grad_norm": 0.0007512461604921746,
+      "learning_rate": 6.418188327962764e-06,
+      "loss": 0.0,
+      "step": 6559
+    },
+    {
+      "epoch": 0.4227621318553844,
+      "grad_norm": 0.001425860074836207,
+      "learning_rate": 6.4174722520587185e-06,
+      "loss": 0.0,
+      "step": 6560
+    },
+    {
+      "epoch": 0.4228265773023136,
+      "grad_norm": 1.1363235783018837,
+      "learning_rate": 6.416756176154673e-06,
+      "loss": 0.0049,
+      "step": 6561
+    },
+    {
+      "epoch": 0.4228910227492428,
+      "grad_norm": 0.001521368662538063,
+      "learning_rate": 6.416040100250627e-06,
+      "loss": 0.0,
+      "step": 6562
+    },
+    {
+      "epoch": 0.4229554681961719,
+      "grad_norm": 0.0006320624923546955,
+      "learning_rate": 6.415324024346581e-06,
+      "loss": 0.0,
+      "step": 6563
+    },
+    {
+      "epoch": 0.4230199136431011,
+      "grad_norm": 0.0170555938090754,
+      "learning_rate": 6.414607948442535e-06,
+      "loss": 0.0,
+      "step": 6564
+    },
+    {
+      "epoch": 0.4230843590900303,
+      "grad_norm": 0.007227821830375058,
+      "learning_rate": 6.413891872538489e-06,
+      "loss": 0.0,
+      "step": 6565
+    },
+    {
+      "epoch": 0.42314880453695947,
+      "grad_norm": 0.3460220977460297,
+      "learning_rate": 6.4131757966344435e-06,
+      "loss": 0.0011,
+      "step": 6566
+    },
+    {
+      "epoch": 0.42321324998388865,
+      "grad_norm": 0.0028084542812121535,
+      "learning_rate": 6.412459720730399e-06,
+      "loss": 0.0,
+      "step": 6567
+    },
+    {
+      "epoch": 0.42327769543081784,
+      "grad_norm": 0.049689493945287815,
+      "learning_rate": 6.411743644826353e-06,
+      "loss": 0.0002,
+      "step": 6568
+    },
+    {
+      "epoch": 0.42334214087774696,
+      "grad_norm": 0.018442280869811974,
+      "learning_rate": 6.411027568922306e-06,
+      "loss": 0.0,
+      "step": 6569
+    },
+    {
+      "epoch": 0.42340658632467615,
+      "grad_norm": 0.016176521402883174,
+      "learning_rate": 6.410311493018261e-06,
+      "loss": 0.0001,
+      "step": 6570
+    },
+    {
+      "epoch": 0.42347103177160533,
+      "grad_norm": 0.0045036594846243954,
+      "learning_rate": 6.409595417114215e-06,
+      "loss": 0.0001,
+      "step": 6571
+    },
+    {
+      "epoch": 0.4235354772185345,
+      "grad_norm": 0.000763490671469306,
+      "learning_rate": 6.408879341210169e-06,
+      "loss": 0.0,
+      "step": 6572
+    },
+    {
+      "epoch": 0.4235999226654637,
+      "grad_norm": 0.01358307692632028,
+      "learning_rate": 6.408163265306124e-06,
+      "loss": 0.0,
+      "step": 6573
+    },
+    {
+      "epoch": 0.4236643681123929,
+      "grad_norm": 0.07061323450311378,
+      "learning_rate": 6.407447189402077e-06,
+      "loss": 0.0001,
+      "step": 6574
+    },
+    {
+      "epoch": 0.423728813559322,
+      "grad_norm": 0.0009675888097842737,
+      "learning_rate": 6.406731113498031e-06,
+      "loss": 0.0,
+      "step": 6575
+    },
+    {
+      "epoch": 0.4237932590062512,
+      "grad_norm": 0.05086051961970249,
+      "learning_rate": 6.406015037593986e-06,
+      "loss": 0.0002,
+      "step": 6576
+    },
+    {
+      "epoch": 0.4238577044531804,
+      "grad_norm": 0.016505252166751583,
+      "learning_rate": 6.40529896168994e-06,
+      "loss": 0.0,
+      "step": 6577
+    },
+    {
+      "epoch": 0.42392214990010957,
+      "grad_norm": 0.0004049548514622685,
+      "learning_rate": 6.404582885785894e-06,
+      "loss": 0.0015,
+      "step": 6578
+    },
+    {
+      "epoch": 0.42398659534703875,
+      "grad_norm": 0.623974793736584,
+      "learning_rate": 6.403866809881848e-06,
+      "loss": 0.0023,
+      "step": 6579
+    },
+    {
+      "epoch": 0.4240510407939679,
+      "grad_norm": 4.7767857672534735,
+      "learning_rate": 6.403150733977802e-06,
+      "loss": 0.0369,
+      "step": 6580
+    },
+    {
+      "epoch": 0.42411548624089707,
+      "grad_norm": 0.0056384048468675904,
+      "learning_rate": 6.402434658073756e-06,
+      "loss": 0.0,
+      "step": 6581
+    },
+    {
+      "epoch": 0.42417993168782625,
+      "grad_norm": 0.041177984595075684,
+      "learning_rate": 6.401718582169711e-06,
+      "loss": 0.0,
+      "step": 6582
+    },
+    {
+      "epoch": 0.42424437713475543,
+      "grad_norm": 0.0028160901279018036,
+      "learning_rate": 6.401002506265664e-06,
+      "loss": 0.0,
+      "step": 6583
+    },
+    {
+      "epoch": 0.4243088225816846,
+      "grad_norm": 0.03627116547304098,
+      "learning_rate": 6.400286430361618e-06,
+      "loss": 0.0003,
+      "step": 6584
+    },
+    {
+      "epoch": 0.4243732680286138,
+      "grad_norm": 0.12108684335516592,
+      "learning_rate": 6.399570354457573e-06,
+      "loss": 0.0001,
+      "step": 6585
+    },
+    {
+      "epoch": 0.42443771347554293,
+      "grad_norm": 0.05032057374908024,
+      "learning_rate": 6.398854278553527e-06,
+      "loss": 0.0004,
+      "step": 6586
+    },
+    {
+      "epoch": 0.4245021589224721,
+      "grad_norm": 0.23586408913024615,
+      "learning_rate": 6.398138202649481e-06,
+      "loss": 0.0008,
+      "step": 6587
+    },
+    {
+      "epoch": 0.4245666043694013,
+      "grad_norm": 0.042724430901290876,
+      "learning_rate": 6.397422126745435e-06,
+      "loss": 0.0,
+      "step": 6588
+    },
+    {
+      "epoch": 0.4246310498163305,
+      "grad_norm": 0.007525076075820635,
+      "learning_rate": 6.396706050841389e-06,
+      "loss": 0.0,
+      "step": 6589
+    },
+    {
+      "epoch": 0.42469549526325967,
+      "grad_norm": 0.03318998658871642,
+      "learning_rate": 6.395989974937344e-06,
+      "loss": 0.0,
+      "step": 6590
+    },
+    {
+      "epoch": 0.4247599407101888,
+      "grad_norm": 0.032235455008593907,
+      "learning_rate": 6.395273899033298e-06,
+      "loss": 0.0002,
+      "step": 6591
+    },
+    {
+      "epoch": 0.424824386157118,
+      "grad_norm": 0.0873344546653782,
+      "learning_rate": 6.394557823129253e-06,
+      "loss": 0.0016,
+      "step": 6592
+    },
+    {
+      "epoch": 0.42488883160404717,
+      "grad_norm": 0.012324949783537603,
+      "learning_rate": 6.393841747225207e-06,
+      "loss": 0.0,
+      "step": 6593
+    },
+    {
+      "epoch": 0.42495327705097635,
+      "grad_norm": 0.029404377397181828,
+      "learning_rate": 6.393125671321161e-06,
+      "loss": 0.0003,
+      "step": 6594
+    },
+    {
+      "epoch": 0.42501772249790554,
+      "grad_norm": 0.004150097631532944,
+      "learning_rate": 6.392409595417115e-06,
+      "loss": 0.0,
+      "step": 6595
+    },
+    {
+      "epoch": 0.4250821679448347,
+      "grad_norm": 0.426125046406366,
+      "learning_rate": 6.391693519513069e-06,
+      "loss": 0.0014,
+      "step": 6596
+    },
+    {
+      "epoch": 0.42514661339176385,
+      "grad_norm": 0.007814042117983378,
+      "learning_rate": 6.390977443609023e-06,
+      "loss": 0.0001,
+      "step": 6597
+    },
+    {
+      "epoch": 0.42521105883869303,
+      "grad_norm": 0.016080993757632295,
+      "learning_rate": 6.390261367704978e-06,
+      "loss": 0.0,
+      "step": 6598
+    },
+    {
+      "epoch": 0.4252755042856222,
+      "grad_norm": 0.001690795443047797,
+      "learning_rate": 6.389545291800931e-06,
+      "loss": 0.0,
+      "step": 6599
+    },
+    {
+      "epoch": 0.4253399497325514,
+      "grad_norm": 0.16194163111905666,
+      "learning_rate": 6.388829215896885e-06,
+      "loss": 0.0003,
+      "step": 6600
+    },
+    {
+      "epoch": 0.4254043951794806,
+      "grad_norm": 0.011428128871439189,
+      "learning_rate": 6.38811313999284e-06,
+      "loss": 0.0,
+      "step": 6601
+    },
+    {
+      "epoch": 0.4254688406264097,
+      "grad_norm": 0.08761187696328585,
+      "learning_rate": 6.387397064088794e-06,
+      "loss": 0.0017,
+      "step": 6602
+    },
+    {
+      "epoch": 0.4255332860733389,
+      "grad_norm": 0.2612283581799962,
+      "learning_rate": 6.386680988184748e-06,
+      "loss": 0.0003,
+      "step": 6603
+    },
+    {
+      "epoch": 0.4255977315202681,
+      "grad_norm": 0.019915802669471502,
+      "learning_rate": 6.385964912280702e-06,
+      "loss": 0.0,
+      "step": 6604
+    },
+    {
+      "epoch": 0.42566217696719727,
+      "grad_norm": 0.02653936073343681,
+      "learning_rate": 6.385248836376656e-06,
+      "loss": 0.0,
+      "step": 6605
+    },
+    {
+      "epoch": 0.42572662241412645,
+      "grad_norm": 0.00020666814500324758,
+      "learning_rate": 6.38453276047261e-06,
+      "loss": 0.0,
+      "step": 6606
+    },
+    {
+      "epoch": 0.42579106786105564,
+      "grad_norm": 0.02526154973886526,
+      "learning_rate": 6.383816684568565e-06,
+      "loss": 0.0,
+      "step": 6607
+    },
+    {
+      "epoch": 0.42585551330798477,
+      "grad_norm": 0.007133051405255623,
+      "learning_rate": 6.383100608664519e-06,
+      "loss": 0.0,
+      "step": 6608
+    },
+    {
+      "epoch": 0.42591995875491395,
+      "grad_norm": 0.0021597927325183465,
+      "learning_rate": 6.382384532760472e-06,
+      "loss": 0.0,
+      "step": 6609
+    },
+    {
+      "epoch": 0.42598440420184314,
+      "grad_norm": 0.023160755933042062,
+      "learning_rate": 6.381668456856427e-06,
+      "loss": 0.0001,
+      "step": 6610
+    },
+    {
+      "epoch": 0.4260488496487723,
+      "grad_norm": 0.07757443802956861,
+      "learning_rate": 6.380952380952381e-06,
+      "loss": 0.0001,
+      "step": 6611
+    },
+    {
+      "epoch": 0.4261132950957015,
+      "grad_norm": 0.06177029019459271,
+      "learning_rate": 6.380236305048335e-06,
+      "loss": 0.0002,
+      "step": 6612
+    },
+    {
+      "epoch": 0.4261777405426307,
+      "grad_norm": 0.22313402180370537,
+      "learning_rate": 6.3795202291442905e-06,
+      "loss": 0.0012,
+      "step": 6613
+    },
+    {
+      "epoch": 0.4262421859895598,
+      "grad_norm": 0.000392622305714873,
+      "learning_rate": 6.378804153240245e-06,
+      "loss": 0.0,
+      "step": 6614
+    },
+    {
+      "epoch": 0.426306631436489,
+      "grad_norm": 0.00409946774879405,
+      "learning_rate": 6.378088077336198e-06,
+      "loss": 0.0,
+      "step": 6615
+    },
+    {
+      "epoch": 0.4263710768834182,
+      "grad_norm": 0.0188214650348023,
+      "learning_rate": 6.3773720014321525e-06,
+      "loss": 0.0001,
+      "step": 6616
+    },
+    {
+      "epoch": 0.42643552233034737,
+      "grad_norm": 0.05915115781377536,
+      "learning_rate": 6.376655925528107e-06,
+      "loss": 0.0005,
+      "step": 6617
+    },
+    {
+      "epoch": 0.42649996777727656,
+      "grad_norm": 0.004750804011270957,
+      "learning_rate": 6.375939849624061e-06,
+      "loss": 0.0,
+      "step": 6618
+    },
+    {
+      "epoch": 0.4265644132242057,
+      "grad_norm": 0.0009950607837926515,
+      "learning_rate": 6.375223773720015e-06,
+      "loss": 0.0,
+      "step": 6619
+    },
+    {
+      "epoch": 0.42662885867113487,
+      "grad_norm": 0.0003103790189822709,
+      "learning_rate": 6.374507697815969e-06,
+      "loss": 0.0,
+      "step": 6620
+    },
+    {
+      "epoch": 0.42669330411806405,
+      "grad_norm": 0.006672502559690551,
+      "learning_rate": 6.373791621911923e-06,
+      "loss": 0.0,
+      "step": 6621
+    },
+    {
+      "epoch": 0.42675774956499324,
+      "grad_norm": 0.0012132214678914587,
+      "learning_rate": 6.3730755460078774e-06,
+      "loss": 0.0,
+      "step": 6622
+    },
+    {
+      "epoch": 0.4268221950119224,
+      "grad_norm": 0.5459795569131483,
+      "learning_rate": 6.372359470103832e-06,
+      "loss": 0.001,
+      "step": 6623
+    },
+    {
+      "epoch": 0.4268866404588516,
+      "grad_norm": 0.1584840858127378,
+      "learning_rate": 6.371643394199786e-06,
+      "loss": 0.0018,
+      "step": 6624
+    },
+    {
+      "epoch": 0.42695108590578074,
+      "grad_norm": 0.02230071107940539,
+      "learning_rate": 6.3709273182957395e-06,
+      "loss": 0.0,
+      "step": 6625
+    },
+    {
+      "epoch": 0.4270155313527099,
+      "grad_norm": 0.2010288563453443,
+      "learning_rate": 6.370211242391694e-06,
+      "loss": 0.0002,
+      "step": 6626
+    },
+    {
+      "epoch": 0.4270799767996391,
+      "grad_norm": 0.0005539057811006905,
+      "learning_rate": 6.369495166487648e-06,
+      "loss": 0.0,
+      "step": 6627
+    },
+    {
+      "epoch": 0.4271444222465683,
+      "grad_norm": 8.601880093504294e-05,
+      "learning_rate": 6.368779090583602e-06,
+      "loss": 0.0,
+      "step": 6628
+    },
+    {
+      "epoch": 0.4272088676934975,
+      "grad_norm": 0.00694744794728398,
+      "learning_rate": 6.368063014679557e-06,
+      "loss": 0.0,
+      "step": 6629
+    },
+    {
+      "epoch": 0.4272733131404266,
+      "grad_norm": 0.00011373609568961924,
+      "learning_rate": 6.36734693877551e-06,
+      "loss": 0.0,
+      "step": 6630
+    },
+    {
+      "epoch": 0.4273377585873558,
+      "grad_norm": 0.2741989161173644,
+      "learning_rate": 6.3666308628714644e-06,
+      "loss": 0.0024,
+      "step": 6631
+    },
+    {
+      "epoch": 0.42740220403428497,
+      "grad_norm": 0.121421209699953,
+      "learning_rate": 6.365914786967419e-06,
+      "loss": 0.0003,
+      "step": 6632
+    },
+    {
+      "epoch": 0.42746664948121416,
+      "grad_norm": 0.0009714433687004506,
+      "learning_rate": 6.365198711063373e-06,
+      "loss": 0.0,
+      "step": 6633
+    },
+    {
+      "epoch": 0.42753109492814334,
+      "grad_norm": 0.10040541532491891,
+      "learning_rate": 6.364482635159327e-06,
+      "loss": 0.0007,
+      "step": 6634
+    },
+    {
+      "epoch": 0.4275955403750725,
+      "grad_norm": 6.250609963284367e-05,
+      "learning_rate": 6.363766559255281e-06,
+      "loss": 0.0,
+      "step": 6635
+    },
+    {
+      "epoch": 0.42765998582200165,
+      "grad_norm": 0.0007349660592156871,
+      "learning_rate": 6.363050483351235e-06,
+      "loss": 0.0,
+      "step": 6636
+    },
+    {
+      "epoch": 0.42772443126893084,
+      "grad_norm": 0.00017060009315793165,
+      "learning_rate": 6.36233440744719e-06,
+      "loss": 0.0,
+      "step": 6637
+    },
+    {
+      "epoch": 0.42778887671586,
+      "grad_norm": 0.002033436154090894,
+      "learning_rate": 6.3616183315431445e-06,
+      "loss": 0.0,
+      "step": 6638
+    },
+    {
+      "epoch": 0.4278533221627892,
+      "grad_norm": 0.001637918147364348,
+      "learning_rate": 6.360902255639099e-06,
+      "loss": 0.0,
+      "step": 6639
+    },
+    {
+      "epoch": 0.4279177676097184,
+      "grad_norm": 0.19539891238751805,
+      "learning_rate": 6.360186179735053e-06,
+      "loss": 0.0004,
+      "step": 6640
+    },
+    {
+      "epoch": 0.4279822130566475,
+      "grad_norm": 0.0018548896809036603,
+      "learning_rate": 6.359470103831007e-06,
+      "loss": 0.0,
+      "step": 6641
+    },
+    {
+      "epoch": 0.4280466585035767,
+      "grad_norm": 0.0013334698909709792,
+      "learning_rate": 6.358754027926961e-06,
+      "loss": 0.0,
+      "step": 6642
+    },
+    {
+      "epoch": 0.4281111039505059,
+      "grad_norm": 0.0028419573034907998,
+      "learning_rate": 6.358037952022915e-06,
+      "loss": 0.0,
+      "step": 6643
+    },
+    {
+      "epoch": 0.4281755493974351,
+      "grad_norm": 0.006881097440326114,
+      "learning_rate": 6.3573218761188695e-06,
+      "loss": 0.0,
+      "step": 6644
+    },
+    {
+      "epoch": 0.42823999484436426,
+      "grad_norm": 0.001116479305550302,
+      "learning_rate": 6.356605800214824e-06,
+      "loss": 0.0,
+      "step": 6645
+    },
+    {
+      "epoch": 0.42830444029129344,
+      "grad_norm": 0.00025453688824276247,
+      "learning_rate": 6.355889724310777e-06,
+      "loss": 0.0,
+      "step": 6646
+    },
+    {
+      "epoch": 0.42836888573822257,
+      "grad_norm": 0.005216877980391289,
+      "learning_rate": 6.3551736484067315e-06,
+      "loss": 0.0,
+      "step": 6647
+    },
+    {
+      "epoch": 0.42843333118515176,
+      "grad_norm": 0.001896177419504992,
+      "learning_rate": 6.354457572502686e-06,
+      "loss": 0.0,
+      "step": 6648
+    },
+    {
+      "epoch": 0.42849777663208094,
+      "grad_norm": 0.0002611695027218609,
+      "learning_rate": 6.35374149659864e-06,
+      "loss": 0.0,
+      "step": 6649
+    },
+    {
+      "epoch": 0.4285622220790101,
+      "grad_norm": 0.22977676603252645,
+      "learning_rate": 6.353025420694594e-06,
+      "loss": 0.0003,
+      "step": 6650
+    },
+    {
+      "epoch": 0.4286266675259393,
+      "grad_norm": 0.0017587722789177331,
+      "learning_rate": 6.352309344790548e-06,
+      "loss": 0.0,
+      "step": 6651
+    },
+    {
+      "epoch": 0.4286911129728685,
+      "grad_norm": 0.0020766966092716993,
+      "learning_rate": 6.351593268886502e-06,
+      "loss": 0.0,
+      "step": 6652
+    },
+    {
+      "epoch": 0.4287555584197976,
+      "grad_norm": 0.40843597414743027,
+      "learning_rate": 6.3508771929824565e-06,
+      "loss": 0.0014,
+      "step": 6653
+    },
+    {
+      "epoch": 0.4288200038667268,
+      "grad_norm": 0.013422847530458746,
+      "learning_rate": 6.350161117078411e-06,
+      "loss": 0.0,
+      "step": 6654
+    },
+    {
+      "epoch": 0.428884449313656,
+      "grad_norm": 0.7281807546593764,
+      "learning_rate": 6.349445041174365e-06,
+      "loss": 0.005,
+      "step": 6655
+    },
+    {
+      "epoch": 0.4289488947605852,
+      "grad_norm": 0.0007956452725589806,
+      "learning_rate": 6.3487289652703185e-06,
+      "loss": 0.0,
+      "step": 6656
+    },
+    {
+      "epoch": 0.42901334020751436,
+      "grad_norm": 0.008950084640330636,
+      "learning_rate": 6.348012889366273e-06,
+      "loss": 0.0,
+      "step": 6657
+    },
+    {
+      "epoch": 0.4290777856544435,
+      "grad_norm": 0.17089821406758554,
+      "learning_rate": 6.347296813462227e-06,
+      "loss": 0.0018,
+      "step": 6658
+    },
+    {
+      "epoch": 0.4291422311013727,
+      "grad_norm": 0.0030694031207953013,
+      "learning_rate": 6.346580737558181e-06,
+      "loss": 0.0,
+      "step": 6659
+    },
+    {
+      "epoch": 0.42920667654830186,
+      "grad_norm": 0.02444291912822385,
+      "learning_rate": 6.3458646616541366e-06,
+      "loss": 0.0,
+      "step": 6660
+    },
+    {
+      "epoch": 0.42927112199523104,
+      "grad_norm": 0.002884947247548892,
+      "learning_rate": 6.345148585750091e-06,
+      "loss": 0.0,
+      "step": 6661
+    },
+    {
+      "epoch": 0.4293355674421602,
+      "grad_norm": 0.005064100957951929,
+      "learning_rate": 6.344432509846044e-06,
+      "loss": 0.0,
+      "step": 6662
+    },
+    {
+      "epoch": 0.4294000128890894,
+      "grad_norm": 0.023081917055650943,
+      "learning_rate": 6.343716433941999e-06,
+      "loss": 0.0002,
+      "step": 6663
+    },
+    {
+      "epoch": 0.42946445833601854,
+      "grad_norm": 0.0009968242300762613,
+      "learning_rate": 6.343000358037953e-06,
+      "loss": 0.0,
+      "step": 6664
+    },
+    {
+      "epoch": 0.4295289037829477,
+      "grad_norm": 0.002290510605043889,
+      "learning_rate": 6.342284282133907e-06,
+      "loss": 0.0,
+      "step": 6665
+    },
+    {
+      "epoch": 0.4295933492298769,
+      "grad_norm": 0.00029481900990017935,
+      "learning_rate": 6.3415682062298615e-06,
+      "loss": 0.0,
+      "step": 6666
+    },
+    {
+      "epoch": 0.4296577946768061,
+      "grad_norm": 0.002503810212365343,
+      "learning_rate": 6.340852130325815e-06,
+      "loss": 0.0,
+      "step": 6667
+    },
+    {
+      "epoch": 0.4297222401237353,
+      "grad_norm": 0.08980436602116373,
+      "learning_rate": 6.340136054421769e-06,
+      "loss": 0.0002,
+      "step": 6668
+    },
+    {
+      "epoch": 0.4297866855706644,
+      "grad_norm": 0.011429338482220037,
+      "learning_rate": 6.3394199785177236e-06,
+      "loss": 0.0,
+      "step": 6669
+    },
+    {
+      "epoch": 0.4298511310175936,
+      "grad_norm": 0.009484022902911424,
+      "learning_rate": 6.338703902613678e-06,
+      "loss": 0.0,
+      "step": 6670
+    },
+    {
+      "epoch": 0.4299155764645228,
+      "grad_norm": 0.02464525902804644,
+      "learning_rate": 6.337987826709632e-06,
+      "loss": 0.0002,
+      "step": 6671
+    },
+    {
+      "epoch": 0.42998002191145196,
+      "grad_norm": 0.0070729571841776796,
+      "learning_rate": 6.337271750805586e-06,
+      "loss": 0.0,
+      "step": 6672
+    },
+    {
+      "epoch": 0.43004446735838114,
+      "grad_norm": 0.0005698593327859449,
+      "learning_rate": 6.33655567490154e-06,
+      "loss": 0.0,
+      "step": 6673
+    },
+    {
+      "epoch": 0.43010891280531033,
+      "grad_norm": 0.007270899548970992,
+      "learning_rate": 6.335839598997494e-06,
+      "loss": 0.0,
+      "step": 6674
+    },
+    {
+      "epoch": 0.43017335825223946,
+      "grad_norm": 0.0009913103341804657,
+      "learning_rate": 6.3351235230934485e-06,
+      "loss": 0.0,
+      "step": 6675
+    },
+    {
+      "epoch": 0.43023780369916864,
+      "grad_norm": 0.002595574146712181,
+      "learning_rate": 6.334407447189402e-06,
+      "loss": 0.0,
+      "step": 6676
+    },
+    {
+      "epoch": 0.4303022491460978,
+      "grad_norm": 0.14062348338474917,
+      "learning_rate": 6.333691371285356e-06,
+      "loss": 0.001,
+      "step": 6677
+    },
+    {
+      "epoch": 0.430366694593027,
+      "grad_norm": 0.000567302968704546,
+      "learning_rate": 6.3329752953813105e-06,
+      "loss": 0.0,
+      "step": 6678
+    },
+    {
+      "epoch": 0.4304311400399562,
+      "grad_norm": 0.0054263714955121115,
+      "learning_rate": 6.332259219477265e-06,
+      "loss": 0.0,
+      "step": 6679
+    },
+    {
+      "epoch": 0.4304955854868854,
+      "grad_norm": 0.016467133939146297,
+      "learning_rate": 6.331543143573219e-06,
+      "loss": 0.0001,
+      "step": 6680
+    },
+    {
+      "epoch": 0.4305600309338145,
+      "grad_norm": 0.0013694414101803056,
+      "learning_rate": 6.330827067669173e-06,
+      "loss": 0.0,
+      "step": 6681
+    },
+    {
+      "epoch": 0.4306244763807437,
+      "grad_norm": 0.0037395349074488956,
+      "learning_rate": 6.330110991765127e-06,
+      "loss": 0.0,
+      "step": 6682
+    },
+    {
+      "epoch": 0.4306889218276729,
+      "grad_norm": 0.29080137264148787,
+      "learning_rate": 6.329394915861082e-06,
+      "loss": 0.0009,
+      "step": 6683
+    },
+    {
+      "epoch": 0.43075336727460206,
+      "grad_norm": 0.002323678893051278,
+      "learning_rate": 6.328678839957036e-06,
+      "loss": 0.0,
+      "step": 6684
+    },
+    {
+      "epoch": 0.43081781272153125,
+      "grad_norm": 0.08255165084440778,
+      "learning_rate": 6.327962764052991e-06,
+      "loss": 0.0001,
+      "step": 6685
+    },
+    {
+      "epoch": 0.4308822581684604,
+      "grad_norm": 0.1910078417668657,
+      "learning_rate": 6.327246688148945e-06,
+      "loss": 0.0004,
+      "step": 6686
+    },
+    {
+      "epoch": 0.43094670361538956,
+      "grad_norm": 0.0015552556160902183,
+      "learning_rate": 6.326530612244899e-06,
+      "loss": 0.0,
+      "step": 6687
+    },
+    {
+      "epoch": 0.43101114906231874,
+      "grad_norm": 0.000931274017975527,
+      "learning_rate": 6.325814536340853e-06,
+      "loss": 0.0,
+      "step": 6688
+    },
+    {
+      "epoch": 0.43107559450924793,
+      "grad_norm": 0.026236450627385367,
+      "learning_rate": 6.325098460436807e-06,
+      "loss": 0.0002,
+      "step": 6689
+    },
+    {
+      "epoch": 0.4311400399561771,
+      "grad_norm": 0.007944311984236513,
+      "learning_rate": 6.324382384532761e-06,
+      "loss": 0.0,
+      "step": 6690
+    },
+    {
+      "epoch": 0.4312044854031063,
+      "grad_norm": 0.0017483911802880752,
+      "learning_rate": 6.323666308628716e-06,
+      "loss": 0.0,
+      "step": 6691
+    },
+    {
+      "epoch": 0.4312689308500354,
+      "grad_norm": 0.8183681514879361,
+      "learning_rate": 6.322950232724669e-06,
+      "loss": 0.0008,
+      "step": 6692
+    },
+    {
+      "epoch": 0.4313333762969646,
+      "grad_norm": 0.013792580667864334,
+      "learning_rate": 6.322234156820623e-06,
+      "loss": 0.0,
+      "step": 6693
+    },
+    {
+      "epoch": 0.4313978217438938,
+      "grad_norm": 0.015097850842103971,
+      "learning_rate": 6.321518080916578e-06,
+      "loss": 0.0,
+      "step": 6694
+    },
+    {
+      "epoch": 0.431462267190823,
+      "grad_norm": 0.006704856775930077,
+      "learning_rate": 6.320802005012532e-06,
+      "loss": 0.0,
+      "step": 6695
+    },
+    {
+      "epoch": 0.43152671263775216,
+      "grad_norm": 0.0008574753945796366,
+      "learning_rate": 6.320085929108486e-06,
+      "loss": 0.0,
+      "step": 6696
+    },
+    {
+      "epoch": 0.4315911580846813,
+      "grad_norm": 0.18016405765260005,
+      "learning_rate": 6.31936985320444e-06,
+      "loss": 0.0019,
+      "step": 6697
+    },
+    {
+      "epoch": 0.4316556035316105,
+      "grad_norm": 4.158763383683872,
+      "learning_rate": 6.318653777300394e-06,
+      "loss": 0.0134,
+      "step": 6698
+    },
+    {
+      "epoch": 0.43172004897853966,
+      "grad_norm": 0.04242575207314043,
+      "learning_rate": 6.317937701396348e-06,
+      "loss": 0.0001,
+      "step": 6699
+    },
+    {
+      "epoch": 0.43178449442546885,
+      "grad_norm": 0.0038851543432019396,
+      "learning_rate": 6.317221625492303e-06,
+      "loss": 0.0,
+      "step": 6700
+    },
+    {
+      "epoch": 0.43184893987239803,
+      "grad_norm": 0.15580466106012691,
+      "learning_rate": 6.316505549588257e-06,
+      "loss": 0.0009,
+      "step": 6701
+    },
+    {
+      "epoch": 0.4319133853193272,
+      "grad_norm": 0.005339445096540471,
+      "learning_rate": 6.31578947368421e-06,
+      "loss": 0.0,
+      "step": 6702
+    },
+    {
+      "epoch": 0.43197783076625634,
+      "grad_norm": 0.0005897817405355401,
+      "learning_rate": 6.315073397780165e-06,
+      "loss": 0.0,
+      "step": 6703
+    },
+    {
+      "epoch": 0.4320422762131855,
+      "grad_norm": 0.00029861629602622337,
+      "learning_rate": 6.314357321876119e-06,
+      "loss": 0.0,
+      "step": 6704
+    },
+    {
+      "epoch": 0.4321067216601147,
+      "grad_norm": 0.0011596472343831097,
+      "learning_rate": 6.313641245972073e-06,
+      "loss": 0.0,
+      "step": 6705
+    },
+    {
+      "epoch": 0.4321711671070439,
+      "grad_norm": 0.21832649072057134,
+      "learning_rate": 6.3129251700680275e-06,
+      "loss": 0.0002,
+      "step": 6706
+    },
+    {
+      "epoch": 0.4322356125539731,
+      "grad_norm": 8.891694569460509e-05,
+      "learning_rate": 6.312209094163983e-06,
+      "loss": 0.0,
+      "step": 6707
+    },
+    {
+      "epoch": 0.4323000580009022,
+      "grad_norm": 0.0010427760358763495,
+      "learning_rate": 6.311493018259936e-06,
+      "loss": 0.0,
+      "step": 6708
+    },
+    {
+      "epoch": 0.4323645034478314,
+      "grad_norm": 0.044251419583747786,
+      "learning_rate": 6.31077694235589e-06,
+      "loss": 0.0,
+      "step": 6709
+    },
+    {
+      "epoch": 0.4324289488947606,
+      "grad_norm": 0.000149714890507064,
+      "learning_rate": 6.310060866451845e-06,
+      "loss": 0.0,
+      "step": 6710
+    },
+    {
+      "epoch": 0.43249339434168976,
+      "grad_norm": 0.20032516548059928,
+      "learning_rate": 6.309344790547799e-06,
+      "loss": 0.0002,
+      "step": 6711
+    },
+    {
+      "epoch": 0.43255783978861895,
+      "grad_norm": 0.002627819770958189,
+      "learning_rate": 6.308628714643753e-06,
+      "loss": 0.0,
+      "step": 6712
+    },
+    {
+      "epoch": 0.43262228523554813,
+      "grad_norm": 0.01592602803687255,
+      "learning_rate": 6.307912638739707e-06,
+      "loss": 0.0,
+      "step": 6713
+    },
+    {
+      "epoch": 0.43268673068247726,
+      "grad_norm": 0.004949941235307722,
+      "learning_rate": 6.307196562835661e-06,
+      "loss": 0.0,
+      "step": 6714
+    },
+    {
+      "epoch": 0.43275117612940645,
+      "grad_norm": 0.4457405776687715,
+      "learning_rate": 6.306480486931615e-06,
+      "loss": 0.0016,
+      "step": 6715
+    },
+    {
+      "epoch": 0.43281562157633563,
+      "grad_norm": 0.0001713559314944813,
+      "learning_rate": 6.30576441102757e-06,
+      "loss": 0.0,
+      "step": 6716
+    },
+    {
+      "epoch": 0.4328800670232648,
+      "grad_norm": 0.52003317085252,
+      "learning_rate": 6.305048335123524e-06,
+      "loss": 0.0034,
+      "step": 6717
+    },
+    {
+      "epoch": 0.432944512470194,
+      "grad_norm": 0.24328987666119062,
+      "learning_rate": 6.304332259219477e-06,
+      "loss": 0.0004,
+      "step": 6718
+    },
+    {
+      "epoch": 0.4330089579171232,
+      "grad_norm": 6.05488712972811e-05,
+      "learning_rate": 6.303616183315432e-06,
+      "loss": 0.0,
+      "step": 6719
+    },
+    {
+      "epoch": 0.4330734033640523,
+      "grad_norm": 0.9154862295455783,
+      "learning_rate": 6.302900107411386e-06,
+      "loss": 0.0044,
+      "step": 6720
+    },
+    {
+      "epoch": 0.4331378488109815,
+      "grad_norm": 0.00011452810699505064,
+      "learning_rate": 6.30218403150734e-06,
+      "loss": 0.0,
+      "step": 6721
+    },
+    {
+      "epoch": 0.4332022942579107,
+      "grad_norm": 0.2684594141416232,
+      "learning_rate": 6.301467955603295e-06,
+      "loss": 0.0047,
+      "step": 6722
+    },
+    {
+      "epoch": 0.43326673970483986,
+      "grad_norm": 0.02207255766112272,
+      "learning_rate": 6.300751879699248e-06,
+      "loss": 0.0,
+      "step": 6723
+    },
+    {
+      "epoch": 0.43333118515176905,
+      "grad_norm": 0.0029896027307358413,
+      "learning_rate": 6.300035803795202e-06,
+      "loss": 0.0,
+      "step": 6724
+    },
+    {
+      "epoch": 0.4333956305986982,
+      "grad_norm": 0.16950898046179497,
+      "learning_rate": 6.299319727891157e-06,
+      "loss": 0.0003,
+      "step": 6725
+    },
+    {
+      "epoch": 0.43346007604562736,
+      "grad_norm": 0.29806840851315397,
+      "learning_rate": 6.298603651987111e-06,
+      "loss": 0.0017,
+      "step": 6726
+    },
+    {
+      "epoch": 0.43352452149255655,
+      "grad_norm": 0.016236714793457938,
+      "learning_rate": 6.297887576083065e-06,
+      "loss": 0.0,
+      "step": 6727
+    },
+    {
+      "epoch": 0.43358896693948573,
+      "grad_norm": 0.009315682470924137,
+      "learning_rate": 6.297171500179019e-06,
+      "loss": 0.0,
+      "step": 6728
+    },
+    {
+      "epoch": 0.4336534123864149,
+      "grad_norm": 0.03930989690288691,
+      "learning_rate": 6.296455424274973e-06,
+      "loss": 0.0001,
+      "step": 6729
+    },
+    {
+      "epoch": 0.4337178578333441,
+      "grad_norm": 0.17229336517397006,
+      "learning_rate": 6.295739348370928e-06,
+      "loss": 0.0018,
+      "step": 6730
+    },
+    {
+      "epoch": 0.43378230328027323,
+      "grad_norm": 0.010794653699986811,
+      "learning_rate": 6.2950232724668825e-06,
+      "loss": 0.0,
+      "step": 6731
+    },
+    {
+      "epoch": 0.4338467487272024,
+      "grad_norm": 0.09056349817328786,
+      "learning_rate": 6.294307196562837e-06,
+      "loss": 0.0001,
+      "step": 6732
+    },
+    {
+      "epoch": 0.4339111941741316,
+      "grad_norm": 0.0032560131138640283,
+      "learning_rate": 6.293591120658791e-06,
+      "loss": 0.0,
+      "step": 6733
+    },
+    {
+      "epoch": 0.4339756396210608,
+      "grad_norm": 0.2929552456604397,
+      "learning_rate": 6.2928750447547445e-06,
+      "loss": 0.0024,
+      "step": 6734
+    },
+    {
+      "epoch": 0.43404008506798997,
+      "grad_norm": 0.004358307485739401,
+      "learning_rate": 6.292158968850699e-06,
+      "loss": 0.0,
+      "step": 6735
+    },
+    {
+      "epoch": 0.4341045305149191,
+      "grad_norm": 0.7401282000325132,
+      "learning_rate": 6.291442892946653e-06,
+      "loss": 0.0031,
+      "step": 6736
+    },
+    {
+      "epoch": 0.4341689759618483,
+      "grad_norm": 0.024909603939854592,
+      "learning_rate": 6.290726817042607e-06,
+      "loss": 0.0001,
+      "step": 6737
+    },
+    {
+      "epoch": 0.43423342140877746,
+      "grad_norm": 0.005705359287954765,
+      "learning_rate": 6.290010741138562e-06,
+      "loss": 0.0,
+      "step": 6738
+    },
+    {
+      "epoch": 0.43429786685570665,
+      "grad_norm": 0.1156991258673197,
+      "learning_rate": 6.289294665234515e-06,
+      "loss": 0.0001,
+      "step": 6739
+    },
+    {
+      "epoch": 0.43436231230263583,
+      "grad_norm": 0.008699033419681219,
+      "learning_rate": 6.2885785893304694e-06,
+      "loss": 0.0,
+      "step": 6740
+    },
+    {
+      "epoch": 0.434426757749565,
+      "grad_norm": 0.6171667602531093,
+      "learning_rate": 6.287862513426424e-06,
+      "loss": 0.0039,
+      "step": 6741
+    },
+    {
+      "epoch": 0.43449120319649415,
+      "grad_norm": 0.16432638744027941,
+      "learning_rate": 6.287146437522378e-06,
+      "loss": 0.0004,
+      "step": 6742
+    },
+    {
+      "epoch": 0.43455564864342333,
+      "grad_norm": 0.02447242080053091,
+      "learning_rate": 6.286430361618332e-06,
+      "loss": 0.0,
+      "step": 6743
+    },
+    {
+      "epoch": 0.4346200940903525,
+      "grad_norm": 0.0921131984266586,
+      "learning_rate": 6.285714285714286e-06,
+      "loss": 0.0001,
+      "step": 6744
+    },
+    {
+      "epoch": 0.4346845395372817,
+      "grad_norm": 0.022735265434989764,
+      "learning_rate": 6.28499820981024e-06,
+      "loss": 0.0,
+      "step": 6745
+    },
+    {
+      "epoch": 0.4347489849842109,
+      "grad_norm": 0.0013172423891951862,
+      "learning_rate": 6.284282133906194e-06,
+      "loss": 0.0,
+      "step": 6746
+    },
+    {
+      "epoch": 0.43481343043114,
+      "grad_norm": 0.007643909237297049,
+      "learning_rate": 6.283566058002149e-06,
+      "loss": 0.0,
+      "step": 6747
+    },
+    {
+      "epoch": 0.4348778758780692,
+      "grad_norm": 0.04607741987970149,
+      "learning_rate": 6.282849982098103e-06,
+      "loss": 0.0001,
+      "step": 6748
+    },
+    {
+      "epoch": 0.4349423213249984,
+      "grad_norm": 0.06911486318143999,
+      "learning_rate": 6.2821339061940564e-06,
+      "loss": 0.0,
+      "step": 6749
+    },
+    {
+      "epoch": 0.43500676677192757,
+      "grad_norm": 0.02393797398342579,
+      "learning_rate": 6.281417830290011e-06,
+      "loss": 0.0,
+      "step": 6750
+    },
+    {
+      "epoch": 0.43507121221885675,
+      "grad_norm": 0.26460427623124094,
+      "learning_rate": 6.280701754385965e-06,
+      "loss": 0.0016,
+      "step": 6751
+    },
+    {
+      "epoch": 0.43513565766578594,
+      "grad_norm": 0.00026218383385574675,
+      "learning_rate": 6.279985678481919e-06,
+      "loss": 0.0,
+      "step": 6752
+    },
+    {
+      "epoch": 0.43520010311271506,
+      "grad_norm": 0.0026633181035528104,
+      "learning_rate": 6.279269602577874e-06,
+      "loss": 0.0,
+      "step": 6753
+    },
+    {
+      "epoch": 0.43526454855964425,
+      "grad_norm": 0.11784700149040153,
+      "learning_rate": 6.278553526673829e-06,
+      "loss": 0.0006,
+      "step": 6754
+    },
+    {
+      "epoch": 0.43532899400657343,
+      "grad_norm": 0.0017846129932184572,
+      "learning_rate": 6.277837450769782e-06,
+      "loss": 0.0,
+      "step": 6755
+    },
+    {
+      "epoch": 0.4353934394535026,
+      "grad_norm": 0.07136146578848694,
+      "learning_rate": 6.2771213748657365e-06,
+      "loss": 0.0001,
+      "step": 6756
+    },
+    {
+      "epoch": 0.4354578849004318,
+      "grad_norm": 0.17367550761601813,
+      "learning_rate": 6.276405298961691e-06,
+      "loss": 0.0003,
+      "step": 6757
+    },
+    {
+      "epoch": 0.435522330347361,
+      "grad_norm": 0.06442347118931648,
+      "learning_rate": 6.275689223057645e-06,
+      "loss": 0.0009,
+      "step": 6758
+    },
+    {
+      "epoch": 0.4355867757942901,
+      "grad_norm": 1.113797680383268,
+      "learning_rate": 6.2749731471535994e-06,
+      "loss": 0.0098,
+      "step": 6759
+    },
+    {
+      "epoch": 0.4356512212412193,
+      "grad_norm": 0.005809118892805713,
+      "learning_rate": 6.274257071249553e-06,
+      "loss": 0.0,
+      "step": 6760
+    },
+    {
+      "epoch": 0.4357156666881485,
+      "grad_norm": 1.4354561913321795,
+      "learning_rate": 6.273540995345507e-06,
+      "loss": 0.0023,
+      "step": 6761
+    },
+    {
+      "epoch": 0.43578011213507767,
+      "grad_norm": 0.052269633952662035,
+      "learning_rate": 6.2728249194414615e-06,
+      "loss": 0.0005,
+      "step": 6762
+    },
+    {
+      "epoch": 0.43584455758200685,
+      "grad_norm": 0.012401707536231203,
+      "learning_rate": 6.272108843537416e-06,
+      "loss": 0.0001,
+      "step": 6763
+    },
+    {
+      "epoch": 0.435909003028936,
+      "grad_norm": 0.0037154901687754456,
+      "learning_rate": 6.27139276763337e-06,
+      "loss": 0.0,
+      "step": 6764
+    },
+    {
+      "epoch": 0.43597344847586517,
+      "grad_norm": 0.0018004172015471341,
+      "learning_rate": 6.2706766917293235e-06,
+      "loss": 0.0,
+      "step": 6765
+    },
+    {
+      "epoch": 0.43603789392279435,
+      "grad_norm": 0.03574004342730524,
+      "learning_rate": 6.269960615825278e-06,
+      "loss": 0.0001,
+      "step": 6766
+    },
+    {
+      "epoch": 0.43610233936972354,
+      "grad_norm": 0.010076841435620492,
+      "learning_rate": 6.269244539921232e-06,
+      "loss": 0.0,
+      "step": 6767
+    },
+    {
+      "epoch": 0.4361667848166527,
+      "grad_norm": 0.15066047090537446,
+      "learning_rate": 6.268528464017186e-06,
+      "loss": 0.0016,
+      "step": 6768
+    },
+    {
+      "epoch": 0.4362312302635819,
+      "grad_norm": 0.15723401401466586,
+      "learning_rate": 6.267812388113141e-06,
+      "loss": 0.0007,
+      "step": 6769
+    },
+    {
+      "epoch": 0.43629567571051103,
+      "grad_norm": 0.29885035164145274,
+      "learning_rate": 6.267096312209094e-06,
+      "loss": 0.0011,
+      "step": 6770
+    },
+    {
+      "epoch": 0.4363601211574402,
+      "grad_norm": 0.06953103863759008,
+      "learning_rate": 6.2663802363050485e-06,
+      "loss": 0.0,
+      "step": 6771
+    },
+    {
+      "epoch": 0.4364245666043694,
+      "grad_norm": 0.0003913824342631887,
+      "learning_rate": 6.265664160401003e-06,
+      "loss": 0.0,
+      "step": 6772
+    },
+    {
+      "epoch": 0.4364890120512986,
+      "grad_norm": 5.8933252323782934e-05,
+      "learning_rate": 6.264948084496957e-06,
+      "loss": 0.0,
+      "step": 6773
+    },
+    {
+      "epoch": 0.43655345749822777,
+      "grad_norm": 0.003684676938295499,
+      "learning_rate": 6.2642320085929105e-06,
+      "loss": 0.0,
+      "step": 6774
+    },
+    {
+      "epoch": 0.4366179029451569,
+      "grad_norm": 0.00030307847703955963,
+      "learning_rate": 6.263515932688865e-06,
+      "loss": 0.0,
+      "step": 6775
+    },
+    {
+      "epoch": 0.4366823483920861,
+      "grad_norm": 0.007013575093070469,
+      "learning_rate": 6.262799856784819e-06,
+      "loss": 0.0,
+      "step": 6776
+    },
+    {
+      "epoch": 0.43674679383901527,
+      "grad_norm": 0.00314157343057512,
+      "learning_rate": 6.262083780880774e-06,
+      "loss": 0.0,
+      "step": 6777
+    },
+    {
+      "epoch": 0.43681123928594445,
+      "grad_norm": 0.0032058113583623357,
+      "learning_rate": 6.2613677049767286e-06,
+      "loss": 0.0,
+      "step": 6778
+    },
+    {
+      "epoch": 0.43687568473287364,
+      "grad_norm": 0.14535620811397174,
+      "learning_rate": 6.260651629072683e-06,
+      "loss": 0.0004,
+      "step": 6779
+    },
+    {
+      "epoch": 0.4369401301798028,
+      "grad_norm": 0.0010641447845441464,
+      "learning_rate": 6.259935553168637e-06,
+      "loss": 0.0,
+      "step": 6780
+    },
+    {
+      "epoch": 0.43700457562673195,
+      "grad_norm": 0.030892537690979292,
+      "learning_rate": 6.259219477264591e-06,
+      "loss": 0.0002,
+      "step": 6781
+    },
+    {
+      "epoch": 0.43706902107366113,
+      "grad_norm": 0.5230201366789412,
+      "learning_rate": 6.258503401360545e-06,
+      "loss": 0.0012,
+      "step": 6782
+    },
+    {
+      "epoch": 0.4371334665205903,
+      "grad_norm": 0.18978095448541774,
+      "learning_rate": 6.257787325456499e-06,
+      "loss": 0.0007,
+      "step": 6783
+    },
+    {
+      "epoch": 0.4371979119675195,
+      "grad_norm": 0.00987531260439076,
+      "learning_rate": 6.2570712495524535e-06,
+      "loss": 0.0,
+      "step": 6784
+    },
+    {
+      "epoch": 0.4372623574144487,
+      "grad_norm": 0.25283649745634335,
+      "learning_rate": 6.256355173648408e-06,
+      "loss": 0.0003,
+      "step": 6785
+    },
+    {
+      "epoch": 0.4373268028613778,
+      "grad_norm": 0.0061907236244896895,
+      "learning_rate": 6.255639097744361e-06,
+      "loss": 0.0,
+      "step": 6786
+    },
+    {
+      "epoch": 0.437391248308307,
+      "grad_norm": 0.2543641582962293,
+      "learning_rate": 6.2549230218403156e-06,
+      "loss": 0.0018,
+      "step": 6787
+    },
+    {
+      "epoch": 0.4374556937552362,
+      "grad_norm": 0.008615261739393556,
+      "learning_rate": 6.25420694593627e-06,
+      "loss": 0.0,
+      "step": 6788
+    },
+    {
+      "epoch": 0.43752013920216537,
+      "grad_norm": 0.01651753520272118,
+      "learning_rate": 6.253490870032224e-06,
+      "loss": 0.0002,
+      "step": 6789
+    },
+    {
+      "epoch": 0.43758458464909455,
+      "grad_norm": 0.21448941922489909,
+      "learning_rate": 6.252774794128178e-06,
+      "loss": 0.0007,
+      "step": 6790
+    },
+    {
+      "epoch": 0.43764903009602374,
+      "grad_norm": 3.2058399888517126e-05,
+      "learning_rate": 6.252058718224132e-06,
+      "loss": 0.0,
+      "step": 6791
+    },
+    {
+      "epoch": 0.43771347554295287,
+      "grad_norm": 0.7354880075909804,
+      "learning_rate": 6.251342642320086e-06,
+      "loss": 0.0028,
+      "step": 6792
+    },
+    {
+      "epoch": 0.43777792098988205,
+      "grad_norm": 0.0030102187090905903,
+      "learning_rate": 6.2506265664160405e-06,
+      "loss": 0.0,
+      "step": 6793
+    },
+    {
+      "epoch": 0.43784236643681124,
+      "grad_norm": 0.01540114091165498,
+      "learning_rate": 6.249910490511995e-06,
+      "loss": 0.0001,
+      "step": 6794
+    },
+    {
+      "epoch": 0.4379068118837404,
+      "grad_norm": 0.08756405771852145,
+      "learning_rate": 6.249194414607948e-06,
+      "loss": 0.0001,
+      "step": 6795
+    },
+    {
+      "epoch": 0.4379712573306696,
+      "grad_norm": 0.004207340449658778,
+      "learning_rate": 6.2484783387039025e-06,
+      "loss": 0.0,
+      "step": 6796
+    },
+    {
+      "epoch": 0.4380357027775988,
+      "grad_norm": 0.03770904945438654,
+      "learning_rate": 6.247762262799857e-06,
+      "loss": 0.0001,
+      "step": 6797
+    },
+    {
+      "epoch": 0.4381001482245279,
+      "grad_norm": 0.0003550056645818315,
+      "learning_rate": 6.247046186895811e-06,
+      "loss": 0.0,
+      "step": 6798
+    },
+    {
+      "epoch": 0.4381645936714571,
+      "grad_norm": 0.14162342297415867,
+      "learning_rate": 6.2463301109917654e-06,
+      "loss": 0.0019,
+      "step": 6799
+    },
+    {
+      "epoch": 0.4382290391183863,
+      "grad_norm": 0.00033271973675798647,
+      "learning_rate": 6.245614035087721e-06,
+      "loss": 0.0,
+      "step": 6800
+    },
+    {
+      "epoch": 0.43829348456531547,
+      "grad_norm": 1.0295957533116706,
+      "learning_rate": 6.244897959183675e-06,
+      "loss": 0.007,
+      "step": 6801
+    },
+    {
+      "epoch": 0.43835793001224466,
+      "grad_norm": 0.011322654464795303,
+      "learning_rate": 6.244181883279628e-06,
+      "loss": 0.0001,
+      "step": 6802
+    },
+    {
+      "epoch": 0.4384223754591738,
+      "grad_norm": 0.005792339884773891,
+      "learning_rate": 6.243465807375583e-06,
+      "loss": 0.0,
+      "step": 6803
+    },
+    {
+      "epoch": 0.43848682090610297,
+      "grad_norm": 0.7715247421327011,
+      "learning_rate": 6.242749731471537e-06,
+      "loss": 0.0032,
+      "step": 6804
+    },
+    {
+      "epoch": 0.43855126635303215,
+      "grad_norm": 0.029430314989175126,
+      "learning_rate": 6.242033655567491e-06,
+      "loss": 0.0001,
+      "step": 6805
+    },
+    {
+      "epoch": 0.43861571179996134,
+      "grad_norm": 0.0044577108409088175,
+      "learning_rate": 6.241317579663445e-06,
+      "loss": 0.0,
+      "step": 6806
+    },
+    {
+      "epoch": 0.4386801572468905,
+      "grad_norm": 0.0011134538704909022,
+      "learning_rate": 6.240601503759399e-06,
+      "loss": 0.0,
+      "step": 6807
+    },
+    {
+      "epoch": 0.4387446026938197,
+      "grad_norm": 0.00017366062586490197,
+      "learning_rate": 6.239885427855353e-06,
+      "loss": 0.0,
+      "step": 6808
+    },
+    {
+      "epoch": 0.43880904814074884,
+      "grad_norm": 0.006931424599151439,
+      "learning_rate": 6.239169351951308e-06,
+      "loss": 0.0,
+      "step": 6809
+    },
+    {
+      "epoch": 0.438873493587678,
+      "grad_norm": 0.06903922706188115,
+      "learning_rate": 6.238453276047262e-06,
+      "loss": 0.0001,
+      "step": 6810
+    },
+    {
+      "epoch": 0.4389379390346072,
+      "grad_norm": 0.0007704492971362574,
+      "learning_rate": 6.237737200143215e-06,
+      "loss": 0.0,
+      "step": 6811
+    },
+    {
+      "epoch": 0.4390023844815364,
+      "grad_norm": 0.4028598105360153,
+      "learning_rate": 6.23702112423917e-06,
+      "loss": 0.0006,
+      "step": 6812
+    },
+    {
+      "epoch": 0.4390668299284656,
+      "grad_norm": 0.008059346311853226,
+      "learning_rate": 6.236305048335124e-06,
+      "loss": 0.0,
+      "step": 6813
+    },
+    {
+      "epoch": 0.4391312753753947,
+      "grad_norm": 0.2791621410064719,
+      "learning_rate": 6.235588972431078e-06,
+      "loss": 0.0007,
+      "step": 6814
+    },
+    {
+      "epoch": 0.4391957208223239,
+      "grad_norm": 0.01712343117021725,
+      "learning_rate": 6.2348728965270325e-06,
+      "loss": 0.0,
+      "step": 6815
+    },
+    {
+      "epoch": 0.43926016626925307,
+      "grad_norm": 0.00298623967276542,
+      "learning_rate": 6.234156820622986e-06,
+      "loss": 0.0,
+      "step": 6816
+    },
+    {
+      "epoch": 0.43932461171618226,
+      "grad_norm": 0.0034664114644248696,
+      "learning_rate": 6.23344074471894e-06,
+      "loss": 0.0,
+      "step": 6817
+    },
+    {
+      "epoch": 0.43938905716311144,
+      "grad_norm": 0.03522052287221309,
+      "learning_rate": 6.232724668814895e-06,
+      "loss": 0.0003,
+      "step": 6818
+    },
+    {
+      "epoch": 0.4394535026100406,
+      "grad_norm": 0.05915418664026023,
+      "learning_rate": 6.232008592910849e-06,
+      "loss": 0.0001,
+      "step": 6819
+    },
+    {
+      "epoch": 0.43951794805696975,
+      "grad_norm": 0.00684155053848704,
+      "learning_rate": 6.231292517006803e-06,
+      "loss": 0.0,
+      "step": 6820
+    },
+    {
+      "epoch": 0.43958239350389894,
+      "grad_norm": 0.034032131900648584,
+      "learning_rate": 6.230576441102757e-06,
+      "loss": 0.0,
+      "step": 6821
+    },
+    {
+      "epoch": 0.4396468389508281,
+      "grad_norm": 0.006321578377653734,
+      "learning_rate": 6.229860365198711e-06,
+      "loss": 0.0,
+      "step": 6822
+    },
+    {
+      "epoch": 0.4397112843977573,
+      "grad_norm": 0.0879231922423166,
+      "learning_rate": 6.229144289294665e-06,
+      "loss": 0.0002,
+      "step": 6823
+    },
+    {
+      "epoch": 0.4397757298446865,
+      "grad_norm": 0.05236788173857305,
+      "learning_rate": 6.22842821339062e-06,
+      "loss": 0.0001,
+      "step": 6824
+    },
+    {
+      "epoch": 0.4398401752916156,
+      "grad_norm": 0.002943995240670316,
+      "learning_rate": 6.227712137486575e-06,
+      "loss": 0.0,
+      "step": 6825
+    },
+    {
+      "epoch": 0.4399046207385448,
+      "grad_norm": 0.022349262684986156,
+      "learning_rate": 6.226996061582529e-06,
+      "loss": 0.0001,
+      "step": 6826
+    },
+    {
+      "epoch": 0.439969066185474,
+      "grad_norm": 0.007797418739776279,
+      "learning_rate": 6.226279985678482e-06,
+      "loss": 0.0,
+      "step": 6827
+    },
+    {
+      "epoch": 0.4400335116324032,
+      "grad_norm": 0.749392223462473,
+      "learning_rate": 6.225563909774437e-06,
+      "loss": 0.0043,
+      "step": 6828
+    },
+    {
+      "epoch": 0.44009795707933236,
+      "grad_norm": 0.014145049204418786,
+      "learning_rate": 6.224847833870391e-06,
+      "loss": 0.0001,
+      "step": 6829
+    },
+    {
+      "epoch": 0.44016240252626154,
+      "grad_norm": 0.22435268993820356,
+      "learning_rate": 6.224131757966345e-06,
+      "loss": 0.0003,
+      "step": 6830
+    },
+    {
+      "epoch": 0.44022684797319067,
+      "grad_norm": 0.019355140889198114,
+      "learning_rate": 6.2234156820623e-06,
+      "loss": 0.0001,
+      "step": 6831
+    },
+    {
+      "epoch": 0.44029129342011986,
+      "grad_norm": 0.15658782559037665,
+      "learning_rate": 6.222699606158253e-06,
+      "loss": 0.0001,
+      "step": 6832
+    },
+    {
+      "epoch": 0.44035573886704904,
+      "grad_norm": 0.010001780505805401,
+      "learning_rate": 6.221983530254207e-06,
+      "loss": 0.0,
+      "step": 6833
+    },
+    {
+      "epoch": 0.4404201843139782,
+      "grad_norm": 0.03181574062344294,
+      "learning_rate": 6.221267454350162e-06,
+      "loss": 0.0,
+      "step": 6834
+    },
+    {
+      "epoch": 0.4404846297609074,
+      "grad_norm": 0.5931863116782081,
+      "learning_rate": 6.220551378446116e-06,
+      "loss": 0.0014,
+      "step": 6835
+    },
+    {
+      "epoch": 0.4405490752078366,
+      "grad_norm": 0.4585510156011477,
+      "learning_rate": 6.21983530254207e-06,
+      "loss": 0.0013,
+      "step": 6836
+    },
+    {
+      "epoch": 0.4406135206547657,
+      "grad_norm": 0.013945956708650028,
+      "learning_rate": 6.219119226638024e-06,
+      "loss": 0.0,
+      "step": 6837
+    },
+    {
+      "epoch": 0.4406779661016949,
+      "grad_norm": 0.023095446256535816,
+      "learning_rate": 6.218403150733978e-06,
+      "loss": 0.0,
+      "step": 6838
+    },
+    {
+      "epoch": 0.4407424115486241,
+      "grad_norm": 0.005353204918259331,
+      "learning_rate": 6.217687074829932e-06,
+      "loss": 0.0,
+      "step": 6839
+    },
+    {
+      "epoch": 0.4408068569955533,
+      "grad_norm": 0.03502575011971063,
+      "learning_rate": 6.216970998925887e-06,
+      "loss": 0.0002,
+      "step": 6840
+    },
+    {
+      "epoch": 0.44087130244248246,
+      "grad_norm": 0.12758947956133784,
+      "learning_rate": 6.216254923021841e-06,
+      "loss": 0.0005,
+      "step": 6841
+    },
+    {
+      "epoch": 0.4409357478894116,
+      "grad_norm": 0.006109814671694446,
+      "learning_rate": 6.215538847117794e-06,
+      "loss": 0.0016,
+      "step": 6842
+    },
+    {
+      "epoch": 0.4410001933363408,
+      "grad_norm": 0.0051795290955148395,
+      "learning_rate": 6.214822771213749e-06,
+      "loss": 0.0,
+      "step": 6843
+    },
+    {
+      "epoch": 0.44106463878326996,
+      "grad_norm": 0.0002441407996229203,
+      "learning_rate": 6.214106695309703e-06,
+      "loss": 0.0,
+      "step": 6844
+    },
+    {
+      "epoch": 0.44112908423019914,
+      "grad_norm": 0.1309906813181773,
+      "learning_rate": 6.213390619405657e-06,
+      "loss": 0.001,
+      "step": 6845
+    },
+    {
+      "epoch": 0.4411935296771283,
+      "grad_norm": 0.0005050727517481047,
+      "learning_rate": 6.2126745435016116e-06,
+      "loss": 0.0,
+      "step": 6846
+    },
+    {
+      "epoch": 0.4412579751240575,
+      "grad_norm": 0.005029458636800389,
+      "learning_rate": 6.211958467597567e-06,
+      "loss": 0.0,
+      "step": 6847
+    },
+    {
+      "epoch": 0.44132242057098664,
+      "grad_norm": 0.015254279423934458,
+      "learning_rate": 6.21124239169352e-06,
+      "loss": 0.0001,
+      "step": 6848
+    },
+    {
+      "epoch": 0.4413868660179158,
+      "grad_norm": 0.00022741485742966934,
+      "learning_rate": 6.2105263157894745e-06,
+      "loss": 0.0,
+      "step": 6849
+    },
+    {
+      "epoch": 0.441451311464845,
+      "grad_norm": 0.22921487814058641,
+      "learning_rate": 6.209810239885429e-06,
+      "loss": 0.0005,
+      "step": 6850
+    },
+    {
+      "epoch": 0.4415157569117742,
+      "grad_norm": 0.0032367808859397958,
+      "learning_rate": 6.209094163981383e-06,
+      "loss": 0.0,
+      "step": 6851
+    },
+    {
+      "epoch": 0.4415802023587034,
+      "grad_norm": 0.7078035888784612,
+      "learning_rate": 6.208378088077337e-06,
+      "loss": 0.0009,
+      "step": 6852
+    },
+    {
+      "epoch": 0.4416446478056325,
+      "grad_norm": 0.0015278402776132136,
+      "learning_rate": 6.207662012173291e-06,
+      "loss": 0.0,
+      "step": 6853
+    },
+    {
+      "epoch": 0.4417090932525617,
+      "grad_norm": 0.005985628339181594,
+      "learning_rate": 6.206945936269245e-06,
+      "loss": 0.0001,
+      "step": 6854
+    },
+    {
+      "epoch": 0.4417735386994909,
+      "grad_norm": 0.08176947446269071,
+      "learning_rate": 6.206229860365199e-06,
+      "loss": 0.0001,
+      "step": 6855
+    },
+    {
+      "epoch": 0.44183798414642006,
+      "grad_norm": 3.5829103841858606e-05,
+      "learning_rate": 6.205513784461154e-06,
+      "loss": 0.0,
+      "step": 6856
+    },
+    {
+      "epoch": 0.44190242959334924,
+      "grad_norm": 0.006482419460032692,
+      "learning_rate": 6.204797708557108e-06,
+      "loss": 0.0,
+      "step": 6857
+    },
+    {
+      "epoch": 0.44196687504027843,
+      "grad_norm": 0.0007300640455878939,
+      "learning_rate": 6.2040816326530614e-06,
+      "loss": 0.0,
+      "step": 6858
+    },
+    {
+      "epoch": 0.44203132048720756,
+      "grad_norm": 0.005625053429722351,
+      "learning_rate": 6.203365556749016e-06,
+      "loss": 0.0,
+      "step": 6859
+    },
+    {
+      "epoch": 0.44209576593413674,
+      "grad_norm": 0.0861411175000196,
+      "learning_rate": 6.20264948084497e-06,
+      "loss": 0.0003,
+      "step": 6860
+    },
+    {
+      "epoch": 0.4421602113810659,
+      "grad_norm": 0.004054258970734921,
+      "learning_rate": 6.201933404940924e-06,
+      "loss": 0.0,
+      "step": 6861
+    },
+    {
+      "epoch": 0.4422246568279951,
+      "grad_norm": 0.009220336750607218,
+      "learning_rate": 6.201217329036879e-06,
+      "loss": 0.0001,
+      "step": 6862
+    },
+    {
+      "epoch": 0.4422891022749243,
+      "grad_norm": 0.011453056003512678,
+      "learning_rate": 6.200501253132832e-06,
+      "loss": 0.0001,
+      "step": 6863
+    },
+    {
+      "epoch": 0.4423535477218534,
+      "grad_norm": 0.0011940250474132966,
+      "learning_rate": 6.199785177228786e-06,
+      "loss": 0.0,
+      "step": 6864
+    },
+    {
+      "epoch": 0.4424179931687826,
+      "grad_norm": 6.666298560305661e-05,
+      "learning_rate": 6.199069101324741e-06,
+      "loss": 0.0,
+      "step": 6865
+    },
+    {
+      "epoch": 0.4424824386157118,
+      "grad_norm": 0.02623876561333548,
+      "learning_rate": 6.198353025420695e-06,
+      "loss": 0.0001,
+      "step": 6866
+    },
+    {
+      "epoch": 0.442546884062641,
+      "grad_norm": 0.04930259728432476,
+      "learning_rate": 6.1976369495166484e-06,
+      "loss": 0.0005,
+      "step": 6867
+    },
+    {
+      "epoch": 0.44261132950957016,
+      "grad_norm": 0.006099219300208031,
+      "learning_rate": 6.196920873612603e-06,
+      "loss": 0.0,
+      "step": 6868
+    },
+    {
+      "epoch": 0.44267577495649935,
+      "grad_norm": 0.0012266434546431698,
+      "learning_rate": 6.196204797708557e-06,
+      "loss": 0.0,
+      "step": 6869
+    },
+    {
+      "epoch": 0.4427402204034285,
+      "grad_norm": 0.9596212412780556,
+      "learning_rate": 6.195488721804512e-06,
+      "loss": 0.0027,
+      "step": 6870
+    },
+    {
+      "epoch": 0.44280466585035766,
+      "grad_norm": 0.004923342711700275,
+      "learning_rate": 6.1947726459004665e-06,
+      "loss": 0.0,
+      "step": 6871
+    },
+    {
+      "epoch": 0.44286911129728684,
+      "grad_norm": 0.001109321347760395,
+      "learning_rate": 6.194056569996421e-06,
+      "loss": 0.0,
+      "step": 6872
+    },
+    {
+      "epoch": 0.44293355674421603,
+      "grad_norm": 0.0483371247384216,
+      "learning_rate": 6.193340494092375e-06,
+      "loss": 0.0002,
+      "step": 6873
+    },
+    {
+      "epoch": 0.4429980021911452,
+      "grad_norm": 0.005611801523459228,
+      "learning_rate": 6.1926244181883285e-06,
+      "loss": 0.0,
+      "step": 6874
+    },
+    {
+      "epoch": 0.4430624476380744,
+      "grad_norm": 0.004807293352621903,
+      "learning_rate": 6.191908342284283e-06,
+      "loss": 0.0,
+      "step": 6875
+    },
+    {
+      "epoch": 0.4431268930850035,
+      "grad_norm": 0.0006191547115299766,
+      "learning_rate": 6.191192266380237e-06,
+      "loss": 0.0,
+      "step": 6876
+    },
+    {
+      "epoch": 0.4431913385319327,
+      "grad_norm": 0.020956838583349454,
+      "learning_rate": 6.1904761904761914e-06,
+      "loss": 0.0,
+      "step": 6877
+    },
+    {
+      "epoch": 0.4432557839788619,
+      "grad_norm": 0.0281989780759363,
+      "learning_rate": 6.189760114572146e-06,
+      "loss": 0.0003,
+      "step": 6878
+    },
+    {
+      "epoch": 0.4433202294257911,
+      "grad_norm": 0.0013422409540706092,
+      "learning_rate": 6.189044038668099e-06,
+      "loss": 0.0,
+      "step": 6879
+    },
+    {
+      "epoch": 0.44338467487272026,
+      "grad_norm": 0.20020187181101248,
+      "learning_rate": 6.1883279627640535e-06,
+      "loss": 0.0001,
+      "step": 6880
+    },
+    {
+      "epoch": 0.4434491203196494,
+      "grad_norm": 0.0016931356119931837,
+      "learning_rate": 6.187611886860008e-06,
+      "loss": 0.0,
+      "step": 6881
+    },
+    {
+      "epoch": 0.4435135657665786,
+      "grad_norm": 0.03510104421217682,
+      "learning_rate": 6.186895810955962e-06,
+      "loss": 0.0001,
+      "step": 6882
+    },
+    {
+      "epoch": 0.44357801121350776,
+      "grad_norm": 0.0005879157791024848,
+      "learning_rate": 6.1861797350519155e-06,
+      "loss": 0.0,
+      "step": 6883
+    },
+    {
+      "epoch": 0.44364245666043695,
+      "grad_norm": 0.015057599267392894,
+      "learning_rate": 6.18546365914787e-06,
+      "loss": 0.0001,
+      "step": 6884
+    },
+    {
+      "epoch": 0.44370690210736613,
+      "grad_norm": 0.004394785597182588,
+      "learning_rate": 6.184747583243824e-06,
+      "loss": 0.0,
+      "step": 6885
+    },
+    {
+      "epoch": 0.4437713475542953,
+      "grad_norm": 0.20035424785335368,
+      "learning_rate": 6.184031507339778e-06,
+      "loss": 0.001,
+      "step": 6886
+    },
+    {
+      "epoch": 0.44383579300122444,
+      "grad_norm": 0.002954961677625816,
+      "learning_rate": 6.183315431435733e-06,
+      "loss": 0.0,
+      "step": 6887
+    },
+    {
+      "epoch": 0.44390023844815363,
+      "grad_norm": 0.14626720861118214,
+      "learning_rate": 6.182599355531686e-06,
+      "loss": 0.0003,
+      "step": 6888
+    },
+    {
+      "epoch": 0.4439646838950828,
+      "grad_norm": 0.004259351645734143,
+      "learning_rate": 6.1818832796276405e-06,
+      "loss": 0.0,
+      "step": 6889
+    },
+    {
+      "epoch": 0.444029129342012,
+      "grad_norm": 0.0006302645757763032,
+      "learning_rate": 6.181167203723595e-06,
+      "loss": 0.0,
+      "step": 6890
+    },
+    {
+      "epoch": 0.4440935747889412,
+      "grad_norm": 0.0005572306616088863,
+      "learning_rate": 6.180451127819549e-06,
+      "loss": 0.0,
+      "step": 6891
+    },
+    {
+      "epoch": 0.4441580202358703,
+      "grad_norm": 0.0032139911123859584,
+      "learning_rate": 6.179735051915503e-06,
+      "loss": 0.0,
+      "step": 6892
+    },
+    {
+      "epoch": 0.4442224656827995,
+      "grad_norm": 0.1437218464876338,
+      "learning_rate": 6.179018976011457e-06,
+      "loss": 0.0004,
+      "step": 6893
+    },
+    {
+      "epoch": 0.4442869111297287,
+      "grad_norm": 0.00045007094295306676,
+      "learning_rate": 6.178302900107413e-06,
+      "loss": 0.0,
+      "step": 6894
+    },
+    {
+      "epoch": 0.44435135657665786,
+      "grad_norm": 0.0007318578584300964,
+      "learning_rate": 6.177586824203366e-06,
+      "loss": 0.0,
+      "step": 6895
+    },
+    {
+      "epoch": 0.44441580202358705,
+      "grad_norm": 0.003502809933033344,
+      "learning_rate": 6.1768707482993206e-06,
+      "loss": 0.0,
+      "step": 6896
+    },
+    {
+      "epoch": 0.44448024747051623,
+      "grad_norm": 0.06071923318971043,
+      "learning_rate": 6.176154672395275e-06,
+      "loss": 0.0001,
+      "step": 6897
+    },
+    {
+      "epoch": 0.44454469291744536,
+      "grad_norm": 0.0004200751505464457,
+      "learning_rate": 6.175438596491229e-06,
+      "loss": 0.0,
+      "step": 6898
+    },
+    {
+      "epoch": 0.44460913836437455,
+      "grad_norm": 0.03086781277013602,
+      "learning_rate": 6.174722520587183e-06,
+      "loss": 0.0001,
+      "step": 6899
+    },
+    {
+      "epoch": 0.44467358381130373,
+      "grad_norm": 0.18499814256496003,
+      "learning_rate": 6.174006444683137e-06,
+      "loss": 0.0007,
+      "step": 6900
+    },
+    {
+      "epoch": 0.4447380292582329,
+      "grad_norm": 0.0012860809106687132,
+      "learning_rate": 6.173290368779091e-06,
+      "loss": 0.0,
+      "step": 6901
+    },
+    {
+      "epoch": 0.4448024747051621,
+      "grad_norm": 0.011116800170123724,
+      "learning_rate": 6.1725742928750455e-06,
+      "loss": 0.0,
+      "step": 6902
+    },
+    {
+      "epoch": 0.4448669201520912,
+      "grad_norm": 0.0315108827698995,
+      "learning_rate": 6.171858216971e-06,
+      "loss": 0.0002,
+      "step": 6903
+    },
+    {
+      "epoch": 0.4449313655990204,
+      "grad_norm": 0.00132297479172432,
+      "learning_rate": 6.171142141066953e-06,
+      "loss": 0.0,
+      "step": 6904
+    },
+    {
+      "epoch": 0.4449958110459496,
+      "grad_norm": 0.06148439258298901,
+      "learning_rate": 6.1704260651629076e-06,
+      "loss": 0.0002,
+      "step": 6905
+    },
+    {
+      "epoch": 0.4450602564928788,
+      "grad_norm": 0.00796708085715272,
+      "learning_rate": 6.169709989258862e-06,
+      "loss": 0.0,
+      "step": 6906
+    },
+    {
+      "epoch": 0.44512470193980797,
+      "grad_norm": 0.21879294518851708,
+      "learning_rate": 6.168993913354816e-06,
+      "loss": 0.0004,
+      "step": 6907
+    },
+    {
+      "epoch": 0.44518914738673715,
+      "grad_norm": 6.952659856934143e-05,
+      "learning_rate": 6.1682778374507705e-06,
+      "loss": 0.0,
+      "step": 6908
+    },
+    {
+      "epoch": 0.4452535928336663,
+      "grad_norm": 0.0029485427620246254,
+      "learning_rate": 6.167561761546724e-06,
+      "loss": 0.0,
+      "step": 6909
+    },
+    {
+      "epoch": 0.44531803828059546,
+      "grad_norm": 0.00095567100834392,
+      "learning_rate": 6.166845685642678e-06,
+      "loss": 0.0,
+      "step": 6910
+    },
+    {
+      "epoch": 0.44538248372752465,
+      "grad_norm": 0.00023555212651686218,
+      "learning_rate": 6.1661296097386325e-06,
+      "loss": 0.0,
+      "step": 6911
+    },
+    {
+      "epoch": 0.44544692917445383,
+      "grad_norm": 0.003228572952052681,
+      "learning_rate": 6.165413533834587e-06,
+      "loss": 0.0,
+      "step": 6912
+    },
+    {
+      "epoch": 0.445511374621383,
+      "grad_norm": 0.008805185668932146,
+      "learning_rate": 6.164697457930541e-06,
+      "loss": 0.0,
+      "step": 6913
+    },
+    {
+      "epoch": 0.4455758200683122,
+      "grad_norm": 0.00209328623167358,
+      "learning_rate": 6.1639813820264945e-06,
+      "loss": 0.0,
+      "step": 6914
+    },
+    {
+      "epoch": 0.44564026551524133,
+      "grad_norm": 4.6677382605160086e-05,
+      "learning_rate": 6.163265306122449e-06,
+      "loss": 0.0,
+      "step": 6915
+    },
+    {
+      "epoch": 0.4457047109621705,
+      "grad_norm": 6.16949104243802e-05,
+      "learning_rate": 6.162549230218403e-06,
+      "loss": 0.0,
+      "step": 6916
+    },
+    {
+      "epoch": 0.4457691564090997,
+      "grad_norm": 0.034629646127696145,
+      "learning_rate": 6.161833154314358e-06,
+      "loss": 0.0001,
+      "step": 6917
+    },
+    {
+      "epoch": 0.4458336018560289,
+      "grad_norm": 0.0003335354281628114,
+      "learning_rate": 6.161117078410313e-06,
+      "loss": 0.0,
+      "step": 6918
+    },
+    {
+      "epoch": 0.44589804730295807,
+      "grad_norm": 0.0004405627311754201,
+      "learning_rate": 6.160401002506267e-06,
+      "loss": 0.0,
+      "step": 6919
+    },
+    {
+      "epoch": 0.4459624927498872,
+      "grad_norm": 0.0006141990923472445,
+      "learning_rate": 6.15968492660222e-06,
+      "loss": 0.0,
+      "step": 6920
+    },
+    {
+      "epoch": 0.4460269381968164,
+      "grad_norm": 0.004484797434344207,
+      "learning_rate": 6.158968850698175e-06,
+      "loss": 0.0001,
+      "step": 6921
+    },
+    {
+      "epoch": 0.44609138364374556,
+      "grad_norm": 0.0006767052006918171,
+      "learning_rate": 6.158252774794129e-06,
+      "loss": 0.0,
+      "step": 6922
+    },
+    {
+      "epoch": 0.44615582909067475,
+      "grad_norm": 0.0006632304789780592,
+      "learning_rate": 6.157536698890083e-06,
+      "loss": 0.0,
+      "step": 6923
+    },
+    {
+      "epoch": 0.44622027453760393,
+      "grad_norm": 0.6850470870255448,
+      "learning_rate": 6.1568206229860375e-06,
+      "loss": 0.002,
+      "step": 6924
+    },
+    {
+      "epoch": 0.4462847199845331,
+      "grad_norm": 0.01536181478069005,
+      "learning_rate": 6.156104547081991e-06,
+      "loss": 0.0,
+      "step": 6925
+    },
+    {
+      "epoch": 0.44634916543146225,
+      "grad_norm": 0.000819613932033979,
+      "learning_rate": 6.155388471177945e-06,
+      "loss": 0.0,
+      "step": 6926
+    },
+    {
+      "epoch": 0.44641361087839143,
+      "grad_norm": 0.0013558397768012507,
+      "learning_rate": 6.1546723952739e-06,
+      "loss": 0.0,
+      "step": 6927
+    },
+    {
+      "epoch": 0.4464780563253206,
+      "grad_norm": 0.00011668090304934545,
+      "learning_rate": 6.153956319369854e-06,
+      "loss": 0.0,
+      "step": 6928
+    },
+    {
+      "epoch": 0.4465425017722498,
+      "grad_norm": 0.02503661579297557,
+      "learning_rate": 6.153240243465808e-06,
+      "loss": 0.0002,
+      "step": 6929
+    },
+    {
+      "epoch": 0.446606947219179,
+      "grad_norm": 0.003287645970191306,
+      "learning_rate": 6.152524167561762e-06,
+      "loss": 0.0,
+      "step": 6930
+    },
+    {
+      "epoch": 0.4466713926661081,
+      "grad_norm": 0.003287645970191306,
+      "learning_rate": 6.152524167561762e-06,
+      "loss": 0.0059,
+      "step": 6931
+    },
+    {
+      "epoch": 0.4467358381130373,
+      "grad_norm": 0.0013659329784108567,
+      "learning_rate": 6.151808091657716e-06,
+      "loss": 0.0,
+      "step": 6932
+    },
+    {
+      "epoch": 0.4468002835599665,
+      "grad_norm": 0.02730988103668324,
+      "learning_rate": 6.15109201575367e-06,
+      "loss": 0.0002,
+      "step": 6933
+    },
+    {
+      "epoch": 0.44686472900689567,
+      "grad_norm": 0.0002492699167945417,
+      "learning_rate": 6.1503759398496245e-06,
+      "loss": 0.0,
+      "step": 6934
+    },
+    {
+      "epoch": 0.44692917445382485,
+      "grad_norm": 0.0004506669312678534,
+      "learning_rate": 6.149659863945579e-06,
+      "loss": 0.0,
+      "step": 6935
+    },
+    {
+      "epoch": 0.44699361990075404,
+      "grad_norm": 0.8233311651435545,
+      "learning_rate": 6.148943788041532e-06,
+      "loss": 0.0042,
+      "step": 6936
+    },
+    {
+      "epoch": 0.44705806534768316,
+      "grad_norm": 0.00037711894076924535,
+      "learning_rate": 6.148227712137487e-06,
+      "loss": 0.0,
+      "step": 6937
+    },
+    {
+      "epoch": 0.44712251079461235,
+      "grad_norm": 0.022929007680163573,
+      "learning_rate": 6.147511636233441e-06,
+      "loss": 0.0001,
+      "step": 6938
+    },
+    {
+      "epoch": 0.44718695624154153,
+      "grad_norm": 0.03777062891841532,
+      "learning_rate": 6.146795560329395e-06,
+      "loss": 0.0001,
+      "step": 6939
+    },
+    {
+      "epoch": 0.4472514016884707,
+      "grad_norm": 0.1337892295669292,
+      "learning_rate": 6.1460794844253495e-06,
+      "loss": 0.0004,
+      "step": 6940
+    },
+    {
+      "epoch": 0.4473158471353999,
+      "grad_norm": 0.024741520669856795,
+      "learning_rate": 6.145363408521305e-06,
+      "loss": 0.0001,
+      "step": 6941
+    },
+    {
+      "epoch": 0.44738029258232903,
+      "grad_norm": 0.006022601529771838,
+      "learning_rate": 6.144647332617258e-06,
+      "loss": 0.0,
+      "step": 6942
+    },
+    {
+      "epoch": 0.4474447380292582,
+      "grad_norm": 0.03185848800646121,
+      "learning_rate": 6.143931256713212e-06,
+      "loss": 0.0003,
+      "step": 6943
+    },
+    {
+      "epoch": 0.4475091834761874,
+      "grad_norm": 0.04492633735220352,
+      "learning_rate": 6.143215180809167e-06,
+      "loss": 0.0005,
+      "step": 6944
+    },
+    {
+      "epoch": 0.4475736289231166,
+      "grad_norm": 0.0009602092421418444,
+      "learning_rate": 6.142499104905121e-06,
+      "loss": 0.0,
+      "step": 6945
+    },
+    {
+      "epoch": 0.44763807437004577,
+      "grad_norm": 0.00023928778128441458,
+      "learning_rate": 6.141783029001075e-06,
+      "loss": 0.0,
+      "step": 6946
+    },
+    {
+      "epoch": 0.44770251981697495,
+      "grad_norm": 0.09243969869578006,
+      "learning_rate": 6.141066953097029e-06,
+      "loss": 0.0001,
+      "step": 6947
+    },
+    {
+      "epoch": 0.4477669652639041,
+      "grad_norm": 0.020509047130477586,
+      "learning_rate": 6.140350877192983e-06,
+      "loss": 0.0002,
+      "step": 6948
+    },
+    {
+      "epoch": 0.44783141071083327,
+      "grad_norm": 0.5742998130825934,
+      "learning_rate": 6.139634801288937e-06,
+      "loss": 0.0026,
+      "step": 6949
+    },
+    {
+      "epoch": 0.44789585615776245,
+      "grad_norm": 0.0006601064763192048,
+      "learning_rate": 6.138918725384892e-06,
+      "loss": 0.0,
+      "step": 6950
+    },
+    {
+      "epoch": 0.44796030160469164,
+      "grad_norm": 0.4013238219805938,
+      "learning_rate": 6.138202649480846e-06,
+      "loss": 0.0024,
+      "step": 6951
+    },
+    {
+      "epoch": 0.4480247470516208,
+      "grad_norm": 0.19748266008165033,
+      "learning_rate": 6.137486573576799e-06,
+      "loss": 0.0003,
+      "step": 6952
+    },
+    {
+      "epoch": 0.44808919249855,
+      "grad_norm": 0.005614501578779723,
+      "learning_rate": 6.136770497672754e-06,
+      "loss": 0.0,
+      "step": 6953
+    },
+    {
+      "epoch": 0.44815363794547913,
+      "grad_norm": 0.0018398338943028784,
+      "learning_rate": 6.136054421768708e-06,
+      "loss": 0.0,
+      "step": 6954
+    },
+    {
+      "epoch": 0.4482180833924083,
+      "grad_norm": 0.006431620033200353,
+      "learning_rate": 6.135338345864662e-06,
+      "loss": 0.0,
+      "step": 6955
+    },
+    {
+      "epoch": 0.4482825288393375,
+      "grad_norm": 5.3012932827666375e-05,
+      "learning_rate": 6.1346222699606166e-06,
+      "loss": 0.0,
+      "step": 6956
+    },
+    {
+      "epoch": 0.4483469742862667,
+      "grad_norm": 0.0054546841273027434,
+      "learning_rate": 6.13390619405657e-06,
+      "loss": 0.0,
+      "step": 6957
+    },
+    {
+      "epoch": 0.44841141973319587,
+      "grad_norm": 0.008955214094707408,
+      "learning_rate": 6.133190118152524e-06,
+      "loss": 0.0,
+      "step": 6958
+    },
+    {
+      "epoch": 0.448475865180125,
+      "grad_norm": 0.012356203690497503,
+      "learning_rate": 6.132474042248479e-06,
+      "loss": 0.0,
+      "step": 6959
+    },
+    {
+      "epoch": 0.4485403106270542,
+      "grad_norm": 0.38077338788568676,
+      "learning_rate": 6.131757966344433e-06,
+      "loss": 0.0019,
+      "step": 6960
+    },
+    {
+      "epoch": 0.44860475607398337,
+      "grad_norm": 0.3053363835218518,
+      "learning_rate": 6.131041890440386e-06,
+      "loss": 0.0017,
+      "step": 6961
+    },
+    {
+      "epoch": 0.44866920152091255,
+      "grad_norm": 0.0024220880437371517,
+      "learning_rate": 6.130325814536341e-06,
+      "loss": 0.0,
+      "step": 6962
+    },
+    {
+      "epoch": 0.44873364696784174,
+      "grad_norm": 0.0015772679082248853,
+      "learning_rate": 6.129609738632295e-06,
+      "loss": 0.0,
+      "step": 6963
+    },
+    {
+      "epoch": 0.4487980924147709,
+      "grad_norm": 0.00033190575499703344,
+      "learning_rate": 6.128893662728249e-06,
+      "loss": 0.0,
+      "step": 6964
+    },
+    {
+      "epoch": 0.44886253786170005,
+      "grad_norm": 0.00012355856685873574,
+      "learning_rate": 6.128177586824204e-06,
+      "loss": 0.0,
+      "step": 6965
+    },
+    {
+      "epoch": 0.44892698330862923,
+      "grad_norm": 0.0008402414323257852,
+      "learning_rate": 6.127461510920159e-06,
+      "loss": 0.0,
+      "step": 6966
+    },
+    {
+      "epoch": 0.4489914287555584,
+      "grad_norm": 0.3629500817741936,
+      "learning_rate": 6.126745435016113e-06,
+      "loss": 0.0032,
+      "step": 6967
+    },
+    {
+      "epoch": 0.4490558742024876,
+      "grad_norm": 2.327572136077638,
+      "learning_rate": 6.1260293591120665e-06,
+      "loss": 0.009,
+      "step": 6968
+    },
+    {
+      "epoch": 0.4491203196494168,
+      "grad_norm": 0.02112311101313786,
+      "learning_rate": 6.125313283208021e-06,
+      "loss": 0.0002,
+      "step": 6969
+    },
+    {
+      "epoch": 0.4491847650963459,
+      "grad_norm": 0.014130007366206468,
+      "learning_rate": 6.124597207303975e-06,
+      "loss": 0.0,
+      "step": 6970
+    },
+    {
+      "epoch": 0.4492492105432751,
+      "grad_norm": 0.0053104441258905654,
+      "learning_rate": 6.123881131399929e-06,
+      "loss": 0.0001,
+      "step": 6971
+    },
+    {
+      "epoch": 0.4493136559902043,
+      "grad_norm": 0.10688876165916378,
+      "learning_rate": 6.123165055495884e-06,
+      "loss": 0.0002,
+      "step": 6972
+    },
+    {
+      "epoch": 0.44937810143713347,
+      "grad_norm": 0.004312108841936607,
+      "learning_rate": 6.122448979591837e-06,
+      "loss": 0.0,
+      "step": 6973
+    },
+    {
+      "epoch": 0.44944254688406265,
+      "grad_norm": 0.0020536256520046975,
+      "learning_rate": 6.121732903687791e-06,
+      "loss": 0.0,
+      "step": 6974
+    },
+    {
+      "epoch": 0.44950699233099184,
+      "grad_norm": 0.4778005434065899,
+      "learning_rate": 6.121016827783746e-06,
+      "loss": 0.0004,
+      "step": 6975
+    },
+    {
+      "epoch": 0.44957143777792097,
+      "grad_norm": 0.3517645890823827,
+      "learning_rate": 6.1203007518797e-06,
+      "loss": 0.0019,
+      "step": 6976
+    },
+    {
+      "epoch": 0.44963588322485015,
+      "grad_norm": 0.00961769649574364,
+      "learning_rate": 6.1195846759756534e-06,
+      "loss": 0.0,
+      "step": 6977
+    },
+    {
+      "epoch": 0.44970032867177934,
+      "grad_norm": 0.02834457866455974,
+      "learning_rate": 6.118868600071608e-06,
+      "loss": 0.0,
+      "step": 6978
+    },
+    {
+      "epoch": 0.4497647741187085,
+      "grad_norm": 0.3496750604736784,
+      "learning_rate": 6.118152524167562e-06,
+      "loss": 0.0005,
+      "step": 6979
+    },
+    {
+      "epoch": 0.4498292195656377,
+      "grad_norm": 0.5185237666966998,
+      "learning_rate": 6.117436448263516e-06,
+      "loss": 0.0009,
+      "step": 6980
+    },
+    {
+      "epoch": 0.44989366501256683,
+      "grad_norm": 0.0019981095150621467,
+      "learning_rate": 6.116720372359471e-06,
+      "loss": 0.0,
+      "step": 6981
+    },
+    {
+      "epoch": 0.449958110459496,
+      "grad_norm": 0.001970621272454851,
+      "learning_rate": 6.116004296455424e-06,
+      "loss": 0.0,
+      "step": 6982
+    },
+    {
+      "epoch": 0.4500225559064252,
+      "grad_norm": 0.14547553760578696,
+      "learning_rate": 6.115288220551378e-06,
+      "loss": 0.0003,
+      "step": 6983
+    },
+    {
+      "epoch": 0.4500870013533544,
+      "grad_norm": 0.04114063122287838,
+      "learning_rate": 6.114572144647333e-06,
+      "loss": 0.0,
+      "step": 6984
+    },
+    {
+      "epoch": 0.4501514468002836,
+      "grad_norm": 0.007233528345344776,
+      "learning_rate": 6.113856068743287e-06,
+      "loss": 0.0,
+      "step": 6985
+    },
+    {
+      "epoch": 0.45021589224721276,
+      "grad_norm": 0.0026296207033014393,
+      "learning_rate": 6.113139992839241e-06,
+      "loss": 0.0,
+      "step": 6986
+    },
+    {
+      "epoch": 0.4502803376941419,
+      "grad_norm": 0.03319251543124218,
+      "learning_rate": 6.112423916935195e-06,
+      "loss": 0.0002,
+      "step": 6987
+    },
+    {
+      "epoch": 0.45034478314107107,
+      "grad_norm": 0.0024029586808359893,
+      "learning_rate": 6.111707841031151e-06,
+      "loss": 0.0,
+      "step": 6988
+    },
+    {
+      "epoch": 0.45040922858800025,
+      "grad_norm": 0.001096244893310349,
+      "learning_rate": 6.110991765127104e-06,
+      "loss": 0.0,
+      "step": 6989
+    },
+    {
+      "epoch": 0.45047367403492944,
+      "grad_norm": 0.47845158051066183,
+      "learning_rate": 6.1102756892230585e-06,
+      "loss": 0.0026,
+      "step": 6990
+    },
+    {
+      "epoch": 0.4505381194818586,
+      "grad_norm": 0.005456332068239849,
+      "learning_rate": 6.109559613319013e-06,
+      "loss": 0.0001,
+      "step": 6991
+    },
+    {
+      "epoch": 0.4506025649287878,
+      "grad_norm": 0.023044872550530067,
+      "learning_rate": 6.108843537414967e-06,
+      "loss": 0.0,
+      "step": 6992
+    },
+    {
+      "epoch": 0.45066701037571694,
+      "grad_norm": 9.10151264291464e-05,
+      "learning_rate": 6.1081274615109205e-06,
+      "loss": 0.0,
+      "step": 6993
+    },
+    {
+      "epoch": 0.4507314558226461,
+      "grad_norm": 0.0011771712542816663,
+      "learning_rate": 6.107411385606875e-06,
+      "loss": 0.0,
+      "step": 6994
+    },
+    {
+      "epoch": 0.4507959012695753,
+      "grad_norm": 0.004017593902929894,
+      "learning_rate": 6.106695309702829e-06,
+      "loss": 0.0,
+      "step": 6995
+    },
+    {
+      "epoch": 0.4508603467165045,
+      "grad_norm": 0.16323827150517303,
+      "learning_rate": 6.1059792337987834e-06,
+      "loss": 0.0008,
+      "step": 6996
+    },
+    {
+      "epoch": 0.4509247921634337,
+      "grad_norm": 0.027595080223723742,
+      "learning_rate": 6.105263157894738e-06,
+      "loss": 0.0,
+      "step": 6997
+    },
+    {
+      "epoch": 0.4509892376103628,
+      "grad_norm": 0.0021043324610431602,
+      "learning_rate": 6.104547081990691e-06,
+      "loss": 0.0,
+      "step": 6998
+    },
+    {
+      "epoch": 0.451053683057292,
+      "grad_norm": 0.03236437900145626,
+      "learning_rate": 6.1038310060866455e-06,
+      "loss": 0.0001,
+      "step": 6999
+    },
+    {
+      "epoch": 0.45111812850422117,
+      "grad_norm": 0.4955685155462287,
+      "learning_rate": 6.1031149301826e-06,
+      "loss": 0.0036,
+      "step": 7000
+    },
+    {
+      "epoch": 0.45118257395115036,
+      "grad_norm": 0.0005264324608317756,
+      "learning_rate": 6.102398854278554e-06,
+      "loss": 0.0,
+      "step": 7001
+    },
+    {
+      "epoch": 0.45124701939807954,
+      "grad_norm": 0.03288913141526833,
+      "learning_rate": 6.101682778374508e-06,
+      "loss": 0.0001,
+      "step": 7002
+    },
+    {
+      "epoch": 0.4513114648450087,
+      "grad_norm": 0.001246752781181037,
+      "learning_rate": 6.100966702470462e-06,
+      "loss": 0.0015,
+      "step": 7003
+    },
+    {
+      "epoch": 0.45137591029193785,
+      "grad_norm": 0.0005902236479268267,
+      "learning_rate": 6.100250626566416e-06,
+      "loss": 0.0,
+      "step": 7004
+    },
+    {
+      "epoch": 0.45144035573886704,
+      "grad_norm": 0.2308556251414973,
+      "learning_rate": 6.09953455066237e-06,
+      "loss": 0.0011,
+      "step": 7005
+    },
+    {
+      "epoch": 0.4515048011857962,
+      "grad_norm": 0.0017562515855380186,
+      "learning_rate": 6.098818474758325e-06,
+      "loss": 0.0,
+      "step": 7006
+    },
+    {
+      "epoch": 0.4515692466327254,
+      "grad_norm": 0.01532581818107362,
+      "learning_rate": 6.098102398854279e-06,
+      "loss": 0.0,
+      "step": 7007
+    },
+    {
+      "epoch": 0.4516336920796546,
+      "grad_norm": 0.08927833375484105,
+      "learning_rate": 6.0973863229502325e-06,
+      "loss": 0.0003,
+      "step": 7008
+    },
+    {
+      "epoch": 0.4516981375265837,
+      "grad_norm": 0.0284605009468016,
+      "learning_rate": 6.096670247046187e-06,
+      "loss": 0.0,
+      "step": 7009
+    },
+    {
+      "epoch": 0.4517625829735129,
+      "grad_norm": 0.0016943289754871211,
+      "learning_rate": 6.095954171142141e-06,
+      "loss": 0.0,
+      "step": 7010
+    },
+    {
+      "epoch": 0.4518270284204421,
+      "grad_norm": 0.02218787044607103,
+      "learning_rate": 6.095238095238096e-06,
+      "loss": 0.0,
+      "step": 7011
+    },
+    {
+      "epoch": 0.4518914738673713,
+      "grad_norm": 0.014370586866855587,
+      "learning_rate": 6.0945220193340505e-06,
+      "loss": 0.0001,
+      "step": 7012
+    },
+    {
+      "epoch": 0.45195591931430046,
+      "grad_norm": 0.0029353212413927815,
+      "learning_rate": 6.093805943430005e-06,
+      "loss": 0.0,
+      "step": 7013
+    },
+    {
+      "epoch": 0.45202036476122964,
+      "grad_norm": 0.009811081740184684,
+      "learning_rate": 6.093089867525958e-06,
+      "loss": 0.0,
+      "step": 7014
+    },
+    {
+      "epoch": 0.45208481020815877,
+      "grad_norm": 0.05393490263073934,
+      "learning_rate": 6.0923737916219126e-06,
+      "loss": 0.0002,
+      "step": 7015
+    },
+    {
+      "epoch": 0.45214925565508796,
+      "grad_norm": 2.5784644625886863,
+      "learning_rate": 6.091657715717867e-06,
+      "loss": 0.0189,
+      "step": 7016
+    },
+    {
+      "epoch": 0.45221370110201714,
+      "grad_norm": 0.0029356539960954304,
+      "learning_rate": 6.090941639813821e-06,
+      "loss": 0.0,
+      "step": 7017
+    },
+    {
+      "epoch": 0.4522781465489463,
+      "grad_norm": 0.00022978794218516798,
+      "learning_rate": 6.0902255639097755e-06,
+      "loss": 0.0,
+      "step": 7018
+    },
+    {
+      "epoch": 0.4523425919958755,
+      "grad_norm": 0.029644232123119035,
+      "learning_rate": 6.089509488005729e-06,
+      "loss": 0.0002,
+      "step": 7019
+    },
+    {
+      "epoch": 0.45240703744280464,
+      "grad_norm": 0.0034083100602888576,
+      "learning_rate": 6.088793412101683e-06,
+      "loss": 0.0,
+      "step": 7020
+    },
+    {
+      "epoch": 0.4524714828897338,
+      "grad_norm": 0.004447110035412839,
+      "learning_rate": 6.0880773361976375e-06,
+      "loss": 0.0,
+      "step": 7021
+    },
+    {
+      "epoch": 0.452535928336663,
+      "grad_norm": 0.13061681728651678,
+      "learning_rate": 6.087361260293592e-06,
+      "loss": 0.0003,
+      "step": 7022
+    },
+    {
+      "epoch": 0.4526003737835922,
+      "grad_norm": 0.03177697500649157,
+      "learning_rate": 6.086645184389546e-06,
+      "loss": 0.0,
+      "step": 7023
+    },
+    {
+      "epoch": 0.4526648192305214,
+      "grad_norm": 0.06309334915132048,
+      "learning_rate": 6.0859291084854996e-06,
+      "loss": 0.0003,
+      "step": 7024
+    },
+    {
+      "epoch": 0.45272926467745056,
+      "grad_norm": 0.04666786572863354,
+      "learning_rate": 6.085213032581454e-06,
+      "loss": 0.0001,
+      "step": 7025
+    },
+    {
+      "epoch": 0.4527937101243797,
+      "grad_norm": 0.030821576426459607,
+      "learning_rate": 6.084496956677408e-06,
+      "loss": 0.0001,
+      "step": 7026
+    },
+    {
+      "epoch": 0.4528581555713089,
+      "grad_norm": 0.010487922371730633,
+      "learning_rate": 6.0837808807733625e-06,
+      "loss": 0.0,
+      "step": 7027
+    },
+    {
+      "epoch": 0.45292260101823806,
+      "grad_norm": 0.2080304755817776,
+      "learning_rate": 6.083064804869317e-06,
+      "loss": 0.0021,
+      "step": 7028
+    },
+    {
+      "epoch": 0.45298704646516724,
+      "grad_norm": 0.009692660458068866,
+      "learning_rate": 6.08234872896527e-06,
+      "loss": 0.0016,
+      "step": 7029
+    },
+    {
+      "epoch": 0.4530514919120964,
+      "grad_norm": 0.004954783683027185,
+      "learning_rate": 6.0816326530612245e-06,
+      "loss": 0.0,
+      "step": 7030
+    },
+    {
+      "epoch": 0.4531159373590256,
+      "grad_norm": 0.003751176911784929,
+      "learning_rate": 6.080916577157179e-06,
+      "loss": 0.0,
+      "step": 7031
+    },
+    {
+      "epoch": 0.45318038280595474,
+      "grad_norm": 0.04467092785293174,
+      "learning_rate": 6.080200501253133e-06,
+      "loss": 0.0004,
+      "step": 7032
+    },
+    {
+      "epoch": 0.4532448282528839,
+      "grad_norm": 0.20824716792808376,
+      "learning_rate": 6.079484425349087e-06,
+      "loss": 0.0005,
+      "step": 7033
+    },
+    {
+      "epoch": 0.4533092736998131,
+      "grad_norm": 7.700961625131927e-05,
+      "learning_rate": 6.078768349445041e-06,
+      "loss": 0.0,
+      "step": 7034
+    },
+    {
+      "epoch": 0.4533737191467423,
+      "grad_norm": 0.03432253720668772,
+      "learning_rate": 6.078052273540996e-06,
+      "loss": 0.0,
+      "step": 7035
+    },
+    {
+      "epoch": 0.4534381645936715,
+      "grad_norm": 0.00040768886861480716,
+      "learning_rate": 6.07733619763695e-06,
+      "loss": 0.0,
+      "step": 7036
+    },
+    {
+      "epoch": 0.4535026100406006,
+      "grad_norm": 0.009148011904105437,
+      "learning_rate": 6.076620121732905e-06,
+      "loss": 0.0,
+      "step": 7037
+    },
+    {
+      "epoch": 0.4535670554875298,
+      "grad_norm": 0.42105234257474816,
+      "learning_rate": 6.075904045828859e-06,
+      "loss": 0.0032,
+      "step": 7038
+    },
+    {
+      "epoch": 0.453631500934459,
+      "grad_norm": 0.003082251194585121,
+      "learning_rate": 6.075187969924813e-06,
+      "loss": 0.0,
+      "step": 7039
+    },
+    {
+      "epoch": 0.45369594638138816,
+      "grad_norm": 0.0007968014546186246,
+      "learning_rate": 6.074471894020767e-06,
+      "loss": 0.0,
+      "step": 7040
+    },
+    {
+      "epoch": 0.45376039182831734,
+      "grad_norm": 0.0013500992183111478,
+      "learning_rate": 6.073755818116721e-06,
+      "loss": 0.0,
+      "step": 7041
+    },
+    {
+      "epoch": 0.45382483727524653,
+      "grad_norm": 0.020758522620687073,
+      "learning_rate": 6.073039742212675e-06,
+      "loss": 0.0001,
+      "step": 7042
+    },
+    {
+      "epoch": 0.45388928272217566,
+      "grad_norm": 0.003780350314592039,
+      "learning_rate": 6.0723236663086295e-06,
+      "loss": 0.0,
+      "step": 7043
+    },
+    {
+      "epoch": 0.45395372816910484,
+      "grad_norm": 0.0025714049356480063,
+      "learning_rate": 6.071607590404584e-06,
+      "loss": 0.0,
+      "step": 7044
+    },
+    {
+      "epoch": 0.454018173616034,
+      "grad_norm": 0.000688096630702408,
+      "learning_rate": 6.070891514500537e-06,
+      "loss": 0.0,
+      "step": 7045
+    },
+    {
+      "epoch": 0.4540826190629632,
+      "grad_norm": 0.0006434500711874743,
+      "learning_rate": 6.070175438596492e-06,
+      "loss": 0.0,
+      "step": 7046
+    },
+    {
+      "epoch": 0.4541470645098924,
+      "grad_norm": 0.35732653654378393,
+      "learning_rate": 6.069459362692446e-06,
+      "loss": 0.0023,
+      "step": 7047
+    },
+    {
+      "epoch": 0.4542115099568215,
+      "grad_norm": 0.01280457731163026,
+      "learning_rate": 6.0687432867884e-06,
+      "loss": 0.0,
+      "step": 7048
+    },
+    {
+      "epoch": 0.4542759554037507,
+      "grad_norm": 0.1197270360691481,
+      "learning_rate": 6.0680272108843545e-06,
+      "loss": 0.0004,
+      "step": 7049
+    },
+    {
+      "epoch": 0.4543404008506799,
+      "grad_norm": 0.0010014052039343515,
+      "learning_rate": 6.067311134980308e-06,
+      "loss": 0.0,
+      "step": 7050
+    },
+    {
+      "epoch": 0.4544048462976091,
+      "grad_norm": 0.03635723904725859,
+      "learning_rate": 6.066595059076262e-06,
+      "loss": 0.0002,
+      "step": 7051
+    },
+    {
+      "epoch": 0.45446929174453826,
+      "grad_norm": 0.00575358947033153,
+      "learning_rate": 6.0658789831722165e-06,
+      "loss": 0.0,
+      "step": 7052
+    },
+    {
+      "epoch": 0.45453373719146745,
+      "grad_norm": 0.013448601461061216,
+      "learning_rate": 6.065162907268171e-06,
+      "loss": 0.0001,
+      "step": 7053
+    },
+    {
+      "epoch": 0.4545981826383966,
+      "grad_norm": 0.00041888456590139113,
+      "learning_rate": 6.064446831364125e-06,
+      "loss": 0.0,
+      "step": 7054
+    },
+    {
+      "epoch": 0.45466262808532576,
+      "grad_norm": 0.002986199622612587,
+      "learning_rate": 6.063730755460079e-06,
+      "loss": 0.0,
+      "step": 7055
+    },
+    {
+      "epoch": 0.45472707353225494,
+      "grad_norm": 0.0008021949438291668,
+      "learning_rate": 6.063014679556033e-06,
+      "loss": 0.0,
+      "step": 7056
+    },
+    {
+      "epoch": 0.45479151897918413,
+      "grad_norm": 0.3501131487920616,
+      "learning_rate": 6.062298603651987e-06,
+      "loss": 0.0012,
+      "step": 7057
+    },
+    {
+      "epoch": 0.4548559644261133,
+      "grad_norm": 0.001603722886898084,
+      "learning_rate": 6.061582527747942e-06,
+      "loss": 0.0,
+      "step": 7058
+    },
+    {
+      "epoch": 0.4549204098730425,
+      "grad_norm": 0.1883643315255252,
+      "learning_rate": 6.060866451843897e-06,
+      "loss": 0.0003,
+      "step": 7059
+    },
+    {
+      "epoch": 0.4549848553199716,
+      "grad_norm": 0.0004600154112903025,
+      "learning_rate": 6.060150375939851e-06,
+      "loss": 0.0,
+      "step": 7060
+    },
+    {
+      "epoch": 0.4550493007669008,
+      "grad_norm": 0.008173502769027983,
+      "learning_rate": 6.059434300035804e-06,
+      "loss": 0.0,
+      "step": 7061
+    },
+    {
+      "epoch": 0.45511374621383,
+      "grad_norm": 0.02577972412109375,
+      "learning_rate": 6.058718224131759e-06,
+      "loss": 0.0001,
+      "step": 7062
+    },
+    {
+      "epoch": 0.4551781916607592,
+      "grad_norm": 0.010985050193318776,
+      "learning_rate": 6.058002148227713e-06,
+      "loss": 0.0,
+      "step": 7063
+    },
+    {
+      "epoch": 0.45524263710768836,
+      "grad_norm": 0.00025358798746525133,
+      "learning_rate": 6.057286072323667e-06,
+      "loss": 0.0,
+      "step": 7064
+    },
+    {
+      "epoch": 0.4553070825546175,
+      "grad_norm": 0.001464238886327982,
+      "learning_rate": 6.0565699964196216e-06,
+      "loss": 0.0,
+      "step": 7065
+    },
+    {
+      "epoch": 0.4553715280015467,
+      "grad_norm": 0.018947541797622022,
+      "learning_rate": 6.055853920515575e-06,
+      "loss": 0.0001,
+      "step": 7066
+    },
+    {
+      "epoch": 0.45543597344847586,
+      "grad_norm": 0.0005999839663420513,
+      "learning_rate": 6.055137844611529e-06,
+      "loss": 0.0,
+      "step": 7067
+    },
+    {
+      "epoch": 0.45550041889540505,
+      "grad_norm": 0.1648804562727237,
+      "learning_rate": 6.054421768707484e-06,
+      "loss": 0.0012,
+      "step": 7068
+    },
+    {
+      "epoch": 0.45556486434233423,
+      "grad_norm": 0.5202259495779435,
+      "learning_rate": 6.053705692803438e-06,
+      "loss": 0.0021,
+      "step": 7069
+    },
+    {
+      "epoch": 0.4556293097892634,
+      "grad_norm": 0.003008793348343684,
+      "learning_rate": 6.052989616899392e-06,
+      "loss": 0.0,
+      "step": 7070
+    },
+    {
+      "epoch": 0.45569375523619254,
+      "grad_norm": 0.07646857151479122,
+      "learning_rate": 6.052273540995346e-06,
+      "loss": 0.0002,
+      "step": 7071
+    },
+    {
+      "epoch": 0.45575820068312173,
+      "grad_norm": 0.2913056023367716,
+      "learning_rate": 6.0515574650913e-06,
+      "loss": 0.002,
+      "step": 7072
+    },
+    {
+      "epoch": 0.4558226461300509,
+      "grad_norm": 0.09931329124014251,
+      "learning_rate": 6.050841389187254e-06,
+      "loss": 0.0005,
+      "step": 7073
+    },
+    {
+      "epoch": 0.4558870915769801,
+      "grad_norm": 0.2881209788590056,
+      "learning_rate": 6.0501253132832086e-06,
+      "loss": 0.0089,
+      "step": 7074
+    },
+    {
+      "epoch": 0.4559515370239093,
+      "grad_norm": 0.00044719869471577915,
+      "learning_rate": 6.049409237379162e-06,
+      "loss": 0.0,
+      "step": 7075
+    },
+    {
+      "epoch": 0.4560159824708384,
+      "grad_norm": 0.011173389599324287,
+      "learning_rate": 6.048693161475116e-06,
+      "loss": 0.0,
+      "step": 7076
+    },
+    {
+      "epoch": 0.4560804279177676,
+      "grad_norm": 0.0010514280422994013,
+      "learning_rate": 6.047977085571071e-06,
+      "loss": 0.0,
+      "step": 7077
+    },
+    {
+      "epoch": 0.4561448733646968,
+      "grad_norm": 0.0006105020922317698,
+      "learning_rate": 6.047261009667025e-06,
+      "loss": 0.0,
+      "step": 7078
+    },
+    {
+      "epoch": 0.45620931881162596,
+      "grad_norm": 0.20810970980521848,
+      "learning_rate": 6.046544933762979e-06,
+      "loss": 0.0003,
+      "step": 7079
+    },
+    {
+      "epoch": 0.45627376425855515,
+      "grad_norm": 0.00014491230861546294,
+      "learning_rate": 6.045828857858933e-06,
+      "loss": 0.0,
+      "step": 7080
+    },
+    {
+      "epoch": 0.45633820970548433,
+      "grad_norm": 0.0008180527801640292,
+      "learning_rate": 6.045112781954889e-06,
+      "loss": 0.0,
+      "step": 7081
+    },
+    {
+      "epoch": 0.45640265515241346,
+      "grad_norm": 0.015151065614577604,
+      "learning_rate": 6.044396706050842e-06,
+      "loss": 0.0,
+      "step": 7082
+    },
+    {
+      "epoch": 0.45646710059934265,
+      "grad_norm": 0.00016913031952718266,
+      "learning_rate": 6.043680630146796e-06,
+      "loss": 0.0,
+      "step": 7083
+    },
+    {
+      "epoch": 0.45653154604627183,
+      "grad_norm": 0.0003290981387745927,
+      "learning_rate": 6.042964554242751e-06,
+      "loss": 0.0,
+      "step": 7084
+    },
+    {
+      "epoch": 0.456595991493201,
+      "grad_norm": 0.00033238180508284173,
+      "learning_rate": 6.042248478338705e-06,
+      "loss": 0.0,
+      "step": 7085
+    },
+    {
+      "epoch": 0.4566604369401302,
+      "grad_norm": 0.0005725779904502857,
+      "learning_rate": 6.041532402434659e-06,
+      "loss": 0.0,
+      "step": 7086
+    },
+    {
+      "epoch": 0.4567248823870593,
+      "grad_norm": 0.0006684580100672684,
+      "learning_rate": 6.040816326530613e-06,
+      "loss": 0.0,
+      "step": 7087
+    },
+    {
+      "epoch": 0.4567893278339885,
+      "grad_norm": 0.0015680347505695766,
+      "learning_rate": 6.040100250626567e-06,
+      "loss": 0.0,
+      "step": 7088
+    },
+    {
+      "epoch": 0.4568537732809177,
+      "grad_norm": 0.00016943457185629763,
+      "learning_rate": 6.039384174722521e-06,
+      "loss": 0.0,
+      "step": 7089
+    },
+    {
+      "epoch": 0.4569182187278469,
+      "grad_norm": 0.005080476489226899,
+      "learning_rate": 6.038668098818476e-06,
+      "loss": 0.0,
+      "step": 7090
+    },
+    {
+      "epoch": 0.45698266417477607,
+      "grad_norm": 0.011049966479080763,
+      "learning_rate": 6.037952022914429e-06,
+      "loss": 0.0002,
+      "step": 7091
+    },
+    {
+      "epoch": 0.45704710962170525,
+      "grad_norm": 0.0004842771125507965,
+      "learning_rate": 6.037235947010383e-06,
+      "loss": 0.0,
+      "step": 7092
+    },
+    {
+      "epoch": 0.4571115550686344,
+      "grad_norm": 3.9384781038392525e-05,
+      "learning_rate": 6.036519871106338e-06,
+      "loss": 0.0,
+      "step": 7093
+    },
+    {
+      "epoch": 0.45717600051556356,
+      "grad_norm": 0.005493382939640472,
+      "learning_rate": 6.035803795202292e-06,
+      "loss": 0.0,
+      "step": 7094
+    },
+    {
+      "epoch": 0.45724044596249275,
+      "grad_norm": 0.016864852974129232,
+      "learning_rate": 6.035087719298246e-06,
+      "loss": 0.0001,
+      "step": 7095
+    },
+    {
+      "epoch": 0.45730489140942193,
+      "grad_norm": 0.23987302508294103,
+      "learning_rate": 6.0343716433942e-06,
+      "loss": 0.0055,
+      "step": 7096
+    },
+    {
+      "epoch": 0.4573693368563511,
+      "grad_norm": 0.000661137041265991,
+      "learning_rate": 6.033655567490154e-06,
+      "loss": 0.0,
+      "step": 7097
+    },
+    {
+      "epoch": 0.4574337823032803,
+      "grad_norm": 0.1169914314041797,
+      "learning_rate": 6.032939491586108e-06,
+      "loss": 0.0018,
+      "step": 7098
+    },
+    {
+      "epoch": 0.45749822775020943,
+      "grad_norm": 0.09465142347291379,
+      "learning_rate": 6.032223415682063e-06,
+      "loss": 0.0002,
+      "step": 7099
+    },
+    {
+      "epoch": 0.4575626731971386,
+      "grad_norm": 0.0004188542693633139,
+      "learning_rate": 6.031507339778017e-06,
+      "loss": 0.0,
+      "step": 7100
+    },
+    {
+      "epoch": 0.4576271186440678,
+      "grad_norm": 0.0006338906192689149,
+      "learning_rate": 6.03079126387397e-06,
+      "loss": 0.0,
+      "step": 7101
+    },
+    {
+      "epoch": 0.457691564090997,
+      "grad_norm": 0.24523413098807104,
+      "learning_rate": 6.030075187969925e-06,
+      "loss": 0.0021,
+      "step": 7102
+    },
+    {
+      "epoch": 0.45775600953792617,
+      "grad_norm": 0.005114259183133491,
+      "learning_rate": 6.029359112065879e-06,
+      "loss": 0.0,
+      "step": 7103
+    },
+    {
+      "epoch": 0.4578204549848553,
+      "grad_norm": 0.00010116153271195064,
+      "learning_rate": 6.028643036161833e-06,
+      "loss": 0.0,
+      "step": 7104
+    },
+    {
+      "epoch": 0.4578849004317845,
+      "grad_norm": 0.0012903867681714094,
+      "learning_rate": 6.0279269602577884e-06,
+      "loss": 0.0,
+      "step": 7105
+    },
+    {
+      "epoch": 0.45794934587871367,
+      "grad_norm": 0.0012879498751937105,
+      "learning_rate": 6.027210884353743e-06,
+      "loss": 0.0,
+      "step": 7106
+    },
+    {
+      "epoch": 0.45801379132564285,
+      "grad_norm": 0.02153369947324202,
+      "learning_rate": 6.026494808449696e-06,
+      "loss": 0.0001,
+      "step": 7107
+    },
+    {
+      "epoch": 0.45807823677257203,
+      "grad_norm": 0.0064330967485958694,
+      "learning_rate": 6.0257787325456505e-06,
+      "loss": 0.0,
+      "step": 7108
+    },
+    {
+      "epoch": 0.4581426822195012,
+      "grad_norm": 0.00024972203715889577,
+      "learning_rate": 6.025062656641605e-06,
+      "loss": 0.0,
+      "step": 7109
+    },
+    {
+      "epoch": 0.45820712766643035,
+      "grad_norm": 0.0016818562335592555,
+      "learning_rate": 6.024346580737559e-06,
+      "loss": 0.0,
+      "step": 7110
+    },
+    {
+      "epoch": 0.45827157311335953,
+      "grad_norm": 0.0129347845643351,
+      "learning_rate": 6.023630504833513e-06,
+      "loss": 0.0,
+      "step": 7111
+    },
+    {
+      "epoch": 0.4583360185602887,
+      "grad_norm": 5.749007999088142e-05,
+      "learning_rate": 6.022914428929467e-06,
+      "loss": 0.0,
+      "step": 7112
+    },
+    {
+      "epoch": 0.4584004640072179,
+      "grad_norm": 0.0006387188346334157,
+      "learning_rate": 6.022198353025421e-06,
+      "loss": 0.0,
+      "step": 7113
+    },
+    {
+      "epoch": 0.4584649094541471,
+      "grad_norm": 0.00012755880001202847,
+      "learning_rate": 6.0214822771213754e-06,
+      "loss": 0.0,
+      "step": 7114
+    },
+    {
+      "epoch": 0.4585293549010762,
+      "grad_norm": 0.10691374328866679,
+      "learning_rate": 6.02076620121733e-06,
+      "loss": 0.0002,
+      "step": 7115
+    },
+    {
+      "epoch": 0.4585938003480054,
+      "grad_norm": 0.9480175376851292,
+      "learning_rate": 6.020050125313284e-06,
+      "loss": 0.0168,
+      "step": 7116
+    },
+    {
+      "epoch": 0.4586582457949346,
+      "grad_norm": 0.0006098464369457968,
+      "learning_rate": 6.0193340494092375e-06,
+      "loss": 0.0,
+      "step": 7117
+    },
+    {
+      "epoch": 0.45872269124186377,
+      "grad_norm": 0.0047673415025089256,
+      "learning_rate": 6.018617973505192e-06,
+      "loss": 0.0,
+      "step": 7118
+    },
+    {
+      "epoch": 0.45878713668879295,
+      "grad_norm": 0.01061895586109674,
+      "learning_rate": 6.017901897601146e-06,
+      "loss": 0.0001,
+      "step": 7119
+    },
+    {
+      "epoch": 0.45885158213572214,
+      "grad_norm": 0.00033589021288196004,
+      "learning_rate": 6.0171858216971e-06,
+      "loss": 0.0,
+      "step": 7120
+    },
+    {
+      "epoch": 0.45891602758265126,
+      "grad_norm": 0.0008578054963712787,
+      "learning_rate": 6.016469745793055e-06,
+      "loss": 0.0,
+      "step": 7121
+    },
+    {
+      "epoch": 0.45898047302958045,
+      "grad_norm": 0.07172960847940046,
+      "learning_rate": 6.015753669889008e-06,
+      "loss": 0.0001,
+      "step": 7122
+    },
+    {
+      "epoch": 0.45904491847650963,
+      "grad_norm": 0.0005887507011680352,
+      "learning_rate": 6.015037593984962e-06,
+      "loss": 0.0,
+      "step": 7123
+    },
+    {
+      "epoch": 0.4591093639234388,
+      "grad_norm": 0.002714727540164013,
+      "learning_rate": 6.014321518080917e-06,
+      "loss": 0.0,
+      "step": 7124
+    },
+    {
+      "epoch": 0.459173809370368,
+      "grad_norm": 0.004810577714593613,
+      "learning_rate": 6.013605442176871e-06,
+      "loss": 0.0,
+      "step": 7125
+    },
+    {
+      "epoch": 0.45923825481729713,
+      "grad_norm": 0.09211610013209857,
+      "learning_rate": 6.012889366272825e-06,
+      "loss": 0.0001,
+      "step": 7126
+    },
+    {
+      "epoch": 0.4593027002642263,
+      "grad_norm": 0.04762401536160834,
+      "learning_rate": 6.012173290368779e-06,
+      "loss": 0.0001,
+      "step": 7127
+    },
+    {
+      "epoch": 0.4593671457111555,
+      "grad_norm": 0.0002858773479675718,
+      "learning_rate": 6.011457214464734e-06,
+      "loss": 0.0,
+      "step": 7128
+    },
+    {
+      "epoch": 0.4594315911580847,
+      "grad_norm": 0.0007525722764745775,
+      "learning_rate": 6.010741138560688e-06,
+      "loss": 0.0,
+      "step": 7129
+    },
+    {
+      "epoch": 0.45949603660501387,
+      "grad_norm": 1.6808105972308205,
+      "learning_rate": 6.0100250626566425e-06,
+      "loss": 0.018,
+      "step": 7130
+    },
+    {
+      "epoch": 0.45956048205194305,
+      "grad_norm": 0.001529241299410118,
+      "learning_rate": 6.009308986752597e-06,
+      "loss": 0.0,
+      "step": 7131
+    },
+    {
+      "epoch": 0.4596249274988722,
+      "grad_norm": 0.0038925792922026225,
+      "learning_rate": 6.008592910848551e-06,
+      "loss": 0.0,
+      "step": 7132
+    },
+    {
+      "epoch": 0.45968937294580137,
+      "grad_norm": 0.010454839195876581,
+      "learning_rate": 6.0078768349445046e-06,
+      "loss": 0.0001,
+      "step": 7133
+    },
+    {
+      "epoch": 0.45975381839273055,
+      "grad_norm": 0.0007752854575114376,
+      "learning_rate": 6.007160759040459e-06,
+      "loss": 0.0,
+      "step": 7134
+    },
+    {
+      "epoch": 0.45981826383965974,
+      "grad_norm": 0.007434410935910509,
+      "learning_rate": 6.006444683136413e-06,
+      "loss": 0.0001,
+      "step": 7135
+    },
+    {
+      "epoch": 0.4598827092865889,
+      "grad_norm": 0.6166326950570551,
+      "learning_rate": 6.0057286072323675e-06,
+      "loss": 0.0026,
+      "step": 7136
+    },
+    {
+      "epoch": 0.4599471547335181,
+      "grad_norm": 0.0003005682636650433,
+      "learning_rate": 6.005012531328322e-06,
+      "loss": 0.0,
+      "step": 7137
+    },
+    {
+      "epoch": 0.46001160018044723,
+      "grad_norm": 0.00016686412122810733,
+      "learning_rate": 6.004296455424275e-06,
+      "loss": 0.0,
+      "step": 7138
+    },
+    {
+      "epoch": 0.4600760456273764,
+      "grad_norm": 0.0008105308466117006,
+      "learning_rate": 6.0035803795202295e-06,
+      "loss": 0.0,
+      "step": 7139
+    },
+    {
+      "epoch": 0.4601404910743056,
+      "grad_norm": 0.002631408133243594,
+      "learning_rate": 6.002864303616184e-06,
+      "loss": 0.0,
+      "step": 7140
+    },
+    {
+      "epoch": 0.4602049365212348,
+      "grad_norm": 0.00022067880593967695,
+      "learning_rate": 6.002148227712138e-06,
+      "loss": 0.0,
+      "step": 7141
+    },
+    {
+      "epoch": 0.46026938196816397,
+      "grad_norm": 0.001211258225825929,
+      "learning_rate": 6.001432151808092e-06,
+      "loss": 0.0,
+      "step": 7142
+    },
+    {
+      "epoch": 0.4603338274150931,
+      "grad_norm": 9.802932496178768e-05,
+      "learning_rate": 6.000716075904046e-06,
+      "loss": 0.0,
+      "step": 7143
+    },
+    {
+      "epoch": 0.4603982728620223,
+      "grad_norm": 0.26474249319163073,
+      "learning_rate": 6e-06,
+      "loss": 0.0026,
+      "step": 7144
+    },
+    {
+      "epoch": 0.46046271830895147,
+      "grad_norm": 1.265212050520982,
+      "learning_rate": 5.9992839240959545e-06,
+      "loss": 0.0032,
+      "step": 7145
+    },
+    {
+      "epoch": 0.46052716375588065,
+      "grad_norm": 0.004371471218572817,
+      "learning_rate": 5.998567848191909e-06,
+      "loss": 0.0,
+      "step": 7146
+    },
+    {
+      "epoch": 0.46059160920280984,
+      "grad_norm": 0.00043372477915415046,
+      "learning_rate": 5.997851772287863e-06,
+      "loss": 0.0,
+      "step": 7147
+    },
+    {
+      "epoch": 0.460656054649739,
+      "grad_norm": 0.008678506284997306,
+      "learning_rate": 5.9971356963838165e-06,
+      "loss": 0.0,
+      "step": 7148
+    },
+    {
+      "epoch": 0.46072050009666815,
+      "grad_norm": 0.0003419930026615307,
+      "learning_rate": 5.996419620479771e-06,
+      "loss": 0.0,
+      "step": 7149
+    },
+    {
+      "epoch": 0.46078494554359734,
+      "grad_norm": 0.002855349159979639,
+      "learning_rate": 5.995703544575725e-06,
+      "loss": 0.0,
+      "step": 7150
+    },
+    {
+      "epoch": 0.4608493909905265,
+      "grad_norm": 0.003908428454548936,
+      "learning_rate": 5.994987468671679e-06,
+      "loss": 0.0,
+      "step": 7151
+    },
+    {
+      "epoch": 0.4609138364374557,
+      "grad_norm": 0.0010189799207077413,
+      "learning_rate": 5.9942713927676345e-06,
+      "loss": 0.0,
+      "step": 7152
+    },
+    {
+      "epoch": 0.4609782818843849,
+      "grad_norm": 0.02421507927741126,
+      "learning_rate": 5.993555316863589e-06,
+      "loss": 0.0001,
+      "step": 7153
+    },
+    {
+      "epoch": 0.461042727331314,
+      "grad_norm": 8.896939169050152e-05,
+      "learning_rate": 5.992839240959542e-06,
+      "loss": 0.0,
+      "step": 7154
+    },
+    {
+      "epoch": 0.4611071727782432,
+      "grad_norm": 0.029285788425192732,
+      "learning_rate": 5.992123165055497e-06,
+      "loss": 0.0001,
+      "step": 7155
+    },
+    {
+      "epoch": 0.4611716182251724,
+      "grad_norm": 0.034353822516338714,
+      "learning_rate": 5.991407089151451e-06,
+      "loss": 0.0,
+      "step": 7156
+    },
+    {
+      "epoch": 0.46123606367210157,
+      "grad_norm": 0.06933937937293207,
+      "learning_rate": 5.990691013247405e-06,
+      "loss": 0.0001,
+      "step": 7157
+    },
+    {
+      "epoch": 0.46130050911903075,
+      "grad_norm": 0.002533742700116563,
+      "learning_rate": 5.9899749373433595e-06,
+      "loss": 0.0,
+      "step": 7158
+    },
+    {
+      "epoch": 0.46136495456595994,
+      "grad_norm": 0.03438283799772264,
+      "learning_rate": 5.989258861439313e-06,
+      "loss": 0.0002,
+      "step": 7159
+    },
+    {
+      "epoch": 0.46142940001288907,
+      "grad_norm": 0.009976091964890367,
+      "learning_rate": 5.988542785535267e-06,
+      "loss": 0.0,
+      "step": 7160
+    },
+    {
+      "epoch": 0.46149384545981825,
+      "grad_norm": 0.013725964591191813,
+      "learning_rate": 5.9878267096312215e-06,
+      "loss": 0.0,
+      "step": 7161
+    },
+    {
+      "epoch": 0.46155829090674744,
+      "grad_norm": 0.022082188905863047,
+      "learning_rate": 5.987110633727176e-06,
+      "loss": 0.0001,
+      "step": 7162
+    },
+    {
+      "epoch": 0.4616227363536766,
+      "grad_norm": 0.01672156348965313,
+      "learning_rate": 5.98639455782313e-06,
+      "loss": 0.0,
+      "step": 7163
+    },
+    {
+      "epoch": 0.4616871818006058,
+      "grad_norm": 0.007685841599548759,
+      "learning_rate": 5.985678481919084e-06,
+      "loss": 0.0,
+      "step": 7164
+    },
+    {
+      "epoch": 0.46175162724753493,
+      "grad_norm": 0.010648891229751439,
+      "learning_rate": 5.984962406015038e-06,
+      "loss": 0.0,
+      "step": 7165
+    },
+    {
+      "epoch": 0.4618160726944641,
+      "grad_norm": 0.08365923854229963,
+      "learning_rate": 5.984246330110992e-06,
+      "loss": 0.0017,
+      "step": 7166
+    },
+    {
+      "epoch": 0.4618805181413933,
+      "grad_norm": 0.08341798140164661,
+      "learning_rate": 5.9835302542069465e-06,
+      "loss": 0.0017,
+      "step": 7167
+    },
+    {
+      "epoch": 0.4619449635883225,
+      "grad_norm": 0.10906040554440197,
+      "learning_rate": 5.9828141783029e-06,
+      "loss": 0.0001,
+      "step": 7168
+    },
+    {
+      "epoch": 0.4620094090352517,
+      "grad_norm": 0.0016372742490571707,
+      "learning_rate": 5.982098102398854e-06,
+      "loss": 0.0,
+      "step": 7169
+    },
+    {
+      "epoch": 0.46207385448218086,
+      "grad_norm": 0.0005011299067149563,
+      "learning_rate": 5.9813820264948085e-06,
+      "loss": 0.0,
+      "step": 7170
+    },
+    {
+      "epoch": 0.46213829992911,
+      "grad_norm": 0.0173133371179596,
+      "learning_rate": 5.980665950590763e-06,
+      "loss": 0.0,
+      "step": 7171
+    },
+    {
+      "epoch": 0.46220274537603917,
+      "grad_norm": 0.0011703360208209618,
+      "learning_rate": 5.979949874686717e-06,
+      "loss": 0.0,
+      "step": 7172
+    },
+    {
+      "epoch": 0.46226719082296835,
+      "grad_norm": 0.0040470612125930484,
+      "learning_rate": 5.979233798782671e-06,
+      "loss": 0.0,
+      "step": 7173
+    },
+    {
+      "epoch": 0.46233163626989754,
+      "grad_norm": 0.0007731407965276777,
+      "learning_rate": 5.978517722878625e-06,
+      "loss": 0.0,
+      "step": 7174
+    },
+    {
+      "epoch": 0.4623960817168267,
+      "grad_norm": 0.005962191522255315,
+      "learning_rate": 5.97780164697458e-06,
+      "loss": 0.0001,
+      "step": 7175
+    },
+    {
+      "epoch": 0.4624605271637559,
+      "grad_norm": 0.04625911729908732,
+      "learning_rate": 5.977085571070534e-06,
+      "loss": 0.0001,
+      "step": 7176
+    },
+    {
+      "epoch": 0.46252497261068504,
+      "grad_norm": 0.002839065374832336,
+      "learning_rate": 5.976369495166489e-06,
+      "loss": 0.0,
+      "step": 7177
+    },
+    {
+      "epoch": 0.4625894180576142,
+      "grad_norm": 0.0056488077167803855,
+      "learning_rate": 5.975653419262443e-06,
+      "loss": 0.0001,
+      "step": 7178
+    },
+    {
+      "epoch": 0.4626538635045434,
+      "grad_norm": 2.9853239298381764,
+      "learning_rate": 5.974937343358397e-06,
+      "loss": 0.0079,
+      "step": 7179
+    },
+    {
+      "epoch": 0.4627183089514726,
+      "grad_norm": 0.15088591204430143,
+      "learning_rate": 5.974221267454351e-06,
+      "loss": 0.0007,
+      "step": 7180
+    },
+    {
+      "epoch": 0.4627827543984018,
+      "grad_norm": 0.004880895834493149,
+      "learning_rate": 5.973505191550305e-06,
+      "loss": 0.0,
+      "step": 7181
+    },
+    {
+      "epoch": 0.4628471998453309,
+      "grad_norm": 0.00042946334087172727,
+      "learning_rate": 5.972789115646259e-06,
+      "loss": 0.0,
+      "step": 7182
+    },
+    {
+      "epoch": 0.4629116452922601,
+      "grad_norm": 0.035591940752936105,
+      "learning_rate": 5.9720730397422136e-06,
+      "loss": 0.0001,
+      "step": 7183
+    },
+    {
+      "epoch": 0.46297609073918927,
+      "grad_norm": 0.0010547751112222126,
+      "learning_rate": 5.971356963838167e-06,
+      "loss": 0.0,
+      "step": 7184
+    },
+    {
+      "epoch": 0.46304053618611846,
+      "grad_norm": 0.06196757870758435,
+      "learning_rate": 5.970640887934121e-06,
+      "loss": 0.0001,
+      "step": 7185
+    },
+    {
+      "epoch": 0.46310498163304764,
+      "grad_norm": 0.0026980154325516037,
+      "learning_rate": 5.969924812030076e-06,
+      "loss": 0.0,
+      "step": 7186
+    },
+    {
+      "epoch": 0.4631694270799768,
+      "grad_norm": 0.22517366429317423,
+      "learning_rate": 5.96920873612603e-06,
+      "loss": 0.0002,
+      "step": 7187
+    },
+    {
+      "epoch": 0.46323387252690595,
+      "grad_norm": 0.004543474395993957,
+      "learning_rate": 5.968492660221984e-06,
+      "loss": 0.0,
+      "step": 7188
+    },
+    {
+      "epoch": 0.46329831797383514,
+      "grad_norm": 0.008096930609374857,
+      "learning_rate": 5.967776584317938e-06,
+      "loss": 0.0,
+      "step": 7189
+    },
+    {
+      "epoch": 0.4633627634207643,
+      "grad_norm": 8.040914560860149e-05,
+      "learning_rate": 5.967060508413892e-06,
+      "loss": 0.0,
+      "step": 7190
+    },
+    {
+      "epoch": 0.4634272088676935,
+      "grad_norm": 0.0038613975120519836,
+      "learning_rate": 5.966344432509846e-06,
+      "loss": 0.0,
+      "step": 7191
+    },
+    {
+      "epoch": 0.4634916543146227,
+      "grad_norm": 0.009150825453134418,
+      "learning_rate": 5.9656283566058006e-06,
+      "loss": 0.0,
+      "step": 7192
+    },
+    {
+      "epoch": 0.4635560997615518,
+      "grad_norm": 0.006888794586705892,
+      "learning_rate": 5.964912280701755e-06,
+      "loss": 0.0,
+      "step": 7193
+    },
+    {
+      "epoch": 0.463620545208481,
+      "grad_norm": 0.0015190788830026104,
+      "learning_rate": 5.964196204797708e-06,
+      "loss": 0.0,
+      "step": 7194
+    },
+    {
+      "epoch": 0.4636849906554102,
+      "grad_norm": 0.0010511826214208134,
+      "learning_rate": 5.963480128893663e-06,
+      "loss": 0.0,
+      "step": 7195
+    },
+    {
+      "epoch": 0.4637494361023394,
+      "grad_norm": 0.006848881538616532,
+      "learning_rate": 5.962764052989617e-06,
+      "loss": 0.0,
+      "step": 7196
+    },
+    {
+      "epoch": 0.46381388154926856,
+      "grad_norm": 0.19715999381198668,
+      "learning_rate": 5.962047977085571e-06,
+      "loss": 0.0018,
+      "step": 7197
+    },
+    {
+      "epoch": 0.46387832699619774,
+      "grad_norm": 9.671389949897078e-05,
+      "learning_rate": 5.961331901181526e-06,
+      "loss": 0.0,
+      "step": 7198
+    },
+    {
+      "epoch": 0.46394277244312687,
+      "grad_norm": 0.010491111385630709,
+      "learning_rate": 5.960615825277481e-06,
+      "loss": 0.0,
+      "step": 7199
+    },
+    {
+      "epoch": 0.46400721789005606,
+      "grad_norm": 0.000629562530891665,
+      "learning_rate": 5.959899749373434e-06,
+      "loss": 0.0,
+      "step": 7200
+    },
+    {
+      "epoch": 0.46407166333698524,
+      "grad_norm": 0.002112236082025321,
+      "learning_rate": 5.959183673469388e-06,
+      "loss": 0.0,
+      "step": 7201
+    },
+    {
+      "epoch": 0.4641361087839144,
+      "grad_norm": 0.2710224448099513,
+      "learning_rate": 5.958467597565343e-06,
+      "loss": 0.0002,
+      "step": 7202
+    },
+    {
+      "epoch": 0.4642005542308436,
+      "grad_norm": 0.02375578089601274,
+      "learning_rate": 5.957751521661297e-06,
+      "loss": 0.0017,
+      "step": 7203
+    },
+    {
+      "epoch": 0.46426499967777274,
+      "grad_norm": 0.0011988023769871627,
+      "learning_rate": 5.957035445757251e-06,
+      "loss": 0.0,
+      "step": 7204
+    },
+    {
+      "epoch": 0.4643294451247019,
+      "grad_norm": 0.003208027864138399,
+      "learning_rate": 5.956319369853205e-06,
+      "loss": 0.0,
+      "step": 7205
+    },
+    {
+      "epoch": 0.4643938905716311,
+      "grad_norm": 0.0009489984323486449,
+      "learning_rate": 5.955603293949159e-06,
+      "loss": 0.0,
+      "step": 7206
+    },
+    {
+      "epoch": 0.4644583360185603,
+      "grad_norm": 0.00045291492100388596,
+      "learning_rate": 5.954887218045113e-06,
+      "loss": 0.0,
+      "step": 7207
+    },
+    {
+      "epoch": 0.4645227814654895,
+      "grad_norm": 0.015716493476830035,
+      "learning_rate": 5.954171142141068e-06,
+      "loss": 0.0001,
+      "step": 7208
+    },
+    {
+      "epoch": 0.46458722691241866,
+      "grad_norm": 0.010753894111213871,
+      "learning_rate": 5.953455066237022e-06,
+      "loss": 0.0,
+      "step": 7209
+    },
+    {
+      "epoch": 0.4646516723593478,
+      "grad_norm": 0.0021181276175598375,
+      "learning_rate": 5.952738990332975e-06,
+      "loss": 0.0,
+      "step": 7210
+    },
+    {
+      "epoch": 0.464716117806277,
+      "grad_norm": 0.0005319157505790057,
+      "learning_rate": 5.95202291442893e-06,
+      "loss": 0.0,
+      "step": 7211
+    },
+    {
+      "epoch": 0.46478056325320616,
+      "grad_norm": 4.463178627000733e-05,
+      "learning_rate": 5.951306838524884e-06,
+      "loss": 0.0,
+      "step": 7212
+    },
+    {
+      "epoch": 0.46484500870013534,
+      "grad_norm": 0.0013662977275578781,
+      "learning_rate": 5.950590762620838e-06,
+      "loss": 0.0,
+      "step": 7213
+    },
+    {
+      "epoch": 0.4649094541470645,
+      "grad_norm": 0.010369585677639376,
+      "learning_rate": 5.949874686716793e-06,
+      "loss": 0.0,
+      "step": 7214
+    },
+    {
+      "epoch": 0.4649738995939937,
+      "grad_norm": 0.03391908857152616,
+      "learning_rate": 5.949158610812746e-06,
+      "loss": 0.0004,
+      "step": 7215
+    },
+    {
+      "epoch": 0.46503834504092284,
+      "grad_norm": 0.06855665758350678,
+      "learning_rate": 5.9484425349087e-06,
+      "loss": 0.0001,
+      "step": 7216
+    },
+    {
+      "epoch": 0.465102790487852,
+      "grad_norm": 0.21561523982086514,
+      "learning_rate": 5.947726459004655e-06,
+      "loss": 0.001,
+      "step": 7217
+    },
+    {
+      "epoch": 0.4651672359347812,
+      "grad_norm": 0.0034353563543390304,
+      "learning_rate": 5.947010383100609e-06,
+      "loss": 0.0,
+      "step": 7218
+    },
+    {
+      "epoch": 0.4652316813817104,
+      "grad_norm": 0.00018852596750020293,
+      "learning_rate": 5.946294307196563e-06,
+      "loss": 0.0,
+      "step": 7219
+    },
+    {
+      "epoch": 0.4652961268286396,
+      "grad_norm": 0.009362767432748718,
+      "learning_rate": 5.945578231292517e-06,
+      "loss": 0.0,
+      "step": 7220
+    },
+    {
+      "epoch": 0.4653605722755687,
+      "grad_norm": 0.022573569619495887,
+      "learning_rate": 5.944862155388471e-06,
+      "loss": 0.0001,
+      "step": 7221
+    },
+    {
+      "epoch": 0.4654250177224979,
+      "grad_norm": 0.006736289318203268,
+      "learning_rate": 5.944146079484426e-06,
+      "loss": 0.0,
+      "step": 7222
+    },
+    {
+      "epoch": 0.4654894631694271,
+      "grad_norm": 0.5131869329603391,
+      "learning_rate": 5.9434300035803804e-06,
+      "loss": 0.0035,
+      "step": 7223
+    },
+    {
+      "epoch": 0.46555390861635626,
+      "grad_norm": 0.15121273651669012,
+      "learning_rate": 5.942713927676335e-06,
+      "loss": 0.0005,
+      "step": 7224
+    },
+    {
+      "epoch": 0.46561835406328544,
+      "grad_norm": 0.000143468918569542,
+      "learning_rate": 5.941997851772289e-06,
+      "loss": 0.0,
+      "step": 7225
+    },
+    {
+      "epoch": 0.46568279951021463,
+      "grad_norm": 0.003423290574634767,
+      "learning_rate": 5.9412817758682425e-06,
+      "loss": 0.0,
+      "step": 7226
+    },
+    {
+      "epoch": 0.46574724495714376,
+      "grad_norm": 0.05099336860487578,
+      "learning_rate": 5.940565699964197e-06,
+      "loss": 0.0004,
+      "step": 7227
+    },
+    {
+      "epoch": 0.46581169040407294,
+      "grad_norm": 0.07674497419352205,
+      "learning_rate": 5.939849624060151e-06,
+      "loss": 0.0017,
+      "step": 7228
+    },
+    {
+      "epoch": 0.4658761358510021,
+      "grad_norm": 0.16468423587705824,
+      "learning_rate": 5.939133548156105e-06,
+      "loss": 0.0013,
+      "step": 7229
+    },
+    {
+      "epoch": 0.4659405812979313,
+      "grad_norm": 0.0024757373314606267,
+      "learning_rate": 5.93841747225206e-06,
+      "loss": 0.0,
+      "step": 7230
+    },
+    {
+      "epoch": 0.4660050267448605,
+      "grad_norm": 0.0039586155872088965,
+      "learning_rate": 5.937701396348013e-06,
+      "loss": 0.0,
+      "step": 7231
+    },
+    {
+      "epoch": 0.4660694721917896,
+      "grad_norm": 0.01858411356660055,
+      "learning_rate": 5.9369853204439674e-06,
+      "loss": 0.0,
+      "step": 7232
+    },
+    {
+      "epoch": 0.4661339176387188,
+      "grad_norm": 0.21115758681710384,
+      "learning_rate": 5.936269244539922e-06,
+      "loss": 0.0026,
+      "step": 7233
+    },
+    {
+      "epoch": 0.466198363085648,
+      "grad_norm": 0.10160640537815353,
+      "learning_rate": 5.935553168635876e-06,
+      "loss": 0.0002,
+      "step": 7234
+    },
+    {
+      "epoch": 0.4662628085325772,
+      "grad_norm": 0.045725959999042595,
+      "learning_rate": 5.93483709273183e-06,
+      "loss": 0.0,
+      "step": 7235
+    },
+    {
+      "epoch": 0.46632725397950636,
+      "grad_norm": 0.01414354404282994,
+      "learning_rate": 5.934121016827784e-06,
+      "loss": 0.0,
+      "step": 7236
+    },
+    {
+      "epoch": 0.46639169942643555,
+      "grad_norm": 0.01251770166379137,
+      "learning_rate": 5.933404940923738e-06,
+      "loss": 0.0001,
+      "step": 7237
+    },
+    {
+      "epoch": 0.4664561448733647,
+      "grad_norm": 0.0007334224021861777,
+      "learning_rate": 5.932688865019692e-06,
+      "loss": 0.0,
+      "step": 7238
+    },
+    {
+      "epoch": 0.46652059032029386,
+      "grad_norm": 0.0022743016070544573,
+      "learning_rate": 5.931972789115647e-06,
+      "loss": 0.0,
+      "step": 7239
+    },
+    {
+      "epoch": 0.46658503576722304,
+      "grad_norm": 0.00039270465816523245,
+      "learning_rate": 5.931256713211601e-06,
+      "loss": 0.0,
+      "step": 7240
+    },
+    {
+      "epoch": 0.46664948121415223,
+      "grad_norm": 0.00010237405302461057,
+      "learning_rate": 5.930540637307554e-06,
+      "loss": 0.0,
+      "step": 7241
+    },
+    {
+      "epoch": 0.4667139266610814,
+      "grad_norm": 0.004509319091979509,
+      "learning_rate": 5.929824561403509e-06,
+      "loss": 0.0,
+      "step": 7242
+    },
+    {
+      "epoch": 0.46677837210801054,
+      "grad_norm": 0.12337622011710904,
+      "learning_rate": 5.929108485499463e-06,
+      "loss": 0.0001,
+      "step": 7243
+    },
+    {
+      "epoch": 0.4668428175549397,
+      "grad_norm": 0.005279184753878209,
+      "learning_rate": 5.928392409595417e-06,
+      "loss": 0.0,
+      "step": 7244
+    },
+    {
+      "epoch": 0.4669072630018689,
+      "grad_norm": 0.013528141218017584,
+      "learning_rate": 5.9276763336913725e-06,
+      "loss": 0.0,
+      "step": 7245
+    },
+    {
+      "epoch": 0.4669717084487981,
+      "grad_norm": 7.838986385026141e-05,
+      "learning_rate": 5.926960257787327e-06,
+      "loss": 0.0,
+      "step": 7246
+    },
+    {
+      "epoch": 0.4670361538957273,
+      "grad_norm": 0.09289410924673282,
+      "learning_rate": 5.92624418188328e-06,
+      "loss": 0.0,
+      "step": 7247
+    },
+    {
+      "epoch": 0.46710059934265646,
+      "grad_norm": 0.005147704458191427,
+      "learning_rate": 5.9255281059792345e-06,
+      "loss": 0.0,
+      "step": 7248
+    },
+    {
+      "epoch": 0.4671650447895856,
+      "grad_norm": 0.0005078345397182034,
+      "learning_rate": 5.924812030075189e-06,
+      "loss": 0.0,
+      "step": 7249
+    },
+    {
+      "epoch": 0.4672294902365148,
+      "grad_norm": 0.01140905101926907,
+      "learning_rate": 5.924095954171143e-06,
+      "loss": 0.0001,
+      "step": 7250
+    },
+    {
+      "epoch": 0.46729393568344396,
+      "grad_norm": 0.013532734656177478,
+      "learning_rate": 5.923379878267097e-06,
+      "loss": 0.0001,
+      "step": 7251
+    },
+    {
+      "epoch": 0.46735838113037315,
+      "grad_norm": 0.00450951372079799,
+      "learning_rate": 5.922663802363051e-06,
+      "loss": 0.0,
+      "step": 7252
+    },
+    {
+      "epoch": 0.46742282657730233,
+      "grad_norm": 0.00013456809584500636,
+      "learning_rate": 5.921947726459005e-06,
+      "loss": 0.0,
+      "step": 7253
+    },
+    {
+      "epoch": 0.4674872720242315,
+      "grad_norm": 0.5415825747969696,
+      "learning_rate": 5.9212316505549595e-06,
+      "loss": 0.0032,
+      "step": 7254
+    },
+    {
+      "epoch": 0.46755171747116064,
+      "grad_norm": 0.06273022360028993,
+      "learning_rate": 5.920515574650914e-06,
+      "loss": 0.0002,
+      "step": 7255
+    },
+    {
+      "epoch": 0.46761616291808983,
+      "grad_norm": 0.001915030388477191,
+      "learning_rate": 5.919799498746868e-06,
+      "loss": 0.0,
+      "step": 7256
+    },
+    {
+      "epoch": 0.467680608365019,
+      "grad_norm": 0.0018932465533284114,
+      "learning_rate": 5.9190834228428215e-06,
+      "loss": 0.0,
+      "step": 7257
+    },
+    {
+      "epoch": 0.4677450538119482,
+      "grad_norm": 0.01038187083953511,
+      "learning_rate": 5.918367346938776e-06,
+      "loss": 0.0,
+      "step": 7258
+    },
+    {
+      "epoch": 0.4678094992588774,
+      "grad_norm": 0.00021900495299448367,
+      "learning_rate": 5.91765127103473e-06,
+      "loss": 0.0,
+      "step": 7259
+    },
+    {
+      "epoch": 0.4678739447058065,
+      "grad_norm": 0.0002388793545880022,
+      "learning_rate": 5.916935195130684e-06,
+      "loss": 0.0,
+      "step": 7260
+    },
+    {
+      "epoch": 0.4679383901527357,
+      "grad_norm": 0.0033289252996606057,
+      "learning_rate": 5.916219119226638e-06,
+      "loss": 0.0,
+      "step": 7261
+    },
+    {
+      "epoch": 0.4680028355996649,
+      "grad_norm": 0.00018771824976225135,
+      "learning_rate": 5.915503043322592e-06,
+      "loss": 0.0,
+      "step": 7262
+    },
+    {
+      "epoch": 0.46806728104659406,
+      "grad_norm": 0.07404355653908364,
+      "learning_rate": 5.9147869674185465e-06,
+      "loss": 0.0008,
+      "step": 7263
+    },
+    {
+      "epoch": 0.46813172649352325,
+      "grad_norm": 0.003018865199686008,
+      "learning_rate": 5.914070891514501e-06,
+      "loss": 0.0,
+      "step": 7264
+    },
+    {
+      "epoch": 0.46819617194045243,
+      "grad_norm": 0.04145828291970412,
+      "learning_rate": 5.913354815610455e-06,
+      "loss": 0.0006,
+      "step": 7265
+    },
+    {
+      "epoch": 0.46826061738738156,
+      "grad_norm": 0.002445120612271508,
+      "learning_rate": 5.9126387397064085e-06,
+      "loss": 0.0,
+      "step": 7266
+    },
+    {
+      "epoch": 0.46832506283431075,
+      "grad_norm": 0.0041065608074551405,
+      "learning_rate": 5.911922663802363e-06,
+      "loss": 0.0,
+      "step": 7267
+    },
+    {
+      "epoch": 0.46838950828123993,
+      "grad_norm": 0.03801747734846256,
+      "learning_rate": 5.911206587898318e-06,
+      "loss": 0.0001,
+      "step": 7268
+    },
+    {
+      "epoch": 0.4684539537281691,
+      "grad_norm": 0.00017918018740986495,
+      "learning_rate": 5.910490511994272e-06,
+      "loss": 0.0,
+      "step": 7269
+    },
+    {
+      "epoch": 0.4685183991750983,
+      "grad_norm": 0.008305111975488961,
+      "learning_rate": 5.9097744360902265e-06,
+      "loss": 0.0001,
+      "step": 7270
+    },
+    {
+      "epoch": 0.46858284462202743,
+      "grad_norm": 3.966849481567199e-05,
+      "learning_rate": 5.909058360186181e-06,
+      "loss": 0.0,
+      "step": 7271
+    },
+    {
+      "epoch": 0.4686472900689566,
+      "grad_norm": 0.0004967530768057412,
+      "learning_rate": 5.908342284282135e-06,
+      "loss": 0.0,
+      "step": 7272
+    },
+    {
+      "epoch": 0.4687117355158858,
+      "grad_norm": 0.0757589549519109,
+      "learning_rate": 5.907626208378089e-06,
+      "loss": 0.0002,
+      "step": 7273
+    },
+    {
+      "epoch": 0.468776180962815,
+      "grad_norm": 0.0013153308624478225,
+      "learning_rate": 5.906910132474043e-06,
+      "loss": 0.0,
+      "step": 7274
+    },
+    {
+      "epoch": 0.46884062640974417,
+      "grad_norm": 0.004487714706241631,
+      "learning_rate": 5.906194056569997e-06,
+      "loss": 0.0,
+      "step": 7275
+    },
+    {
+      "epoch": 0.46890507185667335,
+      "grad_norm": 0.0024251842236077178,
+      "learning_rate": 5.9054779806659515e-06,
+      "loss": 0.0,
+      "step": 7276
+    },
+    {
+      "epoch": 0.4689695173036025,
+      "grad_norm": 0.0162244810554148,
+      "learning_rate": 5.904761904761905e-06,
+      "loss": 0.0001,
+      "step": 7277
+    },
+    {
+      "epoch": 0.46903396275053166,
+      "grad_norm": 0.00018407225832327176,
+      "learning_rate": 5.904045828857859e-06,
+      "loss": 0.0,
+      "step": 7278
+    },
+    {
+      "epoch": 0.46909840819746085,
+      "grad_norm": 0.0008793784795991423,
+      "learning_rate": 5.9033297529538135e-06,
+      "loss": 0.0,
+      "step": 7279
+    },
+    {
+      "epoch": 0.46916285364439003,
+      "grad_norm": 0.045148619317254515,
+      "learning_rate": 5.902613677049768e-06,
+      "loss": 0.0002,
+      "step": 7280
+    },
+    {
+      "epoch": 0.4692272990913192,
+      "grad_norm": 0.0007002317138202523,
+      "learning_rate": 5.901897601145722e-06,
+      "loss": 0.0,
+      "step": 7281
+    },
+    {
+      "epoch": 0.46929174453824835,
+      "grad_norm": 0.0076014641681480645,
+      "learning_rate": 5.901181525241676e-06,
+      "loss": 0.0,
+      "step": 7282
+    },
+    {
+      "epoch": 0.46935618998517753,
+      "grad_norm": 0.058544814019823664,
+      "learning_rate": 5.90046544933763e-06,
+      "loss": 0.0002,
+      "step": 7283
+    },
+    {
+      "epoch": 0.4694206354321067,
+      "grad_norm": 0.2589595880628915,
+      "learning_rate": 5.899749373433584e-06,
+      "loss": 0.0019,
+      "step": 7284
+    },
+    {
+      "epoch": 0.4694850808790359,
+      "grad_norm": 0.0009875575563172685,
+      "learning_rate": 5.8990332975295385e-06,
+      "loss": 0.0,
+      "step": 7285
+    },
+    {
+      "epoch": 0.4695495263259651,
+      "grad_norm": 0.0001364893657174719,
+      "learning_rate": 5.898317221625493e-06,
+      "loss": 0.0,
+      "step": 7286
+    },
+    {
+      "epoch": 0.46961397177289427,
+      "grad_norm": 0.08705876659342433,
+      "learning_rate": 5.897601145721446e-06,
+      "loss": 0.0007,
+      "step": 7287
+    },
+    {
+      "epoch": 0.4696784172198234,
+      "grad_norm": 0.01562531292125164,
+      "learning_rate": 5.8968850698174005e-06,
+      "loss": 0.0001,
+      "step": 7288
+    },
+    {
+      "epoch": 0.4697428626667526,
+      "grad_norm": 0.04099352701129466,
+      "learning_rate": 5.896168993913355e-06,
+      "loss": 0.0001,
+      "step": 7289
+    },
+    {
+      "epoch": 0.46980730811368177,
+      "grad_norm": 0.0005863534760000729,
+      "learning_rate": 5.895452918009309e-06,
+      "loss": 0.0,
+      "step": 7290
+    },
+    {
+      "epoch": 0.46987175356061095,
+      "grad_norm": 0.0015909966007760811,
+      "learning_rate": 5.8947368421052634e-06,
+      "loss": 0.0,
+      "step": 7291
+    },
+    {
+      "epoch": 0.46993619900754013,
+      "grad_norm": 0.00274800952267154,
+      "learning_rate": 5.8940207662012186e-06,
+      "loss": 0.0,
+      "step": 7292
+    },
+    {
+      "epoch": 0.4700006444544693,
+      "grad_norm": 0.000492930772659004,
+      "learning_rate": 5.893304690297173e-06,
+      "loss": 0.0,
+      "step": 7293
+    },
+    {
+      "epoch": 0.47006508990139845,
+      "grad_norm": 0.0025907573420449314,
+      "learning_rate": 5.892588614393126e-06,
+      "loss": 0.0,
+      "step": 7294
+    },
+    {
+      "epoch": 0.47012953534832763,
+      "grad_norm": 0.0007235664111460894,
+      "learning_rate": 5.891872538489081e-06,
+      "loss": 0.0,
+      "step": 7295
+    },
+    {
+      "epoch": 0.4701939807952568,
+      "grad_norm": 0.003975885038461543,
+      "learning_rate": 5.891156462585035e-06,
+      "loss": 0.0,
+      "step": 7296
+    },
+    {
+      "epoch": 0.470258426242186,
+      "grad_norm": 0.0025255628655265244,
+      "learning_rate": 5.890440386680989e-06,
+      "loss": 0.0,
+      "step": 7297
+    },
+    {
+      "epoch": 0.4703228716891152,
+      "grad_norm": 0.005506941121865925,
+      "learning_rate": 5.889724310776943e-06,
+      "loss": 0.0001,
+      "step": 7298
+    },
+    {
+      "epoch": 0.4703873171360443,
+      "grad_norm": 0.0001799831213966565,
+      "learning_rate": 5.889008234872897e-06,
+      "loss": 0.0,
+      "step": 7299
+    },
+    {
+      "epoch": 0.4704517625829735,
+      "grad_norm": 0.004868009039231323,
+      "learning_rate": 5.888292158968851e-06,
+      "loss": 0.0015,
+      "step": 7300
+    },
+    {
+      "epoch": 0.4705162080299027,
+      "grad_norm": 0.4330565587309389,
+      "learning_rate": 5.8875760830648056e-06,
+      "loss": 0.0004,
+      "step": 7301
+    },
+    {
+      "epoch": 0.47058065347683187,
+      "grad_norm": 0.017751362441838763,
+      "learning_rate": 5.88686000716076e-06,
+      "loss": 0.0,
+      "step": 7302
+    },
+    {
+      "epoch": 0.47064509892376105,
+      "grad_norm": 0.00020847490324210836,
+      "learning_rate": 5.886143931256713e-06,
+      "loss": 0.0,
+      "step": 7303
+    },
+    {
+      "epoch": 0.47070954437069024,
+      "grad_norm": 0.007923028419122423,
+      "learning_rate": 5.885427855352668e-06,
+      "loss": 0.0,
+      "step": 7304
+    },
+    {
+      "epoch": 0.47077398981761936,
+      "grad_norm": 0.005294704298977462,
+      "learning_rate": 5.884711779448622e-06,
+      "loss": 0.0,
+      "step": 7305
+    },
+    {
+      "epoch": 0.47083843526454855,
+      "grad_norm": 0.0007913525318603742,
+      "learning_rate": 5.883995703544576e-06,
+      "loss": 0.0,
+      "step": 7306
+    },
+    {
+      "epoch": 0.47090288071147773,
+      "grad_norm": 0.009260764334228884,
+      "learning_rate": 5.8832796276405305e-06,
+      "loss": 0.0,
+      "step": 7307
+    },
+    {
+      "epoch": 0.4709673261584069,
+      "grad_norm": 0.14998565371771327,
+      "learning_rate": 5.882563551736484e-06,
+      "loss": 0.0003,
+      "step": 7308
+    },
+    {
+      "epoch": 0.4710317716053361,
+      "grad_norm": 0.002568590896597837,
+      "learning_rate": 5.881847475832438e-06,
+      "loss": 0.0,
+      "step": 7309
+    },
+    {
+      "epoch": 0.47109621705226523,
+      "grad_norm": 0.0001356063438065277,
+      "learning_rate": 5.8811313999283926e-06,
+      "loss": 0.0,
+      "step": 7310
+    },
+    {
+      "epoch": 0.4711606624991944,
+      "grad_norm": 3.737375653479307e-05,
+      "learning_rate": 5.880415324024347e-06,
+      "loss": 0.0,
+      "step": 7311
+    },
+    {
+      "epoch": 0.4712251079461236,
+      "grad_norm": 0.0018837255037034548,
+      "learning_rate": 5.879699248120301e-06,
+      "loss": 0.0,
+      "step": 7312
+    },
+    {
+      "epoch": 0.4712895533930528,
+      "grad_norm": 9.369458189287065e-05,
+      "learning_rate": 5.878983172216255e-06,
+      "loss": 0.0,
+      "step": 7313
+    },
+    {
+      "epoch": 0.47135399883998197,
+      "grad_norm": 0.004181875470268751,
+      "learning_rate": 5.878267096312209e-06,
+      "loss": 0.0,
+      "step": 7314
+    },
+    {
+      "epoch": 0.47141844428691115,
+      "grad_norm": 0.003124371841280807,
+      "learning_rate": 5.877551020408164e-06,
+      "loss": 0.0001,
+      "step": 7315
+    },
+    {
+      "epoch": 0.4714828897338403,
+      "grad_norm": 0.00016681583676826113,
+      "learning_rate": 5.876834944504118e-06,
+      "loss": 0.0,
+      "step": 7316
+    },
+    {
+      "epoch": 0.47154733518076947,
+      "grad_norm": 0.013830684057848068,
+      "learning_rate": 5.876118868600073e-06,
+      "loss": 0.0,
+      "step": 7317
+    },
+    {
+      "epoch": 0.47161178062769865,
+      "grad_norm": 0.013160734314818012,
+      "learning_rate": 5.875402792696027e-06,
+      "loss": 0.0001,
+      "step": 7318
+    },
+    {
+      "epoch": 0.47167622607462784,
+      "grad_norm": 0.005155490320092713,
+      "learning_rate": 5.87468671679198e-06,
+      "loss": 0.0,
+      "step": 7319
+    },
+    {
+      "epoch": 0.471740671521557,
+      "grad_norm": 0.0018278137970520977,
+      "learning_rate": 5.873970640887935e-06,
+      "loss": 0.0,
+      "step": 7320
+    },
+    {
+      "epoch": 0.47180511696848615,
+      "grad_norm": 0.003185959559079041,
+      "learning_rate": 5.873254564983889e-06,
+      "loss": 0.0,
+      "step": 7321
+    },
+    {
+      "epoch": 0.47186956241541533,
+      "grad_norm": 0.2694356997754317,
+      "learning_rate": 5.872538489079843e-06,
+      "loss": 0.0024,
+      "step": 7322
+    },
+    {
+      "epoch": 0.4719340078623445,
+      "grad_norm": 0.01864031511807212,
+      "learning_rate": 5.871822413175798e-06,
+      "loss": 0.0,
+      "step": 7323
+    },
+    {
+      "epoch": 0.4719984533092737,
+      "grad_norm": 0.015568650232301293,
+      "learning_rate": 5.871106337271751e-06,
+      "loss": 0.0001,
+      "step": 7324
+    },
+    {
+      "epoch": 0.4720628987562029,
+      "grad_norm": 0.0058521904086244305,
+      "learning_rate": 5.870390261367705e-06,
+      "loss": 0.0,
+      "step": 7325
+    },
+    {
+      "epoch": 0.47212734420313207,
+      "grad_norm": 0.09893792709546372,
+      "learning_rate": 5.86967418546366e-06,
+      "loss": 0.0001,
+      "step": 7326
+    },
+    {
+      "epoch": 0.4721917896500612,
+      "grad_norm": 0.3626702410404047,
+      "learning_rate": 5.868958109559614e-06,
+      "loss": 0.0022,
+      "step": 7327
+    },
+    {
+      "epoch": 0.4722562350969904,
+      "grad_norm": 0.24238985590879442,
+      "learning_rate": 5.868242033655568e-06,
+      "loss": 0.0003,
+      "step": 7328
+    },
+    {
+      "epoch": 0.47232068054391957,
+      "grad_norm": 0.2921772246160459,
+      "learning_rate": 5.867525957751522e-06,
+      "loss": 0.0022,
+      "step": 7329
+    },
+    {
+      "epoch": 0.47238512599084875,
+      "grad_norm": 0.10111720327515572,
+      "learning_rate": 5.866809881847476e-06,
+      "loss": 0.0017,
+      "step": 7330
+    },
+    {
+      "epoch": 0.47244957143777794,
+      "grad_norm": 0.0007373990260450756,
+      "learning_rate": 5.86609380594343e-06,
+      "loss": 0.0,
+      "step": 7331
+    },
+    {
+      "epoch": 0.4725140168847071,
+      "grad_norm": 0.014592344951044412,
+      "learning_rate": 5.865377730039385e-06,
+      "loss": 0.0002,
+      "step": 7332
+    },
+    {
+      "epoch": 0.47257846233163625,
+      "grad_norm": 0.004748893437132448,
+      "learning_rate": 5.864661654135339e-06,
+      "loss": 0.0,
+      "step": 7333
+    },
+    {
+      "epoch": 0.47264290777856544,
+      "grad_norm": 0.04088056414933486,
+      "learning_rate": 5.863945578231292e-06,
+      "loss": 0.0001,
+      "step": 7334
+    },
+    {
+      "epoch": 0.4727073532254946,
+      "grad_norm": 0.0006974730055049175,
+      "learning_rate": 5.863229502327247e-06,
+      "loss": 0.0,
+      "step": 7335
+    },
+    {
+      "epoch": 0.4727717986724238,
+      "grad_norm": 0.0060704542827779,
+      "learning_rate": 5.862513426423201e-06,
+      "loss": 0.0,
+      "step": 7336
+    },
+    {
+      "epoch": 0.472836244119353,
+      "grad_norm": 6.82556858305384e-05,
+      "learning_rate": 5.861797350519155e-06,
+      "loss": 0.0,
+      "step": 7337
+    },
+    {
+      "epoch": 0.4729006895662821,
+      "grad_norm": 0.01232797728501841,
+      "learning_rate": 5.86108127461511e-06,
+      "loss": 0.0001,
+      "step": 7338
+    },
+    {
+      "epoch": 0.4729651350132113,
+      "grad_norm": 0.14040671069080024,
+      "learning_rate": 5.860365198711065e-06,
+      "loss": 0.0013,
+      "step": 7339
+    },
+    {
+      "epoch": 0.4730295804601405,
+      "grad_norm": 0.0069158175830979236,
+      "learning_rate": 5.859649122807018e-06,
+      "loss": 0.0,
+      "step": 7340
+    },
+    {
+      "epoch": 0.47309402590706967,
+      "grad_norm": 0.15098211468237663,
+      "learning_rate": 5.8589330469029724e-06,
+      "loss": 0.0014,
+      "step": 7341
+    },
+    {
+      "epoch": 0.47315847135399886,
+      "grad_norm": 0.0005686017696016296,
+      "learning_rate": 5.858216970998927e-06,
+      "loss": 0.0,
+      "step": 7342
+    },
+    {
+      "epoch": 0.47322291680092804,
+      "grad_norm": 0.06047954006858101,
+      "learning_rate": 5.857500895094881e-06,
+      "loss": 0.0001,
+      "step": 7343
+    },
+    {
+      "epoch": 0.47328736224785717,
+      "grad_norm": 0.07052647298794551,
+      "learning_rate": 5.856784819190835e-06,
+      "loss": 0.0008,
+      "step": 7344
+    },
+    {
+      "epoch": 0.47335180769478635,
+      "grad_norm": 8.880895053995195e-05,
+      "learning_rate": 5.856068743286789e-06,
+      "loss": 0.0,
+      "step": 7345
+    },
+    {
+      "epoch": 0.47341625314171554,
+      "grad_norm": 0.0009487713006594181,
+      "learning_rate": 5.855352667382743e-06,
+      "loss": 0.0,
+      "step": 7346
+    },
+    {
+      "epoch": 0.4734806985886447,
+      "grad_norm": 0.2606289649908769,
+      "learning_rate": 5.854636591478697e-06,
+      "loss": 0.0014,
+      "step": 7347
+    },
+    {
+      "epoch": 0.4735451440355739,
+      "grad_norm": 0.19189682199600822,
+      "learning_rate": 5.853920515574652e-06,
+      "loss": 0.0015,
+      "step": 7348
+    },
+    {
+      "epoch": 0.47360958948250304,
+      "grad_norm": 0.00025634485228848494,
+      "learning_rate": 5.853204439670606e-06,
+      "loss": 0.0,
+      "step": 7349
+    },
+    {
+      "epoch": 0.4736740349294322,
+      "grad_norm": 0.0021115394978189358,
+      "learning_rate": 5.8524883637665594e-06,
+      "loss": 0.0,
+      "step": 7350
+    },
+    {
+      "epoch": 0.4737384803763614,
+      "grad_norm": 0.008847833623355518,
+      "learning_rate": 5.851772287862514e-06,
+      "loss": 0.0001,
+      "step": 7351
+    },
+    {
+      "epoch": 0.4738029258232906,
+      "grad_norm": 0.1189699239903375,
+      "learning_rate": 5.851056211958468e-06,
+      "loss": 0.0004,
+      "step": 7352
+    },
+    {
+      "epoch": 0.4738673712702198,
+      "grad_norm": 0.005188761681522458,
+      "learning_rate": 5.850340136054422e-06,
+      "loss": 0.0,
+      "step": 7353
+    },
+    {
+      "epoch": 0.47393181671714896,
+      "grad_norm": 0.009026635366319133,
+      "learning_rate": 5.849624060150377e-06,
+      "loss": 0.0,
+      "step": 7354
+    },
+    {
+      "epoch": 0.4739962621640781,
+      "grad_norm": 0.00019702236665921518,
+      "learning_rate": 5.84890798424633e-06,
+      "loss": 0.0,
+      "step": 7355
+    },
+    {
+      "epoch": 0.47406070761100727,
+      "grad_norm": 3.311441360306938,
+      "learning_rate": 5.848191908342284e-06,
+      "loss": 0.0171,
+      "step": 7356
+    },
+    {
+      "epoch": 0.47412515305793645,
+      "grad_norm": 0.001216040447727667,
+      "learning_rate": 5.847475832438239e-06,
+      "loss": 0.0,
+      "step": 7357
+    },
+    {
+      "epoch": 0.47418959850486564,
+      "grad_norm": 0.17368396932281072,
+      "learning_rate": 5.846759756534193e-06,
+      "loss": 0.0028,
+      "step": 7358
+    },
+    {
+      "epoch": 0.4742540439517948,
+      "grad_norm": 0.156202195722558,
+      "learning_rate": 5.846043680630146e-06,
+      "loss": 0.0011,
+      "step": 7359
+    },
+    {
+      "epoch": 0.47431848939872395,
+      "grad_norm": 0.04618093117781858,
+      "learning_rate": 5.845327604726101e-06,
+      "loss": 0.0002,
+      "step": 7360
+    },
+    {
+      "epoch": 0.47438293484565314,
+      "grad_norm": 0.01748293277886574,
+      "learning_rate": 5.844611528822055e-06,
+      "loss": 0.0,
+      "step": 7361
+    },
+    {
+      "epoch": 0.4744473802925823,
+      "grad_norm": 0.0077936942194119655,
+      "learning_rate": 5.84389545291801e-06,
+      "loss": 0.0,
+      "step": 7362
+    },
+    {
+      "epoch": 0.4745118257395115,
+      "grad_norm": 0.0030220017533081076,
+      "learning_rate": 5.8431793770139645e-06,
+      "loss": 0.0,
+      "step": 7363
+    },
+    {
+      "epoch": 0.4745762711864407,
+      "grad_norm": 0.020481092922284175,
+      "learning_rate": 5.842463301109919e-06,
+      "loss": 0.0,
+      "step": 7364
+    },
+    {
+      "epoch": 0.4746407166333699,
+      "grad_norm": 0.047249753805527046,
+      "learning_rate": 5.841747225205873e-06,
+      "loss": 0.0004,
+      "step": 7365
+    },
+    {
+      "epoch": 0.474705162080299,
+      "grad_norm": 0.21330669411996359,
+      "learning_rate": 5.8410311493018265e-06,
+      "loss": 0.0006,
+      "step": 7366
+    },
+    {
+      "epoch": 0.4747696075272282,
+      "grad_norm": 2.6491441151497485e-05,
+      "learning_rate": 5.840315073397781e-06,
+      "loss": 0.0,
+      "step": 7367
+    },
+    {
+      "epoch": 0.4748340529741574,
+      "grad_norm": 0.0006470988794403943,
+      "learning_rate": 5.839598997493735e-06,
+      "loss": 0.0,
+      "step": 7368
+    },
+    {
+      "epoch": 0.47489849842108656,
+      "grad_norm": 6.626651095656053e-05,
+      "learning_rate": 5.838882921589689e-06,
+      "loss": 0.0,
+      "step": 7369
+    },
+    {
+      "epoch": 0.47496294386801574,
+      "grad_norm": 0.0010031601606788707,
+      "learning_rate": 5.838166845685644e-06,
+      "loss": 0.0,
+      "step": 7370
+    },
+    {
+      "epoch": 0.4750273893149449,
+      "grad_norm": 0.0006513101402277502,
+      "learning_rate": 5.837450769781597e-06,
+      "loss": 0.0,
+      "step": 7371
+    },
+    {
+      "epoch": 0.47509183476187405,
+      "grad_norm": 0.007908438327495618,
+      "learning_rate": 5.8367346938775515e-06,
+      "loss": 0.0,
+      "step": 7372
+    },
+    {
+      "epoch": 0.47515628020880324,
+      "grad_norm": 0.000893835609315527,
+      "learning_rate": 5.836018617973506e-06,
+      "loss": 0.0,
+      "step": 7373
+    },
+    {
+      "epoch": 0.4752207256557324,
+      "grad_norm": 0.0033915587171132375,
+      "learning_rate": 5.83530254206946e-06,
+      "loss": 0.0,
+      "step": 7374
+    },
+    {
+      "epoch": 0.4752851711026616,
+      "grad_norm": 0.01097830771040835,
+      "learning_rate": 5.8345864661654135e-06,
+      "loss": 0.0,
+      "step": 7375
+    },
+    {
+      "epoch": 0.4753496165495908,
+      "grad_norm": 0.0011861722306901786,
+      "learning_rate": 5.833870390261368e-06,
+      "loss": 0.0,
+      "step": 7376
+    },
+    {
+      "epoch": 0.4754140619965199,
+      "grad_norm": 0.039472695668676026,
+      "learning_rate": 5.833154314357322e-06,
+      "loss": 0.0001,
+      "step": 7377
+    },
+    {
+      "epoch": 0.4754785074434491,
+      "grad_norm": 0.00013474575537218255,
+      "learning_rate": 5.832438238453276e-06,
+      "loss": 0.0,
+      "step": 7378
+    },
+    {
+      "epoch": 0.4755429528903783,
+      "grad_norm": 0.004523991526979276,
+      "learning_rate": 5.831722162549231e-06,
+      "loss": 0.0,
+      "step": 7379
+    },
+    {
+      "epoch": 0.4756073983373075,
+      "grad_norm": 0.006835058807200215,
+      "learning_rate": 5.831006086645184e-06,
+      "loss": 0.0,
+      "step": 7380
+    },
+    {
+      "epoch": 0.47567184378423666,
+      "grad_norm": 0.05759036522811417,
+      "learning_rate": 5.8302900107411385e-06,
+      "loss": 0.0001,
+      "step": 7381
+    },
+    {
+      "epoch": 0.47573628923116584,
+      "grad_norm": 0.4775274076385219,
+      "learning_rate": 5.829573934837093e-06,
+      "loss": 0.0034,
+      "step": 7382
+    },
+    {
+      "epoch": 0.47580073467809497,
+      "grad_norm": 0.0005595710209434688,
+      "learning_rate": 5.828857858933047e-06,
+      "loss": 0.0,
+      "step": 7383
+    },
+    {
+      "epoch": 0.47586518012502416,
+      "grad_norm": 0.002950148607580095,
+      "learning_rate": 5.828141783029001e-06,
+      "loss": 0.0,
+      "step": 7384
+    },
+    {
+      "epoch": 0.47592962557195334,
+      "grad_norm": 0.10752587919006612,
+      "learning_rate": 5.8274257071249565e-06,
+      "loss": 0.0007,
+      "step": 7385
+    },
+    {
+      "epoch": 0.4759940710188825,
+      "grad_norm": 0.005091958235020094,
+      "learning_rate": 5.826709631220911e-06,
+      "loss": 0.0001,
+      "step": 7386
+    },
+    {
+      "epoch": 0.4760585164658117,
+      "grad_norm": 0.003841617842285557,
+      "learning_rate": 5.825993555316864e-06,
+      "loss": 0.0,
+      "step": 7387
+    },
+    {
+      "epoch": 0.47612296191274084,
+      "grad_norm": 0.0006383723237262383,
+      "learning_rate": 5.8252774794128185e-06,
+      "loss": 0.0,
+      "step": 7388
+    },
+    {
+      "epoch": 0.47618740735967,
+      "grad_norm": 0.004039966120740329,
+      "learning_rate": 5.824561403508773e-06,
+      "loss": 0.0,
+      "step": 7389
+    },
+    {
+      "epoch": 0.4762518528065992,
+      "grad_norm": 0.0010477007617992421,
+      "learning_rate": 5.823845327604727e-06,
+      "loss": 0.0,
+      "step": 7390
+    },
+    {
+      "epoch": 0.4763162982535284,
+      "grad_norm": 1.387736674906414,
+      "learning_rate": 5.823129251700681e-06,
+      "loss": 0.0068,
+      "step": 7391
+    },
+    {
+      "epoch": 0.4763807437004576,
+      "grad_norm": 0.07167502330915866,
+      "learning_rate": 5.822413175796635e-06,
+      "loss": 0.0002,
+      "step": 7392
+    },
+    {
+      "epoch": 0.47644518914738676,
+      "grad_norm": 0.014098552902959926,
+      "learning_rate": 5.821697099892589e-06,
+      "loss": 0.0,
+      "step": 7393
+    },
+    {
+      "epoch": 0.4765096345943159,
+      "grad_norm": 0.08331748195728007,
+      "learning_rate": 5.8209810239885435e-06,
+      "loss": 0.0001,
+      "step": 7394
+    },
+    {
+      "epoch": 0.4765740800412451,
+      "grad_norm": 0.08048932115789725,
+      "learning_rate": 5.820264948084498e-06,
+      "loss": 0.0002,
+      "step": 7395
+    },
+    {
+      "epoch": 0.47663852548817426,
+      "grad_norm": 0.8311032058054546,
+      "learning_rate": 5.819548872180451e-06,
+      "loss": 0.0028,
+      "step": 7396
+    },
+    {
+      "epoch": 0.47670297093510344,
+      "grad_norm": 0.20586471478306612,
+      "learning_rate": 5.8188327962764055e-06,
+      "loss": 0.0005,
+      "step": 7397
+    },
+    {
+      "epoch": 0.4767674163820326,
+      "grad_norm": 0.322638659886776,
+      "learning_rate": 5.81811672037236e-06,
+      "loss": 0.0031,
+      "step": 7398
+    },
+    {
+      "epoch": 0.4768318618289618,
+      "grad_norm": 0.011732446097699342,
+      "learning_rate": 5.817400644468314e-06,
+      "loss": 0.0001,
+      "step": 7399
+    },
+    {
+      "epoch": 0.47689630727589094,
+      "grad_norm": 0.15127922701807855,
+      "learning_rate": 5.8166845685642684e-06,
+      "loss": 0.0001,
+      "step": 7400
+    },
+    {
+      "epoch": 0.4769607527228201,
+      "grad_norm": 0.216969181012013,
+      "learning_rate": 5.815968492660222e-06,
+      "loss": 0.0025,
+      "step": 7401
+    },
+    {
+      "epoch": 0.4770251981697493,
+      "grad_norm": 0.30436968734146624,
+      "learning_rate": 5.815252416756176e-06,
+      "loss": 0.0004,
+      "step": 7402
+    },
+    {
+      "epoch": 0.4770896436166785,
+      "grad_norm": 0.07910840864342857,
+      "learning_rate": 5.8145363408521305e-06,
+      "loss": 0.0001,
+      "step": 7403
+    },
+    {
+      "epoch": 0.4771540890636077,
+      "grad_norm": 0.001726383251496407,
+      "learning_rate": 5.813820264948085e-06,
+      "loss": 0.0,
+      "step": 7404
+    },
+    {
+      "epoch": 0.4772185345105368,
+      "grad_norm": 0.0003675535003440943,
+      "learning_rate": 5.813104189044039e-06,
+      "loss": 0.0,
+      "step": 7405
+    },
+    {
+      "epoch": 0.477282979957466,
+      "grad_norm": 0.001882370654428137,
+      "learning_rate": 5.8123881131399925e-06,
+      "loss": 0.0,
+      "step": 7406
+    },
+    {
+      "epoch": 0.4773474254043952,
+      "grad_norm": 0.04665171072155993,
+      "learning_rate": 5.811672037235947e-06,
+      "loss": 0.0002,
+      "step": 7407
+    },
+    {
+      "epoch": 0.47741187085132436,
+      "grad_norm": 0.006549475831629919,
+      "learning_rate": 5.810955961331902e-06,
+      "loss": 0.0,
+      "step": 7408
+    },
+    {
+      "epoch": 0.47747631629825354,
+      "grad_norm": 0.004427687294087215,
+      "learning_rate": 5.810239885427856e-06,
+      "loss": 0.0,
+      "step": 7409
+    },
+    {
+      "epoch": 0.47754076174518273,
+      "grad_norm": 0.0017522578296534845,
+      "learning_rate": 5.8095238095238106e-06,
+      "loss": 0.0,
+      "step": 7410
+    },
+    {
+      "epoch": 0.47760520719211186,
+      "grad_norm": 0.3626807387078549,
+      "learning_rate": 5.808807733619765e-06,
+      "loss": 0.0017,
+      "step": 7411
+    },
+    {
+      "epoch": 0.47766965263904104,
+      "grad_norm": 0.0008852981396097059,
+      "learning_rate": 5.808091657715718e-06,
+      "loss": 0.0,
+      "step": 7412
+    },
+    {
+      "epoch": 0.4777340980859702,
+      "grad_norm": 0.023258903396705985,
+      "learning_rate": 5.807375581811673e-06,
+      "loss": 0.0,
+      "step": 7413
+    },
+    {
+      "epoch": 0.4777985435328994,
+      "grad_norm": 0.005061872213993539,
+      "learning_rate": 5.806659505907627e-06,
+      "loss": 0.0001,
+      "step": 7414
+    },
+    {
+      "epoch": 0.4778629889798286,
+      "grad_norm": 0.0009047437859875983,
+      "learning_rate": 5.805943430003581e-06,
+      "loss": 0.0,
+      "step": 7415
+    },
+    {
+      "epoch": 0.4779274344267577,
+      "grad_norm": 0.00862082379217804,
+      "learning_rate": 5.8052273540995355e-06,
+      "loss": 0.0,
+      "step": 7416
+    },
+    {
+      "epoch": 0.4779918798736869,
+      "grad_norm": 0.001022643294842138,
+      "learning_rate": 5.804511278195489e-06,
+      "loss": 0.0,
+      "step": 7417
+    },
+    {
+      "epoch": 0.4780563253206161,
+      "grad_norm": 0.00018940896883695436,
+      "learning_rate": 5.803795202291443e-06,
+      "loss": 0.0,
+      "step": 7418
+    },
+    {
+      "epoch": 0.4781207707675453,
+      "grad_norm": 0.002504940827306221,
+      "learning_rate": 5.8030791263873976e-06,
+      "loss": 0.0,
+      "step": 7419
+    },
+    {
+      "epoch": 0.47818521621447446,
+      "grad_norm": 0.00035435915281225466,
+      "learning_rate": 5.802363050483352e-06,
+      "loss": 0.0,
+      "step": 7420
+    },
+    {
+      "epoch": 0.47824966166140365,
+      "grad_norm": 0.0021682544293959963,
+      "learning_rate": 5.801646974579306e-06,
+      "loss": 0.0,
+      "step": 7421
+    },
+    {
+      "epoch": 0.4783141071083328,
+      "grad_norm": 0.01594937015862623,
+      "learning_rate": 5.80093089867526e-06,
+      "loss": 0.0001,
+      "step": 7422
+    },
+    {
+      "epoch": 0.47837855255526196,
+      "grad_norm": 0.0018651042764992,
+      "learning_rate": 5.800214822771214e-06,
+      "loss": 0.0,
+      "step": 7423
+    },
+    {
+      "epoch": 0.47844299800219114,
+      "grad_norm": 0.001665721738655338,
+      "learning_rate": 5.799498746867168e-06,
+      "loss": 0.0,
+      "step": 7424
+    },
+    {
+      "epoch": 0.47850744344912033,
+      "grad_norm": 0.0003566205222117585,
+      "learning_rate": 5.7987826709631225e-06,
+      "loss": 0.0,
+      "step": 7425
+    },
+    {
+      "epoch": 0.4785718888960495,
+      "grad_norm": 0.0008448971831598651,
+      "learning_rate": 5.798066595059077e-06,
+      "loss": 0.0,
+      "step": 7426
+    },
+    {
+      "epoch": 0.47863633434297864,
+      "grad_norm": 7.599035001878121e-05,
+      "learning_rate": 5.79735051915503e-06,
+      "loss": 0.0,
+      "step": 7427
+    },
+    {
+      "epoch": 0.4787007797899078,
+      "grad_norm": 0.06246002789145786,
+      "learning_rate": 5.7966344432509846e-06,
+      "loss": 0.0006,
+      "step": 7428
+    },
+    {
+      "epoch": 0.478765225236837,
+      "grad_norm": 0.0005704727887440845,
+      "learning_rate": 5.795918367346939e-06,
+      "loss": 0.0,
+      "step": 7429
+    },
+    {
+      "epoch": 0.4788296706837662,
+      "grad_norm": 9.001205747052077e-05,
+      "learning_rate": 5.795202291442893e-06,
+      "loss": 0.0,
+      "step": 7430
+    },
+    {
+      "epoch": 0.4788941161306954,
+      "grad_norm": 1.6705578758492294,
+      "learning_rate": 5.7944862155388475e-06,
+      "loss": 0.0206,
+      "step": 7431
+    },
+    {
+      "epoch": 0.47895856157762456,
+      "grad_norm": 0.0011345647547070011,
+      "learning_rate": 5.793770139634803e-06,
+      "loss": 0.0,
+      "step": 7432
+    },
+    {
+      "epoch": 0.4790230070245537,
+      "grad_norm": 0.14501953125,
+      "learning_rate": 5.793054063730756e-06,
+      "loss": 0.0003,
+      "step": 7433
+    },
+    {
+      "epoch": 0.4790874524714829,
+      "grad_norm": 0.8865466181852298,
+      "learning_rate": 5.79233798782671e-06,
+      "loss": 0.0128,
+      "step": 7434
+    },
+    {
+      "epoch": 0.47915189791841206,
+      "grad_norm": 0.3714381777882189,
+      "learning_rate": 5.791621911922665e-06,
+      "loss": 0.003,
+      "step": 7435
+    },
+    {
+      "epoch": 0.47921634336534125,
+      "grad_norm": 0.5642342588606207,
+      "learning_rate": 5.790905836018619e-06,
+      "loss": 0.0019,
+      "step": 7436
+    },
+    {
+      "epoch": 0.47928078881227043,
+      "grad_norm": 0.016275023984882227,
+      "learning_rate": 5.790189760114573e-06,
+      "loss": 0.0002,
+      "step": 7437
+    },
+    {
+      "epoch": 0.4793452342591996,
+      "grad_norm": 0.008938399006915438,
+      "learning_rate": 5.789473684210527e-06,
+      "loss": 0.0,
+      "step": 7438
+    },
+    {
+      "epoch": 0.47940967970612874,
+      "grad_norm": 0.0010685410566123547,
+      "learning_rate": 5.788757608306481e-06,
+      "loss": 0.0,
+      "step": 7439
+    },
+    {
+      "epoch": 0.47947412515305793,
+      "grad_norm": 0.23940940258627813,
+      "learning_rate": 5.788041532402435e-06,
+      "loss": 0.0005,
+      "step": 7440
+    },
+    {
+      "epoch": 0.4795385705999871,
+      "grad_norm": 1.1296119464180416,
+      "learning_rate": 5.78732545649839e-06,
+      "loss": 0.0082,
+      "step": 7441
+    },
+    {
+      "epoch": 0.4796030160469163,
+      "grad_norm": 0.0049990550644581685,
+      "learning_rate": 5.786609380594344e-06,
+      "loss": 0.0,
+      "step": 7442
+    },
+    {
+      "epoch": 0.4796674614938455,
+      "grad_norm": 0.0003853361686928715,
+      "learning_rate": 5.785893304690297e-06,
+      "loss": 0.0,
+      "step": 7443
+    },
+    {
+      "epoch": 0.4797319069407746,
+      "grad_norm": 0.012548110204785096,
+      "learning_rate": 5.785177228786252e-06,
+      "loss": 0.0001,
+      "step": 7444
+    },
+    {
+      "epoch": 0.4797963523877038,
+      "grad_norm": 0.01280574895032373,
+      "learning_rate": 5.784461152882206e-06,
+      "loss": 0.0,
+      "step": 7445
+    },
+    {
+      "epoch": 0.479860797834633,
+      "grad_norm": 0.04450895395433987,
+      "learning_rate": 5.78374507697816e-06,
+      "loss": 0.0001,
+      "step": 7446
+    },
+    {
+      "epoch": 0.47992524328156216,
+      "grad_norm": 0.21640394877508198,
+      "learning_rate": 5.7830290010741145e-06,
+      "loss": 0.002,
+      "step": 7447
+    },
+    {
+      "epoch": 0.47998968872849135,
+      "grad_norm": 0.023149261464048686,
+      "learning_rate": 5.782312925170068e-06,
+      "loss": 0.0,
+      "step": 7448
+    },
+    {
+      "epoch": 0.48005413417542053,
+      "grad_norm": 0.07784700405213824,
+      "learning_rate": 5.781596849266022e-06,
+      "loss": 0.0001,
+      "step": 7449
+    },
+    {
+      "epoch": 0.48011857962234966,
+      "grad_norm": 0.004143772080952122,
+      "learning_rate": 5.780880773361977e-06,
+      "loss": 0.0,
+      "step": 7450
+    },
+    {
+      "epoch": 0.48018302506927885,
+      "grad_norm": 0.02690683351179182,
+      "learning_rate": 5.780164697457931e-06,
+      "loss": 0.0001,
+      "step": 7451
+    },
+    {
+      "epoch": 0.48024747051620803,
+      "grad_norm": 0.15777312226534332,
+      "learning_rate": 5.779448621553884e-06,
+      "loss": 0.0001,
+      "step": 7452
+    },
+    {
+      "epoch": 0.4803119159631372,
+      "grad_norm": 0.011854422022662701,
+      "learning_rate": 5.778732545649839e-06,
+      "loss": 0.0001,
+      "step": 7453
+    },
+    {
+      "epoch": 0.4803763614100664,
+      "grad_norm": 0.0246195620217399,
+      "learning_rate": 5.778016469745793e-06,
+      "loss": 0.0,
+      "step": 7454
+    },
+    {
+      "epoch": 0.48044080685699553,
+      "grad_norm": 0.004674714612297223,
+      "learning_rate": 5.777300393841748e-06,
+      "loss": 0.0,
+      "step": 7455
+    },
+    {
+      "epoch": 0.4805052523039247,
+      "grad_norm": 0.08145114731413798,
+      "learning_rate": 5.776584317937702e-06,
+      "loss": 0.0001,
+      "step": 7456
+    },
+    {
+      "epoch": 0.4805696977508539,
+      "grad_norm": 0.0011909823012698556,
+      "learning_rate": 5.775868242033657e-06,
+      "loss": 0.0,
+      "step": 7457
+    },
+    {
+      "epoch": 0.4806341431977831,
+      "grad_norm": 0.0018303589815469893,
+      "learning_rate": 5.775152166129611e-06,
+      "loss": 0.0,
+      "step": 7458
+    },
+    {
+      "epoch": 0.48069858864471227,
+      "grad_norm": 0.010239236917028723,
+      "learning_rate": 5.7744360902255644e-06,
+      "loss": 0.0,
+      "step": 7459
+    },
+    {
+      "epoch": 0.48076303409164145,
+      "grad_norm": 0.09835660615437058,
+      "learning_rate": 5.773720014321519e-06,
+      "loss": 0.0001,
+      "step": 7460
+    },
+    {
+      "epoch": 0.4808274795385706,
+      "grad_norm": 0.014118260546048035,
+      "learning_rate": 5.773003938417473e-06,
+      "loss": 0.0001,
+      "step": 7461
+    },
+    {
+      "epoch": 0.48089192498549976,
+      "grad_norm": 0.000602907625883965,
+      "learning_rate": 5.772287862513427e-06,
+      "loss": 0.0,
+      "step": 7462
+    },
+    {
+      "epoch": 0.48095637043242895,
+      "grad_norm": 0.06684985026054438,
+      "learning_rate": 5.771571786609382e-06,
+      "loss": 0.0001,
+      "step": 7463
+    },
+    {
+      "epoch": 0.48102081587935813,
+      "grad_norm": 0.17200650256293518,
+      "learning_rate": 5.770855710705335e-06,
+      "loss": 0.0016,
+      "step": 7464
+    },
+    {
+      "epoch": 0.4810852613262873,
+      "grad_norm": 0.021725604178362307,
+      "learning_rate": 5.770139634801289e-06,
+      "loss": 0.0,
+      "step": 7465
+    },
+    {
+      "epoch": 0.48114970677321645,
+      "grad_norm": 0.01244272677275712,
+      "learning_rate": 5.769423558897244e-06,
+      "loss": 0.0,
+      "step": 7466
+    },
+    {
+      "epoch": 0.48121415222014563,
+      "grad_norm": 0.0024778840976099475,
+      "learning_rate": 5.768707482993198e-06,
+      "loss": 0.0,
+      "step": 7467
+    },
+    {
+      "epoch": 0.4812785976670748,
+      "grad_norm": 0.00267979224785924,
+      "learning_rate": 5.7679914070891514e-06,
+      "loss": 0.0,
+      "step": 7468
+    },
+    {
+      "epoch": 0.481343043114004,
+      "grad_norm": 0.003459435935471662,
+      "learning_rate": 5.767275331185106e-06,
+      "loss": 0.0,
+      "step": 7469
+    },
+    {
+      "epoch": 0.4814074885609332,
+      "grad_norm": 0.44937154410443325,
+      "learning_rate": 5.76655925528106e-06,
+      "loss": 0.0028,
+      "step": 7470
+    },
+    {
+      "epoch": 0.48147193400786237,
+      "grad_norm": 0.1190897293505687,
+      "learning_rate": 5.765843179377014e-06,
+      "loss": 0.0003,
+      "step": 7471
+    },
+    {
+      "epoch": 0.4815363794547915,
+      "grad_norm": 0.00193991626554029,
+      "learning_rate": 5.765127103472969e-06,
+      "loss": 0.0,
+      "step": 7472
+    },
+    {
+      "epoch": 0.4816008249017207,
+      "grad_norm": 0.0002928079372750345,
+      "learning_rate": 5.764411027568922e-06,
+      "loss": 0.0,
+      "step": 7473
+    },
+    {
+      "epoch": 0.48166527034864987,
+      "grad_norm": 0.0026289588066395926,
+      "learning_rate": 5.763694951664876e-06,
+      "loss": 0.0,
+      "step": 7474
+    },
+    {
+      "epoch": 0.48172971579557905,
+      "grad_norm": 0.016364185159167613,
+      "learning_rate": 5.762978875760831e-06,
+      "loss": 0.0002,
+      "step": 7475
+    },
+    {
+      "epoch": 0.48179416124250823,
+      "grad_norm": 0.002886073119532732,
+      "learning_rate": 5.762262799856785e-06,
+      "loss": 0.0,
+      "step": 7476
+    },
+    {
+      "epoch": 0.4818586066894374,
+      "grad_norm": 0.0005045980107262203,
+      "learning_rate": 5.761546723952739e-06,
+      "loss": 0.0,
+      "step": 7477
+    },
+    {
+      "epoch": 0.48192305213636655,
+      "grad_norm": 0.012034130486031228,
+      "learning_rate": 5.760830648048694e-06,
+      "loss": 0.0001,
+      "step": 7478
+    },
+    {
+      "epoch": 0.48198749758329573,
+      "grad_norm": 0.0009000554618051204,
+      "learning_rate": 5.760114572144649e-06,
+      "loss": 0.0,
+      "step": 7479
+    },
+    {
+      "epoch": 0.4820519430302249,
+      "grad_norm": 0.00965203255101447,
+      "learning_rate": 5.759398496240602e-06,
+      "loss": 0.0,
+      "step": 7480
+    },
+    {
+      "epoch": 0.4821163884771541,
+      "grad_norm": 0.05475154813132867,
+      "learning_rate": 5.7586824203365565e-06,
+      "loss": 0.0001,
+      "step": 7481
+    },
+    {
+      "epoch": 0.4821808339240833,
+      "grad_norm": 0.020473049849451586,
+      "learning_rate": 5.757966344432511e-06,
+      "loss": 0.0002,
+      "step": 7482
+    },
+    {
+      "epoch": 0.4822452793710124,
+      "grad_norm": 0.0010466034143123027,
+      "learning_rate": 5.757250268528465e-06,
+      "loss": 0.0,
+      "step": 7483
+    },
+    {
+      "epoch": 0.4823097248179416,
+      "grad_norm": 0.018919086369527306,
+      "learning_rate": 5.7565341926244185e-06,
+      "loss": 0.0001,
+      "step": 7484
+    },
+    {
+      "epoch": 0.4823741702648708,
+      "grad_norm": 0.0019212618847136709,
+      "learning_rate": 5.755818116720373e-06,
+      "loss": 0.0,
+      "step": 7485
+    },
+    {
+      "epoch": 0.48243861571179997,
+      "grad_norm": 0.00398771180909543,
+      "learning_rate": 5.755102040816327e-06,
+      "loss": 0.0,
+      "step": 7486
+    },
+    {
+      "epoch": 0.48250306115872915,
+      "grad_norm": 0.009733597569587743,
+      "learning_rate": 5.754385964912281e-06,
+      "loss": 0.0,
+      "step": 7487
+    },
+    {
+      "epoch": 0.48256750660565834,
+      "grad_norm": 0.00010193001177631241,
+      "learning_rate": 5.753669889008236e-06,
+      "loss": 0.0,
+      "step": 7488
+    },
+    {
+      "epoch": 0.48263195205258747,
+      "grad_norm": 0.0004880647422603623,
+      "learning_rate": 5.752953813104189e-06,
+      "loss": 0.0,
+      "step": 7489
+    },
+    {
+      "epoch": 0.48269639749951665,
+      "grad_norm": 0.00360417876237237,
+      "learning_rate": 5.7522377372001435e-06,
+      "loss": 0.0,
+      "step": 7490
+    },
+    {
+      "epoch": 0.48276084294644583,
+      "grad_norm": 0.0004049090707836464,
+      "learning_rate": 5.751521661296098e-06,
+      "loss": 0.0,
+      "step": 7491
+    },
+    {
+      "epoch": 0.482825288393375,
+      "grad_norm": 0.0013167952446200002,
+      "learning_rate": 5.750805585392052e-06,
+      "loss": 0.0,
+      "step": 7492
+    },
+    {
+      "epoch": 0.4828897338403042,
+      "grad_norm": 0.025174958117207765,
+      "learning_rate": 5.750089509488006e-06,
+      "loss": 0.0002,
+      "step": 7493
+    },
+    {
+      "epoch": 0.48295417928723333,
+      "grad_norm": 0.06014975165900907,
+      "learning_rate": 5.74937343358396e-06,
+      "loss": 0.0016,
+      "step": 7494
+    },
+    {
+      "epoch": 0.4830186247341625,
+      "grad_norm": 0.012060249113546211,
+      "learning_rate": 5.748657357679914e-06,
+      "loss": 0.0,
+      "step": 7495
+    },
+    {
+      "epoch": 0.4830830701810917,
+      "grad_norm": 0.12713788599629475,
+      "learning_rate": 5.747941281775868e-06,
+      "loss": 0.0028,
+      "step": 7496
+    },
+    {
+      "epoch": 0.4831475156280209,
+      "grad_norm": 0.005913938901135187,
+      "learning_rate": 5.747225205871823e-06,
+      "loss": 0.0,
+      "step": 7497
+    },
+    {
+      "epoch": 0.48321196107495007,
+      "grad_norm": 0.00039260180087013206,
+      "learning_rate": 5.746509129967777e-06,
+      "loss": 0.0,
+      "step": 7498
+    },
+    {
+      "epoch": 0.48327640652187925,
+      "grad_norm": 0.014482109685575993,
+      "learning_rate": 5.7457930540637305e-06,
+      "loss": 0.0001,
+      "step": 7499
+    },
+    {
+      "epoch": 0.4833408519688084,
+      "grad_norm": 0.00447910521815926,
+      "learning_rate": 5.745076978159685e-06,
+      "loss": 0.0,
+      "step": 7500
+    },
+    {
+      "epoch": 0.48340529741573757,
+      "grad_norm": 0.026631144486073698,
+      "learning_rate": 5.744360902255639e-06,
+      "loss": 0.0001,
+      "step": 7501
+    },
+    {
+      "epoch": 0.48346974286266675,
+      "grad_norm": 0.012365717619568232,
+      "learning_rate": 5.743644826351594e-06,
+      "loss": 0.0001,
+      "step": 7502
+    },
+    {
+      "epoch": 0.48353418830959594,
+      "grad_norm": 0.000551421378977375,
+      "learning_rate": 5.7429287504475485e-06,
+      "loss": 0.0,
+      "step": 7503
+    },
+    {
+      "epoch": 0.4835986337565251,
+      "grad_norm": 0.007876399816532313,
+      "learning_rate": 5.742212674543503e-06,
+      "loss": 0.0001,
+      "step": 7504
+    },
+    {
+      "epoch": 0.48366307920345425,
+      "grad_norm": 0.001752821937067081,
+      "learning_rate": 5.741496598639456e-06,
+      "loss": 0.0,
+      "step": 7505
+    },
+    {
+      "epoch": 0.48372752465038343,
+      "grad_norm": 0.002184141243038027,
+      "learning_rate": 5.7407805227354105e-06,
+      "loss": 0.0,
+      "step": 7506
+    },
+    {
+      "epoch": 0.4837919700973126,
+      "grad_norm": 0.19810261064370907,
+      "learning_rate": 5.740064446831365e-06,
+      "loss": 0.0009,
+      "step": 7507
+    },
+    {
+      "epoch": 0.4838564155442418,
+      "grad_norm": 0.0006581602418458326,
+      "learning_rate": 5.739348370927319e-06,
+      "loss": 0.0,
+      "step": 7508
+    },
+    {
+      "epoch": 0.483920860991171,
+      "grad_norm": 0.00032189703491337544,
+      "learning_rate": 5.7386322950232734e-06,
+      "loss": 0.0,
+      "step": 7509
+    },
+    {
+      "epoch": 0.48398530643810017,
+      "grad_norm": 0.0007444887274812117,
+      "learning_rate": 5.737916219119227e-06,
+      "loss": 0.0,
+      "step": 7510
+    },
+    {
+      "epoch": 0.4840497518850293,
+      "grad_norm": 0.0661404739703034,
+      "learning_rate": 5.737200143215181e-06,
+      "loss": 0.0002,
+      "step": 7511
+    },
+    {
+      "epoch": 0.4841141973319585,
+      "grad_norm": 0.020757209588938764,
+      "learning_rate": 5.7364840673111355e-06,
+      "loss": 0.0001,
+      "step": 7512
+    },
+    {
+      "epoch": 0.48417864277888767,
+      "grad_norm": 0.005867169331984068,
+      "learning_rate": 5.73576799140709e-06,
+      "loss": 0.0,
+      "step": 7513
+    },
+    {
+      "epoch": 0.48424308822581685,
+      "grad_norm": 0.00047232161720890626,
+      "learning_rate": 5.735051915503044e-06,
+      "loss": 0.0,
+      "step": 7514
+    },
+    {
+      "epoch": 0.48430753367274604,
+      "grad_norm": 0.006958581875660543,
+      "learning_rate": 5.7343358395989975e-06,
+      "loss": 0.0,
+      "step": 7515
+    },
+    {
+      "epoch": 0.4843719791196752,
+      "grad_norm": 0.0033243687316280453,
+      "learning_rate": 5.733619763694952e-06,
+      "loss": 0.0,
+      "step": 7516
+    },
+    {
+      "epoch": 0.48443642456660435,
+      "grad_norm": 0.00025208480732531735,
+      "learning_rate": 5.732903687790906e-06,
+      "loss": 0.0,
+      "step": 7517
+    },
+    {
+      "epoch": 0.48450087001353354,
+      "grad_norm": 0.014381225038362599,
+      "learning_rate": 5.7321876118868604e-06,
+      "loss": 0.0,
+      "step": 7518
+    },
+    {
+      "epoch": 0.4845653154604627,
+      "grad_norm": 0.2902895555370628,
+      "learning_rate": 5.731471535982815e-06,
+      "loss": 0.0006,
+      "step": 7519
+    },
+    {
+      "epoch": 0.4846297609073919,
+      "grad_norm": 0.004500232171590274,
+      "learning_rate": 5.730755460078768e-06,
+      "loss": 0.0,
+      "step": 7520
+    },
+    {
+      "epoch": 0.4846942063543211,
+      "grad_norm": 0.005529903035349548,
+      "learning_rate": 5.7300393841747225e-06,
+      "loss": 0.0,
+      "step": 7521
+    },
+    {
+      "epoch": 0.4847586518012502,
+      "grad_norm": 0.0014221680833822147,
+      "learning_rate": 5.729323308270677e-06,
+      "loss": 0.0,
+      "step": 7522
+    },
+    {
+      "epoch": 0.4848230972481794,
+      "grad_norm": 0.0029934765826047783,
+      "learning_rate": 5.728607232366631e-06,
+      "loss": 0.0,
+      "step": 7523
+    },
+    {
+      "epoch": 0.4848875426951086,
+      "grad_norm": 0.0015283611343524148,
+      "learning_rate": 5.727891156462585e-06,
+      "loss": 0.0,
+      "step": 7524
+    },
+    {
+      "epoch": 0.48495198814203777,
+      "grad_norm": 0.0019634857246066248,
+      "learning_rate": 5.7271750805585405e-06,
+      "loss": 0.0,
+      "step": 7525
+    },
+    {
+      "epoch": 0.48501643358896696,
+      "grad_norm": 0.0034480413744822237,
+      "learning_rate": 5.726459004654494e-06,
+      "loss": 0.0,
+      "step": 7526
+    },
+    {
+      "epoch": 0.48508087903589614,
+      "grad_norm": 0.0026394266389676234,
+      "learning_rate": 5.725742928750448e-06,
+      "loss": 0.0,
+      "step": 7527
+    },
+    {
+      "epoch": 0.48514532448282527,
+      "grad_norm": 0.014411129078745335,
+      "learning_rate": 5.7250268528464026e-06,
+      "loss": 0.0001,
+      "step": 7528
+    },
+    {
+      "epoch": 0.48520976992975445,
+      "grad_norm": 0.0067901182709101905,
+      "learning_rate": 5.724310776942357e-06,
+      "loss": 0.0,
+      "step": 7529
+    },
+    {
+      "epoch": 0.48527421537668364,
+      "grad_norm": 0.002589366357544819,
+      "learning_rate": 5.723594701038311e-06,
+      "loss": 0.0,
+      "step": 7530
+    },
+    {
+      "epoch": 0.4853386608236128,
+      "grad_norm": 0.2295686716096039,
+      "learning_rate": 5.722878625134265e-06,
+      "loss": 0.0002,
+      "step": 7531
+    },
+    {
+      "epoch": 0.485403106270542,
+      "grad_norm": 0.005061924499310147,
+      "learning_rate": 5.722162549230219e-06,
+      "loss": 0.0,
+      "step": 7532
+    },
+    {
+      "epoch": 0.48546755171747114,
+      "grad_norm": 0.011882524196712532,
+      "learning_rate": 5.721446473326173e-06,
+      "loss": 0.0001,
+      "step": 7533
+    },
+    {
+      "epoch": 0.4855319971644003,
+      "grad_norm": 0.16392287489124577,
+      "learning_rate": 5.7207303974221275e-06,
+      "loss": 0.0008,
+      "step": 7534
+    },
+    {
+      "epoch": 0.4855964426113295,
+      "grad_norm": 0.002512980230831525,
+      "learning_rate": 5.720014321518082e-06,
+      "loss": 0.0,
+      "step": 7535
+    },
+    {
+      "epoch": 0.4856608880582587,
+      "grad_norm": 0.006464325307909982,
+      "learning_rate": 5.719298245614035e-06,
+      "loss": 0.0,
+      "step": 7536
+    },
+    {
+      "epoch": 0.4857253335051879,
+      "grad_norm": 0.001662718931743277,
+      "learning_rate": 5.7185821697099896e-06,
+      "loss": 0.0,
+      "step": 7537
+    },
+    {
+      "epoch": 0.48578977895211706,
+      "grad_norm": 0.00312735401489503,
+      "learning_rate": 5.717866093805944e-06,
+      "loss": 0.0,
+      "step": 7538
+    },
+    {
+      "epoch": 0.4858542243990462,
+      "grad_norm": 0.002918325996476907,
+      "learning_rate": 5.717150017901898e-06,
+      "loss": 0.0,
+      "step": 7539
+    },
+    {
+      "epoch": 0.48591866984597537,
+      "grad_norm": 0.0013583321002518408,
+      "learning_rate": 5.7164339419978525e-06,
+      "loss": 0.0,
+      "step": 7540
+    },
+    {
+      "epoch": 0.48598311529290456,
+      "grad_norm": 0.03570830831840738,
+      "learning_rate": 5.715717866093806e-06,
+      "loss": 0.0001,
+      "step": 7541
+    },
+    {
+      "epoch": 0.48604756073983374,
+      "grad_norm": 0.06065757499082424,
+      "learning_rate": 5.71500179018976e-06,
+      "loss": 0.0001,
+      "step": 7542
+    },
+    {
+      "epoch": 0.4861120061867629,
+      "grad_norm": 0.00027401369607215015,
+      "learning_rate": 5.7142857142857145e-06,
+      "loss": 0.0,
+      "step": 7543
+    },
+    {
+      "epoch": 0.48617645163369205,
+      "grad_norm": 0.40300723802844124,
+      "learning_rate": 5.713569638381669e-06,
+      "loss": 0.0053,
+      "step": 7544
+    },
+    {
+      "epoch": 0.48624089708062124,
+      "grad_norm": 0.0010218532965069267,
+      "learning_rate": 5.712853562477622e-06,
+      "loss": 0.0,
+      "step": 7545
+    },
+    {
+      "epoch": 0.4863053425275504,
+      "grad_norm": 0.00012427245394306858,
+      "learning_rate": 5.7121374865735766e-06,
+      "loss": 0.0,
+      "step": 7546
+    },
+    {
+      "epoch": 0.4863697879744796,
+      "grad_norm": 0.00135554513866112,
+      "learning_rate": 5.711421410669531e-06,
+      "loss": 0.0,
+      "step": 7547
+    },
+    {
+      "epoch": 0.4864342334214088,
+      "grad_norm": 0.009465878989607369,
+      "learning_rate": 5.710705334765485e-06,
+      "loss": 0.0,
+      "step": 7548
+    },
+    {
+      "epoch": 0.486498678868338,
+      "grad_norm": 0.003342775812316762,
+      "learning_rate": 5.70998925886144e-06,
+      "loss": 0.0001,
+      "step": 7549
+    },
+    {
+      "epoch": 0.4865631243152671,
+      "grad_norm": 0.722024424887394,
+      "learning_rate": 5.709273182957395e-06,
+      "loss": 0.0034,
+      "step": 7550
+    },
+    {
+      "epoch": 0.4866275697621963,
+      "grad_norm": 0.0004114479573313148,
+      "learning_rate": 5.708557107053349e-06,
+      "loss": 0.0,
+      "step": 7551
+    },
+    {
+      "epoch": 0.4866920152091255,
+      "grad_norm": 0.0004321755849311948,
+      "learning_rate": 5.707841031149302e-06,
+      "loss": 0.0,
+      "step": 7552
+    },
+    {
+      "epoch": 0.48675646065605466,
+      "grad_norm": 0.0028718780582452424,
+      "learning_rate": 5.707124955245257e-06,
+      "loss": 0.0,
+      "step": 7553
+    },
+    {
+      "epoch": 0.48682090610298384,
+      "grad_norm": 0.9203624601679466,
+      "learning_rate": 5.706408879341211e-06,
+      "loss": 0.0046,
+      "step": 7554
+    },
+    {
+      "epoch": 0.486885351549913,
+      "grad_norm": 0.0048342640045114255,
+      "learning_rate": 5.705692803437165e-06,
+      "loss": 0.0,
+      "step": 7555
+    },
+    {
+      "epoch": 0.48694979699684215,
+      "grad_norm": 0.0002541909663165527,
+      "learning_rate": 5.7049767275331195e-06,
+      "loss": 0.0,
+      "step": 7556
+    },
+    {
+      "epoch": 0.48701424244377134,
+      "grad_norm": 0.0004487357840125776,
+      "learning_rate": 5.704260651629073e-06,
+      "loss": 0.0,
+      "step": 7557
+    },
+    {
+      "epoch": 0.4870786878907005,
+      "grad_norm": 0.0008346864542774917,
+      "learning_rate": 5.703544575725027e-06,
+      "loss": 0.0,
+      "step": 7558
+    },
+    {
+      "epoch": 0.4871431333376297,
+      "grad_norm": 0.0018916479246447236,
+      "learning_rate": 5.702828499820982e-06,
+      "loss": 0.0,
+      "step": 7559
+    },
+    {
+      "epoch": 0.4872075787845589,
+      "grad_norm": 0.018288931008625598,
+      "learning_rate": 5.702112423916936e-06,
+      "loss": 0.0,
+      "step": 7560
+    },
+    {
+      "epoch": 0.487272024231488,
+      "grad_norm": 0.0010530625038883763,
+      "learning_rate": 5.701396348012889e-06,
+      "loss": 0.0,
+      "step": 7561
+    },
+    {
+      "epoch": 0.4873364696784172,
+      "grad_norm": 0.007322550871804302,
+      "learning_rate": 5.700680272108844e-06,
+      "loss": 0.0,
+      "step": 7562
+    },
+    {
+      "epoch": 0.4874009151253464,
+      "grad_norm": 0.003065408485538047,
+      "learning_rate": 5.699964196204798e-06,
+      "loss": 0.0,
+      "step": 7563
+    },
+    {
+      "epoch": 0.4874653605722756,
+      "grad_norm": 0.0006956848235609778,
+      "learning_rate": 5.699248120300752e-06,
+      "loss": 0.0,
+      "step": 7564
+    },
+    {
+      "epoch": 0.48752980601920476,
+      "grad_norm": 0.1693866565782438,
+      "learning_rate": 5.6985320443967065e-06,
+      "loss": 0.001,
+      "step": 7565
+    },
+    {
+      "epoch": 0.48759425146613394,
+      "grad_norm": 0.1538983159029183,
+      "learning_rate": 5.69781596849266e-06,
+      "loss": 0.0002,
+      "step": 7566
+    },
+    {
+      "epoch": 0.48765869691306307,
+      "grad_norm": 0.0006069839695752836,
+      "learning_rate": 5.697099892588614e-06,
+      "loss": 0.0,
+      "step": 7567
+    },
+    {
+      "epoch": 0.48772314235999226,
+      "grad_norm": 0.0033471887203188194,
+      "learning_rate": 5.696383816684569e-06,
+      "loss": 0.0,
+      "step": 7568
+    },
+    {
+      "epoch": 0.48778758780692144,
+      "grad_norm": 0.06231746804088351,
+      "learning_rate": 5.695667740780523e-06,
+      "loss": 0.0006,
+      "step": 7569
+    },
+    {
+      "epoch": 0.4878520332538506,
+      "grad_norm": 0.016381977460167466,
+      "learning_rate": 5.694951664876477e-06,
+      "loss": 0.0001,
+      "step": 7570
+    },
+    {
+      "epoch": 0.4879164787007798,
+      "grad_norm": 0.003621855047769446,
+      "learning_rate": 5.694235588972431e-06,
+      "loss": 0.0,
+      "step": 7571
+    },
+    {
+      "epoch": 0.48798092414770894,
+      "grad_norm": 0.00014088177846377143,
+      "learning_rate": 5.693519513068387e-06,
+      "loss": 0.0,
+      "step": 7572
+    },
+    {
+      "epoch": 0.4880453695946381,
+      "grad_norm": 0.0014828029465861036,
+      "learning_rate": 5.69280343716434e-06,
+      "loss": 0.0,
+      "step": 7573
+    },
+    {
+      "epoch": 0.4881098150415673,
+      "grad_norm": 0.012641590761369413,
+      "learning_rate": 5.692087361260294e-06,
+      "loss": 0.0001,
+      "step": 7574
+    },
+    {
+      "epoch": 0.4881742604884965,
+      "grad_norm": 0.001850971111009934,
+      "learning_rate": 5.691371285356249e-06,
+      "loss": 0.0,
+      "step": 7575
+    },
+    {
+      "epoch": 0.4882387059354257,
+      "grad_norm": 0.046230391549455706,
+      "learning_rate": 5.690655209452203e-06,
+      "loss": 0.0001,
+      "step": 7576
+    },
+    {
+      "epoch": 0.48830315138235486,
+      "grad_norm": 0.01488584860814521,
+      "learning_rate": 5.689939133548157e-06,
+      "loss": 0.0001,
+      "step": 7577
+    },
+    {
+      "epoch": 0.488367596829284,
+      "grad_norm": 0.000514240191141193,
+      "learning_rate": 5.689223057644111e-06,
+      "loss": 0.0,
+      "step": 7578
+    },
+    {
+      "epoch": 0.4884320422762132,
+      "grad_norm": 0.008130484260185939,
+      "learning_rate": 5.688506981740065e-06,
+      "loss": 0.0001,
+      "step": 7579
+    },
+    {
+      "epoch": 0.48849648772314236,
+      "grad_norm": 0.0030046709164151357,
+      "learning_rate": 5.687790905836019e-06,
+      "loss": 0.0,
+      "step": 7580
+    },
+    {
+      "epoch": 0.48856093317007154,
+      "grad_norm": 0.015805491853670892,
+      "learning_rate": 5.687074829931974e-06,
+      "loss": 0.0002,
+      "step": 7581
+    },
+    {
+      "epoch": 0.4886253786170007,
+      "grad_norm": 0.011213436765041867,
+      "learning_rate": 5.686358754027927e-06,
+      "loss": 0.0,
+      "step": 7582
+    },
+    {
+      "epoch": 0.48868982406392986,
+      "grad_norm": 0.0007413171695860233,
+      "learning_rate": 5.685642678123881e-06,
+      "loss": 0.0,
+      "step": 7583
+    },
+    {
+      "epoch": 0.48875426951085904,
+      "grad_norm": 0.032078999380731485,
+      "learning_rate": 5.684926602219836e-06,
+      "loss": 0.0001,
+      "step": 7584
+    },
+    {
+      "epoch": 0.4888187149577882,
+      "grad_norm": 0.015428289726630193,
+      "learning_rate": 5.68421052631579e-06,
+      "loss": 0.0001,
+      "step": 7585
+    },
+    {
+      "epoch": 0.4888831604047174,
+      "grad_norm": 0.7430837339681301,
+      "learning_rate": 5.683494450411744e-06,
+      "loss": 0.003,
+      "step": 7586
+    },
+    {
+      "epoch": 0.4889476058516466,
+      "grad_norm": 0.16889645305455697,
+      "learning_rate": 5.682778374507698e-06,
+      "loss": 0.0006,
+      "step": 7587
+    },
+    {
+      "epoch": 0.4890120512985758,
+      "grad_norm": 0.01572568683559918,
+      "learning_rate": 5.682062298603652e-06,
+      "loss": 0.0,
+      "step": 7588
+    },
+    {
+      "epoch": 0.4890764967455049,
+      "grad_norm": 0.06375657091566687,
+      "learning_rate": 5.681346222699606e-06,
+      "loss": 0.0001,
+      "step": 7589
+    },
+    {
+      "epoch": 0.4891409421924341,
+      "grad_norm": 0.0013207851323213868,
+      "learning_rate": 5.680630146795561e-06,
+      "loss": 0.0,
+      "step": 7590
+    },
+    {
+      "epoch": 0.4892053876393633,
+      "grad_norm": 0.0013833316345360025,
+      "learning_rate": 5.679914070891515e-06,
+      "loss": 0.0,
+      "step": 7591
+    },
+    {
+      "epoch": 0.48926983308629246,
+      "grad_norm": 0.002289494076961141,
+      "learning_rate": 5.679197994987468e-06,
+      "loss": 0.0,
+      "step": 7592
+    },
+    {
+      "epoch": 0.48933427853322164,
+      "grad_norm": 0.0002545245494613917,
+      "learning_rate": 5.678481919083423e-06,
+      "loss": 0.0,
+      "step": 7593
+    },
+    {
+      "epoch": 0.48939872398015083,
+      "grad_norm": 0.001060228307932159,
+      "learning_rate": 5.677765843179377e-06,
+      "loss": 0.0,
+      "step": 7594
+    },
+    {
+      "epoch": 0.48946316942707996,
+      "grad_norm": 0.03487591100642847,
+      "learning_rate": 5.677049767275332e-06,
+      "loss": 0.0,
+      "step": 7595
+    },
+    {
+      "epoch": 0.48952761487400914,
+      "grad_norm": 0.0168614849164812,
+      "learning_rate": 5.676333691371286e-06,
+      "loss": 0.0001,
+      "step": 7596
+    },
+    {
+      "epoch": 0.4895920603209383,
+      "grad_norm": 0.0006910856059586328,
+      "learning_rate": 5.675617615467241e-06,
+      "loss": 0.0,
+      "step": 7597
+    },
+    {
+      "epoch": 0.4896565057678675,
+      "grad_norm": 0.07767658589564769,
+      "learning_rate": 5.674901539563194e-06,
+      "loss": 0.0,
+      "step": 7598
+    },
+    {
+      "epoch": 0.4897209512147967,
+      "grad_norm": 0.3195438699631821,
+      "learning_rate": 5.6741854636591485e-06,
+      "loss": 0.0018,
+      "step": 7599
+    },
+    {
+      "epoch": 0.4897853966617258,
+      "grad_norm": 0.002169986000435176,
+      "learning_rate": 5.673469387755103e-06,
+      "loss": 0.0,
+      "step": 7600
+    },
+    {
+      "epoch": 0.489849842108655,
+      "grad_norm": 0.012983810109652043,
+      "learning_rate": 5.672753311851057e-06,
+      "loss": 0.0,
+      "step": 7601
+    },
+    {
+      "epoch": 0.4899142875555842,
+      "grad_norm": 0.014353244432383392,
+      "learning_rate": 5.672037235947011e-06,
+      "loss": 0.0,
+      "step": 7602
+    },
+    {
+      "epoch": 0.4899787330025134,
+      "grad_norm": 0.00011045339360843307,
+      "learning_rate": 5.671321160042965e-06,
+      "loss": 0.0,
+      "step": 7603
+    },
+    {
+      "epoch": 0.49004317844944256,
+      "grad_norm": 0.006119750306557415,
+      "learning_rate": 5.670605084138919e-06,
+      "loss": 0.0,
+      "step": 7604
+    },
+    {
+      "epoch": 0.49010762389637175,
+      "grad_norm": 0.003246663160360342,
+      "learning_rate": 5.669889008234873e-06,
+      "loss": 0.0,
+      "step": 7605
+    },
+    {
+      "epoch": 0.4901720693433009,
+      "grad_norm": 0.062110801071613486,
+      "learning_rate": 5.669172932330828e-06,
+      "loss": 0.0001,
+      "step": 7606
+    },
+    {
+      "epoch": 0.49023651479023006,
+      "grad_norm": 0.007737362321899938,
+      "learning_rate": 5.668456856426782e-06,
+      "loss": 0.0,
+      "step": 7607
+    },
+    {
+      "epoch": 0.49030096023715924,
+      "grad_norm": 0.020342826642074038,
+      "learning_rate": 5.6677407805227355e-06,
+      "loss": 0.0001,
+      "step": 7608
+    },
+    {
+      "epoch": 0.49036540568408843,
+      "grad_norm": 0.002374675857821564,
+      "learning_rate": 5.66702470461869e-06,
+      "loss": 0.0,
+      "step": 7609
+    },
+    {
+      "epoch": 0.4904298511310176,
+      "grad_norm": 0.002491284824095385,
+      "learning_rate": 5.666308628714644e-06,
+      "loss": 0.0,
+      "step": 7610
+    },
+    {
+      "epoch": 0.49049429657794674,
+      "grad_norm": 1.0187017932555893,
+      "learning_rate": 5.665592552810598e-06,
+      "loss": 0.003,
+      "step": 7611
+    },
+    {
+      "epoch": 0.4905587420248759,
+      "grad_norm": 0.05487809644575838,
+      "learning_rate": 5.664876476906553e-06,
+      "loss": 0.0003,
+      "step": 7612
+    },
+    {
+      "epoch": 0.4906231874718051,
+      "grad_norm": 0.0018616346784588265,
+      "learning_rate": 5.664160401002506e-06,
+      "loss": 0.0,
+      "step": 7613
+    },
+    {
+      "epoch": 0.4906876329187343,
+      "grad_norm": 0.0015321618605294287,
+      "learning_rate": 5.66344432509846e-06,
+      "loss": 0.0,
+      "step": 7614
+    },
+    {
+      "epoch": 0.4907520783656635,
+      "grad_norm": 0.025696763498570488,
+      "learning_rate": 5.662728249194415e-06,
+      "loss": 0.0,
+      "step": 7615
+    },
+    {
+      "epoch": 0.49081652381259266,
+      "grad_norm": 0.05154074278488259,
+      "learning_rate": 5.662012173290369e-06,
+      "loss": 0.0003,
+      "step": 7616
+    },
+    {
+      "epoch": 0.4908809692595218,
+      "grad_norm": 0.26117784201754574,
+      "learning_rate": 5.661296097386323e-06,
+      "loss": 0.0008,
+      "step": 7617
+    },
+    {
+      "epoch": 0.490945414706451,
+      "grad_norm": 0.21229055685764633,
+      "learning_rate": 5.660580021482277e-06,
+      "loss": 0.0006,
+      "step": 7618
+    },
+    {
+      "epoch": 0.49100986015338016,
+      "grad_norm": 0.0019154365838764172,
+      "learning_rate": 5.659863945578232e-06,
+      "loss": 0.0,
+      "step": 7619
+    },
+    {
+      "epoch": 0.49107430560030935,
+      "grad_norm": 0.000521643832761043,
+      "learning_rate": 5.659147869674186e-06,
+      "loss": 0.0,
+      "step": 7620
+    },
+    {
+      "epoch": 0.49113875104723853,
+      "grad_norm": 0.05246131011336328,
+      "learning_rate": 5.6584317937701405e-06,
+      "loss": 0.0002,
+      "step": 7621
+    },
+    {
+      "epoch": 0.49120319649416766,
+      "grad_norm": 0.023859120947491402,
+      "learning_rate": 5.657715717866095e-06,
+      "loss": 0.0001,
+      "step": 7622
+    },
+    {
+      "epoch": 0.49126764194109684,
+      "grad_norm": 0.07066070248157924,
+      "learning_rate": 5.656999641962049e-06,
+      "loss": 0.0002,
+      "step": 7623
+    },
+    {
+      "epoch": 0.49133208738802603,
+      "grad_norm": 0.0058702492762727244,
+      "learning_rate": 5.6562835660580025e-06,
+      "loss": 0.0,
+      "step": 7624
+    },
+    {
+      "epoch": 0.4913965328349552,
+      "grad_norm": 0.45628538974263216,
+      "learning_rate": 5.655567490153957e-06,
+      "loss": 0.001,
+      "step": 7625
+    },
+    {
+      "epoch": 0.4914609782818844,
+      "grad_norm": 0.016563145807952723,
+      "learning_rate": 5.654851414249911e-06,
+      "loss": 0.0,
+      "step": 7626
+    },
+    {
+      "epoch": 0.4915254237288136,
+      "grad_norm": 0.0007029439357273081,
+      "learning_rate": 5.6541353383458654e-06,
+      "loss": 0.0,
+      "step": 7627
+    },
+    {
+      "epoch": 0.4915898691757427,
+      "grad_norm": 0.07840164793822463,
+      "learning_rate": 5.65341926244182e-06,
+      "loss": 0.0,
+      "step": 7628
+    },
+    {
+      "epoch": 0.4916543146226719,
+      "grad_norm": 0.3056592361995422,
+      "learning_rate": 5.652703186537773e-06,
+      "loss": 0.0023,
+      "step": 7629
+    },
+    {
+      "epoch": 0.4917187600696011,
+      "grad_norm": 0.6774211016599851,
+      "learning_rate": 5.6519871106337275e-06,
+      "loss": 0.0035,
+      "step": 7630
+    },
+    {
+      "epoch": 0.49178320551653026,
+      "grad_norm": 0.001993808920682025,
+      "learning_rate": 5.651271034729682e-06,
+      "loss": 0.0,
+      "step": 7631
+    },
+    {
+      "epoch": 0.49184765096345945,
+      "grad_norm": 0.1596749738737264,
+      "learning_rate": 5.650554958825636e-06,
+      "loss": 0.0003,
+      "step": 7632
+    },
+    {
+      "epoch": 0.49191209641038863,
+      "grad_norm": 0.018059339301017427,
+      "learning_rate": 5.64983888292159e-06,
+      "loss": 0.0001,
+      "step": 7633
+    },
+    {
+      "epoch": 0.49197654185731776,
+      "grad_norm": 0.0013676482324627582,
+      "learning_rate": 5.649122807017544e-06,
+      "loss": 0.0,
+      "step": 7634
+    },
+    {
+      "epoch": 0.49204098730424695,
+      "grad_norm": 0.001527761549636789,
+      "learning_rate": 5.648406731113498e-06,
+      "loss": 0.0,
+      "step": 7635
+    },
+    {
+      "epoch": 0.49210543275117613,
+      "grad_norm": 0.002563567830487596,
+      "learning_rate": 5.6476906552094524e-06,
+      "loss": 0.0,
+      "step": 7636
+    },
+    {
+      "epoch": 0.4921698781981053,
+      "grad_norm": 0.0020392195153546198,
+      "learning_rate": 5.646974579305407e-06,
+      "loss": 0.0,
+      "step": 7637
+    },
+    {
+      "epoch": 0.4922343236450345,
+      "grad_norm": 0.0025428419743069013,
+      "learning_rate": 5.646258503401361e-06,
+      "loss": 0.0,
+      "step": 7638
+    },
+    {
+      "epoch": 0.49229876909196363,
+      "grad_norm": 0.0010233870824352465,
+      "learning_rate": 5.6455424274973145e-06,
+      "loss": 0.0,
+      "step": 7639
+    },
+    {
+      "epoch": 0.4923632145388928,
+      "grad_norm": 0.16308028959442436,
+      "learning_rate": 5.644826351593269e-06,
+      "loss": 0.0005,
+      "step": 7640
+    },
+    {
+      "epoch": 0.492427659985822,
+      "grad_norm": 0.0002236146238672205,
+      "learning_rate": 5.644110275689223e-06,
+      "loss": 0.0,
+      "step": 7641
+    },
+    {
+      "epoch": 0.4924921054327512,
+      "grad_norm": 0.003699660007362483,
+      "learning_rate": 5.643394199785178e-06,
+      "loss": 0.0,
+      "step": 7642
+    },
+    {
+      "epoch": 0.49255655087968037,
+      "grad_norm": 0.6511107573406089,
+      "learning_rate": 5.6426781238811325e-06,
+      "loss": 0.0012,
+      "step": 7643
+    },
+    {
+      "epoch": 0.49262099632660955,
+      "grad_norm": 0.0006555271681799672,
+      "learning_rate": 5.641962047977087e-06,
+      "loss": 0.0,
+      "step": 7644
+    },
+    {
+      "epoch": 0.4926854417735387,
+      "grad_norm": 0.011247004639613586,
+      "learning_rate": 5.64124597207304e-06,
+      "loss": 0.0,
+      "step": 7645
+    },
+    {
+      "epoch": 0.49274988722046786,
+      "grad_norm": 0.04122928475379016,
+      "learning_rate": 5.6405298961689946e-06,
+      "loss": 0.0,
+      "step": 7646
+    },
+    {
+      "epoch": 0.49281433266739705,
+      "grad_norm": 0.09820465599572177,
+      "learning_rate": 5.639813820264949e-06,
+      "loss": 0.0003,
+      "step": 7647
+    },
+    {
+      "epoch": 0.49287877811432623,
+      "grad_norm": 0.0005418565121424015,
+      "learning_rate": 5.639097744360903e-06,
+      "loss": 0.0,
+      "step": 7648
+    },
+    {
+      "epoch": 0.4929432235612554,
+      "grad_norm": 0.25866958451243477,
+      "learning_rate": 5.6383816684568575e-06,
+      "loss": 0.0004,
+      "step": 7649
+    },
+    {
+      "epoch": 0.49300766900818455,
+      "grad_norm": 0.014861191542048305,
+      "learning_rate": 5.637665592552811e-06,
+      "loss": 0.0,
+      "step": 7650
+    },
+    {
+      "epoch": 0.49307211445511373,
+      "grad_norm": 0.06393503994418094,
+      "learning_rate": 5.636949516648765e-06,
+      "loss": 0.0002,
+      "step": 7651
+    },
+    {
+      "epoch": 0.4931365599020429,
+      "grad_norm": 0.001295256191406105,
+      "learning_rate": 5.6362334407447195e-06,
+      "loss": 0.0,
+      "step": 7652
+    },
+    {
+      "epoch": 0.4932010053489721,
+      "grad_norm": 0.0013651152885221776,
+      "learning_rate": 5.635517364840674e-06,
+      "loss": 0.0,
+      "step": 7653
+    },
+    {
+      "epoch": 0.4932654507959013,
+      "grad_norm": 0.034053694591434605,
+      "learning_rate": 5.634801288936628e-06,
+      "loss": 0.0,
+      "step": 7654
+    },
+    {
+      "epoch": 0.49332989624283047,
+      "grad_norm": 0.025483034313675483,
+      "learning_rate": 5.6340852130325816e-06,
+      "loss": 0.0001,
+      "step": 7655
+    },
+    {
+      "epoch": 0.4933943416897596,
+      "grad_norm": 0.0867213163984139,
+      "learning_rate": 5.633369137128536e-06,
+      "loss": 0.0002,
+      "step": 7656
+    },
+    {
+      "epoch": 0.4934587871366888,
+      "grad_norm": 0.036528468614227996,
+      "learning_rate": 5.63265306122449e-06,
+      "loss": 0.0005,
+      "step": 7657
+    },
+    {
+      "epoch": 0.49352323258361797,
+      "grad_norm": 0.014582704541776105,
+      "learning_rate": 5.6319369853204445e-06,
+      "loss": 0.0001,
+      "step": 7658
+    },
+    {
+      "epoch": 0.49358767803054715,
+      "grad_norm": 0.006312064480264101,
+      "learning_rate": 5.631220909416398e-06,
+      "loss": 0.0,
+      "step": 7659
+    },
+    {
+      "epoch": 0.49365212347747633,
+      "grad_norm": 0.05966863374292694,
+      "learning_rate": 5.630504833512352e-06,
+      "loss": 0.0001,
+      "step": 7660
+    },
+    {
+      "epoch": 0.49371656892440546,
+      "grad_norm": 0.01641082586326572,
+      "learning_rate": 5.6297887576083065e-06,
+      "loss": 0.0001,
+      "step": 7661
+    },
+    {
+      "epoch": 0.49378101437133465,
+      "grad_norm": 0.003940255907692563,
+      "learning_rate": 5.629072681704261e-06,
+      "loss": 0.0,
+      "step": 7662
+    },
+    {
+      "epoch": 0.49384545981826383,
+      "grad_norm": 0.006613115214583254,
+      "learning_rate": 5.628356605800215e-06,
+      "loss": 0.0001,
+      "step": 7663
+    },
+    {
+      "epoch": 0.493909905265193,
+      "grad_norm": 0.05100384626993441,
+      "learning_rate": 5.6276405298961686e-06,
+      "loss": 0.0001,
+      "step": 7664
+    },
+    {
+      "epoch": 0.4939743507121222,
+      "grad_norm": 0.003475097216405046,
+      "learning_rate": 5.6269244539921246e-06,
+      "loss": 0.0,
+      "step": 7665
+    },
+    {
+      "epoch": 0.4940387961590514,
+      "grad_norm": 0.002349933148403365,
+      "learning_rate": 5.626208378088078e-06,
+      "loss": 0.0,
+      "step": 7666
+    },
+    {
+      "epoch": 0.4941032416059805,
+      "grad_norm": 0.1691805409207571,
+      "learning_rate": 5.625492302184032e-06,
+      "loss": 0.0004,
+      "step": 7667
+    },
+    {
+      "epoch": 0.4941676870529097,
+      "grad_norm": 0.04184740078663928,
+      "learning_rate": 5.624776226279987e-06,
+      "loss": 0.0001,
+      "step": 7668
+    },
+    {
+      "epoch": 0.4942321324998389,
+      "grad_norm": 0.015622903452228139,
+      "learning_rate": 5.624060150375941e-06,
+      "loss": 0.0,
+      "step": 7669
+    },
+    {
+      "epoch": 0.49429657794676807,
+      "grad_norm": 0.006198852245547676,
+      "learning_rate": 5.623344074471895e-06,
+      "loss": 0.0,
+      "step": 7670
+    },
+    {
+      "epoch": 0.49436102339369725,
+      "grad_norm": 0.03239944533858724,
+      "learning_rate": 5.622627998567849e-06,
+      "loss": 0.0001,
+      "step": 7671
+    },
+    {
+      "epoch": 0.49442546884062644,
+      "grad_norm": 0.23790876792287802,
+      "learning_rate": 5.621911922663803e-06,
+      "loss": 0.004,
+      "step": 7672
+    },
+    {
+      "epoch": 0.49448991428755557,
+      "grad_norm": 0.002371005246239983,
+      "learning_rate": 5.621195846759757e-06,
+      "loss": 0.0,
+      "step": 7673
+    },
+    {
+      "epoch": 0.49455435973448475,
+      "grad_norm": 0.00392991030615517,
+      "learning_rate": 5.6204797708557115e-06,
+      "loss": 0.0,
+      "step": 7674
+    },
+    {
+      "epoch": 0.49461880518141393,
+      "grad_norm": 0.4288448915211329,
+      "learning_rate": 5.619763694951665e-06,
+      "loss": 0.0046,
+      "step": 7675
+    },
+    {
+      "epoch": 0.4946832506283431,
+      "grad_norm": 0.05516306764421135,
+      "learning_rate": 5.619047619047619e-06,
+      "loss": 0.0003,
+      "step": 7676
+    },
+    {
+      "epoch": 0.4947476960752723,
+      "grad_norm": 0.010246722793248482,
+      "learning_rate": 5.618331543143574e-06,
+      "loss": 0.0,
+      "step": 7677
+    },
+    {
+      "epoch": 0.49481214152220143,
+      "grad_norm": 0.007946841214768584,
+      "learning_rate": 5.617615467239528e-06,
+      "loss": 0.0,
+      "step": 7678
+    },
+    {
+      "epoch": 0.4948765869691306,
+      "grad_norm": 0.0548576853204545,
+      "learning_rate": 5.616899391335482e-06,
+      "loss": 0.0,
+      "step": 7679
+    },
+    {
+      "epoch": 0.4949410324160598,
+      "grad_norm": 0.07455733516548389,
+      "learning_rate": 5.616183315431436e-06,
+      "loss": 0.0001,
+      "step": 7680
+    },
+    {
+      "epoch": 0.495005477862989,
+      "grad_norm": 0.001946500852743602,
+      "learning_rate": 5.61546723952739e-06,
+      "loss": 0.0,
+      "step": 7681
+    },
+    {
+      "epoch": 0.49506992330991817,
+      "grad_norm": 0.11966813281237758,
+      "learning_rate": 5.614751163623344e-06,
+      "loss": 0.0003,
+      "step": 7682
+    },
+    {
+      "epoch": 0.49513436875684735,
+      "grad_norm": 0.22977987073403688,
+      "learning_rate": 5.6140350877192985e-06,
+      "loss": 0.0012,
+      "step": 7683
+    },
+    {
+      "epoch": 0.4951988142037765,
+      "grad_norm": 0.004786080663721793,
+      "learning_rate": 5.613319011815253e-06,
+      "loss": 0.0,
+      "step": 7684
+    },
+    {
+      "epoch": 0.49526325965070567,
+      "grad_norm": 0.3058660284807045,
+      "learning_rate": 5.612602935911206e-06,
+      "loss": 0.0025,
+      "step": 7685
+    },
+    {
+      "epoch": 0.49532770509763485,
+      "grad_norm": 0.0006387131833626681,
+      "learning_rate": 5.611886860007161e-06,
+      "loss": 0.0,
+      "step": 7686
+    },
+    {
+      "epoch": 0.49539215054456404,
+      "grad_norm": 0.007263654905943296,
+      "learning_rate": 5.611170784103115e-06,
+      "loss": 0.0,
+      "step": 7687
+    },
+    {
+      "epoch": 0.4954565959914932,
+      "grad_norm": 0.00047506028153626915,
+      "learning_rate": 5.610454708199069e-06,
+      "loss": 0.0,
+      "step": 7688
+    },
+    {
+      "epoch": 0.49552104143842235,
+      "grad_norm": 0.0012289411413441314,
+      "learning_rate": 5.609738632295024e-06,
+      "loss": 0.0,
+      "step": 7689
+    },
+    {
+      "epoch": 0.49558548688535153,
+      "grad_norm": 0.040993396378150186,
+      "learning_rate": 5.609022556390979e-06,
+      "loss": 0.0001,
+      "step": 7690
+    },
+    {
+      "epoch": 0.4956499323322807,
+      "grad_norm": 0.0013355882791165105,
+      "learning_rate": 5.608306480486932e-06,
+      "loss": 0.0,
+      "step": 7691
+    },
+    {
+      "epoch": 0.4957143777792099,
+      "grad_norm": 0.010719311534290885,
+      "learning_rate": 5.607590404582886e-06,
+      "loss": 0.0,
+      "step": 7692
+    },
+    {
+      "epoch": 0.4957788232261391,
+      "grad_norm": 0.0038407025845055135,
+      "learning_rate": 5.606874328678841e-06,
+      "loss": 0.0,
+      "step": 7693
+    },
+    {
+      "epoch": 0.49584326867306827,
+      "grad_norm": 0.4340892057394482,
+      "learning_rate": 5.606158252774795e-06,
+      "loss": 0.0044,
+      "step": 7694
+    },
+    {
+      "epoch": 0.4959077141199974,
+      "grad_norm": 0.005255564610214897,
+      "learning_rate": 5.605442176870749e-06,
+      "loss": 0.0,
+      "step": 7695
+    },
+    {
+      "epoch": 0.4959721595669266,
+      "grad_norm": 0.018044455183344257,
+      "learning_rate": 5.604726100966703e-06,
+      "loss": 0.0002,
+      "step": 7696
+    },
+    {
+      "epoch": 0.49603660501385577,
+      "grad_norm": 0.0006821521166229848,
+      "learning_rate": 5.604010025062657e-06,
+      "loss": 0.0,
+      "step": 7697
+    },
+    {
+      "epoch": 0.49610105046078495,
+      "grad_norm": 0.12202850007171458,
+      "learning_rate": 5.603293949158611e-06,
+      "loss": 0.0002,
+      "step": 7698
+    },
+    {
+      "epoch": 0.49616549590771414,
+      "grad_norm": 0.036351122742084956,
+      "learning_rate": 5.602577873254566e-06,
+      "loss": 0.0,
+      "step": 7699
+    },
+    {
+      "epoch": 0.49622994135464327,
+      "grad_norm": 0.00016657967509917158,
+      "learning_rate": 5.60186179735052e-06,
+      "loss": 0.0,
+      "step": 7700
+    },
+    {
+      "epoch": 0.49629438680157245,
+      "grad_norm": 0.1313316661645494,
+      "learning_rate": 5.601145721446473e-06,
+      "loss": 0.0003,
+      "step": 7701
+    },
+    {
+      "epoch": 0.49635883224850164,
+      "grad_norm": 0.5427525590777412,
+      "learning_rate": 5.600429645542428e-06,
+      "loss": 0.001,
+      "step": 7702
+    },
+    {
+      "epoch": 0.4964232776954308,
+      "grad_norm": 0.004619553697276954,
+      "learning_rate": 5.599713569638382e-06,
+      "loss": 0.0,
+      "step": 7703
+    },
+    {
+      "epoch": 0.49648772314236,
+      "grad_norm": 0.12349671304702227,
+      "learning_rate": 5.598997493734336e-06,
+      "loss": 0.0016,
+      "step": 7704
+    },
+    {
+      "epoch": 0.4965521685892892,
+      "grad_norm": 0.001022775411367935,
+      "learning_rate": 5.5982814178302906e-06,
+      "loss": 0.0,
+      "step": 7705
+    },
+    {
+      "epoch": 0.4966166140362183,
+      "grad_norm": 0.11159717941088838,
+      "learning_rate": 5.597565341926244e-06,
+      "loss": 0.0002,
+      "step": 7706
+    },
+    {
+      "epoch": 0.4966810594831475,
+      "grad_norm": 0.006720570718083302,
+      "learning_rate": 5.596849266022198e-06,
+      "loss": 0.0,
+      "step": 7707
+    },
+    {
+      "epoch": 0.4967455049300767,
+      "grad_norm": 0.0001619117415859008,
+      "learning_rate": 5.596133190118153e-06,
+      "loss": 0.0,
+      "step": 7708
+    },
+    {
+      "epoch": 0.49680995037700587,
+      "grad_norm": 1.5955664810823738,
+      "learning_rate": 5.595417114214107e-06,
+      "loss": 0.0081,
+      "step": 7709
+    },
+    {
+      "epoch": 0.49687439582393506,
+      "grad_norm": 0.008624999726623077,
+      "learning_rate": 5.594701038310061e-06,
+      "loss": 0.0,
+      "step": 7710
+    },
+    {
+      "epoch": 0.49693884127086424,
+      "grad_norm": 0.011486251104675875,
+      "learning_rate": 5.593984962406015e-06,
+      "loss": 0.0,
+      "step": 7711
+    },
+    {
+      "epoch": 0.49700328671779337,
+      "grad_norm": 0.2950930319796222,
+      "learning_rate": 5.59326888650197e-06,
+      "loss": 0.0009,
+      "step": 7712
+    },
+    {
+      "epoch": 0.49706773216472255,
+      "grad_norm": 0.0014471563901175158,
+      "learning_rate": 5.592552810597924e-06,
+      "loss": 0.0,
+      "step": 7713
+    },
+    {
+      "epoch": 0.49713217761165174,
+      "grad_norm": 0.0038007229908679743,
+      "learning_rate": 5.591836734693878e-06,
+      "loss": 0.0,
+      "step": 7714
+    },
+    {
+      "epoch": 0.4971966230585809,
+      "grad_norm": 0.0018911992897892018,
+      "learning_rate": 5.591120658789833e-06,
+      "loss": 0.0,
+      "step": 7715
+    },
+    {
+      "epoch": 0.4972610685055101,
+      "grad_norm": 0.017360700248121624,
+      "learning_rate": 5.590404582885787e-06,
+      "loss": 0.0,
+      "step": 7716
+    },
+    {
+      "epoch": 0.49732551395243924,
+      "grad_norm": 0.000842689886037626,
+      "learning_rate": 5.5896885069817405e-06,
+      "loss": 0.0,
+      "step": 7717
+    },
+    {
+      "epoch": 0.4973899593993684,
+      "grad_norm": 0.10780282530014913,
+      "learning_rate": 5.588972431077695e-06,
+      "loss": 0.0002,
+      "step": 7718
+    },
+    {
+      "epoch": 0.4974544048462976,
+      "grad_norm": 0.009681021889705044,
+      "learning_rate": 5.588256355173649e-06,
+      "loss": 0.0001,
+      "step": 7719
+    },
+    {
+      "epoch": 0.4975188502932268,
+      "grad_norm": 0.0017426617240457378,
+      "learning_rate": 5.587540279269603e-06,
+      "loss": 0.0,
+      "step": 7720
+    },
+    {
+      "epoch": 0.497583295740156,
+      "grad_norm": 1.5181758690231042,
+      "learning_rate": 5.586824203365558e-06,
+      "loss": 0.0113,
+      "step": 7721
+    },
+    {
+      "epoch": 0.49764774118708516,
+      "grad_norm": 0.006614844277389507,
+      "learning_rate": 5.586108127461511e-06,
+      "loss": 0.0001,
+      "step": 7722
+    },
+    {
+      "epoch": 0.4977121866340143,
+      "grad_norm": 0.2634574442064215,
+      "learning_rate": 5.585392051557465e-06,
+      "loss": 0.0012,
+      "step": 7723
+    },
+    {
+      "epoch": 0.49777663208094347,
+      "grad_norm": 0.26991512770438303,
+      "learning_rate": 5.58467597565342e-06,
+      "loss": 0.0023,
+      "step": 7724
+    },
+    {
+      "epoch": 0.49784107752787266,
+      "grad_norm": 0.009235564462089122,
+      "learning_rate": 5.583959899749374e-06,
+      "loss": 0.0,
+      "step": 7725
+    },
+    {
+      "epoch": 0.49790552297480184,
+      "grad_norm": 0.2566201641688113,
+      "learning_rate": 5.583243823845328e-06,
+      "loss": 0.0005,
+      "step": 7726
+    },
+    {
+      "epoch": 0.497969968421731,
+      "grad_norm": 0.005090404771266792,
+      "learning_rate": 5.582527747941282e-06,
+      "loss": 0.0,
+      "step": 7727
+    },
+    {
+      "epoch": 0.49803441386866015,
+      "grad_norm": 0.27343225474095073,
+      "learning_rate": 5.581811672037236e-06,
+      "loss": 0.002,
+      "step": 7728
+    },
+    {
+      "epoch": 0.49809885931558934,
+      "grad_norm": 0.031385309490975594,
+      "learning_rate": 5.58109559613319e-06,
+      "loss": 0.0,
+      "step": 7729
+    },
+    {
+      "epoch": 0.4981633047625185,
+      "grad_norm": 0.023604299895657106,
+      "learning_rate": 5.580379520229145e-06,
+      "loss": 0.0,
+      "step": 7730
+    },
+    {
+      "epoch": 0.4982277502094477,
+      "grad_norm": 0.010524595936297206,
+      "learning_rate": 5.579663444325099e-06,
+      "loss": 0.0,
+      "step": 7731
+    },
+    {
+      "epoch": 0.4982921956563769,
+      "grad_norm": 0.08313716178764882,
+      "learning_rate": 5.578947368421052e-06,
+      "loss": 0.0001,
+      "step": 7732
+    },
+    {
+      "epoch": 0.4983566411033061,
+      "grad_norm": 0.20455374205723056,
+      "learning_rate": 5.578231292517007e-06,
+      "loss": 0.0002,
+      "step": 7733
+    },
+    {
+      "epoch": 0.4984210865502352,
+      "grad_norm": 0.006954444476851246,
+      "learning_rate": 5.577515216612961e-06,
+      "loss": 0.0,
+      "step": 7734
+    },
+    {
+      "epoch": 0.4984855319971644,
+      "grad_norm": 0.1379288863464268,
+      "learning_rate": 5.576799140708916e-06,
+      "loss": 0.0012,
+      "step": 7735
+    },
+    {
+      "epoch": 0.4985499774440936,
+      "grad_norm": 0.22782729888330902,
+      "learning_rate": 5.5760830648048704e-06,
+      "loss": 0.0003,
+      "step": 7736
+    },
+    {
+      "epoch": 0.49861442289102276,
+      "grad_norm": 0.004750422647156393,
+      "learning_rate": 5.575366988900825e-06,
+      "loss": 0.0,
+      "step": 7737
+    },
+    {
+      "epoch": 0.49867886833795194,
+      "grad_norm": 0.02539140787201507,
+      "learning_rate": 5.574650912996778e-06,
+      "loss": 0.0,
+      "step": 7738
+    },
+    {
+      "epoch": 0.49874331378488107,
+      "grad_norm": 0.0074664136120375325,
+      "learning_rate": 5.5739348370927325e-06,
+      "loss": 0.0,
+      "step": 7739
+    },
+    {
+      "epoch": 0.49880775923181025,
+      "grad_norm": 0.040856026752530014,
+      "learning_rate": 5.573218761188687e-06,
+      "loss": 0.0005,
+      "step": 7740
+    },
+    {
+      "epoch": 0.49887220467873944,
+      "grad_norm": 0.023396444870685918,
+      "learning_rate": 5.572502685284641e-06,
+      "loss": 0.0016,
+      "step": 7741
+    },
+    {
+      "epoch": 0.4989366501256686,
+      "grad_norm": 0.003881107778788916,
+      "learning_rate": 5.571786609380595e-06,
+      "loss": 0.0,
+      "step": 7742
+    },
+    {
+      "epoch": 0.4990010955725978,
+      "grad_norm": 0.004024760296975562,
+      "learning_rate": 5.571070533476549e-06,
+      "loss": 0.0,
+      "step": 7743
+    },
+    {
+      "epoch": 0.499065541019527,
+      "grad_norm": 0.005052507845867952,
+      "learning_rate": 5.570354457572503e-06,
+      "loss": 0.0001,
+      "step": 7744
+    },
+    {
+      "epoch": 0.4991299864664561,
+      "grad_norm": 0.006618753422869714,
+      "learning_rate": 5.5696383816684574e-06,
+      "loss": 0.0,
+      "step": 7745
+    },
+    {
+      "epoch": 0.4991944319133853,
+      "grad_norm": 0.013298532013776608,
+      "learning_rate": 5.568922305764412e-06,
+      "loss": 0.0,
+      "step": 7746
+    },
+    {
+      "epoch": 0.4992588773603145,
+      "grad_norm": 0.008402784516505377,
+      "learning_rate": 5.568206229860366e-06,
+      "loss": 0.0,
+      "step": 7747
+    },
+    {
+      "epoch": 0.4993233228072437,
+      "grad_norm": 0.028681718139751507,
+      "learning_rate": 5.5674901539563195e-06,
+      "loss": 0.0,
+      "step": 7748
+    },
+    {
+      "epoch": 0.49938776825417286,
+      "grad_norm": 0.0004790018685773785,
+      "learning_rate": 5.566774078052274e-06,
+      "loss": 0.0,
+      "step": 7749
+    },
+    {
+      "epoch": 0.49945221370110204,
+      "grad_norm": 0.00017217928028040734,
+      "learning_rate": 5.566058002148228e-06,
+      "loss": 0.0,
+      "step": 7750
+    },
+    {
+      "epoch": 0.4995166591480312,
+      "grad_norm": 0.004187983856702524,
+      "learning_rate": 5.565341926244182e-06,
+      "loss": 0.0,
+      "step": 7751
+    },
+    {
+      "epoch": 0.49958110459496036,
+      "grad_norm": 0.00661046670208516,
+      "learning_rate": 5.564625850340136e-06,
+      "loss": 0.0001,
+      "step": 7752
+    },
+    {
+      "epoch": 0.49964555004188954,
+      "grad_norm": 0.0028452297455472866,
+      "learning_rate": 5.56390977443609e-06,
+      "loss": 0.0,
+      "step": 7753
+    },
+    {
+      "epoch": 0.4997099954888187,
+      "grad_norm": 0.08567421038959883,
+      "learning_rate": 5.5631936985320444e-06,
+      "loss": 0.0003,
+      "step": 7754
+    },
+    {
+      "epoch": 0.4997744409357479,
+      "grad_norm": 0.000638038899302093,
+      "learning_rate": 5.562477622627999e-06,
+      "loss": 0.0,
+      "step": 7755
+    },
+    {
+      "epoch": 0.49983888638267704,
+      "grad_norm": 0.00032599765353913423,
+      "learning_rate": 5.561761546723953e-06,
+      "loss": 0.0,
+      "step": 7756
+    },
+    {
+      "epoch": 0.4999033318296062,
+      "grad_norm": 0.005907647457967332,
+      "learning_rate": 5.5610454708199065e-06,
+      "loss": 0.0,
+      "step": 7757
+    },
+    {
+      "epoch": 0.4999677772765354,
+      "grad_norm": 0.005853467747557248,
+      "learning_rate": 5.560329394915861e-06,
+      "loss": 0.0,
+      "step": 7758
+    },
+    {
+      "epoch": 0.5000322227234646,
+      "grad_norm": 0.001166157403481343,
+      "learning_rate": 5.559613319011816e-06,
+      "loss": 0.0,
+      "step": 7759
+    },
+    {
+      "epoch": 0.5000966681703938,
+      "grad_norm": 0.006691685794746546,
+      "learning_rate": 5.55889724310777e-06,
+      "loss": 0.0,
+      "step": 7760
+    },
+    {
+      "epoch": 0.500161113617323,
+      "grad_norm": 0.00032993997002244194,
+      "learning_rate": 5.5581811672037245e-06,
+      "loss": 0.0,
+      "step": 7761
+    },
+    {
+      "epoch": 0.5002255590642521,
+      "grad_norm": 0.010774148663307965,
+      "learning_rate": 5.557465091299679e-06,
+      "loss": 0.0001,
+      "step": 7762
+    },
+    {
+      "epoch": 0.5002900045111813,
+      "grad_norm": 0.04316844108864388,
+      "learning_rate": 5.556749015395633e-06,
+      "loss": 0.0001,
+      "step": 7763
+    },
+    {
+      "epoch": 0.5003544499581105,
+      "grad_norm": 0.1035563865516481,
+      "learning_rate": 5.5560329394915866e-06,
+      "loss": 0.0002,
+      "step": 7764
+    },
+    {
+      "epoch": 0.5004188954050396,
+      "grad_norm": 0.0013261507056000824,
+      "learning_rate": 5.555316863587541e-06,
+      "loss": 0.0,
+      "step": 7765
+    },
+    {
+      "epoch": 0.5004833408519688,
+      "grad_norm": 0.012714541640589294,
+      "learning_rate": 5.554600787683495e-06,
+      "loss": 0.0,
+      "step": 7766
+    },
+    {
+      "epoch": 0.500547786298898,
+      "grad_norm": 0.0031739563583639747,
+      "learning_rate": 5.5538847117794495e-06,
+      "loss": 0.0,
+      "step": 7767
+    },
+    {
+      "epoch": 0.5006122317458271,
+      "grad_norm": 0.11012735207591887,
+      "learning_rate": 5.553168635875403e-06,
+      "loss": 0.0001,
+      "step": 7768
+    },
+    {
+      "epoch": 0.5006766771927563,
+      "grad_norm": 0.015187950687640926,
+      "learning_rate": 5.552452559971357e-06,
+      "loss": 0.0001,
+      "step": 7769
+    },
+    {
+      "epoch": 0.5007411226396855,
+      "grad_norm": 0.013135091208567084,
+      "learning_rate": 5.5517364840673115e-06,
+      "loss": 0.0,
+      "step": 7770
+    },
+    {
+      "epoch": 0.5008055680866147,
+      "grad_norm": 0.2000327962569176,
+      "learning_rate": 5.551020408163266e-06,
+      "loss": 0.0003,
+      "step": 7771
+    },
+    {
+      "epoch": 0.5008700135335439,
+      "grad_norm": 0.00024363627308754418,
+      "learning_rate": 5.55030433225922e-06,
+      "loss": 0.0,
+      "step": 7772
+    },
+    {
+      "epoch": 0.5009344589804731,
+      "grad_norm": 4.0486426170484195e-05,
+      "learning_rate": 5.5495882563551736e-06,
+      "loss": 0.0,
+      "step": 7773
+    },
+    {
+      "epoch": 0.5009989044274022,
+      "grad_norm": 0.0044095567704248275,
+      "learning_rate": 5.548872180451128e-06,
+      "loss": 0.0,
+      "step": 7774
+    },
+    {
+      "epoch": 0.5010633498743314,
+      "grad_norm": 0.06624869983000084,
+      "learning_rate": 5.548156104547082e-06,
+      "loss": 0.0001,
+      "step": 7775
+    },
+    {
+      "epoch": 0.5011277953212605,
+      "grad_norm": 0.26731744305720906,
+      "learning_rate": 5.5474400286430365e-06,
+      "loss": 0.0012,
+      "step": 7776
+    },
+    {
+      "epoch": 0.5011922407681897,
+      "grad_norm": 0.0031713670464031084,
+      "learning_rate": 5.546723952738991e-06,
+      "loss": 0.0,
+      "step": 7777
+    },
+    {
+      "epoch": 0.5012566862151189,
+      "grad_norm": 0.10094297798692436,
+      "learning_rate": 5.546007876834944e-06,
+      "loss": 0.0002,
+      "step": 7778
+    },
+    {
+      "epoch": 0.5013211316620481,
+      "grad_norm": 0.0006766827932490896,
+      "learning_rate": 5.5452918009308985e-06,
+      "loss": 0.0,
+      "step": 7779
+    },
+    {
+      "epoch": 0.5013855771089772,
+      "grad_norm": 0.006504950923653288,
+      "learning_rate": 5.544575725026853e-06,
+      "loss": 0.0,
+      "step": 7780
+    },
+    {
+      "epoch": 0.5014500225559064,
+      "grad_norm": 0.002359928133011971,
+      "learning_rate": 5.543859649122807e-06,
+      "loss": 0.0,
+      "step": 7781
+    },
+    {
+      "epoch": 0.5015144680028356,
+      "grad_norm": 0.001259427859195719,
+      "learning_rate": 5.543143573218762e-06,
+      "loss": 0.0,
+      "step": 7782
+    },
+    {
+      "epoch": 0.5015789134497648,
+      "grad_norm": 4.103638976019595e-05,
+      "learning_rate": 5.5424274973147166e-06,
+      "loss": 0.0,
+      "step": 7783
+    },
+    {
+      "epoch": 0.501643358896694,
+      "grad_norm": 0.008854051233046384,
+      "learning_rate": 5.54171142141067e-06,
+      "loss": 0.0,
+      "step": 7784
+    },
+    {
+      "epoch": 0.5017078043436232,
+      "grad_norm": 0.0033936376934568737,
+      "learning_rate": 5.540995345506624e-06,
+      "loss": 0.0,
+      "step": 7785
+    },
+    {
+      "epoch": 0.5017722497905523,
+      "grad_norm": 0.023507155038777785,
+      "learning_rate": 5.540279269602579e-06,
+      "loss": 0.0,
+      "step": 7786
+    },
+    {
+      "epoch": 0.5018366952374814,
+      "grad_norm": 0.005942011229836717,
+      "learning_rate": 5.539563193698533e-06,
+      "loss": 0.0,
+      "step": 7787
+    },
+    {
+      "epoch": 0.5019011406844106,
+      "grad_norm": 0.02232280920114333,
+      "learning_rate": 5.538847117794487e-06,
+      "loss": 0.0001,
+      "step": 7788
+    },
+    {
+      "epoch": 0.5019655861313398,
+      "grad_norm": 0.0016590937530407013,
+      "learning_rate": 5.538131041890441e-06,
+      "loss": 0.0,
+      "step": 7789
+    },
+    {
+      "epoch": 0.502030031578269,
+      "grad_norm": 0.011643822116127455,
+      "learning_rate": 5.537414965986395e-06,
+      "loss": 0.0001,
+      "step": 7790
+    },
+    {
+      "epoch": 0.5020944770251982,
+      "grad_norm": 0.0005359369657540056,
+      "learning_rate": 5.536698890082349e-06,
+      "loss": 0.0,
+      "step": 7791
+    },
+    {
+      "epoch": 0.5021589224721273,
+      "grad_norm": 0.06418886703555608,
+      "learning_rate": 5.5359828141783035e-06,
+      "loss": 0.0,
+      "step": 7792
+    },
+    {
+      "epoch": 0.5022233679190565,
+      "grad_norm": 0.0025205363208524005,
+      "learning_rate": 5.535266738274258e-06,
+      "loss": 0.0,
+      "step": 7793
+    },
+    {
+      "epoch": 0.5022878133659857,
+      "grad_norm": 0.47469825069465993,
+      "learning_rate": 5.534550662370211e-06,
+      "loss": 0.0027,
+      "step": 7794
+    },
+    {
+      "epoch": 0.5023522588129149,
+      "grad_norm": 0.0067098508005370774,
+      "learning_rate": 5.533834586466166e-06,
+      "loss": 0.0,
+      "step": 7795
+    },
+    {
+      "epoch": 0.5024167042598441,
+      "grad_norm": 7.736586171454487e-05,
+      "learning_rate": 5.53311851056212e-06,
+      "loss": 0.0,
+      "step": 7796
+    },
+    {
+      "epoch": 0.5024811497067733,
+      "grad_norm": 0.0212810879829306,
+      "learning_rate": 5.532402434658074e-06,
+      "loss": 0.0,
+      "step": 7797
+    },
+    {
+      "epoch": 0.5025455951537023,
+      "grad_norm": 0.027915426011045134,
+      "learning_rate": 5.5316863587540285e-06,
+      "loss": 0.0001,
+      "step": 7798
+    },
+    {
+      "epoch": 0.5026100406006315,
+      "grad_norm": 0.0005632651672147035,
+      "learning_rate": 5.530970282849982e-06,
+      "loss": 0.0,
+      "step": 7799
+    },
+    {
+      "epoch": 0.5026744860475607,
+      "grad_norm": 0.20493163371728026,
+      "learning_rate": 5.530254206945936e-06,
+      "loss": 0.0008,
+      "step": 7800
+    },
+    {
+      "epoch": 0.5027389314944899,
+      "grad_norm": 0.015775780970922863,
+      "learning_rate": 5.5295381310418905e-06,
+      "loss": 0.0,
+      "step": 7801
+    },
+    {
+      "epoch": 0.5028033769414191,
+      "grad_norm": 0.06814150491643632,
+      "learning_rate": 5.528822055137845e-06,
+      "loss": 0.0001,
+      "step": 7802
+    },
+    {
+      "epoch": 0.5028678223883483,
+      "grad_norm": 0.18345450132435454,
+      "learning_rate": 5.528105979233799e-06,
+      "loss": 0.0005,
+      "step": 7803
+    },
+    {
+      "epoch": 0.5029322678352774,
+      "grad_norm": 0.008875562764333902,
+      "learning_rate": 5.527389903329753e-06,
+      "loss": 0.0001,
+      "step": 7804
+    },
+    {
+      "epoch": 0.5029967132822066,
+      "grad_norm": 0.00011833865029197304,
+      "learning_rate": 5.526673827425708e-06,
+      "loss": 0.0,
+      "step": 7805
+    },
+    {
+      "epoch": 0.5030611587291358,
+      "grad_norm": 0.001142839209205189,
+      "learning_rate": 5.525957751521662e-06,
+      "loss": 0.0,
+      "step": 7806
+    },
+    {
+      "epoch": 0.503125604176065,
+      "grad_norm": 0.0007741564340708393,
+      "learning_rate": 5.525241675617616e-06,
+      "loss": 0.0,
+      "step": 7807
+    },
+    {
+      "epoch": 0.5031900496229942,
+      "grad_norm": 0.06842654198181584,
+      "learning_rate": 5.524525599713571e-06,
+      "loss": 0.0002,
+      "step": 7808
+    },
+    {
+      "epoch": 0.5032544950699233,
+      "grad_norm": 0.0018370367967294632,
+      "learning_rate": 5.523809523809525e-06,
+      "loss": 0.0,
+      "step": 7809
+    },
+    {
+      "epoch": 0.5033189405168524,
+      "grad_norm": 0.0042387088958715,
+      "learning_rate": 5.523093447905478e-06,
+      "loss": 0.0,
+      "step": 7810
+    },
+    {
+      "epoch": 0.5033833859637816,
+      "grad_norm": 0.0014034877772028946,
+      "learning_rate": 5.522377372001433e-06,
+      "loss": 0.0,
+      "step": 7811
+    },
+    {
+      "epoch": 0.5034478314107108,
+      "grad_norm": 0.047558174581854164,
+      "learning_rate": 5.521661296097387e-06,
+      "loss": 0.0003,
+      "step": 7812
+    },
+    {
+      "epoch": 0.50351227685764,
+      "grad_norm": 0.11587947569488101,
+      "learning_rate": 5.520945220193341e-06,
+      "loss": 0.0001,
+      "step": 7813
+    },
+    {
+      "epoch": 0.5035767223045692,
+      "grad_norm": 0.0008404481830644785,
+      "learning_rate": 5.520229144289296e-06,
+      "loss": 0.0,
+      "step": 7814
+    },
+    {
+      "epoch": 0.5036411677514984,
+      "grad_norm": 0.027745073397463797,
+      "learning_rate": 5.519513068385249e-06,
+      "loss": 0.0002,
+      "step": 7815
+    },
+    {
+      "epoch": 0.5037056131984275,
+      "grad_norm": 0.0015664983895771825,
+      "learning_rate": 5.518796992481203e-06,
+      "loss": 0.0,
+      "step": 7816
+    },
+    {
+      "epoch": 0.5037700586453567,
+      "grad_norm": 4.358205768194176e-05,
+      "learning_rate": 5.518080916577158e-06,
+      "loss": 0.0,
+      "step": 7817
+    },
+    {
+      "epoch": 0.5038345040922859,
+      "grad_norm": 0.7133459022193046,
+      "learning_rate": 5.517364840673112e-06,
+      "loss": 0.0042,
+      "step": 7818
+    },
+    {
+      "epoch": 0.5038989495392151,
+      "grad_norm": 0.00675914167301981,
+      "learning_rate": 5.516648764769066e-06,
+      "loss": 0.0,
+      "step": 7819
+    },
+    {
+      "epoch": 0.5039633949861442,
+      "grad_norm": 0.4145019286793012,
+      "learning_rate": 5.51593268886502e-06,
+      "loss": 0.0018,
+      "step": 7820
+    },
+    {
+      "epoch": 0.5040278404330734,
+      "grad_norm": 0.0005312810452736112,
+      "learning_rate": 5.515216612960974e-06,
+      "loss": 0.0,
+      "step": 7821
+    },
+    {
+      "epoch": 0.5040922858800025,
+      "grad_norm": 0.24971380839961985,
+      "learning_rate": 5.514500537056928e-06,
+      "loss": 0.0043,
+      "step": 7822
+    },
+    {
+      "epoch": 0.5041567313269317,
+      "grad_norm": 0.11865125663479083,
+      "learning_rate": 5.5137844611528826e-06,
+      "loss": 0.0014,
+      "step": 7823
+    },
+    {
+      "epoch": 0.5042211767738609,
+      "grad_norm": 0.005578891068497698,
+      "learning_rate": 5.513068385248837e-06,
+      "loss": 0.0,
+      "step": 7824
+    },
+    {
+      "epoch": 0.5042856222207901,
+      "grad_norm": 0.00012931750021159341,
+      "learning_rate": 5.51235230934479e-06,
+      "loss": 0.0,
+      "step": 7825
+    },
+    {
+      "epoch": 0.5043500676677193,
+      "grad_norm": 0.6063908845861173,
+      "learning_rate": 5.511636233440745e-06,
+      "loss": 0.0016,
+      "step": 7826
+    },
+    {
+      "epoch": 0.5044145131146485,
+      "grad_norm": 0.00775264018639695,
+      "learning_rate": 5.510920157536699e-06,
+      "loss": 0.0,
+      "step": 7827
+    },
+    {
+      "epoch": 0.5044789585615777,
+      "grad_norm": 0.1251495777689951,
+      "learning_rate": 5.510204081632653e-06,
+      "loss": 0.0017,
+      "step": 7828
+    },
+    {
+      "epoch": 0.5045434040085068,
+      "grad_norm": 0.08034870037365004,
+      "learning_rate": 5.509488005728608e-06,
+      "loss": 0.0001,
+      "step": 7829
+    },
+    {
+      "epoch": 0.504607849455436,
+      "grad_norm": 0.18931312894291646,
+      "learning_rate": 5.508771929824563e-06,
+      "loss": 0.001,
+      "step": 7830
+    },
+    {
+      "epoch": 0.5046722949023651,
+      "grad_norm": 0.0001198111221057338,
+      "learning_rate": 5.508055853920516e-06,
+      "loss": 0.0,
+      "step": 7831
+    },
+    {
+      "epoch": 0.5047367403492943,
+      "grad_norm": 0.0008712099359854654,
+      "learning_rate": 5.50733977801647e-06,
+      "loss": 0.0,
+      "step": 7832
+    },
+    {
+      "epoch": 0.5048011857962235,
+      "grad_norm": 0.024833433088498218,
+      "learning_rate": 5.506623702112425e-06,
+      "loss": 0.0002,
+      "step": 7833
+    },
+    {
+      "epoch": 0.5048656312431526,
+      "grad_norm": 0.0011160004670191623,
+      "learning_rate": 5.505907626208379e-06,
+      "loss": 0.0,
+      "step": 7834
+    },
+    {
+      "epoch": 0.5049300766900818,
+      "grad_norm": 0.2044019086283718,
+      "learning_rate": 5.505191550304333e-06,
+      "loss": 0.0021,
+      "step": 7835
+    },
+    {
+      "epoch": 0.504994522137011,
+      "grad_norm": 0.0008533538186981164,
+      "learning_rate": 5.504475474400287e-06,
+      "loss": 0.0,
+      "step": 7836
+    },
+    {
+      "epoch": 0.5050589675839402,
+      "grad_norm": 0.000822287982651814,
+      "learning_rate": 5.503759398496241e-06,
+      "loss": 0.0,
+      "step": 7837
+    },
+    {
+      "epoch": 0.5051234130308694,
+      "grad_norm": 0.000461616928902722,
+      "learning_rate": 5.503043322592195e-06,
+      "loss": 0.0,
+      "step": 7838
+    },
+    {
+      "epoch": 0.5051878584777986,
+      "grad_norm": 0.0006680393711834995,
+      "learning_rate": 5.50232724668815e-06,
+      "loss": 0.0,
+      "step": 7839
+    },
+    {
+      "epoch": 0.5052523039247278,
+      "grad_norm": 8.234770240082646e-05,
+      "learning_rate": 5.501611170784104e-06,
+      "loss": 0.0,
+      "step": 7840
+    },
+    {
+      "epoch": 0.5053167493716569,
+      "grad_norm": 0.011275463051318033,
+      "learning_rate": 5.500895094880057e-06,
+      "loss": 0.0001,
+      "step": 7841
+    },
+    {
+      "epoch": 0.5053811948185861,
+      "grad_norm": 8.605929278122782e-05,
+      "learning_rate": 5.500179018976012e-06,
+      "loss": 0.0,
+      "step": 7842
+    },
+    {
+      "epoch": 0.5054456402655152,
+      "grad_norm": 0.678472155824014,
+      "learning_rate": 5.499462943071966e-06,
+      "loss": 0.0057,
+      "step": 7843
+    },
+    {
+      "epoch": 0.5055100857124444,
+      "grad_norm": 0.0004380668417542172,
+      "learning_rate": 5.49874686716792e-06,
+      "loss": 0.0,
+      "step": 7844
+    },
+    {
+      "epoch": 0.5055745311593736,
+      "grad_norm": 0.00023279857462329666,
+      "learning_rate": 5.498030791263875e-06,
+      "loss": 0.0,
+      "step": 7845
+    },
+    {
+      "epoch": 0.5056389766063027,
+      "grad_norm": 0.351107005576601,
+      "learning_rate": 5.497314715359828e-06,
+      "loss": 0.0008,
+      "step": 7846
+    },
+    {
+      "epoch": 0.5057034220532319,
+      "grad_norm": 0.014867512331617283,
+      "learning_rate": 5.496598639455782e-06,
+      "loss": 0.0,
+      "step": 7847
+    },
+    {
+      "epoch": 0.5057678675001611,
+      "grad_norm": 0.00021601504494369152,
+      "learning_rate": 5.495882563551737e-06,
+      "loss": 0.0,
+      "step": 7848
+    },
+    {
+      "epoch": 0.5058323129470903,
+      "grad_norm": 0.000571257955598517,
+      "learning_rate": 5.495166487647691e-06,
+      "loss": 0.0,
+      "step": 7849
+    },
+    {
+      "epoch": 0.5058967583940195,
+      "grad_norm": 0.10286979470954646,
+      "learning_rate": 5.494450411743644e-06,
+      "loss": 0.0003,
+      "step": 7850
+    },
+    {
+      "epoch": 0.5059612038409487,
+      "grad_norm": 0.011815902087395896,
+      "learning_rate": 5.493734335839599e-06,
+      "loss": 0.0001,
+      "step": 7851
+    },
+    {
+      "epoch": 0.5060256492878779,
+      "grad_norm": 0.004256208408072257,
+      "learning_rate": 5.493018259935554e-06,
+      "loss": 0.0,
+      "step": 7852
+    },
+    {
+      "epoch": 0.506090094734807,
+      "grad_norm": 0.0010554367991475037,
+      "learning_rate": 5.492302184031508e-06,
+      "loss": 0.0,
+      "step": 7853
+    },
+    {
+      "epoch": 0.5061545401817361,
+      "grad_norm": 0.0008940419510676588,
+      "learning_rate": 5.4915861081274624e-06,
+      "loss": 0.0,
+      "step": 7854
+    },
+    {
+      "epoch": 0.5062189856286653,
+      "grad_norm": 0.15635816883053835,
+      "learning_rate": 5.490870032223417e-06,
+      "loss": 0.0007,
+      "step": 7855
+    },
+    {
+      "epoch": 0.5062834310755945,
+      "grad_norm": 0.002856546237428052,
+      "learning_rate": 5.490153956319371e-06,
+      "loss": 0.0,
+      "step": 7856
+    },
+    {
+      "epoch": 0.5063478765225237,
+      "grad_norm": 0.13532431522261668,
+      "learning_rate": 5.4894378804153245e-06,
+      "loss": 0.0003,
+      "step": 7857
+    },
+    {
+      "epoch": 0.5064123219694528,
+      "grad_norm": 0.8924651293527359,
+      "learning_rate": 5.488721804511279e-06,
+      "loss": 0.0041,
+      "step": 7858
+    },
+    {
+      "epoch": 0.506476767416382,
+      "grad_norm": 0.0003431231341416375,
+      "learning_rate": 5.488005728607233e-06,
+      "loss": 0.0,
+      "step": 7859
+    },
+    {
+      "epoch": 0.5065412128633112,
+      "grad_norm": 0.14257876513611784,
+      "learning_rate": 5.487289652703187e-06,
+      "loss": 0.0004,
+      "step": 7860
+    },
+    {
+      "epoch": 0.5066056583102404,
+      "grad_norm": 0.5874093706149864,
+      "learning_rate": 5.486573576799142e-06,
+      "loss": 0.0022,
+      "step": 7861
+    },
+    {
+      "epoch": 0.5066701037571696,
+      "grad_norm": 0.0017878685428287893,
+      "learning_rate": 5.485857500895095e-06,
+      "loss": 0.0,
+      "step": 7862
+    },
+    {
+      "epoch": 0.5067345492040988,
+      "grad_norm": 0.041144057837388136,
+      "learning_rate": 5.4851414249910494e-06,
+      "loss": 0.0004,
+      "step": 7863
+    },
+    {
+      "epoch": 0.506798994651028,
+      "grad_norm": 0.003932462138094369,
+      "learning_rate": 5.484425349087004e-06,
+      "loss": 0.0,
+      "step": 7864
+    },
+    {
+      "epoch": 0.506863440097957,
+      "grad_norm": 0.0006095626648205814,
+      "learning_rate": 5.483709273182958e-06,
+      "loss": 0.0,
+      "step": 7865
+    },
+    {
+      "epoch": 0.5069278855448862,
+      "grad_norm": 0.0048959003447374555,
+      "learning_rate": 5.4829931972789115e-06,
+      "loss": 0.0,
+      "step": 7866
+    },
+    {
+      "epoch": 0.5069923309918154,
+      "grad_norm": 0.018113076815429438,
+      "learning_rate": 5.482277121374866e-06,
+      "loss": 0.0001,
+      "step": 7867
+    },
+    {
+      "epoch": 0.5070567764387446,
+      "grad_norm": 0.010513576606503637,
+      "learning_rate": 5.48156104547082e-06,
+      "loss": 0.0,
+      "step": 7868
+    },
+    {
+      "epoch": 0.5071212218856738,
+      "grad_norm": 0.003912950009431698,
+      "learning_rate": 5.480844969566774e-06,
+      "loss": 0.0,
+      "step": 7869
+    },
+    {
+      "epoch": 0.507185667332603,
+      "grad_norm": 0.0023382469396162698,
+      "learning_rate": 5.480128893662729e-06,
+      "loss": 0.0,
+      "step": 7870
+    },
+    {
+      "epoch": 0.5072501127795321,
+      "grad_norm": 0.021264051917344883,
+      "learning_rate": 5.479412817758682e-06,
+      "loss": 0.0,
+      "step": 7871
+    },
+    {
+      "epoch": 0.5073145582264613,
+      "grad_norm": 0.04142688589003656,
+      "learning_rate": 5.4786967418546364e-06,
+      "loss": 0.0001,
+      "step": 7872
+    },
+    {
+      "epoch": 0.5073790036733905,
+      "grad_norm": 0.34118988635976827,
+      "learning_rate": 5.477980665950591e-06,
+      "loss": 0.0026,
+      "step": 7873
+    },
+    {
+      "epoch": 0.5074434491203197,
+      "grad_norm": 0.1676696231890971,
+      "learning_rate": 5.477264590046545e-06,
+      "loss": 0.0003,
+      "step": 7874
+    },
+    {
+      "epoch": 0.5075078945672489,
+      "grad_norm": 0.07809003405108669,
+      "learning_rate": 5.4765485141425e-06,
+      "loss": 0.0005,
+      "step": 7875
+    },
+    {
+      "epoch": 0.507572340014178,
+      "grad_norm": 0.016567306202172546,
+      "learning_rate": 5.4758324382384545e-06,
+      "loss": 0.0001,
+      "step": 7876
+    },
+    {
+      "epoch": 0.5076367854611071,
+      "grad_norm": 0.7576097826830628,
+      "learning_rate": 5.475116362334409e-06,
+      "loss": 0.0029,
+      "step": 7877
+    },
+    {
+      "epoch": 0.5077012309080363,
+      "grad_norm": 0.14010409729777418,
+      "learning_rate": 5.474400286430362e-06,
+      "loss": 0.0001,
+      "step": 7878
+    },
+    {
+      "epoch": 0.5077656763549655,
+      "grad_norm": 0.005552410463870406,
+      "learning_rate": 5.4736842105263165e-06,
+      "loss": 0.0,
+      "step": 7879
+    },
+    {
+      "epoch": 0.5078301218018947,
+      "grad_norm": 0.006620169711987851,
+      "learning_rate": 5.472968134622271e-06,
+      "loss": 0.0,
+      "step": 7880
+    },
+    {
+      "epoch": 0.5078945672488239,
+      "grad_norm": 1.0878805985322575,
+      "learning_rate": 5.472252058718225e-06,
+      "loss": 0.0043,
+      "step": 7881
+    },
+    {
+      "epoch": 0.507959012695753,
+      "grad_norm": 0.003664526663129976,
+      "learning_rate": 5.4715359828141786e-06,
+      "loss": 0.0,
+      "step": 7882
+    },
+    {
+      "epoch": 0.5080234581426822,
+      "grad_norm": 0.051065456594272696,
+      "learning_rate": 5.470819906910133e-06,
+      "loss": 0.0001,
+      "step": 7883
+    },
+    {
+      "epoch": 0.5080879035896114,
+      "grad_norm": 0.0010883020255493963,
+      "learning_rate": 5.470103831006087e-06,
+      "loss": 0.0,
+      "step": 7884
+    },
+    {
+      "epoch": 0.5081523490365406,
+      "grad_norm": 0.3437544432266218,
+      "learning_rate": 5.4693877551020415e-06,
+      "loss": 0.0007,
+      "step": 7885
+    },
+    {
+      "epoch": 0.5082167944834698,
+      "grad_norm": 0.09143956119452665,
+      "learning_rate": 5.468671679197996e-06,
+      "loss": 0.0001,
+      "step": 7886
+    },
+    {
+      "epoch": 0.5082812399303989,
+      "grad_norm": 0.0005676533329750051,
+      "learning_rate": 5.467955603293949e-06,
+      "loss": 0.0,
+      "step": 7887
+    },
+    {
+      "epoch": 0.508345685377328,
+      "grad_norm": 0.14704188747214672,
+      "learning_rate": 5.4672395273899035e-06,
+      "loss": 0.0001,
+      "step": 7888
+    },
+    {
+      "epoch": 0.5084101308242572,
+      "grad_norm": 0.0025846929900272945,
+      "learning_rate": 5.466523451485858e-06,
+      "loss": 0.0,
+      "step": 7889
+    },
+    {
+      "epoch": 0.5084745762711864,
+      "grad_norm": 0.009612976558434051,
+      "learning_rate": 5.465807375581812e-06,
+      "loss": 0.0,
+      "step": 7890
+    },
+    {
+      "epoch": 0.5085390217181156,
+      "grad_norm": 0.2334955803368132,
+      "learning_rate": 5.465091299677766e-06,
+      "loss": 0.0007,
+      "step": 7891
+    },
+    {
+      "epoch": 0.5086034671650448,
+      "grad_norm": 0.039201922748705605,
+      "learning_rate": 5.46437522377372e-06,
+      "loss": 0.0001,
+      "step": 7892
+    },
+    {
+      "epoch": 0.508667912611974,
+      "grad_norm": 0.06696498251734098,
+      "learning_rate": 5.463659147869674e-06,
+      "loss": 0.0001,
+      "step": 7893
+    },
+    {
+      "epoch": 0.5087323580589032,
+      "grad_norm": 0.020123397663725916,
+      "learning_rate": 5.4629430719656285e-06,
+      "loss": 0.0001,
+      "step": 7894
+    },
+    {
+      "epoch": 0.5087968035058323,
+      "grad_norm": 0.010950162090299425,
+      "learning_rate": 5.462226996061583e-06,
+      "loss": 0.0001,
+      "step": 7895
+    },
+    {
+      "epoch": 0.5088612489527615,
+      "grad_norm": 0.005105569536457743,
+      "learning_rate": 5.461510920157537e-06,
+      "loss": 0.0,
+      "step": 7896
+    },
+    {
+      "epoch": 0.5089256943996907,
+      "grad_norm": 0.0021593518982048537,
+      "learning_rate": 5.4607948442534905e-06,
+      "loss": 0.0,
+      "step": 7897
+    },
+    {
+      "epoch": 0.5089901398466198,
+      "grad_norm": 0.007026752353424155,
+      "learning_rate": 5.460078768349445e-06,
+      "loss": 0.0,
+      "step": 7898
+    },
+    {
+      "epoch": 0.509054585293549,
+      "grad_norm": 0.0012650142913674938,
+      "learning_rate": 5.4593626924454e-06,
+      "loss": 0.0,
+      "step": 7899
+    },
+    {
+      "epoch": 0.5091190307404782,
+      "grad_norm": 0.014156076340900966,
+      "learning_rate": 5.458646616541354e-06,
+      "loss": 0.0001,
+      "step": 7900
+    },
+    {
+      "epoch": 0.5091834761874073,
+      "grad_norm": 0.17339007764593933,
+      "learning_rate": 5.4579305406373086e-06,
+      "loss": 0.0008,
+      "step": 7901
+    },
+    {
+      "epoch": 0.5092479216343365,
+      "grad_norm": 0.0112315500039506,
+      "learning_rate": 5.457214464733263e-06,
+      "loss": 0.0,
+      "step": 7902
+    },
+    {
+      "epoch": 0.5093123670812657,
+      "grad_norm": 0.01594386663439557,
+      "learning_rate": 5.456498388829216e-06,
+      "loss": 0.0002,
+      "step": 7903
+    },
+    {
+      "epoch": 0.5093768125281949,
+      "grad_norm": 0.024997468723359482,
+      "learning_rate": 5.455782312925171e-06,
+      "loss": 0.0001,
+      "step": 7904
+    },
+    {
+      "epoch": 0.5094412579751241,
+      "grad_norm": 0.005159837593019308,
+      "learning_rate": 5.455066237021125e-06,
+      "loss": 0.0,
+      "step": 7905
+    },
+    {
+      "epoch": 0.5095057034220533,
+      "grad_norm": 0.013976639008599155,
+      "learning_rate": 5.454350161117079e-06,
+      "loss": 0.0001,
+      "step": 7906
+    },
+    {
+      "epoch": 0.5095701488689824,
+      "grad_norm": 1.2407112226143289,
+      "learning_rate": 5.4536340852130335e-06,
+      "loss": 0.0017,
+      "step": 7907
+    },
+    {
+      "epoch": 0.5096345943159116,
+      "grad_norm": 0.0009534155375335451,
+      "learning_rate": 5.452918009308987e-06,
+      "loss": 0.0,
+      "step": 7908
+    },
+    {
+      "epoch": 0.5096990397628407,
+      "grad_norm": 0.0010277082009812432,
+      "learning_rate": 5.452201933404941e-06,
+      "loss": 0.0,
+      "step": 7909
+    },
+    {
+      "epoch": 0.5097634852097699,
+      "grad_norm": 0.012268255597643768,
+      "learning_rate": 5.4514858575008955e-06,
+      "loss": 0.0,
+      "step": 7910
+    },
+    {
+      "epoch": 0.5098279306566991,
+      "grad_norm": 0.05539452184990833,
+      "learning_rate": 5.45076978159685e-06,
+      "loss": 0.0003,
+      "step": 7911
+    },
+    {
+      "epoch": 0.5098923761036283,
+      "grad_norm": 0.002352447691784602,
+      "learning_rate": 5.450053705692804e-06,
+      "loss": 0.0,
+      "step": 7912
+    },
+    {
+      "epoch": 0.5099568215505574,
+      "grad_norm": 0.000641193713048714,
+      "learning_rate": 5.449337629788758e-06,
+      "loss": 0.0,
+      "step": 7913
+    },
+    {
+      "epoch": 0.5100212669974866,
+      "grad_norm": 0.026912041162248948,
+      "learning_rate": 5.448621553884712e-06,
+      "loss": 0.0001,
+      "step": 7914
+    },
+    {
+      "epoch": 0.5100857124444158,
+      "grad_norm": 0.007920404774023341,
+      "learning_rate": 5.447905477980666e-06,
+      "loss": 0.0001,
+      "step": 7915
+    },
+    {
+      "epoch": 0.510150157891345,
+      "grad_norm": 0.252571013657588,
+      "learning_rate": 5.4471894020766205e-06,
+      "loss": 0.0003,
+      "step": 7916
+    },
+    {
+      "epoch": 0.5102146033382742,
+      "grad_norm": 0.006732229568060559,
+      "learning_rate": 5.446473326172575e-06,
+      "loss": 0.0,
+      "step": 7917
+    },
+    {
+      "epoch": 0.5102790487852034,
+      "grad_norm": 0.008280544471657686,
+      "learning_rate": 5.445757250268528e-06,
+      "loss": 0.0,
+      "step": 7918
+    },
+    {
+      "epoch": 0.5103434942321325,
+      "grad_norm": 0.003950611251451125,
+      "learning_rate": 5.4450411743644825e-06,
+      "loss": 0.0,
+      "step": 7919
+    },
+    {
+      "epoch": 0.5104079396790617,
+      "grad_norm": 0.20482560914732528,
+      "learning_rate": 5.444325098460437e-06,
+      "loss": 0.0017,
+      "step": 7920
+    },
+    {
+      "epoch": 0.5104723851259908,
+      "grad_norm": 0.13401776278697505,
+      "learning_rate": 5.443609022556391e-06,
+      "loss": 0.0002,
+      "step": 7921
+    },
+    {
+      "epoch": 0.51053683057292,
+      "grad_norm": 0.07180763760137501,
+      "learning_rate": 5.442892946652346e-06,
+      "loss": 0.0002,
+      "step": 7922
+    },
+    {
+      "epoch": 0.5106012760198492,
+      "grad_norm": 0.015309862017110912,
+      "learning_rate": 5.442176870748301e-06,
+      "loss": 0.0,
+      "step": 7923
+    },
+    {
+      "epoch": 0.5106657214667784,
+      "grad_norm": 0.0011142132569021373,
+      "learning_rate": 5.441460794844254e-06,
+      "loss": 0.0,
+      "step": 7924
+    },
+    {
+      "epoch": 0.5107301669137075,
+      "grad_norm": 0.0024562087064645146,
+      "learning_rate": 5.440744718940208e-06,
+      "loss": 0.0,
+      "step": 7925
+    },
+    {
+      "epoch": 0.5107946123606367,
+      "grad_norm": 0.0013992305215337778,
+      "learning_rate": 5.440028643036163e-06,
+      "loss": 0.0,
+      "step": 7926
+    },
+    {
+      "epoch": 0.5108590578075659,
+      "grad_norm": 0.0032199139412421437,
+      "learning_rate": 5.439312567132117e-06,
+      "loss": 0.0,
+      "step": 7927
+    },
+    {
+      "epoch": 0.5109235032544951,
+      "grad_norm": 0.23834775559403137,
+      "learning_rate": 5.438596491228071e-06,
+      "loss": 0.0019,
+      "step": 7928
+    },
+    {
+      "epoch": 0.5109879487014243,
+      "grad_norm": 0.0745276654029238,
+      "learning_rate": 5.437880415324025e-06,
+      "loss": 0.0001,
+      "step": 7929
+    },
+    {
+      "epoch": 0.5110523941483535,
+      "grad_norm": 0.004944950771004873,
+      "learning_rate": 5.437164339419979e-06,
+      "loss": 0.0,
+      "step": 7930
+    },
+    {
+      "epoch": 0.5111168395952826,
+      "grad_norm": 0.00038942084221351607,
+      "learning_rate": 5.436448263515933e-06,
+      "loss": 0.0,
+      "step": 7931
+    },
+    {
+      "epoch": 0.5111812850422117,
+      "grad_norm": 0.0005447171405046528,
+      "learning_rate": 5.435732187611888e-06,
+      "loss": 0.0,
+      "step": 7932
+    },
+    {
+      "epoch": 0.5112457304891409,
+      "grad_norm": 0.002356542261505252,
+      "learning_rate": 5.435016111707842e-06,
+      "loss": 0.0,
+      "step": 7933
+    },
+    {
+      "epoch": 0.5113101759360701,
+      "grad_norm": 0.0012706526169710428,
+      "learning_rate": 5.434300035803795e-06,
+      "loss": 0.0,
+      "step": 7934
+    },
+    {
+      "epoch": 0.5113746213829993,
+      "grad_norm": 0.07874067125905353,
+      "learning_rate": 5.43358395989975e-06,
+      "loss": 0.0001,
+      "step": 7935
+    },
+    {
+      "epoch": 0.5114390668299285,
+      "grad_norm": 0.001029995249023795,
+      "learning_rate": 5.432867883995704e-06,
+      "loss": 0.0,
+      "step": 7936
+    },
+    {
+      "epoch": 0.5115035122768576,
+      "grad_norm": 0.001838396980398961,
+      "learning_rate": 5.432151808091658e-06,
+      "loss": 0.0,
+      "step": 7937
+    },
+    {
+      "epoch": 0.5115679577237868,
+      "grad_norm": 0.0005883734590697582,
+      "learning_rate": 5.4314357321876125e-06,
+      "loss": 0.0,
+      "step": 7938
+    },
+    {
+      "epoch": 0.511632403170716,
+      "grad_norm": 0.0172669531723459,
+      "learning_rate": 5.430719656283566e-06,
+      "loss": 0.0,
+      "step": 7939
+    },
+    {
+      "epoch": 0.5116968486176452,
+      "grad_norm": 0.0009709076387414066,
+      "learning_rate": 5.43000358037952e-06,
+      "loss": 0.0,
+      "step": 7940
+    },
+    {
+      "epoch": 0.5117612940645744,
+      "grad_norm": 0.008330469998503403,
+      "learning_rate": 5.4292875044754746e-06,
+      "loss": 0.0001,
+      "step": 7941
+    },
+    {
+      "epoch": 0.5118257395115036,
+      "grad_norm": 0.008222742107848081,
+      "learning_rate": 5.428571428571429e-06,
+      "loss": 0.0,
+      "step": 7942
+    },
+    {
+      "epoch": 0.5118901849584326,
+      "grad_norm": 0.0035724153632476397,
+      "learning_rate": 5.427855352667382e-06,
+      "loss": 0.0,
+      "step": 7943
+    },
+    {
+      "epoch": 0.5119546304053618,
+      "grad_norm": 0.0916269789696049,
+      "learning_rate": 5.427139276763337e-06,
+      "loss": 0.0002,
+      "step": 7944
+    },
+    {
+      "epoch": 0.512019075852291,
+      "grad_norm": 0.005102800168367895,
+      "learning_rate": 5.426423200859291e-06,
+      "loss": 0.0,
+      "step": 7945
+    },
+    {
+      "epoch": 0.5120835212992202,
+      "grad_norm": 0.11785411775680807,
+      "learning_rate": 5.425707124955246e-06,
+      "loss": 0.0001,
+      "step": 7946
+    },
+    {
+      "epoch": 0.5121479667461494,
+      "grad_norm": 0.006188772612846142,
+      "learning_rate": 5.4249910490512e-06,
+      "loss": 0.0,
+      "step": 7947
+    },
+    {
+      "epoch": 0.5122124121930786,
+      "grad_norm": 0.0010159282271246758,
+      "learning_rate": 5.424274973147155e-06,
+      "loss": 0.0,
+      "step": 7948
+    },
+    {
+      "epoch": 0.5122768576400077,
+      "grad_norm": 0.05386086260127591,
+      "learning_rate": 5.423558897243109e-06,
+      "loss": 0.0002,
+      "step": 7949
+    },
+    {
+      "epoch": 0.5123413030869369,
+      "grad_norm": 0.15529134635092198,
+      "learning_rate": 5.422842821339062e-06,
+      "loss": 0.0002,
+      "step": 7950
+    },
+    {
+      "epoch": 0.5124057485338661,
+      "grad_norm": 0.01658001019660191,
+      "learning_rate": 5.422126745435017e-06,
+      "loss": 0.0001,
+      "step": 7951
+    },
+    {
+      "epoch": 0.5124701939807953,
+      "grad_norm": 0.0009362448979712939,
+      "learning_rate": 5.421410669530971e-06,
+      "loss": 0.0,
+      "step": 7952
+    },
+    {
+      "epoch": 0.5125346394277245,
+      "grad_norm": 0.00837422541313603,
+      "learning_rate": 5.420694593626925e-06,
+      "loss": 0.0001,
+      "step": 7953
+    },
+    {
+      "epoch": 0.5125990848746536,
+      "grad_norm": 0.015030309507146898,
+      "learning_rate": 5.41997851772288e-06,
+      "loss": 0.0,
+      "step": 7954
+    },
+    {
+      "epoch": 0.5126635303215827,
+      "grad_norm": 0.00020856912114475138,
+      "learning_rate": 5.419262441818833e-06,
+      "loss": 0.0,
+      "step": 7955
+    },
+    {
+      "epoch": 0.5127279757685119,
+      "grad_norm": 0.04200720795641049,
+      "learning_rate": 5.418546365914787e-06,
+      "loss": 0.0001,
+      "step": 7956
+    },
+    {
+      "epoch": 0.5127924212154411,
+      "grad_norm": 3.0653002645703564,
+      "learning_rate": 5.417830290010742e-06,
+      "loss": 0.0219,
+      "step": 7957
+    },
+    {
+      "epoch": 0.5128568666623703,
+      "grad_norm": 0.0005120299741213488,
+      "learning_rate": 5.417114214106696e-06,
+      "loss": 0.0,
+      "step": 7958
+    },
+    {
+      "epoch": 0.5129213121092995,
+      "grad_norm": 0.03933544465942909,
+      "learning_rate": 5.416398138202649e-06,
+      "loss": 0.0001,
+      "step": 7959
+    },
+    {
+      "epoch": 0.5129857575562287,
+      "grad_norm": 0.0012020741314226874,
+      "learning_rate": 5.415682062298604e-06,
+      "loss": 0.0,
+      "step": 7960
+    },
+    {
+      "epoch": 0.5130502030031578,
+      "grad_norm": 0.0005658398741220849,
+      "learning_rate": 5.414965986394558e-06,
+      "loss": 0.0,
+      "step": 7961
+    },
+    {
+      "epoch": 0.513114648450087,
+      "grad_norm": 0.017119137317332913,
+      "learning_rate": 5.414249910490512e-06,
+      "loss": 0.0002,
+      "step": 7962
+    },
+    {
+      "epoch": 0.5131790938970162,
+      "grad_norm": 0.002558508419807508,
+      "learning_rate": 5.413533834586467e-06,
+      "loss": 0.0,
+      "step": 7963
+    },
+    {
+      "epoch": 0.5132435393439454,
+      "grad_norm": 0.0348350584973022,
+      "learning_rate": 5.41281775868242e-06,
+      "loss": 0.0002,
+      "step": 7964
+    },
+    {
+      "epoch": 0.5133079847908745,
+      "grad_norm": 0.0015763089749261437,
+      "learning_rate": 5.412101682778374e-06,
+      "loss": 0.0,
+      "step": 7965
+    },
+    {
+      "epoch": 0.5133724302378037,
+      "grad_norm": 0.073376589482896,
+      "learning_rate": 5.411385606874329e-06,
+      "loss": 0.0001,
+      "step": 7966
+    },
+    {
+      "epoch": 0.5134368756847328,
+      "grad_norm": 0.008660317254924959,
+      "learning_rate": 5.410669530970283e-06,
+      "loss": 0.0,
+      "step": 7967
+    },
+    {
+      "epoch": 0.513501321131662,
+      "grad_norm": 0.00013401005881552363,
+      "learning_rate": 5.409953455066237e-06,
+      "loss": 0.0,
+      "step": 7968
+    },
+    {
+      "epoch": 0.5135657665785912,
+      "grad_norm": 0.009542936602791867,
+      "learning_rate": 5.409237379162192e-06,
+      "loss": 0.0,
+      "step": 7969
+    },
+    {
+      "epoch": 0.5136302120255204,
+      "grad_norm": 0.06508845696056229,
+      "learning_rate": 5.408521303258147e-06,
+      "loss": 0.0002,
+      "step": 7970
+    },
+    {
+      "epoch": 0.5136946574724496,
+      "grad_norm": 0.00033976029903329643,
+      "learning_rate": 5.4078052273541e-06,
+      "loss": 0.0,
+      "step": 7971
+    },
+    {
+      "epoch": 0.5137591029193788,
+      "grad_norm": 0.0012616412870784784,
+      "learning_rate": 5.4070891514500544e-06,
+      "loss": 0.0,
+      "step": 7972
+    },
+    {
+      "epoch": 0.513823548366308,
+      "grad_norm": 0.03339331166490907,
+      "learning_rate": 5.406373075546009e-06,
+      "loss": 0.0001,
+      "step": 7973
+    },
+    {
+      "epoch": 0.5138879938132371,
+      "grad_norm": 0.11922746760816738,
+      "learning_rate": 5.405656999641963e-06,
+      "loss": 0.0002,
+      "step": 7974
+    },
+    {
+      "epoch": 0.5139524392601663,
+      "grad_norm": 0.018733981766930537,
+      "learning_rate": 5.4049409237379165e-06,
+      "loss": 0.0001,
+      "step": 7975
+    },
+    {
+      "epoch": 0.5140168847070954,
+      "grad_norm": 0.1374454530979648,
+      "learning_rate": 5.404224847833871e-06,
+      "loss": 0.0002,
+      "step": 7976
+    },
+    {
+      "epoch": 0.5140813301540246,
+      "grad_norm": 0.2097619022628654,
+      "learning_rate": 5.403508771929825e-06,
+      "loss": 0.001,
+      "step": 7977
+    },
+    {
+      "epoch": 0.5141457756009538,
+      "grad_norm": 0.0008893098012604791,
+      "learning_rate": 5.402792696025779e-06,
+      "loss": 0.0,
+      "step": 7978
+    },
+    {
+      "epoch": 0.5142102210478829,
+      "grad_norm": 0.00014497031483067468,
+      "learning_rate": 5.402076620121734e-06,
+      "loss": 0.0,
+      "step": 7979
+    },
+    {
+      "epoch": 0.5142746664948121,
+      "grad_norm": 0.002657034458652046,
+      "learning_rate": 5.401360544217687e-06,
+      "loss": 0.0,
+      "step": 7980
+    },
+    {
+      "epoch": 0.5143391119417413,
+      "grad_norm": 0.001948556525665166,
+      "learning_rate": 5.4006444683136414e-06,
+      "loss": 0.0,
+      "step": 7981
+    },
+    {
+      "epoch": 0.5144035573886705,
+      "grad_norm": 0.0023215001057442755,
+      "learning_rate": 5.399928392409596e-06,
+      "loss": 0.0,
+      "step": 7982
+    },
+    {
+      "epoch": 0.5144680028355997,
+      "grad_norm": 0.008492666356822762,
+      "learning_rate": 5.39921231650555e-06,
+      "loss": 0.0,
+      "step": 7983
+    },
+    {
+      "epoch": 0.5145324482825289,
+      "grad_norm": 0.006237242050533532,
+      "learning_rate": 5.398496240601504e-06,
+      "loss": 0.0,
+      "step": 7984
+    },
+    {
+      "epoch": 0.514596893729458,
+      "grad_norm": 0.2957292205100141,
+      "learning_rate": 5.397780164697458e-06,
+      "loss": 0.0025,
+      "step": 7985
+    },
+    {
+      "epoch": 0.5146613391763872,
+      "grad_norm": 0.0006000581207181238,
+      "learning_rate": 5.397064088793412e-06,
+      "loss": 0.0,
+      "step": 7986
+    },
+    {
+      "epoch": 0.5147257846233164,
+      "grad_norm": 0.0009272405372127528,
+      "learning_rate": 5.396348012889366e-06,
+      "loss": 0.0,
+      "step": 7987
+    },
+    {
+      "epoch": 0.5147902300702455,
+      "grad_norm": 0.22983715021964995,
+      "learning_rate": 5.395631936985321e-06,
+      "loss": 0.0005,
+      "step": 7988
+    },
+    {
+      "epoch": 0.5148546755171747,
+      "grad_norm": 0.0018526735076136297,
+      "learning_rate": 5.394915861081275e-06,
+      "loss": 0.0,
+      "step": 7989
+    },
+    {
+      "epoch": 0.5149191209641039,
+      "grad_norm": 5.234039816139064e-05,
+      "learning_rate": 5.3941997851772284e-06,
+      "loss": 0.0,
+      "step": 7990
+    },
+    {
+      "epoch": 0.514983566411033,
+      "grad_norm": 1.0890406449039203,
+      "learning_rate": 5.393483709273183e-06,
+      "loss": 0.0013,
+      "step": 7991
+    },
+    {
+      "epoch": 0.5150480118579622,
+      "grad_norm": 0.0010511426046532501,
+      "learning_rate": 5.392767633369138e-06,
+      "loss": 0.0,
+      "step": 7992
+    },
+    {
+      "epoch": 0.5151124573048914,
+      "grad_norm": 0.0011704403448952318,
+      "learning_rate": 5.392051557465092e-06,
+      "loss": 0.0,
+      "step": 7993
+    },
+    {
+      "epoch": 0.5151769027518206,
+      "grad_norm": 0.001237749221018552,
+      "learning_rate": 5.3913354815610465e-06,
+      "loss": 0.0,
+      "step": 7994
+    },
+    {
+      "epoch": 0.5152413481987498,
+      "grad_norm": 0.03349309412923939,
+      "learning_rate": 5.390619405657001e-06,
+      "loss": 0.0002,
+      "step": 7995
+    },
+    {
+      "epoch": 0.515305793645679,
+      "grad_norm": 0.16808593343973655,
+      "learning_rate": 5.389903329752954e-06,
+      "loss": 0.0034,
+      "step": 7996
+    },
+    {
+      "epoch": 0.5153702390926082,
+      "grad_norm": 0.0013906726617820943,
+      "learning_rate": 5.3891872538489085e-06,
+      "loss": 0.0,
+      "step": 7997
+    },
+    {
+      "epoch": 0.5154346845395373,
+      "grad_norm": 0.0045494031450296355,
+      "learning_rate": 5.388471177944863e-06,
+      "loss": 0.0,
+      "step": 7998
+    },
+    {
+      "epoch": 0.5154991299864664,
+      "grad_norm": 0.007636372474605849,
+      "learning_rate": 5.387755102040817e-06,
+      "loss": 0.0,
+      "step": 7999
+    },
+    {
+      "epoch": 0.5155635754333956,
+      "grad_norm": 0.0015319027312418868,
+      "learning_rate": 5.387039026136771e-06,
+      "loss": 0.0,
+      "step": 8000
+    },
+    {
+      "epoch": 0.5156280208803248,
+      "grad_norm": 0.004652866648377999,
+      "learning_rate": 5.386322950232725e-06,
+      "loss": 0.0,
+      "step": 8001
+    },
+    {
+      "epoch": 0.515692466327254,
+      "grad_norm": 0.0018567965467686757,
+      "learning_rate": 5.385606874328679e-06,
+      "loss": 0.0,
+      "step": 8002
+    },
+    {
+      "epoch": 0.5157569117741831,
+      "grad_norm": 0.038382413997460466,
+      "learning_rate": 5.3848907984246335e-06,
+      "loss": 0.0001,
+      "step": 8003
+    },
+    {
+      "epoch": 0.5158213572211123,
+      "grad_norm": 0.0003577640989233756,
+      "learning_rate": 5.384174722520588e-06,
+      "loss": 0.0,
+      "step": 8004
+    },
+    {
+      "epoch": 0.5158858026680415,
+      "grad_norm": 0.19394153080352525,
+      "learning_rate": 5.383458646616542e-06,
+      "loss": 0.0002,
+      "step": 8005
+    },
+    {
+      "epoch": 0.5159502481149707,
+      "grad_norm": 0.0043557760796160425,
+      "learning_rate": 5.3827425707124955e-06,
+      "loss": 0.0,
+      "step": 8006
+    },
+    {
+      "epoch": 0.5160146935618999,
+      "grad_norm": 0.26720070499559867,
+      "learning_rate": 5.38202649480845e-06,
+      "loss": 0.0005,
+      "step": 8007
+    },
+    {
+      "epoch": 0.5160791390088291,
+      "grad_norm": 0.1252867271194741,
+      "learning_rate": 5.381310418904404e-06,
+      "loss": 0.0002,
+      "step": 8008
+    },
+    {
+      "epoch": 0.5161435844557583,
+      "grad_norm": 0.00033245680958482986,
+      "learning_rate": 5.380594343000358e-06,
+      "loss": 0.0,
+      "step": 8009
+    },
+    {
+      "epoch": 0.5162080299026873,
+      "grad_norm": 0.0025579589688075007,
+      "learning_rate": 5.379878267096313e-06,
+      "loss": 0.0,
+      "step": 8010
+    },
+    {
+      "epoch": 0.5162724753496165,
+      "grad_norm": 0.009223071042598662,
+      "learning_rate": 5.379162191192266e-06,
+      "loss": 0.0,
+      "step": 8011
+    },
+    {
+      "epoch": 0.5163369207965457,
+      "grad_norm": 0.5300760762477967,
+      "learning_rate": 5.3784461152882205e-06,
+      "loss": 0.0006,
+      "step": 8012
+    },
+    {
+      "epoch": 0.5164013662434749,
+      "grad_norm": 0.011043418132426576,
+      "learning_rate": 5.377730039384175e-06,
+      "loss": 0.0,
+      "step": 8013
+    },
+    {
+      "epoch": 0.5164658116904041,
+      "grad_norm": 0.00021837471094835563,
+      "learning_rate": 5.377013963480129e-06,
+      "loss": 0.0,
+      "step": 8014
+    },
+    {
+      "epoch": 0.5165302571373332,
+      "grad_norm": 0.3276306356442377,
+      "learning_rate": 5.376297887576083e-06,
+      "loss": 0.0008,
+      "step": 8015
+    },
+    {
+      "epoch": 0.5165947025842624,
+      "grad_norm": 0.2765679994672126,
+      "learning_rate": 5.3755818116720385e-06,
+      "loss": 0.0008,
+      "step": 8016
+    },
+    {
+      "epoch": 0.5166591480311916,
+      "grad_norm": 0.18362511711818005,
+      "learning_rate": 5.374865735767992e-06,
+      "loss": 0.001,
+      "step": 8017
+    },
+    {
+      "epoch": 0.5167235934781208,
+      "grad_norm": 0.002038531219077466,
+      "learning_rate": 5.374149659863946e-06,
+      "loss": 0.0,
+      "step": 8018
+    },
+    {
+      "epoch": 0.51678803892505,
+      "grad_norm": 0.00524951653881258,
+      "learning_rate": 5.3734335839599006e-06,
+      "loss": 0.0,
+      "step": 8019
+    },
+    {
+      "epoch": 0.5168524843719792,
+      "grad_norm": 0.003735285914029635,
+      "learning_rate": 5.372717508055855e-06,
+      "loss": 0.0,
+      "step": 8020
+    },
+    {
+      "epoch": 0.5169169298189082,
+      "grad_norm": 0.048634071984891634,
+      "learning_rate": 5.372001432151809e-06,
+      "loss": 0.0001,
+      "step": 8021
+    },
+    {
+      "epoch": 0.5169813752658374,
+      "grad_norm": 0.007575055206033924,
+      "learning_rate": 5.371285356247763e-06,
+      "loss": 0.0,
+      "step": 8022
+    },
+    {
+      "epoch": 0.5170458207127666,
+      "grad_norm": 0.0896404079922022,
+      "learning_rate": 5.370569280343717e-06,
+      "loss": 0.001,
+      "step": 8023
+    },
+    {
+      "epoch": 0.5171102661596958,
+      "grad_norm": 0.0009642160243532904,
+      "learning_rate": 5.369853204439671e-06,
+      "loss": 0.0,
+      "step": 8024
+    },
+    {
+      "epoch": 0.517174711606625,
+      "grad_norm": 9.580468636606124e-05,
+      "learning_rate": 5.3691371285356255e-06,
+      "loss": 0.0,
+      "step": 8025
+    },
+    {
+      "epoch": 0.5172391570535542,
+      "grad_norm": 0.00038877102939403603,
+      "learning_rate": 5.36842105263158e-06,
+      "loss": 0.0,
+      "step": 8026
+    },
+    {
+      "epoch": 0.5173036025004834,
+      "grad_norm": 0.017995206744111487,
+      "learning_rate": 5.367704976727533e-06,
+      "loss": 0.0,
+      "step": 8027
+    },
+    {
+      "epoch": 0.5173680479474125,
+      "grad_norm": 0.00025071646488426043,
+      "learning_rate": 5.3669889008234875e-06,
+      "loss": 0.0,
+      "step": 8028
+    },
+    {
+      "epoch": 0.5174324933943417,
+      "grad_norm": 0.5368235711987875,
+      "learning_rate": 5.366272824919442e-06,
+      "loss": 0.003,
+      "step": 8029
+    },
+    {
+      "epoch": 0.5174969388412709,
+      "grad_norm": 0.5099628110958586,
+      "learning_rate": 5.365556749015396e-06,
+      "loss": 0.0036,
+      "step": 8030
+    },
+    {
+      "epoch": 0.5175613842882001,
+      "grad_norm": 0.016237080453299548,
+      "learning_rate": 5.3648406731113504e-06,
+      "loss": 0.0001,
+      "step": 8031
+    },
+    {
+      "epoch": 0.5176258297351292,
+      "grad_norm": 0.037514827573428736,
+      "learning_rate": 5.364124597207304e-06,
+      "loss": 0.0002,
+      "step": 8032
+    },
+    {
+      "epoch": 0.5176902751820583,
+      "grad_norm": 0.007461747262392446,
+      "learning_rate": 5.363408521303258e-06,
+      "loss": 0.0,
+      "step": 8033
+    },
+    {
+      "epoch": 0.5177547206289875,
+      "grad_norm": 0.003969606536505866,
+      "learning_rate": 5.3626924453992125e-06,
+      "loss": 0.0,
+      "step": 8034
+    },
+    {
+      "epoch": 0.5178191660759167,
+      "grad_norm": 0.0012055136054845725,
+      "learning_rate": 5.361976369495167e-06,
+      "loss": 0.0,
+      "step": 8035
+    },
+    {
+      "epoch": 0.5178836115228459,
+      "grad_norm": 0.04324137427823979,
+      "learning_rate": 5.36126029359112e-06,
+      "loss": 0.0002,
+      "step": 8036
+    },
+    {
+      "epoch": 0.5179480569697751,
+      "grad_norm": 0.022892261141917725,
+      "learning_rate": 5.3605442176870745e-06,
+      "loss": 0.0,
+      "step": 8037
+    },
+    {
+      "epoch": 0.5180125024167043,
+      "grad_norm": 0.0011791094418522764,
+      "learning_rate": 5.359828141783029e-06,
+      "loss": 0.0,
+      "step": 8038
+    },
+    {
+      "epoch": 0.5180769478636335,
+      "grad_norm": 0.01512018866252014,
+      "learning_rate": 5.359112065878984e-06,
+      "loss": 0.0,
+      "step": 8039
+    },
+    {
+      "epoch": 0.5181413933105626,
+      "grad_norm": 0.17008817142783822,
+      "learning_rate": 5.358395989974938e-06,
+      "loss": 0.0019,
+      "step": 8040
+    },
+    {
+      "epoch": 0.5182058387574918,
+      "grad_norm": 0.06252434748028268,
+      "learning_rate": 5.357679914070893e-06,
+      "loss": 0.0003,
+      "step": 8041
+    },
+    {
+      "epoch": 0.518270284204421,
+      "grad_norm": 0.0023996196919451067,
+      "learning_rate": 5.356963838166847e-06,
+      "loss": 0.0,
+      "step": 8042
+    },
+    {
+      "epoch": 0.5183347296513501,
+      "grad_norm": 0.0007908011827102855,
+      "learning_rate": 5.3562477622628e-06,
+      "loss": 0.0,
+      "step": 8043
+    },
+    {
+      "epoch": 0.5183991750982793,
+      "grad_norm": 0.02134104283681624,
+      "learning_rate": 5.355531686358755e-06,
+      "loss": 0.0003,
+      "step": 8044
+    },
+    {
+      "epoch": 0.5184636205452084,
+      "grad_norm": 0.03731752513019512,
+      "learning_rate": 5.354815610454709e-06,
+      "loss": 0.0001,
+      "step": 8045
+    },
+    {
+      "epoch": 0.5185280659921376,
+      "grad_norm": 0.04644805216348859,
+      "learning_rate": 5.354099534550663e-06,
+      "loss": 0.0003,
+      "step": 8046
+    },
+    {
+      "epoch": 0.5185925114390668,
+      "grad_norm": 0.35393408777423346,
+      "learning_rate": 5.3533834586466175e-06,
+      "loss": 0.0017,
+      "step": 8047
+    },
+    {
+      "epoch": 0.518656956885996,
+      "grad_norm": 0.007836727577587311,
+      "learning_rate": 5.352667382742571e-06,
+      "loss": 0.0,
+      "step": 8048
+    },
+    {
+      "epoch": 0.5187214023329252,
+      "grad_norm": 0.002620074872959019,
+      "learning_rate": 5.351951306838525e-06,
+      "loss": 0.0,
+      "step": 8049
+    },
+    {
+      "epoch": 0.5187858477798544,
+      "grad_norm": 0.0004816107082796707,
+      "learning_rate": 5.35123523093448e-06,
+      "loss": 0.0,
+      "step": 8050
+    },
+    {
+      "epoch": 0.5188502932267836,
+      "grad_norm": 0.00032189141713238927,
+      "learning_rate": 5.350519155030434e-06,
+      "loss": 0.0,
+      "step": 8051
+    },
+    {
+      "epoch": 0.5189147386737127,
+      "grad_norm": 0.001981909159571749,
+      "learning_rate": 5.349803079126387e-06,
+      "loss": 0.0,
+      "step": 8052
+    },
+    {
+      "epoch": 0.5189791841206419,
+      "grad_norm": 0.004482247778658294,
+      "learning_rate": 5.349087003222342e-06,
+      "loss": 0.0,
+      "step": 8053
+    },
+    {
+      "epoch": 0.519043629567571,
+      "grad_norm": 0.0429455602448583,
+      "learning_rate": 5.348370927318296e-06,
+      "loss": 0.0001,
+      "step": 8054
+    },
+    {
+      "epoch": 0.5191080750145002,
+      "grad_norm": 0.00010109813583239728,
+      "learning_rate": 5.34765485141425e-06,
+      "loss": 0.0,
+      "step": 8055
+    },
+    {
+      "epoch": 0.5191725204614294,
+      "grad_norm": 0.005013169175907346,
+      "learning_rate": 5.3469387755102045e-06,
+      "loss": 0.0,
+      "step": 8056
+    },
+    {
+      "epoch": 0.5192369659083585,
+      "grad_norm": 0.0026910242392564347,
+      "learning_rate": 5.346222699606158e-06,
+      "loss": 0.0,
+      "step": 8057
+    },
+    {
+      "epoch": 0.5193014113552877,
+      "grad_norm": 0.001115195455531372,
+      "learning_rate": 5.345506623702112e-06,
+      "loss": 0.0,
+      "step": 8058
+    },
+    {
+      "epoch": 0.5193658568022169,
+      "grad_norm": 0.0036148613287218584,
+      "learning_rate": 5.3447905477980666e-06,
+      "loss": 0.0,
+      "step": 8059
+    },
+    {
+      "epoch": 0.5194303022491461,
+      "grad_norm": 0.012834000600551449,
+      "learning_rate": 5.344074471894021e-06,
+      "loss": 0.0,
+      "step": 8060
+    },
+    {
+      "epoch": 0.5194947476960753,
+      "grad_norm": 0.00019800514969051393,
+      "learning_rate": 5.343358395989975e-06,
+      "loss": 0.0015,
+      "step": 8061
+    },
+    {
+      "epoch": 0.5195591931430045,
+      "grad_norm": 7.261156807335496e-05,
+      "learning_rate": 5.34264232008593e-06,
+      "loss": 0.0,
+      "step": 8062
+    },
+    {
+      "epoch": 0.5196236385899337,
+      "grad_norm": 0.00010452706061990569,
+      "learning_rate": 5.341926244181885e-06,
+      "loss": 0.0,
+      "step": 8063
+    },
+    {
+      "epoch": 0.5196880840368628,
+      "grad_norm": 0.0015938797537969481,
+      "learning_rate": 5.341210168277838e-06,
+      "loss": 0.0,
+      "step": 8064
+    },
+    {
+      "epoch": 0.519752529483792,
+      "grad_norm": 0.0006986058601608112,
+      "learning_rate": 5.340494092373792e-06,
+      "loss": 0.0,
+      "step": 8065
+    },
+    {
+      "epoch": 0.5198169749307211,
+      "grad_norm": 0.00018862228638009165,
+      "learning_rate": 5.339778016469747e-06,
+      "loss": 0.0,
+      "step": 8066
+    },
+    {
+      "epoch": 0.5198814203776503,
+      "grad_norm": 0.04569480241493098,
+      "learning_rate": 5.339061940565701e-06,
+      "loss": 0.0004,
+      "step": 8067
+    },
+    {
+      "epoch": 0.5199458658245795,
+      "grad_norm": 0.00048435932877395973,
+      "learning_rate": 5.338345864661654e-06,
+      "loss": 0.0,
+      "step": 8068
+    },
+    {
+      "epoch": 0.5200103112715087,
+      "grad_norm": 4.264877117970066e-05,
+      "learning_rate": 5.337629788757609e-06,
+      "loss": 0.0,
+      "step": 8069
+    },
+    {
+      "epoch": 0.5200747567184378,
+      "grad_norm": 0.022674035570682753,
+      "learning_rate": 5.336913712853563e-06,
+      "loss": 0.0001,
+      "step": 8070
+    },
+    {
+      "epoch": 0.520139202165367,
+      "grad_norm": 7.506351649127817e-05,
+      "learning_rate": 5.336197636949517e-06,
+      "loss": 0.0,
+      "step": 8071
+    },
+    {
+      "epoch": 0.5202036476122962,
+      "grad_norm": 0.0002121658204572475,
+      "learning_rate": 5.335481561045472e-06,
+      "loss": 0.0,
+      "step": 8072
+    },
+    {
+      "epoch": 0.5202680930592254,
+      "grad_norm": 0.0005612859992496004,
+      "learning_rate": 5.334765485141425e-06,
+      "loss": 0.0,
+      "step": 8073
+    },
+    {
+      "epoch": 0.5203325385061546,
+      "grad_norm": 0.0005723835744671824,
+      "learning_rate": 5.334049409237379e-06,
+      "loss": 0.0,
+      "step": 8074
+    },
+    {
+      "epoch": 0.5203969839530838,
+      "grad_norm": 4.74043528609633e-05,
+      "learning_rate": 5.333333333333334e-06,
+      "loss": 0.0,
+      "step": 8075
+    },
+    {
+      "epoch": 0.5204614294000129,
+      "grad_norm": 2.987999870269243e-05,
+      "learning_rate": 5.332617257429288e-06,
+      "loss": 0.0,
+      "step": 8076
+    },
+    {
+      "epoch": 0.520525874846942,
+      "grad_norm": 0.00460482292988574,
+      "learning_rate": 5.331901181525242e-06,
+      "loss": 0.0,
+      "step": 8077
+    },
+    {
+      "epoch": 0.5205903202938712,
+      "grad_norm": 0.0010893569519421812,
+      "learning_rate": 5.331185105621196e-06,
+      "loss": 0.0015,
+      "step": 8078
+    },
+    {
+      "epoch": 0.5206547657408004,
+      "grad_norm": 0.0019560104342554994,
+      "learning_rate": 5.33046902971715e-06,
+      "loss": 0.0,
+      "step": 8079
+    },
+    {
+      "epoch": 0.5207192111877296,
+      "grad_norm": 0.00010963162534686023,
+      "learning_rate": 5.329752953813104e-06,
+      "loss": 0.0,
+      "step": 8080
+    },
+    {
+      "epoch": 0.5207836566346588,
+      "grad_norm": 0.0014649577549688212,
+      "learning_rate": 5.329036877909059e-06,
+      "loss": 0.0,
+      "step": 8081
+    },
+    {
+      "epoch": 0.5208481020815879,
+      "grad_norm": 0.0045492230178957975,
+      "learning_rate": 5.328320802005013e-06,
+      "loss": 0.0,
+      "step": 8082
+    },
+    {
+      "epoch": 0.5209125475285171,
+      "grad_norm": 0.0012380836922672056,
+      "learning_rate": 5.327604726100966e-06,
+      "loss": 0.0,
+      "step": 8083
+    },
+    {
+      "epoch": 0.5209769929754463,
+      "grad_norm": 0.006471011870845555,
+      "learning_rate": 5.326888650196921e-06,
+      "loss": 0.0,
+      "step": 8084
+    },
+    {
+      "epoch": 0.5210414384223755,
+      "grad_norm": 0.0273098650513487,
+      "learning_rate": 5.326172574292875e-06,
+      "loss": 0.0001,
+      "step": 8085
+    },
+    {
+      "epoch": 0.5211058838693047,
+      "grad_norm": 0.2248162519329055,
+      "learning_rate": 5.32545649838883e-06,
+      "loss": 0.0005,
+      "step": 8086
+    },
+    {
+      "epoch": 0.5211703293162339,
+      "grad_norm": 0.0010722583432270054,
+      "learning_rate": 5.324740422484784e-06,
+      "loss": 0.0,
+      "step": 8087
+    },
+    {
+      "epoch": 0.5212347747631629,
+      "grad_norm": 0.0011142265211501097,
+      "learning_rate": 5.324024346580739e-06,
+      "loss": 0.0,
+      "step": 8088
+    },
+    {
+      "epoch": 0.5212992202100921,
+      "grad_norm": 0.11535493960735047,
+      "learning_rate": 5.323308270676692e-06,
+      "loss": 0.0003,
+      "step": 8089
+    },
+    {
+      "epoch": 0.5213636656570213,
+      "grad_norm": 0.009380588396358299,
+      "learning_rate": 5.3225921947726464e-06,
+      "loss": 0.0,
+      "step": 8090
+    },
+    {
+      "epoch": 0.5214281111039505,
+      "grad_norm": 0.002236862676547502,
+      "learning_rate": 5.321876118868601e-06,
+      "loss": 0.0,
+      "step": 8091
+    },
+    {
+      "epoch": 0.5214925565508797,
+      "grad_norm": 0.0032438128252277815,
+      "learning_rate": 5.321160042964555e-06,
+      "loss": 0.0,
+      "step": 8092
+    },
+    {
+      "epoch": 0.5215570019978089,
+      "grad_norm": 0.0006600547852672644,
+      "learning_rate": 5.320443967060509e-06,
+      "loss": 0.0,
+      "step": 8093
+    },
+    {
+      "epoch": 0.521621447444738,
+      "grad_norm": 0.00015368027513390677,
+      "learning_rate": 5.319727891156463e-06,
+      "loss": 0.0,
+      "step": 8094
+    },
+    {
+      "epoch": 0.5216858928916672,
+      "grad_norm": 0.02774108911226277,
+      "learning_rate": 5.319011815252417e-06,
+      "loss": 0.0002,
+      "step": 8095
+    },
+    {
+      "epoch": 0.5217503383385964,
+      "grad_norm": 0.0006598481528583075,
+      "learning_rate": 5.318295739348371e-06,
+      "loss": 0.0,
+      "step": 8096
+    },
+    {
+      "epoch": 0.5218147837855256,
+      "grad_norm": 0.0009686613851743172,
+      "learning_rate": 5.317579663444326e-06,
+      "loss": 0.0,
+      "step": 8097
+    },
+    {
+      "epoch": 0.5218792292324548,
+      "grad_norm": 0.0003071446257228951,
+      "learning_rate": 5.31686358754028e-06,
+      "loss": 0.0,
+      "step": 8098
+    },
+    {
+      "epoch": 0.5219436746793839,
+      "grad_norm": 0.0007522031735325496,
+      "learning_rate": 5.3161475116362334e-06,
+      "loss": 0.0,
+      "step": 8099
+    },
+    {
+      "epoch": 0.522008120126313,
+      "grad_norm": 0.00565657191007196,
+      "learning_rate": 5.315431435732188e-06,
+      "loss": 0.0,
+      "step": 8100
+    },
+    {
+      "epoch": 0.5220725655732422,
+      "grad_norm": 0.011783286177244336,
+      "learning_rate": 5.314715359828142e-06,
+      "loss": 0.0002,
+      "step": 8101
+    },
+    {
+      "epoch": 0.5221370110201714,
+      "grad_norm": 0.0004662790406560697,
+      "learning_rate": 5.313999283924096e-06,
+      "loss": 0.0,
+      "step": 8102
+    },
+    {
+      "epoch": 0.5222014564671006,
+      "grad_norm": 0.001986001404947262,
+      "learning_rate": 5.313283208020051e-06,
+      "loss": 0.0,
+      "step": 8103
+    },
+    {
+      "epoch": 0.5222659019140298,
+      "grad_norm": 0.0005110061619616537,
+      "learning_rate": 5.312567132116004e-06,
+      "loss": 0.0,
+      "step": 8104
+    },
+    {
+      "epoch": 0.522330347360959,
+      "grad_norm": 6.456255465957766e-05,
+      "learning_rate": 5.311851056211958e-06,
+      "loss": 0.0,
+      "step": 8105
+    },
+    {
+      "epoch": 0.5223947928078881,
+      "grad_norm": 0.003485680539695944,
+      "learning_rate": 5.311134980307913e-06,
+      "loss": 0.0,
+      "step": 8106
+    },
+    {
+      "epoch": 0.5224592382548173,
+      "grad_norm": 0.027200001693703183,
+      "learning_rate": 5.310418904403867e-06,
+      "loss": 0.0003,
+      "step": 8107
+    },
+    {
+      "epoch": 0.5225236837017465,
+      "grad_norm": 0.00010678184588055318,
+      "learning_rate": 5.309702828499821e-06,
+      "loss": 0.0,
+      "step": 8108
+    },
+    {
+      "epoch": 0.5225881291486757,
+      "grad_norm": 0.009661019632250483,
+      "learning_rate": 5.3089867525957764e-06,
+      "loss": 0.0,
+      "step": 8109
+    },
+    {
+      "epoch": 0.5226525745956048,
+      "grad_norm": 0.16898934666960477,
+      "learning_rate": 5.30827067669173e-06,
+      "loss": 0.0021,
+      "step": 8110
+    },
+    {
+      "epoch": 0.522717020042534,
+      "grad_norm": 0.648830777139313,
+      "learning_rate": 5.307554600787684e-06,
+      "loss": 0.0027,
+      "step": 8111
+    },
+    {
+      "epoch": 0.5227814654894631,
+      "grad_norm": 2.3984561151492033e-05,
+      "learning_rate": 5.3068385248836385e-06,
+      "loss": 0.0,
+      "step": 8112
+    },
+    {
+      "epoch": 0.5228459109363923,
+      "grad_norm": 0.1974333531613506,
+      "learning_rate": 5.306122448979593e-06,
+      "loss": 0.0011,
+      "step": 8113
+    },
+    {
+      "epoch": 0.5229103563833215,
+      "grad_norm": 0.00015204341681974146,
+      "learning_rate": 5.305406373075547e-06,
+      "loss": 0.0,
+      "step": 8114
+    },
+    {
+      "epoch": 0.5229748018302507,
+      "grad_norm": 0.0036347013021713263,
+      "learning_rate": 5.3046902971715005e-06,
+      "loss": 0.0,
+      "step": 8115
+    },
+    {
+      "epoch": 0.5230392472771799,
+      "grad_norm": 0.0037800083078743346,
+      "learning_rate": 5.303974221267455e-06,
+      "loss": 0.0,
+      "step": 8116
+    },
+    {
+      "epoch": 0.5231036927241091,
+      "grad_norm": 0.0013677929468042499,
+      "learning_rate": 5.303258145363409e-06,
+      "loss": 0.0,
+      "step": 8117
+    },
+    {
+      "epoch": 0.5231681381710382,
+      "grad_norm": 0.024809124082378647,
+      "learning_rate": 5.302542069459363e-06,
+      "loss": 0.0002,
+      "step": 8118
+    },
+    {
+      "epoch": 0.5232325836179674,
+      "grad_norm": 0.017704296459074436,
+      "learning_rate": 5.301825993555318e-06,
+      "loss": 0.0001,
+      "step": 8119
+    },
+    {
+      "epoch": 0.5232970290648966,
+      "grad_norm": 0.008421566509266936,
+      "learning_rate": 5.301109917651271e-06,
+      "loss": 0.0,
+      "step": 8120
+    },
+    {
+      "epoch": 0.5233614745118257,
+      "grad_norm": 0.00030410408926729963,
+      "learning_rate": 5.3003938417472255e-06,
+      "loss": 0.0,
+      "step": 8121
+    },
+    {
+      "epoch": 0.5234259199587549,
+      "grad_norm": 0.031790182885038674,
+      "learning_rate": 5.29967776584318e-06,
+      "loss": 0.0,
+      "step": 8122
+    },
+    {
+      "epoch": 0.523490365405684,
+      "grad_norm": 0.0006871531393244011,
+      "learning_rate": 5.298961689939134e-06,
+      "loss": 0.0,
+      "step": 8123
+    },
+    {
+      "epoch": 0.5235548108526132,
+      "grad_norm": 0.004459484505413135,
+      "learning_rate": 5.298245614035088e-06,
+      "loss": 0.0,
+      "step": 8124
+    },
+    {
+      "epoch": 0.5236192562995424,
+      "grad_norm": 0.0018307072091210933,
+      "learning_rate": 5.297529538131042e-06,
+      "loss": 0.0,
+      "step": 8125
+    },
+    {
+      "epoch": 0.5236837017464716,
+      "grad_norm": 0.0038162593027634127,
+      "learning_rate": 5.296813462226996e-06,
+      "loss": 0.0,
+      "step": 8126
+    },
+    {
+      "epoch": 0.5237481471934008,
+      "grad_norm": 7.791748082360972e-05,
+      "learning_rate": 5.29609738632295e-06,
+      "loss": 0.0,
+      "step": 8127
+    },
+    {
+      "epoch": 0.52381259264033,
+      "grad_norm": 0.0016177164245403997,
+      "learning_rate": 5.295381310418905e-06,
+      "loss": 0.0,
+      "step": 8128
+    },
+    {
+      "epoch": 0.5238770380872592,
+      "grad_norm": 0.018440304189078296,
+      "learning_rate": 5.294665234514859e-06,
+      "loss": 0.0001,
+      "step": 8129
+    },
+    {
+      "epoch": 0.5239414835341883,
+      "grad_norm": 0.00045727140499551784,
+      "learning_rate": 5.2939491586108125e-06,
+      "loss": 0.0,
+      "step": 8130
+    },
+    {
+      "epoch": 0.5240059289811175,
+      "grad_norm": 0.001961379100123223,
+      "learning_rate": 5.293233082706767e-06,
+      "loss": 0.0,
+      "step": 8131
+    },
+    {
+      "epoch": 0.5240703744280466,
+      "grad_norm": 0.0017229937831158926,
+      "learning_rate": 5.292517006802722e-06,
+      "loss": 0.0,
+      "step": 8132
+    },
+    {
+      "epoch": 0.5241348198749758,
+      "grad_norm": 0.0035674278089422567,
+      "learning_rate": 5.291800930898676e-06,
+      "loss": 0.0,
+      "step": 8133
+    },
+    {
+      "epoch": 0.524199265321905,
+      "grad_norm": 0.05173934526094626,
+      "learning_rate": 5.2910848549946305e-06,
+      "loss": 0.0003,
+      "step": 8134
+    },
+    {
+      "epoch": 0.5242637107688342,
+      "grad_norm": 0.0022714186955937123,
+      "learning_rate": 5.290368779090585e-06,
+      "loss": 0.0,
+      "step": 8135
+    },
+    {
+      "epoch": 0.5243281562157633,
+      "grad_norm": 0.04544742654464294,
+      "learning_rate": 5.289652703186538e-06,
+      "loss": 0.0001,
+      "step": 8136
+    },
+    {
+      "epoch": 0.5243926016626925,
+      "grad_norm": 0.0011263308024880082,
+      "learning_rate": 5.2889366272824926e-06,
+      "loss": 0.0,
+      "step": 8137
+    },
+    {
+      "epoch": 0.5244570471096217,
+      "grad_norm": 0.0005061771864201109,
+      "learning_rate": 5.288220551378447e-06,
+      "loss": 0.0,
+      "step": 8138
+    },
+    {
+      "epoch": 0.5245214925565509,
+      "grad_norm": 0.0009123932009742532,
+      "learning_rate": 5.287504475474401e-06,
+      "loss": 0.0,
+      "step": 8139
+    },
+    {
+      "epoch": 0.5245859380034801,
+      "grad_norm": 0.001865122440905928,
+      "learning_rate": 5.2867883995703554e-06,
+      "loss": 0.0,
+      "step": 8140
+    },
+    {
+      "epoch": 0.5246503834504093,
+      "grad_norm": 0.005351867999890206,
+      "learning_rate": 5.286072323666309e-06,
+      "loss": 0.0,
+      "step": 8141
+    },
+    {
+      "epoch": 0.5247148288973384,
+      "grad_norm": 0.0012923710822036933,
+      "learning_rate": 5.285356247762263e-06,
+      "loss": 0.0,
+      "step": 8142
+    },
+    {
+      "epoch": 0.5247792743442676,
+      "grad_norm": 0.07148435050672101,
+      "learning_rate": 5.2846401718582175e-06,
+      "loss": 0.0002,
+      "step": 8143
+    },
+    {
+      "epoch": 0.5248437197911967,
+      "grad_norm": 0.0003850860287028189,
+      "learning_rate": 5.283924095954172e-06,
+      "loss": 0.0,
+      "step": 8144
+    },
+    {
+      "epoch": 0.5249081652381259,
+      "grad_norm": 0.29063587886178854,
+      "learning_rate": 5.283208020050126e-06,
+      "loss": 0.0009,
+      "step": 8145
+    },
+    {
+      "epoch": 0.5249726106850551,
+      "grad_norm": 0.0005251628566615367,
+      "learning_rate": 5.2824919441460795e-06,
+      "loss": 0.0,
+      "step": 8146
+    },
+    {
+      "epoch": 0.5250370561319843,
+      "grad_norm": 0.0034821841120536135,
+      "learning_rate": 5.281775868242034e-06,
+      "loss": 0.0001,
+      "step": 8147
+    },
+    {
+      "epoch": 0.5251015015789134,
+      "grad_norm": 0.030426072261242363,
+      "learning_rate": 5.281059792337988e-06,
+      "loss": 0.0,
+      "step": 8148
+    },
+    {
+      "epoch": 0.5251659470258426,
+      "grad_norm": 0.05582442107006137,
+      "learning_rate": 5.2803437164339424e-06,
+      "loss": 0.0001,
+      "step": 8149
+    },
+    {
+      "epoch": 0.5252303924727718,
+      "grad_norm": 0.009250913006427195,
+      "learning_rate": 5.279627640529896e-06,
+      "loss": 0.0,
+      "step": 8150
+    },
+    {
+      "epoch": 0.525294837919701,
+      "grad_norm": 0.0034705167538564173,
+      "learning_rate": 5.27891156462585e-06,
+      "loss": 0.0,
+      "step": 8151
+    },
+    {
+      "epoch": 0.5253592833666302,
+      "grad_norm": 0.0016301703975253937,
+      "learning_rate": 5.2781954887218045e-06,
+      "loss": 0.0,
+      "step": 8152
+    },
+    {
+      "epoch": 0.5254237288135594,
+      "grad_norm": 0.004285375626315951,
+      "learning_rate": 5.277479412817759e-06,
+      "loss": 0.0,
+      "step": 8153
+    },
+    {
+      "epoch": 0.5254881742604885,
+      "grad_norm": 0.0001443931223535041,
+      "learning_rate": 5.276763336913713e-06,
+      "loss": 0.0,
+      "step": 8154
+    },
+    {
+      "epoch": 0.5255526197074176,
+      "grad_norm": 0.0012174254517221168,
+      "learning_rate": 5.2760472610096665e-06,
+      "loss": 0.0,
+      "step": 8155
+    },
+    {
+      "epoch": 0.5256170651543468,
+      "grad_norm": 0.1638833555172645,
+      "learning_rate": 5.2753311851056225e-06,
+      "loss": 0.0018,
+      "step": 8156
+    },
+    {
+      "epoch": 0.525681510601276,
+      "grad_norm": 0.011755288426830504,
+      "learning_rate": 5.274615109201576e-06,
+      "loss": 0.0,
+      "step": 8157
+    },
+    {
+      "epoch": 0.5257459560482052,
+      "grad_norm": 0.0003608457318295168,
+      "learning_rate": 5.27389903329753e-06,
+      "loss": 0.0,
+      "step": 8158
+    },
+    {
+      "epoch": 0.5258104014951344,
+      "grad_norm": 0.0014071326382772225,
+      "learning_rate": 5.273182957393485e-06,
+      "loss": 0.0,
+      "step": 8159
+    },
+    {
+      "epoch": 0.5258748469420635,
+      "grad_norm": 0.0002052865912161992,
+      "learning_rate": 5.272466881489439e-06,
+      "loss": 0.0,
+      "step": 8160
+    },
+    {
+      "epoch": 0.5259392923889927,
+      "grad_norm": 0.11002314063391534,
+      "learning_rate": 5.271750805585393e-06,
+      "loss": 0.0002,
+      "step": 8161
+    },
+    {
+      "epoch": 0.5260037378359219,
+      "grad_norm": 0.1543719917652916,
+      "learning_rate": 5.271034729681347e-06,
+      "loss": 0.0016,
+      "step": 8162
+    },
+    {
+      "epoch": 0.5260681832828511,
+      "grad_norm": 0.0013602805663139417,
+      "learning_rate": 5.270318653777301e-06,
+      "loss": 0.0,
+      "step": 8163
+    },
+    {
+      "epoch": 0.5261326287297803,
+      "grad_norm": 0.028567334380291513,
+      "learning_rate": 5.269602577873255e-06,
+      "loss": 0.0,
+      "step": 8164
+    },
+    {
+      "epoch": 0.5261970741767095,
+      "grad_norm": 0.030342511622000475,
+      "learning_rate": 5.2688865019692095e-06,
+      "loss": 0.0,
+      "step": 8165
+    },
+    {
+      "epoch": 0.5262615196236385,
+      "grad_norm": 0.0024130707987652417,
+      "learning_rate": 5.268170426065163e-06,
+      "loss": 0.0,
+      "step": 8166
+    },
+    {
+      "epoch": 0.5263259650705677,
+      "grad_norm": 0.0004052014762314221,
+      "learning_rate": 5.267454350161117e-06,
+      "loss": 0.0,
+      "step": 8167
+    },
+    {
+      "epoch": 0.5263904105174969,
+      "grad_norm": 0.000613029688713524,
+      "learning_rate": 5.266738274257072e-06,
+      "loss": 0.0,
+      "step": 8168
+    },
+    {
+      "epoch": 0.5264548559644261,
+      "grad_norm": 0.007337527303468706,
+      "learning_rate": 5.266022198353026e-06,
+      "loss": 0.0,
+      "step": 8169
+    },
+    {
+      "epoch": 0.5265193014113553,
+      "grad_norm": 0.007629737369463775,
+      "learning_rate": 5.26530612244898e-06,
+      "loss": 0.0,
+      "step": 8170
+    },
+    {
+      "epoch": 0.5265837468582845,
+      "grad_norm": 0.019203614179758664,
+      "learning_rate": 5.264590046544934e-06,
+      "loss": 0.0,
+      "step": 8171
+    },
+    {
+      "epoch": 0.5266481923052136,
+      "grad_norm": 0.009774924112548477,
+      "learning_rate": 5.263873970640888e-06,
+      "loss": 0.0,
+      "step": 8172
+    },
+    {
+      "epoch": 0.5267126377521428,
+      "grad_norm": 0.13492138472105183,
+      "learning_rate": 5.263157894736842e-06,
+      "loss": 0.0001,
+      "step": 8173
+    },
+    {
+      "epoch": 0.526777083199072,
+      "grad_norm": 0.17494896310555966,
+      "learning_rate": 5.2624418188327965e-06,
+      "loss": 0.0026,
+      "step": 8174
+    },
+    {
+      "epoch": 0.5268415286460012,
+      "grad_norm": 0.0030930372243162135,
+      "learning_rate": 5.261725742928751e-06,
+      "loss": 0.0,
+      "step": 8175
+    },
+    {
+      "epoch": 0.5269059740929304,
+      "grad_norm": 0.00018265507632032822,
+      "learning_rate": 5.261009667024704e-06,
+      "loss": 0.0,
+      "step": 8176
+    },
+    {
+      "epoch": 0.5269704195398595,
+      "grad_norm": 0.005233237911499148,
+      "learning_rate": 5.2602935911206586e-06,
+      "loss": 0.0,
+      "step": 8177
+    },
+    {
+      "epoch": 0.5270348649867886,
+      "grad_norm": 0.005108504047076498,
+      "learning_rate": 5.259577515216613e-06,
+      "loss": 0.0,
+      "step": 8178
+    },
+    {
+      "epoch": 0.5270993104337178,
+      "grad_norm": 0.001656930633442291,
+      "learning_rate": 5.258861439312568e-06,
+      "loss": 0.0,
+      "step": 8179
+    },
+    {
+      "epoch": 0.527163755880647,
+      "grad_norm": 0.0015388414270932735,
+      "learning_rate": 5.258145363408522e-06,
+      "loss": 0.0,
+      "step": 8180
+    },
+    {
+      "epoch": 0.5272282013275762,
+      "grad_norm": 0.001493593354986821,
+      "learning_rate": 5.257429287504477e-06,
+      "loss": 0.0,
+      "step": 8181
+    },
+    {
+      "epoch": 0.5272926467745054,
+      "grad_norm": 0.0015624622198609356,
+      "learning_rate": 5.25671321160043e-06,
+      "loss": 0.0,
+      "step": 8182
+    },
+    {
+      "epoch": 0.5273570922214346,
+      "grad_norm": 0.00023563587226363962,
+      "learning_rate": 5.255997135696384e-06,
+      "loss": 0.0,
+      "step": 8183
+    },
+    {
+      "epoch": 0.5274215376683637,
+      "grad_norm": 0.0008716355655753914,
+      "learning_rate": 5.255281059792339e-06,
+      "loss": 0.0,
+      "step": 8184
+    },
+    {
+      "epoch": 0.5274859831152929,
+      "grad_norm": 0.00041412078705236374,
+      "learning_rate": 5.254564983888293e-06,
+      "loss": 0.0,
+      "step": 8185
+    },
+    {
+      "epoch": 0.5275504285622221,
+      "grad_norm": 0.02280689810042121,
+      "learning_rate": 5.253848907984247e-06,
+      "loss": 0.0001,
+      "step": 8186
+    },
+    {
+      "epoch": 0.5276148740091513,
+      "grad_norm": 0.0002442367796725076,
+      "learning_rate": 5.253132832080201e-06,
+      "loss": 0.0,
+      "step": 8187
+    },
+    {
+      "epoch": 0.5276793194560804,
+      "grad_norm": 0.0001824094471575746,
+      "learning_rate": 5.252416756176155e-06,
+      "loss": 0.0,
+      "step": 8188
+    },
+    {
+      "epoch": 0.5277437649030096,
+      "grad_norm": 0.10367689975011432,
+      "learning_rate": 5.251700680272109e-06,
+      "loss": 0.0006,
+      "step": 8189
+    },
+    {
+      "epoch": 0.5278082103499387,
+      "grad_norm": 0.04179084686455239,
+      "learning_rate": 5.250984604368064e-06,
+      "loss": 0.0003,
+      "step": 8190
+    },
+    {
+      "epoch": 0.5278726557968679,
+      "grad_norm": 0.054257698556740556,
+      "learning_rate": 5.250268528464018e-06,
+      "loss": 0.0002,
+      "step": 8191
+    },
+    {
+      "epoch": 0.5279371012437971,
+      "grad_norm": 0.00045334327345424616,
+      "learning_rate": 5.249552452559971e-06,
+      "loss": 0.0,
+      "step": 8192
+    },
+    {
+      "epoch": 0.5280015466907263,
+      "grad_norm": 0.002689039349698105,
+      "learning_rate": 5.248836376655926e-06,
+      "loss": 0.0,
+      "step": 8193
+    },
+    {
+      "epoch": 0.5280659921376555,
+      "grad_norm": 0.003820872434692222,
+      "learning_rate": 5.24812030075188e-06,
+      "loss": 0.0,
+      "step": 8194
+    },
+    {
+      "epoch": 0.5281304375845847,
+      "grad_norm": 0.0031079245168158607,
+      "learning_rate": 5.247404224847834e-06,
+      "loss": 0.0,
+      "step": 8195
+    },
+    {
+      "epoch": 0.5281948830315139,
+      "grad_norm": 0.008044136523780283,
+      "learning_rate": 5.2466881489437886e-06,
+      "loss": 0.0,
+      "step": 8196
+    },
+    {
+      "epoch": 0.528259328478443,
+      "grad_norm": 0.0013573262889266815,
+      "learning_rate": 5.245972073039742e-06,
+      "loss": 0.0,
+      "step": 8197
+    },
+    {
+      "epoch": 0.5283237739253722,
+      "grad_norm": 0.00013557121331184024,
+      "learning_rate": 5.245255997135696e-06,
+      "loss": 0.0,
+      "step": 8198
+    },
+    {
+      "epoch": 0.5283882193723013,
+      "grad_norm": 0.006549067277275383,
+      "learning_rate": 5.244539921231651e-06,
+      "loss": 0.0,
+      "step": 8199
+    },
+    {
+      "epoch": 0.5284526648192305,
+      "grad_norm": 0.00038068579052549697,
+      "learning_rate": 5.243823845327605e-06,
+      "loss": 0.0,
+      "step": 8200
+    },
+    {
+      "epoch": 0.5285171102661597,
+      "grad_norm": 0.001630613111818217,
+      "learning_rate": 5.243107769423559e-06,
+      "loss": 0.0,
+      "step": 8201
+    },
+    {
+      "epoch": 0.5285815557130888,
+      "grad_norm": 0.005277173181096403,
+      "learning_rate": 5.242391693519514e-06,
+      "loss": 0.0,
+      "step": 8202
+    },
+    {
+      "epoch": 0.528646001160018,
+      "grad_norm": 0.0022489153954235027,
+      "learning_rate": 5.241675617615468e-06,
+      "loss": 0.0,
+      "step": 8203
+    },
+    {
+      "epoch": 0.5287104466069472,
+      "grad_norm": 0.0007811213053585651,
+      "learning_rate": 5.240959541711422e-06,
+      "loss": 0.0,
+      "step": 8204
+    },
+    {
+      "epoch": 0.5287748920538764,
+      "grad_norm": 0.15758120777290088,
+      "learning_rate": 5.240243465807376e-06,
+      "loss": 0.0008,
+      "step": 8205
+    },
+    {
+      "epoch": 0.5288393375008056,
+      "grad_norm": 0.226937066591774,
+      "learning_rate": 5.239527389903331e-06,
+      "loss": 0.001,
+      "step": 8206
+    },
+    {
+      "epoch": 0.5289037829477348,
+      "grad_norm": 0.0004585199950160562,
+      "learning_rate": 5.238811313999285e-06,
+      "loss": 0.0,
+      "step": 8207
+    },
+    {
+      "epoch": 0.528968228394664,
+      "grad_norm": 0.007970610709894329,
+      "learning_rate": 5.2380952380952384e-06,
+      "loss": 0.0,
+      "step": 8208
+    },
+    {
+      "epoch": 0.5290326738415931,
+      "grad_norm": 0.0008556513303001802,
+      "learning_rate": 5.237379162191193e-06,
+      "loss": 0.0,
+      "step": 8209
+    },
+    {
+      "epoch": 0.5290971192885222,
+      "grad_norm": 0.0013647344791251014,
+      "learning_rate": 5.236663086287147e-06,
+      "loss": 0.0,
+      "step": 8210
+    },
+    {
+      "epoch": 0.5291615647354514,
+      "grad_norm": 0.0008613043011201751,
+      "learning_rate": 5.235947010383101e-06,
+      "loss": 0.0,
+      "step": 8211
+    },
+    {
+      "epoch": 0.5292260101823806,
+      "grad_norm": 0.0011533930435283846,
+      "learning_rate": 5.235230934479056e-06,
+      "loss": 0.0,
+      "step": 8212
+    },
+    {
+      "epoch": 0.5292904556293098,
+      "grad_norm": 0.004177880975011757,
+      "learning_rate": 5.234514858575009e-06,
+      "loss": 0.0,
+      "step": 8213
+    },
+    {
+      "epoch": 0.529354901076239,
+      "grad_norm": 8.555679965205073e-05,
+      "learning_rate": 5.233798782670963e-06,
+      "loss": 0.0,
+      "step": 8214
+    },
+    {
+      "epoch": 0.5294193465231681,
+      "grad_norm": 0.06403724198652394,
+      "learning_rate": 5.233082706766918e-06,
+      "loss": 0.0002,
+      "step": 8215
+    },
+    {
+      "epoch": 0.5294837919700973,
+      "grad_norm": 0.0004445335666127929,
+      "learning_rate": 5.232366630862872e-06,
+      "loss": 0.0,
+      "step": 8216
+    },
+    {
+      "epoch": 0.5295482374170265,
+      "grad_norm": 0.0018955943196318668,
+      "learning_rate": 5.231650554958826e-06,
+      "loss": 0.0,
+      "step": 8217
+    },
+    {
+      "epoch": 0.5296126828639557,
+      "grad_norm": 0.00044187092058860976,
+      "learning_rate": 5.23093447905478e-06,
+      "loss": 0.0,
+      "step": 8218
+    },
+    {
+      "epoch": 0.5296771283108849,
+      "grad_norm": 0.015893673273748708,
+      "learning_rate": 5.230218403150734e-06,
+      "loss": 0.0001,
+      "step": 8219
+    },
+    {
+      "epoch": 0.529741573757814,
+      "grad_norm": 0.0035433591799342505,
+      "learning_rate": 5.229502327246688e-06,
+      "loss": 0.0,
+      "step": 8220
+    },
+    {
+      "epoch": 0.5298060192047432,
+      "grad_norm": 0.0010208689897191162,
+      "learning_rate": 5.228786251342643e-06,
+      "loss": 0.0,
+      "step": 8221
+    },
+    {
+      "epoch": 0.5298704646516723,
+      "grad_norm": 7.476281600967891e-05,
+      "learning_rate": 5.228070175438597e-06,
+      "loss": 0.0,
+      "step": 8222
+    },
+    {
+      "epoch": 0.5299349100986015,
+      "grad_norm": 0.03769369368956137,
+      "learning_rate": 5.22735409953455e-06,
+      "loss": 0.0002,
+      "step": 8223
+    },
+    {
+      "epoch": 0.5299993555455307,
+      "grad_norm": 0.5087324466067144,
+      "learning_rate": 5.226638023630505e-06,
+      "loss": 0.001,
+      "step": 8224
+    },
+    {
+      "epoch": 0.5300638009924599,
+      "grad_norm": 0.0004116131397795408,
+      "learning_rate": 5.225921947726459e-06,
+      "loss": 0.0,
+      "step": 8225
+    },
+    {
+      "epoch": 0.530128246439389,
+      "grad_norm": 0.12127374146854736,
+      "learning_rate": 5.225205871822414e-06,
+      "loss": 0.0003,
+      "step": 8226
+    },
+    {
+      "epoch": 0.5301926918863182,
+      "grad_norm": 0.43693211006198096,
+      "learning_rate": 5.2244897959183684e-06,
+      "loss": 0.0006,
+      "step": 8227
+    },
+    {
+      "epoch": 0.5302571373332474,
+      "grad_norm": 0.0016617944626236237,
+      "learning_rate": 5.223773720014323e-06,
+      "loss": 0.0,
+      "step": 8228
+    },
+    {
+      "epoch": 0.5303215827801766,
+      "grad_norm": 0.274345728826985,
+      "learning_rate": 5.223057644110276e-06,
+      "loss": 0.0006,
+      "step": 8229
+    },
+    {
+      "epoch": 0.5303860282271058,
+      "grad_norm": 0.005449729470843885,
+      "learning_rate": 5.2223415682062305e-06,
+      "loss": 0.0,
+      "step": 8230
+    },
+    {
+      "epoch": 0.530450473674035,
+      "grad_norm": 0.0013428693602989992,
+      "learning_rate": 5.221625492302185e-06,
+      "loss": 0.0,
+      "step": 8231
+    },
+    {
+      "epoch": 0.5305149191209642,
+      "grad_norm": 0.025936283526849124,
+      "learning_rate": 5.220909416398139e-06,
+      "loss": 0.0,
+      "step": 8232
+    },
+    {
+      "epoch": 0.5305793645678932,
+      "grad_norm": 0.015376020230364216,
+      "learning_rate": 5.220193340494093e-06,
+      "loss": 0.0001,
+      "step": 8233
+    },
+    {
+      "epoch": 0.5306438100148224,
+      "grad_norm": 0.00035719175370790775,
+      "learning_rate": 5.219477264590047e-06,
+      "loss": 0.0,
+      "step": 8234
+    },
+    {
+      "epoch": 0.5307082554617516,
+      "grad_norm": 0.6565876046945677,
+      "learning_rate": 5.218761188686001e-06,
+      "loss": 0.005,
+      "step": 8235
+    },
+    {
+      "epoch": 0.5307727009086808,
+      "grad_norm": 0.011372042509931924,
+      "learning_rate": 5.218045112781955e-06,
+      "loss": 0.0,
+      "step": 8236
+    },
+    {
+      "epoch": 0.53083714635561,
+      "grad_norm": 0.002884196527175304,
+      "learning_rate": 5.21732903687791e-06,
+      "loss": 0.0,
+      "step": 8237
+    },
+    {
+      "epoch": 0.5309015918025392,
+      "grad_norm": 0.0003135085556914037,
+      "learning_rate": 5.216612960973864e-06,
+      "loss": 0.0,
+      "step": 8238
+    },
+    {
+      "epoch": 0.5309660372494683,
+      "grad_norm": 0.01590281361846434,
+      "learning_rate": 5.2158968850698175e-06,
+      "loss": 0.0002,
+      "step": 8239
+    },
+    {
+      "epoch": 0.5310304826963975,
+      "grad_norm": 0.011128835166307135,
+      "learning_rate": 5.215180809165772e-06,
+      "loss": 0.0,
+      "step": 8240
+    },
+    {
+      "epoch": 0.5310949281433267,
+      "grad_norm": 0.08628985929142845,
+      "learning_rate": 5.214464733261726e-06,
+      "loss": 0.0004,
+      "step": 8241
+    },
+    {
+      "epoch": 0.5311593735902559,
+      "grad_norm": 0.0002728034723986173,
+      "learning_rate": 5.21374865735768e-06,
+      "loss": 0.0,
+      "step": 8242
+    },
+    {
+      "epoch": 0.5312238190371851,
+      "grad_norm": 0.0019110246639428413,
+      "learning_rate": 5.213032581453634e-06,
+      "loss": 0.0,
+      "step": 8243
+    },
+    {
+      "epoch": 0.5312882644841141,
+      "grad_norm": 0.12502401836431717,
+      "learning_rate": 5.212316505549588e-06,
+      "loss": 0.0056,
+      "step": 8244
+    },
+    {
+      "epoch": 0.5313527099310433,
+      "grad_norm": 0.0008021568203203074,
+      "learning_rate": 5.211600429645542e-06,
+      "loss": 0.0,
+      "step": 8245
+    },
+    {
+      "epoch": 0.5314171553779725,
+      "grad_norm": 0.012489563526513548,
+      "learning_rate": 5.210884353741497e-06,
+      "loss": 0.0,
+      "step": 8246
+    },
+    {
+      "epoch": 0.5314816008249017,
+      "grad_norm": 0.566265930684311,
+      "learning_rate": 5.210168277837451e-06,
+      "loss": 0.002,
+      "step": 8247
+    },
+    {
+      "epoch": 0.5315460462718309,
+      "grad_norm": 0.16108968909072638,
+      "learning_rate": 5.2094522019334045e-06,
+      "loss": 0.0002,
+      "step": 8248
+    },
+    {
+      "epoch": 0.5316104917187601,
+      "grad_norm": 0.0007321626886337238,
+      "learning_rate": 5.2087361260293605e-06,
+      "loss": 0.0,
+      "step": 8249
+    },
+    {
+      "epoch": 0.5316749371656893,
+      "grad_norm": 0.10953813121065938,
+      "learning_rate": 5.208020050125314e-06,
+      "loss": 0.0001,
+      "step": 8250
+    },
+    {
+      "epoch": 0.5317393826126184,
+      "grad_norm": 0.007629119868102849,
+      "learning_rate": 5.207303974221268e-06,
+      "loss": 0.0,
+      "step": 8251
+    },
+    {
+      "epoch": 0.5318038280595476,
+      "grad_norm": 0.0022120117547238275,
+      "learning_rate": 5.2065878983172225e-06,
+      "loss": 0.0,
+      "step": 8252
+    },
+    {
+      "epoch": 0.5318682735064768,
+      "grad_norm": 9.754262784845069e-05,
+      "learning_rate": 5.205871822413177e-06,
+      "loss": 0.0,
+      "step": 8253
+    },
+    {
+      "epoch": 0.531932718953406,
+      "grad_norm": 0.03304436715002747,
+      "learning_rate": 5.205155746509131e-06,
+      "loss": 0.0,
+      "step": 8254
+    },
+    {
+      "epoch": 0.5319971644003351,
+      "grad_norm": 0.002980140074258487,
+      "learning_rate": 5.2044396706050846e-06,
+      "loss": 0.0,
+      "step": 8255
+    },
+    {
+      "epoch": 0.5320616098472642,
+      "grad_norm": 0.00030013782604176344,
+      "learning_rate": 5.203723594701039e-06,
+      "loss": 0.0,
+      "step": 8256
+    },
+    {
+      "epoch": 0.5321260552941934,
+      "grad_norm": 0.0005443334783101526,
+      "learning_rate": 5.203007518796993e-06,
+      "loss": 0.0,
+      "step": 8257
+    },
+    {
+      "epoch": 0.5321905007411226,
+      "grad_norm": 0.03163681132386296,
+      "learning_rate": 5.2022914428929474e-06,
+      "loss": 0.0001,
+      "step": 8258
+    },
+    {
+      "epoch": 0.5322549461880518,
+      "grad_norm": 0.04303355504829882,
+      "learning_rate": 5.201575366988901e-06,
+      "loss": 0.0001,
+      "step": 8259
+    },
+    {
+      "epoch": 0.532319391634981,
+      "grad_norm": 0.17966033896124106,
+      "learning_rate": 5.200859291084855e-06,
+      "loss": 0.0001,
+      "step": 8260
+    },
+    {
+      "epoch": 0.5323838370819102,
+      "grad_norm": 0.001579268255010264,
+      "learning_rate": 5.2001432151808095e-06,
+      "loss": 0.0,
+      "step": 8261
+    },
+    {
+      "epoch": 0.5324482825288394,
+      "grad_norm": 0.0009424939141334193,
+      "learning_rate": 5.199427139276764e-06,
+      "loss": 0.0,
+      "step": 8262
+    },
+    {
+      "epoch": 0.5325127279757685,
+      "grad_norm": 0.0009968826784912033,
+      "learning_rate": 5.198711063372718e-06,
+      "loss": 0.0,
+      "step": 8263
+    },
+    {
+      "epoch": 0.5325771734226977,
+      "grad_norm": 0.008419342363879062,
+      "learning_rate": 5.1979949874686715e-06,
+      "loss": 0.0001,
+      "step": 8264
+    },
+    {
+      "epoch": 0.5326416188696269,
+      "grad_norm": 0.003526839644485888,
+      "learning_rate": 5.197278911564626e-06,
+      "loss": 0.0,
+      "step": 8265
+    },
+    {
+      "epoch": 0.532706064316556,
+      "grad_norm": 0.0004590648824850025,
+      "learning_rate": 5.19656283566058e-06,
+      "loss": 0.0,
+      "step": 8266
+    },
+    {
+      "epoch": 0.5327705097634852,
+      "grad_norm": 0.3312528124276014,
+      "learning_rate": 5.1958467597565344e-06,
+      "loss": 0.0009,
+      "step": 8267
+    },
+    {
+      "epoch": 0.5328349552104144,
+      "grad_norm": 0.018709853089610762,
+      "learning_rate": 5.195130683852489e-06,
+      "loss": 0.0001,
+      "step": 8268
+    },
+    {
+      "epoch": 0.5328994006573435,
+      "grad_norm": 0.014009747720029147,
+      "learning_rate": 5.194414607948442e-06,
+      "loss": 0.0,
+      "step": 8269
+    },
+    {
+      "epoch": 0.5329638461042727,
+      "grad_norm": 0.0016731185805280674,
+      "learning_rate": 5.1936985320443965e-06,
+      "loss": 0.0,
+      "step": 8270
+    },
+    {
+      "epoch": 0.5330282915512019,
+      "grad_norm": 0.0074413456102401035,
+      "learning_rate": 5.192982456140351e-06,
+      "loss": 0.0,
+      "step": 8271
+    },
+    {
+      "epoch": 0.5330927369981311,
+      "grad_norm": 0.2735715265142258,
+      "learning_rate": 5.192266380236306e-06,
+      "loss": 0.0021,
+      "step": 8272
+    },
+    {
+      "epoch": 0.5331571824450603,
+      "grad_norm": 0.07160542783589822,
+      "learning_rate": 5.19155030433226e-06,
+      "loss": 0.0008,
+      "step": 8273
+    },
+    {
+      "epoch": 0.5332216278919895,
+      "grad_norm": 0.00013488290769220058,
+      "learning_rate": 5.1908342284282145e-06,
+      "loss": 0.0,
+      "step": 8274
+    },
+    {
+      "epoch": 0.5332860733389186,
+      "grad_norm": 2.876226943198003e-05,
+      "learning_rate": 5.190118152524168e-06,
+      "loss": 0.0,
+      "step": 8275
+    },
+    {
+      "epoch": 0.5333505187858478,
+      "grad_norm": 0.0012743140065523632,
+      "learning_rate": 5.189402076620122e-06,
+      "loss": 0.0,
+      "step": 8276
+    },
+    {
+      "epoch": 0.5334149642327769,
+      "grad_norm": 0.004809027917468691,
+      "learning_rate": 5.188686000716077e-06,
+      "loss": 0.0,
+      "step": 8277
+    },
+    {
+      "epoch": 0.5334794096797061,
+      "grad_norm": 0.003701628229133183,
+      "learning_rate": 5.187969924812031e-06,
+      "loss": 0.0,
+      "step": 8278
+    },
+    {
+      "epoch": 0.5335438551266353,
+      "grad_norm": 0.00039285142383473234,
+      "learning_rate": 5.187253848907985e-06,
+      "loss": 0.0,
+      "step": 8279
+    },
+    {
+      "epoch": 0.5336083005735645,
+      "grad_norm": 0.0047367452853752864,
+      "learning_rate": 5.186537773003939e-06,
+      "loss": 0.0,
+      "step": 8280
+    },
+    {
+      "epoch": 0.5336727460204936,
+      "grad_norm": 0.003703470161721052,
+      "learning_rate": 5.185821697099893e-06,
+      "loss": 0.0,
+      "step": 8281
+    },
+    {
+      "epoch": 0.5337371914674228,
+      "grad_norm": 4.95614077371797e-05,
+      "learning_rate": 5.185105621195847e-06,
+      "loss": 0.0,
+      "step": 8282
+    },
+    {
+      "epoch": 0.533801636914352,
+      "grad_norm": 3.902805141580079e-05,
+      "learning_rate": 5.1843895452918015e-06,
+      "loss": 0.0,
+      "step": 8283
+    },
+    {
+      "epoch": 0.5338660823612812,
+      "grad_norm": 0.2618037697781461,
+      "learning_rate": 5.183673469387756e-06,
+      "loss": 0.0004,
+      "step": 8284
+    },
+    {
+      "epoch": 0.5339305278082104,
+      "grad_norm": 0.10652270259820498,
+      "learning_rate": 5.182957393483709e-06,
+      "loss": 0.0002,
+      "step": 8285
+    },
+    {
+      "epoch": 0.5339949732551396,
+      "grad_norm": 0.05615630343200612,
+      "learning_rate": 5.182241317579664e-06,
+      "loss": 0.0001,
+      "step": 8286
+    },
+    {
+      "epoch": 0.5340594187020687,
+      "grad_norm": 8.943660066176195e-05,
+      "learning_rate": 5.181525241675618e-06,
+      "loss": 0.0,
+      "step": 8287
+    },
+    {
+      "epoch": 0.5341238641489978,
+      "grad_norm": 0.00048379422864961943,
+      "learning_rate": 5.180809165771572e-06,
+      "loss": 0.0,
+      "step": 8288
+    },
+    {
+      "epoch": 0.534188309595927,
+      "grad_norm": 0.0003209484811010145,
+      "learning_rate": 5.1800930898675265e-06,
+      "loss": 0.0,
+      "step": 8289
+    },
+    {
+      "epoch": 0.5342527550428562,
+      "grad_norm": 0.000557616408584314,
+      "learning_rate": 5.17937701396348e-06,
+      "loss": 0.0,
+      "step": 8290
+    },
+    {
+      "epoch": 0.5343172004897854,
+      "grad_norm": 0.06915313841843637,
+      "learning_rate": 5.178660938059434e-06,
+      "loss": 0.0001,
+      "step": 8291
+    },
+    {
+      "epoch": 0.5343816459367146,
+      "grad_norm": 7.130504738155417e-05,
+      "learning_rate": 5.1779448621553885e-06,
+      "loss": 0.0,
+      "step": 8292
+    },
+    {
+      "epoch": 0.5344460913836437,
+      "grad_norm": 0.026762604545379487,
+      "learning_rate": 5.177228786251343e-06,
+      "loss": 0.0002,
+      "step": 8293
+    },
+    {
+      "epoch": 0.5345105368305729,
+      "grad_norm": 0.1844287713085006,
+      "learning_rate": 5.176512710347297e-06,
+      "loss": 0.0007,
+      "step": 8294
+    },
+    {
+      "epoch": 0.5345749822775021,
+      "grad_norm": 0.006867867960209021,
+      "learning_rate": 5.1757966344432506e-06,
+      "loss": 0.0,
+      "step": 8295
+    },
+    {
+      "epoch": 0.5346394277244313,
+      "grad_norm": 0.03740969652855573,
+      "learning_rate": 5.175080558539206e-06,
+      "loss": 0.0002,
+      "step": 8296
+    },
+    {
+      "epoch": 0.5347038731713605,
+      "grad_norm": 0.0006198577460679115,
+      "learning_rate": 5.17436448263516e-06,
+      "loss": 0.0,
+      "step": 8297
+    },
+    {
+      "epoch": 0.5347683186182897,
+      "grad_norm": 0.0001554785987894876,
+      "learning_rate": 5.173648406731114e-06,
+      "loss": 0.0,
+      "step": 8298
+    },
+    {
+      "epoch": 0.5348327640652188,
+      "grad_norm": 0.0020449257013498948,
+      "learning_rate": 5.172932330827069e-06,
+      "loss": 0.0,
+      "step": 8299
+    },
+    {
+      "epoch": 0.5348972095121479,
+      "grad_norm": 9.068199803593524e-05,
+      "learning_rate": 5.172216254923023e-06,
+      "loss": 0.0,
+      "step": 8300
+    },
+    {
+      "epoch": 0.5349616549590771,
+      "grad_norm": 0.0018346562331147103,
+      "learning_rate": 5.171500179018976e-06,
+      "loss": 0.0,
+      "step": 8301
+    },
+    {
+      "epoch": 0.5350261004060063,
+      "grad_norm": 0.06669650386462536,
+      "learning_rate": 5.170784103114931e-06,
+      "loss": 0.0005,
+      "step": 8302
+    },
+    {
+      "epoch": 0.5350905458529355,
+      "grad_norm": 3.8839867272744435,
+      "learning_rate": 5.170068027210885e-06,
+      "loss": 0.0224,
+      "step": 8303
+    },
+    {
+      "epoch": 0.5351549912998647,
+      "grad_norm": 0.0004139125755270844,
+      "learning_rate": 5.169351951306839e-06,
+      "loss": 0.0,
+      "step": 8304
+    },
+    {
+      "epoch": 0.5352194367467938,
+      "grad_norm": 0.23567786488833725,
+      "learning_rate": 5.1686358754027936e-06,
+      "loss": 0.0005,
+      "step": 8305
+    },
+    {
+      "epoch": 0.535283882193723,
+      "grad_norm": 0.0032149324484725223,
+      "learning_rate": 5.167919799498747e-06,
+      "loss": 0.0,
+      "step": 8306
+    },
+    {
+      "epoch": 0.5353483276406522,
+      "grad_norm": 0.001745209537508723,
+      "learning_rate": 5.167203723594701e-06,
+      "loss": 0.0,
+      "step": 8307
+    },
+    {
+      "epoch": 0.5354127730875814,
+      "grad_norm": 0.0003493430009983679,
+      "learning_rate": 5.166487647690656e-06,
+      "loss": 0.0,
+      "step": 8308
+    },
+    {
+      "epoch": 0.5354772185345106,
+      "grad_norm": 0.0006592212059325943,
+      "learning_rate": 5.16577157178661e-06,
+      "loss": 0.0,
+      "step": 8309
+    },
+    {
+      "epoch": 0.5355416639814398,
+      "grad_norm": 0.009202076173033253,
+      "learning_rate": 5.165055495882564e-06,
+      "loss": 0.0001,
+      "step": 8310
+    },
+    {
+      "epoch": 0.5356061094283688,
+      "grad_norm": 0.002701766323453557,
+      "learning_rate": 5.164339419978518e-06,
+      "loss": 0.0,
+      "step": 8311
+    },
+    {
+      "epoch": 0.535670554875298,
+      "grad_norm": 0.0005170194706688815,
+      "learning_rate": 5.163623344074472e-06,
+      "loss": 0.0,
+      "step": 8312
+    },
+    {
+      "epoch": 0.5357350003222272,
+      "grad_norm": 0.12185486281755308,
+      "learning_rate": 5.162907268170426e-06,
+      "loss": 0.0001,
+      "step": 8313
+    },
+    {
+      "epoch": 0.5357994457691564,
+      "grad_norm": 0.00047911325761194646,
+      "learning_rate": 5.1621911922663806e-06,
+      "loss": 0.0,
+      "step": 8314
+    },
+    {
+      "epoch": 0.5358638912160856,
+      "grad_norm": 0.0002798710818400487,
+      "learning_rate": 5.161475116362335e-06,
+      "loss": 0.0,
+      "step": 8315
+    },
+    {
+      "epoch": 0.5359283366630148,
+      "grad_norm": 0.01226373081079961,
+      "learning_rate": 5.160759040458288e-06,
+      "loss": 0.0001,
+      "step": 8316
+    },
+    {
+      "epoch": 0.5359927821099439,
+      "grad_norm": 0.00038180829833394726,
+      "learning_rate": 5.160042964554243e-06,
+      "loss": 0.0,
+      "step": 8317
+    },
+    {
+      "epoch": 0.5360572275568731,
+      "grad_norm": 4.5729150975741886e-05,
+      "learning_rate": 5.159326888650197e-06,
+      "loss": 0.0,
+      "step": 8318
+    },
+    {
+      "epoch": 0.5361216730038023,
+      "grad_norm": 0.7364370331298571,
+      "learning_rate": 5.158610812746152e-06,
+      "loss": 0.0017,
+      "step": 8319
+    },
+    {
+      "epoch": 0.5361861184507315,
+      "grad_norm": 0.0003008430119444691,
+      "learning_rate": 5.157894736842106e-06,
+      "loss": 0.0,
+      "step": 8320
+    },
+    {
+      "epoch": 0.5362505638976607,
+      "grad_norm": 0.0008920872326398447,
+      "learning_rate": 5.157178660938061e-06,
+      "loss": 0.0,
+      "step": 8321
+    },
+    {
+      "epoch": 0.5363150093445898,
+      "grad_norm": 0.030279863544600275,
+      "learning_rate": 5.156462585034014e-06,
+      "loss": 0.0002,
+      "step": 8322
+    },
+    {
+      "epoch": 0.5363794547915189,
+      "grad_norm": 0.00022730484292783038,
+      "learning_rate": 5.155746509129968e-06,
+      "loss": 0.0,
+      "step": 8323
+    },
+    {
+      "epoch": 0.5364439002384481,
+      "grad_norm": 6.012959763087097e-05,
+      "learning_rate": 5.155030433225923e-06,
+      "loss": 0.0,
+      "step": 8324
+    },
+    {
+      "epoch": 0.5365083456853773,
+      "grad_norm": 0.004307225892368384,
+      "learning_rate": 5.154314357321877e-06,
+      "loss": 0.0,
+      "step": 8325
+    },
+    {
+      "epoch": 0.5365727911323065,
+      "grad_norm": 0.14500555619742644,
+      "learning_rate": 5.153598281417831e-06,
+      "loss": 0.0013,
+      "step": 8326
+    },
+    {
+      "epoch": 0.5366372365792357,
+      "grad_norm": 0.003633286377291637,
+      "learning_rate": 5.152882205513785e-06,
+      "loss": 0.0,
+      "step": 8327
+    },
+    {
+      "epoch": 0.5367016820261649,
+      "grad_norm": 0.0030499388817333785,
+      "learning_rate": 5.152166129609739e-06,
+      "loss": 0.0,
+      "step": 8328
+    },
+    {
+      "epoch": 0.536766127473094,
+      "grad_norm": 0.003460639557092233,
+      "learning_rate": 5.151450053705693e-06,
+      "loss": 0.0,
+      "step": 8329
+    },
+    {
+      "epoch": 0.5368305729200232,
+      "grad_norm": 0.0654827941475252,
+      "learning_rate": 5.150733977801648e-06,
+      "loss": 0.0002,
+      "step": 8330
+    },
+    {
+      "epoch": 0.5368950183669524,
+      "grad_norm": 0.0006291733494983527,
+      "learning_rate": 5.150017901897602e-06,
+      "loss": 0.0,
+      "step": 8331
+    },
+    {
+      "epoch": 0.5369594638138816,
+      "grad_norm": 0.004459620943301213,
+      "learning_rate": 5.149301825993555e-06,
+      "loss": 0.0001,
+      "step": 8332
+    },
+    {
+      "epoch": 0.5370239092608107,
+      "grad_norm": 0.00040403846333492225,
+      "learning_rate": 5.14858575008951e-06,
+      "loss": 0.0,
+      "step": 8333
+    },
+    {
+      "epoch": 0.5370883547077399,
+      "grad_norm": 0.0007446600423081939,
+      "learning_rate": 5.147869674185464e-06,
+      "loss": 0.0,
+      "step": 8334
+    },
+    {
+      "epoch": 0.537152800154669,
+      "grad_norm": 0.0001650185458354627,
+      "learning_rate": 5.147153598281418e-06,
+      "loss": 0.0,
+      "step": 8335
+    },
+    {
+      "epoch": 0.5372172456015982,
+      "grad_norm": 1.7263584494515336e-05,
+      "learning_rate": 5.146437522377372e-06,
+      "loss": 0.0,
+      "step": 8336
+    },
+    {
+      "epoch": 0.5372816910485274,
+      "grad_norm": 0.002230091483141834,
+      "learning_rate": 5.145721446473326e-06,
+      "loss": 0.0,
+      "step": 8337
+    },
+    {
+      "epoch": 0.5373461364954566,
+      "grad_norm": 0.0005651914951587413,
+      "learning_rate": 5.14500537056928e-06,
+      "loss": 0.0,
+      "step": 8338
+    },
+    {
+      "epoch": 0.5374105819423858,
+      "grad_norm": 0.00022583757860816512,
+      "learning_rate": 5.144289294665235e-06,
+      "loss": 0.0,
+      "step": 8339
+    },
+    {
+      "epoch": 0.537475027389315,
+      "grad_norm": 0.170381111795957,
+      "learning_rate": 5.143573218761189e-06,
+      "loss": 0.0004,
+      "step": 8340
+    },
+    {
+      "epoch": 0.5375394728362441,
+      "grad_norm": 0.002303433150644348,
+      "learning_rate": 5.142857142857142e-06,
+      "loss": 0.0,
+      "step": 8341
+    },
+    {
+      "epoch": 0.5376039182831733,
+      "grad_norm": 0.002125040189832157,
+      "learning_rate": 5.142141066953097e-06,
+      "loss": 0.0,
+      "step": 8342
+    },
+    {
+      "epoch": 0.5376683637301025,
+      "grad_norm": 0.00013921987677024776,
+      "learning_rate": 5.141424991049052e-06,
+      "loss": 0.0,
+      "step": 8343
+    },
+    {
+      "epoch": 0.5377328091770316,
+      "grad_norm": 0.00022435381485150807,
+      "learning_rate": 5.140708915145006e-06,
+      "loss": 0.0,
+      "step": 8344
+    },
+    {
+      "epoch": 0.5377972546239608,
+      "grad_norm": 0.003466804477732005,
+      "learning_rate": 5.1399928392409604e-06,
+      "loss": 0.0,
+      "step": 8345
+    },
+    {
+      "epoch": 0.53786170007089,
+      "grad_norm": 0.00026267158997037504,
+      "learning_rate": 5.139276763336915e-06,
+      "loss": 0.0,
+      "step": 8346
+    },
+    {
+      "epoch": 0.5379261455178191,
+      "grad_norm": 0.0022633420166437274,
+      "learning_rate": 5.138560687432869e-06,
+      "loss": 0.0,
+      "step": 8347
+    },
+    {
+      "epoch": 0.5379905909647483,
+      "grad_norm": 0.025317271810819834,
+      "learning_rate": 5.1378446115288225e-06,
+      "loss": 0.0001,
+      "step": 8348
+    },
+    {
+      "epoch": 0.5380550364116775,
+      "grad_norm": 0.000673680858811099,
+      "learning_rate": 5.137128535624777e-06,
+      "loss": 0.0,
+      "step": 8349
+    },
+    {
+      "epoch": 0.5381194818586067,
+      "grad_norm": 0.0018941198791382194,
+      "learning_rate": 5.136412459720731e-06,
+      "loss": 0.0,
+      "step": 8350
+    },
+    {
+      "epoch": 0.5381839273055359,
+      "grad_norm": 0.5314760288276836,
+      "learning_rate": 5.135696383816685e-06,
+      "loss": 0.0032,
+      "step": 8351
+    },
+    {
+      "epoch": 0.5382483727524651,
+      "grad_norm": 0.010573634830043211,
+      "learning_rate": 5.134980307912639e-06,
+      "loss": 0.0001,
+      "step": 8352
+    },
+    {
+      "epoch": 0.5383128181993942,
+      "grad_norm": 0.00018203131654043764,
+      "learning_rate": 5.134264232008593e-06,
+      "loss": 0.0,
+      "step": 8353
+    },
+    {
+      "epoch": 0.5383772636463234,
+      "grad_norm": 1.1062119213382617,
+      "learning_rate": 5.133548156104547e-06,
+      "loss": 0.0072,
+      "step": 8354
+    },
+    {
+      "epoch": 0.5384417090932525,
+      "grad_norm": 0.010292150074225268,
+      "learning_rate": 5.132832080200502e-06,
+      "loss": 0.0,
+      "step": 8355
+    },
+    {
+      "epoch": 0.5385061545401817,
+      "grad_norm": 0.011232534638379256,
+      "learning_rate": 5.132116004296456e-06,
+      "loss": 0.0001,
+      "step": 8356
+    },
+    {
+      "epoch": 0.5385705999871109,
+      "grad_norm": 0.006481248957277151,
+      "learning_rate": 5.1313999283924095e-06,
+      "loss": 0.0,
+      "step": 8357
+    },
+    {
+      "epoch": 0.5386350454340401,
+      "grad_norm": 0.0103300801049018,
+      "learning_rate": 5.130683852488364e-06,
+      "loss": 0.0,
+      "step": 8358
+    },
+    {
+      "epoch": 0.5386994908809692,
+      "grad_norm": 0.0007205325030744387,
+      "learning_rate": 5.129967776584318e-06,
+      "loss": 0.0,
+      "step": 8359
+    },
+    {
+      "epoch": 0.5387639363278984,
+      "grad_norm": 0.001971069018050957,
+      "learning_rate": 5.129251700680272e-06,
+      "loss": 0.0,
+      "step": 8360
+    },
+    {
+      "epoch": 0.5388283817748276,
+      "grad_norm": 0.00025740076583384244,
+      "learning_rate": 5.128535624776227e-06,
+      "loss": 0.0,
+      "step": 8361
+    },
+    {
+      "epoch": 0.5388928272217568,
+      "grad_norm": 0.004754505978913037,
+      "learning_rate": 5.12781954887218e-06,
+      "loss": 0.0001,
+      "step": 8362
+    },
+    {
+      "epoch": 0.538957272668686,
+      "grad_norm": 6.622904868150729e-05,
+      "learning_rate": 5.127103472968134e-06,
+      "loss": 0.0,
+      "step": 8363
+    },
+    {
+      "epoch": 0.5390217181156152,
+      "grad_norm": 0.3479271808147232,
+      "learning_rate": 5.126387397064089e-06,
+      "loss": 0.0217,
+      "step": 8364
+    },
+    {
+      "epoch": 0.5390861635625444,
+      "grad_norm": 8.444964625094108e-05,
+      "learning_rate": 5.125671321160043e-06,
+      "loss": 0.0,
+      "step": 8365
+    },
+    {
+      "epoch": 0.5391506090094735,
+      "grad_norm": 0.00021665471768633301,
+      "learning_rate": 5.124955245255998e-06,
+      "loss": 0.0,
+      "step": 8366
+    },
+    {
+      "epoch": 0.5392150544564026,
+      "grad_norm": 0.00021705941240832152,
+      "learning_rate": 5.1242391693519525e-06,
+      "loss": 0.0,
+      "step": 8367
+    },
+    {
+      "epoch": 0.5392794999033318,
+      "grad_norm": 0.05381334252700333,
+      "learning_rate": 5.123523093447906e-06,
+      "loss": 0.0002,
+      "step": 8368
+    },
+    {
+      "epoch": 0.539343945350261,
+      "grad_norm": 0.0011084411287508335,
+      "learning_rate": 5.12280701754386e-06,
+      "loss": 0.0,
+      "step": 8369
+    },
+    {
+      "epoch": 0.5394083907971902,
+      "grad_norm": 0.002692409573749917,
+      "learning_rate": 5.1220909416398145e-06,
+      "loss": 0.0,
+      "step": 8370
+    },
+    {
+      "epoch": 0.5394728362441193,
+      "grad_norm": 0.0019404386497965766,
+      "learning_rate": 5.121374865735769e-06,
+      "loss": 0.0,
+      "step": 8371
+    },
+    {
+      "epoch": 0.5395372816910485,
+      "grad_norm": 0.000830474063169443,
+      "learning_rate": 5.120658789831723e-06,
+      "loss": 0.0,
+      "step": 8372
+    },
+    {
+      "epoch": 0.5396017271379777,
+      "grad_norm": 0.0001285680201385665,
+      "learning_rate": 5.1199427139276766e-06,
+      "loss": 0.0,
+      "step": 8373
+    },
+    {
+      "epoch": 0.5396661725849069,
+      "grad_norm": 1.688643150650448e-05,
+      "learning_rate": 5.119226638023631e-06,
+      "loss": 0.0,
+      "step": 8374
+    },
+    {
+      "epoch": 0.5397306180318361,
+      "grad_norm": 0.010633813313598334,
+      "learning_rate": 5.118510562119585e-06,
+      "loss": 0.0002,
+      "step": 8375
+    },
+    {
+      "epoch": 0.5397950634787653,
+      "grad_norm": 0.2513697768569286,
+      "learning_rate": 5.1177944862155394e-06,
+      "loss": 0.0004,
+      "step": 8376
+    },
+    {
+      "epoch": 0.5398595089256945,
+      "grad_norm": 0.011243210437085592,
+      "learning_rate": 5.117078410311494e-06,
+      "loss": 0.0,
+      "step": 8377
+    },
+    {
+      "epoch": 0.5399239543726235,
+      "grad_norm": 0.0634902939249464,
+      "learning_rate": 5.116362334407447e-06,
+      "loss": 0.0001,
+      "step": 8378
+    },
+    {
+      "epoch": 0.5399883998195527,
+      "grad_norm": 0.009778168936793704,
+      "learning_rate": 5.1156462585034015e-06,
+      "loss": 0.0001,
+      "step": 8379
+    },
+    {
+      "epoch": 0.5400528452664819,
+      "grad_norm": 4.753302020719344e-05,
+      "learning_rate": 5.114930182599356e-06,
+      "loss": 0.0,
+      "step": 8380
+    },
+    {
+      "epoch": 0.5401172907134111,
+      "grad_norm": 0.0010928865291910395,
+      "learning_rate": 5.11421410669531e-06,
+      "loss": 0.0,
+      "step": 8381
+    },
+    {
+      "epoch": 0.5401817361603403,
+      "grad_norm": 6.587222646677347e-06,
+      "learning_rate": 5.113498030791264e-06,
+      "loss": 0.0,
+      "step": 8382
+    },
+    {
+      "epoch": 0.5402461816072694,
+      "grad_norm": 0.0003763542620900478,
+      "learning_rate": 5.112781954887218e-06,
+      "loss": 0.0,
+      "step": 8383
+    },
+    {
+      "epoch": 0.5403106270541986,
+      "grad_norm": 0.0005508190320432071,
+      "learning_rate": 5.112065878983172e-06,
+      "loss": 0.0,
+      "step": 8384
+    },
+    {
+      "epoch": 0.5403750725011278,
+      "grad_norm": 0.0001657109248957889,
+      "learning_rate": 5.1113498030791264e-06,
+      "loss": 0.0,
+      "step": 8385
+    },
+    {
+      "epoch": 0.540439517948057,
+      "grad_norm": 0.000585394327920387,
+      "learning_rate": 5.110633727175081e-06,
+      "loss": 0.0,
+      "step": 8386
+    },
+    {
+      "epoch": 0.5405039633949862,
+      "grad_norm": 0.12383700047162453,
+      "learning_rate": 5.109917651271035e-06,
+      "loss": 0.0003,
+      "step": 8387
+    },
+    {
+      "epoch": 0.5405684088419154,
+      "grad_norm": 0.0012521166326916959,
+      "learning_rate": 5.1092015753669885e-06,
+      "loss": 0.0,
+      "step": 8388
+    },
+    {
+      "epoch": 0.5406328542888444,
+      "grad_norm": 0.024399225972084052,
+      "learning_rate": 5.108485499462944e-06,
+      "loss": 0.0002,
+      "step": 8389
+    },
+    {
+      "epoch": 0.5406972997357736,
+      "grad_norm": 0.0009896365504723618,
+      "learning_rate": 5.107769423558898e-06,
+      "loss": 0.0,
+      "step": 8390
+    },
+    {
+      "epoch": 0.5407617451827028,
+      "grad_norm": 0.41062065354460814,
+      "learning_rate": 5.107053347654852e-06,
+      "loss": 0.003,
+      "step": 8391
+    },
+    {
+      "epoch": 0.540826190629632,
+      "grad_norm": 0.00020028325207377276,
+      "learning_rate": 5.1063372717508065e-06,
+      "loss": 0.0,
+      "step": 8392
+    },
+    {
+      "epoch": 0.5408906360765612,
+      "grad_norm": 0.0021295931855767553,
+      "learning_rate": 5.105621195846761e-06,
+      "loss": 0.0,
+      "step": 8393
+    },
+    {
+      "epoch": 0.5409550815234904,
+      "grad_norm": 0.003080648669282957,
+      "learning_rate": 5.104905119942714e-06,
+      "loss": 0.0,
+      "step": 8394
+    },
+    {
+      "epoch": 0.5410195269704196,
+      "grad_norm": 0.037808258822707816,
+      "learning_rate": 5.104189044038669e-06,
+      "loss": 0.0001,
+      "step": 8395
+    },
+    {
+      "epoch": 0.5410839724173487,
+      "grad_norm": 1.003356723819026,
+      "learning_rate": 5.103472968134623e-06,
+      "loss": 0.006,
+      "step": 8396
+    },
+    {
+      "epoch": 0.5411484178642779,
+      "grad_norm": 1.2780228888800633,
+      "learning_rate": 5.102756892230577e-06,
+      "loss": 0.0034,
+      "step": 8397
+    },
+    {
+      "epoch": 0.5412128633112071,
+      "grad_norm": 0.006901590997431245,
+      "learning_rate": 5.1020408163265315e-06,
+      "loss": 0.0,
+      "step": 8398
+    },
+    {
+      "epoch": 0.5412773087581363,
+      "grad_norm": 0.020858812785563983,
+      "learning_rate": 5.101324740422485e-06,
+      "loss": 0.0001,
+      "step": 8399
+    },
+    {
+      "epoch": 0.5413417542050654,
+      "grad_norm": 0.0006388117841880102,
+      "learning_rate": 5.100608664518439e-06,
+      "loss": 0.0,
+      "step": 8400
+    },
+    {
+      "epoch": 0.5414061996519945,
+      "grad_norm": 0.09992779997527523,
+      "learning_rate": 5.0998925886143935e-06,
+      "loss": 0.0002,
+      "step": 8401
+    },
+    {
+      "epoch": 0.5414706450989237,
+      "grad_norm": 0.01728350633489314,
+      "learning_rate": 5.099176512710348e-06,
+      "loss": 0.0,
+      "step": 8402
+    },
+    {
+      "epoch": 0.5415350905458529,
+      "grad_norm": 0.010191441552771597,
+      "learning_rate": 5.098460436806302e-06,
+      "loss": 0.0001,
+      "step": 8403
+    },
+    {
+      "epoch": 0.5415995359927821,
+      "grad_norm": 0.0006010007948885225,
+      "learning_rate": 5.097744360902256e-06,
+      "loss": 0.0,
+      "step": 8404
+    },
+    {
+      "epoch": 0.5416639814397113,
+      "grad_norm": 0.002163450281916386,
+      "learning_rate": 5.09702828499821e-06,
+      "loss": 0.0,
+      "step": 8405
+    },
+    {
+      "epoch": 0.5417284268866405,
+      "grad_norm": 0.2445328803815511,
+      "learning_rate": 5.096312209094164e-06,
+      "loss": 0.0004,
+      "step": 8406
+    },
+    {
+      "epoch": 0.5417928723335697,
+      "grad_norm": 0.00222522598216728,
+      "learning_rate": 5.0955961331901185e-06,
+      "loss": 0.0,
+      "step": 8407
+    },
+    {
+      "epoch": 0.5418573177804988,
+      "grad_norm": 0.016346940554171265,
+      "learning_rate": 5.094880057286073e-06,
+      "loss": 0.0,
+      "step": 8408
+    },
+    {
+      "epoch": 0.541921763227428,
+      "grad_norm": 0.0017265815881555501,
+      "learning_rate": 5.094163981382026e-06,
+      "loss": 0.0,
+      "step": 8409
+    },
+    {
+      "epoch": 0.5419862086743572,
+      "grad_norm": 0.002953750292921564,
+      "learning_rate": 5.0934479054779805e-06,
+      "loss": 0.0,
+      "step": 8410
+    },
+    {
+      "epoch": 0.5420506541212863,
+      "grad_norm": 0.0017401369012034575,
+      "learning_rate": 5.092731829573935e-06,
+      "loss": 0.0,
+      "step": 8411
+    },
+    {
+      "epoch": 0.5421150995682155,
+      "grad_norm": 0.17858228586602923,
+      "learning_rate": 5.092015753669889e-06,
+      "loss": 0.0002,
+      "step": 8412
+    },
+    {
+      "epoch": 0.5421795450151446,
+      "grad_norm": 0.07280302949534724,
+      "learning_rate": 5.091299677765844e-06,
+      "loss": 0.0003,
+      "step": 8413
+    },
+    {
+      "epoch": 0.5422439904620738,
+      "grad_norm": 0.02793595399615248,
+      "learning_rate": 5.0905836018617986e-06,
+      "loss": 0.0,
+      "step": 8414
+    },
+    {
+      "epoch": 0.542308435909003,
+      "grad_norm": 0.07168111059143023,
+      "learning_rate": 5.089867525957752e-06,
+      "loss": 0.0002,
+      "step": 8415
+    },
+    {
+      "epoch": 0.5423728813559322,
+      "grad_norm": 0.013141592770366813,
+      "learning_rate": 5.089151450053706e-06,
+      "loss": 0.0,
+      "step": 8416
+    },
+    {
+      "epoch": 0.5424373268028614,
+      "grad_norm": 0.008050688347397276,
+      "learning_rate": 5.088435374149661e-06,
+      "loss": 0.0,
+      "step": 8417
+    },
+    {
+      "epoch": 0.5425017722497906,
+      "grad_norm": 0.01882217632422738,
+      "learning_rate": 5.087719298245615e-06,
+      "loss": 0.0,
+      "step": 8418
+    },
+    {
+      "epoch": 0.5425662176967198,
+      "grad_norm": 0.0018837017851688694,
+      "learning_rate": 5.087003222341569e-06,
+      "loss": 0.0,
+      "step": 8419
+    },
+    {
+      "epoch": 0.5426306631436489,
+      "grad_norm": 0.0027659480170431224,
+      "learning_rate": 5.086287146437523e-06,
+      "loss": 0.0,
+      "step": 8420
+    },
+    {
+      "epoch": 0.5426951085905781,
+      "grad_norm": 0.02249851271354615,
+      "learning_rate": 5.085571070533477e-06,
+      "loss": 0.0,
+      "step": 8421
+    },
+    {
+      "epoch": 0.5427595540375072,
+      "grad_norm": 0.24873905024797605,
+      "learning_rate": 5.084854994629431e-06,
+      "loss": 0.0004,
+      "step": 8422
+    },
+    {
+      "epoch": 0.5428239994844364,
+      "grad_norm": 0.0633032486789857,
+      "learning_rate": 5.0841389187253856e-06,
+      "loss": 0.0003,
+      "step": 8423
+    },
+    {
+      "epoch": 0.5428884449313656,
+      "grad_norm": 0.00014077413343572848,
+      "learning_rate": 5.08342284282134e-06,
+      "loss": 0.0,
+      "step": 8424
+    },
+    {
+      "epoch": 0.5429528903782948,
+      "grad_norm": 0.16562388707632744,
+      "learning_rate": 5.082706766917293e-06,
+      "loss": 0.0004,
+      "step": 8425
+    },
+    {
+      "epoch": 0.5430173358252239,
+      "grad_norm": 0.020381830748196907,
+      "learning_rate": 5.081990691013248e-06,
+      "loss": 0.0001,
+      "step": 8426
+    },
+    {
+      "epoch": 0.5430817812721531,
+      "grad_norm": 0.0020281902919562383,
+      "learning_rate": 5.081274615109202e-06,
+      "loss": 0.0,
+      "step": 8427
+    },
+    {
+      "epoch": 0.5431462267190823,
+      "grad_norm": 0.0002616038828169522,
+      "learning_rate": 5.080558539205156e-06,
+      "loss": 0.0,
+      "step": 8428
+    },
+    {
+      "epoch": 0.5432106721660115,
+      "grad_norm": 0.043275989112033196,
+      "learning_rate": 5.0798424633011105e-06,
+      "loss": 0.0002,
+      "step": 8429
+    },
+    {
+      "epoch": 0.5432751176129407,
+      "grad_norm": 0.00038925996925701165,
+      "learning_rate": 5.079126387397064e-06,
+      "loss": 0.0,
+      "step": 8430
+    },
+    {
+      "epoch": 0.5433395630598699,
+      "grad_norm": 0.1549475627115091,
+      "learning_rate": 5.078410311493018e-06,
+      "loss": 0.0007,
+      "step": 8431
+    },
+    {
+      "epoch": 0.543404008506799,
+      "grad_norm": 0.22845921268417974,
+      "learning_rate": 5.0776942355889726e-06,
+      "loss": 0.0004,
+      "step": 8432
+    },
+    {
+      "epoch": 0.5434684539537281,
+      "grad_norm": 0.02679208840814703,
+      "learning_rate": 5.076978159684927e-06,
+      "loss": 0.0001,
+      "step": 8433
+    },
+    {
+      "epoch": 0.5435328994006573,
+      "grad_norm": 0.004714868407184482,
+      "learning_rate": 5.07626208378088e-06,
+      "loss": 0.0,
+      "step": 8434
+    },
+    {
+      "epoch": 0.5435973448475865,
+      "grad_norm": 0.00257031806453596,
+      "learning_rate": 5.075546007876835e-06,
+      "loss": 0.0,
+      "step": 8435
+    },
+    {
+      "epoch": 0.5436617902945157,
+      "grad_norm": 0.021800592707464768,
+      "learning_rate": 5.07482993197279e-06,
+      "loss": 0.0,
+      "step": 8436
+    },
+    {
+      "epoch": 0.5437262357414449,
+      "grad_norm": 0.006127748959277537,
+      "learning_rate": 5.074113856068744e-06,
+      "loss": 0.0,
+      "step": 8437
+    },
+    {
+      "epoch": 0.543790681188374,
+      "grad_norm": 0.6883689850433048,
+      "learning_rate": 5.073397780164698e-06,
+      "loss": 0.0046,
+      "step": 8438
+    },
+    {
+      "epoch": 0.5438551266353032,
+      "grad_norm": 0.001150318426266875,
+      "learning_rate": 5.072681704260653e-06,
+      "loss": 0.0,
+      "step": 8439
+    },
+    {
+      "epoch": 0.5439195720822324,
+      "grad_norm": 0.002072532651337089,
+      "learning_rate": 5.071965628356607e-06,
+      "loss": 0.0,
+      "step": 8440
+    },
+    {
+      "epoch": 0.5439840175291616,
+      "grad_norm": 0.0002500489370783692,
+      "learning_rate": 5.07124955245256e-06,
+      "loss": 0.0,
+      "step": 8441
+    },
+    {
+      "epoch": 0.5440484629760908,
+      "grad_norm": 0.3168147475990133,
+      "learning_rate": 5.070533476548515e-06,
+      "loss": 0.0007,
+      "step": 8442
+    },
+    {
+      "epoch": 0.54411290842302,
+      "grad_norm": 0.8963605674738248,
+      "learning_rate": 5.069817400644469e-06,
+      "loss": 0.0054,
+      "step": 8443
+    },
+    {
+      "epoch": 0.5441773538699491,
+      "grad_norm": 0.001800277014813923,
+      "learning_rate": 5.069101324740423e-06,
+      "loss": 0.0,
+      "step": 8444
+    },
+    {
+      "epoch": 0.5442417993168782,
+      "grad_norm": 0.010620852971074788,
+      "learning_rate": 5.068385248836378e-06,
+      "loss": 0.0,
+      "step": 8445
+    },
+    {
+      "epoch": 0.5443062447638074,
+      "grad_norm": 0.012435870960374875,
+      "learning_rate": 5.067669172932331e-06,
+      "loss": 0.0,
+      "step": 8446
+    },
+    {
+      "epoch": 0.5443706902107366,
+      "grad_norm": 0.016274537573299074,
+      "learning_rate": 5.066953097028285e-06,
+      "loss": 0.0002,
+      "step": 8447
+    },
+    {
+      "epoch": 0.5444351356576658,
+      "grad_norm": 0.046486270889920786,
+      "learning_rate": 5.06623702112424e-06,
+      "loss": 0.0,
+      "step": 8448
+    },
+    {
+      "epoch": 0.544499581104595,
+      "grad_norm": 0.2916972351517615,
+      "learning_rate": 5.065520945220194e-06,
+      "loss": 0.0006,
+      "step": 8449
+    },
+    {
+      "epoch": 0.5445640265515241,
+      "grad_norm": 0.006940782184560781,
+      "learning_rate": 5.064804869316147e-06,
+      "loss": 0.0,
+      "step": 8450
+    },
+    {
+      "epoch": 0.5446284719984533,
+      "grad_norm": 0.016796576003806782,
+      "learning_rate": 5.064088793412102e-06,
+      "loss": 0.0,
+      "step": 8451
+    },
+    {
+      "epoch": 0.5446929174453825,
+      "grad_norm": 0.1291454725284565,
+      "learning_rate": 5.063372717508056e-06,
+      "loss": 0.0004,
+      "step": 8452
+    },
+    {
+      "epoch": 0.5447573628923117,
+      "grad_norm": 0.21752034007255436,
+      "learning_rate": 5.06265664160401e-06,
+      "loss": 0.0037,
+      "step": 8453
+    },
+    {
+      "epoch": 0.5448218083392409,
+      "grad_norm": 0.21748016693135402,
+      "learning_rate": 5.061940565699965e-06,
+      "loss": 0.0007,
+      "step": 8454
+    },
+    {
+      "epoch": 0.5448862537861701,
+      "grad_norm": 0.013149752274337866,
+      "learning_rate": 5.061224489795918e-06,
+      "loss": 0.0,
+      "step": 8455
+    },
+    {
+      "epoch": 0.5449506992330991,
+      "grad_norm": 0.00546176216623649,
+      "learning_rate": 5.060508413891872e-06,
+      "loss": 0.0,
+      "step": 8456
+    },
+    {
+      "epoch": 0.5450151446800283,
+      "grad_norm": 0.1679995530738449,
+      "learning_rate": 5.059792337987827e-06,
+      "loss": 0.0003,
+      "step": 8457
+    },
+    {
+      "epoch": 0.5450795901269575,
+      "grad_norm": 0.005963703038636505,
+      "learning_rate": 5.059076262083781e-06,
+      "loss": 0.0,
+      "step": 8458
+    },
+    {
+      "epoch": 0.5451440355738867,
+      "grad_norm": 0.0013005704425181387,
+      "learning_rate": 5.058360186179736e-06,
+      "loss": 0.0,
+      "step": 8459
+    },
+    {
+      "epoch": 0.5452084810208159,
+      "grad_norm": 0.28935846435973755,
+      "learning_rate": 5.05764411027569e-06,
+      "loss": 0.0004,
+      "step": 8460
+    },
+    {
+      "epoch": 0.5452729264677451,
+      "grad_norm": 0.05144240661139117,
+      "learning_rate": 5.056928034371645e-06,
+      "loss": 0.0001,
+      "step": 8461
+    },
+    {
+      "epoch": 0.5453373719146742,
+      "grad_norm": 0.25995588863752056,
+      "learning_rate": 5.056211958467598e-06,
+      "loss": 0.0014,
+      "step": 8462
+    },
+    {
+      "epoch": 0.5454018173616034,
+      "grad_norm": 0.0034649733590266948,
+      "learning_rate": 5.0554958825635524e-06,
+      "loss": 0.0,
+      "step": 8463
+    },
+    {
+      "epoch": 0.5454662628085326,
+      "grad_norm": 0.00736018183046089,
+      "learning_rate": 5.054779806659507e-06,
+      "loss": 0.0,
+      "step": 8464
+    },
+    {
+      "epoch": 0.5455307082554618,
+      "grad_norm": 0.20396441547188474,
+      "learning_rate": 5.054063730755461e-06,
+      "loss": 0.0009,
+      "step": 8465
+    },
+    {
+      "epoch": 0.545595153702391,
+      "grad_norm": 0.0818718599215992,
+      "learning_rate": 5.0533476548514145e-06,
+      "loss": 0.0001,
+      "step": 8466
+    },
+    {
+      "epoch": 0.54565959914932,
+      "grad_norm": 0.571786751738957,
+      "learning_rate": 5.052631578947369e-06,
+      "loss": 0.0005,
+      "step": 8467
+    },
+    {
+      "epoch": 0.5457240445962492,
+      "grad_norm": 0.03189593553225645,
+      "learning_rate": 5.051915503043323e-06,
+      "loss": 0.0,
+      "step": 8468
+    },
+    {
+      "epoch": 0.5457884900431784,
+      "grad_norm": 0.2562977042803086,
+      "learning_rate": 5.051199427139277e-06,
+      "loss": 0.0055,
+      "step": 8469
+    },
+    {
+      "epoch": 0.5458529354901076,
+      "grad_norm": 0.1811014294086341,
+      "learning_rate": 5.050483351235232e-06,
+      "loss": 0.0001,
+      "step": 8470
+    },
+    {
+      "epoch": 0.5459173809370368,
+      "grad_norm": 0.00438817026625853,
+      "learning_rate": 5.049767275331185e-06,
+      "loss": 0.0,
+      "step": 8471
+    },
+    {
+      "epoch": 0.545981826383966,
+      "grad_norm": 0.05164839767589543,
+      "learning_rate": 5.049051199427139e-06,
+      "loss": 0.0001,
+      "step": 8472
+    },
+    {
+      "epoch": 0.5460462718308952,
+      "grad_norm": 0.0001833333554516947,
+      "learning_rate": 5.048335123523094e-06,
+      "loss": 0.0,
+      "step": 8473
+    },
+    {
+      "epoch": 0.5461107172778243,
+      "grad_norm": 0.014345555426445568,
+      "learning_rate": 5.047619047619048e-06,
+      "loss": 0.0,
+      "step": 8474
+    },
+    {
+      "epoch": 0.5461751627247535,
+      "grad_norm": 0.0011238233457995053,
+      "learning_rate": 5.046902971715002e-06,
+      "loss": 0.0,
+      "step": 8475
+    },
+    {
+      "epoch": 0.5462396081716827,
+      "grad_norm": 1.2545085185504081,
+      "learning_rate": 5.046186895810956e-06,
+      "loss": 0.0033,
+      "step": 8476
+    },
+    {
+      "epoch": 0.5463040536186119,
+      "grad_norm": 0.5972288261339238,
+      "learning_rate": 5.04547081990691e-06,
+      "loss": 0.0017,
+      "step": 8477
+    },
+    {
+      "epoch": 0.546368499065541,
+      "grad_norm": 0.00844234305207491,
+      "learning_rate": 5.044754744002864e-06,
+      "loss": 0.0,
+      "step": 8478
+    },
+    {
+      "epoch": 0.5464329445124702,
+      "grad_norm": 0.0011451975272744536,
+      "learning_rate": 5.044038668098819e-06,
+      "loss": 0.0,
+      "step": 8479
+    },
+    {
+      "epoch": 0.5464973899593993,
+      "grad_norm": 0.05387772975290664,
+      "learning_rate": 5.043322592194773e-06,
+      "loss": 0.0002,
+      "step": 8480
+    },
+    {
+      "epoch": 0.5465618354063285,
+      "grad_norm": 0.0005193757903388635,
+      "learning_rate": 5.042606516290726e-06,
+      "loss": 0.0,
+      "step": 8481
+    },
+    {
+      "epoch": 0.5466262808532577,
+      "grad_norm": 5.494419434045538e-05,
+      "learning_rate": 5.041890440386681e-06,
+      "loss": 0.0,
+      "step": 8482
+    },
+    {
+      "epoch": 0.5466907263001869,
+      "grad_norm": 0.027016486452575494,
+      "learning_rate": 5.041174364482636e-06,
+      "loss": 0.0,
+      "step": 8483
+    },
+    {
+      "epoch": 0.5467551717471161,
+      "grad_norm": 0.0043791513808389085,
+      "learning_rate": 5.04045828857859e-06,
+      "loss": 0.0,
+      "step": 8484
+    },
+    {
+      "epoch": 0.5468196171940453,
+      "grad_norm": 0.29107433725311255,
+      "learning_rate": 5.0397422126745445e-06,
+      "loss": 0.0023,
+      "step": 8485
+    },
+    {
+      "epoch": 0.5468840626409744,
+      "grad_norm": 0.1543428437047449,
+      "learning_rate": 5.039026136770499e-06,
+      "loss": 0.001,
+      "step": 8486
+    },
+    {
+      "epoch": 0.5469485080879036,
+      "grad_norm": 0.008881782261700416,
+      "learning_rate": 5.038310060866452e-06,
+      "loss": 0.0,
+      "step": 8487
+    },
+    {
+      "epoch": 0.5470129535348328,
+      "grad_norm": 0.06710262718182129,
+      "learning_rate": 5.0375939849624065e-06,
+      "loss": 0.0003,
+      "step": 8488
+    },
+    {
+      "epoch": 0.5470773989817619,
+      "grad_norm": 0.04351480667082749,
+      "learning_rate": 5.036877909058361e-06,
+      "loss": 0.0016,
+      "step": 8489
+    },
+    {
+      "epoch": 0.5471418444286911,
+      "grad_norm": 0.0027709001929434106,
+      "learning_rate": 5.036161833154315e-06,
+      "loss": 0.0,
+      "step": 8490
+    },
+    {
+      "epoch": 0.5472062898756203,
+      "grad_norm": 0.008246434111247947,
+      "learning_rate": 5.035445757250269e-06,
+      "loss": 0.0,
+      "step": 8491
+    },
+    {
+      "epoch": 0.5472707353225494,
+      "grad_norm": 0.001855429937951199,
+      "learning_rate": 5.034729681346223e-06,
+      "loss": 0.0,
+      "step": 8492
+    },
+    {
+      "epoch": 0.5473351807694786,
+      "grad_norm": 0.054930165588919964,
+      "learning_rate": 5.034013605442177e-06,
+      "loss": 0.0002,
+      "step": 8493
+    },
+    {
+      "epoch": 0.5473996262164078,
+      "grad_norm": 0.005980705986750074,
+      "learning_rate": 5.0332975295381314e-06,
+      "loss": 0.0,
+      "step": 8494
+    },
+    {
+      "epoch": 0.547464071663337,
+      "grad_norm": 0.0031414896183599274,
+      "learning_rate": 5.032581453634086e-06,
+      "loss": 0.0,
+      "step": 8495
+    },
+    {
+      "epoch": 0.5475285171102662,
+      "grad_norm": 0.0005091821035342204,
+      "learning_rate": 5.03186537773004e-06,
+      "loss": 0.0,
+      "step": 8496
+    },
+    {
+      "epoch": 0.5475929625571954,
+      "grad_norm": 0.0017150736231015301,
+      "learning_rate": 5.0311493018259935e-06,
+      "loss": 0.0,
+      "step": 8497
+    },
+    {
+      "epoch": 0.5476574080041245,
+      "grad_norm": 0.19840486551000913,
+      "learning_rate": 5.030433225921948e-06,
+      "loss": 0.0002,
+      "step": 8498
+    },
+    {
+      "epoch": 0.5477218534510537,
+      "grad_norm": 1.3321290190046273,
+      "learning_rate": 5.029717150017902e-06,
+      "loss": 0.0099,
+      "step": 8499
+    },
+    {
+      "epoch": 0.5477862988979828,
+      "grad_norm": 0.00012402391058209097,
+      "learning_rate": 5.029001074113856e-06,
+      "loss": 0.0,
+      "step": 8500
+    },
+    {
+      "epoch": 0.547850744344912,
+      "grad_norm": 0.1594086270292522,
+      "learning_rate": 5.028284998209811e-06,
+      "loss": 0.0033,
+      "step": 8501
+    },
+    {
+      "epoch": 0.5479151897918412,
+      "grad_norm": 0.005608211860773984,
+      "learning_rate": 5.027568922305764e-06,
+      "loss": 0.0,
+      "step": 8502
+    },
+    {
+      "epoch": 0.5479796352387704,
+      "grad_norm": 0.000615421787577941,
+      "learning_rate": 5.0268528464017184e-06,
+      "loss": 0.0,
+      "step": 8503
+    },
+    {
+      "epoch": 0.5480440806856995,
+      "grad_norm": 0.002077126853143437,
+      "learning_rate": 5.026136770497673e-06,
+      "loss": 0.0,
+      "step": 8504
+    },
+    {
+      "epoch": 0.5481085261326287,
+      "grad_norm": 0.06163211903730178,
+      "learning_rate": 5.025420694593627e-06,
+      "loss": 0.0036,
+      "step": 8505
+    },
+    {
+      "epoch": 0.5481729715795579,
+      "grad_norm": 0.016333306029010678,
+      "learning_rate": 5.024704618689582e-06,
+      "loss": 0.0,
+      "step": 8506
+    },
+    {
+      "epoch": 0.5482374170264871,
+      "grad_norm": 0.002153469672369918,
+      "learning_rate": 5.0239885427855365e-06,
+      "loss": 0.0,
+      "step": 8507
+    },
+    {
+      "epoch": 0.5483018624734163,
+      "grad_norm": 0.1651638620042547,
+      "learning_rate": 5.02327246688149e-06,
+      "loss": 0.0011,
+      "step": 8508
+    },
+    {
+      "epoch": 0.5483663079203455,
+      "grad_norm": 0.00636054384750868,
+      "learning_rate": 5.022556390977444e-06,
+      "loss": 0.0001,
+      "step": 8509
+    },
+    {
+      "epoch": 0.5484307533672746,
+      "grad_norm": 0.0033881972217375076,
+      "learning_rate": 5.0218403150733985e-06,
+      "loss": 0.0,
+      "step": 8510
+    },
+    {
+      "epoch": 0.5484951988142037,
+      "grad_norm": 0.0018993068083641498,
+      "learning_rate": 5.021124239169353e-06,
+      "loss": 0.0,
+      "step": 8511
+    },
+    {
+      "epoch": 0.5485596442611329,
+      "grad_norm": 0.005748847183468727,
+      "learning_rate": 5.020408163265307e-06,
+      "loss": 0.0,
+      "step": 8512
+    },
+    {
+      "epoch": 0.5486240897080621,
+      "grad_norm": 0.27499401671229495,
+      "learning_rate": 5.019692087361261e-06,
+      "loss": 0.0009,
+      "step": 8513
+    },
+    {
+      "epoch": 0.5486885351549913,
+      "grad_norm": 0.017018584289137592,
+      "learning_rate": 5.018976011457215e-06,
+      "loss": 0.0,
+      "step": 8514
+    },
+    {
+      "epoch": 0.5487529806019205,
+      "grad_norm": 0.0014632281008861163,
+      "learning_rate": 5.018259935553169e-06,
+      "loss": 0.0,
+      "step": 8515
+    },
+    {
+      "epoch": 0.5488174260488496,
+      "grad_norm": 0.012847125648462757,
+      "learning_rate": 5.0175438596491235e-06,
+      "loss": 0.0,
+      "step": 8516
+    },
+    {
+      "epoch": 0.5488818714957788,
+      "grad_norm": 0.007008510181476973,
+      "learning_rate": 5.016827783745078e-06,
+      "loss": 0.0,
+      "step": 8517
+    },
+    {
+      "epoch": 0.548946316942708,
+      "grad_norm": 0.11917408488960375,
+      "learning_rate": 5.016111707841031e-06,
+      "loss": 0.0003,
+      "step": 8518
+    },
+    {
+      "epoch": 0.5490107623896372,
+      "grad_norm": 0.007272636048923761,
+      "learning_rate": 5.0153956319369855e-06,
+      "loss": 0.0001,
+      "step": 8519
+    },
+    {
+      "epoch": 0.5490752078365664,
+      "grad_norm": 0.004351227458944417,
+      "learning_rate": 5.01467955603294e-06,
+      "loss": 0.0,
+      "step": 8520
+    },
+    {
+      "epoch": 0.5491396532834956,
+      "grad_norm": 0.19051117742905466,
+      "learning_rate": 5.013963480128894e-06,
+      "loss": 0.0034,
+      "step": 8521
+    },
+    {
+      "epoch": 0.5492040987304248,
+      "grad_norm": 0.00026021408293215944,
+      "learning_rate": 5.0132474042248484e-06,
+      "loss": 0.0,
+      "step": 8522
+    },
+    {
+      "epoch": 0.5492685441773538,
+      "grad_norm": 0.002791641585007276,
+      "learning_rate": 5.012531328320802e-06,
+      "loss": 0.0,
+      "step": 8523
+    },
+    {
+      "epoch": 0.549332989624283,
+      "grad_norm": 0.005355082463114203,
+      "learning_rate": 5.011815252416756e-06,
+      "loss": 0.0,
+      "step": 8524
+    },
+    {
+      "epoch": 0.5493974350712122,
+      "grad_norm": 0.005519173063785302,
+      "learning_rate": 5.0110991765127105e-06,
+      "loss": 0.0,
+      "step": 8525
+    },
+    {
+      "epoch": 0.5494618805181414,
+      "grad_norm": 0.0020975079614027133,
+      "learning_rate": 5.010383100608665e-06,
+      "loss": 0.0,
+      "step": 8526
+    },
+    {
+      "epoch": 0.5495263259650706,
+      "grad_norm": 0.015470825862644465,
+      "learning_rate": 5.009667024704618e-06,
+      "loss": 0.0001,
+      "step": 8527
+    },
+    {
+      "epoch": 0.5495907714119997,
+      "grad_norm": 0.0014627428871791936,
+      "learning_rate": 5.0089509488005725e-06,
+      "loss": 0.0,
+      "step": 8528
+    },
+    {
+      "epoch": 0.5496552168589289,
+      "grad_norm": 0.012332215954442869,
+      "learning_rate": 5.008234872896528e-06,
+      "loss": 0.0001,
+      "step": 8529
+    },
+    {
+      "epoch": 0.5497196623058581,
+      "grad_norm": 0.000407497928952683,
+      "learning_rate": 5.007518796992482e-06,
+      "loss": 0.0,
+      "step": 8530
+    },
+    {
+      "epoch": 0.5497841077527873,
+      "grad_norm": 0.0013019234173674507,
+      "learning_rate": 5.006802721088436e-06,
+      "loss": 0.0,
+      "step": 8531
+    },
+    {
+      "epoch": 0.5498485531997165,
+      "grad_norm": 0.0013845580062636382,
+      "learning_rate": 5.0060866451843906e-06,
+      "loss": 0.0,
+      "step": 8532
+    },
+    {
+      "epoch": 0.5499129986466457,
+      "grad_norm": 0.2245196370524226,
+      "learning_rate": 5.005370569280345e-06,
+      "loss": 0.0005,
+      "step": 8533
+    },
+    {
+      "epoch": 0.5499774440935747,
+      "grad_norm": 8.993623459536706e-05,
+      "learning_rate": 5.004654493376298e-06,
+      "loss": 0.0,
+      "step": 8534
+    },
+    {
+      "epoch": 0.5500418895405039,
+      "grad_norm": 0.3626066527828885,
+      "learning_rate": 5.003938417472253e-06,
+      "loss": 0.0022,
+      "step": 8535
+    },
+    {
+      "epoch": 0.5501063349874331,
+      "grad_norm": 0.020626980673355562,
+      "learning_rate": 5.003222341568207e-06,
+      "loss": 0.0,
+      "step": 8536
+    },
+    {
+      "epoch": 0.5501707804343623,
+      "grad_norm": 0.10646967652793571,
+      "learning_rate": 5.002506265664161e-06,
+      "loss": 0.0004,
+      "step": 8537
+    },
+    {
+      "epoch": 0.5502352258812915,
+      "grad_norm": 0.09317456412157121,
+      "learning_rate": 5.0017901897601155e-06,
+      "loss": 0.0002,
+      "step": 8538
+    },
+    {
+      "epoch": 0.5502996713282207,
+      "grad_norm": 0.036085198902033974,
+      "learning_rate": 5.001074113856069e-06,
+      "loss": 0.0002,
+      "step": 8539
+    },
+    {
+      "epoch": 0.5503641167751498,
+      "grad_norm": 0.0037343965925228746,
+      "learning_rate": 5.000358037952023e-06,
+      "loss": 0.0,
+      "step": 8540
+    },
+    {
+      "epoch": 0.550428562222079,
+      "grad_norm": 0.009326869675734522,
+      "learning_rate": 4.9996419620479776e-06,
+      "loss": 0.0,
+      "step": 8541
+    },
+    {
+      "epoch": 0.5504930076690082,
+      "grad_norm": 0.021823792039814445,
+      "learning_rate": 4.998925886143932e-06,
+      "loss": 0.0001,
+      "step": 8542
+    },
+    {
+      "epoch": 0.5505574531159374,
+      "grad_norm": 0.0001219656029408,
+      "learning_rate": 4.998209810239885e-06,
+      "loss": 0.0,
+      "step": 8543
+    },
+    {
+      "epoch": 0.5506218985628666,
+      "grad_norm": 0.03387806160686626,
+      "learning_rate": 4.99749373433584e-06,
+      "loss": 0.0,
+      "step": 8544
+    },
+    {
+      "epoch": 0.5506863440097957,
+      "grad_norm": 0.00012018551306236546,
+      "learning_rate": 4.996777658431794e-06,
+      "loss": 0.0,
+      "step": 8545
+    },
+    {
+      "epoch": 0.5507507894567248,
+      "grad_norm": 0.007050738607783566,
+      "learning_rate": 4.996061582527748e-06,
+      "loss": 0.0,
+      "step": 8546
+    },
+    {
+      "epoch": 0.550815234903654,
+      "grad_norm": 0.0021130126048716142,
+      "learning_rate": 4.9953455066237025e-06,
+      "loss": 0.0,
+      "step": 8547
+    },
+    {
+      "epoch": 0.5508796803505832,
+      "grad_norm": 0.0006315345436932797,
+      "learning_rate": 4.994629430719657e-06,
+      "loss": 0.0,
+      "step": 8548
+    },
+    {
+      "epoch": 0.5509441257975124,
+      "grad_norm": 0.0007102576706969377,
+      "learning_rate": 4.993913354815611e-06,
+      "loss": 0.0,
+      "step": 8549
+    },
+    {
+      "epoch": 0.5510085712444416,
+      "grad_norm": 0.001955949521665762,
+      "learning_rate": 4.993197278911565e-06,
+      "loss": 0.0,
+      "step": 8550
+    },
+    {
+      "epoch": 0.5510730166913708,
+      "grad_norm": 0.009190210993093906,
+      "learning_rate": 4.992481203007519e-06,
+      "loss": 0.0,
+      "step": 8551
+    },
+    {
+      "epoch": 0.5511374621383,
+      "grad_norm": 0.401591291768431,
+      "learning_rate": 4.991765127103473e-06,
+      "loss": 0.0007,
+      "step": 8552
+    },
+    {
+      "epoch": 0.5512019075852291,
+      "grad_norm": 0.0005988339569598564,
+      "learning_rate": 4.9910490511994274e-06,
+      "loss": 0.0,
+      "step": 8553
+    },
+    {
+      "epoch": 0.5512663530321583,
+      "grad_norm": 0.001051486590964813,
+      "learning_rate": 4.990332975295382e-06,
+      "loss": 0.0,
+      "step": 8554
+    },
+    {
+      "epoch": 0.5513307984790875,
+      "grad_norm": 0.006394037454805913,
+      "learning_rate": 4.989616899391336e-06,
+      "loss": 0.0,
+      "step": 8555
+    },
+    {
+      "epoch": 0.5513952439260166,
+      "grad_norm": 5.165007726001237e-05,
+      "learning_rate": 4.9889008234872895e-06,
+      "loss": 0.0,
+      "step": 8556
+    },
+    {
+      "epoch": 0.5514596893729458,
+      "grad_norm": 0.002047369514226012,
+      "learning_rate": 4.988184747583244e-06,
+      "loss": 0.0,
+      "step": 8557
+    },
+    {
+      "epoch": 0.551524134819875,
+      "grad_norm": 0.0012906349742674791,
+      "learning_rate": 4.987468671679198e-06,
+      "loss": 0.0,
+      "step": 8558
+    },
+    {
+      "epoch": 0.5515885802668041,
+      "grad_norm": 0.0029618502007701924,
+      "learning_rate": 4.986752595775152e-06,
+      "loss": 0.0,
+      "step": 8559
+    },
+    {
+      "epoch": 0.5516530257137333,
+      "grad_norm": 0.01238850996303165,
+      "learning_rate": 4.986036519871107e-06,
+      "loss": 0.0,
+      "step": 8560
+    },
+    {
+      "epoch": 0.5517174711606625,
+      "grad_norm": 0.0009714262530897783,
+      "learning_rate": 4.985320443967061e-06,
+      "loss": 0.0,
+      "step": 8561
+    },
+    {
+      "epoch": 0.5517819166075917,
+      "grad_norm": 0.0010781651410204844,
+      "learning_rate": 4.984604368063015e-06,
+      "loss": 0.0,
+      "step": 8562
+    },
+    {
+      "epoch": 0.5518463620545209,
+      "grad_norm": 0.15714035869658133,
+      "learning_rate": 4.98388829215897e-06,
+      "loss": 0.0004,
+      "step": 8563
+    },
+    {
+      "epoch": 0.55191080750145,
+      "grad_norm": 0.0034370084976843578,
+      "learning_rate": 4.983172216254923e-06,
+      "loss": 0.0,
+      "step": 8564
+    },
+    {
+      "epoch": 0.5519752529483792,
+      "grad_norm": 0.0002537463466807135,
+      "learning_rate": 4.982456140350877e-06,
+      "loss": 0.0,
+      "step": 8565
+    },
+    {
+      "epoch": 0.5520396983953084,
+      "grad_norm": 0.0035772505393363426,
+      "learning_rate": 4.981740064446832e-06,
+      "loss": 0.0,
+      "step": 8566
+    },
+    {
+      "epoch": 0.5521041438422375,
+      "grad_norm": 0.0007178845574760114,
+      "learning_rate": 4.981023988542786e-06,
+      "loss": 0.0,
+      "step": 8567
+    },
+    {
+      "epoch": 0.5521685892891667,
+      "grad_norm": 0.005066510748939021,
+      "learning_rate": 4.98030791263874e-06,
+      "loss": 0.0,
+      "step": 8568
+    },
+    {
+      "epoch": 0.5522330347360959,
+      "grad_norm": 0.0005125547267131619,
+      "learning_rate": 4.979591836734694e-06,
+      "loss": 0.0,
+      "step": 8569
+    },
+    {
+      "epoch": 0.552297480183025,
+      "grad_norm": 0.003153537373562369,
+      "learning_rate": 4.978875760830649e-06,
+      "loss": 0.0,
+      "step": 8570
+    },
+    {
+      "epoch": 0.5523619256299542,
+      "grad_norm": 0.006392792439787252,
+      "learning_rate": 4.978159684926603e-06,
+      "loss": 0.0,
+      "step": 8571
+    },
+    {
+      "epoch": 0.5524263710768834,
+      "grad_norm": 4.826716838034971,
+      "learning_rate": 4.977443609022557e-06,
+      "loss": 0.0123,
+      "step": 8572
+    },
+    {
+      "epoch": 0.5524908165238126,
+      "grad_norm": 9.228853949010394e-05,
+      "learning_rate": 4.976727533118511e-06,
+      "loss": 0.0,
+      "step": 8573
+    },
+    {
+      "epoch": 0.5525552619707418,
+      "grad_norm": 0.007953029032488415,
+      "learning_rate": 4.976011457214465e-06,
+      "loss": 0.0,
+      "step": 8574
+    },
+    {
+      "epoch": 0.552619707417671,
+      "grad_norm": 0.0009689809464016709,
+      "learning_rate": 4.9752953813104195e-06,
+      "loss": 0.0,
+      "step": 8575
+    },
+    {
+      "epoch": 0.5526841528646002,
+      "grad_norm": 0.0024152789138099672,
+      "learning_rate": 4.974579305406374e-06,
+      "loss": 0.0,
+      "step": 8576
+    },
+    {
+      "epoch": 0.5527485983115293,
+      "grad_norm": 9.91049513405085e-05,
+      "learning_rate": 4.973863229502327e-06,
+      "loss": 0.0,
+      "step": 8577
+    },
+    {
+      "epoch": 0.5528130437584584,
+      "grad_norm": 0.025710516224901427,
+      "learning_rate": 4.9731471535982815e-06,
+      "loss": 0.0001,
+      "step": 8578
+    },
+    {
+      "epoch": 0.5528774892053876,
+      "grad_norm": 0.0009076876494428708,
+      "learning_rate": 4.972431077694236e-06,
+      "loss": 0.0,
+      "step": 8579
+    },
+    {
+      "epoch": 0.5529419346523168,
+      "grad_norm": 0.3465730529980872,
+      "learning_rate": 4.97171500179019e-06,
+      "loss": 0.0018,
+      "step": 8580
+    },
+    {
+      "epoch": 0.553006380099246,
+      "grad_norm": 0.0037399123513236405,
+      "learning_rate": 4.9709989258861444e-06,
+      "loss": 0.0,
+      "step": 8581
+    },
+    {
+      "epoch": 0.5530708255461751,
+      "grad_norm": 0.0002229589558730772,
+      "learning_rate": 4.970282849982099e-06,
+      "loss": 0.0,
+      "step": 8582
+    },
+    {
+      "epoch": 0.5531352709931043,
+      "grad_norm": 0.004994821684799948,
+      "learning_rate": 4.969566774078053e-06,
+      "loss": 0.0,
+      "step": 8583
+    },
+    {
+      "epoch": 0.5531997164400335,
+      "grad_norm": 0.0001422551604455913,
+      "learning_rate": 4.968850698174007e-06,
+      "loss": 0.0,
+      "step": 8584
+    },
+    {
+      "epoch": 0.5532641618869627,
+      "grad_norm": 3.6293142231062747,
+      "learning_rate": 4.968134622269961e-06,
+      "loss": 0.0243,
+      "step": 8585
+    },
+    {
+      "epoch": 0.5533286073338919,
+      "grad_norm": 0.0006342448217312984,
+      "learning_rate": 4.967418546365915e-06,
+      "loss": 0.0,
+      "step": 8586
+    },
+    {
+      "epoch": 0.5533930527808211,
+      "grad_norm": 0.011555769561746025,
+      "learning_rate": 4.966702470461869e-06,
+      "loss": 0.0,
+      "step": 8587
+    },
+    {
+      "epoch": 0.5534574982277503,
+      "grad_norm": 0.01095507667123741,
+      "learning_rate": 4.965986394557824e-06,
+      "loss": 0.0001,
+      "step": 8588
+    },
+    {
+      "epoch": 0.5535219436746793,
+      "grad_norm": 1.259846859798293e-05,
+      "learning_rate": 4.965270318653778e-06,
+      "loss": 0.0,
+      "step": 8589
+    },
+    {
+      "epoch": 0.5535863891216085,
+      "grad_norm": 0.005156511472907512,
+      "learning_rate": 4.964554242749731e-06,
+      "loss": 0.0,
+      "step": 8590
+    },
+    {
+      "epoch": 0.5536508345685377,
+      "grad_norm": 0.2383249586696943,
+      "learning_rate": 4.963838166845686e-06,
+      "loss": 0.0009,
+      "step": 8591
+    },
+    {
+      "epoch": 0.5537152800154669,
+      "grad_norm": 0.012284478435519698,
+      "learning_rate": 4.96312209094164e-06,
+      "loss": 0.0,
+      "step": 8592
+    },
+    {
+      "epoch": 0.5537797254623961,
+      "grad_norm": 0.0055611878877656114,
+      "learning_rate": 4.962406015037594e-06,
+      "loss": 0.0,
+      "step": 8593
+    },
+    {
+      "epoch": 0.5538441709093253,
+      "grad_norm": 0.32462081338167953,
+      "learning_rate": 4.961689939133549e-06,
+      "loss": 0.0023,
+      "step": 8594
+    },
+    {
+      "epoch": 0.5539086163562544,
+      "grad_norm": 0.0036361403139943778,
+      "learning_rate": 4.960973863229503e-06,
+      "loss": 0.0,
+      "step": 8595
+    },
+    {
+      "epoch": 0.5539730618031836,
+      "grad_norm": 0.01589423451383375,
+      "learning_rate": 4.960257787325457e-06,
+      "loss": 0.0,
+      "step": 8596
+    },
+    {
+      "epoch": 0.5540375072501128,
+      "grad_norm": 0.0011489116491658889,
+      "learning_rate": 4.9595417114214115e-06,
+      "loss": 0.0,
+      "step": 8597
+    },
+    {
+      "epoch": 0.554101952697042,
+      "grad_norm": 0.0010378343495381565,
+      "learning_rate": 4.958825635517365e-06,
+      "loss": 0.0,
+      "step": 8598
+    },
+    {
+      "epoch": 0.5541663981439712,
+      "grad_norm": 0.00014715845418476452,
+      "learning_rate": 4.958109559613319e-06,
+      "loss": 0.0,
+      "step": 8599
+    },
+    {
+      "epoch": 0.5542308435909004,
+      "grad_norm": 0.0020139445294210146,
+      "learning_rate": 4.9573934837092736e-06,
+      "loss": 0.0,
+      "step": 8600
+    },
+    {
+      "epoch": 0.5542952890378294,
+      "grad_norm": 0.0007146691062408835,
+      "learning_rate": 4.956677407805228e-06,
+      "loss": 0.0,
+      "step": 8601
+    },
+    {
+      "epoch": 0.5543597344847586,
+      "grad_norm": 0.006505887901441304,
+      "learning_rate": 4.955961331901182e-06,
+      "loss": 0.0,
+      "step": 8602
+    },
+    {
+      "epoch": 0.5544241799316878,
+      "grad_norm": 0.009405139009455533,
+      "learning_rate": 4.955245255997136e-06,
+      "loss": 0.0,
+      "step": 8603
+    },
+    {
+      "epoch": 0.554488625378617,
+      "grad_norm": 0.003237377086330774,
+      "learning_rate": 4.95452918009309e-06,
+      "loss": 0.0,
+      "step": 8604
+    },
+    {
+      "epoch": 0.5545530708255462,
+      "grad_norm": 0.0031621215893174203,
+      "learning_rate": 4.953813104189045e-06,
+      "loss": 0.0,
+      "step": 8605
+    },
+    {
+      "epoch": 0.5546175162724754,
+      "grad_norm": 0.19976517593339282,
+      "learning_rate": 4.9530970282849985e-06,
+      "loss": 0.0004,
+      "step": 8606
+    },
+    {
+      "epoch": 0.5546819617194045,
+      "grad_norm": 0.0014662636392100076,
+      "learning_rate": 4.952380952380953e-06,
+      "loss": 0.0,
+      "step": 8607
+    },
+    {
+      "epoch": 0.5547464071663337,
+      "grad_norm": 0.3341552899371035,
+      "learning_rate": 4.951664876476907e-06,
+      "loss": 0.0034,
+      "step": 8608
+    },
+    {
+      "epoch": 0.5548108526132629,
+      "grad_norm": 0.004519190199484597,
+      "learning_rate": 4.950948800572861e-06,
+      "loss": 0.0,
+      "step": 8609
+    },
+    {
+      "epoch": 0.5548752980601921,
+      "grad_norm": 0.25093081164918085,
+      "learning_rate": 4.950232724668816e-06,
+      "loss": 0.0005,
+      "step": 8610
+    },
+    {
+      "epoch": 0.5549397435071213,
+      "grad_norm": 0.005922211232974903,
+      "learning_rate": 4.949516648764769e-06,
+      "loss": 0.0,
+      "step": 8611
+    },
+    {
+      "epoch": 0.5550041889540503,
+      "grad_norm": 0.0003468587612752949,
+      "learning_rate": 4.9488005728607234e-06,
+      "loss": 0.0,
+      "step": 8612
+    },
+    {
+      "epoch": 0.5550686344009795,
+      "grad_norm": 0.14501370633433025,
+      "learning_rate": 4.948084496956678e-06,
+      "loss": 0.0002,
+      "step": 8613
+    },
+    {
+      "epoch": 0.5551330798479087,
+      "grad_norm": 0.1744257960372159,
+      "learning_rate": 4.947368421052632e-06,
+      "loss": 0.0004,
+      "step": 8614
+    },
+    {
+      "epoch": 0.5551975252948379,
+      "grad_norm": 0.0036943899184547248,
+      "learning_rate": 4.946652345148586e-06,
+      "loss": 0.0,
+      "step": 8615
+    },
+    {
+      "epoch": 0.5552619707417671,
+      "grad_norm": 0.0005746585162564398,
+      "learning_rate": 4.94593626924454e-06,
+      "loss": 0.0,
+      "step": 8616
+    },
+    {
+      "epoch": 0.5553264161886963,
+      "grad_norm": 0.0054436829334644915,
+      "learning_rate": 4.945220193340495e-06,
+      "loss": 0.0,
+      "step": 8617
+    },
+    {
+      "epoch": 0.5553908616356255,
+      "grad_norm": 0.00019123731125635397,
+      "learning_rate": 4.944504117436449e-06,
+      "loss": 0.0,
+      "step": 8618
+    },
+    {
+      "epoch": 0.5554553070825546,
+      "grad_norm": 0.31377869303838934,
+      "learning_rate": 4.943788041532403e-06,
+      "loss": 0.0007,
+      "step": 8619
+    },
+    {
+      "epoch": 0.5555197525294838,
+      "grad_norm": 0.009279498678199789,
+      "learning_rate": 4.943071965628357e-06,
+      "loss": 0.0,
+      "step": 8620
+    },
+    {
+      "epoch": 0.555584197976413,
+      "grad_norm": 0.0013721914906434604,
+      "learning_rate": 4.942355889724311e-06,
+      "loss": 0.0,
+      "step": 8621
+    },
+    {
+      "epoch": 0.5556486434233422,
+      "grad_norm": 0.00026775923044403474,
+      "learning_rate": 4.941639813820266e-06,
+      "loss": 0.0,
+      "step": 8622
+    },
+    {
+      "epoch": 0.5557130888702713,
+      "grad_norm": 0.22156785673329588,
+      "learning_rate": 4.94092373791622e-06,
+      "loss": 0.0007,
+      "step": 8623
+    },
+    {
+      "epoch": 0.5557775343172005,
+      "grad_norm": 0.01439473978710645,
+      "learning_rate": 4.940207662012173e-06,
+      "loss": 0.0,
+      "step": 8624
+    },
+    {
+      "epoch": 0.5558419797641296,
+      "grad_norm": 0.0032194399396147095,
+      "learning_rate": 4.939491586108128e-06,
+      "loss": 0.0,
+      "step": 8625
+    },
+    {
+      "epoch": 0.5559064252110588,
+      "grad_norm": 0.00022568911306749794,
+      "learning_rate": 4.938775510204082e-06,
+      "loss": 0.0,
+      "step": 8626
+    },
+    {
+      "epoch": 0.555970870657988,
+      "grad_norm": 0.02743558809261413,
+      "learning_rate": 4.938059434300036e-06,
+      "loss": 0.0001,
+      "step": 8627
+    },
+    {
+      "epoch": 0.5560353161049172,
+      "grad_norm": 0.03291318148000441,
+      "learning_rate": 4.93734335839599e-06,
+      "loss": 0.0003,
+      "step": 8628
+    },
+    {
+      "epoch": 0.5560997615518464,
+      "grad_norm": 0.26339233556561864,
+      "learning_rate": 4.936627282491945e-06,
+      "loss": 0.0016,
+      "step": 8629
+    },
+    {
+      "epoch": 0.5561642069987756,
+      "grad_norm": 0.03465554861101963,
+      "learning_rate": 4.935911206587899e-06,
+      "loss": 0.0001,
+      "step": 8630
+    },
+    {
+      "epoch": 0.5562286524457047,
+      "grad_norm": 0.0010887964390989104,
+      "learning_rate": 4.9351951306838534e-06,
+      "loss": 0.0,
+      "step": 8631
+    },
+    {
+      "epoch": 0.5562930978926339,
+      "grad_norm": 0.0033196397485056733,
+      "learning_rate": 4.934479054779807e-06,
+      "loss": 0.0,
+      "step": 8632
+    },
+    {
+      "epoch": 0.5563575433395631,
+      "grad_norm": 0.021989216789134153,
+      "learning_rate": 4.933762978875761e-06,
+      "loss": 0.0002,
+      "step": 8633
+    },
+    {
+      "epoch": 0.5564219887864922,
+      "grad_norm": 1.53336188593831,
+      "learning_rate": 4.9330469029717155e-06,
+      "loss": 0.003,
+      "step": 8634
+    },
+    {
+      "epoch": 0.5564864342334214,
+      "grad_norm": 0.03507442853934759,
+      "learning_rate": 4.93233082706767e-06,
+      "loss": 0.0001,
+      "step": 8635
+    },
+    {
+      "epoch": 0.5565508796803506,
+      "grad_norm": 0.671429575159169,
+      "learning_rate": 4.931614751163623e-06,
+      "loss": 0.0026,
+      "step": 8636
+    },
+    {
+      "epoch": 0.5566153251272797,
+      "grad_norm": 0.0023686506544780763,
+      "learning_rate": 4.9308986752595775e-06,
+      "loss": 0.0,
+      "step": 8637
+    },
+    {
+      "epoch": 0.5566797705742089,
+      "grad_norm": 0.0010374830972492118,
+      "learning_rate": 4.930182599355532e-06,
+      "loss": 0.0,
+      "step": 8638
+    },
+    {
+      "epoch": 0.5567442160211381,
+      "grad_norm": 0.019772198207453226,
+      "learning_rate": 4.929466523451486e-06,
+      "loss": 0.0002,
+      "step": 8639
+    },
+    {
+      "epoch": 0.5568086614680673,
+      "grad_norm": 0.006079940266471607,
+      "learning_rate": 4.9287504475474404e-06,
+      "loss": 0.0,
+      "step": 8640
+    },
+    {
+      "epoch": 0.5568731069149965,
+      "grad_norm": 0.022085923738554603,
+      "learning_rate": 4.928034371643395e-06,
+      "loss": 0.0001,
+      "step": 8641
+    },
+    {
+      "epoch": 0.5569375523619257,
+      "grad_norm": 3.3427854198404874,
+      "learning_rate": 4.927318295739349e-06,
+      "loss": 0.0662,
+      "step": 8642
+    },
+    {
+      "epoch": 0.5570019978088548,
+      "grad_norm": 0.01589109570954559,
+      "learning_rate": 4.926602219835303e-06,
+      "loss": 0.0,
+      "step": 8643
+    },
+    {
+      "epoch": 0.557066443255784,
+      "grad_norm": 0.0008106748129718368,
+      "learning_rate": 4.925886143931257e-06,
+      "loss": 0.0,
+      "step": 8644
+    },
+    {
+      "epoch": 0.5571308887027131,
+      "grad_norm": 0.06000058737916828,
+      "learning_rate": 4.925170068027211e-06,
+      "loss": 0.0038,
+      "step": 8645
+    },
+    {
+      "epoch": 0.5571953341496423,
+      "grad_norm": 0.0015374185399924462,
+      "learning_rate": 4.924453992123165e-06,
+      "loss": 0.0,
+      "step": 8646
+    },
+    {
+      "epoch": 0.5572597795965715,
+      "grad_norm": 0.001674477621343504,
+      "learning_rate": 4.92373791621912e-06,
+      "loss": 0.0,
+      "step": 8647
+    },
+    {
+      "epoch": 0.5573242250435007,
+      "grad_norm": 0.0012146313544239115,
+      "learning_rate": 4.923021840315074e-06,
+      "loss": 0.0,
+      "step": 8648
+    },
+    {
+      "epoch": 0.5573886704904298,
+      "grad_norm": 0.012782928730319724,
+      "learning_rate": 4.922305764411027e-06,
+      "loss": 0.0001,
+      "step": 8649
+    },
+    {
+      "epoch": 0.557453115937359,
+      "grad_norm": 0.07260694958982807,
+      "learning_rate": 4.921589688506982e-06,
+      "loss": 0.001,
+      "step": 8650
+    },
+    {
+      "epoch": 0.5575175613842882,
+      "grad_norm": 0.004229980046474155,
+      "learning_rate": 4.920873612602936e-06,
+      "loss": 0.0,
+      "step": 8651
+    },
+    {
+      "epoch": 0.5575820068312174,
+      "grad_norm": 0.09238504624841795,
+      "learning_rate": 4.92015753669889e-06,
+      "loss": 0.0002,
+      "step": 8652
+    },
+    {
+      "epoch": 0.5576464522781466,
+      "grad_norm": 0.019205153148170562,
+      "learning_rate": 4.919441460794845e-06,
+      "loss": 0.0001,
+      "step": 8653
+    },
+    {
+      "epoch": 0.5577108977250758,
+      "grad_norm": 0.004936052066062885,
+      "learning_rate": 4.918725384890799e-06,
+      "loss": 0.0,
+      "step": 8654
+    },
+    {
+      "epoch": 0.557775343172005,
+      "grad_norm": 0.006712840825438057,
+      "learning_rate": 4.918009308986753e-06,
+      "loss": 0.0,
+      "step": 8655
+    },
+    {
+      "epoch": 0.557839788618934,
+      "grad_norm": 0.002762241064491735,
+      "learning_rate": 4.9172932330827075e-06,
+      "loss": 0.0,
+      "step": 8656
+    },
+    {
+      "epoch": 0.5579042340658632,
+      "grad_norm": 0.01698880984180648,
+      "learning_rate": 4.916577157178661e-06,
+      "loss": 0.0001,
+      "step": 8657
+    },
+    {
+      "epoch": 0.5579686795127924,
+      "grad_norm": 0.00016716416716050216,
+      "learning_rate": 4.915861081274615e-06,
+      "loss": 0.0,
+      "step": 8658
+    },
+    {
+      "epoch": 0.5580331249597216,
+      "grad_norm": 0.07974039068577432,
+      "learning_rate": 4.9151450053705696e-06,
+      "loss": 0.0002,
+      "step": 8659
+    },
+    {
+      "epoch": 0.5580975704066508,
+      "grad_norm": 0.013269370565432554,
+      "learning_rate": 4.914428929466524e-06,
+      "loss": 0.0,
+      "step": 8660
+    },
+    {
+      "epoch": 0.5581620158535799,
+      "grad_norm": 0.005166149279363144,
+      "learning_rate": 4.913712853562478e-06,
+      "loss": 0.0,
+      "step": 8661
+    },
+    {
+      "epoch": 0.5582264613005091,
+      "grad_norm": 0.003897579659758297,
+      "learning_rate": 4.912996777658432e-06,
+      "loss": 0.0,
+      "step": 8662
+    },
+    {
+      "epoch": 0.5582909067474383,
+      "grad_norm": 0.006220268919483848,
+      "learning_rate": 4.912280701754386e-06,
+      "loss": 0.0,
+      "step": 8663
+    },
+    {
+      "epoch": 0.5583553521943675,
+      "grad_norm": 0.005305744753739531,
+      "learning_rate": 4.911564625850341e-06,
+      "loss": 0.0,
+      "step": 8664
+    },
+    {
+      "epoch": 0.5584197976412967,
+      "grad_norm": 0.0005758029329325327,
+      "learning_rate": 4.9108485499462945e-06,
+      "loss": 0.0,
+      "step": 8665
+    },
+    {
+      "epoch": 0.5584842430882259,
+      "grad_norm": 0.000827279042110068,
+      "learning_rate": 4.910132474042249e-06,
+      "loss": 0.0,
+      "step": 8666
+    },
+    {
+      "epoch": 0.558548688535155,
+      "grad_norm": 0.004102911737880749,
+      "learning_rate": 4.909416398138203e-06,
+      "loss": 0.0,
+      "step": 8667
+    },
+    {
+      "epoch": 0.5586131339820841,
+      "grad_norm": 0.0004710363920028894,
+      "learning_rate": 4.908700322234157e-06,
+      "loss": 0.0,
+      "step": 8668
+    },
+    {
+      "epoch": 0.5586775794290133,
+      "grad_norm": 0.00015958688110151233,
+      "learning_rate": 4.907984246330112e-06,
+      "loss": 0.0,
+      "step": 8669
+    },
+    {
+      "epoch": 0.5587420248759425,
+      "grad_norm": 0.02376513779259615,
+      "learning_rate": 4.907268170426065e-06,
+      "loss": 0.0,
+      "step": 8670
+    },
+    {
+      "epoch": 0.5588064703228717,
+      "grad_norm": 0.002388357422968757,
+      "learning_rate": 4.9065520945220194e-06,
+      "loss": 0.0,
+      "step": 8671
+    },
+    {
+      "epoch": 0.5588709157698009,
+      "grad_norm": 0.0016862129693133454,
+      "learning_rate": 4.905836018617974e-06,
+      "loss": 0.0,
+      "step": 8672
+    },
+    {
+      "epoch": 0.55893536121673,
+      "grad_norm": 0.004936221946703141,
+      "learning_rate": 4.905119942713928e-06,
+      "loss": 0.0,
+      "step": 8673
+    },
+    {
+      "epoch": 0.5589998066636592,
+      "grad_norm": 0.0015385476593219008,
+      "learning_rate": 4.904403866809882e-06,
+      "loss": 0.0,
+      "step": 8674
+    },
+    {
+      "epoch": 0.5590642521105884,
+      "grad_norm": 0.09754154621517012,
+      "learning_rate": 4.903687790905837e-06,
+      "loss": 0.0013,
+      "step": 8675
+    },
+    {
+      "epoch": 0.5591286975575176,
+      "grad_norm": 0.0025453102097556697,
+      "learning_rate": 4.902971715001791e-06,
+      "loss": 0.0,
+      "step": 8676
+    },
+    {
+      "epoch": 0.5591931430044468,
+      "grad_norm": 0.000572278693487006,
+      "learning_rate": 4.902255639097745e-06,
+      "loss": 0.0,
+      "step": 8677
+    },
+    {
+      "epoch": 0.559257588451376,
+      "grad_norm": 3.171789459428063,
+      "learning_rate": 4.901539563193699e-06,
+      "loss": 0.0115,
+      "step": 8678
+    },
+    {
+      "epoch": 0.559322033898305,
+      "grad_norm": 0.005036475146354532,
+      "learning_rate": 4.900823487289653e-06,
+      "loss": 0.0,
+      "step": 8679
+    },
+    {
+      "epoch": 0.5593864793452342,
+      "grad_norm": 0.039560586102517295,
+      "learning_rate": 4.900107411385607e-06,
+      "loss": 0.0001,
+      "step": 8680
+    },
+    {
+      "epoch": 0.5594509247921634,
+      "grad_norm": 0.1343451450139708,
+      "learning_rate": 4.899391335481562e-06,
+      "loss": 0.0006,
+      "step": 8681
+    },
+    {
+      "epoch": 0.5595153702390926,
+      "grad_norm": 0.012012192767179632,
+      "learning_rate": 4.898675259577516e-06,
+      "loss": 0.0001,
+      "step": 8682
+    },
+    {
+      "epoch": 0.5595798156860218,
+      "grad_norm": 0.002232516437988932,
+      "learning_rate": 4.897959183673469e-06,
+      "loss": 0.0,
+      "step": 8683
+    },
+    {
+      "epoch": 0.559644261132951,
+      "grad_norm": 0.0005160028869192366,
+      "learning_rate": 4.897243107769424e-06,
+      "loss": 0.0,
+      "step": 8684
+    },
+    {
+      "epoch": 0.5597087065798801,
+      "grad_norm": 0.17146186593991028,
+      "learning_rate": 4.896527031865378e-06,
+      "loss": 0.001,
+      "step": 8685
+    },
+    {
+      "epoch": 0.5597731520268093,
+      "grad_norm": 0.24822877236891366,
+      "learning_rate": 4.895810955961332e-06,
+      "loss": 0.0006,
+      "step": 8686
+    },
+    {
+      "epoch": 0.5598375974737385,
+      "grad_norm": 0.039314318138136045,
+      "learning_rate": 4.8950948800572865e-06,
+      "loss": 0.0004,
+      "step": 8687
+    },
+    {
+      "epoch": 0.5599020429206677,
+      "grad_norm": 0.8952721272098672,
+      "learning_rate": 4.894378804153241e-06,
+      "loss": 0.0025,
+      "step": 8688
+    },
+    {
+      "epoch": 0.5599664883675969,
+      "grad_norm": 0.000244431402831098,
+      "learning_rate": 4.893662728249195e-06,
+      "loss": 0.0,
+      "step": 8689
+    },
+    {
+      "epoch": 0.560030933814526,
+      "grad_norm": 0.003176688002738185,
+      "learning_rate": 4.8929466523451494e-06,
+      "loss": 0.0,
+      "step": 8690
+    },
+    {
+      "epoch": 0.5600953792614551,
+      "grad_norm": 0.0003265274312593406,
+      "learning_rate": 4.892230576441103e-06,
+      "loss": 0.0,
+      "step": 8691
+    },
+    {
+      "epoch": 0.5601598247083843,
+      "grad_norm": 0.0016430742264626347,
+      "learning_rate": 4.891514500537057e-06,
+      "loss": 0.0,
+      "step": 8692
+    },
+    {
+      "epoch": 0.5602242701553135,
+      "grad_norm": 0.0009334472852909937,
+      "learning_rate": 4.8907984246330115e-06,
+      "loss": 0.0,
+      "step": 8693
+    },
+    {
+      "epoch": 0.5602887156022427,
+      "grad_norm": 7.442905227648027e-05,
+      "learning_rate": 4.890082348728966e-06,
+      "loss": 0.0,
+      "step": 8694
+    },
+    {
+      "epoch": 0.5603531610491719,
+      "grad_norm": 0.16791826420947753,
+      "learning_rate": 4.88936627282492e-06,
+      "loss": 0.0007,
+      "step": 8695
+    },
+    {
+      "epoch": 0.5604176064961011,
+      "grad_norm": 0.3082393107668559,
+      "learning_rate": 4.8886501969208735e-06,
+      "loss": 0.0007,
+      "step": 8696
+    },
+    {
+      "epoch": 0.5604820519430302,
+      "grad_norm": 0.00016087311292047117,
+      "learning_rate": 4.887934121016828e-06,
+      "loss": 0.0,
+      "step": 8697
+    },
+    {
+      "epoch": 0.5605464973899594,
+      "grad_norm": 0.16002245291919298,
+      "learning_rate": 4.887218045112782e-06,
+      "loss": 0.0008,
+      "step": 8698
+    },
+    {
+      "epoch": 0.5606109428368886,
+      "grad_norm": 0.004248675914345726,
+      "learning_rate": 4.8865019692087364e-06,
+      "loss": 0.0,
+      "step": 8699
+    },
+    {
+      "epoch": 0.5606753882838178,
+      "grad_norm": 0.006593991119702283,
+      "learning_rate": 4.885785893304691e-06,
+      "loss": 0.0,
+      "step": 8700
+    },
+    {
+      "epoch": 0.5607398337307469,
+      "grad_norm": 0.32584103920711627,
+      "learning_rate": 4.885069817400645e-06,
+      "loss": 0.0014,
+      "step": 8701
+    },
+    {
+      "epoch": 0.5608042791776761,
+      "grad_norm": 0.0002263789195140541,
+      "learning_rate": 4.884353741496599e-06,
+      "loss": 0.0,
+      "step": 8702
+    },
+    {
+      "epoch": 0.5608687246246052,
+      "grad_norm": 0.7640283245408404,
+      "learning_rate": 4.883637665592554e-06,
+      "loss": 0.0059,
+      "step": 8703
+    },
+    {
+      "epoch": 0.5609331700715344,
+      "grad_norm": 7.212601342396701e-05,
+      "learning_rate": 4.882921589688507e-06,
+      "loss": 0.0,
+      "step": 8704
+    },
+    {
+      "epoch": 0.5609976155184636,
+      "grad_norm": 0.010067607272151306,
+      "learning_rate": 4.882205513784461e-06,
+      "loss": 0.0,
+      "step": 8705
+    },
+    {
+      "epoch": 0.5610620609653928,
+      "grad_norm": 0.010785156772366373,
+      "learning_rate": 4.881489437880416e-06,
+      "loss": 0.0001,
+      "step": 8706
+    },
+    {
+      "epoch": 0.561126506412322,
+      "grad_norm": 0.020743954743965017,
+      "learning_rate": 4.88077336197637e-06,
+      "loss": 0.0001,
+      "step": 8707
+    },
+    {
+      "epoch": 0.5611909518592512,
+      "grad_norm": 0.01784634922520059,
+      "learning_rate": 4.880057286072324e-06,
+      "loss": 0.0,
+      "step": 8708
+    },
+    {
+      "epoch": 0.5612553973061803,
+      "grad_norm": 0.030889615161481417,
+      "learning_rate": 4.879341210168278e-06,
+      "loss": 0.0001,
+      "step": 8709
+    },
+    {
+      "epoch": 0.5613198427531095,
+      "grad_norm": 0.08717037831813229,
+      "learning_rate": 4.878625134264233e-06,
+      "loss": 0.0001,
+      "step": 8710
+    },
+    {
+      "epoch": 0.5613842882000387,
+      "grad_norm": 4.0069996870915824e-05,
+      "learning_rate": 4.877909058360187e-06,
+      "loss": 0.0,
+      "step": 8711
+    },
+    {
+      "epoch": 0.5614487336469678,
+      "grad_norm": 0.21334305206683088,
+      "learning_rate": 4.877192982456141e-06,
+      "loss": 0.0004,
+      "step": 8712
+    },
+    {
+      "epoch": 0.561513179093897,
+      "grad_norm": 0.0009662374374246946,
+      "learning_rate": 4.876476906552095e-06,
+      "loss": 0.0,
+      "step": 8713
+    },
+    {
+      "epoch": 0.5615776245408262,
+      "grad_norm": 0.001994360683181997,
+      "learning_rate": 4.875760830648049e-06,
+      "loss": 0.0,
+      "step": 8714
+    },
+    {
+      "epoch": 0.5616420699877553,
+      "grad_norm": 0.0007313893970778114,
+      "learning_rate": 4.8750447547440035e-06,
+      "loss": 0.0,
+      "step": 8715
+    },
+    {
+      "epoch": 0.5617065154346845,
+      "grad_norm": 0.12489726270253314,
+      "learning_rate": 4.874328678839958e-06,
+      "loss": 0.0003,
+      "step": 8716
+    },
+    {
+      "epoch": 0.5617709608816137,
+      "grad_norm": 0.023161239719397465,
+      "learning_rate": 4.873612602935911e-06,
+      "loss": 0.0,
+      "step": 8717
+    },
+    {
+      "epoch": 0.5618354063285429,
+      "grad_norm": 0.12134647530902526,
+      "learning_rate": 4.8728965270318656e-06,
+      "loss": 0.0004,
+      "step": 8718
+    },
+    {
+      "epoch": 0.5618998517754721,
+      "grad_norm": 0.0005077160448647168,
+      "learning_rate": 4.87218045112782e-06,
+      "loss": 0.0,
+      "step": 8719
+    },
+    {
+      "epoch": 0.5619642972224013,
+      "grad_norm": 0.4305684855102253,
+      "learning_rate": 4.871464375223774e-06,
+      "loss": 0.0031,
+      "step": 8720
+    },
+    {
+      "epoch": 0.5620287426693304,
+      "grad_norm": 0.03279392093936876,
+      "learning_rate": 4.8707482993197285e-06,
+      "loss": 0.0001,
+      "step": 8721
+    },
+    {
+      "epoch": 0.5620931881162596,
+      "grad_norm": 0.012182210483527876,
+      "learning_rate": 4.870032223415683e-06,
+      "loss": 0.0001,
+      "step": 8722
+    },
+    {
+      "epoch": 0.5621576335631887,
+      "grad_norm": 0.01315964680711356,
+      "learning_rate": 4.869316147511637e-06,
+      "loss": 0.0,
+      "step": 8723
+    },
+    {
+      "epoch": 0.5622220790101179,
+      "grad_norm": 0.011839825426168998,
+      "learning_rate": 4.868600071607591e-06,
+      "loss": 0.0,
+      "step": 8724
+    },
+    {
+      "epoch": 0.5622865244570471,
+      "grad_norm": 0.003285467399204521,
+      "learning_rate": 4.867883995703545e-06,
+      "loss": 0.0,
+      "step": 8725
+    },
+    {
+      "epoch": 0.5623509699039763,
+      "grad_norm": 0.008998222780521375,
+      "learning_rate": 4.867167919799499e-06,
+      "loss": 0.0,
+      "step": 8726
+    },
+    {
+      "epoch": 0.5624154153509054,
+      "grad_norm": 0.006015495470261791,
+      "learning_rate": 4.866451843895453e-06,
+      "loss": 0.0,
+      "step": 8727
+    },
+    {
+      "epoch": 0.5624798607978346,
+      "grad_norm": 0.0036835391157751737,
+      "learning_rate": 4.865735767991408e-06,
+      "loss": 0.0,
+      "step": 8728
+    },
+    {
+      "epoch": 0.5625443062447638,
+      "grad_norm": 0.000797614702290517,
+      "learning_rate": 4.865019692087362e-06,
+      "loss": 0.0,
+      "step": 8729
+    },
+    {
+      "epoch": 0.562608751691693,
+      "grad_norm": 0.002094505358365688,
+      "learning_rate": 4.8643036161833154e-06,
+      "loss": 0.0,
+      "step": 8730
+    },
+    {
+      "epoch": 0.5626731971386222,
+      "grad_norm": 0.0011268998344081452,
+      "learning_rate": 4.86358754027927e-06,
+      "loss": 0.0,
+      "step": 8731
+    },
+    {
+      "epoch": 0.5627376425855514,
+      "grad_norm": 0.0005851215754860948,
+      "learning_rate": 4.862871464375224e-06,
+      "loss": 0.0,
+      "step": 8732
+    },
+    {
+      "epoch": 0.5628020880324806,
+      "grad_norm": 0.002780093783662651,
+      "learning_rate": 4.862155388471178e-06,
+      "loss": 0.0,
+      "step": 8733
+    },
+    {
+      "epoch": 0.5628665334794096,
+      "grad_norm": 0.0036051815660962795,
+      "learning_rate": 4.861439312567133e-06,
+      "loss": 0.0,
+      "step": 8734
+    },
+    {
+      "epoch": 0.5629309789263388,
+      "grad_norm": 0.17071126224493394,
+      "learning_rate": 4.860723236663087e-06,
+      "loss": 0.0007,
+      "step": 8735
+    },
+    {
+      "epoch": 0.562995424373268,
+      "grad_norm": 0.0009409020897952011,
+      "learning_rate": 4.860007160759041e-06,
+      "loss": 0.0,
+      "step": 8736
+    },
+    {
+      "epoch": 0.5630598698201972,
+      "grad_norm": 0.017214624380895726,
+      "learning_rate": 4.8592910848549955e-06,
+      "loss": 0.0002,
+      "step": 8737
+    },
+    {
+      "epoch": 0.5631243152671264,
+      "grad_norm": 0.0001412999669650116,
+      "learning_rate": 4.858575008950949e-06,
+      "loss": 0.0,
+      "step": 8738
+    },
+    {
+      "epoch": 0.5631887607140555,
+      "grad_norm": 0.5667308534939416,
+      "learning_rate": 4.857858933046903e-06,
+      "loss": 0.0017,
+      "step": 8739
+    },
+    {
+      "epoch": 0.5632532061609847,
+      "grad_norm": 0.0006867199061076417,
+      "learning_rate": 4.857142857142858e-06,
+      "loss": 0.0,
+      "step": 8740
+    },
+    {
+      "epoch": 0.5633176516079139,
+      "grad_norm": 0.0001116138075197791,
+      "learning_rate": 4.856426781238812e-06,
+      "loss": 0.0,
+      "step": 8741
+    },
+    {
+      "epoch": 0.5633820970548431,
+      "grad_norm": 0.23073332358003387,
+      "learning_rate": 4.855710705334765e-06,
+      "loss": 0.0003,
+      "step": 8742
+    },
+    {
+      "epoch": 0.5634465425017723,
+      "grad_norm": 0.0007989920972693515,
+      "learning_rate": 4.85499462943072e-06,
+      "loss": 0.0,
+      "step": 8743
+    },
+    {
+      "epoch": 0.5635109879487015,
+      "grad_norm": 0.001161135735589687,
+      "learning_rate": 4.854278553526674e-06,
+      "loss": 0.0,
+      "step": 8744
+    },
+    {
+      "epoch": 0.5635754333956307,
+      "grad_norm": 0.019831510081458233,
+      "learning_rate": 4.853562477622629e-06,
+      "loss": 0.0001,
+      "step": 8745
+    },
+    {
+      "epoch": 0.5636398788425597,
+      "grad_norm": 0.0007333301662223194,
+      "learning_rate": 4.8528464017185825e-06,
+      "loss": 0.0,
+      "step": 8746
+    },
+    {
+      "epoch": 0.5637043242894889,
+      "grad_norm": 0.0004206232000712696,
+      "learning_rate": 4.852130325814537e-06,
+      "loss": 0.0,
+      "step": 8747
+    },
+    {
+      "epoch": 0.5637687697364181,
+      "grad_norm": 0.00020189099769625215,
+      "learning_rate": 4.851414249910491e-06,
+      "loss": 0.0,
+      "step": 8748
+    },
+    {
+      "epoch": 0.5638332151833473,
+      "grad_norm": 0.23797480623700412,
+      "learning_rate": 4.8506981740064454e-06,
+      "loss": 0.0018,
+      "step": 8749
+    },
+    {
+      "epoch": 0.5638976606302765,
+      "grad_norm": 0.002936163898924804,
+      "learning_rate": 4.849982098102399e-06,
+      "loss": 0.0,
+      "step": 8750
+    },
+    {
+      "epoch": 0.5639621060772056,
+      "grad_norm": 0.001416732870031649,
+      "learning_rate": 4.849266022198353e-06,
+      "loss": 0.0,
+      "step": 8751
+    },
+    {
+      "epoch": 0.5640265515241348,
+      "grad_norm": 0.0005018175610769106,
+      "learning_rate": 4.8485499462943075e-06,
+      "loss": 0.0,
+      "step": 8752
+    },
+    {
+      "epoch": 0.564090996971064,
+      "grad_norm": 0.20247256512586775,
+      "learning_rate": 4.847833870390262e-06,
+      "loss": 0.004,
+      "step": 8753
+    },
+    {
+      "epoch": 0.5641554424179932,
+      "grad_norm": 0.004722470804156485,
+      "learning_rate": 4.847117794486216e-06,
+      "loss": 0.0001,
+      "step": 8754
+    },
+    {
+      "epoch": 0.5642198878649224,
+      "grad_norm": 0.0189753211178013,
+      "learning_rate": 4.8464017185821695e-06,
+      "loss": 0.0,
+      "step": 8755
+    },
+    {
+      "epoch": 0.5642843333118516,
+      "grad_norm": 0.1526467964911563,
+      "learning_rate": 4.845685642678124e-06,
+      "loss": 0.0005,
+      "step": 8756
+    },
+    {
+      "epoch": 0.5643487787587806,
+      "grad_norm": 0.002193407182671908,
+      "learning_rate": 4.844969566774079e-06,
+      "loss": 0.0,
+      "step": 8757
+    },
+    {
+      "epoch": 0.5644132242057098,
+      "grad_norm": 0.0024595912356725554,
+      "learning_rate": 4.8442534908700324e-06,
+      "loss": 0.0,
+      "step": 8758
+    },
+    {
+      "epoch": 0.564477669652639,
+      "grad_norm": 0.16334430991807178,
+      "learning_rate": 4.843537414965987e-06,
+      "loss": 0.0001,
+      "step": 8759
+    },
+    {
+      "epoch": 0.5645421150995682,
+      "grad_norm": 6.383292916162537e-05,
+      "learning_rate": 4.842821339061941e-06,
+      "loss": 0.0,
+      "step": 8760
+    },
+    {
+      "epoch": 0.5646065605464974,
+      "grad_norm": 0.0006328202426825101,
+      "learning_rate": 4.842105263157895e-06,
+      "loss": 0.0,
+      "step": 8761
+    },
+    {
+      "epoch": 0.5646710059934266,
+      "grad_norm": 0.0036656038926473366,
+      "learning_rate": 4.84138918725385e-06,
+      "loss": 0.0,
+      "step": 8762
+    },
+    {
+      "epoch": 0.5647354514403558,
+      "grad_norm": 0.00020467357590308498,
+      "learning_rate": 4.840673111349803e-06,
+      "loss": 0.0,
+      "step": 8763
+    },
+    {
+      "epoch": 0.5647998968872849,
+      "grad_norm": 0.00020467357590308498,
+      "learning_rate": 4.840673111349803e-06,
+      "loss": 0.0068,
+      "step": 8764
+    },
+    {
+      "epoch": 0.5648643423342141,
+      "grad_norm": 0.3235758531595586,
+      "learning_rate": 4.839957035445757e-06,
+      "loss": 0.0026,
+      "step": 8765
+    },
+    {
+      "epoch": 0.5649287877811433,
+      "grad_norm": 0.004006022682251447,
+      "learning_rate": 4.839240959541712e-06,
+      "loss": 0.0,
+      "step": 8766
+    },
+    {
+      "epoch": 0.5649932332280725,
+      "grad_norm": 0.004125043265111191,
+      "learning_rate": 4.838524883637666e-06,
+      "loss": 0.0,
+      "step": 8767
+    },
+    {
+      "epoch": 0.5650576786750016,
+      "grad_norm": 0.5681165458293576,
+      "learning_rate": 4.83780880773362e-06,
+      "loss": 0.003,
+      "step": 8768
+    },
+    {
+      "epoch": 0.5651221241219307,
+      "grad_norm": 0.007982689949264836,
+      "learning_rate": 4.837092731829574e-06,
+      "loss": 0.0,
+      "step": 8769
+    },
+    {
+      "epoch": 0.5651865695688599,
+      "grad_norm": 0.006878737484773677,
+      "learning_rate": 4.836376655925529e-06,
+      "loss": 0.0001,
+      "step": 8770
+    },
+    {
+      "epoch": 0.5652510150157891,
+      "grad_norm": 0.0037331182480409924,
+      "learning_rate": 4.835660580021483e-06,
+      "loss": 0.0,
+      "step": 8771
+    },
+    {
+      "epoch": 0.5653154604627183,
+      "grad_norm": 0.2563085617045643,
+      "learning_rate": 4.834944504117437e-06,
+      "loss": 0.0004,
+      "step": 8772
+    },
+    {
+      "epoch": 0.5653799059096475,
+      "grad_norm": 0.009190612379956239,
+      "learning_rate": 4.834228428213391e-06,
+      "loss": 0.0001,
+      "step": 8773
+    },
+    {
+      "epoch": 0.5654443513565767,
+      "grad_norm": 0.006375923284434245,
+      "learning_rate": 4.833512352309345e-06,
+      "loss": 0.0,
+      "step": 8774
+    },
+    {
+      "epoch": 0.5655087968035059,
+      "grad_norm": 0.796118919865634,
+      "learning_rate": 4.8327962764052995e-06,
+      "loss": 0.0015,
+      "step": 8775
+    },
+    {
+      "epoch": 0.565573242250435,
+      "grad_norm": 0.00342534803484227,
+      "learning_rate": 4.832080200501254e-06,
+      "loss": 0.0,
+      "step": 8776
+    },
+    {
+      "epoch": 0.5656376876973642,
+      "grad_norm": 0.0312884269577183,
+      "learning_rate": 4.831364124597207e-06,
+      "loss": 0.0016,
+      "step": 8777
+    },
+    {
+      "epoch": 0.5657021331442934,
+      "grad_norm": 0.051568192290613435,
+      "learning_rate": 4.8306480486931616e-06,
+      "loss": 0.0005,
+      "step": 8778
+    },
+    {
+      "epoch": 0.5657665785912225,
+      "grad_norm": 0.0001771488648946319,
+      "learning_rate": 4.829931972789116e-06,
+      "loss": 0.0,
+      "step": 8779
+    },
+    {
+      "epoch": 0.5658310240381517,
+      "grad_norm": 0.005487712499418206,
+      "learning_rate": 4.82921589688507e-06,
+      "loss": 0.0001,
+      "step": 8780
+    },
+    {
+      "epoch": 0.5658954694850808,
+      "grad_norm": 0.001248246749537538,
+      "learning_rate": 4.8284998209810245e-06,
+      "loss": 0.0,
+      "step": 8781
+    },
+    {
+      "epoch": 0.56595991493201,
+      "grad_norm": 0.009656013818239738,
+      "learning_rate": 4.827783745076979e-06,
+      "loss": 0.0,
+      "step": 8782
+    },
+    {
+      "epoch": 0.5660243603789392,
+      "grad_norm": 0.001758210468587272,
+      "learning_rate": 4.827067669172933e-06,
+      "loss": 0.0,
+      "step": 8783
+    },
+    {
+      "epoch": 0.5660888058258684,
+      "grad_norm": 0.3298186687581501,
+      "learning_rate": 4.826351593268887e-06,
+      "loss": 0.0013,
+      "step": 8784
+    },
+    {
+      "epoch": 0.5661532512727976,
+      "grad_norm": 1.0338854780988707,
+      "learning_rate": 4.825635517364841e-06,
+      "loss": 0.0159,
+      "step": 8785
+    },
+    {
+      "epoch": 0.5662176967197268,
+      "grad_norm": 0.013471690213689116,
+      "learning_rate": 4.824919441460795e-06,
+      "loss": 0.0,
+      "step": 8786
+    },
+    {
+      "epoch": 0.566282142166656,
+      "grad_norm": 0.00011633145196054886,
+      "learning_rate": 4.824203365556749e-06,
+      "loss": 0.0,
+      "step": 8787
+    },
+    {
+      "epoch": 0.5663465876135851,
+      "grad_norm": 0.005370732892713071,
+      "learning_rate": 4.823487289652704e-06,
+      "loss": 0.0,
+      "step": 8788
+    },
+    {
+      "epoch": 0.5664110330605143,
+      "grad_norm": 1.9052808205184348,
+      "learning_rate": 4.822771213748658e-06,
+      "loss": 0.0155,
+      "step": 8789
+    },
+    {
+      "epoch": 0.5664754785074434,
+      "grad_norm": 0.0016993295556823635,
+      "learning_rate": 4.8220551378446114e-06,
+      "loss": 0.0,
+      "step": 8790
+    },
+    {
+      "epoch": 0.5665399239543726,
+      "grad_norm": 0.21898209656732134,
+      "learning_rate": 4.821339061940566e-06,
+      "loss": 0.0008,
+      "step": 8791
+    },
+    {
+      "epoch": 0.5666043694013018,
+      "grad_norm": 0.0029171804599055747,
+      "learning_rate": 4.82062298603652e-06,
+      "loss": 0.0,
+      "step": 8792
+    },
+    {
+      "epoch": 0.566668814848231,
+      "grad_norm": 0.0011682263683127836,
+      "learning_rate": 4.819906910132474e-06,
+      "loss": 0.0,
+      "step": 8793
+    },
+    {
+      "epoch": 0.5667332602951601,
+      "grad_norm": 0.02120118758689934,
+      "learning_rate": 4.819190834228429e-06,
+      "loss": 0.0002,
+      "step": 8794
+    },
+    {
+      "epoch": 0.5667977057420893,
+      "grad_norm": 0.016868041783098704,
+      "learning_rate": 4.818474758324383e-06,
+      "loss": 0.0001,
+      "step": 8795
+    },
+    {
+      "epoch": 0.5668621511890185,
+      "grad_norm": 0.0024044416183440355,
+      "learning_rate": 4.817758682420337e-06,
+      "loss": 0.0,
+      "step": 8796
+    },
+    {
+      "epoch": 0.5669265966359477,
+      "grad_norm": 1.069360351283807,
+      "learning_rate": 4.8170426065162915e-06,
+      "loss": 0.0072,
+      "step": 8797
+    },
+    {
+      "epoch": 0.5669910420828769,
+      "grad_norm": 0.012245140194462465,
+      "learning_rate": 4.816326530612245e-06,
+      "loss": 0.0,
+      "step": 8798
+    },
+    {
+      "epoch": 0.5670554875298061,
+      "grad_norm": 0.00043831334825429955,
+      "learning_rate": 4.815610454708199e-06,
+      "loss": 0.0,
+      "step": 8799
+    },
+    {
+      "epoch": 0.5671199329767352,
+      "grad_norm": 0.0032662984748576328,
+      "learning_rate": 4.814894378804154e-06,
+      "loss": 0.0,
+      "step": 8800
+    },
+    {
+      "epoch": 0.5671843784236643,
+      "grad_norm": 0.06767947089195707,
+      "learning_rate": 4.814178302900108e-06,
+      "loss": 0.0,
+      "step": 8801
+    },
+    {
+      "epoch": 0.5672488238705935,
+      "grad_norm": 0.040585778805621656,
+      "learning_rate": 4.813462226996062e-06,
+      "loss": 0.0001,
+      "step": 8802
+    },
+    {
+      "epoch": 0.5673132693175227,
+      "grad_norm": 0.0005241659708540031,
+      "learning_rate": 4.812746151092016e-06,
+      "loss": 0.0,
+      "step": 8803
+    },
+    {
+      "epoch": 0.5673777147644519,
+      "grad_norm": 0.06185363157920903,
+      "learning_rate": 4.81203007518797e-06,
+      "loss": 0.0002,
+      "step": 8804
+    },
+    {
+      "epoch": 0.567442160211381,
+      "grad_norm": 0.012243238039889377,
+      "learning_rate": 4.811313999283925e-06,
+      "loss": 0.0,
+      "step": 8805
+    },
+    {
+      "epoch": 0.5675066056583102,
+      "grad_norm": 0.04625774070281256,
+      "learning_rate": 4.8105979233798785e-06,
+      "loss": 0.0003,
+      "step": 8806
+    },
+    {
+      "epoch": 0.5675710511052394,
+      "grad_norm": 0.00486547121355555,
+      "learning_rate": 4.809881847475833e-06,
+      "loss": 0.0,
+      "step": 8807
+    },
+    {
+      "epoch": 0.5676354965521686,
+      "grad_norm": 0.0404367972194789,
+      "learning_rate": 4.809165771571787e-06,
+      "loss": 0.0001,
+      "step": 8808
+    },
+    {
+      "epoch": 0.5676999419990978,
+      "grad_norm": 0.21987166805627087,
+      "learning_rate": 4.8084496956677414e-06,
+      "loss": 0.0002,
+      "step": 8809
+    },
+    {
+      "epoch": 0.567764387446027,
+      "grad_norm": 0.00850806290664799,
+      "learning_rate": 4.807733619763696e-06,
+      "loss": 0.0,
+      "step": 8810
+    },
+    {
+      "epoch": 0.5678288328929562,
+      "grad_norm": 0.046005889230199014,
+      "learning_rate": 4.807017543859649e-06,
+      "loss": 0.0002,
+      "step": 8811
+    },
+    {
+      "epoch": 0.5678932783398852,
+      "grad_norm": 0.33100387723458397,
+      "learning_rate": 4.8063014679556035e-06,
+      "loss": 0.0003,
+      "step": 8812
+    },
+    {
+      "epoch": 0.5679577237868144,
+      "grad_norm": 0.11855455379698503,
+      "learning_rate": 4.805585392051558e-06,
+      "loss": 0.0001,
+      "step": 8813
+    },
+    {
+      "epoch": 0.5680221692337436,
+      "grad_norm": 0.016615890357752884,
+      "learning_rate": 4.804869316147512e-06,
+      "loss": 0.0,
+      "step": 8814
+    },
+    {
+      "epoch": 0.5680866146806728,
+      "grad_norm": 0.669933129930625,
+      "learning_rate": 4.804153240243466e-06,
+      "loss": 0.0017,
+      "step": 8815
+    },
+    {
+      "epoch": 0.568151060127602,
+      "grad_norm": 0.3022767576768994,
+      "learning_rate": 4.803437164339421e-06,
+      "loss": 0.0009,
+      "step": 8816
+    },
+    {
+      "epoch": 0.5682155055745312,
+      "grad_norm": 0.09943305098401042,
+      "learning_rate": 4.802721088435375e-06,
+      "loss": 0.0001,
+      "step": 8817
+    },
+    {
+      "epoch": 0.5682799510214603,
+      "grad_norm": 0.0142782273969829,
+      "learning_rate": 4.802005012531329e-06,
+      "loss": 0.0,
+      "step": 8818
+    },
+    {
+      "epoch": 0.5683443964683895,
+      "grad_norm": 0.23304219371967497,
+      "learning_rate": 4.801288936627283e-06,
+      "loss": 0.0008,
+      "step": 8819
+    },
+    {
+      "epoch": 0.5684088419153187,
+      "grad_norm": 0.02275632003333956,
+      "learning_rate": 4.800572860723237e-06,
+      "loss": 0.0,
+      "step": 8820
+    },
+    {
+      "epoch": 0.5684732873622479,
+      "grad_norm": 0.01564516688304527,
+      "learning_rate": 4.799856784819191e-06,
+      "loss": 0.0,
+      "step": 8821
+    },
+    {
+      "epoch": 0.5685377328091771,
+      "grad_norm": 0.001371723968120945,
+      "learning_rate": 4.799140708915146e-06,
+      "loss": 0.0,
+      "step": 8822
+    },
+    {
+      "epoch": 0.5686021782561063,
+      "grad_norm": 0.012505581892835328,
+      "learning_rate": 4.7984246330111e-06,
+      "loss": 0.0,
+      "step": 8823
+    },
+    {
+      "epoch": 0.5686666237030353,
+      "grad_norm": 0.008130933934212818,
+      "learning_rate": 4.797708557107053e-06,
+      "loss": 0.0,
+      "step": 8824
+    },
+    {
+      "epoch": 0.5687310691499645,
+      "grad_norm": 0.030915509771977535,
+      "learning_rate": 4.796992481203008e-06,
+      "loss": 0.0001,
+      "step": 8825
+    },
+    {
+      "epoch": 0.5687955145968937,
+      "grad_norm": 0.8509190779648602,
+      "learning_rate": 4.796276405298962e-06,
+      "loss": 0.0059,
+      "step": 8826
+    },
+    {
+      "epoch": 0.5688599600438229,
+      "grad_norm": 3.1909198740549973,
+      "learning_rate": 4.795560329394916e-06,
+      "loss": 0.0556,
+      "step": 8827
+    },
+    {
+      "epoch": 0.5689244054907521,
+      "grad_norm": 0.23147505557006737,
+      "learning_rate": 4.7948442534908706e-06,
+      "loss": 0.0005,
+      "step": 8828
+    },
+    {
+      "epoch": 0.5689888509376813,
+      "grad_norm": 0.027221213537766676,
+      "learning_rate": 4.794128177586825e-06,
+      "loss": 0.0002,
+      "step": 8829
+    },
+    {
+      "epoch": 0.5690532963846104,
+      "grad_norm": 0.8506112691652553,
+      "learning_rate": 4.793412101682779e-06,
+      "loss": 0.003,
+      "step": 8830
+    },
+    {
+      "epoch": 0.5691177418315396,
+      "grad_norm": 0.5891452973962943,
+      "learning_rate": 4.7926960257787335e-06,
+      "loss": 0.0032,
+      "step": 8831
+    },
+    {
+      "epoch": 0.5691821872784688,
+      "grad_norm": 0.006008513606025011,
+      "learning_rate": 4.791979949874687e-06,
+      "loss": 0.0001,
+      "step": 8832
+    },
+    {
+      "epoch": 0.569246632725398,
+      "grad_norm": 0.0008348159052347049,
+      "learning_rate": 4.791263873970641e-06,
+      "loss": 0.0,
+      "step": 8833
+    },
+    {
+      "epoch": 0.5693110781723272,
+      "grad_norm": 0.019444771138941552,
+      "learning_rate": 4.7905477980665955e-06,
+      "loss": 0.0,
+      "step": 8834
+    },
+    {
+      "epoch": 0.5693755236192563,
+      "grad_norm": 0.06374701691057286,
+      "learning_rate": 4.78983172216255e-06,
+      "loss": 0.0002,
+      "step": 8835
+    },
+    {
+      "epoch": 0.5694399690661854,
+      "grad_norm": 0.004703231092509179,
+      "learning_rate": 4.789115646258503e-06,
+      "loss": 0.0001,
+      "step": 8836
+    },
+    {
+      "epoch": 0.5695044145131146,
+      "grad_norm": 0.001044352791871368,
+      "learning_rate": 4.7883995703544576e-06,
+      "loss": 0.0,
+      "step": 8837
+    },
+    {
+      "epoch": 0.5695688599600438,
+      "grad_norm": 0.0020537983650882334,
+      "learning_rate": 4.787683494450412e-06,
+      "loss": 0.0,
+      "step": 8838
+    },
+    {
+      "epoch": 0.569633305406973,
+      "grad_norm": 0.00620996218740712,
+      "learning_rate": 4.786967418546366e-06,
+      "loss": 0.0,
+      "step": 8839
+    },
+    {
+      "epoch": 0.5696977508539022,
+      "grad_norm": 0.005034392248727269,
+      "learning_rate": 4.7862513426423205e-06,
+      "loss": 0.0,
+      "step": 8840
+    },
+    {
+      "epoch": 0.5697621963008314,
+      "grad_norm": 0.004523870500418789,
+      "learning_rate": 4.785535266738275e-06,
+      "loss": 0.0,
+      "step": 8841
+    },
+    {
+      "epoch": 0.5698266417477605,
+      "grad_norm": 0.04456405299394487,
+      "learning_rate": 4.784819190834229e-06,
+      "loss": 0.0005,
+      "step": 8842
+    },
+    {
+      "epoch": 0.5698910871946897,
+      "grad_norm": 5.05333046130126e-05,
+      "learning_rate": 4.784103114930183e-06,
+      "loss": 0.0,
+      "step": 8843
+    },
+    {
+      "epoch": 0.5699555326416189,
+      "grad_norm": 0.000751981987644549,
+      "learning_rate": 4.783387039026137e-06,
+      "loss": 0.0,
+      "step": 8844
+    },
+    {
+      "epoch": 0.5700199780885481,
+      "grad_norm": 0.00312416821242081,
+      "learning_rate": 4.782670963122091e-06,
+      "loss": 0.0,
+      "step": 8845
+    },
+    {
+      "epoch": 0.5700844235354772,
+      "grad_norm": 0.0016447265891001916,
+      "learning_rate": 4.781954887218045e-06,
+      "loss": 0.0,
+      "step": 8846
+    },
+    {
+      "epoch": 0.5701488689824064,
+      "grad_norm": 0.005598044842178874,
+      "learning_rate": 4.781238811314e-06,
+      "loss": 0.0001,
+      "step": 8847
+    },
+    {
+      "epoch": 0.5702133144293355,
+      "grad_norm": 0.01762638982733986,
+      "learning_rate": 4.780522735409954e-06,
+      "loss": 0.0,
+      "step": 8848
+    },
+    {
+      "epoch": 0.5702777598762647,
+      "grad_norm": 0.010449044572078174,
+      "learning_rate": 4.7798066595059074e-06,
+      "loss": 0.0,
+      "step": 8849
+    },
+    {
+      "epoch": 0.5703422053231939,
+      "grad_norm": 0.0002665962713856165,
+      "learning_rate": 4.779090583601862e-06,
+      "loss": 0.0,
+      "step": 8850
+    },
+    {
+      "epoch": 0.5704066507701231,
+      "grad_norm": 0.06352423931488883,
+      "learning_rate": 4.778374507697817e-06,
+      "loss": 0.0007,
+      "step": 8851
+    },
+    {
+      "epoch": 0.5704710962170523,
+      "grad_norm": 0.38984084577376893,
+      "learning_rate": 4.77765843179377e-06,
+      "loss": 0.0017,
+      "step": 8852
+    },
+    {
+      "epoch": 0.5705355416639815,
+      "grad_norm": 0.3330555543209313,
+      "learning_rate": 4.776942355889725e-06,
+      "loss": 0.0025,
+      "step": 8853
+    },
+    {
+      "epoch": 0.5705999871109106,
+      "grad_norm": 0.0028222350733292994,
+      "learning_rate": 4.776226279985679e-06,
+      "loss": 0.0,
+      "step": 8854
+    },
+    {
+      "epoch": 0.5706644325578398,
+      "grad_norm": 0.001275800740573039,
+      "learning_rate": 4.775510204081633e-06,
+      "loss": 0.0,
+      "step": 8855
+    },
+    {
+      "epoch": 0.570728878004769,
+      "grad_norm": 0.0010790753775213522,
+      "learning_rate": 4.7747941281775875e-06,
+      "loss": 0.0,
+      "step": 8856
+    },
+    {
+      "epoch": 0.5707933234516981,
+      "grad_norm": 0.04619489207028863,
+      "learning_rate": 4.774078052273541e-06,
+      "loss": 0.0,
+      "step": 8857
+    },
+    {
+      "epoch": 0.5708577688986273,
+      "grad_norm": 0.020743763233173337,
+      "learning_rate": 4.773361976369495e-06,
+      "loss": 0.0001,
+      "step": 8858
+    },
+    {
+      "epoch": 0.5709222143455565,
+      "grad_norm": 0.0006698125467998965,
+      "learning_rate": 4.77264590046545e-06,
+      "loss": 0.0,
+      "step": 8859
+    },
+    {
+      "epoch": 0.5709866597924856,
+      "grad_norm": 0.0014080113945894896,
+      "learning_rate": 4.771929824561404e-06,
+      "loss": 0.0,
+      "step": 8860
+    },
+    {
+      "epoch": 0.5710511052394148,
+      "grad_norm": 0.27785739883434873,
+      "learning_rate": 4.771213748657358e-06,
+      "loss": 0.0003,
+      "step": 8861
+    },
+    {
+      "epoch": 0.571115550686344,
+      "grad_norm": 0.0027483904368733616,
+      "learning_rate": 4.770497672753312e-06,
+      "loss": 0.0,
+      "step": 8862
+    },
+    {
+      "epoch": 0.5711799961332732,
+      "grad_norm": 0.5726181350659572,
+      "learning_rate": 4.769781596849267e-06,
+      "loss": 0.002,
+      "step": 8863
+    },
+    {
+      "epoch": 0.5712444415802024,
+      "grad_norm": 0.00208850930236962,
+      "learning_rate": 4.769065520945221e-06,
+      "loss": 0.0,
+      "step": 8864
+    },
+    {
+      "epoch": 0.5713088870271316,
+      "grad_norm": 0.0028141424681147045,
+      "learning_rate": 4.7683494450411745e-06,
+      "loss": 0.0,
+      "step": 8865
+    },
+    {
+      "epoch": 0.5713733324740607,
+      "grad_norm": 0.05520224991306442,
+      "learning_rate": 4.767633369137129e-06,
+      "loss": 0.0001,
+      "step": 8866
+    },
+    {
+      "epoch": 0.5714377779209899,
+      "grad_norm": 0.0017825190273745063,
+      "learning_rate": 4.766917293233083e-06,
+      "loss": 0.0,
+      "step": 8867
+    },
+    {
+      "epoch": 0.571502223367919,
+      "grad_norm": 0.2973295170699638,
+      "learning_rate": 4.7662012173290374e-06,
+      "loss": 0.0032,
+      "step": 8868
+    },
+    {
+      "epoch": 0.5715666688148482,
+      "grad_norm": 0.004041302913021656,
+      "learning_rate": 4.765485141424992e-06,
+      "loss": 0.0,
+      "step": 8869
+    },
+    {
+      "epoch": 0.5716311142617774,
+      "grad_norm": 0.003423076298219085,
+      "learning_rate": 4.764769065520945e-06,
+      "loss": 0.0,
+      "step": 8870
+    },
+    {
+      "epoch": 0.5716955597087066,
+      "grad_norm": 0.001335184445791643,
+      "learning_rate": 4.7640529896168995e-06,
+      "loss": 0.0,
+      "step": 8871
+    },
+    {
+      "epoch": 0.5717600051556357,
+      "grad_norm": 0.006451075758302448,
+      "learning_rate": 4.763336913712854e-06,
+      "loss": 0.0,
+      "step": 8872
+    },
+    {
+      "epoch": 0.5718244506025649,
+      "grad_norm": 0.2841419575082844,
+      "learning_rate": 4.762620837808808e-06,
+      "loss": 0.0005,
+      "step": 8873
+    },
+    {
+      "epoch": 0.5718888960494941,
+      "grad_norm": 0.0009039202676003914,
+      "learning_rate": 4.761904761904762e-06,
+      "loss": 0.0,
+      "step": 8874
+    },
+    {
+      "epoch": 0.5719533414964233,
+      "grad_norm": 0.0006223950228870037,
+      "learning_rate": 4.761188686000717e-06,
+      "loss": 0.0,
+      "step": 8875
+    },
+    {
+      "epoch": 0.5720177869433525,
+      "grad_norm": 0.0020699808052104343,
+      "learning_rate": 4.760472610096671e-06,
+      "loss": 0.0,
+      "step": 8876
+    },
+    {
+      "epoch": 0.5720822323902817,
+      "grad_norm": 0.33732607430765,
+      "learning_rate": 4.759756534192625e-06,
+      "loss": 0.0015,
+      "step": 8877
+    },
+    {
+      "epoch": 0.5721466778372108,
+      "grad_norm": 0.005303730736312988,
+      "learning_rate": 4.759040458288579e-06,
+      "loss": 0.0,
+      "step": 8878
+    },
+    {
+      "epoch": 0.5722111232841399,
+      "grad_norm": 0.012692532555947952,
+      "learning_rate": 4.758324382384533e-06,
+      "loss": 0.0001,
+      "step": 8879
+    },
+    {
+      "epoch": 0.5722755687310691,
+      "grad_norm": 0.00019626527068705832,
+      "learning_rate": 4.757608306480487e-06,
+      "loss": 0.0,
+      "step": 8880
+    },
+    {
+      "epoch": 0.5723400141779983,
+      "grad_norm": 0.002095381014958743,
+      "learning_rate": 4.756892230576442e-06,
+      "loss": 0.0,
+      "step": 8881
+    },
+    {
+      "epoch": 0.5724044596249275,
+      "grad_norm": 0.1159781279800337,
+      "learning_rate": 4.756176154672396e-06,
+      "loss": 0.0001,
+      "step": 8882
+    },
+    {
+      "epoch": 0.5724689050718567,
+      "grad_norm": 0.01648991749511099,
+      "learning_rate": 4.755460078768349e-06,
+      "loss": 0.0,
+      "step": 8883
+    },
+    {
+      "epoch": 0.5725333505187858,
+      "grad_norm": 0.20105396444792106,
+      "learning_rate": 4.754744002864304e-06,
+      "loss": 0.0002,
+      "step": 8884
+    },
+    {
+      "epoch": 0.572597795965715,
+      "grad_norm": 0.17178626370759914,
+      "learning_rate": 4.754027926960258e-06,
+      "loss": 0.0011,
+      "step": 8885
+    },
+    {
+      "epoch": 0.5726622414126442,
+      "grad_norm": 0.010614898100775799,
+      "learning_rate": 4.753311851056212e-06,
+      "loss": 0.0001,
+      "step": 8886
+    },
+    {
+      "epoch": 0.5727266868595734,
+      "grad_norm": 0.001217281913593414,
+      "learning_rate": 4.7525957751521666e-06,
+      "loss": 0.0,
+      "step": 8887
+    },
+    {
+      "epoch": 0.5727911323065026,
+      "grad_norm": 0.0057027341480303745,
+      "learning_rate": 4.751879699248121e-06,
+      "loss": 0.0,
+      "step": 8888
+    },
+    {
+      "epoch": 0.5728555777534318,
+      "grad_norm": 0.8818495232939141,
+      "learning_rate": 4.751163623344075e-06,
+      "loss": 0.001,
+      "step": 8889
+    },
+    {
+      "epoch": 0.5729200232003608,
+      "grad_norm": 0.00742720014691036,
+      "learning_rate": 4.7504475474400295e-06,
+      "loss": 0.0001,
+      "step": 8890
+    },
+    {
+      "epoch": 0.57298446864729,
+      "grad_norm": 0.027254013778446516,
+      "learning_rate": 4.749731471535983e-06,
+      "loss": 0.0,
+      "step": 8891
+    },
+    {
+      "epoch": 0.5730489140942192,
+      "grad_norm": 0.03194781425699038,
+      "learning_rate": 4.749015395631937e-06,
+      "loss": 0.0,
+      "step": 8892
+    },
+    {
+      "epoch": 0.5731133595411484,
+      "grad_norm": 0.0026582619085988384,
+      "learning_rate": 4.7482993197278915e-06,
+      "loss": 0.0,
+      "step": 8893
+    },
+    {
+      "epoch": 0.5731778049880776,
+      "grad_norm": 0.127752705813193,
+      "learning_rate": 4.747583243823846e-06,
+      "loss": 0.0001,
+      "step": 8894
+    },
+    {
+      "epoch": 0.5732422504350068,
+      "grad_norm": 0.002056805451891573,
+      "learning_rate": 4.7468671679198e-06,
+      "loss": 0.0,
+      "step": 8895
+    },
+    {
+      "epoch": 0.573306695881936,
+      "grad_norm": 0.004686356561184799,
+      "learning_rate": 4.7461510920157536e-06,
+      "loss": 0.0,
+      "step": 8896
+    },
+    {
+      "epoch": 0.5733711413288651,
+      "grad_norm": 0.0038974480489338153,
+      "learning_rate": 4.745435016111708e-06,
+      "loss": 0.0,
+      "step": 8897
+    },
+    {
+      "epoch": 0.5734355867757943,
+      "grad_norm": 0.1602997195389966,
+      "learning_rate": 4.744718940207663e-06,
+      "loss": 0.0011,
+      "step": 8898
+    },
+    {
+      "epoch": 0.5735000322227235,
+      "grad_norm": 0.0026400329446987615,
+      "learning_rate": 4.7440028643036165e-06,
+      "loss": 0.0,
+      "step": 8899
+    },
+    {
+      "epoch": 0.5735644776696527,
+      "grad_norm": 0.003769746465072043,
+      "learning_rate": 4.743286788399571e-06,
+      "loss": 0.0,
+      "step": 8900
+    },
+    {
+      "epoch": 0.5736289231165819,
+      "grad_norm": 0.010710992003722329,
+      "learning_rate": 4.742570712495525e-06,
+      "loss": 0.0,
+      "step": 8901
+    },
+    {
+      "epoch": 0.5736933685635109,
+      "grad_norm": 0.0006582057125751989,
+      "learning_rate": 4.741854636591479e-06,
+      "loss": 0.0,
+      "step": 8902
+    },
+    {
+      "epoch": 0.5737578140104401,
+      "grad_norm": 0.0024639063269701047,
+      "learning_rate": 4.741138560687434e-06,
+      "loss": 0.0,
+      "step": 8903
+    },
+    {
+      "epoch": 0.5738222594573693,
+      "grad_norm": 0.8917694769321234,
+      "learning_rate": 4.740422484783387e-06,
+      "loss": 0.0007,
+      "step": 8904
+    },
+    {
+      "epoch": 0.5738867049042985,
+      "grad_norm": 0.0032197677652261165,
+      "learning_rate": 4.739706408879341e-06,
+      "loss": 0.0,
+      "step": 8905
+    },
+    {
+      "epoch": 0.5739511503512277,
+      "grad_norm": 0.22933696610797213,
+      "learning_rate": 4.738990332975296e-06,
+      "loss": 0.001,
+      "step": 8906
+    },
+    {
+      "epoch": 0.5740155957981569,
+      "grad_norm": 0.06729571615328032,
+      "learning_rate": 4.73827425707125e-06,
+      "loss": 0.0002,
+      "step": 8907
+    },
+    {
+      "epoch": 0.574080041245086,
+      "grad_norm": 0.19443921359345376,
+      "learning_rate": 4.737558181167204e-06,
+      "loss": 0.0006,
+      "step": 8908
+    },
+    {
+      "epoch": 0.5741444866920152,
+      "grad_norm": 0.007514377513820292,
+      "learning_rate": 4.736842105263158e-06,
+      "loss": 0.0,
+      "step": 8909
+    },
+    {
+      "epoch": 0.5742089321389444,
+      "grad_norm": 0.00025992034944802154,
+      "learning_rate": 4.736126029359113e-06,
+      "loss": 0.0,
+      "step": 8910
+    },
+    {
+      "epoch": 0.5742733775858736,
+      "grad_norm": 0.0007463775014227858,
+      "learning_rate": 4.735409953455067e-06,
+      "loss": 0.0,
+      "step": 8911
+    },
+    {
+      "epoch": 0.5743378230328028,
+      "grad_norm": 0.0008458128520093252,
+      "learning_rate": 4.734693877551021e-06,
+      "loss": 0.0,
+      "step": 8912
+    },
+    {
+      "epoch": 0.5744022684797319,
+      "grad_norm": 0.02946843736749207,
+      "learning_rate": 4.733977801646975e-06,
+      "loss": 0.0001,
+      "step": 8913
+    },
+    {
+      "epoch": 0.574466713926661,
+      "grad_norm": 0.0015844954372983143,
+      "learning_rate": 4.733261725742929e-06,
+      "loss": 0.0,
+      "step": 8914
+    },
+    {
+      "epoch": 0.5745311593735902,
+      "grad_norm": 0.0006650693208472371,
+      "learning_rate": 4.7325456498388835e-06,
+      "loss": 0.0,
+      "step": 8915
+    },
+    {
+      "epoch": 0.5745956048205194,
+      "grad_norm": 0.00024925103154753225,
+      "learning_rate": 4.731829573934838e-06,
+      "loss": 0.0,
+      "step": 8916
+    },
+    {
+      "epoch": 0.5746600502674486,
+      "grad_norm": 0.0006270557073641862,
+      "learning_rate": 4.731113498030791e-06,
+      "loss": 0.0,
+      "step": 8917
+    },
+    {
+      "epoch": 0.5747244957143778,
+      "grad_norm": 0.000898050790659976,
+      "learning_rate": 4.730397422126746e-06,
+      "loss": 0.0015,
+      "step": 8918
+    },
+    {
+      "epoch": 0.574788941161307,
+      "grad_norm": 0.0010283628729795369,
+      "learning_rate": 4.7296813462227e-06,
+      "loss": 0.0,
+      "step": 8919
+    },
+    {
+      "epoch": 0.5748533866082361,
+      "grad_norm": 0.014245992428902015,
+      "learning_rate": 4.728965270318654e-06,
+      "loss": 0.0,
+      "step": 8920
+    },
+    {
+      "epoch": 0.5749178320551653,
+      "grad_norm": 0.004830281237210356,
+      "learning_rate": 4.7282491944146085e-06,
+      "loss": 0.0,
+      "step": 8921
+    },
+    {
+      "epoch": 0.5749822775020945,
+      "grad_norm": 0.02278564499924832,
+      "learning_rate": 4.727533118510563e-06,
+      "loss": 0.0,
+      "step": 8922
+    },
+    {
+      "epoch": 0.5750467229490237,
+      "grad_norm": 0.0002993158678736612,
+      "learning_rate": 4.726817042606517e-06,
+      "loss": 0.0,
+      "step": 8923
+    },
+    {
+      "epoch": 0.5751111683959528,
+      "grad_norm": 0.05430118317287579,
+      "learning_rate": 4.726100966702471e-06,
+      "loss": 0.0001,
+      "step": 8924
+    },
+    {
+      "epoch": 0.575175613842882,
+      "grad_norm": 0.0014384505586056613,
+      "learning_rate": 4.725384890798425e-06,
+      "loss": 0.0,
+      "step": 8925
+    },
+    {
+      "epoch": 0.5752400592898111,
+      "grad_norm": 0.5587276218293904,
+      "learning_rate": 4.724668814894379e-06,
+      "loss": 0.0008,
+      "step": 8926
+    },
+    {
+      "epoch": 0.5753045047367403,
+      "grad_norm": 0.009065701688840216,
+      "learning_rate": 4.7239527389903334e-06,
+      "loss": 0.0,
+      "step": 8927
+    },
+    {
+      "epoch": 0.5753689501836695,
+      "grad_norm": 0.04705428814279142,
+      "learning_rate": 4.723236663086288e-06,
+      "loss": 0.0005,
+      "step": 8928
+    },
+    {
+      "epoch": 0.5754333956305987,
+      "grad_norm": 0.009048597003513747,
+      "learning_rate": 4.722520587182241e-06,
+      "loss": 0.0,
+      "step": 8929
+    },
+    {
+      "epoch": 0.5754978410775279,
+      "grad_norm": 0.00033590933556921425,
+      "learning_rate": 4.7218045112781955e-06,
+      "loss": 0.0,
+      "step": 8930
+    },
+    {
+      "epoch": 0.5755622865244571,
+      "grad_norm": 0.002176598827963511,
+      "learning_rate": 4.72108843537415e-06,
+      "loss": 0.0,
+      "step": 8931
+    },
+    {
+      "epoch": 0.5756267319713863,
+      "grad_norm": 0.00026799353923762494,
+      "learning_rate": 4.720372359470104e-06,
+      "loss": 0.0,
+      "step": 8932
+    },
+    {
+      "epoch": 0.5756911774183154,
+      "grad_norm": 0.03386055598732095,
+      "learning_rate": 4.719656283566058e-06,
+      "loss": 0.0001,
+      "step": 8933
+    },
+    {
+      "epoch": 0.5757556228652446,
+      "grad_norm": 0.0008778612228956532,
+      "learning_rate": 4.718940207662013e-06,
+      "loss": 0.0,
+      "step": 8934
+    },
+    {
+      "epoch": 0.5758200683121737,
+      "grad_norm": 0.0007188580320088237,
+      "learning_rate": 4.718224131757967e-06,
+      "loss": 0.0,
+      "step": 8935
+    },
+    {
+      "epoch": 0.5758845137591029,
+      "grad_norm": 0.022633746857609334,
+      "learning_rate": 4.717508055853921e-06,
+      "loss": 0.0001,
+      "step": 8936
+    },
+    {
+      "epoch": 0.5759489592060321,
+      "grad_norm": 0.0006597787799743524,
+      "learning_rate": 4.716791979949875e-06,
+      "loss": 0.0,
+      "step": 8937
+    },
+    {
+      "epoch": 0.5760134046529612,
+      "grad_norm": 0.017968162787472524,
+      "learning_rate": 4.716075904045829e-06,
+      "loss": 0.0002,
+      "step": 8938
+    },
+    {
+      "epoch": 0.5760778500998904,
+      "grad_norm": 0.003721095668540121,
+      "learning_rate": 4.715359828141783e-06,
+      "loss": 0.0,
+      "step": 8939
+    },
+    {
+      "epoch": 0.5761422955468196,
+      "grad_norm": 0.1334403917347702,
+      "learning_rate": 4.714643752237738e-06,
+      "loss": 0.0002,
+      "step": 8940
+    },
+    {
+      "epoch": 0.5762067409937488,
+      "grad_norm": 0.049999945890129134,
+      "learning_rate": 4.713927676333692e-06,
+      "loss": 0.0005,
+      "step": 8941
+    },
+    {
+      "epoch": 0.576271186440678,
+      "grad_norm": 0.04438637225912091,
+      "learning_rate": 4.713211600429645e-06,
+      "loss": 0.0001,
+      "step": 8942
+    },
+    {
+      "epoch": 0.5763356318876072,
+      "grad_norm": 0.004939305130353833,
+      "learning_rate": 4.7124955245256e-06,
+      "loss": 0.0,
+      "step": 8943
+    },
+    {
+      "epoch": 0.5764000773345364,
+      "grad_norm": 0.010423595187445532,
+      "learning_rate": 4.711779448621554e-06,
+      "loss": 0.0001,
+      "step": 8944
+    },
+    {
+      "epoch": 0.5764645227814655,
+      "grad_norm": 0.30387026664487055,
+      "learning_rate": 4.711063372717508e-06,
+      "loss": 0.0003,
+      "step": 8945
+    },
+    {
+      "epoch": 0.5765289682283946,
+      "grad_norm": 0.002739366849377702,
+      "learning_rate": 4.7103472968134626e-06,
+      "loss": 0.0,
+      "step": 8946
+    },
+    {
+      "epoch": 0.5765934136753238,
+      "grad_norm": 0.21060028318841664,
+      "learning_rate": 4.709631220909417e-06,
+      "loss": 0.0042,
+      "step": 8947
+    },
+    {
+      "epoch": 0.576657859122253,
+      "grad_norm": 0.0018765208304858886,
+      "learning_rate": 4.708915145005371e-06,
+      "loss": 0.0,
+      "step": 8948
+    },
+    {
+      "epoch": 0.5767223045691822,
+      "grad_norm": 0.0007770944081688407,
+      "learning_rate": 4.7081990691013255e-06,
+      "loss": 0.0,
+      "step": 8949
+    },
+    {
+      "epoch": 0.5767867500161113,
+      "grad_norm": 0.002056512481472225,
+      "learning_rate": 4.707482993197279e-06,
+      "loss": 0.0,
+      "step": 8950
+    },
+    {
+      "epoch": 0.5768511954630405,
+      "grad_norm": 0.0034826728857168528,
+      "learning_rate": 4.706766917293233e-06,
+      "loss": 0.0,
+      "step": 8951
+    },
+    {
+      "epoch": 0.5769156409099697,
+      "grad_norm": 0.0003830165904485001,
+      "learning_rate": 4.7060508413891875e-06,
+      "loss": 0.0,
+      "step": 8952
+    },
+    {
+      "epoch": 0.5769800863568989,
+      "grad_norm": 0.043901532936228796,
+      "learning_rate": 4.705334765485142e-06,
+      "loss": 0.0001,
+      "step": 8953
+    },
+    {
+      "epoch": 0.5770445318038281,
+      "grad_norm": 0.003480280335613063,
+      "learning_rate": 4.704618689581096e-06,
+      "loss": 0.0,
+      "step": 8954
+    },
+    {
+      "epoch": 0.5771089772507573,
+      "grad_norm": 0.09117785697310841,
+      "learning_rate": 4.7039026136770496e-06,
+      "loss": 0.0001,
+      "step": 8955
+    },
+    {
+      "epoch": 0.5771734226976865,
+      "grad_norm": 0.0001285398936270668,
+      "learning_rate": 4.703186537773004e-06,
+      "loss": 0.0,
+      "step": 8956
+    },
+    {
+      "epoch": 0.5772378681446155,
+      "grad_norm": 0.0008863763737434807,
+      "learning_rate": 4.702470461868959e-06,
+      "loss": 0.0,
+      "step": 8957
+    },
+    {
+      "epoch": 0.5773023135915447,
+      "grad_norm": 0.38130637127350175,
+      "learning_rate": 4.7017543859649125e-06,
+      "loss": 0.0016,
+      "step": 8958
+    },
+    {
+      "epoch": 0.5773667590384739,
+      "grad_norm": 0.008510612682685685,
+      "learning_rate": 4.701038310060867e-06,
+      "loss": 0.0,
+      "step": 8959
+    },
+    {
+      "epoch": 0.5774312044854031,
+      "grad_norm": 0.009348145843828558,
+      "learning_rate": 4.700322234156821e-06,
+      "loss": 0.0001,
+      "step": 8960
+    },
+    {
+      "epoch": 0.5774956499323323,
+      "grad_norm": 0.006941331467167952,
+      "learning_rate": 4.699606158252775e-06,
+      "loss": 0.0,
+      "step": 8961
+    },
+    {
+      "epoch": 0.5775600953792615,
+      "grad_norm": 0.0013218582692286602,
+      "learning_rate": 4.69889008234873e-06,
+      "loss": 0.0,
+      "step": 8962
+    },
+    {
+      "epoch": 0.5776245408261906,
+      "grad_norm": 0.0005961525107431034,
+      "learning_rate": 4.698174006444683e-06,
+      "loss": 0.0,
+      "step": 8963
+    },
+    {
+      "epoch": 0.5776889862731198,
+      "grad_norm": 0.5694010257321127,
+      "learning_rate": 4.697457930540637e-06,
+      "loss": 0.0026,
+      "step": 8964
+    },
+    {
+      "epoch": 0.577753431720049,
+      "grad_norm": 0.0002869682815712005,
+      "learning_rate": 4.696741854636592e-06,
+      "loss": 0.0,
+      "step": 8965
+    },
+    {
+      "epoch": 0.5778178771669782,
+      "grad_norm": 0.006472222447300945,
+      "learning_rate": 4.696025778732546e-06,
+      "loss": 0.0,
+      "step": 8966
+    },
+    {
+      "epoch": 0.5778823226139074,
+      "grad_norm": 0.05890671080692376,
+      "learning_rate": 4.6953097028285e-06,
+      "loss": 0.0001,
+      "step": 8967
+    },
+    {
+      "epoch": 0.5779467680608364,
+      "grad_norm": 0.0015918262072770278,
+      "learning_rate": 4.694593626924455e-06,
+      "loss": 0.0,
+      "step": 8968
+    },
+    {
+      "epoch": 0.5780112135077656,
+      "grad_norm": 0.0012047950646157887,
+      "learning_rate": 4.693877551020409e-06,
+      "loss": 0.0,
+      "step": 8969
+    },
+    {
+      "epoch": 0.5780756589546948,
+      "grad_norm": 0.0001741613348393895,
+      "learning_rate": 4.693161475116363e-06,
+      "loss": 0.0,
+      "step": 8970
+    },
+    {
+      "epoch": 0.578140104401624,
+      "grad_norm": 9.804336294408929e-05,
+      "learning_rate": 4.692445399212317e-06,
+      "loss": 0.0,
+      "step": 8971
+    },
+    {
+      "epoch": 0.5782045498485532,
+      "grad_norm": 0.0002919895338205372,
+      "learning_rate": 4.691729323308271e-06,
+      "loss": 0.0,
+      "step": 8972
+    },
+    {
+      "epoch": 0.5782689952954824,
+      "grad_norm": 0.33817765277223377,
+      "learning_rate": 4.691013247404225e-06,
+      "loss": 0.001,
+      "step": 8973
+    },
+    {
+      "epoch": 0.5783334407424116,
+      "grad_norm": 0.0005003706800080808,
+      "learning_rate": 4.6902971715001795e-06,
+      "loss": 0.0,
+      "step": 8974
+    },
+    {
+      "epoch": 0.5783978861893407,
+      "grad_norm": 0.01697098235105734,
+      "learning_rate": 4.689581095596134e-06,
+      "loss": 0.0,
+      "step": 8975
+    },
+    {
+      "epoch": 0.5784623316362699,
+      "grad_norm": 0.01526346802499094,
+      "learning_rate": 4.688865019692087e-06,
+      "loss": 0.0001,
+      "step": 8976
+    },
+    {
+      "epoch": 0.5785267770831991,
+      "grad_norm": 0.00011173870781894097,
+      "learning_rate": 4.688148943788042e-06,
+      "loss": 0.0,
+      "step": 8977
+    },
+    {
+      "epoch": 0.5785912225301283,
+      "grad_norm": 0.19004203268711622,
+      "learning_rate": 4.687432867883996e-06,
+      "loss": 0.0005,
+      "step": 8978
+    },
+    {
+      "epoch": 0.5786556679770575,
+      "grad_norm": 6.244763941013428e-05,
+      "learning_rate": 4.68671679197995e-06,
+      "loss": 0.0,
+      "step": 8979
+    },
+    {
+      "epoch": 0.5787201134239865,
+      "grad_norm": 9.843372332926753e-05,
+      "learning_rate": 4.6860007160759045e-06,
+      "loss": 0.0,
+      "step": 8980
+    },
+    {
+      "epoch": 0.5787845588709157,
+      "grad_norm": 0.16569536958932962,
+      "learning_rate": 4.685284640171859e-06,
+      "loss": 0.0004,
+      "step": 8981
+    },
+    {
+      "epoch": 0.5788490043178449,
+      "grad_norm": 0.0005437621685816029,
+      "learning_rate": 4.684568564267813e-06,
+      "loss": 0.0,
+      "step": 8982
+    },
+    {
+      "epoch": 0.5789134497647741,
+      "grad_norm": 0.08751011709068818,
+      "learning_rate": 4.683852488363767e-06,
+      "loss": 0.0047,
+      "step": 8983
+    },
+    {
+      "epoch": 0.5789778952117033,
+      "grad_norm": 0.0361179598361197,
+      "learning_rate": 4.683136412459721e-06,
+      "loss": 0.0002,
+      "step": 8984
+    },
+    {
+      "epoch": 0.5790423406586325,
+      "grad_norm": 0.0009243908624880609,
+      "learning_rate": 4.682420336555675e-06,
+      "loss": 0.0,
+      "step": 8985
+    },
+    {
+      "epoch": 0.5791067861055617,
+      "grad_norm": 0.00028813802643748844,
+      "learning_rate": 4.6817042606516294e-06,
+      "loss": 0.0,
+      "step": 8986
+    },
+    {
+      "epoch": 0.5791712315524908,
+      "grad_norm": 0.0014871085523719119,
+      "learning_rate": 4.680988184747584e-06,
+      "loss": 0.0,
+      "step": 8987
+    },
+    {
+      "epoch": 0.57923567699942,
+      "grad_norm": 0.0011678639568634536,
+      "learning_rate": 4.680272108843538e-06,
+      "loss": 0.0,
+      "step": 8988
+    },
+    {
+      "epoch": 0.5793001224463492,
+      "grad_norm": 0.022646964211210575,
+      "learning_rate": 4.6795560329394915e-06,
+      "loss": 0.0,
+      "step": 8989
+    },
+    {
+      "epoch": 0.5793645678932784,
+      "grad_norm": 6.162092306559458e-05,
+      "learning_rate": 4.678839957035446e-06,
+      "loss": 0.0,
+      "step": 8990
+    },
+    {
+      "epoch": 0.5794290133402075,
+      "grad_norm": 0.02449790624310379,
+      "learning_rate": 4.6781238811314e-06,
+      "loss": 0.0,
+      "step": 8991
+    },
+    {
+      "epoch": 0.5794934587871367,
+      "grad_norm": 0.12489244930395811,
+      "learning_rate": 4.677407805227354e-06,
+      "loss": 0.0001,
+      "step": 8992
+    },
+    {
+      "epoch": 0.5795579042340658,
+      "grad_norm": 0.012767780350679178,
+      "learning_rate": 4.676691729323309e-06,
+      "loss": 0.0001,
+      "step": 8993
+    },
+    {
+      "epoch": 0.579622349680995,
+      "grad_norm": 0.009526083138262352,
+      "learning_rate": 4.675975653419263e-06,
+      "loss": 0.0,
+      "step": 8994
+    },
+    {
+      "epoch": 0.5796867951279242,
+      "grad_norm": 0.021985102148627596,
+      "learning_rate": 4.675259577515217e-06,
+      "loss": 0.0002,
+      "step": 8995
+    },
+    {
+      "epoch": 0.5797512405748534,
+      "grad_norm": 0.0004470049212301294,
+      "learning_rate": 4.6745435016111716e-06,
+      "loss": 0.0,
+      "step": 8996
+    },
+    {
+      "epoch": 0.5798156860217826,
+      "grad_norm": 0.0004562537429128238,
+      "learning_rate": 4.673827425707125e-06,
+      "loss": 0.0,
+      "step": 8997
+    },
+    {
+      "epoch": 0.5798801314687118,
+      "grad_norm": 0.0009392800325101437,
+      "learning_rate": 4.673111349803079e-06,
+      "loss": 0.0,
+      "step": 8998
+    },
+    {
+      "epoch": 0.5799445769156409,
+      "grad_norm": 0.008200601472724514,
+      "learning_rate": 4.672395273899034e-06,
+      "loss": 0.0,
+      "step": 8999
+    },
+    {
+      "epoch": 0.5800090223625701,
+      "grad_norm": 0.00019648518137536374,
+      "learning_rate": 4.671679197994988e-06,
+      "loss": 0.0,
+      "step": 9000
+    },
+    {
+      "epoch": 0.5800734678094993,
+      "grad_norm": 0.001726909860329572,
+      "learning_rate": 4.670963122090942e-06,
+      "loss": 0.0,
+      "step": 9001
+    },
+    {
+      "epoch": 0.5801379132564284,
+      "grad_norm": 7.409479499881678e-05,
+      "learning_rate": 4.670247046186896e-06,
+      "loss": 0.0,
+      "step": 9002
+    },
+    {
+      "epoch": 0.5802023587033576,
+      "grad_norm": 0.0003329872254956827,
+      "learning_rate": 4.669530970282851e-06,
+      "loss": 0.0,
+      "step": 9003
+    },
+    {
+      "epoch": 0.5802668041502868,
+      "grad_norm": 0.0010982760062732727,
+      "learning_rate": 4.668814894378805e-06,
+      "loss": 0.0,
+      "step": 9004
+    },
+    {
+      "epoch": 0.5803312495972159,
+      "grad_norm": 0.0035950464528479,
+      "learning_rate": 4.6680988184747586e-06,
+      "loss": 0.0,
+      "step": 9005
+    },
+    {
+      "epoch": 0.5803956950441451,
+      "grad_norm": 0.021047836132908507,
+      "learning_rate": 4.667382742570713e-06,
+      "loss": 0.0002,
+      "step": 9006
+    },
+    {
+      "epoch": 0.5804601404910743,
+      "grad_norm": 0.0011425220805333551,
+      "learning_rate": 4.666666666666667e-06,
+      "loss": 0.0,
+      "step": 9007
+    },
+    {
+      "epoch": 0.5805245859380035,
+      "grad_norm": 0.0008882349360519453,
+      "learning_rate": 4.6659505907626215e-06,
+      "loss": 0.0,
+      "step": 9008
+    },
+    {
+      "epoch": 0.5805890313849327,
+      "grad_norm": 0.0002802394128841394,
+      "learning_rate": 4.665234514858576e-06,
+      "loss": 0.0,
+      "step": 9009
+    },
+    {
+      "epoch": 0.5806534768318619,
+      "grad_norm": 0.0009635131074359399,
+      "learning_rate": 4.664518438954529e-06,
+      "loss": 0.0,
+      "step": 9010
+    },
+    {
+      "epoch": 0.580717922278791,
+      "grad_norm": 0.0009012143762942716,
+      "learning_rate": 4.6638023630504835e-06,
+      "loss": 0.0,
+      "step": 9011
+    },
+    {
+      "epoch": 0.5807823677257202,
+      "grad_norm": 0.006541857974311703,
+      "learning_rate": 4.663086287146438e-06,
+      "loss": 0.0,
+      "step": 9012
+    },
+    {
+      "epoch": 0.5808468131726493,
+      "grad_norm": 0.06217128394273282,
+      "learning_rate": 4.662370211242392e-06,
+      "loss": 0.0002,
+      "step": 9013
+    },
+    {
+      "epoch": 0.5809112586195785,
+      "grad_norm": 0.01966216794474465,
+      "learning_rate": 4.661654135338346e-06,
+      "loss": 0.0,
+      "step": 9014
+    },
+    {
+      "epoch": 0.5809757040665077,
+      "grad_norm": 0.0007292552460633633,
+      "learning_rate": 4.660938059434301e-06,
+      "loss": 0.0,
+      "step": 9015
+    },
+    {
+      "epoch": 0.5810401495134369,
+      "grad_norm": 0.047384355713321426,
+      "learning_rate": 4.660221983530255e-06,
+      "loss": 0.0001,
+      "step": 9016
+    },
+    {
+      "epoch": 0.581104594960366,
+      "grad_norm": 0.004729452570084791,
+      "learning_rate": 4.659505907626209e-06,
+      "loss": 0.0,
+      "step": 9017
+    },
+    {
+      "epoch": 0.5811690404072952,
+      "grad_norm": 0.0005329612298227006,
+      "learning_rate": 4.658789831722163e-06,
+      "loss": 0.0,
+      "step": 9018
+    },
+    {
+      "epoch": 0.5812334858542244,
+      "grad_norm": 0.09264701461117533,
+      "learning_rate": 4.658073755818117e-06,
+      "loss": 0.001,
+      "step": 9019
+    },
+    {
+      "epoch": 0.5812979313011536,
+      "grad_norm": 0.0002257829058000882,
+      "learning_rate": 4.657357679914071e-06,
+      "loss": 0.0,
+      "step": 9020
+    },
+    {
+      "epoch": 0.5813623767480828,
+      "grad_norm": 0.010905818191302687,
+      "learning_rate": 4.656641604010026e-06,
+      "loss": 0.0,
+      "step": 9021
+    },
+    {
+      "epoch": 0.581426822195012,
+      "grad_norm": 0.005600216922704845,
+      "learning_rate": 4.65592552810598e-06,
+      "loss": 0.0,
+      "step": 9022
+    },
+    {
+      "epoch": 0.5814912676419411,
+      "grad_norm": 0.031750284272755426,
+      "learning_rate": 4.655209452201933e-06,
+      "loss": 0.0001,
+      "step": 9023
+    },
+    {
+      "epoch": 0.5815557130888702,
+      "grad_norm": 0.0077528691803133536,
+      "learning_rate": 4.654493376297888e-06,
+      "loss": 0.0,
+      "step": 9024
+    },
+    {
+      "epoch": 0.5816201585357994,
+      "grad_norm": 0.04543225714500262,
+      "learning_rate": 4.653777300393842e-06,
+      "loss": 0.0001,
+      "step": 9025
+    },
+    {
+      "epoch": 0.5816846039827286,
+      "grad_norm": 0.013056369682744871,
+      "learning_rate": 4.653061224489796e-06,
+      "loss": 0.0001,
+      "step": 9026
+    },
+    {
+      "epoch": 0.5817490494296578,
+      "grad_norm": 0.17819283640927924,
+      "learning_rate": 4.652345148585751e-06,
+      "loss": 0.0003,
+      "step": 9027
+    },
+    {
+      "epoch": 0.581813494876587,
+      "grad_norm": 0.08917160086074113,
+      "learning_rate": 4.651629072681705e-06,
+      "loss": 0.0003,
+      "step": 9028
+    },
+    {
+      "epoch": 0.5818779403235161,
+      "grad_norm": 0.006331248638778778,
+      "learning_rate": 4.650912996777659e-06,
+      "loss": 0.0,
+      "step": 9029
+    },
+    {
+      "epoch": 0.5819423857704453,
+      "grad_norm": 0.002775633504862486,
+      "learning_rate": 4.6501969208736135e-06,
+      "loss": 0.0,
+      "step": 9030
+    },
+    {
+      "epoch": 0.5820068312173745,
+      "grad_norm": 6.992336507404066e-05,
+      "learning_rate": 4.649480844969567e-06,
+      "loss": 0.0,
+      "step": 9031
+    },
+    {
+      "epoch": 0.5820712766643037,
+      "grad_norm": 0.00017943615738269465,
+      "learning_rate": 4.648764769065521e-06,
+      "loss": 0.0,
+      "step": 9032
+    },
+    {
+      "epoch": 0.5821357221112329,
+      "grad_norm": 0.0058394424584071415,
+      "learning_rate": 4.6480486931614755e-06,
+      "loss": 0.0,
+      "step": 9033
+    },
+    {
+      "epoch": 0.5822001675581621,
+      "grad_norm": 0.050562310960973855,
+      "learning_rate": 4.64733261725743e-06,
+      "loss": 0.0002,
+      "step": 9034
+    },
+    {
+      "epoch": 0.5822646130050911,
+      "grad_norm": 0.8657970450295585,
+      "learning_rate": 4.646616541353383e-06,
+      "loss": 0.0037,
+      "step": 9035
+    },
+    {
+      "epoch": 0.5823290584520203,
+      "grad_norm": 0.00211204704726517,
+      "learning_rate": 4.645900465449338e-06,
+      "loss": 0.0,
+      "step": 9036
+    },
+    {
+      "epoch": 0.5823935038989495,
+      "grad_norm": 0.004920183719435545,
+      "learning_rate": 4.645184389545292e-06,
+      "loss": 0.0,
+      "step": 9037
+    },
+    {
+      "epoch": 0.5824579493458787,
+      "grad_norm": 0.00663491009308444,
+      "learning_rate": 4.644468313641247e-06,
+      "loss": 0.0,
+      "step": 9038
+    },
+    {
+      "epoch": 0.5825223947928079,
+      "grad_norm": 0.0022066924291144698,
+      "learning_rate": 4.6437522377372005e-06,
+      "loss": 0.0,
+      "step": 9039
+    },
+    {
+      "epoch": 0.5825868402397371,
+      "grad_norm": 0.0022171824438060417,
+      "learning_rate": 4.643036161833155e-06,
+      "loss": 0.0,
+      "step": 9040
+    },
+    {
+      "epoch": 0.5826512856866662,
+      "grad_norm": 0.00032441297467396377,
+      "learning_rate": 4.642320085929109e-06,
+      "loss": 0.0,
+      "step": 9041
+    },
+    {
+      "epoch": 0.5827157311335954,
+      "grad_norm": 0.030769737840866715,
+      "learning_rate": 4.641604010025063e-06,
+      "loss": 0.0,
+      "step": 9042
+    },
+    {
+      "epoch": 0.5827801765805246,
+      "grad_norm": 0.0019483906462393567,
+      "learning_rate": 4.640887934121017e-06,
+      "loss": 0.0,
+      "step": 9043
+    },
+    {
+      "epoch": 0.5828446220274538,
+      "grad_norm": 0.00029194905052751225,
+      "learning_rate": 4.640171858216971e-06,
+      "loss": 0.0,
+      "step": 9044
+    },
+    {
+      "epoch": 0.582909067474383,
+      "grad_norm": 0.0002021999263205449,
+      "learning_rate": 4.6394557823129254e-06,
+      "loss": 0.0,
+      "step": 9045
+    },
+    {
+      "epoch": 0.5829735129213122,
+      "grad_norm": 0.0014373682518094646,
+      "learning_rate": 4.63873970640888e-06,
+      "loss": 0.0,
+      "step": 9046
+    },
+    {
+      "epoch": 0.5830379583682412,
+      "grad_norm": 0.059762924731719465,
+      "learning_rate": 4.638023630504834e-06,
+      "loss": 0.0001,
+      "step": 9047
+    },
+    {
+      "epoch": 0.5831024038151704,
+      "grad_norm": 0.4903177897093207,
+      "learning_rate": 4.6373075546007875e-06,
+      "loss": 0.0041,
+      "step": 9048
+    },
+    {
+      "epoch": 0.5831668492620996,
+      "grad_norm": 0.0014177705494520779,
+      "learning_rate": 4.636591478696742e-06,
+      "loss": 0.0,
+      "step": 9049
+    },
+    {
+      "epoch": 0.5832312947090288,
+      "grad_norm": 0.2566934777913586,
+      "learning_rate": 4.635875402792697e-06,
+      "loss": 0.0021,
+      "step": 9050
+    },
+    {
+      "epoch": 0.583295740155958,
+      "grad_norm": 0.018140133354513432,
+      "learning_rate": 4.63515932688865e-06,
+      "loss": 0.0001,
+      "step": 9051
+    },
+    {
+      "epoch": 0.5833601856028872,
+      "grad_norm": 0.00788615014582874,
+      "learning_rate": 4.634443250984605e-06,
+      "loss": 0.0,
+      "step": 9052
+    },
+    {
+      "epoch": 0.5834246310498163,
+      "grad_norm": 0.01840864843113238,
+      "learning_rate": 4.633727175080559e-06,
+      "loss": 0.0001,
+      "step": 9053
+    },
+    {
+      "epoch": 0.5834890764967455,
+      "grad_norm": 0.001822052435323416,
+      "learning_rate": 4.633011099176513e-06,
+      "loss": 0.0,
+      "step": 9054
+    },
+    {
+      "epoch": 0.5835535219436747,
+      "grad_norm": 0.017397195278423098,
+      "learning_rate": 4.6322950232724676e-06,
+      "loss": 0.0,
+      "step": 9055
+    },
+    {
+      "epoch": 0.5836179673906039,
+      "grad_norm": 0.6295143409970986,
+      "learning_rate": 4.631578947368421e-06,
+      "loss": 0.0036,
+      "step": 9056
+    },
+    {
+      "epoch": 0.5836824128375331,
+      "grad_norm": 0.004145135076455736,
+      "learning_rate": 4.630862871464375e-06,
+      "loss": 0.0,
+      "step": 9057
+    },
+    {
+      "epoch": 0.5837468582844622,
+      "grad_norm": 0.01922043763141991,
+      "learning_rate": 4.63014679556033e-06,
+      "loss": 0.0001,
+      "step": 9058
+    },
+    {
+      "epoch": 0.5838113037313913,
+      "grad_norm": 0.21877145661980707,
+      "learning_rate": 4.629430719656284e-06,
+      "loss": 0.0003,
+      "step": 9059
+    },
+    {
+      "epoch": 0.5838757491783205,
+      "grad_norm": 0.06723857036500273,
+      "learning_rate": 4.628714643752238e-06,
+      "loss": 0.0001,
+      "step": 9060
+    },
+    {
+      "epoch": 0.5839401946252497,
+      "grad_norm": 0.037172463083185225,
+      "learning_rate": 4.627998567848192e-06,
+      "loss": 0.0,
+      "step": 9061
+    },
+    {
+      "epoch": 0.5840046400721789,
+      "grad_norm": 0.0024169997396676244,
+      "learning_rate": 4.627282491944147e-06,
+      "loss": 0.0,
+      "step": 9062
+    },
+    {
+      "epoch": 0.5840690855191081,
+      "grad_norm": 0.06419768070761167,
+      "learning_rate": 4.626566416040101e-06,
+      "loss": 0.0,
+      "step": 9063
+    },
+    {
+      "epoch": 0.5841335309660373,
+      "grad_norm": 0.01399367865517338,
+      "learning_rate": 4.6258503401360546e-06,
+      "loss": 0.0,
+      "step": 9064
+    },
+    {
+      "epoch": 0.5841979764129664,
+      "grad_norm": 0.000952353755274673,
+      "learning_rate": 4.625134264232009e-06,
+      "loss": 0.0,
+      "step": 9065
+    },
+    {
+      "epoch": 0.5842624218598956,
+      "grad_norm": 0.0015975798325585998,
+      "learning_rate": 4.624418188327963e-06,
+      "loss": 0.0,
+      "step": 9066
+    },
+    {
+      "epoch": 0.5843268673068248,
+      "grad_norm": 0.0026882272384257705,
+      "learning_rate": 4.6237021124239175e-06,
+      "loss": 0.0,
+      "step": 9067
+    },
+    {
+      "epoch": 0.584391312753754,
+      "grad_norm": 0.001748996446030553,
+      "learning_rate": 4.622986036519872e-06,
+      "loss": 0.0,
+      "step": 9068
+    },
+    {
+      "epoch": 0.5844557582006831,
+      "grad_norm": 0.18220017498674793,
+      "learning_rate": 4.622269960615825e-06,
+      "loss": 0.0017,
+      "step": 9069
+    },
+    {
+      "epoch": 0.5845202036476123,
+      "grad_norm": 0.009680095160558684,
+      "learning_rate": 4.6215538847117795e-06,
+      "loss": 0.0,
+      "step": 9070
+    },
+    {
+      "epoch": 0.5845846490945414,
+      "grad_norm": 1.2011762913568456,
+      "learning_rate": 4.620837808807734e-06,
+      "loss": 0.0104,
+      "step": 9071
+    },
+    {
+      "epoch": 0.5846490945414706,
+      "grad_norm": 0.009852353560086855,
+      "learning_rate": 4.620121732903688e-06,
+      "loss": 0.0,
+      "step": 9072
+    },
+    {
+      "epoch": 0.5847135399883998,
+      "grad_norm": 0.0034802751090491258,
+      "learning_rate": 4.619405656999642e-06,
+      "loss": 0.0,
+      "step": 9073
+    },
+    {
+      "epoch": 0.584777985435329,
+      "grad_norm": 0.02462778481389003,
+      "learning_rate": 4.618689581095597e-06,
+      "loss": 0.0,
+      "step": 9074
+    },
+    {
+      "epoch": 0.5848424308822582,
+      "grad_norm": 0.07292170024709481,
+      "learning_rate": 4.617973505191551e-06,
+      "loss": 0.0001,
+      "step": 9075
+    },
+    {
+      "epoch": 0.5849068763291874,
+      "grad_norm": 0.01276836844157095,
+      "learning_rate": 4.617257429287505e-06,
+      "loss": 0.0001,
+      "step": 9076
+    },
+    {
+      "epoch": 0.5849713217761165,
+      "grad_norm": 0.0059313510178414935,
+      "learning_rate": 4.616541353383459e-06,
+      "loss": 0.0,
+      "step": 9077
+    },
+    {
+      "epoch": 0.5850357672230457,
+      "grad_norm": 0.0050054168634305505,
+      "learning_rate": 4.615825277479413e-06,
+      "loss": 0.0,
+      "step": 9078
+    },
+    {
+      "epoch": 0.5851002126699749,
+      "grad_norm": 0.004330369079694027,
+      "learning_rate": 4.615109201575367e-06,
+      "loss": 0.0,
+      "step": 9079
+    },
+    {
+      "epoch": 0.585164658116904,
+      "grad_norm": 0.0014310698214295584,
+      "learning_rate": 4.614393125671322e-06,
+      "loss": 0.0,
+      "step": 9080
+    },
+    {
+      "epoch": 0.5852291035638332,
+      "grad_norm": 0.09305590243071263,
+      "learning_rate": 4.613677049767276e-06,
+      "loss": 0.0003,
+      "step": 9081
+    },
+    {
+      "epoch": 0.5852935490107624,
+      "grad_norm": 0.003215462836571615,
+      "learning_rate": 4.612960973863229e-06,
+      "loss": 0.0,
+      "step": 9082
+    },
+    {
+      "epoch": 0.5853579944576915,
+      "grad_norm": 1.087399822312692,
+      "learning_rate": 4.612244897959184e-06,
+      "loss": 0.0105,
+      "step": 9083
+    },
+    {
+      "epoch": 0.5854224399046207,
+      "grad_norm": 0.7281104794474664,
+      "learning_rate": 4.611528822055138e-06,
+      "loss": 0.0062,
+      "step": 9084
+    },
+    {
+      "epoch": 0.5854868853515499,
+      "grad_norm": 0.00043356516939963686,
+      "learning_rate": 4.610812746151092e-06,
+      "loss": 0.0,
+      "step": 9085
+    },
+    {
+      "epoch": 0.5855513307984791,
+      "grad_norm": 0.00033685958252393,
+      "learning_rate": 4.610096670247047e-06,
+      "loss": 0.0,
+      "step": 9086
+    },
+    {
+      "epoch": 0.5856157762454083,
+      "grad_norm": 0.002318961319503037,
+      "learning_rate": 4.609380594343001e-06,
+      "loss": 0.0,
+      "step": 9087
+    },
+    {
+      "epoch": 0.5856802216923375,
+      "grad_norm": 0.005592278210797931,
+      "learning_rate": 4.608664518438955e-06,
+      "loss": 0.0,
+      "step": 9088
+    },
+    {
+      "epoch": 0.5857446671392667,
+      "grad_norm": 0.0243394892930314,
+      "learning_rate": 4.6079484425349095e-06,
+      "loss": 0.0,
+      "step": 9089
+    },
+    {
+      "epoch": 0.5858091125861958,
+      "grad_norm": 0.11083153764968887,
+      "learning_rate": 4.607232366630863e-06,
+      "loss": 0.0001,
+      "step": 9090
+    },
+    {
+      "epoch": 0.5858735580331249,
+      "grad_norm": 0.0007002490856994532,
+      "learning_rate": 4.606516290726817e-06,
+      "loss": 0.0,
+      "step": 9091
+    },
+    {
+      "epoch": 0.5859380034800541,
+      "grad_norm": 0.0017745563231980528,
+      "learning_rate": 4.6058002148227715e-06,
+      "loss": 0.0,
+      "step": 9092
+    },
+    {
+      "epoch": 0.5860024489269833,
+      "grad_norm": 0.005978146684738015,
+      "learning_rate": 4.605084138918726e-06,
+      "loss": 0.0,
+      "step": 9093
+    },
+    {
+      "epoch": 0.5860668943739125,
+      "grad_norm": 0.05288187124635033,
+      "learning_rate": 4.60436806301468e-06,
+      "loss": 0.0001,
+      "step": 9094
+    },
+    {
+      "epoch": 0.5861313398208416,
+      "grad_norm": 0.016551244789661204,
+      "learning_rate": 4.603651987110634e-06,
+      "loss": 0.0002,
+      "step": 9095
+    },
+    {
+      "epoch": 0.5861957852677708,
+      "grad_norm": 0.044669416309915146,
+      "learning_rate": 4.602935911206588e-06,
+      "loss": 0.0001,
+      "step": 9096
+    },
+    {
+      "epoch": 0.5862602307147,
+      "grad_norm": 0.2125491982896501,
+      "learning_rate": 4.602219835302543e-06,
+      "loss": 0.0002,
+      "step": 9097
+    },
+    {
+      "epoch": 0.5863246761616292,
+      "grad_norm": 0.49376460609078615,
+      "learning_rate": 4.6015037593984965e-06,
+      "loss": 0.001,
+      "step": 9098
+    },
+    {
+      "epoch": 0.5863891216085584,
+      "grad_norm": 0.016966827455400634,
+      "learning_rate": 4.600787683494451e-06,
+      "loss": 0.0,
+      "step": 9099
+    },
+    {
+      "epoch": 0.5864535670554876,
+      "grad_norm": 0.02803030621426667,
+      "learning_rate": 4.600071607590405e-06,
+      "loss": 0.0,
+      "step": 9100
+    },
+    {
+      "epoch": 0.5865180125024168,
+      "grad_norm": 0.00504899862022767,
+      "learning_rate": 4.599355531686359e-06,
+      "loss": 0.0,
+      "step": 9101
+    },
+    {
+      "epoch": 0.5865824579493458,
+      "grad_norm": 5.1629753872988315,
+      "learning_rate": 4.598639455782314e-06,
+      "loss": 0.0183,
+      "step": 9102
+    },
+    {
+      "epoch": 0.586646903396275,
+      "grad_norm": 0.0011573463084333388,
+      "learning_rate": 4.597923379878267e-06,
+      "loss": 0.0,
+      "step": 9103
+    },
+    {
+      "epoch": 0.5867113488432042,
+      "grad_norm": 0.009361562436643706,
+      "learning_rate": 4.5972073039742214e-06,
+      "loss": 0.0,
+      "step": 9104
+    },
+    {
+      "epoch": 0.5867757942901334,
+      "grad_norm": 0.027741470990613967,
+      "learning_rate": 4.596491228070176e-06,
+      "loss": 0.0,
+      "step": 9105
+    },
+    {
+      "epoch": 0.5868402397370626,
+      "grad_norm": 0.00832458852908692,
+      "learning_rate": 4.59577515216613e-06,
+      "loss": 0.0,
+      "step": 9106
+    },
+    {
+      "epoch": 0.5869046851839917,
+      "grad_norm": 0.001290437778751822,
+      "learning_rate": 4.595059076262084e-06,
+      "loss": 0.0,
+      "step": 9107
+    },
+    {
+      "epoch": 0.5869691306309209,
+      "grad_norm": 0.015002341091636952,
+      "learning_rate": 4.594343000358039e-06,
+      "loss": 0.0,
+      "step": 9108
+    },
+    {
+      "epoch": 0.5870335760778501,
+      "grad_norm": 0.004408846988619795,
+      "learning_rate": 4.593626924453993e-06,
+      "loss": 0.0,
+      "step": 9109
+    },
+    {
+      "epoch": 0.5870980215247793,
+      "grad_norm": 0.039488889693771616,
+      "learning_rate": 4.592910848549947e-06,
+      "loss": 0.0,
+      "step": 9110
+    },
+    {
+      "epoch": 0.5871624669717085,
+      "grad_norm": 0.026583034234040636,
+      "learning_rate": 4.592194772645901e-06,
+      "loss": 0.0003,
+      "step": 9111
+    },
+    {
+      "epoch": 0.5872269124186377,
+      "grad_norm": 0.000849249254509577,
+      "learning_rate": 4.591478696741855e-06,
+      "loss": 0.0,
+      "step": 9112
+    },
+    {
+      "epoch": 0.5872913578655667,
+      "grad_norm": 0.004329321968183173,
+      "learning_rate": 4.590762620837809e-06,
+      "loss": 0.0,
+      "step": 9113
+    },
+    {
+      "epoch": 0.5873558033124959,
+      "grad_norm": 0.04470970489953232,
+      "learning_rate": 4.5900465449337636e-06,
+      "loss": 0.0,
+      "step": 9114
+    },
+    {
+      "epoch": 0.5874202487594251,
+      "grad_norm": 0.004520625298732878,
+      "learning_rate": 4.589330469029718e-06,
+      "loss": 0.0,
+      "step": 9115
+    },
+    {
+      "epoch": 0.5874846942063543,
+      "grad_norm": 0.00766393875770787,
+      "learning_rate": 4.588614393125671e-06,
+      "loss": 0.0,
+      "step": 9116
+    },
+    {
+      "epoch": 0.5875491396532835,
+      "grad_norm": 0.002323407439271178,
+      "learning_rate": 4.587898317221626e-06,
+      "loss": 0.0,
+      "step": 9117
+    },
+    {
+      "epoch": 0.5876135851002127,
+      "grad_norm": 0.06852094765522453,
+      "learning_rate": 4.58718224131758e-06,
+      "loss": 0.0001,
+      "step": 9118
+    },
+    {
+      "epoch": 0.5876780305471418,
+      "grad_norm": 0.004350183486519918,
+      "learning_rate": 4.586466165413534e-06,
+      "loss": 0.0001,
+      "step": 9119
+    },
+    {
+      "epoch": 0.587742475994071,
+      "grad_norm": 0.03648216598243733,
+      "learning_rate": 4.5857500895094885e-06,
+      "loss": 0.0001,
+      "step": 9120
+    },
+    {
+      "epoch": 0.5878069214410002,
+      "grad_norm": 0.31417986446623686,
+      "learning_rate": 4.585034013605443e-06,
+      "loss": 0.0014,
+      "step": 9121
+    },
+    {
+      "epoch": 0.5878713668879294,
+      "grad_norm": 0.0414216516748464,
+      "learning_rate": 4.584317937701397e-06,
+      "loss": 0.0001,
+      "step": 9122
+    },
+    {
+      "epoch": 0.5879358123348586,
+      "grad_norm": 0.05218019741684195,
+      "learning_rate": 4.583601861797351e-06,
+      "loss": 0.0001,
+      "step": 9123
+    },
+    {
+      "epoch": 0.5880002577817878,
+      "grad_norm": 0.00462391310842688,
+      "learning_rate": 4.582885785893305e-06,
+      "loss": 0.0,
+      "step": 9124
+    },
+    {
+      "epoch": 0.5880647032287168,
+      "grad_norm": 0.020700351989345004,
+      "learning_rate": 4.582169709989259e-06,
+      "loss": 0.0001,
+      "step": 9125
+    },
+    {
+      "epoch": 0.588129148675646,
+      "grad_norm": 0.0029324404378681776,
+      "learning_rate": 4.5814536340852135e-06,
+      "loss": 0.0,
+      "step": 9126
+    },
+    {
+      "epoch": 0.5881935941225752,
+      "grad_norm": 0.00026724816470433214,
+      "learning_rate": 4.580737558181168e-06,
+      "loss": 0.0,
+      "step": 9127
+    },
+    {
+      "epoch": 0.5882580395695044,
+      "grad_norm": 0.00047139101971960793,
+      "learning_rate": 4.580021482277121e-06,
+      "loss": 0.0,
+      "step": 9128
+    },
+    {
+      "epoch": 0.5883224850164336,
+      "grad_norm": 0.001048866333167167,
+      "learning_rate": 4.5793054063730755e-06,
+      "loss": 0.0,
+      "step": 9129
+    },
+    {
+      "epoch": 0.5883869304633628,
+      "grad_norm": 0.00039698803690931634,
+      "learning_rate": 4.57858933046903e-06,
+      "loss": 0.0,
+      "step": 9130
+    },
+    {
+      "epoch": 0.588451375910292,
+      "grad_norm": 0.006695436789118914,
+      "learning_rate": 4.577873254564984e-06,
+      "loss": 0.0001,
+      "step": 9131
+    },
+    {
+      "epoch": 0.5885158213572211,
+      "grad_norm": 0.0019064073027897633,
+      "learning_rate": 4.577157178660938e-06,
+      "loss": 0.0,
+      "step": 9132
+    },
+    {
+      "epoch": 0.5885802668041503,
+      "grad_norm": 0.0002214748274238106,
+      "learning_rate": 4.576441102756893e-06,
+      "loss": 0.0,
+      "step": 9133
+    },
+    {
+      "epoch": 0.5886447122510795,
+      "grad_norm": 0.1768014864419196,
+      "learning_rate": 4.575725026852847e-06,
+      "loss": 0.0032,
+      "step": 9134
+    },
+    {
+      "epoch": 0.5887091576980087,
+      "grad_norm": 0.00627439198323087,
+      "learning_rate": 4.575008950948801e-06,
+      "loss": 0.0,
+      "step": 9135
+    },
+    {
+      "epoch": 0.5887736031449378,
+      "grad_norm": 0.0019090848411585798,
+      "learning_rate": 4.574292875044755e-06,
+      "loss": 0.0,
+      "step": 9136
+    },
+    {
+      "epoch": 0.588838048591867,
+      "grad_norm": 0.023734394041223905,
+      "learning_rate": 4.573576799140709e-06,
+      "loss": 0.0001,
+      "step": 9137
+    },
+    {
+      "epoch": 0.5889024940387961,
+      "grad_norm": 0.0012559579821498448,
+      "learning_rate": 4.572860723236663e-06,
+      "loss": 0.0,
+      "step": 9138
+    },
+    {
+      "epoch": 0.5889669394857253,
+      "grad_norm": 0.005693433432522691,
+      "learning_rate": 4.572144647332618e-06,
+      "loss": 0.0001,
+      "step": 9139
+    },
+    {
+      "epoch": 0.5890313849326545,
+      "grad_norm": 0.009353012652150373,
+      "learning_rate": 4.571428571428572e-06,
+      "loss": 0.0,
+      "step": 9140
+    },
+    {
+      "epoch": 0.5890958303795837,
+      "grad_norm": 0.029658588021311515,
+      "learning_rate": 4.570712495524525e-06,
+      "loss": 0.0001,
+      "step": 9141
+    },
+    {
+      "epoch": 0.5891602758265129,
+      "grad_norm": 0.003302594123403248,
+      "learning_rate": 4.56999641962048e-06,
+      "loss": 0.0,
+      "step": 9142
+    },
+    {
+      "epoch": 0.589224721273442,
+      "grad_norm": 0.001705376414646427,
+      "learning_rate": 4.569280343716435e-06,
+      "loss": 0.0,
+      "step": 9143
+    },
+    {
+      "epoch": 0.5892891667203712,
+      "grad_norm": 0.18220126885195806,
+      "learning_rate": 4.568564267812388e-06,
+      "loss": 0.0004,
+      "step": 9144
+    },
+    {
+      "epoch": 0.5893536121673004,
+      "grad_norm": 0.0019109073460819275,
+      "learning_rate": 4.567848191908343e-06,
+      "loss": 0.0,
+      "step": 9145
+    },
+    {
+      "epoch": 0.5894180576142296,
+      "grad_norm": 0.005024864271140364,
+      "learning_rate": 4.567132116004297e-06,
+      "loss": 0.0,
+      "step": 9146
+    },
+    {
+      "epoch": 0.5894825030611587,
+      "grad_norm": 0.00905942092399999,
+      "learning_rate": 4.566416040100251e-06,
+      "loss": 0.0,
+      "step": 9147
+    },
+    {
+      "epoch": 0.5895469485080879,
+      "grad_norm": 0.005881856218996886,
+      "learning_rate": 4.5656999641962055e-06,
+      "loss": 0.0,
+      "step": 9148
+    },
+    {
+      "epoch": 0.589611393955017,
+      "grad_norm": 0.0008304460678939578,
+      "learning_rate": 4.564983888292159e-06,
+      "loss": 0.0,
+      "step": 9149
+    },
+    {
+      "epoch": 0.5896758394019462,
+      "grad_norm": 0.007501626420824486,
+      "learning_rate": 4.564267812388113e-06,
+      "loss": 0.0,
+      "step": 9150
+    },
+    {
+      "epoch": 0.5897402848488754,
+      "grad_norm": 0.0012510811332456175,
+      "learning_rate": 4.5635517364840675e-06,
+      "loss": 0.0,
+      "step": 9151
+    },
+    {
+      "epoch": 0.5898047302958046,
+      "grad_norm": 0.0746794568702863,
+      "learning_rate": 4.562835660580022e-06,
+      "loss": 0.001,
+      "step": 9152
+    },
+    {
+      "epoch": 0.5898691757427338,
+      "grad_norm": 0.26361614627359486,
+      "learning_rate": 4.562119584675976e-06,
+      "loss": 0.0021,
+      "step": 9153
+    },
+    {
+      "epoch": 0.589933621189663,
+      "grad_norm": 0.0002156212009346355,
+      "learning_rate": 4.56140350877193e-06,
+      "loss": 0.0,
+      "step": 9154
+    },
+    {
+      "epoch": 0.5899980666365922,
+      "grad_norm": 0.0015590836780957843,
+      "learning_rate": 4.560687432867885e-06,
+      "loss": 0.0,
+      "step": 9155
+    },
+    {
+      "epoch": 0.5900625120835213,
+      "grad_norm": 0.004308015541784625,
+      "learning_rate": 4.559971356963839e-06,
+      "loss": 0.0,
+      "step": 9156
+    },
+    {
+      "epoch": 0.5901269575304505,
+      "grad_norm": 0.0717703984591805,
+      "learning_rate": 4.5592552810597925e-06,
+      "loss": 0.0003,
+      "step": 9157
+    },
+    {
+      "epoch": 0.5901914029773796,
+      "grad_norm": 0.0016221557899336572,
+      "learning_rate": 4.558539205155747e-06,
+      "loss": 0.0,
+      "step": 9158
+    },
+    {
+      "epoch": 0.5902558484243088,
+      "grad_norm": 0.02843239715369182,
+      "learning_rate": 4.557823129251701e-06,
+      "loss": 0.0001,
+      "step": 9159
+    },
+    {
+      "epoch": 0.590320293871238,
+      "grad_norm": 0.08906306174586306,
+      "learning_rate": 4.557107053347655e-06,
+      "loss": 0.0002,
+      "step": 9160
+    },
+    {
+      "epoch": 0.5903847393181672,
+      "grad_norm": 0.041065458269552234,
+      "learning_rate": 4.55639097744361e-06,
+      "loss": 0.0001,
+      "step": 9161
+    },
+    {
+      "epoch": 0.5904491847650963,
+      "grad_norm": 0.0008033243988639701,
+      "learning_rate": 4.555674901539563e-06,
+      "loss": 0.0,
+      "step": 9162
+    },
+    {
+      "epoch": 0.5905136302120255,
+      "grad_norm": 0.05046193202663287,
+      "learning_rate": 4.5549588256355174e-06,
+      "loss": 0.0,
+      "step": 9163
+    },
+    {
+      "epoch": 0.5905780756589547,
+      "grad_norm": 0.00023855639736130923,
+      "learning_rate": 4.554242749731472e-06,
+      "loss": 0.0,
+      "step": 9164
+    },
+    {
+      "epoch": 0.5906425211058839,
+      "grad_norm": 0.0006409486529008727,
+      "learning_rate": 4.553526673827426e-06,
+      "loss": 0.0,
+      "step": 9165
+    },
+    {
+      "epoch": 0.5907069665528131,
+      "grad_norm": 0.00037044303511188413,
+      "learning_rate": 4.55281059792338e-06,
+      "loss": 0.0,
+      "step": 9166
+    },
+    {
+      "epoch": 0.5907714119997423,
+      "grad_norm": 0.012088132812897083,
+      "learning_rate": 4.552094522019335e-06,
+      "loss": 0.0,
+      "step": 9167
+    },
+    {
+      "epoch": 0.5908358574466714,
+      "grad_norm": 0.0016999619869035114,
+      "learning_rate": 4.551378446115289e-06,
+      "loss": 0.0,
+      "step": 9168
+    },
+    {
+      "epoch": 0.5909003028936005,
+      "grad_norm": 0.38509600634394847,
+      "learning_rate": 4.550662370211243e-06,
+      "loss": 0.0014,
+      "step": 9169
+    },
+    {
+      "epoch": 0.5909647483405297,
+      "grad_norm": 0.0015224179117437268,
+      "learning_rate": 4.549946294307197e-06,
+      "loss": 0.0,
+      "step": 9170
+    },
+    {
+      "epoch": 0.5910291937874589,
+      "grad_norm": 0.0011083900503956407,
+      "learning_rate": 4.549230218403151e-06,
+      "loss": 0.0,
+      "step": 9171
+    },
+    {
+      "epoch": 0.5910936392343881,
+      "grad_norm": 0.00030231327758239914,
+      "learning_rate": 4.548514142499105e-06,
+      "loss": 0.0,
+      "step": 9172
+    },
+    {
+      "epoch": 0.5911580846813173,
+      "grad_norm": 0.17115920743812696,
+      "learning_rate": 4.5477980665950596e-06,
+      "loss": 0.0002,
+      "step": 9173
+    },
+    {
+      "epoch": 0.5912225301282464,
+      "grad_norm": 0.0030166942177998085,
+      "learning_rate": 4.547081990691014e-06,
+      "loss": 0.0,
+      "step": 9174
+    },
+    {
+      "epoch": 0.5912869755751756,
+      "grad_norm": 0.013242115384337326,
+      "learning_rate": 4.546365914786967e-06,
+      "loss": 0.0001,
+      "step": 9175
+    },
+    {
+      "epoch": 0.5913514210221048,
+      "grad_norm": 0.23101641353282965,
+      "learning_rate": 4.545649838882922e-06,
+      "loss": 0.0006,
+      "step": 9176
+    },
+    {
+      "epoch": 0.591415866469034,
+      "grad_norm": 0.03984360700703188,
+      "learning_rate": 4.544933762978876e-06,
+      "loss": 0.0016,
+      "step": 9177
+    },
+    {
+      "epoch": 0.5914803119159632,
+      "grad_norm": 0.03024855675695247,
+      "learning_rate": 4.54421768707483e-06,
+      "loss": 0.0002,
+      "step": 9178
+    },
+    {
+      "epoch": 0.5915447573628924,
+      "grad_norm": 0.04043637977027526,
+      "learning_rate": 4.5435016111707845e-06,
+      "loss": 0.0004,
+      "step": 9179
+    },
+    {
+      "epoch": 0.5916092028098214,
+      "grad_norm": 0.10753774462695014,
+      "learning_rate": 4.542785535266739e-06,
+      "loss": 0.0001,
+      "step": 9180
+    },
+    {
+      "epoch": 0.5916736482567506,
+      "grad_norm": 0.0025644812183412474,
+      "learning_rate": 4.542069459362693e-06,
+      "loss": 0.0,
+      "step": 9181
+    },
+    {
+      "epoch": 0.5917380937036798,
+      "grad_norm": 0.04065201438857625,
+      "learning_rate": 4.541353383458647e-06,
+      "loss": 0.0001,
+      "step": 9182
+    },
+    {
+      "epoch": 0.591802539150609,
+      "grad_norm": 0.3754980038895867,
+      "learning_rate": 4.540637307554601e-06,
+      "loss": 0.0003,
+      "step": 9183
+    },
+    {
+      "epoch": 0.5918669845975382,
+      "grad_norm": 0.0013786835065639602,
+      "learning_rate": 4.539921231650555e-06,
+      "loss": 0.0,
+      "step": 9184
+    },
+    {
+      "epoch": 0.5919314300444674,
+      "grad_norm": 0.038676694668396946,
+      "learning_rate": 4.5392051557465095e-06,
+      "loss": 0.0018,
+      "step": 9185
+    },
+    {
+      "epoch": 0.5919958754913965,
+      "grad_norm": 0.123739661317091,
+      "learning_rate": 4.538489079842464e-06,
+      "loss": 0.0001,
+      "step": 9186
+    },
+    {
+      "epoch": 0.5920603209383257,
+      "grad_norm": 0.002075990619147738,
+      "learning_rate": 4.537773003938418e-06,
+      "loss": 0.0,
+      "step": 9187
+    },
+    {
+      "epoch": 0.5921247663852549,
+      "grad_norm": 0.0011510265777243546,
+      "learning_rate": 4.5370569280343715e-06,
+      "loss": 0.0,
+      "step": 9188
+    },
+    {
+      "epoch": 0.5921892118321841,
+      "grad_norm": 0.006128596688916639,
+      "learning_rate": 4.536340852130326e-06,
+      "loss": 0.0,
+      "step": 9189
+    },
+    {
+      "epoch": 0.5922536572791133,
+      "grad_norm": 0.3238451195135463,
+      "learning_rate": 4.535624776226281e-06,
+      "loss": 0.0008,
+      "step": 9190
+    },
+    {
+      "epoch": 0.5923181027260424,
+      "grad_norm": 0.033411018542313785,
+      "learning_rate": 4.534908700322234e-06,
+      "loss": 0.0,
+      "step": 9191
+    },
+    {
+      "epoch": 0.5923825481729715,
+      "grad_norm": 0.009916791814087991,
+      "learning_rate": 4.534192624418189e-06,
+      "loss": 0.0,
+      "step": 9192
+    },
+    {
+      "epoch": 0.5924469936199007,
+      "grad_norm": 0.0006763817433519578,
+      "learning_rate": 4.533476548514143e-06,
+      "loss": 0.0,
+      "step": 9193
+    },
+    {
+      "epoch": 0.5925114390668299,
+      "grad_norm": 0.2030172612492698,
+      "learning_rate": 4.532760472610097e-06,
+      "loss": 0.0042,
+      "step": 9194
+    },
+    {
+      "epoch": 0.5925758845137591,
+      "grad_norm": 0.0010908219317394157,
+      "learning_rate": 4.532044396706052e-06,
+      "loss": 0.0,
+      "step": 9195
+    },
+    {
+      "epoch": 0.5926403299606883,
+      "grad_norm": 0.0005182772170067717,
+      "learning_rate": 4.531328320802005e-06,
+      "loss": 0.0,
+      "step": 9196
+    },
+    {
+      "epoch": 0.5927047754076175,
+      "grad_norm": 0.0014441613798425383,
+      "learning_rate": 4.530612244897959e-06,
+      "loss": 0.0,
+      "step": 9197
+    },
+    {
+      "epoch": 0.5927692208545466,
+      "grad_norm": 0.0008751122361331298,
+      "learning_rate": 4.529896168993914e-06,
+      "loss": 0.0,
+      "step": 9198
+    },
+    {
+      "epoch": 0.5928336663014758,
+      "grad_norm": 0.15613020834851898,
+      "learning_rate": 4.529180093089868e-06,
+      "loss": 0.0012,
+      "step": 9199
+    },
+    {
+      "epoch": 0.592898111748405,
+      "grad_norm": 0.023299958553483276,
+      "learning_rate": 4.528464017185822e-06,
+      "loss": 0.0,
+      "step": 9200
+    },
+    {
+      "epoch": 0.5929625571953342,
+      "grad_norm": 0.0002505282700823623,
+      "learning_rate": 4.527747941281776e-06,
+      "loss": 0.0,
+      "step": 9201
+    },
+    {
+      "epoch": 0.5930270026422634,
+      "grad_norm": 0.06823508339202548,
+      "learning_rate": 4.527031865377731e-06,
+      "loss": 0.0002,
+      "step": 9202
+    },
+    {
+      "epoch": 0.5930914480891925,
+      "grad_norm": 0.0023710727571781822,
+      "learning_rate": 4.526315789473685e-06,
+      "loss": 0.0,
+      "step": 9203
+    },
+    {
+      "epoch": 0.5931558935361216,
+      "grad_norm": 0.005031136311208585,
+      "learning_rate": 4.525599713569639e-06,
+      "loss": 0.0,
+      "step": 9204
+    },
+    {
+      "epoch": 0.5932203389830508,
+      "grad_norm": 0.00313223262834699,
+      "learning_rate": 4.524883637665593e-06,
+      "loss": 0.0,
+      "step": 9205
+    },
+    {
+      "epoch": 0.59328478442998,
+      "grad_norm": 0.0023746098856013845,
+      "learning_rate": 4.524167561761547e-06,
+      "loss": 0.0,
+      "step": 9206
+    },
+    {
+      "epoch": 0.5933492298769092,
+      "grad_norm": 0.00026919858289471487,
+      "learning_rate": 4.5234514858575015e-06,
+      "loss": 0.0,
+      "step": 9207
+    },
+    {
+      "epoch": 0.5934136753238384,
+      "grad_norm": 0.004509877341564837,
+      "learning_rate": 4.522735409953456e-06,
+      "loss": 0.0,
+      "step": 9208
+    },
+    {
+      "epoch": 0.5934781207707676,
+      "grad_norm": 0.42020257838400926,
+      "learning_rate": 4.522019334049409e-06,
+      "loss": 0.0015,
+      "step": 9209
+    },
+    {
+      "epoch": 0.5935425662176967,
+      "grad_norm": 0.0028492342946140163,
+      "learning_rate": 4.5213032581453635e-06,
+      "loss": 0.0,
+      "step": 9210
+    },
+    {
+      "epoch": 0.5936070116646259,
+      "grad_norm": 0.09158406548041811,
+      "learning_rate": 4.520587182241318e-06,
+      "loss": 0.0007,
+      "step": 9211
+    },
+    {
+      "epoch": 0.5936714571115551,
+      "grad_norm": 0.005189893702205593,
+      "learning_rate": 4.519871106337272e-06,
+      "loss": 0.0,
+      "step": 9212
+    },
+    {
+      "epoch": 0.5937359025584843,
+      "grad_norm": 0.002655863031524807,
+      "learning_rate": 4.5191550304332264e-06,
+      "loss": 0.0,
+      "step": 9213
+    },
+    {
+      "epoch": 0.5938003480054134,
+      "grad_norm": 0.00017574247321199604,
+      "learning_rate": 4.518438954529181e-06,
+      "loss": 0.0,
+      "step": 9214
+    },
+    {
+      "epoch": 0.5938647934523426,
+      "grad_norm": 0.004720604057724035,
+      "learning_rate": 4.517722878625135e-06,
+      "loss": 0.0,
+      "step": 9215
+    },
+    {
+      "epoch": 0.5939292388992717,
+      "grad_norm": 0.000470269239913421,
+      "learning_rate": 4.517006802721089e-06,
+      "loss": 0.0,
+      "step": 9216
+    },
+    {
+      "epoch": 0.5939936843462009,
+      "grad_norm": 0.004527785334700599,
+      "learning_rate": 4.516290726817043e-06,
+      "loss": 0.0,
+      "step": 9217
+    },
+    {
+      "epoch": 0.5940581297931301,
+      "grad_norm": 0.22733169733458167,
+      "learning_rate": 4.515574650912997e-06,
+      "loss": 0.0008,
+      "step": 9218
+    },
+    {
+      "epoch": 0.5941225752400593,
+      "grad_norm": 0.0009303410068623072,
+      "learning_rate": 4.514858575008951e-06,
+      "loss": 0.0,
+      "step": 9219
+    },
+    {
+      "epoch": 0.5941870206869885,
+      "grad_norm": 0.006910788414918203,
+      "learning_rate": 4.514142499104906e-06,
+      "loss": 0.0001,
+      "step": 9220
+    },
+    {
+      "epoch": 0.5942514661339177,
+      "grad_norm": 0.29944842913354613,
+      "learning_rate": 4.513426423200859e-06,
+      "loss": 0.0009,
+      "step": 9221
+    },
+    {
+      "epoch": 0.5943159115808468,
+      "grad_norm": 0.015923637568431896,
+      "learning_rate": 4.5127103472968134e-06,
+      "loss": 0.0,
+      "step": 9222
+    },
+    {
+      "epoch": 0.594380357027776,
+      "grad_norm": 0.1276660123801875,
+      "learning_rate": 4.511994271392768e-06,
+      "loss": 0.0002,
+      "step": 9223
+    },
+    {
+      "epoch": 0.5944448024747052,
+      "grad_norm": 0.05243302043805129,
+      "learning_rate": 4.511278195488722e-06,
+      "loss": 0.0004,
+      "step": 9224
+    },
+    {
+      "epoch": 0.5945092479216343,
+      "grad_norm": 0.02153507462951738,
+      "learning_rate": 4.510562119584676e-06,
+      "loss": 0.0001,
+      "step": 9225
+    },
+    {
+      "epoch": 0.5945736933685635,
+      "grad_norm": 0.0004915653704900115,
+      "learning_rate": 4.509846043680631e-06,
+      "loss": 0.0,
+      "step": 9226
+    },
+    {
+      "epoch": 0.5946381388154927,
+      "grad_norm": 0.00825432485504778,
+      "learning_rate": 4.509129967776585e-06,
+      "loss": 0.0,
+      "step": 9227
+    },
+    {
+      "epoch": 0.5947025842624218,
+      "grad_norm": 0.08825993565435779,
+      "learning_rate": 4.508413891872539e-06,
+      "loss": 0.0006,
+      "step": 9228
+    },
+    {
+      "epoch": 0.594767029709351,
+      "grad_norm": 8.340152364850253e-05,
+      "learning_rate": 4.507697815968493e-06,
+      "loss": 0.0,
+      "step": 9229
+    },
+    {
+      "epoch": 0.5948314751562802,
+      "grad_norm": 0.013485439351184434,
+      "learning_rate": 4.506981740064447e-06,
+      "loss": 0.0001,
+      "step": 9230
+    },
+    {
+      "epoch": 0.5948959206032094,
+      "grad_norm": 0.005780882656582949,
+      "learning_rate": 4.506265664160401e-06,
+      "loss": 0.0001,
+      "step": 9231
+    },
+    {
+      "epoch": 0.5949603660501386,
+      "grad_norm": 0.01674859071076599,
+      "learning_rate": 4.5055495882563556e-06,
+      "loss": 0.0,
+      "step": 9232
+    },
+    {
+      "epoch": 0.5950248114970678,
+      "grad_norm": 0.0019716342943576176,
+      "learning_rate": 4.50483351235231e-06,
+      "loss": 0.0,
+      "step": 9233
+    },
+    {
+      "epoch": 0.595089256943997,
+      "grad_norm": 8.759305110141916e-05,
+      "learning_rate": 4.504117436448263e-06,
+      "loss": 0.0,
+      "step": 9234
+    },
+    {
+      "epoch": 0.5951537023909261,
+      "grad_norm": 0.002496013202718065,
+      "learning_rate": 4.503401360544218e-06,
+      "loss": 0.0,
+      "step": 9235
+    },
+    {
+      "epoch": 0.5952181478378552,
+      "grad_norm": 0.00025275087997574504,
+      "learning_rate": 4.502685284640172e-06,
+      "loss": 0.0,
+      "step": 9236
+    },
+    {
+      "epoch": 0.5952825932847844,
+      "grad_norm": 0.0024015228303083213,
+      "learning_rate": 4.501969208736126e-06,
+      "loss": 0.0,
+      "step": 9237
+    },
+    {
+      "epoch": 0.5953470387317136,
+      "grad_norm": 0.1258258953590908,
+      "learning_rate": 4.5012531328320805e-06,
+      "loss": 0.0003,
+      "step": 9238
+    },
+    {
+      "epoch": 0.5954114841786428,
+      "grad_norm": 0.0032277228050439774,
+      "learning_rate": 4.500537056928035e-06,
+      "loss": 0.0,
+      "step": 9239
+    },
+    {
+      "epoch": 0.5954759296255719,
+      "grad_norm": 0.0014017478054440399,
+      "learning_rate": 4.499820981023989e-06,
+      "loss": 0.0,
+      "step": 9240
+    },
+    {
+      "epoch": 0.5955403750725011,
+      "grad_norm": 0.0043084187561121135,
+      "learning_rate": 4.499104905119943e-06,
+      "loss": 0.0,
+      "step": 9241
+    },
+    {
+      "epoch": 0.5956048205194303,
+      "grad_norm": 0.014204112734319778,
+      "learning_rate": 4.498388829215897e-06,
+      "loss": 0.0,
+      "step": 9242
+    },
+    {
+      "epoch": 0.5956692659663595,
+      "grad_norm": 0.0012018417368002514,
+      "learning_rate": 4.497672753311851e-06,
+      "loss": 0.0,
+      "step": 9243
+    },
+    {
+      "epoch": 0.5957337114132887,
+      "grad_norm": 0.001397111472659201,
+      "learning_rate": 4.4969566774078055e-06,
+      "loss": 0.0,
+      "step": 9244
+    },
+    {
+      "epoch": 0.5957981568602179,
+      "grad_norm": 0.004663769686834882,
+      "learning_rate": 4.49624060150376e-06,
+      "loss": 0.0,
+      "step": 9245
+    },
+    {
+      "epoch": 0.595862602307147,
+      "grad_norm": 0.056992404426177555,
+      "learning_rate": 4.495524525599714e-06,
+      "loss": 0.0004,
+      "step": 9246
+    },
+    {
+      "epoch": 0.5959270477540761,
+      "grad_norm": 0.002981061286054127,
+      "learning_rate": 4.4948084496956675e-06,
+      "loss": 0.0,
+      "step": 9247
+    },
+    {
+      "epoch": 0.5959914932010053,
+      "grad_norm": 0.903715832698863,
+      "learning_rate": 4.494092373791623e-06,
+      "loss": 0.0048,
+      "step": 9248
+    },
+    {
+      "epoch": 0.5960559386479345,
+      "grad_norm": 0.04485061684727686,
+      "learning_rate": 4.493376297887577e-06,
+      "loss": 0.0001,
+      "step": 9249
+    },
+    {
+      "epoch": 0.5961203840948637,
+      "grad_norm": 0.014483704247325296,
+      "learning_rate": 4.49266022198353e-06,
+      "loss": 0.0,
+      "step": 9250
+    },
+    {
+      "epoch": 0.5961848295417929,
+      "grad_norm": 0.0025339210037765942,
+      "learning_rate": 4.491944146079485e-06,
+      "loss": 0.0,
+      "step": 9251
+    },
+    {
+      "epoch": 0.596249274988722,
+      "grad_norm": 0.02972256912942362,
+      "learning_rate": 4.491228070175439e-06,
+      "loss": 0.0,
+      "step": 9252
+    },
+    {
+      "epoch": 0.5963137204356512,
+      "grad_norm": 0.003333447946224778,
+      "learning_rate": 4.490511994271393e-06,
+      "loss": 0.0,
+      "step": 9253
+    },
+    {
+      "epoch": 0.5963781658825804,
+      "grad_norm": 0.002218025045587023,
+      "learning_rate": 4.489795918367348e-06,
+      "loss": 0.0,
+      "step": 9254
+    },
+    {
+      "epoch": 0.5964426113295096,
+      "grad_norm": 0.00025620211906760514,
+      "learning_rate": 4.489079842463301e-06,
+      "loss": 0.0,
+      "step": 9255
+    },
+    {
+      "epoch": 0.5965070567764388,
+      "grad_norm": 0.3525647500102783,
+      "learning_rate": 4.488363766559255e-06,
+      "loss": 0.0013,
+      "step": 9256
+    },
+    {
+      "epoch": 0.596571502223368,
+      "grad_norm": 0.0009345009565512394,
+      "learning_rate": 4.48764769065521e-06,
+      "loss": 0.0,
+      "step": 9257
+    },
+    {
+      "epoch": 0.596635947670297,
+      "grad_norm": 0.2165891494152304,
+      "learning_rate": 4.486931614751164e-06,
+      "loss": 0.0006,
+      "step": 9258
+    },
+    {
+      "epoch": 0.5967003931172262,
+      "grad_norm": 0.05442176698580331,
+      "learning_rate": 4.486215538847118e-06,
+      "loss": 0.0001,
+      "step": 9259
+    },
+    {
+      "epoch": 0.5967648385641554,
+      "grad_norm": 0.027872295910078077,
+      "learning_rate": 4.4854994629430725e-06,
+      "loss": 0.0,
+      "step": 9260
+    },
+    {
+      "epoch": 0.5968292840110846,
+      "grad_norm": 0.0002799742786045959,
+      "learning_rate": 4.484783387039027e-06,
+      "loss": 0.0,
+      "step": 9261
+    },
+    {
+      "epoch": 0.5968937294580138,
+      "grad_norm": 0.00018996156454566542,
+      "learning_rate": 4.484067311134981e-06,
+      "loss": 0.0,
+      "step": 9262
+    },
+    {
+      "epoch": 0.596958174904943,
+      "grad_norm": 0.14084470328682747,
+      "learning_rate": 4.483351235230935e-06,
+      "loss": 0.0012,
+      "step": 9263
+    },
+    {
+      "epoch": 0.5970226203518721,
+      "grad_norm": 0.001003408319839176,
+      "learning_rate": 4.482635159326889e-06,
+      "loss": 0.0,
+      "step": 9264
+    },
+    {
+      "epoch": 0.5970870657988013,
+      "grad_norm": 0.006807802621411491,
+      "learning_rate": 4.481919083422843e-06,
+      "loss": 0.0,
+      "step": 9265
+    },
+    {
+      "epoch": 0.5971515112457305,
+      "grad_norm": 0.020114278322576007,
+      "learning_rate": 4.4812030075187975e-06,
+      "loss": 0.0001,
+      "step": 9266
+    },
+    {
+      "epoch": 0.5972159566926597,
+      "grad_norm": 0.00017411804282206113,
+      "learning_rate": 4.480486931614752e-06,
+      "loss": 0.0,
+      "step": 9267
+    },
+    {
+      "epoch": 0.5972804021395889,
+      "grad_norm": 0.01873928089012201,
+      "learning_rate": 4.479770855710705e-06,
+      "loss": 0.0,
+      "step": 9268
+    },
+    {
+      "epoch": 0.597344847586518,
+      "grad_norm": 0.0011347830260341888,
+      "learning_rate": 4.4790547798066595e-06,
+      "loss": 0.0,
+      "step": 9269
+    },
+    {
+      "epoch": 0.5974092930334471,
+      "grad_norm": 0.153683499525407,
+      "learning_rate": 4.478338703902614e-06,
+      "loss": 0.0005,
+      "step": 9270
+    },
+    {
+      "epoch": 0.5974737384803763,
+      "grad_norm": 0.005827371173133551,
+      "learning_rate": 4.477622627998568e-06,
+      "loss": 0.0,
+      "step": 9271
+    },
+    {
+      "epoch": 0.5975381839273055,
+      "grad_norm": 0.0005971115847287093,
+      "learning_rate": 4.4769065520945224e-06,
+      "loss": 0.0,
+      "step": 9272
+    },
+    {
+      "epoch": 0.5976026293742347,
+      "grad_norm": 0.004836487626164628,
+      "learning_rate": 4.476190476190477e-06,
+      "loss": 0.0,
+      "step": 9273
+    },
+    {
+      "epoch": 0.5976670748211639,
+      "grad_norm": 0.0005959039040827599,
+      "learning_rate": 4.475474400286431e-06,
+      "loss": 0.0,
+      "step": 9274
+    },
+    {
+      "epoch": 0.5977315202680931,
+      "grad_norm": 0.3020565196447107,
+      "learning_rate": 4.474758324382385e-06,
+      "loss": 0.0003,
+      "step": 9275
+    },
+    {
+      "epoch": 0.5977959657150222,
+      "grad_norm": 0.001559705042993861,
+      "learning_rate": 4.474042248478339e-06,
+      "loss": 0.0,
+      "step": 9276
+    },
+    {
+      "epoch": 0.5978604111619514,
+      "grad_norm": 1.8227666892697204e-05,
+      "learning_rate": 4.473326172574293e-06,
+      "loss": 0.0,
+      "step": 9277
+    },
+    {
+      "epoch": 0.5979248566088806,
+      "grad_norm": 0.0013650415837087972,
+      "learning_rate": 4.472610096670247e-06,
+      "loss": 0.0,
+      "step": 9278
+    },
+    {
+      "epoch": 0.5979893020558098,
+      "grad_norm": 0.16759155859822403,
+      "learning_rate": 4.471894020766202e-06,
+      "loss": 0.0009,
+      "step": 9279
+    },
+    {
+      "epoch": 0.598053747502739,
+      "grad_norm": 0.0008573025217973782,
+      "learning_rate": 4.471177944862156e-06,
+      "loss": 0.0,
+      "step": 9280
+    },
+    {
+      "epoch": 0.5981181929496681,
+      "grad_norm": 4.933914884938549e-05,
+      "learning_rate": 4.4704618689581094e-06,
+      "loss": 0.0,
+      "step": 9281
+    },
+    {
+      "epoch": 0.5981826383965972,
+      "grad_norm": 0.0037086977972102943,
+      "learning_rate": 4.469745793054064e-06,
+      "loss": 0.0,
+      "step": 9282
+    },
+    {
+      "epoch": 0.5982470838435264,
+      "grad_norm": 0.002770381211495419,
+      "learning_rate": 4.469029717150019e-06,
+      "loss": 0.0,
+      "step": 9283
+    },
+    {
+      "epoch": 0.5983115292904556,
+      "grad_norm": 0.002774506415338846,
+      "learning_rate": 4.468313641245972e-06,
+      "loss": 0.0,
+      "step": 9284
+    },
+    {
+      "epoch": 0.5983759747373848,
+      "grad_norm": 0.0010348561652971477,
+      "learning_rate": 4.467597565341927e-06,
+      "loss": 0.0,
+      "step": 9285
+    },
+    {
+      "epoch": 0.598440420184314,
+      "grad_norm": 0.00225808661724457,
+      "learning_rate": 4.466881489437881e-06,
+      "loss": 0.0,
+      "step": 9286
+    },
+    {
+      "epoch": 0.5985048656312432,
+      "grad_norm": 0.0001941368847980672,
+      "learning_rate": 4.466165413533835e-06,
+      "loss": 0.0,
+      "step": 9287
+    },
+    {
+      "epoch": 0.5985693110781724,
+      "grad_norm": 6.751922171693977e-05,
+      "learning_rate": 4.4654493376297895e-06,
+      "loss": 0.0,
+      "step": 9288
+    },
+    {
+      "epoch": 0.5986337565251015,
+      "grad_norm": 0.0003746486823726859,
+      "learning_rate": 4.464733261725743e-06,
+      "loss": 0.0,
+      "step": 9289
+    },
+    {
+      "epoch": 0.5986982019720307,
+      "grad_norm": 0.0026928507882407063,
+      "learning_rate": 4.464017185821697e-06,
+      "loss": 0.0,
+      "step": 9290
+    },
+    {
+      "epoch": 0.5987626474189599,
+      "grad_norm": 0.018445121237624786,
+      "learning_rate": 4.4633011099176516e-06,
+      "loss": 0.0002,
+      "step": 9291
+    },
+    {
+      "epoch": 0.598827092865889,
+      "grad_norm": 0.00025725082956279275,
+      "learning_rate": 4.462585034013606e-06,
+      "loss": 0.0,
+      "step": 9292
+    },
+    {
+      "epoch": 0.5988915383128182,
+      "grad_norm": 0.0010799365281890174,
+      "learning_rate": 4.46186895810956e-06,
+      "loss": 0.0,
+      "step": 9293
+    },
+    {
+      "epoch": 0.5989559837597473,
+      "grad_norm": 0.007159703054942445,
+      "learning_rate": 4.461152882205514e-06,
+      "loss": 0.0,
+      "step": 9294
+    },
+    {
+      "epoch": 0.5990204292066765,
+      "grad_norm": 1.2268386049962541e-05,
+      "learning_rate": 4.460436806301469e-06,
+      "loss": 0.0,
+      "step": 9295
+    },
+    {
+      "epoch": 0.5990848746536057,
+      "grad_norm": 0.06045411872724877,
+      "learning_rate": 4.459720730397423e-06,
+      "loss": 0.0006,
+      "step": 9296
+    },
+    {
+      "epoch": 0.5991493201005349,
+      "grad_norm": 0.19484259344276214,
+      "learning_rate": 4.4590046544933765e-06,
+      "loss": 0.0031,
+      "step": 9297
+    },
+    {
+      "epoch": 0.5992137655474641,
+      "grad_norm": 0.00011573336964747802,
+      "learning_rate": 4.458288578589331e-06,
+      "loss": 0.0,
+      "step": 9298
+    },
+    {
+      "epoch": 0.5992782109943933,
+      "grad_norm": 0.00013179948953989136,
+      "learning_rate": 4.457572502685285e-06,
+      "loss": 0.0,
+      "step": 9299
+    },
+    {
+      "epoch": 0.5993426564413225,
+      "grad_norm": 0.08617986833949201,
+      "learning_rate": 4.456856426781239e-06,
+      "loss": 0.0001,
+      "step": 9300
+    },
+    {
+      "epoch": 0.5994071018882516,
+      "grad_norm": 2.1917965791246905,
+      "learning_rate": 4.456140350877194e-06,
+      "loss": 0.0069,
+      "step": 9301
+    },
+    {
+      "epoch": 0.5994715473351808,
+      "grad_norm": 0.005079315964603255,
+      "learning_rate": 4.455424274973147e-06,
+      "loss": 0.0,
+      "step": 9302
+    },
+    {
+      "epoch": 0.5995359927821099,
+      "grad_norm": 0.0005311949623312273,
+      "learning_rate": 4.4547081990691015e-06,
+      "loss": 0.0,
+      "step": 9303
+    },
+    {
+      "epoch": 0.5996004382290391,
+      "grad_norm": 0.00018466529970892427,
+      "learning_rate": 4.453992123165056e-06,
+      "loss": 0.0,
+      "step": 9304
+    },
+    {
+      "epoch": 0.5996648836759683,
+      "grad_norm": 0.0024224931746407233,
+      "learning_rate": 4.45327604726101e-06,
+      "loss": 0.0,
+      "step": 9305
+    },
+    {
+      "epoch": 0.5997293291228974,
+      "grad_norm": 9.214679526500649e-05,
+      "learning_rate": 4.452559971356964e-06,
+      "loss": 0.0,
+      "step": 9306
+    },
+    {
+      "epoch": 0.5997937745698266,
+      "grad_norm": 0.00013752331314721667,
+      "learning_rate": 4.451843895452919e-06,
+      "loss": 0.0,
+      "step": 9307
+    },
+    {
+      "epoch": 0.5998582200167558,
+      "grad_norm": 0.0012078464524044698,
+      "learning_rate": 4.451127819548873e-06,
+      "loss": 0.0,
+      "step": 9308
+    },
+    {
+      "epoch": 0.599922665463685,
+      "grad_norm": 0.02017807043125452,
+      "learning_rate": 4.450411743644827e-06,
+      "loss": 0.0001,
+      "step": 9309
+    },
+    {
+      "epoch": 0.5999871109106142,
+      "grad_norm": 0.12984587208589446,
+      "learning_rate": 4.449695667740781e-06,
+      "loss": 0.0004,
+      "step": 9310
+    },
+    {
+      "epoch": 0.6000515563575434,
+      "grad_norm": 0.04357864342243002,
+      "learning_rate": 4.448979591836735e-06,
+      "loss": 0.0001,
+      "step": 9311
+    },
+    {
+      "epoch": 0.6001160018044726,
+      "grad_norm": 0.009136213091560104,
+      "learning_rate": 4.448263515932689e-06,
+      "loss": 0.0,
+      "step": 9312
+    },
+    {
+      "epoch": 0.6001804472514017,
+      "grad_norm": 0.03765791702092579,
+      "learning_rate": 4.447547440028644e-06,
+      "loss": 0.0,
+      "step": 9313
+    },
+    {
+      "epoch": 0.6002448926983308,
+      "grad_norm": 0.13066622728000005,
+      "learning_rate": 4.446831364124598e-06,
+      "loss": 0.0004,
+      "step": 9314
+    },
+    {
+      "epoch": 0.60030933814526,
+      "grad_norm": 0.033811319537168136,
+      "learning_rate": 4.446115288220551e-06,
+      "loss": 0.0001,
+      "step": 9315
+    },
+    {
+      "epoch": 0.6003737835921892,
+      "grad_norm": 0.003377651363750598,
+      "learning_rate": 4.445399212316506e-06,
+      "loss": 0.0,
+      "step": 9316
+    },
+    {
+      "epoch": 0.6004382290391184,
+      "grad_norm": 0.0012151443025219713,
+      "learning_rate": 4.44468313641246e-06,
+      "loss": 0.0,
+      "step": 9317
+    },
+    {
+      "epoch": 0.6005026744860475,
+      "grad_norm": 0.00030076058428306606,
+      "learning_rate": 4.443967060508414e-06,
+      "loss": 0.0,
+      "step": 9318
+    },
+    {
+      "epoch": 0.6005671199329767,
+      "grad_norm": 0.0015305195960151965,
+      "learning_rate": 4.4432509846043685e-06,
+      "loss": 0.0,
+      "step": 9319
+    },
+    {
+      "epoch": 0.6006315653799059,
+      "grad_norm": 0.439756161792432,
+      "learning_rate": 4.442534908700323e-06,
+      "loss": 0.0012,
+      "step": 9320
+    },
+    {
+      "epoch": 0.6006960108268351,
+      "grad_norm": 0.00840660182838595,
+      "learning_rate": 4.441818832796277e-06,
+      "loss": 0.0,
+      "step": 9321
+    },
+    {
+      "epoch": 0.6007604562737643,
+      "grad_norm": 0.024425231047921048,
+      "learning_rate": 4.4411027568922314e-06,
+      "loss": 0.0002,
+      "step": 9322
+    },
+    {
+      "epoch": 0.6008249017206935,
+      "grad_norm": 0.00420921162090846,
+      "learning_rate": 4.440386680988185e-06,
+      "loss": 0.0,
+      "step": 9323
+    },
+    {
+      "epoch": 0.6008893471676227,
+      "grad_norm": 0.03988730496138384,
+      "learning_rate": 4.439670605084139e-06,
+      "loss": 0.0001,
+      "step": 9324
+    },
+    {
+      "epoch": 0.6009537926145517,
+      "grad_norm": 0.0012243449380229714,
+      "learning_rate": 4.4389545291800935e-06,
+      "loss": 0.0,
+      "step": 9325
+    },
+    {
+      "epoch": 0.6010182380614809,
+      "grad_norm": 0.0006668839996335907,
+      "learning_rate": 4.438238453276048e-06,
+      "loss": 0.0,
+      "step": 9326
+    },
+    {
+      "epoch": 0.6010826835084101,
+      "grad_norm": 0.0001269798514468688,
+      "learning_rate": 4.437522377372001e-06,
+      "loss": 0.0,
+      "step": 9327
+    },
+    {
+      "epoch": 0.6011471289553393,
+      "grad_norm": 0.006069018507372652,
+      "learning_rate": 4.4368063014679555e-06,
+      "loss": 0.0,
+      "step": 9328
+    },
+    {
+      "epoch": 0.6012115744022685,
+      "grad_norm": 0.009952276861537372,
+      "learning_rate": 4.43609022556391e-06,
+      "loss": 0.0,
+      "step": 9329
+    },
+    {
+      "epoch": 0.6012760198491977,
+      "grad_norm": 0.0001698301808685413,
+      "learning_rate": 4.435374149659865e-06,
+      "loss": 0.0,
+      "step": 9330
+    },
+    {
+      "epoch": 0.6013404652961268,
+      "grad_norm": 0.0007671255886284143,
+      "learning_rate": 4.4346580737558184e-06,
+      "loss": 0.0,
+      "step": 9331
+    },
+    {
+      "epoch": 0.601404910743056,
+      "grad_norm": 0.019085248753486946,
+      "learning_rate": 4.433941997851773e-06,
+      "loss": 0.0002,
+      "step": 9332
+    },
+    {
+      "epoch": 0.6014693561899852,
+      "grad_norm": 0.0005585553011050799,
+      "learning_rate": 4.433225921947727e-06,
+      "loss": 0.0,
+      "step": 9333
+    },
+    {
+      "epoch": 0.6015338016369144,
+      "grad_norm": 0.010856243724535,
+      "learning_rate": 4.432509846043681e-06,
+      "loss": 0.0001,
+      "step": 9334
+    },
+    {
+      "epoch": 0.6015982470838436,
+      "grad_norm": 0.0002317457563166157,
+      "learning_rate": 4.431793770139635e-06,
+      "loss": 0.0,
+      "step": 9335
+    },
+    {
+      "epoch": 0.6016626925307726,
+      "grad_norm": 0.0065507252221229315,
+      "learning_rate": 4.431077694235589e-06,
+      "loss": 0.0001,
+      "step": 9336
+    },
+    {
+      "epoch": 0.6017271379777018,
+      "grad_norm": 0.00010838378621555254,
+      "learning_rate": 4.430361618331543e-06,
+      "loss": 0.0,
+      "step": 9337
+    },
+    {
+      "epoch": 0.601791583424631,
+      "grad_norm": 0.0017090922054476857,
+      "learning_rate": 4.429645542427498e-06,
+      "loss": 0.0,
+      "step": 9338
+    },
+    {
+      "epoch": 0.6018560288715602,
+      "grad_norm": 0.009171633804983492,
+      "learning_rate": 4.428929466523452e-06,
+      "loss": 0.0,
+      "step": 9339
+    },
+    {
+      "epoch": 0.6019204743184894,
+      "grad_norm": 0.0022138650096616835,
+      "learning_rate": 4.4282133906194054e-06,
+      "loss": 0.0,
+      "step": 9340
+    },
+    {
+      "epoch": 0.6019849197654186,
+      "grad_norm": 0.00016321892076394534,
+      "learning_rate": 4.42749731471536e-06,
+      "loss": 0.0,
+      "step": 9341
+    },
+    {
+      "epoch": 0.6020493652123478,
+      "grad_norm": 0.7931170395375743,
+      "learning_rate": 4.426781238811315e-06,
+      "loss": 0.0087,
+      "step": 9342
+    },
+    {
+      "epoch": 0.6021138106592769,
+      "grad_norm": 0.0027087375142905547,
+      "learning_rate": 4.426065162907268e-06,
+      "loss": 0.0,
+      "step": 9343
+    },
+    {
+      "epoch": 0.6021782561062061,
+      "grad_norm": 0.00030620151578539744,
+      "learning_rate": 4.425349087003223e-06,
+      "loss": 0.0,
+      "step": 9344
+    },
+    {
+      "epoch": 0.6022427015531353,
+      "grad_norm": 0.35033655461765545,
+      "learning_rate": 4.424633011099177e-06,
+      "loss": 0.007,
+      "step": 9345
+    },
+    {
+      "epoch": 0.6023071470000645,
+      "grad_norm": 0.0008401919777756227,
+      "learning_rate": 4.423916935195131e-06,
+      "loss": 0.0,
+      "step": 9346
+    },
+    {
+      "epoch": 0.6023715924469936,
+      "grad_norm": 0.00012728167519971065,
+      "learning_rate": 4.4232008592910855e-06,
+      "loss": 0.0,
+      "step": 9347
+    },
+    {
+      "epoch": 0.6024360378939227,
+      "grad_norm": 0.012302640572469041,
+      "learning_rate": 4.422484783387039e-06,
+      "loss": 0.0,
+      "step": 9348
+    },
+    {
+      "epoch": 0.6025004833408519,
+      "grad_norm": 0.08447763946476083,
+      "learning_rate": 4.421768707482993e-06,
+      "loss": 0.001,
+      "step": 9349
+    },
+    {
+      "epoch": 0.6025649287877811,
+      "grad_norm": 0.0011682809611789449,
+      "learning_rate": 4.4210526315789476e-06,
+      "loss": 0.0,
+      "step": 9350
+    },
+    {
+      "epoch": 0.6026293742347103,
+      "grad_norm": 0.0011277098483875475,
+      "learning_rate": 4.420336555674902e-06,
+      "loss": 0.0,
+      "step": 9351
+    },
+    {
+      "epoch": 0.6026938196816395,
+      "grad_norm": 0.011400610017446352,
+      "learning_rate": 4.419620479770856e-06,
+      "loss": 0.0,
+      "step": 9352
+    },
+    {
+      "epoch": 0.6027582651285687,
+      "grad_norm": 0.0541042169718465,
+      "learning_rate": 4.41890440386681e-06,
+      "loss": 0.0001,
+      "step": 9353
+    },
+    {
+      "epoch": 0.6028227105754979,
+      "grad_norm": 0.0004733979353402136,
+      "learning_rate": 4.418188327962765e-06,
+      "loss": 0.0,
+      "step": 9354
+    },
+    {
+      "epoch": 0.602887156022427,
+      "grad_norm": 0.00089674663328231,
+      "learning_rate": 4.417472252058719e-06,
+      "loss": 0.0,
+      "step": 9355
+    },
+    {
+      "epoch": 0.6029516014693562,
+      "grad_norm": 0.0038075149394037884,
+      "learning_rate": 4.4167561761546725e-06,
+      "loss": 0.0,
+      "step": 9356
+    },
+    {
+      "epoch": 0.6030160469162854,
+      "grad_norm": 0.0006955252692623161,
+      "learning_rate": 4.416040100250627e-06,
+      "loss": 0.0,
+      "step": 9357
+    },
+    {
+      "epoch": 0.6030804923632146,
+      "grad_norm": 0.00020624013289641623,
+      "learning_rate": 4.415324024346581e-06,
+      "loss": 0.0,
+      "step": 9358
+    },
+    {
+      "epoch": 0.6031449378101437,
+      "grad_norm": 0.048323787565792246,
+      "learning_rate": 4.414607948442535e-06,
+      "loss": 0.0002,
+      "step": 9359
+    },
+    {
+      "epoch": 0.6032093832570729,
+      "grad_norm": 0.0012364503824618905,
+      "learning_rate": 4.41389187253849e-06,
+      "loss": 0.0,
+      "step": 9360
+    },
+    {
+      "epoch": 0.603273828704002,
+      "grad_norm": 0.6295609472432298,
+      "learning_rate": 4.413175796634443e-06,
+      "loss": 0.001,
+      "step": 9361
+    },
+    {
+      "epoch": 0.6033382741509312,
+      "grad_norm": 7.311658614915109e-05,
+      "learning_rate": 4.4124597207303975e-06,
+      "loss": 0.0,
+      "step": 9362
+    },
+    {
+      "epoch": 0.6034027195978604,
+      "grad_norm": 0.00035291927788144885,
+      "learning_rate": 4.411743644826352e-06,
+      "loss": 0.0,
+      "step": 9363
+    },
+    {
+      "epoch": 0.6034671650447896,
+      "grad_norm": 0.0010846684443867236,
+      "learning_rate": 4.411027568922306e-06,
+      "loss": 0.0,
+      "step": 9364
+    },
+    {
+      "epoch": 0.6035316104917188,
+      "grad_norm": 0.00017466946654724934,
+      "learning_rate": 4.41031149301826e-06,
+      "loss": 0.0,
+      "step": 9365
+    },
+    {
+      "epoch": 0.603596055938648,
+      "grad_norm": 0.001453067649005503,
+      "learning_rate": 4.409595417114215e-06,
+      "loss": 0.0,
+      "step": 9366
+    },
+    {
+      "epoch": 0.6036605013855771,
+      "grad_norm": 0.00018182376128211283,
+      "learning_rate": 4.408879341210169e-06,
+      "loss": 0.0,
+      "step": 9367
+    },
+    {
+      "epoch": 0.6037249468325063,
+      "grad_norm": 0.011400766377271708,
+      "learning_rate": 4.408163265306123e-06,
+      "loss": 0.0,
+      "step": 9368
+    },
+    {
+      "epoch": 0.6037893922794355,
+      "grad_norm": 0.4758228127251253,
+      "learning_rate": 4.407447189402077e-06,
+      "loss": 0.0008,
+      "step": 9369
+    },
+    {
+      "epoch": 0.6038538377263646,
+      "grad_norm": 7.137859350053377e-05,
+      "learning_rate": 4.406731113498031e-06,
+      "loss": 0.0,
+      "step": 9370
+    },
+    {
+      "epoch": 0.6039182831732938,
+      "grad_norm": 0.002106704556151511,
+      "learning_rate": 4.406015037593985e-06,
+      "loss": 0.0,
+      "step": 9371
+    },
+    {
+      "epoch": 0.603982728620223,
+      "grad_norm": 0.01561994489465668,
+      "learning_rate": 4.40529896168994e-06,
+      "loss": 0.0,
+      "step": 9372
+    },
+    {
+      "epoch": 0.6040471740671521,
+      "grad_norm": 0.009708016318482603,
+      "learning_rate": 4.404582885785894e-06,
+      "loss": 0.0,
+      "step": 9373
+    },
+    {
+      "epoch": 0.6041116195140813,
+      "grad_norm": 0.001466915021623313,
+      "learning_rate": 4.403866809881847e-06,
+      "loss": 0.0,
+      "step": 9374
+    },
+    {
+      "epoch": 0.6041760649610105,
+      "grad_norm": 0.0007412942422267652,
+      "learning_rate": 4.403150733977802e-06,
+      "loss": 0.0,
+      "step": 9375
+    },
+    {
+      "epoch": 0.6042405104079397,
+      "grad_norm": 0.005260947110871978,
+      "learning_rate": 4.402434658073756e-06,
+      "loss": 0.0,
+      "step": 9376
+    },
+    {
+      "epoch": 0.6043049558548689,
+      "grad_norm": 5.846787475238024e-05,
+      "learning_rate": 4.40171858216971e-06,
+      "loss": 0.0,
+      "step": 9377
+    },
+    {
+      "epoch": 0.6043694013017981,
+      "grad_norm": 0.0029260585270888422,
+      "learning_rate": 4.4010025062656645e-06,
+      "loss": 0.0,
+      "step": 9378
+    },
+    {
+      "epoch": 0.6044338467487272,
+      "grad_norm": 0.0002419209646407248,
+      "learning_rate": 4.400286430361619e-06,
+      "loss": 0.0,
+      "step": 9379
+    },
+    {
+      "epoch": 0.6044982921956564,
+      "grad_norm": 0.0003612351482923959,
+      "learning_rate": 4.399570354457573e-06,
+      "loss": 0.0,
+      "step": 9380
+    },
+    {
+      "epoch": 0.6045627376425855,
+      "grad_norm": 9.690475230580101e-05,
+      "learning_rate": 4.3988542785535274e-06,
+      "loss": 0.0,
+      "step": 9381
+    },
+    {
+      "epoch": 0.6046271830895147,
+      "grad_norm": 0.11312480103886216,
+      "learning_rate": 4.398138202649481e-06,
+      "loss": 0.0004,
+      "step": 9382
+    },
+    {
+      "epoch": 0.6046916285364439,
+      "grad_norm": 0.0003024319588343002,
+      "learning_rate": 4.397422126745435e-06,
+      "loss": 0.0,
+      "step": 9383
+    },
+    {
+      "epoch": 0.604756073983373,
+      "grad_norm": 0.0012353313973243075,
+      "learning_rate": 4.3967060508413895e-06,
+      "loss": 0.0,
+      "step": 9384
+    },
+    {
+      "epoch": 0.6048205194303022,
+      "grad_norm": 0.012779457318109045,
+      "learning_rate": 4.395989974937344e-06,
+      "loss": 0.0002,
+      "step": 9385
+    },
+    {
+      "epoch": 0.6048849648772314,
+      "grad_norm": 0.0009381798352755683,
+      "learning_rate": 4.395273899033298e-06,
+      "loss": 0.0,
+      "step": 9386
+    },
+    {
+      "epoch": 0.6049494103241606,
+      "grad_norm": 0.00010893321625617296,
+      "learning_rate": 4.3945578231292515e-06,
+      "loss": 0.0,
+      "step": 9387
+    },
+    {
+      "epoch": 0.6050138557710898,
+      "grad_norm": 0.003997170920714122,
+      "learning_rate": 4.393841747225206e-06,
+      "loss": 0.0,
+      "step": 9388
+    },
+    {
+      "epoch": 0.605078301218019,
+      "grad_norm": 8.38234992955942e-05,
+      "learning_rate": 4.393125671321161e-06,
+      "loss": 0.0,
+      "step": 9389
+    },
+    {
+      "epoch": 0.6051427466649482,
+      "grad_norm": 0.0005577757944864189,
+      "learning_rate": 4.3924095954171144e-06,
+      "loss": 0.0,
+      "step": 9390
+    },
+    {
+      "epoch": 0.6052071921118773,
+      "grad_norm": 0.0007627417636095448,
+      "learning_rate": 4.391693519513069e-06,
+      "loss": 0.0,
+      "step": 9391
+    },
+    {
+      "epoch": 0.6052716375588064,
+      "grad_norm": 0.00022876330997271556,
+      "learning_rate": 4.390977443609023e-06,
+      "loss": 0.0,
+      "step": 9392
+    },
+    {
+      "epoch": 0.6053360830057356,
+      "grad_norm": 0.3514795311405156,
+      "learning_rate": 4.390261367704977e-06,
+      "loss": 0.0034,
+      "step": 9393
+    },
+    {
+      "epoch": 0.6054005284526648,
+      "grad_norm": 0.06014872393965382,
+      "learning_rate": 4.389545291800932e-06,
+      "loss": 0.0,
+      "step": 9394
+    },
+    {
+      "epoch": 0.605464973899594,
+      "grad_norm": 0.0012982233820627276,
+      "learning_rate": 4.388829215896885e-06,
+      "loss": 0.0,
+      "step": 9395
+    },
+    {
+      "epoch": 0.6055294193465232,
+      "grad_norm": 0.0006104295092222888,
+      "learning_rate": 4.388113139992839e-06,
+      "loss": 0.0,
+      "step": 9396
+    },
+    {
+      "epoch": 0.6055938647934523,
+      "grad_norm": 0.0015226621546786093,
+      "learning_rate": 4.387397064088794e-06,
+      "loss": 0.0,
+      "step": 9397
+    },
+    {
+      "epoch": 0.6056583102403815,
+      "grad_norm": 0.004909857157835074,
+      "learning_rate": 4.386680988184748e-06,
+      "loss": 0.0,
+      "step": 9398
+    },
+    {
+      "epoch": 0.6057227556873107,
+      "grad_norm": 0.0016943522584871827,
+      "learning_rate": 4.385964912280702e-06,
+      "loss": 0.0,
+      "step": 9399
+    },
+    {
+      "epoch": 0.6057872011342399,
+      "grad_norm": 0.145636188277844,
+      "learning_rate": 4.3852488363766566e-06,
+      "loss": 0.0014,
+      "step": 9400
+    },
+    {
+      "epoch": 0.6058516465811691,
+      "grad_norm": 0.0037991052504417187,
+      "learning_rate": 4.384532760472611e-06,
+      "loss": 0.0,
+      "step": 9401
+    },
+    {
+      "epoch": 0.6059160920280983,
+      "grad_norm": 0.0005582835381395184,
+      "learning_rate": 4.383816684568565e-06,
+      "loss": 0.0,
+      "step": 9402
+    },
+    {
+      "epoch": 0.6059805374750273,
+      "grad_norm": 0.14714135028611341,
+      "learning_rate": 4.383100608664519e-06,
+      "loss": 0.004,
+      "step": 9403
+    },
+    {
+      "epoch": 0.6060449829219565,
+      "grad_norm": 0.00026380090583077884,
+      "learning_rate": 4.382384532760473e-06,
+      "loss": 0.0,
+      "step": 9404
+    },
+    {
+      "epoch": 0.6061094283688857,
+      "grad_norm": 0.007371777117651726,
+      "learning_rate": 4.381668456856427e-06,
+      "loss": 0.0,
+      "step": 9405
+    },
+    {
+      "epoch": 0.6061738738158149,
+      "grad_norm": 0.0010503819422914768,
+      "learning_rate": 4.3809523809523815e-06,
+      "loss": 0.0,
+      "step": 9406
+    },
+    {
+      "epoch": 0.6062383192627441,
+      "grad_norm": 0.06254363771771412,
+      "learning_rate": 4.380236305048336e-06,
+      "loss": 0.0001,
+      "step": 9407
+    },
+    {
+      "epoch": 0.6063027647096733,
+      "grad_norm": 0.0305562041932073,
+      "learning_rate": 4.379520229144289e-06,
+      "loss": 0.0001,
+      "step": 9408
+    },
+    {
+      "epoch": 0.6063672101566024,
+      "grad_norm": 0.00900166756321749,
+      "learning_rate": 4.3788041532402436e-06,
+      "loss": 0.0,
+      "step": 9409
+    },
+    {
+      "epoch": 0.6064316556035316,
+      "grad_norm": 0.30886882883342337,
+      "learning_rate": 4.378088077336198e-06,
+      "loss": 0.001,
+      "step": 9410
+    },
+    {
+      "epoch": 0.6064961010504608,
+      "grad_norm": 0.5044701548161394,
+      "learning_rate": 4.377372001432152e-06,
+      "loss": 0.0022,
+      "step": 9411
+    },
+    {
+      "epoch": 0.60656054649739,
+      "grad_norm": 0.0034124956418660653,
+      "learning_rate": 4.3766559255281065e-06,
+      "loss": 0.0,
+      "step": 9412
+    },
+    {
+      "epoch": 0.6066249919443192,
+      "grad_norm": 0.0024336553075388167,
+      "learning_rate": 4.375939849624061e-06,
+      "loss": 0.0,
+      "step": 9413
+    },
+    {
+      "epoch": 0.6066894373912483,
+      "grad_norm": 0.0006861778038973162,
+      "learning_rate": 4.375223773720015e-06,
+      "loss": 0.0,
+      "step": 9414
+    },
+    {
+      "epoch": 0.6067538828381774,
+      "grad_norm": 0.0007463060227097989,
+      "learning_rate": 4.374507697815969e-06,
+      "loss": 0.0,
+      "step": 9415
+    },
+    {
+      "epoch": 0.6068183282851066,
+      "grad_norm": 0.008138348403934246,
+      "learning_rate": 4.373791621911923e-06,
+      "loss": 0.0,
+      "step": 9416
+    },
+    {
+      "epoch": 0.6068827737320358,
+      "grad_norm": 0.0015037820156294577,
+      "learning_rate": 4.373075546007877e-06,
+      "loss": 0.0,
+      "step": 9417
+    },
+    {
+      "epoch": 0.606947219178965,
+      "grad_norm": 7.03840792272183e-05,
+      "learning_rate": 4.372359470103831e-06,
+      "loss": 0.0,
+      "step": 9418
+    },
+    {
+      "epoch": 0.6070116646258942,
+      "grad_norm": 0.724938618593745,
+      "learning_rate": 4.371643394199786e-06,
+      "loss": 0.006,
+      "step": 9419
+    },
+    {
+      "epoch": 0.6070761100728234,
+      "grad_norm": 0.00116128567512363,
+      "learning_rate": 4.370927318295739e-06,
+      "loss": 0.0,
+      "step": 9420
+    },
+    {
+      "epoch": 0.6071405555197525,
+      "grad_norm": 0.0850829120345377,
+      "learning_rate": 4.3702112423916935e-06,
+      "loss": 0.0002,
+      "step": 9421
+    },
+    {
+      "epoch": 0.6072050009666817,
+      "grad_norm": 0.0010536833509948063,
+      "learning_rate": 4.369495166487648e-06,
+      "loss": 0.0,
+      "step": 9422
+    },
+    {
+      "epoch": 0.6072694464136109,
+      "grad_norm": 0.00385163720268815,
+      "learning_rate": 4.368779090583602e-06,
+      "loss": 0.0,
+      "step": 9423
+    },
+    {
+      "epoch": 0.6073338918605401,
+      "grad_norm": 0.11442118970579021,
+      "learning_rate": 4.368063014679556e-06,
+      "loss": 0.0001,
+      "step": 9424
+    },
+    {
+      "epoch": 0.6073983373074693,
+      "grad_norm": 0.0007916495679545597,
+      "learning_rate": 4.367346938775511e-06,
+      "loss": 0.0,
+      "step": 9425
+    },
+    {
+      "epoch": 0.6074627827543984,
+      "grad_norm": 0.0011092425353548807,
+      "learning_rate": 4.366630862871465e-06,
+      "loss": 0.0,
+      "step": 9426
+    },
+    {
+      "epoch": 0.6075272282013275,
+      "grad_norm": 9.429646011118711e-05,
+      "learning_rate": 4.365914786967419e-06,
+      "loss": 0.0,
+      "step": 9427
+    },
+    {
+      "epoch": 0.6075916736482567,
+      "grad_norm": 0.002318340581239123,
+      "learning_rate": 4.365198711063373e-06,
+      "loss": 0.0,
+      "step": 9428
+    },
+    {
+      "epoch": 0.6076561190951859,
+      "grad_norm": 0.0031013551819890905,
+      "learning_rate": 4.364482635159327e-06,
+      "loss": 0.0,
+      "step": 9429
+    },
+    {
+      "epoch": 0.6077205645421151,
+      "grad_norm": 0.0011860604719637813,
+      "learning_rate": 4.363766559255281e-06,
+      "loss": 0.0,
+      "step": 9430
+    },
+    {
+      "epoch": 0.6077850099890443,
+      "grad_norm": 0.0868075929508591,
+      "learning_rate": 4.363050483351236e-06,
+      "loss": 0.0017,
+      "step": 9431
+    },
+    {
+      "epoch": 0.6078494554359735,
+      "grad_norm": 0.0016228980177938656,
+      "learning_rate": 4.36233440744719e-06,
+      "loss": 0.0,
+      "step": 9432
+    },
+    {
+      "epoch": 0.6079139008829026,
+      "grad_norm": 0.002918866658195965,
+      "learning_rate": 4.361618331543143e-06,
+      "loss": 0.0,
+      "step": 9433
+    },
+    {
+      "epoch": 0.6079783463298318,
+      "grad_norm": 0.40538073119100626,
+      "learning_rate": 4.360902255639098e-06,
+      "loss": 0.0006,
+      "step": 9434
+    },
+    {
+      "epoch": 0.608042791776761,
+      "grad_norm": 0.006536840513148788,
+      "learning_rate": 4.360186179735053e-06,
+      "loss": 0.0,
+      "step": 9435
+    },
+    {
+      "epoch": 0.6081072372236902,
+      "grad_norm": 0.02315314722042714,
+      "learning_rate": 4.359470103831006e-06,
+      "loss": 0.0001,
+      "step": 9436
+    },
+    {
+      "epoch": 0.6081716826706193,
+      "grad_norm": 0.0005369910501333005,
+      "learning_rate": 4.3587540279269605e-06,
+      "loss": 0.0,
+      "step": 9437
+    },
+    {
+      "epoch": 0.6082361281175485,
+      "grad_norm": 0.014507810777531585,
+      "learning_rate": 4.358037952022915e-06,
+      "loss": 0.0001,
+      "step": 9438
+    },
+    {
+      "epoch": 0.6083005735644776,
+      "grad_norm": 0.0014420113067096622,
+      "learning_rate": 4.357321876118869e-06,
+      "loss": 0.0,
+      "step": 9439
+    },
+    {
+      "epoch": 0.6083650190114068,
+      "grad_norm": 0.0003051545929258759,
+      "learning_rate": 4.3566058002148234e-06,
+      "loss": 0.0,
+      "step": 9440
+    },
+    {
+      "epoch": 0.608429464458336,
+      "grad_norm": 0.10313037041927729,
+      "learning_rate": 4.355889724310777e-06,
+      "loss": 0.0001,
+      "step": 9441
+    },
+    {
+      "epoch": 0.6084939099052652,
+      "grad_norm": 0.00023097012118933073,
+      "learning_rate": 4.355173648406731e-06,
+      "loss": 0.0,
+      "step": 9442
+    },
+    {
+      "epoch": 0.6085583553521944,
+      "grad_norm": 0.004176636239720323,
+      "learning_rate": 4.3544575725026855e-06,
+      "loss": 0.0,
+      "step": 9443
+    },
+    {
+      "epoch": 0.6086228007991236,
+      "grad_norm": 0.031410970579226795,
+      "learning_rate": 4.35374149659864e-06,
+      "loss": 0.0001,
+      "step": 9444
+    },
+    {
+      "epoch": 0.6086872462460527,
+      "grad_norm": 0.011897834685541422,
+      "learning_rate": 4.353025420694594e-06,
+      "loss": 0.0,
+      "step": 9445
+    },
+    {
+      "epoch": 0.6087516916929819,
+      "grad_norm": 0.00046606655128807525,
+      "learning_rate": 4.3523093447905475e-06,
+      "loss": 0.0,
+      "step": 9446
+    },
+    {
+      "epoch": 0.6088161371399111,
+      "grad_norm": 0.024800149341012708,
+      "learning_rate": 4.351593268886503e-06,
+      "loss": 0.0015,
+      "step": 9447
+    },
+    {
+      "epoch": 0.6088805825868402,
+      "grad_norm": 0.0221088822473336,
+      "learning_rate": 4.350877192982457e-06,
+      "loss": 0.0,
+      "step": 9448
+    },
+    {
+      "epoch": 0.6089450280337694,
+      "grad_norm": 6.0055351174664875e-05,
+      "learning_rate": 4.3501611170784104e-06,
+      "loss": 0.0,
+      "step": 9449
+    },
+    {
+      "epoch": 0.6090094734806986,
+      "grad_norm": 0.05522244514116825,
+      "learning_rate": 4.349445041174365e-06,
+      "loss": 0.0,
+      "step": 9450
+    },
+    {
+      "epoch": 0.6090739189276277,
+      "grad_norm": 0.0020661568106822036,
+      "learning_rate": 4.348728965270319e-06,
+      "loss": 0.0,
+      "step": 9451
+    },
+    {
+      "epoch": 0.6091383643745569,
+      "grad_norm": 0.00014003216050082992,
+      "learning_rate": 4.348012889366273e-06,
+      "loss": 0.0,
+      "step": 9452
+    },
+    {
+      "epoch": 0.6092028098214861,
+      "grad_norm": 5.3691884195080886e-05,
+      "learning_rate": 4.347296813462228e-06,
+      "loss": 0.0,
+      "step": 9453
+    },
+    {
+      "epoch": 0.6092672552684153,
+      "grad_norm": 2.2470153142569833e-05,
+      "learning_rate": 4.346580737558181e-06,
+      "loss": 0.0,
+      "step": 9454
+    },
+    {
+      "epoch": 0.6093317007153445,
+      "grad_norm": 0.005263171562567179,
+      "learning_rate": 4.345864661654135e-06,
+      "loss": 0.0,
+      "step": 9455
+    },
+    {
+      "epoch": 0.6093961461622737,
+      "grad_norm": 0.028391588065871397,
+      "learning_rate": 4.34514858575009e-06,
+      "loss": 0.0,
+      "step": 9456
+    },
+    {
+      "epoch": 0.6094605916092029,
+      "grad_norm": 0.0032687334314929254,
+      "learning_rate": 4.344432509846044e-06,
+      "loss": 0.0,
+      "step": 9457
+    },
+    {
+      "epoch": 0.609525037056132,
+      "grad_norm": 0.003059619076663178,
+      "learning_rate": 4.343716433941998e-06,
+      "loss": 0.0,
+      "step": 9458
+    },
+    {
+      "epoch": 0.6095894825030611,
+      "grad_norm": 0.00011531079670687498,
+      "learning_rate": 4.3430003580379526e-06,
+      "loss": 0.0,
+      "step": 9459
+    },
+    {
+      "epoch": 0.6096539279499903,
+      "grad_norm": 0.7228147590624995,
+      "learning_rate": 4.342284282133907e-06,
+      "loss": 0.0041,
+      "step": 9460
+    },
+    {
+      "epoch": 0.6097183733969195,
+      "grad_norm": 0.002931431348183814,
+      "learning_rate": 4.341568206229861e-06,
+      "loss": 0.0,
+      "step": 9461
+    },
+    {
+      "epoch": 0.6097828188438487,
+      "grad_norm": 2.102471073261476e-05,
+      "learning_rate": 4.340852130325815e-06,
+      "loss": 0.0,
+      "step": 9462
+    },
+    {
+      "epoch": 0.6098472642907778,
+      "grad_norm": 0.00016897046354369543,
+      "learning_rate": 4.340136054421769e-06,
+      "loss": 0.0,
+      "step": 9463
+    },
+    {
+      "epoch": 0.609911709737707,
+      "grad_norm": 0.0014111495206503312,
+      "learning_rate": 4.339419978517723e-06,
+      "loss": 0.0,
+      "step": 9464
+    },
+    {
+      "epoch": 0.6099761551846362,
+      "grad_norm": 0.00035216732375646924,
+      "learning_rate": 4.3387039026136775e-06,
+      "loss": 0.0,
+      "step": 9465
+    },
+    {
+      "epoch": 0.6100406006315654,
+      "grad_norm": 0.0001382897188749735,
+      "learning_rate": 4.337987826709632e-06,
+      "loss": 0.0,
+      "step": 9466
+    },
+    {
+      "epoch": 0.6101050460784946,
+      "grad_norm": 0.009698598664787292,
+      "learning_rate": 4.337271750805585e-06,
+      "loss": 0.0,
+      "step": 9467
+    },
+    {
+      "epoch": 0.6101694915254238,
+      "grad_norm": 0.06168647575651353,
+      "learning_rate": 4.3365556749015396e-06,
+      "loss": 0.0,
+      "step": 9468
+    },
+    {
+      "epoch": 0.610233936972353,
+      "grad_norm": 0.004649457971022612,
+      "learning_rate": 4.335839598997494e-06,
+      "loss": 0.0,
+      "step": 9469
+    },
+    {
+      "epoch": 0.610298382419282,
+      "grad_norm": 0.006819369683324906,
+      "learning_rate": 4.335123523093448e-06,
+      "loss": 0.0,
+      "step": 9470
+    },
+    {
+      "epoch": 0.6103628278662112,
+      "grad_norm": 0.0038879490160562474,
+      "learning_rate": 4.3344074471894025e-06,
+      "loss": 0.0,
+      "step": 9471
+    },
+    {
+      "epoch": 0.6104272733131404,
+      "grad_norm": 0.0001479780592880403,
+      "learning_rate": 4.333691371285357e-06,
+      "loss": 0.0,
+      "step": 9472
+    },
+    {
+      "epoch": 0.6104917187600696,
+      "grad_norm": 0.0005454773247210619,
+      "learning_rate": 4.332975295381311e-06,
+      "loss": 0.0,
+      "step": 9473
+    },
+    {
+      "epoch": 0.6105561642069988,
+      "grad_norm": 0.041149449013538716,
+      "learning_rate": 4.332259219477265e-06,
+      "loss": 0.0001,
+      "step": 9474
+    },
+    {
+      "epoch": 0.610620609653928,
+      "grad_norm": 0.0016966869856339295,
+      "learning_rate": 4.331543143573219e-06,
+      "loss": 0.0,
+      "step": 9475
+    },
+    {
+      "epoch": 0.6106850551008571,
+      "grad_norm": 0.10398651031290745,
+      "learning_rate": 4.330827067669173e-06,
+      "loss": 0.0004,
+      "step": 9476
+    },
+    {
+      "epoch": 0.6107495005477863,
+      "grad_norm": 0.0028450556875423104,
+      "learning_rate": 4.330110991765127e-06,
+      "loss": 0.0,
+      "step": 9477
+    },
+    {
+      "epoch": 0.6108139459947155,
+      "grad_norm": 0.0033837477527026193,
+      "learning_rate": 4.329394915861082e-06,
+      "loss": 0.0,
+      "step": 9478
+    },
+    {
+      "epoch": 0.6108783914416447,
+      "grad_norm": 0.006324049606286477,
+      "learning_rate": 4.328678839957036e-06,
+      "loss": 0.0,
+      "step": 9479
+    },
+    {
+      "epoch": 0.6109428368885739,
+      "grad_norm": 0.07765264767556082,
+      "learning_rate": 4.3279627640529895e-06,
+      "loss": 0.0001,
+      "step": 9480
+    },
+    {
+      "epoch": 0.6110072823355029,
+      "grad_norm": 0.01570204834010833,
+      "learning_rate": 4.327246688148944e-06,
+      "loss": 0.0001,
+      "step": 9481
+    },
+    {
+      "epoch": 0.6110717277824321,
+      "grad_norm": 0.0018498621459318526,
+      "learning_rate": 4.326530612244899e-06,
+      "loss": 0.0,
+      "step": 9482
+    },
+    {
+      "epoch": 0.6111361732293613,
+      "grad_norm": 0.037915579330810155,
+      "learning_rate": 4.325814536340852e-06,
+      "loss": 0.0,
+      "step": 9483
+    },
+    {
+      "epoch": 0.6112006186762905,
+      "grad_norm": 0.18641078003385522,
+      "learning_rate": 4.325098460436807e-06,
+      "loss": 0.0009,
+      "step": 9484
+    },
+    {
+      "epoch": 0.6112650641232197,
+      "grad_norm": 0.004106447632898952,
+      "learning_rate": 4.324382384532761e-06,
+      "loss": 0.0015,
+      "step": 9485
+    },
+    {
+      "epoch": 0.6113295095701489,
+      "grad_norm": 0.00024362199685292236,
+      "learning_rate": 4.323666308628715e-06,
+      "loss": 0.0,
+      "step": 9486
+    },
+    {
+      "epoch": 0.611393955017078,
+      "grad_norm": 0.3062199665445999,
+      "learning_rate": 4.3229502327246696e-06,
+      "loss": 0.0016,
+      "step": 9487
+    },
+    {
+      "epoch": 0.6114584004640072,
+      "grad_norm": 0.010514023663567535,
+      "learning_rate": 4.322234156820623e-06,
+      "loss": 0.0001,
+      "step": 9488
+    },
+    {
+      "epoch": 0.6115228459109364,
+      "grad_norm": 0.0024672399660492102,
+      "learning_rate": 4.321518080916577e-06,
+      "loss": 0.0,
+      "step": 9489
+    },
+    {
+      "epoch": 0.6115872913578656,
+      "grad_norm": 0.0035961915359396693,
+      "learning_rate": 4.320802005012532e-06,
+      "loss": 0.0,
+      "step": 9490
+    },
+    {
+      "epoch": 0.6116517368047948,
+      "grad_norm": 0.0008687014780787742,
+      "learning_rate": 4.320085929108486e-06,
+      "loss": 0.0,
+      "step": 9491
+    },
+    {
+      "epoch": 0.6117161822517239,
+      "grad_norm": 0.00013052206685491074,
+      "learning_rate": 4.31936985320444e-06,
+      "loss": 0.0,
+      "step": 9492
+    },
+    {
+      "epoch": 0.611780627698653,
+      "grad_norm": 0.0008957571000338346,
+      "learning_rate": 4.318653777300394e-06,
+      "loss": 0.0,
+      "step": 9493
+    },
+    {
+      "epoch": 0.6118450731455822,
+      "grad_norm": 0.0026848465137663054,
+      "learning_rate": 4.317937701396349e-06,
+      "loss": 0.0,
+      "step": 9494
+    },
+    {
+      "epoch": 0.6119095185925114,
+      "grad_norm": 0.32030492866685867,
+      "learning_rate": 4.317221625492303e-06,
+      "loss": 0.0006,
+      "step": 9495
+    },
+    {
+      "epoch": 0.6119739640394406,
+      "grad_norm": 0.06446050110427554,
+      "learning_rate": 4.3165055495882565e-06,
+      "loss": 0.0003,
+      "step": 9496
+    },
+    {
+      "epoch": 0.6120384094863698,
+      "grad_norm": 0.0011579000983677743,
+      "learning_rate": 4.315789473684211e-06,
+      "loss": 0.0,
+      "step": 9497
+    },
+    {
+      "epoch": 0.612102854933299,
+      "grad_norm": 0.009338821450512087,
+      "learning_rate": 4.315073397780165e-06,
+      "loss": 0.0,
+      "step": 9498
+    },
+    {
+      "epoch": 0.6121673003802282,
+      "grad_norm": 0.0014420975540785869,
+      "learning_rate": 4.3143573218761194e-06,
+      "loss": 0.0,
+      "step": 9499
+    },
+    {
+      "epoch": 0.6122317458271573,
+      "grad_norm": 2.6957402428930943,
+      "learning_rate": 4.313641245972074e-06,
+      "loss": 0.0235,
+      "step": 9500
+    },
+    {
+      "epoch": 0.6122961912740865,
+      "grad_norm": 0.042654696893015,
+      "learning_rate": 4.312925170068027e-06,
+      "loss": 0.0004,
+      "step": 9501
+    },
+    {
+      "epoch": 0.6123606367210157,
+      "grad_norm": 0.3342348687492541,
+      "learning_rate": 4.3122090941639815e-06,
+      "loss": 0.0006,
+      "step": 9502
+    },
+    {
+      "epoch": 0.6124250821679449,
+      "grad_norm": 0.01680525990444114,
+      "learning_rate": 4.311493018259936e-06,
+      "loss": 0.0,
+      "step": 9503
+    },
+    {
+      "epoch": 0.612489527614874,
+      "grad_norm": 0.0285541931701851,
+      "learning_rate": 4.31077694235589e-06,
+      "loss": 0.0001,
+      "step": 9504
+    },
+    {
+      "epoch": 0.6125539730618031,
+      "grad_norm": 0.0024596945855401516,
+      "learning_rate": 4.310060866451844e-06,
+      "loss": 0.0,
+      "step": 9505
+    },
+    {
+      "epoch": 0.6126184185087323,
+      "grad_norm": 0.16790719345938113,
+      "learning_rate": 4.309344790547799e-06,
+      "loss": 0.0014,
+      "step": 9506
+    },
+    {
+      "epoch": 0.6126828639556615,
+      "grad_norm": 0.0001499272645216901,
+      "learning_rate": 4.308628714643753e-06,
+      "loss": 0.0,
+      "step": 9507
+    },
+    {
+      "epoch": 0.6127473094025907,
+      "grad_norm": 0.000488662623177265,
+      "learning_rate": 4.307912638739707e-06,
+      "loss": 0.0,
+      "step": 9508
+    },
+    {
+      "epoch": 0.6128117548495199,
+      "grad_norm": 0.00029549820058340356,
+      "learning_rate": 4.307196562835661e-06,
+      "loss": 0.0,
+      "step": 9509
+    },
+    {
+      "epoch": 0.6128762002964491,
+      "grad_norm": 0.16494169692561012,
+      "learning_rate": 4.306480486931615e-06,
+      "loss": 0.0021,
+      "step": 9510
+    },
+    {
+      "epoch": 0.6129406457433783,
+      "grad_norm": 0.0022252980179280958,
+      "learning_rate": 4.305764411027569e-06,
+      "loss": 0.0,
+      "step": 9511
+    },
+    {
+      "epoch": 0.6130050911903074,
+      "grad_norm": 0.0031953051167575635,
+      "learning_rate": 4.305048335123524e-06,
+      "loss": 0.0,
+      "step": 9512
+    },
+    {
+      "epoch": 0.6130695366372366,
+      "grad_norm": 0.0798065214137776,
+      "learning_rate": 4.304332259219477e-06,
+      "loss": 0.0002,
+      "step": 9513
+    },
+    {
+      "epoch": 0.6131339820841658,
+      "grad_norm": 0.0032883889415415957,
+      "learning_rate": 4.303616183315431e-06,
+      "loss": 0.0,
+      "step": 9514
+    },
+    {
+      "epoch": 0.6131984275310949,
+      "grad_norm": 0.026226701485447677,
+      "learning_rate": 4.302900107411386e-06,
+      "loss": 0.0001,
+      "step": 9515
+    },
+    {
+      "epoch": 0.6132628729780241,
+      "grad_norm": 0.027247142216119388,
+      "learning_rate": 4.30218403150734e-06,
+      "loss": 0.0002,
+      "step": 9516
+    },
+    {
+      "epoch": 0.6133273184249532,
+      "grad_norm": 0.03939227437205754,
+      "learning_rate": 4.301467955603294e-06,
+      "loss": 0.0,
+      "step": 9517
+    },
+    {
+      "epoch": 0.6133917638718824,
+      "grad_norm": 0.00019537654302919405,
+      "learning_rate": 4.3007518796992486e-06,
+      "loss": 0.0,
+      "step": 9518
+    },
+    {
+      "epoch": 0.6134562093188116,
+      "grad_norm": 0.02171230850239767,
+      "learning_rate": 4.300035803795203e-06,
+      "loss": 0.0001,
+      "step": 9519
+    },
+    {
+      "epoch": 0.6135206547657408,
+      "grad_norm": 0.205370531067778,
+      "learning_rate": 4.299319727891157e-06,
+      "loss": 0.0008,
+      "step": 9520
+    },
+    {
+      "epoch": 0.61358510021267,
+      "grad_norm": 0.0006344435985339772,
+      "learning_rate": 4.2986036519871115e-06,
+      "loss": 0.0,
+      "step": 9521
+    },
+    {
+      "epoch": 0.6136495456595992,
+      "grad_norm": 0.031260613412393105,
+      "learning_rate": 4.297887576083065e-06,
+      "loss": 0.0001,
+      "step": 9522
+    },
+    {
+      "epoch": 0.6137139911065284,
+      "grad_norm": 0.048878088660678165,
+      "learning_rate": 4.297171500179019e-06,
+      "loss": 0.0001,
+      "step": 9523
+    },
+    {
+      "epoch": 0.6137784365534575,
+      "grad_norm": 0.5282191040348414,
+      "learning_rate": 4.2964554242749735e-06,
+      "loss": 0.0025,
+      "step": 9524
+    },
+    {
+      "epoch": 0.6138428820003867,
+      "grad_norm": 0.12579375317430874,
+      "learning_rate": 4.295739348370928e-06,
+      "loss": 0.0005,
+      "step": 9525
+    },
+    {
+      "epoch": 0.6139073274473158,
+      "grad_norm": 0.04641811663453107,
+      "learning_rate": 4.295023272466881e-06,
+      "loss": 0.0001,
+      "step": 9526
+    },
+    {
+      "epoch": 0.613971772894245,
+      "grad_norm": 0.0002580502539569169,
+      "learning_rate": 4.2943071965628356e-06,
+      "loss": 0.0,
+      "step": 9527
+    },
+    {
+      "epoch": 0.6140362183411742,
+      "grad_norm": 0.0033282476706218336,
+      "learning_rate": 4.29359112065879e-06,
+      "loss": 0.0,
+      "step": 9528
+    },
+    {
+      "epoch": 0.6141006637881034,
+      "grad_norm": 0.00047556915669403905,
+      "learning_rate": 4.292875044754745e-06,
+      "loss": 0.0,
+      "step": 9529
+    },
+    {
+      "epoch": 0.6141651092350325,
+      "grad_norm": 0.0013045353118663315,
+      "learning_rate": 4.2921589688506985e-06,
+      "loss": 0.0,
+      "step": 9530
+    },
+    {
+      "epoch": 0.6142295546819617,
+      "grad_norm": 1.2612551377705028,
+      "learning_rate": 4.291442892946653e-06,
+      "loss": 0.0021,
+      "step": 9531
+    },
+    {
+      "epoch": 0.6142940001288909,
+      "grad_norm": 0.00489277865661325,
+      "learning_rate": 4.290726817042607e-06,
+      "loss": 0.0,
+      "step": 9532
+    },
+    {
+      "epoch": 0.6143584455758201,
+      "grad_norm": 0.0012519150045497776,
+      "learning_rate": 4.290010741138561e-06,
+      "loss": 0.0,
+      "step": 9533
+    },
+    {
+      "epoch": 0.6144228910227493,
+      "grad_norm": 0.00015498001967760488,
+      "learning_rate": 4.289294665234515e-06,
+      "loss": 0.0,
+      "step": 9534
+    },
+    {
+      "epoch": 0.6144873364696785,
+      "grad_norm": 0.28547099784379487,
+      "learning_rate": 4.288578589330469e-06,
+      "loss": 0.0015,
+      "step": 9535
+    },
+    {
+      "epoch": 0.6145517819166076,
+      "grad_norm": 0.004607738427160797,
+      "learning_rate": 4.287862513426423e-06,
+      "loss": 0.0,
+      "step": 9536
+    },
+    {
+      "epoch": 0.6146162273635367,
+      "grad_norm": 0.03986350100567441,
+      "learning_rate": 4.287146437522378e-06,
+      "loss": 0.0,
+      "step": 9537
+    },
+    {
+      "epoch": 0.6146806728104659,
+      "grad_norm": 0.030827555896124095,
+      "learning_rate": 4.286430361618332e-06,
+      "loss": 0.0,
+      "step": 9538
+    },
+    {
+      "epoch": 0.6147451182573951,
+      "grad_norm": 0.00016944203631999406,
+      "learning_rate": 4.2857142857142855e-06,
+      "loss": 0.0,
+      "step": 9539
+    },
+    {
+      "epoch": 0.6148095637043243,
+      "grad_norm": 0.3045021862960025,
+      "learning_rate": 4.284998209810241e-06,
+      "loss": 0.0008,
+      "step": 9540
+    },
+    {
+      "epoch": 0.6148740091512535,
+      "grad_norm": 0.14254954456604813,
+      "learning_rate": 4.284282133906195e-06,
+      "loss": 0.0017,
+      "step": 9541
+    },
+    {
+      "epoch": 0.6149384545981826,
+      "grad_norm": 0.015514097077617816,
+      "learning_rate": 4.283566058002148e-06,
+      "loss": 0.0001,
+      "step": 9542
+    },
+    {
+      "epoch": 0.6150029000451118,
+      "grad_norm": 0.00337946789828538,
+      "learning_rate": 4.282849982098103e-06,
+      "loss": 0.0,
+      "step": 9543
+    },
+    {
+      "epoch": 0.615067345492041,
+      "grad_norm": 0.10997992786917365,
+      "learning_rate": 4.282133906194057e-06,
+      "loss": 0.0001,
+      "step": 9544
+    },
+    {
+      "epoch": 0.6151317909389702,
+      "grad_norm": 0.08706040598996456,
+      "learning_rate": 4.281417830290011e-06,
+      "loss": 0.0002,
+      "step": 9545
+    },
+    {
+      "epoch": 0.6151962363858994,
+      "grad_norm": 0.0017644996784956883,
+      "learning_rate": 4.2807017543859656e-06,
+      "loss": 0.0,
+      "step": 9546
+    },
+    {
+      "epoch": 0.6152606818328286,
+      "grad_norm": 0.06708993009425664,
+      "learning_rate": 4.279985678481919e-06,
+      "loss": 0.0001,
+      "step": 9547
+    },
+    {
+      "epoch": 0.6153251272797576,
+      "grad_norm": 0.24657193153772985,
+      "learning_rate": 4.279269602577873e-06,
+      "loss": 0.0002,
+      "step": 9548
+    },
+    {
+      "epoch": 0.6153895727266868,
+      "grad_norm": 0.0006839220314001171,
+      "learning_rate": 4.278553526673828e-06,
+      "loss": 0.0,
+      "step": 9549
+    },
+    {
+      "epoch": 0.615454018173616,
+      "grad_norm": 0.028784731907356236,
+      "learning_rate": 4.277837450769782e-06,
+      "loss": 0.0002,
+      "step": 9550
+    },
+    {
+      "epoch": 0.6155184636205452,
+      "grad_norm": 0.0002460372595801576,
+      "learning_rate": 4.277121374865736e-06,
+      "loss": 0.0,
+      "step": 9551
+    },
+    {
+      "epoch": 0.6155829090674744,
+      "grad_norm": 0.08505335527793859,
+      "learning_rate": 4.2764052989616905e-06,
+      "loss": 0.0001,
+      "step": 9552
+    },
+    {
+      "epoch": 0.6156473545144036,
+      "grad_norm": 0.0007359800036842759,
+      "learning_rate": 4.275689223057645e-06,
+      "loss": 0.0,
+      "step": 9553
+    },
+    {
+      "epoch": 0.6157117999613327,
+      "grad_norm": 0.0001408191174706121,
+      "learning_rate": 4.274973147153599e-06,
+      "loss": 0.0,
+      "step": 9554
+    },
+    {
+      "epoch": 0.6157762454082619,
+      "grad_norm": 0.0023287981858707055,
+      "learning_rate": 4.2742570712495525e-06,
+      "loss": 0.0,
+      "step": 9555
+    },
+    {
+      "epoch": 0.6158406908551911,
+      "grad_norm": 0.0019438432122889477,
+      "learning_rate": 4.273540995345507e-06,
+      "loss": 0.0,
+      "step": 9556
+    },
+    {
+      "epoch": 0.6159051363021203,
+      "grad_norm": 1.152815331124407,
+      "learning_rate": 4.272824919441461e-06,
+      "loss": 0.0041,
+      "step": 9557
+    },
+    {
+      "epoch": 0.6159695817490495,
+      "grad_norm": 0.009054800502372921,
+      "learning_rate": 4.2721088435374154e-06,
+      "loss": 0.0001,
+      "step": 9558
+    },
+    {
+      "epoch": 0.6160340271959786,
+      "grad_norm": 0.41673807088647546,
+      "learning_rate": 4.27139276763337e-06,
+      "loss": 0.0025,
+      "step": 9559
+    },
+    {
+      "epoch": 0.6160984726429077,
+      "grad_norm": 0.001291795997738671,
+      "learning_rate": 4.270676691729323e-06,
+      "loss": 0.0,
+      "step": 9560
+    },
+    {
+      "epoch": 0.6161629180898369,
+      "grad_norm": 0.00037450827288190545,
+      "learning_rate": 4.2699606158252775e-06,
+      "loss": 0.0,
+      "step": 9561
+    },
+    {
+      "epoch": 0.6162273635367661,
+      "grad_norm": 0.0005643279883164137,
+      "learning_rate": 4.269244539921232e-06,
+      "loss": 0.0,
+      "step": 9562
+    },
+    {
+      "epoch": 0.6162918089836953,
+      "grad_norm": 0.00013711708141211,
+      "learning_rate": 4.268528464017186e-06,
+      "loss": 0.0,
+      "step": 9563
+    },
+    {
+      "epoch": 0.6163562544306245,
+      "grad_norm": 0.00627109371547579,
+      "learning_rate": 4.26781238811314e-06,
+      "loss": 0.0015,
+      "step": 9564
+    },
+    {
+      "epoch": 0.6164206998775537,
+      "grad_norm": 0.025730302441083985,
+      "learning_rate": 4.267096312209095e-06,
+      "loss": 0.0001,
+      "step": 9565
+    },
+    {
+      "epoch": 0.6164851453244828,
+      "grad_norm": 0.0020360262097681384,
+      "learning_rate": 4.266380236305049e-06,
+      "loss": 0.0,
+      "step": 9566
+    },
+    {
+      "epoch": 0.616549590771412,
+      "grad_norm": 0.00888623677442725,
+      "learning_rate": 4.265664160401003e-06,
+      "loss": 0.0,
+      "step": 9567
+    },
+    {
+      "epoch": 0.6166140362183412,
+      "grad_norm": 0.00021110422840300865,
+      "learning_rate": 4.264948084496957e-06,
+      "loss": 0.0,
+      "step": 9568
+    },
+    {
+      "epoch": 0.6166784816652704,
+      "grad_norm": 0.5416255501278395,
+      "learning_rate": 4.264232008592911e-06,
+      "loss": 0.0025,
+      "step": 9569
+    },
+    {
+      "epoch": 0.6167429271121995,
+      "grad_norm": 0.19467164760361685,
+      "learning_rate": 4.263515932688865e-06,
+      "loss": 0.0003,
+      "step": 9570
+    },
+    {
+      "epoch": 0.6168073725591287,
+      "grad_norm": 0.13399001852000092,
+      "learning_rate": 4.26279985678482e-06,
+      "loss": 0.0003,
+      "step": 9571
+    },
+    {
+      "epoch": 0.6168718180060578,
+      "grad_norm": 0.00011533534958016236,
+      "learning_rate": 4.262083780880774e-06,
+      "loss": 0.0,
+      "step": 9572
+    },
+    {
+      "epoch": 0.616936263452987,
+      "grad_norm": 0.08406597431618373,
+      "learning_rate": 4.261367704976727e-06,
+      "loss": 0.0001,
+      "step": 9573
+    },
+    {
+      "epoch": 0.6170007088999162,
+      "grad_norm": 0.006471258951393888,
+      "learning_rate": 4.260651629072682e-06,
+      "loss": 0.0,
+      "step": 9574
+    },
+    {
+      "epoch": 0.6170651543468454,
+      "grad_norm": 0.11885665445735855,
+      "learning_rate": 4.259935553168637e-06,
+      "loss": 0.0016,
+      "step": 9575
+    },
+    {
+      "epoch": 0.6171295997937746,
+      "grad_norm": 0.005192039990253033,
+      "learning_rate": 4.25921947726459e-06,
+      "loss": 0.0001,
+      "step": 9576
+    },
+    {
+      "epoch": 0.6171940452407038,
+      "grad_norm": 0.10412234963006879,
+      "learning_rate": 4.2585034013605446e-06,
+      "loss": 0.0001,
+      "step": 9577
+    },
+    {
+      "epoch": 0.6172584906876329,
+      "grad_norm": 0.03084306802890673,
+      "learning_rate": 4.257787325456499e-06,
+      "loss": 0.0,
+      "step": 9578
+    },
+    {
+      "epoch": 0.6173229361345621,
+      "grad_norm": 0.1886445616070666,
+      "learning_rate": 4.257071249552453e-06,
+      "loss": 0.0003,
+      "step": 9579
+    },
+    {
+      "epoch": 0.6173873815814913,
+      "grad_norm": 0.007571403887036818,
+      "learning_rate": 4.2563551736484075e-06,
+      "loss": 0.0,
+      "step": 9580
+    },
+    {
+      "epoch": 0.6174518270284205,
+      "grad_norm": 0.11695276831761708,
+      "learning_rate": 4.255639097744361e-06,
+      "loss": 0.0015,
+      "step": 9581
+    },
+    {
+      "epoch": 0.6175162724753496,
+      "grad_norm": 0.03454928162524213,
+      "learning_rate": 4.254923021840315e-06,
+      "loss": 0.0001,
+      "step": 9582
+    },
+    {
+      "epoch": 0.6175807179222788,
+      "grad_norm": 0.08363969623079226,
+      "learning_rate": 4.2542069459362695e-06,
+      "loss": 0.0001,
+      "step": 9583
+    },
+    {
+      "epoch": 0.6176451633692079,
+      "grad_norm": 0.0032724227689246324,
+      "learning_rate": 4.253490870032224e-06,
+      "loss": 0.0,
+      "step": 9584
+    },
+    {
+      "epoch": 0.6177096088161371,
+      "grad_norm": 0.16838093020000738,
+      "learning_rate": 4.252774794128178e-06,
+      "loss": 0.002,
+      "step": 9585
+    },
+    {
+      "epoch": 0.6177740542630663,
+      "grad_norm": 0.015356828445240035,
+      "learning_rate": 4.2520587182241316e-06,
+      "loss": 0.0001,
+      "step": 9586
+    },
+    {
+      "epoch": 0.6178384997099955,
+      "grad_norm": 0.24950890436843448,
+      "learning_rate": 4.251342642320087e-06,
+      "loss": 0.0009,
+      "step": 9587
+    },
+    {
+      "epoch": 0.6179029451569247,
+      "grad_norm": 0.0004905781353135952,
+      "learning_rate": 4.250626566416041e-06,
+      "loss": 0.0,
+      "step": 9588
+    },
+    {
+      "epoch": 0.6179673906038539,
+      "grad_norm": 0.0017723076927522122,
+      "learning_rate": 4.2499104905119945e-06,
+      "loss": 0.0,
+      "step": 9589
+    },
+    {
+      "epoch": 0.618031836050783,
+      "grad_norm": 0.00027595294939807507,
+      "learning_rate": 4.249194414607949e-06,
+      "loss": 0.0,
+      "step": 9590
+    },
+    {
+      "epoch": 0.6180962814977122,
+      "grad_norm": 0.44881769604763494,
+      "learning_rate": 4.248478338703903e-06,
+      "loss": 0.0044,
+      "step": 9591
+    },
+    {
+      "epoch": 0.6181607269446414,
+      "grad_norm": 0.0011654496177690313,
+      "learning_rate": 4.247762262799857e-06,
+      "loss": 0.0,
+      "step": 9592
+    },
+    {
+      "epoch": 0.6182251723915705,
+      "grad_norm": 0.0002969354686474064,
+      "learning_rate": 4.247046186895812e-06,
+      "loss": 0.0,
+      "step": 9593
+    },
+    {
+      "epoch": 0.6182896178384997,
+      "grad_norm": 0.06758952912812587,
+      "learning_rate": 4.246330110991765e-06,
+      "loss": 0.0003,
+      "step": 9594
+    },
+    {
+      "epoch": 0.6183540632854289,
+      "grad_norm": 0.0023606559277491745,
+      "learning_rate": 4.245614035087719e-06,
+      "loss": 0.0,
+      "step": 9595
+    },
+    {
+      "epoch": 0.618418508732358,
+      "grad_norm": 0.0005240889689227749,
+      "learning_rate": 4.244897959183674e-06,
+      "loss": 0.0,
+      "step": 9596
+    },
+    {
+      "epoch": 0.6184829541792872,
+      "grad_norm": 0.0028731344088614675,
+      "learning_rate": 4.244181883279628e-06,
+      "loss": 0.0,
+      "step": 9597
+    },
+    {
+      "epoch": 0.6185473996262164,
+      "grad_norm": 0.003774219453522588,
+      "learning_rate": 4.243465807375582e-06,
+      "loss": 0.0,
+      "step": 9598
+    },
+    {
+      "epoch": 0.6186118450731456,
+      "grad_norm": 0.04812190750716624,
+      "learning_rate": 4.242749731471537e-06,
+      "loss": 0.0002,
+      "step": 9599
+    },
+    {
+      "epoch": 0.6186762905200748,
+      "grad_norm": 0.0009303746721064348,
+      "learning_rate": 4.242033655567491e-06,
+      "loss": 0.0,
+      "step": 9600
+    },
+    {
+      "epoch": 0.618740735967004,
+      "grad_norm": 0.000261596983827811,
+      "learning_rate": 4.241317579663445e-06,
+      "loss": 0.0,
+      "step": 9601
+    },
+    {
+      "epoch": 0.6188051814139331,
+      "grad_norm": 4.060820420659366e-05,
+      "learning_rate": 4.240601503759399e-06,
+      "loss": 0.0,
+      "step": 9602
+    },
+    {
+      "epoch": 0.6188696268608623,
+      "grad_norm": 0.05925821210830824,
+      "learning_rate": 4.239885427855353e-06,
+      "loss": 0.0016,
+      "step": 9603
+    },
+    {
+      "epoch": 0.6189340723077914,
+      "grad_norm": 0.0003562257420606521,
+      "learning_rate": 4.239169351951307e-06,
+      "loss": 0.0,
+      "step": 9604
+    },
+    {
+      "epoch": 0.6189985177547206,
+      "grad_norm": 0.031397710705029135,
+      "learning_rate": 4.2384532760472616e-06,
+      "loss": 0.0,
+      "step": 9605
+    },
+    {
+      "epoch": 0.6190629632016498,
+      "grad_norm": 2.4042094265960856e-05,
+      "learning_rate": 4.237737200143216e-06,
+      "loss": 0.0,
+      "step": 9606
+    },
+    {
+      "epoch": 0.619127408648579,
+      "grad_norm": 0.021245099475145683,
+      "learning_rate": 4.237021124239169e-06,
+      "loss": 0.0002,
+      "step": 9607
+    },
+    {
+      "epoch": 0.6191918540955081,
+      "grad_norm": 0.0018329740066415307,
+      "learning_rate": 4.236305048335124e-06,
+      "loss": 0.0,
+      "step": 9608
+    },
+    {
+      "epoch": 0.6192562995424373,
+      "grad_norm": 0.003174532843030457,
+      "learning_rate": 4.235588972431078e-06,
+      "loss": 0.0,
+      "step": 9609
+    },
+    {
+      "epoch": 0.6193207449893665,
+      "grad_norm": 0.12445039922168191,
+      "learning_rate": 4.234872896527032e-06,
+      "loss": 0.0004,
+      "step": 9610
+    },
+    {
+      "epoch": 0.6193851904362957,
+      "grad_norm": 0.6708809129849929,
+      "learning_rate": 4.2341568206229865e-06,
+      "loss": 0.0025,
+      "step": 9611
+    },
+    {
+      "epoch": 0.6194496358832249,
+      "grad_norm": 0.06722900209427916,
+      "learning_rate": 4.233440744718941e-06,
+      "loss": 0.0002,
+      "step": 9612
+    },
+    {
+      "epoch": 0.6195140813301541,
+      "grad_norm": 0.02487789501266384,
+      "learning_rate": 4.232724668814895e-06,
+      "loss": 0.0001,
+      "step": 9613
+    },
+    {
+      "epoch": 0.6195785267770832,
+      "grad_norm": 0.0070891124843938435,
+      "learning_rate": 4.232008592910849e-06,
+      "loss": 0.0,
+      "step": 9614
+    },
+    {
+      "epoch": 0.6196429722240123,
+      "grad_norm": 0.03626208442549633,
+      "learning_rate": 4.231292517006803e-06,
+      "loss": 0.0001,
+      "step": 9615
+    },
+    {
+      "epoch": 0.6197074176709415,
+      "grad_norm": 0.13685853487861604,
+      "learning_rate": 4.230576441102757e-06,
+      "loss": 0.0004,
+      "step": 9616
+    },
+    {
+      "epoch": 0.6197718631178707,
+      "grad_norm": 0.02757626052057049,
+      "learning_rate": 4.2298603651987114e-06,
+      "loss": 0.0001,
+      "step": 9617
+    },
+    {
+      "epoch": 0.6198363085647999,
+      "grad_norm": 0.0014798321427789366,
+      "learning_rate": 4.229144289294666e-06,
+      "loss": 0.0,
+      "step": 9618
+    },
+    {
+      "epoch": 0.6199007540117291,
+      "grad_norm": 0.01049815395489131,
+      "learning_rate": 4.228428213390619e-06,
+      "loss": 0.0001,
+      "step": 9619
+    },
+    {
+      "epoch": 0.6199651994586582,
+      "grad_norm": 0.6844524113320779,
+      "learning_rate": 4.2277121374865735e-06,
+      "loss": 0.0044,
+      "step": 9620
+    },
+    {
+      "epoch": 0.6200296449055874,
+      "grad_norm": 0.0002147181528283652,
+      "learning_rate": 4.226996061582528e-06,
+      "loss": 0.0,
+      "step": 9621
+    },
+    {
+      "epoch": 0.6200940903525166,
+      "grad_norm": 0.2500937256360808,
+      "learning_rate": 4.226279985678483e-06,
+      "loss": 0.001,
+      "step": 9622
+    },
+    {
+      "epoch": 0.6201585357994458,
+      "grad_norm": 0.04441692228177357,
+      "learning_rate": 4.225563909774436e-06,
+      "loss": 0.0001,
+      "step": 9623
+    },
+    {
+      "epoch": 0.620222981246375,
+      "grad_norm": 0.0028260496702314057,
+      "learning_rate": 4.224847833870391e-06,
+      "loss": 0.0,
+      "step": 9624
+    },
+    {
+      "epoch": 0.6202874266933042,
+      "grad_norm": 0.027488300872075832,
+      "learning_rate": 4.224131757966345e-06,
+      "loss": 0.0,
+      "step": 9625
+    },
+    {
+      "epoch": 0.6203518721402332,
+      "grad_norm": 7.346583784405198e-05,
+      "learning_rate": 4.223415682062299e-06,
+      "loss": 0.0,
+      "step": 9626
+    },
+    {
+      "epoch": 0.6204163175871624,
+      "grad_norm": 0.06306652635555225,
+      "learning_rate": 4.222699606158253e-06,
+      "loss": 0.0002,
+      "step": 9627
+    },
+    {
+      "epoch": 0.6204807630340916,
+      "grad_norm": 0.00592616535198368,
+      "learning_rate": 4.221983530254207e-06,
+      "loss": 0.0001,
+      "step": 9628
+    },
+    {
+      "epoch": 0.6205452084810208,
+      "grad_norm": 0.0001044583771192941,
+      "learning_rate": 4.221267454350161e-06,
+      "loss": 0.0,
+      "step": 9629
+    },
+    {
+      "epoch": 0.62060965392795,
+      "grad_norm": 5.9252248572389913e-05,
+      "learning_rate": 4.220551378446116e-06,
+      "loss": 0.0,
+      "step": 9630
+    },
+    {
+      "epoch": 0.6206740993748792,
+      "grad_norm": 0.04576163754311494,
+      "learning_rate": 4.21983530254207e-06,
+      "loss": 0.0017,
+      "step": 9631
+    },
+    {
+      "epoch": 0.6207385448218083,
+      "grad_norm": 0.011120442526384637,
+      "learning_rate": 4.219119226638023e-06,
+      "loss": 0.0001,
+      "step": 9632
+    },
+    {
+      "epoch": 0.6208029902687375,
+      "grad_norm": 0.0012373695941060873,
+      "learning_rate": 4.218403150733978e-06,
+      "loss": 0.0,
+      "step": 9633
+    },
+    {
+      "epoch": 0.6208674357156667,
+      "grad_norm": 0.0033271694642447464,
+      "learning_rate": 4.217687074829933e-06,
+      "loss": 0.0,
+      "step": 9634
+    },
+    {
+      "epoch": 0.6209318811625959,
+      "grad_norm": 0.0023266009334657824,
+      "learning_rate": 4.216970998925886e-06,
+      "loss": 0.0,
+      "step": 9635
+    },
+    {
+      "epoch": 0.6209963266095251,
+      "grad_norm": 0.0006776249456023286,
+      "learning_rate": 4.2162549230218406e-06,
+      "loss": 0.0,
+      "step": 9636
+    },
+    {
+      "epoch": 0.6210607720564542,
+      "grad_norm": 0.01654521411218407,
+      "learning_rate": 4.215538847117795e-06,
+      "loss": 0.0,
+      "step": 9637
+    },
+    {
+      "epoch": 0.6211252175033833,
+      "grad_norm": 4.852683199395689e-05,
+      "learning_rate": 4.214822771213749e-06,
+      "loss": 0.0,
+      "step": 9638
+    },
+    {
+      "epoch": 0.6211896629503125,
+      "grad_norm": 0.0021993879120563247,
+      "learning_rate": 4.2141066953097035e-06,
+      "loss": 0.0,
+      "step": 9639
+    },
+    {
+      "epoch": 0.6212541083972417,
+      "grad_norm": 0.0007637886447388776,
+      "learning_rate": 4.213390619405657e-06,
+      "loss": 0.0,
+      "step": 9640
+    },
+    {
+      "epoch": 0.6213185538441709,
+      "grad_norm": 0.007377042865803844,
+      "learning_rate": 4.212674543501611e-06,
+      "loss": 0.0,
+      "step": 9641
+    },
+    {
+      "epoch": 0.6213829992911001,
+      "grad_norm": 0.0005533541648899976,
+      "learning_rate": 4.2119584675975655e-06,
+      "loss": 0.0,
+      "step": 9642
+    },
+    {
+      "epoch": 0.6214474447380293,
+      "grad_norm": 0.42125994387586024,
+      "learning_rate": 4.21124239169352e-06,
+      "loss": 0.0025,
+      "step": 9643
+    },
+    {
+      "epoch": 0.6215118901849584,
+      "grad_norm": 0.03175686695537614,
+      "learning_rate": 4.210526315789474e-06,
+      "loss": 0.0001,
+      "step": 9644
+    },
+    {
+      "epoch": 0.6215763356318876,
+      "grad_norm": 0.299370056123468,
+      "learning_rate": 4.209810239885428e-06,
+      "loss": 0.0024,
+      "step": 9645
+    },
+    {
+      "epoch": 0.6216407810788168,
+      "grad_norm": 0.26325604179437306,
+      "learning_rate": 4.209094163981383e-06,
+      "loss": 0.0007,
+      "step": 9646
+    },
+    {
+      "epoch": 0.621705226525746,
+      "grad_norm": 0.01473776976325703,
+      "learning_rate": 4.208378088077337e-06,
+      "loss": 0.0,
+      "step": 9647
+    },
+    {
+      "epoch": 0.6217696719726751,
+      "grad_norm": 0.0005109300137223879,
+      "learning_rate": 4.2076620121732905e-06,
+      "loss": 0.0,
+      "step": 9648
+    },
+    {
+      "epoch": 0.6218341174196043,
+      "grad_norm": 0.0002493537642654104,
+      "learning_rate": 4.206945936269245e-06,
+      "loss": 0.0,
+      "step": 9649
+    },
+    {
+      "epoch": 0.6218985628665334,
+      "grad_norm": 0.07706493977662457,
+      "learning_rate": 4.206229860365199e-06,
+      "loss": 0.0001,
+      "step": 9650
+    },
+    {
+      "epoch": 0.6219630083134626,
+      "grad_norm": 0.0003456695922869574,
+      "learning_rate": 4.205513784461153e-06,
+      "loss": 0.0,
+      "step": 9651
+    },
+    {
+      "epoch": 0.6220274537603918,
+      "grad_norm": 0.0018471179024505506,
+      "learning_rate": 4.204797708557108e-06,
+      "loss": 0.0,
+      "step": 9652
+    },
+    {
+      "epoch": 0.622091899207321,
+      "grad_norm": 0.4382099794044541,
+      "learning_rate": 4.204081632653061e-06,
+      "loss": 0.0014,
+      "step": 9653
+    },
+    {
+      "epoch": 0.6221563446542502,
+      "grad_norm": 0.00014206190967321758,
+      "learning_rate": 4.203365556749015e-06,
+      "loss": 0.0,
+      "step": 9654
+    },
+    {
+      "epoch": 0.6222207901011794,
+      "grad_norm": 0.023678985918828478,
+      "learning_rate": 4.20264948084497e-06,
+      "loss": 0.0,
+      "step": 9655
+    },
+    {
+      "epoch": 0.6222852355481086,
+      "grad_norm": 0.0015190579278471688,
+      "learning_rate": 4.201933404940924e-06,
+      "loss": 0.0,
+      "step": 9656
+    },
+    {
+      "epoch": 0.6223496809950377,
+      "grad_norm": 2.9540092315116344,
+      "learning_rate": 4.201217329036878e-06,
+      "loss": 0.0244,
+      "step": 9657
+    },
+    {
+      "epoch": 0.6224141264419669,
+      "grad_norm": 0.00012206465167372003,
+      "learning_rate": 4.200501253132833e-06,
+      "loss": 0.0,
+      "step": 9658
+    },
+    {
+      "epoch": 0.6224785718888961,
+      "grad_norm": 0.0002953703826734249,
+      "learning_rate": 4.199785177228787e-06,
+      "loss": 0.0,
+      "step": 9659
+    },
+    {
+      "epoch": 0.6225430173358252,
+      "grad_norm": 0.0006869575541465469,
+      "learning_rate": 4.199069101324741e-06,
+      "loss": 0.0,
+      "step": 9660
+    },
+    {
+      "epoch": 0.6226074627827544,
+      "grad_norm": 0.005004019924537658,
+      "learning_rate": 4.198353025420695e-06,
+      "loss": 0.0,
+      "step": 9661
+    },
+    {
+      "epoch": 0.6226719082296835,
+      "grad_norm": 0.0036624435182955875,
+      "learning_rate": 4.197636949516649e-06,
+      "loss": 0.0,
+      "step": 9662
+    },
+    {
+      "epoch": 0.6227363536766127,
+      "grad_norm": 0.00772927658587039,
+      "learning_rate": 4.196920873612603e-06,
+      "loss": 0.0001,
+      "step": 9663
+    },
+    {
+      "epoch": 0.6228007991235419,
+      "grad_norm": 0.007641090487948875,
+      "learning_rate": 4.1962047977085576e-06,
+      "loss": 0.0,
+      "step": 9664
+    },
+    {
+      "epoch": 0.6228652445704711,
+      "grad_norm": 0.12874149139306848,
+      "learning_rate": 4.195488721804512e-06,
+      "loss": 0.0005,
+      "step": 9665
+    },
+    {
+      "epoch": 0.6229296900174003,
+      "grad_norm": 0.043186854746607224,
+      "learning_rate": 4.194772645900465e-06,
+      "loss": 0.0001,
+      "step": 9666
+    },
+    {
+      "epoch": 0.6229941354643295,
+      "grad_norm": 0.0034495539036033776,
+      "learning_rate": 4.19405656999642e-06,
+      "loss": 0.0,
+      "step": 9667
+    },
+    {
+      "epoch": 0.6230585809112587,
+      "grad_norm": 0.0015735462216765262,
+      "learning_rate": 4.193340494092374e-06,
+      "loss": 0.0,
+      "step": 9668
+    },
+    {
+      "epoch": 0.6231230263581878,
+      "grad_norm": 0.0028003798244138406,
+      "learning_rate": 4.192624418188328e-06,
+      "loss": 0.0,
+      "step": 9669
+    },
+    {
+      "epoch": 0.623187471805117,
+      "grad_norm": 0.0021314886590723505,
+      "learning_rate": 4.1919083422842825e-06,
+      "loss": 0.0,
+      "step": 9670
+    },
+    {
+      "epoch": 0.6232519172520461,
+      "grad_norm": 0.020338036201677363,
+      "learning_rate": 4.191192266380237e-06,
+      "loss": 0.0,
+      "step": 9671
+    },
+    {
+      "epoch": 0.6233163626989753,
+      "grad_norm": 0.15354739217184882,
+      "learning_rate": 4.190476190476191e-06,
+      "loss": 0.0016,
+      "step": 9672
+    },
+    {
+      "epoch": 0.6233808081459045,
+      "grad_norm": 0.015586767135734498,
+      "learning_rate": 4.189760114572145e-06,
+      "loss": 0.0,
+      "step": 9673
+    },
+    {
+      "epoch": 0.6234452535928336,
+      "grad_norm": 0.016917160072103598,
+      "learning_rate": 4.189044038668099e-06,
+      "loss": 0.0,
+      "step": 9674
+    },
+    {
+      "epoch": 0.6235096990397628,
+      "grad_norm": 0.04137864002051029,
+      "learning_rate": 4.188327962764053e-06,
+      "loss": 0.0,
+      "step": 9675
+    },
+    {
+      "epoch": 0.623574144486692,
+      "grad_norm": 0.005969930532603671,
+      "learning_rate": 4.1876118868600074e-06,
+      "loss": 0.0,
+      "step": 9676
+    },
+    {
+      "epoch": 0.6236385899336212,
+      "grad_norm": 0.01609130556531648,
+      "learning_rate": 4.186895810955962e-06,
+      "loss": 0.0,
+      "step": 9677
+    },
+    {
+      "epoch": 0.6237030353805504,
+      "grad_norm": 0.0025735075923348596,
+      "learning_rate": 4.186179735051916e-06,
+      "loss": 0.0,
+      "step": 9678
+    },
+    {
+      "epoch": 0.6237674808274796,
+      "grad_norm": 0.0016625419018875228,
+      "learning_rate": 4.1854636591478695e-06,
+      "loss": 0.0,
+      "step": 9679
+    },
+    {
+      "epoch": 0.6238319262744088,
+      "grad_norm": 0.0003346156449337851,
+      "learning_rate": 4.184747583243825e-06,
+      "loss": 0.0,
+      "step": 9680
+    },
+    {
+      "epoch": 0.6238963717213379,
+      "grad_norm": 0.007099106515354779,
+      "learning_rate": 4.184031507339779e-06,
+      "loss": 0.0,
+      "step": 9681
+    },
+    {
+      "epoch": 0.623960817168267,
+      "grad_norm": 0.003438129501815967,
+      "learning_rate": 4.183315431435732e-06,
+      "loss": 0.0,
+      "step": 9682
+    },
+    {
+      "epoch": 0.6240252626151962,
+      "grad_norm": 0.03816232143606552,
+      "learning_rate": 4.182599355531687e-06,
+      "loss": 0.0005,
+      "step": 9683
+    },
+    {
+      "epoch": 0.6240897080621254,
+      "grad_norm": 0.07756988779862854,
+      "learning_rate": 4.181883279627641e-06,
+      "loss": 0.0004,
+      "step": 9684
+    },
+    {
+      "epoch": 0.6241541535090546,
+      "grad_norm": 0.4092447233751244,
+      "learning_rate": 4.181167203723595e-06,
+      "loss": 0.001,
+      "step": 9685
+    },
+    {
+      "epoch": 0.6242185989559838,
+      "grad_norm": 0.0007207571886176135,
+      "learning_rate": 4.18045112781955e-06,
+      "loss": 0.0,
+      "step": 9686
+    },
+    {
+      "epoch": 0.6242830444029129,
+      "grad_norm": 0.0015483125744365732,
+      "learning_rate": 4.179735051915503e-06,
+      "loss": 0.0,
+      "step": 9687
+    },
+    {
+      "epoch": 0.6243474898498421,
+      "grad_norm": 0.0021180640674535764,
+      "learning_rate": 4.179018976011457e-06,
+      "loss": 0.0,
+      "step": 9688
+    },
+    {
+      "epoch": 0.6244119352967713,
+      "grad_norm": 0.0004169306514595993,
+      "learning_rate": 4.178302900107412e-06,
+      "loss": 0.0,
+      "step": 9689
+    },
+    {
+      "epoch": 0.6244763807437005,
+      "grad_norm": 0.2161596745728091,
+      "learning_rate": 4.177586824203366e-06,
+      "loss": 0.0003,
+      "step": 9690
+    },
+    {
+      "epoch": 0.6245408261906297,
+      "grad_norm": 0.0011705409204292648,
+      "learning_rate": 4.17687074829932e-06,
+      "loss": 0.0,
+      "step": 9691
+    },
+    {
+      "epoch": 0.6246052716375589,
+      "grad_norm": 0.0003713461406357066,
+      "learning_rate": 4.1761546723952745e-06,
+      "loss": 0.0,
+      "step": 9692
+    },
+    {
+      "epoch": 0.6246697170844879,
+      "grad_norm": 0.021227762380770367,
+      "learning_rate": 4.175438596491229e-06,
+      "loss": 0.0,
+      "step": 9693
+    },
+    {
+      "epoch": 0.6247341625314171,
+      "grad_norm": 0.002796829726804933,
+      "learning_rate": 4.174722520587183e-06,
+      "loss": 0.0,
+      "step": 9694
+    },
+    {
+      "epoch": 0.6247986079783463,
+      "grad_norm": 0.029885802088271313,
+      "learning_rate": 4.1740064446831366e-06,
+      "loss": 0.0,
+      "step": 9695
+    },
+    {
+      "epoch": 0.6248630534252755,
+      "grad_norm": 0.058992337485208504,
+      "learning_rate": 4.173290368779091e-06,
+      "loss": 0.0001,
+      "step": 9696
+    },
+    {
+      "epoch": 0.6249274988722047,
+      "grad_norm": 0.006196414162415305,
+      "learning_rate": 4.172574292875045e-06,
+      "loss": 0.0,
+      "step": 9697
+    },
+    {
+      "epoch": 0.6249919443191339,
+      "grad_norm": 0.0003641721267636872,
+      "learning_rate": 4.1718582169709995e-06,
+      "loss": 0.0,
+      "step": 9698
+    },
+    {
+      "epoch": 0.625056389766063,
+      "grad_norm": 0.000895571559838155,
+      "learning_rate": 4.171142141066954e-06,
+      "loss": 0.0,
+      "step": 9699
+    },
+    {
+      "epoch": 0.6251208352129922,
+      "grad_norm": 0.035813595296406875,
+      "learning_rate": 4.170426065162907e-06,
+      "loss": 0.0004,
+      "step": 9700
+    },
+    {
+      "epoch": 0.6251852806599214,
+      "grad_norm": 0.014866533526694621,
+      "learning_rate": 4.1697099892588615e-06,
+      "loss": 0.0,
+      "step": 9701
+    },
+    {
+      "epoch": 0.6252497261068506,
+      "grad_norm": 0.0004485504683336761,
+      "learning_rate": 4.168993913354816e-06,
+      "loss": 0.0,
+      "step": 9702
+    },
+    {
+      "epoch": 0.6253141715537798,
+      "grad_norm": 0.003314286639645609,
+      "learning_rate": 4.16827783745077e-06,
+      "loss": 0.0,
+      "step": 9703
+    },
+    {
+      "epoch": 0.6253786170007088,
+      "grad_norm": 0.0038918480210237234,
+      "learning_rate": 4.167561761546724e-06,
+      "loss": 0.0,
+      "step": 9704
+    },
+    {
+      "epoch": 0.625443062447638,
+      "grad_norm": 0.12356112889405628,
+      "learning_rate": 4.166845685642679e-06,
+      "loss": 0.0005,
+      "step": 9705
+    },
+    {
+      "epoch": 0.6255075078945672,
+      "grad_norm": 0.010094385634850529,
+      "learning_rate": 4.166129609738633e-06,
+      "loss": 0.0,
+      "step": 9706
+    },
+    {
+      "epoch": 0.6255719533414964,
+      "grad_norm": 0.0004073141393491479,
+      "learning_rate": 4.165413533834587e-06,
+      "loss": 0.0,
+      "step": 9707
+    },
+    {
+      "epoch": 0.6256363987884256,
+      "grad_norm": 0.017301657748644873,
+      "learning_rate": 4.164697457930541e-06,
+      "loss": 0.0001,
+      "step": 9708
+    },
+    {
+      "epoch": 0.6257008442353548,
+      "grad_norm": 0.008330549915594329,
+      "learning_rate": 4.163981382026495e-06,
+      "loss": 0.0,
+      "step": 9709
+    },
+    {
+      "epoch": 0.625765289682284,
+      "grad_norm": 0.0012220208405838166,
+      "learning_rate": 4.163265306122449e-06,
+      "loss": 0.0,
+      "step": 9710
+    },
+    {
+      "epoch": 0.6258297351292131,
+      "grad_norm": 0.0024680865602183513,
+      "learning_rate": 4.162549230218404e-06,
+      "loss": 0.0,
+      "step": 9711
+    },
+    {
+      "epoch": 0.6258941805761423,
+      "grad_norm": 0.0005113877056839675,
+      "learning_rate": 4.161833154314357e-06,
+      "loss": 0.0,
+      "step": 9712
+    },
+    {
+      "epoch": 0.6259586260230715,
+      "grad_norm": 0.22660793473691426,
+      "learning_rate": 4.161117078410311e-06,
+      "loss": 0.0002,
+      "step": 9713
+    },
+    {
+      "epoch": 0.6260230714700007,
+      "grad_norm": 0.0678046505039315,
+      "learning_rate": 4.160401002506266e-06,
+      "loss": 0.0,
+      "step": 9714
+    },
+    {
+      "epoch": 0.6260875169169298,
+      "grad_norm": 0.0009742323449192132,
+      "learning_rate": 4.159684926602221e-06,
+      "loss": 0.0,
+      "step": 9715
+    },
+    {
+      "epoch": 0.626151962363859,
+      "grad_norm": 0.006071902899983566,
+      "learning_rate": 4.158968850698174e-06,
+      "loss": 0.0,
+      "step": 9716
+    },
+    {
+      "epoch": 0.6262164078107881,
+      "grad_norm": 0.2007009679693991,
+      "learning_rate": 4.158252774794129e-06,
+      "loss": 0.0006,
+      "step": 9717
+    },
+    {
+      "epoch": 0.6262808532577173,
+      "grad_norm": 0.6364653615568687,
+      "learning_rate": 4.157536698890083e-06,
+      "loss": 0.0149,
+      "step": 9718
+    },
+    {
+      "epoch": 0.6263452987046465,
+      "grad_norm": 0.002847480991120004,
+      "learning_rate": 4.156820622986037e-06,
+      "loss": 0.0,
+      "step": 9719
+    },
+    {
+      "epoch": 0.6264097441515757,
+      "grad_norm": 0.0029042838819600516,
+      "learning_rate": 4.156104547081991e-06,
+      "loss": 0.0,
+      "step": 9720
+    },
+    {
+      "epoch": 0.6264741895985049,
+      "grad_norm": 0.0010571381934525565,
+      "learning_rate": 4.155388471177945e-06,
+      "loss": 0.0,
+      "step": 9721
+    },
+    {
+      "epoch": 0.6265386350454341,
+      "grad_norm": 0.0006312867694469445,
+      "learning_rate": 4.154672395273899e-06,
+      "loss": 0.0,
+      "step": 9722
+    },
+    {
+      "epoch": 0.6266030804923632,
+      "grad_norm": 0.501257536919454,
+      "learning_rate": 4.1539563193698536e-06,
+      "loss": 0.0035,
+      "step": 9723
+    },
+    {
+      "epoch": 0.6266675259392924,
+      "grad_norm": 0.0003598110398831839,
+      "learning_rate": 4.153240243465808e-06,
+      "loss": 0.0,
+      "step": 9724
+    },
+    {
+      "epoch": 0.6267319713862216,
+      "grad_norm": 0.4419211195223192,
+      "learning_rate": 4.152524167561761e-06,
+      "loss": 0.0013,
+      "step": 9725
+    },
+    {
+      "epoch": 0.6267964168331507,
+      "grad_norm": 0.005140032610661535,
+      "learning_rate": 4.151808091657716e-06,
+      "loss": 0.0,
+      "step": 9726
+    },
+    {
+      "epoch": 0.6268608622800799,
+      "grad_norm": 0.006610240234879396,
+      "learning_rate": 4.151092015753671e-06,
+      "loss": 0.0,
+      "step": 9727
+    },
+    {
+      "epoch": 0.626925307727009,
+      "grad_norm": 0.012762952650590107,
+      "learning_rate": 4.150375939849624e-06,
+      "loss": 0.0,
+      "step": 9728
+    },
+    {
+      "epoch": 0.6269897531739382,
+      "grad_norm": 0.010514889004076607,
+      "learning_rate": 4.1496598639455785e-06,
+      "loss": 0.0001,
+      "step": 9729
+    },
+    {
+      "epoch": 0.6270541986208674,
+      "grad_norm": 3.305536977108447,
+      "learning_rate": 4.148943788041533e-06,
+      "loss": 0.0243,
+      "step": 9730
+    },
+    {
+      "epoch": 0.6271186440677966,
+      "grad_norm": 0.009644390248062271,
+      "learning_rate": 4.148227712137487e-06,
+      "loss": 0.0001,
+      "step": 9731
+    },
+    {
+      "epoch": 0.6271830895147258,
+      "grad_norm": 0.004172765545080662,
+      "learning_rate": 4.147511636233441e-06,
+      "loss": 0.0,
+      "step": 9732
+    },
+    {
+      "epoch": 0.627247534961655,
+      "grad_norm": 0.6687569011795615,
+      "learning_rate": 4.146795560329395e-06,
+      "loss": 0.0072,
+      "step": 9733
+    },
+    {
+      "epoch": 0.6273119804085842,
+      "grad_norm": 0.005077958161566104,
+      "learning_rate": 4.146079484425349e-06,
+      "loss": 0.0,
+      "step": 9734
+    },
+    {
+      "epoch": 0.6273764258555133,
+      "grad_norm": 0.0011856766644601346,
+      "learning_rate": 4.1453634085213034e-06,
+      "loss": 0.0,
+      "step": 9735
+    },
+    {
+      "epoch": 0.6274408713024425,
+      "grad_norm": 0.0011318798438648293,
+      "learning_rate": 4.144647332617258e-06,
+      "loss": 0.0,
+      "step": 9736
+    },
+    {
+      "epoch": 0.6275053167493717,
+      "grad_norm": 0.0022622025217353646,
+      "learning_rate": 4.143931256713212e-06,
+      "loss": 0.0,
+      "step": 9737
+    },
+    {
+      "epoch": 0.6275697621963008,
+      "grad_norm": 0.0031208555963091454,
+      "learning_rate": 4.1432151808091655e-06,
+      "loss": 0.0,
+      "step": 9738
+    },
+    {
+      "epoch": 0.62763420764323,
+      "grad_norm": 0.11720895173505456,
+      "learning_rate": 4.142499104905121e-06,
+      "loss": 0.0001,
+      "step": 9739
+    },
+    {
+      "epoch": 0.6276986530901592,
+      "grad_norm": 0.005784035279351796,
+      "learning_rate": 4.141783029001075e-06,
+      "loss": 0.0,
+      "step": 9740
+    },
+    {
+      "epoch": 0.6277630985370883,
+      "grad_norm": 0.001448754885450368,
+      "learning_rate": 4.141066953097028e-06,
+      "loss": 0.0,
+      "step": 9741
+    },
+    {
+      "epoch": 0.6278275439840175,
+      "grad_norm": 0.0007223212627194857,
+      "learning_rate": 4.140350877192983e-06,
+      "loss": 0.0,
+      "step": 9742
+    },
+    {
+      "epoch": 0.6278919894309467,
+      "grad_norm": 0.14653846377902094,
+      "learning_rate": 4.139634801288937e-06,
+      "loss": 0.0009,
+      "step": 9743
+    },
+    {
+      "epoch": 0.6279564348778759,
+      "grad_norm": 0.06317755627566499,
+      "learning_rate": 4.138918725384891e-06,
+      "loss": 0.0,
+      "step": 9744
+    },
+    {
+      "epoch": 0.6280208803248051,
+      "grad_norm": 0.3560164288228587,
+      "learning_rate": 4.138202649480846e-06,
+      "loss": 0.0018,
+      "step": 9745
+    },
+    {
+      "epoch": 0.6280853257717343,
+      "grad_norm": 0.005363883013006638,
+      "learning_rate": 4.137486573576799e-06,
+      "loss": 0.0,
+      "step": 9746
+    },
+    {
+      "epoch": 0.6281497712186634,
+      "grad_norm": 0.033138692973302124,
+      "learning_rate": 4.136770497672753e-06,
+      "loss": 0.0,
+      "step": 9747
+    },
+    {
+      "epoch": 0.6282142166655926,
+      "grad_norm": 0.28489794855303097,
+      "learning_rate": 4.136054421768708e-06,
+      "loss": 0.0004,
+      "step": 9748
+    },
+    {
+      "epoch": 0.6282786621125217,
+      "grad_norm": 0.003932125381898977,
+      "learning_rate": 4.135338345864662e-06,
+      "loss": 0.0,
+      "step": 9749
+    },
+    {
+      "epoch": 0.6283431075594509,
+      "grad_norm": 0.019557865455868052,
+      "learning_rate": 4.134622269960616e-06,
+      "loss": 0.0001,
+      "step": 9750
+    },
+    {
+      "epoch": 0.6284075530063801,
+      "grad_norm": 0.000930857796089239,
+      "learning_rate": 4.1339061940565705e-06,
+      "loss": 0.0,
+      "step": 9751
+    },
+    {
+      "epoch": 0.6284719984533093,
+      "grad_norm": 0.004989099766665328,
+      "learning_rate": 4.133190118152525e-06,
+      "loss": 0.0,
+      "step": 9752
+    },
+    {
+      "epoch": 0.6285364439002384,
+      "grad_norm": 0.0035396782047819827,
+      "learning_rate": 4.132474042248479e-06,
+      "loss": 0.0,
+      "step": 9753
+    },
+    {
+      "epoch": 0.6286008893471676,
+      "grad_norm": 0.05278230405212935,
+      "learning_rate": 4.1317579663444326e-06,
+      "loss": 0.0001,
+      "step": 9754
+    },
+    {
+      "epoch": 0.6286653347940968,
+      "grad_norm": 0.003403609391087778,
+      "learning_rate": 4.131041890440387e-06,
+      "loss": 0.0,
+      "step": 9755
+    },
+    {
+      "epoch": 0.628729780241026,
+      "grad_norm": 0.07996074866573993,
+      "learning_rate": 4.130325814536341e-06,
+      "loss": 0.0002,
+      "step": 9756
+    },
+    {
+      "epoch": 0.6287942256879552,
+      "grad_norm": 0.2723902128574586,
+      "learning_rate": 4.1296097386322955e-06,
+      "loss": 0.0012,
+      "step": 9757
+    },
+    {
+      "epoch": 0.6288586711348844,
+      "grad_norm": 0.0025250918803746652,
+      "learning_rate": 4.12889366272825e-06,
+      "loss": 0.0,
+      "step": 9758
+    },
+    {
+      "epoch": 0.6289231165818135,
+      "grad_norm": 0.0360516493613523,
+      "learning_rate": 4.128177586824203e-06,
+      "loss": 0.0001,
+      "step": 9759
+    },
+    {
+      "epoch": 0.6289875620287426,
+      "grad_norm": 0.051373972553429056,
+      "learning_rate": 4.1274615109201575e-06,
+      "loss": 0.0002,
+      "step": 9760
+    },
+    {
+      "epoch": 0.6290520074756718,
+      "grad_norm": 0.007127486987450691,
+      "learning_rate": 4.126745435016112e-06,
+      "loss": 0.0,
+      "step": 9761
+    },
+    {
+      "epoch": 0.629116452922601,
+      "grad_norm": 0.04414521157648746,
+      "learning_rate": 4.126029359112066e-06,
+      "loss": 0.0001,
+      "step": 9762
+    },
+    {
+      "epoch": 0.6291808983695302,
+      "grad_norm": 0.015933055726976987,
+      "learning_rate": 4.12531328320802e-06,
+      "loss": 0.0,
+      "step": 9763
+    },
+    {
+      "epoch": 0.6292453438164594,
+      "grad_norm": 0.0004764234417019966,
+      "learning_rate": 4.124597207303975e-06,
+      "loss": 0.0,
+      "step": 9764
+    },
+    {
+      "epoch": 0.6293097892633885,
+      "grad_norm": 0.005602980997375922,
+      "learning_rate": 4.123881131399929e-06,
+      "loss": 0.0,
+      "step": 9765
+    },
+    {
+      "epoch": 0.6293742347103177,
+      "grad_norm": 0.0022225477643587023,
+      "learning_rate": 4.123165055495883e-06,
+      "loss": 0.0,
+      "step": 9766
+    },
+    {
+      "epoch": 0.6294386801572469,
+      "grad_norm": 0.3022448735374261,
+      "learning_rate": 4.122448979591837e-06,
+      "loss": 0.0012,
+      "step": 9767
+    },
+    {
+      "epoch": 0.6295031256041761,
+      "grad_norm": 0.02914737365516086,
+      "learning_rate": 4.121732903687791e-06,
+      "loss": 0.0001,
+      "step": 9768
+    },
+    {
+      "epoch": 0.6295675710511053,
+      "grad_norm": 0.05615636769679371,
+      "learning_rate": 4.121016827783745e-06,
+      "loss": 0.0001,
+      "step": 9769
+    },
+    {
+      "epoch": 0.6296320164980345,
+      "grad_norm": 0.0003076192540867642,
+      "learning_rate": 4.1203007518797e-06,
+      "loss": 0.0,
+      "step": 9770
+    },
+    {
+      "epoch": 0.6296964619449635,
+      "grad_norm": 0.06890702224480363,
+      "learning_rate": 4.119584675975654e-06,
+      "loss": 0.0003,
+      "step": 9771
+    },
+    {
+      "epoch": 0.6297609073918927,
+      "grad_norm": 0.006440353235390373,
+      "learning_rate": 4.118868600071607e-06,
+      "loss": 0.0,
+      "step": 9772
+    },
+    {
+      "epoch": 0.6298253528388219,
+      "grad_norm": 0.001049338266341266,
+      "learning_rate": 4.118152524167562e-06,
+      "loss": 0.0,
+      "step": 9773
+    },
+    {
+      "epoch": 0.6298897982857511,
+      "grad_norm": 0.09006338639124273,
+      "learning_rate": 4.117436448263517e-06,
+      "loss": 0.0017,
+      "step": 9774
+    },
+    {
+      "epoch": 0.6299542437326803,
+      "grad_norm": 0.32334044423878544,
+      "learning_rate": 4.11672037235947e-06,
+      "loss": 0.0022,
+      "step": 9775
+    },
+    {
+      "epoch": 0.6300186891796095,
+      "grad_norm": 0.0014797895047647244,
+      "learning_rate": 4.116004296455425e-06,
+      "loss": 0.0,
+      "step": 9776
+    },
+    {
+      "epoch": 0.6300831346265386,
+      "grad_norm": 0.002647025070808818,
+      "learning_rate": 4.115288220551379e-06,
+      "loss": 0.0,
+      "step": 9777
+    },
+    {
+      "epoch": 0.6301475800734678,
+      "grad_norm": 0.0002881831873300413,
+      "learning_rate": 4.114572144647333e-06,
+      "loss": 0.0,
+      "step": 9778
+    },
+    {
+      "epoch": 0.630212025520397,
+      "grad_norm": 0.0025381103557760654,
+      "learning_rate": 4.1138560687432875e-06,
+      "loss": 0.0,
+      "step": 9779
+    },
+    {
+      "epoch": 0.6302764709673262,
+      "grad_norm": 0.0006100557027432721,
+      "learning_rate": 4.113139992839241e-06,
+      "loss": 0.0,
+      "step": 9780
+    },
+    {
+      "epoch": 0.6303409164142554,
+      "grad_norm": 0.00111819279469302,
+      "learning_rate": 4.112423916935195e-06,
+      "loss": 0.0,
+      "step": 9781
+    },
+    {
+      "epoch": 0.6304053618611845,
+      "grad_norm": 0.009024577275247043,
+      "learning_rate": 4.1117078410311496e-06,
+      "loss": 0.0,
+      "step": 9782
+    },
+    {
+      "epoch": 0.6304698073081136,
+      "grad_norm": 0.040998943660520266,
+      "learning_rate": 4.110991765127104e-06,
+      "loss": 0.0005,
+      "step": 9783
+    },
+    {
+      "epoch": 0.6305342527550428,
+      "grad_norm": 0.2744994794192702,
+      "learning_rate": 4.110275689223058e-06,
+      "loss": 0.0017,
+      "step": 9784
+    },
+    {
+      "epoch": 0.630598698201972,
+      "grad_norm": 0.47197642973060894,
+      "learning_rate": 4.109559613319012e-06,
+      "loss": 0.0008,
+      "step": 9785
+    },
+    {
+      "epoch": 0.6306631436489012,
+      "grad_norm": 0.011569277487227457,
+      "learning_rate": 4.108843537414967e-06,
+      "loss": 0.0001,
+      "step": 9786
+    },
+    {
+      "epoch": 0.6307275890958304,
+      "grad_norm": 0.00016951053258177544,
+      "learning_rate": 4.108127461510921e-06,
+      "loss": 0.0,
+      "step": 9787
+    },
+    {
+      "epoch": 0.6307920345427596,
+      "grad_norm": 0.0002565186560096678,
+      "learning_rate": 4.1074113856068745e-06,
+      "loss": 0.0,
+      "step": 9788
+    },
+    {
+      "epoch": 0.6308564799896887,
+      "grad_norm": 0.015760302528762305,
+      "learning_rate": 4.106695309702829e-06,
+      "loss": 0.0001,
+      "step": 9789
+    },
+    {
+      "epoch": 0.6309209254366179,
+      "grad_norm": 0.0005307053243126115,
+      "learning_rate": 4.105979233798783e-06,
+      "loss": 0.0,
+      "step": 9790
+    },
+    {
+      "epoch": 0.6309853708835471,
+      "grad_norm": 0.009203236432359213,
+      "learning_rate": 4.105263157894737e-06,
+      "loss": 0.0,
+      "step": 9791
+    },
+    {
+      "epoch": 0.6310498163304763,
+      "grad_norm": 0.00010736600066678872,
+      "learning_rate": 4.104547081990692e-06,
+      "loss": 0.0,
+      "step": 9792
+    },
+    {
+      "epoch": 0.6311142617774054,
+      "grad_norm": 0.01727893981573304,
+      "learning_rate": 4.103831006086645e-06,
+      "loss": 0.0,
+      "step": 9793
+    },
+    {
+      "epoch": 0.6311787072243346,
+      "grad_norm": 0.005858238365772789,
+      "learning_rate": 4.1031149301825994e-06,
+      "loss": 0.0,
+      "step": 9794
+    },
+    {
+      "epoch": 0.6312431526712637,
+      "grad_norm": 0.03891247460398242,
+      "learning_rate": 4.102398854278554e-06,
+      "loss": 0.0001,
+      "step": 9795
+    },
+    {
+      "epoch": 0.6313075981181929,
+      "grad_norm": 0.03522325296595622,
+      "learning_rate": 4.101682778374508e-06,
+      "loss": 0.0001,
+      "step": 9796
+    },
+    {
+      "epoch": 0.6313720435651221,
+      "grad_norm": 0.0002808154914054595,
+      "learning_rate": 4.100966702470462e-06,
+      "loss": 0.0,
+      "step": 9797
+    },
+    {
+      "epoch": 0.6314364890120513,
+      "grad_norm": 0.0385147856123727,
+      "learning_rate": 4.100250626566417e-06,
+      "loss": 0.0004,
+      "step": 9798
+    },
+    {
+      "epoch": 0.6315009344589805,
+      "grad_norm": 0.005532733142487714,
+      "learning_rate": 4.099534550662371e-06,
+      "loss": 0.0,
+      "step": 9799
+    },
+    {
+      "epoch": 0.6315653799059097,
+      "grad_norm": 0.034152268583626674,
+      "learning_rate": 4.098818474758325e-06,
+      "loss": 0.0001,
+      "step": 9800
+    },
+    {
+      "epoch": 0.6316298253528388,
+      "grad_norm": 0.39788263618123626,
+      "learning_rate": 4.098102398854279e-06,
+      "loss": 0.0004,
+      "step": 9801
+    },
+    {
+      "epoch": 0.631694270799768,
+      "grad_norm": 0.008583379104093024,
+      "learning_rate": 4.097386322950233e-06,
+      "loss": 0.0,
+      "step": 9802
+    },
+    {
+      "epoch": 0.6317587162466972,
+      "grad_norm": 0.0025010910330443806,
+      "learning_rate": 4.096670247046187e-06,
+      "loss": 0.0,
+      "step": 9803
+    },
+    {
+      "epoch": 0.6318231616936264,
+      "grad_norm": 0.01292406597771796,
+      "learning_rate": 4.095954171142142e-06,
+      "loss": 0.0,
+      "step": 9804
+    },
+    {
+      "epoch": 0.6318876071405555,
+      "grad_norm": 0.011911608905516258,
+      "learning_rate": 4.095238095238096e-06,
+      "loss": 0.0,
+      "step": 9805
+    },
+    {
+      "epoch": 0.6319520525874847,
+      "grad_norm": 0.021996962163873125,
+      "learning_rate": 4.094522019334049e-06,
+      "loss": 0.0,
+      "step": 9806
+    },
+    {
+      "epoch": 0.6320164980344138,
+      "grad_norm": 0.0006577608885751697,
+      "learning_rate": 4.093805943430004e-06,
+      "loss": 0.0,
+      "step": 9807
+    },
+    {
+      "epoch": 0.632080943481343,
+      "grad_norm": 0.0022851470470816276,
+      "learning_rate": 4.093089867525958e-06,
+      "loss": 0.0,
+      "step": 9808
+    },
+    {
+      "epoch": 0.6321453889282722,
+      "grad_norm": 0.00044632028055951176,
+      "learning_rate": 4.092373791621912e-06,
+      "loss": 0.0,
+      "step": 9809
+    },
+    {
+      "epoch": 0.6322098343752014,
+      "grad_norm": 0.0005976803320070037,
+      "learning_rate": 4.0916577157178665e-06,
+      "loss": 0.0,
+      "step": 9810
+    },
+    {
+      "epoch": 0.6322742798221306,
+      "grad_norm": 0.0005299045073304586,
+      "learning_rate": 4.090941639813821e-06,
+      "loss": 0.0,
+      "step": 9811
+    },
+    {
+      "epoch": 0.6323387252690598,
+      "grad_norm": 0.00022852497761108778,
+      "learning_rate": 4.090225563909775e-06,
+      "loss": 0.0,
+      "step": 9812
+    },
+    {
+      "epoch": 0.632403170715989,
+      "grad_norm": 0.21186302379079988,
+      "learning_rate": 4.089509488005729e-06,
+      "loss": 0.0031,
+      "step": 9813
+    },
+    {
+      "epoch": 0.6324676161629181,
+      "grad_norm": 0.12596821572501107,
+      "learning_rate": 4.088793412101683e-06,
+      "loss": 0.0002,
+      "step": 9814
+    },
+    {
+      "epoch": 0.6325320616098473,
+      "grad_norm": 0.0013975578937536427,
+      "learning_rate": 4.088077336197637e-06,
+      "loss": 0.0,
+      "step": 9815
+    },
+    {
+      "epoch": 0.6325965070567764,
+      "grad_norm": 0.005904553442262322,
+      "learning_rate": 4.0873612602935915e-06,
+      "loss": 0.0,
+      "step": 9816
+    },
+    {
+      "epoch": 0.6326609525037056,
+      "grad_norm": 0.016000071829722146,
+      "learning_rate": 4.086645184389546e-06,
+      "loss": 0.0002,
+      "step": 9817
+    },
+    {
+      "epoch": 0.6327253979506348,
+      "grad_norm": 0.0028822450328563408,
+      "learning_rate": 4.085929108485499e-06,
+      "loss": 0.0,
+      "step": 9818
+    },
+    {
+      "epoch": 0.632789843397564,
+      "grad_norm": 0.05139031940953903,
+      "learning_rate": 4.0852130325814535e-06,
+      "loss": 0.0002,
+      "step": 9819
+    },
+    {
+      "epoch": 0.6328542888444931,
+      "grad_norm": 9.16404113373634e-05,
+      "learning_rate": 4.084496956677408e-06,
+      "loss": 0.0,
+      "step": 9820
+    },
+    {
+      "epoch": 0.6329187342914223,
+      "grad_norm": 0.0016909506227782403,
+      "learning_rate": 4.083780880773363e-06,
+      "loss": 0.0,
+      "step": 9821
+    },
+    {
+      "epoch": 0.6329831797383515,
+      "grad_norm": 0.025381268785659834,
+      "learning_rate": 4.083064804869316e-06,
+      "loss": 0.0,
+      "step": 9822
+    },
+    {
+      "epoch": 0.6330476251852807,
+      "grad_norm": 0.004422708829880086,
+      "learning_rate": 4.082348728965271e-06,
+      "loss": 0.0001,
+      "step": 9823
+    },
+    {
+      "epoch": 0.6331120706322099,
+      "grad_norm": 0.002109393715706444,
+      "learning_rate": 4.081632653061225e-06,
+      "loss": 0.0,
+      "step": 9824
+    },
+    {
+      "epoch": 0.633176516079139,
+      "grad_norm": 0.002699339732813906,
+      "learning_rate": 4.080916577157179e-06,
+      "loss": 0.0,
+      "step": 9825
+    },
+    {
+      "epoch": 0.6332409615260682,
+      "grad_norm": 0.0017968199788469807,
+      "learning_rate": 4.080200501253133e-06,
+      "loss": 0.0,
+      "step": 9826
+    },
+    {
+      "epoch": 0.6333054069729973,
+      "grad_norm": 0.004023536004873301,
+      "learning_rate": 4.079484425349087e-06,
+      "loss": 0.0,
+      "step": 9827
+    },
+    {
+      "epoch": 0.6333698524199265,
+      "grad_norm": 0.0003069514670206831,
+      "learning_rate": 4.078768349445041e-06,
+      "loss": 0.0,
+      "step": 9828
+    },
+    {
+      "epoch": 0.6334342978668557,
+      "grad_norm": 0.0028215493809561425,
+      "learning_rate": 4.078052273540996e-06,
+      "loss": 0.0,
+      "step": 9829
+    },
+    {
+      "epoch": 0.6334987433137849,
+      "grad_norm": 0.11994944466776976,
+      "learning_rate": 4.07733619763695e-06,
+      "loss": 0.0003,
+      "step": 9830
+    },
+    {
+      "epoch": 0.633563188760714,
+      "grad_norm": 0.10761587225219324,
+      "learning_rate": 4.076620121732903e-06,
+      "loss": 0.0003,
+      "step": 9831
+    },
+    {
+      "epoch": 0.6336276342076432,
+      "grad_norm": 0.011345525980090358,
+      "learning_rate": 4.0759040458288586e-06,
+      "loss": 0.0,
+      "step": 9832
+    },
+    {
+      "epoch": 0.6336920796545724,
+      "grad_norm": 0.02846167078207075,
+      "learning_rate": 4.075187969924813e-06,
+      "loss": 0.0,
+      "step": 9833
+    },
+    {
+      "epoch": 0.6337565251015016,
+      "grad_norm": 0.0062590929192869165,
+      "learning_rate": 4.074471894020766e-06,
+      "loss": 0.0,
+      "step": 9834
+    },
+    {
+      "epoch": 0.6338209705484308,
+      "grad_norm": 0.3454879802491421,
+      "learning_rate": 4.073755818116721e-06,
+      "loss": 0.0009,
+      "step": 9835
+    },
+    {
+      "epoch": 0.63388541599536,
+      "grad_norm": 0.00633899276324277,
+      "learning_rate": 4.073039742212675e-06,
+      "loss": 0.0,
+      "step": 9836
+    },
+    {
+      "epoch": 0.6339498614422892,
+      "grad_norm": 0.014193635549446313,
+      "learning_rate": 4.072323666308629e-06,
+      "loss": 0.0,
+      "step": 9837
+    },
+    {
+      "epoch": 0.6340143068892182,
+      "grad_norm": 0.11098855918860485,
+      "learning_rate": 4.0716075904045835e-06,
+      "loss": 0.0002,
+      "step": 9838
+    },
+    {
+      "epoch": 0.6340787523361474,
+      "grad_norm": 0.0023207273082203215,
+      "learning_rate": 4.070891514500537e-06,
+      "loss": 0.0,
+      "step": 9839
+    },
+    {
+      "epoch": 0.6341431977830766,
+      "grad_norm": 0.0006344213112119258,
+      "learning_rate": 4.070175438596491e-06,
+      "loss": 0.0,
+      "step": 9840
+    },
+    {
+      "epoch": 0.6342076432300058,
+      "grad_norm": 0.08611299195893168,
+      "learning_rate": 4.0694593626924456e-06,
+      "loss": 0.0004,
+      "step": 9841
+    },
+    {
+      "epoch": 0.634272088676935,
+      "grad_norm": 8.981659056621043e-05,
+      "learning_rate": 4.0687432867884e-06,
+      "loss": 0.0,
+      "step": 9842
+    },
+    {
+      "epoch": 0.6343365341238641,
+      "grad_norm": 0.00012564965903519854,
+      "learning_rate": 4.068027210884354e-06,
+      "loss": 0.0,
+      "step": 9843
+    },
+    {
+      "epoch": 0.6344009795707933,
+      "grad_norm": 0.0006654507491736668,
+      "learning_rate": 4.0673111349803084e-06,
+      "loss": 0.0,
+      "step": 9844
+    },
+    {
+      "epoch": 0.6344654250177225,
+      "grad_norm": 0.0034752734788544896,
+      "learning_rate": 4.066595059076263e-06,
+      "loss": 0.0,
+      "step": 9845
+    },
+    {
+      "epoch": 0.6345298704646517,
+      "grad_norm": 0.0011339178097354364,
+      "learning_rate": 4.065878983172217e-06,
+      "loss": 0.0,
+      "step": 9846
+    },
+    {
+      "epoch": 0.6345943159115809,
+      "grad_norm": 0.008264259238473211,
+      "learning_rate": 4.0651629072681705e-06,
+      "loss": 0.0,
+      "step": 9847
+    },
+    {
+      "epoch": 0.6346587613585101,
+      "grad_norm": 0.003021064877237617,
+      "learning_rate": 4.064446831364125e-06,
+      "loss": 0.0,
+      "step": 9848
+    },
+    {
+      "epoch": 0.6347232068054391,
+      "grad_norm": 0.0014914980012132165,
+      "learning_rate": 4.063730755460079e-06,
+      "loss": 0.0,
+      "step": 9849
+    },
+    {
+      "epoch": 0.6347876522523683,
+      "grad_norm": 0.12833182633070414,
+      "learning_rate": 4.063014679556033e-06,
+      "loss": 0.0003,
+      "step": 9850
+    },
+    {
+      "epoch": 0.6348520976992975,
+      "grad_norm": 0.010818723560201225,
+      "learning_rate": 4.062298603651988e-06,
+      "loss": 0.0,
+      "step": 9851
+    },
+    {
+      "epoch": 0.6349165431462267,
+      "grad_norm": 0.15318523214957946,
+      "learning_rate": 4.061582527747941e-06,
+      "loss": 0.0004,
+      "step": 9852
+    },
+    {
+      "epoch": 0.6349809885931559,
+      "grad_norm": 0.19271358384033405,
+      "learning_rate": 4.0608664518438954e-06,
+      "loss": 0.0023,
+      "step": 9853
+    },
+    {
+      "epoch": 0.6350454340400851,
+      "grad_norm": 0.10603754438942112,
+      "learning_rate": 4.06015037593985e-06,
+      "loss": 0.0002,
+      "step": 9854
+    },
+    {
+      "epoch": 0.6351098794870143,
+      "grad_norm": 0.00042468406024622174,
+      "learning_rate": 4.059434300035804e-06,
+      "loss": 0.0,
+      "step": 9855
+    },
+    {
+      "epoch": 0.6351743249339434,
+      "grad_norm": 0.0011900971873508306,
+      "learning_rate": 4.058718224131758e-06,
+      "loss": 0.0,
+      "step": 9856
+    },
+    {
+      "epoch": 0.6352387703808726,
+      "grad_norm": 0.2271719250462828,
+      "learning_rate": 4.058002148227713e-06,
+      "loss": 0.0003,
+      "step": 9857
+    },
+    {
+      "epoch": 0.6353032158278018,
+      "grad_norm": 0.0006767117946808032,
+      "learning_rate": 4.057286072323667e-06,
+      "loss": 0.0,
+      "step": 9858
+    },
+    {
+      "epoch": 0.635367661274731,
+      "grad_norm": 0.00020405993774540835,
+      "learning_rate": 4.056569996419621e-06,
+      "loss": 0.0,
+      "step": 9859
+    },
+    {
+      "epoch": 0.6354321067216601,
+      "grad_norm": 0.0038386513119053446,
+      "learning_rate": 4.055853920515575e-06,
+      "loss": 0.0,
+      "step": 9860
+    },
+    {
+      "epoch": 0.6354965521685892,
+      "grad_norm": 0.14175836497304545,
+      "learning_rate": 4.055137844611529e-06,
+      "loss": 0.0041,
+      "step": 9861
+    },
+    {
+      "epoch": 0.6355609976155184,
+      "grad_norm": 0.5609968286873434,
+      "learning_rate": 4.054421768707483e-06,
+      "loss": 0.0015,
+      "step": 9862
+    },
+    {
+      "epoch": 0.6356254430624476,
+      "grad_norm": 0.22833693274969805,
+      "learning_rate": 4.053705692803438e-06,
+      "loss": 0.0037,
+      "step": 9863
+    },
+    {
+      "epoch": 0.6356898885093768,
+      "grad_norm": 0.0004343785643513413,
+      "learning_rate": 4.052989616899392e-06,
+      "loss": 0.0,
+      "step": 9864
+    },
+    {
+      "epoch": 0.635754333956306,
+      "grad_norm": 0.00029451475877984956,
+      "learning_rate": 4.052273540995345e-06,
+      "loss": 0.0,
+      "step": 9865
+    },
+    {
+      "epoch": 0.6358187794032352,
+      "grad_norm": 0.1578427849877735,
+      "learning_rate": 4.0515574650913e-06,
+      "loss": 0.0021,
+      "step": 9866
+    },
+    {
+      "epoch": 0.6358832248501644,
+      "grad_norm": 0.002889111443432167,
+      "learning_rate": 4.050841389187255e-06,
+      "loss": 0.0,
+      "step": 9867
+    },
+    {
+      "epoch": 0.6359476702970935,
+      "grad_norm": 0.015472808998329517,
+      "learning_rate": 4.050125313283208e-06,
+      "loss": 0.0,
+      "step": 9868
+    },
+    {
+      "epoch": 0.6360121157440227,
+      "grad_norm": 0.0006547937423044578,
+      "learning_rate": 4.0494092373791625e-06,
+      "loss": 0.0,
+      "step": 9869
+    },
+    {
+      "epoch": 0.6360765611909519,
+      "grad_norm": 0.0016740075257970334,
+      "learning_rate": 4.048693161475117e-06,
+      "loss": 0.0,
+      "step": 9870
+    },
+    {
+      "epoch": 0.636141006637881,
+      "grad_norm": 0.0032675604364454484,
+      "learning_rate": 4.047977085571071e-06,
+      "loss": 0.0,
+      "step": 9871
+    },
+    {
+      "epoch": 0.6362054520848102,
+      "grad_norm": 0.07995895205773623,
+      "learning_rate": 4.047261009667025e-06,
+      "loss": 0.0001,
+      "step": 9872
+    },
+    {
+      "epoch": 0.6362698975317393,
+      "grad_norm": 0.01585555228427254,
+      "learning_rate": 4.046544933762979e-06,
+      "loss": 0.0002,
+      "step": 9873
+    },
+    {
+      "epoch": 0.6363343429786685,
+      "grad_norm": 0.0005467433965174832,
+      "learning_rate": 4.045828857858933e-06,
+      "loss": 0.0,
+      "step": 9874
+    },
+    {
+      "epoch": 0.6363987884255977,
+      "grad_norm": 0.002782702034882121,
+      "learning_rate": 4.0451127819548875e-06,
+      "loss": 0.0,
+      "step": 9875
+    },
+    {
+      "epoch": 0.6364632338725269,
+      "grad_norm": 0.0010571305041660374,
+      "learning_rate": 4.044396706050842e-06,
+      "loss": 0.0,
+      "step": 9876
+    },
+    {
+      "epoch": 0.6365276793194561,
+      "grad_norm": 0.671455673838229,
+      "learning_rate": 4.043680630146796e-06,
+      "loss": 0.0056,
+      "step": 9877
+    },
+    {
+      "epoch": 0.6365921247663853,
+      "grad_norm": 6.159026134501892e-05,
+      "learning_rate": 4.0429645542427495e-06,
+      "loss": 0.0,
+      "step": 9878
+    },
+    {
+      "epoch": 0.6366565702133145,
+      "grad_norm": 0.11269901540612999,
+      "learning_rate": 4.042248478338705e-06,
+      "loss": 0.0002,
+      "step": 9879
+    },
+    {
+      "epoch": 0.6367210156602436,
+      "grad_norm": 0.13829956983992775,
+      "learning_rate": 4.041532402434659e-06,
+      "loss": 0.0004,
+      "step": 9880
+    },
+    {
+      "epoch": 0.6367854611071728,
+      "grad_norm": 0.00028129003725998087,
+      "learning_rate": 4.040816326530612e-06,
+      "loss": 0.0,
+      "step": 9881
+    },
+    {
+      "epoch": 0.636849906554102,
+      "grad_norm": 0.18228105218956056,
+      "learning_rate": 4.040100250626567e-06,
+      "loss": 0.0002,
+      "step": 9882
+    },
+    {
+      "epoch": 0.6369143520010311,
+      "grad_norm": 0.00044808417187598156,
+      "learning_rate": 4.039384174722521e-06,
+      "loss": 0.0,
+      "step": 9883
+    },
+    {
+      "epoch": 0.6369787974479603,
+      "grad_norm": 0.004541619186065626,
+      "learning_rate": 4.038668098818475e-06,
+      "loss": 0.0,
+      "step": 9884
+    },
+    {
+      "epoch": 0.6370432428948894,
+      "grad_norm": 0.1287355882717353,
+      "learning_rate": 4.03795202291443e-06,
+      "loss": 0.0004,
+      "step": 9885
+    },
+    {
+      "epoch": 0.6371076883418186,
+      "grad_norm": 0.000665257157196762,
+      "learning_rate": 4.037235947010383e-06,
+      "loss": 0.0,
+      "step": 9886
+    },
+    {
+      "epoch": 0.6371721337887478,
+      "grad_norm": 0.00015430858028718633,
+      "learning_rate": 4.036519871106337e-06,
+      "loss": 0.0,
+      "step": 9887
+    },
+    {
+      "epoch": 0.637236579235677,
+      "grad_norm": 0.01693254525602601,
+      "learning_rate": 4.035803795202292e-06,
+      "loss": 0.0001,
+      "step": 9888
+    },
+    {
+      "epoch": 0.6373010246826062,
+      "grad_norm": 0.000977959590454271,
+      "learning_rate": 4.035087719298246e-06,
+      "loss": 0.0,
+      "step": 9889
+    },
+    {
+      "epoch": 0.6373654701295354,
+      "grad_norm": 0.0001807368223541276,
+      "learning_rate": 4.0343716433942e-06,
+      "loss": 0.0,
+      "step": 9890
+    },
+    {
+      "epoch": 0.6374299155764646,
+      "grad_norm": 0.13511406041563384,
+      "learning_rate": 4.0336555674901546e-06,
+      "loss": 0.0004,
+      "step": 9891
+    },
+    {
+      "epoch": 0.6374943610233937,
+      "grad_norm": 0.00462214646127079,
+      "learning_rate": 4.032939491586109e-06,
+      "loss": 0.0,
+      "step": 9892
+    },
+    {
+      "epoch": 0.6375588064703229,
+      "grad_norm": 8.380419679333197e-05,
+      "learning_rate": 4.032223415682063e-06,
+      "loss": 0.0,
+      "step": 9893
+    },
+    {
+      "epoch": 0.637623251917252,
+      "grad_norm": 0.001185555940977872,
+      "learning_rate": 4.031507339778017e-06,
+      "loss": 0.0,
+      "step": 9894
+    },
+    {
+      "epoch": 0.6376876973641812,
+      "grad_norm": 0.01787338779530966,
+      "learning_rate": 4.030791263873971e-06,
+      "loss": 0.0,
+      "step": 9895
+    },
+    {
+      "epoch": 0.6377521428111104,
+      "grad_norm": 0.05857672841469521,
+      "learning_rate": 4.030075187969925e-06,
+      "loss": 0.0003,
+      "step": 9896
+    },
+    {
+      "epoch": 0.6378165882580396,
+      "grad_norm": 0.042201623274486,
+      "learning_rate": 4.0293591120658795e-06,
+      "loss": 0.0002,
+      "step": 9897
+    },
+    {
+      "epoch": 0.6378810337049687,
+      "grad_norm": 0.014896175954414791,
+      "learning_rate": 4.028643036161834e-06,
+      "loss": 0.0,
+      "step": 9898
+    },
+    {
+      "epoch": 0.6379454791518979,
+      "grad_norm": 0.006496097356850603,
+      "learning_rate": 4.027926960257787e-06,
+      "loss": 0.0,
+      "step": 9899
+    },
+    {
+      "epoch": 0.6380099245988271,
+      "grad_norm": 0.0005663189838518361,
+      "learning_rate": 4.0272108843537416e-06,
+      "loss": 0.0,
+      "step": 9900
+    },
+    {
+      "epoch": 0.6380743700457563,
+      "grad_norm": 0.00168549572276779,
+      "learning_rate": 4.026494808449696e-06,
+      "loss": 0.0,
+      "step": 9901
+    },
+    {
+      "epoch": 0.6381388154926855,
+      "grad_norm": 0.021107483516845608,
+      "learning_rate": 4.02577873254565e-06,
+      "loss": 0.0001,
+      "step": 9902
+    },
+    {
+      "epoch": 0.6382032609396147,
+      "grad_norm": 7.06100544471114e-05,
+      "learning_rate": 4.0250626566416044e-06,
+      "loss": 0.0,
+      "step": 9903
+    },
+    {
+      "epoch": 0.6382677063865438,
+      "grad_norm": 0.004029435938904156,
+      "learning_rate": 4.024346580737559e-06,
+      "loss": 0.0,
+      "step": 9904
+    },
+    {
+      "epoch": 0.6383321518334729,
+      "grad_norm": 0.003942796891727747,
+      "learning_rate": 4.023630504833513e-06,
+      "loss": 0.0,
+      "step": 9905
+    },
+    {
+      "epoch": 0.6383965972804021,
+      "grad_norm": 0.10042532347163381,
+      "learning_rate": 4.022914428929467e-06,
+      "loss": 0.0055,
+      "step": 9906
+    },
+    {
+      "epoch": 0.6384610427273313,
+      "grad_norm": 0.013568788087903462,
+      "learning_rate": 4.022198353025421e-06,
+      "loss": 0.0,
+      "step": 9907
+    },
+    {
+      "epoch": 0.6385254881742605,
+      "grad_norm": 0.11221282353071427,
+      "learning_rate": 4.021482277121375e-06,
+      "loss": 0.0001,
+      "step": 9908
+    },
+    {
+      "epoch": 0.6385899336211897,
+      "grad_norm": 0.00044984605912318877,
+      "learning_rate": 4.020766201217329e-06,
+      "loss": 0.0,
+      "step": 9909
+    },
+    {
+      "epoch": 0.6386543790681188,
+      "grad_norm": 0.00013384249196529097,
+      "learning_rate": 4.020050125313284e-06,
+      "loss": 0.0,
+      "step": 9910
+    },
+    {
+      "epoch": 0.638718824515048,
+      "grad_norm": 0.003407667433976282,
+      "learning_rate": 4.019334049409237e-06,
+      "loss": 0.0,
+      "step": 9911
+    },
+    {
+      "epoch": 0.6387832699619772,
+      "grad_norm": 0.011748166466181148,
+      "learning_rate": 4.0186179735051914e-06,
+      "loss": 0.0,
+      "step": 9912
+    },
+    {
+      "epoch": 0.6388477154089064,
+      "grad_norm": 0.027792480039555506,
+      "learning_rate": 4.017901897601146e-06,
+      "loss": 0.0,
+      "step": 9913
+    },
+    {
+      "epoch": 0.6389121608558356,
+      "grad_norm": 0.008272129900484659,
+      "learning_rate": 4.017185821697101e-06,
+      "loss": 0.0001,
+      "step": 9914
+    },
+    {
+      "epoch": 0.6389766063027648,
+      "grad_norm": 0.0002833124323843118,
+      "learning_rate": 4.016469745793054e-06,
+      "loss": 0.0,
+      "step": 9915
+    },
+    {
+      "epoch": 0.6390410517496938,
+      "grad_norm": 0.025964498676430598,
+      "learning_rate": 4.015753669889009e-06,
+      "loss": 0.0,
+      "step": 9916
+    },
+    {
+      "epoch": 0.639105497196623,
+      "grad_norm": 0.17665134300402457,
+      "learning_rate": 4.015037593984963e-06,
+      "loss": 0.0008,
+      "step": 9917
+    },
+    {
+      "epoch": 0.6391699426435522,
+      "grad_norm": 0.010062429170806275,
+      "learning_rate": 4.014321518080917e-06,
+      "loss": 0.0,
+      "step": 9918
+    },
+    {
+      "epoch": 0.6392343880904814,
+      "grad_norm": 0.001940753301590691,
+      "learning_rate": 4.013605442176871e-06,
+      "loss": 0.0,
+      "step": 9919
+    },
+    {
+      "epoch": 0.6392988335374106,
+      "grad_norm": 0.13457927397190214,
+      "learning_rate": 4.012889366272825e-06,
+      "loss": 0.0004,
+      "step": 9920
+    },
+    {
+      "epoch": 0.6393632789843398,
+      "grad_norm": 0.00035575603334679986,
+      "learning_rate": 4.012173290368779e-06,
+      "loss": 0.0,
+      "step": 9921
+    },
+    {
+      "epoch": 0.6394277244312689,
+      "grad_norm": 0.19851455393941003,
+      "learning_rate": 4.011457214464734e-06,
+      "loss": 0.0019,
+      "step": 9922
+    },
+    {
+      "epoch": 0.6394921698781981,
+      "grad_norm": 0.0008800058176206106,
+      "learning_rate": 4.010741138560688e-06,
+      "loss": 0.0,
+      "step": 9923
+    },
+    {
+      "epoch": 0.6395566153251273,
+      "grad_norm": 0.015232629286257246,
+      "learning_rate": 4.010025062656641e-06,
+      "loss": 0.0,
+      "step": 9924
+    },
+    {
+      "epoch": 0.6396210607720565,
+      "grad_norm": 0.005831437638291429,
+      "learning_rate": 4.009308986752596e-06,
+      "loss": 0.0,
+      "step": 9925
+    },
+    {
+      "epoch": 0.6396855062189857,
+      "grad_norm": 0.00974646379763554,
+      "learning_rate": 4.008592910848551e-06,
+      "loss": 0.0,
+      "step": 9926
+    },
+    {
+      "epoch": 0.6397499516659148,
+      "grad_norm": 0.00011297209002473684,
+      "learning_rate": 4.007876834944504e-06,
+      "loss": 0.0,
+      "step": 9927
+    },
+    {
+      "epoch": 0.6398143971128439,
+      "grad_norm": 0.00044290672661680006,
+      "learning_rate": 4.0071607590404585e-06,
+      "loss": 0.0,
+      "step": 9928
+    },
+    {
+      "epoch": 0.6398788425597731,
+      "grad_norm": 0.002342356820183904,
+      "learning_rate": 4.006444683136413e-06,
+      "loss": 0.0,
+      "step": 9929
+    },
+    {
+      "epoch": 0.6399432880067023,
+      "grad_norm": 0.002250269272006791,
+      "learning_rate": 4.005728607232367e-06,
+      "loss": 0.0,
+      "step": 9930
+    },
+    {
+      "epoch": 0.6400077334536315,
+      "grad_norm": 0.007098118431674166,
+      "learning_rate": 4.005012531328321e-06,
+      "loss": 0.0001,
+      "step": 9931
+    },
+    {
+      "epoch": 0.6400721789005607,
+      "grad_norm": 0.0006225908267203256,
+      "learning_rate": 4.004296455424275e-06,
+      "loss": 0.0,
+      "step": 9932
+    },
+    {
+      "epoch": 0.6401366243474899,
+      "grad_norm": 0.002042717291839209,
+      "learning_rate": 4.003580379520229e-06,
+      "loss": 0.0,
+      "step": 9933
+    },
+    {
+      "epoch": 0.640201069794419,
+      "grad_norm": 0.8397476141266291,
+      "learning_rate": 4.0028643036161835e-06,
+      "loss": 0.0014,
+      "step": 9934
+    },
+    {
+      "epoch": 0.6402655152413482,
+      "grad_norm": 0.01909159525474567,
+      "learning_rate": 4.002148227712138e-06,
+      "loss": 0.0001,
+      "step": 9935
+    },
+    {
+      "epoch": 0.6403299606882774,
+      "grad_norm": 0.0063646263341570804,
+      "learning_rate": 4.001432151808092e-06,
+      "loss": 0.0,
+      "step": 9936
+    },
+    {
+      "epoch": 0.6403944061352066,
+      "grad_norm": 0.00024152938378510026,
+      "learning_rate": 4.000716075904046e-06,
+      "loss": 0.0,
+      "step": 9937
+    },
+    {
+      "epoch": 0.6404588515821357,
+      "grad_norm": 0.00029932230104452766,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.0,
+      "step": 9938
+    },
+    {
+      "epoch": 0.6405232970290649,
+      "grad_norm": 0.01575336589201022,
+      "learning_rate": 3.999283924095955e-06,
+      "loss": 0.0001,
+      "step": 9939
+    },
+    {
+      "epoch": 0.640587742475994,
+      "grad_norm": 0.4669428800544325,
+      "learning_rate": 3.998567848191908e-06,
+      "loss": 0.0021,
+      "step": 9940
+    },
+    {
+      "epoch": 0.6406521879229232,
+      "grad_norm": 0.0012947301126589104,
+      "learning_rate": 3.997851772287863e-06,
+      "loss": 0.0,
+      "step": 9941
+    },
+    {
+      "epoch": 0.6407166333698524,
+      "grad_norm": 0.013855897877270457,
+      "learning_rate": 3.997135696383817e-06,
+      "loss": 0.0,
+      "step": 9942
+    },
+    {
+      "epoch": 0.6407810788167816,
+      "grad_norm": 0.00033208404617710225,
+      "learning_rate": 3.996419620479771e-06,
+      "loss": 0.0,
+      "step": 9943
+    },
+    {
+      "epoch": 0.6408455242637108,
+      "grad_norm": 0.10435821286848203,
+      "learning_rate": 3.995703544575726e-06,
+      "loss": 0.0006,
+      "step": 9944
+    },
+    {
+      "epoch": 0.64090996971064,
+      "grad_norm": 0.015503957950853523,
+      "learning_rate": 3.994987468671679e-06,
+      "loss": 0.0,
+      "step": 9945
+    },
+    {
+      "epoch": 0.6409744151575691,
+      "grad_norm": 0.0005417660503162809,
+      "learning_rate": 3.994271392767633e-06,
+      "loss": 0.0,
+      "step": 9946
+    },
+    {
+      "epoch": 0.6410388606044983,
+      "grad_norm": 2.6578900723737707e-05,
+      "learning_rate": 3.993555316863588e-06,
+      "loss": 0.0,
+      "step": 9947
+    },
+    {
+      "epoch": 0.6411033060514275,
+      "grad_norm": 0.00017938237437263428,
+      "learning_rate": 3.992839240959542e-06,
+      "loss": 0.0,
+      "step": 9948
+    },
+    {
+      "epoch": 0.6411677514983566,
+      "grad_norm": 0.34429821818284284,
+      "learning_rate": 3.992123165055496e-06,
+      "loss": 0.0011,
+      "step": 9949
+    },
+    {
+      "epoch": 0.6412321969452858,
+      "grad_norm": 0.0034726296503280736,
+      "learning_rate": 3.9914070891514506e-06,
+      "loss": 0.0,
+      "step": 9950
+    },
+    {
+      "epoch": 0.641296642392215,
+      "grad_norm": 0.00018911348694327,
+      "learning_rate": 3.990691013247405e-06,
+      "loss": 0.0,
+      "step": 9951
+    },
+    {
+      "epoch": 0.6413610878391441,
+      "grad_norm": 0.0021938632494621483,
+      "learning_rate": 3.989974937343359e-06,
+      "loss": 0.0,
+      "step": 9952
+    },
+    {
+      "epoch": 0.6414255332860733,
+      "grad_norm": 0.0012114879226396045,
+      "learning_rate": 3.989258861439313e-06,
+      "loss": 0.0,
+      "step": 9953
+    },
+    {
+      "epoch": 0.6414899787330025,
+      "grad_norm": 0.05394574339195268,
+      "learning_rate": 3.988542785535267e-06,
+      "loss": 0.0005,
+      "step": 9954
+    },
+    {
+      "epoch": 0.6415544241799317,
+      "grad_norm": 0.9579358243969295,
+      "learning_rate": 3.987826709631221e-06,
+      "loss": 0.0048,
+      "step": 9955
+    },
+    {
+      "epoch": 0.6416188696268609,
+      "grad_norm": 0.34143033673758666,
+      "learning_rate": 3.9871106337271755e-06,
+      "loss": 0.0042,
+      "step": 9956
+    },
+    {
+      "epoch": 0.6416833150737901,
+      "grad_norm": 0.17423421059847782,
+      "learning_rate": 3.98639455782313e-06,
+      "loss": 0.0005,
+      "step": 9957
+    },
+    {
+      "epoch": 0.6417477605207192,
+      "grad_norm": 0.0009932734330962791,
+      "learning_rate": 3.985678481919083e-06,
+      "loss": 0.0,
+      "step": 9958
+    },
+    {
+      "epoch": 0.6418122059676484,
+      "grad_norm": 0.000888577600146786,
+      "learning_rate": 3.9849624060150376e-06,
+      "loss": 0.0,
+      "step": 9959
+    },
+    {
+      "epoch": 0.6418766514145776,
+      "grad_norm": 0.012914636043180742,
+      "learning_rate": 3.984246330110992e-06,
+      "loss": 0.0002,
+      "step": 9960
+    },
+    {
+      "epoch": 0.6419410968615067,
+      "grad_norm": 0.04609532181840023,
+      "learning_rate": 3.983530254206946e-06,
+      "loss": 0.0,
+      "step": 9961
+    },
+    {
+      "epoch": 0.6420055423084359,
+      "grad_norm": 0.16728946370704814,
+      "learning_rate": 3.9828141783029004e-06,
+      "loss": 0.0015,
+      "step": 9962
+    },
+    {
+      "epoch": 0.6420699877553651,
+      "grad_norm": 0.0006231815176938104,
+      "learning_rate": 3.982098102398855e-06,
+      "loss": 0.0,
+      "step": 9963
+    },
+    {
+      "epoch": 0.6421344332022942,
+      "grad_norm": 0.04196601012932234,
+      "learning_rate": 3.981382026494809e-06,
+      "loss": 0.0001,
+      "step": 9964
+    },
+    {
+      "epoch": 0.6421988786492234,
+      "grad_norm": 0.6129773352744001,
+      "learning_rate": 3.980665950590763e-06,
+      "loss": 0.0047,
+      "step": 9965
+    },
+    {
+      "epoch": 0.6422633240961526,
+      "grad_norm": 0.006540673913466794,
+      "learning_rate": 3.979949874686717e-06,
+      "loss": 0.0,
+      "step": 9966
+    },
+    {
+      "epoch": 0.6423277695430818,
+      "grad_norm": 0.010330284011252865,
+      "learning_rate": 3.979233798782671e-06,
+      "loss": 0.0001,
+      "step": 9967
+    },
+    {
+      "epoch": 0.642392214990011,
+      "grad_norm": 0.0012068267576809393,
+      "learning_rate": 3.978517722878625e-06,
+      "loss": 0.0,
+      "step": 9968
+    },
+    {
+      "epoch": 0.6424566604369402,
+      "grad_norm": 0.004279481705147905,
+      "learning_rate": 3.97780164697458e-06,
+      "loss": 0.0,
+      "step": 9969
+    },
+    {
+      "epoch": 0.6425211058838693,
+      "grad_norm": 0.015038512520982627,
+      "learning_rate": 3.977085571070534e-06,
+      "loss": 0.0,
+      "step": 9970
+    },
+    {
+      "epoch": 0.6425855513307985,
+      "grad_norm": 0.0005527755708677373,
+      "learning_rate": 3.9763694951664874e-06,
+      "loss": 0.0,
+      "step": 9971
+    },
+    {
+      "epoch": 0.6426499967777276,
+      "grad_norm": 0.002589969194584389,
+      "learning_rate": 3.975653419262443e-06,
+      "loss": 0.0,
+      "step": 9972
+    },
+    {
+      "epoch": 0.6427144422246568,
+      "grad_norm": 0.4182773591616916,
+      "learning_rate": 3.974937343358397e-06,
+      "loss": 0.0002,
+      "step": 9973
+    },
+    {
+      "epoch": 0.642778887671586,
+      "grad_norm": 0.09478272415081301,
+      "learning_rate": 3.97422126745435e-06,
+      "loss": 0.0001,
+      "step": 9974
+    },
+    {
+      "epoch": 0.6428433331185152,
+      "grad_norm": 0.06265709550030224,
+      "learning_rate": 3.973505191550305e-06,
+      "loss": 0.0,
+      "step": 9975
+    },
+    {
+      "epoch": 0.6429077785654443,
+      "grad_norm": 0.004482413553391734,
+      "learning_rate": 3.972789115646259e-06,
+      "loss": 0.0,
+      "step": 9976
+    },
+    {
+      "epoch": 0.6429722240123735,
+      "grad_norm": 0.012129023568842196,
+      "learning_rate": 3.972073039742213e-06,
+      "loss": 0.0001,
+      "step": 9977
+    },
+    {
+      "epoch": 0.6430366694593027,
+      "grad_norm": 0.02426212837729917,
+      "learning_rate": 3.9713569638381675e-06,
+      "loss": 0.0001,
+      "step": 9978
+    },
+    {
+      "epoch": 0.6431011149062319,
+      "grad_norm": 0.388016344424232,
+      "learning_rate": 3.970640887934121e-06,
+      "loss": 0.0015,
+      "step": 9979
+    },
+    {
+      "epoch": 0.6431655603531611,
+      "grad_norm": 0.009141367046433759,
+      "learning_rate": 3.969924812030075e-06,
+      "loss": 0.0,
+      "step": 9980
+    },
+    {
+      "epoch": 0.6432300058000903,
+      "grad_norm": 0.0032954234977543386,
+      "learning_rate": 3.96920873612603e-06,
+      "loss": 0.0,
+      "step": 9981
+    },
+    {
+      "epoch": 0.6432944512470194,
+      "grad_norm": 0.0022175645149016825,
+      "learning_rate": 3.968492660221984e-06,
+      "loss": 0.0,
+      "step": 9982
+    },
+    {
+      "epoch": 0.6433588966939485,
+      "grad_norm": 0.0007280509024169878,
+      "learning_rate": 3.967776584317938e-06,
+      "loss": 0.0,
+      "step": 9983
+    },
+    {
+      "epoch": 0.6434233421408777,
+      "grad_norm": 0.056778146835200966,
+      "learning_rate": 3.9670605084138925e-06,
+      "loss": 0.0001,
+      "step": 9984
+    },
+    {
+      "epoch": 0.6434877875878069,
+      "grad_norm": 0.26815644299933966,
+      "learning_rate": 3.966344432509847e-06,
+      "loss": 0.0022,
+      "step": 9985
+    },
+    {
+      "epoch": 0.6435522330347361,
+      "grad_norm": 0.007594321936676571,
+      "learning_rate": 3.965628356605801e-06,
+      "loss": 0.0,
+      "step": 9986
+    },
+    {
+      "epoch": 0.6436166784816653,
+      "grad_norm": 0.001696701726712196,
+      "learning_rate": 3.9649122807017545e-06,
+      "loss": 0.0,
+      "step": 9987
+    },
+    {
+      "epoch": 0.6436811239285944,
+      "grad_norm": 0.001928534918932584,
+      "learning_rate": 3.964196204797709e-06,
+      "loss": 0.0,
+      "step": 9988
+    },
+    {
+      "epoch": 0.6437455693755236,
+      "grad_norm": 0.08768668886361478,
+      "learning_rate": 3.963480128893663e-06,
+      "loss": 0.0008,
+      "step": 9989
+    },
+    {
+      "epoch": 0.6438100148224528,
+      "grad_norm": 0.001866720703754344,
+      "learning_rate": 3.962764052989617e-06,
+      "loss": 0.0,
+      "step": 9990
+    },
+    {
+      "epoch": 0.643874460269382,
+      "grad_norm": 0.11344296324003514,
+      "learning_rate": 3.962047977085572e-06,
+      "loss": 0.0017,
+      "step": 9991
+    },
+    {
+      "epoch": 0.6439389057163112,
+      "grad_norm": 0.0041534937274514475,
+      "learning_rate": 3.961331901181525e-06,
+      "loss": 0.0,
+      "step": 9992
+    },
+    {
+      "epoch": 0.6440033511632404,
+      "grad_norm": 0.0043058563308134915,
+      "learning_rate": 3.9606158252774795e-06,
+      "loss": 0.0,
+      "step": 9993
+    },
+    {
+      "epoch": 0.6440677966101694,
+      "grad_norm": 0.0028722903588658793,
+      "learning_rate": 3.959899749373434e-06,
+      "loss": 0.0,
+      "step": 9994
+    },
+    {
+      "epoch": 0.6441322420570986,
+      "grad_norm": 0.0002852931476689053,
+      "learning_rate": 3.959183673469388e-06,
+      "loss": 0.0,
+      "step": 9995
+    },
+    {
+      "epoch": 0.6441966875040278,
+      "grad_norm": 1.4214037439131955,
+      "learning_rate": 3.958467597565342e-06,
+      "loss": 0.0013,
+      "step": 9996
+    },
+    {
+      "epoch": 0.644261132950957,
+      "grad_norm": 0.002912964128315929,
+      "learning_rate": 3.957751521661297e-06,
+      "loss": 0.0,
+      "step": 9997
+    },
+    {
+      "epoch": 0.6443255783978862,
+      "grad_norm": 0.008286350517975941,
+      "learning_rate": 3.957035445757251e-06,
+      "loss": 0.0,
+      "step": 9998
+    },
+    {
+      "epoch": 0.6443900238448154,
+      "grad_norm": 0.0009665192786302733,
+      "learning_rate": 3.956319369853205e-06,
+      "loss": 0.0,
+      "step": 9999
+    },
+    {
+      "epoch": 0.6444544692917445,
+      "grad_norm": 0.0026490413554310444,
+      "learning_rate": 3.955603293949159e-06,
+      "loss": 0.0,
+      "step": 10000
+    },
+    {
+      "epoch": 0.6445189147386737,
+      "grad_norm": 0.01759335401256255,
+      "learning_rate": 3.954887218045113e-06,
+      "loss": 0.0,
+      "step": 10001
+    },
+    {
+      "epoch": 0.6445833601856029,
+      "grad_norm": 0.009520969315805258,
+      "learning_rate": 3.954171142141067e-06,
+      "loss": 0.0,
+      "step": 10002
+    },
+    {
+      "epoch": 0.6446478056325321,
+      "grad_norm": 0.001728311984742966,
+      "learning_rate": 3.953455066237022e-06,
+      "loss": 0.0,
+      "step": 10003
+    },
+    {
+      "epoch": 0.6447122510794613,
+      "grad_norm": 0.0009198832699478046,
+      "learning_rate": 3.952738990332975e-06,
+      "loss": 0.0,
+      "step": 10004
+    },
+    {
+      "epoch": 0.6447766965263904,
+      "grad_norm": 0.0011534950561722203,
+      "learning_rate": 3.952022914428929e-06,
+      "loss": 0.0,
+      "step": 10005
+    },
+    {
+      "epoch": 0.6448411419733195,
+      "grad_norm": 0.0003031917733232299,
+      "learning_rate": 3.951306838524884e-06,
+      "loss": 0.0,
+      "step": 10006
+    },
+    {
+      "epoch": 0.6449055874202487,
+      "grad_norm": 0.001510916530904056,
+      "learning_rate": 3.950590762620839e-06,
+      "loss": 0.0,
+      "step": 10007
+    },
+    {
+      "epoch": 0.6449700328671779,
+      "grad_norm": 0.0451465203765521,
+      "learning_rate": 3.949874686716792e-06,
+      "loss": 0.0004,
+      "step": 10008
+    },
+    {
+      "epoch": 0.6450344783141071,
+      "grad_norm": 0.0009382365449063471,
+      "learning_rate": 3.9491586108127466e-06,
+      "loss": 0.0,
+      "step": 10009
+    },
+    {
+      "epoch": 0.6450989237610363,
+      "grad_norm": 0.00043472888646857314,
+      "learning_rate": 3.948442534908701e-06,
+      "loss": 0.0,
+      "step": 10010
+    },
+    {
+      "epoch": 0.6451633692079655,
+      "grad_norm": 0.0012430078323916194,
+      "learning_rate": 3.947726459004655e-06,
+      "loss": 0.0,
+      "step": 10011
+    },
+    {
+      "epoch": 0.6452278146548946,
+      "grad_norm": 0.0014639720055084765,
+      "learning_rate": 3.947010383100609e-06,
+      "loss": 0.0,
+      "step": 10012
+    },
+    {
+      "epoch": 0.6452922601018238,
+      "grad_norm": 0.002401018043739887,
+      "learning_rate": 3.946294307196563e-06,
+      "loss": 0.0,
+      "step": 10013
+    },
+    {
+      "epoch": 0.645356705548753,
+      "grad_norm": 0.015478704234006003,
+      "learning_rate": 3.945578231292517e-06,
+      "loss": 0.0,
+      "step": 10014
+    },
+    {
+      "epoch": 0.6454211509956822,
+      "grad_norm": 0.001973197331748077,
+      "learning_rate": 3.9448621553884715e-06,
+      "loss": 0.0,
+      "step": 10015
+    },
+    {
+      "epoch": 0.6454855964426113,
+      "grad_norm": 0.05995384634129306,
+      "learning_rate": 3.944146079484426e-06,
+      "loss": 0.0016,
+      "step": 10016
+    },
+    {
+      "epoch": 0.6455500418895405,
+      "grad_norm": 0.04199333799249915,
+      "learning_rate": 3.943430003580379e-06,
+      "loss": 0.0,
+      "step": 10017
+    },
+    {
+      "epoch": 0.6456144873364696,
+      "grad_norm": 0.05694073473606429,
+      "learning_rate": 3.9427139276763336e-06,
+      "loss": 0.0001,
+      "step": 10018
+    },
+    {
+      "epoch": 0.6456789327833988,
+      "grad_norm": 0.0034993164557464897,
+      "learning_rate": 3.941997851772289e-06,
+      "loss": 0.0,
+      "step": 10019
+    },
+    {
+      "epoch": 0.645743378230328,
+      "grad_norm": 0.009195842661245979,
+      "learning_rate": 3.941281775868242e-06,
+      "loss": 0.0001,
+      "step": 10020
+    },
+    {
+      "epoch": 0.6458078236772572,
+      "grad_norm": 0.00044558898441164694,
+      "learning_rate": 3.9405656999641964e-06,
+      "loss": 0.0,
+      "step": 10021
+    },
+    {
+      "epoch": 0.6458722691241864,
+      "grad_norm": 0.0009647395321825095,
+      "learning_rate": 3.939849624060151e-06,
+      "loss": 0.0,
+      "step": 10022
+    },
+    {
+      "epoch": 0.6459367145711156,
+      "grad_norm": 0.05492106017896077,
+      "learning_rate": 3.939133548156105e-06,
+      "loss": 0.0001,
+      "step": 10023
+    },
+    {
+      "epoch": 0.6460011600180448,
+      "grad_norm": 0.11704149687743283,
+      "learning_rate": 3.938417472252059e-06,
+      "loss": 0.0001,
+      "step": 10024
+    },
+    {
+      "epoch": 0.6460656054649739,
+      "grad_norm": 0.005956494344295038,
+      "learning_rate": 3.937701396348013e-06,
+      "loss": 0.0,
+      "step": 10025
+    },
+    {
+      "epoch": 0.6461300509119031,
+      "grad_norm": 0.00031369930774159714,
+      "learning_rate": 3.936985320443967e-06,
+      "loss": 0.0,
+      "step": 10026
+    },
+    {
+      "epoch": 0.6461944963588322,
+      "grad_norm": 0.06431663756990698,
+      "learning_rate": 3.936269244539921e-06,
+      "loss": 0.0007,
+      "step": 10027
+    },
+    {
+      "epoch": 0.6462589418057614,
+      "grad_norm": 0.0011454047402656453,
+      "learning_rate": 3.935553168635876e-06,
+      "loss": 0.0,
+      "step": 10028
+    },
+    {
+      "epoch": 0.6463233872526906,
+      "grad_norm": 0.005924979209156877,
+      "learning_rate": 3.93483709273183e-06,
+      "loss": 0.0,
+      "step": 10029
+    },
+    {
+      "epoch": 0.6463878326996197,
+      "grad_norm": 0.021247387098242476,
+      "learning_rate": 3.9341210168277834e-06,
+      "loss": 0.0003,
+      "step": 10030
+    },
+    {
+      "epoch": 0.6464522781465489,
+      "grad_norm": 0.009782796901761426,
+      "learning_rate": 3.933404940923739e-06,
+      "loss": 0.0,
+      "step": 10031
+    },
+    {
+      "epoch": 0.6465167235934781,
+      "grad_norm": 0.034589421586336565,
+      "learning_rate": 3.932688865019693e-06,
+      "loss": 0.0,
+      "step": 10032
+    },
+    {
+      "epoch": 0.6465811690404073,
+      "grad_norm": 0.18486351863565736,
+      "learning_rate": 3.931972789115646e-06,
+      "loss": 0.002,
+      "step": 10033
+    },
+    {
+      "epoch": 0.6466456144873365,
+      "grad_norm": 7.558595765490377e-05,
+      "learning_rate": 3.931256713211601e-06,
+      "loss": 0.0,
+      "step": 10034
+    },
+    {
+      "epoch": 0.6467100599342657,
+      "grad_norm": 0.001811319970900262,
+      "learning_rate": 3.930540637307555e-06,
+      "loss": 0.0,
+      "step": 10035
+    },
+    {
+      "epoch": 0.6467745053811949,
+      "grad_norm": 0.00010906105212224324,
+      "learning_rate": 3.929824561403509e-06,
+      "loss": 0.0,
+      "step": 10036
+    },
+    {
+      "epoch": 0.646838950828124,
+      "grad_norm": 3.836100339832203,
+      "learning_rate": 3.9291084854994635e-06,
+      "loss": 0.0449,
+      "step": 10037
+    },
+    {
+      "epoch": 0.6469033962750532,
+      "grad_norm": 0.0005711962089788333,
+      "learning_rate": 3.928392409595417e-06,
+      "loss": 0.0,
+      "step": 10038
+    },
+    {
+      "epoch": 0.6469678417219823,
+      "grad_norm": 0.0012421915486166405,
+      "learning_rate": 3.927676333691371e-06,
+      "loss": 0.0,
+      "step": 10039
+    },
+    {
+      "epoch": 0.6470322871689115,
+      "grad_norm": 0.10516009377572318,
+      "learning_rate": 3.926960257787326e-06,
+      "loss": 0.0004,
+      "step": 10040
+    },
+    {
+      "epoch": 0.6470967326158407,
+      "grad_norm": 0.011991362360826062,
+      "learning_rate": 3.92624418188328e-06,
+      "loss": 0.0,
+      "step": 10041
+    },
+    {
+      "epoch": 0.6471611780627698,
+      "grad_norm": 0.0014761180281003693,
+      "learning_rate": 3.925528105979234e-06,
+      "loss": 0.0,
+      "step": 10042
+    },
+    {
+      "epoch": 0.647225623509699,
+      "grad_norm": 0.0014376881504091783,
+      "learning_rate": 3.9248120300751885e-06,
+      "loss": 0.0,
+      "step": 10043
+    },
+    {
+      "epoch": 0.6472900689566282,
+      "grad_norm": 0.32301565801310034,
+      "learning_rate": 3.924095954171143e-06,
+      "loss": 0.0005,
+      "step": 10044
+    },
+    {
+      "epoch": 0.6473545144035574,
+      "grad_norm": 0.0014243909713996896,
+      "learning_rate": 3.923379878267097e-06,
+      "loss": 0.0,
+      "step": 10045
+    },
+    {
+      "epoch": 0.6474189598504866,
+      "grad_norm": 0.000623841882694804,
+      "learning_rate": 3.9226638023630505e-06,
+      "loss": 0.0,
+      "step": 10046
+    },
+    {
+      "epoch": 0.6474834052974158,
+      "grad_norm": 0.021735964259168947,
+      "learning_rate": 3.921947726459005e-06,
+      "loss": 0.0,
+      "step": 10047
+    },
+    {
+      "epoch": 0.647547850744345,
+      "grad_norm": 0.003005139298277695,
+      "learning_rate": 3.921231650554959e-06,
+      "loss": 0.0,
+      "step": 10048
+    },
+    {
+      "epoch": 0.6476122961912741,
+      "grad_norm": 0.009583776841016588,
+      "learning_rate": 3.920515574650913e-06,
+      "loss": 0.0,
+      "step": 10049
+    },
+    {
+      "epoch": 0.6476767416382032,
+      "grad_norm": 0.0008914135898324495,
+      "learning_rate": 3.919799498746868e-06,
+      "loss": 0.0,
+      "step": 10050
+    },
+    {
+      "epoch": 0.6477411870851324,
+      "grad_norm": 0.0013635385532760897,
+      "learning_rate": 3.919083422842821e-06,
+      "loss": 0.0,
+      "step": 10051
+    },
+    {
+      "epoch": 0.6478056325320616,
+      "grad_norm": 0.09818460111478787,
+      "learning_rate": 3.9183673469387755e-06,
+      "loss": 0.0012,
+      "step": 10052
+    },
+    {
+      "epoch": 0.6478700779789908,
+      "grad_norm": 0.0008677872964371634,
+      "learning_rate": 3.91765127103473e-06,
+      "loss": 0.0,
+      "step": 10053
+    },
+    {
+      "epoch": 0.64793452342592,
+      "grad_norm": 0.013500065410627189,
+      "learning_rate": 3.916935195130684e-06,
+      "loss": 0.0001,
+      "step": 10054
+    },
+    {
+      "epoch": 0.6479989688728491,
+      "grad_norm": 0.00023195192506884497,
+      "learning_rate": 3.916219119226638e-06,
+      "loss": 0.0,
+      "step": 10055
+    },
+    {
+      "epoch": 0.6480634143197783,
+      "grad_norm": 0.00956348066167725,
+      "learning_rate": 3.915503043322593e-06,
+      "loss": 0.0,
+      "step": 10056
+    },
+    {
+      "epoch": 0.6481278597667075,
+      "grad_norm": 0.0011233392884167606,
+      "learning_rate": 3.914786967418547e-06,
+      "loss": 0.0,
+      "step": 10057
+    },
+    {
+      "epoch": 0.6481923052136367,
+      "grad_norm": 0.00217687030993054,
+      "learning_rate": 3.914070891514501e-06,
+      "loss": 0.0,
+      "step": 10058
+    },
+    {
+      "epoch": 0.6482567506605659,
+      "grad_norm": 0.0007690180828168407,
+      "learning_rate": 3.913354815610455e-06,
+      "loss": 0.0,
+      "step": 10059
+    },
+    {
+      "epoch": 0.6483211961074951,
+      "grad_norm": 0.00018957142718229443,
+      "learning_rate": 3.912638739706409e-06,
+      "loss": 0.0,
+      "step": 10060
+    },
+    {
+      "epoch": 0.6483856415544241,
+      "grad_norm": 0.0033156219554045124,
+      "learning_rate": 3.911922663802363e-06,
+      "loss": 0.0,
+      "step": 10061
+    },
+    {
+      "epoch": 0.6484500870013533,
+      "grad_norm": 0.0193806849948235,
+      "learning_rate": 3.911206587898318e-06,
+      "loss": 0.0,
+      "step": 10062
+    },
+    {
+      "epoch": 0.6485145324482825,
+      "grad_norm": 0.0018474096180259869,
+      "learning_rate": 3.910490511994272e-06,
+      "loss": 0.0,
+      "step": 10063
+    },
+    {
+      "epoch": 0.6485789778952117,
+      "grad_norm": 0.0027109400656019204,
+      "learning_rate": 3.909774436090225e-06,
+      "loss": 0.0,
+      "step": 10064
+    },
+    {
+      "epoch": 0.6486434233421409,
+      "grad_norm": 0.17048972712294916,
+      "learning_rate": 3.90905836018618e-06,
+      "loss": 0.0036,
+      "step": 10065
+    },
+    {
+      "epoch": 0.64870786878907,
+      "grad_norm": 0.001639427937846142,
+      "learning_rate": 3.908342284282135e-06,
+      "loss": 0.0,
+      "step": 10066
+    },
+    {
+      "epoch": 0.6487723142359992,
+      "grad_norm": 0.0003054587604726002,
+      "learning_rate": 3.907626208378088e-06,
+      "loss": 0.0,
+      "step": 10067
+    },
+    {
+      "epoch": 0.6488367596829284,
+      "grad_norm": 0.0027535541277126873,
+      "learning_rate": 3.9069101324740426e-06,
+      "loss": 0.0,
+      "step": 10068
+    },
+    {
+      "epoch": 0.6489012051298576,
+      "grad_norm": 0.0006657856217193417,
+      "learning_rate": 3.906194056569997e-06,
+      "loss": 0.0,
+      "step": 10069
+    },
+    {
+      "epoch": 0.6489656505767868,
+      "grad_norm": 0.0060369187223777215,
+      "learning_rate": 3.905477980665951e-06,
+      "loss": 0.0,
+      "step": 10070
+    },
+    {
+      "epoch": 0.649030096023716,
+      "grad_norm": 0.0019431457633511454,
+      "learning_rate": 3.9047619047619055e-06,
+      "loss": 0.0,
+      "step": 10071
+    },
+    {
+      "epoch": 0.649094541470645,
+      "grad_norm": 0.3679831994007836,
+      "learning_rate": 3.904045828857859e-06,
+      "loss": 0.0011,
+      "step": 10072
+    },
+    {
+      "epoch": 0.6491589869175742,
+      "grad_norm": 0.007803615252833467,
+      "learning_rate": 3.903329752953813e-06,
+      "loss": 0.0001,
+      "step": 10073
+    },
+    {
+      "epoch": 0.6492234323645034,
+      "grad_norm": 0.11488098575653381,
+      "learning_rate": 3.9026136770497675e-06,
+      "loss": 0.0004,
+      "step": 10074
+    },
+    {
+      "epoch": 0.6492878778114326,
+      "grad_norm": 0.00033450819125250343,
+      "learning_rate": 3.901897601145722e-06,
+      "loss": 0.0,
+      "step": 10075
+    },
+    {
+      "epoch": 0.6493523232583618,
+      "grad_norm": 0.15049065941398837,
+      "learning_rate": 3.901181525241676e-06,
+      "loss": 0.0002,
+      "step": 10076
+    },
+    {
+      "epoch": 0.649416768705291,
+      "grad_norm": 0.001570508333555772,
+      "learning_rate": 3.90046544933763e-06,
+      "loss": 0.0,
+      "step": 10077
+    },
+    {
+      "epoch": 0.6494812141522202,
+      "grad_norm": 0.030068070105821104,
+      "learning_rate": 3.899749373433585e-06,
+      "loss": 0.0,
+      "step": 10078
+    },
+    {
+      "epoch": 0.6495456595991493,
+      "grad_norm": 0.07123143989605957,
+      "learning_rate": 3.899033297529539e-06,
+      "loss": 0.0001,
+      "step": 10079
+    },
+    {
+      "epoch": 0.6496101050460785,
+      "grad_norm": 0.007017937085895609,
+      "learning_rate": 3.8983172216254924e-06,
+      "loss": 0.0,
+      "step": 10080
+    },
+    {
+      "epoch": 0.6496745504930077,
+      "grad_norm": 0.0017217782930513232,
+      "learning_rate": 3.897601145721447e-06,
+      "loss": 0.0,
+      "step": 10081
+    },
+    {
+      "epoch": 0.6497389959399369,
+      "grad_norm": 0.46246827248698735,
+      "learning_rate": 3.896885069817401e-06,
+      "loss": 0.0006,
+      "step": 10082
+    },
+    {
+      "epoch": 0.649803441386866,
+      "grad_norm": 0.020012365780105686,
+      "learning_rate": 3.896168993913355e-06,
+      "loss": 0.0,
+      "step": 10083
+    },
+    {
+      "epoch": 0.6498678868337951,
+      "grad_norm": 0.0037748346140321957,
+      "learning_rate": 3.89545291800931e-06,
+      "loss": 0.0,
+      "step": 10084
+    },
+    {
+      "epoch": 0.6499323322807243,
+      "grad_norm": 7.76592282656403e-05,
+      "learning_rate": 3.894736842105263e-06,
+      "loss": 0.0,
+      "step": 10085
+    },
+    {
+      "epoch": 0.6499967777276535,
+      "grad_norm": 0.00016100952285215606,
+      "learning_rate": 3.894020766201217e-06,
+      "loss": 0.0,
+      "step": 10086
+    },
+    {
+      "epoch": 0.6500612231745827,
+      "grad_norm": 0.20804745112879044,
+      "learning_rate": 3.893304690297172e-06,
+      "loss": 0.0007,
+      "step": 10087
+    },
+    {
+      "epoch": 0.6501256686215119,
+      "grad_norm": 0.0005138267208881703,
+      "learning_rate": 3.892588614393126e-06,
+      "loss": 0.0,
+      "step": 10088
+    },
+    {
+      "epoch": 0.6501901140684411,
+      "grad_norm": 0.00140696667912771,
+      "learning_rate": 3.89187253848908e-06,
+      "loss": 0.0,
+      "step": 10089
+    },
+    {
+      "epoch": 0.6502545595153703,
+      "grad_norm": 0.0014816001929669846,
+      "learning_rate": 3.891156462585035e-06,
+      "loss": 0.0,
+      "step": 10090
+    },
+    {
+      "epoch": 0.6503190049622994,
+      "grad_norm": 0.011444484703948571,
+      "learning_rate": 3.890440386680989e-06,
+      "loss": 0.0,
+      "step": 10091
+    },
+    {
+      "epoch": 0.6503834504092286,
+      "grad_norm": 0.002593092375976146,
+      "learning_rate": 3.889724310776943e-06,
+      "loss": 0.0,
+      "step": 10092
+    },
+    {
+      "epoch": 0.6504478958561578,
+      "grad_norm": 0.0008823114322447084,
+      "learning_rate": 3.889008234872897e-06,
+      "loss": 0.0,
+      "step": 10093
+    },
+    {
+      "epoch": 0.6505123413030869,
+      "grad_norm": 0.45643913189772434,
+      "learning_rate": 3.888292158968851e-06,
+      "loss": 0.0005,
+      "step": 10094
+    },
+    {
+      "epoch": 0.6505767867500161,
+      "grad_norm": 0.001553883683300925,
+      "learning_rate": 3.887576083064805e-06,
+      "loss": 0.0,
+      "step": 10095
+    },
+    {
+      "epoch": 0.6506412321969453,
+      "grad_norm": 5.411795877500031e-05,
+      "learning_rate": 3.8868600071607595e-06,
+      "loss": 0.0,
+      "step": 10096
+    },
+    {
+      "epoch": 0.6507056776438744,
+      "grad_norm": 0.02975246910597393,
+      "learning_rate": 3.886143931256714e-06,
+      "loss": 0.0003,
+      "step": 10097
+    },
+    {
+      "epoch": 0.6507701230908036,
+      "grad_norm": 0.0008351681676013702,
+      "learning_rate": 3.885427855352667e-06,
+      "loss": 0.0,
+      "step": 10098
+    },
+    {
+      "epoch": 0.6508345685377328,
+      "grad_norm": 0.3116308164738484,
+      "learning_rate": 3.884711779448622e-06,
+      "loss": 0.0023,
+      "step": 10099
+    },
+    {
+      "epoch": 0.650899013984662,
+      "grad_norm": 0.003933262974662413,
+      "learning_rate": 3.883995703544576e-06,
+      "loss": 0.0,
+      "step": 10100
+    },
+    {
+      "epoch": 0.6509634594315912,
+      "grad_norm": 0.07911621652664459,
+      "learning_rate": 3.88327962764053e-06,
+      "loss": 0.0001,
+      "step": 10101
+    },
+    {
+      "epoch": 0.6510279048785204,
+      "grad_norm": 0.0004710866212675768,
+      "learning_rate": 3.8825635517364845e-06,
+      "loss": 0.0,
+      "step": 10102
+    },
+    {
+      "epoch": 0.6510923503254495,
+      "grad_norm": 0.0062199568895057135,
+      "learning_rate": 3.881847475832439e-06,
+      "loss": 0.0,
+      "step": 10103
+    },
+    {
+      "epoch": 0.6511567957723787,
+      "grad_norm": 0.47167855019758076,
+      "learning_rate": 3.881131399928393e-06,
+      "loss": 0.0011,
+      "step": 10104
+    },
+    {
+      "epoch": 0.6512212412193079,
+      "grad_norm": 0.32272571896552665,
+      "learning_rate": 3.880415324024347e-06,
+      "loss": 0.0016,
+      "step": 10105
+    },
+    {
+      "epoch": 0.651285686666237,
+      "grad_norm": 0.07696105436920875,
+      "learning_rate": 3.879699248120301e-06,
+      "loss": 0.0001,
+      "step": 10106
+    },
+    {
+      "epoch": 0.6513501321131662,
+      "grad_norm": 0.0015881369909475057,
+      "learning_rate": 3.878983172216255e-06,
+      "loss": 0.0,
+      "step": 10107
+    },
+    {
+      "epoch": 0.6514145775600954,
+      "grad_norm": 0.00023232975344055165,
+      "learning_rate": 3.878267096312209e-06,
+      "loss": 0.0,
+      "step": 10108
+    },
+    {
+      "epoch": 0.6514790230070245,
+      "grad_norm": 0.0025092237827565697,
+      "learning_rate": 3.877551020408164e-06,
+      "loss": 0.0,
+      "step": 10109
+    },
+    {
+      "epoch": 0.6515434684539537,
+      "grad_norm": 0.000561346000613679,
+      "learning_rate": 3.876834944504117e-06,
+      "loss": 0.0,
+      "step": 10110
+    },
+    {
+      "epoch": 0.6516079139008829,
+      "grad_norm": 0.08851294074908397,
+      "learning_rate": 3.8761188686000715e-06,
+      "loss": 0.0008,
+      "step": 10111
+    },
+    {
+      "epoch": 0.6516723593478121,
+      "grad_norm": 9.094776729926713e-05,
+      "learning_rate": 3.875402792696027e-06,
+      "loss": 0.0,
+      "step": 10112
+    },
+    {
+      "epoch": 0.6517368047947413,
+      "grad_norm": 0.009651514658336328,
+      "learning_rate": 3.874686716791981e-06,
+      "loss": 0.0001,
+      "step": 10113
+    },
+    {
+      "epoch": 0.6518012502416705,
+      "grad_norm": 8.080717123989275e-05,
+      "learning_rate": 3.873970640887934e-06,
+      "loss": 0.0,
+      "step": 10114
+    },
+    {
+      "epoch": 0.6518656956885996,
+      "grad_norm": 0.005040657665753197,
+      "learning_rate": 3.873254564983889e-06,
+      "loss": 0.0,
+      "step": 10115
+    },
+    {
+      "epoch": 0.6519301411355288,
+      "grad_norm": 3.942586078710562e-05,
+      "learning_rate": 3.872538489079843e-06,
+      "loss": 0.0,
+      "step": 10116
+    },
+    {
+      "epoch": 0.6519945865824579,
+      "grad_norm": 0.0007002757109208436,
+      "learning_rate": 3.871822413175797e-06,
+      "loss": 0.0,
+      "step": 10117
+    },
+    {
+      "epoch": 0.6520590320293871,
+      "grad_norm": 0.0019366055380994323,
+      "learning_rate": 3.871106337271751e-06,
+      "loss": 0.0,
+      "step": 10118
+    },
+    {
+      "epoch": 0.6521234774763163,
+      "grad_norm": 0.0035408849877017687,
+      "learning_rate": 3.870390261367705e-06,
+      "loss": 0.0,
+      "step": 10119
+    },
+    {
+      "epoch": 0.6521879229232455,
+      "grad_norm": 0.6270142760119335,
+      "learning_rate": 3.869674185463659e-06,
+      "loss": 0.0046,
+      "step": 10120
+    },
+    {
+      "epoch": 0.6522523683701746,
+      "grad_norm": 0.00039780686488753573,
+      "learning_rate": 3.868958109559614e-06,
+      "loss": 0.0,
+      "step": 10121
+    },
+    {
+      "epoch": 0.6523168138171038,
+      "grad_norm": 0.003640746475764863,
+      "learning_rate": 3.868242033655568e-06,
+      "loss": 0.0,
+      "step": 10122
+    },
+    {
+      "epoch": 0.652381259264033,
+      "grad_norm": 0.4130364426705445,
+      "learning_rate": 3.867525957751521e-06,
+      "loss": 0.0016,
+      "step": 10123
+    },
+    {
+      "epoch": 0.6524457047109622,
+      "grad_norm": 0.059154666829826646,
+      "learning_rate": 3.8668098818474765e-06,
+      "loss": 0.0004,
+      "step": 10124
+    },
+    {
+      "epoch": 0.6525101501578914,
+      "grad_norm": 0.0013061928478204214,
+      "learning_rate": 3.866093805943431e-06,
+      "loss": 0.0,
+      "step": 10125
+    },
+    {
+      "epoch": 0.6525745956048206,
+      "grad_norm": 0.28045806386242605,
+      "learning_rate": 3.865377730039384e-06,
+      "loss": 0.0004,
+      "step": 10126
+    },
+    {
+      "epoch": 0.6526390410517497,
+      "grad_norm": 0.01899712727240623,
+      "learning_rate": 3.8646616541353386e-06,
+      "loss": 0.0001,
+      "step": 10127
+    },
+    {
+      "epoch": 0.6527034864986788,
+      "grad_norm": 0.051157784159917466,
+      "learning_rate": 3.863945578231293e-06,
+      "loss": 0.0001,
+      "step": 10128
+    },
+    {
+      "epoch": 0.652767931945608,
+      "grad_norm": 0.025598438954331462,
+      "learning_rate": 3.863229502327247e-06,
+      "loss": 0.0001,
+      "step": 10129
+    },
+    {
+      "epoch": 0.6528323773925372,
+      "grad_norm": 0.0016328159009993447,
+      "learning_rate": 3.8625134264232015e-06,
+      "loss": 0.0,
+      "step": 10130
+    },
+    {
+      "epoch": 0.6528968228394664,
+      "grad_norm": 2.6339330807109222,
+      "learning_rate": 3.861797350519155e-06,
+      "loss": 0.0112,
+      "step": 10131
+    },
+    {
+      "epoch": 0.6529612682863956,
+      "grad_norm": 0.05387536153884999,
+      "learning_rate": 3.861081274615109e-06,
+      "loss": 0.0007,
+      "step": 10132
+    },
+    {
+      "epoch": 0.6530257137333247,
+      "grad_norm": 0.4103096084899589,
+      "learning_rate": 3.8603651987110635e-06,
+      "loss": 0.002,
+      "step": 10133
+    },
+    {
+      "epoch": 0.6530901591802539,
+      "grad_norm": 0.0023990892949300193,
+      "learning_rate": 3.859649122807018e-06,
+      "loss": 0.0,
+      "step": 10134
+    },
+    {
+      "epoch": 0.6531546046271831,
+      "grad_norm": 0.0025769823262489213,
+      "learning_rate": 3.858933046902972e-06,
+      "loss": 0.0,
+      "step": 10135
+    },
+    {
+      "epoch": 0.6532190500741123,
+      "grad_norm": 0.008294945393946193,
+      "learning_rate": 3.858216970998926e-06,
+      "loss": 0.0,
+      "step": 10136
+    },
+    {
+      "epoch": 0.6532834955210415,
+      "grad_norm": 0.0038449482692927794,
+      "learning_rate": 3.857500895094881e-06,
+      "loss": 0.0,
+      "step": 10137
+    },
+    {
+      "epoch": 0.6533479409679707,
+      "grad_norm": 0.016722334512031662,
+      "learning_rate": 3.856784819190835e-06,
+      "loss": 0.0002,
+      "step": 10138
+    },
+    {
+      "epoch": 0.6534123864148997,
+      "grad_norm": 0.45878159933893453,
+      "learning_rate": 3.8560687432867884e-06,
+      "loss": 0.0022,
+      "step": 10139
+    },
+    {
+      "epoch": 0.6534768318618289,
+      "grad_norm": 0.003303008831653256,
+      "learning_rate": 3.855352667382743e-06,
+      "loss": 0.0,
+      "step": 10140
+    },
+    {
+      "epoch": 0.6535412773087581,
+      "grad_norm": 0.01095606656160586,
+      "learning_rate": 3.854636591478697e-06,
+      "loss": 0.0,
+      "step": 10141
+    },
+    {
+      "epoch": 0.6536057227556873,
+      "grad_norm": 0.0007942663676417572,
+      "learning_rate": 3.853920515574651e-06,
+      "loss": 0.0,
+      "step": 10142
+    },
+    {
+      "epoch": 0.6536701682026165,
+      "grad_norm": 0.011422861115300881,
+      "learning_rate": 3.853204439670606e-06,
+      "loss": 0.0,
+      "step": 10143
+    },
+    {
+      "epoch": 0.6537346136495457,
+      "grad_norm": 0.13214908426043642,
+      "learning_rate": 3.852488363766559e-06,
+      "loss": 0.0003,
+      "step": 10144
+    },
+    {
+      "epoch": 0.6537990590964748,
+      "grad_norm": 0.00021781995305225694,
+      "learning_rate": 3.851772287862513e-06,
+      "loss": 0.0,
+      "step": 10145
+    },
+    {
+      "epoch": 0.653863504543404,
+      "grad_norm": 0.002687471652329774,
+      "learning_rate": 3.851056211958468e-06,
+      "loss": 0.0,
+      "step": 10146
+    },
+    {
+      "epoch": 0.6539279499903332,
+      "grad_norm": 0.0005095891078883199,
+      "learning_rate": 3.850340136054422e-06,
+      "loss": 0.0,
+      "step": 10147
+    },
+    {
+      "epoch": 0.6539923954372624,
+      "grad_norm": 0.00152943308971141,
+      "learning_rate": 3.849624060150376e-06,
+      "loss": 0.0,
+      "step": 10148
+    },
+    {
+      "epoch": 0.6540568408841916,
+      "grad_norm": 0.0010116890607374909,
+      "learning_rate": 3.848907984246331e-06,
+      "loss": 0.0,
+      "step": 10149
+    },
+    {
+      "epoch": 0.6541212863311207,
+      "grad_norm": 0.008247315056338593,
+      "learning_rate": 3.848191908342285e-06,
+      "loss": 0.0,
+      "step": 10150
+    },
+    {
+      "epoch": 0.6541857317780498,
+      "grad_norm": 0.03783836195338165,
+      "learning_rate": 3.847475832438239e-06,
+      "loss": 0.0001,
+      "step": 10151
+    },
+    {
+      "epoch": 0.654250177224979,
+      "grad_norm": 0.03685747469146827,
+      "learning_rate": 3.846759756534193e-06,
+      "loss": 0.0,
+      "step": 10152
+    },
+    {
+      "epoch": 0.6543146226719082,
+      "grad_norm": 0.01833364381008545,
+      "learning_rate": 3.846043680630147e-06,
+      "loss": 0.0,
+      "step": 10153
+    },
+    {
+      "epoch": 0.6543790681188374,
+      "grad_norm": 0.5712872105755907,
+      "learning_rate": 3.845327604726101e-06,
+      "loss": 0.0033,
+      "step": 10154
+    },
+    {
+      "epoch": 0.6544435135657666,
+      "grad_norm": 0.004128894377295956,
+      "learning_rate": 3.8446115288220555e-06,
+      "loss": 0.0,
+      "step": 10155
+    },
+    {
+      "epoch": 0.6545079590126958,
+      "grad_norm": 0.026285459872411314,
+      "learning_rate": 3.84389545291801e-06,
+      "loss": 0.0,
+      "step": 10156
+    },
+    {
+      "epoch": 0.654572404459625,
+      "grad_norm": 0.010629958355985208,
+      "learning_rate": 3.843179377013963e-06,
+      "loss": 0.0001,
+      "step": 10157
+    },
+    {
+      "epoch": 0.6546368499065541,
+      "grad_norm": 0.0003458473109674389,
+      "learning_rate": 3.842463301109918e-06,
+      "loss": 0.0,
+      "step": 10158
+    },
+    {
+      "epoch": 0.6547012953534833,
+      "grad_norm": 0.019025562351784445,
+      "learning_rate": 3.841747225205873e-06,
+      "loss": 0.0,
+      "step": 10159
+    },
+    {
+      "epoch": 0.6547657408004125,
+      "grad_norm": 0.0031882815792937374,
+      "learning_rate": 3.841031149301826e-06,
+      "loss": 0.0,
+      "step": 10160
+    },
+    {
+      "epoch": 0.6548301862473416,
+      "grad_norm": 0.01813268104120978,
+      "learning_rate": 3.8403150733977805e-06,
+      "loss": 0.0,
+      "step": 10161
+    },
+    {
+      "epoch": 0.6548946316942708,
+      "grad_norm": 0.02995992804919363,
+      "learning_rate": 3.839598997493735e-06,
+      "loss": 0.0,
+      "step": 10162
+    },
+    {
+      "epoch": 0.6549590771411999,
+      "grad_norm": 0.006443501348050104,
+      "learning_rate": 3.838882921589689e-06,
+      "loss": 0.0,
+      "step": 10163
+    },
+    {
+      "epoch": 0.6550235225881291,
+      "grad_norm": 0.00042109950495271373,
+      "learning_rate": 3.838166845685643e-06,
+      "loss": 0.0,
+      "step": 10164
+    },
+    {
+      "epoch": 0.6550879680350583,
+      "grad_norm": 0.0008626208900784901,
+      "learning_rate": 3.837450769781597e-06,
+      "loss": 0.0,
+      "step": 10165
+    },
+    {
+      "epoch": 0.6551524134819875,
+      "grad_norm": 0.003391952493412297,
+      "learning_rate": 3.836734693877551e-06,
+      "loss": 0.0,
+      "step": 10166
+    },
+    {
+      "epoch": 0.6552168589289167,
+      "grad_norm": 0.005285101991014211,
+      "learning_rate": 3.836018617973505e-06,
+      "loss": 0.0,
+      "step": 10167
+    },
+    {
+      "epoch": 0.6552813043758459,
+      "grad_norm": 0.01946311256116589,
+      "learning_rate": 3.83530254206946e-06,
+      "loss": 0.0,
+      "step": 10168
+    },
+    {
+      "epoch": 0.655345749822775,
+      "grad_norm": 0.0026740134172567116,
+      "learning_rate": 3.834586466165414e-06,
+      "loss": 0.0,
+      "step": 10169
+    },
+    {
+      "epoch": 0.6554101952697042,
+      "grad_norm": 0.0033936852631130097,
+      "learning_rate": 3.8338703902613675e-06,
+      "loss": 0.0,
+      "step": 10170
+    },
+    {
+      "epoch": 0.6554746407166334,
+      "grad_norm": 0.0009185211832299836,
+      "learning_rate": 3.833154314357323e-06,
+      "loss": 0.0,
+      "step": 10171
+    },
+    {
+      "epoch": 0.6555390861635625,
+      "grad_norm": 0.06459203475290538,
+      "learning_rate": 3.832438238453277e-06,
+      "loss": 0.0005,
+      "step": 10172
+    },
+    {
+      "epoch": 0.6556035316104917,
+      "grad_norm": 0.0026117542544247074,
+      "learning_rate": 3.83172216254923e-06,
+      "loss": 0.0,
+      "step": 10173
+    },
+    {
+      "epoch": 0.6556679770574209,
+      "grad_norm": 0.1658535446637217,
+      "learning_rate": 3.831006086645185e-06,
+      "loss": 0.0001,
+      "step": 10174
+    },
+    {
+      "epoch": 0.65573242250435,
+      "grad_norm": 0.12451728359577288,
+      "learning_rate": 3.830290010741139e-06,
+      "loss": 0.002,
+      "step": 10175
+    },
+    {
+      "epoch": 0.6557968679512792,
+      "grad_norm": 0.30647530928600114,
+      "learning_rate": 3.829573934837093e-06,
+      "loss": 0.0006,
+      "step": 10176
+    },
+    {
+      "epoch": 0.6558613133982084,
+      "grad_norm": 0.05586064285560092,
+      "learning_rate": 3.8288578589330476e-06,
+      "loss": 0.0002,
+      "step": 10177
+    },
+    {
+      "epoch": 0.6559257588451376,
+      "grad_norm": 0.0025995801811718247,
+      "learning_rate": 3.828141783029001e-06,
+      "loss": 0.0,
+      "step": 10178
+    },
+    {
+      "epoch": 0.6559902042920668,
+      "grad_norm": 0.0038249163839321164,
+      "learning_rate": 3.827425707124955e-06,
+      "loss": 0.0,
+      "step": 10179
+    },
+    {
+      "epoch": 0.656054649738996,
+      "grad_norm": 0.4108950183217354,
+      "learning_rate": 3.82670963122091e-06,
+      "loss": 0.0029,
+      "step": 10180
+    },
+    {
+      "epoch": 0.6561190951859251,
+      "grad_norm": 0.004374683406390487,
+      "learning_rate": 3.825993555316864e-06,
+      "loss": 0.0,
+      "step": 10181
+    },
+    {
+      "epoch": 0.6561835406328543,
+      "grad_norm": 0.0014969097910061068,
+      "learning_rate": 3.825277479412818e-06,
+      "loss": 0.0,
+      "step": 10182
+    },
+    {
+      "epoch": 0.6562479860797835,
+      "grad_norm": 0.0012723266838122523,
+      "learning_rate": 3.8245614035087725e-06,
+      "loss": 0.0,
+      "step": 10183
+    },
+    {
+      "epoch": 0.6563124315267126,
+      "grad_norm": 0.008525120834268957,
+      "learning_rate": 3.823845327604727e-06,
+      "loss": 0.0,
+      "step": 10184
+    },
+    {
+      "epoch": 0.6563768769736418,
+      "grad_norm": 0.0030668060505652244,
+      "learning_rate": 3.823129251700681e-06,
+      "loss": 0.0,
+      "step": 10185
+    },
+    {
+      "epoch": 0.656441322420571,
+      "grad_norm": 0.00534143900713158,
+      "learning_rate": 3.8224131757966346e-06,
+      "loss": 0.0,
+      "step": 10186
+    },
+    {
+      "epoch": 0.6565057678675001,
+      "grad_norm": 0.0037717083756212658,
+      "learning_rate": 3.821697099892589e-06,
+      "loss": 0.0,
+      "step": 10187
+    },
+    {
+      "epoch": 0.6565702133144293,
+      "grad_norm": 0.0012179376897736268,
+      "learning_rate": 3.820981023988543e-06,
+      "loss": 0.0,
+      "step": 10188
+    },
+    {
+      "epoch": 0.6566346587613585,
+      "grad_norm": 0.00037249200636945535,
+      "learning_rate": 3.8202649480844975e-06,
+      "loss": 0.0,
+      "step": 10189
+    },
+    {
+      "epoch": 0.6566991042082877,
+      "grad_norm": 0.017734283475125033,
+      "learning_rate": 3.819548872180452e-06,
+      "loss": 0.0002,
+      "step": 10190
+    },
+    {
+      "epoch": 0.6567635496552169,
+      "grad_norm": 0.020177854077664724,
+      "learning_rate": 3.818832796276405e-06,
+      "loss": 0.0,
+      "step": 10191
+    },
+    {
+      "epoch": 0.6568279951021461,
+      "grad_norm": 0.007316823046694329,
+      "learning_rate": 3.8181167203723595e-06,
+      "loss": 0.0,
+      "step": 10192
+    },
+    {
+      "epoch": 0.6568924405490753,
+      "grad_norm": 0.0005207608387712419,
+      "learning_rate": 3.817400644468314e-06,
+      "loss": 0.0,
+      "step": 10193
+    },
+    {
+      "epoch": 0.6569568859960044,
+      "grad_norm": 0.0009445133087726743,
+      "learning_rate": 3.816684568564268e-06,
+      "loss": 0.0,
+      "step": 10194
+    },
+    {
+      "epoch": 0.6570213314429335,
+      "grad_norm": 0.02219222355146514,
+      "learning_rate": 3.815968492660222e-06,
+      "loss": 0.0,
+      "step": 10195
+    },
+    {
+      "epoch": 0.6570857768898627,
+      "grad_norm": 0.0012817471509581164,
+      "learning_rate": 3.815252416756177e-06,
+      "loss": 0.0,
+      "step": 10196
+    },
+    {
+      "epoch": 0.6571502223367919,
+      "grad_norm": 0.010393691367853066,
+      "learning_rate": 3.8145363408521306e-06,
+      "loss": 0.0,
+      "step": 10197
+    },
+    {
+      "epoch": 0.6572146677837211,
+      "grad_norm": 0.016148501614400432,
+      "learning_rate": 3.813820264948085e-06,
+      "loss": 0.0,
+      "step": 10198
+    },
+    {
+      "epoch": 0.6572791132306502,
+      "grad_norm": 0.028812178897839727,
+      "learning_rate": 3.813104189044039e-06,
+      "loss": 0.0,
+      "step": 10199
+    },
+    {
+      "epoch": 0.6573435586775794,
+      "grad_norm": 0.0002261249635128685,
+      "learning_rate": 3.812388113139993e-06,
+      "loss": 0.0,
+      "step": 10200
+    },
+    {
+      "epoch": 0.6574080041245086,
+      "grad_norm": 0.001779168386697235,
+      "learning_rate": 3.8116720372359473e-06,
+      "loss": 0.0,
+      "step": 10201
+    },
+    {
+      "epoch": 0.6574724495714378,
+      "grad_norm": 0.0011019397221192367,
+      "learning_rate": 3.8109559613319012e-06,
+      "loss": 0.0,
+      "step": 10202
+    },
+    {
+      "epoch": 0.657536895018367,
+      "grad_norm": 0.00014219508472730657,
+      "learning_rate": 3.8102398854278555e-06,
+      "loss": 0.0,
+      "step": 10203
+    },
+    {
+      "epoch": 0.6576013404652962,
+      "grad_norm": 0.0265215081898783,
+      "learning_rate": 3.80952380952381e-06,
+      "loss": 0.0002,
+      "step": 10204
+    },
+    {
+      "epoch": 0.6576657859122254,
+      "grad_norm": 0.03401318937206382,
+      "learning_rate": 3.8088077336197637e-06,
+      "loss": 0.0002,
+      "step": 10205
+    },
+    {
+      "epoch": 0.6577302313591544,
+      "grad_norm": 0.07009893009647121,
+      "learning_rate": 3.8080916577157184e-06,
+      "loss": 0.0,
+      "step": 10206
+    },
+    {
+      "epoch": 0.6577946768060836,
+      "grad_norm": 0.008663384096209157,
+      "learning_rate": 3.8073755818116727e-06,
+      "loss": 0.0,
+      "step": 10207
+    },
+    {
+      "epoch": 0.6578591222530128,
+      "grad_norm": 9.288414808061254e-05,
+      "learning_rate": 3.8066595059076266e-06,
+      "loss": 0.0,
+      "step": 10208
+    },
+    {
+      "epoch": 0.657923567699942,
+      "grad_norm": 0.007428766665651455,
+      "learning_rate": 3.805943430003581e-06,
+      "loss": 0.0,
+      "step": 10209
+    },
+    {
+      "epoch": 0.6579880131468712,
+      "grad_norm": 0.026942313183641934,
+      "learning_rate": 3.8052273540995348e-06,
+      "loss": 0.0,
+      "step": 10210
+    },
+    {
+      "epoch": 0.6580524585938003,
+      "grad_norm": 0.0005140052428280983,
+      "learning_rate": 3.804511278195489e-06,
+      "loss": 0.0,
+      "step": 10211
+    },
+    {
+      "epoch": 0.6581169040407295,
+      "grad_norm": 0.07336983683314319,
+      "learning_rate": 3.8037952022914434e-06,
+      "loss": 0.0001,
+      "step": 10212
+    },
+    {
+      "epoch": 0.6581813494876587,
+      "grad_norm": 0.03837160424420366,
+      "learning_rate": 3.8030791263873972e-06,
+      "loss": 0.0001,
+      "step": 10213
+    },
+    {
+      "epoch": 0.6582457949345879,
+      "grad_norm": 0.03194277067175515,
+      "learning_rate": 3.8023630504833515e-06,
+      "loss": 0.0,
+      "step": 10214
+    },
+    {
+      "epoch": 0.6583102403815171,
+      "grad_norm": 0.008725165434776942,
+      "learning_rate": 3.8016469745793054e-06,
+      "loss": 0.0,
+      "step": 10215
+    },
+    {
+      "epoch": 0.6583746858284463,
+      "grad_norm": 0.011985939675556128,
+      "learning_rate": 3.8009308986752597e-06,
+      "loss": 0.0,
+      "step": 10216
+    },
+    {
+      "epoch": 0.6584391312753753,
+      "grad_norm": 0.09138096272130088,
+      "learning_rate": 3.8002148227712136e-06,
+      "loss": 0.0001,
+      "step": 10217
+    },
+    {
+      "epoch": 0.6585035767223045,
+      "grad_norm": 0.1891860258878625,
+      "learning_rate": 3.7994987468671683e-06,
+      "loss": 0.0002,
+      "step": 10218
+    },
+    {
+      "epoch": 0.6585680221692337,
+      "grad_norm": 0.08866577064407546,
+      "learning_rate": 3.7987826709631226e-06,
+      "loss": 0.0001,
+      "step": 10219
+    },
+    {
+      "epoch": 0.6586324676161629,
+      "grad_norm": 0.006611086351549965,
+      "learning_rate": 3.798066595059077e-06,
+      "loss": 0.0,
+      "step": 10220
+    },
+    {
+      "epoch": 0.6586969130630921,
+      "grad_norm": 0.00027910876529814774,
+      "learning_rate": 3.7973505191550308e-06,
+      "loss": 0.0,
+      "step": 10221
+    },
+    {
+      "epoch": 0.6587613585100213,
+      "grad_norm": 0.007999838552962334,
+      "learning_rate": 3.796634443250985e-06,
+      "loss": 0.0001,
+      "step": 10222
+    },
+    {
+      "epoch": 0.6588258039569505,
+      "grad_norm": 0.2032794548559423,
+      "learning_rate": 3.795918367346939e-06,
+      "loss": 0.0005,
+      "step": 10223
+    },
+    {
+      "epoch": 0.6588902494038796,
+      "grad_norm": 0.0003595132206011157,
+      "learning_rate": 3.7952022914428932e-06,
+      "loss": 0.0,
+      "step": 10224
+    },
+    {
+      "epoch": 0.6589546948508088,
+      "grad_norm": 0.0008375620930174066,
+      "learning_rate": 3.794486215538847e-06,
+      "loss": 0.0,
+      "step": 10225
+    },
+    {
+      "epoch": 0.659019140297738,
+      "grad_norm": 0.020827546998719356,
+      "learning_rate": 3.7937701396348014e-06,
+      "loss": 0.0,
+      "step": 10226
+    },
+    {
+      "epoch": 0.6590835857446672,
+      "grad_norm": 0.03600446840366598,
+      "learning_rate": 3.7930540637307557e-06,
+      "loss": 0.0001,
+      "step": 10227
+    },
+    {
+      "epoch": 0.6591480311915963,
+      "grad_norm": 0.004241069782651026,
+      "learning_rate": 3.7923379878267096e-06,
+      "loss": 0.0,
+      "step": 10228
+    },
+    {
+      "epoch": 0.6592124766385254,
+      "grad_norm": 0.019451562174052747,
+      "learning_rate": 3.7916219119226643e-06,
+      "loss": 0.0001,
+      "step": 10229
+    },
+    {
+      "epoch": 0.6592769220854546,
+      "grad_norm": 0.002471784102794206,
+      "learning_rate": 3.7909058360186186e-06,
+      "loss": 0.0,
+      "step": 10230
+    },
+    {
+      "epoch": 0.6593413675323838,
+      "grad_norm": 0.006852716105307031,
+      "learning_rate": 3.7901897601145725e-06,
+      "loss": 0.0,
+      "step": 10231
+    },
+    {
+      "epoch": 0.659405812979313,
+      "grad_norm": 0.06718536487225223,
+      "learning_rate": 3.789473684210527e-06,
+      "loss": 0.0003,
+      "step": 10232
+    },
+    {
+      "epoch": 0.6594702584262422,
+      "grad_norm": 0.15153008586304992,
+      "learning_rate": 3.7887576083064807e-06,
+      "loss": 0.0007,
+      "step": 10233
+    },
+    {
+      "epoch": 0.6595347038731714,
+      "grad_norm": 0.10961949745578928,
+      "learning_rate": 3.788041532402435e-06,
+      "loss": 0.0001,
+      "step": 10234
+    },
+    {
+      "epoch": 0.6595991493201006,
+      "grad_norm": 0.00020195634307442343,
+      "learning_rate": 3.7873254564983893e-06,
+      "loss": 0.0,
+      "step": 10235
+    },
+    {
+      "epoch": 0.6596635947670297,
+      "grad_norm": 0.008753787719257302,
+      "learning_rate": 3.786609380594343e-06,
+      "loss": 0.0001,
+      "step": 10236
+    },
+    {
+      "epoch": 0.6597280402139589,
+      "grad_norm": 0.00017843003587698393,
+      "learning_rate": 3.7858933046902974e-06,
+      "loss": 0.0,
+      "step": 10237
+    },
+    {
+      "epoch": 0.6597924856608881,
+      "grad_norm": 0.003163697938415987,
+      "learning_rate": 3.7851772287862513e-06,
+      "loss": 0.0,
+      "step": 10238
+    },
+    {
+      "epoch": 0.6598569311078172,
+      "grad_norm": 0.0016409395954385996,
+      "learning_rate": 3.7844611528822056e-06,
+      "loss": 0.0,
+      "step": 10239
+    },
+    {
+      "epoch": 0.6599213765547464,
+      "grad_norm": 0.0016194677414049581,
+      "learning_rate": 3.78374507697816e-06,
+      "loss": 0.0,
+      "step": 10240
+    },
+    {
+      "epoch": 0.6599858220016755,
+      "grad_norm": 0.2341164513889306,
+      "learning_rate": 3.783029001074114e-06,
+      "loss": 0.002,
+      "step": 10241
+    },
+    {
+      "epoch": 0.6600502674486047,
+      "grad_norm": 0.00014059234136107755,
+      "learning_rate": 3.7823129251700685e-06,
+      "loss": 0.0,
+      "step": 10242
+    },
+    {
+      "epoch": 0.6601147128955339,
+      "grad_norm": 0.012938931855012437,
+      "learning_rate": 3.781596849266023e-06,
+      "loss": 0.0001,
+      "step": 10243
+    },
+    {
+      "epoch": 0.6601791583424631,
+      "grad_norm": 0.00014948623886025877,
+      "learning_rate": 3.7808807733619767e-06,
+      "loss": 0.0,
+      "step": 10244
+    },
+    {
+      "epoch": 0.6602436037893923,
+      "grad_norm": 0.0002305776851317747,
+      "learning_rate": 3.780164697457931e-06,
+      "loss": 0.0,
+      "step": 10245
+    },
+    {
+      "epoch": 0.6603080492363215,
+      "grad_norm": 0.0003601501449476667,
+      "learning_rate": 3.779448621553885e-06,
+      "loss": 0.0,
+      "step": 10246
+    },
+    {
+      "epoch": 0.6603724946832507,
+      "grad_norm": 0.006198069005480283,
+      "learning_rate": 3.778732545649839e-06,
+      "loss": 0.0,
+      "step": 10247
+    },
+    {
+      "epoch": 0.6604369401301798,
+      "grad_norm": 0.003078498213774524,
+      "learning_rate": 3.7780164697457935e-06,
+      "loss": 0.0,
+      "step": 10248
+    },
+    {
+      "epoch": 0.660501385577109,
+      "grad_norm": 0.0014853955072896012,
+      "learning_rate": 3.7773003938417473e-06,
+      "loss": 0.0,
+      "step": 10249
+    },
+    {
+      "epoch": 0.6605658310240381,
+      "grad_norm": 0.0002482578605181744,
+      "learning_rate": 3.7765843179377016e-06,
+      "loss": 0.0,
+      "step": 10250
+    },
+    {
+      "epoch": 0.6606302764709673,
+      "grad_norm": 0.8442832180262457,
+      "learning_rate": 3.7758682420336555e-06,
+      "loss": 0.0057,
+      "step": 10251
+    },
+    {
+      "epoch": 0.6606947219178965,
+      "grad_norm": 9.143114696421962e-06,
+      "learning_rate": 3.77515216612961e-06,
+      "loss": 0.0,
+      "step": 10252
+    },
+    {
+      "epoch": 0.6607591673648257,
+      "grad_norm": 0.05428738982897947,
+      "learning_rate": 3.7744360902255645e-06,
+      "loss": 0.0001,
+      "step": 10253
+    },
+    {
+      "epoch": 0.6608236128117548,
+      "grad_norm": 0.0005961785885248,
+      "learning_rate": 3.7737200143215184e-06,
+      "loss": 0.0,
+      "step": 10254
+    },
+    {
+      "epoch": 0.660888058258684,
+      "grad_norm": 0.003677904195814865,
+      "learning_rate": 3.7730039384174727e-06,
+      "loss": 0.0,
+      "step": 10255
+    },
+    {
+      "epoch": 0.6609525037056132,
+      "grad_norm": 0.03861703337575527,
+      "learning_rate": 3.772287862513427e-06,
+      "loss": 0.0001,
+      "step": 10256
+    },
+    {
+      "epoch": 0.6610169491525424,
+      "grad_norm": 0.000588057262521846,
+      "learning_rate": 3.771571786609381e-06,
+      "loss": 0.0,
+      "step": 10257
+    },
+    {
+      "epoch": 0.6610813945994716,
+      "grad_norm": 0.0016087935946511298,
+      "learning_rate": 3.770855710705335e-06,
+      "loss": 0.0,
+      "step": 10258
+    },
+    {
+      "epoch": 0.6611458400464008,
+      "grad_norm": 0.0012067817277527145,
+      "learning_rate": 3.770139634801289e-06,
+      "loss": 0.0,
+      "step": 10259
+    },
+    {
+      "epoch": 0.6612102854933299,
+      "grad_norm": 0.00023797434159941603,
+      "learning_rate": 3.7694235588972433e-06,
+      "loss": 0.0,
+      "step": 10260
+    },
+    {
+      "epoch": 0.6612747309402591,
+      "grad_norm": 0.0037044470691454294,
+      "learning_rate": 3.7687074829931976e-06,
+      "loss": 0.0,
+      "step": 10261
+    },
+    {
+      "epoch": 0.6613391763871882,
+      "grad_norm": 8.325739223354846e-05,
+      "learning_rate": 3.7679914070891515e-06,
+      "loss": 0.0,
+      "step": 10262
+    },
+    {
+      "epoch": 0.6614036218341174,
+      "grad_norm": 0.45104876403275584,
+      "learning_rate": 3.767275331185106e-06,
+      "loss": 0.0009,
+      "step": 10263
+    },
+    {
+      "epoch": 0.6614680672810466,
+      "grad_norm": 0.03306756149105825,
+      "learning_rate": 3.7665592552810605e-06,
+      "loss": 0.0002,
+      "step": 10264
+    },
+    {
+      "epoch": 0.6615325127279758,
+      "grad_norm": 0.0009416338385984487,
+      "learning_rate": 3.7658431793770144e-06,
+      "loss": 0.0,
+      "step": 10265
+    },
+    {
+      "epoch": 0.6615969581749049,
+      "grad_norm": 0.000314439940892325,
+      "learning_rate": 3.7651271034729687e-06,
+      "loss": 0.0,
+      "step": 10266
+    },
+    {
+      "epoch": 0.6616614036218341,
+      "grad_norm": 0.000485502264331188,
+      "learning_rate": 3.7644110275689226e-06,
+      "loss": 0.0,
+      "step": 10267
+    },
+    {
+      "epoch": 0.6617258490687633,
+      "grad_norm": 0.0014131418138240244,
+      "learning_rate": 3.763694951664877e-06,
+      "loss": 0.0,
+      "step": 10268
+    },
+    {
+      "epoch": 0.6617902945156925,
+      "grad_norm": 0.0012902404209281864,
+      "learning_rate": 3.762978875760831e-06,
+      "loss": 0.0,
+      "step": 10269
+    },
+    {
+      "epoch": 0.6618547399626217,
+      "grad_norm": 0.5734556928030273,
+      "learning_rate": 3.762262799856785e-06,
+      "loss": 0.0018,
+      "step": 10270
+    },
+    {
+      "epoch": 0.6619191854095509,
+      "grad_norm": 0.016855807832785976,
+      "learning_rate": 3.7615467239527394e-06,
+      "loss": 0.0,
+      "step": 10271
+    },
+    {
+      "epoch": 0.66198363085648,
+      "grad_norm": 0.06585408753475606,
+      "learning_rate": 3.7608306480486932e-06,
+      "loss": 0.0001,
+      "step": 10272
+    },
+    {
+      "epoch": 0.6620480763034091,
+      "grad_norm": 0.0023837119585305598,
+      "learning_rate": 3.7601145721446475e-06,
+      "loss": 0.0,
+      "step": 10273
+    },
+    {
+      "epoch": 0.6621125217503383,
+      "grad_norm": 0.03470923425898855,
+      "learning_rate": 3.7593984962406014e-06,
+      "loss": 0.0,
+      "step": 10274
+    },
+    {
+      "epoch": 0.6621769671972675,
+      "grad_norm": 0.004313669130485635,
+      "learning_rate": 3.7586824203365557e-06,
+      "loss": 0.0,
+      "step": 10275
+    },
+    {
+      "epoch": 0.6622414126441967,
+      "grad_norm": 0.21375927750213114,
+      "learning_rate": 3.7579663444325104e-06,
+      "loss": 0.0007,
+      "step": 10276
+    },
+    {
+      "epoch": 0.6623058580911259,
+      "grad_norm": 0.00922187225142663,
+      "learning_rate": 3.7572502685284647e-06,
+      "loss": 0.0,
+      "step": 10277
+    },
+    {
+      "epoch": 0.662370303538055,
+      "grad_norm": 0.0019563550661400313,
+      "learning_rate": 3.7565341926244186e-06,
+      "loss": 0.0,
+      "step": 10278
+    },
+    {
+      "epoch": 0.6624347489849842,
+      "grad_norm": 0.012985174580154232,
+      "learning_rate": 3.755818116720373e-06,
+      "loss": 0.0,
+      "step": 10279
+    },
+    {
+      "epoch": 0.6624991944319134,
+      "grad_norm": 0.1419093634249794,
+      "learning_rate": 3.7551020408163268e-06,
+      "loss": 0.0015,
+      "step": 10280
+    },
+    {
+      "epoch": 0.6625636398788426,
+      "grad_norm": 7.844805718374012e-05,
+      "learning_rate": 3.754385964912281e-06,
+      "loss": 0.0,
+      "step": 10281
+    },
+    {
+      "epoch": 0.6626280853257718,
+      "grad_norm": 0.000882768606512407,
+      "learning_rate": 3.753669889008235e-06,
+      "loss": 0.0,
+      "step": 10282
+    },
+    {
+      "epoch": 0.662692530772701,
+      "grad_norm": 0.0002696189077644942,
+      "learning_rate": 3.7529538131041892e-06,
+      "loss": 0.0,
+      "step": 10283
+    },
+    {
+      "epoch": 0.66275697621963,
+      "grad_norm": 0.064502581492965,
+      "learning_rate": 3.7522377372001435e-06,
+      "loss": 0.0011,
+      "step": 10284
+    },
+    {
+      "epoch": 0.6628214216665592,
+      "grad_norm": 0.00019324913286995657,
+      "learning_rate": 3.7515216612960974e-06,
+      "loss": 0.0,
+      "step": 10285
+    },
+    {
+      "epoch": 0.6628858671134884,
+      "grad_norm": 0.005003145435338595,
+      "learning_rate": 3.7508055853920517e-06,
+      "loss": 0.0,
+      "step": 10286
+    },
+    {
+      "epoch": 0.6629503125604176,
+      "grad_norm": 0.005832136003543727,
+      "learning_rate": 3.7500895094880056e-06,
+      "loss": 0.0,
+      "step": 10287
+    },
+    {
+      "epoch": 0.6630147580073468,
+      "grad_norm": 0.008914760200775532,
+      "learning_rate": 3.7493734335839603e-06,
+      "loss": 0.0,
+      "step": 10288
+    },
+    {
+      "epoch": 0.663079203454276,
+      "grad_norm": 0.06776215709505574,
+      "learning_rate": 3.7486573576799146e-06,
+      "loss": 0.0001,
+      "step": 10289
+    },
+    {
+      "epoch": 0.6631436489012051,
+      "grad_norm": 0.0010801679433969562,
+      "learning_rate": 3.7479412817758685e-06,
+      "loss": 0.0,
+      "step": 10290
+    },
+    {
+      "epoch": 0.6632080943481343,
+      "grad_norm": 0.3396181804918623,
+      "learning_rate": 3.747225205871823e-06,
+      "loss": 0.0009,
+      "step": 10291
+    },
+    {
+      "epoch": 0.6632725397950635,
+      "grad_norm": 0.1640217764584769,
+      "learning_rate": 3.746509129967777e-06,
+      "loss": 0.0024,
+      "step": 10292
+    },
+    {
+      "epoch": 0.6633369852419927,
+      "grad_norm": 0.0013012118961681762,
+      "learning_rate": 3.745793054063731e-06,
+      "loss": 0.0,
+      "step": 10293
+    },
+    {
+      "epoch": 0.6634014306889219,
+      "grad_norm": 0.05860680792265722,
+      "learning_rate": 3.7450769781596853e-06,
+      "loss": 0.0002,
+      "step": 10294
+    },
+    {
+      "epoch": 0.663465876135851,
+      "grad_norm": 0.8549390572584554,
+      "learning_rate": 3.744360902255639e-06,
+      "loss": 0.0212,
+      "step": 10295
+    },
+    {
+      "epoch": 0.6635303215827801,
+      "grad_norm": 0.0017172884277614458,
+      "learning_rate": 3.7436448263515934e-06,
+      "loss": 0.0,
+      "step": 10296
+    },
+    {
+      "epoch": 0.6635947670297093,
+      "grad_norm": 0.08668761005700598,
+      "learning_rate": 3.7429287504475477e-06,
+      "loss": 0.0001,
+      "step": 10297
+    },
+    {
+      "epoch": 0.6636592124766385,
+      "grad_norm": 0.0023396848876374708,
+      "learning_rate": 3.7422126745435016e-06,
+      "loss": 0.0,
+      "step": 10298
+    },
+    {
+      "epoch": 0.6637236579235677,
+      "grad_norm": 0.0006625158127470572,
+      "learning_rate": 3.7414965986394563e-06,
+      "loss": 0.0,
+      "step": 10299
+    },
+    {
+      "epoch": 0.6637881033704969,
+      "grad_norm": 0.006708177138559468,
+      "learning_rate": 3.7407805227354106e-06,
+      "loss": 0.0,
+      "step": 10300
+    },
+    {
+      "epoch": 0.6638525488174261,
+      "grad_norm": 0.0007932890888525091,
+      "learning_rate": 3.7400644468313645e-06,
+      "loss": 0.0,
+      "step": 10301
+    },
+    {
+      "epoch": 0.6639169942643552,
+      "grad_norm": 0.0015757591637932753,
+      "learning_rate": 3.739348370927319e-06,
+      "loss": 0.0,
+      "step": 10302
+    },
+    {
+      "epoch": 0.6639814397112844,
+      "grad_norm": 0.002359474580749016,
+      "learning_rate": 3.7386322950232727e-06,
+      "loss": 0.0,
+      "step": 10303
+    },
+    {
+      "epoch": 0.6640458851582136,
+      "grad_norm": 0.047649399272796086,
+      "learning_rate": 3.737916219119227e-06,
+      "loss": 0.0002,
+      "step": 10304
+    },
+    {
+      "epoch": 0.6641103306051428,
+      "grad_norm": 0.02729586683217372,
+      "learning_rate": 3.7372001432151813e-06,
+      "loss": 0.0001,
+      "step": 10305
+    },
+    {
+      "epoch": 0.6641747760520719,
+      "grad_norm": 0.0031866841413431637,
+      "learning_rate": 3.736484067311135e-06,
+      "loss": 0.0,
+      "step": 10306
+    },
+    {
+      "epoch": 0.664239221499001,
+      "grad_norm": 0.0036964935657331686,
+      "learning_rate": 3.7357679914070895e-06,
+      "loss": 0.0,
+      "step": 10307
+    },
+    {
+      "epoch": 0.6643036669459302,
+      "grad_norm": 0.003219235261201563,
+      "learning_rate": 3.7350519155030433e-06,
+      "loss": 0.0,
+      "step": 10308
+    },
+    {
+      "epoch": 0.6643681123928594,
+      "grad_norm": 0.22426532434435242,
+      "learning_rate": 3.7343358395989976e-06,
+      "loss": 0.0003,
+      "step": 10309
+    },
+    {
+      "epoch": 0.6644325578397886,
+      "grad_norm": 0.0018211682093739476,
+      "learning_rate": 3.733619763694952e-06,
+      "loss": 0.0,
+      "step": 10310
+    },
+    {
+      "epoch": 0.6644970032867178,
+      "grad_norm": 0.2506031122662469,
+      "learning_rate": 3.7329036877909062e-06,
+      "loss": 0.0008,
+      "step": 10311
+    },
+    {
+      "epoch": 0.664561448733647,
+      "grad_norm": 0.0004966368882332178,
+      "learning_rate": 3.7321876118868605e-06,
+      "loss": 0.0,
+      "step": 10312
+    },
+    {
+      "epoch": 0.6646258941805762,
+      "grad_norm": 0.03001260390392036,
+      "learning_rate": 3.731471535982815e-06,
+      "loss": 0.0,
+      "step": 10313
+    },
+    {
+      "epoch": 0.6646903396275053,
+      "grad_norm": 0.0038538890073602046,
+      "learning_rate": 3.7307554600787687e-06,
+      "loss": 0.0,
+      "step": 10314
+    },
+    {
+      "epoch": 0.6647547850744345,
+      "grad_norm": 0.02009810895984538,
+      "learning_rate": 3.730039384174723e-06,
+      "loss": 0.0,
+      "step": 10315
+    },
+    {
+      "epoch": 0.6648192305213637,
+      "grad_norm": 0.0017787021194924147,
+      "learning_rate": 3.729323308270677e-06,
+      "loss": 0.0,
+      "step": 10316
+    },
+    {
+      "epoch": 0.6648836759682928,
+      "grad_norm": 0.2367472360801086,
+      "learning_rate": 3.728607232366631e-06,
+      "loss": 0.0008,
+      "step": 10317
+    },
+    {
+      "epoch": 0.664948121415222,
+      "grad_norm": 0.001101869151755424,
+      "learning_rate": 3.7278911564625855e-06,
+      "loss": 0.0,
+      "step": 10318
+    },
+    {
+      "epoch": 0.6650125668621512,
+      "grad_norm": 6.84407851724217e-05,
+      "learning_rate": 3.7271750805585393e-06,
+      "loss": 0.0,
+      "step": 10319
+    },
+    {
+      "epoch": 0.6650770123090803,
+      "grad_norm": 0.04718631936326828,
+      "learning_rate": 3.7264590046544936e-06,
+      "loss": 0.0001,
+      "step": 10320
+    },
+    {
+      "epoch": 0.6651414577560095,
+      "grad_norm": 0.002101132604809257,
+      "learning_rate": 3.7257429287504475e-06,
+      "loss": 0.0,
+      "step": 10321
+    },
+    {
+      "epoch": 0.6652059032029387,
+      "grad_norm": 0.054981715372835395,
+      "learning_rate": 3.725026852846402e-06,
+      "loss": 0.0,
+      "step": 10322
+    },
+    {
+      "epoch": 0.6652703486498679,
+      "grad_norm": 0.0002458432285043592,
+      "learning_rate": 3.7243107769423565e-06,
+      "loss": 0.0,
+      "step": 10323
+    },
+    {
+      "epoch": 0.6653347940967971,
+      "grad_norm": 0.14842282398353007,
+      "learning_rate": 3.7235947010383104e-06,
+      "loss": 0.0018,
+      "step": 10324
+    },
+    {
+      "epoch": 0.6653992395437263,
+      "grad_norm": 0.06344344302113403,
+      "learning_rate": 3.7228786251342647e-06,
+      "loss": 0.0,
+      "step": 10325
+    },
+    {
+      "epoch": 0.6654636849906554,
+      "grad_norm": 0.0023448721186625793,
+      "learning_rate": 3.722162549230219e-06,
+      "loss": 0.0,
+      "step": 10326
+    },
+    {
+      "epoch": 0.6655281304375846,
+      "grad_norm": 0.019546412006886026,
+      "learning_rate": 3.721446473326173e-06,
+      "loss": 0.0,
+      "step": 10327
+    },
+    {
+      "epoch": 0.6655925758845137,
+      "grad_norm": 0.16962522064579758,
+      "learning_rate": 3.720730397422127e-06,
+      "loss": 0.0005,
+      "step": 10328
+    },
+    {
+      "epoch": 0.6656570213314429,
+      "grad_norm": 0.01771669440975255,
+      "learning_rate": 3.720014321518081e-06,
+      "loss": 0.0,
+      "step": 10329
+    },
+    {
+      "epoch": 0.6657214667783721,
+      "grad_norm": 0.00024993857282373024,
+      "learning_rate": 3.7192982456140354e-06,
+      "loss": 0.0,
+      "step": 10330
+    },
+    {
+      "epoch": 0.6657859122253013,
+      "grad_norm": 0.3408004951727263,
+      "learning_rate": 3.7185821697099892e-06,
+      "loss": 0.0013,
+      "step": 10331
+    },
+    {
+      "epoch": 0.6658503576722304,
+      "grad_norm": 0.0005357757778321938,
+      "learning_rate": 3.7178660938059435e-06,
+      "loss": 0.0,
+      "step": 10332
+    },
+    {
+      "epoch": 0.6659148031191596,
+      "grad_norm": 0.03956123790735388,
+      "learning_rate": 3.717150017901898e-06,
+      "loss": 0.0001,
+      "step": 10333
+    },
+    {
+      "epoch": 0.6659792485660888,
+      "grad_norm": 0.0036982926831180154,
+      "learning_rate": 3.7164339419978526e-06,
+      "loss": 0.0,
+      "step": 10334
+    },
+    {
+      "epoch": 0.666043694013018,
+      "grad_norm": 0.0012340388580778937,
+      "learning_rate": 3.7157178660938064e-06,
+      "loss": 0.0,
+      "step": 10335
+    },
+    {
+      "epoch": 0.6661081394599472,
+      "grad_norm": 0.007880068164499425,
+      "learning_rate": 3.7150017901897607e-06,
+      "loss": 0.0,
+      "step": 10336
+    },
+    {
+      "epoch": 0.6661725849068764,
+      "grad_norm": 0.03773377443955219,
+      "learning_rate": 3.7142857142857146e-06,
+      "loss": 0.0004,
+      "step": 10337
+    },
+    {
+      "epoch": 0.6662370303538055,
+      "grad_norm": 0.0001938463646608074,
+      "learning_rate": 3.713569638381669e-06,
+      "loss": 0.0,
+      "step": 10338
+    },
+    {
+      "epoch": 0.6663014758007347,
+      "grad_norm": 0.0012649033419837533,
+      "learning_rate": 3.7128535624776228e-06,
+      "loss": 0.0,
+      "step": 10339
+    },
+    {
+      "epoch": 0.6663659212476638,
+      "grad_norm": 0.0038339946928142945,
+      "learning_rate": 3.712137486573577e-06,
+      "loss": 0.0,
+      "step": 10340
+    },
+    {
+      "epoch": 0.666430366694593,
+      "grad_norm": 0.7119580583622275,
+      "learning_rate": 3.7114214106695314e-06,
+      "loss": 0.0041,
+      "step": 10341
+    },
+    {
+      "epoch": 0.6664948121415222,
+      "grad_norm": 0.06270795364927652,
+      "learning_rate": 3.7107053347654852e-06,
+      "loss": 0.0,
+      "step": 10342
+    },
+    {
+      "epoch": 0.6665592575884514,
+      "grad_norm": 0.0682359125476423,
+      "learning_rate": 3.7099892588614395e-06,
+      "loss": 0.0017,
+      "step": 10343
+    },
+    {
+      "epoch": 0.6666237030353805,
+      "grad_norm": 0.027771674137431628,
+      "learning_rate": 3.7092731829573934e-06,
+      "loss": 0.0003,
+      "step": 10344
+    },
+    {
+      "epoch": 0.6666881484823097,
+      "grad_norm": 0.06264411062499124,
+      "learning_rate": 3.7085571070533477e-06,
+      "loss": 0.001,
+      "step": 10345
+    },
+    {
+      "epoch": 0.6667525939292389,
+      "grad_norm": 0.25018068579057157,
+      "learning_rate": 3.7078410311493024e-06,
+      "loss": 0.003,
+      "step": 10346
+    },
+    {
+      "epoch": 0.6668170393761681,
+      "grad_norm": 0.0006350143128432878,
+      "learning_rate": 3.7071249552452563e-06,
+      "loss": 0.0,
+      "step": 10347
+    },
+    {
+      "epoch": 0.6668814848230973,
+      "grad_norm": 0.004212846550878931,
+      "learning_rate": 3.7064088793412106e-06,
+      "loss": 0.0,
+      "step": 10348
+    },
+    {
+      "epoch": 0.6669459302700265,
+      "grad_norm": 0.00048795961950414513,
+      "learning_rate": 3.705692803437165e-06,
+      "loss": 0.0,
+      "step": 10349
+    },
+    {
+      "epoch": 0.6670103757169557,
+      "grad_norm": 0.2391944801432455,
+      "learning_rate": 3.704976727533119e-06,
+      "loss": 0.0008,
+      "step": 10350
+    },
+    {
+      "epoch": 0.6670748211638847,
+      "grad_norm": 0.000245488891439777,
+      "learning_rate": 3.704260651629073e-06,
+      "loss": 0.0,
+      "step": 10351
+    },
+    {
+      "epoch": 0.6671392666108139,
+      "grad_norm": 2.1831801820844412e-05,
+      "learning_rate": 3.703544575725027e-06,
+      "loss": 0.0,
+      "step": 10352
+    },
+    {
+      "epoch": 0.6672037120577431,
+      "grad_norm": 0.06861953276863456,
+      "learning_rate": 3.7028284998209813e-06,
+      "loss": 0.0001,
+      "step": 10353
+    },
+    {
+      "epoch": 0.6672681575046723,
+      "grad_norm": 0.21130783632373465,
+      "learning_rate": 3.7021124239169356e-06,
+      "loss": 0.0009,
+      "step": 10354
+    },
+    {
+      "epoch": 0.6673326029516015,
+      "grad_norm": 0.0006989204759455222,
+      "learning_rate": 3.7013963480128894e-06,
+      "loss": 0.0,
+      "step": 10355
+    },
+    {
+      "epoch": 0.6673970483985306,
+      "grad_norm": 0.00032966569814352545,
+      "learning_rate": 3.7006802721088437e-06,
+      "loss": 0.0,
+      "step": 10356
+    },
+    {
+      "epoch": 0.6674614938454598,
+      "grad_norm": 0.29515427786156617,
+      "learning_rate": 3.6999641962047976e-06,
+      "loss": 0.0017,
+      "step": 10357
+    },
+    {
+      "epoch": 0.667525939292389,
+      "grad_norm": 0.003538618157151033,
+      "learning_rate": 3.6992481203007523e-06,
+      "loss": 0.0,
+      "step": 10358
+    },
+    {
+      "epoch": 0.6675903847393182,
+      "grad_norm": 0.02578721018192703,
+      "learning_rate": 3.6985320443967066e-06,
+      "loss": 0.0,
+      "step": 10359
+    },
+    {
+      "epoch": 0.6676548301862474,
+      "grad_norm": 0.000487793444415082,
+      "learning_rate": 3.6978159684926605e-06,
+      "loss": 0.0,
+      "step": 10360
+    },
+    {
+      "epoch": 0.6677192756331766,
+      "grad_norm": 0.0018810635425694684,
+      "learning_rate": 3.697099892588615e-06,
+      "loss": 0.0,
+      "step": 10361
+    },
+    {
+      "epoch": 0.6677837210801056,
+      "grad_norm": 0.0018085405660967183,
+      "learning_rate": 3.696383816684569e-06,
+      "loss": 0.0,
+      "step": 10362
+    },
+    {
+      "epoch": 0.6678481665270348,
+      "grad_norm": 0.1033503724846543,
+      "learning_rate": 3.695667740780523e-06,
+      "loss": 0.0001,
+      "step": 10363
+    },
+    {
+      "epoch": 0.667912611973964,
+      "grad_norm": 0.03163602936890037,
+      "learning_rate": 3.6949516648764773e-06,
+      "loss": 0.0001,
+      "step": 10364
+    },
+    {
+      "epoch": 0.6679770574208932,
+      "grad_norm": 0.0005767042099623167,
+      "learning_rate": 3.694235588972431e-06,
+      "loss": 0.0,
+      "step": 10365
+    },
+    {
+      "epoch": 0.6680415028678224,
+      "grad_norm": 0.17640823233059116,
+      "learning_rate": 3.6935195130683855e-06,
+      "loss": 0.0018,
+      "step": 10366
+    },
+    {
+      "epoch": 0.6681059483147516,
+      "grad_norm": 0.0022090416866323596,
+      "learning_rate": 3.6928034371643393e-06,
+      "loss": 0.0,
+      "step": 10367
+    },
+    {
+      "epoch": 0.6681703937616807,
+      "grad_norm": 0.044256572962756785,
+      "learning_rate": 3.6920873612602936e-06,
+      "loss": 0.0,
+      "step": 10368
+    },
+    {
+      "epoch": 0.6682348392086099,
+      "grad_norm": 0.011968673310753146,
+      "learning_rate": 3.6913712853562483e-06,
+      "loss": 0.0001,
+      "step": 10369
+    },
+    {
+      "epoch": 0.6682992846555391,
+      "grad_norm": 0.002222004775121942,
+      "learning_rate": 3.6906552094522026e-06,
+      "loss": 0.0,
+      "step": 10370
+    },
+    {
+      "epoch": 0.6683637301024683,
+      "grad_norm": 0.003974169480094497,
+      "learning_rate": 3.6899391335481565e-06,
+      "loss": 0.0,
+      "step": 10371
+    },
+    {
+      "epoch": 0.6684281755493975,
+      "grad_norm": 0.011160728574854355,
+      "learning_rate": 3.689223057644111e-06,
+      "loss": 0.0001,
+      "step": 10372
+    },
+    {
+      "epoch": 0.6684926209963266,
+      "grad_norm": 0.03456428294752379,
+      "learning_rate": 3.6885069817400647e-06,
+      "loss": 0.0004,
+      "step": 10373
+    },
+    {
+      "epoch": 0.6685570664432557,
+      "grad_norm": 0.004038779991600407,
+      "learning_rate": 3.687790905836019e-06,
+      "loss": 0.0,
+      "step": 10374
+    },
+    {
+      "epoch": 0.6686215118901849,
+      "grad_norm": 0.000989251636449544,
+      "learning_rate": 3.687074829931973e-06,
+      "loss": 0.0,
+      "step": 10375
+    },
+    {
+      "epoch": 0.6686859573371141,
+      "grad_norm": 0.004165436098205323,
+      "learning_rate": 3.686358754027927e-06,
+      "loss": 0.0,
+      "step": 10376
+    },
+    {
+      "epoch": 0.6687504027840433,
+      "grad_norm": 0.14913321884813732,
+      "learning_rate": 3.6856426781238815e-06,
+      "loss": 0.0003,
+      "step": 10377
+    },
+    {
+      "epoch": 0.6688148482309725,
+      "grad_norm": 0.00021340746164696231,
+      "learning_rate": 3.6849266022198353e-06,
+      "loss": 0.0,
+      "step": 10378
+    },
+    {
+      "epoch": 0.6688792936779017,
+      "grad_norm": 0.001224805831090876,
+      "learning_rate": 3.6842105263157896e-06,
+      "loss": 0.0,
+      "step": 10379
+    },
+    {
+      "epoch": 0.6689437391248308,
+      "grad_norm": 0.059466480614118154,
+      "learning_rate": 3.6834944504117435e-06,
+      "loss": 0.0004,
+      "step": 10380
+    },
+    {
+      "epoch": 0.66900818457176,
+      "grad_norm": 0.022219569955405607,
+      "learning_rate": 3.6827783745076982e-06,
+      "loss": 0.0,
+      "step": 10381
+    },
+    {
+      "epoch": 0.6690726300186892,
+      "grad_norm": 0.010246315911304519,
+      "learning_rate": 3.6820622986036525e-06,
+      "loss": 0.0001,
+      "step": 10382
+    },
+    {
+      "epoch": 0.6691370754656184,
+      "grad_norm": 0.18016104909322808,
+      "learning_rate": 3.6813462226996064e-06,
+      "loss": 0.0015,
+      "step": 10383
+    },
+    {
+      "epoch": 0.6692015209125475,
+      "grad_norm": 0.0039480985634492635,
+      "learning_rate": 3.6806301467955607e-06,
+      "loss": 0.0,
+      "step": 10384
+    },
+    {
+      "epoch": 0.6692659663594767,
+      "grad_norm": 0.00021953914315241395,
+      "learning_rate": 3.679914070891515e-06,
+      "loss": 0.0,
+      "step": 10385
+    },
+    {
+      "epoch": 0.6693304118064058,
+      "grad_norm": 0.00140818974422586,
+      "learning_rate": 3.679197994987469e-06,
+      "loss": 0.0,
+      "step": 10386
+    },
+    {
+      "epoch": 0.669394857253335,
+      "grad_norm": 0.06190358518524449,
+      "learning_rate": 3.678481919083423e-06,
+      "loss": 0.0002,
+      "step": 10387
+    },
+    {
+      "epoch": 0.6694593027002642,
+      "grad_norm": 0.00048093340199941554,
+      "learning_rate": 3.677765843179377e-06,
+      "loss": 0.0,
+      "step": 10388
+    },
+    {
+      "epoch": 0.6695237481471934,
+      "grad_norm": 0.010102305479308519,
+      "learning_rate": 3.6770497672753314e-06,
+      "loss": 0.0,
+      "step": 10389
+    },
+    {
+      "epoch": 0.6695881935941226,
+      "grad_norm": 0.012027537219855382,
+      "learning_rate": 3.6763336913712857e-06,
+      "loss": 0.0,
+      "step": 10390
+    },
+    {
+      "epoch": 0.6696526390410518,
+      "grad_norm": 0.00474196388602497,
+      "learning_rate": 3.6756176154672395e-06,
+      "loss": 0.0,
+      "step": 10391
+    },
+    {
+      "epoch": 0.669717084487981,
+      "grad_norm": 0.00038679097703264,
+      "learning_rate": 3.674901539563194e-06,
+      "loss": 0.0,
+      "step": 10392
+    },
+    {
+      "epoch": 0.6697815299349101,
+      "grad_norm": 7.549534460174918e-05,
+      "learning_rate": 3.6741854636591486e-06,
+      "loss": 0.0,
+      "step": 10393
+    },
+    {
+      "epoch": 0.6698459753818393,
+      "grad_norm": 0.0027491068659299105,
+      "learning_rate": 3.6734693877551024e-06,
+      "loss": 0.0,
+      "step": 10394
+    },
+    {
+      "epoch": 0.6699104208287684,
+      "grad_norm": 0.05399843993640478,
+      "learning_rate": 3.6727533118510567e-06,
+      "loss": 0.0001,
+      "step": 10395
+    },
+    {
+      "epoch": 0.6699748662756976,
+      "grad_norm": 0.004283962775883192,
+      "learning_rate": 3.6720372359470106e-06,
+      "loss": 0.0,
+      "step": 10396
+    },
+    {
+      "epoch": 0.6700393117226268,
+      "grad_norm": 0.0001430948322371208,
+      "learning_rate": 3.671321160042965e-06,
+      "loss": 0.0,
+      "step": 10397
+    },
+    {
+      "epoch": 0.670103757169556,
+      "grad_norm": 0.00017731739681702292,
+      "learning_rate": 3.670605084138919e-06,
+      "loss": 0.0,
+      "step": 10398
+    },
+    {
+      "epoch": 0.6701682026164851,
+      "grad_norm": 0.000991244561069628,
+      "learning_rate": 3.669889008234873e-06,
+      "loss": 0.0,
+      "step": 10399
+    },
+    {
+      "epoch": 0.6702326480634143,
+      "grad_norm": 0.0001893566769277467,
+      "learning_rate": 3.6691729323308274e-06,
+      "loss": 0.0,
+      "step": 10400
+    },
+    {
+      "epoch": 0.6702970935103435,
+      "grad_norm": 0.004589059406335244,
+      "learning_rate": 3.6684568564267812e-06,
+      "loss": 0.0,
+      "step": 10401
+    },
+    {
+      "epoch": 0.6703615389572727,
+      "grad_norm": 3.647468846965012e-05,
+      "learning_rate": 3.6677407805227355e-06,
+      "loss": 0.0,
+      "step": 10402
+    },
+    {
+      "epoch": 0.6704259844042019,
+      "grad_norm": 6.37729973034846e-05,
+      "learning_rate": 3.66702470461869e-06,
+      "loss": 0.0,
+      "step": 10403
+    },
+    {
+      "epoch": 0.670490429851131,
+      "grad_norm": 0.2196042122873555,
+      "learning_rate": 3.666308628714644e-06,
+      "loss": 0.0008,
+      "step": 10404
+    },
+    {
+      "epoch": 0.6705548752980602,
+      "grad_norm": 0.007662267436804598,
+      "learning_rate": 3.6655925528105984e-06,
+      "loss": 0.0,
+      "step": 10405
+    },
+    {
+      "epoch": 0.6706193207449893,
+      "grad_norm": 0.0004067778012657678,
+      "learning_rate": 3.6648764769065527e-06,
+      "loss": 0.0,
+      "step": 10406
+    },
+    {
+      "epoch": 0.6706837661919185,
+      "grad_norm": 0.010651062407556951,
+      "learning_rate": 3.6641604010025066e-06,
+      "loss": 0.0,
+      "step": 10407
+    },
+    {
+      "epoch": 0.6707482116388477,
+      "grad_norm": 0.0031969810332983287,
+      "learning_rate": 3.663444325098461e-06,
+      "loss": 0.0,
+      "step": 10408
+    },
+    {
+      "epoch": 0.6708126570857769,
+      "grad_norm": 0.00044501171077525185,
+      "learning_rate": 3.662728249194415e-06,
+      "loss": 0.0,
+      "step": 10409
+    },
+    {
+      "epoch": 0.670877102532706,
+      "grad_norm": 0.00013680105479665556,
+      "learning_rate": 3.662012173290369e-06,
+      "loss": 0.0,
+      "step": 10410
+    },
+    {
+      "epoch": 0.6709415479796352,
+      "grad_norm": 0.025411179136745145,
+      "learning_rate": 3.6612960973863234e-06,
+      "loss": 0.0,
+      "step": 10411
+    },
+    {
+      "epoch": 0.6710059934265644,
+      "grad_norm": 0.05550540515503469,
+      "learning_rate": 3.6605800214822773e-06,
+      "loss": 0.0001,
+      "step": 10412
+    },
+    {
+      "epoch": 0.6710704388734936,
+      "grad_norm": 3.929251421293004e-05,
+      "learning_rate": 3.6598639455782316e-06,
+      "loss": 0.0,
+      "step": 10413
+    },
+    {
+      "epoch": 0.6711348843204228,
+      "grad_norm": 0.09189680423013248,
+      "learning_rate": 3.6591478696741854e-06,
+      "loss": 0.0002,
+      "step": 10414
+    },
+    {
+      "epoch": 0.671199329767352,
+      "grad_norm": 0.0006463515367147947,
+      "learning_rate": 3.6584317937701397e-06,
+      "loss": 0.0,
+      "step": 10415
+    },
+    {
+      "epoch": 0.6712637752142812,
+      "grad_norm": 0.00046602857314811133,
+      "learning_rate": 3.6577157178660945e-06,
+      "loss": 0.0,
+      "step": 10416
+    },
+    {
+      "epoch": 0.6713282206612103,
+      "grad_norm": 8.019380777964712e-05,
+      "learning_rate": 3.6569996419620483e-06,
+      "loss": 0.0,
+      "step": 10417
+    },
+    {
+      "epoch": 0.6713926661081394,
+      "grad_norm": 0.035754243974769166,
+      "learning_rate": 3.6562835660580026e-06,
+      "loss": 0.0,
+      "step": 10418
+    },
+    {
+      "epoch": 0.6714571115550686,
+      "grad_norm": 0.03264983678887585,
+      "learning_rate": 3.655567490153957e-06,
+      "loss": 0.0001,
+      "step": 10419
+    },
+    {
+      "epoch": 0.6715215570019978,
+      "grad_norm": 0.00046047314390313494,
+      "learning_rate": 3.654851414249911e-06,
+      "loss": 0.0,
+      "step": 10420
+    },
+    {
+      "epoch": 0.671586002448927,
+      "grad_norm": 0.14819997681575406,
+      "learning_rate": 3.654135338345865e-06,
+      "loss": 0.0023,
+      "step": 10421
+    },
+    {
+      "epoch": 0.6716504478958562,
+      "grad_norm": 0.12865062133924046,
+      "learning_rate": 3.653419262441819e-06,
+      "loss": 0.0002,
+      "step": 10422
+    },
+    {
+      "epoch": 0.6717148933427853,
+      "grad_norm": 0.000465925691879833,
+      "learning_rate": 3.6527031865377733e-06,
+      "loss": 0.0,
+      "step": 10423
+    },
+    {
+      "epoch": 0.6717793387897145,
+      "grad_norm": 0.10362138383915132,
+      "learning_rate": 3.651987110633727e-06,
+      "loss": 0.0008,
+      "step": 10424
+    },
+    {
+      "epoch": 0.6718437842366437,
+      "grad_norm": 0.11788996914326605,
+      "learning_rate": 3.6512710347296815e-06,
+      "loss": 0.0001,
+      "step": 10425
+    },
+    {
+      "epoch": 0.6719082296835729,
+      "grad_norm": 0.0004748098370773154,
+      "learning_rate": 3.6505549588256358e-06,
+      "loss": 0.0,
+      "step": 10426
+    },
+    {
+      "epoch": 0.6719726751305021,
+      "grad_norm": 0.06221575286604827,
+      "learning_rate": 3.6498388829215896e-06,
+      "loss": 0.0002,
+      "step": 10427
+    },
+    {
+      "epoch": 0.6720371205774313,
+      "grad_norm": 0.0006763647039382567,
+      "learning_rate": 3.6491228070175443e-06,
+      "loss": 0.0,
+      "step": 10428
+    },
+    {
+      "epoch": 0.6721015660243603,
+      "grad_norm": 0.013047591917105377,
+      "learning_rate": 3.6484067311134986e-06,
+      "loss": 0.0,
+      "step": 10429
+    },
+    {
+      "epoch": 0.6721660114712895,
+      "grad_norm": 0.0018885810922089606,
+      "learning_rate": 3.6476906552094525e-06,
+      "loss": 0.0,
+      "step": 10430
+    },
+    {
+      "epoch": 0.6722304569182187,
+      "grad_norm": 0.00018979485411064437,
+      "learning_rate": 3.646974579305407e-06,
+      "loss": 0.0,
+      "step": 10431
+    },
+    {
+      "epoch": 0.6722949023651479,
+      "grad_norm": 0.25609584975065497,
+      "learning_rate": 3.6462585034013607e-06,
+      "loss": 0.0009,
+      "step": 10432
+    },
+    {
+      "epoch": 0.6723593478120771,
+      "grad_norm": 0.024535905068322705,
+      "learning_rate": 3.645542427497315e-06,
+      "loss": 0.0001,
+      "step": 10433
+    },
+    {
+      "epoch": 0.6724237932590063,
+      "grad_norm": 0.0027585886411952447,
+      "learning_rate": 3.6448263515932693e-06,
+      "loss": 0.0,
+      "step": 10434
+    },
+    {
+      "epoch": 0.6724882387059354,
+      "grad_norm": 0.028292844046307754,
+      "learning_rate": 3.644110275689223e-06,
+      "loss": 0.0001,
+      "step": 10435
+    },
+    {
+      "epoch": 0.6725526841528646,
+      "grad_norm": 0.027540698239601234,
+      "learning_rate": 3.6433941997851775e-06,
+      "loss": 0.0001,
+      "step": 10436
+    },
+    {
+      "epoch": 0.6726171295997938,
+      "grad_norm": 0.00963730360489511,
+      "learning_rate": 3.6426781238811313e-06,
+      "loss": 0.0001,
+      "step": 10437
+    },
+    {
+      "epoch": 0.672681575046723,
+      "grad_norm": 0.0018280268136798214,
+      "learning_rate": 3.6419620479770856e-06,
+      "loss": 0.0,
+      "step": 10438
+    },
+    {
+      "epoch": 0.6727460204936522,
+      "grad_norm": 0.0001900797635635939,
+      "learning_rate": 3.6412459720730404e-06,
+      "loss": 0.0,
+      "step": 10439
+    },
+    {
+      "epoch": 0.6728104659405812,
+      "grad_norm": 0.00023532304022595498,
+      "learning_rate": 3.6405298961689942e-06,
+      "loss": 0.0,
+      "step": 10440
+    },
+    {
+      "epoch": 0.6728749113875104,
+      "grad_norm": 0.0019142251629258287,
+      "learning_rate": 3.6398138202649485e-06,
+      "loss": 0.0,
+      "step": 10441
+    },
+    {
+      "epoch": 0.6729393568344396,
+      "grad_norm": 0.00029465788696560115,
+      "learning_rate": 3.639097744360903e-06,
+      "loss": 0.0,
+      "step": 10442
+    },
+    {
+      "epoch": 0.6730038022813688,
+      "grad_norm": 0.00023300991071189333,
+      "learning_rate": 3.6383816684568567e-06,
+      "loss": 0.0,
+      "step": 10443
+    },
+    {
+      "epoch": 0.673068247728298,
+      "grad_norm": 0.22997037699398992,
+      "learning_rate": 3.637665592552811e-06,
+      "loss": 0.0006,
+      "step": 10444
+    },
+    {
+      "epoch": 0.6731326931752272,
+      "grad_norm": 0.41474730166085766,
+      "learning_rate": 3.636949516648765e-06,
+      "loss": 0.0018,
+      "step": 10445
+    },
+    {
+      "epoch": 0.6731971386221564,
+      "grad_norm": 0.0018304284832990753,
+      "learning_rate": 3.636233440744719e-06,
+      "loss": 0.0,
+      "step": 10446
+    },
+    {
+      "epoch": 0.6732615840690855,
+      "grad_norm": 0.2089151285548803,
+      "learning_rate": 3.6355173648406735e-06,
+      "loss": 0.0019,
+      "step": 10447
+    },
+    {
+      "epoch": 0.6733260295160147,
+      "grad_norm": 0.0003634330347498069,
+      "learning_rate": 3.6348012889366274e-06,
+      "loss": 0.0,
+      "step": 10448
+    },
+    {
+      "epoch": 0.6733904749629439,
+      "grad_norm": 0.001954738083069696,
+      "learning_rate": 3.6340852130325817e-06,
+      "loss": 0.0,
+      "step": 10449
+    },
+    {
+      "epoch": 0.6734549204098731,
+      "grad_norm": 0.001967566134604329,
+      "learning_rate": 3.6333691371285355e-06,
+      "loss": 0.0,
+      "step": 10450
+    },
+    {
+      "epoch": 0.6735193658568022,
+      "grad_norm": 0.01803615169688362,
+      "learning_rate": 3.6326530612244903e-06,
+      "loss": 0.0002,
+      "step": 10451
+    },
+    {
+      "epoch": 0.6735838113037314,
+      "grad_norm": 0.18593643853341382,
+      "learning_rate": 3.6319369853204446e-06,
+      "loss": 0.0004,
+      "step": 10452
+    },
+    {
+      "epoch": 0.6736482567506605,
+      "grad_norm": 0.0021236017655750163,
+      "learning_rate": 3.6312209094163984e-06,
+      "loss": 0.0,
+      "step": 10453
+    },
+    {
+      "epoch": 0.6737127021975897,
+      "grad_norm": 0.0017699274242623294,
+      "learning_rate": 3.6305048335123527e-06,
+      "loss": 0.0,
+      "step": 10454
+    },
+    {
+      "epoch": 0.6737771476445189,
+      "grad_norm": 0.0001289320386872569,
+      "learning_rate": 3.629788757608307e-06,
+      "loss": 0.0,
+      "step": 10455
+    },
+    {
+      "epoch": 0.6738415930914481,
+      "grad_norm": 0.017909558811695953,
+      "learning_rate": 3.629072681704261e-06,
+      "loss": 0.0,
+      "step": 10456
+    },
+    {
+      "epoch": 0.6739060385383773,
+      "grad_norm": 0.8763559597886815,
+      "learning_rate": 3.628356605800215e-06,
+      "loss": 0.0046,
+      "step": 10457
+    },
+    {
+      "epoch": 0.6739704839853065,
+      "grad_norm": 0.00027276141796046864,
+      "learning_rate": 3.627640529896169e-06,
+      "loss": 0.0,
+      "step": 10458
+    },
+    {
+      "epoch": 0.6740349294322356,
+      "grad_norm": 2.0164551910408486,
+      "learning_rate": 3.6269244539921234e-06,
+      "loss": 0.0096,
+      "step": 10459
+    },
+    {
+      "epoch": 0.6740993748791648,
+      "grad_norm": 0.00017352627869547739,
+      "learning_rate": 3.6262083780880777e-06,
+      "loss": 0.0,
+      "step": 10460
+    },
+    {
+      "epoch": 0.674163820326094,
+      "grad_norm": 0.0023188071801069775,
+      "learning_rate": 3.6254923021840315e-06,
+      "loss": 0.0,
+      "step": 10461
+    },
+    {
+      "epoch": 0.6742282657730231,
+      "grad_norm": 0.004480081186474678,
+      "learning_rate": 3.624776226279986e-06,
+      "loss": 0.0,
+      "step": 10462
+    },
+    {
+      "epoch": 0.6742927112199523,
+      "grad_norm": 0.003887006294521679,
+      "learning_rate": 3.6240601503759406e-06,
+      "loss": 0.0,
+      "step": 10463
+    },
+    {
+      "epoch": 0.6743571566668815,
+      "grad_norm": 0.00026653596353573024,
+      "learning_rate": 3.6233440744718944e-06,
+      "loss": 0.0,
+      "step": 10464
+    },
+    {
+      "epoch": 0.6744216021138106,
+      "grad_norm": 0.00012657650139262516,
+      "learning_rate": 3.6226279985678487e-06,
+      "loss": 0.0,
+      "step": 10465
+    },
+    {
+      "epoch": 0.6744860475607398,
+      "grad_norm": 0.0031173247361041124,
+      "learning_rate": 3.6219119226638026e-06,
+      "loss": 0.0,
+      "step": 10466
+    },
+    {
+      "epoch": 0.674550493007669,
+      "grad_norm": 0.23030574335774806,
+      "learning_rate": 3.621195846759757e-06,
+      "loss": 0.0005,
+      "step": 10467
+    },
+    {
+      "epoch": 0.6746149384545982,
+      "grad_norm": 0.05862473026472335,
+      "learning_rate": 3.6204797708557112e-06,
+      "loss": 0.0006,
+      "step": 10468
+    },
+    {
+      "epoch": 0.6746793839015274,
+      "grad_norm": 0.0003351622123052417,
+      "learning_rate": 3.619763694951665e-06,
+      "loss": 0.0,
+      "step": 10469
+    },
+    {
+      "epoch": 0.6747438293484566,
+      "grad_norm": 0.00023233519721705282,
+      "learning_rate": 3.6190476190476194e-06,
+      "loss": 0.0,
+      "step": 10470
+    },
+    {
+      "epoch": 0.6748082747953857,
+      "grad_norm": 0.0035822470136483175,
+      "learning_rate": 3.6183315431435733e-06,
+      "loss": 0.0,
+      "step": 10471
+    },
+    {
+      "epoch": 0.6748727202423149,
+      "grad_norm": 0.0004719819756121723,
+      "learning_rate": 3.6176154672395276e-06,
+      "loss": 0.0,
+      "step": 10472
+    },
+    {
+      "epoch": 0.674937165689244,
+      "grad_norm": 0.004377407497624656,
+      "learning_rate": 3.6168993913354814e-06,
+      "loss": 0.0,
+      "step": 10473
+    },
+    {
+      "epoch": 0.6750016111361732,
+      "grad_norm": 0.00190075138693745,
+      "learning_rate": 3.616183315431436e-06,
+      "loss": 0.0,
+      "step": 10474
+    },
+    {
+      "epoch": 0.6750660565831024,
+      "grad_norm": 0.004317345459707869,
+      "learning_rate": 3.6154672395273905e-06,
+      "loss": 0.0,
+      "step": 10475
+    },
+    {
+      "epoch": 0.6751305020300316,
+      "grad_norm": 0.0014907583759944283,
+      "learning_rate": 3.6147511636233448e-06,
+      "loss": 0.0,
+      "step": 10476
+    },
+    {
+      "epoch": 0.6751949474769607,
+      "grad_norm": 0.0018248634672141091,
+      "learning_rate": 3.6140350877192986e-06,
+      "loss": 0.0,
+      "step": 10477
+    },
+    {
+      "epoch": 0.6752593929238899,
+      "grad_norm": 0.000946823727749897,
+      "learning_rate": 3.613319011815253e-06,
+      "loss": 0.0,
+      "step": 10478
+    },
+    {
+      "epoch": 0.6753238383708191,
+      "grad_norm": 0.014769051121375172,
+      "learning_rate": 3.612602935911207e-06,
+      "loss": 0.0001,
+      "step": 10479
+    },
+    {
+      "epoch": 0.6753882838177483,
+      "grad_norm": 0.0037277822552123324,
+      "learning_rate": 3.611886860007161e-06,
+      "loss": 0.0,
+      "step": 10480
+    },
+    {
+      "epoch": 0.6754527292646775,
+      "grad_norm": 0.003775872304534592,
+      "learning_rate": 3.611170784103115e-06,
+      "loss": 0.0,
+      "step": 10481
+    },
+    {
+      "epoch": 0.6755171747116067,
+      "grad_norm": 0.0004943616280413878,
+      "learning_rate": 3.6104547081990693e-06,
+      "loss": 0.0,
+      "step": 10482
+    },
+    {
+      "epoch": 0.6755816201585358,
+      "grad_norm": 1.1965923065115471,
+      "learning_rate": 3.6097386322950236e-06,
+      "loss": 0.0035,
+      "step": 10483
+    },
+    {
+      "epoch": 0.675646065605465,
+      "grad_norm": 0.04179766004707495,
+      "learning_rate": 3.6090225563909775e-06,
+      "loss": 0.0001,
+      "step": 10484
+    },
+    {
+      "epoch": 0.6757105110523941,
+      "grad_norm": 0.006248109641524533,
+      "learning_rate": 3.6083064804869318e-06,
+      "loss": 0.0,
+      "step": 10485
+    },
+    {
+      "epoch": 0.6757749564993233,
+      "grad_norm": 0.0003839978250398899,
+      "learning_rate": 3.6075904045828865e-06,
+      "loss": 0.0,
+      "step": 10486
+    },
+    {
+      "epoch": 0.6758394019462525,
+      "grad_norm": 0.006934538377296952,
+      "learning_rate": 3.6068743286788403e-06,
+      "loss": 0.0,
+      "step": 10487
+    },
+    {
+      "epoch": 0.6759038473931817,
+      "grad_norm": 0.00016873140626715315,
+      "learning_rate": 3.6061582527747946e-06,
+      "loss": 0.0,
+      "step": 10488
+    },
+    {
+      "epoch": 0.6759682928401108,
+      "grad_norm": 0.01762084273842604,
+      "learning_rate": 3.6054421768707485e-06,
+      "loss": 0.0,
+      "step": 10489
+    },
+    {
+      "epoch": 0.67603273828704,
+      "grad_norm": 0.46920061704573907,
+      "learning_rate": 3.604726100966703e-06,
+      "loss": 0.0009,
+      "step": 10490
+    },
+    {
+      "epoch": 0.6760971837339692,
+      "grad_norm": 0.05753273553305687,
+      "learning_rate": 3.604010025062657e-06,
+      "loss": 0.0004,
+      "step": 10491
+    },
+    {
+      "epoch": 0.6761616291808984,
+      "grad_norm": 0.0003436129141898001,
+      "learning_rate": 3.603293949158611e-06,
+      "loss": 0.0,
+      "step": 10492
+    },
+    {
+      "epoch": 0.6762260746278276,
+      "grad_norm": 0.0037809435500818145,
+      "learning_rate": 3.6025778732545653e-06,
+      "loss": 0.0,
+      "step": 10493
+    },
+    {
+      "epoch": 0.6762905200747568,
+      "grad_norm": 0.0003521812249406442,
+      "learning_rate": 3.601861797350519e-06,
+      "loss": 0.0,
+      "step": 10494
+    },
+    {
+      "epoch": 0.676354965521686,
+      "grad_norm": 0.00025709992147468773,
+      "learning_rate": 3.6011457214464735e-06,
+      "loss": 0.0,
+      "step": 10495
+    },
+    {
+      "epoch": 0.676419410968615,
+      "grad_norm": 0.005573763881191372,
+      "learning_rate": 3.6004296455424278e-06,
+      "loss": 0.0,
+      "step": 10496
+    },
+    {
+      "epoch": 0.6764838564155442,
+      "grad_norm": 0.02430756228068615,
+      "learning_rate": 3.5997135696383816e-06,
+      "loss": 0.0,
+      "step": 10497
+    },
+    {
+      "epoch": 0.6765483018624734,
+      "grad_norm": 0.017898788840930383,
+      "learning_rate": 3.5989974937343364e-06,
+      "loss": 0.0,
+      "step": 10498
+    },
+    {
+      "epoch": 0.6766127473094026,
+      "grad_norm": 0.008393354787299115,
+      "learning_rate": 3.5982814178302907e-06,
+      "loss": 0.0,
+      "step": 10499
+    },
+    {
+      "epoch": 0.6766771927563318,
+      "grad_norm": 0.003671404471434068,
+      "learning_rate": 3.5975653419262445e-06,
+      "loss": 0.0,
+      "step": 10500
+    },
+    {
+      "epoch": 0.6767416382032609,
+      "grad_norm": 0.0050457432757222654,
+      "learning_rate": 3.596849266022199e-06,
+      "loss": 0.0,
+      "step": 10501
+    },
+    {
+      "epoch": 0.6768060836501901,
+      "grad_norm": 0.00114763842394845,
+      "learning_rate": 3.5961331901181527e-06,
+      "loss": 0.0,
+      "step": 10502
+    },
+    {
+      "epoch": 0.6768705290971193,
+      "grad_norm": 0.0008545279761466147,
+      "learning_rate": 3.595417114214107e-06,
+      "loss": 0.0,
+      "step": 10503
+    },
+    {
+      "epoch": 0.6769349745440485,
+      "grad_norm": 0.00034751624162334137,
+      "learning_rate": 3.5947010383100613e-06,
+      "loss": 0.0,
+      "step": 10504
+    },
+    {
+      "epoch": 0.6769994199909777,
+      "grad_norm": 0.015582994439986578,
+      "learning_rate": 3.593984962406015e-06,
+      "loss": 0.0,
+      "step": 10505
+    },
+    {
+      "epoch": 0.6770638654379069,
+      "grad_norm": 0.040964320145442906,
+      "learning_rate": 3.5932688865019695e-06,
+      "loss": 0.0001,
+      "step": 10506
+    },
+    {
+      "epoch": 0.6771283108848359,
+      "grad_norm": 0.010853452613408077,
+      "learning_rate": 3.5925528105979234e-06,
+      "loss": 0.0001,
+      "step": 10507
+    },
+    {
+      "epoch": 0.6771927563317651,
+      "grad_norm": 0.02525209498935422,
+      "learning_rate": 3.5918367346938777e-06,
+      "loss": 0.0001,
+      "step": 10508
+    },
+    {
+      "epoch": 0.6772572017786943,
+      "grad_norm": 0.022178140745299144,
+      "learning_rate": 3.5911206587898324e-06,
+      "loss": 0.0,
+      "step": 10509
+    },
+    {
+      "epoch": 0.6773216472256235,
+      "grad_norm": 0.008376311699932986,
+      "learning_rate": 3.5904045828857863e-06,
+      "loss": 0.0001,
+      "step": 10510
+    },
+    {
+      "epoch": 0.6773860926725527,
+      "grad_norm": 0.001207447161221422,
+      "learning_rate": 3.5896885069817406e-06,
+      "loss": 0.0,
+      "step": 10511
+    },
+    {
+      "epoch": 0.6774505381194819,
+      "grad_norm": 0.0029356591079545027,
+      "learning_rate": 3.588972431077695e-06,
+      "loss": 0.0,
+      "step": 10512
+    },
+    {
+      "epoch": 0.677514983566411,
+      "grad_norm": 0.017201078670219305,
+      "learning_rate": 3.5882563551736487e-06,
+      "loss": 0.0,
+      "step": 10513
+    },
+    {
+      "epoch": 0.6775794290133402,
+      "grad_norm": 0.00014996079096439788,
+      "learning_rate": 3.587540279269603e-06,
+      "loss": 0.0,
+      "step": 10514
+    },
+    {
+      "epoch": 0.6776438744602694,
+      "grad_norm": 0.01063396315857339,
+      "learning_rate": 3.586824203365557e-06,
+      "loss": 0.0001,
+      "step": 10515
+    },
+    {
+      "epoch": 0.6777083199071986,
+      "grad_norm": 0.000691680774477271,
+      "learning_rate": 3.586108127461511e-06,
+      "loss": 0.0,
+      "step": 10516
+    },
+    {
+      "epoch": 0.6777727653541278,
+      "grad_norm": 0.057157545980472034,
+      "learning_rate": 3.585392051557465e-06,
+      "loss": 0.0002,
+      "step": 10517
+    },
+    {
+      "epoch": 0.6778372108010569,
+      "grad_norm": 0.00029040852215511486,
+      "learning_rate": 3.5846759756534194e-06,
+      "loss": 0.0,
+      "step": 10518
+    },
+    {
+      "epoch": 0.677901656247986,
+      "grad_norm": 0.003804059215826307,
+      "learning_rate": 3.5839598997493737e-06,
+      "loss": 0.0,
+      "step": 10519
+    },
+    {
+      "epoch": 0.6779661016949152,
+      "grad_norm": 0.009584564853263446,
+      "learning_rate": 3.5832438238453275e-06,
+      "loss": 0.0001,
+      "step": 10520
+    },
+    {
+      "epoch": 0.6780305471418444,
+      "grad_norm": 0.0015142631223244963,
+      "learning_rate": 3.5825277479412823e-06,
+      "loss": 0.0,
+      "step": 10521
+    },
+    {
+      "epoch": 0.6780949925887736,
+      "grad_norm": 0.0005718388729823543,
+      "learning_rate": 3.5818116720372366e-06,
+      "loss": 0.0,
+      "step": 10522
+    },
+    {
+      "epoch": 0.6781594380357028,
+      "grad_norm": 0.012310340151570972,
+      "learning_rate": 3.5810955961331904e-06,
+      "loss": 0.0,
+      "step": 10523
+    },
+    {
+      "epoch": 0.678223883482632,
+      "grad_norm": 0.00043170374826357976,
+      "learning_rate": 3.5803795202291447e-06,
+      "loss": 0.0,
+      "step": 10524
+    },
+    {
+      "epoch": 0.6782883289295611,
+      "grad_norm": 0.027632992934753346,
+      "learning_rate": 3.5796634443250986e-06,
+      "loss": 0.0,
+      "step": 10525
+    },
+    {
+      "epoch": 0.6783527743764903,
+      "grad_norm": 0.001345522934734624,
+      "learning_rate": 3.578947368421053e-06,
+      "loss": 0.0,
+      "step": 10526
+    },
+    {
+      "epoch": 0.6784172198234195,
+      "grad_norm": 0.0008807635063323086,
+      "learning_rate": 3.5782312925170072e-06,
+      "loss": 0.0,
+      "step": 10527
+    },
+    {
+      "epoch": 0.6784816652703487,
+      "grad_norm": 0.002128459962488065,
+      "learning_rate": 3.577515216612961e-06,
+      "loss": 0.0,
+      "step": 10528
+    },
+    {
+      "epoch": 0.6785461107172778,
+      "grad_norm": 0.0006520468088653372,
+      "learning_rate": 3.5767991407089154e-06,
+      "loss": 0.0,
+      "step": 10529
+    },
+    {
+      "epoch": 0.678610556164207,
+      "grad_norm": 0.02735517573911248,
+      "learning_rate": 3.5760830648048693e-06,
+      "loss": 0.0,
+      "step": 10530
+    },
+    {
+      "epoch": 0.6786750016111361,
+      "grad_norm": 0.4512681734435724,
+      "learning_rate": 3.5753669889008236e-06,
+      "loss": 0.0032,
+      "step": 10531
+    },
+    {
+      "epoch": 0.6787394470580653,
+      "grad_norm": 0.0036808522723295995,
+      "learning_rate": 3.574650912996778e-06,
+      "loss": 0.0,
+      "step": 10532
+    },
+    {
+      "epoch": 0.6788038925049945,
+      "grad_norm": 7.517311356908132e-05,
+      "learning_rate": 3.573934837092732e-06,
+      "loss": 0.0,
+      "step": 10533
+    },
+    {
+      "epoch": 0.6788683379519237,
+      "grad_norm": 0.11774557663446128,
+      "learning_rate": 3.5732187611886865e-06,
+      "loss": 0.0018,
+      "step": 10534
+    },
+    {
+      "epoch": 0.6789327833988529,
+      "grad_norm": 0.02647121095954798,
+      "learning_rate": 3.5725026852846408e-06,
+      "loss": 0.0002,
+      "step": 10535
+    },
+    {
+      "epoch": 0.6789972288457821,
+      "grad_norm": 0.035277858321916856,
+      "learning_rate": 3.5717866093805946e-06,
+      "loss": 0.0,
+      "step": 10536
+    },
+    {
+      "epoch": 0.6790616742927112,
+      "grad_norm": 0.41236802431897623,
+      "learning_rate": 3.571070533476549e-06,
+      "loss": 0.0012,
+      "step": 10537
+    },
+    {
+      "epoch": 0.6791261197396404,
+      "grad_norm": 0.0002817103551778652,
+      "learning_rate": 3.570354457572503e-06,
+      "loss": 0.0,
+      "step": 10538
+    },
+    {
+      "epoch": 0.6791905651865696,
+      "grad_norm": 0.0067230817027691785,
+      "learning_rate": 3.569638381668457e-06,
+      "loss": 0.0,
+      "step": 10539
+    },
+    {
+      "epoch": 0.6792550106334987,
+      "grad_norm": 0.0019842109186937367,
+      "learning_rate": 3.5689223057644114e-06,
+      "loss": 0.0,
+      "step": 10540
+    },
+    {
+      "epoch": 0.6793194560804279,
+      "grad_norm": 0.0008768804890963424,
+      "learning_rate": 3.5682062298603653e-06,
+      "loss": 0.0,
+      "step": 10541
+    },
+    {
+      "epoch": 0.6793839015273571,
+      "grad_norm": 0.052039909390406425,
+      "learning_rate": 3.5674901539563196e-06,
+      "loss": 0.0,
+      "step": 10542
+    },
+    {
+      "epoch": 0.6794483469742862,
+      "grad_norm": 0.021989190979891722,
+      "learning_rate": 3.5667740780522735e-06,
+      "loss": 0.0,
+      "step": 10543
+    },
+    {
+      "epoch": 0.6795127924212154,
+      "grad_norm": 0.0007638219482915713,
+      "learning_rate": 3.5660580021482278e-06,
+      "loss": 0.0,
+      "step": 10544
+    },
+    {
+      "epoch": 0.6795772378681446,
+      "grad_norm": 1.0232287477942066,
+      "learning_rate": 3.5653419262441825e-06,
+      "loss": 0.0047,
+      "step": 10545
+    },
+    {
+      "epoch": 0.6796416833150738,
+      "grad_norm": 0.001587062494048603,
+      "learning_rate": 3.5646258503401363e-06,
+      "loss": 0.0,
+      "step": 10546
+    },
+    {
+      "epoch": 0.679706128762003,
+      "grad_norm": 0.018462534965671778,
+      "learning_rate": 3.5639097744360906e-06,
+      "loss": 0.0,
+      "step": 10547
+    },
+    {
+      "epoch": 0.6797705742089322,
+      "grad_norm": 0.013377734314796297,
+      "learning_rate": 3.563193698532045e-06,
+      "loss": 0.0,
+      "step": 10548
+    },
+    {
+      "epoch": 0.6798350196558614,
+      "grad_norm": 0.9623252028734142,
+      "learning_rate": 3.562477622627999e-06,
+      "loss": 0.0018,
+      "step": 10549
+    },
+    {
+      "epoch": 0.6798994651027905,
+      "grad_norm": 0.0017705625053728948,
+      "learning_rate": 3.561761546723953e-06,
+      "loss": 0.0,
+      "step": 10550
+    },
+    {
+      "epoch": 0.6799639105497196,
+      "grad_norm": 0.00017713197781208631,
+      "learning_rate": 3.561045470819907e-06,
+      "loss": 0.0,
+      "step": 10551
+    },
+    {
+      "epoch": 0.6800283559966488,
+      "grad_norm": 0.020427811779944585,
+      "learning_rate": 3.5603293949158613e-06,
+      "loss": 0.0,
+      "step": 10552
+    },
+    {
+      "epoch": 0.680092801443578,
+      "grad_norm": 2.4890916304191784,
+      "learning_rate": 3.5596133190118156e-06,
+      "loss": 0.0203,
+      "step": 10553
+    },
+    {
+      "epoch": 0.6801572468905072,
+      "grad_norm": 0.001329291279453643,
+      "learning_rate": 3.5588972431077695e-06,
+      "loss": 0.0,
+      "step": 10554
+    },
+    {
+      "epoch": 0.6802216923374363,
+      "grad_norm": 0.1980014711359068,
+      "learning_rate": 3.5581811672037238e-06,
+      "loss": 0.0014,
+      "step": 10555
+    },
+    {
+      "epoch": 0.6802861377843655,
+      "grad_norm": 0.0051990412329180015,
+      "learning_rate": 3.5574650912996785e-06,
+      "loss": 0.0,
+      "step": 10556
+    },
+    {
+      "epoch": 0.6803505832312947,
+      "grad_norm": 0.017665047112924433,
+      "learning_rate": 3.5567490153956324e-06,
+      "loss": 0.0,
+      "step": 10557
+    },
+    {
+      "epoch": 0.6804150286782239,
+      "grad_norm": 0.001657024012985856,
+      "learning_rate": 3.5560329394915867e-06,
+      "loss": 0.0,
+      "step": 10558
+    },
+    {
+      "epoch": 0.6804794741251531,
+      "grad_norm": 0.0008793134489039452,
+      "learning_rate": 3.5553168635875405e-06,
+      "loss": 0.0,
+      "step": 10559
+    },
+    {
+      "epoch": 0.6805439195720823,
+      "grad_norm": 0.0007335920395671671,
+      "learning_rate": 3.554600787683495e-06,
+      "loss": 0.0,
+      "step": 10560
+    },
+    {
+      "epoch": 0.6806083650190115,
+      "grad_norm": 0.0045903698338804495,
+      "learning_rate": 3.553884711779449e-06,
+      "loss": 0.0,
+      "step": 10561
+    },
+    {
+      "epoch": 0.6806728104659406,
+      "grad_norm": 0.000662578100139206,
+      "learning_rate": 3.553168635875403e-06,
+      "loss": 0.0,
+      "step": 10562
+    },
+    {
+      "epoch": 0.6807372559128697,
+      "grad_norm": 0.0009688112778837717,
+      "learning_rate": 3.5524525599713573e-06,
+      "loss": 0.0,
+      "step": 10563
+    },
+    {
+      "epoch": 0.6808017013597989,
+      "grad_norm": 0.0014027335978677121,
+      "learning_rate": 3.551736484067311e-06,
+      "loss": 0.0,
+      "step": 10564
+    },
+    {
+      "epoch": 0.6808661468067281,
+      "grad_norm": 6.014109765455335e-05,
+      "learning_rate": 3.5510204081632655e-06,
+      "loss": 0.0,
+      "step": 10565
+    },
+    {
+      "epoch": 0.6809305922536573,
+      "grad_norm": 0.0012696360857282496,
+      "learning_rate": 3.5503043322592194e-06,
+      "loss": 0.0,
+      "step": 10566
+    },
+    {
+      "epoch": 0.6809950377005864,
+      "grad_norm": 0.0020264019625621287,
+      "learning_rate": 3.5495882563551737e-06,
+      "loss": 0.0,
+      "step": 10567
+    },
+    {
+      "epoch": 0.6810594831475156,
+      "grad_norm": 0.0001300260332313125,
+      "learning_rate": 3.5488721804511284e-06,
+      "loss": 0.0,
+      "step": 10568
+    },
+    {
+      "epoch": 0.6811239285944448,
+      "grad_norm": 0.11144132046079638,
+      "learning_rate": 3.5481561045470827e-06,
+      "loss": 0.0001,
+      "step": 10569
+    },
+    {
+      "epoch": 0.681188374041374,
+      "grad_norm": 0.0040325540813843305,
+      "learning_rate": 3.5474400286430366e-06,
+      "loss": 0.0,
+      "step": 10570
+    },
+    {
+      "epoch": 0.6812528194883032,
+      "grad_norm": 0.00011549701540412366,
+      "learning_rate": 3.546723952738991e-06,
+      "loss": 0.0,
+      "step": 10571
+    },
+    {
+      "epoch": 0.6813172649352324,
+      "grad_norm": 0.011268333555497053,
+      "learning_rate": 3.5460078768349447e-06,
+      "loss": 0.0001,
+      "step": 10572
+    },
+    {
+      "epoch": 0.6813817103821616,
+      "grad_norm": 0.0006847428079941398,
+      "learning_rate": 3.545291800930899e-06,
+      "loss": 0.0,
+      "step": 10573
+    },
+    {
+      "epoch": 0.6814461558290906,
+      "grad_norm": 0.009468708355515205,
+      "learning_rate": 3.544575725026853e-06,
+      "loss": 0.0,
+      "step": 10574
+    },
+    {
+      "epoch": 0.6815106012760198,
+      "grad_norm": 0.006315914786455601,
+      "learning_rate": 3.543859649122807e-06,
+      "loss": 0.0,
+      "step": 10575
+    },
+    {
+      "epoch": 0.681575046722949,
+      "grad_norm": 0.006827846703311107,
+      "learning_rate": 3.5431435732187615e-06,
+      "loss": 0.0,
+      "step": 10576
+    },
+    {
+      "epoch": 0.6816394921698782,
+      "grad_norm": 0.029604628539176416,
+      "learning_rate": 3.5424274973147154e-06,
+      "loss": 0.0002,
+      "step": 10577
+    },
+    {
+      "epoch": 0.6817039376168074,
+      "grad_norm": 0.051215081592810865,
+      "learning_rate": 3.5417114214106697e-06,
+      "loss": 0.0002,
+      "step": 10578
+    },
+    {
+      "epoch": 0.6817683830637365,
+      "grad_norm": 0.016480723087823855,
+      "learning_rate": 3.5409953455066235e-06,
+      "loss": 0.0001,
+      "step": 10579
+    },
+    {
+      "epoch": 0.6818328285106657,
+      "grad_norm": 0.0006766804201587516,
+      "learning_rate": 3.5402792696025783e-06,
+      "loss": 0.0,
+      "step": 10580
+    },
+    {
+      "epoch": 0.6818972739575949,
+      "grad_norm": 0.0019866411763065327,
+      "learning_rate": 3.5395631936985326e-06,
+      "loss": 0.0,
+      "step": 10581
+    },
+    {
+      "epoch": 0.6819617194045241,
+      "grad_norm": 0.0004252701808851787,
+      "learning_rate": 3.5388471177944864e-06,
+      "loss": 0.0,
+      "step": 10582
+    },
+    {
+      "epoch": 0.6820261648514533,
+      "grad_norm": 0.00031296475467953647,
+      "learning_rate": 3.5381310418904407e-06,
+      "loss": 0.0,
+      "step": 10583
+    },
+    {
+      "epoch": 0.6820906102983825,
+      "grad_norm": 0.00832683055652773,
+      "learning_rate": 3.537414965986395e-06,
+      "loss": 0.0,
+      "step": 10584
+    },
+    {
+      "epoch": 0.6821550557453115,
+      "grad_norm": 0.021867903048846355,
+      "learning_rate": 3.536698890082349e-06,
+      "loss": 0.0002,
+      "step": 10585
+    },
+    {
+      "epoch": 0.6822195011922407,
+      "grad_norm": 0.48420196949728816,
+      "learning_rate": 3.5359828141783032e-06,
+      "loss": 0.0034,
+      "step": 10586
+    },
+    {
+      "epoch": 0.6822839466391699,
+      "grad_norm": 0.0024605339805774456,
+      "learning_rate": 3.535266738274257e-06,
+      "loss": 0.0,
+      "step": 10587
+    },
+    {
+      "epoch": 0.6823483920860991,
+      "grad_norm": 0.011360759604440345,
+      "learning_rate": 3.5345506623702114e-06,
+      "loss": 0.0,
+      "step": 10588
+    },
+    {
+      "epoch": 0.6824128375330283,
+      "grad_norm": 0.0009775462378554197,
+      "learning_rate": 3.5338345864661657e-06,
+      "loss": 0.0,
+      "step": 10589
+    },
+    {
+      "epoch": 0.6824772829799575,
+      "grad_norm": 0.001092895735309418,
+      "learning_rate": 3.5331185105621196e-06,
+      "loss": 0.0,
+      "step": 10590
+    },
+    {
+      "epoch": 0.6825417284268867,
+      "grad_norm": 0.0027246687183617134,
+      "learning_rate": 3.5324024346580743e-06,
+      "loss": 0.0,
+      "step": 10591
+    },
+    {
+      "epoch": 0.6826061738738158,
+      "grad_norm": 0.0007618257843385489,
+      "learning_rate": 3.5316863587540286e-06,
+      "loss": 0.0,
+      "step": 10592
+    },
+    {
+      "epoch": 0.682670619320745,
+      "grad_norm": 0.18731919153917567,
+      "learning_rate": 3.5309702828499825e-06,
+      "loss": 0.0006,
+      "step": 10593
+    },
+    {
+      "epoch": 0.6827350647676742,
+      "grad_norm": 0.00033679473596064693,
+      "learning_rate": 3.5302542069459368e-06,
+      "loss": 0.0,
+      "step": 10594
+    },
+    {
+      "epoch": 0.6827995102146034,
+      "grad_norm": 0.3369809402210824,
+      "learning_rate": 3.5295381310418906e-06,
+      "loss": 0.0029,
+      "step": 10595
+    },
+    {
+      "epoch": 0.6828639556615325,
+      "grad_norm": 0.007815817597920173,
+      "learning_rate": 3.528822055137845e-06,
+      "loss": 0.0,
+      "step": 10596
+    },
+    {
+      "epoch": 0.6829284011084616,
+      "grad_norm": 0.01382235643620594,
+      "learning_rate": 3.5281059792337992e-06,
+      "loss": 0.0,
+      "step": 10597
+    },
+    {
+      "epoch": 0.6829928465553908,
+      "grad_norm": 0.0020800628314902774,
+      "learning_rate": 3.527389903329753e-06,
+      "loss": 0.0,
+      "step": 10598
+    },
+    {
+      "epoch": 0.68305729200232,
+      "grad_norm": 0.014260892033320325,
+      "learning_rate": 3.5266738274257074e-06,
+      "loss": 0.0,
+      "step": 10599
+    },
+    {
+      "epoch": 0.6831217374492492,
+      "grad_norm": 0.0028445670215190093,
+      "learning_rate": 3.5259577515216613e-06,
+      "loss": 0.0,
+      "step": 10600
+    },
+    {
+      "epoch": 0.6831861828961784,
+      "grad_norm": 0.0008771312919833636,
+      "learning_rate": 3.5252416756176156e-06,
+      "loss": 0.0,
+      "step": 10601
+    },
+    {
+      "epoch": 0.6832506283431076,
+      "grad_norm": 0.00018122247011554844,
+      "learning_rate": 3.52452559971357e-06,
+      "loss": 0.0,
+      "step": 10602
+    },
+    {
+      "epoch": 0.6833150737900368,
+      "grad_norm": 0.0008672090977392164,
+      "learning_rate": 3.523809523809524e-06,
+      "loss": 0.0,
+      "step": 10603
+    },
+    {
+      "epoch": 0.6833795192369659,
+      "grad_norm": 0.0035352795664838213,
+      "learning_rate": 3.5230934479054785e-06,
+      "loss": 0.0,
+      "step": 10604
+    },
+    {
+      "epoch": 0.6834439646838951,
+      "grad_norm": 0.0002724304476376662,
+      "learning_rate": 3.5223773720014328e-06,
+      "loss": 0.0,
+      "step": 10605
+    },
+    {
+      "epoch": 0.6835084101308243,
+      "grad_norm": 0.5370559665829012,
+      "learning_rate": 3.5216612960973866e-06,
+      "loss": 0.0063,
+      "step": 10606
+    },
+    {
+      "epoch": 0.6835728555777534,
+      "grad_norm": 0.0007439298862113308,
+      "learning_rate": 3.520945220193341e-06,
+      "loss": 0.0,
+      "step": 10607
+    },
+    {
+      "epoch": 0.6836373010246826,
+      "grad_norm": 0.009702558574781294,
+      "learning_rate": 3.520229144289295e-06,
+      "loss": 0.0,
+      "step": 10608
+    },
+    {
+      "epoch": 0.6837017464716117,
+      "grad_norm": 0.0011016534404801282,
+      "learning_rate": 3.519513068385249e-06,
+      "loss": 0.0,
+      "step": 10609
+    },
+    {
+      "epoch": 0.6837661919185409,
+      "grad_norm": 0.017491655293344472,
+      "learning_rate": 3.5187969924812034e-06,
+      "loss": 0.0015,
+      "step": 10610
+    },
+    {
+      "epoch": 0.6838306373654701,
+      "grad_norm": 0.173947823491574,
+      "learning_rate": 3.5180809165771573e-06,
+      "loss": 0.0015,
+      "step": 10611
+    },
+    {
+      "epoch": 0.6838950828123993,
+      "grad_norm": 0.0004411864460499935,
+      "learning_rate": 3.5173648406731116e-06,
+      "loss": 0.0,
+      "step": 10612
+    },
+    {
+      "epoch": 0.6839595282593285,
+      "grad_norm": 0.05571853331626954,
+      "learning_rate": 3.5166487647690655e-06,
+      "loss": 0.0004,
+      "step": 10613
+    },
+    {
+      "epoch": 0.6840239737062577,
+      "grad_norm": 0.0010041793745665642,
+      "learning_rate": 3.5159326888650198e-06,
+      "loss": 0.0,
+      "step": 10614
+    },
+    {
+      "epoch": 0.6840884191531869,
+      "grad_norm": 0.0983773975207399,
+      "learning_rate": 3.5152166129609745e-06,
+      "loss": 0.0002,
+      "step": 10615
+    },
+    {
+      "epoch": 0.684152864600116,
+      "grad_norm": 0.028895436137705906,
+      "learning_rate": 3.5145005370569284e-06,
+      "loss": 0.0,
+      "step": 10616
+    },
+    {
+      "epoch": 0.6842173100470452,
+      "grad_norm": 0.03385962597373734,
+      "learning_rate": 3.5137844611528827e-06,
+      "loss": 0.0,
+      "step": 10617
+    },
+    {
+      "epoch": 0.6842817554939743,
+      "grad_norm": 0.023507120372402705,
+      "learning_rate": 3.513068385248837e-06,
+      "loss": 0.0002,
+      "step": 10618
+    },
+    {
+      "epoch": 0.6843462009409035,
+      "grad_norm": 0.024889008607317492,
+      "learning_rate": 3.512352309344791e-06,
+      "loss": 0.0,
+      "step": 10619
+    },
+    {
+      "epoch": 0.6844106463878327,
+      "grad_norm": 0.0014179847128868335,
+      "learning_rate": 3.511636233440745e-06,
+      "loss": 0.0,
+      "step": 10620
+    },
+    {
+      "epoch": 0.6844750918347619,
+      "grad_norm": 0.013406477162943287,
+      "learning_rate": 3.510920157536699e-06,
+      "loss": 0.0,
+      "step": 10621
+    },
+    {
+      "epoch": 0.684539537281691,
+      "grad_norm": 0.28027808953122846,
+      "learning_rate": 3.5102040816326533e-06,
+      "loss": 0.0023,
+      "step": 10622
+    },
+    {
+      "epoch": 0.6846039827286202,
+      "grad_norm": 0.00861103756734106,
+      "learning_rate": 3.509488005728607e-06,
+      "loss": 0.0,
+      "step": 10623
+    },
+    {
+      "epoch": 0.6846684281755494,
+      "grad_norm": 0.09979447762360293,
+      "learning_rate": 3.5087719298245615e-06,
+      "loss": 0.0017,
+      "step": 10624
+    },
+    {
+      "epoch": 0.6847328736224786,
+      "grad_norm": 0.08866655842139125,
+      "learning_rate": 3.5080558539205158e-06,
+      "loss": 0.0017,
+      "step": 10625
+    },
+    {
+      "epoch": 0.6847973190694078,
+      "grad_norm": 0.09441540926634437,
+      "learning_rate": 3.5073397780164705e-06,
+      "loss": 0.0024,
+      "step": 10626
+    },
+    {
+      "epoch": 0.684861764516337,
+      "grad_norm": 0.10564041736905677,
+      "learning_rate": 3.5066237021124244e-06,
+      "loss": 0.0009,
+      "step": 10627
+    },
+    {
+      "epoch": 0.6849262099632661,
+      "grad_norm": 0.00668683870026609,
+      "learning_rate": 3.5059076262083787e-06,
+      "loss": 0.0,
+      "step": 10628
+    },
+    {
+      "epoch": 0.6849906554101952,
+      "grad_norm": 0.007881264256661852,
+      "learning_rate": 3.5051915503043326e-06,
+      "loss": 0.0,
+      "step": 10629
+    },
+    {
+      "epoch": 0.6850551008571244,
+      "grad_norm": 0.009405162217880085,
+      "learning_rate": 3.504475474400287e-06,
+      "loss": 0.0,
+      "step": 10630
+    },
+    {
+      "epoch": 0.6851195463040536,
+      "grad_norm": 0.002327160651186275,
+      "learning_rate": 3.5037593984962407e-06,
+      "loss": 0.0,
+      "step": 10631
+    },
+    {
+      "epoch": 0.6851839917509828,
+      "grad_norm": 0.02192513837913387,
+      "learning_rate": 3.503043322592195e-06,
+      "loss": 0.0,
+      "step": 10632
+    },
+    {
+      "epoch": 0.685248437197912,
+      "grad_norm": 4.1137151297167716e-05,
+      "learning_rate": 3.5023272466881493e-06,
+      "loss": 0.0,
+      "step": 10633
+    },
+    {
+      "epoch": 0.6853128826448411,
+      "grad_norm": 0.26734327883653797,
+      "learning_rate": 3.501611170784103e-06,
+      "loss": 0.0002,
+      "step": 10634
+    },
+    {
+      "epoch": 0.6853773280917703,
+      "grad_norm": 0.03286824451174531,
+      "learning_rate": 3.5008950948800575e-06,
+      "loss": 0.0001,
+      "step": 10635
+    },
+    {
+      "epoch": 0.6854417735386995,
+      "grad_norm": 0.03350731756160981,
+      "learning_rate": 3.5001790189760114e-06,
+      "loss": 0.0001,
+      "step": 10636
+    },
+    {
+      "epoch": 0.6855062189856287,
+      "grad_norm": 0.003075251300806112,
+      "learning_rate": 3.4994629430719657e-06,
+      "loss": 0.0,
+      "step": 10637
+    },
+    {
+      "epoch": 0.6855706644325579,
+      "grad_norm": 0.010674466356599537,
+      "learning_rate": 3.4987468671679204e-06,
+      "loss": 0.0,
+      "step": 10638
+    },
+    {
+      "epoch": 0.6856351098794871,
+      "grad_norm": 0.013242521425976893,
+      "learning_rate": 3.4980307912638743e-06,
+      "loss": 0.0,
+      "step": 10639
+    },
+    {
+      "epoch": 0.6856995553264162,
+      "grad_norm": 0.23402005179956256,
+      "learning_rate": 3.4973147153598286e-06,
+      "loss": 0.0005,
+      "step": 10640
+    },
+    {
+      "epoch": 0.6857640007733453,
+      "grad_norm": 0.006823108867421351,
+      "learning_rate": 3.496598639455783e-06,
+      "loss": 0.0,
+      "step": 10641
+    },
+    {
+      "epoch": 0.6858284462202745,
+      "grad_norm": 0.005193235394536426,
+      "learning_rate": 3.4958825635517367e-06,
+      "loss": 0.0,
+      "step": 10642
+    },
+    {
+      "epoch": 0.6858928916672037,
+      "grad_norm": 0.05758601897356446,
+      "learning_rate": 3.495166487647691e-06,
+      "loss": 0.0001,
+      "step": 10643
+    },
+    {
+      "epoch": 0.6859573371141329,
+      "grad_norm": 0.0002420785467362924,
+      "learning_rate": 3.494450411743645e-06,
+      "loss": 0.0,
+      "step": 10644
+    },
+    {
+      "epoch": 0.686021782561062,
+      "grad_norm": 0.013903106800684462,
+      "learning_rate": 3.4937343358395992e-06,
+      "loss": 0.0,
+      "step": 10645
+    },
+    {
+      "epoch": 0.6860862280079912,
+      "grad_norm": 0.0013727656492007864,
+      "learning_rate": 3.4930182599355535e-06,
+      "loss": 0.0,
+      "step": 10646
+    },
+    {
+      "epoch": 0.6861506734549204,
+      "grad_norm": 0.005639060991519206,
+      "learning_rate": 3.4923021840315074e-06,
+      "loss": 0.0,
+      "step": 10647
+    },
+    {
+      "epoch": 0.6862151189018496,
+      "grad_norm": 0.00012059895211766672,
+      "learning_rate": 3.4915861081274617e-06,
+      "loss": 0.0,
+      "step": 10648
+    },
+    {
+      "epoch": 0.6862795643487788,
+      "grad_norm": 0.000295021304383494,
+      "learning_rate": 3.4908700322234156e-06,
+      "loss": 0.0,
+      "step": 10649
+    },
+    {
+      "epoch": 0.686344009795708,
+      "grad_norm": 0.006692717847118855,
+      "learning_rate": 3.4901539563193703e-06,
+      "loss": 0.0,
+      "step": 10650
+    },
+    {
+      "epoch": 0.6864084552426372,
+      "grad_norm": 0.0008308969576633938,
+      "learning_rate": 3.4894378804153246e-06,
+      "loss": 0.0,
+      "step": 10651
+    },
+    {
+      "epoch": 0.6864729006895662,
+      "grad_norm": 0.02864746925349982,
+      "learning_rate": 3.4887218045112785e-06,
+      "loss": 0.0001,
+      "step": 10652
+    },
+    {
+      "epoch": 0.6865373461364954,
+      "grad_norm": 0.0004295749878789247,
+      "learning_rate": 3.4880057286072328e-06,
+      "loss": 0.0,
+      "step": 10653
+    },
+    {
+      "epoch": 0.6866017915834246,
+      "grad_norm": 0.0003032108960250809,
+      "learning_rate": 3.487289652703187e-06,
+      "loss": 0.0,
+      "step": 10654
+    },
+    {
+      "epoch": 0.6866662370303538,
+      "grad_norm": 0.0032508764152708682,
+      "learning_rate": 3.486573576799141e-06,
+      "loss": 0.0,
+      "step": 10655
+    },
+    {
+      "epoch": 0.686730682477283,
+      "grad_norm": 0.3218022338317263,
+      "learning_rate": 3.4858575008950952e-06,
+      "loss": 0.0021,
+      "step": 10656
+    },
+    {
+      "epoch": 0.6867951279242122,
+      "grad_norm": 0.013413374439972325,
+      "learning_rate": 3.485141424991049e-06,
+      "loss": 0.0001,
+      "step": 10657
+    },
+    {
+      "epoch": 0.6868595733711413,
+      "grad_norm": 0.0022459060523140166,
+      "learning_rate": 3.4844253490870034e-06,
+      "loss": 0.0,
+      "step": 10658
+    },
+    {
+      "epoch": 0.6869240188180705,
+      "grad_norm": 0.0025463195329138914,
+      "learning_rate": 3.4837092731829573e-06,
+      "loss": 0.0,
+      "step": 10659
+    },
+    {
+      "epoch": 0.6869884642649997,
+      "grad_norm": 0.07051437260233849,
+      "learning_rate": 3.4829931972789116e-06,
+      "loss": 0.0017,
+      "step": 10660
+    },
+    {
+      "epoch": 0.6870529097119289,
+      "grad_norm": 0.002290644215445257,
+      "learning_rate": 3.4822771213748663e-06,
+      "loss": 0.0,
+      "step": 10661
+    },
+    {
+      "epoch": 0.6871173551588581,
+      "grad_norm": 0.00039243013710876817,
+      "learning_rate": 3.4815610454708206e-06,
+      "loss": 0.0,
+      "step": 10662
+    },
+    {
+      "epoch": 0.6871818006057872,
+      "grad_norm": 0.001251349809574215,
+      "learning_rate": 3.4808449695667745e-06,
+      "loss": 0.0,
+      "step": 10663
+    },
+    {
+      "epoch": 0.6872462460527163,
+      "grad_norm": 0.02047947219666481,
+      "learning_rate": 3.4801288936627288e-06,
+      "loss": 0.0001,
+      "step": 10664
+    },
+    {
+      "epoch": 0.6873106914996455,
+      "grad_norm": 0.030119726822941218,
+      "learning_rate": 3.4794128177586826e-06,
+      "loss": 0.0001,
+      "step": 10665
+    },
+    {
+      "epoch": 0.6873751369465747,
+      "grad_norm": 0.005128161573191081,
+      "learning_rate": 3.478696741854637e-06,
+      "loss": 0.0,
+      "step": 10666
+    },
+    {
+      "epoch": 0.6874395823935039,
+      "grad_norm": 0.06837941625274693,
+      "learning_rate": 3.477980665950591e-06,
+      "loss": 0.0001,
+      "step": 10667
+    },
+    {
+      "epoch": 0.6875040278404331,
+      "grad_norm": 0.03753905404742778,
+      "learning_rate": 3.477264590046545e-06,
+      "loss": 0.0,
+      "step": 10668
+    },
+    {
+      "epoch": 0.6875684732873623,
+      "grad_norm": 0.0034440145309386637,
+      "learning_rate": 3.4765485141424994e-06,
+      "loss": 0.0,
+      "step": 10669
+    },
+    {
+      "epoch": 0.6876329187342914,
+      "grad_norm": 0.14563323382401655,
+      "learning_rate": 3.4758324382384533e-06,
+      "loss": 0.0002,
+      "step": 10670
+    },
+    {
+      "epoch": 0.6876973641812206,
+      "grad_norm": 0.007722652524045524,
+      "learning_rate": 3.4751163623344076e-06,
+      "loss": 0.0,
+      "step": 10671
+    },
+    {
+      "epoch": 0.6877618096281498,
+      "grad_norm": 0.0032592152884733613,
+      "learning_rate": 3.4744002864303615e-06,
+      "loss": 0.0,
+      "step": 10672
+    },
+    {
+      "epoch": 0.687826255075079,
+      "grad_norm": 0.008455749548395619,
+      "learning_rate": 3.473684210526316e-06,
+      "loss": 0.0,
+      "step": 10673
+    },
+    {
+      "epoch": 0.6878907005220081,
+      "grad_norm": 0.09676174091931598,
+      "learning_rate": 3.4729681346222705e-06,
+      "loss": 0.0003,
+      "step": 10674
+    },
+    {
+      "epoch": 0.6879551459689373,
+      "grad_norm": 0.0010924824758405294,
+      "learning_rate": 3.4722520587182244e-06,
+      "loss": 0.0,
+      "step": 10675
+    },
+    {
+      "epoch": 0.6880195914158664,
+      "grad_norm": 0.010774355308097124,
+      "learning_rate": 3.4715359828141787e-06,
+      "loss": 0.0016,
+      "step": 10676
+    },
+    {
+      "epoch": 0.6880840368627956,
+      "grad_norm": 0.01795689798363324,
+      "learning_rate": 3.470819906910133e-06,
+      "loss": 0.0,
+      "step": 10677
+    },
+    {
+      "epoch": 0.6881484823097248,
+      "grad_norm": 0.0027896545235474085,
+      "learning_rate": 3.470103831006087e-06,
+      "loss": 0.0,
+      "step": 10678
+    },
+    {
+      "epoch": 0.688212927756654,
+      "grad_norm": 0.22212767183159907,
+      "learning_rate": 3.469387755102041e-06,
+      "loss": 0.0005,
+      "step": 10679
+    },
+    {
+      "epoch": 0.6882773732035832,
+      "grad_norm": 0.0003912980966343014,
+      "learning_rate": 3.468671679197995e-06,
+      "loss": 0.0,
+      "step": 10680
+    },
+    {
+      "epoch": 0.6883418186505124,
+      "grad_norm": 0.0005156493416187653,
+      "learning_rate": 3.4679556032939493e-06,
+      "loss": 0.0,
+      "step": 10681
+    },
+    {
+      "epoch": 0.6884062640974415,
+      "grad_norm": 0.021590134423169444,
+      "learning_rate": 3.4672395273899036e-06,
+      "loss": 0.0,
+      "step": 10682
+    },
+    {
+      "epoch": 0.6884707095443707,
+      "grad_norm": 0.023141501828001405,
+      "learning_rate": 3.4665234514858575e-06,
+      "loss": 0.0,
+      "step": 10683
+    },
+    {
+      "epoch": 0.6885351549912999,
+      "grad_norm": 0.007118016829719887,
+      "learning_rate": 3.4658073755818118e-06,
+      "loss": 0.0001,
+      "step": 10684
+    },
+    {
+      "epoch": 0.688599600438229,
+      "grad_norm": 0.25918209531883035,
+      "learning_rate": 3.4650912996777665e-06,
+      "loss": 0.0005,
+      "step": 10685
+    },
+    {
+      "epoch": 0.6886640458851582,
+      "grad_norm": 0.004237713823522767,
+      "learning_rate": 3.4643752237737204e-06,
+      "loss": 0.0,
+      "step": 10686
+    },
+    {
+      "epoch": 0.6887284913320874,
+      "grad_norm": 0.038278050860711475,
+      "learning_rate": 3.4636591478696747e-06,
+      "loss": 0.0,
+      "step": 10687
+    },
+    {
+      "epoch": 0.6887929367790165,
+      "grad_norm": 0.000426746379782578,
+      "learning_rate": 3.4629430719656286e-06,
+      "loss": 0.0,
+      "step": 10688
+    },
+    {
+      "epoch": 0.6888573822259457,
+      "grad_norm": 0.0034051439268017562,
+      "learning_rate": 3.462226996061583e-06,
+      "loss": 0.0,
+      "step": 10689
+    },
+    {
+      "epoch": 0.6889218276728749,
+      "grad_norm": 0.018312988913133918,
+      "learning_rate": 3.461510920157537e-06,
+      "loss": 0.0,
+      "step": 10690
+    },
+    {
+      "epoch": 0.6889862731198041,
+      "grad_norm": 0.0008704069031677928,
+      "learning_rate": 3.460794844253491e-06,
+      "loss": 0.0,
+      "step": 10691
+    },
+    {
+      "epoch": 0.6890507185667333,
+      "grad_norm": 0.00011138326842831115,
+      "learning_rate": 3.4600787683494453e-06,
+      "loss": 0.0,
+      "step": 10692
+    },
+    {
+      "epoch": 0.6891151640136625,
+      "grad_norm": 6.343770674378296e-05,
+      "learning_rate": 3.459362692445399e-06,
+      "loss": 0.0,
+      "step": 10693
+    },
+    {
+      "epoch": 0.6891796094605916,
+      "grad_norm": 0.000850738362306896,
+      "learning_rate": 3.4586466165413535e-06,
+      "loss": 0.0,
+      "step": 10694
+    },
+    {
+      "epoch": 0.6892440549075208,
+      "grad_norm": 0.15241558875373135,
+      "learning_rate": 3.457930540637308e-06,
+      "loss": 0.0005,
+      "step": 10695
+    },
+    {
+      "epoch": 0.6893085003544499,
+      "grad_norm": 0.0037602512408664913,
+      "learning_rate": 3.457214464733262e-06,
+      "loss": 0.0,
+      "step": 10696
+    },
+    {
+      "epoch": 0.6893729458013791,
+      "grad_norm": 0.16921284611583368,
+      "learning_rate": 3.4564983888292164e-06,
+      "loss": 0.0014,
+      "step": 10697
+    },
+    {
+      "epoch": 0.6894373912483083,
+      "grad_norm": 0.04126671726871299,
+      "learning_rate": 3.4557823129251707e-06,
+      "loss": 0.0001,
+      "step": 10698
+    },
+    {
+      "epoch": 0.6895018366952375,
+      "grad_norm": 0.24241588960094448,
+      "learning_rate": 3.4550662370211246e-06,
+      "loss": 0.0006,
+      "step": 10699
+    },
+    {
+      "epoch": 0.6895662821421666,
+      "grad_norm": 0.0011653326035232546,
+      "learning_rate": 3.454350161117079e-06,
+      "loss": 0.0,
+      "step": 10700
+    },
+    {
+      "epoch": 0.6896307275890958,
+      "grad_norm": 0.0053412736709698705,
+      "learning_rate": 3.4536340852130327e-06,
+      "loss": 0.0001,
+      "step": 10701
+    },
+    {
+      "epoch": 0.689695173036025,
+      "grad_norm": 0.0027695711960786054,
+      "learning_rate": 3.452918009308987e-06,
+      "loss": 0.0,
+      "step": 10702
+    },
+    {
+      "epoch": 0.6897596184829542,
+      "grad_norm": 0.000903931335369121,
+      "learning_rate": 3.4522019334049413e-06,
+      "loss": 0.0,
+      "step": 10703
+    },
+    {
+      "epoch": 0.6898240639298834,
+      "grad_norm": 0.0032606382890580307,
+      "learning_rate": 3.4514858575008952e-06,
+      "loss": 0.0,
+      "step": 10704
+    },
+    {
+      "epoch": 0.6898885093768126,
+      "grad_norm": 0.0019891495099166236,
+      "learning_rate": 3.4507697815968495e-06,
+      "loss": 0.0,
+      "step": 10705
+    },
+    {
+      "epoch": 0.6899529548237417,
+      "grad_norm": 0.014469597284214894,
+      "learning_rate": 3.4500537056928034e-06,
+      "loss": 0.0001,
+      "step": 10706
+    },
+    {
+      "epoch": 0.6900174002706708,
+      "grad_norm": 0.0036697474437936265,
+      "learning_rate": 3.4493376297887577e-06,
+      "loss": 0.0001,
+      "step": 10707
+    },
+    {
+      "epoch": 0.6900818457176,
+      "grad_norm": 0.001639690389296713,
+      "learning_rate": 3.4486215538847124e-06,
+      "loss": 0.0,
+      "step": 10708
+    },
+    {
+      "epoch": 0.6901462911645292,
+      "grad_norm": 0.001005555053943566,
+      "learning_rate": 3.4479054779806663e-06,
+      "loss": 0.0,
+      "step": 10709
+    },
+    {
+      "epoch": 0.6902107366114584,
+      "grad_norm": 0.09674031832156237,
+      "learning_rate": 3.4471894020766206e-06,
+      "loss": 0.0002,
+      "step": 10710
+    },
+    {
+      "epoch": 0.6902751820583876,
+      "grad_norm": 0.17584697766118745,
+      "learning_rate": 3.446473326172575e-06,
+      "loss": 0.0001,
+      "step": 10711
+    },
+    {
+      "epoch": 0.6903396275053167,
+      "grad_norm": 0.0003299470982010914,
+      "learning_rate": 3.4457572502685288e-06,
+      "loss": 0.0,
+      "step": 10712
+    },
+    {
+      "epoch": 0.6904040729522459,
+      "grad_norm": 0.0012676769435483054,
+      "learning_rate": 3.445041174364483e-06,
+      "loss": 0.0,
+      "step": 10713
+    },
+    {
+      "epoch": 0.6904685183991751,
+      "grad_norm": 0.0001250973702220075,
+      "learning_rate": 3.444325098460437e-06,
+      "loss": 0.0,
+      "step": 10714
+    },
+    {
+      "epoch": 0.6905329638461043,
+      "grad_norm": 0.0016230453999726533,
+      "learning_rate": 3.4436090225563912e-06,
+      "loss": 0.0,
+      "step": 10715
+    },
+    {
+      "epoch": 0.6905974092930335,
+      "grad_norm": 0.7712714178984845,
+      "learning_rate": 3.442892946652345e-06,
+      "loss": 0.0016,
+      "step": 10716
+    },
+    {
+      "epoch": 0.6906618547399627,
+      "grad_norm": 0.012430636333727142,
+      "learning_rate": 3.4421768707482994e-06,
+      "loss": 0.0,
+      "step": 10717
+    },
+    {
+      "epoch": 0.6907263001868919,
+      "grad_norm": 0.00038744205676761773,
+      "learning_rate": 3.4414607948442537e-06,
+      "loss": 0.0,
+      "step": 10718
+    },
+    {
+      "epoch": 0.6907907456338209,
+      "grad_norm": 0.001783591718322277,
+      "learning_rate": 3.4407447189402076e-06,
+      "loss": 0.0,
+      "step": 10719
+    },
+    {
+      "epoch": 0.6908551910807501,
+      "grad_norm": 0.00012095653921781966,
+      "learning_rate": 3.4400286430361623e-06,
+      "loss": 0.0,
+      "step": 10720
+    },
+    {
+      "epoch": 0.6909196365276793,
+      "grad_norm": 1.0810774582805374,
+      "learning_rate": 3.4393125671321166e-06,
+      "loss": 0.0064,
+      "step": 10721
+    },
+    {
+      "epoch": 0.6909840819746085,
+      "grad_norm": 0.006299809688504188,
+      "learning_rate": 3.4385964912280705e-06,
+      "loss": 0.0,
+      "step": 10722
+    },
+    {
+      "epoch": 0.6910485274215377,
+      "grad_norm": 0.002780990180928392,
+      "learning_rate": 3.4378804153240248e-06,
+      "loss": 0.0,
+      "step": 10723
+    },
+    {
+      "epoch": 0.6911129728684668,
+      "grad_norm": 0.057636782983209595,
+      "learning_rate": 3.4371643394199786e-06,
+      "loss": 0.0001,
+      "step": 10724
+    },
+    {
+      "epoch": 0.691177418315396,
+      "grad_norm": 0.0008725221319251466,
+      "learning_rate": 3.436448263515933e-06,
+      "loss": 0.0,
+      "step": 10725
+    },
+    {
+      "epoch": 0.6912418637623252,
+      "grad_norm": 0.026326179625309587,
+      "learning_rate": 3.4357321876118872e-06,
+      "loss": 0.0002,
+      "step": 10726
+    },
+    {
+      "epoch": 0.6913063092092544,
+      "grad_norm": 0.521066305189868,
+      "learning_rate": 3.435016111707841e-06,
+      "loss": 0.0036,
+      "step": 10727
+    },
+    {
+      "epoch": 0.6913707546561836,
+      "grad_norm": 0.00021508146730564562,
+      "learning_rate": 3.4343000358037954e-06,
+      "loss": 0.0,
+      "step": 10728
+    },
+    {
+      "epoch": 0.6914352001031128,
+      "grad_norm": 5.2332481397573e-05,
+      "learning_rate": 3.4335839598997493e-06,
+      "loss": 0.0,
+      "step": 10729
+    },
+    {
+      "epoch": 0.6914996455500418,
+      "grad_norm": 0.0025788359544813106,
+      "learning_rate": 3.4328678839957036e-06,
+      "loss": 0.0,
+      "step": 10730
+    },
+    {
+      "epoch": 0.691564090996971,
+      "grad_norm": 0.7060573678077907,
+      "learning_rate": 3.4321518080916583e-06,
+      "loss": 0.0049,
+      "step": 10731
+    },
+    {
+      "epoch": 0.6916285364439002,
+      "grad_norm": 0.44814332690464703,
+      "learning_rate": 3.431435732187612e-06,
+      "loss": 0.0036,
+      "step": 10732
+    },
+    {
+      "epoch": 0.6916929818908294,
+      "grad_norm": 0.3525222816594591,
+      "learning_rate": 3.4307196562835665e-06,
+      "loss": 0.0026,
+      "step": 10733
+    },
+    {
+      "epoch": 0.6917574273377586,
+      "grad_norm": 0.10370206238152077,
+      "learning_rate": 3.4300035803795208e-06,
+      "loss": 0.0001,
+      "step": 10734
+    },
+    {
+      "epoch": 0.6918218727846878,
+      "grad_norm": 0.00335698601833503,
+      "learning_rate": 3.4292875044754747e-06,
+      "loss": 0.0,
+      "step": 10735
+    },
+    {
+      "epoch": 0.691886318231617,
+      "grad_norm": 0.006926952458820839,
+      "learning_rate": 3.428571428571429e-06,
+      "loss": 0.0001,
+      "step": 10736
+    },
+    {
+      "epoch": 0.6919507636785461,
+      "grad_norm": 0.0003903578948580782,
+      "learning_rate": 3.427855352667383e-06,
+      "loss": 0.0,
+      "step": 10737
+    },
+    {
+      "epoch": 0.6920152091254753,
+      "grad_norm": 0.06039492128259943,
+      "learning_rate": 3.427139276763337e-06,
+      "loss": 0.0016,
+      "step": 10738
+    },
+    {
+      "epoch": 0.6920796545724045,
+      "grad_norm": 0.21314970053730722,
+      "learning_rate": 3.4264232008592914e-06,
+      "loss": 0.0006,
+      "step": 10739
+    },
+    {
+      "epoch": 0.6921441000193337,
+      "grad_norm": 0.009887192339525711,
+      "learning_rate": 3.4257071249552453e-06,
+      "loss": 0.0001,
+      "step": 10740
+    },
+    {
+      "epoch": 0.6922085454662628,
+      "grad_norm": 0.0023324859903385378,
+      "learning_rate": 3.4249910490511996e-06,
+      "loss": 0.0,
+      "step": 10741
+    },
+    {
+      "epoch": 0.6922729909131919,
+      "grad_norm": 0.0017754964193178054,
+      "learning_rate": 3.4242749731471535e-06,
+      "loss": 0.0,
+      "step": 10742
+    },
+    {
+      "epoch": 0.6923374363601211,
+      "grad_norm": 0.11862733365427512,
+      "learning_rate": 3.423558897243108e-06,
+      "loss": 0.0002,
+      "step": 10743
+    },
+    {
+      "epoch": 0.6924018818070503,
+      "grad_norm": 0.12538965409042613,
+      "learning_rate": 3.4228428213390625e-06,
+      "loss": 0.0017,
+      "step": 10744
+    },
+    {
+      "epoch": 0.6924663272539795,
+      "grad_norm": 0.1790398036759841,
+      "learning_rate": 3.4221267454350164e-06,
+      "loss": 0.0012,
+      "step": 10745
+    },
+    {
+      "epoch": 0.6925307727009087,
+      "grad_norm": 0.016628649121499365,
+      "learning_rate": 3.4214106695309707e-06,
+      "loss": 0.0001,
+      "step": 10746
+    },
+    {
+      "epoch": 0.6925952181478379,
+      "grad_norm": 0.003371619886892593,
+      "learning_rate": 3.420694593626925e-06,
+      "loss": 0.0,
+      "step": 10747
+    },
+    {
+      "epoch": 0.692659663594767,
+      "grad_norm": 0.24389006154688553,
+      "learning_rate": 3.419978517722879e-06,
+      "loss": 0.0006,
+      "step": 10748
+    },
+    {
+      "epoch": 0.6927241090416962,
+      "grad_norm": 0.14232802337674683,
+      "learning_rate": 3.419262441818833e-06,
+      "loss": 0.0003,
+      "step": 10749
+    },
+    {
+      "epoch": 0.6927885544886254,
+      "grad_norm": 0.019736813074936816,
+      "learning_rate": 3.418546365914787e-06,
+      "loss": 0.0,
+      "step": 10750
+    },
+    {
+      "epoch": 0.6928529999355546,
+      "grad_norm": 0.05097154578008529,
+      "learning_rate": 3.4178302900107413e-06,
+      "loss": 0.0,
+      "step": 10751
+    },
+    {
+      "epoch": 0.6929174453824837,
+      "grad_norm": 0.00107446008995132,
+      "learning_rate": 3.4171142141066956e-06,
+      "loss": 0.0,
+      "step": 10752
+    },
+    {
+      "epoch": 0.6929818908294129,
+      "grad_norm": 0.012435354325548929,
+      "learning_rate": 3.4163981382026495e-06,
+      "loss": 0.0,
+      "step": 10753
+    },
+    {
+      "epoch": 0.693046336276342,
+      "grad_norm": 0.021957133028454156,
+      "learning_rate": 3.415682062298604e-06,
+      "loss": 0.0,
+      "step": 10754
+    },
+    {
+      "epoch": 0.6931107817232712,
+      "grad_norm": 0.0002480213227601337,
+      "learning_rate": 3.4149659863945585e-06,
+      "loss": 0.0,
+      "step": 10755
+    },
+    {
+      "epoch": 0.6931752271702004,
+      "grad_norm": 0.0623140810818036,
+      "learning_rate": 3.4142499104905124e-06,
+      "loss": 0.0,
+      "step": 10756
+    },
+    {
+      "epoch": 0.6932396726171296,
+      "grad_norm": 0.0017141559011994213,
+      "learning_rate": 3.4135338345864667e-06,
+      "loss": 0.0,
+      "step": 10757
+    },
+    {
+      "epoch": 0.6933041180640588,
+      "grad_norm": 0.014615192039343805,
+      "learning_rate": 3.4128177586824206e-06,
+      "loss": 0.0,
+      "step": 10758
+    },
+    {
+      "epoch": 0.693368563510988,
+      "grad_norm": 0.0017439914087220587,
+      "learning_rate": 3.412101682778375e-06,
+      "loss": 0.0,
+      "step": 10759
+    },
+    {
+      "epoch": 0.6934330089579172,
+      "grad_norm": 0.0012153270549224956,
+      "learning_rate": 3.411385606874329e-06,
+      "loss": 0.0,
+      "step": 10760
+    },
+    {
+      "epoch": 0.6934974544048463,
+      "grad_norm": 0.014303665596357573,
+      "learning_rate": 3.410669530970283e-06,
+      "loss": 0.0,
+      "step": 10761
+    },
+    {
+      "epoch": 0.6935618998517755,
+      "grad_norm": 7.439093973990937e-05,
+      "learning_rate": 3.4099534550662373e-06,
+      "loss": 0.0,
+      "step": 10762
+    },
+    {
+      "epoch": 0.6936263452987046,
+      "grad_norm": 0.19960271073953648,
+      "learning_rate": 3.4092373791621912e-06,
+      "loss": 0.0003,
+      "step": 10763
+    },
+    {
+      "epoch": 0.6936907907456338,
+      "grad_norm": 0.018784679469903552,
+      "learning_rate": 3.4085213032581455e-06,
+      "loss": 0.0001,
+      "step": 10764
+    },
+    {
+      "epoch": 0.693755236192563,
+      "grad_norm": 0.06285531773678558,
+      "learning_rate": 3.4078052273540994e-06,
+      "loss": 0.0003,
+      "step": 10765
+    },
+    {
+      "epoch": 0.6938196816394921,
+      "grad_norm": 0.02711289526911007,
+      "learning_rate": 3.407089151450054e-06,
+      "loss": 0.0,
+      "step": 10766
+    },
+    {
+      "epoch": 0.6938841270864213,
+      "grad_norm": 0.005589801722090483,
+      "learning_rate": 3.4063730755460084e-06,
+      "loss": 0.0,
+      "step": 10767
+    },
+    {
+      "epoch": 0.6939485725333505,
+      "grad_norm": 0.012897300775219056,
+      "learning_rate": 3.4056569996419627e-06,
+      "loss": 0.0001,
+      "step": 10768
+    },
+    {
+      "epoch": 0.6940130179802797,
+      "grad_norm": 0.01099284164409652,
+      "learning_rate": 3.4049409237379166e-06,
+      "loss": 0.0001,
+      "step": 10769
+    },
+    {
+      "epoch": 0.6940774634272089,
+      "grad_norm": 0.0016891558399507187,
+      "learning_rate": 3.404224847833871e-06,
+      "loss": 0.0,
+      "step": 10770
+    },
+    {
+      "epoch": 0.6941419088741381,
+      "grad_norm": 0.006634654576318617,
+      "learning_rate": 3.4035087719298248e-06,
+      "loss": 0.0,
+      "step": 10771
+    },
+    {
+      "epoch": 0.6942063543210673,
+      "grad_norm": 0.13746926170005874,
+      "learning_rate": 3.402792696025779e-06,
+      "loss": 0.0001,
+      "step": 10772
+    },
+    {
+      "epoch": 0.6942707997679964,
+      "grad_norm": 0.02645937380972612,
+      "learning_rate": 3.402076620121733e-06,
+      "loss": 0.0001,
+      "step": 10773
+    },
+    {
+      "epoch": 0.6943352452149255,
+      "grad_norm": 0.0033664323788412985,
+      "learning_rate": 3.4013605442176872e-06,
+      "loss": 0.0,
+      "step": 10774
+    },
+    {
+      "epoch": 0.6943996906618547,
+      "grad_norm": 0.0022188361754276117,
+      "learning_rate": 3.4006444683136415e-06,
+      "loss": 0.0,
+      "step": 10775
+    },
+    {
+      "epoch": 0.6944641361087839,
+      "grad_norm": 0.0004594087068661742,
+      "learning_rate": 3.3999283924095954e-06,
+      "loss": 0.0,
+      "step": 10776
+    },
+    {
+      "epoch": 0.6945285815557131,
+      "grad_norm": 0.0032890205841026223,
+      "learning_rate": 3.3992123165055497e-06,
+      "loss": 0.0,
+      "step": 10777
+    },
+    {
+      "epoch": 0.6945930270026422,
+      "grad_norm": 0.07355642935563832,
+      "learning_rate": 3.3984962406015044e-06,
+      "loss": 0.0001,
+      "step": 10778
+    },
+    {
+      "epoch": 0.6946574724495714,
+      "grad_norm": 0.019127169783181307,
+      "learning_rate": 3.3977801646974583e-06,
+      "loss": 0.0,
+      "step": 10779
+    },
+    {
+      "epoch": 0.6947219178965006,
+      "grad_norm": 0.0010884441376112783,
+      "learning_rate": 3.3970640887934126e-06,
+      "loss": 0.0,
+      "step": 10780
+    },
+    {
+      "epoch": 0.6947863633434298,
+      "grad_norm": 0.0023750329760530457,
+      "learning_rate": 3.3963480128893665e-06,
+      "loss": 0.0,
+      "step": 10781
+    },
+    {
+      "epoch": 0.694850808790359,
+      "grad_norm": 0.0007830149276151957,
+      "learning_rate": 3.3956319369853208e-06,
+      "loss": 0.0,
+      "step": 10782
+    },
+    {
+      "epoch": 0.6949152542372882,
+      "grad_norm": 0.04364455540834853,
+      "learning_rate": 3.394915861081275e-06,
+      "loss": 0.0001,
+      "step": 10783
+    },
+    {
+      "epoch": 0.6949796996842174,
+      "grad_norm": 0.016637574609427337,
+      "learning_rate": 3.394199785177229e-06,
+      "loss": 0.0002,
+      "step": 10784
+    },
+    {
+      "epoch": 0.6950441451311464,
+      "grad_norm": 0.002434681407404538,
+      "learning_rate": 3.3934837092731832e-06,
+      "loss": 0.0,
+      "step": 10785
+    },
+    {
+      "epoch": 0.6951085905780756,
+      "grad_norm": 0.006948871061763611,
+      "learning_rate": 3.392767633369137e-06,
+      "loss": 0.0,
+      "step": 10786
+    },
+    {
+      "epoch": 0.6951730360250048,
+      "grad_norm": 0.004640961971091515,
+      "learning_rate": 3.3920515574650914e-06,
+      "loss": 0.0,
+      "step": 10787
+    },
+    {
+      "epoch": 0.695237481471934,
+      "grad_norm": 0.00041903898312994556,
+      "learning_rate": 3.3913354815610457e-06,
+      "loss": 0.0,
+      "step": 10788
+    },
+    {
+      "epoch": 0.6953019269188632,
+      "grad_norm": 0.053400623619324995,
+      "learning_rate": 3.3906194056569996e-06,
+      "loss": 0.0001,
+      "step": 10789
+    },
+    {
+      "epoch": 0.6953663723657924,
+      "grad_norm": 0.0006922881994538185,
+      "learning_rate": 3.3899033297529543e-06,
+      "loss": 0.0,
+      "step": 10790
+    },
+    {
+      "epoch": 0.6954308178127215,
+      "grad_norm": 0.009405572997517235,
+      "learning_rate": 3.3891872538489086e-06,
+      "loss": 0.0,
+      "step": 10791
+    },
+    {
+      "epoch": 0.6954952632596507,
+      "grad_norm": 0.00023536262169622017,
+      "learning_rate": 3.3884711779448625e-06,
+      "loss": 0.0,
+      "step": 10792
+    },
+    {
+      "epoch": 0.6955597087065799,
+      "grad_norm": 0.000701178056448638,
+      "learning_rate": 3.3877551020408168e-06,
+      "loss": 0.0,
+      "step": 10793
+    },
+    {
+      "epoch": 0.6956241541535091,
+      "grad_norm": 0.04493621669798482,
+      "learning_rate": 3.3870390261367707e-06,
+      "loss": 0.0001,
+      "step": 10794
+    },
+    {
+      "epoch": 0.6956885996004383,
+      "grad_norm": 0.03422308278740534,
+      "learning_rate": 3.386322950232725e-06,
+      "loss": 0.0,
+      "step": 10795
+    },
+    {
+      "epoch": 0.6957530450473675,
+      "grad_norm": 0.0007628641240363434,
+      "learning_rate": 3.3856068743286793e-06,
+      "loss": 0.0,
+      "step": 10796
+    },
+    {
+      "epoch": 0.6958174904942965,
+      "grad_norm": 9.180779913812038e-05,
+      "learning_rate": 3.384890798424633e-06,
+      "loss": 0.0,
+      "step": 10797
+    },
+    {
+      "epoch": 0.6958819359412257,
+      "grad_norm": 0.10470937724474787,
+      "learning_rate": 3.3841747225205874e-06,
+      "loss": 0.0004,
+      "step": 10798
+    },
+    {
+      "epoch": 0.6959463813881549,
+      "grad_norm": 0.05540156375470692,
+      "learning_rate": 3.3834586466165413e-06,
+      "loss": 0.0002,
+      "step": 10799
+    },
+    {
+      "epoch": 0.6960108268350841,
+      "grad_norm": 0.0035984583674340755,
+      "learning_rate": 3.3827425707124956e-06,
+      "loss": 0.0,
+      "step": 10800
+    },
+    {
+      "epoch": 0.6960752722820133,
+      "grad_norm": 0.0015088185612061752,
+      "learning_rate": 3.3820264948084503e-06,
+      "loss": 0.0,
+      "step": 10801
+    },
+    {
+      "epoch": 0.6961397177289425,
+      "grad_norm": 0.04853284722866676,
+      "learning_rate": 3.381310418904404e-06,
+      "loss": 0.0001,
+      "step": 10802
+    },
+    {
+      "epoch": 0.6962041631758716,
+      "grad_norm": 0.0012557955823111358,
+      "learning_rate": 3.3805943430003585e-06,
+      "loss": 0.0,
+      "step": 10803
+    },
+    {
+      "epoch": 0.6962686086228008,
+      "grad_norm": 0.01631435553766888,
+      "learning_rate": 3.379878267096313e-06,
+      "loss": 0.0,
+      "step": 10804
+    },
+    {
+      "epoch": 0.69633305406973,
+      "grad_norm": 0.003211445832762659,
+      "learning_rate": 3.3791621911922667e-06,
+      "loss": 0.0,
+      "step": 10805
+    },
+    {
+      "epoch": 0.6963974995166592,
+      "grad_norm": 0.0006773092919922552,
+      "learning_rate": 3.378446115288221e-06,
+      "loss": 0.0,
+      "step": 10806
+    },
+    {
+      "epoch": 0.6964619449635884,
+      "grad_norm": 0.097851148669192,
+      "learning_rate": 3.377730039384175e-06,
+      "loss": 0.0002,
+      "step": 10807
+    },
+    {
+      "epoch": 0.6965263904105174,
+      "grad_norm": 0.000700090125051526,
+      "learning_rate": 3.377013963480129e-06,
+      "loss": 0.0,
+      "step": 10808
+    },
+    {
+      "epoch": 0.6965908358574466,
+      "grad_norm": 0.01792088747671209,
+      "learning_rate": 3.376297887576083e-06,
+      "loss": 0.0001,
+      "step": 10809
+    },
+    {
+      "epoch": 0.6966552813043758,
+      "grad_norm": 0.004783641588541656,
+      "learning_rate": 3.3755818116720373e-06,
+      "loss": 0.0,
+      "step": 10810
+    },
+    {
+      "epoch": 0.696719726751305,
+      "grad_norm": 0.0020453910864311918,
+      "learning_rate": 3.3748657357679916e-06,
+      "loss": 0.0,
+      "step": 10811
+    },
+    {
+      "epoch": 0.6967841721982342,
+      "grad_norm": 0.0832148634326752,
+      "learning_rate": 3.3741496598639455e-06,
+      "loss": 0.0005,
+      "step": 10812
+    },
+    {
+      "epoch": 0.6968486176451634,
+      "grad_norm": 0.0001007275952549603,
+      "learning_rate": 3.3734335839599002e-06,
+      "loss": 0.0,
+      "step": 10813
+    },
+    {
+      "epoch": 0.6969130630920926,
+      "grad_norm": 0.005044496693702052,
+      "learning_rate": 3.3727175080558545e-06,
+      "loss": 0.0,
+      "step": 10814
+    },
+    {
+      "epoch": 0.6969775085390217,
+      "grad_norm": 0.17172673180819398,
+      "learning_rate": 3.3720014321518084e-06,
+      "loss": 0.0019,
+      "step": 10815
+    },
+    {
+      "epoch": 0.6970419539859509,
+      "grad_norm": 0.0009186030854356705,
+      "learning_rate": 3.3712853562477627e-06,
+      "loss": 0.0,
+      "step": 10816
+    },
+    {
+      "epoch": 0.6971063994328801,
+      "grad_norm": 0.012453707151726626,
+      "learning_rate": 3.3705692803437166e-06,
+      "loss": 0.0,
+      "step": 10817
+    },
+    {
+      "epoch": 0.6971708448798093,
+      "grad_norm": 0.0009751430932883996,
+      "learning_rate": 3.369853204439671e-06,
+      "loss": 0.0,
+      "step": 10818
+    },
+    {
+      "epoch": 0.6972352903267384,
+      "grad_norm": 0.0004079109002148108,
+      "learning_rate": 3.369137128535625e-06,
+      "loss": 0.0,
+      "step": 10819
+    },
+    {
+      "epoch": 0.6972997357736676,
+      "grad_norm": 0.2528612640796825,
+      "learning_rate": 3.368421052631579e-06,
+      "loss": 0.0003,
+      "step": 10820
+    },
+    {
+      "epoch": 0.6973641812205967,
+      "grad_norm": 0.1139363960253434,
+      "learning_rate": 3.3677049767275333e-06,
+      "loss": 0.0013,
+      "step": 10821
+    },
+    {
+      "epoch": 0.6974286266675259,
+      "grad_norm": 0.005500716077919033,
+      "learning_rate": 3.3669889008234872e-06,
+      "loss": 0.0,
+      "step": 10822
+    },
+    {
+      "epoch": 0.6974930721144551,
+      "grad_norm": 0.08116801795662411,
+      "learning_rate": 3.3662728249194415e-06,
+      "loss": 0.0001,
+      "step": 10823
+    },
+    {
+      "epoch": 0.6975575175613843,
+      "grad_norm": 0.0035024566652384916,
+      "learning_rate": 3.365556749015396e-06,
+      "loss": 0.0,
+      "step": 10824
+    },
+    {
+      "epoch": 0.6976219630083135,
+      "grad_norm": 0.015370411768365658,
+      "learning_rate": 3.36484067311135e-06,
+      "loss": 0.0001,
+      "step": 10825
+    },
+    {
+      "epoch": 0.6976864084552427,
+      "grad_norm": 0.06178277284897271,
+      "learning_rate": 3.3641245972073044e-06,
+      "loss": 0.0002,
+      "step": 10826
+    },
+    {
+      "epoch": 0.6977508539021718,
+      "grad_norm": 0.00037744400225636377,
+      "learning_rate": 3.3634085213032587e-06,
+      "loss": 0.0,
+      "step": 10827
+    },
+    {
+      "epoch": 0.697815299349101,
+      "grad_norm": 0.005827135498293183,
+      "learning_rate": 3.3626924453992126e-06,
+      "loss": 0.0,
+      "step": 10828
+    },
+    {
+      "epoch": 0.6978797447960302,
+      "grad_norm": 0.22508163560634026,
+      "learning_rate": 3.361976369495167e-06,
+      "loss": 0.0002,
+      "step": 10829
+    },
+    {
+      "epoch": 0.6979441902429593,
+      "grad_norm": 0.00026929893020132337,
+      "learning_rate": 3.3612602935911208e-06,
+      "loss": 0.0,
+      "step": 10830
+    },
+    {
+      "epoch": 0.6980086356898885,
+      "grad_norm": 0.0020361584289357894,
+      "learning_rate": 3.360544217687075e-06,
+      "loss": 0.0,
+      "step": 10831
+    },
+    {
+      "epoch": 0.6980730811368177,
+      "grad_norm": 0.0004221505435830251,
+      "learning_rate": 3.3598281417830294e-06,
+      "loss": 0.0,
+      "step": 10832
+    },
+    {
+      "epoch": 0.6981375265837468,
+      "grad_norm": 0.05349353661611049,
+      "learning_rate": 3.3591120658789832e-06,
+      "loss": 0.0001,
+      "step": 10833
+    },
+    {
+      "epoch": 0.698201972030676,
+      "grad_norm": 0.05797448742304965,
+      "learning_rate": 3.3583959899749375e-06,
+      "loss": 0.0008,
+      "step": 10834
+    },
+    {
+      "epoch": 0.6982664174776052,
+      "grad_norm": 0.0008506722447421768,
+      "learning_rate": 3.3576799140708914e-06,
+      "loss": 0.0,
+      "step": 10835
+    },
+    {
+      "epoch": 0.6983308629245344,
+      "grad_norm": 0.00021320946359430795,
+      "learning_rate": 3.356963838166846e-06,
+      "loss": 0.0,
+      "step": 10836
+    },
+    {
+      "epoch": 0.6983953083714636,
+      "grad_norm": 0.39711350760972824,
+      "learning_rate": 3.3562477622628004e-06,
+      "loss": 0.0011,
+      "step": 10837
+    },
+    {
+      "epoch": 0.6984597538183928,
+      "grad_norm": 0.04127152969316703,
+      "learning_rate": 3.3555316863587543e-06,
+      "loss": 0.0001,
+      "step": 10838
+    },
+    {
+      "epoch": 0.6985241992653219,
+      "grad_norm": 0.006618943598043729,
+      "learning_rate": 3.3548156104547086e-06,
+      "loss": 0.0,
+      "step": 10839
+    },
+    {
+      "epoch": 0.6985886447122511,
+      "grad_norm": 0.04176915064793223,
+      "learning_rate": 3.354099534550663e-06,
+      "loss": 0.0001,
+      "step": 10840
+    },
+    {
+      "epoch": 0.6986530901591802,
+      "grad_norm": 0.0001714131345175587,
+      "learning_rate": 3.3533834586466168e-06,
+      "loss": 0.0,
+      "step": 10841
+    },
+    {
+      "epoch": 0.6987175356061094,
+      "grad_norm": 0.0010024964582433073,
+      "learning_rate": 3.352667382742571e-06,
+      "loss": 0.0,
+      "step": 10842
+    },
+    {
+      "epoch": 0.6987819810530386,
+      "grad_norm": 0.00040623033050484755,
+      "learning_rate": 3.351951306838525e-06,
+      "loss": 0.0,
+      "step": 10843
+    },
+    {
+      "epoch": 0.6988464264999678,
+      "grad_norm": 0.00402618346109113,
+      "learning_rate": 3.3512352309344792e-06,
+      "loss": 0.0,
+      "step": 10844
+    },
+    {
+      "epoch": 0.6989108719468969,
+      "grad_norm": 0.22191415897871572,
+      "learning_rate": 3.3505191550304335e-06,
+      "loss": 0.0002,
+      "step": 10845
+    },
+    {
+      "epoch": 0.6989753173938261,
+      "grad_norm": 0.014029645455560074,
+      "learning_rate": 3.3498030791263874e-06,
+      "loss": 0.0015,
+      "step": 10846
+    },
+    {
+      "epoch": 0.6990397628407553,
+      "grad_norm": 8.324158896226437e-05,
+      "learning_rate": 3.3490870032223417e-06,
+      "loss": 0.0,
+      "step": 10847
+    },
+    {
+      "epoch": 0.6991042082876845,
+      "grad_norm": 0.0006412068556455778,
+      "learning_rate": 3.3483709273182964e-06,
+      "loss": 0.0,
+      "step": 10848
+    },
+    {
+      "epoch": 0.6991686537346137,
+      "grad_norm": 4.787087265715539e-05,
+      "learning_rate": 3.3476548514142503e-06,
+      "loss": 0.0,
+      "step": 10849
+    },
+    {
+      "epoch": 0.6992330991815429,
+      "grad_norm": 0.00030164510962597276,
+      "learning_rate": 3.3469387755102046e-06,
+      "loss": 0.0,
+      "step": 10850
+    },
+    {
+      "epoch": 0.699297544628472,
+      "grad_norm": 0.30012512478553754,
+      "learning_rate": 3.3462226996061585e-06,
+      "loss": 0.0024,
+      "step": 10851
+    },
+    {
+      "epoch": 0.6993619900754011,
+      "grad_norm": 0.0059113082282672945,
+      "learning_rate": 3.3455066237021128e-06,
+      "loss": 0.0001,
+      "step": 10852
+    },
+    {
+      "epoch": 0.6994264355223303,
+      "grad_norm": 0.000438918178855173,
+      "learning_rate": 3.344790547798067e-06,
+      "loss": 0.0,
+      "step": 10853
+    },
+    {
+      "epoch": 0.6994908809692595,
+      "grad_norm": 0.0004580738369121127,
+      "learning_rate": 3.344074471894021e-06,
+      "loss": 0.0,
+      "step": 10854
+    },
+    {
+      "epoch": 0.6995553264161887,
+      "grad_norm": 0.0008452896284506048,
+      "learning_rate": 3.3433583959899753e-06,
+      "loss": 0.0,
+      "step": 10855
+    },
+    {
+      "epoch": 0.6996197718631179,
+      "grad_norm": 0.0020973779835961976,
+      "learning_rate": 3.342642320085929e-06,
+      "loss": 0.0,
+      "step": 10856
+    },
+    {
+      "epoch": 0.699684217310047,
+      "grad_norm": 0.0008297177221649112,
+      "learning_rate": 3.3419262441818834e-06,
+      "loss": 0.0,
+      "step": 10857
+    },
+    {
+      "epoch": 0.6997486627569762,
+      "grad_norm": 0.0027529704284585976,
+      "learning_rate": 3.3412101682778373e-06,
+      "loss": 0.0,
+      "step": 10858
+    },
+    {
+      "epoch": 0.6998131082039054,
+      "grad_norm": 0.0008741259795175414,
+      "learning_rate": 3.3404940923737916e-06,
+      "loss": 0.0,
+      "step": 10859
+    },
+    {
+      "epoch": 0.6998775536508346,
+      "grad_norm": 0.0035309142146040173,
+      "learning_rate": 3.3397780164697463e-06,
+      "loss": 0.0,
+      "step": 10860
+    },
+    {
+      "epoch": 0.6999419990977638,
+      "grad_norm": 0.4958865867061118,
+      "learning_rate": 3.3390619405657006e-06,
+      "loss": 0.0028,
+      "step": 10861
+    },
+    {
+      "epoch": 0.700006444544693,
+      "grad_norm": 0.0072535462915399975,
+      "learning_rate": 3.3383458646616545e-06,
+      "loss": 0.0,
+      "step": 10862
+    },
+    {
+      "epoch": 0.7000708899916221,
+      "grad_norm": 0.02617736765886348,
+      "learning_rate": 3.337629788757609e-06,
+      "loss": 0.0,
+      "step": 10863
+    },
+    {
+      "epoch": 0.7001353354385512,
+      "grad_norm": 0.025339570434611148,
+      "learning_rate": 3.3369137128535627e-06,
+      "loss": 0.0,
+      "step": 10864
+    },
+    {
+      "epoch": 0.7001997808854804,
+      "grad_norm": 0.0001549518610992421,
+      "learning_rate": 3.336197636949517e-06,
+      "loss": 0.0,
+      "step": 10865
+    },
+    {
+      "epoch": 0.7002642263324096,
+      "grad_norm": 0.0006099219318847603,
+      "learning_rate": 3.335481561045471e-06,
+      "loss": 0.0,
+      "step": 10866
+    },
+    {
+      "epoch": 0.7003286717793388,
+      "grad_norm": 0.00019973177616691002,
+      "learning_rate": 3.334765485141425e-06,
+      "loss": 0.0,
+      "step": 10867
+    },
+    {
+      "epoch": 0.700393117226268,
+      "grad_norm": 0.0203871412425311,
+      "learning_rate": 3.3340494092373794e-06,
+      "loss": 0.0,
+      "step": 10868
+    },
+    {
+      "epoch": 0.7004575626731971,
+      "grad_norm": 0.00012341835821285828,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 0.0,
+      "step": 10869
+    },
+    {
+      "epoch": 0.7005220081201263,
+      "grad_norm": 3.691681563023068e-05,
+      "learning_rate": 3.3326172574292876e-06,
+      "loss": 0.0,
+      "step": 10870
+    },
+    {
+      "epoch": 0.7005864535670555,
+      "grad_norm": 0.041492466191393934,
+      "learning_rate": 3.3319011815252423e-06,
+      "loss": 0.0002,
+      "step": 10871
+    },
+    {
+      "epoch": 0.7006508990139847,
+      "grad_norm": 0.16212739040050764,
+      "learning_rate": 3.3311851056211962e-06,
+      "loss": 0.0005,
+      "step": 10872
+    },
+    {
+      "epoch": 0.7007153444609139,
+      "grad_norm": 0.001993052236712686,
+      "learning_rate": 3.3304690297171505e-06,
+      "loss": 0.0,
+      "step": 10873
+    },
+    {
+      "epoch": 0.7007797899078431,
+      "grad_norm": 0.0021670367114100574,
+      "learning_rate": 3.3297529538131044e-06,
+      "loss": 0.0,
+      "step": 10874
+    },
+    {
+      "epoch": 0.7008442353547721,
+      "grad_norm": 0.0013950064005321817,
+      "learning_rate": 3.3290368779090587e-06,
+      "loss": 0.0,
+      "step": 10875
+    },
+    {
+      "epoch": 0.7009086808017013,
+      "grad_norm": 0.0007154109309629973,
+      "learning_rate": 3.328320802005013e-06,
+      "loss": 0.0,
+      "step": 10876
+    },
+    {
+      "epoch": 0.7009731262486305,
+      "grad_norm": 0.03324730647386008,
+      "learning_rate": 3.327604726100967e-06,
+      "loss": 0.0003,
+      "step": 10877
+    },
+    {
+      "epoch": 0.7010375716955597,
+      "grad_norm": 0.001851781740293082,
+      "learning_rate": 3.326888650196921e-06,
+      "loss": 0.0,
+      "step": 10878
+    },
+    {
+      "epoch": 0.7011020171424889,
+      "grad_norm": 0.47544556412340017,
+      "learning_rate": 3.326172574292875e-06,
+      "loss": 0.0027,
+      "step": 10879
+    },
+    {
+      "epoch": 0.7011664625894181,
+      "grad_norm": 0.03185670473705229,
+      "learning_rate": 3.3254564983888293e-06,
+      "loss": 0.0,
+      "step": 10880
+    },
+    {
+      "epoch": 0.7012309080363472,
+      "grad_norm": 7.054423748793358e-05,
+      "learning_rate": 3.3247404224847836e-06,
+      "loss": 0.0,
+      "step": 10881
+    },
+    {
+      "epoch": 0.7012953534832764,
+      "grad_norm": 0.006720652727550893,
+      "learning_rate": 3.3240243465807375e-06,
+      "loss": 0.0001,
+      "step": 10882
+    },
+    {
+      "epoch": 0.7013597989302056,
+      "grad_norm": 0.00926893960988273,
+      "learning_rate": 3.3233082706766922e-06,
+      "loss": 0.0,
+      "step": 10883
+    },
+    {
+      "epoch": 0.7014242443771348,
+      "grad_norm": 0.0037271138191373935,
+      "learning_rate": 3.3225921947726465e-06,
+      "loss": 0.0,
+      "step": 10884
+    },
+    {
+      "epoch": 0.701488689824064,
+      "grad_norm": 0.26028400697964355,
+      "learning_rate": 3.3218761188686004e-06,
+      "loss": 0.0001,
+      "step": 10885
+    },
+    {
+      "epoch": 0.7015531352709931,
+      "grad_norm": 0.008150517832203049,
+      "learning_rate": 3.3211600429645547e-06,
+      "loss": 0.0,
+      "step": 10886
+    },
+    {
+      "epoch": 0.7016175807179222,
+      "grad_norm": 0.016123935218947682,
+      "learning_rate": 3.3204439670605086e-06,
+      "loss": 0.0,
+      "step": 10887
+    },
+    {
+      "epoch": 0.7016820261648514,
+      "grad_norm": 0.0025847379419570832,
+      "learning_rate": 3.319727891156463e-06,
+      "loss": 0.0,
+      "step": 10888
+    },
+    {
+      "epoch": 0.7017464716117806,
+      "grad_norm": 0.02264905884294404,
+      "learning_rate": 3.319011815252417e-06,
+      "loss": 0.0,
+      "step": 10889
+    },
+    {
+      "epoch": 0.7018109170587098,
+      "grad_norm": 0.012890156146756319,
+      "learning_rate": 3.318295739348371e-06,
+      "loss": 0.0,
+      "step": 10890
+    },
+    {
+      "epoch": 0.701875362505639,
+      "grad_norm": 0.0005745350547465109,
+      "learning_rate": 3.3175796634443254e-06,
+      "loss": 0.0,
+      "step": 10891
+    },
+    {
+      "epoch": 0.7019398079525682,
+      "grad_norm": 0.015874661805627098,
+      "learning_rate": 3.3168635875402792e-06,
+      "loss": 0.0,
+      "step": 10892
+    },
+    {
+      "epoch": 0.7020042533994973,
+      "grad_norm": 0.0008491966764384103,
+      "learning_rate": 3.3161475116362335e-06,
+      "loss": 0.0,
+      "step": 10893
+    },
+    {
+      "epoch": 0.7020686988464265,
+      "grad_norm": 0.004415929786913111,
+      "learning_rate": 3.315431435732188e-06,
+      "loss": 0.0,
+      "step": 10894
+    },
+    {
+      "epoch": 0.7021331442933557,
+      "grad_norm": 0.04686867154635354,
+      "learning_rate": 3.314715359828142e-06,
+      "loss": 0.0001,
+      "step": 10895
+    },
+    {
+      "epoch": 0.7021975897402849,
+      "grad_norm": 0.026635296990076995,
+      "learning_rate": 3.3139992839240964e-06,
+      "loss": 0.0001,
+      "step": 10896
+    },
+    {
+      "epoch": 0.702262035187214,
+      "grad_norm": 0.0002269513361057032,
+      "learning_rate": 3.3132832080200507e-06,
+      "loss": 0.0,
+      "step": 10897
+    },
+    {
+      "epoch": 0.7023264806341432,
+      "grad_norm": 0.0008907517858752973,
+      "learning_rate": 3.3125671321160046e-06,
+      "loss": 0.0,
+      "step": 10898
+    },
+    {
+      "epoch": 0.7023909260810723,
+      "grad_norm": 0.13432873522726482,
+      "learning_rate": 3.311851056211959e-06,
+      "loss": 0.0003,
+      "step": 10899
+    },
+    {
+      "epoch": 0.7024553715280015,
+      "grad_norm": 0.005984730876438511,
+      "learning_rate": 3.3111349803079128e-06,
+      "loss": 0.0,
+      "step": 10900
+    },
+    {
+      "epoch": 0.7025198169749307,
+      "grad_norm": 0.17898883995504597,
+      "learning_rate": 3.310418904403867e-06,
+      "loss": 0.0002,
+      "step": 10901
+    },
+    {
+      "epoch": 0.7025842624218599,
+      "grad_norm": 0.00810524339822239,
+      "learning_rate": 3.3097028284998214e-06,
+      "loss": 0.0,
+      "step": 10902
+    },
+    {
+      "epoch": 0.7026487078687891,
+      "grad_norm": 0.002444304481964341,
+      "learning_rate": 3.3089867525957752e-06,
+      "loss": 0.0,
+      "step": 10903
+    },
+    {
+      "epoch": 0.7027131533157183,
+      "grad_norm": 0.0017770346103979729,
+      "learning_rate": 3.3082706766917295e-06,
+      "loss": 0.0,
+      "step": 10904
+    },
+    {
+      "epoch": 0.7027775987626474,
+      "grad_norm": 0.001104809674803772,
+      "learning_rate": 3.3075546007876834e-06,
+      "loss": 0.0,
+      "step": 10905
+    },
+    {
+      "epoch": 0.7028420442095766,
+      "grad_norm": 0.009800180749657906,
+      "learning_rate": 3.306838524883638e-06,
+      "loss": 0.0,
+      "step": 10906
+    },
+    {
+      "epoch": 0.7029064896565058,
+      "grad_norm": 0.0026749097474474373,
+      "learning_rate": 3.3061224489795924e-06,
+      "loss": 0.0,
+      "step": 10907
+    },
+    {
+      "epoch": 0.7029709351034349,
+      "grad_norm": 0.01313033705099765,
+      "learning_rate": 3.3054063730755463e-06,
+      "loss": 0.0001,
+      "step": 10908
+    },
+    {
+      "epoch": 0.7030353805503641,
+      "grad_norm": 0.00436655342451471,
+      "learning_rate": 3.3046902971715006e-06,
+      "loss": 0.0,
+      "step": 10909
+    },
+    {
+      "epoch": 0.7030998259972933,
+      "grad_norm": 0.007957100060090051,
+      "learning_rate": 3.303974221267455e-06,
+      "loss": 0.0,
+      "step": 10910
+    },
+    {
+      "epoch": 0.7031642714442224,
+      "grad_norm": 0.6992021270992388,
+      "learning_rate": 3.3032581453634088e-06,
+      "loss": 0.0096,
+      "step": 10911
+    },
+    {
+      "epoch": 0.7032287168911516,
+      "grad_norm": 0.004975517335532034,
+      "learning_rate": 3.302542069459363e-06,
+      "loss": 0.0,
+      "step": 10912
+    },
+    {
+      "epoch": 0.7032931623380808,
+      "grad_norm": 0.013508631034990674,
+      "learning_rate": 3.301825993555317e-06,
+      "loss": 0.0,
+      "step": 10913
+    },
+    {
+      "epoch": 0.70335760778501,
+      "grad_norm": 0.8323221906874271,
+      "learning_rate": 3.3011099176512713e-06,
+      "loss": 0.0153,
+      "step": 10914
+    },
+    {
+      "epoch": 0.7034220532319392,
+      "grad_norm": 0.0002831184487363148,
+      "learning_rate": 3.300393841747225e-06,
+      "loss": 0.0,
+      "step": 10915
+    },
+    {
+      "epoch": 0.7034864986788684,
+      "grad_norm": 0.014769501888290858,
+      "learning_rate": 3.2996777658431794e-06,
+      "loss": 0.0,
+      "step": 10916
+    },
+    {
+      "epoch": 0.7035509441257976,
+      "grad_norm": 0.4236323875892939,
+      "learning_rate": 3.2989616899391337e-06,
+      "loss": 0.0031,
+      "step": 10917
+    },
+    {
+      "epoch": 0.7036153895727267,
+      "grad_norm": 0.011189639856460357,
+      "learning_rate": 3.2982456140350885e-06,
+      "loss": 0.0,
+      "step": 10918
+    },
+    {
+      "epoch": 0.7036798350196558,
+      "grad_norm": 0.00829430680010469,
+      "learning_rate": 3.2975295381310423e-06,
+      "loss": 0.0,
+      "step": 10919
+    },
+    {
+      "epoch": 0.703744280466585,
+      "grad_norm": 0.0039546963976976515,
+      "learning_rate": 3.2968134622269966e-06,
+      "loss": 0.0,
+      "step": 10920
+    },
+    {
+      "epoch": 0.7038087259135142,
+      "grad_norm": 0.01591126280223268,
+      "learning_rate": 3.2960973863229505e-06,
+      "loss": 0.0002,
+      "step": 10921
+    },
+    {
+      "epoch": 0.7038731713604434,
+      "grad_norm": 0.012323091831577818,
+      "learning_rate": 3.295381310418905e-06,
+      "loss": 0.0,
+      "step": 10922
+    },
+    {
+      "epoch": 0.7039376168073725,
+      "grad_norm": 0.0007462944453321257,
+      "learning_rate": 3.2946652345148587e-06,
+      "loss": 0.0,
+      "step": 10923
+    },
+    {
+      "epoch": 0.7040020622543017,
+      "grad_norm": 0.0035263945187977517,
+      "learning_rate": 3.293949158610813e-06,
+      "loss": 0.0,
+      "step": 10924
+    },
+    {
+      "epoch": 0.7040665077012309,
+      "grad_norm": 0.006692252532699696,
+      "learning_rate": 3.2932330827067673e-06,
+      "loss": 0.0,
+      "step": 10925
+    },
+    {
+      "epoch": 0.7041309531481601,
+      "grad_norm": 0.011862744356579047,
+      "learning_rate": 3.292517006802721e-06,
+      "loss": 0.0,
+      "step": 10926
+    },
+    {
+      "epoch": 0.7041953985950893,
+      "grad_norm": 0.0009588384878591355,
+      "learning_rate": 3.2918009308986754e-06,
+      "loss": 0.0,
+      "step": 10927
+    },
+    {
+      "epoch": 0.7042598440420185,
+      "grad_norm": 0.06882280888848988,
+      "learning_rate": 3.2910848549946293e-06,
+      "loss": 0.0002,
+      "step": 10928
+    },
+    {
+      "epoch": 0.7043242894889477,
+      "grad_norm": 0.03040901424031329,
+      "learning_rate": 3.2903687790905836e-06,
+      "loss": 0.0,
+      "step": 10929
+    },
+    {
+      "epoch": 0.7043887349358767,
+      "grad_norm": 0.003522038167252063,
+      "learning_rate": 3.2896527031865383e-06,
+      "loss": 0.0,
+      "step": 10930
+    },
+    {
+      "epoch": 0.7044531803828059,
+      "grad_norm": 0.15972103322031145,
+      "learning_rate": 3.2889366272824922e-06,
+      "loss": 0.0018,
+      "step": 10931
+    },
+    {
+      "epoch": 0.7045176258297351,
+      "grad_norm": 0.06893447717889657,
+      "learning_rate": 3.2882205513784465e-06,
+      "loss": 0.0001,
+      "step": 10932
+    },
+    {
+      "epoch": 0.7045820712766643,
+      "grad_norm": 3.0882135037847522,
+      "learning_rate": 3.287504475474401e-06,
+      "loss": 0.0268,
+      "step": 10933
+    },
+    {
+      "epoch": 0.7046465167235935,
+      "grad_norm": 0.139790129637713,
+      "learning_rate": 3.2867883995703547e-06,
+      "loss": 0.0003,
+      "step": 10934
+    },
+    {
+      "epoch": 0.7047109621705226,
+      "grad_norm": 0.01926119261248183,
+      "learning_rate": 3.286072323666309e-06,
+      "loss": 0.0,
+      "step": 10935
+    },
+    {
+      "epoch": 0.7047754076174518,
+      "grad_norm": 0.003360752891597998,
+      "learning_rate": 3.285356247762263e-06,
+      "loss": 0.0,
+      "step": 10936
+    },
+    {
+      "epoch": 0.704839853064381,
+      "grad_norm": 0.02846292441579501,
+      "learning_rate": 3.284640171858217e-06,
+      "loss": 0.0,
+      "step": 10937
+    },
+    {
+      "epoch": 0.7049042985113102,
+      "grad_norm": 0.010289595574202861,
+      "learning_rate": 3.2839240959541715e-06,
+      "loss": 0.0,
+      "step": 10938
+    },
+    {
+      "epoch": 0.7049687439582394,
+      "grad_norm": 0.022801437024493606,
+      "learning_rate": 3.2832080200501253e-06,
+      "loss": 0.0,
+      "step": 10939
+    },
+    {
+      "epoch": 0.7050331894051686,
+      "grad_norm": 0.00020393405837112934,
+      "learning_rate": 3.2824919441460796e-06,
+      "loss": 0.0,
+      "step": 10940
+    },
+    {
+      "epoch": 0.7050976348520978,
+      "grad_norm": 0.012538392880569913,
+      "learning_rate": 3.2817758682420335e-06,
+      "loss": 0.0,
+      "step": 10941
+    },
+    {
+      "epoch": 0.7051620802990268,
+      "grad_norm": 0.02402381009882897,
+      "learning_rate": 3.2810597923379882e-06,
+      "loss": 0.0,
+      "step": 10942
+    },
+    {
+      "epoch": 0.705226525745956,
+      "grad_norm": 0.03254472748774419,
+      "learning_rate": 3.2803437164339425e-06,
+      "loss": 0.0,
+      "step": 10943
+    },
+    {
+      "epoch": 0.7052909711928852,
+      "grad_norm": 0.004207380008318214,
+      "learning_rate": 3.2796276405298964e-06,
+      "loss": 0.0,
+      "step": 10944
+    },
+    {
+      "epoch": 0.7053554166398144,
+      "grad_norm": 0.020072217846481624,
+      "learning_rate": 3.2789115646258507e-06,
+      "loss": 0.0,
+      "step": 10945
+    },
+    {
+      "epoch": 0.7054198620867436,
+      "grad_norm": 0.016132958636072284,
+      "learning_rate": 3.278195488721805e-06,
+      "loss": 0.0,
+      "step": 10946
+    },
+    {
+      "epoch": 0.7054843075336727,
+      "grad_norm": 0.04939703204957824,
+      "learning_rate": 3.277479412817759e-06,
+      "loss": 0.0001,
+      "step": 10947
+    },
+    {
+      "epoch": 0.7055487529806019,
+      "grad_norm": 0.0008245602376543607,
+      "learning_rate": 3.276763336913713e-06,
+      "loss": 0.0,
+      "step": 10948
+    },
+    {
+      "epoch": 0.7056131984275311,
+      "grad_norm": 0.0005412021396897577,
+      "learning_rate": 3.276047261009667e-06,
+      "loss": 0.0,
+      "step": 10949
+    },
+    {
+      "epoch": 0.7056776438744603,
+      "grad_norm": 0.0034232693202888597,
+      "learning_rate": 3.2753311851056214e-06,
+      "loss": 0.0,
+      "step": 10950
+    },
+    {
+      "epoch": 0.7057420893213895,
+      "grad_norm": 0.005485021994327029,
+      "learning_rate": 3.2746151092015752e-06,
+      "loss": 0.0,
+      "step": 10951
+    },
+    {
+      "epoch": 0.7058065347683187,
+      "grad_norm": 0.017868621028440626,
+      "learning_rate": 3.2738990332975295e-06,
+      "loss": 0.0001,
+      "step": 10952
+    },
+    {
+      "epoch": 0.7058709802152477,
+      "grad_norm": 0.002559377594809387,
+      "learning_rate": 3.2731829573934843e-06,
+      "loss": 0.0,
+      "step": 10953
+    },
+    {
+      "epoch": 0.7059354256621769,
+      "grad_norm": 0.10073643138781171,
+      "learning_rate": 3.2724668814894385e-06,
+      "loss": 0.0001,
+      "step": 10954
+    },
+    {
+      "epoch": 0.7059998711091061,
+      "grad_norm": 0.005557456868990858,
+      "learning_rate": 3.2717508055853924e-06,
+      "loss": 0.0,
+      "step": 10955
+    },
+    {
+      "epoch": 0.7060643165560353,
+      "grad_norm": 0.018022847987917433,
+      "learning_rate": 3.2710347296813467e-06,
+      "loss": 0.0,
+      "step": 10956
+    },
+    {
+      "epoch": 0.7061287620029645,
+      "grad_norm": 0.0010845042955528442,
+      "learning_rate": 3.2703186537773006e-06,
+      "loss": 0.0,
+      "step": 10957
+    },
+    {
+      "epoch": 0.7061932074498937,
+      "grad_norm": 0.006678590974714848,
+      "learning_rate": 3.269602577873255e-06,
+      "loss": 0.0,
+      "step": 10958
+    },
+    {
+      "epoch": 0.7062576528968229,
+      "grad_norm": 0.24238967916520135,
+      "learning_rate": 3.2688865019692088e-06,
+      "loss": 0.0003,
+      "step": 10959
+    },
+    {
+      "epoch": 0.706322098343752,
+      "grad_norm": 0.0005615450731238256,
+      "learning_rate": 3.268170426065163e-06,
+      "loss": 0.0,
+      "step": 10960
+    },
+    {
+      "epoch": 0.7063865437906812,
+      "grad_norm": 0.001237721068779436,
+      "learning_rate": 3.2674543501611174e-06,
+      "loss": 0.0,
+      "step": 10961
+    },
+    {
+      "epoch": 0.7064509892376104,
+      "grad_norm": 0.0022154238217625168,
+      "learning_rate": 3.2667382742570712e-06,
+      "loss": 0.0,
+      "step": 10962
+    },
+    {
+      "epoch": 0.7065154346845396,
+      "grad_norm": 0.00036969182107827645,
+      "learning_rate": 3.2660221983530255e-06,
+      "loss": 0.0,
+      "step": 10963
+    },
+    {
+      "epoch": 0.7065798801314687,
+      "grad_norm": 0.14243230155867326,
+      "learning_rate": 3.2653061224489794e-06,
+      "loss": 0.0003,
+      "step": 10964
+    },
+    {
+      "epoch": 0.7066443255783978,
+      "grad_norm": 0.005058644391544198,
+      "learning_rate": 3.264590046544934e-06,
+      "loss": 0.0,
+      "step": 10965
+    },
+    {
+      "epoch": 0.706708771025327,
+      "grad_norm": 0.0032059561848995183,
+      "learning_rate": 3.2638739706408884e-06,
+      "loss": 0.0,
+      "step": 10966
+    },
+    {
+      "epoch": 0.7067732164722562,
+      "grad_norm": 0.013179660912521117,
+      "learning_rate": 3.2631578947368423e-06,
+      "loss": 0.0001,
+      "step": 10967
+    },
+    {
+      "epoch": 0.7068376619191854,
+      "grad_norm": 0.11744684687956847,
+      "learning_rate": 3.2624418188327966e-06,
+      "loss": 0.0002,
+      "step": 10968
+    },
+    {
+      "epoch": 0.7069021073661146,
+      "grad_norm": 0.004042410681713448,
+      "learning_rate": 3.261725742928751e-06,
+      "loss": 0.0,
+      "step": 10969
+    },
+    {
+      "epoch": 0.7069665528130438,
+      "grad_norm": 4.283922259766499e-05,
+      "learning_rate": 3.2610096670247048e-06,
+      "loss": 0.0,
+      "step": 10970
+    },
+    {
+      "epoch": 0.707030998259973,
+      "grad_norm": 0.9901557003270737,
+      "learning_rate": 3.260293591120659e-06,
+      "loss": 0.006,
+      "step": 10971
+    },
+    {
+      "epoch": 0.7070954437069021,
+      "grad_norm": 0.037523802452588115,
+      "learning_rate": 3.259577515216613e-06,
+      "loss": 0.0,
+      "step": 10972
+    },
+    {
+      "epoch": 0.7071598891538313,
+      "grad_norm": 0.015195530897357019,
+      "learning_rate": 3.2588614393125673e-06,
+      "loss": 0.0,
+      "step": 10973
+    },
+    {
+      "epoch": 0.7072243346007605,
+      "grad_norm": 0.0022295571638922154,
+      "learning_rate": 3.2581453634085216e-06,
+      "loss": 0.0,
+      "step": 10974
+    },
+    {
+      "epoch": 0.7072887800476896,
+      "grad_norm": 0.8156517229050663,
+      "learning_rate": 3.2574292875044754e-06,
+      "loss": 0.0035,
+      "step": 10975
+    },
+    {
+      "epoch": 0.7073532254946188,
+      "grad_norm": 0.004237609517313484,
+      "learning_rate": 3.2567132116004297e-06,
+      "loss": 0.0,
+      "step": 10976
+    },
+    {
+      "epoch": 0.707417670941548,
+      "grad_norm": 0.005571668982925098,
+      "learning_rate": 3.2559971356963845e-06,
+      "loss": 0.0,
+      "step": 10977
+    },
+    {
+      "epoch": 0.7074821163884771,
+      "grad_norm": 0.0036520828137762886,
+      "learning_rate": 3.2552810597923383e-06,
+      "loss": 0.0,
+      "step": 10978
+    },
+    {
+      "epoch": 0.7075465618354063,
+      "grad_norm": 0.016016935812411235,
+      "learning_rate": 3.2545649838882926e-06,
+      "loss": 0.0,
+      "step": 10979
+    },
+    {
+      "epoch": 0.7076110072823355,
+      "grad_norm": 0.3980306343157112,
+      "learning_rate": 3.2538489079842465e-06,
+      "loss": 0.0041,
+      "step": 10980
+    },
+    {
+      "epoch": 0.7076754527292647,
+      "grad_norm": 0.0009197268556172491,
+      "learning_rate": 3.253132832080201e-06,
+      "loss": 0.0,
+      "step": 10981
+    },
+    {
+      "epoch": 0.7077398981761939,
+      "grad_norm": 0.00043302879034731197,
+      "learning_rate": 3.252416756176155e-06,
+      "loss": 0.0,
+      "step": 10982
+    },
+    {
+      "epoch": 0.7078043436231231,
+      "grad_norm": 0.004272269559729301,
+      "learning_rate": 3.251700680272109e-06,
+      "loss": 0.0,
+      "step": 10983
+    },
+    {
+      "epoch": 0.7078687890700522,
+      "grad_norm": 0.0024739338398368465,
+      "learning_rate": 3.2509846043680633e-06,
+      "loss": 0.0,
+      "step": 10984
+    },
+    {
+      "epoch": 0.7079332345169814,
+      "grad_norm": 0.1845882851059872,
+      "learning_rate": 3.250268528464017e-06,
+      "loss": 0.0028,
+      "step": 10985
+    },
+    {
+      "epoch": 0.7079976799639105,
+      "grad_norm": 0.08929369829022431,
+      "learning_rate": 3.2495524525599714e-06,
+      "loss": 0.0002,
+      "step": 10986
+    },
+    {
+      "epoch": 0.7080621254108397,
+      "grad_norm": 0.0027870529419560435,
+      "learning_rate": 3.2488363766559257e-06,
+      "loss": 0.0,
+      "step": 10987
+    },
+    {
+      "epoch": 0.7081265708577689,
+      "grad_norm": 0.0017530348572803891,
+      "learning_rate": 3.24812030075188e-06,
+      "loss": 0.0,
+      "step": 10988
+    },
+    {
+      "epoch": 0.708191016304698,
+      "grad_norm": 0.006516597776420439,
+      "learning_rate": 3.2474042248478343e-06,
+      "loss": 0.0001,
+      "step": 10989
+    },
+    {
+      "epoch": 0.7082554617516272,
+      "grad_norm": 0.05491665317800124,
+      "learning_rate": 3.2466881489437886e-06,
+      "loss": 0.0006,
+      "step": 10990
+    },
+    {
+      "epoch": 0.7083199071985564,
+      "grad_norm": 0.007344119285479627,
+      "learning_rate": 3.2459720730397425e-06,
+      "loss": 0.0,
+      "step": 10991
+    },
+    {
+      "epoch": 0.7083843526454856,
+      "grad_norm": 0.004823036421794768,
+      "learning_rate": 3.245255997135697e-06,
+      "loss": 0.0,
+      "step": 10992
+    },
+    {
+      "epoch": 0.7084487980924148,
+      "grad_norm": 0.001133053437795517,
+      "learning_rate": 3.2445399212316507e-06,
+      "loss": 0.0,
+      "step": 10993
+    },
+    {
+      "epoch": 0.708513243539344,
+      "grad_norm": 0.010246483494655423,
+      "learning_rate": 3.243823845327605e-06,
+      "loss": 0.0001,
+      "step": 10994
+    },
+    {
+      "epoch": 0.7085776889862732,
+      "grad_norm": 0.00031286769307990606,
+      "learning_rate": 3.2431077694235593e-06,
+      "loss": 0.0,
+      "step": 10995
+    },
+    {
+      "epoch": 0.7086421344332023,
+      "grad_norm": 0.025983279433569137,
+      "learning_rate": 3.242391693519513e-06,
+      "loss": 0.0003,
+      "step": 10996
+    },
+    {
+      "epoch": 0.7087065798801314,
+      "grad_norm": 0.002991161752113345,
+      "learning_rate": 3.2416756176154675e-06,
+      "loss": 0.0,
+      "step": 10997
+    },
+    {
+      "epoch": 0.7087710253270606,
+      "grad_norm": 0.00019414406743275953,
+      "learning_rate": 3.2409595417114213e-06,
+      "loss": 0.0,
+      "step": 10998
+    },
+    {
+      "epoch": 0.7088354707739898,
+      "grad_norm": 0.0051728910727569175,
+      "learning_rate": 3.2402434658073756e-06,
+      "loss": 0.0,
+      "step": 10999
+    },
+    {
+      "epoch": 0.708899916220919,
+      "grad_norm": 0.032033257450606374,
+      "learning_rate": 3.2395273899033304e-06,
+      "loss": 0.0,
+      "step": 11000
+    },
+    {
+      "epoch": 0.7089643616678482,
+      "grad_norm": 0.0013291294574410593,
+      "learning_rate": 3.2388113139992842e-06,
+      "loss": 0.0,
+      "step": 11001
+    },
+    {
+      "epoch": 0.7090288071147773,
+      "grad_norm": 0.0030029524874015485,
+      "learning_rate": 3.2380952380952385e-06,
+      "loss": 0.0,
+      "step": 11002
+    },
+    {
+      "epoch": 0.7090932525617065,
+      "grad_norm": 0.018705699350202824,
+      "learning_rate": 3.237379162191193e-06,
+      "loss": 0.0,
+      "step": 11003
+    },
+    {
+      "epoch": 0.7091576980086357,
+      "grad_norm": 0.12145905230771745,
+      "learning_rate": 3.2366630862871467e-06,
+      "loss": 0.0004,
+      "step": 11004
+    },
+    {
+      "epoch": 0.7092221434555649,
+      "grad_norm": 0.002445939444972882,
+      "learning_rate": 3.235947010383101e-06,
+      "loss": 0.0,
+      "step": 11005
+    },
+    {
+      "epoch": 0.7092865889024941,
+      "grad_norm": 0.0030465461515954964,
+      "learning_rate": 3.235230934479055e-06,
+      "loss": 0.0,
+      "step": 11006
+    },
+    {
+      "epoch": 0.7093510343494233,
+      "grad_norm": 0.0022634303186151597,
+      "learning_rate": 3.234514858575009e-06,
+      "loss": 0.0,
+      "step": 11007
+    },
+    {
+      "epoch": 0.7094154797963523,
+      "grad_norm": 0.1156427131436469,
+      "learning_rate": 3.233798782670963e-06,
+      "loss": 0.0002,
+      "step": 11008
+    },
+    {
+      "epoch": 0.7094799252432815,
+      "grad_norm": 0.002947625272147337,
+      "learning_rate": 3.2330827067669174e-06,
+      "loss": 0.0,
+      "step": 11009
+    },
+    {
+      "epoch": 0.7095443706902107,
+      "grad_norm": 0.04937201848564881,
+      "learning_rate": 3.2323666308628717e-06,
+      "loss": 0.0001,
+      "step": 11010
+    },
+    {
+      "epoch": 0.7096088161371399,
+      "grad_norm": 0.011345745946028784,
+      "learning_rate": 3.2316505549588255e-06,
+      "loss": 0.0,
+      "step": 11011
+    },
+    {
+      "epoch": 0.7096732615840691,
+      "grad_norm": 0.13372508837535974,
+      "learning_rate": 3.2309344790547803e-06,
+      "loss": 0.0005,
+      "step": 11012
+    },
+    {
+      "epoch": 0.7097377070309983,
+      "grad_norm": 0.02591297992435188,
+      "learning_rate": 3.2302184031507345e-06,
+      "loss": 0.0,
+      "step": 11013
+    },
+    {
+      "epoch": 0.7098021524779274,
+      "grad_norm": 0.0005216657897116044,
+      "learning_rate": 3.2295023272466884e-06,
+      "loss": 0.0,
+      "step": 11014
+    },
+    {
+      "epoch": 0.7098665979248566,
+      "grad_norm": 0.001228820551302051,
+      "learning_rate": 3.2287862513426427e-06,
+      "loss": 0.0,
+      "step": 11015
+    },
+    {
+      "epoch": 0.7099310433717858,
+      "grad_norm": 0.004049530687622813,
+      "learning_rate": 3.2280701754385966e-06,
+      "loss": 0.0,
+      "step": 11016
+    },
+    {
+      "epoch": 0.709995488818715,
+      "grad_norm": 0.01042890442965104,
+      "learning_rate": 3.227354099534551e-06,
+      "loss": 0.0,
+      "step": 11017
+    },
+    {
+      "epoch": 0.7100599342656442,
+      "grad_norm": 0.20473263111957016,
+      "learning_rate": 3.226638023630505e-06,
+      "loss": 0.0013,
+      "step": 11018
+    },
+    {
+      "epoch": 0.7101243797125734,
+      "grad_norm": 0.003618323682590707,
+      "learning_rate": 3.225921947726459e-06,
+      "loss": 0.0,
+      "step": 11019
+    },
+    {
+      "epoch": 0.7101888251595024,
+      "grad_norm": 0.00013886623571113355,
+      "learning_rate": 3.2252058718224134e-06,
+      "loss": 0.0,
+      "step": 11020
+    },
+    {
+      "epoch": 0.7102532706064316,
+      "grad_norm": 0.00031239600076566764,
+      "learning_rate": 3.2244897959183672e-06,
+      "loss": 0.0,
+      "step": 11021
+    },
+    {
+      "epoch": 0.7103177160533608,
+      "grad_norm": 0.015242570841153947,
+      "learning_rate": 3.2237737200143215e-06,
+      "loss": 0.0001,
+      "step": 11022
+    },
+    {
+      "epoch": 0.71038216150029,
+      "grad_norm": 5.747915247272944e-05,
+      "learning_rate": 3.2230576441102763e-06,
+      "loss": 0.0,
+      "step": 11023
+    },
+    {
+      "epoch": 0.7104466069472192,
+      "grad_norm": 0.0062475296909040865,
+      "learning_rate": 3.22234156820623e-06,
+      "loss": 0.0,
+      "step": 11024
+    },
+    {
+      "epoch": 0.7105110523941484,
+      "grad_norm": 0.0017142098867375052,
+      "learning_rate": 3.2216254923021844e-06,
+      "loss": 0.0,
+      "step": 11025
+    },
+    {
+      "epoch": 0.7105754978410775,
+      "grad_norm": 0.0011041837520175915,
+      "learning_rate": 3.2209094163981387e-06,
+      "loss": 0.0,
+      "step": 11026
+    },
+    {
+      "epoch": 0.7106399432880067,
+      "grad_norm": 0.00122298031984066,
+      "learning_rate": 3.2201933404940926e-06,
+      "loss": 0.0,
+      "step": 11027
+    },
+    {
+      "epoch": 0.7107043887349359,
+      "grad_norm": 0.00011782227449732348,
+      "learning_rate": 3.219477264590047e-06,
+      "loss": 0.0,
+      "step": 11028
+    },
+    {
+      "epoch": 0.7107688341818651,
+      "grad_norm": 0.000478511478482095,
+      "learning_rate": 3.2187611886860008e-06,
+      "loss": 0.0,
+      "step": 11029
+    },
+    {
+      "epoch": 0.7108332796287943,
+      "grad_norm": 0.0075666354507896,
+      "learning_rate": 3.218045112781955e-06,
+      "loss": 0.0,
+      "step": 11030
+    },
+    {
+      "epoch": 0.7108977250757234,
+      "grad_norm": 0.040470238174043854,
+      "learning_rate": 3.2173290368779094e-06,
+      "loss": 0.0003,
+      "step": 11031
+    },
+    {
+      "epoch": 0.7109621705226525,
+      "grad_norm": 0.049252596260576695,
+      "learning_rate": 3.2166129609738633e-06,
+      "loss": 0.0005,
+      "step": 11032
+    },
+    {
+      "epoch": 0.7110266159695817,
+      "grad_norm": 0.5214163060714664,
+      "learning_rate": 3.2158968850698176e-06,
+      "loss": 0.0017,
+      "step": 11033
+    },
+    {
+      "epoch": 0.7110910614165109,
+      "grad_norm": 0.004091547353671431,
+      "learning_rate": 3.2151808091657714e-06,
+      "loss": 0.0,
+      "step": 11034
+    },
+    {
+      "epoch": 0.7111555068634401,
+      "grad_norm": 0.8728484540582238,
+      "learning_rate": 3.214464733261726e-06,
+      "loss": 0.0037,
+      "step": 11035
+    },
+    {
+      "epoch": 0.7112199523103693,
+      "grad_norm": 0.002028648083962951,
+      "learning_rate": 3.2137486573576805e-06,
+      "loss": 0.0,
+      "step": 11036
+    },
+    {
+      "epoch": 0.7112843977572985,
+      "grad_norm": 0.0017844158187033214,
+      "learning_rate": 3.2130325814536343e-06,
+      "loss": 0.0,
+      "step": 11037
+    },
+    {
+      "epoch": 0.7113488432042276,
+      "grad_norm": 0.29092505665033047,
+      "learning_rate": 3.2123165055495886e-06,
+      "loss": 0.0003,
+      "step": 11038
+    },
+    {
+      "epoch": 0.7114132886511568,
+      "grad_norm": 0.03781926653402004,
+      "learning_rate": 3.211600429645543e-06,
+      "loss": 0.0001,
+      "step": 11039
+    },
+    {
+      "epoch": 0.711477734098086,
+      "grad_norm": 0.023017320635540486,
+      "learning_rate": 3.210884353741497e-06,
+      "loss": 0.0,
+      "step": 11040
+    },
+    {
+      "epoch": 0.7115421795450152,
+      "grad_norm": 0.2787249919125269,
+      "learning_rate": 3.210168277837451e-06,
+      "loss": 0.001,
+      "step": 11041
+    },
+    {
+      "epoch": 0.7116066249919443,
+      "grad_norm": 0.0002637866973357019,
+      "learning_rate": 3.209452201933405e-06,
+      "loss": 0.0,
+      "step": 11042
+    },
+    {
+      "epoch": 0.7116710704388735,
+      "grad_norm": 0.003413300700603889,
+      "learning_rate": 3.2087361260293593e-06,
+      "loss": 0.0,
+      "step": 11043
+    },
+    {
+      "epoch": 0.7117355158858026,
+      "grad_norm": 0.008288206540353161,
+      "learning_rate": 3.2080200501253136e-06,
+      "loss": 0.0,
+      "step": 11044
+    },
+    {
+      "epoch": 0.7117999613327318,
+      "grad_norm": 0.0009564137452828161,
+      "learning_rate": 3.2073039742212674e-06,
+      "loss": 0.0,
+      "step": 11045
+    },
+    {
+      "epoch": 0.711864406779661,
+      "grad_norm": 0.011072656092947879,
+      "learning_rate": 3.2065878983172217e-06,
+      "loss": 0.0,
+      "step": 11046
+    },
+    {
+      "epoch": 0.7119288522265902,
+      "grad_norm": 0.1796167379448892,
+      "learning_rate": 3.2058718224131765e-06,
+      "loss": 0.0003,
+      "step": 11047
+    },
+    {
+      "epoch": 0.7119932976735194,
+      "grad_norm": 0.07329979538349857,
+      "learning_rate": 3.2051557465091303e-06,
+      "loss": 0.0001,
+      "step": 11048
+    },
+    {
+      "epoch": 0.7120577431204486,
+      "grad_norm": 0.1903261245429498,
+      "learning_rate": 3.2044396706050846e-06,
+      "loss": 0.002,
+      "step": 11049
+    },
+    {
+      "epoch": 0.7121221885673777,
+      "grad_norm": 0.0020960073602700443,
+      "learning_rate": 3.2037235947010385e-06,
+      "loss": 0.0,
+      "step": 11050
+    },
+    {
+      "epoch": 0.7121866340143069,
+      "grad_norm": 0.15443077795960122,
+      "learning_rate": 3.203007518796993e-06,
+      "loss": 0.002,
+      "step": 11051
+    },
+    {
+      "epoch": 0.7122510794612361,
+      "grad_norm": 0.006115973097704426,
+      "learning_rate": 3.202291442892947e-06,
+      "loss": 0.0,
+      "step": 11052
+    },
+    {
+      "epoch": 0.7123155249081652,
+      "grad_norm": 0.00581395781401189,
+      "learning_rate": 3.201575366988901e-06,
+      "loss": 0.0,
+      "step": 11053
+    },
+    {
+      "epoch": 0.7123799703550944,
+      "grad_norm": 0.0022123917383665612,
+      "learning_rate": 3.2008592910848553e-06,
+      "loss": 0.0,
+      "step": 11054
+    },
+    {
+      "epoch": 0.7124444158020236,
+      "grad_norm": 0.00022742322294811237,
+      "learning_rate": 3.200143215180809e-06,
+      "loss": 0.0,
+      "step": 11055
+    },
+    {
+      "epoch": 0.7125088612489527,
+      "grad_norm": 0.030732401054467395,
+      "learning_rate": 3.1994271392767635e-06,
+      "loss": 0.0,
+      "step": 11056
+    },
+    {
+      "epoch": 0.7125733066958819,
+      "grad_norm": 0.0022031101795498163,
+      "learning_rate": 3.1987110633727173e-06,
+      "loss": 0.0,
+      "step": 11057
+    },
+    {
+      "epoch": 0.7126377521428111,
+      "grad_norm": 0.05268874666893051,
+      "learning_rate": 3.197994987468672e-06,
+      "loss": 0.0001,
+      "step": 11058
+    },
+    {
+      "epoch": 0.7127021975897403,
+      "grad_norm": 0.000596779920373873,
+      "learning_rate": 3.1972789115646264e-06,
+      "loss": 0.0,
+      "step": 11059
+    },
+    {
+      "epoch": 0.7127666430366695,
+      "grad_norm": 0.015269021418092311,
+      "learning_rate": 3.1965628356605807e-06,
+      "loss": 0.0,
+      "step": 11060
+    },
+    {
+      "epoch": 0.7128310884835987,
+      "grad_norm": 0.0007433832035599262,
+      "learning_rate": 3.1958467597565345e-06,
+      "loss": 0.0,
+      "step": 11061
+    },
+    {
+      "epoch": 0.7128955339305278,
+      "grad_norm": 0.00024632248124767153,
+      "learning_rate": 3.195130683852489e-06,
+      "loss": 0.0,
+      "step": 11062
+    },
+    {
+      "epoch": 0.712959979377457,
+      "grad_norm": 6.792495538918946e-05,
+      "learning_rate": 3.1944146079484427e-06,
+      "loss": 0.0,
+      "step": 11063
+    },
+    {
+      "epoch": 0.7130244248243861,
+      "grad_norm": 0.0007723524883490751,
+      "learning_rate": 3.193698532044397e-06,
+      "loss": 0.0,
+      "step": 11064
+    },
+    {
+      "epoch": 0.7130888702713153,
+      "grad_norm": 0.00046314913899469146,
+      "learning_rate": 3.192982456140351e-06,
+      "loss": 0.0,
+      "step": 11065
+    },
+    {
+      "epoch": 0.7131533157182445,
+      "grad_norm": 0.0002443464187616548,
+      "learning_rate": 3.192266380236305e-06,
+      "loss": 0.0,
+      "step": 11066
+    },
+    {
+      "epoch": 0.7132177611651737,
+      "grad_norm": 0.03511749555454775,
+      "learning_rate": 3.1915503043322595e-06,
+      "loss": 0.0,
+      "step": 11067
+    },
+    {
+      "epoch": 0.7132822066121028,
+      "grad_norm": 0.08750064745305847,
+      "learning_rate": 3.1908342284282134e-06,
+      "loss": 0.0001,
+      "step": 11068
+    },
+    {
+      "epoch": 0.713346652059032,
+      "grad_norm": 0.0007083306043677525,
+      "learning_rate": 3.1901181525241677e-06,
+      "loss": 0.0,
+      "step": 11069
+    },
+    {
+      "epoch": 0.7134110975059612,
+      "grad_norm": 0.004821037321704811,
+      "learning_rate": 3.1894020766201224e-06,
+      "loss": 0.0,
+      "step": 11070
+    },
+    {
+      "epoch": 0.7134755429528904,
+      "grad_norm": 6.842244899806434e-05,
+      "learning_rate": 3.1886860007160762e-06,
+      "loss": 0.0,
+      "step": 11071
+    },
+    {
+      "epoch": 0.7135399883998196,
+      "grad_norm": 0.19849096388992776,
+      "learning_rate": 3.1879699248120305e-06,
+      "loss": 0.0017,
+      "step": 11072
+    },
+    {
+      "epoch": 0.7136044338467488,
+      "grad_norm": 0.001744117216499498,
+      "learning_rate": 3.1872538489079844e-06,
+      "loss": 0.0,
+      "step": 11073
+    },
+    {
+      "epoch": 0.713668879293678,
+      "grad_norm": 0.0007771459762669267,
+      "learning_rate": 3.1865377730039387e-06,
+      "loss": 0.0,
+      "step": 11074
+    },
+    {
+      "epoch": 0.713733324740607,
+      "grad_norm": 0.0016081975587549844,
+      "learning_rate": 3.185821697099893e-06,
+      "loss": 0.0,
+      "step": 11075
+    },
+    {
+      "epoch": 0.7137977701875362,
+      "grad_norm": 0.09325613134932795,
+      "learning_rate": 3.185105621195847e-06,
+      "loss": 0.0002,
+      "step": 11076
+    },
+    {
+      "epoch": 0.7138622156344654,
+      "grad_norm": 0.332418115417522,
+      "learning_rate": 3.184389545291801e-06,
+      "loss": 0.0014,
+      "step": 11077
+    },
+    {
+      "epoch": 0.7139266610813946,
+      "grad_norm": 0.04041215457695259,
+      "learning_rate": 3.183673469387755e-06,
+      "loss": 0.0001,
+      "step": 11078
+    },
+    {
+      "epoch": 0.7139911065283238,
+      "grad_norm": 0.08855809440335718,
+      "learning_rate": 3.1829573934837094e-06,
+      "loss": 0.0016,
+      "step": 11079
+    },
+    {
+      "epoch": 0.714055551975253,
+      "grad_norm": 0.037115998367209524,
+      "learning_rate": 3.1822413175796637e-06,
+      "loss": 0.0001,
+      "step": 11080
+    },
+    {
+      "epoch": 0.7141199974221821,
+      "grad_norm": 0.0009066699956542551,
+      "learning_rate": 3.1815252416756175e-06,
+      "loss": 0.0,
+      "step": 11081
+    },
+    {
+      "epoch": 0.7141844428691113,
+      "grad_norm": 0.019780705026331263,
+      "learning_rate": 3.1808091657715723e-06,
+      "loss": 0.0,
+      "step": 11082
+    },
+    {
+      "epoch": 0.7142488883160405,
+      "grad_norm": 0.00261057496389828,
+      "learning_rate": 3.1800930898675266e-06,
+      "loss": 0.0,
+      "step": 11083
+    },
+    {
+      "epoch": 0.7143133337629697,
+      "grad_norm": 0.004270109738808761,
+      "learning_rate": 3.1793770139634804e-06,
+      "loss": 0.0,
+      "step": 11084
+    },
+    {
+      "epoch": 0.7143777792098989,
+      "grad_norm": 0.00031775442414158883,
+      "learning_rate": 3.1786609380594347e-06,
+      "loss": 0.0,
+      "step": 11085
+    },
+    {
+      "epoch": 0.7144422246568279,
+      "grad_norm": 0.12453918901981105,
+      "learning_rate": 3.1779448621553886e-06,
+      "loss": 0.0002,
+      "step": 11086
+    },
+    {
+      "epoch": 0.7145066701037571,
+      "grad_norm": 0.02119295088790315,
+      "learning_rate": 3.177228786251343e-06,
+      "loss": 0.0,
+      "step": 11087
+    },
+    {
+      "epoch": 0.7145711155506863,
+      "grad_norm": 0.0006572972137359756,
+      "learning_rate": 3.176512710347297e-06,
+      "loss": 0.0,
+      "step": 11088
+    },
+    {
+      "epoch": 0.7146355609976155,
+      "grad_norm": 0.0015856123993797646,
+      "learning_rate": 3.175796634443251e-06,
+      "loss": 0.0,
+      "step": 11089
+    },
+    {
+      "epoch": 0.7147000064445447,
+      "grad_norm": 0.005097678348488337,
+      "learning_rate": 3.1750805585392054e-06,
+      "loss": 0.0,
+      "step": 11090
+    },
+    {
+      "epoch": 0.7147644518914739,
+      "grad_norm": 0.03935784061800326,
+      "learning_rate": 3.1743644826351593e-06,
+      "loss": 0.0001,
+      "step": 11091
+    },
+    {
+      "epoch": 0.714828897338403,
+      "grad_norm": 0.00048666767231215894,
+      "learning_rate": 3.1736484067311136e-06,
+      "loss": 0.0,
+      "step": 11092
+    },
+    {
+      "epoch": 0.7148933427853322,
+      "grad_norm": 0.004039960942877218,
+      "learning_rate": 3.1729323308270683e-06,
+      "loss": 0.0,
+      "step": 11093
+    },
+    {
+      "epoch": 0.7149577882322614,
+      "grad_norm": 0.05651111224218124,
+      "learning_rate": 3.172216254923022e-06,
+      "loss": 0.0001,
+      "step": 11094
+    },
+    {
+      "epoch": 0.7150222336791906,
+      "grad_norm": 0.0012988435437589815,
+      "learning_rate": 3.1715001790189765e-06,
+      "loss": 0.0,
+      "step": 11095
+    },
+    {
+      "epoch": 0.7150866791261198,
+      "grad_norm": 0.0006677970549519835,
+      "learning_rate": 3.1707841031149308e-06,
+      "loss": 0.0,
+      "step": 11096
+    },
+    {
+      "epoch": 0.715151124573049,
+      "grad_norm": 0.0030971611143586165,
+      "learning_rate": 3.1700680272108846e-06,
+      "loss": 0.0,
+      "step": 11097
+    },
+    {
+      "epoch": 0.715215570019978,
+      "grad_norm": 0.05772528771724839,
+      "learning_rate": 3.169351951306839e-06,
+      "loss": 0.0001,
+      "step": 11098
+    },
+    {
+      "epoch": 0.7152800154669072,
+      "grad_norm": 0.0017996675808251306,
+      "learning_rate": 3.168635875402793e-06,
+      "loss": 0.0,
+      "step": 11099
+    },
+    {
+      "epoch": 0.7153444609138364,
+      "grad_norm": 0.0022183191323614845,
+      "learning_rate": 3.167919799498747e-06,
+      "loss": 0.0,
+      "step": 11100
+    },
+    {
+      "epoch": 0.7154089063607656,
+      "grad_norm": 0.23467454047809808,
+      "learning_rate": 3.167203723594701e-06,
+      "loss": 0.0058,
+      "step": 11101
+    },
+    {
+      "epoch": 0.7154733518076948,
+      "grad_norm": 0.002101152624454598,
+      "learning_rate": 3.1664876476906553e-06,
+      "loss": 0.0,
+      "step": 11102
+    },
+    {
+      "epoch": 0.715537797254624,
+      "grad_norm": 0.10132137891567196,
+      "learning_rate": 3.1657715717866096e-06,
+      "loss": 0.0002,
+      "step": 11103
+    },
+    {
+      "epoch": 0.7156022427015531,
+      "grad_norm": 0.0013626302825582128,
+      "learning_rate": 3.1650554958825634e-06,
+      "loss": 0.0,
+      "step": 11104
+    },
+    {
+      "epoch": 0.7156666881484823,
+      "grad_norm": 0.013922212605411412,
+      "learning_rate": 3.164339419978518e-06,
+      "loss": 0.0,
+      "step": 11105
+    },
+    {
+      "epoch": 0.7157311335954115,
+      "grad_norm": 0.5795495550330169,
+      "learning_rate": 3.1636233440744725e-06,
+      "loss": 0.0019,
+      "step": 11106
+    },
+    {
+      "epoch": 0.7157955790423407,
+      "grad_norm": 0.019280375892706227,
+      "learning_rate": 3.1629072681704263e-06,
+      "loss": 0.0,
+      "step": 11107
+    },
+    {
+      "epoch": 0.7158600244892699,
+      "grad_norm": 3.448348121566918e-05,
+      "learning_rate": 3.1621911922663806e-06,
+      "loss": 0.0,
+      "step": 11108
+    },
+    {
+      "epoch": 0.715924469936199,
+      "grad_norm": 0.37455556282125935,
+      "learning_rate": 3.1614751163623345e-06,
+      "loss": 0.0016,
+      "step": 11109
+    },
+    {
+      "epoch": 0.7159889153831281,
+      "grad_norm": 0.002586349361368294,
+      "learning_rate": 3.160759040458289e-06,
+      "loss": 0.0,
+      "step": 11110
+    },
+    {
+      "epoch": 0.7160533608300573,
+      "grad_norm": 0.008448127471088534,
+      "learning_rate": 3.160042964554243e-06,
+      "loss": 0.0,
+      "step": 11111
+    },
+    {
+      "epoch": 0.7161178062769865,
+      "grad_norm": 0.0010937285941726932,
+      "learning_rate": 3.159326888650197e-06,
+      "loss": 0.0,
+      "step": 11112
+    },
+    {
+      "epoch": 0.7161822517239157,
+      "grad_norm": 0.0012277663763215767,
+      "learning_rate": 3.1586108127461513e-06,
+      "loss": 0.0,
+      "step": 11113
+    },
+    {
+      "epoch": 0.7162466971708449,
+      "grad_norm": 0.0018488032473446409,
+      "learning_rate": 3.157894736842105e-06,
+      "loss": 0.0,
+      "step": 11114
+    },
+    {
+      "epoch": 0.7163111426177741,
+      "grad_norm": 0.08111634839048355,
+      "learning_rate": 3.1571786609380595e-06,
+      "loss": 0.0001,
+      "step": 11115
+    },
+    {
+      "epoch": 0.7163755880647033,
+      "grad_norm": 0.0031572357244101274,
+      "learning_rate": 3.1564625850340138e-06,
+      "loss": 0.0,
+      "step": 11116
+    },
+    {
+      "epoch": 0.7164400335116324,
+      "grad_norm": 0.0014554546438801737,
+      "learning_rate": 3.155746509129968e-06,
+      "loss": 0.0,
+      "step": 11117
+    },
+    {
+      "epoch": 0.7165044789585616,
+      "grad_norm": 0.028527429420330017,
+      "learning_rate": 3.1550304332259224e-06,
+      "loss": 0.0,
+      "step": 11118
+    },
+    {
+      "epoch": 0.7165689244054908,
+      "grad_norm": 0.0013177336847458132,
+      "learning_rate": 3.1543143573218767e-06,
+      "loss": 0.0,
+      "step": 11119
+    },
+    {
+      "epoch": 0.7166333698524199,
+      "grad_norm": 0.02391522038213578,
+      "learning_rate": 3.1535982814178305e-06,
+      "loss": 0.0001,
+      "step": 11120
+    },
+    {
+      "epoch": 0.7166978152993491,
+      "grad_norm": 0.001059440528141261,
+      "learning_rate": 3.152882205513785e-06,
+      "loss": 0.0,
+      "step": 11121
+    },
+    {
+      "epoch": 0.7167622607462782,
+      "grad_norm": 0.001678356185974784,
+      "learning_rate": 3.1521661296097387e-06,
+      "loss": 0.0,
+      "step": 11122
+    },
+    {
+      "epoch": 0.7168267061932074,
+      "grad_norm": 0.00032899135591734686,
+      "learning_rate": 3.151450053705693e-06,
+      "loss": 0.0,
+      "step": 11123
+    },
+    {
+      "epoch": 0.7168911516401366,
+      "grad_norm": 0.0042420537766143995,
+      "learning_rate": 3.1507339778016473e-06,
+      "loss": 0.0,
+      "step": 11124
+    },
+    {
+      "epoch": 0.7169555970870658,
+      "grad_norm": 0.06737302971743062,
+      "learning_rate": 3.150017901897601e-06,
+      "loss": 0.0001,
+      "step": 11125
+    },
+    {
+      "epoch": 0.717020042533995,
+      "grad_norm": 0.10045379921951969,
+      "learning_rate": 3.1493018259935555e-06,
+      "loss": 0.0002,
+      "step": 11126
+    },
+    {
+      "epoch": 0.7170844879809242,
+      "grad_norm": 0.03187420971150182,
+      "learning_rate": 3.1485857500895094e-06,
+      "loss": 0.0002,
+      "step": 11127
+    },
+    {
+      "epoch": 0.7171489334278534,
+      "grad_norm": 0.001211750648026592,
+      "learning_rate": 3.147869674185464e-06,
+      "loss": 0.0,
+      "step": 11128
+    },
+    {
+      "epoch": 0.7172133788747825,
+      "grad_norm": 0.008495406179175139,
+      "learning_rate": 3.1471535982814184e-06,
+      "loss": 0.0,
+      "step": 11129
+    },
+    {
+      "epoch": 0.7172778243217117,
+      "grad_norm": 0.0019832293144138556,
+      "learning_rate": 3.1464375223773722e-06,
+      "loss": 0.0,
+      "step": 11130
+    },
+    {
+      "epoch": 0.7173422697686408,
+      "grad_norm": 0.001240237132151751,
+      "learning_rate": 3.1457214464733265e-06,
+      "loss": 0.0,
+      "step": 11131
+    },
+    {
+      "epoch": 0.71740671521557,
+      "grad_norm": 0.0025074924361898395,
+      "learning_rate": 3.145005370569281e-06,
+      "loss": 0.0,
+      "step": 11132
+    },
+    {
+      "epoch": 0.7174711606624992,
+      "grad_norm": 0.0027266612771010145,
+      "learning_rate": 3.1442892946652347e-06,
+      "loss": 0.0,
+      "step": 11133
+    },
+    {
+      "epoch": 0.7175356061094283,
+      "grad_norm": 0.05183947841060524,
+      "learning_rate": 3.143573218761189e-06,
+      "loss": 0.0001,
+      "step": 11134
+    },
+    {
+      "epoch": 0.7176000515563575,
+      "grad_norm": 0.00013604534610558102,
+      "learning_rate": 3.142857142857143e-06,
+      "loss": 0.0,
+      "step": 11135
+    },
+    {
+      "epoch": 0.7176644970032867,
+      "grad_norm": 0.004464684547835725,
+      "learning_rate": 3.142141066953097e-06,
+      "loss": 0.0,
+      "step": 11136
+    },
+    {
+      "epoch": 0.7177289424502159,
+      "grad_norm": 0.043388709463875526,
+      "learning_rate": 3.1414249910490515e-06,
+      "loss": 0.0002,
+      "step": 11137
+    },
+    {
+      "epoch": 0.7177933878971451,
+      "grad_norm": 0.563788289759631,
+      "learning_rate": 3.1407089151450054e-06,
+      "loss": 0.0048,
+      "step": 11138
+    },
+    {
+      "epoch": 0.7178578333440743,
+      "grad_norm": 0.00031376440967617104,
+      "learning_rate": 3.1399928392409597e-06,
+      "loss": 0.0,
+      "step": 11139
+    },
+    {
+      "epoch": 0.7179222787910035,
+      "grad_norm": 0.0009322943557485454,
+      "learning_rate": 3.1392767633369144e-06,
+      "loss": 0.0,
+      "step": 11140
+    },
+    {
+      "epoch": 0.7179867242379326,
+      "grad_norm": 0.2445635070487584,
+      "learning_rate": 3.1385606874328683e-06,
+      "loss": 0.0011,
+      "step": 11141
+    },
+    {
+      "epoch": 0.7180511696848617,
+      "grad_norm": 0.0009303590616047754,
+      "learning_rate": 3.1378446115288226e-06,
+      "loss": 0.0,
+      "step": 11142
+    },
+    {
+      "epoch": 0.7181156151317909,
+      "grad_norm": 0.002144644973596857,
+      "learning_rate": 3.1371285356247764e-06,
+      "loss": 0.0,
+      "step": 11143
+    },
+    {
+      "epoch": 0.7181800605787201,
+      "grad_norm": 0.24049082303382727,
+      "learning_rate": 3.1364124597207307e-06,
+      "loss": 0.0005,
+      "step": 11144
+    },
+    {
+      "epoch": 0.7182445060256493,
+      "grad_norm": 0.010361287644884275,
+      "learning_rate": 3.135696383816685e-06,
+      "loss": 0.0001,
+      "step": 11145
+    },
+    {
+      "epoch": 0.7183089514725784,
+      "grad_norm": 0.00045569104002404434,
+      "learning_rate": 3.134980307912639e-06,
+      "loss": 0.0,
+      "step": 11146
+    },
+    {
+      "epoch": 0.7183733969195076,
+      "grad_norm": 0.1184363656128077,
+      "learning_rate": 3.134264232008593e-06,
+      "loss": 0.0013,
+      "step": 11147
+    },
+    {
+      "epoch": 0.7184378423664368,
+      "grad_norm": 0.0018075368988239252,
+      "learning_rate": 3.133548156104547e-06,
+      "loss": 0.0,
+      "step": 11148
+    },
+    {
+      "epoch": 0.718502287813366,
+      "grad_norm": 0.00025330471947959053,
+      "learning_rate": 3.1328320802005014e-06,
+      "loss": 0.0,
+      "step": 11149
+    },
+    {
+      "epoch": 0.7185667332602952,
+      "grad_norm": 0.0018817507730820291,
+      "learning_rate": 3.1321160042964553e-06,
+      "loss": 0.0,
+      "step": 11150
+    },
+    {
+      "epoch": 0.7186311787072244,
+      "grad_norm": 0.14699625858626997,
+      "learning_rate": 3.1313999283924096e-06,
+      "loss": 0.001,
+      "step": 11151
+    },
+    {
+      "epoch": 0.7186956241541536,
+      "grad_norm": 0.010185056320389333,
+      "learning_rate": 3.1306838524883643e-06,
+      "loss": 0.0015,
+      "step": 11152
+    },
+    {
+      "epoch": 0.7187600696010826,
+      "grad_norm": 0.00017482119536334955,
+      "learning_rate": 3.1299677765843186e-06,
+      "loss": 0.0,
+      "step": 11153
+    },
+    {
+      "epoch": 0.7188245150480118,
+      "grad_norm": 0.008587165702478385,
+      "learning_rate": 3.1292517006802725e-06,
+      "loss": 0.0,
+      "step": 11154
+    },
+    {
+      "epoch": 0.718888960494941,
+      "grad_norm": 0.0009752747086168306,
+      "learning_rate": 3.1285356247762268e-06,
+      "loss": 0.0,
+      "step": 11155
+    },
+    {
+      "epoch": 0.7189534059418702,
+      "grad_norm": 0.00273619332827515,
+      "learning_rate": 3.1278195488721806e-06,
+      "loss": 0.0,
+      "step": 11156
+    },
+    {
+      "epoch": 0.7190178513887994,
+      "grad_norm": 0.0003457374189358038,
+      "learning_rate": 3.127103472968135e-06,
+      "loss": 0.0,
+      "step": 11157
+    },
+    {
+      "epoch": 0.7190822968357286,
+      "grad_norm": 0.001673470044109342,
+      "learning_rate": 3.126387397064089e-06,
+      "loss": 0.0,
+      "step": 11158
+    },
+    {
+      "epoch": 0.7191467422826577,
+      "grad_norm": 0.00021614544359388704,
+      "learning_rate": 3.125671321160043e-06,
+      "loss": 0.0,
+      "step": 11159
+    },
+    {
+      "epoch": 0.7192111877295869,
+      "grad_norm": 0.0036123325054397515,
+      "learning_rate": 3.1249552452559974e-06,
+      "loss": 0.0,
+      "step": 11160
+    },
+    {
+      "epoch": 0.7192756331765161,
+      "grad_norm": 0.01439849637720543,
+      "learning_rate": 3.1242391693519513e-06,
+      "loss": 0.0,
+      "step": 11161
+    },
+    {
+      "epoch": 0.7193400786234453,
+      "grad_norm": 0.0011721288297011038,
+      "learning_rate": 3.1235230934479056e-06,
+      "loss": 0.0,
+      "step": 11162
+    },
+    {
+      "epoch": 0.7194045240703745,
+      "grad_norm": 0.004723643909055465,
+      "learning_rate": 3.1228070175438603e-06,
+      "loss": 0.0,
+      "step": 11163
+    },
+    {
+      "epoch": 0.7194689695173035,
+      "grad_norm": 0.0035602629306015162,
+      "learning_rate": 3.122090941639814e-06,
+      "loss": 0.0,
+      "step": 11164
+    },
+    {
+      "epoch": 0.7195334149642327,
+      "grad_norm": 0.0003956756912840429,
+      "learning_rate": 3.1213748657357685e-06,
+      "loss": 0.0,
+      "step": 11165
+    },
+    {
+      "epoch": 0.7195978604111619,
+      "grad_norm": 0.001660037121350968,
+      "learning_rate": 3.1206587898317223e-06,
+      "loss": 0.0,
+      "step": 11166
+    },
+    {
+      "epoch": 0.7196623058580911,
+      "grad_norm": 0.00039859561483115464,
+      "learning_rate": 3.1199427139276766e-06,
+      "loss": 0.0,
+      "step": 11167
+    },
+    {
+      "epoch": 0.7197267513050203,
+      "grad_norm": 0.0060584983061866965,
+      "learning_rate": 3.119226638023631e-06,
+      "loss": 0.0001,
+      "step": 11168
+    },
+    {
+      "epoch": 0.7197911967519495,
+      "grad_norm": 0.037109889478630684,
+      "learning_rate": 3.118510562119585e-06,
+      "loss": 0.0,
+      "step": 11169
+    },
+    {
+      "epoch": 0.7198556421988787,
+      "grad_norm": 0.00727883548794619,
+      "learning_rate": 3.117794486215539e-06,
+      "loss": 0.0001,
+      "step": 11170
+    },
+    {
+      "epoch": 0.7199200876458078,
+      "grad_norm": 0.03501727862860073,
+      "learning_rate": 3.117078410311493e-06,
+      "loss": 0.0,
+      "step": 11171
+    },
+    {
+      "epoch": 0.719984533092737,
+      "grad_norm": 8.61872471854203e-05,
+      "learning_rate": 3.1163623344074473e-06,
+      "loss": 0.0,
+      "step": 11172
+    },
+    {
+      "epoch": 0.7200489785396662,
+      "grad_norm": 0.045333885416277295,
+      "learning_rate": 3.1156462585034016e-06,
+      "loss": 0.0001,
+      "step": 11173
+    },
+    {
+      "epoch": 0.7201134239865954,
+      "grad_norm": 0.0004629402319689133,
+      "learning_rate": 3.1149301825993555e-06,
+      "loss": 0.0,
+      "step": 11174
+    },
+    {
+      "epoch": 0.7201778694335246,
+      "grad_norm": 0.001063566655592568,
+      "learning_rate": 3.11421410669531e-06,
+      "loss": 0.0,
+      "step": 11175
+    },
+    {
+      "epoch": 0.7202423148804536,
+      "grad_norm": 0.0014662245610195432,
+      "learning_rate": 3.1134980307912645e-06,
+      "loss": 0.0,
+      "step": 11176
+    },
+    {
+      "epoch": 0.7203067603273828,
+      "grad_norm": 0.01931647362060871,
+      "learning_rate": 3.1127819548872184e-06,
+      "loss": 0.0,
+      "step": 11177
+    },
+    {
+      "epoch": 0.720371205774312,
+      "grad_norm": 0.0004281946405760085,
+      "learning_rate": 3.1120658789831727e-06,
+      "loss": 0.0,
+      "step": 11178
+    },
+    {
+      "epoch": 0.7204356512212412,
+      "grad_norm": 7.321520534062253e-05,
+      "learning_rate": 3.1113498030791265e-06,
+      "loss": 0.0,
+      "step": 11179
+    },
+    {
+      "epoch": 0.7205000966681704,
+      "grad_norm": 8.563287008322675e-05,
+      "learning_rate": 3.110633727175081e-06,
+      "loss": 0.0,
+      "step": 11180
+    },
+    {
+      "epoch": 0.7205645421150996,
+      "grad_norm": 0.00019715790289080482,
+      "learning_rate": 3.109917651271035e-06,
+      "loss": 0.0,
+      "step": 11181
+    },
+    {
+      "epoch": 0.7206289875620288,
+      "grad_norm": 0.002634901520461059,
+      "learning_rate": 3.109201575366989e-06,
+      "loss": 0.0,
+      "step": 11182
+    },
+    {
+      "epoch": 0.7206934330089579,
+      "grad_norm": 0.0008865863108786099,
+      "learning_rate": 3.1084854994629433e-06,
+      "loss": 0.0,
+      "step": 11183
+    },
+    {
+      "epoch": 0.7207578784558871,
+      "grad_norm": 0.025815671687311366,
+      "learning_rate": 3.107769423558897e-06,
+      "loss": 0.0,
+      "step": 11184
+    },
+    {
+      "epoch": 0.7208223239028163,
+      "grad_norm": 0.0032002885997212812,
+      "learning_rate": 3.1070533476548515e-06,
+      "loss": 0.0,
+      "step": 11185
+    },
+    {
+      "epoch": 0.7208867693497455,
+      "grad_norm": 0.0004901728854833662,
+      "learning_rate": 3.1063372717508058e-06,
+      "loss": 0.0,
+      "step": 11186
+    },
+    {
+      "epoch": 0.7209512147966746,
+      "grad_norm": 0.00753785072550071,
+      "learning_rate": 3.10562119584676e-06,
+      "loss": 0.0,
+      "step": 11187
+    },
+    {
+      "epoch": 0.7210156602436038,
+      "grad_norm": 0.001151215509133265,
+      "learning_rate": 3.1049051199427144e-06,
+      "loss": 0.0,
+      "step": 11188
+    },
+    {
+      "epoch": 0.7210801056905329,
+      "grad_norm": 0.00012703756509971546,
+      "learning_rate": 3.1041890440386687e-06,
+      "loss": 0.0,
+      "step": 11189
+    },
+    {
+      "epoch": 0.7211445511374621,
+      "grad_norm": 0.00027725036286674646,
+      "learning_rate": 3.1034729681346225e-06,
+      "loss": 0.0,
+      "step": 11190
+    },
+    {
+      "epoch": 0.7212089965843913,
+      "grad_norm": 0.00032789403026681533,
+      "learning_rate": 3.102756892230577e-06,
+      "loss": 0.0,
+      "step": 11191
+    },
+    {
+      "epoch": 0.7212734420313205,
+      "grad_norm": 0.5477644907254653,
+      "learning_rate": 3.1020408163265307e-06,
+      "loss": 0.0013,
+      "step": 11192
+    },
+    {
+      "epoch": 0.7213378874782497,
+      "grad_norm": 0.022567781263925505,
+      "learning_rate": 3.101324740422485e-06,
+      "loss": 0.0003,
+      "step": 11193
+    },
+    {
+      "epoch": 0.7214023329251789,
+      "grad_norm": 0.00021412049244247774,
+      "learning_rate": 3.1006086645184393e-06,
+      "loss": 0.0,
+      "step": 11194
+    },
+    {
+      "epoch": 0.721466778372108,
+      "grad_norm": 0.04202279788293058,
+      "learning_rate": 3.099892588614393e-06,
+      "loss": 0.0001,
+      "step": 11195
+    },
+    {
+      "epoch": 0.7215312238190372,
+      "grad_norm": 0.004843800760419462,
+      "learning_rate": 3.0991765127103475e-06,
+      "loss": 0.0,
+      "step": 11196
+    },
+    {
+      "epoch": 0.7215956692659664,
+      "grad_norm": 0.002510341090575078,
+      "learning_rate": 3.0984604368063014e-06,
+      "loss": 0.0,
+      "step": 11197
+    },
+    {
+      "epoch": 0.7216601147128955,
+      "grad_norm": 0.01219112790806607,
+      "learning_rate": 3.097744360902256e-06,
+      "loss": 0.0001,
+      "step": 11198
+    },
+    {
+      "epoch": 0.7217245601598247,
+      "grad_norm": 6.304089065724722e-05,
+      "learning_rate": 3.0970282849982104e-06,
+      "loss": 0.0,
+      "step": 11199
+    },
+    {
+      "epoch": 0.7217890056067539,
+      "grad_norm": 0.5504341113723744,
+      "learning_rate": 3.0963122090941643e-06,
+      "loss": 0.002,
+      "step": 11200
+    },
+    {
+      "epoch": 0.721853451053683,
+      "grad_norm": 0.01023833832935001,
+      "learning_rate": 3.0955961331901186e-06,
+      "loss": 0.0001,
+      "step": 11201
+    },
+    {
+      "epoch": 0.7219178965006122,
+      "grad_norm": 0.1783995912772252,
+      "learning_rate": 3.094880057286073e-06,
+      "loss": 0.0037,
+      "step": 11202
+    },
+    {
+      "epoch": 0.7219823419475414,
+      "grad_norm": 0.0008514347736365616,
+      "learning_rate": 3.0941639813820267e-06,
+      "loss": 0.0,
+      "step": 11203
+    },
+    {
+      "epoch": 0.7220467873944706,
+      "grad_norm": 0.0005467954296606304,
+      "learning_rate": 3.093447905477981e-06,
+      "loss": 0.0,
+      "step": 11204
+    },
+    {
+      "epoch": 0.7221112328413998,
+      "grad_norm": 0.12932414173611667,
+      "learning_rate": 3.092731829573935e-06,
+      "loss": 0.0003,
+      "step": 11205
+    },
+    {
+      "epoch": 0.722175678288329,
+      "grad_norm": 0.00013929640553021798,
+      "learning_rate": 3.092015753669889e-06,
+      "loss": 0.0,
+      "step": 11206
+    },
+    {
+      "epoch": 0.7222401237352581,
+      "grad_norm": 0.14328180054046125,
+      "learning_rate": 3.091299677765843e-06,
+      "loss": 0.0006,
+      "step": 11207
+    },
+    {
+      "epoch": 0.7223045691821873,
+      "grad_norm": 0.0005452546115663261,
+      "learning_rate": 3.0905836018617974e-06,
+      "loss": 0.0,
+      "step": 11208
+    },
+    {
+      "epoch": 0.7223690146291164,
+      "grad_norm": 0.00024169194708091167,
+      "learning_rate": 3.0898675259577517e-06,
+      "loss": 0.0,
+      "step": 11209
+    },
+    {
+      "epoch": 0.7224334600760456,
+      "grad_norm": 0.004371841536006174,
+      "learning_rate": 3.0891514500537064e-06,
+      "loss": 0.0,
+      "step": 11210
+    },
+    {
+      "epoch": 0.7224979055229748,
+      "grad_norm": 0.0039695361977091495,
+      "learning_rate": 3.0884353741496603e-06,
+      "loss": 0.0,
+      "step": 11211
+    },
+    {
+      "epoch": 0.722562350969904,
+      "grad_norm": 0.001597297508301907,
+      "learning_rate": 3.0877192982456146e-06,
+      "loss": 0.0,
+      "step": 11212
+    },
+    {
+      "epoch": 0.7226267964168331,
+      "grad_norm": 0.0149499392375087,
+      "learning_rate": 3.0870032223415685e-06,
+      "loss": 0.0001,
+      "step": 11213
+    },
+    {
+      "epoch": 0.7226912418637623,
+      "grad_norm": 0.0009184210155997939,
+      "learning_rate": 3.0862871464375228e-06,
+      "loss": 0.0,
+      "step": 11214
+    },
+    {
+      "epoch": 0.7227556873106915,
+      "grad_norm": 0.011828507087317438,
+      "learning_rate": 3.0855710705334766e-06,
+      "loss": 0.0,
+      "step": 11215
+    },
+    {
+      "epoch": 0.7228201327576207,
+      "grad_norm": 0.004237269538513051,
+      "learning_rate": 3.084854994629431e-06,
+      "loss": 0.0,
+      "step": 11216
+    },
+    {
+      "epoch": 0.7228845782045499,
+      "grad_norm": 0.000772409782086021,
+      "learning_rate": 3.0841389187253852e-06,
+      "loss": 0.0,
+      "step": 11217
+    },
+    {
+      "epoch": 0.7229490236514791,
+      "grad_norm": 0.00760526489143851,
+      "learning_rate": 3.083422842821339e-06,
+      "loss": 0.0001,
+      "step": 11218
+    },
+    {
+      "epoch": 0.7230134690984082,
+      "grad_norm": 0.03619062946544116,
+      "learning_rate": 3.0827067669172934e-06,
+      "loss": 0.0003,
+      "step": 11219
+    },
+    {
+      "epoch": 0.7230779145453373,
+      "grad_norm": 0.0007644458517121425,
+      "learning_rate": 3.0819906910132473e-06,
+      "loss": 0.0,
+      "step": 11220
+    },
+    {
+      "epoch": 0.7231423599922665,
+      "grad_norm": 0.0006187374178444931,
+      "learning_rate": 3.0812746151092016e-06,
+      "loss": 0.0,
+      "step": 11221
+    },
+    {
+      "epoch": 0.7232068054391957,
+      "grad_norm": 0.008779720301754806,
+      "learning_rate": 3.0805585392051563e-06,
+      "loss": 0.0,
+      "step": 11222
+    },
+    {
+      "epoch": 0.7232712508861249,
+      "grad_norm": 0.33317532619060475,
+      "learning_rate": 3.07984246330111e-06,
+      "loss": 0.0028,
+      "step": 11223
+    },
+    {
+      "epoch": 0.7233356963330541,
+      "grad_norm": 0.008626626856931026,
+      "learning_rate": 3.0791263873970645e-06,
+      "loss": 0.0,
+      "step": 11224
+    },
+    {
+      "epoch": 0.7234001417799832,
+      "grad_norm": 0.0015988935817651294,
+      "learning_rate": 3.0784103114930188e-06,
+      "loss": 0.0,
+      "step": 11225
+    },
+    {
+      "epoch": 0.7234645872269124,
+      "grad_norm": 0.019773563400669943,
+      "learning_rate": 3.0776942355889726e-06,
+      "loss": 0.0,
+      "step": 11226
+    },
+    {
+      "epoch": 0.7235290326738416,
+      "grad_norm": 0.0044985562525182805,
+      "learning_rate": 3.076978159684927e-06,
+      "loss": 0.0,
+      "step": 11227
+    },
+    {
+      "epoch": 0.7235934781207708,
+      "grad_norm": 0.001497283255147728,
+      "learning_rate": 3.076262083780881e-06,
+      "loss": 0.0,
+      "step": 11228
+    },
+    {
+      "epoch": 0.7236579235677,
+      "grad_norm": 0.00011553391408470788,
+      "learning_rate": 3.075546007876835e-06,
+      "loss": 0.0,
+      "step": 11229
+    },
+    {
+      "epoch": 0.7237223690146292,
+      "grad_norm": 0.0025702919683202085,
+      "learning_rate": 3.0748299319727894e-06,
+      "loss": 0.0,
+      "step": 11230
+    },
+    {
+      "epoch": 0.7237868144615582,
+      "grad_norm": 3.5742902256819365e-05,
+      "learning_rate": 3.0741138560687433e-06,
+      "loss": 0.0,
+      "step": 11231
+    },
+    {
+      "epoch": 0.7238512599084874,
+      "grad_norm": 0.0008341897101581062,
+      "learning_rate": 3.0733977801646976e-06,
+      "loss": 0.0,
+      "step": 11232
+    },
+    {
+      "epoch": 0.7239157053554166,
+      "grad_norm": 0.0004687527850827818,
+      "learning_rate": 3.0726817042606523e-06,
+      "loss": 0.0,
+      "step": 11233
+    },
+    {
+      "epoch": 0.7239801508023458,
+      "grad_norm": 0.06597886275171361,
+      "learning_rate": 3.071965628356606e-06,
+      "loss": 0.0002,
+      "step": 11234
+    },
+    {
+      "epoch": 0.724044596249275,
+      "grad_norm": 0.006935657033462403,
+      "learning_rate": 3.0712495524525605e-06,
+      "loss": 0.0,
+      "step": 11235
+    },
+    {
+      "epoch": 0.7241090416962042,
+      "grad_norm": 0.08593750812790572,
+      "learning_rate": 3.0705334765485144e-06,
+      "loss": 0.0005,
+      "step": 11236
+    },
+    {
+      "epoch": 0.7241734871431333,
+      "grad_norm": 0.002378809302678618,
+      "learning_rate": 3.0698174006444687e-06,
+      "loss": 0.0,
+      "step": 11237
+    },
+    {
+      "epoch": 0.7242379325900625,
+      "grad_norm": 0.45399070124829183,
+      "learning_rate": 3.069101324740423e-06,
+      "loss": 0.0022,
+      "step": 11238
+    },
+    {
+      "epoch": 0.7243023780369917,
+      "grad_norm": 0.26184424907155546,
+      "learning_rate": 3.068385248836377e-06,
+      "loss": 0.0009,
+      "step": 11239
+    },
+    {
+      "epoch": 0.7243668234839209,
+      "grad_norm": 0.0033589809308402887,
+      "learning_rate": 3.067669172932331e-06,
+      "loss": 0.0,
+      "step": 11240
+    },
+    {
+      "epoch": 0.7244312689308501,
+      "grad_norm": 0.005790182700251807,
+      "learning_rate": 3.066953097028285e-06,
+      "loss": 0.0,
+      "step": 11241
+    },
+    {
+      "epoch": 0.7244957143777793,
+      "grad_norm": 0.0002524290695434493,
+      "learning_rate": 3.0662370211242393e-06,
+      "loss": 0.0,
+      "step": 11242
+    },
+    {
+      "epoch": 0.7245601598247083,
+      "grad_norm": 0.0005162926392841873,
+      "learning_rate": 3.065520945220193e-06,
+      "loss": 0.0,
+      "step": 11243
+    },
+    {
+      "epoch": 0.7246246052716375,
+      "grad_norm": 0.0017082764232077385,
+      "learning_rate": 3.0648048693161475e-06,
+      "loss": 0.0,
+      "step": 11244
+    },
+    {
+      "epoch": 0.7246890507185667,
+      "grad_norm": 0.004266903234189341,
+      "learning_rate": 3.064088793412102e-06,
+      "loss": 0.0,
+      "step": 11245
+    },
+    {
+      "epoch": 0.7247534961654959,
+      "grad_norm": 0.0001998843517710784,
+      "learning_rate": 3.0633727175080565e-06,
+      "loss": 0.0,
+      "step": 11246
+    },
+    {
+      "epoch": 0.7248179416124251,
+      "grad_norm": 0.16482393362610753,
+      "learning_rate": 3.0626566416040104e-06,
+      "loss": 0.0002,
+      "step": 11247
+    },
+    {
+      "epoch": 0.7248823870593543,
+      "grad_norm": 0.00958134520982764,
+      "learning_rate": 3.0619405656999647e-06,
+      "loss": 0.0,
+      "step": 11248
+    },
+    {
+      "epoch": 0.7249468325062834,
+      "grad_norm": 0.005892490520614762,
+      "learning_rate": 3.0612244897959185e-06,
+      "loss": 0.0,
+      "step": 11249
+    },
+    {
+      "epoch": 0.7250112779532126,
+      "grad_norm": 0.01922783843529163,
+      "learning_rate": 3.060508413891873e-06,
+      "loss": 0.0,
+      "step": 11250
+    },
+    {
+      "epoch": 0.7250757234001418,
+      "grad_norm": 9.092396979500205e-05,
+      "learning_rate": 3.0597923379878267e-06,
+      "loss": 0.0,
+      "step": 11251
+    },
+    {
+      "epoch": 0.725140168847071,
+      "grad_norm": 0.0004333292273602369,
+      "learning_rate": 3.059076262083781e-06,
+      "loss": 0.0,
+      "step": 11252
+    },
+    {
+      "epoch": 0.7252046142940002,
+      "grad_norm": 0.017930935702477148,
+      "learning_rate": 3.0583601861797353e-06,
+      "loss": 0.0,
+      "step": 11253
+    },
+    {
+      "epoch": 0.7252690597409293,
+      "grad_norm": 0.31085538834256204,
+      "learning_rate": 3.057644110275689e-06,
+      "loss": 0.0065,
+      "step": 11254
+    },
+    {
+      "epoch": 0.7253335051878584,
+      "grad_norm": 0.014556196484398517,
+      "learning_rate": 3.0569280343716435e-06,
+      "loss": 0.0001,
+      "step": 11255
+    },
+    {
+      "epoch": 0.7253979506347876,
+      "grad_norm": 0.00015255389938576366,
+      "learning_rate": 3.0562119584675974e-06,
+      "loss": 0.0,
+      "step": 11256
+    },
+    {
+      "epoch": 0.7254623960817168,
+      "grad_norm": 0.00020395906453181,
+      "learning_rate": 3.055495882563552e-06,
+      "loss": 0.0,
+      "step": 11257
+    },
+    {
+      "epoch": 0.725526841528646,
+      "grad_norm": 0.0003205706918907401,
+      "learning_rate": 3.0547798066595064e-06,
+      "loss": 0.0,
+      "step": 11258
+    },
+    {
+      "epoch": 0.7255912869755752,
+      "grad_norm": 4.8430433611672415e-05,
+      "learning_rate": 3.0540637307554603e-06,
+      "loss": 0.0,
+      "step": 11259
+    },
+    {
+      "epoch": 0.7256557324225044,
+      "grad_norm": 0.0034224555781167698,
+      "learning_rate": 3.0533476548514146e-06,
+      "loss": 0.0,
+      "step": 11260
+    },
+    {
+      "epoch": 0.7257201778694335,
+      "grad_norm": 0.0007747417139003617,
+      "learning_rate": 3.052631578947369e-06,
+      "loss": 0.0,
+      "step": 11261
+    },
+    {
+      "epoch": 0.7257846233163627,
+      "grad_norm": 0.005629201743521242,
+      "learning_rate": 3.0519155030433227e-06,
+      "loss": 0.0016,
+      "step": 11262
+    },
+    {
+      "epoch": 0.7258490687632919,
+      "grad_norm": 0.05939066877792539,
+      "learning_rate": 3.051199427139277e-06,
+      "loss": 0.0002,
+      "step": 11263
+    },
+    {
+      "epoch": 0.7259135142102211,
+      "grad_norm": 0.024681152626680736,
+      "learning_rate": 3.050483351235231e-06,
+      "loss": 0.0002,
+      "step": 11264
+    },
+    {
+      "epoch": 0.7259779596571502,
+      "grad_norm": 0.16541939988288423,
+      "learning_rate": 3.049767275331185e-06,
+      "loss": 0.0003,
+      "step": 11265
+    },
+    {
+      "epoch": 0.7260424051040794,
+      "grad_norm": 0.006080011470615735,
+      "learning_rate": 3.0490511994271395e-06,
+      "loss": 0.0001,
+      "step": 11266
+    },
+    {
+      "epoch": 0.7261068505510085,
+      "grad_norm": 0.0047498769675451856,
+      "learning_rate": 3.0483351235230934e-06,
+      "loss": 0.0,
+      "step": 11267
+    },
+    {
+      "epoch": 0.7261712959979377,
+      "grad_norm": 0.04948049102232195,
+      "learning_rate": 3.047619047619048e-06,
+      "loss": 0.0001,
+      "step": 11268
+    },
+    {
+      "epoch": 0.7262357414448669,
+      "grad_norm": 0.0006189984344180252,
+      "learning_rate": 3.0469029717150024e-06,
+      "loss": 0.0,
+      "step": 11269
+    },
+    {
+      "epoch": 0.7263001868917961,
+      "grad_norm": 0.005652951470014697,
+      "learning_rate": 3.0461868958109563e-06,
+      "loss": 0.0,
+      "step": 11270
+    },
+    {
+      "epoch": 0.7263646323387253,
+      "grad_norm": 0.0005117226456541811,
+      "learning_rate": 3.0454708199069106e-06,
+      "loss": 0.0,
+      "step": 11271
+    },
+    {
+      "epoch": 0.7264290777856545,
+      "grad_norm": 0.0007810068636911916,
+      "learning_rate": 3.0447547440028645e-06,
+      "loss": 0.0,
+      "step": 11272
+    },
+    {
+      "epoch": 0.7264935232325836,
+      "grad_norm": 0.06590527306262674,
+      "learning_rate": 3.0440386680988188e-06,
+      "loss": 0.0004,
+      "step": 11273
+    },
+    {
+      "epoch": 0.7265579686795128,
+      "grad_norm": 0.0005164033597064126,
+      "learning_rate": 3.043322592194773e-06,
+      "loss": 0.0,
+      "step": 11274
+    },
+    {
+      "epoch": 0.726622414126442,
+      "grad_norm": 0.0003381609834740342,
+      "learning_rate": 3.042606516290727e-06,
+      "loss": 0.0,
+      "step": 11275
+    },
+    {
+      "epoch": 0.7266868595733711,
+      "grad_norm": 0.021830852884517263,
+      "learning_rate": 3.0418904403866812e-06,
+      "loss": 0.0,
+      "step": 11276
+    },
+    {
+      "epoch": 0.7267513050203003,
+      "grad_norm": 6.346864446019381e-05,
+      "learning_rate": 3.041174364482635e-06,
+      "loss": 0.0,
+      "step": 11277
+    },
+    {
+      "epoch": 0.7268157504672295,
+      "grad_norm": 0.03467683947264384,
+      "learning_rate": 3.0404582885785894e-06,
+      "loss": 0.0,
+      "step": 11278
+    },
+    {
+      "epoch": 0.7268801959141586,
+      "grad_norm": 0.004254908353937039,
+      "learning_rate": 3.0397422126745437e-06,
+      "loss": 0.0,
+      "step": 11279
+    },
+    {
+      "epoch": 0.7269446413610878,
+      "grad_norm": 0.003063179042156029,
+      "learning_rate": 3.039026136770498e-06,
+      "loss": 0.0,
+      "step": 11280
+    },
+    {
+      "epoch": 0.727009086808017,
+      "grad_norm": 0.0001314564867009563,
+      "learning_rate": 3.0383100608664523e-06,
+      "loss": 0.0,
+      "step": 11281
+    },
+    {
+      "epoch": 0.7270735322549462,
+      "grad_norm": 0.004146407912709877,
+      "learning_rate": 3.0375939849624066e-06,
+      "loss": 0.0,
+      "step": 11282
+    },
+    {
+      "epoch": 0.7271379777018754,
+      "grad_norm": 0.00010302604658438643,
+      "learning_rate": 3.0368779090583605e-06,
+      "loss": 0.0,
+      "step": 11283
+    },
+    {
+      "epoch": 0.7272024231488046,
+      "grad_norm": 0.00013965483383936777,
+      "learning_rate": 3.0361618331543148e-06,
+      "loss": 0.0,
+      "step": 11284
+    },
+    {
+      "epoch": 0.7272668685957338,
+      "grad_norm": 0.00018378616446001092,
+      "learning_rate": 3.0354457572502686e-06,
+      "loss": 0.0,
+      "step": 11285
+    },
+    {
+      "epoch": 0.7273313140426629,
+      "grad_norm": 0.001957099914498733,
+      "learning_rate": 3.034729681346223e-06,
+      "loss": 0.0,
+      "step": 11286
+    },
+    {
+      "epoch": 0.727395759489592,
+      "grad_norm": 0.0008828377932276057,
+      "learning_rate": 3.0340136054421772e-06,
+      "loss": 0.0,
+      "step": 11287
+    },
+    {
+      "epoch": 0.7274602049365212,
+      "grad_norm": 0.0029698710575783546,
+      "learning_rate": 3.033297529538131e-06,
+      "loss": 0.0,
+      "step": 11288
+    },
+    {
+      "epoch": 0.7275246503834504,
+      "grad_norm": 0.00019433618893119242,
+      "learning_rate": 3.0325814536340854e-06,
+      "loss": 0.0,
+      "step": 11289
+    },
+    {
+      "epoch": 0.7275890958303796,
+      "grad_norm": 0.0006416192589271509,
+      "learning_rate": 3.0318653777300393e-06,
+      "loss": 0.0,
+      "step": 11290
+    },
+    {
+      "epoch": 0.7276535412773087,
+      "grad_norm": 0.00838866934093486,
+      "learning_rate": 3.0311493018259936e-06,
+      "loss": 0.0,
+      "step": 11291
+    },
+    {
+      "epoch": 0.7277179867242379,
+      "grad_norm": 0.000245455502279775,
+      "learning_rate": 3.0304332259219483e-06,
+      "loss": 0.0,
+      "step": 11292
+    },
+    {
+      "epoch": 0.7277824321711671,
+      "grad_norm": 0.004930906444727167,
+      "learning_rate": 3.029717150017902e-06,
+      "loss": 0.0,
+      "step": 11293
+    },
+    {
+      "epoch": 0.7278468776180963,
+      "grad_norm": 4.776988774088927e-05,
+      "learning_rate": 3.0290010741138565e-06,
+      "loss": 0.0,
+      "step": 11294
+    },
+    {
+      "epoch": 0.7279113230650255,
+      "grad_norm": 0.27195599982658003,
+      "learning_rate": 3.0282849982098108e-06,
+      "loss": 0.001,
+      "step": 11295
+    },
+    {
+      "epoch": 0.7279757685119547,
+      "grad_norm": 0.0030353622788795963,
+      "learning_rate": 3.0275689223057647e-06,
+      "loss": 0.0,
+      "step": 11296
+    },
+    {
+      "epoch": 0.7280402139588839,
+      "grad_norm": 0.5768505948030167,
+      "learning_rate": 3.026852846401719e-06,
+      "loss": 0.003,
+      "step": 11297
+    },
+    {
+      "epoch": 0.7281046594058129,
+      "grad_norm": 0.0025597431430050125,
+      "learning_rate": 3.026136770497673e-06,
+      "loss": 0.0,
+      "step": 11298
+    },
+    {
+      "epoch": 0.7281691048527421,
+      "grad_norm": 0.0016598811747845761,
+      "learning_rate": 3.025420694593627e-06,
+      "loss": 0.0,
+      "step": 11299
+    },
+    {
+      "epoch": 0.7282335502996713,
+      "grad_norm": 0.00014893458879150577,
+      "learning_rate": 3.024704618689581e-06,
+      "loss": 0.0,
+      "step": 11300
+    },
+    {
+      "epoch": 0.7282979957466005,
+      "grad_norm": 0.0021818727213651556,
+      "learning_rate": 3.0239885427855353e-06,
+      "loss": 0.0,
+      "step": 11301
+    },
+    {
+      "epoch": 0.7283624411935297,
+      "grad_norm": 0.008553871083575986,
+      "learning_rate": 3.0232724668814896e-06,
+      "loss": 0.0,
+      "step": 11302
+    },
+    {
+      "epoch": 0.7284268866404588,
+      "grad_norm": 0.001350344278458923,
+      "learning_rate": 3.0225563909774443e-06,
+      "loss": 0.0,
+      "step": 11303
+    },
+    {
+      "epoch": 0.728491332087388,
+      "grad_norm": 0.00034752832518681015,
+      "learning_rate": 3.021840315073398e-06,
+      "loss": 0.0,
+      "step": 11304
+    },
+    {
+      "epoch": 0.7285557775343172,
+      "grad_norm": 0.004296723301807933,
+      "learning_rate": 3.0211242391693525e-06,
+      "loss": 0.0,
+      "step": 11305
+    },
+    {
+      "epoch": 0.7286202229812464,
+      "grad_norm": 0.003349462100284539,
+      "learning_rate": 3.0204081632653064e-06,
+      "loss": 0.0,
+      "step": 11306
+    },
+    {
+      "epoch": 0.7286846684281756,
+      "grad_norm": 0.13197393612356048,
+      "learning_rate": 3.0196920873612607e-06,
+      "loss": 0.0005,
+      "step": 11307
+    },
+    {
+      "epoch": 0.7287491138751048,
+      "grad_norm": 0.000510281932853362,
+      "learning_rate": 3.0189760114572145e-06,
+      "loss": 0.0,
+      "step": 11308
+    },
+    {
+      "epoch": 0.7288135593220338,
+      "grad_norm": 0.0011085601485967163,
+      "learning_rate": 3.018259935553169e-06,
+      "loss": 0.0,
+      "step": 11309
+    },
+    {
+      "epoch": 0.728878004768963,
+      "grad_norm": 0.005277812025546389,
+      "learning_rate": 3.017543859649123e-06,
+      "loss": 0.0,
+      "step": 11310
+    },
+    {
+      "epoch": 0.7289424502158922,
+      "grad_norm": 0.0002395444898397595,
+      "learning_rate": 3.016827783745077e-06,
+      "loss": 0.0,
+      "step": 11311
+    },
+    {
+      "epoch": 0.7290068956628214,
+      "grad_norm": 0.0831582389340663,
+      "learning_rate": 3.0161117078410313e-06,
+      "loss": 0.0011,
+      "step": 11312
+    },
+    {
+      "epoch": 0.7290713411097506,
+      "grad_norm": 0.0017825745778012954,
+      "learning_rate": 3.015395631936985e-06,
+      "loss": 0.0,
+      "step": 11313
+    },
+    {
+      "epoch": 0.7291357865566798,
+      "grad_norm": 0.0011956396270402388,
+      "learning_rate": 3.0146795560329395e-06,
+      "loss": 0.0,
+      "step": 11314
+    },
+    {
+      "epoch": 0.729200232003609,
+      "grad_norm": 0.00043885243242865044,
+      "learning_rate": 3.0139634801288942e-06,
+      "loss": 0.0,
+      "step": 11315
+    },
+    {
+      "epoch": 0.7292646774505381,
+      "grad_norm": 0.0009921284919679632,
+      "learning_rate": 3.013247404224848e-06,
+      "loss": 0.0,
+      "step": 11316
+    },
+    {
+      "epoch": 0.7293291228974673,
+      "grad_norm": 0.0023250831159840257,
+      "learning_rate": 3.0125313283208024e-06,
+      "loss": 0.0,
+      "step": 11317
+    },
+    {
+      "epoch": 0.7293935683443965,
+      "grad_norm": 0.15707155844292203,
+      "learning_rate": 3.0118152524167567e-06,
+      "loss": 0.0004,
+      "step": 11318
+    },
+    {
+      "epoch": 0.7294580137913257,
+      "grad_norm": 0.013931934554137165,
+      "learning_rate": 3.0110991765127106e-06,
+      "loss": 0.0,
+      "step": 11319
+    },
+    {
+      "epoch": 0.7295224592382549,
+      "grad_norm": 0.0014724451177281826,
+      "learning_rate": 3.010383100608665e-06,
+      "loss": 0.0,
+      "step": 11320
+    },
+    {
+      "epoch": 0.729586904685184,
+      "grad_norm": 9.713692903118522e-05,
+      "learning_rate": 3.0096670247046187e-06,
+      "loss": 0.0,
+      "step": 11321
+    },
+    {
+      "epoch": 0.7296513501321131,
+      "grad_norm": 0.0010043175423519069,
+      "learning_rate": 3.008950948800573e-06,
+      "loss": 0.0,
+      "step": 11322
+    },
+    {
+      "epoch": 0.7297157955790423,
+      "grad_norm": 0.003977753280873503,
+      "learning_rate": 3.0082348728965273e-06,
+      "loss": 0.0,
+      "step": 11323
+    },
+    {
+      "epoch": 0.7297802410259715,
+      "grad_norm": 7.66866136069101e-05,
+      "learning_rate": 3.007518796992481e-06,
+      "loss": 0.0,
+      "step": 11324
+    },
+    {
+      "epoch": 0.7298446864729007,
+      "grad_norm": 0.03245453503710955,
+      "learning_rate": 3.0068027210884355e-06,
+      "loss": 0.0001,
+      "step": 11325
+    },
+    {
+      "epoch": 0.7299091319198299,
+      "grad_norm": 0.017230089378299205,
+      "learning_rate": 3.0060866451843894e-06,
+      "loss": 0.0002,
+      "step": 11326
+    },
+    {
+      "epoch": 0.729973577366759,
+      "grad_norm": 0.0011237248614723047,
+      "learning_rate": 3.005370569280344e-06,
+      "loss": 0.0,
+      "step": 11327
+    },
+    {
+      "epoch": 0.7300380228136882,
+      "grad_norm": 0.001596873892730706,
+      "learning_rate": 3.0046544933762984e-06,
+      "loss": 0.0,
+      "step": 11328
+    },
+    {
+      "epoch": 0.7301024682606174,
+      "grad_norm": 0.0025830303630132935,
+      "learning_rate": 3.0039384174722523e-06,
+      "loss": 0.0,
+      "step": 11329
+    },
+    {
+      "epoch": 0.7301669137075466,
+      "grad_norm": 0.0003530421299883205,
+      "learning_rate": 3.0032223415682066e-06,
+      "loss": 0.0,
+      "step": 11330
+    },
+    {
+      "epoch": 0.7302313591544758,
+      "grad_norm": 0.002322914063727893,
+      "learning_rate": 3.002506265664161e-06,
+      "loss": 0.0,
+      "step": 11331
+    },
+    {
+      "epoch": 0.7302958046014049,
+      "grad_norm": 0.015656113380858726,
+      "learning_rate": 3.0017901897601148e-06,
+      "loss": 0.0,
+      "step": 11332
+    },
+    {
+      "epoch": 0.730360250048334,
+      "grad_norm": 0.002232419987401452,
+      "learning_rate": 3.001074113856069e-06,
+      "loss": 0.0,
+      "step": 11333
+    },
+    {
+      "epoch": 0.7304246954952632,
+      "grad_norm": 0.0006180026645144517,
+      "learning_rate": 3.000358037952023e-06,
+      "loss": 0.0,
+      "step": 11334
+    },
+    {
+      "epoch": 0.7304891409421924,
+      "grad_norm": 0.0053043607240062145,
+      "learning_rate": 2.9996419620479772e-06,
+      "loss": 0.0,
+      "step": 11335
+    },
+    {
+      "epoch": 0.7305535863891216,
+      "grad_norm": 0.0007339025781432594,
+      "learning_rate": 2.9989258861439315e-06,
+      "loss": 0.0,
+      "step": 11336
+    },
+    {
+      "epoch": 0.7306180318360508,
+      "grad_norm": 0.014264670100293255,
+      "learning_rate": 2.9982098102398854e-06,
+      "loss": 0.0,
+      "step": 11337
+    },
+    {
+      "epoch": 0.73068247728298,
+      "grad_norm": 0.0012291208711675198,
+      "learning_rate": 2.9974937343358397e-06,
+      "loss": 0.0,
+      "step": 11338
+    },
+    {
+      "epoch": 0.7307469227299092,
+      "grad_norm": 0.06005159391534337,
+      "learning_rate": 2.9967776584317944e-06,
+      "loss": 0.0001,
+      "step": 11339
+    },
+    {
+      "epoch": 0.7308113681768383,
+      "grad_norm": 0.0003933094764526526,
+      "learning_rate": 2.9960615825277483e-06,
+      "loss": 0.0,
+      "step": 11340
+    },
+    {
+      "epoch": 0.7308758136237675,
+      "grad_norm": 0.021090790890940164,
+      "learning_rate": 2.9953455066237026e-06,
+      "loss": 0.0002,
+      "step": 11341
+    },
+    {
+      "epoch": 0.7309402590706967,
+      "grad_norm": 0.025931634133735976,
+      "learning_rate": 2.9946294307196565e-06,
+      "loss": 0.0,
+      "step": 11342
+    },
+    {
+      "epoch": 0.7310047045176258,
+      "grad_norm": 0.02852376562835928,
+      "learning_rate": 2.9939133548156108e-06,
+      "loss": 0.0,
+      "step": 11343
+    },
+    {
+      "epoch": 0.731069149964555,
+      "grad_norm": 0.015667323444314415,
+      "learning_rate": 2.993197278911565e-06,
+      "loss": 0.0,
+      "step": 11344
+    },
+    {
+      "epoch": 0.7311335954114841,
+      "grad_norm": 0.00218994004770224,
+      "learning_rate": 2.992481203007519e-06,
+      "loss": 0.0,
+      "step": 11345
+    },
+    {
+      "epoch": 0.7311980408584133,
+      "grad_norm": 0.16275101952445878,
+      "learning_rate": 2.9917651271034732e-06,
+      "loss": 0.0028,
+      "step": 11346
+    },
+    {
+      "epoch": 0.7312624863053425,
+      "grad_norm": 0.0003862983051810963,
+      "learning_rate": 2.991049051199427e-06,
+      "loss": 0.0,
+      "step": 11347
+    },
+    {
+      "epoch": 0.7313269317522717,
+      "grad_norm": 0.003829513862433199,
+      "learning_rate": 2.9903329752953814e-06,
+      "loss": 0.0,
+      "step": 11348
+    },
+    {
+      "epoch": 0.7313913771992009,
+      "grad_norm": 0.0034865630005024286,
+      "learning_rate": 2.9896168993913353e-06,
+      "loss": 0.0,
+      "step": 11349
+    },
+    {
+      "epoch": 0.7314558226461301,
+      "grad_norm": 0.0009932220979224295,
+      "learning_rate": 2.98890082348729e-06,
+      "loss": 0.0,
+      "step": 11350
+    },
+    {
+      "epoch": 0.7315202680930593,
+      "grad_norm": 0.002630709002841604,
+      "learning_rate": 2.9881847475832443e-06,
+      "loss": 0.0,
+      "step": 11351
+    },
+    {
+      "epoch": 0.7315847135399884,
+      "grad_norm": 0.008694154945653666,
+      "learning_rate": 2.9874686716791986e-06,
+      "loss": 0.0,
+      "step": 11352
+    },
+    {
+      "epoch": 0.7316491589869176,
+      "grad_norm": 0.001717471994335099,
+      "learning_rate": 2.9867525957751525e-06,
+      "loss": 0.0,
+      "step": 11353
+    },
+    {
+      "epoch": 0.7317136044338467,
+      "grad_norm": 0.08697505214756895,
+      "learning_rate": 2.9860365198711068e-06,
+      "loss": 0.0017,
+      "step": 11354
+    },
+    {
+      "epoch": 0.7317780498807759,
+      "grad_norm": 0.0014773768805635785,
+      "learning_rate": 2.9853204439670607e-06,
+      "loss": 0.0,
+      "step": 11355
+    },
+    {
+      "epoch": 0.7318424953277051,
+      "grad_norm": 0.01304011785695086,
+      "learning_rate": 2.984604368063015e-06,
+      "loss": 0.0,
+      "step": 11356
+    },
+    {
+      "epoch": 0.7319069407746343,
+      "grad_norm": 0.006473541821742216,
+      "learning_rate": 2.983888292158969e-06,
+      "loss": 0.0,
+      "step": 11357
+    },
+    {
+      "epoch": 0.7319713862215634,
+      "grad_norm": 0.0013719339669280896,
+      "learning_rate": 2.983172216254923e-06,
+      "loss": 0.0,
+      "step": 11358
+    },
+    {
+      "epoch": 0.7320358316684926,
+      "grad_norm": 0.0002129689555400633,
+      "learning_rate": 2.9824561403508774e-06,
+      "loss": 0.0,
+      "step": 11359
+    },
+    {
+      "epoch": 0.7321002771154218,
+      "grad_norm": 0.0001586731060494783,
+      "learning_rate": 2.9817400644468313e-06,
+      "loss": 0.0,
+      "step": 11360
+    },
+    {
+      "epoch": 0.732164722562351,
+      "grad_norm": 0.013050724962506173,
+      "learning_rate": 2.9810239885427856e-06,
+      "loss": 0.0,
+      "step": 11361
+    },
+    {
+      "epoch": 0.7322291680092802,
+      "grad_norm": 0.019111154363726266,
+      "learning_rate": 2.9803079126387403e-06,
+      "loss": 0.0001,
+      "step": 11362
+    },
+    {
+      "epoch": 0.7322936134562094,
+      "grad_norm": 0.023404573866207774,
+      "learning_rate": 2.979591836734694e-06,
+      "loss": 0.0001,
+      "step": 11363
+    },
+    {
+      "epoch": 0.7323580589031385,
+      "grad_norm": 0.0005499391081068312,
+      "learning_rate": 2.9788757608306485e-06,
+      "loss": 0.0,
+      "step": 11364
+    },
+    {
+      "epoch": 0.7324225043500676,
+      "grad_norm": 0.0021230133353467673,
+      "learning_rate": 2.9781596849266024e-06,
+      "loss": 0.0,
+      "step": 11365
+    },
+    {
+      "epoch": 0.7324869497969968,
+      "grad_norm": 0.014250397877147453,
+      "learning_rate": 2.9774436090225567e-06,
+      "loss": 0.0001,
+      "step": 11366
+    },
+    {
+      "epoch": 0.732551395243926,
+      "grad_norm": 0.00023056297783971086,
+      "learning_rate": 2.976727533118511e-06,
+      "loss": 0.0,
+      "step": 11367
+    },
+    {
+      "epoch": 0.7326158406908552,
+      "grad_norm": 0.073680816757015,
+      "learning_rate": 2.976011457214465e-06,
+      "loss": 0.0002,
+      "step": 11368
+    },
+    {
+      "epoch": 0.7326802861377844,
+      "grad_norm": 0.011644553199364758,
+      "learning_rate": 2.975295381310419e-06,
+      "loss": 0.0,
+      "step": 11369
+    },
+    {
+      "epoch": 0.7327447315847135,
+      "grad_norm": 1.2168431158448905,
+      "learning_rate": 2.974579305406373e-06,
+      "loss": 0.0042,
+      "step": 11370
+    },
+    {
+      "epoch": 0.7328091770316427,
+      "grad_norm": 0.0011119148824099833,
+      "learning_rate": 2.9738632295023273e-06,
+      "loss": 0.0,
+      "step": 11371
+    },
+    {
+      "epoch": 0.7328736224785719,
+      "grad_norm": 0.14873476128698843,
+      "learning_rate": 2.9731471535982816e-06,
+      "loss": 0.0002,
+      "step": 11372
+    },
+    {
+      "epoch": 0.7329380679255011,
+      "grad_norm": 0.0005180624511212282,
+      "learning_rate": 2.9724310776942355e-06,
+      "loss": 0.0,
+      "step": 11373
+    },
+    {
+      "epoch": 0.7330025133724303,
+      "grad_norm": 0.08457680714713561,
+      "learning_rate": 2.9717150017901902e-06,
+      "loss": 0.0002,
+      "step": 11374
+    },
+    {
+      "epoch": 0.7330669588193595,
+      "grad_norm": 3.189354502948727e-05,
+      "learning_rate": 2.9709989258861445e-06,
+      "loss": 0.0,
+      "step": 11375
+    },
+    {
+      "epoch": 0.7331314042662885,
+      "grad_norm": 0.00015374242569765145,
+      "learning_rate": 2.9702828499820984e-06,
+      "loss": 0.0,
+      "step": 11376
+    },
+    {
+      "epoch": 0.7331958497132177,
+      "grad_norm": 0.004303427852003826,
+      "learning_rate": 2.9695667740780527e-06,
+      "loss": 0.0,
+      "step": 11377
+    },
+    {
+      "epoch": 0.7332602951601469,
+      "grad_norm": 0.01627340553711377,
+      "learning_rate": 2.9688506981740066e-06,
+      "loss": 0.0,
+      "step": 11378
+    },
+    {
+      "epoch": 0.7333247406070761,
+      "grad_norm": 0.1106417562165362,
+      "learning_rate": 2.968134622269961e-06,
+      "loss": 0.0001,
+      "step": 11379
+    },
+    {
+      "epoch": 0.7333891860540053,
+      "grad_norm": 0.11460116547485791,
+      "learning_rate": 2.967418546365915e-06,
+      "loss": 0.0018,
+      "step": 11380
+    },
+    {
+      "epoch": 0.7334536315009345,
+      "grad_norm": 0.0073095201484695465,
+      "learning_rate": 2.966702470461869e-06,
+      "loss": 0.0,
+      "step": 11381
+    },
+    {
+      "epoch": 0.7335180769478636,
+      "grad_norm": 0.014872503234698634,
+      "learning_rate": 2.9659863945578233e-06,
+      "loss": 0.0001,
+      "step": 11382
+    },
+    {
+      "epoch": 0.7335825223947928,
+      "grad_norm": 0.042749717691514845,
+      "learning_rate": 2.965270318653777e-06,
+      "loss": 0.0001,
+      "step": 11383
+    },
+    {
+      "epoch": 0.733646967841722,
+      "grad_norm": 0.0008445524787041037,
+      "learning_rate": 2.9645542427497315e-06,
+      "loss": 0.0,
+      "step": 11384
+    },
+    {
+      "epoch": 0.7337114132886512,
+      "grad_norm": 0.005167055323404655,
+      "learning_rate": 2.9638381668456862e-06,
+      "loss": 0.0,
+      "step": 11385
+    },
+    {
+      "epoch": 0.7337758587355804,
+      "grad_norm": 0.026027118945750397,
+      "learning_rate": 2.96312209094164e-06,
+      "loss": 0.0,
+      "step": 11386
+    },
+    {
+      "epoch": 0.7338403041825095,
+      "grad_norm": 0.0004145667714222251,
+      "learning_rate": 2.9624060150375944e-06,
+      "loss": 0.0,
+      "step": 11387
+    },
+    {
+      "epoch": 0.7339047496294386,
+      "grad_norm": 5.982031954885564e-05,
+      "learning_rate": 2.9616899391335487e-06,
+      "loss": 0.0,
+      "step": 11388
+    },
+    {
+      "epoch": 0.7339691950763678,
+      "grad_norm": 0.04745125465754833,
+      "learning_rate": 2.9609738632295026e-06,
+      "loss": 0.0003,
+      "step": 11389
+    },
+    {
+      "epoch": 0.734033640523297,
+      "grad_norm": 0.0004990294223088284,
+      "learning_rate": 2.960257787325457e-06,
+      "loss": 0.0,
+      "step": 11390
+    },
+    {
+      "epoch": 0.7340980859702262,
+      "grad_norm": 0.0006998216447846934,
+      "learning_rate": 2.9595417114214108e-06,
+      "loss": 0.0,
+      "step": 11391
+    },
+    {
+      "epoch": 0.7341625314171554,
+      "grad_norm": 0.01677587625693549,
+      "learning_rate": 2.958825635517365e-06,
+      "loss": 0.0,
+      "step": 11392
+    },
+    {
+      "epoch": 0.7342269768640846,
+      "grad_norm": 0.0003467193528171593,
+      "learning_rate": 2.958109559613319e-06,
+      "loss": 0.0,
+      "step": 11393
+    },
+    {
+      "epoch": 0.7342914223110137,
+      "grad_norm": 0.0005840219191719138,
+      "learning_rate": 2.9573934837092732e-06,
+      "loss": 0.0,
+      "step": 11394
+    },
+    {
+      "epoch": 0.7343558677579429,
+      "grad_norm": 0.0032256336054706956,
+      "learning_rate": 2.9566774078052275e-06,
+      "loss": 0.0,
+      "step": 11395
+    },
+    {
+      "epoch": 0.7344203132048721,
+      "grad_norm": 0.00010625610456919803,
+      "learning_rate": 2.9559613319011814e-06,
+      "loss": 0.0,
+      "step": 11396
+    },
+    {
+      "epoch": 0.7344847586518013,
+      "grad_norm": 0.00030531125861322296,
+      "learning_rate": 2.955245255997136e-06,
+      "loss": 0.0,
+      "step": 11397
+    },
+    {
+      "epoch": 0.7345492040987305,
+      "grad_norm": 0.2009604092585717,
+      "learning_rate": 2.9545291800930904e-06,
+      "loss": 0.0005,
+      "step": 11398
+    },
+    {
+      "epoch": 0.7346136495456596,
+      "grad_norm": 0.33045670282215517,
+      "learning_rate": 2.9538131041890443e-06,
+      "loss": 0.0024,
+      "step": 11399
+    },
+    {
+      "epoch": 0.7346780949925887,
+      "grad_norm": 0.0005136405285533002,
+      "learning_rate": 2.9530970282849986e-06,
+      "loss": 0.0,
+      "step": 11400
+    },
+    {
+      "epoch": 0.7347425404395179,
+      "grad_norm": 0.009765878316454831,
+      "learning_rate": 2.9523809523809525e-06,
+      "loss": 0.0,
+      "step": 11401
+    },
+    {
+      "epoch": 0.7348069858864471,
+      "grad_norm": 0.33457723500160075,
+      "learning_rate": 2.9516648764769068e-06,
+      "loss": 0.0013,
+      "step": 11402
+    },
+    {
+      "epoch": 0.7348714313333763,
+      "grad_norm": 0.08505562734413115,
+      "learning_rate": 2.950948800572861e-06,
+      "loss": 0.0001,
+      "step": 11403
+    },
+    {
+      "epoch": 0.7349358767803055,
+      "grad_norm": 0.0005396746602388885,
+      "learning_rate": 2.950232724668815e-06,
+      "loss": 0.0,
+      "step": 11404
+    },
+    {
+      "epoch": 0.7350003222272347,
+      "grad_norm": 0.0007219231506156975,
+      "learning_rate": 2.9495166487647692e-06,
+      "loss": 0.0,
+      "step": 11405
+    },
+    {
+      "epoch": 0.7350647676741638,
+      "grad_norm": 0.00080918341992736,
+      "learning_rate": 2.948800572860723e-06,
+      "loss": 0.0,
+      "step": 11406
+    },
+    {
+      "epoch": 0.735129213121093,
+      "grad_norm": 0.008291451385984977,
+      "learning_rate": 2.9480844969566774e-06,
+      "loss": 0.0001,
+      "step": 11407
+    },
+    {
+      "epoch": 0.7351936585680222,
+      "grad_norm": 3.7420126980391375e-05,
+      "learning_rate": 2.9473684210526317e-06,
+      "loss": 0.0,
+      "step": 11408
+    },
+    {
+      "epoch": 0.7352581040149514,
+      "grad_norm": 0.00547087372155844,
+      "learning_rate": 2.9466523451485864e-06,
+      "loss": 0.0,
+      "step": 11409
+    },
+    {
+      "epoch": 0.7353225494618805,
+      "grad_norm": 0.0017479068818298368,
+      "learning_rate": 2.9459362692445403e-06,
+      "loss": 0.0,
+      "step": 11410
+    },
+    {
+      "epoch": 0.7353869949088097,
+      "grad_norm": 0.00347103415906202,
+      "learning_rate": 2.9452201933404946e-06,
+      "loss": 0.0,
+      "step": 11411
+    },
+    {
+      "epoch": 0.7354514403557388,
+      "grad_norm": 0.004133792163996183,
+      "learning_rate": 2.9445041174364485e-06,
+      "loss": 0.0,
+      "step": 11412
+    },
+    {
+      "epoch": 0.735515885802668,
+      "grad_norm": 0.00042971752764806866,
+      "learning_rate": 2.9437880415324028e-06,
+      "loss": 0.0,
+      "step": 11413
+    },
+    {
+      "epoch": 0.7355803312495972,
+      "grad_norm": 0.00010637823235444148,
+      "learning_rate": 2.9430719656283567e-06,
+      "loss": 0.0,
+      "step": 11414
+    },
+    {
+      "epoch": 0.7356447766965264,
+      "grad_norm": 0.00012087427014443836,
+      "learning_rate": 2.942355889724311e-06,
+      "loss": 0.0,
+      "step": 11415
+    },
+    {
+      "epoch": 0.7357092221434556,
+      "grad_norm": 0.00217104518436385,
+      "learning_rate": 2.9416398138202653e-06,
+      "loss": 0.0,
+      "step": 11416
+    },
+    {
+      "epoch": 0.7357736675903848,
+      "grad_norm": 6.510885686800687e-05,
+      "learning_rate": 2.940923737916219e-06,
+      "loss": 0.0,
+      "step": 11417
+    },
+    {
+      "epoch": 0.735838113037314,
+      "grad_norm": 0.013142905983270175,
+      "learning_rate": 2.9402076620121734e-06,
+      "loss": 0.0001,
+      "step": 11418
+    },
+    {
+      "epoch": 0.7359025584842431,
+      "grad_norm": 0.002358491054728065,
+      "learning_rate": 2.9394915861081273e-06,
+      "loss": 0.0,
+      "step": 11419
+    },
+    {
+      "epoch": 0.7359670039311723,
+      "grad_norm": 0.0026521542996968076,
+      "learning_rate": 2.938775510204082e-06,
+      "loss": 0.0,
+      "step": 11420
+    },
+    {
+      "epoch": 0.7360314493781014,
+      "grad_norm": 0.003946334973341138,
+      "learning_rate": 2.9380594343000363e-06,
+      "loss": 0.0,
+      "step": 11421
+    },
+    {
+      "epoch": 0.7360958948250306,
+      "grad_norm": 0.0088235968052502,
+      "learning_rate": 2.93734335839599e-06,
+      "loss": 0.0,
+      "step": 11422
+    },
+    {
+      "epoch": 0.7361603402719598,
+      "grad_norm": 0.004964299051000416,
+      "learning_rate": 2.9366272824919445e-06,
+      "loss": 0.0,
+      "step": 11423
+    },
+    {
+      "epoch": 0.7362247857188889,
+      "grad_norm": 0.008458645840074462,
+      "learning_rate": 2.935911206587899e-06,
+      "loss": 0.0,
+      "step": 11424
+    },
+    {
+      "epoch": 0.7362892311658181,
+      "grad_norm": 0.003313165458895752,
+      "learning_rate": 2.9351951306838527e-06,
+      "loss": 0.0,
+      "step": 11425
+    },
+    {
+      "epoch": 0.7363536766127473,
+      "grad_norm": 0.0005009463709045276,
+      "learning_rate": 2.934479054779807e-06,
+      "loss": 0.0,
+      "step": 11426
+    },
+    {
+      "epoch": 0.7364181220596765,
+      "grad_norm": 0.0006078475946347758,
+      "learning_rate": 2.933762978875761e-06,
+      "loss": 0.0,
+      "step": 11427
+    },
+    {
+      "epoch": 0.7364825675066057,
+      "grad_norm": 0.007668545880177946,
+      "learning_rate": 2.933046902971715e-06,
+      "loss": 0.0,
+      "step": 11428
+    },
+    {
+      "epoch": 0.7365470129535349,
+      "grad_norm": 0.16581019987401066,
+      "learning_rate": 2.9323308270676694e-06,
+      "loss": 0.0004,
+      "step": 11429
+    },
+    {
+      "epoch": 0.736611458400464,
+      "grad_norm": 9.086367411350327e-05,
+      "learning_rate": 2.9316147511636233e-06,
+      "loss": 0.0,
+      "step": 11430
+    },
+    {
+      "epoch": 0.7366759038473932,
+      "grad_norm": 0.002234896439144746,
+      "learning_rate": 2.9308986752595776e-06,
+      "loss": 0.0,
+      "step": 11431
+    },
+    {
+      "epoch": 0.7367403492943223,
+      "grad_norm": 0.005501577603166098,
+      "learning_rate": 2.9301825993555323e-06,
+      "loss": 0.0,
+      "step": 11432
+    },
+    {
+      "epoch": 0.7368047947412515,
+      "grad_norm": 0.014186799638279296,
+      "learning_rate": 2.9294665234514862e-06,
+      "loss": 0.0,
+      "step": 11433
+    },
+    {
+      "epoch": 0.7368692401881807,
+      "grad_norm": 0.007490603212982399,
+      "learning_rate": 2.9287504475474405e-06,
+      "loss": 0.0,
+      "step": 11434
+    },
+    {
+      "epoch": 0.7369336856351099,
+      "grad_norm": 0.003929949417439596,
+      "learning_rate": 2.9280343716433944e-06,
+      "loss": 0.0,
+      "step": 11435
+    },
+    {
+      "epoch": 0.736998131082039,
+      "grad_norm": 0.5977107253835082,
+      "learning_rate": 2.9273182957393487e-06,
+      "loss": 0.0005,
+      "step": 11436
+    },
+    {
+      "epoch": 0.7370625765289682,
+      "grad_norm": 0.018352840375388126,
+      "learning_rate": 2.926602219835303e-06,
+      "loss": 0.0,
+      "step": 11437
+    },
+    {
+      "epoch": 0.7371270219758974,
+      "grad_norm": 0.00020872345792618345,
+      "learning_rate": 2.925886143931257e-06,
+      "loss": 0.0,
+      "step": 11438
+    },
+    {
+      "epoch": 0.7371914674228266,
+      "grad_norm": 0.00234018454065431,
+      "learning_rate": 2.925170068027211e-06,
+      "loss": 0.0,
+      "step": 11439
+    },
+    {
+      "epoch": 0.7372559128697558,
+      "grad_norm": 0.0016579721245929928,
+      "learning_rate": 2.924453992123165e-06,
+      "loss": 0.0,
+      "step": 11440
+    },
+    {
+      "epoch": 0.737320358316685,
+      "grad_norm": 0.08998472100608822,
+      "learning_rate": 2.9237379162191193e-06,
+      "loss": 0.0001,
+      "step": 11441
+    },
+    {
+      "epoch": 0.7373848037636141,
+      "grad_norm": 0.0016083044416349665,
+      "learning_rate": 2.923021840315073e-06,
+      "loss": 0.0,
+      "step": 11442
+    },
+    {
+      "epoch": 0.7374492492105432,
+      "grad_norm": 0.4860207992251301,
+      "learning_rate": 2.9223057644110275e-06,
+      "loss": 0.0011,
+      "step": 11443
+    },
+    {
+      "epoch": 0.7375136946574724,
+      "grad_norm": 0.0023780665069745095,
+      "learning_rate": 2.9215896885069822e-06,
+      "loss": 0.0,
+      "step": 11444
+    },
+    {
+      "epoch": 0.7375781401044016,
+      "grad_norm": 0.18297514282903246,
+      "learning_rate": 2.9208736126029365e-06,
+      "loss": 0.0008,
+      "step": 11445
+    },
+    {
+      "epoch": 0.7376425855513308,
+      "grad_norm": 7.226204698247018e-05,
+      "learning_rate": 2.9201575366988904e-06,
+      "loss": 0.0,
+      "step": 11446
+    },
+    {
+      "epoch": 0.73770703099826,
+      "grad_norm": 0.005099393862582168,
+      "learning_rate": 2.9194414607948447e-06,
+      "loss": 0.0,
+      "step": 11447
+    },
+    {
+      "epoch": 0.7377714764451891,
+      "grad_norm": 0.04962834057641299,
+      "learning_rate": 2.9187253848907986e-06,
+      "loss": 0.0001,
+      "step": 11448
+    },
+    {
+      "epoch": 0.7378359218921183,
+      "grad_norm": 0.0006057873141577277,
+      "learning_rate": 2.918009308986753e-06,
+      "loss": 0.0,
+      "step": 11449
+    },
+    {
+      "epoch": 0.7379003673390475,
+      "grad_norm": 0.0028990582394420833,
+      "learning_rate": 2.9172932330827068e-06,
+      "loss": 0.0,
+      "step": 11450
+    },
+    {
+      "epoch": 0.7379648127859767,
+      "grad_norm": 0.050842658381563356,
+      "learning_rate": 2.916577157178661e-06,
+      "loss": 0.0007,
+      "step": 11451
+    },
+    {
+      "epoch": 0.7380292582329059,
+      "grad_norm": 0.0024151748871815318,
+      "learning_rate": 2.9158610812746154e-06,
+      "loss": 0.0,
+      "step": 11452
+    },
+    {
+      "epoch": 0.7380937036798351,
+      "grad_norm": 0.0003905186066450872,
+      "learning_rate": 2.9151450053705692e-06,
+      "loss": 0.0,
+      "step": 11453
+    },
+    {
+      "epoch": 0.7381581491267641,
+      "grad_norm": 0.09842042434272515,
+      "learning_rate": 2.9144289294665235e-06,
+      "loss": 0.0003,
+      "step": 11454
+    },
+    {
+      "epoch": 0.7382225945736933,
+      "grad_norm": 0.03346659978636565,
+      "learning_rate": 2.9137128535624782e-06,
+      "loss": 0.0001,
+      "step": 11455
+    },
+    {
+      "epoch": 0.7382870400206225,
+      "grad_norm": 0.006367096281842634,
+      "learning_rate": 2.912996777658432e-06,
+      "loss": 0.0,
+      "step": 11456
+    },
+    {
+      "epoch": 0.7383514854675517,
+      "grad_norm": 0.00023461946393520187,
+      "learning_rate": 2.9122807017543864e-06,
+      "loss": 0.0,
+      "step": 11457
+    },
+    {
+      "epoch": 0.7384159309144809,
+      "grad_norm": 0.00038369322384038427,
+      "learning_rate": 2.9115646258503403e-06,
+      "loss": 0.0,
+      "step": 11458
+    },
+    {
+      "epoch": 0.7384803763614101,
+      "grad_norm": 0.005769438618029025,
+      "learning_rate": 2.9108485499462946e-06,
+      "loss": 0.0,
+      "step": 11459
+    },
+    {
+      "epoch": 0.7385448218083392,
+      "grad_norm": 0.0017036441431630288,
+      "learning_rate": 2.910132474042249e-06,
+      "loss": 0.0,
+      "step": 11460
+    },
+    {
+      "epoch": 0.7386092672552684,
+      "grad_norm": 0.0001072216240296,
+      "learning_rate": 2.9094163981382028e-06,
+      "loss": 0.0,
+      "step": 11461
+    },
+    {
+      "epoch": 0.7386737127021976,
+      "grad_norm": 0.01801058241218933,
+      "learning_rate": 2.908700322234157e-06,
+      "loss": 0.0001,
+      "step": 11462
+    },
+    {
+      "epoch": 0.7387381581491268,
+      "grad_norm": 0.22848583907627634,
+      "learning_rate": 2.907984246330111e-06,
+      "loss": 0.0015,
+      "step": 11463
+    },
+    {
+      "epoch": 0.738802603596056,
+      "grad_norm": 0.01842134254047434,
+      "learning_rate": 2.9072681704260652e-06,
+      "loss": 0.0002,
+      "step": 11464
+    },
+    {
+      "epoch": 0.7388670490429851,
+      "grad_norm": 0.1558477289825116,
+      "learning_rate": 2.9065520945220195e-06,
+      "loss": 0.0003,
+      "step": 11465
+    },
+    {
+      "epoch": 0.7389314944899142,
+      "grad_norm": 0.0013242711822388608,
+      "learning_rate": 2.9058360186179734e-06,
+      "loss": 0.0,
+      "step": 11466
+    },
+    {
+      "epoch": 0.7389959399368434,
+      "grad_norm": 0.00503076842425939,
+      "learning_rate": 2.905119942713928e-06,
+      "loss": 0.0,
+      "step": 11467
+    },
+    {
+      "epoch": 0.7390603853837726,
+      "grad_norm": 0.05842548289512627,
+      "learning_rate": 2.9044038668098824e-06,
+      "loss": 0.0006,
+      "step": 11468
+    },
+    {
+      "epoch": 0.7391248308307018,
+      "grad_norm": 0.003572431656853577,
+      "learning_rate": 2.9036877909058363e-06,
+      "loss": 0.0,
+      "step": 11469
+    },
+    {
+      "epoch": 0.739189276277631,
+      "grad_norm": 0.0209967272130641,
+      "learning_rate": 2.9029717150017906e-06,
+      "loss": 0.0001,
+      "step": 11470
+    },
+    {
+      "epoch": 0.7392537217245602,
+      "grad_norm": 0.0006626698904435314,
+      "learning_rate": 2.9022556390977445e-06,
+      "loss": 0.0,
+      "step": 11471
+    },
+    {
+      "epoch": 0.7393181671714893,
+      "grad_norm": 0.0793006035205354,
+      "learning_rate": 2.9015395631936988e-06,
+      "loss": 0.0007,
+      "step": 11472
+    },
+    {
+      "epoch": 0.7393826126184185,
+      "grad_norm": 0.0037272037398564804,
+      "learning_rate": 2.900823487289653e-06,
+      "loss": 0.0,
+      "step": 11473
+    },
+    {
+      "epoch": 0.7394470580653477,
+      "grad_norm": 0.2464952448338863,
+      "learning_rate": 2.900107411385607e-06,
+      "loss": 0.0058,
+      "step": 11474
+    },
+    {
+      "epoch": 0.7395115035122769,
+      "grad_norm": 4.4250068883029826e-05,
+      "learning_rate": 2.8993913354815613e-06,
+      "loss": 0.0,
+      "step": 11475
+    },
+    {
+      "epoch": 0.7395759489592061,
+      "grad_norm": 0.00012204855908644014,
+      "learning_rate": 2.898675259577515e-06,
+      "loss": 0.0,
+      "step": 11476
+    },
+    {
+      "epoch": 0.7396403944061352,
+      "grad_norm": 0.002904846725812882,
+      "learning_rate": 2.8979591836734694e-06,
+      "loss": 0.0,
+      "step": 11477
+    },
+    {
+      "epoch": 0.7397048398530643,
+      "grad_norm": 0.002904846725812882,
+      "learning_rate": 2.8979591836734694e-06,
+      "loss": 0.0034,
+      "step": 11478
+    },
+    {
+      "epoch": 0.7397692852999935,
+      "grad_norm": 0.004992793364236146,
+      "learning_rate": 2.8972431077694237e-06,
+      "loss": 0.0,
+      "step": 11479
+    },
+    {
+      "epoch": 0.7398337307469227,
+      "grad_norm": 0.04042075975410935,
+      "learning_rate": 2.896527031865378e-06,
+      "loss": 0.0001,
+      "step": 11480
+    },
+    {
+      "epoch": 0.7398981761938519,
+      "grad_norm": 0.006872133714897192,
+      "learning_rate": 2.8958109559613323e-06,
+      "loss": 0.0,
+      "step": 11481
+    },
+    {
+      "epoch": 0.7399626216407811,
+      "grad_norm": 0.10795343834250004,
+      "learning_rate": 2.8950948800572866e-06,
+      "loss": 0.0002,
+      "step": 11482
+    },
+    {
+      "epoch": 0.7400270670877103,
+      "grad_norm": 0.0005270637838438341,
+      "learning_rate": 2.8943788041532405e-06,
+      "loss": 0.0,
+      "step": 11483
+    },
+    {
+      "epoch": 0.7400915125346395,
+      "grad_norm": 0.015242437661451919,
+      "learning_rate": 2.893662728249195e-06,
+      "loss": 0.0,
+      "step": 11484
+    },
+    {
+      "epoch": 0.7401559579815686,
+      "grad_norm": 0.11215733937589789,
+      "learning_rate": 2.8929466523451487e-06,
+      "loss": 0.0001,
+      "step": 11485
+    },
+    {
+      "epoch": 0.7402204034284978,
+      "grad_norm": 5.972577802370192e-05,
+      "learning_rate": 2.892230576441103e-06,
+      "loss": 0.0,
+      "step": 11486
+    },
+    {
+      "epoch": 0.740284848875427,
+      "grad_norm": 0.0011349598866917305,
+      "learning_rate": 2.8915145005370573e-06,
+      "loss": 0.0,
+      "step": 11487
+    },
+    {
+      "epoch": 0.7403492943223561,
+      "grad_norm": 0.0022451144249306092,
+      "learning_rate": 2.890798424633011e-06,
+      "loss": 0.0,
+      "step": 11488
+    },
+    {
+      "epoch": 0.7404137397692853,
+      "grad_norm": 0.0007307866650342127,
+      "learning_rate": 2.8900823487289654e-06,
+      "loss": 0.0,
+      "step": 11489
+    },
+    {
+      "epoch": 0.7404781852162144,
+      "grad_norm": 0.0025448175047036894,
+      "learning_rate": 2.8893662728249193e-06,
+      "loss": 0.0,
+      "step": 11490
+    },
+    {
+      "epoch": 0.7405426306631436,
+      "grad_norm": 7.213562714107801e-05,
+      "learning_rate": 2.888650196920874e-06,
+      "loss": 0.0,
+      "step": 11491
+    },
+    {
+      "epoch": 0.7406070761100728,
+      "grad_norm": 0.00037888931948071514,
+      "learning_rate": 2.8879341210168283e-06,
+      "loss": 0.0,
+      "step": 11492
+    },
+    {
+      "epoch": 0.740671521557002,
+      "grad_norm": 0.0003706806676715015,
+      "learning_rate": 2.8872180451127822e-06,
+      "loss": 0.0,
+      "step": 11493
+    },
+    {
+      "epoch": 0.7407359670039312,
+      "grad_norm": 0.011404830342325069,
+      "learning_rate": 2.8865019692087365e-06,
+      "loss": 0.0,
+      "step": 11494
+    },
+    {
+      "epoch": 0.7408004124508604,
+      "grad_norm": 0.0014902201044040247,
+      "learning_rate": 2.885785893304691e-06,
+      "loss": 0.0,
+      "step": 11495
+    },
+    {
+      "epoch": 0.7408648578977896,
+      "grad_norm": 0.00031089468463307054,
+      "learning_rate": 2.8850698174006447e-06,
+      "loss": 0.0,
+      "step": 11496
+    },
+    {
+      "epoch": 0.7409293033447187,
+      "grad_norm": 0.0004114111374888369,
+      "learning_rate": 2.884353741496599e-06,
+      "loss": 0.0,
+      "step": 11497
+    },
+    {
+      "epoch": 0.7409937487916479,
+      "grad_norm": 0.002545742978935913,
+      "learning_rate": 2.883637665592553e-06,
+      "loss": 0.0,
+      "step": 11498
+    },
+    {
+      "epoch": 0.741058194238577,
+      "grad_norm": 8.01331414854357e-05,
+      "learning_rate": 2.882921589688507e-06,
+      "loss": 0.0,
+      "step": 11499
+    },
+    {
+      "epoch": 0.7411226396855062,
+      "grad_norm": 0.826731859534718,
+      "learning_rate": 2.882205513784461e-06,
+      "loss": 0.0006,
+      "step": 11500
+    },
+    {
+      "epoch": 0.7411870851324354,
+      "grad_norm": 0.1948683552474743,
+      "learning_rate": 2.8814894378804153e-06,
+      "loss": 0.0004,
+      "step": 11501
+    },
+    {
+      "epoch": 0.7412515305793645,
+      "grad_norm": 0.008470631267963194,
+      "learning_rate": 2.8807733619763696e-06,
+      "loss": 0.0,
+      "step": 11502
+    },
+    {
+      "epoch": 0.7413159760262937,
+      "grad_norm": 0.008737941093447015,
+      "learning_rate": 2.8800572860723244e-06,
+      "loss": 0.0001,
+      "step": 11503
+    },
+    {
+      "epoch": 0.7413804214732229,
+      "grad_norm": 0.01340398639142613,
+      "learning_rate": 2.8793412101682782e-06,
+      "loss": 0.0002,
+      "step": 11504
+    },
+    {
+      "epoch": 0.7414448669201521,
+      "grad_norm": 0.000581258970536343,
+      "learning_rate": 2.8786251342642325e-06,
+      "loss": 0.0,
+      "step": 11505
+    },
+    {
+      "epoch": 0.7415093123670813,
+      "grad_norm": 0.004788253913425937,
+      "learning_rate": 2.8779090583601864e-06,
+      "loss": 0.0,
+      "step": 11506
+    },
+    {
+      "epoch": 0.7415737578140105,
+      "grad_norm": 0.00012619639194410797,
+      "learning_rate": 2.8771929824561407e-06,
+      "loss": 0.0,
+      "step": 11507
+    },
+    {
+      "epoch": 0.7416382032609397,
+      "grad_norm": 0.011955899024996343,
+      "learning_rate": 2.8764769065520946e-06,
+      "loss": 0.0,
+      "step": 11508
+    },
+    {
+      "epoch": 0.7417026487078688,
+      "grad_norm": 0.0011804670870373755,
+      "learning_rate": 2.875760830648049e-06,
+      "loss": 0.0,
+      "step": 11509
+    },
+    {
+      "epoch": 0.7417670941547979,
+      "grad_norm": 0.002210108798975343,
+      "learning_rate": 2.875044754744003e-06,
+      "loss": 0.0,
+      "step": 11510
+    },
+    {
+      "epoch": 0.7418315396017271,
+      "grad_norm": 0.0004053548305945269,
+      "learning_rate": 2.874328678839957e-06,
+      "loss": 0.0,
+      "step": 11511
+    },
+    {
+      "epoch": 0.7418959850486563,
+      "grad_norm": 0.00013120753274026113,
+      "learning_rate": 2.8736126029359114e-06,
+      "loss": 0.0,
+      "step": 11512
+    },
+    {
+      "epoch": 0.7419604304955855,
+      "grad_norm": 0.12246954292956626,
+      "learning_rate": 2.8728965270318652e-06,
+      "loss": 0.0002,
+      "step": 11513
+    },
+    {
+      "epoch": 0.7420248759425147,
+      "grad_norm": 0.0031773706184475043,
+      "learning_rate": 2.8721804511278195e-06,
+      "loss": 0.0,
+      "step": 11514
+    },
+    {
+      "epoch": 0.7420893213894438,
+      "grad_norm": 0.0003136670518566507,
+      "learning_rate": 2.8714643752237742e-06,
+      "loss": 0.0,
+      "step": 11515
+    },
+    {
+      "epoch": 0.742153766836373,
+      "grad_norm": 0.05410939153081128,
+      "learning_rate": 2.870748299319728e-06,
+      "loss": 0.0001,
+      "step": 11516
+    },
+    {
+      "epoch": 0.7422182122833022,
+      "grad_norm": 7.286480718874506e-05,
+      "learning_rate": 2.8700322234156824e-06,
+      "loss": 0.0,
+      "step": 11517
+    },
+    {
+      "epoch": 0.7422826577302314,
+      "grad_norm": 0.005174802579942772,
+      "learning_rate": 2.8693161475116367e-06,
+      "loss": 0.0,
+      "step": 11518
+    },
+    {
+      "epoch": 0.7423471031771606,
+      "grad_norm": 0.001427121198095007,
+      "learning_rate": 2.8686000716075906e-06,
+      "loss": 0.0,
+      "step": 11519
+    },
+    {
+      "epoch": 0.7424115486240898,
+      "grad_norm": 0.021820159725184658,
+      "learning_rate": 2.867883995703545e-06,
+      "loss": 0.0001,
+      "step": 11520
+    },
+    {
+      "epoch": 0.7424759940710188,
+      "grad_norm": 0.0030847242572352293,
+      "learning_rate": 2.8671679197994988e-06,
+      "loss": 0.0,
+      "step": 11521
+    },
+    {
+      "epoch": 0.742540439517948,
+      "grad_norm": 0.10641283012131907,
+      "learning_rate": 2.866451843895453e-06,
+      "loss": 0.0017,
+      "step": 11522
+    },
+    {
+      "epoch": 0.7426048849648772,
+      "grad_norm": 0.009833028385451683,
+      "learning_rate": 2.8657357679914074e-06,
+      "loss": 0.0001,
+      "step": 11523
+    },
+    {
+      "epoch": 0.7426693304118064,
+      "grad_norm": 0.003582891327112572,
+      "learning_rate": 2.8650196920873612e-06,
+      "loss": 0.0,
+      "step": 11524
+    },
+    {
+      "epoch": 0.7427337758587356,
+      "grad_norm": 0.00897878326744134,
+      "learning_rate": 2.8643036161833155e-06,
+      "loss": 0.0001,
+      "step": 11525
+    },
+    {
+      "epoch": 0.7427982213056648,
+      "grad_norm": 0.00011438453578087207,
+      "learning_rate": 2.8635875402792703e-06,
+      "loss": 0.0,
+      "step": 11526
+    },
+    {
+      "epoch": 0.7428626667525939,
+      "grad_norm": 1.8637766949118943,
+      "learning_rate": 2.862871464375224e-06,
+      "loss": 0.0185,
+      "step": 11527
+    },
+    {
+      "epoch": 0.7429271121995231,
+      "grad_norm": 8.769021280880539e-05,
+      "learning_rate": 2.8621553884711784e-06,
+      "loss": 0.0,
+      "step": 11528
+    },
+    {
+      "epoch": 0.7429915576464523,
+      "grad_norm": 0.009708635368675824,
+      "learning_rate": 2.8614393125671323e-06,
+      "loss": 0.0001,
+      "step": 11529
+    },
+    {
+      "epoch": 0.7430560030933815,
+      "grad_norm": 0.25208524448416436,
+      "learning_rate": 2.8607232366630866e-06,
+      "loss": 0.001,
+      "step": 11530
+    },
+    {
+      "epoch": 0.7431204485403107,
+      "grad_norm": 0.23889954660465637,
+      "learning_rate": 2.860007160759041e-06,
+      "loss": 0.0013,
+      "step": 11531
+    },
+    {
+      "epoch": 0.7431848939872397,
+      "grad_norm": 0.003267420010600027,
+      "learning_rate": 2.8592910848549948e-06,
+      "loss": 0.0,
+      "step": 11532
+    },
+    {
+      "epoch": 0.7432493394341689,
+      "grad_norm": 8.740468378403458e-05,
+      "learning_rate": 2.858575008950949e-06,
+      "loss": 0.0,
+      "step": 11533
+    },
+    {
+      "epoch": 0.7433137848810981,
+      "grad_norm": 6.484494334233853e-05,
+      "learning_rate": 2.857858933046903e-06,
+      "loss": 0.0,
+      "step": 11534
+    },
+    {
+      "epoch": 0.7433782303280273,
+      "grad_norm": 0.33328327409046465,
+      "learning_rate": 2.8571428571428573e-06,
+      "loss": 0.0026,
+      "step": 11535
+    },
+    {
+      "epoch": 0.7434426757749565,
+      "grad_norm": 0.03705969421250206,
+      "learning_rate": 2.856426781238811e-06,
+      "loss": 0.0001,
+      "step": 11536
+    },
+    {
+      "epoch": 0.7435071212218857,
+      "grad_norm": 0.08214135387497098,
+      "learning_rate": 2.8557107053347654e-06,
+      "loss": 0.0001,
+      "step": 11537
+    },
+    {
+      "epoch": 0.7435715666688149,
+      "grad_norm": 0.0009779501742388384,
+      "learning_rate": 2.85499462943072e-06,
+      "loss": 0.0,
+      "step": 11538
+    },
+    {
+      "epoch": 0.743636012115744,
+      "grad_norm": 0.0019099608653063945,
+      "learning_rate": 2.8542785535266745e-06,
+      "loss": 0.0,
+      "step": 11539
+    },
+    {
+      "epoch": 0.7437004575626732,
+      "grad_norm": 6.36266112464164e-05,
+      "learning_rate": 2.8535624776226283e-06,
+      "loss": 0.0,
+      "step": 11540
+    },
+    {
+      "epoch": 0.7437649030096024,
+      "grad_norm": 0.01838611981256415,
+      "learning_rate": 2.8528464017185826e-06,
+      "loss": 0.0,
+      "step": 11541
+    },
+    {
+      "epoch": 0.7438293484565316,
+      "grad_norm": 0.00011629883493401809,
+      "learning_rate": 2.8521303258145365e-06,
+      "loss": 0.0,
+      "step": 11542
+    },
+    {
+      "epoch": 0.7438937939034607,
+      "grad_norm": 0.002767961297251372,
+      "learning_rate": 2.851414249910491e-06,
+      "loss": 0.0,
+      "step": 11543
+    },
+    {
+      "epoch": 0.7439582393503898,
+      "grad_norm": 0.054474319327986465,
+      "learning_rate": 2.8506981740064447e-06,
+      "loss": 0.0004,
+      "step": 11544
+    },
+    {
+      "epoch": 0.744022684797319,
+      "grad_norm": 0.0006248430618284908,
+      "learning_rate": 2.849982098102399e-06,
+      "loss": 0.0,
+      "step": 11545
+    },
+    {
+      "epoch": 0.7440871302442482,
+      "grad_norm": 0.00481655779190564,
+      "learning_rate": 2.8492660221983533e-06,
+      "loss": 0.0,
+      "step": 11546
+    },
+    {
+      "epoch": 0.7441515756911774,
+      "grad_norm": 0.015192968024787968,
+      "learning_rate": 2.848549946294307e-06,
+      "loss": 0.0001,
+      "step": 11547
+    },
+    {
+      "epoch": 0.7442160211381066,
+      "grad_norm": 0.02535298427612439,
+      "learning_rate": 2.8478338703902614e-06,
+      "loss": 0.0001,
+      "step": 11548
+    },
+    {
+      "epoch": 0.7442804665850358,
+      "grad_norm": 0.008441149317855135,
+      "learning_rate": 2.8471177944862153e-06,
+      "loss": 0.0001,
+      "step": 11549
+    },
+    {
+      "epoch": 0.744344912031965,
+      "grad_norm": 0.00043440331282244664,
+      "learning_rate": 2.84640171858217e-06,
+      "loss": 0.0,
+      "step": 11550
+    },
+    {
+      "epoch": 0.7444093574788941,
+      "grad_norm": 0.0028974444813782536,
+      "learning_rate": 2.8456856426781243e-06,
+      "loss": 0.0,
+      "step": 11551
+    },
+    {
+      "epoch": 0.7444738029258233,
+      "grad_norm": 0.00011725375211290053,
+      "learning_rate": 2.8449695667740786e-06,
+      "loss": 0.0,
+      "step": 11552
+    },
+    {
+      "epoch": 0.7445382483727525,
+      "grad_norm": 0.0003650860736984975,
+      "learning_rate": 2.8442534908700325e-06,
+      "loss": 0.0,
+      "step": 11553
+    },
+    {
+      "epoch": 0.7446026938196817,
+      "grad_norm": 4.315019431511173e-05,
+      "learning_rate": 2.843537414965987e-06,
+      "loss": 0.0,
+      "step": 11554
+    },
+    {
+      "epoch": 0.7446671392666108,
+      "grad_norm": 0.002278690880074027,
+      "learning_rate": 2.8428213390619407e-06,
+      "loss": 0.0,
+      "step": 11555
+    },
+    {
+      "epoch": 0.74473158471354,
+      "grad_norm": 0.0001178197491432631,
+      "learning_rate": 2.842105263157895e-06,
+      "loss": 0.0,
+      "step": 11556
+    },
+    {
+      "epoch": 0.7447960301604691,
+      "grad_norm": 0.003881942521512389,
+      "learning_rate": 2.841389187253849e-06,
+      "loss": 0.0,
+      "step": 11557
+    },
+    {
+      "epoch": 0.7448604756073983,
+      "grad_norm": 0.0021296588472513025,
+      "learning_rate": 2.840673111349803e-06,
+      "loss": 0.0,
+      "step": 11558
+    },
+    {
+      "epoch": 0.7449249210543275,
+      "grad_norm": 0.00046791558014023247,
+      "learning_rate": 2.8399570354457575e-06,
+      "loss": 0.0,
+      "step": 11559
+    },
+    {
+      "epoch": 0.7449893665012567,
+      "grad_norm": 0.00020211463966043466,
+      "learning_rate": 2.8392409595417113e-06,
+      "loss": 0.0,
+      "step": 11560
+    },
+    {
+      "epoch": 0.7450538119481859,
+      "grad_norm": 0.0003985674663379224,
+      "learning_rate": 2.838524883637666e-06,
+      "loss": 0.0,
+      "step": 11561
+    },
+    {
+      "epoch": 0.7451182573951151,
+      "grad_norm": 0.33669691966511234,
+      "learning_rate": 2.8378088077336204e-06,
+      "loss": 0.0031,
+      "step": 11562
+    },
+    {
+      "epoch": 0.7451827028420442,
+      "grad_norm": 0.012881644125098787,
+      "learning_rate": 2.8370927318295742e-06,
+      "loss": 0.0001,
+      "step": 11563
+    },
+    {
+      "epoch": 0.7452471482889734,
+      "grad_norm": 0.0004095707305752869,
+      "learning_rate": 2.8363766559255285e-06,
+      "loss": 0.0,
+      "step": 11564
+    },
+    {
+      "epoch": 0.7453115937359026,
+      "grad_norm": 0.0553894190092041,
+      "learning_rate": 2.8356605800214824e-06,
+      "loss": 0.0002,
+      "step": 11565
+    },
+    {
+      "epoch": 0.7453760391828317,
+      "grad_norm": 0.0015646026292246813,
+      "learning_rate": 2.8349445041174367e-06,
+      "loss": 0.0,
+      "step": 11566
+    },
+    {
+      "epoch": 0.7454404846297609,
+      "grad_norm": 0.0007490978582765871,
+      "learning_rate": 2.834228428213391e-06,
+      "loss": 0.0,
+      "step": 11567
+    },
+    {
+      "epoch": 0.74550493007669,
+      "grad_norm": 0.0008107603181492814,
+      "learning_rate": 2.833512352309345e-06,
+      "loss": 0.0,
+      "step": 11568
+    },
+    {
+      "epoch": 0.7455693755236192,
+      "grad_norm": 0.0005066625926412042,
+      "learning_rate": 2.832796276405299e-06,
+      "loss": 0.0,
+      "step": 11569
+    },
+    {
+      "epoch": 0.7456338209705484,
+      "grad_norm": 0.0031491577925305417,
+      "learning_rate": 2.832080200501253e-06,
+      "loss": 0.0,
+      "step": 11570
+    },
+    {
+      "epoch": 0.7456982664174776,
+      "grad_norm": 0.0008086278143267719,
+      "learning_rate": 2.8313641245972074e-06,
+      "loss": 0.0,
+      "step": 11571
+    },
+    {
+      "epoch": 0.7457627118644068,
+      "grad_norm": 0.0003411885658767725,
+      "learning_rate": 2.8306480486931616e-06,
+      "loss": 0.0,
+      "step": 11572
+    },
+    {
+      "epoch": 0.745827157311336,
+      "grad_norm": 0.0005964478571595986,
+      "learning_rate": 2.829931972789116e-06,
+      "loss": 0.0,
+      "step": 11573
+    },
+    {
+      "epoch": 0.7458916027582652,
+      "grad_norm": 0.12147032343687166,
+      "learning_rate": 2.8292158968850702e-06,
+      "loss": 0.0002,
+      "step": 11574
+    },
+    {
+      "epoch": 0.7459560482051943,
+      "grad_norm": 0.03803393591887575,
+      "learning_rate": 2.8284998209810245e-06,
+      "loss": 0.0016,
+      "step": 11575
+    },
+    {
+      "epoch": 0.7460204936521235,
+      "grad_norm": 0.00016790024335527423,
+      "learning_rate": 2.8277837450769784e-06,
+      "loss": 0.0,
+      "step": 11576
+    },
+    {
+      "epoch": 0.7460849390990526,
+      "grad_norm": 0.0024457187488064213,
+      "learning_rate": 2.8270676691729327e-06,
+      "loss": 0.0,
+      "step": 11577
+    },
+    {
+      "epoch": 0.7461493845459818,
+      "grad_norm": 6.03264194184613e-05,
+      "learning_rate": 2.8263515932688866e-06,
+      "loss": 0.0,
+      "step": 11578
+    },
+    {
+      "epoch": 0.746213829992911,
+      "grad_norm": 0.0009593868231441341,
+      "learning_rate": 2.825635517364841e-06,
+      "loss": 0.0,
+      "step": 11579
+    },
+    {
+      "epoch": 0.7462782754398402,
+      "grad_norm": 0.00015227808015561272,
+      "learning_rate": 2.824919441460795e-06,
+      "loss": 0.0,
+      "step": 11580
+    },
+    {
+      "epoch": 0.7463427208867693,
+      "grad_norm": 0.0005544944467892892,
+      "learning_rate": 2.824203365556749e-06,
+      "loss": 0.0,
+      "step": 11581
+    },
+    {
+      "epoch": 0.7464071663336985,
+      "grad_norm": 0.14816729522203623,
+      "learning_rate": 2.8234872896527034e-06,
+      "loss": 0.0003,
+      "step": 11582
+    },
+    {
+      "epoch": 0.7464716117806277,
+      "grad_norm": 0.12938729032013577,
+      "learning_rate": 2.8227712137486572e-06,
+      "loss": 0.0004,
+      "step": 11583
+    },
+    {
+      "epoch": 0.7465360572275569,
+      "grad_norm": 0.0844716364187078,
+      "learning_rate": 2.8220551378446115e-06,
+      "loss": 0.0006,
+      "step": 11584
+    },
+    {
+      "epoch": 0.7466005026744861,
+      "grad_norm": 0.002823985210489743,
+      "learning_rate": 2.8213390619405663e-06,
+      "loss": 0.0,
+      "step": 11585
+    },
+    {
+      "epoch": 0.7466649481214153,
+      "grad_norm": 0.05066442807138474,
+      "learning_rate": 2.82062298603652e-06,
+      "loss": 0.0002,
+      "step": 11586
+    },
+    {
+      "epoch": 0.7467293935683444,
+      "grad_norm": 0.004276016365310969,
+      "learning_rate": 2.8199069101324744e-06,
+      "loss": 0.0,
+      "step": 11587
+    },
+    {
+      "epoch": 0.7467938390152735,
+      "grad_norm": 0.0028644723760939557,
+      "learning_rate": 2.8191908342284287e-06,
+      "loss": 0.0,
+      "step": 11588
+    },
+    {
+      "epoch": 0.7468582844622027,
+      "grad_norm": 0.0002588828785945799,
+      "learning_rate": 2.8184747583243826e-06,
+      "loss": 0.0,
+      "step": 11589
+    },
+    {
+      "epoch": 0.7469227299091319,
+      "grad_norm": 0.00039665747800869124,
+      "learning_rate": 2.817758682420337e-06,
+      "loss": 0.0,
+      "step": 11590
+    },
+    {
+      "epoch": 0.7469871753560611,
+      "grad_norm": 0.0003482603860227377,
+      "learning_rate": 2.8170426065162908e-06,
+      "loss": 0.0,
+      "step": 11591
+    },
+    {
+      "epoch": 0.7470516208029903,
+      "grad_norm": 0.0008336352241551569,
+      "learning_rate": 2.816326530612245e-06,
+      "loss": 0.0,
+      "step": 11592
+    },
+    {
+      "epoch": 0.7471160662499194,
+      "grad_norm": 5.235047905893616e-05,
+      "learning_rate": 2.815610454708199e-06,
+      "loss": 0.0,
+      "step": 11593
+    },
+    {
+      "epoch": 0.7471805116968486,
+      "grad_norm": 0.0014324225534880688,
+      "learning_rate": 2.8148943788041533e-06,
+      "loss": 0.0,
+      "step": 11594
+    },
+    {
+      "epoch": 0.7472449571437778,
+      "grad_norm": 0.0017366095073440056,
+      "learning_rate": 2.8141783029001076e-06,
+      "loss": 0.0,
+      "step": 11595
+    },
+    {
+      "epoch": 0.747309402590707,
+      "grad_norm": 0.0004592679869086307,
+      "learning_rate": 2.8134622269960623e-06,
+      "loss": 0.0,
+      "step": 11596
+    },
+    {
+      "epoch": 0.7473738480376362,
+      "grad_norm": 0.0057053165657561505,
+      "learning_rate": 2.812746151092016e-06,
+      "loss": 0.0,
+      "step": 11597
+    },
+    {
+      "epoch": 0.7474382934845654,
+      "grad_norm": 0.0006819042703925117,
+      "learning_rate": 2.8120300751879705e-06,
+      "loss": 0.0015,
+      "step": 11598
+    },
+    {
+      "epoch": 0.7475027389314944,
+      "grad_norm": 0.6504672810054322,
+      "learning_rate": 2.8113139992839243e-06,
+      "loss": 0.002,
+      "step": 11599
+    },
+    {
+      "epoch": 0.7475671843784236,
+      "grad_norm": 0.00035804196067088125,
+      "learning_rate": 2.8105979233798786e-06,
+      "loss": 0.0,
+      "step": 11600
+    },
+    {
+      "epoch": 0.7476316298253528,
+      "grad_norm": 0.0016188882058790126,
+      "learning_rate": 2.8098818474758325e-06,
+      "loss": 0.0,
+      "step": 11601
+    },
+    {
+      "epoch": 0.747696075272282,
+      "grad_norm": 0.00036099919084685566,
+      "learning_rate": 2.809165771571787e-06,
+      "loss": 0.0,
+      "step": 11602
+    },
+    {
+      "epoch": 0.7477605207192112,
+      "grad_norm": 0.0029460432117195917,
+      "learning_rate": 2.808449695667741e-06,
+      "loss": 0.0,
+      "step": 11603
+    },
+    {
+      "epoch": 0.7478249661661404,
+      "grad_norm": 0.00011651976692317094,
+      "learning_rate": 2.807733619763695e-06,
+      "loss": 0.0,
+      "step": 11604
+    },
+    {
+      "epoch": 0.7478894116130695,
+      "grad_norm": 0.0025859682305249014,
+      "learning_rate": 2.8070175438596493e-06,
+      "loss": 0.0,
+      "step": 11605
+    },
+    {
+      "epoch": 0.7479538570599987,
+      "grad_norm": 4.958741962410607e-05,
+      "learning_rate": 2.806301467955603e-06,
+      "loss": 0.0,
+      "step": 11606
+    },
+    {
+      "epoch": 0.7480183025069279,
+      "grad_norm": 0.0029761885102953744,
+      "learning_rate": 2.8055853920515574e-06,
+      "loss": 0.0,
+      "step": 11607
+    },
+    {
+      "epoch": 0.7480827479538571,
+      "grad_norm": 0.0008891637672764099,
+      "learning_rate": 2.804869316147512e-06,
+      "loss": 0.0,
+      "step": 11608
+    },
+    {
+      "epoch": 0.7481471934007863,
+      "grad_norm": 2.754389762637177e-05,
+      "learning_rate": 2.804153240243466e-06,
+      "loss": 0.0,
+      "step": 11609
+    },
+    {
+      "epoch": 0.7482116388477154,
+      "grad_norm": 0.006316527622726746,
+      "learning_rate": 2.8034371643394203e-06,
+      "loss": 0.0,
+      "step": 11610
+    },
+    {
+      "epoch": 0.7482760842946445,
+      "grad_norm": 0.02277983004237064,
+      "learning_rate": 2.8027210884353746e-06,
+      "loss": 0.0001,
+      "step": 11611
+    },
+    {
+      "epoch": 0.7483405297415737,
+      "grad_norm": 5.643851932003774e-05,
+      "learning_rate": 2.8020050125313285e-06,
+      "loss": 0.0,
+      "step": 11612
+    },
+    {
+      "epoch": 0.7484049751885029,
+      "grad_norm": 0.00037393311065744746,
+      "learning_rate": 2.801288936627283e-06,
+      "loss": 0.0,
+      "step": 11613
+    },
+    {
+      "epoch": 0.7484694206354321,
+      "grad_norm": 0.002534550843795288,
+      "learning_rate": 2.8005728607232367e-06,
+      "loss": 0.0,
+      "step": 11614
+    },
+    {
+      "epoch": 0.7485338660823613,
+      "grad_norm": 0.024649051818809553,
+      "learning_rate": 2.799856784819191e-06,
+      "loss": 0.0002,
+      "step": 11615
+    },
+    {
+      "epoch": 0.7485983115292905,
+      "grad_norm": 0.041082903201940656,
+      "learning_rate": 2.7991407089151453e-06,
+      "loss": 0.0002,
+      "step": 11616
+    },
+    {
+      "epoch": 0.7486627569762196,
+      "grad_norm": 0.1127003747936899,
+      "learning_rate": 2.798424633011099e-06,
+      "loss": 0.0002,
+      "step": 11617
+    },
+    {
+      "epoch": 0.7487272024231488,
+      "grad_norm": 0.0001859838946770348,
+      "learning_rate": 2.7977085571070535e-06,
+      "loss": 0.0,
+      "step": 11618
+    },
+    {
+      "epoch": 0.748791647870078,
+      "grad_norm": 5.065890990128626e-05,
+      "learning_rate": 2.7969924812030073e-06,
+      "loss": 0.0,
+      "step": 11619
+    },
+    {
+      "epoch": 0.7488560933170072,
+      "grad_norm": 9.265103468572512e-05,
+      "learning_rate": 2.796276405298962e-06,
+      "loss": 0.0,
+      "step": 11620
+    },
+    {
+      "epoch": 0.7489205387639364,
+      "grad_norm": 0.0008966920858239456,
+      "learning_rate": 2.7955603293949164e-06,
+      "loss": 0.0,
+      "step": 11621
+    },
+    {
+      "epoch": 0.7489849842108655,
+      "grad_norm": 0.00012183848043519522,
+      "learning_rate": 2.7948442534908702e-06,
+      "loss": 0.0,
+      "step": 11622
+    },
+    {
+      "epoch": 0.7490494296577946,
+      "grad_norm": 0.0016598313811174246,
+      "learning_rate": 2.7941281775868245e-06,
+      "loss": 0.0,
+      "step": 11623
+    },
+    {
+      "epoch": 0.7491138751047238,
+      "grad_norm": 1.1238977335514882e-05,
+      "learning_rate": 2.793412101682779e-06,
+      "loss": 0.0,
+      "step": 11624
+    },
+    {
+      "epoch": 0.749178320551653,
+      "grad_norm": 0.00034594887045504333,
+      "learning_rate": 2.7926960257787327e-06,
+      "loss": 0.0,
+      "step": 11625
+    },
+    {
+      "epoch": 0.7492427659985822,
+      "grad_norm": 0.04615189947607823,
+      "learning_rate": 2.791979949874687e-06,
+      "loss": 0.0001,
+      "step": 11626
+    },
+    {
+      "epoch": 0.7493072114455114,
+      "grad_norm": 0.0002986722078818211,
+      "learning_rate": 2.791263873970641e-06,
+      "loss": 0.0,
+      "step": 11627
+    },
+    {
+      "epoch": 0.7493716568924406,
+      "grad_norm": 0.005092092729537614,
+      "learning_rate": 2.790547798066595e-06,
+      "loss": 0.0,
+      "step": 11628
+    },
+    {
+      "epoch": 0.7494361023393697,
+      "grad_norm": 0.1200249868569926,
+      "learning_rate": 2.7898317221625495e-06,
+      "loss": 0.0002,
+      "step": 11629
+    },
+    {
+      "epoch": 0.7495005477862989,
+      "grad_norm": 0.0049599157186567015,
+      "learning_rate": 2.7891156462585034e-06,
+      "loss": 0.0,
+      "step": 11630
+    },
+    {
+      "epoch": 0.7495649932332281,
+      "grad_norm": 8.649864165149563e-05,
+      "learning_rate": 2.788399570354458e-06,
+      "loss": 0.0,
+      "step": 11631
+    },
+    {
+      "epoch": 0.7496294386801573,
+      "grad_norm": 0.0016241807887423438,
+      "learning_rate": 2.7876834944504124e-06,
+      "loss": 0.0,
+      "step": 11632
+    },
+    {
+      "epoch": 0.7496938841270864,
+      "grad_norm": 0.000766953844427467,
+      "learning_rate": 2.7869674185463662e-06,
+      "loss": 0.0,
+      "step": 11633
+    },
+    {
+      "epoch": 0.7497583295740156,
+      "grad_norm": 0.00042285343101394534,
+      "learning_rate": 2.7862513426423205e-06,
+      "loss": 0.0,
+      "step": 11634
+    },
+    {
+      "epoch": 0.7498227750209447,
+      "grad_norm": 0.000205183922965492,
+      "learning_rate": 2.7855352667382744e-06,
+      "loss": 0.0,
+      "step": 11635
+    },
+    {
+      "epoch": 0.7498872204678739,
+      "grad_norm": 5.591482784916908e-05,
+      "learning_rate": 2.7848191908342287e-06,
+      "loss": 0.0,
+      "step": 11636
+    },
+    {
+      "epoch": 0.7499516659148031,
+      "grad_norm": 0.0005972326634359615,
+      "learning_rate": 2.784103114930183e-06,
+      "loss": 0.0,
+      "step": 11637
+    },
+    {
+      "epoch": 0.7500161113617323,
+      "grad_norm": 0.08351007143777367,
+      "learning_rate": 2.783387039026137e-06,
+      "loss": 0.0016,
+      "step": 11638
+    },
+    {
+      "epoch": 0.7500805568086615,
+      "grad_norm": 1.1101820253498027,
+      "learning_rate": 2.782670963122091e-06,
+      "loss": 0.0085,
+      "step": 11639
+    },
+    {
+      "epoch": 0.7501450022555907,
+      "grad_norm": 4.642928998663188e-05,
+      "learning_rate": 2.781954887218045e-06,
+      "loss": 0.0,
+      "step": 11640
+    },
+    {
+      "epoch": 0.7502094477025198,
+      "grad_norm": 0.003023120807805911,
+      "learning_rate": 2.7812388113139994e-06,
+      "loss": 0.0,
+      "step": 11641
+    },
+    {
+      "epoch": 0.750273893149449,
+      "grad_norm": 0.00029685534288482725,
+      "learning_rate": 2.7805227354099532e-06,
+      "loss": 0.0,
+      "step": 11642
+    },
+    {
+      "epoch": 0.7503383385963782,
+      "grad_norm": 0.00021844772164633,
+      "learning_rate": 2.779806659505908e-06,
+      "loss": 0.0,
+      "step": 11643
+    },
+    {
+      "epoch": 0.7504027840433073,
+      "grad_norm": 0.00017051293991199247,
+      "learning_rate": 2.7790905836018623e-06,
+      "loss": 0.0,
+      "step": 11644
+    },
+    {
+      "epoch": 0.7504672294902365,
+      "grad_norm": 0.0012412810367730285,
+      "learning_rate": 2.7783745076978166e-06,
+      "loss": 0.0,
+      "step": 11645
+    },
+    {
+      "epoch": 0.7505316749371657,
+      "grad_norm": 9.528078393517384e-05,
+      "learning_rate": 2.7776584317937704e-06,
+      "loss": 0.0,
+      "step": 11646
+    },
+    {
+      "epoch": 0.7505961203840948,
+      "grad_norm": 0.0011691385906994708,
+      "learning_rate": 2.7769423558897247e-06,
+      "loss": 0.0,
+      "step": 11647
+    },
+    {
+      "epoch": 0.750660565831024,
+      "grad_norm": 0.00023046299115114463,
+      "learning_rate": 2.7762262799856786e-06,
+      "loss": 0.0,
+      "step": 11648
+    },
+    {
+      "epoch": 0.7507250112779532,
+      "grad_norm": 0.0002655424185747253,
+      "learning_rate": 2.775510204081633e-06,
+      "loss": 0.0,
+      "step": 11649
+    },
+    {
+      "epoch": 0.7507894567248824,
+      "grad_norm": 0.19039229979656985,
+      "learning_rate": 2.7747941281775868e-06,
+      "loss": 0.0004,
+      "step": 11650
+    },
+    {
+      "epoch": 0.7508539021718116,
+      "grad_norm": 0.0019108833699937504,
+      "learning_rate": 2.774078052273541e-06,
+      "loss": 0.0,
+      "step": 11651
+    },
+    {
+      "epoch": 0.7509183476187408,
+      "grad_norm": 0.0007730187024007991,
+      "learning_rate": 2.7733619763694954e-06,
+      "loss": 0.0,
+      "step": 11652
+    },
+    {
+      "epoch": 0.75098279306567,
+      "grad_norm": 0.003987731423439629,
+      "learning_rate": 2.7726459004654493e-06,
+      "loss": 0.0,
+      "step": 11653
+    },
+    {
+      "epoch": 0.7510472385125991,
+      "grad_norm": 0.000375934201446267,
+      "learning_rate": 2.7719298245614036e-06,
+      "loss": 0.0,
+      "step": 11654
+    },
+    {
+      "epoch": 0.7511116839595282,
+      "grad_norm": 0.31494069653808116,
+      "learning_rate": 2.7712137486573583e-06,
+      "loss": 0.0014,
+      "step": 11655
+    },
+    {
+      "epoch": 0.7511761294064574,
+      "grad_norm": 0.005691559960223658,
+      "learning_rate": 2.770497672753312e-06,
+      "loss": 0.0001,
+      "step": 11656
+    },
+    {
+      "epoch": 0.7512405748533866,
+      "grad_norm": 0.0004438557435561563,
+      "learning_rate": 2.7697815968492665e-06,
+      "loss": 0.0,
+      "step": 11657
+    },
+    {
+      "epoch": 0.7513050203003158,
+      "grad_norm": 0.05629256404691263,
+      "learning_rate": 2.7690655209452203e-06,
+      "loss": 0.0001,
+      "step": 11658
+    },
+    {
+      "epoch": 0.751369465747245,
+      "grad_norm": 0.00039722302627628154,
+      "learning_rate": 2.7683494450411746e-06,
+      "loss": 0.0,
+      "step": 11659
+    },
+    {
+      "epoch": 0.7514339111941741,
+      "grad_norm": 0.025686982877658825,
+      "learning_rate": 2.767633369137129e-06,
+      "loss": 0.0002,
+      "step": 11660
+    },
+    {
+      "epoch": 0.7514983566411033,
+      "grad_norm": 0.0007727245090655069,
+      "learning_rate": 2.766917293233083e-06,
+      "loss": 0.0,
+      "step": 11661
+    },
+    {
+      "epoch": 0.7515628020880325,
+      "grad_norm": 0.000900820580803089,
+      "learning_rate": 2.766201217329037e-06,
+      "loss": 0.0,
+      "step": 11662
+    },
+    {
+      "epoch": 0.7516272475349617,
+      "grad_norm": 0.00027621965420460257,
+      "learning_rate": 2.765485141424991e-06,
+      "loss": 0.0,
+      "step": 11663
+    },
+    {
+      "epoch": 0.7516916929818909,
+      "grad_norm": 0.0007073996963652853,
+      "learning_rate": 2.7647690655209453e-06,
+      "loss": 0.0,
+      "step": 11664
+    },
+    {
+      "epoch": 0.75175613842882,
+      "grad_norm": 0.0006071636553376335,
+      "learning_rate": 2.7640529896168996e-06,
+      "loss": 0.0,
+      "step": 11665
+    },
+    {
+      "epoch": 0.7518205838757491,
+      "grad_norm": 0.012462280848083922,
+      "learning_rate": 2.763336913712854e-06,
+      "loss": 0.0001,
+      "step": 11666
+    },
+    {
+      "epoch": 0.7518850293226783,
+      "grad_norm": 0.00018560450923334242,
+      "learning_rate": 2.762620837808808e-06,
+      "loss": 0.0,
+      "step": 11667
+    },
+    {
+      "epoch": 0.7519494747696075,
+      "grad_norm": 0.0006644943512278006,
+      "learning_rate": 2.7619047619047625e-06,
+      "loss": 0.0,
+      "step": 11668
+    },
+    {
+      "epoch": 0.7520139202165367,
+      "grad_norm": 0.0020807138950169435,
+      "learning_rate": 2.7611886860007163e-06,
+      "loss": 0.0,
+      "step": 11669
+    },
+    {
+      "epoch": 0.7520783656634659,
+      "grad_norm": 5.37856004501817e-05,
+      "learning_rate": 2.7604726100966706e-06,
+      "loss": 0.0,
+      "step": 11670
+    },
+    {
+      "epoch": 0.752142811110395,
+      "grad_norm": 0.0005517811580377726,
+      "learning_rate": 2.7597565341926245e-06,
+      "loss": 0.0,
+      "step": 11671
+    },
+    {
+      "epoch": 0.7522072565573242,
+      "grad_norm": 0.018654850971019456,
+      "learning_rate": 2.759040458288579e-06,
+      "loss": 0.0002,
+      "step": 11672
+    },
+    {
+      "epoch": 0.7522717020042534,
+      "grad_norm": 0.0015801575484079477,
+      "learning_rate": 2.758324382384533e-06,
+      "loss": 0.0,
+      "step": 11673
+    },
+    {
+      "epoch": 0.7523361474511826,
+      "grad_norm": 0.004039334597610442,
+      "learning_rate": 2.757608306480487e-06,
+      "loss": 0.0,
+      "step": 11674
+    },
+    {
+      "epoch": 0.7524005928981118,
+      "grad_norm": 0.06624770873534068,
+      "learning_rate": 2.7568922305764413e-06,
+      "loss": 0.0005,
+      "step": 11675
+    },
+    {
+      "epoch": 0.752465038345041,
+      "grad_norm": 0.0010852293084164862,
+      "learning_rate": 2.756176154672395e-06,
+      "loss": 0.0,
+      "step": 11676
+    },
+    {
+      "epoch": 0.75252948379197,
+      "grad_norm": 0.0002506029780493605,
+      "learning_rate": 2.7554600787683495e-06,
+      "loss": 0.0,
+      "step": 11677
+    },
+    {
+      "epoch": 0.7525939292388992,
+      "grad_norm": 0.0009374336268016534,
+      "learning_rate": 2.754744002864304e-06,
+      "loss": 0.0,
+      "step": 11678
+    },
+    {
+      "epoch": 0.7526583746858284,
+      "grad_norm": 8.921424638537497e-05,
+      "learning_rate": 2.754027926960258e-06,
+      "loss": 0.0,
+      "step": 11679
+    },
+    {
+      "epoch": 0.7527228201327576,
+      "grad_norm": 0.0015448375941823727,
+      "learning_rate": 2.7533118510562124e-06,
+      "loss": 0.0,
+      "step": 11680
+    },
+    {
+      "epoch": 0.7527872655796868,
+      "grad_norm": 0.0010091009396157359,
+      "learning_rate": 2.7525957751521667e-06,
+      "loss": 0.0,
+      "step": 11681
+    },
+    {
+      "epoch": 0.752851711026616,
+      "grad_norm": 0.0004459042079651145,
+      "learning_rate": 2.7518796992481205e-06,
+      "loss": 0.0,
+      "step": 11682
+    },
+    {
+      "epoch": 0.7529161564735452,
+      "grad_norm": 0.01627399213125485,
+      "learning_rate": 2.751163623344075e-06,
+      "loss": 0.0,
+      "step": 11683
+    },
+    {
+      "epoch": 0.7529806019204743,
+      "grad_norm": 0.013965691610753837,
+      "learning_rate": 2.7504475474400287e-06,
+      "loss": 0.0001,
+      "step": 11684
+    },
+    {
+      "epoch": 0.7530450473674035,
+      "grad_norm": 0.0001156550801431203,
+      "learning_rate": 2.749731471535983e-06,
+      "loss": 0.0,
+      "step": 11685
+    },
+    {
+      "epoch": 0.7531094928143327,
+      "grad_norm": 0.3935840113011604,
+      "learning_rate": 2.7490153956319373e-06,
+      "loss": 0.0004,
+      "step": 11686
+    },
+    {
+      "epoch": 0.7531739382612619,
+      "grad_norm": 0.0007292888015694302,
+      "learning_rate": 2.748299319727891e-06,
+      "loss": 0.0,
+      "step": 11687
+    },
+    {
+      "epoch": 0.753238383708191,
+      "grad_norm": 0.00018506515606916054,
+      "learning_rate": 2.7475832438238455e-06,
+      "loss": 0.0,
+      "step": 11688
+    },
+    {
+      "epoch": 0.7533028291551201,
+      "grad_norm": 0.0002825209862627467,
+      "learning_rate": 2.7468671679197994e-06,
+      "loss": 0.0,
+      "step": 11689
+    },
+    {
+      "epoch": 0.7533672746020493,
+      "grad_norm": 0.0049305875242178976,
+      "learning_rate": 2.746151092015754e-06,
+      "loss": 0.0,
+      "step": 11690
+    },
+    {
+      "epoch": 0.7534317200489785,
+      "grad_norm": 0.0003643963192400323,
+      "learning_rate": 2.7454350161117084e-06,
+      "loss": 0.0,
+      "step": 11691
+    },
+    {
+      "epoch": 0.7534961654959077,
+      "grad_norm": 0.0035886233576998884,
+      "learning_rate": 2.7447189402076622e-06,
+      "loss": 0.0,
+      "step": 11692
+    },
+    {
+      "epoch": 0.7535606109428369,
+      "grad_norm": 0.03080716919407748,
+      "learning_rate": 2.7440028643036165e-06,
+      "loss": 0.0001,
+      "step": 11693
+    },
+    {
+      "epoch": 0.7536250563897661,
+      "grad_norm": 0.00470493868439932,
+      "learning_rate": 2.743286788399571e-06,
+      "loss": 0.0,
+      "step": 11694
+    },
+    {
+      "epoch": 0.7536895018366953,
+      "grad_norm": 0.00019118115247683613,
+      "learning_rate": 2.7425707124955247e-06,
+      "loss": 0.0,
+      "step": 11695
+    },
+    {
+      "epoch": 0.7537539472836244,
+      "grad_norm": 5.5976844224649946e-05,
+      "learning_rate": 2.741854636591479e-06,
+      "loss": 0.0,
+      "step": 11696
+    },
+    {
+      "epoch": 0.7538183927305536,
+      "grad_norm": 0.0237676543025234,
+      "learning_rate": 2.741138560687433e-06,
+      "loss": 0.0,
+      "step": 11697
+    },
+    {
+      "epoch": 0.7538828381774828,
+      "grad_norm": 0.38799647013666066,
+      "learning_rate": 2.740422484783387e-06,
+      "loss": 0.0029,
+      "step": 11698
+    },
+    {
+      "epoch": 0.753947283624412,
+      "grad_norm": 0.003417863242828486,
+      "learning_rate": 2.739706408879341e-06,
+      "loss": 0.0,
+      "step": 11699
+    },
+    {
+      "epoch": 0.7540117290713411,
+      "grad_norm": 3.090706308481155e-05,
+      "learning_rate": 2.7389903329752954e-06,
+      "loss": 0.0,
+      "step": 11700
+    },
+    {
+      "epoch": 0.7540761745182702,
+      "grad_norm": 6.625108551354473e-05,
+      "learning_rate": 2.73827425707125e-06,
+      "loss": 0.0,
+      "step": 11701
+    },
+    {
+      "epoch": 0.7541406199651994,
+      "grad_norm": 0.00021590261960305907,
+      "learning_rate": 2.7375581811672044e-06,
+      "loss": 0.0,
+      "step": 11702
+    },
+    {
+      "epoch": 0.7542050654121286,
+      "grad_norm": 0.0006234648806642372,
+      "learning_rate": 2.7368421052631583e-06,
+      "loss": 0.0,
+      "step": 11703
+    },
+    {
+      "epoch": 0.7542695108590578,
+      "grad_norm": 0.0071332380691034,
+      "learning_rate": 2.7361260293591126e-06,
+      "loss": 0.0,
+      "step": 11704
+    },
+    {
+      "epoch": 0.754333956305987,
+      "grad_norm": 1.2562028050453755,
+      "learning_rate": 2.7354099534550664e-06,
+      "loss": 0.0093,
+      "step": 11705
+    },
+    {
+      "epoch": 0.7543984017529162,
+      "grad_norm": 0.0029726964662291043,
+      "learning_rate": 2.7346938775510207e-06,
+      "loss": 0.0,
+      "step": 11706
+    },
+    {
+      "epoch": 0.7544628471998454,
+      "grad_norm": 0.014132551918677235,
+      "learning_rate": 2.7339778016469746e-06,
+      "loss": 0.0002,
+      "step": 11707
+    },
+    {
+      "epoch": 0.7545272926467745,
+      "grad_norm": 0.001203151290063077,
+      "learning_rate": 2.733261725742929e-06,
+      "loss": 0.0,
+      "step": 11708
+    },
+    {
+      "epoch": 0.7545917380937037,
+      "grad_norm": 0.009634512422446848,
+      "learning_rate": 2.732545649838883e-06,
+      "loss": 0.0001,
+      "step": 11709
+    },
+    {
+      "epoch": 0.7546561835406329,
+      "grad_norm": 0.16926328691142287,
+      "learning_rate": 2.731829573934837e-06,
+      "loss": 0.0003,
+      "step": 11710
+    },
+    {
+      "epoch": 0.754720628987562,
+      "grad_norm": 0.021506676559323645,
+      "learning_rate": 2.7311134980307914e-06,
+      "loss": 0.0001,
+      "step": 11711
+    },
+    {
+      "epoch": 0.7547850744344912,
+      "grad_norm": 0.12201709350159476,
+      "learning_rate": 2.7303974221267453e-06,
+      "loss": 0.0001,
+      "step": 11712
+    },
+    {
+      "epoch": 0.7548495198814203,
+      "grad_norm": 0.010025205131539256,
+      "learning_rate": 2.7296813462227e-06,
+      "loss": 0.0001,
+      "step": 11713
+    },
+    {
+      "epoch": 0.7549139653283495,
+      "grad_norm": 0.00034131962690424583,
+      "learning_rate": 2.7289652703186543e-06,
+      "loss": 0.0,
+      "step": 11714
+    },
+    {
+      "epoch": 0.7549784107752787,
+      "grad_norm": 0.0005576713005746142,
+      "learning_rate": 2.728249194414608e-06,
+      "loss": 0.0,
+      "step": 11715
+    },
+    {
+      "epoch": 0.7550428562222079,
+      "grad_norm": 0.00021151105237174136,
+      "learning_rate": 2.7275331185105625e-06,
+      "loss": 0.0,
+      "step": 11716
+    },
+    {
+      "epoch": 0.7551073016691371,
+      "grad_norm": 0.0005511650236481126,
+      "learning_rate": 2.7268170426065167e-06,
+      "loss": 0.0,
+      "step": 11717
+    },
+    {
+      "epoch": 0.7551717471160663,
+      "grad_norm": 0.0005032934039541653,
+      "learning_rate": 2.7261009667024706e-06,
+      "loss": 0.0,
+      "step": 11718
+    },
+    {
+      "epoch": 0.7552361925629955,
+      "grad_norm": 0.004268815835783067,
+      "learning_rate": 2.725384890798425e-06,
+      "loss": 0.0,
+      "step": 11719
+    },
+    {
+      "epoch": 0.7553006380099246,
+      "grad_norm": 0.029657345675551724,
+      "learning_rate": 2.724668814894379e-06,
+      "loss": 0.0001,
+      "step": 11720
+    },
+    {
+      "epoch": 0.7553650834568538,
+      "grad_norm": 0.00019803287793178183,
+      "learning_rate": 2.723952738990333e-06,
+      "loss": 0.0,
+      "step": 11721
+    },
+    {
+      "epoch": 0.7554295289037829,
+      "grad_norm": 0.017483873308816737,
+      "learning_rate": 2.7232366630862874e-06,
+      "loss": 0.0001,
+      "step": 11722
+    },
+    {
+      "epoch": 0.7554939743507121,
+      "grad_norm": 0.0002098614959974441,
+      "learning_rate": 2.7225205871822413e-06,
+      "loss": 0.0,
+      "step": 11723
+    },
+    {
+      "epoch": 0.7555584197976413,
+      "grad_norm": 0.0066547328197811,
+      "learning_rate": 2.7218045112781956e-06,
+      "loss": 0.0,
+      "step": 11724
+    },
+    {
+      "epoch": 0.7556228652445705,
+      "grad_norm": 0.00320167998285512,
+      "learning_rate": 2.7210884353741503e-06,
+      "loss": 0.0,
+      "step": 11725
+    },
+    {
+      "epoch": 0.7556873106914996,
+      "grad_norm": 0.00010177039287106444,
+      "learning_rate": 2.720372359470104e-06,
+      "loss": 0.0,
+      "step": 11726
+    },
+    {
+      "epoch": 0.7557517561384288,
+      "grad_norm": 0.00154407389412439,
+      "learning_rate": 2.7196562835660585e-06,
+      "loss": 0.0,
+      "step": 11727
+    },
+    {
+      "epoch": 0.755816201585358,
+      "grad_norm": 0.0033565564377196987,
+      "learning_rate": 2.7189402076620123e-06,
+      "loss": 0.0,
+      "step": 11728
+    },
+    {
+      "epoch": 0.7558806470322872,
+      "grad_norm": 0.1146090764871986,
+      "learning_rate": 2.7182241317579666e-06,
+      "loss": 0.0001,
+      "step": 11729
+    },
+    {
+      "epoch": 0.7559450924792164,
+      "grad_norm": 0.0142495845011723,
+      "learning_rate": 2.717508055853921e-06,
+      "loss": 0.0,
+      "step": 11730
+    },
+    {
+      "epoch": 0.7560095379261456,
+      "grad_norm": 0.028604510943794178,
+      "learning_rate": 2.716791979949875e-06,
+      "loss": 0.0,
+      "step": 11731
+    },
+    {
+      "epoch": 0.7560739833730747,
+      "grad_norm": 0.0015702490887233036,
+      "learning_rate": 2.716075904045829e-06,
+      "loss": 0.0,
+      "step": 11732
+    },
+    {
+      "epoch": 0.7561384288200038,
+      "grad_norm": 0.0037016426026286998,
+      "learning_rate": 2.715359828141783e-06,
+      "loss": 0.0,
+      "step": 11733
+    },
+    {
+      "epoch": 0.756202874266933,
+      "grad_norm": 0.05622580946870829,
+      "learning_rate": 2.7146437522377373e-06,
+      "loss": 0.0016,
+      "step": 11734
+    },
+    {
+      "epoch": 0.7562673197138622,
+      "grad_norm": 0.016572480640073047,
+      "learning_rate": 2.713927676333691e-06,
+      "loss": 0.0001,
+      "step": 11735
+    },
+    {
+      "epoch": 0.7563317651607914,
+      "grad_norm": 0.011137781346677419,
+      "learning_rate": 2.7132116004296455e-06,
+      "loss": 0.0,
+      "step": 11736
+    },
+    {
+      "epoch": 0.7563962106077206,
+      "grad_norm": 0.05387597953175312,
+      "learning_rate": 2.7124955245256e-06,
+      "loss": 0.0017,
+      "step": 11737
+    },
+    {
+      "epoch": 0.7564606560546497,
+      "grad_norm": 0.007463667476841625,
+      "learning_rate": 2.7117794486215545e-06,
+      "loss": 0.0,
+      "step": 11738
+    },
+    {
+      "epoch": 0.7565251015015789,
+      "grad_norm": 0.006396773303127624,
+      "learning_rate": 2.7110633727175084e-06,
+      "loss": 0.0,
+      "step": 11739
+    },
+    {
+      "epoch": 0.7565895469485081,
+      "grad_norm": 0.21658050634270629,
+      "learning_rate": 2.7103472968134627e-06,
+      "loss": 0.0003,
+      "step": 11740
+    },
+    {
+      "epoch": 0.7566539923954373,
+      "grad_norm": 0.009636629385455547,
+      "learning_rate": 2.7096312209094165e-06,
+      "loss": 0.0,
+      "step": 11741
+    },
+    {
+      "epoch": 0.7567184378423665,
+      "grad_norm": 0.003825152018230289,
+      "learning_rate": 2.708915145005371e-06,
+      "loss": 0.0,
+      "step": 11742
+    },
+    {
+      "epoch": 0.7567828832892957,
+      "grad_norm": 0.003434779293330581,
+      "learning_rate": 2.7081990691013247e-06,
+      "loss": 0.0,
+      "step": 11743
+    },
+    {
+      "epoch": 0.7568473287362247,
+      "grad_norm": 0.005043666369474453,
+      "learning_rate": 2.707482993197279e-06,
+      "loss": 0.0,
+      "step": 11744
+    },
+    {
+      "epoch": 0.7569117741831539,
+      "grad_norm": 0.20329336378052815,
+      "learning_rate": 2.7067669172932333e-06,
+      "loss": 0.0007,
+      "step": 11745
+    },
+    {
+      "epoch": 0.7569762196300831,
+      "grad_norm": 0.08628995642796942,
+      "learning_rate": 2.706050841389187e-06,
+      "loss": 0.0001,
+      "step": 11746
+    },
+    {
+      "epoch": 0.7570406650770123,
+      "grad_norm": 0.0038641870067376086,
+      "learning_rate": 2.7053347654851415e-06,
+      "loss": 0.0,
+      "step": 11747
+    },
+    {
+      "epoch": 0.7571051105239415,
+      "grad_norm": 0.00018371592263419093,
+      "learning_rate": 2.704618689581096e-06,
+      "loss": 0.0,
+      "step": 11748
+    },
+    {
+      "epoch": 0.7571695559708707,
+      "grad_norm": 0.15424518081402985,
+      "learning_rate": 2.70390261367705e-06,
+      "loss": 0.002,
+      "step": 11749
+    },
+    {
+      "epoch": 0.7572340014177998,
+      "grad_norm": 0.013526881003292895,
+      "learning_rate": 2.7031865377730044e-06,
+      "loss": 0.0,
+      "step": 11750
+    },
+    {
+      "epoch": 0.757298446864729,
+      "grad_norm": 0.12452717105498468,
+      "learning_rate": 2.7024704618689582e-06,
+      "loss": 0.0019,
+      "step": 11751
+    },
+    {
+      "epoch": 0.7573628923116582,
+      "grad_norm": 0.003727796039911149,
+      "learning_rate": 2.7017543859649125e-06,
+      "loss": 0.0,
+      "step": 11752
+    },
+    {
+      "epoch": 0.7574273377585874,
+      "grad_norm": 0.0013507419381989824,
+      "learning_rate": 2.701038310060867e-06,
+      "loss": 0.0,
+      "step": 11753
+    },
+    {
+      "epoch": 0.7574917832055166,
+      "grad_norm": 0.003996855090206607,
+      "learning_rate": 2.7003222341568207e-06,
+      "loss": 0.0,
+      "step": 11754
+    },
+    {
+      "epoch": 0.7575562286524457,
+      "grad_norm": 0.1571776864987052,
+      "learning_rate": 2.699606158252775e-06,
+      "loss": 0.0003,
+      "step": 11755
+    },
+    {
+      "epoch": 0.7576206740993748,
+      "grad_norm": 0.012616492852407405,
+      "learning_rate": 2.698890082348729e-06,
+      "loss": 0.0,
+      "step": 11756
+    },
+    {
+      "epoch": 0.757685119546304,
+      "grad_norm": 0.0265738536170251,
+      "learning_rate": 2.698174006444683e-06,
+      "loss": 0.0,
+      "step": 11757
+    },
+    {
+      "epoch": 0.7577495649932332,
+      "grad_norm": 0.001051202791457292,
+      "learning_rate": 2.6974579305406375e-06,
+      "loss": 0.0,
+      "step": 11758
+    },
+    {
+      "epoch": 0.7578140104401624,
+      "grad_norm": 0.009863241518408,
+      "learning_rate": 2.6967418546365914e-06,
+      "loss": 0.0001,
+      "step": 11759
+    },
+    {
+      "epoch": 0.7578784558870916,
+      "grad_norm": 0.00017777405807486947,
+      "learning_rate": 2.696025778732546e-06,
+      "loss": 0.0,
+      "step": 11760
+    },
+    {
+      "epoch": 0.7579429013340208,
+      "grad_norm": 0.0215673841155166,
+      "learning_rate": 2.6953097028285004e-06,
+      "loss": 0.0,
+      "step": 11761
+    },
+    {
+      "epoch": 0.7580073467809499,
+      "grad_norm": 0.00022594751348228815,
+      "learning_rate": 2.6945936269244543e-06,
+      "loss": 0.0,
+      "step": 11762
+    },
+    {
+      "epoch": 0.7580717922278791,
+      "grad_norm": 0.03331891898553955,
+      "learning_rate": 2.6938775510204086e-06,
+      "loss": 0.0001,
+      "step": 11763
+    },
+    {
+      "epoch": 0.7581362376748083,
+      "grad_norm": 0.001087133402002619,
+      "learning_rate": 2.6931614751163624e-06,
+      "loss": 0.0,
+      "step": 11764
+    },
+    {
+      "epoch": 0.7582006831217375,
+      "grad_norm": 1.1937595826408476,
+      "learning_rate": 2.6924453992123167e-06,
+      "loss": 0.0228,
+      "step": 11765
+    },
+    {
+      "epoch": 0.7582651285686666,
+      "grad_norm": 2.6517257757050547,
+      "learning_rate": 2.691729323308271e-06,
+      "loss": 0.0089,
+      "step": 11766
+    },
+    {
+      "epoch": 0.7583295740155958,
+      "grad_norm": 0.0009098672745745888,
+      "learning_rate": 2.691013247404225e-06,
+      "loss": 0.0,
+      "step": 11767
+    },
+    {
+      "epoch": 0.7583940194625249,
+      "grad_norm": 0.0005929477012532038,
+      "learning_rate": 2.690297171500179e-06,
+      "loss": 0.0,
+      "step": 11768
+    },
+    {
+      "epoch": 0.7584584649094541,
+      "grad_norm": 0.04149749930801474,
+      "learning_rate": 2.689581095596133e-06,
+      "loss": 0.0,
+      "step": 11769
+    },
+    {
+      "epoch": 0.7585229103563833,
+      "grad_norm": 0.0042939977993353496,
+      "learning_rate": 2.6888650196920874e-06,
+      "loss": 0.0,
+      "step": 11770
+    },
+    {
+      "epoch": 0.7585873558033125,
+      "grad_norm": 0.000777963915366081,
+      "learning_rate": 2.6881489437880417e-06,
+      "loss": 0.0,
+      "step": 11771
+    },
+    {
+      "epoch": 0.7586518012502417,
+      "grad_norm": 0.0033128294424841676,
+      "learning_rate": 2.687432867883996e-06,
+      "loss": 0.0,
+      "step": 11772
+    },
+    {
+      "epoch": 0.7587162466971709,
+      "grad_norm": 0.0009968778887012326,
+      "learning_rate": 2.6867167919799503e-06,
+      "loss": 0.0,
+      "step": 11773
+    },
+    {
+      "epoch": 0.7587806921441,
+      "grad_norm": 0.008245827496948636,
+      "learning_rate": 2.6860007160759046e-06,
+      "loss": 0.0001,
+      "step": 11774
+    },
+    {
+      "epoch": 0.7588451375910292,
+      "grad_norm": 0.006805220256750272,
+      "learning_rate": 2.6852846401718585e-06,
+      "loss": 0.0001,
+      "step": 11775
+    },
+    {
+      "epoch": 0.7589095830379584,
+      "grad_norm": 0.006761178031938303,
+      "learning_rate": 2.6845685642678127e-06,
+      "loss": 0.0,
+      "step": 11776
+    },
+    {
+      "epoch": 0.7589740284848876,
+      "grad_norm": 0.24914090247152243,
+      "learning_rate": 2.6838524883637666e-06,
+      "loss": 0.0019,
+      "step": 11777
+    },
+    {
+      "epoch": 0.7590384739318167,
+      "grad_norm": 0.0001379510648379672,
+      "learning_rate": 2.683136412459721e-06,
+      "loss": 0.0,
+      "step": 11778
+    },
+    {
+      "epoch": 0.7591029193787459,
+      "grad_norm": 0.49096497583264964,
+      "learning_rate": 2.6824203365556752e-06,
+      "loss": 0.0012,
+      "step": 11779
+    },
+    {
+      "epoch": 0.759167364825675,
+      "grad_norm": 0.010198572628584911,
+      "learning_rate": 2.681704260651629e-06,
+      "loss": 0.0,
+      "step": 11780
+    },
+    {
+      "epoch": 0.7592318102726042,
+      "grad_norm": 0.02135744720818544,
+      "learning_rate": 2.6809881847475834e-06,
+      "loss": 0.0,
+      "step": 11781
+    },
+    {
+      "epoch": 0.7592962557195334,
+      "grad_norm": 0.0610014097381115,
+      "learning_rate": 2.6802721088435373e-06,
+      "loss": 0.0001,
+      "step": 11782
+    },
+    {
+      "epoch": 0.7593607011664626,
+      "grad_norm": 0.003880145932573114,
+      "learning_rate": 2.679556032939492e-06,
+      "loss": 0.0,
+      "step": 11783
+    },
+    {
+      "epoch": 0.7594251466133918,
+      "grad_norm": 0.00020370923103518696,
+      "learning_rate": 2.6788399570354463e-06,
+      "loss": 0.0,
+      "step": 11784
+    },
+    {
+      "epoch": 0.759489592060321,
+      "grad_norm": 0.0014196499306884838,
+      "learning_rate": 2.6781238811314e-06,
+      "loss": 0.0,
+      "step": 11785
+    },
+    {
+      "epoch": 0.7595540375072501,
+      "grad_norm": 0.0004107186318848994,
+      "learning_rate": 2.6774078052273545e-06,
+      "loss": 0.0,
+      "step": 11786
+    },
+    {
+      "epoch": 0.7596184829541793,
+      "grad_norm": 0.004399499841082977,
+      "learning_rate": 2.6766917293233088e-06,
+      "loss": 0.0,
+      "step": 11787
+    },
+    {
+      "epoch": 0.7596829284011085,
+      "grad_norm": 0.043867588185750386,
+      "learning_rate": 2.6759756534192626e-06,
+      "loss": 0.0004,
+      "step": 11788
+    },
+    {
+      "epoch": 0.7597473738480376,
+      "grad_norm": 0.0009829067833408381,
+      "learning_rate": 2.675259577515217e-06,
+      "loss": 0.0,
+      "step": 11789
+    },
+    {
+      "epoch": 0.7598118192949668,
+      "grad_norm": 0.23802523069820278,
+      "learning_rate": 2.674543501611171e-06,
+      "loss": 0.001,
+      "step": 11790
+    },
+    {
+      "epoch": 0.759876264741896,
+      "grad_norm": 0.0033577163553118712,
+      "learning_rate": 2.673827425707125e-06,
+      "loss": 0.0,
+      "step": 11791
+    },
+    {
+      "epoch": 0.7599407101888251,
+      "grad_norm": 0.02319084706000951,
+      "learning_rate": 2.673111349803079e-06,
+      "loss": 0.0001,
+      "step": 11792
+    },
+    {
+      "epoch": 0.7600051556357543,
+      "grad_norm": 0.056277099181222555,
+      "learning_rate": 2.6723952738990333e-06,
+      "loss": 0.0001,
+      "step": 11793
+    },
+    {
+      "epoch": 0.7600696010826835,
+      "grad_norm": 0.00015177896495238968,
+      "learning_rate": 2.6716791979949876e-06,
+      "loss": 0.0,
+      "step": 11794
+    },
+    {
+      "epoch": 0.7601340465296127,
+      "grad_norm": 0.0015311593093557537,
+      "learning_rate": 2.6709631220909423e-06,
+      "loss": 0.0,
+      "step": 11795
+    },
+    {
+      "epoch": 0.7601984919765419,
+      "grad_norm": 0.021320862762892245,
+      "learning_rate": 2.670247046186896e-06,
+      "loss": 0.0,
+      "step": 11796
+    },
+    {
+      "epoch": 0.7602629374234711,
+      "grad_norm": 0.031283424056328626,
+      "learning_rate": 2.6695309702828505e-06,
+      "loss": 0.0001,
+      "step": 11797
+    },
+    {
+      "epoch": 0.7603273828704002,
+      "grad_norm": 0.23894887948035884,
+      "learning_rate": 2.6688148943788044e-06,
+      "loss": 0.0002,
+      "step": 11798
+    },
+    {
+      "epoch": 0.7603918283173294,
+      "grad_norm": 0.04260379057087094,
+      "learning_rate": 2.6680988184747587e-06,
+      "loss": 0.0003,
+      "step": 11799
+    },
+    {
+      "epoch": 0.7604562737642585,
+      "grad_norm": 9.578836850924028e-05,
+      "learning_rate": 2.6673827425707125e-06,
+      "loss": 0.0,
+      "step": 11800
+    },
+    {
+      "epoch": 0.7605207192111877,
+      "grad_norm": 0.006114623869051579,
+      "learning_rate": 2.666666666666667e-06,
+      "loss": 0.0,
+      "step": 11801
+    },
+    {
+      "epoch": 0.7605851646581169,
+      "grad_norm": 0.0006305918796649802,
+      "learning_rate": 2.665950590762621e-06,
+      "loss": 0.0,
+      "step": 11802
+    },
+    {
+      "epoch": 0.7606496101050461,
+      "grad_norm": 0.0018584589838389813,
+      "learning_rate": 2.665234514858575e-06,
+      "loss": 0.0,
+      "step": 11803
+    },
+    {
+      "epoch": 0.7607140555519752,
+      "grad_norm": 0.0934540428233906,
+      "learning_rate": 2.6645184389545293e-06,
+      "loss": 0.0003,
+      "step": 11804
+    },
+    {
+      "epoch": 0.7607785009989044,
+      "grad_norm": 0.017028167671529763,
+      "learning_rate": 2.663802363050483e-06,
+      "loss": 0.0,
+      "step": 11805
+    },
+    {
+      "epoch": 0.7608429464458336,
+      "grad_norm": 0.3192495760082157,
+      "learning_rate": 2.6630862871464375e-06,
+      "loss": 0.004,
+      "step": 11806
+    },
+    {
+      "epoch": 0.7609073918927628,
+      "grad_norm": 0.000859402811223285,
+      "learning_rate": 2.662370211242392e-06,
+      "loss": 0.0,
+      "step": 11807
+    },
+    {
+      "epoch": 0.760971837339692,
+      "grad_norm": 0.001708086676881105,
+      "learning_rate": 2.661654135338346e-06,
+      "loss": 0.0,
+      "step": 11808
+    },
+    {
+      "epoch": 0.7610362827866212,
+      "grad_norm": 0.0002548697259070124,
+      "learning_rate": 2.6609380594343004e-06,
+      "loss": 0.0,
+      "step": 11809
+    },
+    {
+      "epoch": 0.7611007282335503,
+      "grad_norm": 0.032245765729999896,
+      "learning_rate": 2.6602219835302547e-06,
+      "loss": 0.0001,
+      "step": 11810
+    },
+    {
+      "epoch": 0.7611651736804794,
+      "grad_norm": 0.07445310934893193,
+      "learning_rate": 2.6595059076262085e-06,
+      "loss": 0.0001,
+      "step": 11811
+    },
+    {
+      "epoch": 0.7612296191274086,
+      "grad_norm": 0.17762862192819528,
+      "learning_rate": 2.658789831722163e-06,
+      "loss": 0.0005,
+      "step": 11812
+    },
+    {
+      "epoch": 0.7612940645743378,
+      "grad_norm": 0.011117645753348048,
+      "learning_rate": 2.6580737558181167e-06,
+      "loss": 0.0001,
+      "step": 11813
+    },
+    {
+      "epoch": 0.761358510021267,
+      "grad_norm": 0.34508473103897397,
+      "learning_rate": 2.657357679914071e-06,
+      "loss": 0.0005,
+      "step": 11814
+    },
+    {
+      "epoch": 0.7614229554681962,
+      "grad_norm": 0.1738423116659773,
+      "learning_rate": 2.6566416040100253e-06,
+      "loss": 0.0006,
+      "step": 11815
+    },
+    {
+      "epoch": 0.7614874009151253,
+      "grad_norm": 0.005119695379919854,
+      "learning_rate": 2.655925528105979e-06,
+      "loss": 0.0,
+      "step": 11816
+    },
+    {
+      "epoch": 0.7615518463620545,
+      "grad_norm": 1.4379237835705299,
+      "learning_rate": 2.6552094522019335e-06,
+      "loss": 0.0012,
+      "step": 11817
+    },
+    {
+      "epoch": 0.7616162918089837,
+      "grad_norm": 0.0022575272986593542,
+      "learning_rate": 2.6544933762978882e-06,
+      "loss": 0.0,
+      "step": 11818
+    },
+    {
+      "epoch": 0.7616807372559129,
+      "grad_norm": 0.5371539565571194,
+      "learning_rate": 2.653777300393842e-06,
+      "loss": 0.0042,
+      "step": 11819
+    },
+    {
+      "epoch": 0.7617451827028421,
+      "grad_norm": 0.00024939560620323515,
+      "learning_rate": 2.6530612244897964e-06,
+      "loss": 0.0,
+      "step": 11820
+    },
+    {
+      "epoch": 0.7618096281497713,
+      "grad_norm": 0.0006033900395677962,
+      "learning_rate": 2.6523451485857503e-06,
+      "loss": 0.0,
+      "step": 11821
+    },
+    {
+      "epoch": 0.7618740735967003,
+      "grad_norm": 8.636202424418066e-05,
+      "learning_rate": 2.6516290726817046e-06,
+      "loss": 0.0,
+      "step": 11822
+    },
+    {
+      "epoch": 0.7619385190436295,
+      "grad_norm": 0.00012699654655637103,
+      "learning_rate": 2.650912996777659e-06,
+      "loss": 0.0,
+      "step": 11823
+    },
+    {
+      "epoch": 0.7620029644905587,
+      "grad_norm": 0.001349256260770123,
+      "learning_rate": 2.6501969208736127e-06,
+      "loss": 0.0,
+      "step": 11824
+    },
+    {
+      "epoch": 0.7620674099374879,
+      "grad_norm": 0.00020679210237969375,
+      "learning_rate": 2.649480844969567e-06,
+      "loss": 0.0,
+      "step": 11825
+    },
+    {
+      "epoch": 0.7621318553844171,
+      "grad_norm": 0.002953117774268297,
+      "learning_rate": 2.648764769065521e-06,
+      "loss": 0.0,
+      "step": 11826
+    },
+    {
+      "epoch": 0.7621963008313463,
+      "grad_norm": 0.003053642823270433,
+      "learning_rate": 2.648048693161475e-06,
+      "loss": 0.0,
+      "step": 11827
+    },
+    {
+      "epoch": 0.7622607462782754,
+      "grad_norm": 0.01158112876131613,
+      "learning_rate": 2.6473326172574295e-06,
+      "loss": 0.0,
+      "step": 11828
+    },
+    {
+      "epoch": 0.7623251917252046,
+      "grad_norm": 0.00046034852405999084,
+      "learning_rate": 2.6466165413533834e-06,
+      "loss": 0.0,
+      "step": 11829
+    },
+    {
+      "epoch": 0.7623896371721338,
+      "grad_norm": 0.0006405258341291647,
+      "learning_rate": 2.645900465449338e-06,
+      "loss": 0.0,
+      "step": 11830
+    },
+    {
+      "epoch": 0.762454082619063,
+      "grad_norm": 0.0016408110038443101,
+      "learning_rate": 2.6451843895452924e-06,
+      "loss": 0.0,
+      "step": 11831
+    },
+    {
+      "epoch": 0.7625185280659922,
+      "grad_norm": 0.001069109636870012,
+      "learning_rate": 2.6444683136412463e-06,
+      "loss": 0.0,
+      "step": 11832
+    },
+    {
+      "epoch": 0.7625829735129213,
+      "grad_norm": 0.015143400205832954,
+      "learning_rate": 2.6437522377372006e-06,
+      "loss": 0.0,
+      "step": 11833
+    },
+    {
+      "epoch": 0.7626474189598504,
+      "grad_norm": 0.0010125760276858443,
+      "learning_rate": 2.6430361618331545e-06,
+      "loss": 0.0,
+      "step": 11834
+    },
+    {
+      "epoch": 0.7627118644067796,
+      "grad_norm": 0.022163290000592962,
+      "learning_rate": 2.6423200859291087e-06,
+      "loss": 0.0016,
+      "step": 11835
+    },
+    {
+      "epoch": 0.7627763098537088,
+      "grad_norm": 0.7956127941337039,
+      "learning_rate": 2.641604010025063e-06,
+      "loss": 0.0027,
+      "step": 11836
+    },
+    {
+      "epoch": 0.762840755300638,
+      "grad_norm": 0.16291320241804333,
+      "learning_rate": 2.640887934121017e-06,
+      "loss": 0.0019,
+      "step": 11837
+    },
+    {
+      "epoch": 0.7629052007475672,
+      "grad_norm": 0.00014900202124249368,
+      "learning_rate": 2.6401718582169712e-06,
+      "loss": 0.0,
+      "step": 11838
+    },
+    {
+      "epoch": 0.7629696461944964,
+      "grad_norm": 0.0017136274959623572,
+      "learning_rate": 2.639455782312925e-06,
+      "loss": 0.0,
+      "step": 11839
+    },
+    {
+      "epoch": 0.7630340916414255,
+      "grad_norm": 0.05000114029835755,
+      "learning_rate": 2.6387397064088794e-06,
+      "loss": 0.0,
+      "step": 11840
+    },
+    {
+      "epoch": 0.7630985370883547,
+      "grad_norm": 7.336247695839558e-05,
+      "learning_rate": 2.6380236305048333e-06,
+      "loss": 0.0,
+      "step": 11841
+    },
+    {
+      "epoch": 0.7631629825352839,
+      "grad_norm": 0.6437919769674801,
+      "learning_rate": 2.637307554600788e-06,
+      "loss": 0.0017,
+      "step": 11842
+    },
+    {
+      "epoch": 0.7632274279822131,
+      "grad_norm": 0.009178616250083763,
+      "learning_rate": 2.6365914786967423e-06,
+      "loss": 0.0,
+      "step": 11843
+    },
+    {
+      "epoch": 0.7632918734291422,
+      "grad_norm": 0.0001536315240165565,
+      "learning_rate": 2.6358754027926966e-06,
+      "loss": 0.0,
+      "step": 11844
+    },
+    {
+      "epoch": 0.7633563188760714,
+      "grad_norm": 0.04400854197836655,
+      "learning_rate": 2.6351593268886505e-06,
+      "loss": 0.0,
+      "step": 11845
+    },
+    {
+      "epoch": 0.7634207643230005,
+      "grad_norm": 0.05531086042366422,
+      "learning_rate": 2.6344432509846048e-06,
+      "loss": 0.0001,
+      "step": 11846
+    },
+    {
+      "epoch": 0.7634852097699297,
+      "grad_norm": 0.00016720311908218334,
+      "learning_rate": 2.6337271750805586e-06,
+      "loss": 0.0,
+      "step": 11847
+    },
+    {
+      "epoch": 0.7635496552168589,
+      "grad_norm": 0.0006551267521307725,
+      "learning_rate": 2.633011099176513e-06,
+      "loss": 0.0,
+      "step": 11848
+    },
+    {
+      "epoch": 0.7636141006637881,
+      "grad_norm": 6.0717022634338646e-05,
+      "learning_rate": 2.632295023272467e-06,
+      "loss": 0.0,
+      "step": 11849
+    },
+    {
+      "epoch": 0.7636785461107173,
+      "grad_norm": 0.010353891275486336,
+      "learning_rate": 2.631578947368421e-06,
+      "loss": 0.0,
+      "step": 11850
+    },
+    {
+      "epoch": 0.7637429915576465,
+      "grad_norm": 0.0015369640711080093,
+      "learning_rate": 2.6308628714643754e-06,
+      "loss": 0.0,
+      "step": 11851
+    },
+    {
+      "epoch": 0.7638074370045757,
+      "grad_norm": 0.0003941849795398379,
+      "learning_rate": 2.6301467955603293e-06,
+      "loss": 0.0,
+      "step": 11852
+    },
+    {
+      "epoch": 0.7638718824515048,
+      "grad_norm": 0.0003463927405450879,
+      "learning_rate": 2.629430719656284e-06,
+      "loss": 0.0,
+      "step": 11853
+    },
+    {
+      "epoch": 0.763936327898434,
+      "grad_norm": 0.0008114464169782369,
+      "learning_rate": 2.6287146437522383e-06,
+      "loss": 0.0,
+      "step": 11854
+    },
+    {
+      "epoch": 0.7640007733453632,
+      "grad_norm": 6.755891019105164e-05,
+      "learning_rate": 2.627998567848192e-06,
+      "loss": 0.0,
+      "step": 11855
+    },
+    {
+      "epoch": 0.7640652187922923,
+      "grad_norm": 0.0011958914319152606,
+      "learning_rate": 2.6272824919441465e-06,
+      "loss": 0.0,
+      "step": 11856
+    },
+    {
+      "epoch": 0.7641296642392215,
+      "grad_norm": 0.0038164852243763306,
+      "learning_rate": 2.6265664160401004e-06,
+      "loss": 0.0,
+      "step": 11857
+    },
+    {
+      "epoch": 0.7641941096861506,
+      "grad_norm": 0.00030386703955184003,
+      "learning_rate": 2.6258503401360547e-06,
+      "loss": 0.0,
+      "step": 11858
+    },
+    {
+      "epoch": 0.7642585551330798,
+      "grad_norm": 0.00043240885358736416,
+      "learning_rate": 2.625134264232009e-06,
+      "loss": 0.0,
+      "step": 11859
+    },
+    {
+      "epoch": 0.764323000580009,
+      "grad_norm": 0.002957734448481022,
+      "learning_rate": 2.624418188327963e-06,
+      "loss": 0.0,
+      "step": 11860
+    },
+    {
+      "epoch": 0.7643874460269382,
+      "grad_norm": 0.0002885876430917981,
+      "learning_rate": 2.623702112423917e-06,
+      "loss": 0.0,
+      "step": 11861
+    },
+    {
+      "epoch": 0.7644518914738674,
+      "grad_norm": 0.005890731307086754,
+      "learning_rate": 2.622986036519871e-06,
+      "loss": 0.0,
+      "step": 11862
+    },
+    {
+      "epoch": 0.7645163369207966,
+      "grad_norm": 6.651252772496995e-05,
+      "learning_rate": 2.6222699606158253e-06,
+      "loss": 0.0,
+      "step": 11863
+    },
+    {
+      "epoch": 0.7645807823677258,
+      "grad_norm": 0.09255987482417974,
+      "learning_rate": 2.6215538847117796e-06,
+      "loss": 0.0017,
+      "step": 11864
+    },
+    {
+      "epoch": 0.7646452278146549,
+      "grad_norm": 0.0030729239305532265,
+      "learning_rate": 2.620837808807734e-06,
+      "loss": 0.0,
+      "step": 11865
+    },
+    {
+      "epoch": 0.7647096732615841,
+      "grad_norm": 0.0004729470478401344,
+      "learning_rate": 2.620121732903688e-06,
+      "loss": 0.0,
+      "step": 11866
+    },
+    {
+      "epoch": 0.7647741187085132,
+      "grad_norm": 0.00031197574419735606,
+      "learning_rate": 2.6194056569996425e-06,
+      "loss": 0.0,
+      "step": 11867
+    },
+    {
+      "epoch": 0.7648385641554424,
+      "grad_norm": 0.554805689295591,
+      "learning_rate": 2.6186895810955964e-06,
+      "loss": 0.0024,
+      "step": 11868
+    },
+    {
+      "epoch": 0.7649030096023716,
+      "grad_norm": 0.0005885395825914211,
+      "learning_rate": 2.6179735051915507e-06,
+      "loss": 0.0,
+      "step": 11869
+    },
+    {
+      "epoch": 0.7649674550493007,
+      "grad_norm": 0.0012661613214064538,
+      "learning_rate": 2.6172574292875045e-06,
+      "loss": 0.0,
+      "step": 11870
+    },
+    {
+      "epoch": 0.7650319004962299,
+      "grad_norm": 8.82429252709678e-05,
+      "learning_rate": 2.616541353383459e-06,
+      "loss": 0.0,
+      "step": 11871
+    },
+    {
+      "epoch": 0.7650963459431591,
+      "grad_norm": 0.1790998115578762,
+      "learning_rate": 2.615825277479413e-06,
+      "loss": 0.0002,
+      "step": 11872
+    },
+    {
+      "epoch": 0.7651607913900883,
+      "grad_norm": 0.08301635025098202,
+      "learning_rate": 2.615109201575367e-06,
+      "loss": 0.0006,
+      "step": 11873
+    },
+    {
+      "epoch": 0.7652252368370175,
+      "grad_norm": 0.0008072109844447568,
+      "learning_rate": 2.6143931256713213e-06,
+      "loss": 0.0,
+      "step": 11874
+    },
+    {
+      "epoch": 0.7652896822839467,
+      "grad_norm": 5.068903277494801e-05,
+      "learning_rate": 2.613677049767275e-06,
+      "loss": 0.0,
+      "step": 11875
+    },
+    {
+      "epoch": 0.7653541277308759,
+      "grad_norm": 0.003124247831575227,
+      "learning_rate": 2.6129609738632295e-06,
+      "loss": 0.0,
+      "step": 11876
+    },
+    {
+      "epoch": 0.765418573177805,
+      "grad_norm": 0.00019207895262184278,
+      "learning_rate": 2.6122448979591842e-06,
+      "loss": 0.0,
+      "step": 11877
+    },
+    {
+      "epoch": 0.7654830186247341,
+      "grad_norm": 0.2378545285856134,
+      "learning_rate": 2.611528822055138e-06,
+      "loss": 0.0004,
+      "step": 11878
+    },
+    {
+      "epoch": 0.7655474640716633,
+      "grad_norm": 0.008514530768873717,
+      "learning_rate": 2.6108127461510924e-06,
+      "loss": 0.0,
+      "step": 11879
+    },
+    {
+      "epoch": 0.7656119095185925,
+      "grad_norm": 0.00046250907666416433,
+      "learning_rate": 2.6100966702470467e-06,
+      "loss": 0.0,
+      "step": 11880
+    },
+    {
+      "epoch": 0.7656763549655217,
+      "grad_norm": 0.13777567500132695,
+      "learning_rate": 2.6093805943430006e-06,
+      "loss": 0.0023,
+      "step": 11881
+    },
+    {
+      "epoch": 0.7657408004124509,
+      "grad_norm": 0.0003629261576732833,
+      "learning_rate": 2.608664518438955e-06,
+      "loss": 0.0,
+      "step": 11882
+    },
+    {
+      "epoch": 0.76580524585938,
+      "grad_norm": 0.0009412332498742197,
+      "learning_rate": 2.6079484425349087e-06,
+      "loss": 0.0,
+      "step": 11883
+    },
+    {
+      "epoch": 0.7658696913063092,
+      "grad_norm": 0.00012120714432166618,
+      "learning_rate": 2.607232366630863e-06,
+      "loss": 0.0,
+      "step": 11884
+    },
+    {
+      "epoch": 0.7659341367532384,
+      "grad_norm": 0.0012524471309115777,
+      "learning_rate": 2.606516290726817e-06,
+      "loss": 0.0,
+      "step": 11885
+    },
+    {
+      "epoch": 0.7659985822001676,
+      "grad_norm": 0.0004439158991068107,
+      "learning_rate": 2.605800214822771e-06,
+      "loss": 0.0,
+      "step": 11886
+    },
+    {
+      "epoch": 0.7660630276470968,
+      "grad_norm": 0.014125048239822585,
+      "learning_rate": 2.6050841389187255e-06,
+      "loss": 0.0,
+      "step": 11887
+    },
+    {
+      "epoch": 0.766127473094026,
+      "grad_norm": 0.01923080340759084,
+      "learning_rate": 2.6043680630146802e-06,
+      "loss": 0.0,
+      "step": 11888
+    },
+    {
+      "epoch": 0.766191918540955,
+      "grad_norm": 0.00011607705395213511,
+      "learning_rate": 2.603651987110634e-06,
+      "loss": 0.0,
+      "step": 11889
+    },
+    {
+      "epoch": 0.7662563639878842,
+      "grad_norm": 0.0006579140278466091,
+      "learning_rate": 2.6029359112065884e-06,
+      "loss": 0.0,
+      "step": 11890
+    },
+    {
+      "epoch": 0.7663208094348134,
+      "grad_norm": 0.0002124806354313648,
+      "learning_rate": 2.6022198353025423e-06,
+      "loss": 0.0,
+      "step": 11891
+    },
+    {
+      "epoch": 0.7663852548817426,
+      "grad_norm": 0.0013674333354162045,
+      "learning_rate": 2.6015037593984966e-06,
+      "loss": 0.0,
+      "step": 11892
+    },
+    {
+      "epoch": 0.7664497003286718,
+      "grad_norm": 0.014675526109410085,
+      "learning_rate": 2.6007876834944505e-06,
+      "loss": 0.0001,
+      "step": 11893
+    },
+    {
+      "epoch": 0.766514145775601,
+      "grad_norm": 0.03573676645358599,
+      "learning_rate": 2.6000716075904047e-06,
+      "loss": 0.0001,
+      "step": 11894
+    },
+    {
+      "epoch": 0.7665785912225301,
+      "grad_norm": 0.0008220368009597546,
+      "learning_rate": 2.599355531686359e-06,
+      "loss": 0.0,
+      "step": 11895
+    },
+    {
+      "epoch": 0.7666430366694593,
+      "grad_norm": 0.021377558030884054,
+      "learning_rate": 2.598639455782313e-06,
+      "loss": 0.0,
+      "step": 11896
+    },
+    {
+      "epoch": 0.7667074821163885,
+      "grad_norm": 0.01966843926829134,
+      "learning_rate": 2.5979233798782672e-06,
+      "loss": 0.0001,
+      "step": 11897
+    },
+    {
+      "epoch": 0.7667719275633177,
+      "grad_norm": 0.00019038251652734773,
+      "learning_rate": 2.597207303974221e-06,
+      "loss": 0.0,
+      "step": 11898
+    },
+    {
+      "epoch": 0.7668363730102469,
+      "grad_norm": 0.002648309532818556,
+      "learning_rate": 2.5964912280701754e-06,
+      "loss": 0.0,
+      "step": 11899
+    },
+    {
+      "epoch": 0.766900818457176,
+      "grad_norm": 0.0001730522603491945,
+      "learning_rate": 2.59577515216613e-06,
+      "loss": 0.0,
+      "step": 11900
+    },
+    {
+      "epoch": 0.7669652639041051,
+      "grad_norm": 1.6432124179162014e-05,
+      "learning_rate": 2.595059076262084e-06,
+      "loss": 0.0,
+      "step": 11901
+    },
+    {
+      "epoch": 0.7670297093510343,
+      "grad_norm": 1.7618660333607035e-05,
+      "learning_rate": 2.5943430003580383e-06,
+      "loss": 0.0,
+      "step": 11902
+    },
+    {
+      "epoch": 0.7670941547979635,
+      "grad_norm": 0.03666116270808565,
+      "learning_rate": 2.5936269244539926e-06,
+      "loss": 0.0,
+      "step": 11903
+    },
+    {
+      "epoch": 0.7671586002448927,
+      "grad_norm": 0.11028554089182246,
+      "learning_rate": 2.5929108485499465e-06,
+      "loss": 0.0002,
+      "step": 11904
+    },
+    {
+      "epoch": 0.7672230456918219,
+      "grad_norm": 0.22051480791536138,
+      "learning_rate": 2.5921947726459008e-06,
+      "loss": 0.0013,
+      "step": 11905
+    },
+    {
+      "epoch": 0.767287491138751,
+      "grad_norm": 0.006515029704666616,
+      "learning_rate": 2.5914786967418546e-06,
+      "loss": 0.0,
+      "step": 11906
+    },
+    {
+      "epoch": 0.7673519365856802,
+      "grad_norm": 0.006650815831030154,
+      "learning_rate": 2.590762620837809e-06,
+      "loss": 0.0,
+      "step": 11907
+    },
+    {
+      "epoch": 0.7674163820326094,
+      "grad_norm": 0.0051438193964023025,
+      "learning_rate": 2.5900465449337632e-06,
+      "loss": 0.0,
+      "step": 11908
+    },
+    {
+      "epoch": 0.7674808274795386,
+      "grad_norm": 0.4898605593241484,
+      "learning_rate": 2.589330469029717e-06,
+      "loss": 0.0023,
+      "step": 11909
+    },
+    {
+      "epoch": 0.7675452729264678,
+      "grad_norm": 0.0006288678826158701,
+      "learning_rate": 2.5886143931256714e-06,
+      "loss": 0.0,
+      "step": 11910
+    },
+    {
+      "epoch": 0.7676097183733969,
+      "grad_norm": 0.003386373418255314,
+      "learning_rate": 2.5878983172216253e-06,
+      "loss": 0.0,
+      "step": 11911
+    },
+    {
+      "epoch": 0.767674163820326,
+      "grad_norm": 0.002668168384645223,
+      "learning_rate": 2.58718224131758e-06,
+      "loss": 0.0,
+      "step": 11912
+    },
+    {
+      "epoch": 0.7677386092672552,
+      "grad_norm": 0.001082914875932513,
+      "learning_rate": 2.5864661654135343e-06,
+      "loss": 0.0,
+      "step": 11913
+    },
+    {
+      "epoch": 0.7678030547141844,
+      "grad_norm": 0.06223376012771815,
+      "learning_rate": 2.585750089509488e-06,
+      "loss": 0.0003,
+      "step": 11914
+    },
+    {
+      "epoch": 0.7678675001611136,
+      "grad_norm": 0.31937400054402,
+      "learning_rate": 2.5850340136054425e-06,
+      "loss": 0.0019,
+      "step": 11915
+    },
+    {
+      "epoch": 0.7679319456080428,
+      "grad_norm": 0.018148016411417802,
+      "learning_rate": 2.5843179377013968e-06,
+      "loss": 0.0001,
+      "step": 11916
+    },
+    {
+      "epoch": 0.767996391054972,
+      "grad_norm": 0.0038383521751228934,
+      "learning_rate": 2.5836018617973507e-06,
+      "loss": 0.0,
+      "step": 11917
+    },
+    {
+      "epoch": 0.7680608365019012,
+      "grad_norm": 0.1600973730490817,
+      "learning_rate": 2.582885785893305e-06,
+      "loss": 0.0012,
+      "step": 11918
+    },
+    {
+      "epoch": 0.7681252819488303,
+      "grad_norm": 0.0460683108904585,
+      "learning_rate": 2.582169709989259e-06,
+      "loss": 0.0001,
+      "step": 11919
+    },
+    {
+      "epoch": 0.7681897273957595,
+      "grad_norm": 0.0016058862460527552,
+      "learning_rate": 2.581453634085213e-06,
+      "loss": 0.0,
+      "step": 11920
+    },
+    {
+      "epoch": 0.7682541728426887,
+      "grad_norm": 0.002115022812249171,
+      "learning_rate": 2.5807375581811674e-06,
+      "loss": 0.0,
+      "step": 11921
+    },
+    {
+      "epoch": 0.7683186182896179,
+      "grad_norm": 0.004045692377456239,
+      "learning_rate": 2.5800214822771213e-06,
+      "loss": 0.0,
+      "step": 11922
+    },
+    {
+      "epoch": 0.768383063736547,
+      "grad_norm": 0.0005268503339750655,
+      "learning_rate": 2.579305406373076e-06,
+      "loss": 0.0,
+      "step": 11923
+    },
+    {
+      "epoch": 0.7684475091834762,
+      "grad_norm": 0.0009562755514328496,
+      "learning_rate": 2.5785893304690303e-06,
+      "loss": 0.0,
+      "step": 11924
+    },
+    {
+      "epoch": 0.7685119546304053,
+      "grad_norm": 0.21633146350250407,
+      "learning_rate": 2.577873254564984e-06,
+      "loss": 0.0003,
+      "step": 11925
+    },
+    {
+      "epoch": 0.7685764000773345,
+      "grad_norm": 0.0018381836193263551,
+      "learning_rate": 2.5771571786609385e-06,
+      "loss": 0.0,
+      "step": 11926
+    },
+    {
+      "epoch": 0.7686408455242637,
+      "grad_norm": 0.007529660661730795,
+      "learning_rate": 2.5764411027568924e-06,
+      "loss": 0.0001,
+      "step": 11927
+    },
+    {
+      "epoch": 0.7687052909711929,
+      "grad_norm": 0.0025654502952239984,
+      "learning_rate": 2.5757250268528467e-06,
+      "loss": 0.0,
+      "step": 11928
+    },
+    {
+      "epoch": 0.7687697364181221,
+      "grad_norm": 0.0316317751952979,
+      "learning_rate": 2.575008950948801e-06,
+      "loss": 0.0002,
+      "step": 11929
+    },
+    {
+      "epoch": 0.7688341818650513,
+      "grad_norm": 0.1274536097738183,
+      "learning_rate": 2.574292875044755e-06,
+      "loss": 0.0004,
+      "step": 11930
+    },
+    {
+      "epoch": 0.7688986273119804,
+      "grad_norm": 0.0013855358441066388,
+      "learning_rate": 2.573576799140709e-06,
+      "loss": 0.0,
+      "step": 11931
+    },
+    {
+      "epoch": 0.7689630727589096,
+      "grad_norm": 0.03422381413667952,
+      "learning_rate": 2.572860723236663e-06,
+      "loss": 0.0001,
+      "step": 11932
+    },
+    {
+      "epoch": 0.7690275182058388,
+      "grad_norm": 0.00036604118401955133,
+      "learning_rate": 2.5721446473326173e-06,
+      "loss": 0.0,
+      "step": 11933
+    },
+    {
+      "epoch": 0.7690919636527679,
+      "grad_norm": 0.007119691489192297,
+      "learning_rate": 2.571428571428571e-06,
+      "loss": 0.0,
+      "step": 11934
+    },
+    {
+      "epoch": 0.7691564090996971,
+      "grad_norm": 0.0007796408203194449,
+      "learning_rate": 2.570712495524526e-06,
+      "loss": 0.0,
+      "step": 11935
+    },
+    {
+      "epoch": 0.7692208545466263,
+      "grad_norm": 0.0046414742114352535,
+      "learning_rate": 2.5699964196204802e-06,
+      "loss": 0.0,
+      "step": 11936
+    },
+    {
+      "epoch": 0.7692852999935554,
+      "grad_norm": 0.0026985613912551645,
+      "learning_rate": 2.5692803437164345e-06,
+      "loss": 0.0,
+      "step": 11937
+    },
+    {
+      "epoch": 0.7693497454404846,
+      "grad_norm": 0.015896688905246856,
+      "learning_rate": 2.5685642678123884e-06,
+      "loss": 0.0001,
+      "step": 11938
+    },
+    {
+      "epoch": 0.7694141908874138,
+      "grad_norm": 0.002225209735481927,
+      "learning_rate": 2.5678481919083427e-06,
+      "loss": 0.0,
+      "step": 11939
+    },
+    {
+      "epoch": 0.769478636334343,
+      "grad_norm": 0.000544120900602758,
+      "learning_rate": 2.5671321160042966e-06,
+      "loss": 0.0,
+      "step": 11940
+    },
+    {
+      "epoch": 0.7695430817812722,
+      "grad_norm": 0.000987951875979495,
+      "learning_rate": 2.566416040100251e-06,
+      "loss": 0.0,
+      "step": 11941
+    },
+    {
+      "epoch": 0.7696075272282014,
+      "grad_norm": 0.2139565355227848,
+      "learning_rate": 2.5656999641962047e-06,
+      "loss": 0.0011,
+      "step": 11942
+    },
+    {
+      "epoch": 0.7696719726751305,
+      "grad_norm": 0.0014181229277454124,
+      "learning_rate": 2.564983888292159e-06,
+      "loss": 0.0,
+      "step": 11943
+    },
+    {
+      "epoch": 0.7697364181220597,
+      "grad_norm": 0.0011511183313510068,
+      "learning_rate": 2.5642678123881133e-06,
+      "loss": 0.0,
+      "step": 11944
+    },
+    {
+      "epoch": 0.7698008635689888,
+      "grad_norm": 0.021332500647752366,
+      "learning_rate": 2.563551736484067e-06,
+      "loss": 0.0,
+      "step": 11945
+    },
+    {
+      "epoch": 0.769865309015918,
+      "grad_norm": 0.0006443705067154068,
+      "learning_rate": 2.5628356605800215e-06,
+      "loss": 0.0,
+      "step": 11946
+    },
+    {
+      "epoch": 0.7699297544628472,
+      "grad_norm": 0.018157607163298645,
+      "learning_rate": 2.5621195846759762e-06,
+      "loss": 0.0,
+      "step": 11947
+    },
+    {
+      "epoch": 0.7699941999097764,
+      "grad_norm": 0.0006373038105464806,
+      "learning_rate": 2.56140350877193e-06,
+      "loss": 0.0,
+      "step": 11948
+    },
+    {
+      "epoch": 0.7700586453567055,
+      "grad_norm": 0.1702311649693865,
+      "learning_rate": 2.5606874328678844e-06,
+      "loss": 0.0007,
+      "step": 11949
+    },
+    {
+      "epoch": 0.7701230908036347,
+      "grad_norm": 0.006839913357284363,
+      "learning_rate": 2.5599713569638383e-06,
+      "loss": 0.0,
+      "step": 11950
+    },
+    {
+      "epoch": 0.7701875362505639,
+      "grad_norm": 0.020449848798393098,
+      "learning_rate": 2.5592552810597926e-06,
+      "loss": 0.0,
+      "step": 11951
+    },
+    {
+      "epoch": 0.7702519816974931,
+      "grad_norm": 0.0186318491828164,
+      "learning_rate": 2.558539205155747e-06,
+      "loss": 0.0,
+      "step": 11952
+    },
+    {
+      "epoch": 0.7703164271444223,
+      "grad_norm": 0.012205018831311388,
+      "learning_rate": 2.5578231292517007e-06,
+      "loss": 0.0,
+      "step": 11953
+    },
+    {
+      "epoch": 0.7703808725913515,
+      "grad_norm": 0.014573322368295935,
+      "learning_rate": 2.557107053347655e-06,
+      "loss": 0.0001,
+      "step": 11954
+    },
+    {
+      "epoch": 0.7704453180382806,
+      "grad_norm": 0.3025764545657545,
+      "learning_rate": 2.556390977443609e-06,
+      "loss": 0.0009,
+      "step": 11955
+    },
+    {
+      "epoch": 0.7705097634852097,
+      "grad_norm": 0.003638116650934389,
+      "learning_rate": 2.5556749015395632e-06,
+      "loss": 0.0,
+      "step": 11956
+    },
+    {
+      "epoch": 0.7705742089321389,
+      "grad_norm": 0.005688319311506136,
+      "learning_rate": 2.5549588256355175e-06,
+      "loss": 0.0,
+      "step": 11957
+    },
+    {
+      "epoch": 0.7706386543790681,
+      "grad_norm": 0.00278424510278735,
+      "learning_rate": 2.554242749731472e-06,
+      "loss": 0.0,
+      "step": 11958
+    },
+    {
+      "epoch": 0.7707030998259973,
+      "grad_norm": 0.01585624886430006,
+      "learning_rate": 2.553526673827426e-06,
+      "loss": 0.0,
+      "step": 11959
+    },
+    {
+      "epoch": 0.7707675452729265,
+      "grad_norm": 0.0003486352458013042,
+      "learning_rate": 2.5528105979233804e-06,
+      "loss": 0.0,
+      "step": 11960
+    },
+    {
+      "epoch": 0.7708319907198556,
+      "grad_norm": 0.013720875341040538,
+      "learning_rate": 2.5520945220193343e-06,
+      "loss": 0.0,
+      "step": 11961
+    },
+    {
+      "epoch": 0.7708964361667848,
+      "grad_norm": 0.0011977463656281952,
+      "learning_rate": 2.5513784461152886e-06,
+      "loss": 0.0,
+      "step": 11962
+    },
+    {
+      "epoch": 0.770960881613714,
+      "grad_norm": 0.0006557547530678418,
+      "learning_rate": 2.5506623702112425e-06,
+      "loss": 0.0,
+      "step": 11963
+    },
+    {
+      "epoch": 0.7710253270606432,
+      "grad_norm": 7.640697653079052e-05,
+      "learning_rate": 2.5499462943071968e-06,
+      "loss": 0.0,
+      "step": 11964
+    },
+    {
+      "epoch": 0.7710897725075724,
+      "grad_norm": 0.0005535518755850447,
+      "learning_rate": 2.549230218403151e-06,
+      "loss": 0.0,
+      "step": 11965
+    },
+    {
+      "epoch": 0.7711542179545016,
+      "grad_norm": 0.014170017898537866,
+      "learning_rate": 2.548514142499105e-06,
+      "loss": 0.0,
+      "step": 11966
+    },
+    {
+      "epoch": 0.7712186634014306,
+      "grad_norm": 0.025082221368806706,
+      "learning_rate": 2.5477980665950592e-06,
+      "loss": 0.0,
+      "step": 11967
+    },
+    {
+      "epoch": 0.7712831088483598,
+      "grad_norm": 0.06604501701956178,
+      "learning_rate": 2.547081990691013e-06,
+      "loss": 0.0001,
+      "step": 11968
+    },
+    {
+      "epoch": 0.771347554295289,
+      "grad_norm": 0.5134007423695695,
+      "learning_rate": 2.5463659147869674e-06,
+      "loss": 0.0026,
+      "step": 11969
+    },
+    {
+      "epoch": 0.7714119997422182,
+      "grad_norm": 9.906826757261537e-05,
+      "learning_rate": 2.545649838882922e-06,
+      "loss": 0.0,
+      "step": 11970
+    },
+    {
+      "epoch": 0.7714764451891474,
+      "grad_norm": 0.0019939027732718065,
+      "learning_rate": 2.544933762978876e-06,
+      "loss": 0.0,
+      "step": 11971
+    },
+    {
+      "epoch": 0.7715408906360766,
+      "grad_norm": 0.0009953462564640401,
+      "learning_rate": 2.5442176870748303e-06,
+      "loss": 0.0,
+      "step": 11972
+    },
+    {
+      "epoch": 0.7716053360830057,
+      "grad_norm": 0.00022765940038362478,
+      "learning_rate": 2.5435016111707846e-06,
+      "loss": 0.0,
+      "step": 11973
+    },
+    {
+      "epoch": 0.7716697815299349,
+      "grad_norm": 0.0034598503765954076,
+      "learning_rate": 2.5427855352667385e-06,
+      "loss": 0.0,
+      "step": 11974
+    },
+    {
+      "epoch": 0.7717342269768641,
+      "grad_norm": 0.0015874933110641958,
+      "learning_rate": 2.5420694593626928e-06,
+      "loss": 0.0,
+      "step": 11975
+    },
+    {
+      "epoch": 0.7717986724237933,
+      "grad_norm": 0.19490331702864275,
+      "learning_rate": 2.5413533834586467e-06,
+      "loss": 0.0002,
+      "step": 11976
+    },
+    {
+      "epoch": 0.7718631178707225,
+      "grad_norm": 0.046350962259317584,
+      "learning_rate": 2.540637307554601e-06,
+      "loss": 0.0002,
+      "step": 11977
+    },
+    {
+      "epoch": 0.7719275633176516,
+      "grad_norm": 0.0029893966150884525,
+      "learning_rate": 2.5399212316505553e-06,
+      "loss": 0.0,
+      "step": 11978
+    },
+    {
+      "epoch": 0.7719920087645807,
+      "grad_norm": 0.006386238741863867,
+      "learning_rate": 2.539205155746509e-06,
+      "loss": 0.0,
+      "step": 11979
+    },
+    {
+      "epoch": 0.7720564542115099,
+      "grad_norm": 0.08005569888313953,
+      "learning_rate": 2.5384890798424634e-06,
+      "loss": 0.0001,
+      "step": 11980
+    },
+    {
+      "epoch": 0.7721208996584391,
+      "grad_norm": 0.00017401220544152146,
+      "learning_rate": 2.5377730039384173e-06,
+      "loss": 0.0,
+      "step": 11981
+    },
+    {
+      "epoch": 0.7721853451053683,
+      "grad_norm": 0.00920950918241963,
+      "learning_rate": 2.537056928034372e-06,
+      "loss": 0.0001,
+      "step": 11982
+    },
+    {
+      "epoch": 0.7722497905522975,
+      "grad_norm": 4.005209131921557e-05,
+      "learning_rate": 2.5363408521303263e-06,
+      "loss": 0.0,
+      "step": 11983
+    },
+    {
+      "epoch": 0.7723142359992267,
+      "grad_norm": 0.0041205538081291486,
+      "learning_rate": 2.53562477622628e-06,
+      "loss": 0.0,
+      "step": 11984
+    },
+    {
+      "epoch": 0.7723786814461558,
+      "grad_norm": 0.21781045231342352,
+      "learning_rate": 2.5349087003222345e-06,
+      "loss": 0.0006,
+      "step": 11985
+    },
+    {
+      "epoch": 0.772443126893085,
+      "grad_norm": 0.4514950507189504,
+      "learning_rate": 2.534192624418189e-06,
+      "loss": 0.0019,
+      "step": 11986
+    },
+    {
+      "epoch": 0.7725075723400142,
+      "grad_norm": 0.0022551838848633467,
+      "learning_rate": 2.5334765485141427e-06,
+      "loss": 0.0,
+      "step": 11987
+    },
+    {
+      "epoch": 0.7725720177869434,
+      "grad_norm": 0.03850665996937269,
+      "learning_rate": 2.532760472610097e-06,
+      "loss": 0.0,
+      "step": 11988
+    },
+    {
+      "epoch": 0.7726364632338725,
+      "grad_norm": 0.003876283490998469,
+      "learning_rate": 2.532044396706051e-06,
+      "loss": 0.0,
+      "step": 11989
+    },
+    {
+      "epoch": 0.7727009086808017,
+      "grad_norm": 0.001072320260391634,
+      "learning_rate": 2.531328320802005e-06,
+      "loss": 0.0,
+      "step": 11990
+    },
+    {
+      "epoch": 0.7727653541277308,
+      "grad_norm": 0.0027809967217183487,
+      "learning_rate": 2.530612244897959e-06,
+      "loss": 0.0,
+      "step": 11991
+    },
+    {
+      "epoch": 0.77282979957466,
+      "grad_norm": 8.871064129087865e-05,
+      "learning_rate": 2.5298961689939133e-06,
+      "loss": 0.0,
+      "step": 11992
+    },
+    {
+      "epoch": 0.7728942450215892,
+      "grad_norm": 0.004071476358003262,
+      "learning_rate": 2.529180093089868e-06,
+      "loss": 0.0,
+      "step": 11993
+    },
+    {
+      "epoch": 0.7729586904685184,
+      "grad_norm": 0.0002506879102149424,
+      "learning_rate": 2.5284640171858223e-06,
+      "loss": 0.0,
+      "step": 11994
+    },
+    {
+      "epoch": 0.7730231359154476,
+      "grad_norm": 0.0063177136721583385,
+      "learning_rate": 2.5277479412817762e-06,
+      "loss": 0.0,
+      "step": 11995
+    },
+    {
+      "epoch": 0.7730875813623768,
+      "grad_norm": 0.00446458534063843,
+      "learning_rate": 2.5270318653777305e-06,
+      "loss": 0.0,
+      "step": 11996
+    },
+    {
+      "epoch": 0.773152026809306,
+      "grad_norm": 0.0008947799086842924,
+      "learning_rate": 2.5263157894736844e-06,
+      "loss": 0.0,
+      "step": 11997
+    },
+    {
+      "epoch": 0.7732164722562351,
+      "grad_norm": 0.0004906066965026694,
+      "learning_rate": 2.5255997135696387e-06,
+      "loss": 0.0,
+      "step": 11998
+    },
+    {
+      "epoch": 0.7732809177031643,
+      "grad_norm": 0.0006771847353210726,
+      "learning_rate": 2.5248836376655926e-06,
+      "loss": 0.0,
+      "step": 11999
+    },
+    {
+      "epoch": 0.7733453631500935,
+      "grad_norm": 0.03204525715449161,
+      "learning_rate": 2.524167561761547e-06,
+      "loss": 0.0,
+      "step": 12000
+    },
+    {
+      "epoch": 0.7734098085970226,
+      "grad_norm": 0.17269605440999874,
+      "learning_rate": 2.523451485857501e-06,
+      "loss": 0.0004,
+      "step": 12001
+    },
+    {
+      "epoch": 0.7734742540439518,
+      "grad_norm": 0.0002331402822554568,
+      "learning_rate": 2.522735409953455e-06,
+      "loss": 0.0,
+      "step": 12002
+    },
+    {
+      "epoch": 0.7735386994908809,
+      "grad_norm": 0.0018706894203856107,
+      "learning_rate": 2.5220193340494093e-06,
+      "loss": 0.0,
+      "step": 12003
+    },
+    {
+      "epoch": 0.7736031449378101,
+      "grad_norm": 0.23577475587510444,
+      "learning_rate": 2.521303258145363e-06,
+      "loss": 0.003,
+      "step": 12004
+    },
+    {
+      "epoch": 0.7736675903847393,
+      "grad_norm": 0.09998982366549632,
+      "learning_rate": 2.520587182241318e-06,
+      "loss": 0.0002,
+      "step": 12005
+    },
+    {
+      "epoch": 0.7737320358316685,
+      "grad_norm": 0.00016423824356981455,
+      "learning_rate": 2.5198711063372722e-06,
+      "loss": 0.0,
+      "step": 12006
+    },
+    {
+      "epoch": 0.7737964812785977,
+      "grad_norm": 0.03779216557272751,
+      "learning_rate": 2.519155030433226e-06,
+      "loss": 0.0,
+      "step": 12007
+    },
+    {
+      "epoch": 0.7738609267255269,
+      "grad_norm": 0.0012791941326287728,
+      "learning_rate": 2.5184389545291804e-06,
+      "loss": 0.0,
+      "step": 12008
+    },
+    {
+      "epoch": 0.773925372172456,
+      "grad_norm": 0.012356869073728675,
+      "learning_rate": 2.5177228786251347e-06,
+      "loss": 0.0,
+      "step": 12009
+    },
+    {
+      "epoch": 0.7739898176193852,
+      "grad_norm": 0.4994760987231003,
+      "learning_rate": 2.5170068027210886e-06,
+      "loss": 0.0029,
+      "step": 12010
+    },
+    {
+      "epoch": 0.7740542630663144,
+      "grad_norm": 0.06908226621424454,
+      "learning_rate": 2.516290726817043e-06,
+      "loss": 0.0016,
+      "step": 12011
+    },
+    {
+      "epoch": 0.7741187085132435,
+      "grad_norm": 0.28788372637119103,
+      "learning_rate": 2.5155746509129967e-06,
+      "loss": 0.0004,
+      "step": 12012
+    },
+    {
+      "epoch": 0.7741831539601727,
+      "grad_norm": 0.0007493520973933247,
+      "learning_rate": 2.514858575008951e-06,
+      "loss": 0.0,
+      "step": 12013
+    },
+    {
+      "epoch": 0.7742475994071019,
+      "grad_norm": 0.2166681766839472,
+      "learning_rate": 2.5141424991049053e-06,
+      "loss": 0.0003,
+      "step": 12014
+    },
+    {
+      "epoch": 0.774312044854031,
+      "grad_norm": 0.01656720079967917,
+      "learning_rate": 2.5134264232008592e-06,
+      "loss": 0.0,
+      "step": 12015
+    },
+    {
+      "epoch": 0.7743764903009602,
+      "grad_norm": 0.028964880535246422,
+      "learning_rate": 2.5127103472968135e-06,
+      "loss": 0.0,
+      "step": 12016
+    },
+    {
+      "epoch": 0.7744409357478894,
+      "grad_norm": 0.01668855371166636,
+      "learning_rate": 2.5119942713927682e-06,
+      "loss": 0.0,
+      "step": 12017
+    },
+    {
+      "epoch": 0.7745053811948186,
+      "grad_norm": 0.00826921975490535,
+      "learning_rate": 2.511278195488722e-06,
+      "loss": 0.0,
+      "step": 12018
+    },
+    {
+      "epoch": 0.7745698266417478,
+      "grad_norm": 0.0019967600267275804,
+      "learning_rate": 2.5105621195846764e-06,
+      "loss": 0.0,
+      "step": 12019
+    },
+    {
+      "epoch": 0.774634272088677,
+      "grad_norm": 0.1981025072169954,
+      "learning_rate": 2.5098460436806303e-06,
+      "loss": 0.001,
+      "step": 12020
+    },
+    {
+      "epoch": 0.7746987175356062,
+      "grad_norm": 0.44295811245102745,
+      "learning_rate": 2.5091299677765846e-06,
+      "loss": 0.0034,
+      "step": 12021
+    },
+    {
+      "epoch": 0.7747631629825353,
+      "grad_norm": 0.0007132173739187416,
+      "learning_rate": 2.508413891872539e-06,
+      "loss": 0.0,
+      "step": 12022
+    },
+    {
+      "epoch": 0.7748276084294644,
+      "grad_norm": 0.0025241999101835977,
+      "learning_rate": 2.5076978159684928e-06,
+      "loss": 0.0,
+      "step": 12023
+    },
+    {
+      "epoch": 0.7748920538763936,
+      "grad_norm": 0.00020836271478278545,
+      "learning_rate": 2.506981740064447e-06,
+      "loss": 0.0,
+      "step": 12024
+    },
+    {
+      "epoch": 0.7749564993233228,
+      "grad_norm": 0.004587700629129224,
+      "learning_rate": 2.506265664160401e-06,
+      "loss": 0.0,
+      "step": 12025
+    },
+    {
+      "epoch": 0.775020944770252,
+      "grad_norm": 0.017800457394732166,
+      "learning_rate": 2.5055495882563552e-06,
+      "loss": 0.0,
+      "step": 12026
+    },
+    {
+      "epoch": 0.7750853902171811,
+      "grad_norm": 0.005903831600595229,
+      "learning_rate": 2.504833512352309e-06,
+      "loss": 0.0,
+      "step": 12027
+    },
+    {
+      "epoch": 0.7751498356641103,
+      "grad_norm": 0.0019456858051113458,
+      "learning_rate": 2.504117436448264e-06,
+      "loss": 0.0,
+      "step": 12028
+    },
+    {
+      "epoch": 0.7752142811110395,
+      "grad_norm": 0.008377258458590922,
+      "learning_rate": 2.503401360544218e-06,
+      "loss": 0.0,
+      "step": 12029
+    },
+    {
+      "epoch": 0.7752787265579687,
+      "grad_norm": 0.01561660979057207,
+      "learning_rate": 2.5026852846401724e-06,
+      "loss": 0.0,
+      "step": 12030
+    },
+    {
+      "epoch": 0.7753431720048979,
+      "grad_norm": 0.02577598930821863,
+      "learning_rate": 2.5019692087361263e-06,
+      "loss": 0.0,
+      "step": 12031
+    },
+    {
+      "epoch": 0.7754076174518271,
+      "grad_norm": 0.0010712492468354091,
+      "learning_rate": 2.5012531328320806e-06,
+      "loss": 0.0,
+      "step": 12032
+    },
+    {
+      "epoch": 0.7754720628987563,
+      "grad_norm": 0.0944316984142256,
+      "learning_rate": 2.5005370569280345e-06,
+      "loss": 0.0001,
+      "step": 12033
+    },
+    {
+      "epoch": 0.7755365083456853,
+      "grad_norm": 0.0232625944343488,
+      "learning_rate": 2.4998209810239888e-06,
+      "loss": 0.0003,
+      "step": 12034
+    },
+    {
+      "epoch": 0.7756009537926145,
+      "grad_norm": 1.4485049070539173,
+      "learning_rate": 2.4991049051199427e-06,
+      "loss": 0.0025,
+      "step": 12035
+    },
+    {
+      "epoch": 0.7756653992395437,
+      "grad_norm": 0.0016307701840922474,
+      "learning_rate": 2.498388829215897e-06,
+      "loss": 0.0,
+      "step": 12036
+    },
+    {
+      "epoch": 0.7757298446864729,
+      "grad_norm": 0.02336175794591136,
+      "learning_rate": 2.4976727533118513e-06,
+      "loss": 0.0001,
+      "step": 12037
+    },
+    {
+      "epoch": 0.7757942901334021,
+      "grad_norm": 0.16227938458248187,
+      "learning_rate": 2.4969566774078056e-06,
+      "loss": 0.0022,
+      "step": 12038
+    },
+    {
+      "epoch": 0.7758587355803312,
+      "grad_norm": 0.07302846880666206,
+      "learning_rate": 2.4962406015037594e-06,
+      "loss": 0.0004,
+      "step": 12039
+    },
+    {
+      "epoch": 0.7759231810272604,
+      "grad_norm": 0.00041019317192437306,
+      "learning_rate": 2.4955245255997137e-06,
+      "loss": 0.0,
+      "step": 12040
+    },
+    {
+      "epoch": 0.7759876264741896,
+      "grad_norm": 0.002584801277987377,
+      "learning_rate": 2.494808449695668e-06,
+      "loss": 0.0,
+      "step": 12041
+    },
+    {
+      "epoch": 0.7760520719211188,
+      "grad_norm": 0.002223323087132319,
+      "learning_rate": 2.494092373791622e-06,
+      "loss": 0.0,
+      "step": 12042
+    },
+    {
+      "epoch": 0.776116517368048,
+      "grad_norm": 0.01440653692624194,
+      "learning_rate": 2.493376297887576e-06,
+      "loss": 0.0,
+      "step": 12043
+    },
+    {
+      "epoch": 0.7761809628149772,
+      "grad_norm": 0.053300064123326024,
+      "learning_rate": 2.4926602219835305e-06,
+      "loss": 0.0001,
+      "step": 12044
+    },
+    {
+      "epoch": 0.7762454082619062,
+      "grad_norm": 0.20566420595865056,
+      "learning_rate": 2.491944146079485e-06,
+      "loss": 0.0023,
+      "step": 12045
+    },
+    {
+      "epoch": 0.7763098537088354,
+      "grad_norm": 0.0007623805293543374,
+      "learning_rate": 2.4912280701754387e-06,
+      "loss": 0.0,
+      "step": 12046
+    },
+    {
+      "epoch": 0.7763742991557646,
+      "grad_norm": 0.0014681434347622946,
+      "learning_rate": 2.490511994271393e-06,
+      "loss": 0.0,
+      "step": 12047
+    },
+    {
+      "epoch": 0.7764387446026938,
+      "grad_norm": 0.00043488636991413735,
+      "learning_rate": 2.489795918367347e-06,
+      "loss": 0.0,
+      "step": 12048
+    },
+    {
+      "epoch": 0.776503190049623,
+      "grad_norm": 0.0017831550431723615,
+      "learning_rate": 2.4890798424633016e-06,
+      "loss": 0.0,
+      "step": 12049
+    },
+    {
+      "epoch": 0.7765676354965522,
+      "grad_norm": 0.0018114956405963806,
+      "learning_rate": 2.4883637665592554e-06,
+      "loss": 0.0,
+      "step": 12050
+    },
+    {
+      "epoch": 0.7766320809434814,
+      "grad_norm": 0.18950622836851014,
+      "learning_rate": 2.4876476906552097e-06,
+      "loss": 0.0002,
+      "step": 12051
+    },
+    {
+      "epoch": 0.7766965263904105,
+      "grad_norm": 0.00418238718089347,
+      "learning_rate": 2.4869316147511636e-06,
+      "loss": 0.0,
+      "step": 12052
+    },
+    {
+      "epoch": 0.7767609718373397,
+      "grad_norm": 0.0004478533245452791,
+      "learning_rate": 2.486215538847118e-06,
+      "loss": 0.0,
+      "step": 12053
+    },
+    {
+      "epoch": 0.7768254172842689,
+      "grad_norm": 0.00024396889565855113,
+      "learning_rate": 2.4854994629430722e-06,
+      "loss": 0.0,
+      "step": 12054
+    },
+    {
+      "epoch": 0.7768898627311981,
+      "grad_norm": 0.2903369953181515,
+      "learning_rate": 2.4847833870390265e-06,
+      "loss": 0.0008,
+      "step": 12055
+    },
+    {
+      "epoch": 0.7769543081781272,
+      "grad_norm": 0.00020092059146728687,
+      "learning_rate": 2.4840673111349804e-06,
+      "loss": 0.0,
+      "step": 12056
+    },
+    {
+      "epoch": 0.7770187536250563,
+      "grad_norm": 0.0011666064461578786,
+      "learning_rate": 2.4833512352309347e-06,
+      "loss": 0.0,
+      "step": 12057
+    },
+    {
+      "epoch": 0.7770831990719855,
+      "grad_norm": 0.001190503348124305,
+      "learning_rate": 2.482635159326889e-06,
+      "loss": 0.0,
+      "step": 12058
+    },
+    {
+      "epoch": 0.7771476445189147,
+      "grad_norm": 0.3657997162209834,
+      "learning_rate": 2.481919083422843e-06,
+      "loss": 0.0032,
+      "step": 12059
+    },
+    {
+      "epoch": 0.7772120899658439,
+      "grad_norm": 0.001634408520878064,
+      "learning_rate": 2.481203007518797e-06,
+      "loss": 0.0,
+      "step": 12060
+    },
+    {
+      "epoch": 0.7772765354127731,
+      "grad_norm": 0.1469337056910139,
+      "learning_rate": 2.4804869316147515e-06,
+      "loss": 0.0017,
+      "step": 12061
+    },
+    {
+      "epoch": 0.7773409808597023,
+      "grad_norm": 0.06710843879393069,
+      "learning_rate": 2.4797708557107058e-06,
+      "loss": 0.0002,
+      "step": 12062
+    },
+    {
+      "epoch": 0.7774054263066315,
+      "grad_norm": 0.0017892333181064397,
+      "learning_rate": 2.4790547798066596e-06,
+      "loss": 0.0,
+      "step": 12063
+    },
+    {
+      "epoch": 0.7774698717535606,
+      "grad_norm": 0.004690414808615758,
+      "learning_rate": 2.478338703902614e-06,
+      "loss": 0.0,
+      "step": 12064
+    },
+    {
+      "epoch": 0.7775343172004898,
+      "grad_norm": 0.004454153225457359,
+      "learning_rate": 2.477622627998568e-06,
+      "loss": 0.0,
+      "step": 12065
+    },
+    {
+      "epoch": 0.777598762647419,
+      "grad_norm": 0.31551044717576315,
+      "learning_rate": 2.4769065520945225e-06,
+      "loss": 0.0011,
+      "step": 12066
+    },
+    {
+      "epoch": 0.7776632080943481,
+      "grad_norm": 0.0014824010634681267,
+      "learning_rate": 2.4761904761904764e-06,
+      "loss": 0.0,
+      "step": 12067
+    },
+    {
+      "epoch": 0.7777276535412773,
+      "grad_norm": 0.005493271680943317,
+      "learning_rate": 2.4754744002864307e-06,
+      "loss": 0.0,
+      "step": 12068
+    },
+    {
+      "epoch": 0.7777920989882064,
+      "grad_norm": 0.002344001587170874,
+      "learning_rate": 2.4747583243823846e-06,
+      "loss": 0.0,
+      "step": 12069
+    },
+    {
+      "epoch": 0.7778565444351356,
+      "grad_norm": 0.00015707887304980444,
+      "learning_rate": 2.474042248478339e-06,
+      "loss": 0.0,
+      "step": 12070
+    },
+    {
+      "epoch": 0.7779209898820648,
+      "grad_norm": 0.0006823565350090584,
+      "learning_rate": 2.473326172574293e-06,
+      "loss": 0.0,
+      "step": 12071
+    },
+    {
+      "epoch": 0.777985435328994,
+      "grad_norm": 0.0017209300771120863,
+      "learning_rate": 2.4726100966702475e-06,
+      "loss": 0.0,
+      "step": 12072
+    },
+    {
+      "epoch": 0.7780498807759232,
+      "grad_norm": 0.2138218417932704,
+      "learning_rate": 2.4718940207662013e-06,
+      "loss": 0.0006,
+      "step": 12073
+    },
+    {
+      "epoch": 0.7781143262228524,
+      "grad_norm": 0.0065011778796621855,
+      "learning_rate": 2.4711779448621556e-06,
+      "loss": 0.0,
+      "step": 12074
+    },
+    {
+      "epoch": 0.7781787716697816,
+      "grad_norm": 0.0003251508710584543,
+      "learning_rate": 2.47046186895811e-06,
+      "loss": 0.0,
+      "step": 12075
+    },
+    {
+      "epoch": 0.7782432171167107,
+      "grad_norm": 0.0010564877939732219,
+      "learning_rate": 2.469745793054064e-06,
+      "loss": 0.0,
+      "step": 12076
+    },
+    {
+      "epoch": 0.7783076625636399,
+      "grad_norm": 0.0001559384697446135,
+      "learning_rate": 2.469029717150018e-06,
+      "loss": 0.0,
+      "step": 12077
+    },
+    {
+      "epoch": 0.7783721080105691,
+      "grad_norm": 0.0007897786642414037,
+      "learning_rate": 2.4683136412459724e-06,
+      "loss": 0.0,
+      "step": 12078
+    },
+    {
+      "epoch": 0.7784365534574982,
+      "grad_norm": 0.008367158669417396,
+      "learning_rate": 2.4675975653419267e-06,
+      "loss": 0.0,
+      "step": 12079
+    },
+    {
+      "epoch": 0.7785009989044274,
+      "grad_norm": 0.0024458826459670666,
+      "learning_rate": 2.4668814894378806e-06,
+      "loss": 0.0,
+      "step": 12080
+    },
+    {
+      "epoch": 0.7785654443513566,
+      "grad_norm": 0.02535862235484841,
+      "learning_rate": 2.466165413533835e-06,
+      "loss": 0.0001,
+      "step": 12081
+    },
+    {
+      "epoch": 0.7786298897982857,
+      "grad_norm": 0.0008567730125366892,
+      "learning_rate": 2.4654493376297888e-06,
+      "loss": 0.0,
+      "step": 12082
+    },
+    {
+      "epoch": 0.7786943352452149,
+      "grad_norm": 0.0010730528697962104,
+      "learning_rate": 2.464733261725743e-06,
+      "loss": 0.0,
+      "step": 12083
+    },
+    {
+      "epoch": 0.7787587806921441,
+      "grad_norm": 0.05461918233566015,
+      "learning_rate": 2.4640171858216974e-06,
+      "loss": 0.0001,
+      "step": 12084
+    },
+    {
+      "epoch": 0.7788232261390733,
+      "grad_norm": 0.15517757863659667,
+      "learning_rate": 2.4633011099176517e-06,
+      "loss": 0.0015,
+      "step": 12085
+    },
+    {
+      "epoch": 0.7788876715860025,
+      "grad_norm": 0.1818583424467823,
+      "learning_rate": 2.4625850340136055e-06,
+      "loss": 0.0013,
+      "step": 12086
+    },
+    {
+      "epoch": 0.7789521170329317,
+      "grad_norm": 0.001713274648651585,
+      "learning_rate": 2.46186895810956e-06,
+      "loss": 0.0,
+      "step": 12087
+    },
+    {
+      "epoch": 0.7790165624798608,
+      "grad_norm": 0.14944862140584098,
+      "learning_rate": 2.4611528822055137e-06,
+      "loss": 0.0011,
+      "step": 12088
+    },
+    {
+      "epoch": 0.77908100792679,
+      "grad_norm": 0.12680037185162404,
+      "learning_rate": 2.460436806301468e-06,
+      "loss": 0.0012,
+      "step": 12089
+    },
+    {
+      "epoch": 0.7791454533737191,
+      "grad_norm": 2.549845910452418e-05,
+      "learning_rate": 2.4597207303974223e-06,
+      "loss": 0.0,
+      "step": 12090
+    },
+    {
+      "epoch": 0.7792098988206483,
+      "grad_norm": 0.000564051977485508,
+      "learning_rate": 2.4590046544933766e-06,
+      "loss": 0.0,
+      "step": 12091
+    },
+    {
+      "epoch": 0.7792743442675775,
+      "grad_norm": 0.0010766900145023287,
+      "learning_rate": 2.4582885785893305e-06,
+      "loss": 0.0,
+      "step": 12092
+    },
+    {
+      "epoch": 0.7793387897145067,
+      "grad_norm": 0.7329335370994735,
+      "learning_rate": 2.4575725026852848e-06,
+      "loss": 0.001,
+      "step": 12093
+    },
+    {
+      "epoch": 0.7794032351614358,
+      "grad_norm": 0.001353292375859486,
+      "learning_rate": 2.456856426781239e-06,
+      "loss": 0.0,
+      "step": 12094
+    },
+    {
+      "epoch": 0.779467680608365,
+      "grad_norm": 0.053991346554312696,
+      "learning_rate": 2.456140350877193e-06,
+      "loss": 0.0,
+      "step": 12095
+    },
+    {
+      "epoch": 0.7795321260552942,
+      "grad_norm": 0.005993064838952124,
+      "learning_rate": 2.4554242749731473e-06,
+      "loss": 0.0,
+      "step": 12096
+    },
+    {
+      "epoch": 0.7795965715022234,
+      "grad_norm": 0.0021416859490622545,
+      "learning_rate": 2.4547081990691016e-06,
+      "loss": 0.0,
+      "step": 12097
+    },
+    {
+      "epoch": 0.7796610169491526,
+      "grad_norm": 0.014471636175070532,
+      "learning_rate": 2.453992123165056e-06,
+      "loss": 0.0,
+      "step": 12098
+    },
+    {
+      "epoch": 0.7797254623960818,
+      "grad_norm": 0.006637386067261587,
+      "learning_rate": 2.4532760472610097e-06,
+      "loss": 0.0,
+      "step": 12099
+    },
+    {
+      "epoch": 0.7797899078430109,
+      "grad_norm": 0.01238649530879906,
+      "learning_rate": 2.452559971356964e-06,
+      "loss": 0.0001,
+      "step": 12100
+    },
+    {
+      "epoch": 0.77985435328994,
+      "grad_norm": 0.04190091815805419,
+      "learning_rate": 2.4518438954529183e-06,
+      "loss": 0.0016,
+      "step": 12101
+    },
+    {
+      "epoch": 0.7799187987368692,
+      "grad_norm": 0.00034396283472747845,
+      "learning_rate": 2.4511278195488726e-06,
+      "loss": 0.0,
+      "step": 12102
+    },
+    {
+      "epoch": 0.7799832441837984,
+      "grad_norm": 0.00787000151745771,
+      "learning_rate": 2.4504117436448265e-06,
+      "loss": 0.0,
+      "step": 12103
+    },
+    {
+      "epoch": 0.7800476896307276,
+      "grad_norm": 0.0040082247399448896,
+      "learning_rate": 2.449695667740781e-06,
+      "loss": 0.0,
+      "step": 12104
+    },
+    {
+      "epoch": 0.7801121350776568,
+      "grad_norm": 0.4142050407785845,
+      "learning_rate": 2.4489795918367347e-06,
+      "loss": 0.0009,
+      "step": 12105
+    },
+    {
+      "epoch": 0.7801765805245859,
+      "grad_norm": 0.0002606664246261931,
+      "learning_rate": 2.448263515932689e-06,
+      "loss": 0.0,
+      "step": 12106
+    },
+    {
+      "epoch": 0.7802410259715151,
+      "grad_norm": 0.003579972965774035,
+      "learning_rate": 2.4475474400286433e-06,
+      "loss": 0.0,
+      "step": 12107
+    },
+    {
+      "epoch": 0.7803054714184443,
+      "grad_norm": 0.021305318846473693,
+      "learning_rate": 2.4468313641245976e-06,
+      "loss": 0.0002,
+      "step": 12108
+    },
+    {
+      "epoch": 0.7803699168653735,
+      "grad_norm": 0.012691156115001318,
+      "learning_rate": 2.4461152882205514e-06,
+      "loss": 0.0,
+      "step": 12109
+    },
+    {
+      "epoch": 0.7804343623123027,
+      "grad_norm": 0.0017728354092780912,
+      "learning_rate": 2.4453992123165057e-06,
+      "loss": 0.0,
+      "step": 12110
+    },
+    {
+      "epoch": 0.7804988077592319,
+      "grad_norm": 0.0016533355333587835,
+      "learning_rate": 2.44468313641246e-06,
+      "loss": 0.0,
+      "step": 12111
+    },
+    {
+      "epoch": 0.7805632532061609,
+      "grad_norm": 0.0022065467282514017,
+      "learning_rate": 2.443967060508414e-06,
+      "loss": 0.0,
+      "step": 12112
+    },
+    {
+      "epoch": 0.7806276986530901,
+      "grad_norm": 0.004123723032585986,
+      "learning_rate": 2.4432509846043682e-06,
+      "loss": 0.0,
+      "step": 12113
+    },
+    {
+      "epoch": 0.7806921441000193,
+      "grad_norm": 0.0005452104333529033,
+      "learning_rate": 2.4425349087003225e-06,
+      "loss": 0.0,
+      "step": 12114
+    },
+    {
+      "epoch": 0.7807565895469485,
+      "grad_norm": 0.0005542779206685047,
+      "learning_rate": 2.441818832796277e-06,
+      "loss": 0.0,
+      "step": 12115
+    },
+    {
+      "epoch": 0.7808210349938777,
+      "grad_norm": 0.10948264410785608,
+      "learning_rate": 2.4411027568922307e-06,
+      "loss": 0.0004,
+      "step": 12116
+    },
+    {
+      "epoch": 0.7808854804408069,
+      "grad_norm": 0.0001848680504991575,
+      "learning_rate": 2.440386680988185e-06,
+      "loss": 0.0,
+      "step": 12117
+    },
+    {
+      "epoch": 0.780949925887736,
+      "grad_norm": 0.0007432499498959208,
+      "learning_rate": 2.439670605084139e-06,
+      "loss": 0.0,
+      "step": 12118
+    },
+    {
+      "epoch": 0.7810143713346652,
+      "grad_norm": 0.004389440175323089,
+      "learning_rate": 2.4389545291800936e-06,
+      "loss": 0.0,
+      "step": 12119
+    },
+    {
+      "epoch": 0.7810788167815944,
+      "grad_norm": 0.002810186785865301,
+      "learning_rate": 2.4382384532760475e-06,
+      "loss": 0.0,
+      "step": 12120
+    },
+    {
+      "epoch": 0.7811432622285236,
+      "grad_norm": 0.007075445688301599,
+      "learning_rate": 2.4375223773720018e-06,
+      "loss": 0.0,
+      "step": 12121
+    },
+    {
+      "epoch": 0.7812077076754528,
+      "grad_norm": 0.00019089657477344002,
+      "learning_rate": 2.4368063014679556e-06,
+      "loss": 0.0,
+      "step": 12122
+    },
+    {
+      "epoch": 0.7812721531223819,
+      "grad_norm": 0.002233569077508966,
+      "learning_rate": 2.43609022556391e-06,
+      "loss": 0.0,
+      "step": 12123
+    },
+    {
+      "epoch": 0.781336598569311,
+      "grad_norm": 2.4103195654440213,
+      "learning_rate": 2.4353741496598642e-06,
+      "loss": 0.0292,
+      "step": 12124
+    },
+    {
+      "epoch": 0.7814010440162402,
+      "grad_norm": 8.232393805314394e-05,
+      "learning_rate": 2.4346580737558185e-06,
+      "loss": 0.0,
+      "step": 12125
+    },
+    {
+      "epoch": 0.7814654894631694,
+      "grad_norm": 0.16571786769047236,
+      "learning_rate": 2.4339419978517724e-06,
+      "loss": 0.0003,
+      "step": 12126
+    },
+    {
+      "epoch": 0.7815299349100986,
+      "grad_norm": 0.006719157453833545,
+      "learning_rate": 2.4332259219477267e-06,
+      "loss": 0.0,
+      "step": 12127
+    },
+    {
+      "epoch": 0.7815943803570278,
+      "grad_norm": 0.004906313933490102,
+      "learning_rate": 2.432509846043681e-06,
+      "loss": 0.0,
+      "step": 12128
+    },
+    {
+      "epoch": 0.781658825803957,
+      "grad_norm": 0.006681028163229526,
+      "learning_rate": 2.431793770139635e-06,
+      "loss": 0.0,
+      "step": 12129
+    },
+    {
+      "epoch": 0.7817232712508861,
+      "grad_norm": 0.2901713907485199,
+      "learning_rate": 2.431077694235589e-06,
+      "loss": 0.0,
+      "step": 12130
+    },
+    {
+      "epoch": 0.7817877166978153,
+      "grad_norm": 0.0014225891402345502,
+      "learning_rate": 2.4303616183315435e-06,
+      "loss": 0.0,
+      "step": 12131
+    },
+    {
+      "epoch": 0.7818521621447445,
+      "grad_norm": 0.08323650832116088,
+      "learning_rate": 2.4296455424274978e-06,
+      "loss": 0.0001,
+      "step": 12132
+    },
+    {
+      "epoch": 0.7819166075916737,
+      "grad_norm": 0.003190025050149661,
+      "learning_rate": 2.4289294665234516e-06,
+      "loss": 0.0,
+      "step": 12133
+    },
+    {
+      "epoch": 0.7819810530386028,
+      "grad_norm": 0.00029784653909833324,
+      "learning_rate": 2.428213390619406e-06,
+      "loss": 0.0,
+      "step": 12134
+    },
+    {
+      "epoch": 0.782045498485532,
+      "grad_norm": 0.0007026094397731929,
+      "learning_rate": 2.42749731471536e-06,
+      "loss": 0.0,
+      "step": 12135
+    },
+    {
+      "epoch": 0.7821099439324611,
+      "grad_norm": 0.0012636307050004592,
+      "learning_rate": 2.4267812388113145e-06,
+      "loss": 0.0,
+      "step": 12136
+    },
+    {
+      "epoch": 0.7821743893793903,
+      "grad_norm": 0.0029543542002693716,
+      "learning_rate": 2.4260651629072684e-06,
+      "loss": 0.0,
+      "step": 12137
+    },
+    {
+      "epoch": 0.7822388348263195,
+      "grad_norm": 0.00022484532724232343,
+      "learning_rate": 2.4253490870032227e-06,
+      "loss": 0.0,
+      "step": 12138
+    },
+    {
+      "epoch": 0.7823032802732487,
+      "grad_norm": 0.09689424289044542,
+      "learning_rate": 2.4246330110991766e-06,
+      "loss": 0.0002,
+      "step": 12139
+    },
+    {
+      "epoch": 0.7823677257201779,
+      "grad_norm": 0.0004091185705752037,
+      "learning_rate": 2.423916935195131e-06,
+      "loss": 0.0,
+      "step": 12140
+    },
+    {
+      "epoch": 0.7824321711671071,
+      "grad_norm": 0.03076440841027304,
+      "learning_rate": 2.4232008592910848e-06,
+      "loss": 0.0,
+      "step": 12141
+    },
+    {
+      "epoch": 0.7824966166140362,
+      "grad_norm": 0.00017963182636153072,
+      "learning_rate": 2.4224847833870395e-06,
+      "loss": 0.0,
+      "step": 12142
+    },
+    {
+      "epoch": 0.7825610620609654,
+      "grad_norm": 5.2125403499151375e-05,
+      "learning_rate": 2.4217687074829934e-06,
+      "loss": 0.0,
+      "step": 12143
+    },
+    {
+      "epoch": 0.7826255075078946,
+      "grad_norm": 0.000837962959501766,
+      "learning_rate": 2.4210526315789477e-06,
+      "loss": 0.0,
+      "step": 12144
+    },
+    {
+      "epoch": 0.7826899529548237,
+      "grad_norm": 0.0031376415468503094,
+      "learning_rate": 2.4203365556749015e-06,
+      "loss": 0.0,
+      "step": 12145
+    },
+    {
+      "epoch": 0.7827543984017529,
+      "grad_norm": 5.146406773654627e-05,
+      "learning_rate": 2.419620479770856e-06,
+      "loss": 0.0,
+      "step": 12146
+    },
+    {
+      "epoch": 0.782818843848682,
+      "grad_norm": 0.00029061667402159076,
+      "learning_rate": 2.41890440386681e-06,
+      "loss": 0.0,
+      "step": 12147
+    },
+    {
+      "epoch": 0.7828832892956112,
+      "grad_norm": 0.00689310914039741,
+      "learning_rate": 2.4181883279627644e-06,
+      "loss": 0.0,
+      "step": 12148
+    },
+    {
+      "epoch": 0.7829477347425404,
+      "grad_norm": 0.00255918569442203,
+      "learning_rate": 2.4174722520587183e-06,
+      "loss": 0.0,
+      "step": 12149
+    },
+    {
+      "epoch": 0.7830121801894696,
+      "grad_norm": 0.14416244853354604,
+      "learning_rate": 2.4167561761546726e-06,
+      "loss": 0.0012,
+      "step": 12150
+    },
+    {
+      "epoch": 0.7830766256363988,
+      "grad_norm": 0.3272660684480119,
+      "learning_rate": 2.416040100250627e-06,
+      "loss": 0.0005,
+      "step": 12151
+    },
+    {
+      "epoch": 0.783141071083328,
+      "grad_norm": 0.30817576965074384,
+      "learning_rate": 2.4153240243465808e-06,
+      "loss": 0.0019,
+      "step": 12152
+    },
+    {
+      "epoch": 0.7832055165302572,
+      "grad_norm": 0.07889921893529749,
+      "learning_rate": 2.414607948442535e-06,
+      "loss": 0.0001,
+      "step": 12153
+    },
+    {
+      "epoch": 0.7832699619771863,
+      "grad_norm": 0.00043163331430451227,
+      "learning_rate": 2.4138918725384894e-06,
+      "loss": 0.0,
+      "step": 12154
+    },
+    {
+      "epoch": 0.7833344074241155,
+      "grad_norm": 0.0001289400500351095,
+      "learning_rate": 2.4131757966344437e-06,
+      "loss": 0.0,
+      "step": 12155
+    },
+    {
+      "epoch": 0.7833988528710447,
+      "grad_norm": 0.0007038620007471168,
+      "learning_rate": 2.4124597207303976e-06,
+      "loss": 0.0,
+      "step": 12156
+    },
+    {
+      "epoch": 0.7834632983179738,
+      "grad_norm": 0.009012677822004099,
+      "learning_rate": 2.411743644826352e-06,
+      "loss": 0.0,
+      "step": 12157
+    },
+    {
+      "epoch": 0.783527743764903,
+      "grad_norm": 0.015514184778514704,
+      "learning_rate": 2.4110275689223057e-06,
+      "loss": 0.0,
+      "step": 12158
+    },
+    {
+      "epoch": 0.7835921892118322,
+      "grad_norm": 0.0007476084735745873,
+      "learning_rate": 2.41031149301826e-06,
+      "loss": 0.0,
+      "step": 12159
+    },
+    {
+      "epoch": 0.7836566346587613,
+      "grad_norm": 0.004590340510394398,
+      "learning_rate": 2.4095954171142143e-06,
+      "loss": 0.0,
+      "step": 12160
+    },
+    {
+      "epoch": 0.7837210801056905,
+      "grad_norm": 0.010306499838861465,
+      "learning_rate": 2.4088793412101686e-06,
+      "loss": 0.0,
+      "step": 12161
+    },
+    {
+      "epoch": 0.7837855255526197,
+      "grad_norm": 0.0005240210405741492,
+      "learning_rate": 2.4081632653061225e-06,
+      "loss": 0.0,
+      "step": 12162
+    },
+    {
+      "epoch": 0.7838499709995489,
+      "grad_norm": 0.23179992839626576,
+      "learning_rate": 2.407447189402077e-06,
+      "loss": 0.0022,
+      "step": 12163
+    },
+    {
+      "epoch": 0.7839144164464781,
+      "grad_norm": 0.00023400309095611104,
+      "learning_rate": 2.406731113498031e-06,
+      "loss": 0.0,
+      "step": 12164
+    },
+    {
+      "epoch": 0.7839788618934073,
+      "grad_norm": 0.03840447149473234,
+      "learning_rate": 2.406015037593985e-06,
+      "loss": 0.0001,
+      "step": 12165
+    },
+    {
+      "epoch": 0.7840433073403364,
+      "grad_norm": 0.33425590007271,
+      "learning_rate": 2.4052989616899393e-06,
+      "loss": 0.0024,
+      "step": 12166
+    },
+    {
+      "epoch": 0.7841077527872656,
+      "grad_norm": 0.013922142574820724,
+      "learning_rate": 2.4045828857858936e-06,
+      "loss": 0.0001,
+      "step": 12167
+    },
+    {
+      "epoch": 0.7841721982341947,
+      "grad_norm": 0.0012904758371460228,
+      "learning_rate": 2.403866809881848e-06,
+      "loss": 0.0,
+      "step": 12168
+    },
+    {
+      "epoch": 0.7842366436811239,
+      "grad_norm": 0.0009271433348567991,
+      "learning_rate": 2.4031507339778017e-06,
+      "loss": 0.0,
+      "step": 12169
+    },
+    {
+      "epoch": 0.7843010891280531,
+      "grad_norm": 0.029267167884655023,
+      "learning_rate": 2.402434658073756e-06,
+      "loss": 0.0001,
+      "step": 12170
+    },
+    {
+      "epoch": 0.7843655345749823,
+      "grad_norm": 0.006470476356729586,
+      "learning_rate": 2.4017185821697103e-06,
+      "loss": 0.0001,
+      "step": 12171
+    },
+    {
+      "epoch": 0.7844299800219114,
+      "grad_norm": 0.006364758656668254,
+      "learning_rate": 2.4010025062656646e-06,
+      "loss": 0.0,
+      "step": 12172
+    },
+    {
+      "epoch": 0.7844944254688406,
+      "grad_norm": 0.34521296607212343,
+      "learning_rate": 2.4002864303616185e-06,
+      "loss": 0.0032,
+      "step": 12173
+    },
+    {
+      "epoch": 0.7845588709157698,
+      "grad_norm": 0.003681485751642218,
+      "learning_rate": 2.399570354457573e-06,
+      "loss": 0.0,
+      "step": 12174
+    },
+    {
+      "epoch": 0.784623316362699,
+      "grad_norm": 0.00445226304354565,
+      "learning_rate": 2.3988542785535267e-06,
+      "loss": 0.0,
+      "step": 12175
+    },
+    {
+      "epoch": 0.7846877618096282,
+      "grad_norm": 0.0015989638304230221,
+      "learning_rate": 2.398138202649481e-06,
+      "loss": 0.0,
+      "step": 12176
+    },
+    {
+      "epoch": 0.7847522072565574,
+      "grad_norm": 0.0043978620529324984,
+      "learning_rate": 2.3974221267454353e-06,
+      "loss": 0.0,
+      "step": 12177
+    },
+    {
+      "epoch": 0.7848166527034866,
+      "grad_norm": 0.017444955525010995,
+      "learning_rate": 2.3967060508413896e-06,
+      "loss": 0.0,
+      "step": 12178
+    },
+    {
+      "epoch": 0.7848810981504156,
+      "grad_norm": 0.01364086457456686,
+      "learning_rate": 2.3959899749373435e-06,
+      "loss": 0.0015,
+      "step": 12179
+    },
+    {
+      "epoch": 0.7849455435973448,
+      "grad_norm": 0.005634009224861755,
+      "learning_rate": 2.3952738990332978e-06,
+      "loss": 0.0,
+      "step": 12180
+    },
+    {
+      "epoch": 0.785009989044274,
+      "grad_norm": 0.0002739373318359726,
+      "learning_rate": 2.3945578231292516e-06,
+      "loss": 0.0,
+      "step": 12181
+    },
+    {
+      "epoch": 0.7850744344912032,
+      "grad_norm": 0.013928744806136031,
+      "learning_rate": 2.393841747225206e-06,
+      "loss": 0.0,
+      "step": 12182
+    },
+    {
+      "epoch": 0.7851388799381324,
+      "grad_norm": 0.00152916755038011,
+      "learning_rate": 2.3931256713211602e-06,
+      "loss": 0.0,
+      "step": 12183
+    },
+    {
+      "epoch": 0.7852033253850615,
+      "grad_norm": 0.005809892577009027,
+      "learning_rate": 2.3924095954171145e-06,
+      "loss": 0.0,
+      "step": 12184
+    },
+    {
+      "epoch": 0.7852677708319907,
+      "grad_norm": 0.013241687900794809,
+      "learning_rate": 2.3916935195130684e-06,
+      "loss": 0.0001,
+      "step": 12185
+    },
+    {
+      "epoch": 0.7853322162789199,
+      "grad_norm": 0.00023618599780431406,
+      "learning_rate": 2.3909774436090227e-06,
+      "loss": 0.0,
+      "step": 12186
+    },
+    {
+      "epoch": 0.7853966617258491,
+      "grad_norm": 0.030064050017800315,
+      "learning_rate": 2.390261367704977e-06,
+      "loss": 0.0003,
+      "step": 12187
+    },
+    {
+      "epoch": 0.7854611071727783,
+      "grad_norm": 0.01977551719248264,
+      "learning_rate": 2.389545291800931e-06,
+      "loss": 0.0015,
+      "step": 12188
+    },
+    {
+      "epoch": 0.7855255526197075,
+      "grad_norm": 0.0002665702840403745,
+      "learning_rate": 2.388829215896885e-06,
+      "loss": 0.0,
+      "step": 12189
+    },
+    {
+      "epoch": 0.7855899980666365,
+      "grad_norm": 0.016736773736830732,
+      "learning_rate": 2.3881131399928395e-06,
+      "loss": 0.0,
+      "step": 12190
+    },
+    {
+      "epoch": 0.7856544435135657,
+      "grad_norm": 0.011101890454886009,
+      "learning_rate": 2.3873970640887938e-06,
+      "loss": 0.0,
+      "step": 12191
+    },
+    {
+      "epoch": 0.7857188889604949,
+      "grad_norm": 0.23320839139528582,
+      "learning_rate": 2.3866809881847476e-06,
+      "loss": 0.0002,
+      "step": 12192
+    },
+    {
+      "epoch": 0.7857833344074241,
+      "grad_norm": 0.001548839832227801,
+      "learning_rate": 2.385964912280702e-06,
+      "loss": 0.0,
+      "step": 12193
+    },
+    {
+      "epoch": 0.7858477798543533,
+      "grad_norm": 0.008140804390900756,
+      "learning_rate": 2.385248836376656e-06,
+      "loss": 0.0,
+      "step": 12194
+    },
+    {
+      "epoch": 0.7859122253012825,
+      "grad_norm": 0.002026366280862881,
+      "learning_rate": 2.3845327604726105e-06,
+      "loss": 0.0,
+      "step": 12195
+    },
+    {
+      "epoch": 0.7859766707482116,
+      "grad_norm": 0.0012847861580297382,
+      "learning_rate": 2.3838166845685644e-06,
+      "loss": 0.0,
+      "step": 12196
+    },
+    {
+      "epoch": 0.7860411161951408,
+      "grad_norm": 0.0006644772422334051,
+      "learning_rate": 2.3831006086645187e-06,
+      "loss": 0.0,
+      "step": 12197
+    },
+    {
+      "epoch": 0.78610556164207,
+      "grad_norm": 0.03664619062678337,
+      "learning_rate": 2.3823845327604726e-06,
+      "loss": 0.0,
+      "step": 12198
+    },
+    {
+      "epoch": 0.7861700070889992,
+      "grad_norm": 0.004651097115143568,
+      "learning_rate": 2.381668456856427e-06,
+      "loss": 0.0,
+      "step": 12199
+    },
+    {
+      "epoch": 0.7862344525359284,
+      "grad_norm": 0.015155141237009116,
+      "learning_rate": 2.380952380952381e-06,
+      "loss": 0.0,
+      "step": 12200
+    },
+    {
+      "epoch": 0.7862988979828575,
+      "grad_norm": 0.14372060879564724,
+      "learning_rate": 2.3802363050483355e-06,
+      "loss": 0.0021,
+      "step": 12201
+    },
+    {
+      "epoch": 0.7863633434297866,
+      "grad_norm": 0.012198772020524575,
+      "learning_rate": 2.3795202291442894e-06,
+      "loss": 0.0001,
+      "step": 12202
+    },
+    {
+      "epoch": 0.7864277888767158,
+      "grad_norm": 0.003081126606843541,
+      "learning_rate": 2.3788041532402437e-06,
+      "loss": 0.0,
+      "step": 12203
+    },
+    {
+      "epoch": 0.786492234323645,
+      "grad_norm": 0.0045107827367743885,
+      "learning_rate": 2.378088077336198e-06,
+      "loss": 0.0,
+      "step": 12204
+    },
+    {
+      "epoch": 0.7865566797705742,
+      "grad_norm": 0.14999968012139864,
+      "learning_rate": 2.377372001432152e-06,
+      "loss": 0.0013,
+      "step": 12205
+    },
+    {
+      "epoch": 0.7866211252175034,
+      "grad_norm": 0.09314629267320713,
+      "learning_rate": 2.376655925528106e-06,
+      "loss": 0.0001,
+      "step": 12206
+    },
+    {
+      "epoch": 0.7866855706644326,
+      "grad_norm": 0.0022869868699410103,
+      "learning_rate": 2.3759398496240604e-06,
+      "loss": 0.0,
+      "step": 12207
+    },
+    {
+      "epoch": 0.7867500161113617,
+      "grad_norm": 0.00021443939616199012,
+      "learning_rate": 2.3752237737200147e-06,
+      "loss": 0.0,
+      "step": 12208
+    },
+    {
+      "epoch": 0.7868144615582909,
+      "grad_norm": 0.007957057997663745,
+      "learning_rate": 2.3745076978159686e-06,
+      "loss": 0.0,
+      "step": 12209
+    },
+    {
+      "epoch": 0.7868789070052201,
+      "grad_norm": 0.047592136178799455,
+      "learning_rate": 2.373791621911923e-06,
+      "loss": 0.0,
+      "step": 12210
+    },
+    {
+      "epoch": 0.7869433524521493,
+      "grad_norm": 0.004593912436612267,
+      "learning_rate": 2.3730755460078768e-06,
+      "loss": 0.0,
+      "step": 12211
+    },
+    {
+      "epoch": 0.7870077978990784,
+      "grad_norm": 0.012097084100934885,
+      "learning_rate": 2.3723594701038315e-06,
+      "loss": 0.0001,
+      "step": 12212
+    },
+    {
+      "epoch": 0.7870722433460076,
+      "grad_norm": 0.02758894133465858,
+      "learning_rate": 2.3716433941997854e-06,
+      "loss": 0.0,
+      "step": 12213
+    },
+    {
+      "epoch": 0.7871366887929367,
+      "grad_norm": 0.000602953256711079,
+      "learning_rate": 2.3709273182957397e-06,
+      "loss": 0.0,
+      "step": 12214
+    },
+    {
+      "epoch": 0.7872011342398659,
+      "grad_norm": 0.004375077774056914,
+      "learning_rate": 2.3702112423916936e-06,
+      "loss": 0.0,
+      "step": 12215
+    },
+    {
+      "epoch": 0.7872655796867951,
+      "grad_norm": 0.0025716119276872535,
+      "learning_rate": 2.369495166487648e-06,
+      "loss": 0.0,
+      "step": 12216
+    },
+    {
+      "epoch": 0.7873300251337243,
+      "grad_norm": 0.01916180191521439,
+      "learning_rate": 2.368779090583602e-06,
+      "loss": 0.0001,
+      "step": 12217
+    },
+    {
+      "epoch": 0.7873944705806535,
+      "grad_norm": 0.15724028651836391,
+      "learning_rate": 2.3680630146795564e-06,
+      "loss": 0.0004,
+      "step": 12218
+    },
+    {
+      "epoch": 0.7874589160275827,
+      "grad_norm": 0.14482280604470232,
+      "learning_rate": 2.3673469387755103e-06,
+      "loss": 0.0017,
+      "step": 12219
+    },
+    {
+      "epoch": 0.7875233614745119,
+      "grad_norm": 0.17595384919038112,
+      "learning_rate": 2.3666308628714646e-06,
+      "loss": 0.0005,
+      "step": 12220
+    },
+    {
+      "epoch": 0.787587806921441,
+      "grad_norm": 0.11437295345132624,
+      "learning_rate": 2.365914786967419e-06,
+      "loss": 0.0003,
+      "step": 12221
+    },
+    {
+      "epoch": 0.7876522523683702,
+      "grad_norm": 0.027399682842273777,
+      "learning_rate": 2.365198711063373e-06,
+      "loss": 0.0,
+      "step": 12222
+    },
+    {
+      "epoch": 0.7877166978152993,
+      "grad_norm": 0.00710695761001995,
+      "learning_rate": 2.364482635159327e-06,
+      "loss": 0.0,
+      "step": 12223
+    },
+    {
+      "epoch": 0.7877811432622285,
+      "grad_norm": 0.0010690633257261582,
+      "learning_rate": 2.3637665592552814e-06,
+      "loss": 0.0,
+      "step": 12224
+    },
+    {
+      "epoch": 0.7878455887091577,
+      "grad_norm": 0.3357881946120721,
+      "learning_rate": 2.3630504833512357e-06,
+      "loss": 0.0006,
+      "step": 12225
+    },
+    {
+      "epoch": 0.7879100341560868,
+      "grad_norm": 0.011652891799082953,
+      "learning_rate": 2.3623344074471896e-06,
+      "loss": 0.0,
+      "step": 12226
+    },
+    {
+      "epoch": 0.787974479603016,
+      "grad_norm": 0.002222747962640288,
+      "learning_rate": 2.361618331543144e-06,
+      "loss": 0.0,
+      "step": 12227
+    },
+    {
+      "epoch": 0.7880389250499452,
+      "grad_norm": 0.010843309689273661,
+      "learning_rate": 2.3609022556390977e-06,
+      "loss": 0.0001,
+      "step": 12228
+    },
+    {
+      "epoch": 0.7881033704968744,
+      "grad_norm": 0.46336933926030965,
+      "learning_rate": 2.360186179735052e-06,
+      "loss": 0.0068,
+      "step": 12229
+    },
+    {
+      "epoch": 0.7881678159438036,
+      "grad_norm": 0.0005158168464518133,
+      "learning_rate": 2.3594701038310063e-06,
+      "loss": 0.0015,
+      "step": 12230
+    },
+    {
+      "epoch": 0.7882322613907328,
+      "grad_norm": 0.00042800521227334665,
+      "learning_rate": 2.3587540279269606e-06,
+      "loss": 0.0,
+      "step": 12231
+    },
+    {
+      "epoch": 0.788296706837662,
+      "grad_norm": 0.0375830123299696,
+      "learning_rate": 2.3580379520229145e-06,
+      "loss": 0.0003,
+      "step": 12232
+    },
+    {
+      "epoch": 0.7883611522845911,
+      "grad_norm": 0.3342013072829814,
+      "learning_rate": 2.357321876118869e-06,
+      "loss": 0.0003,
+      "step": 12233
+    },
+    {
+      "epoch": 0.7884255977315203,
+      "grad_norm": 0.0008485553032904481,
+      "learning_rate": 2.3566058002148227e-06,
+      "loss": 0.0,
+      "step": 12234
+    },
+    {
+      "epoch": 0.7884900431784494,
+      "grad_norm": 0.4544385418544847,
+      "learning_rate": 2.355889724310777e-06,
+      "loss": 0.0008,
+      "step": 12235
+    },
+    {
+      "epoch": 0.7885544886253786,
+      "grad_norm": 0.0010409078297272856,
+      "learning_rate": 2.3551736484067313e-06,
+      "loss": 0.0,
+      "step": 12236
+    },
+    {
+      "epoch": 0.7886189340723078,
+      "grad_norm": 0.002169620912121323,
+      "learning_rate": 2.3544575725026856e-06,
+      "loss": 0.0,
+      "step": 12237
+    },
+    {
+      "epoch": 0.788683379519237,
+      "grad_norm": 0.0017981680341119573,
+      "learning_rate": 2.3537414965986395e-06,
+      "loss": 0.0,
+      "step": 12238
+    },
+    {
+      "epoch": 0.7887478249661661,
+      "grad_norm": 0.04692421753212694,
+      "learning_rate": 2.3530254206945938e-06,
+      "loss": 0.0004,
+      "step": 12239
+    },
+    {
+      "epoch": 0.7888122704130953,
+      "grad_norm": 0.040648232692013384,
+      "learning_rate": 2.352309344790548e-06,
+      "loss": 0.0001,
+      "step": 12240
+    },
+    {
+      "epoch": 0.7888767158600245,
+      "grad_norm": 0.21630014614031065,
+      "learning_rate": 2.351593268886502e-06,
+      "loss": 0.0003,
+      "step": 12241
+    },
+    {
+      "epoch": 0.7889411613069537,
+      "grad_norm": 0.0016455505937524788,
+      "learning_rate": 2.3508771929824562e-06,
+      "loss": 0.0,
+      "step": 12242
+    },
+    {
+      "epoch": 0.7890056067538829,
+      "grad_norm": 1.7173800802001882e-05,
+      "learning_rate": 2.3501611170784105e-06,
+      "loss": 0.0,
+      "step": 12243
+    },
+    {
+      "epoch": 0.789070052200812,
+      "grad_norm": 0.3804484963166623,
+      "learning_rate": 2.349445041174365e-06,
+      "loss": 0.0029,
+      "step": 12244
+    },
+    {
+      "epoch": 0.7891344976477412,
+      "grad_norm": 0.0006245605744261292,
+      "learning_rate": 2.3487289652703187e-06,
+      "loss": 0.0,
+      "step": 12245
+    },
+    {
+      "epoch": 0.7891989430946703,
+      "grad_norm": 0.0016185882460237843,
+      "learning_rate": 2.348012889366273e-06,
+      "loss": 0.0,
+      "step": 12246
+    },
+    {
+      "epoch": 0.7892633885415995,
+      "grad_norm": 6.474478917892943e-05,
+      "learning_rate": 2.3472968134622273e-06,
+      "loss": 0.0,
+      "step": 12247
+    },
+    {
+      "epoch": 0.7893278339885287,
+      "grad_norm": 0.0010739064109665784,
+      "learning_rate": 2.3465807375581816e-06,
+      "loss": 0.0,
+      "step": 12248
+    },
+    {
+      "epoch": 0.7893922794354579,
+      "grad_norm": 0.005548181129234342,
+      "learning_rate": 2.3458646616541355e-06,
+      "loss": 0.0001,
+      "step": 12249
+    },
+    {
+      "epoch": 0.789456724882387,
+      "grad_norm": 0.008573517040880186,
+      "learning_rate": 2.3451485857500898e-06,
+      "loss": 0.0,
+      "step": 12250
+    },
+    {
+      "epoch": 0.7895211703293162,
+      "grad_norm": 0.0006694098065241521,
+      "learning_rate": 2.3444325098460436e-06,
+      "loss": 0.0,
+      "step": 12251
+    },
+    {
+      "epoch": 0.7895856157762454,
+      "grad_norm": 0.027516989197968755,
+      "learning_rate": 2.343716433941998e-06,
+      "loss": 0.0,
+      "step": 12252
+    },
+    {
+      "epoch": 0.7896500612231746,
+      "grad_norm": 0.0049549690740599,
+      "learning_rate": 2.3430003580379522e-06,
+      "loss": 0.0,
+      "step": 12253
+    },
+    {
+      "epoch": 0.7897145066701038,
+      "grad_norm": 0.0029326547432867644,
+      "learning_rate": 2.3422842821339065e-06,
+      "loss": 0.0,
+      "step": 12254
+    },
+    {
+      "epoch": 0.789778952117033,
+      "grad_norm": 0.025886055570603583,
+      "learning_rate": 2.3415682062298604e-06,
+      "loss": 0.0,
+      "step": 12255
+    },
+    {
+      "epoch": 0.7898433975639622,
+      "grad_norm": 0.012209315086344184,
+      "learning_rate": 2.3408521303258147e-06,
+      "loss": 0.0,
+      "step": 12256
+    },
+    {
+      "epoch": 0.7899078430108912,
+      "grad_norm": 6.541967178543215e-05,
+      "learning_rate": 2.340136054421769e-06,
+      "loss": 0.0,
+      "step": 12257
+    },
+    {
+      "epoch": 0.7899722884578204,
+      "grad_norm": 0.002371812470589662,
+      "learning_rate": 2.339419978517723e-06,
+      "loss": 0.0,
+      "step": 12258
+    },
+    {
+      "epoch": 0.7900367339047496,
+      "grad_norm": 0.11401620408703504,
+      "learning_rate": 2.338703902613677e-06,
+      "loss": 0.0019,
+      "step": 12259
+    },
+    {
+      "epoch": 0.7901011793516788,
+      "grad_norm": 0.012610422179252519,
+      "learning_rate": 2.3379878267096315e-06,
+      "loss": 0.0,
+      "step": 12260
+    },
+    {
+      "epoch": 0.790165624798608,
+      "grad_norm": 0.00046666391484797036,
+      "learning_rate": 2.3372717508055858e-06,
+      "loss": 0.0,
+      "step": 12261
+    },
+    {
+      "epoch": 0.7902300702455372,
+      "grad_norm": 0.00217984363643231,
+      "learning_rate": 2.3365556749015397e-06,
+      "loss": 0.0,
+      "step": 12262
+    },
+    {
+      "epoch": 0.7902945156924663,
+      "grad_norm": 0.008862938690758203,
+      "learning_rate": 2.335839598997494e-06,
+      "loss": 0.0,
+      "step": 12263
+    },
+    {
+      "epoch": 0.7903589611393955,
+      "grad_norm": 0.05935538910762845,
+      "learning_rate": 2.335123523093448e-06,
+      "loss": 0.0001,
+      "step": 12264
+    },
+    {
+      "epoch": 0.7904234065863247,
+      "grad_norm": 0.0005315012449867554,
+      "learning_rate": 2.3344074471894026e-06,
+      "loss": 0.0,
+      "step": 12265
+    },
+    {
+      "epoch": 0.7904878520332539,
+      "grad_norm": 0.0012946143773809717,
+      "learning_rate": 2.3336913712853564e-06,
+      "loss": 0.0,
+      "step": 12266
+    },
+    {
+      "epoch": 0.7905522974801831,
+      "grad_norm": 0.022759214076861673,
+      "learning_rate": 2.3329752953813107e-06,
+      "loss": 0.0,
+      "step": 12267
+    },
+    {
+      "epoch": 0.7906167429271121,
+      "grad_norm": 0.0003268068807733046,
+      "learning_rate": 2.3322592194772646e-06,
+      "loss": 0.0,
+      "step": 12268
+    },
+    {
+      "epoch": 0.7906811883740413,
+      "grad_norm": 0.3537525626450311,
+      "learning_rate": 2.331543143573219e-06,
+      "loss": 0.0012,
+      "step": 12269
+    },
+    {
+      "epoch": 0.7907456338209705,
+      "grad_norm": 0.01784420948702125,
+      "learning_rate": 2.330827067669173e-06,
+      "loss": 0.0001,
+      "step": 12270
+    },
+    {
+      "epoch": 0.7908100792678997,
+      "grad_norm": 0.001048223111041715,
+      "learning_rate": 2.3301109917651275e-06,
+      "loss": 0.0,
+      "step": 12271
+    },
+    {
+      "epoch": 0.7908745247148289,
+      "grad_norm": 0.0011720714576711842,
+      "learning_rate": 2.3293949158610814e-06,
+      "loss": 0.0,
+      "step": 12272
+    },
+    {
+      "epoch": 0.7909389701617581,
+      "grad_norm": 0.09711526275150101,
+      "learning_rate": 2.3286788399570357e-06,
+      "loss": 0.0001,
+      "step": 12273
+    },
+    {
+      "epoch": 0.7910034156086873,
+      "grad_norm": 0.0004784918328230193,
+      "learning_rate": 2.32796276405299e-06,
+      "loss": 0.0,
+      "step": 12274
+    },
+    {
+      "epoch": 0.7910678610556164,
+      "grad_norm": 0.0019445352077134038,
+      "learning_rate": 2.327246688148944e-06,
+      "loss": 0.0,
+      "step": 12275
+    },
+    {
+      "epoch": 0.7911323065025456,
+      "grad_norm": 0.0010747047967742114,
+      "learning_rate": 2.326530612244898e-06,
+      "loss": 0.0,
+      "step": 12276
+    },
+    {
+      "epoch": 0.7911967519494748,
+      "grad_norm": 0.0008913434106858376,
+      "learning_rate": 2.3258145363408524e-06,
+      "loss": 0.0,
+      "step": 12277
+    },
+    {
+      "epoch": 0.791261197396404,
+      "grad_norm": 0.0008809684577119583,
+      "learning_rate": 2.3250984604368067e-06,
+      "loss": 0.0,
+      "step": 12278
+    },
+    {
+      "epoch": 0.7913256428433331,
+      "grad_norm": 0.0033671814651020987,
+      "learning_rate": 2.3243823845327606e-06,
+      "loss": 0.0,
+      "step": 12279
+    },
+    {
+      "epoch": 0.7913900882902623,
+      "grad_norm": 0.0011540542898446035,
+      "learning_rate": 2.323666308628715e-06,
+      "loss": 0.0,
+      "step": 12280
+    },
+    {
+      "epoch": 0.7914545337371914,
+      "grad_norm": 0.003371865755117032,
+      "learning_rate": 2.322950232724669e-06,
+      "loss": 0.0,
+      "step": 12281
+    },
+    {
+      "epoch": 0.7915189791841206,
+      "grad_norm": 0.00016563388657528873,
+      "learning_rate": 2.3222341568206235e-06,
+      "loss": 0.0,
+      "step": 12282
+    },
+    {
+      "epoch": 0.7915834246310498,
+      "grad_norm": 0.09632593516717095,
+      "learning_rate": 2.3215180809165774e-06,
+      "loss": 0.0013,
+      "step": 12283
+    },
+    {
+      "epoch": 0.791647870077979,
+      "grad_norm": 0.0017298544745243945,
+      "learning_rate": 2.3208020050125317e-06,
+      "loss": 0.0,
+      "step": 12284
+    },
+    {
+      "epoch": 0.7917123155249082,
+      "grad_norm": 0.010223582251694568,
+      "learning_rate": 2.3200859291084856e-06,
+      "loss": 0.0001,
+      "step": 12285
+    },
+    {
+      "epoch": 0.7917767609718374,
+      "grad_norm": 0.003706090422825231,
+      "learning_rate": 2.31936985320444e-06,
+      "loss": 0.0,
+      "step": 12286
+    },
+    {
+      "epoch": 0.7918412064187665,
+      "grad_norm": 0.5455833713255104,
+      "learning_rate": 2.3186537773003937e-06,
+      "loss": 0.0033,
+      "step": 12287
+    },
+    {
+      "epoch": 0.7919056518656957,
+      "grad_norm": 0.005634153863743244,
+      "learning_rate": 2.3179377013963485e-06,
+      "loss": 0.0,
+      "step": 12288
+    },
+    {
+      "epoch": 0.7919700973126249,
+      "grad_norm": 0.00811497967985401,
+      "learning_rate": 2.3172216254923023e-06,
+      "loss": 0.0,
+      "step": 12289
+    },
+    {
+      "epoch": 0.792034542759554,
+      "grad_norm": 0.023416422764268287,
+      "learning_rate": 2.3165055495882566e-06,
+      "loss": 0.0,
+      "step": 12290
+    },
+    {
+      "epoch": 0.7920989882064832,
+      "grad_norm": 0.000397575294584166,
+      "learning_rate": 2.3157894736842105e-06,
+      "loss": 0.0,
+      "step": 12291
+    },
+    {
+      "epoch": 0.7921634336534124,
+      "grad_norm": 0.09261146259030603,
+      "learning_rate": 2.315073397780165e-06,
+      "loss": 0.0007,
+      "step": 12292
+    },
+    {
+      "epoch": 0.7922278791003415,
+      "grad_norm": 6.736243921635722e-05,
+      "learning_rate": 2.314357321876119e-06,
+      "loss": 0.0,
+      "step": 12293
+    },
+    {
+      "epoch": 0.7922923245472707,
+      "grad_norm": 8.188848771461251e-05,
+      "learning_rate": 2.3136412459720734e-06,
+      "loss": 0.0,
+      "step": 12294
+    },
+    {
+      "epoch": 0.7923567699941999,
+      "grad_norm": 0.5248949627024653,
+      "learning_rate": 2.3129251700680273e-06,
+      "loss": 0.0005,
+      "step": 12295
+    },
+    {
+      "epoch": 0.7924212154411291,
+      "grad_norm": 0.0022065020062548503,
+      "learning_rate": 2.3122090941639816e-06,
+      "loss": 0.0,
+      "step": 12296
+    },
+    {
+      "epoch": 0.7924856608880583,
+      "grad_norm": 0.0012940578166477242,
+      "learning_rate": 2.311493018259936e-06,
+      "loss": 0.0,
+      "step": 12297
+    },
+    {
+      "epoch": 0.7925501063349875,
+      "grad_norm": 0.0008661635742292246,
+      "learning_rate": 2.3107769423558898e-06,
+      "loss": 0.0,
+      "step": 12298
+    },
+    {
+      "epoch": 0.7926145517819166,
+      "grad_norm": 0.012853991928710864,
+      "learning_rate": 2.310060866451844e-06,
+      "loss": 0.0,
+      "step": 12299
+    },
+    {
+      "epoch": 0.7926789972288458,
+      "grad_norm": 0.0013576044205097536,
+      "learning_rate": 2.3093447905477984e-06,
+      "loss": 0.0,
+      "step": 12300
+    },
+    {
+      "epoch": 0.792743442675775,
+      "grad_norm": 6.910762728236609e-05,
+      "learning_rate": 2.3086287146437527e-06,
+      "loss": 0.0,
+      "step": 12301
+    },
+    {
+      "epoch": 0.7928078881227041,
+      "grad_norm": 0.0067343948025941945,
+      "learning_rate": 2.3079126387397065e-06,
+      "loss": 0.0,
+      "step": 12302
+    },
+    {
+      "epoch": 0.7928723335696333,
+      "grad_norm": 0.13638891214778198,
+      "learning_rate": 2.307196562835661e-06,
+      "loss": 0.0004,
+      "step": 12303
+    },
+    {
+      "epoch": 0.7929367790165625,
+      "grad_norm": 0.32272106701892816,
+      "learning_rate": 2.3064804869316147e-06,
+      "loss": 0.0012,
+      "step": 12304
+    },
+    {
+      "epoch": 0.7930012244634916,
+      "grad_norm": 0.0008684087055918425,
+      "learning_rate": 2.305764411027569e-06,
+      "loss": 0.0,
+      "step": 12305
+    },
+    {
+      "epoch": 0.7930656699104208,
+      "grad_norm": 0.0001796671162931569,
+      "learning_rate": 2.3050483351235233e-06,
+      "loss": 0.0,
+      "step": 12306
+    },
+    {
+      "epoch": 0.79313011535735,
+      "grad_norm": 0.22511224430501187,
+      "learning_rate": 2.3043322592194776e-06,
+      "loss": 0.0007,
+      "step": 12307
+    },
+    {
+      "epoch": 0.7931945608042792,
+      "grad_norm": 0.0004992730688903502,
+      "learning_rate": 2.3036161833154315e-06,
+      "loss": 0.0,
+      "step": 12308
+    },
+    {
+      "epoch": 0.7932590062512084,
+      "grad_norm": 0.006909952409031217,
+      "learning_rate": 2.3029001074113858e-06,
+      "loss": 0.0,
+      "step": 12309
+    },
+    {
+      "epoch": 0.7933234516981376,
+      "grad_norm": 0.0040115938022819,
+      "learning_rate": 2.30218403150734e-06,
+      "loss": 0.0,
+      "step": 12310
+    },
+    {
+      "epoch": 0.7933878971450667,
+      "grad_norm": 0.0014633534190212837,
+      "learning_rate": 2.301467955603294e-06,
+      "loss": 0.0,
+      "step": 12311
+    },
+    {
+      "epoch": 0.7934523425919959,
+      "grad_norm": 0.012486874126142355,
+      "learning_rate": 2.3007518796992482e-06,
+      "loss": 0.0,
+      "step": 12312
+    },
+    {
+      "epoch": 0.793516788038925,
+      "grad_norm": 0.0010191498823039192,
+      "learning_rate": 2.3000358037952025e-06,
+      "loss": 0.0,
+      "step": 12313
+    },
+    {
+      "epoch": 0.7935812334858542,
+      "grad_norm": 0.004574874004510897,
+      "learning_rate": 2.299319727891157e-06,
+      "loss": 0.0,
+      "step": 12314
+    },
+    {
+      "epoch": 0.7936456789327834,
+      "grad_norm": 0.00018292927670207949,
+      "learning_rate": 2.2986036519871107e-06,
+      "loss": 0.0,
+      "step": 12315
+    },
+    {
+      "epoch": 0.7937101243797126,
+      "grad_norm": 0.2433858495795913,
+      "learning_rate": 2.297887576083065e-06,
+      "loss": 0.0025,
+      "step": 12316
+    },
+    {
+      "epoch": 0.7937745698266417,
+      "grad_norm": 0.000507205422647784,
+      "learning_rate": 2.2971715001790193e-06,
+      "loss": 0.0,
+      "step": 12317
+    },
+    {
+      "epoch": 0.7938390152735709,
+      "grad_norm": 0.1609688126555202,
+      "learning_rate": 2.2964554242749736e-06,
+      "loss": 0.0019,
+      "step": 12318
+    },
+    {
+      "epoch": 0.7939034607205001,
+      "grad_norm": 0.0012293050600295469,
+      "learning_rate": 2.2957393483709275e-06,
+      "loss": 0.0,
+      "step": 12319
+    },
+    {
+      "epoch": 0.7939679061674293,
+      "grad_norm": 0.0009011055666482163,
+      "learning_rate": 2.2950232724668818e-06,
+      "loss": 0.0,
+      "step": 12320
+    },
+    {
+      "epoch": 0.7940323516143585,
+      "grad_norm": 0.002458300198121493,
+      "learning_rate": 2.2943071965628357e-06,
+      "loss": 0.0,
+      "step": 12321
+    },
+    {
+      "epoch": 0.7940967970612877,
+      "grad_norm": 0.010545672807899122,
+      "learning_rate": 2.29359112065879e-06,
+      "loss": 0.0001,
+      "step": 12322
+    },
+    {
+      "epoch": 0.7941612425082168,
+      "grad_norm": 0.0006009889248218322,
+      "learning_rate": 2.2928750447547443e-06,
+      "loss": 0.0,
+      "step": 12323
+    },
+    {
+      "epoch": 0.7942256879551459,
+      "grad_norm": 0.0002882234106889472,
+      "learning_rate": 2.2921589688506986e-06,
+      "loss": 0.0,
+      "step": 12324
+    },
+    {
+      "epoch": 0.7942901334020751,
+      "grad_norm": 7.939152417382084e-05,
+      "learning_rate": 2.2914428929466524e-06,
+      "loss": 0.0,
+      "step": 12325
+    },
+    {
+      "epoch": 0.7943545788490043,
+      "grad_norm": 0.0004449329234502113,
+      "learning_rate": 2.2907268170426067e-06,
+      "loss": 0.0,
+      "step": 12326
+    },
+    {
+      "epoch": 0.7944190242959335,
+      "grad_norm": 0.00014424612203982502,
+      "learning_rate": 2.2900107411385606e-06,
+      "loss": 0.0,
+      "step": 12327
+    },
+    {
+      "epoch": 0.7944834697428627,
+      "grad_norm": 0.0031572437902694242,
+      "learning_rate": 2.289294665234515e-06,
+      "loss": 0.0,
+      "step": 12328
+    },
+    {
+      "epoch": 0.7945479151897918,
+      "grad_norm": 0.02013882993644046,
+      "learning_rate": 2.288578589330469e-06,
+      "loss": 0.0,
+      "step": 12329
+    },
+    {
+      "epoch": 0.794612360636721,
+      "grad_norm": 0.0006641821282922399,
+      "learning_rate": 2.2878625134264235e-06,
+      "loss": 0.0,
+      "step": 12330
+    },
+    {
+      "epoch": 0.7946768060836502,
+      "grad_norm": 0.0964850576276222,
+      "learning_rate": 2.2871464375223774e-06,
+      "loss": 0.0003,
+      "step": 12331
+    },
+    {
+      "epoch": 0.7947412515305794,
+      "grad_norm": 0.0007458173316273714,
+      "learning_rate": 2.2864303616183317e-06,
+      "loss": 0.0,
+      "step": 12332
+    },
+    {
+      "epoch": 0.7948056969775086,
+      "grad_norm": 0.044678187758849984,
+      "learning_rate": 2.285714285714286e-06,
+      "loss": 0.0001,
+      "step": 12333
+    },
+    {
+      "epoch": 0.7948701424244378,
+      "grad_norm": 0.012587534549491937,
+      "learning_rate": 2.28499820981024e-06,
+      "loss": 0.0001,
+      "step": 12334
+    },
+    {
+      "epoch": 0.7949345878713668,
+      "grad_norm": 0.11678656098745568,
+      "learning_rate": 2.284282133906194e-06,
+      "loss": 0.0002,
+      "step": 12335
+    },
+    {
+      "epoch": 0.794999033318296,
+      "grad_norm": 0.0010133240072671896,
+      "learning_rate": 2.2835660580021484e-06,
+      "loss": 0.0,
+      "step": 12336
+    },
+    {
+      "epoch": 0.7950634787652252,
+      "grad_norm": 0.01128362238810577,
+      "learning_rate": 2.2828499820981027e-06,
+      "loss": 0.0,
+      "step": 12337
+    },
+    {
+      "epoch": 0.7951279242121544,
+      "grad_norm": 0.325144468806227,
+      "learning_rate": 2.2821339061940566e-06,
+      "loss": 0.0024,
+      "step": 12338
+    },
+    {
+      "epoch": 0.7951923696590836,
+      "grad_norm": 0.0003143201195948108,
+      "learning_rate": 2.281417830290011e-06,
+      "loss": 0.0,
+      "step": 12339
+    },
+    {
+      "epoch": 0.7952568151060128,
+      "grad_norm": 0.00018327716875614307,
+      "learning_rate": 2.280701754385965e-06,
+      "loss": 0.0,
+      "step": 12340
+    },
+    {
+      "epoch": 0.7953212605529419,
+      "grad_norm": 0.00038591333705380033,
+      "learning_rate": 2.2799856784819195e-06,
+      "loss": 0.0,
+      "step": 12341
+    },
+    {
+      "epoch": 0.7953857059998711,
+      "grad_norm": 0.0012024442904717586,
+      "learning_rate": 2.2792696025778734e-06,
+      "loss": 0.0,
+      "step": 12342
+    },
+    {
+      "epoch": 0.7954501514468003,
+      "grad_norm": 0.00036475198609929637,
+      "learning_rate": 2.2785535266738277e-06,
+      "loss": 0.0,
+      "step": 12343
+    },
+    {
+      "epoch": 0.7955145968937295,
+      "grad_norm": 0.009867178122339677,
+      "learning_rate": 2.2778374507697816e-06,
+      "loss": 0.0,
+      "step": 12344
+    },
+    {
+      "epoch": 0.7955790423406587,
+      "grad_norm": 0.00080797752438696,
+      "learning_rate": 2.277121374865736e-06,
+      "loss": 0.0,
+      "step": 12345
+    },
+    {
+      "epoch": 0.7956434877875878,
+      "grad_norm": 0.000350938664247268,
+      "learning_rate": 2.27640529896169e-06,
+      "loss": 0.0,
+      "step": 12346
+    },
+    {
+      "epoch": 0.7957079332345169,
+      "grad_norm": 0.0004998791925092817,
+      "learning_rate": 2.2756892230576445e-06,
+      "loss": 0.0,
+      "step": 12347
+    },
+    {
+      "epoch": 0.7957723786814461,
+      "grad_norm": 0.023357191681495117,
+      "learning_rate": 2.2749731471535983e-06,
+      "loss": 0.0001,
+      "step": 12348
+    },
+    {
+      "epoch": 0.7958368241283753,
+      "grad_norm": 0.3889522787023688,
+      "learning_rate": 2.2742570712495526e-06,
+      "loss": 0.0028,
+      "step": 12349
+    },
+    {
+      "epoch": 0.7959012695753045,
+      "grad_norm": 0.026527525468430544,
+      "learning_rate": 2.273540995345507e-06,
+      "loss": 0.0,
+      "step": 12350
+    },
+    {
+      "epoch": 0.7959657150222337,
+      "grad_norm": 0.0003701205214895632,
+      "learning_rate": 2.272824919441461e-06,
+      "loss": 0.0,
+      "step": 12351
+    },
+    {
+      "epoch": 0.7960301604691629,
+      "grad_norm": 0.048662335749158375,
+      "learning_rate": 2.272108843537415e-06,
+      "loss": 0.0005,
+      "step": 12352
+    },
+    {
+      "epoch": 0.796094605916092,
+      "grad_norm": 0.0028033741150151723,
+      "learning_rate": 2.2713927676333694e-06,
+      "loss": 0.0,
+      "step": 12353
+    },
+    {
+      "epoch": 0.7961590513630212,
+      "grad_norm": 0.005704388075964442,
+      "learning_rate": 2.2706766917293237e-06,
+      "loss": 0.0,
+      "step": 12354
+    },
+    {
+      "epoch": 0.7962234968099504,
+      "grad_norm": 0.002117519304700571,
+      "learning_rate": 2.2699606158252776e-06,
+      "loss": 0.0,
+      "step": 12355
+    },
+    {
+      "epoch": 0.7962879422568796,
+      "grad_norm": 0.02104627910229218,
+      "learning_rate": 2.269244539921232e-06,
+      "loss": 0.0001,
+      "step": 12356
+    },
+    {
+      "epoch": 0.7963523877038087,
+      "grad_norm": 0.0002981008190228777,
+      "learning_rate": 2.2685284640171858e-06,
+      "loss": 0.0,
+      "step": 12357
+    },
+    {
+      "epoch": 0.7964168331507379,
+      "grad_norm": 0.15222814648679983,
+      "learning_rate": 2.2678123881131405e-06,
+      "loss": 0.0034,
+      "step": 12358
+    },
+    {
+      "epoch": 0.796481278597667,
+      "grad_norm": 0.0007378246127458366,
+      "learning_rate": 2.2670963122090944e-06,
+      "loss": 0.0,
+      "step": 12359
+    },
+    {
+      "epoch": 0.7965457240445962,
+      "grad_norm": 0.0003285574545467159,
+      "learning_rate": 2.2663802363050487e-06,
+      "loss": 0.0,
+      "step": 12360
+    },
+    {
+      "epoch": 0.7966101694915254,
+      "grad_norm": 0.011981375678915126,
+      "learning_rate": 2.2656641604010025e-06,
+      "loss": 0.0,
+      "step": 12361
+    },
+    {
+      "epoch": 0.7966746149384546,
+      "grad_norm": 0.0715815572735966,
+      "learning_rate": 2.264948084496957e-06,
+      "loss": 0.0,
+      "step": 12362
+    },
+    {
+      "epoch": 0.7967390603853838,
+      "grad_norm": 0.0077975512422411635,
+      "learning_rate": 2.264232008592911e-06,
+      "loss": 0.0,
+      "step": 12363
+    },
+    {
+      "epoch": 0.796803505832313,
+      "grad_norm": 0.00018230401913338704,
+      "learning_rate": 2.2635159326888654e-06,
+      "loss": 0.0,
+      "step": 12364
+    },
+    {
+      "epoch": 0.7968679512792421,
+      "grad_norm": 0.00044439336476779,
+      "learning_rate": 2.2627998567848193e-06,
+      "loss": 0.0,
+      "step": 12365
+    },
+    {
+      "epoch": 0.7969323967261713,
+      "grad_norm": 0.0558334291996121,
+      "learning_rate": 2.2620837808807736e-06,
+      "loss": 0.0,
+      "step": 12366
+    },
+    {
+      "epoch": 0.7969968421731005,
+      "grad_norm": 0.0017058114736429463,
+      "learning_rate": 2.261367704976728e-06,
+      "loss": 0.0,
+      "step": 12367
+    },
+    {
+      "epoch": 0.7970612876200296,
+      "grad_norm": 0.0031099253657042403,
+      "learning_rate": 2.2606516290726818e-06,
+      "loss": 0.0,
+      "step": 12368
+    },
+    {
+      "epoch": 0.7971257330669588,
+      "grad_norm": 0.07728687070940958,
+      "learning_rate": 2.259935553168636e-06,
+      "loss": 0.0008,
+      "step": 12369
+    },
+    {
+      "epoch": 0.797190178513888,
+      "grad_norm": 0.1450721760764216,
+      "learning_rate": 2.2592194772645904e-06,
+      "loss": 0.0005,
+      "step": 12370
+    },
+    {
+      "epoch": 0.7972546239608171,
+      "grad_norm": 0.2237326802977375,
+      "learning_rate": 2.2585034013605447e-06,
+      "loss": 0.0004,
+      "step": 12371
+    },
+    {
+      "epoch": 0.7973190694077463,
+      "grad_norm": 0.010944975406888244,
+      "learning_rate": 2.2577873254564985e-06,
+      "loss": 0.0,
+      "step": 12372
+    },
+    {
+      "epoch": 0.7973835148546755,
+      "grad_norm": 0.0016884389008774191,
+      "learning_rate": 2.257071249552453e-06,
+      "loss": 0.0,
+      "step": 12373
+    },
+    {
+      "epoch": 0.7974479603016047,
+      "grad_norm": 0.00034746512207015046,
+      "learning_rate": 2.2563551736484067e-06,
+      "loss": 0.0,
+      "step": 12374
+    },
+    {
+      "epoch": 0.7975124057485339,
+      "grad_norm": 0.005475084699684978,
+      "learning_rate": 2.255639097744361e-06,
+      "loss": 0.0,
+      "step": 12375
+    },
+    {
+      "epoch": 0.7975768511954631,
+      "grad_norm": 0.2716014115799345,
+      "learning_rate": 2.2549230218403153e-06,
+      "loss": 0.001,
+      "step": 12376
+    },
+    {
+      "epoch": 0.7976412966423923,
+      "grad_norm": 0.001062484250198185,
+      "learning_rate": 2.2542069459362696e-06,
+      "loss": 0.0,
+      "step": 12377
+    },
+    {
+      "epoch": 0.7977057420893214,
+      "grad_norm": 0.007033136720229323,
+      "learning_rate": 2.2534908700322235e-06,
+      "loss": 0.0001,
+      "step": 12378
+    },
+    {
+      "epoch": 0.7977701875362506,
+      "grad_norm": 0.022493932942061604,
+      "learning_rate": 2.2527747941281778e-06,
+      "loss": 0.0,
+      "step": 12379
+    },
+    {
+      "epoch": 0.7978346329831797,
+      "grad_norm": 0.0005868051093204366,
+      "learning_rate": 2.2520587182241317e-06,
+      "loss": 0.0,
+      "step": 12380
+    },
+    {
+      "epoch": 0.7978990784301089,
+      "grad_norm": 0.01206821548237533,
+      "learning_rate": 2.251342642320086e-06,
+      "loss": 0.0,
+      "step": 12381
+    },
+    {
+      "epoch": 0.7979635238770381,
+      "grad_norm": 0.003972101955416267,
+      "learning_rate": 2.2506265664160403e-06,
+      "loss": 0.0,
+      "step": 12382
+    },
+    {
+      "epoch": 0.7980279693239672,
+      "grad_norm": 0.019091583821505498,
+      "learning_rate": 2.2499104905119946e-06,
+      "loss": 0.0002,
+      "step": 12383
+    },
+    {
+      "epoch": 0.7980924147708964,
+      "grad_norm": 0.0028018271371107813,
+      "learning_rate": 2.2491944146079484e-06,
+      "loss": 0.0,
+      "step": 12384
+    },
+    {
+      "epoch": 0.7981568602178256,
+      "grad_norm": 0.018689290061926607,
+      "learning_rate": 2.2484783387039027e-06,
+      "loss": 0.0,
+      "step": 12385
+    },
+    {
+      "epoch": 0.7982213056647548,
+      "grad_norm": 0.0010994308213438746,
+      "learning_rate": 2.247762262799857e-06,
+      "loss": 0.0,
+      "step": 12386
+    },
+    {
+      "epoch": 0.798285751111684,
+      "grad_norm": 0.0011902206978131014,
+      "learning_rate": 2.2470461868958113e-06,
+      "loss": 0.0,
+      "step": 12387
+    },
+    {
+      "epoch": 0.7983501965586132,
+      "grad_norm": 0.00018473605557963436,
+      "learning_rate": 2.246330110991765e-06,
+      "loss": 0.0,
+      "step": 12388
+    },
+    {
+      "epoch": 0.7984146420055424,
+      "grad_norm": 0.0027828798294523296,
+      "learning_rate": 2.2456140350877195e-06,
+      "loss": 0.0,
+      "step": 12389
+    },
+    {
+      "epoch": 0.7984790874524715,
+      "grad_norm": 0.0026909045095040935,
+      "learning_rate": 2.244897959183674e-06,
+      "loss": 0.0,
+      "step": 12390
+    },
+    {
+      "epoch": 0.7985435328994006,
+      "grad_norm": 0.04215782510421399,
+      "learning_rate": 2.2441818832796277e-06,
+      "loss": 0.0,
+      "step": 12391
+    },
+    {
+      "epoch": 0.7986079783463298,
+      "grad_norm": 0.004400490777957583,
+      "learning_rate": 2.243465807375582e-06,
+      "loss": 0.0,
+      "step": 12392
+    },
+    {
+      "epoch": 0.798672423793259,
+      "grad_norm": 0.0009927995263724551,
+      "learning_rate": 2.2427497314715363e-06,
+      "loss": 0.0,
+      "step": 12393
+    },
+    {
+      "epoch": 0.7987368692401882,
+      "grad_norm": 0.0008656766188128951,
+      "learning_rate": 2.2420336555674906e-06,
+      "loss": 0.0,
+      "step": 12394
+    },
+    {
+      "epoch": 0.7988013146871173,
+      "grad_norm": 0.11693899508875823,
+      "learning_rate": 2.2413175796634444e-06,
+      "loss": 0.0012,
+      "step": 12395
+    },
+    {
+      "epoch": 0.7988657601340465,
+      "grad_norm": 0.015184840303952818,
+      "learning_rate": 2.2406015037593987e-06,
+      "loss": 0.0,
+      "step": 12396
+    },
+    {
+      "epoch": 0.7989302055809757,
+      "grad_norm": 0.0008131608440570534,
+      "learning_rate": 2.2398854278553526e-06,
+      "loss": 0.0,
+      "step": 12397
+    },
+    {
+      "epoch": 0.7989946510279049,
+      "grad_norm": 0.008973112267477039,
+      "learning_rate": 2.239169351951307e-06,
+      "loss": 0.0,
+      "step": 12398
+    },
+    {
+      "epoch": 0.7990590964748341,
+      "grad_norm": 0.4256889173153369,
+      "learning_rate": 2.2384532760472612e-06,
+      "loss": 0.0007,
+      "step": 12399
+    },
+    {
+      "epoch": 0.7991235419217633,
+      "grad_norm": 0.0014311981142707912,
+      "learning_rate": 2.2377372001432155e-06,
+      "loss": 0.0,
+      "step": 12400
+    },
+    {
+      "epoch": 0.7991879873686925,
+      "grad_norm": 0.0038494384225200677,
+      "learning_rate": 2.2370211242391694e-06,
+      "loss": 0.0,
+      "step": 12401
+    },
+    {
+      "epoch": 0.7992524328156215,
+      "grad_norm": 0.00021152556432885007,
+      "learning_rate": 2.2363050483351237e-06,
+      "loss": 0.0,
+      "step": 12402
+    },
+    {
+      "epoch": 0.7993168782625507,
+      "grad_norm": 0.0004490586247319541,
+      "learning_rate": 2.235588972431078e-06,
+      "loss": 0.0,
+      "step": 12403
+    },
+    {
+      "epoch": 0.7993813237094799,
+      "grad_norm": 0.0006710799683837646,
+      "learning_rate": 2.234872896527032e-06,
+      "loss": 0.0,
+      "step": 12404
+    },
+    {
+      "epoch": 0.7994457691564091,
+      "grad_norm": 0.010726106254010393,
+      "learning_rate": 2.234156820622986e-06,
+      "loss": 0.0,
+      "step": 12405
+    },
+    {
+      "epoch": 0.7995102146033383,
+      "grad_norm": 0.0014574245169015954,
+      "learning_rate": 2.2334407447189405e-06,
+      "loss": 0.0,
+      "step": 12406
+    },
+    {
+      "epoch": 0.7995746600502674,
+      "grad_norm": 0.0005909390737096638,
+      "learning_rate": 2.2327246688148948e-06,
+      "loss": 0.0,
+      "step": 12407
+    },
+    {
+      "epoch": 0.7996391054971966,
+      "grad_norm": 0.00010923331761016114,
+      "learning_rate": 2.2320085929108486e-06,
+      "loss": 0.0,
+      "step": 12408
+    },
+    {
+      "epoch": 0.7997035509441258,
+      "grad_norm": 0.03501950265571546,
+      "learning_rate": 2.231292517006803e-06,
+      "loss": 0.0,
+      "step": 12409
+    },
+    {
+      "epoch": 0.799767996391055,
+      "grad_norm": 0.0004077594829991419,
+      "learning_rate": 2.230576441102757e-06,
+      "loss": 0.0,
+      "step": 12410
+    },
+    {
+      "epoch": 0.7998324418379842,
+      "grad_norm": 2.0668252348953236e-05,
+      "learning_rate": 2.2298603651987115e-06,
+      "loss": 0.0,
+      "step": 12411
+    },
+    {
+      "epoch": 0.7998968872849134,
+      "grad_norm": 0.0005704418239510764,
+      "learning_rate": 2.2291442892946654e-06,
+      "loss": 0.0,
+      "step": 12412
+    },
+    {
+      "epoch": 0.7999613327318424,
+      "grad_norm": 0.01770676458062447,
+      "learning_rate": 2.2284282133906197e-06,
+      "loss": 0.0,
+      "step": 12413
+    },
+    {
+      "epoch": 0.8000257781787716,
+      "grad_norm": 0.00979534521749774,
+      "learning_rate": 2.2277121374865736e-06,
+      "loss": 0.0001,
+      "step": 12414
+    },
+    {
+      "epoch": 0.8000902236257008,
+      "grad_norm": 0.006045575204337346,
+      "learning_rate": 2.226996061582528e-06,
+      "loss": 0.0,
+      "step": 12415
+    },
+    {
+      "epoch": 0.80015466907263,
+      "grad_norm": 0.006182677580691062,
+      "learning_rate": 2.226279985678482e-06,
+      "loss": 0.0,
+      "step": 12416
+    },
+    {
+      "epoch": 0.8002191145195592,
+      "grad_norm": 0.01825002921528511,
+      "learning_rate": 2.2255639097744365e-06,
+      "loss": 0.0,
+      "step": 12417
+    },
+    {
+      "epoch": 0.8002835599664884,
+      "grad_norm": 0.11108596716148732,
+      "learning_rate": 2.2248478338703904e-06,
+      "loss": 0.0002,
+      "step": 12418
+    },
+    {
+      "epoch": 0.8003480054134176,
+      "grad_norm": 0.004849966220652724,
+      "learning_rate": 2.2241317579663447e-06,
+      "loss": 0.0,
+      "step": 12419
+    },
+    {
+      "epoch": 0.8004124508603467,
+      "grad_norm": 0.01844083369199926,
+      "learning_rate": 2.223415682062299e-06,
+      "loss": 0.0001,
+      "step": 12420
+    },
+    {
+      "epoch": 0.8004768963072759,
+      "grad_norm": 0.011662986322466401,
+      "learning_rate": 2.222699606158253e-06,
+      "loss": 0.0015,
+      "step": 12421
+    },
+    {
+      "epoch": 0.8005413417542051,
+      "grad_norm": 0.00012785641594670615,
+      "learning_rate": 2.221983530254207e-06,
+      "loss": 0.0,
+      "step": 12422
+    },
+    {
+      "epoch": 0.8006057872011343,
+      "grad_norm": 0.00016684387752677132,
+      "learning_rate": 2.2212674543501614e-06,
+      "loss": 0.0,
+      "step": 12423
+    },
+    {
+      "epoch": 0.8006702326480634,
+      "grad_norm": 0.00021000222066320482,
+      "learning_rate": 2.2205513784461157e-06,
+      "loss": 0.0,
+      "step": 12424
+    },
+    {
+      "epoch": 0.8007346780949925,
+      "grad_norm": 0.00046518345635586757,
+      "learning_rate": 2.2198353025420696e-06,
+      "loss": 0.0,
+      "step": 12425
+    },
+    {
+      "epoch": 0.8007991235419217,
+      "grad_norm": 0.18923775731011344,
+      "learning_rate": 2.219119226638024e-06,
+      "loss": 0.0003,
+      "step": 12426
+    },
+    {
+      "epoch": 0.8008635689888509,
+      "grad_norm": 0.007316504578402545,
+      "learning_rate": 2.2184031507339778e-06,
+      "loss": 0.0001,
+      "step": 12427
+    },
+    {
+      "epoch": 0.8009280144357801,
+      "grad_norm": 0.004194083563563775,
+      "learning_rate": 2.2176870748299325e-06,
+      "loss": 0.0,
+      "step": 12428
+    },
+    {
+      "epoch": 0.8009924598827093,
+      "grad_norm": 0.0009184694452434844,
+      "learning_rate": 2.2169709989258864e-06,
+      "loss": 0.0,
+      "step": 12429
+    },
+    {
+      "epoch": 0.8010569053296385,
+      "grad_norm": 0.02332515788220043,
+      "learning_rate": 2.2162549230218407e-06,
+      "loss": 0.0,
+      "step": 12430
+    },
+    {
+      "epoch": 0.8011213507765677,
+      "grad_norm": 0.031507976952465794,
+      "learning_rate": 2.2155388471177945e-06,
+      "loss": 0.0001,
+      "step": 12431
+    },
+    {
+      "epoch": 0.8011857962234968,
+      "grad_norm": 2.666919954509735,
+      "learning_rate": 2.214822771213749e-06,
+      "loss": 0.0356,
+      "step": 12432
+    },
+    {
+      "epoch": 0.801250241670426,
+      "grad_norm": 0.001449000874599054,
+      "learning_rate": 2.2141066953097027e-06,
+      "loss": 0.0,
+      "step": 12433
+    },
+    {
+      "epoch": 0.8013146871173552,
+      "grad_norm": 0.004870778020393193,
+      "learning_rate": 2.2133906194056574e-06,
+      "loss": 0.0,
+      "step": 12434
+    },
+    {
+      "epoch": 0.8013791325642843,
+      "grad_norm": 7.898662380784237e-05,
+      "learning_rate": 2.2126745435016113e-06,
+      "loss": 0.0,
+      "step": 12435
+    },
+    {
+      "epoch": 0.8014435780112135,
+      "grad_norm": 0.0002028904110712959,
+      "learning_rate": 2.2119584675975656e-06,
+      "loss": 0.0,
+      "step": 12436
+    },
+    {
+      "epoch": 0.8015080234581426,
+      "grad_norm": 0.005889722100197126,
+      "learning_rate": 2.2112423916935195e-06,
+      "loss": 0.0,
+      "step": 12437
+    },
+    {
+      "epoch": 0.8015724689050718,
+      "grad_norm": 0.0014307871393988698,
+      "learning_rate": 2.2105263157894738e-06,
+      "loss": 0.0,
+      "step": 12438
+    },
+    {
+      "epoch": 0.801636914352001,
+      "grad_norm": 0.0011634336621486382,
+      "learning_rate": 2.209810239885428e-06,
+      "loss": 0.0,
+      "step": 12439
+    },
+    {
+      "epoch": 0.8017013597989302,
+      "grad_norm": 0.020376505300973437,
+      "learning_rate": 2.2090941639813824e-06,
+      "loss": 0.0,
+      "step": 12440
+    },
+    {
+      "epoch": 0.8017658052458594,
+      "grad_norm": 0.003821548746330608,
+      "learning_rate": 2.2083780880773363e-06,
+      "loss": 0.0,
+      "step": 12441
+    },
+    {
+      "epoch": 0.8018302506927886,
+      "grad_norm": 0.0014602166162969207,
+      "learning_rate": 2.2076620121732906e-06,
+      "loss": 0.0,
+      "step": 12442
+    },
+    {
+      "epoch": 0.8018946961397178,
+      "grad_norm": 0.14062339066538232,
+      "learning_rate": 2.206945936269245e-06,
+      "loss": 0.0018,
+      "step": 12443
+    },
+    {
+      "epoch": 0.8019591415866469,
+      "grad_norm": 0.020081862810718113,
+      "learning_rate": 2.2062298603651987e-06,
+      "loss": 0.0,
+      "step": 12444
+    },
+    {
+      "epoch": 0.8020235870335761,
+      "grad_norm": 0.13397164659681082,
+      "learning_rate": 2.205513784461153e-06,
+      "loss": 0.0003,
+      "step": 12445
+    },
+    {
+      "epoch": 0.8020880324805052,
+      "grad_norm": 0.0022000253668111184,
+      "learning_rate": 2.2047977085571073e-06,
+      "loss": 0.0,
+      "step": 12446
+    },
+    {
+      "epoch": 0.8021524779274344,
+      "grad_norm": 0.006739471661244481,
+      "learning_rate": 2.2040816326530616e-06,
+      "loss": 0.0,
+      "step": 12447
+    },
+    {
+      "epoch": 0.8022169233743636,
+      "grad_norm": 0.007373562634504472,
+      "learning_rate": 2.2033655567490155e-06,
+      "loss": 0.0,
+      "step": 12448
+    },
+    {
+      "epoch": 0.8022813688212928,
+      "grad_norm": 0.056199122722132976,
+      "learning_rate": 2.20264948084497e-06,
+      "loss": 0.0001,
+      "step": 12449
+    },
+    {
+      "epoch": 0.8023458142682219,
+      "grad_norm": 0.16691897834307712,
+      "learning_rate": 2.2019334049409237e-06,
+      "loss": 0.002,
+      "step": 12450
+    },
+    {
+      "epoch": 0.8024102597151511,
+      "grad_norm": 0.00024640163647747016,
+      "learning_rate": 2.201217329036878e-06,
+      "loss": 0.0,
+      "step": 12451
+    },
+    {
+      "epoch": 0.8024747051620803,
+      "grad_norm": 0.013440612014857863,
+      "learning_rate": 2.2005012531328323e-06,
+      "loss": 0.0,
+      "step": 12452
+    },
+    {
+      "epoch": 0.8025391506090095,
+      "grad_norm": 0.006342392524851095,
+      "learning_rate": 2.1997851772287866e-06,
+      "loss": 0.0,
+      "step": 12453
+    },
+    {
+      "epoch": 0.8026035960559387,
+      "grad_norm": 0.00017604408165197298,
+      "learning_rate": 2.1990691013247404e-06,
+      "loss": 0.0,
+      "step": 12454
+    },
+    {
+      "epoch": 0.8026680415028679,
+      "grad_norm": 0.0022245882873562933,
+      "learning_rate": 2.1983530254206947e-06,
+      "loss": 0.0,
+      "step": 12455
+    },
+    {
+      "epoch": 0.802732486949797,
+      "grad_norm": 5.712391774772362e-05,
+      "learning_rate": 2.197636949516649e-06,
+      "loss": 0.0,
+      "step": 12456
+    },
+    {
+      "epoch": 0.8027969323967262,
+      "grad_norm": 0.0006953470399971264,
+      "learning_rate": 2.196920873612603e-06,
+      "loss": 0.0,
+      "step": 12457
+    },
+    {
+      "epoch": 0.8028613778436553,
+      "grad_norm": 0.0016720927108341782,
+      "learning_rate": 2.1962047977085572e-06,
+      "loss": 0.0,
+      "step": 12458
+    },
+    {
+      "epoch": 0.8029258232905845,
+      "grad_norm": 0.00015475809128655122,
+      "learning_rate": 2.1954887218045115e-06,
+      "loss": 0.0,
+      "step": 12459
+    },
+    {
+      "epoch": 0.8029902687375137,
+      "grad_norm": 0.0013262927475274097,
+      "learning_rate": 2.194772645900466e-06,
+      "loss": 0.0,
+      "step": 12460
+    },
+    {
+      "epoch": 0.8030547141844429,
+      "grad_norm": 0.0011045115282342043,
+      "learning_rate": 2.1940565699964197e-06,
+      "loss": 0.0,
+      "step": 12461
+    },
+    {
+      "epoch": 0.803119159631372,
+      "grad_norm": 0.0021931957013456202,
+      "learning_rate": 2.193340494092374e-06,
+      "loss": 0.0,
+      "step": 12462
+    },
+    {
+      "epoch": 0.8031836050783012,
+      "grad_norm": 5.5322073605114676e-05,
+      "learning_rate": 2.1926244181883283e-06,
+      "loss": 0.0,
+      "step": 12463
+    },
+    {
+      "epoch": 0.8032480505252304,
+      "grad_norm": 0.18646149330729211,
+      "learning_rate": 2.1919083422842826e-06,
+      "loss": 0.0006,
+      "step": 12464
+    },
+    {
+      "epoch": 0.8033124959721596,
+      "grad_norm": 0.00282455793739817,
+      "learning_rate": 2.1911922663802365e-06,
+      "loss": 0.0,
+      "step": 12465
+    },
+    {
+      "epoch": 0.8033769414190888,
+      "grad_norm": 0.004386932538795974,
+      "learning_rate": 2.1904761904761908e-06,
+      "loss": 0.0,
+      "step": 12466
+    },
+    {
+      "epoch": 0.803441386866018,
+      "grad_norm": 1.3727636621027356e-05,
+      "learning_rate": 2.1897601145721446e-06,
+      "loss": 0.0,
+      "step": 12467
+    },
+    {
+      "epoch": 0.8035058323129471,
+      "grad_norm": 0.00011849457349212985,
+      "learning_rate": 2.189044038668099e-06,
+      "loss": 0.0,
+      "step": 12468
+    },
+    {
+      "epoch": 0.8035702777598762,
+      "grad_norm": 0.0011204254283753746,
+      "learning_rate": 2.1883279627640532e-06,
+      "loss": 0.0,
+      "step": 12469
+    },
+    {
+      "epoch": 0.8036347232068054,
+      "grad_norm": 0.005514206749045671,
+      "learning_rate": 2.1876118868600075e-06,
+      "loss": 0.0,
+      "step": 12470
+    },
+    {
+      "epoch": 0.8036991686537346,
+      "grad_norm": 0.002547410752228369,
+      "learning_rate": 2.1868958109559614e-06,
+      "loss": 0.0,
+      "step": 12471
+    },
+    {
+      "epoch": 0.8037636141006638,
+      "grad_norm": 0.004749226856846408,
+      "learning_rate": 2.1861797350519157e-06,
+      "loss": 0.0,
+      "step": 12472
+    },
+    {
+      "epoch": 0.803828059547593,
+      "grad_norm": 0.008761383470593858,
+      "learning_rate": 2.1854636591478696e-06,
+      "loss": 0.0,
+      "step": 12473
+    },
+    {
+      "epoch": 0.8038925049945221,
+      "grad_norm": 0.0002561698488996513,
+      "learning_rate": 2.184747583243824e-06,
+      "loss": 0.0,
+      "step": 12474
+    },
+    {
+      "epoch": 0.8039569504414513,
+      "grad_norm": 0.004750373251371964,
+      "learning_rate": 2.184031507339778e-06,
+      "loss": 0.0,
+      "step": 12475
+    },
+    {
+      "epoch": 0.8040213958883805,
+      "grad_norm": 0.00010574556100120908,
+      "learning_rate": 2.1833154314357325e-06,
+      "loss": 0.0,
+      "step": 12476
+    },
+    {
+      "epoch": 0.8040858413353097,
+      "grad_norm": 0.312174436736941,
+      "learning_rate": 2.1825993555316864e-06,
+      "loss": 0.0056,
+      "step": 12477
+    },
+    {
+      "epoch": 0.8041502867822389,
+      "grad_norm": 0.023472666434982383,
+      "learning_rate": 2.1818832796276407e-06,
+      "loss": 0.0002,
+      "step": 12478
+    },
+    {
+      "epoch": 0.8042147322291681,
+      "grad_norm": 0.0014214595665206025,
+      "learning_rate": 2.181167203723595e-06,
+      "loss": 0.0,
+      "step": 12479
+    },
+    {
+      "epoch": 0.8042791776760971,
+      "grad_norm": 0.004601320748266051,
+      "learning_rate": 2.180451127819549e-06,
+      "loss": 0.0,
+      "step": 12480
+    },
+    {
+      "epoch": 0.8043436231230263,
+      "grad_norm": 0.0004118759793522063,
+      "learning_rate": 2.179735051915503e-06,
+      "loss": 0.0,
+      "step": 12481
+    },
+    {
+      "epoch": 0.8044080685699555,
+      "grad_norm": 0.007549649471261577,
+      "learning_rate": 2.1790189760114574e-06,
+      "loss": 0.0001,
+      "step": 12482
+    },
+    {
+      "epoch": 0.8044725140168847,
+      "grad_norm": 0.0036913314331920907,
+      "learning_rate": 2.1783029001074117e-06,
+      "loss": 0.0,
+      "step": 12483
+    },
+    {
+      "epoch": 0.8045369594638139,
+      "grad_norm": 0.2587022743807844,
+      "learning_rate": 2.1775868242033656e-06,
+      "loss": 0.0009,
+      "step": 12484
+    },
+    {
+      "epoch": 0.8046014049107431,
+      "grad_norm": 0.00025450952995651784,
+      "learning_rate": 2.17687074829932e-06,
+      "loss": 0.0,
+      "step": 12485
+    },
+    {
+      "epoch": 0.8046658503576722,
+      "grad_norm": 0.7104865939347345,
+      "learning_rate": 2.1761546723952738e-06,
+      "loss": 0.002,
+      "step": 12486
+    },
+    {
+      "epoch": 0.8047302958046014,
+      "grad_norm": 0.004899011875170007,
+      "learning_rate": 2.1754385964912285e-06,
+      "loss": 0.0,
+      "step": 12487
+    },
+    {
+      "epoch": 0.8047947412515306,
+      "grad_norm": 0.0022642938689466183,
+      "learning_rate": 2.1747225205871824e-06,
+      "loss": 0.0,
+      "step": 12488
+    },
+    {
+      "epoch": 0.8048591866984598,
+      "grad_norm": 0.0004189518177420199,
+      "learning_rate": 2.1740064446831367e-06,
+      "loss": 0.0,
+      "step": 12489
+    },
+    {
+      "epoch": 0.804923632145389,
+      "grad_norm": 2.2536943495337073e-05,
+      "learning_rate": 2.1732903687790905e-06,
+      "loss": 0.0,
+      "step": 12490
+    },
+    {
+      "epoch": 0.804988077592318,
+      "grad_norm": 0.0035356629953806427,
+      "learning_rate": 2.172574292875045e-06,
+      "loss": 0.0,
+      "step": 12491
+    },
+    {
+      "epoch": 0.8050525230392472,
+      "grad_norm": 0.4910174343224499,
+      "learning_rate": 2.171858216970999e-06,
+      "loss": 0.001,
+      "step": 12492
+    },
+    {
+      "epoch": 0.8051169684861764,
+      "grad_norm": 0.0033785242190478037,
+      "learning_rate": 2.1711421410669534e-06,
+      "loss": 0.0,
+      "step": 12493
+    },
+    {
+      "epoch": 0.8051814139331056,
+      "grad_norm": 0.00046202312693729595,
+      "learning_rate": 2.1704260651629073e-06,
+      "loss": 0.0,
+      "step": 12494
+    },
+    {
+      "epoch": 0.8052458593800348,
+      "grad_norm": 0.0003226347943040989,
+      "learning_rate": 2.1697099892588616e-06,
+      "loss": 0.0,
+      "step": 12495
+    },
+    {
+      "epoch": 0.805310304826964,
+      "grad_norm": 0.00031251158052698306,
+      "learning_rate": 2.168993913354816e-06,
+      "loss": 0.0,
+      "step": 12496
+    },
+    {
+      "epoch": 0.8053747502738932,
+      "grad_norm": 0.023313700776633278,
+      "learning_rate": 2.1682778374507698e-06,
+      "loss": 0.0,
+      "step": 12497
+    },
+    {
+      "epoch": 0.8054391957208223,
+      "grad_norm": 0.01404507452701196,
+      "learning_rate": 2.167561761546724e-06,
+      "loss": 0.0,
+      "step": 12498
+    },
+    {
+      "epoch": 0.8055036411677515,
+      "grad_norm": 0.00015273987081015428,
+      "learning_rate": 2.1668456856426784e-06,
+      "loss": 0.0,
+      "step": 12499
+    },
+    {
+      "epoch": 0.8055680866146807,
+      "grad_norm": 0.002063542098063826,
+      "learning_rate": 2.1661296097386327e-06,
+      "loss": 0.0,
+      "step": 12500
+    },
+    {
+      "epoch": 0.8056325320616099,
+      "grad_norm": 0.0011278421061533598,
+      "learning_rate": 2.1654135338345866e-06,
+      "loss": 0.0,
+      "step": 12501
+    },
+    {
+      "epoch": 0.805696977508539,
+      "grad_norm": 0.006941577005496535,
+      "learning_rate": 2.164697457930541e-06,
+      "loss": 0.0,
+      "step": 12502
+    },
+    {
+      "epoch": 0.8057614229554682,
+      "grad_norm": 0.0012370293237017856,
+      "learning_rate": 2.1639813820264947e-06,
+      "loss": 0.0,
+      "step": 12503
+    },
+    {
+      "epoch": 0.8058258684023973,
+      "grad_norm": 0.017707468873228566,
+      "learning_rate": 2.1632653061224495e-06,
+      "loss": 0.0002,
+      "step": 12504
+    },
+    {
+      "epoch": 0.8058903138493265,
+      "grad_norm": 0.00018853056554883304,
+      "learning_rate": 2.1625492302184033e-06,
+      "loss": 0.0,
+      "step": 12505
+    },
+    {
+      "epoch": 0.8059547592962557,
+      "grad_norm": 0.004347717010727028,
+      "learning_rate": 2.1618331543143576e-06,
+      "loss": 0.0,
+      "step": 12506
+    },
+    {
+      "epoch": 0.8060192047431849,
+      "grad_norm": 0.0013038376412325182,
+      "learning_rate": 2.1611170784103115e-06,
+      "loss": 0.0,
+      "step": 12507
+    },
+    {
+      "epoch": 0.8060836501901141,
+      "grad_norm": 0.09628063265594201,
+      "learning_rate": 2.160401002506266e-06,
+      "loss": 0.0002,
+      "step": 12508
+    },
+    {
+      "epoch": 0.8061480956370433,
+      "grad_norm": 0.0007463655062723487,
+      "learning_rate": 2.15968492660222e-06,
+      "loss": 0.0,
+      "step": 12509
+    },
+    {
+      "epoch": 0.8062125410839724,
+      "grad_norm": 0.00048655696128334684,
+      "learning_rate": 2.1589688506981744e-06,
+      "loss": 0.0,
+      "step": 12510
+    },
+    {
+      "epoch": 0.8062769865309016,
+      "grad_norm": 0.005872971745491133,
+      "learning_rate": 2.1582527747941283e-06,
+      "loss": 0.0,
+      "step": 12511
+    },
+    {
+      "epoch": 0.8063414319778308,
+      "grad_norm": 0.002717922444481002,
+      "learning_rate": 2.1575366988900826e-06,
+      "loss": 0.0,
+      "step": 12512
+    },
+    {
+      "epoch": 0.8064058774247599,
+      "grad_norm": 0.014760318307408935,
+      "learning_rate": 2.156820622986037e-06,
+      "loss": 0.0,
+      "step": 12513
+    },
+    {
+      "epoch": 0.8064703228716891,
+      "grad_norm": 7.437945918253065e-05,
+      "learning_rate": 2.1561045470819907e-06,
+      "loss": 0.0,
+      "step": 12514
+    },
+    {
+      "epoch": 0.8065347683186183,
+      "grad_norm": 0.0002872171139769259,
+      "learning_rate": 2.155388471177945e-06,
+      "loss": 0.0,
+      "step": 12515
+    },
+    {
+      "epoch": 0.8065992137655474,
+      "grad_norm": 0.4398679174420515,
+      "learning_rate": 2.1546723952738993e-06,
+      "loss": 0.0032,
+      "step": 12516
+    },
+    {
+      "epoch": 0.8066636592124766,
+      "grad_norm": 0.0012609922360681577,
+      "learning_rate": 2.1539563193698536e-06,
+      "loss": 0.0,
+      "step": 12517
+    },
+    {
+      "epoch": 0.8067281046594058,
+      "grad_norm": 0.00022379636048257173,
+      "learning_rate": 2.1532402434658075e-06,
+      "loss": 0.0,
+      "step": 12518
+    },
+    {
+      "epoch": 0.806792550106335,
+      "grad_norm": 0.0017465105291618174,
+      "learning_rate": 2.152524167561762e-06,
+      "loss": 0.0,
+      "step": 12519
+    },
+    {
+      "epoch": 0.8068569955532642,
+      "grad_norm": 0.00038133249943826756,
+      "learning_rate": 2.1518080916577157e-06,
+      "loss": 0.0,
+      "step": 12520
+    },
+    {
+      "epoch": 0.8069214410001934,
+      "grad_norm": 0.00132960592924005,
+      "learning_rate": 2.15109201575367e-06,
+      "loss": 0.0,
+      "step": 12521
+    },
+    {
+      "epoch": 0.8069858864471225,
+      "grad_norm": 0.017596323143278837,
+      "learning_rate": 2.1503759398496243e-06,
+      "loss": 0.0,
+      "step": 12522
+    },
+    {
+      "epoch": 0.8070503318940517,
+      "grad_norm": 0.002651803633894323,
+      "learning_rate": 2.1496598639455786e-06,
+      "loss": 0.0,
+      "step": 12523
+    },
+    {
+      "epoch": 0.8071147773409808,
+      "grad_norm": 0.0018630916251092704,
+      "learning_rate": 2.1489437880415325e-06,
+      "loss": 0.0,
+      "step": 12524
+    },
+    {
+      "epoch": 0.80717922278791,
+      "grad_norm": 0.00027101958664413014,
+      "learning_rate": 2.1482277121374868e-06,
+      "loss": 0.0,
+      "step": 12525
+    },
+    {
+      "epoch": 0.8072436682348392,
+      "grad_norm": 0.014118955231454783,
+      "learning_rate": 2.1475116362334406e-06,
+      "loss": 0.0,
+      "step": 12526
+    },
+    {
+      "epoch": 0.8073081136817684,
+      "grad_norm": 0.0004878170706532088,
+      "learning_rate": 2.146795560329395e-06,
+      "loss": 0.0,
+      "step": 12527
+    },
+    {
+      "epoch": 0.8073725591286975,
+      "grad_norm": 0.7035151141142062,
+      "learning_rate": 2.1460794844253492e-06,
+      "loss": 0.0207,
+      "step": 12528
+    },
+    {
+      "epoch": 0.8074370045756267,
+      "grad_norm": 0.00014633861646761186,
+      "learning_rate": 2.1453634085213035e-06,
+      "loss": 0.0,
+      "step": 12529
+    },
+    {
+      "epoch": 0.8075014500225559,
+      "grad_norm": 0.0007063263377159374,
+      "learning_rate": 2.1446473326172574e-06,
+      "loss": 0.0,
+      "step": 12530
+    },
+    {
+      "epoch": 0.8075658954694851,
+      "grad_norm": 0.031044198535053354,
+      "learning_rate": 2.1439312567132117e-06,
+      "loss": 0.0002,
+      "step": 12531
+    },
+    {
+      "epoch": 0.8076303409164143,
+      "grad_norm": 0.1859323512895166,
+      "learning_rate": 2.143215180809166e-06,
+      "loss": 0.0007,
+      "step": 12532
+    },
+    {
+      "epoch": 0.8076947863633435,
+      "grad_norm": 0.003519137570280147,
+      "learning_rate": 2.1424991049051203e-06,
+      "loss": 0.0,
+      "step": 12533
+    },
+    {
+      "epoch": 0.8077592318102726,
+      "grad_norm": 0.007856042322162549,
+      "learning_rate": 2.141783029001074e-06,
+      "loss": 0.0,
+      "step": 12534
+    },
+    {
+      "epoch": 0.8078236772572018,
+      "grad_norm": 0.00011564863298519748,
+      "learning_rate": 2.1410669530970285e-06,
+      "loss": 0.0,
+      "step": 12535
+    },
+    {
+      "epoch": 0.8078881227041309,
+      "grad_norm": 0.006344745271119972,
+      "learning_rate": 2.1403508771929828e-06,
+      "loss": 0.0,
+      "step": 12536
+    },
+    {
+      "epoch": 0.8079525681510601,
+      "grad_norm": 0.033851168725958655,
+      "learning_rate": 2.1396348012889367e-06,
+      "loss": 0.0001,
+      "step": 12537
+    },
+    {
+      "epoch": 0.8080170135979893,
+      "grad_norm": 0.00015541963414884185,
+      "learning_rate": 2.138918725384891e-06,
+      "loss": 0.0,
+      "step": 12538
+    },
+    {
+      "epoch": 0.8080814590449185,
+      "grad_norm": 0.007355932033656503,
+      "learning_rate": 2.1382026494808452e-06,
+      "loss": 0.0,
+      "step": 12539
+    },
+    {
+      "epoch": 0.8081459044918476,
+      "grad_norm": 0.0014587984007397436,
+      "learning_rate": 2.1374865735767995e-06,
+      "loss": 0.0,
+      "step": 12540
+    },
+    {
+      "epoch": 0.8082103499387768,
+      "grad_norm": 0.3055763484401704,
+      "learning_rate": 2.1367704976727534e-06,
+      "loss": 0.0009,
+      "step": 12541
+    },
+    {
+      "epoch": 0.808274795385706,
+      "grad_norm": 0.015262446442021883,
+      "learning_rate": 2.1360544217687077e-06,
+      "loss": 0.0,
+      "step": 12542
+    },
+    {
+      "epoch": 0.8083392408326352,
+      "grad_norm": 0.007324747722120583,
+      "learning_rate": 2.1353383458646616e-06,
+      "loss": 0.0001,
+      "step": 12543
+    },
+    {
+      "epoch": 0.8084036862795644,
+      "grad_norm": 0.03268571168614651,
+      "learning_rate": 2.134622269960616e-06,
+      "loss": 0.0003,
+      "step": 12544
+    },
+    {
+      "epoch": 0.8084681317264936,
+      "grad_norm": 0.002548795640975255,
+      "learning_rate": 2.13390619405657e-06,
+      "loss": 0.0,
+      "step": 12545
+    },
+    {
+      "epoch": 0.8085325771734228,
+      "grad_norm": 0.0005647212933699163,
+      "learning_rate": 2.1331901181525245e-06,
+      "loss": 0.0,
+      "step": 12546
+    },
+    {
+      "epoch": 0.8085970226203518,
+      "grad_norm": 0.011028547755694413,
+      "learning_rate": 2.1324740422484784e-06,
+      "loss": 0.0001,
+      "step": 12547
+    },
+    {
+      "epoch": 0.808661468067281,
+      "grad_norm": 0.0016098602253972305,
+      "learning_rate": 2.1317579663444327e-06,
+      "loss": 0.0,
+      "step": 12548
+    },
+    {
+      "epoch": 0.8087259135142102,
+      "grad_norm": 0.002141363180888285,
+      "learning_rate": 2.131041890440387e-06,
+      "loss": 0.0,
+      "step": 12549
+    },
+    {
+      "epoch": 0.8087903589611394,
+      "grad_norm": 0.009605867876297334,
+      "learning_rate": 2.130325814536341e-06,
+      "loss": 0.0,
+      "step": 12550
+    },
+    {
+      "epoch": 0.8088548044080686,
+      "grad_norm": 0.012447253129473378,
+      "learning_rate": 2.129609738632295e-06,
+      "loss": 0.0,
+      "step": 12551
+    },
+    {
+      "epoch": 0.8089192498549977,
+      "grad_norm": 0.0005868374143533707,
+      "learning_rate": 2.1288936627282494e-06,
+      "loss": 0.0,
+      "step": 12552
+    },
+    {
+      "epoch": 0.8089836953019269,
+      "grad_norm": 0.005273947446889605,
+      "learning_rate": 2.1281775868242037e-06,
+      "loss": 0.0,
+      "step": 12553
+    },
+    {
+      "epoch": 0.8090481407488561,
+      "grad_norm": 0.0002057675618183742,
+      "learning_rate": 2.1274615109201576e-06,
+      "loss": 0.0,
+      "step": 12554
+    },
+    {
+      "epoch": 0.8091125861957853,
+      "grad_norm": 0.00044356842756500183,
+      "learning_rate": 2.126745435016112e-06,
+      "loss": 0.0,
+      "step": 12555
+    },
+    {
+      "epoch": 0.8091770316427145,
+      "grad_norm": 0.0005960369586189239,
+      "learning_rate": 2.1260293591120658e-06,
+      "loss": 0.0,
+      "step": 12556
+    },
+    {
+      "epoch": 0.8092414770896437,
+      "grad_norm": 0.0005960369586189239,
+      "learning_rate": 2.1260293591120658e-06,
+      "loss": 0.0065,
+      "step": 12557
+    },
+    {
+      "epoch": 0.8093059225365727,
+      "grad_norm": 0.0009959009747822732,
+      "learning_rate": 2.1253132832080205e-06,
+      "loss": 0.0,
+      "step": 12558
+    },
+    {
+      "epoch": 0.8093703679835019,
+      "grad_norm": 0.001079138957866863,
+      "learning_rate": 2.1245972073039744e-06,
+      "loss": 0.0,
+      "step": 12559
+    },
+    {
+      "epoch": 0.8094348134304311,
+      "grad_norm": 0.0001700052524290455,
+      "learning_rate": 2.1238811313999287e-06,
+      "loss": 0.0,
+      "step": 12560
+    },
+    {
+      "epoch": 0.8094992588773603,
+      "grad_norm": 0.3267133866120286,
+      "learning_rate": 2.1231650554958826e-06,
+      "loss": 0.0021,
+      "step": 12561
+    },
+    {
+      "epoch": 0.8095637043242895,
+      "grad_norm": 0.026077396622425923,
+      "learning_rate": 2.122448979591837e-06,
+      "loss": 0.0015,
+      "step": 12562
+    },
+    {
+      "epoch": 0.8096281497712187,
+      "grad_norm": 0.004266359877687984,
+      "learning_rate": 2.121732903687791e-06,
+      "loss": 0.0,
+      "step": 12563
+    },
+    {
+      "epoch": 0.8096925952181478,
+      "grad_norm": 0.0020029363056970465,
+      "learning_rate": 2.1210168277837455e-06,
+      "loss": 0.0,
+      "step": 12564
+    },
+    {
+      "epoch": 0.809757040665077,
+      "grad_norm": 0.0022158301051287744,
+      "learning_rate": 2.1203007518796993e-06,
+      "loss": 0.0,
+      "step": 12565
+    },
+    {
+      "epoch": 0.8098214861120062,
+      "grad_norm": 0.02784966132114145,
+      "learning_rate": 2.1195846759756536e-06,
+      "loss": 0.0,
+      "step": 12566
+    },
+    {
+      "epoch": 0.8098859315589354,
+      "grad_norm": 0.003321398332227599,
+      "learning_rate": 2.118868600071608e-06,
+      "loss": 0.0,
+      "step": 12567
+    },
+    {
+      "epoch": 0.8099503770058646,
+      "grad_norm": 0.0006111349772194386,
+      "learning_rate": 2.118152524167562e-06,
+      "loss": 0.0,
+      "step": 12568
+    },
+    {
+      "epoch": 0.8100148224527937,
+      "grad_norm": 8.239420264536209e-05,
+      "learning_rate": 2.117436448263516e-06,
+      "loss": 0.0,
+      "step": 12569
+    },
+    {
+      "epoch": 0.8100792678997228,
+      "grad_norm": 0.011744492043335403,
+      "learning_rate": 2.1167203723594704e-06,
+      "loss": 0.0001,
+      "step": 12570
+    },
+    {
+      "epoch": 0.810143713346652,
+      "grad_norm": 0.00613206029830229,
+      "learning_rate": 2.1160042964554247e-06,
+      "loss": 0.0,
+      "step": 12571
+    },
+    {
+      "epoch": 0.8102081587935812,
+      "grad_norm": 0.0308008172561735,
+      "learning_rate": 2.1152882205513786e-06,
+      "loss": 0.0003,
+      "step": 12572
+    },
+    {
+      "epoch": 0.8102726042405104,
+      "grad_norm": 0.010741848493233576,
+      "learning_rate": 2.114572144647333e-06,
+      "loss": 0.0,
+      "step": 12573
+    },
+    {
+      "epoch": 0.8103370496874396,
+      "grad_norm": 0.0015504692910859963,
+      "learning_rate": 2.1138560687432867e-06,
+      "loss": 0.0,
+      "step": 12574
+    },
+    {
+      "epoch": 0.8104014951343688,
+      "grad_norm": 0.006183430411555323,
+      "learning_rate": 2.1131399928392415e-06,
+      "loss": 0.0,
+      "step": 12575
+    },
+    {
+      "epoch": 0.810465940581298,
+      "grad_norm": 0.00024011503847110936,
+      "learning_rate": 2.1124239169351953e-06,
+      "loss": 0.0,
+      "step": 12576
+    },
+    {
+      "epoch": 0.8105303860282271,
+      "grad_norm": 0.0013300068832935133,
+      "learning_rate": 2.1117078410311496e-06,
+      "loss": 0.0,
+      "step": 12577
+    },
+    {
+      "epoch": 0.8105948314751563,
+      "grad_norm": 0.0001257257657714678,
+      "learning_rate": 2.1109917651271035e-06,
+      "loss": 0.0,
+      "step": 12578
+    },
+    {
+      "epoch": 0.8106592769220855,
+      "grad_norm": 0.0012701698727820548,
+      "learning_rate": 2.110275689223058e-06,
+      "loss": 0.0,
+      "step": 12579
+    },
+    {
+      "epoch": 0.8107237223690146,
+      "grad_norm": 0.002781898222436216,
+      "learning_rate": 2.1095596133190117e-06,
+      "loss": 0.0,
+      "step": 12580
+    },
+    {
+      "epoch": 0.8107881678159438,
+      "grad_norm": 0.008623381992397053,
+      "learning_rate": 2.1088435374149664e-06,
+      "loss": 0.0,
+      "step": 12581
+    },
+    {
+      "epoch": 0.810852613262873,
+      "grad_norm": 0.0035804817932016133,
+      "learning_rate": 2.1081274615109203e-06,
+      "loss": 0.0,
+      "step": 12582
+    },
+    {
+      "epoch": 0.8109170587098021,
+      "grad_norm": 0.0004178044112188488,
+      "learning_rate": 2.1074113856068746e-06,
+      "loss": 0.0,
+      "step": 12583
+    },
+    {
+      "epoch": 0.8109815041567313,
+      "grad_norm": 0.20006863377531295,
+      "learning_rate": 2.1066953097028285e-06,
+      "loss": 0.0039,
+      "step": 12584
+    },
+    {
+      "epoch": 0.8110459496036605,
+      "grad_norm": 0.023746772059720624,
+      "learning_rate": 2.1059792337987828e-06,
+      "loss": 0.0,
+      "step": 12585
+    },
+    {
+      "epoch": 0.8111103950505897,
+      "grad_norm": 0.00017902644761458916,
+      "learning_rate": 2.105263157894737e-06,
+      "loss": 0.0,
+      "step": 12586
+    },
+    {
+      "epoch": 0.8111748404975189,
+      "grad_norm": 0.0015365147211049636,
+      "learning_rate": 2.1045470819906914e-06,
+      "loss": 0.0,
+      "step": 12587
+    },
+    {
+      "epoch": 0.811239285944448,
+      "grad_norm": 0.03381551467867641,
+      "learning_rate": 2.1038310060866452e-06,
+      "loss": 0.0002,
+      "step": 12588
+    },
+    {
+      "epoch": 0.8113037313913772,
+      "grad_norm": 0.01445457605861845,
+      "learning_rate": 2.1031149301825995e-06,
+      "loss": 0.0001,
+      "step": 12589
+    },
+    {
+      "epoch": 0.8113681768383064,
+      "grad_norm": 0.0031470428802809754,
+      "learning_rate": 2.102398854278554e-06,
+      "loss": 0.0,
+      "step": 12590
+    },
+    {
+      "epoch": 0.8114326222852355,
+      "grad_norm": 0.47407611668946104,
+      "learning_rate": 2.1016827783745077e-06,
+      "loss": 0.0027,
+      "step": 12591
+    },
+    {
+      "epoch": 0.8114970677321647,
+      "grad_norm": 0.001997918103711385,
+      "learning_rate": 2.100966702470462e-06,
+      "loss": 0.0,
+      "step": 12592
+    },
+    {
+      "epoch": 0.8115615131790939,
+      "grad_norm": 0.00022226182021700908,
+      "learning_rate": 2.1002506265664163e-06,
+      "loss": 0.0,
+      "step": 12593
+    },
+    {
+      "epoch": 0.811625958626023,
+      "grad_norm": 0.005495734573647927,
+      "learning_rate": 2.0995345506623706e-06,
+      "loss": 0.0,
+      "step": 12594
+    },
+    {
+      "epoch": 0.8116904040729522,
+      "grad_norm": 0.36579584629506817,
+      "learning_rate": 2.0988184747583245e-06,
+      "loss": 0.001,
+      "step": 12595
+    },
+    {
+      "epoch": 0.8117548495198814,
+      "grad_norm": 0.0017201286993632876,
+      "learning_rate": 2.0981023988542788e-06,
+      "loss": 0.0,
+      "step": 12596
+    },
+    {
+      "epoch": 0.8118192949668106,
+      "grad_norm": 0.0012654749260697195,
+      "learning_rate": 2.0973863229502327e-06,
+      "loss": 0.0,
+      "step": 12597
+    },
+    {
+      "epoch": 0.8118837404137398,
+      "grad_norm": 6.77314804816197e-05,
+      "learning_rate": 2.096670247046187e-06,
+      "loss": 0.0,
+      "step": 12598
+    },
+    {
+      "epoch": 0.811948185860669,
+      "grad_norm": 0.12170673011688415,
+      "learning_rate": 2.0959541711421412e-06,
+      "loss": 0.0003,
+      "step": 12599
+    },
+    {
+      "epoch": 0.8120126313075982,
+      "grad_norm": 0.00039734094272957175,
+      "learning_rate": 2.0952380952380955e-06,
+      "loss": 0.0,
+      "step": 12600
+    },
+    {
+      "epoch": 0.8120770767545273,
+      "grad_norm": 0.0040080291412414185,
+      "learning_rate": 2.0945220193340494e-06,
+      "loss": 0.0,
+      "step": 12601
+    },
+    {
+      "epoch": 0.8121415222014564,
+      "grad_norm": 0.011901667177716067,
+      "learning_rate": 2.0938059434300037e-06,
+      "loss": 0.0,
+      "step": 12602
+    },
+    {
+      "epoch": 0.8122059676483856,
+      "grad_norm": 0.27331548420750906,
+      "learning_rate": 2.093089867525958e-06,
+      "loss": 0.0013,
+      "step": 12603
+    },
+    {
+      "epoch": 0.8122704130953148,
+      "grad_norm": 0.004315035369221101,
+      "learning_rate": 2.0923737916219123e-06,
+      "loss": 0.0,
+      "step": 12604
+    },
+    {
+      "epoch": 0.812334858542244,
+      "grad_norm": 1.478788977076481e-05,
+      "learning_rate": 2.091657715717866e-06,
+      "loss": 0.0,
+      "step": 12605
+    },
+    {
+      "epoch": 0.8123993039891731,
+      "grad_norm": 0.018809907453625072,
+      "learning_rate": 2.0909416398138205e-06,
+      "loss": 0.0001,
+      "step": 12606
+    },
+    {
+      "epoch": 0.8124637494361023,
+      "grad_norm": 0.0015582649472403046,
+      "learning_rate": 2.090225563909775e-06,
+      "loss": 0.0,
+      "step": 12607
+    },
+    {
+      "epoch": 0.8125281948830315,
+      "grad_norm": 0.000712515429465975,
+      "learning_rate": 2.0895094880057287e-06,
+      "loss": 0.0,
+      "step": 12608
+    },
+    {
+      "epoch": 0.8125926403299607,
+      "grad_norm": 0.00017508893570977122,
+      "learning_rate": 2.088793412101683e-06,
+      "loss": 0.0,
+      "step": 12609
+    },
+    {
+      "epoch": 0.8126570857768899,
+      "grad_norm": 0.0010479845875821226,
+      "learning_rate": 2.0880773361976373e-06,
+      "loss": 0.0,
+      "step": 12610
+    },
+    {
+      "epoch": 0.8127215312238191,
+      "grad_norm": 0.001556306928741489,
+      "learning_rate": 2.0873612602935916e-06,
+      "loss": 0.0,
+      "step": 12611
+    },
+    {
+      "epoch": 0.8127859766707483,
+      "grad_norm": 0.006236001901142749,
+      "learning_rate": 2.0866451843895454e-06,
+      "loss": 0.0001,
+      "step": 12612
+    },
+    {
+      "epoch": 0.8128504221176774,
+      "grad_norm": 0.0030114578149282394,
+      "learning_rate": 2.0859291084854997e-06,
+      "loss": 0.0,
+      "step": 12613
+    },
+    {
+      "epoch": 0.8129148675646065,
+      "grad_norm": 0.011364972960464367,
+      "learning_rate": 2.0852130325814536e-06,
+      "loss": 0.0,
+      "step": 12614
+    },
+    {
+      "epoch": 0.8129793130115357,
+      "grad_norm": 0.005280251572317013,
+      "learning_rate": 2.084496956677408e-06,
+      "loss": 0.0,
+      "step": 12615
+    },
+    {
+      "epoch": 0.8130437584584649,
+      "grad_norm": 8.507760533251935e-05,
+      "learning_rate": 2.083780880773362e-06,
+      "loss": 0.0,
+      "step": 12616
+    },
+    {
+      "epoch": 0.8131082039053941,
+      "grad_norm": 0.00038017094694071246,
+      "learning_rate": 2.0830648048693165e-06,
+      "loss": 0.0,
+      "step": 12617
+    },
+    {
+      "epoch": 0.8131726493523233,
+      "grad_norm": 0.2922925010749905,
+      "learning_rate": 2.0823487289652704e-06,
+      "loss": 0.0007,
+      "step": 12618
+    },
+    {
+      "epoch": 0.8132370947992524,
+      "grad_norm": 0.23094773223415585,
+      "learning_rate": 2.0816326530612247e-06,
+      "loss": 0.0004,
+      "step": 12619
+    },
+    {
+      "epoch": 0.8133015402461816,
+      "grad_norm": 0.5233092933901883,
+      "learning_rate": 2.0809165771571786e-06,
+      "loss": 0.001,
+      "step": 12620
+    },
+    {
+      "epoch": 0.8133659856931108,
+      "grad_norm": 0.00033577295685188345,
+      "learning_rate": 2.080200501253133e-06,
+      "loss": 0.0,
+      "step": 12621
+    },
+    {
+      "epoch": 0.81343043114004,
+      "grad_norm": 0.03787885973590073,
+      "learning_rate": 2.079484425349087e-06,
+      "loss": 0.0001,
+      "step": 12622
+    },
+    {
+      "epoch": 0.8134948765869692,
+      "grad_norm": 0.00023672040014827814,
+      "learning_rate": 2.0787683494450415e-06,
+      "loss": 0.0,
+      "step": 12623
+    },
+    {
+      "epoch": 0.8135593220338984,
+      "grad_norm": 0.0004785405222872389,
+      "learning_rate": 2.0780522735409953e-06,
+      "loss": 0.0,
+      "step": 12624
+    },
+    {
+      "epoch": 0.8136237674808274,
+      "grad_norm": 0.002556479343831401,
+      "learning_rate": 2.0773361976369496e-06,
+      "loss": 0.0,
+      "step": 12625
+    },
+    {
+      "epoch": 0.8136882129277566,
+      "grad_norm": 0.071685791015625,
+      "learning_rate": 2.076620121732904e-06,
+      "loss": 0.0001,
+      "step": 12626
+    },
+    {
+      "epoch": 0.8137526583746858,
+      "grad_norm": 0.00015047715241365775,
+      "learning_rate": 2.075904045828858e-06,
+      "loss": 0.0,
+      "step": 12627
+    },
+    {
+      "epoch": 0.813817103821615,
+      "grad_norm": 0.039498544859528056,
+      "learning_rate": 2.075187969924812e-06,
+      "loss": 0.0001,
+      "step": 12628
+    },
+    {
+      "epoch": 0.8138815492685442,
+      "grad_norm": 0.007269135108434769,
+      "learning_rate": 2.0744718940207664e-06,
+      "loss": 0.0,
+      "step": 12629
+    },
+    {
+      "epoch": 0.8139459947154734,
+      "grad_norm": 0.00021724833356658388,
+      "learning_rate": 2.0737558181167207e-06,
+      "loss": 0.0,
+      "step": 12630
+    },
+    {
+      "epoch": 0.8140104401624025,
+      "grad_norm": 0.15544795779068946,
+      "learning_rate": 2.0730397422126746e-06,
+      "loss": 0.0005,
+      "step": 12631
+    },
+    {
+      "epoch": 0.8140748856093317,
+      "grad_norm": 0.007632494535068704,
+      "learning_rate": 2.072323666308629e-06,
+      "loss": 0.0,
+      "step": 12632
+    },
+    {
+      "epoch": 0.8141393310562609,
+      "grad_norm": 0.026181218628717726,
+      "learning_rate": 2.0716075904045827e-06,
+      "loss": 0.0,
+      "step": 12633
+    },
+    {
+      "epoch": 0.8142037765031901,
+      "grad_norm": 0.169198243764644,
+      "learning_rate": 2.0708915145005375e-06,
+      "loss": 0.0004,
+      "step": 12634
+    },
+    {
+      "epoch": 0.8142682219501193,
+      "grad_norm": 0.02404185888492924,
+      "learning_rate": 2.0701754385964913e-06,
+      "loss": 0.0001,
+      "step": 12635
+    },
+    {
+      "epoch": 0.8143326673970483,
+      "grad_norm": 0.003228024572993855,
+      "learning_rate": 2.0694593626924456e-06,
+      "loss": 0.0,
+      "step": 12636
+    },
+    {
+      "epoch": 0.8143971128439775,
+      "grad_norm": 0.0023332422248495926,
+      "learning_rate": 2.0687432867883995e-06,
+      "loss": 0.0,
+      "step": 12637
+    },
+    {
+      "epoch": 0.8144615582909067,
+      "grad_norm": 5.473414157304134e-05,
+      "learning_rate": 2.068027210884354e-06,
+      "loss": 0.0,
+      "step": 12638
+    },
+    {
+      "epoch": 0.8145260037378359,
+      "grad_norm": 0.021708693061578042,
+      "learning_rate": 2.067311134980308e-06,
+      "loss": 0.0,
+      "step": 12639
+    },
+    {
+      "epoch": 0.8145904491847651,
+      "grad_norm": 0.45574811805194243,
+      "learning_rate": 2.0665950590762624e-06,
+      "loss": 0.0004,
+      "step": 12640
+    },
+    {
+      "epoch": 0.8146548946316943,
+      "grad_norm": 0.0013192928213251146,
+      "learning_rate": 2.0658789831722163e-06,
+      "loss": 0.0,
+      "step": 12641
+    },
+    {
+      "epoch": 0.8147193400786235,
+      "grad_norm": 0.11894261211468155,
+      "learning_rate": 2.0651629072681706e-06,
+      "loss": 0.0002,
+      "step": 12642
+    },
+    {
+      "epoch": 0.8147837855255526,
+      "grad_norm": 0.14892849528857,
+      "learning_rate": 2.064446831364125e-06,
+      "loss": 0.0002,
+      "step": 12643
+    },
+    {
+      "epoch": 0.8148482309724818,
+      "grad_norm": 0.5862908378806496,
+      "learning_rate": 2.0637307554600788e-06,
+      "loss": 0.0043,
+      "step": 12644
+    },
+    {
+      "epoch": 0.814912676419411,
+      "grad_norm": 0.00046785723467117026,
+      "learning_rate": 2.063014679556033e-06,
+      "loss": 0.0,
+      "step": 12645
+    },
+    {
+      "epoch": 0.8149771218663402,
+      "grad_norm": 0.0002734640965516467,
+      "learning_rate": 2.0622986036519874e-06,
+      "loss": 0.0,
+      "step": 12646
+    },
+    {
+      "epoch": 0.8150415673132693,
+      "grad_norm": 0.009720572095862221,
+      "learning_rate": 2.0615825277479417e-06,
+      "loss": 0.0,
+      "step": 12647
+    },
+    {
+      "epoch": 0.8151060127601985,
+      "grad_norm": 0.0276897489043289,
+      "learning_rate": 2.0608664518438955e-06,
+      "loss": 0.0,
+      "step": 12648
+    },
+    {
+      "epoch": 0.8151704582071276,
+      "grad_norm": 0.12226153845249521,
+      "learning_rate": 2.06015037593985e-06,
+      "loss": 0.0002,
+      "step": 12649
+    },
+    {
+      "epoch": 0.8152349036540568,
+      "grad_norm": 0.045159530174020975,
+      "learning_rate": 2.0594343000358037e-06,
+      "loss": 0.0,
+      "step": 12650
+    },
+    {
+      "epoch": 0.815299349100986,
+      "grad_norm": 0.0007022768080407791,
+      "learning_rate": 2.0587182241317584e-06,
+      "loss": 0.0,
+      "step": 12651
+    },
+    {
+      "epoch": 0.8153637945479152,
+      "grad_norm": 0.023347433204983205,
+      "learning_rate": 2.0580021482277123e-06,
+      "loss": 0.0001,
+      "step": 12652
+    },
+    {
+      "epoch": 0.8154282399948444,
+      "grad_norm": 0.00021818342706713008,
+      "learning_rate": 2.0572860723236666e-06,
+      "loss": 0.0,
+      "step": 12653
+    },
+    {
+      "epoch": 0.8154926854417736,
+      "grad_norm": 0.19678674301530588,
+      "learning_rate": 2.0565699964196205e-06,
+      "loss": 0.0008,
+      "step": 12654
+    },
+    {
+      "epoch": 0.8155571308887027,
+      "grad_norm": 2.9480723597161855,
+      "learning_rate": 2.0558539205155748e-06,
+      "loss": 0.0264,
+      "step": 12655
+    },
+    {
+      "epoch": 0.8156215763356319,
+      "grad_norm": 0.000604122829339633,
+      "learning_rate": 2.055137844611529e-06,
+      "loss": 0.0,
+      "step": 12656
+    },
+    {
+      "epoch": 0.8156860217825611,
+      "grad_norm": 0.004410635766435355,
+      "learning_rate": 2.0544217687074834e-06,
+      "loss": 0.0,
+      "step": 12657
+    },
+    {
+      "epoch": 0.8157504672294902,
+      "grad_norm": 0.007988108148366502,
+      "learning_rate": 2.0537056928034372e-06,
+      "loss": 0.0,
+      "step": 12658
+    },
+    {
+      "epoch": 0.8158149126764194,
+      "grad_norm": 0.00025976478303401106,
+      "learning_rate": 2.0529896168993915e-06,
+      "loss": 0.0,
+      "step": 12659
+    },
+    {
+      "epoch": 0.8158793581233486,
+      "grad_norm": 0.024995079524881312,
+      "learning_rate": 2.052273540995346e-06,
+      "loss": 0.0,
+      "step": 12660
+    },
+    {
+      "epoch": 0.8159438035702777,
+      "grad_norm": 0.041993960354219664,
+      "learning_rate": 2.0515574650912997e-06,
+      "loss": 0.0001,
+      "step": 12661
+    },
+    {
+      "epoch": 0.8160082490172069,
+      "grad_norm": 0.0006448330306277122,
+      "learning_rate": 2.050841389187254e-06,
+      "loss": 0.0,
+      "step": 12662
+    },
+    {
+      "epoch": 0.8160726944641361,
+      "grad_norm": 0.0007699161659463133,
+      "learning_rate": 2.0501253132832083e-06,
+      "loss": 0.0,
+      "step": 12663
+    },
+    {
+      "epoch": 0.8161371399110653,
+      "grad_norm": 0.0001718285197238581,
+      "learning_rate": 2.0494092373791626e-06,
+      "loss": 0.0,
+      "step": 12664
+    },
+    {
+      "epoch": 0.8162015853579945,
+      "grad_norm": 0.00014869667986260737,
+      "learning_rate": 2.0486931614751165e-06,
+      "loss": 0.0,
+      "step": 12665
+    },
+    {
+      "epoch": 0.8162660308049237,
+      "grad_norm": 0.0001513088943245394,
+      "learning_rate": 2.047977085571071e-06,
+      "loss": 0.0,
+      "step": 12666
+    },
+    {
+      "epoch": 0.8163304762518528,
+      "grad_norm": 0.005079574876596566,
+      "learning_rate": 2.0472610096670247e-06,
+      "loss": 0.0,
+      "step": 12667
+    },
+    {
+      "epoch": 0.816394921698782,
+      "grad_norm": 0.0012937261291333171,
+      "learning_rate": 2.046544933762979e-06,
+      "loss": 0.0,
+      "step": 12668
+    },
+    {
+      "epoch": 0.8164593671457111,
+      "grad_norm": 0.013307215880097614,
+      "learning_rate": 2.0458288578589333e-06,
+      "loss": 0.0,
+      "step": 12669
+    },
+    {
+      "epoch": 0.8165238125926403,
+      "grad_norm": 0.0060048266720873255,
+      "learning_rate": 2.0451127819548876e-06,
+      "loss": 0.0,
+      "step": 12670
+    },
+    {
+      "epoch": 0.8165882580395695,
+      "grad_norm": 0.00019912951092864882,
+      "learning_rate": 2.0443967060508414e-06,
+      "loss": 0.0,
+      "step": 12671
+    },
+    {
+      "epoch": 0.8166527034864987,
+      "grad_norm": 0.00549363525012209,
+      "learning_rate": 2.0436806301467957e-06,
+      "loss": 0.0,
+      "step": 12672
+    },
+    {
+      "epoch": 0.8167171489334278,
+      "grad_norm": 0.00042558856689712486,
+      "learning_rate": 2.0429645542427496e-06,
+      "loss": 0.0,
+      "step": 12673
+    },
+    {
+      "epoch": 0.816781594380357,
+      "grad_norm": 0.00016062680426746665,
+      "learning_rate": 2.042248478338704e-06,
+      "loss": 0.0,
+      "step": 12674
+    },
+    {
+      "epoch": 0.8168460398272862,
+      "grad_norm": 0.08333596551979962,
+      "learning_rate": 2.041532402434658e-06,
+      "loss": 0.0004,
+      "step": 12675
+    },
+    {
+      "epoch": 0.8169104852742154,
+      "grad_norm": 0.2981101617579293,
+      "learning_rate": 2.0408163265306125e-06,
+      "loss": 0.0017,
+      "step": 12676
+    },
+    {
+      "epoch": 0.8169749307211446,
+      "grad_norm": 0.0010572690640603976,
+      "learning_rate": 2.0401002506265664e-06,
+      "loss": 0.0,
+      "step": 12677
+    },
+    {
+      "epoch": 0.8170393761680738,
+      "grad_norm": 0.0012552339913406846,
+      "learning_rate": 2.0393841747225207e-06,
+      "loss": 0.0,
+      "step": 12678
+    },
+    {
+      "epoch": 0.817103821615003,
+      "grad_norm": 0.016703592774660905,
+      "learning_rate": 2.038668098818475e-06,
+      "loss": 0.0,
+      "step": 12679
+    },
+    {
+      "epoch": 0.8171682670619321,
+      "grad_norm": 0.004141902966402822,
+      "learning_rate": 2.0379520229144293e-06,
+      "loss": 0.0,
+      "step": 12680
+    },
+    {
+      "epoch": 0.8172327125088612,
+      "grad_norm": 0.0007111984253445308,
+      "learning_rate": 2.037235947010383e-06,
+      "loss": 0.0,
+      "step": 12681
+    },
+    {
+      "epoch": 0.8172971579557904,
+      "grad_norm": 0.0021643333428704877,
+      "learning_rate": 2.0365198711063375e-06,
+      "loss": 0.0,
+      "step": 12682
+    },
+    {
+      "epoch": 0.8173616034027196,
+      "grad_norm": 0.009506596409528539,
+      "learning_rate": 2.0358037952022918e-06,
+      "loss": 0.0,
+      "step": 12683
+    },
+    {
+      "epoch": 0.8174260488496488,
+      "grad_norm": 0.00015312242409299139,
+      "learning_rate": 2.0350877192982456e-06,
+      "loss": 0.0,
+      "step": 12684
+    },
+    {
+      "epoch": 0.8174904942965779,
+      "grad_norm": 0.0015363426845521342,
+      "learning_rate": 2.0343716433942e-06,
+      "loss": 0.0,
+      "step": 12685
+    },
+    {
+      "epoch": 0.8175549397435071,
+      "grad_norm": 0.0038774986870014085,
+      "learning_rate": 2.0336555674901542e-06,
+      "loss": 0.0,
+      "step": 12686
+    },
+    {
+      "epoch": 0.8176193851904363,
+      "grad_norm": 0.0002841460507729212,
+      "learning_rate": 2.0329394915861085e-06,
+      "loss": 0.0,
+      "step": 12687
+    },
+    {
+      "epoch": 0.8176838306373655,
+      "grad_norm": 0.0004165062032192322,
+      "learning_rate": 2.0322234156820624e-06,
+      "loss": 0.0,
+      "step": 12688
+    },
+    {
+      "epoch": 0.8177482760842947,
+      "grad_norm": 0.0006068016544899525,
+      "learning_rate": 2.0315073397780167e-06,
+      "loss": 0.0,
+      "step": 12689
+    },
+    {
+      "epoch": 0.8178127215312239,
+      "grad_norm": 0.056295926580902295,
+      "learning_rate": 2.0307912638739706e-06,
+      "loss": 0.0002,
+      "step": 12690
+    },
+    {
+      "epoch": 0.817877166978153,
+      "grad_norm": 0.0007701938668082081,
+      "learning_rate": 2.030075187969925e-06,
+      "loss": 0.0,
+      "step": 12691
+    },
+    {
+      "epoch": 0.8179416124250821,
+      "grad_norm": 0.0004344841076649519,
+      "learning_rate": 2.029359112065879e-06,
+      "loss": 0.0,
+      "step": 12692
+    },
+    {
+      "epoch": 0.8180060578720113,
+      "grad_norm": 0.000434818477799362,
+      "learning_rate": 2.0286430361618335e-06,
+      "loss": 0.0,
+      "step": 12693
+    },
+    {
+      "epoch": 0.8180705033189405,
+      "grad_norm": 0.0002604699380230901,
+      "learning_rate": 2.0279269602577873e-06,
+      "loss": 0.0,
+      "step": 12694
+    },
+    {
+      "epoch": 0.8181349487658697,
+      "grad_norm": 0.003211286809710176,
+      "learning_rate": 2.0272108843537416e-06,
+      "loss": 0.0,
+      "step": 12695
+    },
+    {
+      "epoch": 0.8181993942127989,
+      "grad_norm": 0.018087344344049478,
+      "learning_rate": 2.026494808449696e-06,
+      "loss": 0.0001,
+      "step": 12696
+    },
+    {
+      "epoch": 0.818263839659728,
+      "grad_norm": 0.02370398735358649,
+      "learning_rate": 2.02577873254565e-06,
+      "loss": 0.0002,
+      "step": 12697
+    },
+    {
+      "epoch": 0.8183282851066572,
+      "grad_norm": 0.0011283720401023612,
+      "learning_rate": 2.025062656641604e-06,
+      "loss": 0.0,
+      "step": 12698
+    },
+    {
+      "epoch": 0.8183927305535864,
+      "grad_norm": 0.0006159527725816931,
+      "learning_rate": 2.0243465807375584e-06,
+      "loss": 0.0,
+      "step": 12699
+    },
+    {
+      "epoch": 0.8184571760005156,
+      "grad_norm": 0.00618308122004743,
+      "learning_rate": 2.0236305048335127e-06,
+      "loss": 0.0,
+      "step": 12700
+    },
+    {
+      "epoch": 0.8185216214474448,
+      "grad_norm": 0.0012521473666674897,
+      "learning_rate": 2.0229144289294666e-06,
+      "loss": 0.0,
+      "step": 12701
+    },
+    {
+      "epoch": 0.818586066894374,
+      "grad_norm": 0.6123207248370139,
+      "learning_rate": 2.022198353025421e-06,
+      "loss": 0.0015,
+      "step": 12702
+    },
+    {
+      "epoch": 0.818650512341303,
+      "grad_norm": 0.002254610635089249,
+      "learning_rate": 2.0214822771213748e-06,
+      "loss": 0.0,
+      "step": 12703
+    },
+    {
+      "epoch": 0.8187149577882322,
+      "grad_norm": 0.1917155355543102,
+      "learning_rate": 2.0207662012173295e-06,
+      "loss": 0.0011,
+      "step": 12704
+    },
+    {
+      "epoch": 0.8187794032351614,
+      "grad_norm": 0.0010368363220421362,
+      "learning_rate": 2.0200501253132834e-06,
+      "loss": 0.0,
+      "step": 12705
+    },
+    {
+      "epoch": 0.8188438486820906,
+      "grad_norm": 0.0002894131131910139,
+      "learning_rate": 2.0193340494092377e-06,
+      "loss": 0.0,
+      "step": 12706
+    },
+    {
+      "epoch": 0.8189082941290198,
+      "grad_norm": 0.00039159992150386276,
+      "learning_rate": 2.0186179735051915e-06,
+      "loss": 0.0,
+      "step": 12707
+    },
+    {
+      "epoch": 0.818972739575949,
+      "grad_norm": 0.08468321098748963,
+      "learning_rate": 2.017901897601146e-06,
+      "loss": 0.0002,
+      "step": 12708
+    },
+    {
+      "epoch": 0.8190371850228781,
+      "grad_norm": 3.6733720040983004,
+      "learning_rate": 2.0171858216971e-06,
+      "loss": 0.0662,
+      "step": 12709
+    },
+    {
+      "epoch": 0.8191016304698073,
+      "grad_norm": 0.002083130458600624,
+      "learning_rate": 2.0164697457930544e-06,
+      "loss": 0.0,
+      "step": 12710
+    },
+    {
+      "epoch": 0.8191660759167365,
+      "grad_norm": 6.762979958088849e-05,
+      "learning_rate": 2.0157536698890083e-06,
+      "loss": 0.0,
+      "step": 12711
+    },
+    {
+      "epoch": 0.8192305213636657,
+      "grad_norm": 0.07021264761194777,
+      "learning_rate": 2.0150375939849626e-06,
+      "loss": 0.0001,
+      "step": 12712
+    },
+    {
+      "epoch": 0.8192949668105949,
+      "grad_norm": 8.095978465353855e-05,
+      "learning_rate": 2.014321518080917e-06,
+      "loss": 0.0,
+      "step": 12713
+    },
+    {
+      "epoch": 0.819359412257524,
+      "grad_norm": 0.00044268987224894545,
+      "learning_rate": 2.0136054421768708e-06,
+      "loss": 0.0,
+      "step": 12714
+    },
+    {
+      "epoch": 0.8194238577044531,
+      "grad_norm": 0.7423377035637856,
+      "learning_rate": 2.012889366272825e-06,
+      "loss": 0.0018,
+      "step": 12715
+    },
+    {
+      "epoch": 0.8194883031513823,
+      "grad_norm": 0.012233627670962024,
+      "learning_rate": 2.0121732903687794e-06,
+      "loss": 0.0,
+      "step": 12716
+    },
+    {
+      "epoch": 0.8195527485983115,
+      "grad_norm": 0.0039064674577683235,
+      "learning_rate": 2.0114572144647337e-06,
+      "loss": 0.0,
+      "step": 12717
+    },
+    {
+      "epoch": 0.8196171940452407,
+      "grad_norm": 6.302641262364943e-05,
+      "learning_rate": 2.0107411385606875e-06,
+      "loss": 0.0,
+      "step": 12718
+    },
+    {
+      "epoch": 0.8196816394921699,
+      "grad_norm": 0.0005544111992086107,
+      "learning_rate": 2.010025062656642e-06,
+      "loss": 0.0,
+      "step": 12719
+    },
+    {
+      "epoch": 0.8197460849390991,
+      "grad_norm": 0.005699905157546856,
+      "learning_rate": 2.0093089867525957e-06,
+      "loss": 0.0,
+      "step": 12720
+    },
+    {
+      "epoch": 0.8198105303860282,
+      "grad_norm": 0.24517601948537127,
+      "learning_rate": 2.0085929108485504e-06,
+      "loss": 0.0037,
+      "step": 12721
+    },
+    {
+      "epoch": 0.8198749758329574,
+      "grad_norm": 1.4732163879247422,
+      "learning_rate": 2.0078768349445043e-06,
+      "loss": 0.0111,
+      "step": 12722
+    },
+    {
+      "epoch": 0.8199394212798866,
+      "grad_norm": 0.0004949487012933176,
+      "learning_rate": 2.0071607590404586e-06,
+      "loss": 0.0,
+      "step": 12723
+    },
+    {
+      "epoch": 0.8200038667268158,
+      "grad_norm": 0.00566935245972329,
+      "learning_rate": 2.0064446831364125e-06,
+      "loss": 0.0,
+      "step": 12724
+    },
+    {
+      "epoch": 0.8200683121737449,
+      "grad_norm": 0.000825286660599744,
+      "learning_rate": 2.005728607232367e-06,
+      "loss": 0.0,
+      "step": 12725
+    },
+    {
+      "epoch": 0.8201327576206741,
+      "grad_norm": 0.0006255925426041789,
+      "learning_rate": 2.0050125313283207e-06,
+      "loss": 0.0,
+      "step": 12726
+    },
+    {
+      "epoch": 0.8201972030676032,
+      "grad_norm": 0.3893231441875631,
+      "learning_rate": 2.0042964554242754e-06,
+      "loss": 0.0011,
+      "step": 12727
+    },
+    {
+      "epoch": 0.8202616485145324,
+      "grad_norm": 0.3826723426061182,
+      "learning_rate": 2.0035803795202293e-06,
+      "loss": 0.0002,
+      "step": 12728
+    },
+    {
+      "epoch": 0.8203260939614616,
+      "grad_norm": 0.002656368265336166,
+      "learning_rate": 2.0028643036161836e-06,
+      "loss": 0.0,
+      "step": 12729
+    },
+    {
+      "epoch": 0.8203905394083908,
+      "grad_norm": 0.0021246091607805234,
+      "learning_rate": 2.0021482277121374e-06,
+      "loss": 0.0,
+      "step": 12730
+    },
+    {
+      "epoch": 0.82045498485532,
+      "grad_norm": 0.09306680573160549,
+      "learning_rate": 2.0014321518080917e-06,
+      "loss": 0.0002,
+      "step": 12731
+    },
+    {
+      "epoch": 0.8205194303022492,
+      "grad_norm": 0.0013687062223432345,
+      "learning_rate": 2.000716075904046e-06,
+      "loss": 0.0,
+      "step": 12732
+    },
+    {
+      "epoch": 0.8205838757491783,
+      "grad_norm": 0.002352388538741838,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 0.0,
+      "step": 12733
+    },
+    {
+      "epoch": 0.8206483211961075,
+      "grad_norm": 0.000925862567441048,
+      "learning_rate": 1.999283924095954e-06,
+      "loss": 0.0,
+      "step": 12734
+    },
+    {
+      "epoch": 0.8207127666430367,
+      "grad_norm": 0.0026647197191653005,
+      "learning_rate": 1.9985678481919085e-06,
+      "loss": 0.0,
+      "step": 12735
+    },
+    {
+      "epoch": 0.8207772120899658,
+      "grad_norm": 0.0005046259473754929,
+      "learning_rate": 1.997851772287863e-06,
+      "loss": 0.0,
+      "step": 12736
+    },
+    {
+      "epoch": 0.820841657536895,
+      "grad_norm": 0.0010993903374720605,
+      "learning_rate": 1.9971356963838167e-06,
+      "loss": 0.0,
+      "step": 12737
+    },
+    {
+      "epoch": 0.8209061029838242,
+      "grad_norm": 0.005264727428518476,
+      "learning_rate": 1.996419620479771e-06,
+      "loss": 0.0,
+      "step": 12738
+    },
+    {
+      "epoch": 0.8209705484307533,
+      "grad_norm": 0.002761264554517652,
+      "learning_rate": 1.9957035445757253e-06,
+      "loss": 0.0,
+      "step": 12739
+    },
+    {
+      "epoch": 0.8210349938776825,
+      "grad_norm": 0.027237537922393484,
+      "learning_rate": 1.9949874686716796e-06,
+      "loss": 0.0,
+      "step": 12740
+    },
+    {
+      "epoch": 0.8210994393246117,
+      "grad_norm": 0.00029789924426951506,
+      "learning_rate": 1.9942713927676335e-06,
+      "loss": 0.0,
+      "step": 12741
+    },
+    {
+      "epoch": 0.8211638847715409,
+      "grad_norm": 0.0032934977863916964,
+      "learning_rate": 1.9935553168635878e-06,
+      "loss": 0.0,
+      "step": 12742
+    },
+    {
+      "epoch": 0.8212283302184701,
+      "grad_norm": 0.05155409225999489,
+      "learning_rate": 1.9928392409595416e-06,
+      "loss": 0.0006,
+      "step": 12743
+    },
+    {
+      "epoch": 0.8212927756653993,
+      "grad_norm": 0.03518922239741282,
+      "learning_rate": 1.992123165055496e-06,
+      "loss": 0.0001,
+      "step": 12744
+    },
+    {
+      "epoch": 0.8213572211123285,
+      "grad_norm": 0.0006332805434407577,
+      "learning_rate": 1.9914070891514502e-06,
+      "loss": 0.0,
+      "step": 12745
+    },
+    {
+      "epoch": 0.8214216665592576,
+      "grad_norm": 0.001911123414774432,
+      "learning_rate": 1.9906910132474045e-06,
+      "loss": 0.0,
+      "step": 12746
+    },
+    {
+      "epoch": 0.8214861120061867,
+      "grad_norm": 0.002936392954632133,
+      "learning_rate": 1.9899749373433584e-06,
+      "loss": 0.0,
+      "step": 12747
+    },
+    {
+      "epoch": 0.8215505574531159,
+      "grad_norm": 0.004308939923328707,
+      "learning_rate": 1.9892588614393127e-06,
+      "loss": 0.0,
+      "step": 12748
+    },
+    {
+      "epoch": 0.8216150029000451,
+      "grad_norm": 0.07829764482853954,
+      "learning_rate": 1.988542785535267e-06,
+      "loss": 0.0001,
+      "step": 12749
+    },
+    {
+      "epoch": 0.8216794483469743,
+      "grad_norm": 0.002653395061435763,
+      "learning_rate": 1.9878267096312213e-06,
+      "loss": 0.0,
+      "step": 12750
+    },
+    {
+      "epoch": 0.8217438937939034,
+      "grad_norm": 0.0009104643964468059,
+      "learning_rate": 1.987110633727175e-06,
+      "loss": 0.0,
+      "step": 12751
+    },
+    {
+      "epoch": 0.8218083392408326,
+      "grad_norm": 0.008782726808995691,
+      "learning_rate": 1.9863945578231295e-06,
+      "loss": 0.0015,
+      "step": 12752
+    },
+    {
+      "epoch": 0.8218727846877618,
+      "grad_norm": 0.01460205373826194,
+      "learning_rate": 1.9856784819190838e-06,
+      "loss": 0.0,
+      "step": 12753
+    },
+    {
+      "epoch": 0.821937230134691,
+      "grad_norm": 0.4726310597162175,
+      "learning_rate": 1.9849624060150376e-06,
+      "loss": 0.0026,
+      "step": 12754
+    },
+    {
+      "epoch": 0.8220016755816202,
+      "grad_norm": 0.04270153103544648,
+      "learning_rate": 1.984246330110992e-06,
+      "loss": 0.0001,
+      "step": 12755
+    },
+    {
+      "epoch": 0.8220661210285494,
+      "grad_norm": 0.5393062261715389,
+      "learning_rate": 1.9835302542069462e-06,
+      "loss": 0.0009,
+      "step": 12756
+    },
+    {
+      "epoch": 0.8221305664754786,
+      "grad_norm": 2.490178844356341,
+      "learning_rate": 1.9828141783029005e-06,
+      "loss": 0.0009,
+      "step": 12757
+    },
+    {
+      "epoch": 0.8221950119224077,
+      "grad_norm": 0.2574040327126001,
+      "learning_rate": 1.9820981023988544e-06,
+      "loss": 0.0014,
+      "step": 12758
+    },
+    {
+      "epoch": 0.8222594573693368,
+      "grad_norm": 0.1584653388532432,
+      "learning_rate": 1.9813820264948087e-06,
+      "loss": 0.0004,
+      "step": 12759
+    },
+    {
+      "epoch": 0.822323902816266,
+      "grad_norm": 0.04805075777073572,
+      "learning_rate": 1.9806659505907626e-06,
+      "loss": 0.0003,
+      "step": 12760
+    },
+    {
+      "epoch": 0.8223883482631952,
+      "grad_norm": 0.0788094696601303,
+      "learning_rate": 1.979949874686717e-06,
+      "loss": 0.0001,
+      "step": 12761
+    },
+    {
+      "epoch": 0.8224527937101244,
+      "grad_norm": 0.004995515025960436,
+      "learning_rate": 1.979233798782671e-06,
+      "loss": 0.0001,
+      "step": 12762
+    },
+    {
+      "epoch": 0.8225172391570535,
+      "grad_norm": 0.0001973465953139601,
+      "learning_rate": 1.9785177228786255e-06,
+      "loss": 0.0,
+      "step": 12763
+    },
+    {
+      "epoch": 0.8225816846039827,
+      "grad_norm": 0.0012884267076477087,
+      "learning_rate": 1.9778016469745794e-06,
+      "loss": 0.0,
+      "step": 12764
+    },
+    {
+      "epoch": 0.8226461300509119,
+      "grad_norm": 0.005440834923307024,
+      "learning_rate": 1.9770855710705337e-06,
+      "loss": 0.0,
+      "step": 12765
+    },
+    {
+      "epoch": 0.8227105754978411,
+      "grad_norm": 0.293044830616161,
+      "learning_rate": 1.9763694951664875e-06,
+      "loss": 0.0007,
+      "step": 12766
+    },
+    {
+      "epoch": 0.8227750209447703,
+      "grad_norm": 0.00020292315307484786,
+      "learning_rate": 1.975653419262442e-06,
+      "loss": 0.0,
+      "step": 12767
+    },
+    {
+      "epoch": 0.8228394663916995,
+      "grad_norm": 0.014746585006457157,
+      "learning_rate": 1.974937343358396e-06,
+      "loss": 0.0001,
+      "step": 12768
+    },
+    {
+      "epoch": 0.8229039118386287,
+      "grad_norm": 0.0029612802244079247,
+      "learning_rate": 1.9742212674543504e-06,
+      "loss": 0.0,
+      "step": 12769
+    },
+    {
+      "epoch": 0.8229683572855577,
+      "grad_norm": 0.0009473144841386316,
+      "learning_rate": 1.9735051915503043e-06,
+      "loss": 0.0,
+      "step": 12770
+    },
+    {
+      "epoch": 0.8230328027324869,
+      "grad_norm": 0.0028644580881699412,
+      "learning_rate": 1.9727891156462586e-06,
+      "loss": 0.0,
+      "step": 12771
+    },
+    {
+      "epoch": 0.8230972481794161,
+      "grad_norm": 0.016254687805129026,
+      "learning_rate": 1.972073039742213e-06,
+      "loss": 0.0,
+      "step": 12772
+    },
+    {
+      "epoch": 0.8231616936263453,
+      "grad_norm": 0.0022775558697995233,
+      "learning_rate": 1.9713569638381668e-06,
+      "loss": 0.0,
+      "step": 12773
+    },
+    {
+      "epoch": 0.8232261390732745,
+      "grad_norm": 0.00014820028206541386,
+      "learning_rate": 1.970640887934121e-06,
+      "loss": 0.0,
+      "step": 12774
+    },
+    {
+      "epoch": 0.8232905845202036,
+      "grad_norm": 0.11497893146273225,
+      "learning_rate": 1.9699248120300754e-06,
+      "loss": 0.0016,
+      "step": 12775
+    },
+    {
+      "epoch": 0.8233550299671328,
+      "grad_norm": 0.02543805923047829,
+      "learning_rate": 1.9692087361260297e-06,
+      "loss": 0.0,
+      "step": 12776
+    },
+    {
+      "epoch": 0.823419475414062,
+      "grad_norm": 0.0007624847949835808,
+      "learning_rate": 1.9684926602219835e-06,
+      "loss": 0.0,
+      "step": 12777
+    },
+    {
+      "epoch": 0.8234839208609912,
+      "grad_norm": 0.0002072227547720038,
+      "learning_rate": 1.967776584317938e-06,
+      "loss": 0.0,
+      "step": 12778
+    },
+    {
+      "epoch": 0.8235483663079204,
+      "grad_norm": 0.0020613742597048756,
+      "learning_rate": 1.9670605084138917e-06,
+      "loss": 0.0,
+      "step": 12779
+    },
+    {
+      "epoch": 0.8236128117548496,
+      "grad_norm": 0.07955951346787625,
+      "learning_rate": 1.9663444325098464e-06,
+      "loss": 0.0002,
+      "step": 12780
+    },
+    {
+      "epoch": 0.8236772572017786,
+      "grad_norm": 0.0011062340486165578,
+      "learning_rate": 1.9656283566058003e-06,
+      "loss": 0.0,
+      "step": 12781
+    },
+    {
+      "epoch": 0.8237417026487078,
+      "grad_norm": 0.0018729547487564492,
+      "learning_rate": 1.9649122807017546e-06,
+      "loss": 0.0,
+      "step": 12782
+    },
+    {
+      "epoch": 0.823806148095637,
+      "grad_norm": 1.9180195290861062,
+      "learning_rate": 1.9641962047977085e-06,
+      "loss": 0.0044,
+      "step": 12783
+    },
+    {
+      "epoch": 0.8238705935425662,
+      "grad_norm": 0.0004110962381387348,
+      "learning_rate": 1.963480128893663e-06,
+      "loss": 0.0,
+      "step": 12784
+    },
+    {
+      "epoch": 0.8239350389894954,
+      "grad_norm": 0.002992485947360091,
+      "learning_rate": 1.962764052989617e-06,
+      "loss": 0.0,
+      "step": 12785
+    },
+    {
+      "epoch": 0.8239994844364246,
+      "grad_norm": 0.23977967782600768,
+      "learning_rate": 1.9620479770855714e-06,
+      "loss": 0.0028,
+      "step": 12786
+    },
+    {
+      "epoch": 0.8240639298833538,
+      "grad_norm": 0.0001052902595383848,
+      "learning_rate": 1.9613319011815253e-06,
+      "loss": 0.0,
+      "step": 12787
+    },
+    {
+      "epoch": 0.8241283753302829,
+      "grad_norm": 0.0009807068807572042,
+      "learning_rate": 1.9606158252774796e-06,
+      "loss": 0.0,
+      "step": 12788
+    },
+    {
+      "epoch": 0.8241928207772121,
+      "grad_norm": 0.00017703910016623285,
+      "learning_rate": 1.959899749373434e-06,
+      "loss": 0.0,
+      "step": 12789
+    },
+    {
+      "epoch": 0.8242572662241413,
+      "grad_norm": 0.0008779980011984871,
+      "learning_rate": 1.9591836734693877e-06,
+      "loss": 0.0,
+      "step": 12790
+    },
+    {
+      "epoch": 0.8243217116710705,
+      "grad_norm": 0.0007441098091688068,
+      "learning_rate": 1.958467597565342e-06,
+      "loss": 0.0,
+      "step": 12791
+    },
+    {
+      "epoch": 0.8243861571179996,
+      "grad_norm": 0.0012419554472922133,
+      "learning_rate": 1.9577515216612963e-06,
+      "loss": 0.0,
+      "step": 12792
+    },
+    {
+      "epoch": 0.8244506025649287,
+      "grad_norm": 0.0016284702447365287,
+      "learning_rate": 1.9570354457572506e-06,
+      "loss": 0.0,
+      "step": 12793
+    },
+    {
+      "epoch": 0.8245150480118579,
+      "grad_norm": 0.0033685588571939636,
+      "learning_rate": 1.9563193698532045e-06,
+      "loss": 0.0,
+      "step": 12794
+    },
+    {
+      "epoch": 0.8245794934587871,
+      "grad_norm": 0.0286288462786047,
+      "learning_rate": 1.955603293949159e-06,
+      "loss": 0.0,
+      "step": 12795
+    },
+    {
+      "epoch": 0.8246439389057163,
+      "grad_norm": 0.0021863046623147764,
+      "learning_rate": 1.9548872180451127e-06,
+      "loss": 0.0,
+      "step": 12796
+    },
+    {
+      "epoch": 0.8247083843526455,
+      "grad_norm": 0.002743261369717264,
+      "learning_rate": 1.9541711421410674e-06,
+      "loss": 0.0,
+      "step": 12797
+    },
+    {
+      "epoch": 0.8247728297995747,
+      "grad_norm": 0.030985487605830843,
+      "learning_rate": 1.9534550662370213e-06,
+      "loss": 0.0,
+      "step": 12798
+    },
+    {
+      "epoch": 0.8248372752465039,
+      "grad_norm": 0.011039880199293242,
+      "learning_rate": 1.9527389903329756e-06,
+      "loss": 0.0,
+      "step": 12799
+    },
+    {
+      "epoch": 0.824901720693433,
+      "grad_norm": 0.0006824560142423226,
+      "learning_rate": 1.9520229144289295e-06,
+      "loss": 0.0,
+      "step": 12800
+    },
+    {
+      "epoch": 0.8249661661403622,
+      "grad_norm": 0.010945881791514213,
+      "learning_rate": 1.9513068385248838e-06,
+      "loss": 0.0001,
+      "step": 12801
+    },
+    {
+      "epoch": 0.8250306115872914,
+      "grad_norm": 0.00721237801787959,
+      "learning_rate": 1.950590762620838e-06,
+      "loss": 0.0,
+      "step": 12802
+    },
+    {
+      "epoch": 0.8250950570342205,
+      "grad_norm": 0.0004965975994171999,
+      "learning_rate": 1.9498746867167923e-06,
+      "loss": 0.0,
+      "step": 12803
+    },
+    {
+      "epoch": 0.8251595024811497,
+      "grad_norm": 0.009016010154456617,
+      "learning_rate": 1.9491586108127462e-06,
+      "loss": 0.0,
+      "step": 12804
+    },
+    {
+      "epoch": 0.8252239479280788,
+      "grad_norm": 0.0012284352036761618,
+      "learning_rate": 1.9484425349087005e-06,
+      "loss": 0.0,
+      "step": 12805
+    },
+    {
+      "epoch": 0.825288393375008,
+      "grad_norm": 0.0028087657996823242,
+      "learning_rate": 1.947726459004655e-06,
+      "loss": 0.0,
+      "step": 12806
+    },
+    {
+      "epoch": 0.8253528388219372,
+      "grad_norm": 0.0011335282820437035,
+      "learning_rate": 1.9470103831006087e-06,
+      "loss": 0.0,
+      "step": 12807
+    },
+    {
+      "epoch": 0.8254172842688664,
+      "grad_norm": 0.024974543026714955,
+      "learning_rate": 1.946294307196563e-06,
+      "loss": 0.0,
+      "step": 12808
+    },
+    {
+      "epoch": 0.8254817297157956,
+      "grad_norm": 0.01026302773500995,
+      "learning_rate": 1.9455782312925173e-06,
+      "loss": 0.0001,
+      "step": 12809
+    },
+    {
+      "epoch": 0.8255461751627248,
+      "grad_norm": 0.00013102626201935377,
+      "learning_rate": 1.9448621553884716e-06,
+      "loss": 0.0,
+      "step": 12810
+    },
+    {
+      "epoch": 0.825610620609654,
+      "grad_norm": 0.05074874360207065,
+      "learning_rate": 1.9441460794844255e-06,
+      "loss": 0.0002,
+      "step": 12811
+    },
+    {
+      "epoch": 0.8256750660565831,
+      "grad_norm": 0.16791835295001642,
+      "learning_rate": 1.9434300035803798e-06,
+      "loss": 0.0002,
+      "step": 12812
+    },
+    {
+      "epoch": 0.8257395115035123,
+      "grad_norm": 0.22231346647938868,
+      "learning_rate": 1.9427139276763336e-06,
+      "loss": 0.0009,
+      "step": 12813
+    },
+    {
+      "epoch": 0.8258039569504414,
+      "grad_norm": 0.005087052383364399,
+      "learning_rate": 1.941997851772288e-06,
+      "loss": 0.0001,
+      "step": 12814
+    },
+    {
+      "epoch": 0.8258684023973706,
+      "grad_norm": 0.29658397413916804,
+      "learning_rate": 1.9412817758682422e-06,
+      "loss": 0.0004,
+      "step": 12815
+    },
+    {
+      "epoch": 0.8259328478442998,
+      "grad_norm": 0.00015184382402304126,
+      "learning_rate": 1.9405656999641965e-06,
+      "loss": 0.0,
+      "step": 12816
+    },
+    {
+      "epoch": 0.825997293291229,
+      "grad_norm": 0.00011728335842496472,
+      "learning_rate": 1.9398496240601504e-06,
+      "loss": 0.0,
+      "step": 12817
+    },
+    {
+      "epoch": 0.8260617387381581,
+      "grad_norm": 0.12153047977766473,
+      "learning_rate": 1.9391335481561047e-06,
+      "loss": 0.0014,
+      "step": 12818
+    },
+    {
+      "epoch": 0.8261261841850873,
+      "grad_norm": 0.0024044143837508186,
+      "learning_rate": 1.9384174722520586e-06,
+      "loss": 0.0,
+      "step": 12819
+    },
+    {
+      "epoch": 0.8261906296320165,
+      "grad_norm": 0.26979896131740444,
+      "learning_rate": 1.9377013963480133e-06,
+      "loss": 0.0059,
+      "step": 12820
+    },
+    {
+      "epoch": 0.8262550750789457,
+      "grad_norm": 0.007687804269723589,
+      "learning_rate": 1.936985320443967e-06,
+      "loss": 0.0,
+      "step": 12821
+    },
+    {
+      "epoch": 0.8263195205258749,
+      "grad_norm": 0.001681070516331721,
+      "learning_rate": 1.9362692445399215e-06,
+      "loss": 0.0,
+      "step": 12822
+    },
+    {
+      "epoch": 0.8263839659728041,
+      "grad_norm": 0.010557766168112366,
+      "learning_rate": 1.9355531686358754e-06,
+      "loss": 0.0,
+      "step": 12823
+    },
+    {
+      "epoch": 0.8264484114197332,
+      "grad_norm": 0.0008512911900243197,
+      "learning_rate": 1.9348370927318297e-06,
+      "loss": 0.0,
+      "step": 12824
+    },
+    {
+      "epoch": 0.8265128568666623,
+      "grad_norm": 0.0013926741667203172,
+      "learning_rate": 1.934121016827784e-06,
+      "loss": 0.0,
+      "step": 12825
+    },
+    {
+      "epoch": 0.8265773023135915,
+      "grad_norm": 0.0006454232476995218,
+      "learning_rate": 1.9334049409237383e-06,
+      "loss": 0.0,
+      "step": 12826
+    },
+    {
+      "epoch": 0.8266417477605207,
+      "grad_norm": 0.007012337352928124,
+      "learning_rate": 1.932688865019692e-06,
+      "loss": 0.0,
+      "step": 12827
+    },
+    {
+      "epoch": 0.8267061932074499,
+      "grad_norm": 0.0016290758214399952,
+      "learning_rate": 1.9319727891156464e-06,
+      "loss": 0.0,
+      "step": 12828
+    },
+    {
+      "epoch": 0.826770638654379,
+      "grad_norm": 0.0004002776358242824,
+      "learning_rate": 1.9312567132116007e-06,
+      "loss": 0.0,
+      "step": 12829
+    },
+    {
+      "epoch": 0.8268350841013082,
+      "grad_norm": 0.0006474920133281832,
+      "learning_rate": 1.9305406373075546e-06,
+      "loss": 0.0,
+      "step": 12830
+    },
+    {
+      "epoch": 0.8268995295482374,
+      "grad_norm": 0.0004922729677889804,
+      "learning_rate": 1.929824561403509e-06,
+      "loss": 0.0,
+      "step": 12831
+    },
+    {
+      "epoch": 0.8269639749951666,
+      "grad_norm": 0.015237039801407486,
+      "learning_rate": 1.929108485499463e-06,
+      "loss": 0.0,
+      "step": 12832
+    },
+    {
+      "epoch": 0.8270284204420958,
+      "grad_norm": 0.001385017502838934,
+      "learning_rate": 1.9283924095954175e-06,
+      "loss": 0.0,
+      "step": 12833
+    },
+    {
+      "epoch": 0.827092865889025,
+      "grad_norm": 0.00036743085902484786,
+      "learning_rate": 1.9276763336913714e-06,
+      "loss": 0.0,
+      "step": 12834
+    },
+    {
+      "epoch": 0.8271573113359542,
+      "grad_norm": 0.21470143199397795,
+      "learning_rate": 1.9269602577873257e-06,
+      "loss": 0.0052,
+      "step": 12835
+    },
+    {
+      "epoch": 0.8272217567828833,
+      "grad_norm": 0.20051220260145272,
+      "learning_rate": 1.9262441818832795e-06,
+      "loss": 0.0009,
+      "step": 12836
+    },
+    {
+      "epoch": 0.8272862022298124,
+      "grad_norm": 0.0013799034224737958,
+      "learning_rate": 1.925528105979234e-06,
+      "loss": 0.0,
+      "step": 12837
+    },
+    {
+      "epoch": 0.8273506476767416,
+      "grad_norm": 0.0011499711319856818,
+      "learning_rate": 1.924812030075188e-06,
+      "loss": 0.0,
+      "step": 12838
+    },
+    {
+      "epoch": 0.8274150931236708,
+      "grad_norm": 0.004180721553757159,
+      "learning_rate": 1.9240959541711424e-06,
+      "loss": 0.0,
+      "step": 12839
+    },
+    {
+      "epoch": 0.8274795385706,
+      "grad_norm": 0.00888103511971256,
+      "learning_rate": 1.9233798782670963e-06,
+      "loss": 0.0,
+      "step": 12840
+    },
+    {
+      "epoch": 0.8275439840175292,
+      "grad_norm": 0.0004716433398487072,
+      "learning_rate": 1.9226638023630506e-06,
+      "loss": 0.0,
+      "step": 12841
+    },
+    {
+      "epoch": 0.8276084294644583,
+      "grad_norm": 3.567135691782576e-05,
+      "learning_rate": 1.921947726459005e-06,
+      "loss": 0.0,
+      "step": 12842
+    },
+    {
+      "epoch": 0.8276728749113875,
+      "grad_norm": 0.006151835977891471,
+      "learning_rate": 1.921231650554959e-06,
+      "loss": 0.0,
+      "step": 12843
+    },
+    {
+      "epoch": 0.8277373203583167,
+      "grad_norm": 0.0017068130805905073,
+      "learning_rate": 1.920515574650913e-06,
+      "loss": 0.0,
+      "step": 12844
+    },
+    {
+      "epoch": 0.8278017658052459,
+      "grad_norm": 0.029947653519642226,
+      "learning_rate": 1.9197994987468674e-06,
+      "loss": 0.0015,
+      "step": 12845
+    },
+    {
+      "epoch": 0.8278662112521751,
+      "grad_norm": 0.03704959829540365,
+      "learning_rate": 1.9190834228428217e-06,
+      "loss": 0.0001,
+      "step": 12846
+    },
+    {
+      "epoch": 0.8279306566991043,
+      "grad_norm": 0.0018888023632662365,
+      "learning_rate": 1.9183673469387756e-06,
+      "loss": 0.0,
+      "step": 12847
+    },
+    {
+      "epoch": 0.8279951021460333,
+      "grad_norm": 0.0032775118244479816,
+      "learning_rate": 1.91765127103473e-06,
+      "loss": 0.0,
+      "step": 12848
+    },
+    {
+      "epoch": 0.8280595475929625,
+      "grad_norm": 0.002627906814267534,
+      "learning_rate": 1.9169351951306837e-06,
+      "loss": 0.0,
+      "step": 12849
+    },
+    {
+      "epoch": 0.8281239930398917,
+      "grad_norm": 0.0008013443872099772,
+      "learning_rate": 1.9162191192266385e-06,
+      "loss": 0.0,
+      "step": 12850
+    },
+    {
+      "epoch": 0.8281884384868209,
+      "grad_norm": 0.0009410073850608853,
+      "learning_rate": 1.9155030433225923e-06,
+      "loss": 0.0,
+      "step": 12851
+    },
+    {
+      "epoch": 0.8282528839337501,
+      "grad_norm": 0.0010344266427138322,
+      "learning_rate": 1.9147869674185466e-06,
+      "loss": 0.0,
+      "step": 12852
+    },
+    {
+      "epoch": 0.8283173293806793,
+      "grad_norm": 0.00014077844258373367,
+      "learning_rate": 1.9140708915145005e-06,
+      "loss": 0.0,
+      "step": 12853
+    },
+    {
+      "epoch": 0.8283817748276084,
+      "grad_norm": 0.00022600811992336105,
+      "learning_rate": 1.913354815610455e-06,
+      "loss": 0.0,
+      "step": 12854
+    },
+    {
+      "epoch": 0.8284462202745376,
+      "grad_norm": 0.0003438217147535301,
+      "learning_rate": 1.912638739706409e-06,
+      "loss": 0.0,
+      "step": 12855
+    },
+    {
+      "epoch": 0.8285106657214668,
+      "grad_norm": 1.3295451481395686,
+      "learning_rate": 1.9119226638023634e-06,
+      "loss": 0.0073,
+      "step": 12856
+    },
+    {
+      "epoch": 0.828575111168396,
+      "grad_norm": 0.09529210403966394,
+      "learning_rate": 1.9112065878983173e-06,
+      "loss": 0.0002,
+      "step": 12857
+    },
+    {
+      "epoch": 0.8286395566153252,
+      "grad_norm": 0.00013075154801560885,
+      "learning_rate": 1.9104905119942716e-06,
+      "loss": 0.0,
+      "step": 12858
+    },
+    {
+      "epoch": 0.8287040020622543,
+      "grad_norm": 0.005347229015423188,
+      "learning_rate": 1.909774436090226e-06,
+      "loss": 0.0,
+      "step": 12859
+    },
+    {
+      "epoch": 0.8287684475091834,
+      "grad_norm": 0.011590017739172188,
+      "learning_rate": 1.9090583601861798e-06,
+      "loss": 0.0,
+      "step": 12860
+    },
+    {
+      "epoch": 0.8288328929561126,
+      "grad_norm": 0.002305119796443571,
+      "learning_rate": 1.908342284282134e-06,
+      "loss": 0.0,
+      "step": 12861
+    },
+    {
+      "epoch": 0.8288973384030418,
+      "grad_norm": 0.0032701660571884997,
+      "learning_rate": 1.9076262083780883e-06,
+      "loss": 0.0,
+      "step": 12862
+    },
+    {
+      "epoch": 0.828961783849971,
+      "grad_norm": 0.00032209619860416687,
+      "learning_rate": 1.9069101324740424e-06,
+      "loss": 0.0,
+      "step": 12863
+    },
+    {
+      "epoch": 0.8290262292969002,
+      "grad_norm": 0.0001742985353381191,
+      "learning_rate": 1.9061940565699965e-06,
+      "loss": 0.0,
+      "step": 12864
+    },
+    {
+      "epoch": 0.8290906747438294,
+      "grad_norm": 0.006867729174768813,
+      "learning_rate": 1.9054779806659506e-06,
+      "loss": 0.0,
+      "step": 12865
+    },
+    {
+      "epoch": 0.8291551201907585,
+      "grad_norm": 6.386471846661756e-05,
+      "learning_rate": 1.904761904761905e-06,
+      "loss": 0.0,
+      "step": 12866
+    },
+    {
+      "epoch": 0.8292195656376877,
+      "grad_norm": 8.795452018101676e-05,
+      "learning_rate": 1.9040458288578592e-06,
+      "loss": 0.0,
+      "step": 12867
+    },
+    {
+      "epoch": 0.8292840110846169,
+      "grad_norm": 0.014635279825817802,
+      "learning_rate": 1.9033297529538133e-06,
+      "loss": 0.0,
+      "step": 12868
+    },
+    {
+      "epoch": 0.8293484565315461,
+      "grad_norm": 0.20185402148521048,
+      "learning_rate": 1.9026136770497674e-06,
+      "loss": 0.0011,
+      "step": 12869
+    },
+    {
+      "epoch": 0.8294129019784752,
+      "grad_norm": 3.5018456418513304e-05,
+      "learning_rate": 1.9018976011457217e-06,
+      "loss": 0.0,
+      "step": 12870
+    },
+    {
+      "epoch": 0.8294773474254044,
+      "grad_norm": 0.0011095530381541555,
+      "learning_rate": 1.9011815252416758e-06,
+      "loss": 0.0,
+      "step": 12871
+    },
+    {
+      "epoch": 0.8295417928723335,
+      "grad_norm": 0.06928320343427943,
+      "learning_rate": 1.9004654493376299e-06,
+      "loss": 0.0001,
+      "step": 12872
+    },
+    {
+      "epoch": 0.8296062383192627,
+      "grad_norm": 0.005472200014042761,
+      "learning_rate": 1.8997493734335842e-06,
+      "loss": 0.0001,
+      "step": 12873
+    },
+    {
+      "epoch": 0.8296706837661919,
+      "grad_norm": 0.01115602003143175,
+      "learning_rate": 1.8990332975295385e-06,
+      "loss": 0.0001,
+      "step": 12874
+    },
+    {
+      "epoch": 0.8297351292131211,
+      "grad_norm": 0.0010032057744439192,
+      "learning_rate": 1.8983172216254925e-06,
+      "loss": 0.0,
+      "step": 12875
+    },
+    {
+      "epoch": 0.8297995746600503,
+      "grad_norm": 0.002481002567431664,
+      "learning_rate": 1.8976011457214466e-06,
+      "loss": 0.0,
+      "step": 12876
+    },
+    {
+      "epoch": 0.8298640201069795,
+      "grad_norm": 0.005168957551730069,
+      "learning_rate": 1.8968850698174007e-06,
+      "loss": 0.0,
+      "step": 12877
+    },
+    {
+      "epoch": 0.8299284655539086,
+      "grad_norm": 0.0038538895973451717,
+      "learning_rate": 1.8961689939133548e-06,
+      "loss": 0.0,
+      "step": 12878
+    },
+    {
+      "epoch": 0.8299929110008378,
+      "grad_norm": 0.12830823844414282,
+      "learning_rate": 1.8954529180093093e-06,
+      "loss": 0.0001,
+      "step": 12879
+    },
+    {
+      "epoch": 0.830057356447767,
+      "grad_norm": 0.0022606158190638017,
+      "learning_rate": 1.8947368421052634e-06,
+      "loss": 0.0,
+      "step": 12880
+    },
+    {
+      "epoch": 0.8301218018946961,
+      "grad_norm": 0.1553001860605195,
+      "learning_rate": 1.8940207662012175e-06,
+      "loss": 0.0007,
+      "step": 12881
+    },
+    {
+      "epoch": 0.8301862473416253,
+      "grad_norm": 9.681461373014582e-05,
+      "learning_rate": 1.8933046902971716e-06,
+      "loss": 0.0,
+      "step": 12882
+    },
+    {
+      "epoch": 0.8302506927885545,
+      "grad_norm": 0.006053399325819987,
+      "learning_rate": 1.8925886143931257e-06,
+      "loss": 0.0,
+      "step": 12883
+    },
+    {
+      "epoch": 0.8303151382354836,
+      "grad_norm": 0.0003135291907860067,
+      "learning_rate": 1.89187253848908e-06,
+      "loss": 0.0,
+      "step": 12884
+    },
+    {
+      "epoch": 0.8303795836824128,
+      "grad_norm": 0.0016962123214610799,
+      "learning_rate": 1.8911564625850343e-06,
+      "loss": 0.0,
+      "step": 12885
+    },
+    {
+      "epoch": 0.830444029129342,
+      "grad_norm": 0.0022821039817459542,
+      "learning_rate": 1.8904403866809883e-06,
+      "loss": 0.0,
+      "step": 12886
+    },
+    {
+      "epoch": 0.8305084745762712,
+      "grad_norm": 0.44118902887440714,
+      "learning_rate": 1.8897243107769424e-06,
+      "loss": 0.0008,
+      "step": 12887
+    },
+    {
+      "epoch": 0.8305729200232004,
+      "grad_norm": 0.0026145982784008394,
+      "learning_rate": 1.8890082348728967e-06,
+      "loss": 0.0,
+      "step": 12888
+    },
+    {
+      "epoch": 0.8306373654701296,
+      "grad_norm": 2.009516340008284e-05,
+      "learning_rate": 1.8882921589688508e-06,
+      "loss": 0.0,
+      "step": 12889
+    },
+    {
+      "epoch": 0.8307018109170587,
+      "grad_norm": 0.00039474701415418735,
+      "learning_rate": 1.887576083064805e-06,
+      "loss": 0.0,
+      "step": 12890
+    },
+    {
+      "epoch": 0.8307662563639879,
+      "grad_norm": 0.02382485355526577,
+      "learning_rate": 1.8868600071607592e-06,
+      "loss": 0.0001,
+      "step": 12891
+    },
+    {
+      "epoch": 0.830830701810917,
+      "grad_norm": 0.005398793179076116,
+      "learning_rate": 1.8861439312567135e-06,
+      "loss": 0.0,
+      "step": 12892
+    },
+    {
+      "epoch": 0.8308951472578462,
+      "grad_norm": 0.0768639292053062,
+      "learning_rate": 1.8854278553526676e-06,
+      "loss": 0.0002,
+      "step": 12893
+    },
+    {
+      "epoch": 0.8309595927047754,
+      "grad_norm": 0.008180770152188349,
+      "learning_rate": 1.8847117794486217e-06,
+      "loss": 0.0,
+      "step": 12894
+    },
+    {
+      "epoch": 0.8310240381517046,
+      "grad_norm": 0.003654351436249659,
+      "learning_rate": 1.8839957035445758e-06,
+      "loss": 0.0,
+      "step": 12895
+    },
+    {
+      "epoch": 0.8310884835986337,
+      "grad_norm": 0.0013665234520897544,
+      "learning_rate": 1.8832796276405303e-06,
+      "loss": 0.0,
+      "step": 12896
+    },
+    {
+      "epoch": 0.8311529290455629,
+      "grad_norm": 0.00821381169757801,
+      "learning_rate": 1.8825635517364844e-06,
+      "loss": 0.0,
+      "step": 12897
+    },
+    {
+      "epoch": 0.8312173744924921,
+      "grad_norm": 0.02980132621626116,
+      "learning_rate": 1.8818474758324384e-06,
+      "loss": 0.0001,
+      "step": 12898
+    },
+    {
+      "epoch": 0.8312818199394213,
+      "grad_norm": 0.15304842610763447,
+      "learning_rate": 1.8811313999283925e-06,
+      "loss": 0.0002,
+      "step": 12899
+    },
+    {
+      "epoch": 0.8313462653863505,
+      "grad_norm": 3.2859700974284204e-05,
+      "learning_rate": 1.8804153240243466e-06,
+      "loss": 0.0,
+      "step": 12900
+    },
+    {
+      "epoch": 0.8314107108332797,
+      "grad_norm": 7.330661953553586e-05,
+      "learning_rate": 1.8796992481203007e-06,
+      "loss": 0.0,
+      "step": 12901
+    },
+    {
+      "epoch": 0.8314751562802088,
+      "grad_norm": 0.0016363356131233538,
+      "learning_rate": 1.8789831722162552e-06,
+      "loss": 0.0,
+      "step": 12902
+    },
+    {
+      "epoch": 0.8315396017271379,
+      "grad_norm": 0.0023635623644680886,
+      "learning_rate": 1.8782670963122093e-06,
+      "loss": 0.0,
+      "step": 12903
+    },
+    {
+      "epoch": 0.8316040471740671,
+      "grad_norm": 0.0011234160999025852,
+      "learning_rate": 1.8775510204081634e-06,
+      "loss": 0.0,
+      "step": 12904
+    },
+    {
+      "epoch": 0.8316684926209963,
+      "grad_norm": 0.00569605487531826,
+      "learning_rate": 1.8768349445041175e-06,
+      "loss": 0.0,
+      "step": 12905
+    },
+    {
+      "epoch": 0.8317329380679255,
+      "grad_norm": 0.11860830172496771,
+      "learning_rate": 1.8761188686000718e-06,
+      "loss": 0.0001,
+      "step": 12906
+    },
+    {
+      "epoch": 0.8317973835148547,
+      "grad_norm": 0.0001888324722297905,
+      "learning_rate": 1.8754027926960259e-06,
+      "loss": 0.0,
+      "step": 12907
+    },
+    {
+      "epoch": 0.8318618289617838,
+      "grad_norm": 0.043075241266047205,
+      "learning_rate": 1.8746867167919802e-06,
+      "loss": 0.0001,
+      "step": 12908
+    },
+    {
+      "epoch": 0.831926274408713,
+      "grad_norm": 0.0002445940055020511,
+      "learning_rate": 1.8739706408879342e-06,
+      "loss": 0.0,
+      "step": 12909
+    },
+    {
+      "epoch": 0.8319907198556422,
+      "grad_norm": 0.002135059555199994,
+      "learning_rate": 1.8732545649838885e-06,
+      "loss": 0.0,
+      "step": 12910
+    },
+    {
+      "epoch": 0.8320551653025714,
+      "grad_norm": 3.074306074790163,
+      "learning_rate": 1.8725384890798426e-06,
+      "loss": 0.0207,
+      "step": 12911
+    },
+    {
+      "epoch": 0.8321196107495006,
+      "grad_norm": 0.0009971946495302751,
+      "learning_rate": 1.8718224131757967e-06,
+      "loss": 0.0,
+      "step": 12912
+    },
+    {
+      "epoch": 0.8321840561964298,
+      "grad_norm": 0.0005966592023226857,
+      "learning_rate": 1.8711063372717508e-06,
+      "loss": 0.0,
+      "step": 12913
+    },
+    {
+      "epoch": 0.832248501643359,
+      "grad_norm": 0.0005944522041070304,
+      "learning_rate": 1.8703902613677053e-06,
+      "loss": 0.0,
+      "step": 12914
+    },
+    {
+      "epoch": 0.832312947090288,
+      "grad_norm": 0.00020677188897750951,
+      "learning_rate": 1.8696741854636594e-06,
+      "loss": 0.0,
+      "step": 12915
+    },
+    {
+      "epoch": 0.8323773925372172,
+      "grad_norm": 0.00010503859283321327,
+      "learning_rate": 1.8689581095596135e-06,
+      "loss": 0.0,
+      "step": 12916
+    },
+    {
+      "epoch": 0.8324418379841464,
+      "grad_norm": 0.06789593276378175,
+      "learning_rate": 1.8682420336555676e-06,
+      "loss": 0.0002,
+      "step": 12917
+    },
+    {
+      "epoch": 0.8325062834310756,
+      "grad_norm": 0.00022078347307954933,
+      "learning_rate": 1.8675259577515217e-06,
+      "loss": 0.0,
+      "step": 12918
+    },
+    {
+      "epoch": 0.8325707288780048,
+      "grad_norm": 0.14633415785093218,
+      "learning_rate": 1.866809881847476e-06,
+      "loss": 0.0004,
+      "step": 12919
+    },
+    {
+      "epoch": 0.832635174324934,
+      "grad_norm": 0.26466572272238886,
+      "learning_rate": 1.8660938059434303e-06,
+      "loss": 0.0044,
+      "step": 12920
+    },
+    {
+      "epoch": 0.8326996197718631,
+      "grad_norm": 0.00018119899261728837,
+      "learning_rate": 1.8653777300393843e-06,
+      "loss": 0.0,
+      "step": 12921
+    },
+    {
+      "epoch": 0.8327640652187923,
+      "grad_norm": 0.49176382570365107,
+      "learning_rate": 1.8646616541353384e-06,
+      "loss": 0.0004,
+      "step": 12922
+    },
+    {
+      "epoch": 0.8328285106657215,
+      "grad_norm": 0.17224296823060695,
+      "learning_rate": 1.8639455782312927e-06,
+      "loss": 0.0004,
+      "step": 12923
+    },
+    {
+      "epoch": 0.8328929561126507,
+      "grad_norm": 0.8200194062285915,
+      "learning_rate": 1.8632295023272468e-06,
+      "loss": 0.0046,
+      "step": 12924
+    },
+    {
+      "epoch": 0.8329574015595799,
+      "grad_norm": 0.022811282350755628,
+      "learning_rate": 1.862513426423201e-06,
+      "loss": 0.0,
+      "step": 12925
+    },
+    {
+      "epoch": 0.8330218470065089,
+      "grad_norm": 0.0002876985247923036,
+      "learning_rate": 1.8617973505191552e-06,
+      "loss": 0.0,
+      "step": 12926
+    },
+    {
+      "epoch": 0.8330862924534381,
+      "grad_norm": 0.00026064448045058826,
+      "learning_rate": 1.8610812746151095e-06,
+      "loss": 0.0,
+      "step": 12927
+    },
+    {
+      "epoch": 0.8331507379003673,
+      "grad_norm": 0.00276155562535339,
+      "learning_rate": 1.8603651987110636e-06,
+      "loss": 0.0,
+      "step": 12928
+    },
+    {
+      "epoch": 0.8332151833472965,
+      "grad_norm": 0.0018431929870960982,
+      "learning_rate": 1.8596491228070177e-06,
+      "loss": 0.0,
+      "step": 12929
+    },
+    {
+      "epoch": 0.8332796287942257,
+      "grad_norm": 0.0007927471892093145,
+      "learning_rate": 1.8589330469029718e-06,
+      "loss": 0.0,
+      "step": 12930
+    },
+    {
+      "epoch": 0.8333440742411549,
+      "grad_norm": 0.602887935199894,
+      "learning_rate": 1.8582169709989263e-06,
+      "loss": 0.0018,
+      "step": 12931
+    },
+    {
+      "epoch": 0.833408519688084,
+      "grad_norm": 0.047878668853832275,
+      "learning_rate": 1.8575008950948804e-06,
+      "loss": 0.0001,
+      "step": 12932
+    },
+    {
+      "epoch": 0.8334729651350132,
+      "grad_norm": 0.03237215842096609,
+      "learning_rate": 1.8567848191908345e-06,
+      "loss": 0.0,
+      "step": 12933
+    },
+    {
+      "epoch": 0.8335374105819424,
+      "grad_norm": 0.014911026549980291,
+      "learning_rate": 1.8560687432867885e-06,
+      "loss": 0.0,
+      "step": 12934
+    },
+    {
+      "epoch": 0.8336018560288716,
+      "grad_norm": 0.01923477641968151,
+      "learning_rate": 1.8553526673827426e-06,
+      "loss": 0.0001,
+      "step": 12935
+    },
+    {
+      "epoch": 0.8336663014758008,
+      "grad_norm": 0.2234138849757082,
+      "learning_rate": 1.8546365914786967e-06,
+      "loss": 0.0002,
+      "step": 12936
+    },
+    {
+      "epoch": 0.8337307469227299,
+      "grad_norm": 0.023034146600674356,
+      "learning_rate": 1.8539205155746512e-06,
+      "loss": 0.0,
+      "step": 12937
+    },
+    {
+      "epoch": 0.833795192369659,
+      "grad_norm": 0.004499721841399995,
+      "learning_rate": 1.8532044396706053e-06,
+      "loss": 0.0,
+      "step": 12938
+    },
+    {
+      "epoch": 0.8338596378165882,
+      "grad_norm": 0.027223810403340872,
+      "learning_rate": 1.8524883637665594e-06,
+      "loss": 0.0,
+      "step": 12939
+    },
+    {
+      "epoch": 0.8339240832635174,
+      "grad_norm": 0.0010825319904780766,
+      "learning_rate": 1.8517722878625135e-06,
+      "loss": 0.0,
+      "step": 12940
+    },
+    {
+      "epoch": 0.8339885287104466,
+      "grad_norm": 0.00017965621067447417,
+      "learning_rate": 1.8510562119584678e-06,
+      "loss": 0.0,
+      "step": 12941
+    },
+    {
+      "epoch": 0.8340529741573758,
+      "grad_norm": 0.3216080980251922,
+      "learning_rate": 1.8503401360544219e-06,
+      "loss": 0.0006,
+      "step": 12942
+    },
+    {
+      "epoch": 0.834117419604305,
+      "grad_norm": 0.0010344771420191343,
+      "learning_rate": 1.8496240601503762e-06,
+      "loss": 0.0,
+      "step": 12943
+    },
+    {
+      "epoch": 0.8341818650512342,
+      "grad_norm": 0.0012871061826992497,
+      "learning_rate": 1.8489079842463303e-06,
+      "loss": 0.0,
+      "step": 12944
+    },
+    {
+      "epoch": 0.8342463104981633,
+      "grad_norm": 0.001207658190563529,
+      "learning_rate": 1.8481919083422846e-06,
+      "loss": 0.0,
+      "step": 12945
+    },
+    {
+      "epoch": 0.8343107559450925,
+      "grad_norm": 0.0007330767288757314,
+      "learning_rate": 1.8474758324382386e-06,
+      "loss": 0.0,
+      "step": 12946
+    },
+    {
+      "epoch": 0.8343752013920217,
+      "grad_norm": 0.004926647550943549,
+      "learning_rate": 1.8467597565341927e-06,
+      "loss": 0.0,
+      "step": 12947
+    },
+    {
+      "epoch": 0.8344396468389508,
+      "grad_norm": 0.01605856973953187,
+      "learning_rate": 1.8460436806301468e-06,
+      "loss": 0.0,
+      "step": 12948
+    },
+    {
+      "epoch": 0.83450409228588,
+      "grad_norm": 0.0005230443666051655,
+      "learning_rate": 1.8453276047261013e-06,
+      "loss": 0.0,
+      "step": 12949
+    },
+    {
+      "epoch": 0.8345685377328091,
+      "grad_norm": 0.0008104200669707647,
+      "learning_rate": 1.8446115288220554e-06,
+      "loss": 0.0,
+      "step": 12950
+    },
+    {
+      "epoch": 0.8346329831797383,
+      "grad_norm": 0.011457724398698297,
+      "learning_rate": 1.8438954529180095e-06,
+      "loss": 0.0,
+      "step": 12951
+    },
+    {
+      "epoch": 0.8346974286266675,
+      "grad_norm": 0.024955868465614733,
+      "learning_rate": 1.8431793770139636e-06,
+      "loss": 0.0003,
+      "step": 12952
+    },
+    {
+      "epoch": 0.8347618740735967,
+      "grad_norm": 0.0011789983636083115,
+      "learning_rate": 1.8424633011099177e-06,
+      "loss": 0.0,
+      "step": 12953
+    },
+    {
+      "epoch": 0.8348263195205259,
+      "grad_norm": 0.00127945397321626,
+      "learning_rate": 1.8417472252058718e-06,
+      "loss": 0.0,
+      "step": 12954
+    },
+    {
+      "epoch": 0.8348907649674551,
+      "grad_norm": 0.0005486498215946787,
+      "learning_rate": 1.8410311493018263e-06,
+      "loss": 0.0,
+      "step": 12955
+    },
+    {
+      "epoch": 0.8349552104143843,
+      "grad_norm": 0.0013076929794172173,
+      "learning_rate": 1.8403150733977804e-06,
+      "loss": 0.0,
+      "step": 12956
+    },
+    {
+      "epoch": 0.8350196558613134,
+      "grad_norm": 0.002699150201750239,
+      "learning_rate": 1.8395989974937344e-06,
+      "loss": 0.0,
+      "step": 12957
+    },
+    {
+      "epoch": 0.8350841013082426,
+      "grad_norm": 0.0014782031120580252,
+      "learning_rate": 1.8388829215896885e-06,
+      "loss": 0.0,
+      "step": 12958
+    },
+    {
+      "epoch": 0.8351485467551717,
+      "grad_norm": 0.00013325395203645215,
+      "learning_rate": 1.8381668456856428e-06,
+      "loss": 0.0,
+      "step": 12959
+    },
+    {
+      "epoch": 0.8352129922021009,
+      "grad_norm": 0.00028691027465177585,
+      "learning_rate": 1.837450769781597e-06,
+      "loss": 0.0,
+      "step": 12960
+    },
+    {
+      "epoch": 0.8352774376490301,
+      "grad_norm": 0.015239722260245718,
+      "learning_rate": 1.8367346938775512e-06,
+      "loss": 0.0001,
+      "step": 12961
+    },
+    {
+      "epoch": 0.8353418830959592,
+      "grad_norm": 0.01791622106760161,
+      "learning_rate": 1.8360186179735053e-06,
+      "loss": 0.0001,
+      "step": 12962
+    },
+    {
+      "epoch": 0.8354063285428884,
+      "grad_norm": 0.0010604469781897421,
+      "learning_rate": 1.8353025420694596e-06,
+      "loss": 0.0,
+      "step": 12963
+    },
+    {
+      "epoch": 0.8354707739898176,
+      "grad_norm": 0.0024505637205691246,
+      "learning_rate": 1.8345864661654137e-06,
+      "loss": 0.0,
+      "step": 12964
+    },
+    {
+      "epoch": 0.8355352194367468,
+      "grad_norm": 0.0013904709710636813,
+      "learning_rate": 1.8338703902613678e-06,
+      "loss": 0.0,
+      "step": 12965
+    },
+    {
+      "epoch": 0.835599664883676,
+      "grad_norm": 0.03134726683056867,
+      "learning_rate": 1.833154314357322e-06,
+      "loss": 0.0001,
+      "step": 12966
+    },
+    {
+      "epoch": 0.8356641103306052,
+      "grad_norm": 0.32471975430680294,
+      "learning_rate": 1.8324382384532764e-06,
+      "loss": 0.0007,
+      "step": 12967
+    },
+    {
+      "epoch": 0.8357285557775344,
+      "grad_norm": 0.46926647978812036,
+      "learning_rate": 1.8317221625492305e-06,
+      "loss": 0.0024,
+      "step": 12968
+    },
+    {
+      "epoch": 0.8357930012244635,
+      "grad_norm": 0.0007497435298024752,
+      "learning_rate": 1.8310060866451845e-06,
+      "loss": 0.0,
+      "step": 12969
+    },
+    {
+      "epoch": 0.8358574466713926,
+      "grad_norm": 8.529497298495229e-05,
+      "learning_rate": 1.8302900107411386e-06,
+      "loss": 0.0,
+      "step": 12970
+    },
+    {
+      "epoch": 0.8359218921183218,
+      "grad_norm": 0.010444437001298015,
+      "learning_rate": 1.8295739348370927e-06,
+      "loss": 0.0001,
+      "step": 12971
+    },
+    {
+      "epoch": 0.835986337565251,
+      "grad_norm": 0.0005033715826079014,
+      "learning_rate": 1.8288578589330472e-06,
+      "loss": 0.0,
+      "step": 12972
+    },
+    {
+      "epoch": 0.8360507830121802,
+      "grad_norm": 0.010451150054464193,
+      "learning_rate": 1.8281417830290013e-06,
+      "loss": 0.0,
+      "step": 12973
+    },
+    {
+      "epoch": 0.8361152284591093,
+      "grad_norm": 0.1669388847722097,
+      "learning_rate": 1.8274257071249554e-06,
+      "loss": 0.0002,
+      "step": 12974
+    },
+    {
+      "epoch": 0.8361796739060385,
+      "grad_norm": 3.708374839204555e-05,
+      "learning_rate": 1.8267096312209095e-06,
+      "loss": 0.0,
+      "step": 12975
+    },
+    {
+      "epoch": 0.8362441193529677,
+      "grad_norm": 0.024097830011570242,
+      "learning_rate": 1.8259935553168636e-06,
+      "loss": 0.0,
+      "step": 12976
+    },
+    {
+      "epoch": 0.8363085647998969,
+      "grad_norm": 0.00035244492069838926,
+      "learning_rate": 1.8252774794128179e-06,
+      "loss": 0.0,
+      "step": 12977
+    },
+    {
+      "epoch": 0.8363730102468261,
+      "grad_norm": 0.0006009073649762303,
+      "learning_rate": 1.8245614035087722e-06,
+      "loss": 0.0,
+      "step": 12978
+    },
+    {
+      "epoch": 0.8364374556937553,
+      "grad_norm": 0.002436471679603375,
+      "learning_rate": 1.8238453276047263e-06,
+      "loss": 0.0,
+      "step": 12979
+    },
+    {
+      "epoch": 0.8365019011406845,
+      "grad_norm": 0.0019328984019827203,
+      "learning_rate": 1.8231292517006803e-06,
+      "loss": 0.0,
+      "step": 12980
+    },
+    {
+      "epoch": 0.8365663465876135,
+      "grad_norm": 0.00047925150121626243,
+      "learning_rate": 1.8224131757966346e-06,
+      "loss": 0.0,
+      "step": 12981
+    },
+    {
+      "epoch": 0.8366307920345427,
+      "grad_norm": 0.0023810597907460897,
+      "learning_rate": 1.8216970998925887e-06,
+      "loss": 0.0,
+      "step": 12982
+    },
+    {
+      "epoch": 0.8366952374814719,
+      "grad_norm": 0.005150080251448669,
+      "learning_rate": 1.8209810239885428e-06,
+      "loss": 0.0,
+      "step": 12983
+    },
+    {
+      "epoch": 0.8367596829284011,
+      "grad_norm": 0.0005138275505945283,
+      "learning_rate": 1.8202649480844971e-06,
+      "loss": 0.0,
+      "step": 12984
+    },
+    {
+      "epoch": 0.8368241283753303,
+      "grad_norm": 0.0001395393045962735,
+      "learning_rate": 1.8195488721804514e-06,
+      "loss": 0.0,
+      "step": 12985
+    },
+    {
+      "epoch": 0.8368885738222595,
+      "grad_norm": 0.015921765880894,
+      "learning_rate": 1.8188327962764055e-06,
+      "loss": 0.0001,
+      "step": 12986
+    },
+    {
+      "epoch": 0.8369530192691886,
+      "grad_norm": 0.10171368698405213,
+      "learning_rate": 1.8181167203723596e-06,
+      "loss": 0.0009,
+      "step": 12987
+    },
+    {
+      "epoch": 0.8370174647161178,
+      "grad_norm": 0.003001050790157522,
+      "learning_rate": 1.8174006444683137e-06,
+      "loss": 0.0,
+      "step": 12988
+    },
+    {
+      "epoch": 0.837081910163047,
+      "grad_norm": 0.0007591269273822723,
+      "learning_rate": 1.8166845685642678e-06,
+      "loss": 0.0,
+      "step": 12989
+    },
+    {
+      "epoch": 0.8371463556099762,
+      "grad_norm": 0.0001454134195656195,
+      "learning_rate": 1.8159684926602223e-06,
+      "loss": 0.0,
+      "step": 12990
+    },
+    {
+      "epoch": 0.8372108010569054,
+      "grad_norm": 0.10966075031034136,
+      "learning_rate": 1.8152524167561764e-06,
+      "loss": 0.0001,
+      "step": 12991
+    },
+    {
+      "epoch": 0.8372752465038346,
+      "grad_norm": 0.004862647588603843,
+      "learning_rate": 1.8145363408521305e-06,
+      "loss": 0.0,
+      "step": 12992
+    },
+    {
+      "epoch": 0.8373396919507636,
+      "grad_norm": 0.19124268335483594,
+      "learning_rate": 1.8138202649480845e-06,
+      "loss": 0.0006,
+      "step": 12993
+    },
+    {
+      "epoch": 0.8374041373976928,
+      "grad_norm": 0.00014524860075802834,
+      "learning_rate": 1.8131041890440388e-06,
+      "loss": 0.0,
+      "step": 12994
+    },
+    {
+      "epoch": 0.837468582844622,
+      "grad_norm": 0.04439166994184201,
+      "learning_rate": 1.812388113139993e-06,
+      "loss": 0.0001,
+      "step": 12995
+    },
+    {
+      "epoch": 0.8375330282915512,
+      "grad_norm": 6.69389849704551e-05,
+      "learning_rate": 1.8116720372359472e-06,
+      "loss": 0.0,
+      "step": 12996
+    },
+    {
+      "epoch": 0.8375974737384804,
+      "grad_norm": 0.001974699603729132,
+      "learning_rate": 1.8109559613319013e-06,
+      "loss": 0.0,
+      "step": 12997
+    },
+    {
+      "epoch": 0.8376619191854096,
+      "grad_norm": 0.00018389771702319728,
+      "learning_rate": 1.8102398854278556e-06,
+      "loss": 0.0,
+      "step": 12998
+    },
+    {
+      "epoch": 0.8377263646323387,
+      "grad_norm": 0.12907701798589907,
+      "learning_rate": 1.8095238095238097e-06,
+      "loss": 0.0002,
+      "step": 12999
+    },
+    {
+      "epoch": 0.8377908100792679,
+      "grad_norm": 0.23746540765057983,
+      "learning_rate": 1.8088077336197638e-06,
+      "loss": 0.0008,
+      "step": 13000
+    },
+    {
+      "epoch": 0.8378552555261971,
+      "grad_norm": 0.05998519156857347,
+      "learning_rate": 1.808091657715718e-06,
+      "loss": 0.0001,
+      "step": 13001
+    },
+    {
+      "epoch": 0.8379197009731263,
+      "grad_norm": 0.001119847624552485,
+      "learning_rate": 1.8073755818116724e-06,
+      "loss": 0.0,
+      "step": 13002
+    },
+    {
+      "epoch": 0.8379841464200555,
+      "grad_norm": 0.46324393715573475,
+      "learning_rate": 1.8066595059076265e-06,
+      "loss": 0.0016,
+      "step": 13003
+    },
+    {
+      "epoch": 0.8380485918669845,
+      "grad_norm": 0.002904501204279146,
+      "learning_rate": 1.8059434300035806e-06,
+      "loss": 0.0,
+      "step": 13004
+    },
+    {
+      "epoch": 0.8381130373139137,
+      "grad_norm": 0.0032653766814766654,
+      "learning_rate": 1.8052273540995346e-06,
+      "loss": 0.0,
+      "step": 13005
+    },
+    {
+      "epoch": 0.8381774827608429,
+      "grad_norm": 0.000118351637666897,
+      "learning_rate": 1.8045112781954887e-06,
+      "loss": 0.0,
+      "step": 13006
+    },
+    {
+      "epoch": 0.8382419282077721,
+      "grad_norm": 0.09376592798824523,
+      "learning_rate": 1.8037952022914432e-06,
+      "loss": 0.0002,
+      "step": 13007
+    },
+    {
+      "epoch": 0.8383063736547013,
+      "grad_norm": 0.0040541966063241675,
+      "learning_rate": 1.8030791263873973e-06,
+      "loss": 0.0,
+      "step": 13008
+    },
+    {
+      "epoch": 0.8383708191016305,
+      "grad_norm": 0.013738120997783352,
+      "learning_rate": 1.8023630504833514e-06,
+      "loss": 0.0001,
+      "step": 13009
+    },
+    {
+      "epoch": 0.8384352645485597,
+      "grad_norm": 0.0031340080052982603,
+      "learning_rate": 1.8016469745793055e-06,
+      "loss": 0.0,
+      "step": 13010
+    },
+    {
+      "epoch": 0.8384997099954888,
+      "grad_norm": 0.0009455405513584888,
+      "learning_rate": 1.8009308986752596e-06,
+      "loss": 0.0,
+      "step": 13011
+    },
+    {
+      "epoch": 0.838564155442418,
+      "grad_norm": 0.0036418512175713817,
+      "learning_rate": 1.8002148227712139e-06,
+      "loss": 0.0,
+      "step": 13012
+    },
+    {
+      "epoch": 0.8386286008893472,
+      "grad_norm": 0.0006205554728098501,
+      "learning_rate": 1.7994987468671682e-06,
+      "loss": 0.0,
+      "step": 13013
+    },
+    {
+      "epoch": 0.8386930463362764,
+      "grad_norm": 0.0025184925414780623,
+      "learning_rate": 1.7987826709631223e-06,
+      "loss": 0.0,
+      "step": 13014
+    },
+    {
+      "epoch": 0.8387574917832055,
+      "grad_norm": 9.977119185934306e-05,
+      "learning_rate": 1.7980665950590764e-06,
+      "loss": 0.0,
+      "step": 13015
+    },
+    {
+      "epoch": 0.8388219372301347,
+      "grad_norm": 0.001322791226674588,
+      "learning_rate": 1.7973505191550307e-06,
+      "loss": 0.0,
+      "step": 13016
+    },
+    {
+      "epoch": 0.8388863826770638,
+      "grad_norm": 9.753431868816942e-05,
+      "learning_rate": 1.7966344432509847e-06,
+      "loss": 0.0,
+      "step": 13017
+    },
+    {
+      "epoch": 0.838950828123993,
+      "grad_norm": 0.024187093953887785,
+      "learning_rate": 1.7959183673469388e-06,
+      "loss": 0.0,
+      "step": 13018
+    },
+    {
+      "epoch": 0.8390152735709222,
+      "grad_norm": 0.0008915200755770839,
+      "learning_rate": 1.7952022914428931e-06,
+      "loss": 0.0,
+      "step": 13019
+    },
+    {
+      "epoch": 0.8390797190178514,
+      "grad_norm": 0.1700030878168758,
+      "learning_rate": 1.7944862155388474e-06,
+      "loss": 0.0021,
+      "step": 13020
+    },
+    {
+      "epoch": 0.8391441644647806,
+      "grad_norm": 0.009521015167901947,
+      "learning_rate": 1.7937701396348015e-06,
+      "loss": 0.0001,
+      "step": 13021
+    },
+    {
+      "epoch": 0.8392086099117098,
+      "grad_norm": 0.0009619655795215219,
+      "learning_rate": 1.7930540637307556e-06,
+      "loss": 0.0,
+      "step": 13022
+    },
+    {
+      "epoch": 0.8392730553586389,
+      "grad_norm": 0.04029302332851765,
+      "learning_rate": 1.7923379878267097e-06,
+      "loss": 0.0,
+      "step": 13023
+    },
+    {
+      "epoch": 0.8393375008055681,
+      "grad_norm": 0.0386175096813077,
+      "learning_rate": 1.7916219119226638e-06,
+      "loss": 0.0,
+      "step": 13024
+    },
+    {
+      "epoch": 0.8394019462524973,
+      "grad_norm": 4.674073385948886e-05,
+      "learning_rate": 1.7909058360186183e-06,
+      "loss": 0.0,
+      "step": 13025
+    },
+    {
+      "epoch": 0.8394663916994264,
+      "grad_norm": 0.13500024579167066,
+      "learning_rate": 1.7901897601145724e-06,
+      "loss": 0.0003,
+      "step": 13026
+    },
+    {
+      "epoch": 0.8395308371463556,
+      "grad_norm": 0.0024757972109323823,
+      "learning_rate": 1.7894736842105265e-06,
+      "loss": 0.0,
+      "step": 13027
+    },
+    {
+      "epoch": 0.8395952825932848,
+      "grad_norm": 0.0021501952153607775,
+      "learning_rate": 1.7887576083064805e-06,
+      "loss": 0.0,
+      "step": 13028
+    },
+    {
+      "epoch": 0.8396597280402139,
+      "grad_norm": 0.0008697595381524505,
+      "learning_rate": 1.7880415324024346e-06,
+      "loss": 0.0,
+      "step": 13029
+    },
+    {
+      "epoch": 0.8397241734871431,
+      "grad_norm": 0.4445862117639593,
+      "learning_rate": 1.787325456498389e-06,
+      "loss": 0.0035,
+      "step": 13030
+    },
+    {
+      "epoch": 0.8397886189340723,
+      "grad_norm": 0.36612350437383684,
+      "learning_rate": 1.7866093805943432e-06,
+      "loss": 0.0016,
+      "step": 13031
+    },
+    {
+      "epoch": 0.8398530643810015,
+      "grad_norm": 0.012165564798806148,
+      "learning_rate": 1.7858933046902973e-06,
+      "loss": 0.0,
+      "step": 13032
+    },
+    {
+      "epoch": 0.8399175098279307,
+      "grad_norm": 0.0779756907391507,
+      "learning_rate": 1.7851772287862514e-06,
+      "loss": 0.0001,
+      "step": 13033
+    },
+    {
+      "epoch": 0.8399819552748599,
+      "grad_norm": 0.026000581433871277,
+      "learning_rate": 1.7844611528822057e-06,
+      "loss": 0.0015,
+      "step": 13034
+    },
+    {
+      "epoch": 0.840046400721789,
+      "grad_norm": 0.2525092647996164,
+      "learning_rate": 1.7837450769781598e-06,
+      "loss": 0.0005,
+      "step": 13035
+    },
+    {
+      "epoch": 0.8401108461687182,
+      "grad_norm": 0.0005119127002491398,
+      "learning_rate": 1.7830290010741139e-06,
+      "loss": 0.0,
+      "step": 13036
+    },
+    {
+      "epoch": 0.8401752916156473,
+      "grad_norm": 0.0020882671639366607,
+      "learning_rate": 1.7823129251700682e-06,
+      "loss": 0.0,
+      "step": 13037
+    },
+    {
+      "epoch": 0.8402397370625765,
+      "grad_norm": 0.01590513585309205,
+      "learning_rate": 1.7815968492660225e-06,
+      "loss": 0.0,
+      "step": 13038
+    },
+    {
+      "epoch": 0.8403041825095057,
+      "grad_norm": 0.0003583728461652098,
+      "learning_rate": 1.7808807733619766e-06,
+      "loss": 0.0,
+      "step": 13039
+    },
+    {
+      "epoch": 0.8403686279564349,
+      "grad_norm": 0.0017618107669094471,
+      "learning_rate": 1.7801646974579306e-06,
+      "loss": 0.0,
+      "step": 13040
+    },
+    {
+      "epoch": 0.840433073403364,
+      "grad_norm": 0.0051314915355801785,
+      "learning_rate": 1.7794486215538847e-06,
+      "loss": 0.0,
+      "step": 13041
+    },
+    {
+      "epoch": 0.8404975188502932,
+      "grad_norm": 0.0015903867467092674,
+      "learning_rate": 1.7787325456498392e-06,
+      "loss": 0.0,
+      "step": 13042
+    },
+    {
+      "epoch": 0.8405619642972224,
+      "grad_norm": 0.0004816465431161624,
+      "learning_rate": 1.7780164697457933e-06,
+      "loss": 0.0,
+      "step": 13043
+    },
+    {
+      "epoch": 0.8406264097441516,
+      "grad_norm": 0.006361106917538149,
+      "learning_rate": 1.7773003938417474e-06,
+      "loss": 0.0,
+      "step": 13044
+    },
+    {
+      "epoch": 0.8406908551910808,
+      "grad_norm": 0.035033549990959596,
+      "learning_rate": 1.7765843179377015e-06,
+      "loss": 0.0001,
+      "step": 13045
+    },
+    {
+      "epoch": 0.84075530063801,
+      "grad_norm": 0.005053168074157976,
+      "learning_rate": 1.7758682420336556e-06,
+      "loss": 0.0,
+      "step": 13046
+    },
+    {
+      "epoch": 0.8408197460849391,
+      "grad_norm": 0.1503405115928112,
+      "learning_rate": 1.7751521661296097e-06,
+      "loss": 0.0003,
+      "step": 13047
+    },
+    {
+      "epoch": 0.8408841915318682,
+      "grad_norm": 0.8358501852193341,
+      "learning_rate": 1.7744360902255642e-06,
+      "loss": 0.0055,
+      "step": 13048
+    },
+    {
+      "epoch": 0.8409486369787974,
+      "grad_norm": 0.0012061131465209045,
+      "learning_rate": 1.7737200143215183e-06,
+      "loss": 0.0,
+      "step": 13049
+    },
+    {
+      "epoch": 0.8410130824257266,
+      "grad_norm": 0.030221909655107952,
+      "learning_rate": 1.7730039384174724e-06,
+      "loss": 0.0,
+      "step": 13050
+    },
+    {
+      "epoch": 0.8410775278726558,
+      "grad_norm": 0.0003382566794423338,
+      "learning_rate": 1.7722878625134265e-06,
+      "loss": 0.0,
+      "step": 13051
+    },
+    {
+      "epoch": 0.841141973319585,
+      "grad_norm": 0.001474604475494994,
+      "learning_rate": 1.7715717866093807e-06,
+      "loss": 0.0,
+      "step": 13052
+    },
+    {
+      "epoch": 0.8412064187665141,
+      "grad_norm": 0.0005242323894890739,
+      "learning_rate": 1.7708557107053348e-06,
+      "loss": 0.0,
+      "step": 13053
+    },
+    {
+      "epoch": 0.8412708642134433,
+      "grad_norm": 0.01598698800507151,
+      "learning_rate": 1.7701396348012891e-06,
+      "loss": 0.0,
+      "step": 13054
+    },
+    {
+      "epoch": 0.8413353096603725,
+      "grad_norm": 0.02646729332155176,
+      "learning_rate": 1.7694235588972432e-06,
+      "loss": 0.0,
+      "step": 13055
+    },
+    {
+      "epoch": 0.8413997551073017,
+      "grad_norm": 0.0004705005211530495,
+      "learning_rate": 1.7687074829931975e-06,
+      "loss": 0.0,
+      "step": 13056
+    },
+    {
+      "epoch": 0.8414642005542309,
+      "grad_norm": 0.2572792781261736,
+      "learning_rate": 1.7679914070891516e-06,
+      "loss": 0.001,
+      "step": 13057
+    },
+    {
+      "epoch": 0.8415286460011601,
+      "grad_norm": 0.003987081818030144,
+      "learning_rate": 1.7672753311851057e-06,
+      "loss": 0.0,
+      "step": 13058
+    },
+    {
+      "epoch": 0.8415930914480892,
+      "grad_norm": 0.017189671808628206,
+      "learning_rate": 1.7665592552810598e-06,
+      "loss": 0.0,
+      "step": 13059
+    },
+    {
+      "epoch": 0.8416575368950183,
+      "grad_norm": 0.038183230144691487,
+      "learning_rate": 1.7658431793770143e-06,
+      "loss": 0.0,
+      "step": 13060
+    },
+    {
+      "epoch": 0.8417219823419475,
+      "grad_norm": 0.041134170528374744,
+      "learning_rate": 1.7651271034729684e-06,
+      "loss": 0.0001,
+      "step": 13061
+    },
+    {
+      "epoch": 0.8417864277888767,
+      "grad_norm": 0.08703766028306478,
+      "learning_rate": 1.7644110275689225e-06,
+      "loss": 0.0002,
+      "step": 13062
+    },
+    {
+      "epoch": 0.8418508732358059,
+      "grad_norm": 0.00010780192104593178,
+      "learning_rate": 1.7636949516648766e-06,
+      "loss": 0.0,
+      "step": 13063
+    },
+    {
+      "epoch": 0.8419153186827351,
+      "grad_norm": 0.11928142763613796,
+      "learning_rate": 1.7629788757608306e-06,
+      "loss": 0.0004,
+      "step": 13064
+    },
+    {
+      "epoch": 0.8419797641296642,
+      "grad_norm": 0.5486655215015411,
+      "learning_rate": 1.762262799856785e-06,
+      "loss": 0.0035,
+      "step": 13065
+    },
+    {
+      "epoch": 0.8420442095765934,
+      "grad_norm": 0.005871428197205636,
+      "learning_rate": 1.7615467239527392e-06,
+      "loss": 0.0,
+      "step": 13066
+    },
+    {
+      "epoch": 0.8421086550235226,
+      "grad_norm": 0.0003421205363785894,
+      "learning_rate": 1.7608306480486933e-06,
+      "loss": 0.0,
+      "step": 13067
+    },
+    {
+      "epoch": 0.8421731004704518,
+      "grad_norm": 0.00034174588045342405,
+      "learning_rate": 1.7601145721446474e-06,
+      "loss": 0.0,
+      "step": 13068
+    },
+    {
+      "epoch": 0.842237545917381,
+      "grad_norm": 0.004250460203295504,
+      "learning_rate": 1.7593984962406017e-06,
+      "loss": 0.0,
+      "step": 13069
+    },
+    {
+      "epoch": 0.8423019913643102,
+      "grad_norm": 0.0015205250632764336,
+      "learning_rate": 1.7586824203365558e-06,
+      "loss": 0.0,
+      "step": 13070
+    },
+    {
+      "epoch": 0.8423664368112392,
+      "grad_norm": 0.004258007696294317,
+      "learning_rate": 1.7579663444325099e-06,
+      "loss": 0.0,
+      "step": 13071
+    },
+    {
+      "epoch": 0.8424308822581684,
+      "grad_norm": 0.0034072375699601702,
+      "learning_rate": 1.7572502685284642e-06,
+      "loss": 0.0,
+      "step": 13072
+    },
+    {
+      "epoch": 0.8424953277050976,
+      "grad_norm": 0.0014717670616816948,
+      "learning_rate": 1.7565341926244185e-06,
+      "loss": 0.0,
+      "step": 13073
+    },
+    {
+      "epoch": 0.8425597731520268,
+      "grad_norm": 0.13174503866161089,
+      "learning_rate": 1.7558181167203726e-06,
+      "loss": 0.0001,
+      "step": 13074
+    },
+    {
+      "epoch": 0.842624218598956,
+      "grad_norm": 0.13783244480645623,
+      "learning_rate": 1.7551020408163267e-06,
+      "loss": 0.0013,
+      "step": 13075
+    },
+    {
+      "epoch": 0.8426886640458852,
+      "grad_norm": 0.005166068120200246,
+      "learning_rate": 1.7543859649122807e-06,
+      "loss": 0.0,
+      "step": 13076
+    },
+    {
+      "epoch": 0.8427531094928143,
+      "grad_norm": 0.0006774612637503077,
+      "learning_rate": 1.7536698890082353e-06,
+      "loss": 0.0,
+      "step": 13077
+    },
+    {
+      "epoch": 0.8428175549397435,
+      "grad_norm": 0.17182032842691972,
+      "learning_rate": 1.7529538131041893e-06,
+      "loss": 0.0017,
+      "step": 13078
+    },
+    {
+      "epoch": 0.8428820003866727,
+      "grad_norm": 0.04182167002943656,
+      "learning_rate": 1.7522377372001434e-06,
+      "loss": 0.0002,
+      "step": 13079
+    },
+    {
+      "epoch": 0.8429464458336019,
+      "grad_norm": 0.0244410842129277,
+      "learning_rate": 1.7515216612960975e-06,
+      "loss": 0.0001,
+      "step": 13080
+    },
+    {
+      "epoch": 0.8430108912805311,
+      "grad_norm": 0.010742733410488706,
+      "learning_rate": 1.7508055853920516e-06,
+      "loss": 0.0,
+      "step": 13081
+    },
+    {
+      "epoch": 0.8430753367274602,
+      "grad_norm": 0.010742733410488706,
+      "learning_rate": 1.7508055853920516e-06,
+      "loss": 0.0257,
+      "step": 13082
+    },
+    {
+      "epoch": 0.8431397821743893,
+      "grad_norm": 7.704857180219094e-05,
+      "learning_rate": 1.7500895094880057e-06,
+      "loss": 0.0,
+      "step": 13083
+    },
+    {
+      "epoch": 0.8432042276213185,
+      "grad_norm": 0.06449464703172729,
+      "learning_rate": 1.7493734335839602e-06,
+      "loss": 0.0001,
+      "step": 13084
+    },
+    {
+      "epoch": 0.8432686730682477,
+      "grad_norm": 0.000153695612872603,
+      "learning_rate": 1.7486573576799143e-06,
+      "loss": 0.0,
+      "step": 13085
+    },
+    {
+      "epoch": 0.8433331185151769,
+      "grad_norm": 0.004558501781863454,
+      "learning_rate": 1.7479412817758684e-06,
+      "loss": 0.0,
+      "step": 13086
+    },
+    {
+      "epoch": 0.8433975639621061,
+      "grad_norm": 0.0014270143678160212,
+      "learning_rate": 1.7472252058718225e-06,
+      "loss": 0.0,
+      "step": 13087
+    },
+    {
+      "epoch": 0.8434620094090353,
+      "grad_norm": 0.0021367750367809737,
+      "learning_rate": 1.7465091299677768e-06,
+      "loss": 0.0,
+      "step": 13088
+    },
+    {
+      "epoch": 0.8435264548559644,
+      "grad_norm": 0.06312020112435841,
+      "learning_rate": 1.7457930540637308e-06,
+      "loss": 0.0001,
+      "step": 13089
+    },
+    {
+      "epoch": 0.8435909003028936,
+      "grad_norm": 0.0005994224308608808,
+      "learning_rate": 1.7450769781596851e-06,
+      "loss": 0.0,
+      "step": 13090
+    },
+    {
+      "epoch": 0.8436553457498228,
+      "grad_norm": 0.008210769671738014,
+      "learning_rate": 1.7443609022556392e-06,
+      "loss": 0.0,
+      "step": 13091
+    },
+    {
+      "epoch": 0.843719791196752,
+      "grad_norm": 0.009704245330285514,
+      "learning_rate": 1.7436448263515935e-06,
+      "loss": 0.0,
+      "step": 13092
+    },
+    {
+      "epoch": 0.8437842366436811,
+      "grad_norm": 0.45894062564205756,
+      "learning_rate": 1.7429287504475476e-06,
+      "loss": 0.0032,
+      "step": 13093
+    },
+    {
+      "epoch": 0.8438486820906103,
+      "grad_norm": 0.0023116636587011403,
+      "learning_rate": 1.7422126745435017e-06,
+      "loss": 0.0,
+      "step": 13094
+    },
+    {
+      "epoch": 0.8439131275375394,
+      "grad_norm": 0.002211421761670453,
+      "learning_rate": 1.7414965986394558e-06,
+      "loss": 0.0,
+      "step": 13095
+    },
+    {
+      "epoch": 0.8439775729844686,
+      "grad_norm": 0.01702725153815147,
+      "learning_rate": 1.7407805227354103e-06,
+      "loss": 0.0,
+      "step": 13096
+    },
+    {
+      "epoch": 0.8440420184313978,
+      "grad_norm": 0.0001140036103517741,
+      "learning_rate": 1.7400644468313644e-06,
+      "loss": 0.0,
+      "step": 13097
+    },
+    {
+      "epoch": 0.844106463878327,
+      "grad_norm": 0.10455236751921528,
+      "learning_rate": 1.7393483709273185e-06,
+      "loss": 0.0018,
+      "step": 13098
+    },
+    {
+      "epoch": 0.8441709093252562,
+      "grad_norm": 0.002027961693557616,
+      "learning_rate": 1.7386322950232726e-06,
+      "loss": 0.0,
+      "step": 13099
+    },
+    {
+      "epoch": 0.8442353547721854,
+      "grad_norm": 0.00024180815422684246,
+      "learning_rate": 1.7379162191192266e-06,
+      "loss": 0.0,
+      "step": 13100
+    },
+    {
+      "epoch": 0.8442998002191145,
+      "grad_norm": 0.079310239055661,
+      "learning_rate": 1.7372001432151807e-06,
+      "loss": 0.0001,
+      "step": 13101
+    },
+    {
+      "epoch": 0.8443642456660437,
+      "grad_norm": 0.00905125777422377,
+      "learning_rate": 1.7364840673111352e-06,
+      "loss": 0.0,
+      "step": 13102
+    },
+    {
+      "epoch": 0.8444286911129729,
+      "grad_norm": 0.2213289704661812,
+      "learning_rate": 1.7357679914070893e-06,
+      "loss": 0.0009,
+      "step": 13103
+    },
+    {
+      "epoch": 0.844493136559902,
+      "grad_norm": 0.013132135644655554,
+      "learning_rate": 1.7350519155030434e-06,
+      "loss": 0.0,
+      "step": 13104
+    },
+    {
+      "epoch": 0.8445575820068312,
+      "grad_norm": 0.0019095839276892347,
+      "learning_rate": 1.7343358395989975e-06,
+      "loss": 0.0,
+      "step": 13105
+    },
+    {
+      "epoch": 0.8446220274537604,
+      "grad_norm": 0.003959361517452852,
+      "learning_rate": 1.7336197636949518e-06,
+      "loss": 0.0,
+      "step": 13106
+    },
+    {
+      "epoch": 0.8446864729006895,
+      "grad_norm": 0.0032159310611906385,
+      "learning_rate": 1.7329036877909059e-06,
+      "loss": 0.0,
+      "step": 13107
+    },
+    {
+      "epoch": 0.8447509183476187,
+      "grad_norm": 0.0019923615717410644,
+      "learning_rate": 1.7321876118868602e-06,
+      "loss": 0.0,
+      "step": 13108
+    },
+    {
+      "epoch": 0.8448153637945479,
+      "grad_norm": 0.004360544594844355,
+      "learning_rate": 1.7314715359828143e-06,
+      "loss": 0.0,
+      "step": 13109
+    },
+    {
+      "epoch": 0.8448798092414771,
+      "grad_norm": 0.0004440588758890354,
+      "learning_rate": 1.7307554600787686e-06,
+      "loss": 0.0,
+      "step": 13110
+    },
+    {
+      "epoch": 0.8449442546884063,
+      "grad_norm": 0.05991632583505031,
+      "learning_rate": 1.7300393841747227e-06,
+      "loss": 0.0001,
+      "step": 13111
+    },
+    {
+      "epoch": 0.8450087001353355,
+      "grad_norm": 0.0005124873215020824,
+      "learning_rate": 1.7293233082706767e-06,
+      "loss": 0.0,
+      "step": 13112
+    },
+    {
+      "epoch": 0.8450731455822647,
+      "grad_norm": 0.004131471238295923,
+      "learning_rate": 1.728607232366631e-06,
+      "loss": 0.0,
+      "step": 13113
+    },
+    {
+      "epoch": 0.8451375910291938,
+      "grad_norm": 0.002933325315730856,
+      "learning_rate": 1.7278911564625853e-06,
+      "loss": 0.0,
+      "step": 13114
+    },
+    {
+      "epoch": 0.8452020364761229,
+      "grad_norm": 0.0019792287577856935,
+      "learning_rate": 1.7271750805585394e-06,
+      "loss": 0.0,
+      "step": 13115
+    },
+    {
+      "epoch": 0.8452664819230521,
+      "grad_norm": 0.022908054626963152,
+      "learning_rate": 1.7264590046544935e-06,
+      "loss": 0.0,
+      "step": 13116
+    },
+    {
+      "epoch": 0.8453309273699813,
+      "grad_norm": 0.001399493420350886,
+      "learning_rate": 1.7257429287504476e-06,
+      "loss": 0.0,
+      "step": 13117
+    },
+    {
+      "epoch": 0.8453953728169105,
+      "grad_norm": 0.001018219180645176,
+      "learning_rate": 1.7250268528464017e-06,
+      "loss": 0.0,
+      "step": 13118
+    },
+    {
+      "epoch": 0.8454598182638396,
+      "grad_norm": 0.04616747542702802,
+      "learning_rate": 1.7243107769423562e-06,
+      "loss": 0.0,
+      "step": 13119
+    },
+    {
+      "epoch": 0.8455242637107688,
+      "grad_norm": 0.009493654566610898,
+      "learning_rate": 1.7235947010383103e-06,
+      "loss": 0.0,
+      "step": 13120
+    },
+    {
+      "epoch": 0.845588709157698,
+      "grad_norm": 0.00967814859014821,
+      "learning_rate": 1.7228786251342644e-06,
+      "loss": 0.0,
+      "step": 13121
+    },
+    {
+      "epoch": 0.8456531546046272,
+      "grad_norm": 0.10150213006238015,
+      "learning_rate": 1.7221625492302185e-06,
+      "loss": 0.0003,
+      "step": 13122
+    },
+    {
+      "epoch": 0.8457176000515564,
+      "grad_norm": 0.6385996833154148,
+      "learning_rate": 1.7214464733261726e-06,
+      "loss": 0.0013,
+      "step": 13123
+    },
+    {
+      "epoch": 0.8457820454984856,
+      "grad_norm": 0.027547564191873486,
+      "learning_rate": 1.7207303974221269e-06,
+      "loss": 0.0001,
+      "step": 13124
+    },
+    {
+      "epoch": 0.8458464909454148,
+      "grad_norm": 0.04341508013828472,
+      "learning_rate": 1.7200143215180811e-06,
+      "loss": 0.0003,
+      "step": 13125
+    },
+    {
+      "epoch": 0.8459109363923438,
+      "grad_norm": 0.00016965293402197205,
+      "learning_rate": 1.7192982456140352e-06,
+      "loss": 0.0,
+      "step": 13126
+    },
+    {
+      "epoch": 0.845975381839273,
+      "grad_norm": 0.002586838111347734,
+      "learning_rate": 1.7185821697099893e-06,
+      "loss": 0.0,
+      "step": 13127
+    },
+    {
+      "epoch": 0.8460398272862022,
+      "grad_norm": 0.02806070046606198,
+      "learning_rate": 1.7178660938059436e-06,
+      "loss": 0.0003,
+      "step": 13128
+    },
+    {
+      "epoch": 0.8461042727331314,
+      "grad_norm": 0.00040487585066894285,
+      "learning_rate": 1.7171500179018977e-06,
+      "loss": 0.0,
+      "step": 13129
+    },
+    {
+      "epoch": 0.8461687181800606,
+      "grad_norm": 0.0041082910374663,
+      "learning_rate": 1.7164339419978518e-06,
+      "loss": 0.0,
+      "step": 13130
+    },
+    {
+      "epoch": 0.8462331636269897,
+      "grad_norm": 0.0015562831145490758,
+      "learning_rate": 1.715717866093806e-06,
+      "loss": 0.0,
+      "step": 13131
+    },
+    {
+      "epoch": 0.8462976090739189,
+      "grad_norm": 0.0009302798442001268,
+      "learning_rate": 1.7150017901897604e-06,
+      "loss": 0.0,
+      "step": 13132
+    },
+    {
+      "epoch": 0.8463620545208481,
+      "grad_norm": 0.01062960002645244,
+      "learning_rate": 1.7142857142857145e-06,
+      "loss": 0.0,
+      "step": 13133
+    },
+    {
+      "epoch": 0.8464264999677773,
+      "grad_norm": 0.00834073110916737,
+      "learning_rate": 1.7135696383816686e-06,
+      "loss": 0.0,
+      "step": 13134
+    },
+    {
+      "epoch": 0.8464909454147065,
+      "grad_norm": 0.0009627436988840297,
+      "learning_rate": 1.7128535624776227e-06,
+      "loss": 0.0,
+      "step": 13135
+    },
+    {
+      "epoch": 0.8465553908616357,
+      "grad_norm": 0.0009720207341823486,
+      "learning_rate": 1.7121374865735767e-06,
+      "loss": 0.0,
+      "step": 13136
+    },
+    {
+      "epoch": 0.8466198363085649,
+      "grad_norm": 0.007233971164138382,
+      "learning_rate": 1.7114214106695313e-06,
+      "loss": 0.0,
+      "step": 13137
+    },
+    {
+      "epoch": 0.8466842817554939,
+      "grad_norm": 0.006121266302367881,
+      "learning_rate": 1.7107053347654853e-06,
+      "loss": 0.0,
+      "step": 13138
+    },
+    {
+      "epoch": 0.8467487272024231,
+      "grad_norm": 0.00019090575891950602,
+      "learning_rate": 1.7099892588614394e-06,
+      "loss": 0.0,
+      "step": 13139
+    },
+    {
+      "epoch": 0.8468131726493523,
+      "grad_norm": 0.3189073772494604,
+      "learning_rate": 1.7092731829573935e-06,
+      "loss": 0.0069,
+      "step": 13140
+    },
+    {
+      "epoch": 0.8468776180962815,
+      "grad_norm": 0.00029818719882822894,
+      "learning_rate": 1.7085571070533478e-06,
+      "loss": 0.0,
+      "step": 13141
+    },
+    {
+      "epoch": 0.8469420635432107,
+      "grad_norm": 0.02607671581948451,
+      "learning_rate": 1.707841031149302e-06,
+      "loss": 0.0001,
+      "step": 13142
+    },
+    {
+      "epoch": 0.8470065089901399,
+      "grad_norm": 0.10383467270417568,
+      "learning_rate": 1.7071249552452562e-06,
+      "loss": 0.0007,
+      "step": 13143
+    },
+    {
+      "epoch": 0.847070954437069,
+      "grad_norm": 0.00030382140353013776,
+      "learning_rate": 1.7064088793412103e-06,
+      "loss": 0.0,
+      "step": 13144
+    },
+    {
+      "epoch": 0.8471353998839982,
+      "grad_norm": 0.006279257922826684,
+      "learning_rate": 1.7056928034371646e-06,
+      "loss": 0.0,
+      "step": 13145
+    },
+    {
+      "epoch": 0.8471998453309274,
+      "grad_norm": 0.45831563402449355,
+      "learning_rate": 1.7049767275331187e-06,
+      "loss": 0.0036,
+      "step": 13146
+    },
+    {
+      "epoch": 0.8472642907778566,
+      "grad_norm": 0.004021349795027649,
+      "learning_rate": 1.7042606516290728e-06,
+      "loss": 0.0,
+      "step": 13147
+    },
+    {
+      "epoch": 0.8473287362247858,
+      "grad_norm": 0.03808135836534344,
+      "learning_rate": 1.703544575725027e-06,
+      "loss": 0.0,
+      "step": 13148
+    },
+    {
+      "epoch": 0.8473931816717148,
+      "grad_norm": 0.11625193826279853,
+      "learning_rate": 1.7028284998209814e-06,
+      "loss": 0.0017,
+      "step": 13149
+    },
+    {
+      "epoch": 0.847457627118644,
+      "grad_norm": 0.00826856773484578,
+      "learning_rate": 1.7021124239169354e-06,
+      "loss": 0.0001,
+      "step": 13150
+    },
+    {
+      "epoch": 0.8475220725655732,
+      "grad_norm": 0.06094455670493373,
+      "learning_rate": 1.7013963480128895e-06,
+      "loss": 0.0001,
+      "step": 13151
+    },
+    {
+      "epoch": 0.8475865180125024,
+      "grad_norm": 0.32370913335997803,
+      "learning_rate": 1.7006802721088436e-06,
+      "loss": 0.0005,
+      "step": 13152
+    },
+    {
+      "epoch": 0.8476509634594316,
+      "grad_norm": 0.004277870466133441,
+      "learning_rate": 1.6999641962047977e-06,
+      "loss": 0.0,
+      "step": 13153
+    },
+    {
+      "epoch": 0.8477154089063608,
+      "grad_norm": 0.006180712845803867,
+      "learning_rate": 1.6992481203007522e-06,
+      "loss": 0.0001,
+      "step": 13154
+    },
+    {
+      "epoch": 0.84777985435329,
+      "grad_norm": 0.00029611523936459857,
+      "learning_rate": 1.6985320443967063e-06,
+      "loss": 0.0,
+      "step": 13155
+    },
+    {
+      "epoch": 0.8478442998002191,
+      "grad_norm": 0.0003400751143154014,
+      "learning_rate": 1.6978159684926604e-06,
+      "loss": 0.0,
+      "step": 13156
+    },
+    {
+      "epoch": 0.8479087452471483,
+      "grad_norm": 0.04685887416980266,
+      "learning_rate": 1.6970998925886145e-06,
+      "loss": 0.0001,
+      "step": 13157
+    },
+    {
+      "epoch": 0.8479731906940775,
+      "grad_norm": 0.004478614412734495,
+      "learning_rate": 1.6963838166845686e-06,
+      "loss": 0.0,
+      "step": 13158
+    },
+    {
+      "epoch": 0.8480376361410067,
+      "grad_norm": 0.0872153273947622,
+      "learning_rate": 1.6956677407805229e-06,
+      "loss": 0.0016,
+      "step": 13159
+    },
+    {
+      "epoch": 0.8481020815879358,
+      "grad_norm": 0.0006567281632697598,
+      "learning_rate": 1.6949516648764772e-06,
+      "loss": 0.0,
+      "step": 13160
+    },
+    {
+      "epoch": 0.848166527034865,
+      "grad_norm": 0.000274595862112701,
+      "learning_rate": 1.6942355889724312e-06,
+      "loss": 0.0,
+      "step": 13161
+    },
+    {
+      "epoch": 0.8482309724817941,
+      "grad_norm": 0.002213287427482091,
+      "learning_rate": 1.6935195130683853e-06,
+      "loss": 0.0,
+      "step": 13162
+    },
+    {
+      "epoch": 0.8482954179287233,
+      "grad_norm": 0.0006916792952072503,
+      "learning_rate": 1.6928034371643396e-06,
+      "loss": 0.0,
+      "step": 13163
+    },
+    {
+      "epoch": 0.8483598633756525,
+      "grad_norm": 0.014104434453009731,
+      "learning_rate": 1.6920873612602937e-06,
+      "loss": 0.0002,
+      "step": 13164
+    },
+    {
+      "epoch": 0.8484243088225817,
+      "grad_norm": 0.37413188666834707,
+      "learning_rate": 1.6913712853562478e-06,
+      "loss": 0.0051,
+      "step": 13165
+    },
+    {
+      "epoch": 0.8484887542695109,
+      "grad_norm": 0.22097166148012748,
+      "learning_rate": 1.690655209452202e-06,
+      "loss": 0.0032,
+      "step": 13166
+    },
+    {
+      "epoch": 0.84855319971644,
+      "grad_norm": 0.05122863864429417,
+      "learning_rate": 1.6899391335481564e-06,
+      "loss": 0.0001,
+      "step": 13167
+    },
+    {
+      "epoch": 0.8486176451633692,
+      "grad_norm": 0.0022830902384754453,
+      "learning_rate": 1.6892230576441105e-06,
+      "loss": 0.0,
+      "step": 13168
+    },
+    {
+      "epoch": 0.8486820906102984,
+      "grad_norm": 0.00552262182764459,
+      "learning_rate": 1.6885069817400646e-06,
+      "loss": 0.0,
+      "step": 13169
+    },
+    {
+      "epoch": 0.8487465360572276,
+      "grad_norm": 0.004210165256711199,
+      "learning_rate": 1.6877909058360187e-06,
+      "loss": 0.0,
+      "step": 13170
+    },
+    {
+      "epoch": 0.8488109815041567,
+      "grad_norm": 0.08326636885960183,
+      "learning_rate": 1.6870748299319727e-06,
+      "loss": 0.0001,
+      "step": 13171
+    },
+    {
+      "epoch": 0.8488754269510859,
+      "grad_norm": 0.002165883292149573,
+      "learning_rate": 1.6863587540279273e-06,
+      "loss": 0.0,
+      "step": 13172
+    },
+    {
+      "epoch": 0.848939872398015,
+      "grad_norm": 0.02688358309133841,
+      "learning_rate": 1.6856426781238813e-06,
+      "loss": 0.0,
+      "step": 13173
+    },
+    {
+      "epoch": 0.8490043178449442,
+      "grad_norm": 0.0035068966026358408,
+      "learning_rate": 1.6849266022198354e-06,
+      "loss": 0.0015,
+      "step": 13174
+    },
+    {
+      "epoch": 0.8490687632918734,
+      "grad_norm": 0.08898056050565528,
+      "learning_rate": 1.6842105263157895e-06,
+      "loss": 0.0001,
+      "step": 13175
+    },
+    {
+      "epoch": 0.8491332087388026,
+      "grad_norm": 0.01142288213511176,
+      "learning_rate": 1.6834944504117436e-06,
+      "loss": 0.0,
+      "step": 13176
+    },
+    {
+      "epoch": 0.8491976541857318,
+      "grad_norm": 0.12086776057633856,
+      "learning_rate": 1.682778374507698e-06,
+      "loss": 0.0002,
+      "step": 13177
+    },
+    {
+      "epoch": 0.849262099632661,
+      "grad_norm": 0.016550340064353578,
+      "learning_rate": 1.6820622986036522e-06,
+      "loss": 0.0002,
+      "step": 13178
+    },
+    {
+      "epoch": 0.8493265450795902,
+      "grad_norm": 0.00171017518821912,
+      "learning_rate": 1.6813462226996063e-06,
+      "loss": 0.0,
+      "step": 13179
+    },
+    {
+      "epoch": 0.8493909905265193,
+      "grad_norm": 0.07949205546168953,
+      "learning_rate": 1.6806301467955604e-06,
+      "loss": 0.0012,
+      "step": 13180
+    },
+    {
+      "epoch": 0.8494554359734485,
+      "grad_norm": 0.09155284118646113,
+      "learning_rate": 1.6799140708915147e-06,
+      "loss": 0.0008,
+      "step": 13181
+    },
+    {
+      "epoch": 0.8495198814203776,
+      "grad_norm": 0.0025166251856908886,
+      "learning_rate": 1.6791979949874688e-06,
+      "loss": 0.0,
+      "step": 13182
+    },
+    {
+      "epoch": 0.8495843268673068,
+      "grad_norm": 0.01984616554006101,
+      "learning_rate": 1.678481919083423e-06,
+      "loss": 0.0002,
+      "step": 13183
+    },
+    {
+      "epoch": 0.849648772314236,
+      "grad_norm": 0.0010042060360331753,
+      "learning_rate": 1.6777658431793771e-06,
+      "loss": 0.0,
+      "step": 13184
+    },
+    {
+      "epoch": 0.8497132177611652,
+      "grad_norm": 0.0009949796621702785,
+      "learning_rate": 1.6770497672753314e-06,
+      "loss": 0.0,
+      "step": 13185
+    },
+    {
+      "epoch": 0.8497776632080943,
+      "grad_norm": 0.053973604706768964,
+      "learning_rate": 1.6763336913712855e-06,
+      "loss": 0.0001,
+      "step": 13186
+    },
+    {
+      "epoch": 0.8498421086550235,
+      "grad_norm": 0.4081985619394894,
+      "learning_rate": 1.6756176154672396e-06,
+      "loss": 0.0036,
+      "step": 13187
+    },
+    {
+      "epoch": 0.8499065541019527,
+      "grad_norm": 0.02374758216156314,
+      "learning_rate": 1.6749015395631937e-06,
+      "loss": 0.0001,
+      "step": 13188
+    },
+    {
+      "epoch": 0.8499709995488819,
+      "grad_norm": 0.05115024449824533,
+      "learning_rate": 1.6741854636591482e-06,
+      "loss": 0.0002,
+      "step": 13189
+    },
+    {
+      "epoch": 0.8500354449958111,
+      "grad_norm": 1.4907648784656893,
+      "learning_rate": 1.6734693877551023e-06,
+      "loss": 0.006,
+      "step": 13190
+    },
+    {
+      "epoch": 0.8500998904427403,
+      "grad_norm": 0.000761882675939549,
+      "learning_rate": 1.6727533118510564e-06,
+      "loss": 0.0,
+      "step": 13191
+    },
+    {
+      "epoch": 0.8501643358896694,
+      "grad_norm": 0.0010225074365764744,
+      "learning_rate": 1.6720372359470105e-06,
+      "loss": 0.0,
+      "step": 13192
+    },
+    {
+      "epoch": 0.8502287813365985,
+      "grad_norm": 0.011092580154141158,
+      "learning_rate": 1.6713211600429646e-06,
+      "loss": 0.0,
+      "step": 13193
+    },
+    {
+      "epoch": 0.8502932267835277,
+      "grad_norm": 0.0006417750320247429,
+      "learning_rate": 1.6706050841389187e-06,
+      "loss": 0.0,
+      "step": 13194
+    },
+    {
+      "epoch": 0.8503576722304569,
+      "grad_norm": 0.0035144805856446135,
+      "learning_rate": 1.6698890082348732e-06,
+      "loss": 0.0,
+      "step": 13195
+    },
+    {
+      "epoch": 0.8504221176773861,
+      "grad_norm": 0.0010827521412211539,
+      "learning_rate": 1.6691729323308273e-06,
+      "loss": 0.0,
+      "step": 13196
+    },
+    {
+      "epoch": 0.8504865631243153,
+      "grad_norm": 0.03509750524593469,
+      "learning_rate": 1.6684568564267813e-06,
+      "loss": 0.0001,
+      "step": 13197
+    },
+    {
+      "epoch": 0.8505510085712444,
+      "grad_norm": 0.00036619498825359834,
+      "learning_rate": 1.6677407805227354e-06,
+      "loss": 0.0,
+      "step": 13198
+    },
+    {
+      "epoch": 0.8506154540181736,
+      "grad_norm": 0.009914766960140308,
+      "learning_rate": 1.6670247046186897e-06,
+      "loss": 0.0,
+      "step": 13199
+    },
+    {
+      "epoch": 0.8506798994651028,
+      "grad_norm": 0.0017422356060475125,
+      "learning_rate": 1.6663086287146438e-06,
+      "loss": 0.0,
+      "step": 13200
+    },
+    {
+      "epoch": 0.850744344912032,
+      "grad_norm": 0.5032569191045334,
+      "learning_rate": 1.6655925528105981e-06,
+      "loss": 0.0029,
+      "step": 13201
+    },
+    {
+      "epoch": 0.8508087903589612,
+      "grad_norm": 0.10452537366147462,
+      "learning_rate": 1.6648764769065522e-06,
+      "loss": 0.0004,
+      "step": 13202
+    },
+    {
+      "epoch": 0.8508732358058904,
+      "grad_norm": 0.5053188131076043,
+      "learning_rate": 1.6641604010025065e-06,
+      "loss": 0.0021,
+      "step": 13203
+    },
+    {
+      "epoch": 0.8509376812528194,
+      "grad_norm": 0.0012992472535777047,
+      "learning_rate": 1.6634443250984606e-06,
+      "loss": 0.0,
+      "step": 13204
+    },
+    {
+      "epoch": 0.8510021266997486,
+      "grad_norm": 0.00011059526996946594,
+      "learning_rate": 1.6627282491944147e-06,
+      "loss": 0.0,
+      "step": 13205
+    },
+    {
+      "epoch": 0.8510665721466778,
+      "grad_norm": 0.005765190230397704,
+      "learning_rate": 1.6620121732903688e-06,
+      "loss": 0.0,
+      "step": 13206
+    },
+    {
+      "epoch": 0.851131017593607,
+      "grad_norm": 0.003982361016511815,
+      "learning_rate": 1.6612960973863233e-06,
+      "loss": 0.0,
+      "step": 13207
+    },
+    {
+      "epoch": 0.8511954630405362,
+      "grad_norm": 0.1607033931876445,
+      "learning_rate": 1.6605800214822774e-06,
+      "loss": 0.0006,
+      "step": 13208
+    },
+    {
+      "epoch": 0.8512599084874654,
+      "grad_norm": 0.0023549343502827992,
+      "learning_rate": 1.6598639455782314e-06,
+      "loss": 0.0,
+      "step": 13209
+    },
+    {
+      "epoch": 0.8513243539343945,
+      "grad_norm": 0.004134231289848903,
+      "learning_rate": 1.6591478696741855e-06,
+      "loss": 0.0,
+      "step": 13210
+    },
+    {
+      "epoch": 0.8513887993813237,
+      "grad_norm": 0.0012640425582206376,
+      "learning_rate": 1.6584317937701396e-06,
+      "loss": 0.0,
+      "step": 13211
+    },
+    {
+      "epoch": 0.8514532448282529,
+      "grad_norm": 0.0021418223676945244,
+      "learning_rate": 1.657715717866094e-06,
+      "loss": 0.0,
+      "step": 13212
+    },
+    {
+      "epoch": 0.8515176902751821,
+      "grad_norm": 0.0007885922196547374,
+      "learning_rate": 1.6569996419620482e-06,
+      "loss": 0.0,
+      "step": 13213
+    },
+    {
+      "epoch": 0.8515821357221113,
+      "grad_norm": 0.0014711894637515442,
+      "learning_rate": 1.6562835660580023e-06,
+      "loss": 0.0,
+      "step": 13214
+    },
+    {
+      "epoch": 0.8516465811690405,
+      "grad_norm": 0.008335604533173686,
+      "learning_rate": 1.6555674901539564e-06,
+      "loss": 0.0,
+      "step": 13215
+    },
+    {
+      "epoch": 0.8517110266159695,
+      "grad_norm": 0.011965576789911826,
+      "learning_rate": 1.6548514142499107e-06,
+      "loss": 0.0,
+      "step": 13216
+    },
+    {
+      "epoch": 0.8517754720628987,
+      "grad_norm": 0.0012711725620480775,
+      "learning_rate": 1.6541353383458648e-06,
+      "loss": 0.0,
+      "step": 13217
+    },
+    {
+      "epoch": 0.8518399175098279,
+      "grad_norm": 0.03274330104237303,
+      "learning_rate": 1.653419262441819e-06,
+      "loss": 0.0,
+      "step": 13218
+    },
+    {
+      "epoch": 0.8519043629567571,
+      "grad_norm": 0.013053572989104686,
+      "learning_rate": 1.6527031865377732e-06,
+      "loss": 0.0,
+      "step": 13219
+    },
+    {
+      "epoch": 0.8519688084036863,
+      "grad_norm": 0.0005108394442840666,
+      "learning_rate": 1.6519871106337275e-06,
+      "loss": 0.0,
+      "step": 13220
+    },
+    {
+      "epoch": 0.8520332538506155,
+      "grad_norm": 0.008659800547823577,
+      "learning_rate": 1.6512710347296815e-06,
+      "loss": 0.0,
+      "step": 13221
+    },
+    {
+      "epoch": 0.8520976992975446,
+      "grad_norm": 0.013328921264361386,
+      "learning_rate": 1.6505549588256356e-06,
+      "loss": 0.0,
+      "step": 13222
+    },
+    {
+      "epoch": 0.8521621447444738,
+      "grad_norm": 0.00534133565134591,
+      "learning_rate": 1.6498388829215897e-06,
+      "loss": 0.0,
+      "step": 13223
+    },
+    {
+      "epoch": 0.852226590191403,
+      "grad_norm": 0.004128411945539397,
+      "learning_rate": 1.6491228070175442e-06,
+      "loss": 0.0,
+      "step": 13224
+    },
+    {
+      "epoch": 0.8522910356383322,
+      "grad_norm": 0.0011464066259772201,
+      "learning_rate": 1.6484067311134983e-06,
+      "loss": 0.0,
+      "step": 13225
+    },
+    {
+      "epoch": 0.8523554810852614,
+      "grad_norm": 0.0012449689486442015,
+      "learning_rate": 1.6476906552094524e-06,
+      "loss": 0.0,
+      "step": 13226
+    },
+    {
+      "epoch": 0.8524199265321905,
+      "grad_norm": 0.01839498996198106,
+      "learning_rate": 1.6469745793054065e-06,
+      "loss": 0.0,
+      "step": 13227
+    },
+    {
+      "epoch": 0.8524843719791196,
+      "grad_norm": 0.02975664278763737,
+      "learning_rate": 1.6462585034013606e-06,
+      "loss": 0.0,
+      "step": 13228
+    },
+    {
+      "epoch": 0.8525488174260488,
+      "grad_norm": 0.004641548279767545,
+      "learning_rate": 1.6455424274973147e-06,
+      "loss": 0.0,
+      "step": 13229
+    },
+    {
+      "epoch": 0.852613262872978,
+      "grad_norm": 0.06830025568807072,
+      "learning_rate": 1.6448263515932692e-06,
+      "loss": 0.0005,
+      "step": 13230
+    },
+    {
+      "epoch": 0.8526777083199072,
+      "grad_norm": 1.172192695787183,
+      "learning_rate": 1.6441102756892233e-06,
+      "loss": 0.0056,
+      "step": 13231
+    },
+    {
+      "epoch": 0.8527421537668364,
+      "grad_norm": 0.011497391242717231,
+      "learning_rate": 1.6433941997851773e-06,
+      "loss": 0.0,
+      "step": 13232
+    },
+    {
+      "epoch": 0.8528065992137656,
+      "grad_norm": 0.05892172851319213,
+      "learning_rate": 1.6426781238811314e-06,
+      "loss": 0.0,
+      "step": 13233
+    },
+    {
+      "epoch": 0.8528710446606947,
+      "grad_norm": 0.004463375528091414,
+      "learning_rate": 1.6419620479770857e-06,
+      "loss": 0.0,
+      "step": 13234
+    },
+    {
+      "epoch": 0.8529354901076239,
+      "grad_norm": 0.0018064123459318863,
+      "learning_rate": 1.6412459720730398e-06,
+      "loss": 0.0,
+      "step": 13235
+    },
+    {
+      "epoch": 0.8529999355545531,
+      "grad_norm": 0.06578452059686321,
+      "learning_rate": 1.6405298961689941e-06,
+      "loss": 0.0001,
+      "step": 13236
+    },
+    {
+      "epoch": 0.8530643810014823,
+      "grad_norm": 0.016703103161657937,
+      "learning_rate": 1.6398138202649482e-06,
+      "loss": 0.0,
+      "step": 13237
+    },
+    {
+      "epoch": 0.8531288264484114,
+      "grad_norm": 0.014220012566819788,
+      "learning_rate": 1.6390977443609025e-06,
+      "loss": 0.0002,
+      "step": 13238
+    },
+    {
+      "epoch": 0.8531932718953406,
+      "grad_norm": 0.09662233925181075,
+      "learning_rate": 1.6383816684568566e-06,
+      "loss": 0.0001,
+      "step": 13239
+    },
+    {
+      "epoch": 0.8532577173422697,
+      "grad_norm": 0.19108335351306138,
+      "learning_rate": 1.6376655925528107e-06,
+      "loss": 0.0005,
+      "step": 13240
+    },
+    {
+      "epoch": 0.8533221627891989,
+      "grad_norm": 0.021867105829644136,
+      "learning_rate": 1.6369495166487648e-06,
+      "loss": 0.0,
+      "step": 13241
+    },
+    {
+      "epoch": 0.8533866082361281,
+      "grad_norm": 0.005161472184663322,
+      "learning_rate": 1.6362334407447193e-06,
+      "loss": 0.0,
+      "step": 13242
+    },
+    {
+      "epoch": 0.8534510536830573,
+      "grad_norm": 0.005717583823887771,
+      "learning_rate": 1.6355173648406734e-06,
+      "loss": 0.0,
+      "step": 13243
+    },
+    {
+      "epoch": 0.8535154991299865,
+      "grad_norm": 0.008121971108982103,
+      "learning_rate": 1.6348012889366274e-06,
+      "loss": 0.0,
+      "step": 13244
+    },
+    {
+      "epoch": 0.8535799445769157,
+      "grad_norm": 0.00037337266763354336,
+      "learning_rate": 1.6340852130325815e-06,
+      "loss": 0.0,
+      "step": 13245
+    },
+    {
+      "epoch": 0.8536443900238448,
+      "grad_norm": 0.002726389582061402,
+      "learning_rate": 1.6333691371285356e-06,
+      "loss": 0.0,
+      "step": 13246
+    },
+    {
+      "epoch": 0.853708835470774,
+      "grad_norm": 0.12864904319243925,
+      "learning_rate": 1.6326530612244897e-06,
+      "loss": 0.0001,
+      "step": 13247
+    },
+    {
+      "epoch": 0.8537732809177032,
+      "grad_norm": 0.01956431971786598,
+      "learning_rate": 1.6319369853204442e-06,
+      "loss": 0.0,
+      "step": 13248
+    },
+    {
+      "epoch": 0.8538377263646323,
+      "grad_norm": 0.0007446259199047396,
+      "learning_rate": 1.6312209094163983e-06,
+      "loss": 0.0,
+      "step": 13249
+    },
+    {
+      "epoch": 0.8539021718115615,
+      "grad_norm": 8.923878846050036e-05,
+      "learning_rate": 1.6305048335123524e-06,
+      "loss": 0.0,
+      "step": 13250
+    },
+    {
+      "epoch": 0.8539666172584907,
+      "grad_norm": 0.00614986020674208,
+      "learning_rate": 1.6297887576083065e-06,
+      "loss": 0.0,
+      "step": 13251
+    },
+    {
+      "epoch": 0.8540310627054198,
+      "grad_norm": 0.0002994068452241638,
+      "learning_rate": 1.6290726817042608e-06,
+      "loss": 0.0,
+      "step": 13252
+    },
+    {
+      "epoch": 0.854095508152349,
+      "grad_norm": 0.019766686431275136,
+      "learning_rate": 1.6283566058002149e-06,
+      "loss": 0.0001,
+      "step": 13253
+    },
+    {
+      "epoch": 0.8541599535992782,
+      "grad_norm": 0.08729203685103124,
+      "learning_rate": 1.6276405298961692e-06,
+      "loss": 0.0016,
+      "step": 13254
+    },
+    {
+      "epoch": 0.8542243990462074,
+      "grad_norm": 0.0016916745618347053,
+      "learning_rate": 1.6269244539921233e-06,
+      "loss": 0.0,
+      "step": 13255
+    },
+    {
+      "epoch": 0.8542888444931366,
+      "grad_norm": 0.0019892775296158497,
+      "learning_rate": 1.6262083780880776e-06,
+      "loss": 0.0,
+      "step": 13256
+    },
+    {
+      "epoch": 0.8543532899400658,
+      "grad_norm": 0.0023993870594850473,
+      "learning_rate": 1.6254923021840316e-06,
+      "loss": 0.0,
+      "step": 13257
+    },
+    {
+      "epoch": 0.854417735386995,
+      "grad_norm": 0.00016196712541359675,
+      "learning_rate": 1.6247762262799857e-06,
+      "loss": 0.0,
+      "step": 13258
+    },
+    {
+      "epoch": 0.8544821808339241,
+      "grad_norm": 6.87116925052651e-05,
+      "learning_rate": 1.62406015037594e-06,
+      "loss": 0.0,
+      "step": 13259
+    },
+    {
+      "epoch": 0.8545466262808532,
+      "grad_norm": 0.004721255990723467,
+      "learning_rate": 1.6233440744718943e-06,
+      "loss": 0.0,
+      "step": 13260
+    },
+    {
+      "epoch": 0.8546110717277824,
+      "grad_norm": 0.0018702730203080486,
+      "learning_rate": 1.6226279985678484e-06,
+      "loss": 0.0,
+      "step": 13261
+    },
+    {
+      "epoch": 0.8546755171747116,
+      "grad_norm": 0.04213422495428291,
+      "learning_rate": 1.6219119226638025e-06,
+      "loss": 0.0004,
+      "step": 13262
+    },
+    {
+      "epoch": 0.8547399626216408,
+      "grad_norm": 0.09176947306452253,
+      "learning_rate": 1.6211958467597566e-06,
+      "loss": 0.0001,
+      "step": 13263
+    },
+    {
+      "epoch": 0.8548044080685699,
+      "grad_norm": 0.0002837684054832139,
+      "learning_rate": 1.6204797708557107e-06,
+      "loss": 0.0,
+      "step": 13264
+    },
+    {
+      "epoch": 0.8548688535154991,
+      "grad_norm": 0.00034699342449560114,
+      "learning_rate": 1.6197636949516652e-06,
+      "loss": 0.0,
+      "step": 13265
+    },
+    {
+      "epoch": 0.8549332989624283,
+      "grad_norm": 0.0001257464769171491,
+      "learning_rate": 1.6190476190476193e-06,
+      "loss": 0.0,
+      "step": 13266
+    },
+    {
+      "epoch": 0.8549977444093575,
+      "grad_norm": 0.015559039084421477,
+      "learning_rate": 1.6183315431435734e-06,
+      "loss": 0.0,
+      "step": 13267
+    },
+    {
+      "epoch": 0.8550621898562867,
+      "grad_norm": 0.0008531191131122104,
+      "learning_rate": 1.6176154672395274e-06,
+      "loss": 0.0,
+      "step": 13268
+    },
+    {
+      "epoch": 0.8551266353032159,
+      "grad_norm": 0.003998335045763489,
+      "learning_rate": 1.6168993913354815e-06,
+      "loss": 0.0,
+      "step": 13269
+    },
+    {
+      "epoch": 0.855191080750145,
+      "grad_norm": 0.163619043841412,
+      "learning_rate": 1.6161833154314358e-06,
+      "loss": 0.0004,
+      "step": 13270
+    },
+    {
+      "epoch": 0.8552555261970741,
+      "grad_norm": 0.03945032064933214,
+      "learning_rate": 1.6154672395273901e-06,
+      "loss": 0.0002,
+      "step": 13271
+    },
+    {
+      "epoch": 0.8553199716440033,
+      "grad_norm": 3.905622064627832e-05,
+      "learning_rate": 1.6147511636233442e-06,
+      "loss": 0.0,
+      "step": 13272
+    },
+    {
+      "epoch": 0.8553844170909325,
+      "grad_norm": 0.08700623339778292,
+      "learning_rate": 1.6140350877192983e-06,
+      "loss": 0.0001,
+      "step": 13273
+    },
+    {
+      "epoch": 0.8554488625378617,
+      "grad_norm": 0.0002938337301690031,
+      "learning_rate": 1.6133190118152526e-06,
+      "loss": 0.0,
+      "step": 13274
+    },
+    {
+      "epoch": 0.8555133079847909,
+      "grad_norm": 0.0016490099642151128,
+      "learning_rate": 1.6126029359112067e-06,
+      "loss": 0.0,
+      "step": 13275
+    },
+    {
+      "epoch": 0.85557775343172,
+      "grad_norm": 0.21113241867889743,
+      "learning_rate": 1.6118868600071608e-06,
+      "loss": 0.0012,
+      "step": 13276
+    },
+    {
+      "epoch": 0.8556421988786492,
+      "grad_norm": 0.00034076801657522657,
+      "learning_rate": 1.611170784103115e-06,
+      "loss": 0.0,
+      "step": 13277
+    },
+    {
+      "epoch": 0.8557066443255784,
+      "grad_norm": 0.0007656847467303316,
+      "learning_rate": 1.6104547081990694e-06,
+      "loss": 0.0,
+      "step": 13278
+    },
+    {
+      "epoch": 0.8557710897725076,
+      "grad_norm": 0.004611446967922683,
+      "learning_rate": 1.6097386322950235e-06,
+      "loss": 0.0,
+      "step": 13279
+    },
+    {
+      "epoch": 0.8558355352194368,
+      "grad_norm": 4.522851739194399e-05,
+      "learning_rate": 1.6090225563909775e-06,
+      "loss": 0.0,
+      "step": 13280
+    },
+    {
+      "epoch": 0.855899980666366,
+      "grad_norm": 0.09413581178952365,
+      "learning_rate": 1.6083064804869316e-06,
+      "loss": 0.0017,
+      "step": 13281
+    },
+    {
+      "epoch": 0.855964426113295,
+      "grad_norm": 0.004960847512303381,
+      "learning_rate": 1.6075904045828857e-06,
+      "loss": 0.0,
+      "step": 13282
+    },
+    {
+      "epoch": 0.8560288715602242,
+      "grad_norm": 0.035130821078171386,
+      "learning_rate": 1.6068743286788402e-06,
+      "loss": 0.0,
+      "step": 13283
+    },
+    {
+      "epoch": 0.8560933170071534,
+      "grad_norm": 0.6814632458239104,
+      "learning_rate": 1.6061582527747943e-06,
+      "loss": 0.0024,
+      "step": 13284
+    },
+    {
+      "epoch": 0.8561577624540826,
+      "grad_norm": 0.0006295020560277422,
+      "learning_rate": 1.6054421768707484e-06,
+      "loss": 0.0,
+      "step": 13285
+    },
+    {
+      "epoch": 0.8562222079010118,
+      "grad_norm": 0.000463167640538834,
+      "learning_rate": 1.6047261009667025e-06,
+      "loss": 0.0,
+      "step": 13286
+    },
+    {
+      "epoch": 0.856286653347941,
+      "grad_norm": 0.003007813351736707,
+      "learning_rate": 1.6040100250626568e-06,
+      "loss": 0.0,
+      "step": 13287
+    },
+    {
+      "epoch": 0.8563510987948701,
+      "grad_norm": 0.008826912320212115,
+      "learning_rate": 1.6032939491586109e-06,
+      "loss": 0.0,
+      "step": 13288
+    },
+    {
+      "epoch": 0.8564155442417993,
+      "grad_norm": 0.004008362923994413,
+      "learning_rate": 1.6025778732545652e-06,
+      "loss": 0.0,
+      "step": 13289
+    },
+    {
+      "epoch": 0.8564799896887285,
+      "grad_norm": 0.00025324509025633033,
+      "learning_rate": 1.6018617973505193e-06,
+      "loss": 0.0,
+      "step": 13290
+    },
+    {
+      "epoch": 0.8565444351356577,
+      "grad_norm": 0.009025551561342875,
+      "learning_rate": 1.6011457214464736e-06,
+      "loss": 0.0,
+      "step": 13291
+    },
+    {
+      "epoch": 0.8566088805825869,
+      "grad_norm": 0.15220108443539926,
+      "learning_rate": 1.6004296455424276e-06,
+      "loss": 0.0002,
+      "step": 13292
+    },
+    {
+      "epoch": 0.8566733260295161,
+      "grad_norm": 0.0005597975580254473,
+      "learning_rate": 1.5997135696383817e-06,
+      "loss": 0.0,
+      "step": 13293
+    },
+    {
+      "epoch": 0.8567377714764451,
+      "grad_norm": 0.2044531426959965,
+      "learning_rate": 1.598997493734336e-06,
+      "loss": 0.0004,
+      "step": 13294
+    },
+    {
+      "epoch": 0.8568022169233743,
+      "grad_norm": 0.0065360515775523985,
+      "learning_rate": 1.5982814178302903e-06,
+      "loss": 0.0,
+      "step": 13295
+    },
+    {
+      "epoch": 0.8568666623703035,
+      "grad_norm": 0.00010121851515030464,
+      "learning_rate": 1.5975653419262444e-06,
+      "loss": 0.0,
+      "step": 13296
+    },
+    {
+      "epoch": 0.8569311078172327,
+      "grad_norm": 0.1027933826978185,
+      "learning_rate": 1.5968492660221985e-06,
+      "loss": 0.0018,
+      "step": 13297
+    },
+    {
+      "epoch": 0.8569955532641619,
+      "grad_norm": 0.31212899118972154,
+      "learning_rate": 1.5961331901181526e-06,
+      "loss": 0.0011,
+      "step": 13298
+    },
+    {
+      "epoch": 0.8570599987110911,
+      "grad_norm": 0.011037713981860054,
+      "learning_rate": 1.5954171142141067e-06,
+      "loss": 0.0,
+      "step": 13299
+    },
+    {
+      "epoch": 0.8571244441580202,
+      "grad_norm": 0.17396643847202156,
+      "learning_rate": 1.5947010383100612e-06,
+      "loss": 0.0002,
+      "step": 13300
+    },
+    {
+      "epoch": 0.8571888896049494,
+      "grad_norm": 0.27335187388744736,
+      "learning_rate": 1.5939849624060153e-06,
+      "loss": 0.0031,
+      "step": 13301
+    },
+    {
+      "epoch": 0.8572533350518786,
+      "grad_norm": 0.5506759164068347,
+      "learning_rate": 1.5932688865019694e-06,
+      "loss": 0.0027,
+      "step": 13302
+    },
+    {
+      "epoch": 0.8573177804988078,
+      "grad_norm": 0.038104221039077645,
+      "learning_rate": 1.5925528105979234e-06,
+      "loss": 0.0002,
+      "step": 13303
+    },
+    {
+      "epoch": 0.857382225945737,
+      "grad_norm": 0.03073348441255423,
+      "learning_rate": 1.5918367346938775e-06,
+      "loss": 0.0,
+      "step": 13304
+    },
+    {
+      "epoch": 0.8574466713926661,
+      "grad_norm": 0.003336774922123674,
+      "learning_rate": 1.5911206587898318e-06,
+      "loss": 0.0,
+      "step": 13305
+    },
+    {
+      "epoch": 0.8575111168395952,
+      "grad_norm": 0.019049603577953083,
+      "learning_rate": 1.5904045828857861e-06,
+      "loss": 0.0,
+      "step": 13306
+    },
+    {
+      "epoch": 0.8575755622865244,
+      "grad_norm": 0.006879444816098133,
+      "learning_rate": 1.5896885069817402e-06,
+      "loss": 0.0,
+      "step": 13307
+    },
+    {
+      "epoch": 0.8576400077334536,
+      "grad_norm": 0.0035166293965754702,
+      "learning_rate": 1.5889724310776943e-06,
+      "loss": 0.0,
+      "step": 13308
+    },
+    {
+      "epoch": 0.8577044531803828,
+      "grad_norm": 0.10649776478809318,
+      "learning_rate": 1.5882563551736486e-06,
+      "loss": 0.0012,
+      "step": 13309
+    },
+    {
+      "epoch": 0.857768898627312,
+      "grad_norm": 0.03601981388202897,
+      "learning_rate": 1.5875402792696027e-06,
+      "loss": 0.0,
+      "step": 13310
+    },
+    {
+      "epoch": 0.8578333440742412,
+      "grad_norm": 0.00010030491278225593,
+      "learning_rate": 1.5868242033655568e-06,
+      "loss": 0.0,
+      "step": 13311
+    },
+    {
+      "epoch": 0.8578977895211704,
+      "grad_norm": 0.004006094877726113,
+      "learning_rate": 1.586108127461511e-06,
+      "loss": 0.0,
+      "step": 13312
+    },
+    {
+      "epoch": 0.8579622349680995,
+      "grad_norm": 0.01206481644215532,
+      "learning_rate": 1.5853920515574654e-06,
+      "loss": 0.0,
+      "step": 13313
+    },
+    {
+      "epoch": 0.8580266804150287,
+      "grad_norm": 0.0012709160823947188,
+      "learning_rate": 1.5846759756534195e-06,
+      "loss": 0.0,
+      "step": 13314
+    },
+    {
+      "epoch": 0.8580911258619579,
+      "grad_norm": 0.0005756146179106836,
+      "learning_rate": 1.5839598997493736e-06,
+      "loss": 0.0,
+      "step": 13315
+    },
+    {
+      "epoch": 0.858155571308887,
+      "grad_norm": 0.005792067297186545,
+      "learning_rate": 1.5832438238453276e-06,
+      "loss": 0.0,
+      "step": 13316
+    },
+    {
+      "epoch": 0.8582200167558162,
+      "grad_norm": 0.01710870502387094,
+      "learning_rate": 1.5825277479412817e-06,
+      "loss": 0.0001,
+      "step": 13317
+    },
+    {
+      "epoch": 0.8582844622027453,
+      "grad_norm": 0.002173941342640871,
+      "learning_rate": 1.5818116720372362e-06,
+      "loss": 0.0,
+      "step": 13318
+    },
+    {
+      "epoch": 0.8583489076496745,
+      "grad_norm": 0.0006425608284127846,
+      "learning_rate": 1.5810955961331903e-06,
+      "loss": 0.0,
+      "step": 13319
+    },
+    {
+      "epoch": 0.8584133530966037,
+      "grad_norm": 0.30172473763244206,
+      "learning_rate": 1.5803795202291444e-06,
+      "loss": 0.0016,
+      "step": 13320
+    },
+    {
+      "epoch": 0.8584777985435329,
+      "grad_norm": 0.01861099902398579,
+      "learning_rate": 1.5796634443250985e-06,
+      "loss": 0.0,
+      "step": 13321
+    },
+    {
+      "epoch": 0.8585422439904621,
+      "grad_norm": 0.0012165400924309016,
+      "learning_rate": 1.5789473684210526e-06,
+      "loss": 0.0,
+      "step": 13322
+    },
+    {
+      "epoch": 0.8586066894373913,
+      "grad_norm": 0.0016069345021319553,
+      "learning_rate": 1.5782312925170069e-06,
+      "loss": 0.0,
+      "step": 13323
+    },
+    {
+      "epoch": 0.8586711348843205,
+      "grad_norm": 0.011594244425208303,
+      "learning_rate": 1.5775152166129612e-06,
+      "loss": 0.0001,
+      "step": 13324
+    },
+    {
+      "epoch": 0.8587355803312496,
+      "grad_norm": 0.0014469452087132976,
+      "learning_rate": 1.5767991407089153e-06,
+      "loss": 0.0,
+      "step": 13325
+    },
+    {
+      "epoch": 0.8588000257781788,
+      "grad_norm": 0.4306278862323567,
+      "learning_rate": 1.5760830648048694e-06,
+      "loss": 0.0004,
+      "step": 13326
+    },
+    {
+      "epoch": 0.8588644712251079,
+      "grad_norm": 0.003471658113443766,
+      "learning_rate": 1.5753669889008237e-06,
+      "loss": 0.0,
+      "step": 13327
+    },
+    {
+      "epoch": 0.8589289166720371,
+      "grad_norm": 0.0004689620724818602,
+      "learning_rate": 1.5746509129967777e-06,
+      "loss": 0.0,
+      "step": 13328
+    },
+    {
+      "epoch": 0.8589933621189663,
+      "grad_norm": 0.02153173085345865,
+      "learning_rate": 1.573934837092732e-06,
+      "loss": 0.0001,
+      "step": 13329
+    },
+    {
+      "epoch": 0.8590578075658954,
+      "grad_norm": 0.2610378517842289,
+      "learning_rate": 1.5732187611886861e-06,
+      "loss": 0.0006,
+      "step": 13330
+    },
+    {
+      "epoch": 0.8591222530128246,
+      "grad_norm": 0.02625268560888587,
+      "learning_rate": 1.5725026852846404e-06,
+      "loss": 0.0001,
+      "step": 13331
+    },
+    {
+      "epoch": 0.8591866984597538,
+      "grad_norm": 0.050066121532948815,
+      "learning_rate": 1.5717866093805945e-06,
+      "loss": 0.0022,
+      "step": 13332
+    },
+    {
+      "epoch": 0.859251143906683,
+      "grad_norm": 0.0002840123484040217,
+      "learning_rate": 1.5710705334765486e-06,
+      "loss": 0.0,
+      "step": 13333
+    },
+    {
+      "epoch": 0.8593155893536122,
+      "grad_norm": 0.016482623120749718,
+      "learning_rate": 1.5703544575725027e-06,
+      "loss": 0.0,
+      "step": 13334
+    },
+    {
+      "epoch": 0.8593800348005414,
+      "grad_norm": 0.0861558848682203,
+      "learning_rate": 1.5696383816684572e-06,
+      "loss": 0.0016,
+      "step": 13335
+    },
+    {
+      "epoch": 0.8594444802474706,
+      "grad_norm": 0.00206119523180686,
+      "learning_rate": 1.5689223057644113e-06,
+      "loss": 0.0,
+      "step": 13336
+    },
+    {
+      "epoch": 0.8595089256943997,
+      "grad_norm": 0.0010139368389927733,
+      "learning_rate": 1.5682062298603654e-06,
+      "loss": 0.0,
+      "step": 13337
+    },
+    {
+      "epoch": 0.8595733711413288,
+      "grad_norm": 0.0006872459690220183,
+      "learning_rate": 1.5674901539563195e-06,
+      "loss": 0.0,
+      "step": 13338
+    },
+    {
+      "epoch": 0.859637816588258,
+      "grad_norm": 0.0007045706704706853,
+      "learning_rate": 1.5667740780522735e-06,
+      "loss": 0.0,
+      "step": 13339
+    },
+    {
+      "epoch": 0.8597022620351872,
+      "grad_norm": 0.05906231015101705,
+      "learning_rate": 1.5660580021482276e-06,
+      "loss": 0.0,
+      "step": 13340
+    },
+    {
+      "epoch": 0.8597667074821164,
+      "grad_norm": 0.0011324638062900777,
+      "learning_rate": 1.5653419262441821e-06,
+      "loss": 0.0,
+      "step": 13341
+    },
+    {
+      "epoch": 0.8598311529290456,
+      "grad_norm": 0.010297346909910793,
+      "learning_rate": 1.5646258503401362e-06,
+      "loss": 0.0,
+      "step": 13342
+    },
+    {
+      "epoch": 0.8598955983759747,
+      "grad_norm": 0.29292520835291497,
+      "learning_rate": 1.5639097744360903e-06,
+      "loss": 0.0003,
+      "step": 13343
+    },
+    {
+      "epoch": 0.8599600438229039,
+      "grad_norm": 0.031157416981430704,
+      "learning_rate": 1.5631936985320444e-06,
+      "loss": 0.0016,
+      "step": 13344
+    },
+    {
+      "epoch": 0.8600244892698331,
+      "grad_norm": 0.005336673025525772,
+      "learning_rate": 1.5624776226279987e-06,
+      "loss": 0.0,
+      "step": 13345
+    },
+    {
+      "epoch": 0.8600889347167623,
+      "grad_norm": 0.0007176590913766336,
+      "learning_rate": 1.5617615467239528e-06,
+      "loss": 0.0,
+      "step": 13346
+    },
+    {
+      "epoch": 0.8601533801636915,
+      "grad_norm": 6.578943500745828e-05,
+      "learning_rate": 1.561045470819907e-06,
+      "loss": 0.0,
+      "step": 13347
+    },
+    {
+      "epoch": 0.8602178256106207,
+      "grad_norm": 0.0002624024066516301,
+      "learning_rate": 1.5603293949158612e-06,
+      "loss": 0.0,
+      "step": 13348
+    },
+    {
+      "epoch": 0.8602822710575497,
+      "grad_norm": 0.09216368719245878,
+      "learning_rate": 1.5596133190118155e-06,
+      "loss": 0.0006,
+      "step": 13349
+    },
+    {
+      "epoch": 0.8603467165044789,
+      "grad_norm": 0.003946083988254043,
+      "learning_rate": 1.5588972431077696e-06,
+      "loss": 0.0,
+      "step": 13350
+    },
+    {
+      "epoch": 0.8604111619514081,
+      "grad_norm": 0.000322491336789557,
+      "learning_rate": 1.5581811672037236e-06,
+      "loss": 0.0,
+      "step": 13351
+    },
+    {
+      "epoch": 0.8604756073983373,
+      "grad_norm": 0.00325187811401634,
+      "learning_rate": 1.5574650912996777e-06,
+      "loss": 0.0,
+      "step": 13352
+    },
+    {
+      "epoch": 0.8605400528452665,
+      "grad_norm": 0.02672152168310301,
+      "learning_rate": 1.5567490153956322e-06,
+      "loss": 0.0001,
+      "step": 13353
+    },
+    {
+      "epoch": 0.8606044982921957,
+      "grad_norm": 0.0010806659731611257,
+      "learning_rate": 1.5560329394915863e-06,
+      "loss": 0.0,
+      "step": 13354
+    },
+    {
+      "epoch": 0.8606689437391248,
+      "grad_norm": 0.0012196018515187096,
+      "learning_rate": 1.5553168635875404e-06,
+      "loss": 0.0,
+      "step": 13355
+    },
+    {
+      "epoch": 0.860733389186054,
+      "grad_norm": 0.03056133185486748,
+      "learning_rate": 1.5546007876834945e-06,
+      "loss": 0.0001,
+      "step": 13356
+    },
+    {
+      "epoch": 0.8607978346329832,
+      "grad_norm": 0.003190366588636452,
+      "learning_rate": 1.5538847117794486e-06,
+      "loss": 0.0,
+      "step": 13357
+    },
+    {
+      "epoch": 0.8608622800799124,
+      "grad_norm": 0.0010321709471120919,
+      "learning_rate": 1.5531686358754029e-06,
+      "loss": 0.0,
+      "step": 13358
+    },
+    {
+      "epoch": 0.8609267255268416,
+      "grad_norm": 0.005754727492837118,
+      "learning_rate": 1.5524525599713572e-06,
+      "loss": 0.0,
+      "step": 13359
+    },
+    {
+      "epoch": 0.8609911709737708,
+      "grad_norm": 0.02314667772246618,
+      "learning_rate": 1.5517364840673113e-06,
+      "loss": 0.0003,
+      "step": 13360
+    },
+    {
+      "epoch": 0.8610556164206998,
+      "grad_norm": 0.006225745757706881,
+      "learning_rate": 1.5510204081632654e-06,
+      "loss": 0.0,
+      "step": 13361
+    },
+    {
+      "epoch": 0.861120061867629,
+      "grad_norm": 0.012019083819958221,
+      "learning_rate": 1.5503043322592197e-06,
+      "loss": 0.0,
+      "step": 13362
+    },
+    {
+      "epoch": 0.8611845073145582,
+      "grad_norm": 0.000692814239205552,
+      "learning_rate": 1.5495882563551737e-06,
+      "loss": 0.0,
+      "step": 13363
+    },
+    {
+      "epoch": 0.8612489527614874,
+      "grad_norm": 0.004901376096575237,
+      "learning_rate": 1.548872180451128e-06,
+      "loss": 0.0,
+      "step": 13364
+    },
+    {
+      "epoch": 0.8613133982084166,
+      "grad_norm": 0.02895489313893144,
+      "learning_rate": 1.5481561045470821e-06,
+      "loss": 0.0001,
+      "step": 13365
+    },
+    {
+      "epoch": 0.8613778436553458,
+      "grad_norm": 0.05815220203678352,
+      "learning_rate": 1.5474400286430364e-06,
+      "loss": 0.0002,
+      "step": 13366
+    },
+    {
+      "epoch": 0.8614422891022749,
+      "grad_norm": 0.0003883691743082568,
+      "learning_rate": 1.5467239527389905e-06,
+      "loss": 0.0,
+      "step": 13367
+    },
+    {
+      "epoch": 0.8615067345492041,
+      "grad_norm": 0.0003275088709743321,
+      "learning_rate": 1.5460078768349446e-06,
+      "loss": 0.0,
+      "step": 13368
+    },
+    {
+      "epoch": 0.8615711799961333,
+      "grad_norm": 0.0009924672732805403,
+      "learning_rate": 1.5452918009308987e-06,
+      "loss": 0.0,
+      "step": 13369
+    },
+    {
+      "epoch": 0.8616356254430625,
+      "grad_norm": 0.0034706501414582956,
+      "learning_rate": 1.5445757250268532e-06,
+      "loss": 0.0,
+      "step": 13370
+    },
+    {
+      "epoch": 0.8617000708899917,
+      "grad_norm": 0.08772039352150697,
+      "learning_rate": 1.5438596491228073e-06,
+      "loss": 0.0017,
+      "step": 13371
+    },
+    {
+      "epoch": 0.8617645163369207,
+      "grad_norm": 0.0026765761249290475,
+      "learning_rate": 1.5431435732187614e-06,
+      "loss": 0.0,
+      "step": 13372
+    },
+    {
+      "epoch": 0.8618289617838499,
+      "grad_norm": 0.0022527401314781322,
+      "learning_rate": 1.5424274973147155e-06,
+      "loss": 0.0,
+      "step": 13373
+    },
+    {
+      "epoch": 0.8618934072307791,
+      "grad_norm": 0.012568387233077847,
+      "learning_rate": 1.5417114214106696e-06,
+      "loss": 0.0,
+      "step": 13374
+    },
+    {
+      "epoch": 0.8619578526777083,
+      "grad_norm": 0.030271575278257603,
+      "learning_rate": 1.5409953455066236e-06,
+      "loss": 0.0004,
+      "step": 13375
+    },
+    {
+      "epoch": 0.8620222981246375,
+      "grad_norm": 4.230431264947589e-05,
+      "learning_rate": 1.5402792696025781e-06,
+      "loss": 0.0,
+      "step": 13376
+    },
+    {
+      "epoch": 0.8620867435715667,
+      "grad_norm": 0.0004940029506976715,
+      "learning_rate": 1.5395631936985322e-06,
+      "loss": 0.0,
+      "step": 13377
+    },
+    {
+      "epoch": 0.8621511890184959,
+      "grad_norm": 0.0011218007381096433,
+      "learning_rate": 1.5388471177944863e-06,
+      "loss": 0.0,
+      "step": 13378
+    },
+    {
+      "epoch": 0.862215634465425,
+      "grad_norm": 0.07873246830828655,
+      "learning_rate": 1.5381310418904404e-06,
+      "loss": 0.0001,
+      "step": 13379
+    },
+    {
+      "epoch": 0.8622800799123542,
+      "grad_norm": 0.025534215775952065,
+      "learning_rate": 1.5374149659863947e-06,
+      "loss": 0.0,
+      "step": 13380
+    },
+    {
+      "epoch": 0.8623445253592834,
+      "grad_norm": 0.01992692039099067,
+      "learning_rate": 1.5366988900823488e-06,
+      "loss": 0.0002,
+      "step": 13381
+    },
+    {
+      "epoch": 0.8624089708062126,
+      "grad_norm": 0.30028876324941983,
+      "learning_rate": 1.535982814178303e-06,
+      "loss": 0.0015,
+      "step": 13382
+    },
+    {
+      "epoch": 0.8624734162531417,
+      "grad_norm": 0.0885966816085772,
+      "learning_rate": 1.5352667382742572e-06,
+      "loss": 0.0001,
+      "step": 13383
+    },
+    {
+      "epoch": 0.8625378617000709,
+      "grad_norm": 7.695888165750718e-05,
+      "learning_rate": 1.5345506623702115e-06,
+      "loss": 0.0,
+      "step": 13384
+    },
+    {
+      "epoch": 0.862602307147,
+      "grad_norm": 0.0010478483803299753,
+      "learning_rate": 1.5338345864661656e-06,
+      "loss": 0.0,
+      "step": 13385
+    },
+    {
+      "epoch": 0.8626667525939292,
+      "grad_norm": 0.0009519866972095988,
+      "learning_rate": 1.5331185105621197e-06,
+      "loss": 0.0,
+      "step": 13386
+    },
+    {
+      "epoch": 0.8627311980408584,
+      "grad_norm": 0.05243510078726089,
+      "learning_rate": 1.5324024346580737e-06,
+      "loss": 0.0001,
+      "step": 13387
+    },
+    {
+      "epoch": 0.8627956434877876,
+      "grad_norm": 0.0004253109465269119,
+      "learning_rate": 1.5316863587540282e-06,
+      "loss": 0.0,
+      "step": 13388
+    },
+    {
+      "epoch": 0.8628600889347168,
+      "grad_norm": 0.003103137673456359,
+      "learning_rate": 1.5309702828499823e-06,
+      "loss": 0.0,
+      "step": 13389
+    },
+    {
+      "epoch": 0.862924534381646,
+      "grad_norm": 0.0005662525075994763,
+      "learning_rate": 1.5302542069459364e-06,
+      "loss": 0.0,
+      "step": 13390
+    },
+    {
+      "epoch": 0.8629889798285751,
+      "grad_norm": 0.0021847116480264747,
+      "learning_rate": 1.5295381310418905e-06,
+      "loss": 0.0,
+      "step": 13391
+    },
+    {
+      "epoch": 0.8630534252755043,
+      "grad_norm": 0.1484782827970648,
+      "learning_rate": 1.5288220551378446e-06,
+      "loss": 0.0001,
+      "step": 13392
+    },
+    {
+      "epoch": 0.8631178707224335,
+      "grad_norm": 0.01737780890753743,
+      "learning_rate": 1.5281059792337987e-06,
+      "loss": 0.0,
+      "step": 13393
+    },
+    {
+      "epoch": 0.8631823161693626,
+      "grad_norm": 0.0061111167426056365,
+      "learning_rate": 1.5273899033297532e-06,
+      "loss": 0.0001,
+      "step": 13394
+    },
+    {
+      "epoch": 0.8632467616162918,
+      "grad_norm": 6.793325152705395e-05,
+      "learning_rate": 1.5266738274257073e-06,
+      "loss": 0.0,
+      "step": 13395
+    },
+    {
+      "epoch": 0.863311207063221,
+      "grad_norm": 0.0013361856110366322,
+      "learning_rate": 1.5259577515216614e-06,
+      "loss": 0.0,
+      "step": 13396
+    },
+    {
+      "epoch": 0.8633756525101501,
+      "grad_norm": 0.04722671877927456,
+      "learning_rate": 1.5252416756176155e-06,
+      "loss": 0.0,
+      "step": 13397
+    },
+    {
+      "epoch": 0.8634400979570793,
+      "grad_norm": 0.03648959870640535,
+      "learning_rate": 1.5245255997135698e-06,
+      "loss": 0.0001,
+      "step": 13398
+    },
+    {
+      "epoch": 0.8635045434040085,
+      "grad_norm": 0.0003418445429188015,
+      "learning_rate": 1.523809523809524e-06,
+      "loss": 0.0,
+      "step": 13399
+    },
+    {
+      "epoch": 0.8635689888509377,
+      "grad_norm": 0.0008608845575165467,
+      "learning_rate": 1.5230934479054781e-06,
+      "loss": 0.0,
+      "step": 13400
+    },
+    {
+      "epoch": 0.8636334342978669,
+      "grad_norm": 0.002113036385751762,
+      "learning_rate": 1.5223773720014322e-06,
+      "loss": 0.0,
+      "step": 13401
+    },
+    {
+      "epoch": 0.8636978797447961,
+      "grad_norm": 0.0015854664134770094,
+      "learning_rate": 1.5216612960973865e-06,
+      "loss": 0.0,
+      "step": 13402
+    },
+    {
+      "epoch": 0.8637623251917252,
+      "grad_norm": 0.0069202574918006445,
+      "learning_rate": 1.5209452201933406e-06,
+      "loss": 0.0,
+      "step": 13403
+    },
+    {
+      "epoch": 0.8638267706386544,
+      "grad_norm": 0.0801661461210914,
+      "learning_rate": 1.5202291442892947e-06,
+      "loss": 0.0005,
+      "step": 13404
+    },
+    {
+      "epoch": 0.8638912160855835,
+      "grad_norm": 0.03243276882707622,
+      "learning_rate": 1.519513068385249e-06,
+      "loss": 0.0017,
+      "step": 13405
+    },
+    {
+      "epoch": 0.8639556615325127,
+      "grad_norm": 0.0045153442530477235,
+      "learning_rate": 1.5187969924812033e-06,
+      "loss": 0.0,
+      "step": 13406
+    },
+    {
+      "epoch": 0.8640201069794419,
+      "grad_norm": 0.0018214858012334234,
+      "learning_rate": 1.5180809165771574e-06,
+      "loss": 0.0,
+      "step": 13407
+    },
+    {
+      "epoch": 0.864084552426371,
+      "grad_norm": 0.006574459754307482,
+      "learning_rate": 1.5173648406731115e-06,
+      "loss": 0.0,
+      "step": 13408
+    },
+    {
+      "epoch": 0.8641489978733002,
+      "grad_norm": 0.000748829527636167,
+      "learning_rate": 1.5166487647690656e-06,
+      "loss": 0.0,
+      "step": 13409
+    },
+    {
+      "epoch": 0.8642134433202294,
+      "grad_norm": 0.003037122112591288,
+      "learning_rate": 1.5159326888650196e-06,
+      "loss": 0.0,
+      "step": 13410
+    },
+    {
+      "epoch": 0.8642778887671586,
+      "grad_norm": 0.0004237258940566272,
+      "learning_rate": 1.5152166129609742e-06,
+      "loss": 0.0,
+      "step": 13411
+    },
+    {
+      "epoch": 0.8643423342140878,
+      "grad_norm": 0.03228855131648255,
+      "learning_rate": 1.5145005370569282e-06,
+      "loss": 0.0,
+      "step": 13412
+    },
+    {
+      "epoch": 0.864406779661017,
+      "grad_norm": 0.04334934425745581,
+      "learning_rate": 1.5137844611528823e-06,
+      "loss": 0.0001,
+      "step": 13413
+    },
+    {
+      "epoch": 0.8644712251079462,
+      "grad_norm": 0.003014186912717113,
+      "learning_rate": 1.5130683852488364e-06,
+      "loss": 0.0,
+      "step": 13414
+    },
+    {
+      "epoch": 0.8645356705548753,
+      "grad_norm": 0.0003104792844514378,
+      "learning_rate": 1.5123523093447905e-06,
+      "loss": 0.0,
+      "step": 13415
+    },
+    {
+      "epoch": 0.8646001160018044,
+      "grad_norm": 3.002535634592271e-05,
+      "learning_rate": 1.5116362334407448e-06,
+      "loss": 0.0,
+      "step": 13416
+    },
+    {
+      "epoch": 0.8646645614487336,
+      "grad_norm": 0.011916289380351773,
+      "learning_rate": 1.510920157536699e-06,
+      "loss": 0.0001,
+      "step": 13417
+    },
+    {
+      "epoch": 0.8647290068956628,
+      "grad_norm": 0.00014985182630863873,
+      "learning_rate": 1.5102040816326532e-06,
+      "loss": 0.0,
+      "step": 13418
+    },
+    {
+      "epoch": 0.864793452342592,
+      "grad_norm": 0.0009954287758235075,
+      "learning_rate": 1.5094880057286073e-06,
+      "loss": 0.0,
+      "step": 13419
+    },
+    {
+      "epoch": 0.8648578977895212,
+      "grad_norm": 0.020676748662305275,
+      "learning_rate": 1.5087719298245616e-06,
+      "loss": 0.0,
+      "step": 13420
+    },
+    {
+      "epoch": 0.8649223432364503,
+      "grad_norm": 0.016235645549665126,
+      "learning_rate": 1.5080558539205157e-06,
+      "loss": 0.0001,
+      "step": 13421
+    },
+    {
+      "epoch": 0.8649867886833795,
+      "grad_norm": 0.004837391867327394,
+      "learning_rate": 1.5073397780164697e-06,
+      "loss": 0.0,
+      "step": 13422
+    },
+    {
+      "epoch": 0.8650512341303087,
+      "grad_norm": 0.024349001979618767,
+      "learning_rate": 1.506623702112424e-06,
+      "loss": 0.0001,
+      "step": 13423
+    },
+    {
+      "epoch": 0.8651156795772379,
+      "grad_norm": 0.00068867812212618,
+      "learning_rate": 1.5059076262083783e-06,
+      "loss": 0.0,
+      "step": 13424
+    },
+    {
+      "epoch": 0.8651801250241671,
+      "grad_norm": 0.0016578578833495088,
+      "learning_rate": 1.5051915503043324e-06,
+      "loss": 0.0,
+      "step": 13425
+    },
+    {
+      "epoch": 0.8652445704710963,
+      "grad_norm": 0.03446109256285739,
+      "learning_rate": 1.5044754744002865e-06,
+      "loss": 0.0001,
+      "step": 13426
+    },
+    {
+      "epoch": 0.8653090159180253,
+      "grad_norm": 0.00011652735111931998,
+      "learning_rate": 1.5037593984962406e-06,
+      "loss": 0.0,
+      "step": 13427
+    },
+    {
+      "epoch": 0.8653734613649545,
+      "grad_norm": 0.115670160063929,
+      "learning_rate": 1.5030433225921947e-06,
+      "loss": 0.0006,
+      "step": 13428
+    },
+    {
+      "epoch": 0.8654379068118837,
+      "grad_norm": 0.0024135343454817215,
+      "learning_rate": 1.5023272466881492e-06,
+      "loss": 0.0,
+      "step": 13429
+    },
+    {
+      "epoch": 0.8655023522588129,
+      "grad_norm": 0.017187191238393462,
+      "learning_rate": 1.5016111707841033e-06,
+      "loss": 0.0001,
+      "step": 13430
+    },
+    {
+      "epoch": 0.8655667977057421,
+      "grad_norm": 0.35311210110829644,
+      "learning_rate": 1.5008950948800574e-06,
+      "loss": 0.0006,
+      "step": 13431
+    },
+    {
+      "epoch": 0.8656312431526713,
+      "grad_norm": 0.3191252663884319,
+      "learning_rate": 1.5001790189760115e-06,
+      "loss": 0.0019,
+      "step": 13432
+    },
+    {
+      "epoch": 0.8656956885996004,
+      "grad_norm": 0.04233723504764846,
+      "learning_rate": 1.4994629430719658e-06,
+      "loss": 0.0016,
+      "step": 13433
+    },
+    {
+      "epoch": 0.8657601340465296,
+      "grad_norm": 0.00145089825630913,
+      "learning_rate": 1.4987468671679198e-06,
+      "loss": 0.0,
+      "step": 13434
+    },
+    {
+      "epoch": 0.8658245794934588,
+      "grad_norm": 0.20710145012059114,
+      "learning_rate": 1.4980307912638741e-06,
+      "loss": 0.0002,
+      "step": 13435
+    },
+    {
+      "epoch": 0.865889024940388,
+      "grad_norm": 0.0031344022196231128,
+      "learning_rate": 1.4973147153598282e-06,
+      "loss": 0.0,
+      "step": 13436
+    },
+    {
+      "epoch": 0.8659534703873172,
+      "grad_norm": 0.0118111356298611,
+      "learning_rate": 1.4965986394557825e-06,
+      "loss": 0.0,
+      "step": 13437
+    },
+    {
+      "epoch": 0.8660179158342464,
+      "grad_norm": 0.0065583660622332745,
+      "learning_rate": 1.4958825635517366e-06,
+      "loss": 0.0,
+      "step": 13438
+    },
+    {
+      "epoch": 0.8660823612811754,
+      "grad_norm": 0.2883475069855709,
+      "learning_rate": 1.4951664876476907e-06,
+      "loss": 0.0025,
+      "step": 13439
+    },
+    {
+      "epoch": 0.8661468067281046,
+      "grad_norm": 0.005772183739680817,
+      "learning_rate": 1.494450411743645e-06,
+      "loss": 0.0,
+      "step": 13440
+    },
+    {
+      "epoch": 0.8662112521750338,
+      "grad_norm": 0.0052695852327754155,
+      "learning_rate": 1.4937343358395993e-06,
+      "loss": 0.0,
+      "step": 13441
+    },
+    {
+      "epoch": 0.866275697621963,
+      "grad_norm": 0.0012456353387858528,
+      "learning_rate": 1.4930182599355534e-06,
+      "loss": 0.0,
+      "step": 13442
+    },
+    {
+      "epoch": 0.8663401430688922,
+      "grad_norm": 0.0015878721758347974,
+      "learning_rate": 1.4923021840315075e-06,
+      "loss": 0.0,
+      "step": 13443
+    },
+    {
+      "epoch": 0.8664045885158214,
+      "grad_norm": 0.014455519841065518,
+      "learning_rate": 1.4915861081274616e-06,
+      "loss": 0.0,
+      "step": 13444
+    },
+    {
+      "epoch": 0.8664690339627505,
+      "grad_norm": 0.04454679008845088,
+      "learning_rate": 1.4908700322234157e-06,
+      "loss": 0.0001,
+      "step": 13445
+    },
+    {
+      "epoch": 0.8665334794096797,
+      "grad_norm": 0.45188975503978307,
+      "learning_rate": 1.4901539563193702e-06,
+      "loss": 0.003,
+      "step": 13446
+    },
+    {
+      "epoch": 0.8665979248566089,
+      "grad_norm": 0.010539871515908213,
+      "learning_rate": 1.4894378804153242e-06,
+      "loss": 0.0001,
+      "step": 13447
+    },
+    {
+      "epoch": 0.8666623703035381,
+      "grad_norm": 0.005161306898547368,
+      "learning_rate": 1.4887218045112783e-06,
+      "loss": 0.0,
+      "step": 13448
+    },
+    {
+      "epoch": 0.8667268157504673,
+      "grad_norm": 0.0016819094308091322,
+      "learning_rate": 1.4880057286072324e-06,
+      "loss": 0.0,
+      "step": 13449
+    },
+    {
+      "epoch": 0.8667912611973964,
+      "grad_norm": 0.13813809970298882,
+      "learning_rate": 1.4872896527031865e-06,
+      "loss": 0.0018,
+      "step": 13450
+    },
+    {
+      "epoch": 0.8668557066443255,
+      "grad_norm": 2.5470248661249853e-05,
+      "learning_rate": 1.4865735767991408e-06,
+      "loss": 0.0,
+      "step": 13451
+    },
+    {
+      "epoch": 0.8669201520912547,
+      "grad_norm": 0.001220750155884022,
+      "learning_rate": 1.4858575008950951e-06,
+      "loss": 0.0,
+      "step": 13452
+    },
+    {
+      "epoch": 0.8669845975381839,
+      "grad_norm": 0.09825203355172227,
+      "learning_rate": 1.4851414249910492e-06,
+      "loss": 0.001,
+      "step": 13453
+    },
+    {
+      "epoch": 0.8670490429851131,
+      "grad_norm": 0.0006788445189597282,
+      "learning_rate": 1.4844253490870033e-06,
+      "loss": 0.0,
+      "step": 13454
+    },
+    {
+      "epoch": 0.8671134884320423,
+      "grad_norm": 0.040948573152339476,
+      "learning_rate": 1.4837092731829576e-06,
+      "loss": 0.0002,
+      "step": 13455
+    },
+    {
+      "epoch": 0.8671779338789715,
+      "grad_norm": 1.2300284915043287,
+      "learning_rate": 1.4829931972789117e-06,
+      "loss": 0.0033,
+      "step": 13456
+    },
+    {
+      "epoch": 0.8672423793259006,
+      "grad_norm": 0.0013821472762745976,
+      "learning_rate": 1.4822771213748658e-06,
+      "loss": 0.0,
+      "step": 13457
+    },
+    {
+      "epoch": 0.8673068247728298,
+      "grad_norm": 0.0001129702660427535,
+      "learning_rate": 1.48156104547082e-06,
+      "loss": 0.0,
+      "step": 13458
+    },
+    {
+      "epoch": 0.867371270219759,
+      "grad_norm": 0.05667991960518738,
+      "learning_rate": 1.4808449695667744e-06,
+      "loss": 0.0002,
+      "step": 13459
+    },
+    {
+      "epoch": 0.8674357156666882,
+      "grad_norm": 0.01425209902570669,
+      "learning_rate": 1.4801288936627284e-06,
+      "loss": 0.0001,
+      "step": 13460
+    },
+    {
+      "epoch": 0.8675001611136173,
+      "grad_norm": 0.002137552431334319,
+      "learning_rate": 1.4794128177586825e-06,
+      "loss": 0.0,
+      "step": 13461
+    },
+    {
+      "epoch": 0.8675646065605465,
+      "grad_norm": 0.13578705766458848,
+      "learning_rate": 1.4786967418546366e-06,
+      "loss": 0.0001,
+      "step": 13462
+    },
+    {
+      "epoch": 0.8676290520074756,
+      "grad_norm": 0.0021114395670508357,
+      "learning_rate": 1.4779806659505907e-06,
+      "loss": 0.0,
+      "step": 13463
+    },
+    {
+      "epoch": 0.8676934974544048,
+      "grad_norm": 0.00019131494950189957,
+      "learning_rate": 1.4772645900465452e-06,
+      "loss": 0.0,
+      "step": 13464
+    },
+    {
+      "epoch": 0.867757942901334,
+      "grad_norm": 0.0035857106761646687,
+      "learning_rate": 1.4765485141424993e-06,
+      "loss": 0.0,
+      "step": 13465
+    },
+    {
+      "epoch": 0.8678223883482632,
+      "grad_norm": 0.0001692712253456178,
+      "learning_rate": 1.4758324382384534e-06,
+      "loss": 0.0,
+      "step": 13466
+    },
+    {
+      "epoch": 0.8678868337951924,
+      "grad_norm": 0.0005669241135073661,
+      "learning_rate": 1.4751163623344075e-06,
+      "loss": 0.0,
+      "step": 13467
+    },
+    {
+      "epoch": 0.8679512792421216,
+      "grad_norm": 0.001889031131247042,
+      "learning_rate": 1.4744002864303616e-06,
+      "loss": 0.0,
+      "step": 13468
+    },
+    {
+      "epoch": 0.8680157246890507,
+      "grad_norm": 0.05634150570667161,
+      "learning_rate": 1.4736842105263159e-06,
+      "loss": 0.0001,
+      "step": 13469
+    },
+    {
+      "epoch": 0.8680801701359799,
+      "grad_norm": 0.016770040901093828,
+      "learning_rate": 1.4729681346222702e-06,
+      "loss": 0.0,
+      "step": 13470
+    },
+    {
+      "epoch": 0.8681446155829091,
+      "grad_norm": 0.007393320511604408,
+      "learning_rate": 1.4722520587182242e-06,
+      "loss": 0.0,
+      "step": 13471
+    },
+    {
+      "epoch": 0.8682090610298382,
+      "grad_norm": 0.005720097209988372,
+      "learning_rate": 1.4715359828141783e-06,
+      "loss": 0.0,
+      "step": 13472
+    },
+    {
+      "epoch": 0.8682735064767674,
+      "grad_norm": 0.0014226270995623246,
+      "learning_rate": 1.4708199069101326e-06,
+      "loss": 0.0,
+      "step": 13473
+    },
+    {
+      "epoch": 0.8683379519236966,
+      "grad_norm": 0.03148733109778701,
+      "learning_rate": 1.4701038310060867e-06,
+      "loss": 0.0001,
+      "step": 13474
+    },
+    {
+      "epoch": 0.8684023973706257,
+      "grad_norm": 0.018190387811027366,
+      "learning_rate": 1.469387755102041e-06,
+      "loss": 0.0,
+      "step": 13475
+    },
+    {
+      "epoch": 0.8684668428175549,
+      "grad_norm": 0.00022876699057328286,
+      "learning_rate": 1.468671679197995e-06,
+      "loss": 0.0,
+      "step": 13476
+    },
+    {
+      "epoch": 0.8685312882644841,
+      "grad_norm": 0.007228422026355488,
+      "learning_rate": 1.4679556032939494e-06,
+      "loss": 0.0,
+      "step": 13477
+    },
+    {
+      "epoch": 0.8685957337114133,
+      "grad_norm": 0.022195556988556204,
+      "learning_rate": 1.4672395273899035e-06,
+      "loss": 0.0,
+      "step": 13478
+    },
+    {
+      "epoch": 0.8686601791583425,
+      "grad_norm": 0.0018545337949908754,
+      "learning_rate": 1.4665234514858576e-06,
+      "loss": 0.0,
+      "step": 13479
+    },
+    {
+      "epoch": 0.8687246246052717,
+      "grad_norm": 0.12720693745515196,
+      "learning_rate": 1.4658073755818117e-06,
+      "loss": 0.0003,
+      "step": 13480
+    },
+    {
+      "epoch": 0.8687890700522009,
+      "grad_norm": 0.14163634432106645,
+      "learning_rate": 1.4650912996777662e-06,
+      "loss": 0.0018,
+      "step": 13481
+    },
+    {
+      "epoch": 0.86885351549913,
+      "grad_norm": 0.4481010131587821,
+      "learning_rate": 1.4643752237737203e-06,
+      "loss": 0.0029,
+      "step": 13482
+    },
+    {
+      "epoch": 0.8689179609460591,
+      "grad_norm": 0.00217015626315937,
+      "learning_rate": 1.4636591478696743e-06,
+      "loss": 0.0,
+      "step": 13483
+    },
+    {
+      "epoch": 0.8689824063929883,
+      "grad_norm": 0.0025758466115319723,
+      "learning_rate": 1.4629430719656284e-06,
+      "loss": 0.0,
+      "step": 13484
+    },
+    {
+      "epoch": 0.8690468518399175,
+      "grad_norm": 0.44933947723859485,
+      "learning_rate": 1.4622269960615825e-06,
+      "loss": 0.0008,
+      "step": 13485
+    },
+    {
+      "epoch": 0.8691112972868467,
+      "grad_norm": 0.006622376254887656,
+      "learning_rate": 1.4615109201575366e-06,
+      "loss": 0.0,
+      "step": 13486
+    },
+    {
+      "epoch": 0.8691757427337758,
+      "grad_norm": 0.010118893463182543,
+      "learning_rate": 1.4607948442534911e-06,
+      "loss": 0.0,
+      "step": 13487
+    },
+    {
+      "epoch": 0.869240188180705,
+      "grad_norm": 0.002533162206517466,
+      "learning_rate": 1.4600787683494452e-06,
+      "loss": 0.0,
+      "step": 13488
+    },
+    {
+      "epoch": 0.8693046336276342,
+      "grad_norm": 0.0046699535377350216,
+      "learning_rate": 1.4593626924453993e-06,
+      "loss": 0.0,
+      "step": 13489
+    },
+    {
+      "epoch": 0.8693690790745634,
+      "grad_norm": 0.0006231500933800507,
+      "learning_rate": 1.4586466165413534e-06,
+      "loss": 0.0,
+      "step": 13490
+    },
+    {
+      "epoch": 0.8694335245214926,
+      "grad_norm": 0.0129105570295586,
+      "learning_rate": 1.4579305406373077e-06,
+      "loss": 0.0,
+      "step": 13491
+    },
+    {
+      "epoch": 0.8694979699684218,
+      "grad_norm": 0.6536391411143632,
+      "learning_rate": 1.4572144647332618e-06,
+      "loss": 0.003,
+      "step": 13492
+    },
+    {
+      "epoch": 0.869562415415351,
+      "grad_norm": 0.00645190186822942,
+      "learning_rate": 1.456498388829216e-06,
+      "loss": 0.0,
+      "step": 13493
+    },
+    {
+      "epoch": 0.86962686086228,
+      "grad_norm": 0.021776802677437902,
+      "learning_rate": 1.4557823129251701e-06,
+      "loss": 0.0002,
+      "step": 13494
+    },
+    {
+      "epoch": 0.8696913063092092,
+      "grad_norm": 0.9354631868333679,
+      "learning_rate": 1.4550662370211244e-06,
+      "loss": 0.0034,
+      "step": 13495
+    },
+    {
+      "epoch": 0.8697557517561384,
+      "grad_norm": 0.24174388126058172,
+      "learning_rate": 1.4543501611170785e-06,
+      "loss": 0.002,
+      "step": 13496
+    },
+    {
+      "epoch": 0.8698201972030676,
+      "grad_norm": 0.002051742025662557,
+      "learning_rate": 1.4536340852130326e-06,
+      "loss": 0.0,
+      "step": 13497
+    },
+    {
+      "epoch": 0.8698846426499968,
+      "grad_norm": 0.03131467117702846,
+      "learning_rate": 1.4529180093089867e-06,
+      "loss": 0.0,
+      "step": 13498
+    },
+    {
+      "epoch": 0.869949088096926,
+      "grad_norm": 0.016429974192892383,
+      "learning_rate": 1.4522019334049412e-06,
+      "loss": 0.0002,
+      "step": 13499
+    },
+    {
+      "epoch": 0.8700135335438551,
+      "grad_norm": 0.0035691864897899376,
+      "learning_rate": 1.4514858575008953e-06,
+      "loss": 0.0,
+      "step": 13500
+    },
+    {
+      "epoch": 0.8700779789907843,
+      "grad_norm": 0.014284719551174541,
+      "learning_rate": 1.4507697815968494e-06,
+      "loss": 0.0,
+      "step": 13501
+    },
+    {
+      "epoch": 0.8701424244377135,
+      "grad_norm": 0.0003508945434332547,
+      "learning_rate": 1.4500537056928035e-06,
+      "loss": 0.0,
+      "step": 13502
+    },
+    {
+      "epoch": 0.8702068698846427,
+      "grad_norm": 0.005536226193423074,
+      "learning_rate": 1.4493376297887576e-06,
+      "loss": 0.0,
+      "step": 13503
+    },
+    {
+      "epoch": 0.8702713153315719,
+      "grad_norm": 0.013431267923642803,
+      "learning_rate": 1.4486215538847119e-06,
+      "loss": 0.0,
+      "step": 13504
+    },
+    {
+      "epoch": 0.8703357607785009,
+      "grad_norm": 0.03924292381700065,
+      "learning_rate": 1.4479054779806662e-06,
+      "loss": 0.0,
+      "step": 13505
+    },
+    {
+      "epoch": 0.8704002062254301,
+      "grad_norm": 0.014671159050005588,
+      "learning_rate": 1.4471894020766202e-06,
+      "loss": 0.0,
+      "step": 13506
+    },
+    {
+      "epoch": 0.8704646516723593,
+      "grad_norm": 0.00044016681878419526,
+      "learning_rate": 1.4464733261725743e-06,
+      "loss": 0.0,
+      "step": 13507
+    },
+    {
+      "epoch": 0.8705290971192885,
+      "grad_norm": 0.0027930283003449025,
+      "learning_rate": 1.4457572502685286e-06,
+      "loss": 0.0,
+      "step": 13508
+    },
+    {
+      "epoch": 0.8705935425662177,
+      "grad_norm": 0.005349489845537028,
+      "learning_rate": 1.4450411743644827e-06,
+      "loss": 0.0,
+      "step": 13509
+    },
+    {
+      "epoch": 0.8706579880131469,
+      "grad_norm": 0.0025331927243357815,
+      "learning_rate": 1.444325098460437e-06,
+      "loss": 0.0,
+      "step": 13510
+    },
+    {
+      "epoch": 0.870722433460076,
+      "grad_norm": 0.1705458464266017,
+      "learning_rate": 1.4436090225563911e-06,
+      "loss": 0.0003,
+      "step": 13511
+    },
+    {
+      "epoch": 0.8707868789070052,
+      "grad_norm": 0.017371828112596428,
+      "learning_rate": 1.4428929466523454e-06,
+      "loss": 0.0,
+      "step": 13512
+    },
+    {
+      "epoch": 0.8708513243539344,
+      "grad_norm": 0.0038565979744293725,
+      "learning_rate": 1.4421768707482995e-06,
+      "loss": 0.0,
+      "step": 13513
+    },
+    {
+      "epoch": 0.8709157698008636,
+      "grad_norm": 0.0015431336010864626,
+      "learning_rate": 1.4414607948442536e-06,
+      "loss": 0.0,
+      "step": 13514
+    },
+    {
+      "epoch": 0.8709802152477928,
+      "grad_norm": 0.38062584393036586,
+      "learning_rate": 1.4407447189402077e-06,
+      "loss": 0.0007,
+      "step": 13515
+    },
+    {
+      "epoch": 0.871044660694722,
+      "grad_norm": 0.002684096329343692,
+      "learning_rate": 1.4400286430361622e-06,
+      "loss": 0.0,
+      "step": 13516
+    },
+    {
+      "epoch": 0.871109106141651,
+      "grad_norm": 0.0004909211941232381,
+      "learning_rate": 1.4393125671321163e-06,
+      "loss": 0.0,
+      "step": 13517
+    },
+    {
+      "epoch": 0.8711735515885802,
+      "grad_norm": 0.001157450722865243,
+      "learning_rate": 1.4385964912280704e-06,
+      "loss": 0.0,
+      "step": 13518
+    },
+    {
+      "epoch": 0.8712379970355094,
+      "grad_norm": 0.0035391587568699614,
+      "learning_rate": 1.4378804153240244e-06,
+      "loss": 0.0,
+      "step": 13519
+    },
+    {
+      "epoch": 0.8713024424824386,
+      "grad_norm": 0.00045686885179854926,
+      "learning_rate": 1.4371643394199785e-06,
+      "loss": 0.0,
+      "step": 13520
+    },
+    {
+      "epoch": 0.8713668879293678,
+      "grad_norm": 0.0021138812403863616,
+      "learning_rate": 1.4364482635159326e-06,
+      "loss": 0.0,
+      "step": 13521
+    },
+    {
+      "epoch": 0.871431333376297,
+      "grad_norm": 0.2128194987620841,
+      "learning_rate": 1.4357321876118871e-06,
+      "loss": 0.0006,
+      "step": 13522
+    },
+    {
+      "epoch": 0.8714957788232262,
+      "grad_norm": 0.01931788608564039,
+      "learning_rate": 1.4350161117078412e-06,
+      "loss": 0.0,
+      "step": 13523
+    },
+    {
+      "epoch": 0.8715602242701553,
+      "grad_norm": 0.0003648647198489776,
+      "learning_rate": 1.4343000358037953e-06,
+      "loss": 0.0,
+      "step": 13524
+    },
+    {
+      "epoch": 0.8716246697170845,
+      "grad_norm": 0.004620281109182782,
+      "learning_rate": 1.4335839598997494e-06,
+      "loss": 0.0,
+      "step": 13525
+    },
+    {
+      "epoch": 0.8716891151640137,
+      "grad_norm": 0.017122646764027347,
+      "learning_rate": 1.4328678839957037e-06,
+      "loss": 0.0,
+      "step": 13526
+    },
+    {
+      "epoch": 0.8717535606109429,
+      "grad_norm": 0.0012365848463775104,
+      "learning_rate": 1.4321518080916578e-06,
+      "loss": 0.0,
+      "step": 13527
+    },
+    {
+      "epoch": 0.871818006057872,
+      "grad_norm": 0.0016554455275155562,
+      "learning_rate": 1.431435732187612e-06,
+      "loss": 0.0,
+      "step": 13528
+    },
+    {
+      "epoch": 0.8718824515048011,
+      "grad_norm": 0.018926993280560703,
+      "learning_rate": 1.4307196562835662e-06,
+      "loss": 0.0015,
+      "step": 13529
+    },
+    {
+      "epoch": 0.8719468969517303,
+      "grad_norm": 0.014421366017851302,
+      "learning_rate": 1.4300035803795205e-06,
+      "loss": 0.0001,
+      "step": 13530
+    },
+    {
+      "epoch": 0.8720113423986595,
+      "grad_norm": 0.0012043756470092263,
+      "learning_rate": 1.4292875044754745e-06,
+      "loss": 0.0,
+      "step": 13531
+    },
+    {
+      "epoch": 0.8720757878455887,
+      "grad_norm": 0.06249889730434415,
+      "learning_rate": 1.4285714285714286e-06,
+      "loss": 0.0001,
+      "step": 13532
+    },
+    {
+      "epoch": 0.8721402332925179,
+      "grad_norm": 0.00483204331224003,
+      "learning_rate": 1.4278553526673827e-06,
+      "loss": 0.0,
+      "step": 13533
+    },
+    {
+      "epoch": 0.8722046787394471,
+      "grad_norm": 0.016188550479828548,
+      "learning_rate": 1.4271392767633372e-06,
+      "loss": 0.0001,
+      "step": 13534
+    },
+    {
+      "epoch": 0.8722691241863763,
+      "grad_norm": 0.2588630282665265,
+      "learning_rate": 1.4264232008592913e-06,
+      "loss": 0.0022,
+      "step": 13535
+    },
+    {
+      "epoch": 0.8723335696333054,
+      "grad_norm": 0.017522219661073486,
+      "learning_rate": 1.4257071249552454e-06,
+      "loss": 0.0,
+      "step": 13536
+    },
+    {
+      "epoch": 0.8723980150802346,
+      "grad_norm": 4.777473903327223,
+      "learning_rate": 1.4249910490511995e-06,
+      "loss": 0.0184,
+      "step": 13537
+    },
+    {
+      "epoch": 0.8724624605271638,
+      "grad_norm": 0.1122914928480347,
+      "learning_rate": 1.4242749731471536e-06,
+      "loss": 0.0,
+      "step": 13538
+    },
+    {
+      "epoch": 0.8725269059740929,
+      "grad_norm": 0.0003457079673620274,
+      "learning_rate": 1.4235588972431077e-06,
+      "loss": 0.0,
+      "step": 13539
+    },
+    {
+      "epoch": 0.8725913514210221,
+      "grad_norm": 0.008842368697230442,
+      "learning_rate": 1.4228428213390622e-06,
+      "loss": 0.0,
+      "step": 13540
+    },
+    {
+      "epoch": 0.8726557968679513,
+      "grad_norm": 0.16889078991419842,
+      "learning_rate": 1.4221267454350163e-06,
+      "loss": 0.0005,
+      "step": 13541
+    },
+    {
+      "epoch": 0.8727202423148804,
+      "grad_norm": 0.09180406278511782,
+      "learning_rate": 1.4214106695309703e-06,
+      "loss": 0.0001,
+      "step": 13542
+    },
+    {
+      "epoch": 0.8727846877618096,
+      "grad_norm": 0.0002542434149114111,
+      "learning_rate": 1.4206945936269244e-06,
+      "loss": 0.0,
+      "step": 13543
+    },
+    {
+      "epoch": 0.8728491332087388,
+      "grad_norm": 0.00021139095458569033,
+      "learning_rate": 1.4199785177228787e-06,
+      "loss": 0.0,
+      "step": 13544
+    },
+    {
+      "epoch": 0.872913578655668,
+      "grad_norm": 0.006701556482451194,
+      "learning_rate": 1.419262441818833e-06,
+      "loss": 0.0,
+      "step": 13545
+    },
+    {
+      "epoch": 0.8729780241025972,
+      "grad_norm": 0.0005409555739583791,
+      "learning_rate": 1.4185463659147871e-06,
+      "loss": 0.0,
+      "step": 13546
+    },
+    {
+      "epoch": 0.8730424695495264,
+      "grad_norm": 0.29029280227176923,
+      "learning_rate": 1.4178302900107412e-06,
+      "loss": 0.0005,
+      "step": 13547
+    },
+    {
+      "epoch": 0.8731069149964555,
+      "grad_norm": 0.003792905193337104,
+      "learning_rate": 1.4171142141066955e-06,
+      "loss": 0.0,
+      "step": 13548
+    },
+    {
+      "epoch": 0.8731713604433847,
+      "grad_norm": 0.027987387467200055,
+      "learning_rate": 1.4163981382026496e-06,
+      "loss": 0.0,
+      "step": 13549
+    },
+    {
+      "epoch": 0.8732358058903138,
+      "grad_norm": 0.004768562503807217,
+      "learning_rate": 1.4156820622986037e-06,
+      "loss": 0.0,
+      "step": 13550
+    },
+    {
+      "epoch": 0.873300251337243,
+      "grad_norm": 0.0008558681731003965,
+      "learning_rate": 1.414965986394558e-06,
+      "loss": 0.0,
+      "step": 13551
+    },
+    {
+      "epoch": 0.8733646967841722,
+      "grad_norm": 0.09874931857320618,
+      "learning_rate": 1.4142499104905123e-06,
+      "loss": 0.0001,
+      "step": 13552
+    },
+    {
+      "epoch": 0.8734291422311014,
+      "grad_norm": 0.5997927258018302,
+      "learning_rate": 1.4135338345864664e-06,
+      "loss": 0.0015,
+      "step": 13553
+    },
+    {
+      "epoch": 0.8734935876780305,
+      "grad_norm": 0.06590779190624765,
+      "learning_rate": 1.4128177586824204e-06,
+      "loss": 0.0001,
+      "step": 13554
+    },
+    {
+      "epoch": 0.8735580331249597,
+      "grad_norm": 0.010813035118637227,
+      "learning_rate": 1.4121016827783745e-06,
+      "loss": 0.0001,
+      "step": 13555
+    },
+    {
+      "epoch": 0.8736224785718889,
+      "grad_norm": 0.0007776535748360822,
+      "learning_rate": 1.4113856068743286e-06,
+      "loss": 0.0,
+      "step": 13556
+    },
+    {
+      "epoch": 0.8736869240188181,
+      "grad_norm": 0.514541151528957,
+      "learning_rate": 1.4106695309702831e-06,
+      "loss": 0.0216,
+      "step": 13557
+    },
+    {
+      "epoch": 0.8737513694657473,
+      "grad_norm": 0.0011680135680204743,
+      "learning_rate": 1.4099534550662372e-06,
+      "loss": 0.0,
+      "step": 13558
+    },
+    {
+      "epoch": 0.8738158149126765,
+      "grad_norm": 0.001274273368860219,
+      "learning_rate": 1.4092373791621913e-06,
+      "loss": 0.0,
+      "step": 13559
+    },
+    {
+      "epoch": 0.8738802603596056,
+      "grad_norm": 0.0016573716185651368,
+      "learning_rate": 1.4085213032581454e-06,
+      "loss": 0.0,
+      "step": 13560
+    },
+    {
+      "epoch": 0.8739447058065347,
+      "grad_norm": 0.0005385912905206142,
+      "learning_rate": 1.4078052273540995e-06,
+      "loss": 0.0,
+      "step": 13561
+    },
+    {
+      "epoch": 0.8740091512534639,
+      "grad_norm": 0.09338146530950199,
+      "learning_rate": 1.4070891514500538e-06,
+      "loss": 0.0002,
+      "step": 13562
+    },
+    {
+      "epoch": 0.8740735967003931,
+      "grad_norm": 0.0030035474103590263,
+      "learning_rate": 1.406373075546008e-06,
+      "loss": 0.0,
+      "step": 13563
+    },
+    {
+      "epoch": 0.8741380421473223,
+      "grad_norm": 0.0005251321157146958,
+      "learning_rate": 1.4056569996419622e-06,
+      "loss": 0.0,
+      "step": 13564
+    },
+    {
+      "epoch": 0.8742024875942515,
+      "grad_norm": 0.0021644242133924342,
+      "learning_rate": 1.4049409237379162e-06,
+      "loss": 0.0,
+      "step": 13565
+    },
+    {
+      "epoch": 0.8742669330411806,
+      "grad_norm": 0.0002113496909306623,
+      "learning_rate": 1.4042248478338705e-06,
+      "loss": 0.0,
+      "step": 13566
+    },
+    {
+      "epoch": 0.8743313784881098,
+      "grad_norm": 0.002869868115987156,
+      "learning_rate": 1.4035087719298246e-06,
+      "loss": 0.0,
+      "step": 13567
+    },
+    {
+      "epoch": 0.874395823935039,
+      "grad_norm": 3.127468659924674,
+      "learning_rate": 1.4027926960257787e-06,
+      "loss": 0.0123,
+      "step": 13568
+    },
+    {
+      "epoch": 0.8744602693819682,
+      "grad_norm": 0.08067815140353134,
+      "learning_rate": 1.402076620121733e-06,
+      "loss": 0.0002,
+      "step": 13569
+    },
+    {
+      "epoch": 0.8745247148288974,
+      "grad_norm": 0.017830586287385566,
+      "learning_rate": 1.4013605442176873e-06,
+      "loss": 0.0,
+      "step": 13570
+    },
+    {
+      "epoch": 0.8745891602758266,
+      "grad_norm": 0.006552162045926843,
+      "learning_rate": 1.4006444683136414e-06,
+      "loss": 0.0,
+      "step": 13571
+    },
+    {
+      "epoch": 0.8746536057227556,
+      "grad_norm": 8.130109325365869e-05,
+      "learning_rate": 1.3999283924095955e-06,
+      "loss": 0.0,
+      "step": 13572
+    },
+    {
+      "epoch": 0.8747180511696848,
+      "grad_norm": 0.00481453068839515,
+      "learning_rate": 1.3992123165055496e-06,
+      "loss": 0.0,
+      "step": 13573
+    },
+    {
+      "epoch": 0.874782496616614,
+      "grad_norm": 0.0016966316384423679,
+      "learning_rate": 1.3984962406015037e-06,
+      "loss": 0.0,
+      "step": 13574
+    },
+    {
+      "epoch": 0.8748469420635432,
+      "grad_norm": 6.507700313948872e-05,
+      "learning_rate": 1.3977801646974582e-06,
+      "loss": 0.0,
+      "step": 13575
+    },
+    {
+      "epoch": 0.8749113875104724,
+      "grad_norm": 0.0010351706398880871,
+      "learning_rate": 1.3970640887934123e-06,
+      "loss": 0.0,
+      "step": 13576
+    },
+    {
+      "epoch": 0.8749758329574016,
+      "grad_norm": 0.002042297500884413,
+      "learning_rate": 1.3963480128893664e-06,
+      "loss": 0.0,
+      "step": 13577
+    },
+    {
+      "epoch": 0.8750402784043307,
+      "grad_norm": 0.00035755874045150096,
+      "learning_rate": 1.3956319369853204e-06,
+      "loss": 0.0,
+      "step": 13578
+    },
+    {
+      "epoch": 0.8751047238512599,
+      "grad_norm": 0.0007519831593113182,
+      "learning_rate": 1.3949158610812747e-06,
+      "loss": 0.0,
+      "step": 13579
+    },
+    {
+      "epoch": 0.8751691692981891,
+      "grad_norm": 0.01997057158629694,
+      "learning_rate": 1.394199785177229e-06,
+      "loss": 0.0015,
+      "step": 13580
+    },
+    {
+      "epoch": 0.8752336147451183,
+      "grad_norm": 0.00042799543401737973,
+      "learning_rate": 1.3934837092731831e-06,
+      "loss": 0.0,
+      "step": 13581
+    },
+    {
+      "epoch": 0.8752980601920475,
+      "grad_norm": 0.003192598232773263,
+      "learning_rate": 1.3927676333691372e-06,
+      "loss": 0.0,
+      "step": 13582
+    },
+    {
+      "epoch": 0.8753625056389766,
+      "grad_norm": 0.00041092294582336747,
+      "learning_rate": 1.3920515574650915e-06,
+      "loss": 0.0,
+      "step": 13583
+    },
+    {
+      "epoch": 0.8754269510859057,
+      "grad_norm": 0.00015048739275314575,
+      "learning_rate": 1.3913354815610456e-06,
+      "loss": 0.0,
+      "step": 13584
+    },
+    {
+      "epoch": 0.8754913965328349,
+      "grad_norm": 0.00040192569485888335,
+      "learning_rate": 1.3906194056569997e-06,
+      "loss": 0.0,
+      "step": 13585
+    },
+    {
+      "epoch": 0.8755558419797641,
+      "grad_norm": 0.003118728490399756,
+      "learning_rate": 1.389903329752954e-06,
+      "loss": 0.0,
+      "step": 13586
+    },
+    {
+      "epoch": 0.8756202874266933,
+      "grad_norm": 0.028921511163939883,
+      "learning_rate": 1.3891872538489083e-06,
+      "loss": 0.0003,
+      "step": 13587
+    },
+    {
+      "epoch": 0.8756847328736225,
+      "grad_norm": 0.0007635765470389747,
+      "learning_rate": 1.3884711779448624e-06,
+      "loss": 0.0,
+      "step": 13588
+    },
+    {
+      "epoch": 0.8757491783205517,
+      "grad_norm": 0.019386178158575473,
+      "learning_rate": 1.3877551020408165e-06,
+      "loss": 0.0001,
+      "step": 13589
+    },
+    {
+      "epoch": 0.8758136237674808,
+      "grad_norm": 0.008952326877814495,
+      "learning_rate": 1.3870390261367705e-06,
+      "loss": 0.0,
+      "step": 13590
+    },
+    {
+      "epoch": 0.87587806921441,
+      "grad_norm": 0.0064969913757071,
+      "learning_rate": 1.3863229502327246e-06,
+      "loss": 0.0,
+      "step": 13591
+    },
+    {
+      "epoch": 0.8759425146613392,
+      "grad_norm": 0.24652262056953084,
+      "learning_rate": 1.3856068743286791e-06,
+      "loss": 0.0098,
+      "step": 13592
+    },
+    {
+      "epoch": 0.8760069601082684,
+      "grad_norm": 0.008717637446990048,
+      "learning_rate": 1.3848907984246332e-06,
+      "loss": 0.0,
+      "step": 13593
+    },
+    {
+      "epoch": 0.8760714055551976,
+      "grad_norm": 0.1055407587350373,
+      "learning_rate": 1.3841747225205873e-06,
+      "loss": 0.0001,
+      "step": 13594
+    },
+    {
+      "epoch": 0.8761358510021267,
+      "grad_norm": 0.013464757423937554,
+      "learning_rate": 1.3834586466165414e-06,
+      "loss": 0.0,
+      "step": 13595
+    },
+    {
+      "epoch": 0.8762002964490558,
+      "grad_norm": 0.0014337143804371206,
+      "learning_rate": 1.3827425707124955e-06,
+      "loss": 0.0,
+      "step": 13596
+    },
+    {
+      "epoch": 0.876264741895985,
+      "grad_norm": 0.4674874787515268,
+      "learning_rate": 1.3820264948084498e-06,
+      "loss": 0.0015,
+      "step": 13597
+    },
+    {
+      "epoch": 0.8763291873429142,
+      "grad_norm": 0.00098483335952236,
+      "learning_rate": 1.381310418904404e-06,
+      "loss": 0.0,
+      "step": 13598
+    },
+    {
+      "epoch": 0.8763936327898434,
+      "grad_norm": 0.09651554507659374,
+      "learning_rate": 1.3805943430003582e-06,
+      "loss": 0.0002,
+      "step": 13599
+    },
+    {
+      "epoch": 0.8764580782367726,
+      "grad_norm": 0.05975753060859266,
+      "learning_rate": 1.3798782670963123e-06,
+      "loss": 0.0006,
+      "step": 13600
+    },
+    {
+      "epoch": 0.8765225236837018,
+      "grad_norm": 0.0008181424472536412,
+      "learning_rate": 1.3791621911922666e-06,
+      "loss": 0.0,
+      "step": 13601
+    },
+    {
+      "epoch": 0.8765869691306309,
+      "grad_norm": 0.0007309612300930143,
+      "learning_rate": 1.3784461152882206e-06,
+      "loss": 0.0,
+      "step": 13602
+    },
+    {
+      "epoch": 0.8766514145775601,
+      "grad_norm": 0.0004453427397565943,
+      "learning_rate": 1.3777300393841747e-06,
+      "loss": 0.0,
+      "step": 13603
+    },
+    {
+      "epoch": 0.8767158600244893,
+      "grad_norm": 0.0013161070099909815,
+      "learning_rate": 1.377013963480129e-06,
+      "loss": 0.0,
+      "step": 13604
+    },
+    {
+      "epoch": 0.8767803054714185,
+      "grad_norm": 8.2985330607004e-05,
+      "learning_rate": 1.3762978875760833e-06,
+      "loss": 0.0,
+      "step": 13605
+    },
+    {
+      "epoch": 0.8768447509183476,
+      "grad_norm": 0.0014163201861122309,
+      "learning_rate": 1.3755818116720374e-06,
+      "loss": 0.0,
+      "step": 13606
+    },
+    {
+      "epoch": 0.8769091963652768,
+      "grad_norm": 0.006504134348321924,
+      "learning_rate": 1.3748657357679915e-06,
+      "loss": 0.0,
+      "step": 13607
+    },
+    {
+      "epoch": 0.8769736418122059,
+      "grad_norm": 0.004427846073773377,
+      "learning_rate": 1.3741496598639456e-06,
+      "loss": 0.0,
+      "step": 13608
+    },
+    {
+      "epoch": 0.8770380872591351,
+      "grad_norm": 8.069661349110574e-05,
+      "learning_rate": 1.3734335839598997e-06,
+      "loss": 0.0,
+      "step": 13609
+    },
+    {
+      "epoch": 0.8771025327060643,
+      "grad_norm": 0.00013724066858743323,
+      "learning_rate": 1.3727175080558542e-06,
+      "loss": 0.0,
+      "step": 13610
+    },
+    {
+      "epoch": 0.8771669781529935,
+      "grad_norm": 1.0793487189499922e-05,
+      "learning_rate": 1.3720014321518083e-06,
+      "loss": 0.0,
+      "step": 13611
+    },
+    {
+      "epoch": 0.8772314235999227,
+      "grad_norm": 0.016116090551314,
+      "learning_rate": 1.3712853562477624e-06,
+      "loss": 0.0001,
+      "step": 13612
+    },
+    {
+      "epoch": 0.8772958690468519,
+      "grad_norm": 0.0044309911209259294,
+      "learning_rate": 1.3705692803437164e-06,
+      "loss": 0.0,
+      "step": 13613
+    },
+    {
+      "epoch": 0.877360314493781,
+      "grad_norm": 0.00038663612199369644,
+      "learning_rate": 1.3698532044396705e-06,
+      "loss": 0.0,
+      "step": 13614
+    },
+    {
+      "epoch": 0.8774247599407102,
+      "grad_norm": 0.00034108201632122604,
+      "learning_rate": 1.369137128535625e-06,
+      "loss": 0.0,
+      "step": 13615
+    },
+    {
+      "epoch": 0.8774892053876394,
+      "grad_norm": 0.0007516695032004047,
+      "learning_rate": 1.3684210526315791e-06,
+      "loss": 0.0,
+      "step": 13616
+    },
+    {
+      "epoch": 0.8775536508345685,
+      "grad_norm": 0.000810677722890353,
+      "learning_rate": 1.3677049767275332e-06,
+      "loss": 0.0,
+      "step": 13617
+    },
+    {
+      "epoch": 0.8776180962814977,
+      "grad_norm": 0.0004293978015281151,
+      "learning_rate": 1.3669889008234873e-06,
+      "loss": 0.0,
+      "step": 13618
+    },
+    {
+      "epoch": 0.8776825417284269,
+      "grad_norm": 0.09996970768416442,
+      "learning_rate": 1.3662728249194416e-06,
+      "loss": 0.0018,
+      "step": 13619
+    },
+    {
+      "epoch": 0.877746987175356,
+      "grad_norm": 0.009210777517721869,
+      "learning_rate": 1.3655567490153957e-06,
+      "loss": 0.0,
+      "step": 13620
+    },
+    {
+      "epoch": 0.8778114326222852,
+      "grad_norm": 2.0031004060201308e-05,
+      "learning_rate": 1.36484067311135e-06,
+      "loss": 0.0,
+      "step": 13621
+    },
+    {
+      "epoch": 0.8778758780692144,
+      "grad_norm": 0.0002399283084197288,
+      "learning_rate": 1.364124597207304e-06,
+      "loss": 0.0,
+      "step": 13622
+    },
+    {
+      "epoch": 0.8779403235161436,
+      "grad_norm": 0.0014824010634681267,
+      "learning_rate": 1.3634085213032584e-06,
+      "loss": 0.0,
+      "step": 13623
+    },
+    {
+      "epoch": 0.8780047689630728,
+      "grad_norm": 0.0006106176764788257,
+      "learning_rate": 1.3626924453992125e-06,
+      "loss": 0.0,
+      "step": 13624
+    },
+    {
+      "epoch": 0.878069214410002,
+      "grad_norm": 0.009545374588239326,
+      "learning_rate": 1.3619763694951665e-06,
+      "loss": 0.0,
+      "step": 13625
+    },
+    {
+      "epoch": 0.8781336598569311,
+      "grad_norm": 0.00035692053622342487,
+      "learning_rate": 1.3612602935911206e-06,
+      "loss": 0.0,
+      "step": 13626
+    },
+    {
+      "epoch": 0.8781981053038603,
+      "grad_norm": 0.0086509276995505,
+      "learning_rate": 1.3605442176870751e-06,
+      "loss": 0.0,
+      "step": 13627
+    },
+    {
+      "epoch": 0.8782625507507894,
+      "grad_norm": 0.00010314129543442811,
+      "learning_rate": 1.3598281417830292e-06,
+      "loss": 0.0,
+      "step": 13628
+    },
+    {
+      "epoch": 0.8783269961977186,
+      "grad_norm": 0.01060045892036667,
+      "learning_rate": 1.3591120658789833e-06,
+      "loss": 0.0,
+      "step": 13629
+    },
+    {
+      "epoch": 0.8783914416446478,
+      "grad_norm": 0.0026332486628218815,
+      "learning_rate": 1.3583959899749374e-06,
+      "loss": 0.0,
+      "step": 13630
+    },
+    {
+      "epoch": 0.878455887091577,
+      "grad_norm": 0.004919190051721374,
+      "learning_rate": 1.3576799140708915e-06,
+      "loss": 0.0,
+      "step": 13631
+    },
+    {
+      "epoch": 0.8785203325385061,
+      "grad_norm": 0.005739669680137454,
+      "learning_rate": 1.3569638381668456e-06,
+      "loss": 0.0,
+      "step": 13632
+    },
+    {
+      "epoch": 0.8785847779854353,
+      "grad_norm": 0.12541555085615214,
+      "learning_rate": 1.3562477622628e-06,
+      "loss": 0.0016,
+      "step": 13633
+    },
+    {
+      "epoch": 0.8786492234323645,
+      "grad_norm": 0.009859135451052567,
+      "learning_rate": 1.3555316863587542e-06,
+      "loss": 0.0001,
+      "step": 13634
+    },
+    {
+      "epoch": 0.8787136688792937,
+      "grad_norm": 0.28341660181746003,
+      "learning_rate": 1.3548156104547083e-06,
+      "loss": 0.0006,
+      "step": 13635
+    },
+    {
+      "epoch": 0.8787781143262229,
+      "grad_norm": 0.0016875942051804806,
+      "learning_rate": 1.3540995345506624e-06,
+      "loss": 0.0,
+      "step": 13636
+    },
+    {
+      "epoch": 0.8788425597731521,
+      "grad_norm": 0.0005612762262681204,
+      "learning_rate": 1.3533834586466167e-06,
+      "loss": 0.0,
+      "step": 13637
+    },
+    {
+      "epoch": 0.8789070052200812,
+      "grad_norm": 0.01833501293870889,
+      "learning_rate": 1.3526673827425707e-06,
+      "loss": 0.0,
+      "step": 13638
+    },
+    {
+      "epoch": 0.8789714506670103,
+      "grad_norm": 0.0016657878041175568,
+      "learning_rate": 1.351951306838525e-06,
+      "loss": 0.0,
+      "step": 13639
+    },
+    {
+      "epoch": 0.8790358961139395,
+      "grad_norm": 0.008593055425421249,
+      "learning_rate": 1.3512352309344791e-06,
+      "loss": 0.0,
+      "step": 13640
+    },
+    {
+      "epoch": 0.8791003415608687,
+      "grad_norm": 0.022741383577491645,
+      "learning_rate": 1.3505191550304334e-06,
+      "loss": 0.0002,
+      "step": 13641
+    },
+    {
+      "epoch": 0.8791647870077979,
+      "grad_norm": 0.00012513328344494333,
+      "learning_rate": 1.3498030791263875e-06,
+      "loss": 0.0,
+      "step": 13642
+    },
+    {
+      "epoch": 0.8792292324547271,
+      "grad_norm": 0.0015158990746431459,
+      "learning_rate": 1.3490870032223416e-06,
+      "loss": 0.0,
+      "step": 13643
+    },
+    {
+      "epoch": 0.8792936779016562,
+      "grad_norm": 0.0001142028649185917,
+      "learning_rate": 1.3483709273182957e-06,
+      "loss": 0.0,
+      "step": 13644
+    },
+    {
+      "epoch": 0.8793581233485854,
+      "grad_norm": 0.00018615067818089466,
+      "learning_rate": 1.3476548514142502e-06,
+      "loss": 0.0,
+      "step": 13645
+    },
+    {
+      "epoch": 0.8794225687955146,
+      "grad_norm": 0.0016702490090621004,
+      "learning_rate": 1.3469387755102043e-06,
+      "loss": 0.0,
+      "step": 13646
+    },
+    {
+      "epoch": 0.8794870142424438,
+      "grad_norm": 0.014081349773854846,
+      "learning_rate": 1.3462226996061584e-06,
+      "loss": 0.0,
+      "step": 13647
+    },
+    {
+      "epoch": 0.879551459689373,
+      "grad_norm": 0.35911500897104526,
+      "learning_rate": 1.3455066237021125e-06,
+      "loss": 0.0003,
+      "step": 13648
+    },
+    {
+      "epoch": 0.8796159051363022,
+      "grad_norm": 0.0004700586450096311,
+      "learning_rate": 1.3447905477980665e-06,
+      "loss": 0.0,
+      "step": 13649
+    },
+    {
+      "epoch": 0.8796803505832312,
+      "grad_norm": 0.007744483732701911,
+      "learning_rate": 1.3440744718940208e-06,
+      "loss": 0.0001,
+      "step": 13650
+    },
+    {
+      "epoch": 0.8797447960301604,
+      "grad_norm": 0.0002823123017943778,
+      "learning_rate": 1.3433583959899751e-06,
+      "loss": 0.0,
+      "step": 13651
+    },
+    {
+      "epoch": 0.8798092414770896,
+      "grad_norm": 0.0020454451113791103,
+      "learning_rate": 1.3426423200859292e-06,
+      "loss": 0.0,
+      "step": 13652
+    },
+    {
+      "epoch": 0.8798736869240188,
+      "grad_norm": 0.011770034371449568,
+      "learning_rate": 1.3419262441818833e-06,
+      "loss": 0.0,
+      "step": 13653
+    },
+    {
+      "epoch": 0.879938132370948,
+      "grad_norm": 0.0009106569832223424,
+      "learning_rate": 1.3412101682778376e-06,
+      "loss": 0.0,
+      "step": 13654
+    },
+    {
+      "epoch": 0.8800025778178772,
+      "grad_norm": 0.00044701932244403723,
+      "learning_rate": 1.3404940923737917e-06,
+      "loss": 0.0,
+      "step": 13655
+    },
+    {
+      "epoch": 0.8800670232648063,
+      "grad_norm": 0.0013901423332051905,
+      "learning_rate": 1.339778016469746e-06,
+      "loss": 0.0,
+      "step": 13656
+    },
+    {
+      "epoch": 0.8801314687117355,
+      "grad_norm": 0.024266204133080428,
+      "learning_rate": 1.3390619405657e-06,
+      "loss": 0.0,
+      "step": 13657
+    },
+    {
+      "epoch": 0.8801959141586647,
+      "grad_norm": 0.003448365666168964,
+      "learning_rate": 1.3383458646616544e-06,
+      "loss": 0.0,
+      "step": 13658
+    },
+    {
+      "epoch": 0.8802603596055939,
+      "grad_norm": 0.25593555568473547,
+      "learning_rate": 1.3376297887576085e-06,
+      "loss": 0.0009,
+      "step": 13659
+    },
+    {
+      "epoch": 0.8803248050525231,
+      "grad_norm": 0.00023793433604057834,
+      "learning_rate": 1.3369137128535626e-06,
+      "loss": 0.0,
+      "step": 13660
+    },
+    {
+      "epoch": 0.8803892504994522,
+      "grad_norm": 0.024291282911824568,
+      "learning_rate": 1.3361976369495166e-06,
+      "loss": 0.0,
+      "step": 13661
+    },
+    {
+      "epoch": 0.8804536959463813,
+      "grad_norm": 0.18787491587994204,
+      "learning_rate": 1.3354815610454712e-06,
+      "loss": 0.0014,
+      "step": 13662
+    },
+    {
+      "epoch": 0.8805181413933105,
+      "grad_norm": 0.00011337929585352012,
+      "learning_rate": 1.3347654851414252e-06,
+      "loss": 0.0,
+      "step": 13663
+    },
+    {
+      "epoch": 0.8805825868402397,
+      "grad_norm": 0.00015175064550529223,
+      "learning_rate": 1.3340494092373793e-06,
+      "loss": 0.0,
+      "step": 13664
+    },
+    {
+      "epoch": 0.8806470322871689,
+      "grad_norm": 0.012094931872168792,
+      "learning_rate": 1.3333333333333334e-06,
+      "loss": 0.0,
+      "step": 13665
+    },
+    {
+      "epoch": 0.8807114777340981,
+      "grad_norm": 0.008657847704660484,
+      "learning_rate": 1.3326172574292875e-06,
+      "loss": 0.0001,
+      "step": 13666
+    },
+    {
+      "epoch": 0.8807759231810273,
+      "grad_norm": 0.0029345671366418132,
+      "learning_rate": 1.3319011815252416e-06,
+      "loss": 0.0,
+      "step": 13667
+    },
+    {
+      "epoch": 0.8808403686279564,
+      "grad_norm": 0.00023515681443448538,
+      "learning_rate": 1.331185105621196e-06,
+      "loss": 0.0,
+      "step": 13668
+    },
+    {
+      "epoch": 0.8809048140748856,
+      "grad_norm": 0.0001054609110638124,
+      "learning_rate": 1.3304690297171502e-06,
+      "loss": 0.0,
+      "step": 13669
+    },
+    {
+      "epoch": 0.8809692595218148,
+      "grad_norm": 0.0027009243656194505,
+      "learning_rate": 1.3297529538131043e-06,
+      "loss": 0.0,
+      "step": 13670
+    },
+    {
+      "epoch": 0.881033704968744,
+      "grad_norm": 0.00012685416549973138,
+      "learning_rate": 1.3290368779090584e-06,
+      "loss": 0.0,
+      "step": 13671
+    },
+    {
+      "epoch": 0.8810981504156732,
+      "grad_norm": 0.3780272996016023,
+      "learning_rate": 1.3283208020050127e-06,
+      "loss": 0.0023,
+      "step": 13672
+    },
+    {
+      "epoch": 0.8811625958626023,
+      "grad_norm": 0.10550351364688096,
+      "learning_rate": 1.3276047261009667e-06,
+      "loss": 0.0001,
+      "step": 13673
+    },
+    {
+      "epoch": 0.8812270413095314,
+      "grad_norm": 0.0005343506373210069,
+      "learning_rate": 1.326888650196921e-06,
+      "loss": 0.0,
+      "step": 13674
+    },
+    {
+      "epoch": 0.8812914867564606,
+      "grad_norm": 0.006842115762950535,
+      "learning_rate": 1.3261725742928751e-06,
+      "loss": 0.0,
+      "step": 13675
+    },
+    {
+      "epoch": 0.8813559322033898,
+      "grad_norm": 0.0013787853414366827,
+      "learning_rate": 1.3254564983888294e-06,
+      "loss": 0.0,
+      "step": 13676
+    },
+    {
+      "epoch": 0.881420377650319,
+      "grad_norm": 0.6943399292392509,
+      "learning_rate": 1.3247404224847835e-06,
+      "loss": 0.0052,
+      "step": 13677
+    },
+    {
+      "epoch": 0.8814848230972482,
+      "grad_norm": 0.001786099160118304,
+      "learning_rate": 1.3240243465807376e-06,
+      "loss": 0.0,
+      "step": 13678
+    },
+    {
+      "epoch": 0.8815492685441774,
+      "grad_norm": 0.0029717481777851596,
+      "learning_rate": 1.3233082706766917e-06,
+      "loss": 0.0,
+      "step": 13679
+    },
+    {
+      "epoch": 0.8816137139911066,
+      "grad_norm": 6.38256307935207e-05,
+      "learning_rate": 1.3225921947726462e-06,
+      "loss": 0.0,
+      "step": 13680
+    },
+    {
+      "epoch": 0.8816781594380357,
+      "grad_norm": 0.0002884480060583603,
+      "learning_rate": 1.3218761188686003e-06,
+      "loss": 0.0,
+      "step": 13681
+    },
+    {
+      "epoch": 0.8817426048849649,
+      "grad_norm": 0.1623689515082662,
+      "learning_rate": 1.3211600429645544e-06,
+      "loss": 0.0002,
+      "step": 13682
+    },
+    {
+      "epoch": 0.8818070503318941,
+      "grad_norm": 0.04362992253845935,
+      "learning_rate": 1.3204439670605085e-06,
+      "loss": 0.0001,
+      "step": 13683
+    },
+    {
+      "epoch": 0.8818714957788232,
+      "grad_norm": 0.22108830118373307,
+      "learning_rate": 1.3197278911564625e-06,
+      "loss": 0.0004,
+      "step": 13684
+    },
+    {
+      "epoch": 0.8819359412257524,
+      "grad_norm": 0.0003036379952213748,
+      "learning_rate": 1.3190118152524166e-06,
+      "loss": 0.0,
+      "step": 13685
+    },
+    {
+      "epoch": 0.8820003866726815,
+      "grad_norm": 0.008215158033441384,
+      "learning_rate": 1.3182957393483711e-06,
+      "loss": 0.0001,
+      "step": 13686
+    },
+    {
+      "epoch": 0.8820648321196107,
+      "grad_norm": 0.000746369885476597,
+      "learning_rate": 1.3175796634443252e-06,
+      "loss": 0.0,
+      "step": 13687
+    },
+    {
+      "epoch": 0.8821292775665399,
+      "grad_norm": 0.2331175370722083,
+      "learning_rate": 1.3168635875402793e-06,
+      "loss": 0.0001,
+      "step": 13688
+    },
+    {
+      "epoch": 0.8821937230134691,
+      "grad_norm": 3.3127325739645425e-05,
+      "learning_rate": 1.3161475116362334e-06,
+      "loss": 0.0,
+      "step": 13689
+    },
+    {
+      "epoch": 0.8822581684603983,
+      "grad_norm": 0.00046292881256141657,
+      "learning_rate": 1.3154314357321877e-06,
+      "loss": 0.0,
+      "step": 13690
+    },
+    {
+      "epoch": 0.8823226139073275,
+      "grad_norm": 2.4879745552462693e-05,
+      "learning_rate": 1.314715359828142e-06,
+      "loss": 0.0,
+      "step": 13691
+    },
+    {
+      "epoch": 0.8823870593542567,
+      "grad_norm": 0.024843685636272067,
+      "learning_rate": 1.313999283924096e-06,
+      "loss": 0.0,
+      "step": 13692
+    },
+    {
+      "epoch": 0.8824515048011858,
+      "grad_norm": 0.0033880595142095345,
+      "learning_rate": 1.3132832080200502e-06,
+      "loss": 0.0,
+      "step": 13693
+    },
+    {
+      "epoch": 0.882515950248115,
+      "grad_norm": 0.005841345105160331,
+      "learning_rate": 1.3125671321160045e-06,
+      "loss": 0.0,
+      "step": 13694
+    },
+    {
+      "epoch": 0.8825803956950441,
+      "grad_norm": 0.0034708352770504813,
+      "learning_rate": 1.3118510562119586e-06,
+      "loss": 0.0,
+      "step": 13695
+    },
+    {
+      "epoch": 0.8826448411419733,
+      "grad_norm": 0.0011531345218125334,
+      "learning_rate": 1.3111349803079127e-06,
+      "loss": 0.0,
+      "step": 13696
+    },
+    {
+      "epoch": 0.8827092865889025,
+      "grad_norm": 0.0008902594119790399,
+      "learning_rate": 1.310418904403867e-06,
+      "loss": 0.0,
+      "step": 13697
+    },
+    {
+      "epoch": 0.8827737320358316,
+      "grad_norm": 0.02761833022301556,
+      "learning_rate": 1.3097028284998212e-06,
+      "loss": 0.0,
+      "step": 13698
+    },
+    {
+      "epoch": 0.8828381774827608,
+      "grad_norm": 7.322041241513758e-05,
+      "learning_rate": 1.3089867525957753e-06,
+      "loss": 0.0,
+      "step": 13699
+    },
+    {
+      "epoch": 0.88290262292969,
+      "grad_norm": 0.0003351661766753556,
+      "learning_rate": 1.3082706766917294e-06,
+      "loss": 0.0,
+      "step": 13700
+    },
+    {
+      "epoch": 0.8829670683766192,
+      "grad_norm": 0.13306850415458327,
+      "learning_rate": 1.3075546007876835e-06,
+      "loss": 0.0003,
+      "step": 13701
+    },
+    {
+      "epoch": 0.8830315138235484,
+      "grad_norm": 0.0009763088853932016,
+      "learning_rate": 1.3068385248836376e-06,
+      "loss": 0.0,
+      "step": 13702
+    },
+    {
+      "epoch": 0.8830959592704776,
+      "grad_norm": 0.030336426954044204,
+      "learning_rate": 1.3061224489795921e-06,
+      "loss": 0.0002,
+      "step": 13703
+    },
+    {
+      "epoch": 0.8831604047174068,
+      "grad_norm": 0.3518987319355953,
+      "learning_rate": 1.3054063730755462e-06,
+      "loss": 0.0013,
+      "step": 13704
+    },
+    {
+      "epoch": 0.8832248501643359,
+      "grad_norm": 0.00021259481191901802,
+      "learning_rate": 1.3046902971715003e-06,
+      "loss": 0.0,
+      "step": 13705
+    },
+    {
+      "epoch": 0.883289295611265,
+      "grad_norm": 0.0012099330641390034,
+      "learning_rate": 1.3039742212674544e-06,
+      "loss": 0.0,
+      "step": 13706
+    },
+    {
+      "epoch": 0.8833537410581942,
+      "grad_norm": 0.01886221954292911,
+      "learning_rate": 1.3032581453634085e-06,
+      "loss": 0.0,
+      "step": 13707
+    },
+    {
+      "epoch": 0.8834181865051234,
+      "grad_norm": 0.09828483459776614,
+      "learning_rate": 1.3025420694593628e-06,
+      "loss": 0.0001,
+      "step": 13708
+    },
+    {
+      "epoch": 0.8834826319520526,
+      "grad_norm": 0.0005948211234448418,
+      "learning_rate": 1.301825993555317e-06,
+      "loss": 0.0,
+      "step": 13709
+    },
+    {
+      "epoch": 0.8835470773989818,
+      "grad_norm": 0.00020818992887561486,
+      "learning_rate": 1.3011099176512711e-06,
+      "loss": 0.0,
+      "step": 13710
+    },
+    {
+      "epoch": 0.8836115228459109,
+      "grad_norm": 0.1842981683540968,
+      "learning_rate": 1.3003938417472252e-06,
+      "loss": 0.0011,
+      "step": 13711
+    },
+    {
+      "epoch": 0.8836759682928401,
+      "grad_norm": 0.008536273547773218,
+      "learning_rate": 1.2996777658431795e-06,
+      "loss": 0.0001,
+      "step": 13712
+    },
+    {
+      "epoch": 0.8837404137397693,
+      "grad_norm": 0.00032051570706314344,
+      "learning_rate": 1.2989616899391336e-06,
+      "loss": 0.0,
+      "step": 13713
+    },
+    {
+      "epoch": 0.8838048591866985,
+      "grad_norm": 0.17510439577339187,
+      "learning_rate": 1.2982456140350877e-06,
+      "loss": 0.0005,
+      "step": 13714
+    },
+    {
+      "epoch": 0.8838693046336277,
+      "grad_norm": 0.01028719993330639,
+      "learning_rate": 1.297529538131042e-06,
+      "loss": 0.0,
+      "step": 13715
+    },
+    {
+      "epoch": 0.8839337500805569,
+      "grad_norm": 0.08046575674952673,
+      "learning_rate": 1.2968134622269963e-06,
+      "loss": 0.0017,
+      "step": 13716
+    },
+    {
+      "epoch": 0.8839981955274859,
+      "grad_norm": 0.0006212313252662225,
+      "learning_rate": 1.2960973863229504e-06,
+      "loss": 0.0,
+      "step": 13717
+    },
+    {
+      "epoch": 0.8840626409744151,
+      "grad_norm": 0.00033610867341460235,
+      "learning_rate": 1.2953813104189045e-06,
+      "loss": 0.0,
+      "step": 13718
+    },
+    {
+      "epoch": 0.8841270864213443,
+      "grad_norm": 0.00023716963033959892,
+      "learning_rate": 1.2946652345148586e-06,
+      "loss": 0.0,
+      "step": 13719
+    },
+    {
+      "epoch": 0.8841915318682735,
+      "grad_norm": 0.0016286595639973542,
+      "learning_rate": 1.2939491586108126e-06,
+      "loss": 0.0,
+      "step": 13720
+    },
+    {
+      "epoch": 0.8842559773152027,
+      "grad_norm": 0.031614340354932574,
+      "learning_rate": 1.2932330827067672e-06,
+      "loss": 0.0001,
+      "step": 13721
+    },
+    {
+      "epoch": 0.8843204227621319,
+      "grad_norm": 0.00017104715739226403,
+      "learning_rate": 1.2925170068027212e-06,
+      "loss": 0.0,
+      "step": 13722
+    },
+    {
+      "epoch": 0.884384868209061,
+      "grad_norm": 9.70356569071254e-05,
+      "learning_rate": 1.2918009308986753e-06,
+      "loss": 0.0,
+      "step": 13723
+    },
+    {
+      "epoch": 0.8844493136559902,
+      "grad_norm": 0.3967907666004509,
+      "learning_rate": 1.2910848549946294e-06,
+      "loss": 0.0011,
+      "step": 13724
+    },
+    {
+      "epoch": 0.8845137591029194,
+      "grad_norm": 0.0008351767774307439,
+      "learning_rate": 1.2903687790905837e-06,
+      "loss": 0.0,
+      "step": 13725
+    },
+    {
+      "epoch": 0.8845782045498486,
+      "grad_norm": 0.0010768701346120296,
+      "learning_rate": 1.289652703186538e-06,
+      "loss": 0.0,
+      "step": 13726
+    },
+    {
+      "epoch": 0.8846426499967778,
+      "grad_norm": 0.00035160258067471443,
+      "learning_rate": 1.288936627282492e-06,
+      "loss": 0.0,
+      "step": 13727
+    },
+    {
+      "epoch": 0.8847070954437068,
+      "grad_norm": 0.21524670698834755,
+      "learning_rate": 1.2882205513784462e-06,
+      "loss": 0.0003,
+      "step": 13728
+    },
+    {
+      "epoch": 0.884771540890636,
+      "grad_norm": 7.928518740038473e-05,
+      "learning_rate": 1.2875044754744005e-06,
+      "loss": 0.0,
+      "step": 13729
+    },
+    {
+      "epoch": 0.8848359863375652,
+      "grad_norm": 0.7751929273734448,
+      "learning_rate": 1.2867883995703546e-06,
+      "loss": 0.0052,
+      "step": 13730
+    },
+    {
+      "epoch": 0.8849004317844944,
+      "grad_norm": 0.000505047569173897,
+      "learning_rate": 1.2860723236663087e-06,
+      "loss": 0.0,
+      "step": 13731
+    },
+    {
+      "epoch": 0.8849648772314236,
+      "grad_norm": 0.0017112831956432482,
+      "learning_rate": 1.285356247762263e-06,
+      "loss": 0.0,
+      "step": 13732
+    },
+    {
+      "epoch": 0.8850293226783528,
+      "grad_norm": 2.923113262558036e-05,
+      "learning_rate": 1.2846401718582173e-06,
+      "loss": 0.0,
+      "step": 13733
+    },
+    {
+      "epoch": 0.885093768125282,
+      "grad_norm": 0.08377971496993508,
+      "learning_rate": 1.2839240959541713e-06,
+      "loss": 0.0017,
+      "step": 13734
+    },
+    {
+      "epoch": 0.8851582135722111,
+      "grad_norm": 0.0018758007745659337,
+      "learning_rate": 1.2832080200501254e-06,
+      "loss": 0.0,
+      "step": 13735
+    },
+    {
+      "epoch": 0.8852226590191403,
+      "grad_norm": 0.005519941912064853,
+      "learning_rate": 1.2824919441460795e-06,
+      "loss": 0.0,
+      "step": 13736
+    },
+    {
+      "epoch": 0.8852871044660695,
+      "grad_norm": 0.007801686846673793,
+      "learning_rate": 1.2817758682420336e-06,
+      "loss": 0.0001,
+      "step": 13737
+    },
+    {
+      "epoch": 0.8853515499129987,
+      "grad_norm": 0.3482728911432587,
+      "learning_rate": 1.2810597923379881e-06,
+      "loss": 0.0008,
+      "step": 13738
+    },
+    {
+      "epoch": 0.8854159953599279,
+      "grad_norm": 0.0009555235755133371,
+      "learning_rate": 1.2803437164339422e-06,
+      "loss": 0.0,
+      "step": 13739
+    },
+    {
+      "epoch": 0.885480440806857,
+      "grad_norm": 0.002469717287478994,
+      "learning_rate": 1.2796276405298963e-06,
+      "loss": 0.0,
+      "step": 13740
+    },
+    {
+      "epoch": 0.8855448862537861,
+      "grad_norm": 0.0026660388959545405,
+      "learning_rate": 1.2789115646258504e-06,
+      "loss": 0.0,
+      "step": 13741
+    },
+    {
+      "epoch": 0.8856093317007153,
+      "grad_norm": 0.005242530312289297,
+      "learning_rate": 1.2781954887218045e-06,
+      "loss": 0.0,
+      "step": 13742
+    },
+    {
+      "epoch": 0.8856737771476445,
+      "grad_norm": 9.516788258618955e-05,
+      "learning_rate": 1.2774794128177588e-06,
+      "loss": 0.0,
+      "step": 13743
+    },
+    {
+      "epoch": 0.8857382225945737,
+      "grad_norm": 2.566322068723762e-05,
+      "learning_rate": 1.276763336913713e-06,
+      "loss": 0.0,
+      "step": 13744
+    },
+    {
+      "epoch": 0.8858026680415029,
+      "grad_norm": 0.00032851554042659717,
+      "learning_rate": 1.2760472610096671e-06,
+      "loss": 0.0,
+      "step": 13745
+    },
+    {
+      "epoch": 0.8858671134884321,
+      "grad_norm": 0.2881616395959894,
+      "learning_rate": 1.2753311851056212e-06,
+      "loss": 0.0024,
+      "step": 13746
+    },
+    {
+      "epoch": 0.8859315589353612,
+      "grad_norm": 0.0005407381204699677,
+      "learning_rate": 1.2746151092015755e-06,
+      "loss": 0.0,
+      "step": 13747
+    },
+    {
+      "epoch": 0.8859960043822904,
+      "grad_norm": 0.00013134327426654014,
+      "learning_rate": 1.2738990332975296e-06,
+      "loss": 0.0,
+      "step": 13748
+    },
+    {
+      "epoch": 0.8860604498292196,
+      "grad_norm": 1.0938324603394522e-05,
+      "learning_rate": 1.2731829573934837e-06,
+      "loss": 0.0,
+      "step": 13749
+    },
+    {
+      "epoch": 0.8861248952761488,
+      "grad_norm": 0.004271410915227554,
+      "learning_rate": 1.272466881489438e-06,
+      "loss": 0.0,
+      "step": 13750
+    },
+    {
+      "epoch": 0.8861893407230779,
+      "grad_norm": 0.0026957620892503614,
+      "learning_rate": 1.2717508055853923e-06,
+      "loss": 0.0,
+      "step": 13751
+    },
+    {
+      "epoch": 0.886253786170007,
+      "grad_norm": 0.0006152756950276613,
+      "learning_rate": 1.2710347296813464e-06,
+      "loss": 0.0,
+      "step": 13752
+    },
+    {
+      "epoch": 0.8863182316169362,
+      "grad_norm": 2.2124446079129158e-05,
+      "learning_rate": 1.2703186537773005e-06,
+      "loss": 0.0,
+      "step": 13753
+    },
+    {
+      "epoch": 0.8863826770638654,
+      "grad_norm": 0.05126629319637686,
+      "learning_rate": 1.2696025778732546e-06,
+      "loss": 0.0001,
+      "step": 13754
+    },
+    {
+      "epoch": 0.8864471225107946,
+      "grad_norm": 0.017536264153307488,
+      "learning_rate": 1.2688865019692087e-06,
+      "loss": 0.0001,
+      "step": 13755
+    },
+    {
+      "epoch": 0.8865115679577238,
+      "grad_norm": 0.04960957108481395,
+      "learning_rate": 1.2681704260651632e-06,
+      "loss": 0.0,
+      "step": 13756
+    },
+    {
+      "epoch": 0.886576013404653,
+      "grad_norm": 0.005953873003306027,
+      "learning_rate": 1.2674543501611172e-06,
+      "loss": 0.0001,
+      "step": 13757
+    },
+    {
+      "epoch": 0.8866404588515822,
+      "grad_norm": 0.0004576293244010673,
+      "learning_rate": 1.2667382742570713e-06,
+      "loss": 0.0,
+      "step": 13758
+    },
+    {
+      "epoch": 0.8867049042985113,
+      "grad_norm": 0.002728465719264111,
+      "learning_rate": 1.2660221983530254e-06,
+      "loss": 0.0,
+      "step": 13759
+    },
+    {
+      "epoch": 0.8867693497454405,
+      "grad_norm": 0.0020225571803567754,
+      "learning_rate": 1.2653061224489795e-06,
+      "loss": 0.0,
+      "step": 13760
+    },
+    {
+      "epoch": 0.8868337951923697,
+      "grad_norm": 0.14241467212327003,
+      "learning_rate": 1.264590046544934e-06,
+      "loss": 0.0004,
+      "step": 13761
+    },
+    {
+      "epoch": 0.8868982406392988,
+      "grad_norm": 0.0014042617102855625,
+      "learning_rate": 1.2638739706408881e-06,
+      "loss": 0.0,
+      "step": 13762
+    },
+    {
+      "epoch": 0.886962686086228,
+      "grad_norm": 0.029573956164921412,
+      "learning_rate": 1.2631578947368422e-06,
+      "loss": 0.0,
+      "step": 13763
+    },
+    {
+      "epoch": 0.8870271315331572,
+      "grad_norm": 0.0030826232614705414,
+      "learning_rate": 1.2624418188327963e-06,
+      "loss": 0.0,
+      "step": 13764
+    },
+    {
+      "epoch": 0.8870915769800863,
+      "grad_norm": 0.000188945426545367,
+      "learning_rate": 1.2617257429287506e-06,
+      "loss": 0.0,
+      "step": 13765
+    },
+    {
+      "epoch": 0.8871560224270155,
+      "grad_norm": 0.008708895375226261,
+      "learning_rate": 1.2610096670247047e-06,
+      "loss": 0.0,
+      "step": 13766
+    },
+    {
+      "epoch": 0.8872204678739447,
+      "grad_norm": 0.17166571446342943,
+      "learning_rate": 1.260293591120659e-06,
+      "loss": 0.0007,
+      "step": 13767
+    },
+    {
+      "epoch": 0.8872849133208739,
+      "grad_norm": 0.0806280824807589,
+      "learning_rate": 1.259577515216613e-06,
+      "loss": 0.0008,
+      "step": 13768
+    },
+    {
+      "epoch": 0.8873493587678031,
+      "grad_norm": 0.00011126534919590591,
+      "learning_rate": 1.2588614393125673e-06,
+      "loss": 0.0,
+      "step": 13769
+    },
+    {
+      "epoch": 0.8874138042147323,
+      "grad_norm": 0.0013592761131405947,
+      "learning_rate": 1.2581453634085214e-06,
+      "loss": 0.0,
+      "step": 13770
+    },
+    {
+      "epoch": 0.8874782496616614,
+      "grad_norm": 0.011567165436922357,
+      "learning_rate": 1.2574292875044755e-06,
+      "loss": 0.0,
+      "step": 13771
+    },
+    {
+      "epoch": 0.8875426951085906,
+      "grad_norm": 0.0010506584973813378,
+      "learning_rate": 1.2567132116004296e-06,
+      "loss": 0.0,
+      "step": 13772
+    },
+    {
+      "epoch": 0.8876071405555197,
+      "grad_norm": 0.001634937844128216,
+      "learning_rate": 1.2559971356963841e-06,
+      "loss": 0.0015,
+      "step": 13773
+    },
+    {
+      "epoch": 0.8876715860024489,
+      "grad_norm": 0.0007561086547375215,
+      "learning_rate": 1.2552810597923382e-06,
+      "loss": 0.0,
+      "step": 13774
+    },
+    {
+      "epoch": 0.8877360314493781,
+      "grad_norm": 0.00021426457144976387,
+      "learning_rate": 1.2545649838882923e-06,
+      "loss": 0.0,
+      "step": 13775
+    },
+    {
+      "epoch": 0.8878004768963073,
+      "grad_norm": 0.0017384501229145124,
+      "learning_rate": 1.2538489079842464e-06,
+      "loss": 0.0,
+      "step": 13776
+    },
+    {
+      "epoch": 0.8878649223432364,
+      "grad_norm": 0.46608758452350224,
+      "learning_rate": 1.2531328320802005e-06,
+      "loss": 0.003,
+      "step": 13777
+    },
+    {
+      "epoch": 0.8879293677901656,
+      "grad_norm": 0.003180161390498234,
+      "learning_rate": 1.2524167561761546e-06,
+      "loss": 0.0,
+      "step": 13778
+    },
+    {
+      "epoch": 0.8879938132370948,
+      "grad_norm": 0.0005952020863398482,
+      "learning_rate": 1.251700680272109e-06,
+      "loss": 0.0,
+      "step": 13779
+    },
+    {
+      "epoch": 0.888058258684024,
+      "grad_norm": 0.0008731909170846803,
+      "learning_rate": 1.2509846043680632e-06,
+      "loss": 0.0,
+      "step": 13780
+    },
+    {
+      "epoch": 0.8881227041309532,
+      "grad_norm": 0.003438820389549209,
+      "learning_rate": 1.2502685284640172e-06,
+      "loss": 0.0,
+      "step": 13781
+    },
+    {
+      "epoch": 0.8881871495778824,
+      "grad_norm": 0.00016075516900061808,
+      "learning_rate": 1.2495524525599713e-06,
+      "loss": 0.0,
+      "step": 13782
+    },
+    {
+      "epoch": 0.8882515950248115,
+      "grad_norm": 0.005191481164772671,
+      "learning_rate": 1.2488363766559256e-06,
+      "loss": 0.0,
+      "step": 13783
+    },
+    {
+      "epoch": 0.8883160404717406,
+      "grad_norm": 0.0009777725279008903,
+      "learning_rate": 1.2481203007518797e-06,
+      "loss": 0.0,
+      "step": 13784
+    },
+    {
+      "epoch": 0.8883804859186698,
+      "grad_norm": 0.18591865636404056,
+      "learning_rate": 1.247404224847834e-06,
+      "loss": 0.0002,
+      "step": 13785
+    },
+    {
+      "epoch": 0.888444931365599,
+      "grad_norm": 0.016696817939353084,
+      "learning_rate": 1.246688148943788e-06,
+      "loss": 0.0001,
+      "step": 13786
+    },
+    {
+      "epoch": 0.8885093768125282,
+      "grad_norm": 0.005742868049428506,
+      "learning_rate": 1.2459720730397424e-06,
+      "loss": 0.0001,
+      "step": 13787
+    },
+    {
+      "epoch": 0.8885738222594574,
+      "grad_norm": 0.0033753578995078114,
+      "learning_rate": 1.2452559971356965e-06,
+      "loss": 0.0,
+      "step": 13788
+    },
+    {
+      "epoch": 0.8886382677063865,
+      "grad_norm": 0.001299590697601426,
+      "learning_rate": 1.2445399212316508e-06,
+      "loss": 0.0,
+      "step": 13789
+    },
+    {
+      "epoch": 0.8887027131533157,
+      "grad_norm": 0.01598992051312426,
+      "learning_rate": 1.2438238453276049e-06,
+      "loss": 0.0002,
+      "step": 13790
+    },
+    {
+      "epoch": 0.8887671586002449,
+      "grad_norm": 0.7328409038968464,
+      "learning_rate": 1.243107769423559e-06,
+      "loss": 0.0008,
+      "step": 13791
+    },
+    {
+      "epoch": 0.8888316040471741,
+      "grad_norm": 0.00030574028092213986,
+      "learning_rate": 1.2423916935195133e-06,
+      "loss": 0.0,
+      "step": 13792
+    },
+    {
+      "epoch": 0.8888960494941033,
+      "grad_norm": 0.0005879745614522931,
+      "learning_rate": 1.2416756176154673e-06,
+      "loss": 0.0,
+      "step": 13793
+    },
+    {
+      "epoch": 0.8889604949410325,
+      "grad_norm": 0.010634271394679871,
+      "learning_rate": 1.2409595417114214e-06,
+      "loss": 0.0,
+      "step": 13794
+    },
+    {
+      "epoch": 0.8890249403879615,
+      "grad_norm": 0.0029319469486168174,
+      "learning_rate": 1.2402434658073757e-06,
+      "loss": 0.0,
+      "step": 13795
+    },
+    {
+      "epoch": 0.8890893858348907,
+      "grad_norm": 0.00044063952636015224,
+      "learning_rate": 1.2395273899033298e-06,
+      "loss": 0.0,
+      "step": 13796
+    },
+    {
+      "epoch": 0.8891538312818199,
+      "grad_norm": 0.03374082286006838,
+      "learning_rate": 1.238811313999284e-06,
+      "loss": 0.0,
+      "step": 13797
+    },
+    {
+      "epoch": 0.8892182767287491,
+      "grad_norm": 0.0031583981479104645,
+      "learning_rate": 1.2380952380952382e-06,
+      "loss": 0.0,
+      "step": 13798
+    },
+    {
+      "epoch": 0.8892827221756783,
+      "grad_norm": 0.0029201196111990247,
+      "learning_rate": 1.2373791621911923e-06,
+      "loss": 0.0,
+      "step": 13799
+    },
+    {
+      "epoch": 0.8893471676226075,
+      "grad_norm": 0.0002927000163405691,
+      "learning_rate": 1.2366630862871466e-06,
+      "loss": 0.0,
+      "step": 13800
+    },
+    {
+      "epoch": 0.8894116130695366,
+      "grad_norm": 0.00024142619848897449,
+      "learning_rate": 1.2359470103831007e-06,
+      "loss": 0.0,
+      "step": 13801
+    },
+    {
+      "epoch": 0.8894760585164658,
+      "grad_norm": 0.7436967061286708,
+      "learning_rate": 1.235230934479055e-06,
+      "loss": 0.0079,
+      "step": 13802
+    },
+    {
+      "epoch": 0.889540503963395,
+      "grad_norm": 0.0003735531843330027,
+      "learning_rate": 1.234514858575009e-06,
+      "loss": 0.0,
+      "step": 13803
+    },
+    {
+      "epoch": 0.8896049494103242,
+      "grad_norm": 0.004015970444965455,
+      "learning_rate": 1.2337987826709634e-06,
+      "loss": 0.0,
+      "step": 13804
+    },
+    {
+      "epoch": 0.8896693948572534,
+      "grad_norm": 0.00976236445414446,
+      "learning_rate": 1.2330827067669174e-06,
+      "loss": 0.0001,
+      "step": 13805
+    },
+    {
+      "epoch": 0.8897338403041825,
+      "grad_norm": 0.2136054593067004,
+      "learning_rate": 1.2323666308628715e-06,
+      "loss": 0.0032,
+      "step": 13806
+    },
+    {
+      "epoch": 0.8897982857511116,
+      "grad_norm": 0.0038618296947997015,
+      "learning_rate": 1.2316505549588258e-06,
+      "loss": 0.0,
+      "step": 13807
+    },
+    {
+      "epoch": 0.8898627311980408,
+      "grad_norm": 0.025397306718503647,
+      "learning_rate": 1.23093447905478e-06,
+      "loss": 0.0001,
+      "step": 13808
+    },
+    {
+      "epoch": 0.88992717664497,
+      "grad_norm": 0.009758540280719333,
+      "learning_rate": 1.230218403150734e-06,
+      "loss": 0.0,
+      "step": 13809
+    },
+    {
+      "epoch": 0.8899916220918992,
+      "grad_norm": 0.00016259548961602632,
+      "learning_rate": 1.2295023272466883e-06,
+      "loss": 0.0,
+      "step": 13810
+    },
+    {
+      "epoch": 0.8900560675388284,
+      "grad_norm": 0.0006507345876399534,
+      "learning_rate": 1.2287862513426424e-06,
+      "loss": 0.0,
+      "step": 13811
+    },
+    {
+      "epoch": 0.8901205129857576,
+      "grad_norm": 0.1792454153276647,
+      "learning_rate": 1.2280701754385965e-06,
+      "loss": 0.0003,
+      "step": 13812
+    },
+    {
+      "epoch": 0.8901849584326867,
+      "grad_norm": 0.00969342986091718,
+      "learning_rate": 1.2273540995345508e-06,
+      "loss": 0.0,
+      "step": 13813
+    },
+    {
+      "epoch": 0.8902494038796159,
+      "grad_norm": 0.0008400377772548953,
+      "learning_rate": 1.2266380236305049e-06,
+      "loss": 0.0,
+      "step": 13814
+    },
+    {
+      "epoch": 0.8903138493265451,
+      "grad_norm": 0.00017475221910045208,
+      "learning_rate": 1.2259219477264592e-06,
+      "loss": 0.0,
+      "step": 13815
+    },
+    {
+      "epoch": 0.8903782947734743,
+      "grad_norm": 0.00044403956214980725,
+      "learning_rate": 1.2252058718224132e-06,
+      "loss": 0.0,
+      "step": 13816
+    },
+    {
+      "epoch": 0.8904427402204035,
+      "grad_norm": 0.0698061899771694,
+      "learning_rate": 1.2244897959183673e-06,
+      "loss": 0.0001,
+      "step": 13817
+    },
+    {
+      "epoch": 0.8905071856673326,
+      "grad_norm": 0.025812009727729505,
+      "learning_rate": 1.2237737200143216e-06,
+      "loss": 0.0,
+      "step": 13818
+    },
+    {
+      "epoch": 0.8905716311142617,
+      "grad_norm": 0.36327571249402313,
+      "learning_rate": 1.2230576441102757e-06,
+      "loss": 0.0034,
+      "step": 13819
+    },
+    {
+      "epoch": 0.8906360765611909,
+      "grad_norm": 0.2055305667179693,
+      "learning_rate": 1.22234156820623e-06,
+      "loss": 0.0007,
+      "step": 13820
+    },
+    {
+      "epoch": 0.8907005220081201,
+      "grad_norm": 0.01355179297404914,
+      "learning_rate": 1.2216254923021841e-06,
+      "loss": 0.0,
+      "step": 13821
+    },
+    {
+      "epoch": 0.8907649674550493,
+      "grad_norm": 0.02007575542711268,
+      "learning_rate": 1.2209094163981384e-06,
+      "loss": 0.0002,
+      "step": 13822
+    },
+    {
+      "epoch": 0.8908294129019785,
+      "grad_norm": 0.07990338578217246,
+      "learning_rate": 1.2201933404940925e-06,
+      "loss": 0.0002,
+      "step": 13823
+    },
+    {
+      "epoch": 0.8908938583489077,
+      "grad_norm": 0.0023076275997369897,
+      "learning_rate": 1.2194772645900468e-06,
+      "loss": 0.0,
+      "step": 13824
+    },
+    {
+      "epoch": 0.8909583037958368,
+      "grad_norm": 0.009210954462485512,
+      "learning_rate": 1.2187611886860009e-06,
+      "loss": 0.0,
+      "step": 13825
+    },
+    {
+      "epoch": 0.891022749242766,
+      "grad_norm": 0.004081220907520122,
+      "learning_rate": 1.218045112781955e-06,
+      "loss": 0.0,
+      "step": 13826
+    },
+    {
+      "epoch": 0.8910871946896952,
+      "grad_norm": 0.008870903584263718,
+      "learning_rate": 1.2173290368779093e-06,
+      "loss": 0.0001,
+      "step": 13827
+    },
+    {
+      "epoch": 0.8911516401366244,
+      "grad_norm": 2.2225392181746675,
+      "learning_rate": 1.2166129609738633e-06,
+      "loss": 0.0042,
+      "step": 13828
+    },
+    {
+      "epoch": 0.8912160855835535,
+      "grad_norm": 7.365705023488865e-05,
+      "learning_rate": 1.2158968850698174e-06,
+      "loss": 0.0,
+      "step": 13829
+    },
+    {
+      "epoch": 0.8912805310304827,
+      "grad_norm": 8.679196143845854e-05,
+      "learning_rate": 1.2151808091657717e-06,
+      "loss": 0.0,
+      "step": 13830
+    },
+    {
+      "epoch": 0.8913449764774118,
+      "grad_norm": 0.0025323594558665664,
+      "learning_rate": 1.2144647332617258e-06,
+      "loss": 0.0,
+      "step": 13831
+    },
+    {
+      "epoch": 0.891409421924341,
+      "grad_norm": 0.12459235478699489,
+      "learning_rate": 1.21374865735768e-06,
+      "loss": 0.0001,
+      "step": 13832
+    },
+    {
+      "epoch": 0.8914738673712702,
+      "grad_norm": 0.011818195017482032,
+      "learning_rate": 1.2130325814536342e-06,
+      "loss": 0.0,
+      "step": 13833
+    },
+    {
+      "epoch": 0.8915383128181994,
+      "grad_norm": 0.14417296540437205,
+      "learning_rate": 1.2123165055495883e-06,
+      "loss": 0.0003,
+      "step": 13834
+    },
+    {
+      "epoch": 0.8916027582651286,
+      "grad_norm": 0.010910567690735996,
+      "learning_rate": 1.2116004296455424e-06,
+      "loss": 0.0001,
+      "step": 13835
+    },
+    {
+      "epoch": 0.8916672037120578,
+      "grad_norm": 7.052803181361057e-05,
+      "learning_rate": 1.2108843537414967e-06,
+      "loss": 0.0,
+      "step": 13836
+    },
+    {
+      "epoch": 0.891731649158987,
+      "grad_norm": 0.17583592412288143,
+      "learning_rate": 1.2101682778374508e-06,
+      "loss": 0.0003,
+      "step": 13837
+    },
+    {
+      "epoch": 0.8917960946059161,
+      "grad_norm": 0.023835514292793722,
+      "learning_rate": 1.209452201933405e-06,
+      "loss": 0.0,
+      "step": 13838
+    },
+    {
+      "epoch": 0.8918605400528453,
+      "grad_norm": 0.0015865956615176357,
+      "learning_rate": 1.2087361260293592e-06,
+      "loss": 0.0,
+      "step": 13839
+    },
+    {
+      "epoch": 0.8919249854997744,
+      "grad_norm": 0.2847122236501794,
+      "learning_rate": 1.2080200501253135e-06,
+      "loss": 0.0024,
+      "step": 13840
+    },
+    {
+      "epoch": 0.8919894309467036,
+      "grad_norm": 0.00039617673521535684,
+      "learning_rate": 1.2073039742212675e-06,
+      "loss": 0.0,
+      "step": 13841
+    },
+    {
+      "epoch": 0.8920538763936328,
+      "grad_norm": 0.0015060300582539626,
+      "learning_rate": 1.2065878983172218e-06,
+      "loss": 0.0,
+      "step": 13842
+    },
+    {
+      "epoch": 0.892118321840562,
+      "grad_norm": 7.080614102933089e-05,
+      "learning_rate": 1.205871822413176e-06,
+      "loss": 0.0,
+      "step": 13843
+    },
+    {
+      "epoch": 0.8921827672874911,
+      "grad_norm": 0.087443366093727,
+      "learning_rate": 1.20515574650913e-06,
+      "loss": 0.0004,
+      "step": 13844
+    },
+    {
+      "epoch": 0.8922472127344203,
+      "grad_norm": 0.0008844615606733612,
+      "learning_rate": 1.2044396706050843e-06,
+      "loss": 0.0,
+      "step": 13845
+    },
+    {
+      "epoch": 0.8923116581813495,
+      "grad_norm": 0.012319154787187002,
+      "learning_rate": 1.2037235947010384e-06,
+      "loss": 0.0,
+      "step": 13846
+    },
+    {
+      "epoch": 0.8923761036282787,
+      "grad_norm": 0.010120784738954119,
+      "learning_rate": 1.2030075187969925e-06,
+      "loss": 0.0,
+      "step": 13847
+    },
+    {
+      "epoch": 0.8924405490752079,
+      "grad_norm": 0.015798697562626713,
+      "learning_rate": 1.2022914428929468e-06,
+      "loss": 0.0001,
+      "step": 13848
+    },
+    {
+      "epoch": 0.892504994522137,
+      "grad_norm": 0.0012742288041931434,
+      "learning_rate": 1.2015753669889009e-06,
+      "loss": 0.0,
+      "step": 13849
+    },
+    {
+      "epoch": 0.8925694399690662,
+      "grad_norm": 0.020043164353019335,
+      "learning_rate": 1.2008592910848552e-06,
+      "loss": 0.0,
+      "step": 13850
+    },
+    {
+      "epoch": 0.8926338854159953,
+      "grad_norm": 0.03441415644112191,
+      "learning_rate": 1.2001432151808093e-06,
+      "loss": 0.0,
+      "step": 13851
+    },
+    {
+      "epoch": 0.8926983308629245,
+      "grad_norm": 0.00010035991690649003,
+      "learning_rate": 1.1994271392767633e-06,
+      "loss": 0.0,
+      "step": 13852
+    },
+    {
+      "epoch": 0.8927627763098537,
+      "grad_norm": 0.0003660068627871594,
+      "learning_rate": 1.1987110633727176e-06,
+      "loss": 0.0,
+      "step": 13853
+    },
+    {
+      "epoch": 0.8928272217567829,
+      "grad_norm": 0.013529318495661974,
+      "learning_rate": 1.1979949874686717e-06,
+      "loss": 0.0,
+      "step": 13854
+    },
+    {
+      "epoch": 0.892891667203712,
+      "grad_norm": 0.00016863076810777927,
+      "learning_rate": 1.1972789115646258e-06,
+      "loss": 0.0,
+      "step": 13855
+    },
+    {
+      "epoch": 0.8929561126506412,
+      "grad_norm": 0.004578985770439536,
+      "learning_rate": 1.1965628356605801e-06,
+      "loss": 0.0,
+      "step": 13856
+    },
+    {
+      "epoch": 0.8930205580975704,
+      "grad_norm": 0.6975465310207425,
+      "learning_rate": 1.1958467597565342e-06,
+      "loss": 0.0011,
+      "step": 13857
+    },
+    {
+      "epoch": 0.8930850035444996,
+      "grad_norm": 0.000828732726699813,
+      "learning_rate": 1.1951306838524885e-06,
+      "loss": 0.0,
+      "step": 13858
+    },
+    {
+      "epoch": 0.8931494489914288,
+      "grad_norm": 0.0008080449898315412,
+      "learning_rate": 1.1944146079484426e-06,
+      "loss": 0.0,
+      "step": 13859
+    },
+    {
+      "epoch": 0.893213894438358,
+      "grad_norm": 0.00027180681347506474,
+      "learning_rate": 1.1936985320443969e-06,
+      "loss": 0.0,
+      "step": 13860
+    },
+    {
+      "epoch": 0.8932783398852872,
+      "grad_norm": 0.018442260354453323,
+      "learning_rate": 1.192982456140351e-06,
+      "loss": 0.0001,
+      "step": 13861
+    },
+    {
+      "epoch": 0.8933427853322162,
+      "grad_norm": 0.4737637997431658,
+      "learning_rate": 1.1922663802363053e-06,
+      "loss": 0.0008,
+      "step": 13862
+    },
+    {
+      "epoch": 0.8934072307791454,
+      "grad_norm": 0.0038302988254236064,
+      "learning_rate": 1.1915503043322594e-06,
+      "loss": 0.0,
+      "step": 13863
+    },
+    {
+      "epoch": 0.8934716762260746,
+      "grad_norm": 0.0001033771169610538,
+      "learning_rate": 1.1908342284282134e-06,
+      "loss": 0.0,
+      "step": 13864
+    },
+    {
+      "epoch": 0.8935361216730038,
+      "grad_norm": 0.00024783354701006966,
+      "learning_rate": 1.1901181525241677e-06,
+      "loss": 0.0,
+      "step": 13865
+    },
+    {
+      "epoch": 0.893600567119933,
+      "grad_norm": 0.004368677639749437,
+      "learning_rate": 1.1894020766201218e-06,
+      "loss": 0.0,
+      "step": 13866
+    },
+    {
+      "epoch": 0.8936650125668621,
+      "grad_norm": 0.0018244756781653027,
+      "learning_rate": 1.188686000716076e-06,
+      "loss": 0.0,
+      "step": 13867
+    },
+    {
+      "epoch": 0.8937294580137913,
+      "grad_norm": 0.00023026815644767366,
+      "learning_rate": 1.1879699248120302e-06,
+      "loss": 0.0,
+      "step": 13868
+    },
+    {
+      "epoch": 0.8937939034607205,
+      "grad_norm": 0.003097842611027994,
+      "learning_rate": 1.1872538489079843e-06,
+      "loss": 0.0,
+      "step": 13869
+    },
+    {
+      "epoch": 0.8938583489076497,
+      "grad_norm": 0.03308828563419499,
+      "learning_rate": 1.1865377730039384e-06,
+      "loss": 0.0001,
+      "step": 13870
+    },
+    {
+      "epoch": 0.8939227943545789,
+      "grad_norm": 0.0006239056166978975,
+      "learning_rate": 1.1858216970998927e-06,
+      "loss": 0.0,
+      "step": 13871
+    },
+    {
+      "epoch": 0.8939872398015081,
+      "grad_norm": 1.2440699958126985e-05,
+      "learning_rate": 1.1851056211958468e-06,
+      "loss": 0.0,
+      "step": 13872
+    },
+    {
+      "epoch": 0.8940516852484371,
+      "grad_norm": 5.976930241618748e-05,
+      "learning_rate": 1.184389545291801e-06,
+      "loss": 0.0,
+      "step": 13873
+    },
+    {
+      "epoch": 0.8941161306953663,
+      "grad_norm": 0.0028886406204160628,
+      "learning_rate": 1.1836734693877552e-06,
+      "loss": 0.0,
+      "step": 13874
+    },
+    {
+      "epoch": 0.8941805761422955,
+      "grad_norm": 0.004454665917691752,
+      "learning_rate": 1.1829573934837095e-06,
+      "loss": 0.0,
+      "step": 13875
+    },
+    {
+      "epoch": 0.8942450215892247,
+      "grad_norm": 0.5158161040383212,
+      "learning_rate": 1.1822413175796635e-06,
+      "loss": 0.0029,
+      "step": 13876
+    },
+    {
+      "epoch": 0.8943094670361539,
+      "grad_norm": 0.02061518097300911,
+      "learning_rate": 1.1815252416756178e-06,
+      "loss": 0.0015,
+      "step": 13877
+    },
+    {
+      "epoch": 0.8943739124830831,
+      "grad_norm": 1.7711709279277926e-05,
+      "learning_rate": 1.180809165771572e-06,
+      "loss": 0.0,
+      "step": 13878
+    },
+    {
+      "epoch": 0.8944383579300123,
+      "grad_norm": 0.18675871185998966,
+      "learning_rate": 1.180093089867526e-06,
+      "loss": 0.0006,
+      "step": 13879
+    },
+    {
+      "epoch": 0.8945028033769414,
+      "grad_norm": 0.00019593830987744068,
+      "learning_rate": 1.1793770139634803e-06,
+      "loss": 0.0,
+      "step": 13880
+    },
+    {
+      "epoch": 0.8945672488238706,
+      "grad_norm": 0.00010785606641666989,
+      "learning_rate": 1.1786609380594344e-06,
+      "loss": 0.0,
+      "step": 13881
+    },
+    {
+      "epoch": 0.8946316942707998,
+      "grad_norm": 0.00014090590342756955,
+      "learning_rate": 1.1779448621553885e-06,
+      "loss": 0.0,
+      "step": 13882
+    },
+    {
+      "epoch": 0.894696139717729,
+      "grad_norm": 0.015730361982695604,
+      "learning_rate": 1.1772287862513428e-06,
+      "loss": 0.0001,
+      "step": 13883
+    },
+    {
+      "epoch": 0.8947605851646581,
+      "grad_norm": 0.165150147923511,
+      "learning_rate": 1.1765127103472969e-06,
+      "loss": 0.0006,
+      "step": 13884
+    },
+    {
+      "epoch": 0.8948250306115872,
+      "grad_norm": 0.001520522969766945,
+      "learning_rate": 1.175796634443251e-06,
+      "loss": 0.0,
+      "step": 13885
+    },
+    {
+      "epoch": 0.8948894760585164,
+      "grad_norm": 0.03972055794668295,
+      "learning_rate": 1.1750805585392053e-06,
+      "loss": 0.0001,
+      "step": 13886
+    },
+    {
+      "epoch": 0.8949539215054456,
+      "grad_norm": 0.0003771624240822517,
+      "learning_rate": 1.1743644826351593e-06,
+      "loss": 0.0,
+      "step": 13887
+    },
+    {
+      "epoch": 0.8950183669523748,
+      "grad_norm": 0.11441051028315816,
+      "learning_rate": 1.1736484067311136e-06,
+      "loss": 0.0001,
+      "step": 13888
+    },
+    {
+      "epoch": 0.895082812399304,
+      "grad_norm": 0.007310869051213783,
+      "learning_rate": 1.1729323308270677e-06,
+      "loss": 0.0,
+      "step": 13889
+    },
+    {
+      "epoch": 0.8951472578462332,
+      "grad_norm": 0.0001329414092162519,
+      "learning_rate": 1.1722162549230218e-06,
+      "loss": 0.0,
+      "step": 13890
+    },
+    {
+      "epoch": 0.8952117032931624,
+      "grad_norm": 0.0014470808931624586,
+      "learning_rate": 1.1715001790189761e-06,
+      "loss": 0.0,
+      "step": 13891
+    },
+    {
+      "epoch": 0.8952761487400915,
+      "grad_norm": 0.0004422742882779398,
+      "learning_rate": 1.1707841031149302e-06,
+      "loss": 0.0,
+      "step": 13892
+    },
+    {
+      "epoch": 0.8953405941870207,
+      "grad_norm": 0.0030581721309221434,
+      "learning_rate": 1.1700680272108845e-06,
+      "loss": 0.0,
+      "step": 13893
+    },
+    {
+      "epoch": 0.8954050396339499,
+      "grad_norm": 2.8451644614638217e-05,
+      "learning_rate": 1.1693519513068386e-06,
+      "loss": 0.0,
+      "step": 13894
+    },
+    {
+      "epoch": 0.8954694850808791,
+      "grad_norm": 0.0007229926952766978,
+      "learning_rate": 1.1686358754027929e-06,
+      "loss": 0.0,
+      "step": 13895
+    },
+    {
+      "epoch": 0.8955339305278082,
+      "grad_norm": 0.0029705727247215812,
+      "learning_rate": 1.167919799498747e-06,
+      "loss": 0.0,
+      "step": 13896
+    },
+    {
+      "epoch": 0.8955983759747373,
+      "grad_norm": 0.019346182931325447,
+      "learning_rate": 1.1672037235947013e-06,
+      "loss": 0.0001,
+      "step": 13897
+    },
+    {
+      "epoch": 0.8956628214216665,
+      "grad_norm": 0.0002895578797201431,
+      "learning_rate": 1.1664876476906554e-06,
+      "loss": 0.0,
+      "step": 13898
+    },
+    {
+      "epoch": 0.8957272668685957,
+      "grad_norm": 0.0006637398536619145,
+      "learning_rate": 1.1657715717866095e-06,
+      "loss": 0.0,
+      "step": 13899
+    },
+    {
+      "epoch": 0.8957917123155249,
+      "grad_norm": 0.00019846648889674596,
+      "learning_rate": 1.1650554958825638e-06,
+      "loss": 0.0,
+      "step": 13900
+    },
+    {
+      "epoch": 0.8958561577624541,
+      "grad_norm": 0.00031634464297703157,
+      "learning_rate": 1.1643394199785178e-06,
+      "loss": 0.0,
+      "step": 13901
+    },
+    {
+      "epoch": 0.8959206032093833,
+      "grad_norm": 0.004311220159345927,
+      "learning_rate": 1.163623344074472e-06,
+      "loss": 0.0,
+      "step": 13902
+    },
+    {
+      "epoch": 0.8959850486563125,
+      "grad_norm": 1.0598536802465255,
+      "learning_rate": 1.1629072681704262e-06,
+      "loss": 0.0031,
+      "step": 13903
+    },
+    {
+      "epoch": 0.8960494941032416,
+      "grad_norm": 0.0008670161043206829,
+      "learning_rate": 1.1621911922663803e-06,
+      "loss": 0.0,
+      "step": 13904
+    },
+    {
+      "epoch": 0.8961139395501708,
+      "grad_norm": 0.0011511655880350254,
+      "learning_rate": 1.1614751163623344e-06,
+      "loss": 0.0,
+      "step": 13905
+    },
+    {
+      "epoch": 0.8961783849971,
+      "grad_norm": 0.003973540317480649,
+      "learning_rate": 1.1607590404582887e-06,
+      "loss": 0.0,
+      "step": 13906
+    },
+    {
+      "epoch": 0.8962428304440291,
+      "grad_norm": 0.14830795080849515,
+      "learning_rate": 1.1600429645542428e-06,
+      "loss": 0.0018,
+      "step": 13907
+    },
+    {
+      "epoch": 0.8963072758909583,
+      "grad_norm": 0.0007323064595173414,
+      "learning_rate": 1.1593268886501969e-06,
+      "loss": 0.0,
+      "step": 13908
+    },
+    {
+      "epoch": 0.8963717213378875,
+      "grad_norm": 0.00023526852555263698,
+      "learning_rate": 1.1586108127461512e-06,
+      "loss": 0.0,
+      "step": 13909
+    },
+    {
+      "epoch": 0.8964361667848166,
+      "grad_norm": 0.002325935800464206,
+      "learning_rate": 1.1578947368421053e-06,
+      "loss": 0.0,
+      "step": 13910
+    },
+    {
+      "epoch": 0.8965006122317458,
+      "grad_norm": 0.4110362286054091,
+      "learning_rate": 1.1571786609380596e-06,
+      "loss": 0.0021,
+      "step": 13911
+    },
+    {
+      "epoch": 0.896565057678675,
+      "grad_norm": 0.021074825949622938,
+      "learning_rate": 1.1564625850340136e-06,
+      "loss": 0.0,
+      "step": 13912
+    },
+    {
+      "epoch": 0.8966295031256042,
+      "grad_norm": 0.0045007280964542195,
+      "learning_rate": 1.155746509129968e-06,
+      "loss": 0.0,
+      "step": 13913
+    },
+    {
+      "epoch": 0.8966939485725334,
+      "grad_norm": 0.0007198015241596646,
+      "learning_rate": 1.155030433225922e-06,
+      "loss": 0.0,
+      "step": 13914
+    },
+    {
+      "epoch": 0.8967583940194626,
+      "grad_norm": 0.019581690119028813,
+      "learning_rate": 1.1543143573218763e-06,
+      "loss": 0.0,
+      "step": 13915
+    },
+    {
+      "epoch": 0.8968228394663917,
+      "grad_norm": 0.00037558322132740917,
+      "learning_rate": 1.1535982814178304e-06,
+      "loss": 0.0,
+      "step": 13916
+    },
+    {
+      "epoch": 0.8968872849133209,
+      "grad_norm": 0.01912842353742819,
+      "learning_rate": 1.1528822055137845e-06,
+      "loss": 0.0001,
+      "step": 13917
+    },
+    {
+      "epoch": 0.89695173036025,
+      "grad_norm": 0.00896517446981582,
+      "learning_rate": 1.1521661296097388e-06,
+      "loss": 0.0,
+      "step": 13918
+    },
+    {
+      "epoch": 0.8970161758071792,
+      "grad_norm": 0.002951599669487491,
+      "learning_rate": 1.1514500537056929e-06,
+      "loss": 0.0,
+      "step": 13919
+    },
+    {
+      "epoch": 0.8970806212541084,
+      "grad_norm": 0.0033833929402165733,
+      "learning_rate": 1.150733977801647e-06,
+      "loss": 0.0,
+      "step": 13920
+    },
+    {
+      "epoch": 0.8971450667010376,
+      "grad_norm": 0.0063744704204274924,
+      "learning_rate": 1.1500179018976013e-06,
+      "loss": 0.0,
+      "step": 13921
+    },
+    {
+      "epoch": 0.8972095121479667,
+      "grad_norm": 0.0013080568903183537,
+      "learning_rate": 1.1493018259935554e-06,
+      "loss": 0.0,
+      "step": 13922
+    },
+    {
+      "epoch": 0.8972739575948959,
+      "grad_norm": 0.005177324218002255,
+      "learning_rate": 1.1485857500895097e-06,
+      "loss": 0.0,
+      "step": 13923
+    },
+    {
+      "epoch": 0.8973384030418251,
+      "grad_norm": 0.00016433341066363132,
+      "learning_rate": 1.1478696741854637e-06,
+      "loss": 0.0,
+      "step": 13924
+    },
+    {
+      "epoch": 0.8974028484887543,
+      "grad_norm": 0.005040815180212994,
+      "learning_rate": 1.1471535982814178e-06,
+      "loss": 0.0,
+      "step": 13925
+    },
+    {
+      "epoch": 0.8974672939356835,
+      "grad_norm": 0.19144006352602805,
+      "learning_rate": 1.1464375223773721e-06,
+      "loss": 0.0004,
+      "step": 13926
+    },
+    {
+      "epoch": 0.8975317393826127,
+      "grad_norm": 0.005144720003540155,
+      "learning_rate": 1.1457214464733262e-06,
+      "loss": 0.0,
+      "step": 13927
+    },
+    {
+      "epoch": 0.8975961848295418,
+      "grad_norm": 0.005620512736102895,
+      "learning_rate": 1.1450053705692803e-06,
+      "loss": 0.0,
+      "step": 13928
+    },
+    {
+      "epoch": 0.8976606302764709,
+      "grad_norm": 0.02314239599743946,
+      "learning_rate": 1.1442892946652346e-06,
+      "loss": 0.0001,
+      "step": 13929
+    },
+    {
+      "epoch": 0.8977250757234001,
+      "grad_norm": 3.417519190465539e-05,
+      "learning_rate": 1.1435732187611887e-06,
+      "loss": 0.0,
+      "step": 13930
+    },
+    {
+      "epoch": 0.8977895211703293,
+      "grad_norm": 0.15925896314589388,
+      "learning_rate": 1.142857142857143e-06,
+      "loss": 0.0025,
+      "step": 13931
+    },
+    {
+      "epoch": 0.8978539666172585,
+      "grad_norm": 0.002229307091129153,
+      "learning_rate": 1.142141066953097e-06,
+      "loss": 0.0,
+      "step": 13932
+    },
+    {
+      "epoch": 0.8979184120641877,
+      "grad_norm": 0.0027272175086743445,
+      "learning_rate": 1.1414249910490514e-06,
+      "loss": 0.0,
+      "step": 13933
+    },
+    {
+      "epoch": 0.8979828575111168,
+      "grad_norm": 0.03168294190225496,
+      "learning_rate": 1.1407089151450055e-06,
+      "loss": 0.0,
+      "step": 13934
+    },
+    {
+      "epoch": 0.898047302958046,
+      "grad_norm": 0.0009569713106729391,
+      "learning_rate": 1.1399928392409598e-06,
+      "loss": 0.0,
+      "step": 13935
+    },
+    {
+      "epoch": 0.8981117484049752,
+      "grad_norm": 0.0008837372743535399,
+      "learning_rate": 1.1392767633369138e-06,
+      "loss": 0.0,
+      "step": 13936
+    },
+    {
+      "epoch": 0.8981761938519044,
+      "grad_norm": 0.00016719703676270603,
+      "learning_rate": 1.138560687432868e-06,
+      "loss": 0.0,
+      "step": 13937
+    },
+    {
+      "epoch": 0.8982406392988336,
+      "grad_norm": 0.0015464604635017866,
+      "learning_rate": 1.1378446115288222e-06,
+      "loss": 0.0,
+      "step": 13938
+    },
+    {
+      "epoch": 0.8983050847457628,
+      "grad_norm": 0.005463581930813619,
+      "learning_rate": 1.1371285356247763e-06,
+      "loss": 0.0,
+      "step": 13939
+    },
+    {
+      "epoch": 0.8983695301926918,
+      "grad_norm": 0.004126363956869999,
+      "learning_rate": 1.1364124597207304e-06,
+      "loss": 0.0,
+      "step": 13940
+    },
+    {
+      "epoch": 0.898433975639621,
+      "grad_norm": 0.00044599899445592567,
+      "learning_rate": 1.1356963838166847e-06,
+      "loss": 0.0,
+      "step": 13941
+    },
+    {
+      "epoch": 0.8984984210865502,
+      "grad_norm": 0.0006417580259393078,
+      "learning_rate": 1.1349803079126388e-06,
+      "loss": 0.0,
+      "step": 13942
+    },
+    {
+      "epoch": 0.8985628665334794,
+      "grad_norm": 0.0028110095254215103,
+      "learning_rate": 1.1342642320085929e-06,
+      "loss": 0.0,
+      "step": 13943
+    },
+    {
+      "epoch": 0.8986273119804086,
+      "grad_norm": 0.0008243069210426579,
+      "learning_rate": 1.1335481561045472e-06,
+      "loss": 0.0,
+      "step": 13944
+    },
+    {
+      "epoch": 0.8986917574273378,
+      "grad_norm": 0.015608134890901078,
+      "learning_rate": 1.1328320802005013e-06,
+      "loss": 0.0001,
+      "step": 13945
+    },
+    {
+      "epoch": 0.8987562028742669,
+      "grad_norm": 0.00016418725029251164,
+      "learning_rate": 1.1321160042964556e-06,
+      "loss": 0.0,
+      "step": 13946
+    },
+    {
+      "epoch": 0.8988206483211961,
+      "grad_norm": 0.0017409146369892678,
+      "learning_rate": 1.1313999283924096e-06,
+      "loss": 0.0,
+      "step": 13947
+    },
+    {
+      "epoch": 0.8988850937681253,
+      "grad_norm": 5.412282878661023e-05,
+      "learning_rate": 1.130683852488364e-06,
+      "loss": 0.0,
+      "step": 13948
+    },
+    {
+      "epoch": 0.8989495392150545,
+      "grad_norm": 0.0004645559859257255,
+      "learning_rate": 1.129967776584318e-06,
+      "loss": 0.0,
+      "step": 13949
+    },
+    {
+      "epoch": 0.8990139846619837,
+      "grad_norm": 0.002543894375187707,
+      "learning_rate": 1.1292517006802723e-06,
+      "loss": 0.0,
+      "step": 13950
+    },
+    {
+      "epoch": 0.8990784301089128,
+      "grad_norm": 0.00021699801752490995,
+      "learning_rate": 1.1285356247762264e-06,
+      "loss": 0.0,
+      "step": 13951
+    },
+    {
+      "epoch": 0.8991428755558419,
+      "grad_norm": 0.00015121543904727036,
+      "learning_rate": 1.1278195488721805e-06,
+      "loss": 0.0,
+      "step": 13952
+    },
+    {
+      "epoch": 0.8992073210027711,
+      "grad_norm": 0.010239032263425,
+      "learning_rate": 1.1271034729681348e-06,
+      "loss": 0.0,
+      "step": 13953
+    },
+    {
+      "epoch": 0.8992717664497003,
+      "grad_norm": 0.018982114515024993,
+      "learning_rate": 1.1263873970640889e-06,
+      "loss": 0.0,
+      "step": 13954
+    },
+    {
+      "epoch": 0.8993362118966295,
+      "grad_norm": 0.0005564598211045715,
+      "learning_rate": 1.125671321160043e-06,
+      "loss": 0.0,
+      "step": 13955
+    },
+    {
+      "epoch": 0.8994006573435587,
+      "grad_norm": 0.0008152348454253998,
+      "learning_rate": 1.1249552452559973e-06,
+      "loss": 0.0,
+      "step": 13956
+    },
+    {
+      "epoch": 0.8994651027904879,
+      "grad_norm": 0.142068940030953,
+      "learning_rate": 1.1242391693519514e-06,
+      "loss": 0.0002,
+      "step": 13957
+    },
+    {
+      "epoch": 0.899529548237417,
+      "grad_norm": 0.017551017927371366,
+      "learning_rate": 1.1235230934479057e-06,
+      "loss": 0.0001,
+      "step": 13958
+    },
+    {
+      "epoch": 0.8995939936843462,
+      "grad_norm": 0.13314984757949708,
+      "learning_rate": 1.1228070175438598e-06,
+      "loss": 0.0002,
+      "step": 13959
+    },
+    {
+      "epoch": 0.8996584391312754,
+      "grad_norm": 0.003780466996403599,
+      "learning_rate": 1.1220909416398138e-06,
+      "loss": 0.0,
+      "step": 13960
+    },
+    {
+      "epoch": 0.8997228845782046,
+      "grad_norm": 3.8431651826954105e-05,
+      "learning_rate": 1.1213748657357681e-06,
+      "loss": 0.0,
+      "step": 13961
+    },
+    {
+      "epoch": 0.8997873300251337,
+      "grad_norm": 0.0007425784620714796,
+      "learning_rate": 1.1206587898317222e-06,
+      "loss": 0.0,
+      "step": 13962
+    },
+    {
+      "epoch": 0.8998517754720629,
+      "grad_norm": 0.0002841375235135086,
+      "learning_rate": 1.1199427139276763e-06,
+      "loss": 0.0,
+      "step": 13963
+    },
+    {
+      "epoch": 0.899916220918992,
+      "grad_norm": 0.0030655431826020446,
+      "learning_rate": 1.1192266380236306e-06,
+      "loss": 0.0,
+      "step": 13964
+    },
+    {
+      "epoch": 0.8999806663659212,
+      "grad_norm": 0.0006356895919749247,
+      "learning_rate": 1.1185105621195847e-06,
+      "loss": 0.0,
+      "step": 13965
+    },
+    {
+      "epoch": 0.9000451118128504,
+      "grad_norm": 0.020512951337609962,
+      "learning_rate": 1.117794486215539e-06,
+      "loss": 0.0001,
+      "step": 13966
+    },
+    {
+      "epoch": 0.9001095572597796,
+      "grad_norm": 0.04083187946678933,
+      "learning_rate": 1.117078410311493e-06,
+      "loss": 0.0003,
+      "step": 13967
+    },
+    {
+      "epoch": 0.9001740027067088,
+      "grad_norm": 0.0001550509178908707,
+      "learning_rate": 1.1163623344074474e-06,
+      "loss": 0.0,
+      "step": 13968
+    },
+    {
+      "epoch": 0.900238448153638,
+      "grad_norm": 0.0004273088370013827,
+      "learning_rate": 1.1156462585034015e-06,
+      "loss": 0.0,
+      "step": 13969
+    },
+    {
+      "epoch": 0.9003028936005671,
+      "grad_norm": 0.0003448674989961782,
+      "learning_rate": 1.1149301825993558e-06,
+      "loss": 0.0,
+      "step": 13970
+    },
+    {
+      "epoch": 0.9003673390474963,
+      "grad_norm": 0.0008768537809582048,
+      "learning_rate": 1.1142141066953099e-06,
+      "loss": 0.0,
+      "step": 13971
+    },
+    {
+      "epoch": 0.9004317844944255,
+      "grad_norm": 5.592022632896799e-05,
+      "learning_rate": 1.113498030791264e-06,
+      "loss": 0.0,
+      "step": 13972
+    },
+    {
+      "epoch": 0.9004962299413547,
+      "grad_norm": 0.6882594855306482,
+      "learning_rate": 1.1127819548872182e-06,
+      "loss": 0.0027,
+      "step": 13973
+    },
+    {
+      "epoch": 0.9005606753882838,
+      "grad_norm": 0.009598311999210143,
+      "learning_rate": 1.1120658789831723e-06,
+      "loss": 0.0,
+      "step": 13974
+    },
+    {
+      "epoch": 0.900625120835213,
+      "grad_norm": 0.02725082065057163,
+      "learning_rate": 1.1113498030791264e-06,
+      "loss": 0.0,
+      "step": 13975
+    },
+    {
+      "epoch": 0.9006895662821421,
+      "grad_norm": 0.00868479275831414,
+      "learning_rate": 1.1106337271750807e-06,
+      "loss": 0.0,
+      "step": 13976
+    },
+    {
+      "epoch": 0.9007540117290713,
+      "grad_norm": 0.007083911823828566,
+      "learning_rate": 1.1099176512710348e-06,
+      "loss": 0.0001,
+      "step": 13977
+    },
+    {
+      "epoch": 0.9008184571760005,
+      "grad_norm": 0.002803608504870315,
+      "learning_rate": 1.1092015753669889e-06,
+      "loss": 0.0,
+      "step": 13978
+    },
+    {
+      "epoch": 0.9008829026229297,
+      "grad_norm": 0.0006188193145472269,
+      "learning_rate": 1.1084854994629432e-06,
+      "loss": 0.0,
+      "step": 13979
+    },
+    {
+      "epoch": 0.9009473480698589,
+      "grad_norm": 0.024428739904172637,
+      "learning_rate": 1.1077694235588973e-06,
+      "loss": 0.0002,
+      "step": 13980
+    },
+    {
+      "epoch": 0.9010117935167881,
+      "grad_norm": 0.0003124864897644293,
+      "learning_rate": 1.1070533476548514e-06,
+      "loss": 0.0,
+      "step": 13981
+    },
+    {
+      "epoch": 0.9010762389637172,
+      "grad_norm": 0.10956738785973091,
+      "learning_rate": 1.1063372717508057e-06,
+      "loss": 0.0001,
+      "step": 13982
+    },
+    {
+      "epoch": 0.9011406844106464,
+      "grad_norm": 0.02323807997631617,
+      "learning_rate": 1.1056211958467597e-06,
+      "loss": 0.0,
+      "step": 13983
+    },
+    {
+      "epoch": 0.9012051298575756,
+      "grad_norm": 7.911839137377666e-05,
+      "learning_rate": 1.104905119942714e-06,
+      "loss": 0.0,
+      "step": 13984
+    },
+    {
+      "epoch": 0.9012695753045047,
+      "grad_norm": 1.153643374127301e-05,
+      "learning_rate": 1.1041890440386681e-06,
+      "loss": 0.0,
+      "step": 13985
+    },
+    {
+      "epoch": 0.9013340207514339,
+      "grad_norm": 0.03433700359423184,
+      "learning_rate": 1.1034729681346224e-06,
+      "loss": 0.0,
+      "step": 13986
+    },
+    {
+      "epoch": 0.9013984661983631,
+      "grad_norm": 0.0001273740311295764,
+      "learning_rate": 1.1027568922305765e-06,
+      "loss": 0.0,
+      "step": 13987
+    },
+    {
+      "epoch": 0.9014629116452922,
+      "grad_norm": 0.00029643403967102475,
+      "learning_rate": 1.1020408163265308e-06,
+      "loss": 0.0,
+      "step": 13988
+    },
+    {
+      "epoch": 0.9015273570922214,
+      "grad_norm": 0.00167223813672472,
+      "learning_rate": 1.101324740422485e-06,
+      "loss": 0.0,
+      "step": 13989
+    },
+    {
+      "epoch": 0.9015918025391506,
+      "grad_norm": 0.049248185514050316,
+      "learning_rate": 1.100608664518439e-06,
+      "loss": 0.0,
+      "step": 13990
+    },
+    {
+      "epoch": 0.9016562479860798,
+      "grad_norm": 0.002362216149147042,
+      "learning_rate": 1.0998925886143933e-06,
+      "loss": 0.0,
+      "step": 13991
+    },
+    {
+      "epoch": 0.901720693433009,
+      "grad_norm": 0.006003178554491009,
+      "learning_rate": 1.0991765127103474e-06,
+      "loss": 0.0001,
+      "step": 13992
+    },
+    {
+      "epoch": 0.9017851388799382,
+      "grad_norm": 0.22652499118620095,
+      "learning_rate": 1.0984604368063015e-06,
+      "loss": 0.0002,
+      "step": 13993
+    },
+    {
+      "epoch": 0.9018495843268673,
+      "grad_norm": 0.02536765650686301,
+      "learning_rate": 1.0977443609022558e-06,
+      "loss": 0.0002,
+      "step": 13994
+    },
+    {
+      "epoch": 0.9019140297737965,
+      "grad_norm": 0.04242994603591199,
+      "learning_rate": 1.0970282849982098e-06,
+      "loss": 0.0001,
+      "step": 13995
+    },
+    {
+      "epoch": 0.9019784752207256,
+      "grad_norm": 0.007177177779843645,
+      "learning_rate": 1.0963122090941641e-06,
+      "loss": 0.0,
+      "step": 13996
+    },
+    {
+      "epoch": 0.9020429206676548,
+      "grad_norm": 0.180822076514872,
+      "learning_rate": 1.0955961331901182e-06,
+      "loss": 0.002,
+      "step": 13997
+    },
+    {
+      "epoch": 0.902107366114584,
+      "grad_norm": 0.0034121207623937933,
+      "learning_rate": 1.0948800572860723e-06,
+      "loss": 0.0,
+      "step": 13998
+    },
+    {
+      "epoch": 0.9021718115615132,
+      "grad_norm": 0.004040904780187532,
+      "learning_rate": 1.0941639813820266e-06,
+      "loss": 0.0,
+      "step": 13999
+    },
+    {
+      "epoch": 0.9022362570084423,
+      "grad_norm": 0.000310484319187804,
+      "learning_rate": 1.0934479054779807e-06,
+      "loss": 0.0,
+      "step": 14000
+    },
+    {
+      "epoch": 0.9023007024553715,
+      "grad_norm": 0.01932705216887116,
+      "learning_rate": 1.0927318295739348e-06,
+      "loss": 0.0001,
+      "step": 14001
+    },
+    {
+      "epoch": 0.9023651479023007,
+      "grad_norm": 0.0008629699703404948,
+      "learning_rate": 1.092015753669889e-06,
+      "loss": 0.0,
+      "step": 14002
+    },
+    {
+      "epoch": 0.9024295933492299,
+      "grad_norm": 0.0003779540630065977,
+      "learning_rate": 1.0912996777658432e-06,
+      "loss": 0.0,
+      "step": 14003
+    },
+    {
+      "epoch": 0.9024940387961591,
+      "grad_norm": 0.004033788037230716,
+      "learning_rate": 1.0905836018617975e-06,
+      "loss": 0.0,
+      "step": 14004
+    },
+    {
+      "epoch": 0.9025584842430883,
+      "grad_norm": 0.00012087015154678254,
+      "learning_rate": 1.0898675259577516e-06,
+      "loss": 0.0,
+      "step": 14005
+    },
+    {
+      "epoch": 0.9026229296900175,
+      "grad_norm": 0.0031684532813541473,
+      "learning_rate": 1.0891514500537059e-06,
+      "loss": 0.0,
+      "step": 14006
+    },
+    {
+      "epoch": 0.9026873751369465,
+      "grad_norm": 0.7884746287853464,
+      "learning_rate": 1.08843537414966e-06,
+      "loss": 0.0014,
+      "step": 14007
+    },
+    {
+      "epoch": 0.9027518205838757,
+      "grad_norm": 2.669712599619591e-05,
+      "learning_rate": 1.0877192982456142e-06,
+      "loss": 0.0,
+      "step": 14008
+    },
+    {
+      "epoch": 0.9028162660308049,
+      "grad_norm": 0.0009535648759989387,
+      "learning_rate": 1.0870032223415683e-06,
+      "loss": 0.0,
+      "step": 14009
+    },
+    {
+      "epoch": 0.9028807114777341,
+      "grad_norm": 0.002809913133161104,
+      "learning_rate": 1.0862871464375224e-06,
+      "loss": 0.0,
+      "step": 14010
+    },
+    {
+      "epoch": 0.9029451569246633,
+      "grad_norm": 0.0012233205019003342,
+      "learning_rate": 1.0855710705334767e-06,
+      "loss": 0.0,
+      "step": 14011
+    },
+    {
+      "epoch": 0.9030096023715924,
+      "grad_norm": 0.005354862179131867,
+      "learning_rate": 1.0848549946294308e-06,
+      "loss": 0.0,
+      "step": 14012
+    },
+    {
+      "epoch": 0.9030740478185216,
+      "grad_norm": 0.0001383134804233665,
+      "learning_rate": 1.0841389187253849e-06,
+      "loss": 0.0,
+      "step": 14013
+    },
+    {
+      "epoch": 0.9031384932654508,
+      "grad_norm": 0.0012410408689786621,
+      "learning_rate": 1.0834228428213392e-06,
+      "loss": 0.0,
+      "step": 14014
+    },
+    {
+      "epoch": 0.90320293871238,
+      "grad_norm": 0.00015145828171334654,
+      "learning_rate": 1.0827067669172933e-06,
+      "loss": 0.0,
+      "step": 14015
+    },
+    {
+      "epoch": 0.9032673841593092,
+      "grad_norm": 0.0029233848246608997,
+      "learning_rate": 1.0819906910132474e-06,
+      "loss": 0.0,
+      "step": 14016
+    },
+    {
+      "epoch": 0.9033318296062384,
+      "grad_norm": 6.762058916841936e-06,
+      "learning_rate": 1.0812746151092017e-06,
+      "loss": 0.0,
+      "step": 14017
+    },
+    {
+      "epoch": 0.9033962750531674,
+      "grad_norm": 0.14695861977500713,
+      "learning_rate": 1.0805585392051558e-06,
+      "loss": 0.0018,
+      "step": 14018
+    },
+    {
+      "epoch": 0.9034607205000966,
+      "grad_norm": 0.0027973229930890954,
+      "learning_rate": 1.07984246330111e-06,
+      "loss": 0.0,
+      "step": 14019
+    },
+    {
+      "epoch": 0.9035251659470258,
+      "grad_norm": 0.004426821608431362,
+      "learning_rate": 1.0791263873970641e-06,
+      "loss": 0.0,
+      "step": 14020
+    },
+    {
+      "epoch": 0.903589611393955,
+      "grad_norm": 0.9546233422531074,
+      "learning_rate": 1.0784103114930184e-06,
+      "loss": 0.0039,
+      "step": 14021
+    },
+    {
+      "epoch": 0.9036540568408842,
+      "grad_norm": 0.0015999861394946403,
+      "learning_rate": 1.0776942355889725e-06,
+      "loss": 0.0,
+      "step": 14022
+    },
+    {
+      "epoch": 0.9037185022878134,
+      "grad_norm": 0.8040267676053892,
+      "learning_rate": 1.0769781596849268e-06,
+      "loss": 0.0073,
+      "step": 14023
+    },
+    {
+      "epoch": 0.9037829477347425,
+      "grad_norm": 0.1496658198797121,
+      "learning_rate": 1.076262083780881e-06,
+      "loss": 0.0009,
+      "step": 14024
+    },
+    {
+      "epoch": 0.9038473931816717,
+      "grad_norm": 0.00028597161914835925,
+      "learning_rate": 1.075546007876835e-06,
+      "loss": 0.0,
+      "step": 14025
+    },
+    {
+      "epoch": 0.9039118386286009,
+      "grad_norm": 0.0010954895000369828,
+      "learning_rate": 1.0748299319727893e-06,
+      "loss": 0.0,
+      "step": 14026
+    },
+    {
+      "epoch": 0.9039762840755301,
+      "grad_norm": 0.00033146946288850447,
+      "learning_rate": 1.0741138560687434e-06,
+      "loss": 0.0,
+      "step": 14027
+    },
+    {
+      "epoch": 0.9040407295224593,
+      "grad_norm": 0.0071175550409161665,
+      "learning_rate": 1.0733977801646975e-06,
+      "loss": 0.0,
+      "step": 14028
+    },
+    {
+      "epoch": 0.9041051749693884,
+      "grad_norm": 0.01369357303133858,
+      "learning_rate": 1.0726817042606518e-06,
+      "loss": 0.0,
+      "step": 14029
+    },
+    {
+      "epoch": 0.9041696204163175,
+      "grad_norm": 0.009619704462363344,
+      "learning_rate": 1.0719656283566059e-06,
+      "loss": 0.0,
+      "step": 14030
+    },
+    {
+      "epoch": 0.9042340658632467,
+      "grad_norm": 0.0007094879850284002,
+      "learning_rate": 1.0712495524525602e-06,
+      "loss": 0.0,
+      "step": 14031
+    },
+    {
+      "epoch": 0.9042985113101759,
+      "grad_norm": 0.013750855588743177,
+      "learning_rate": 1.0705334765485142e-06,
+      "loss": 0.0001,
+      "step": 14032
+    },
+    {
+      "epoch": 0.9043629567571051,
+      "grad_norm": 0.00027600823917411867,
+      "learning_rate": 1.0698174006444683e-06,
+      "loss": 0.0,
+      "step": 14033
+    },
+    {
+      "epoch": 0.9044274022040343,
+      "grad_norm": 0.01503473339922653,
+      "learning_rate": 1.0691013247404226e-06,
+      "loss": 0.0,
+      "step": 14034
+    },
+    {
+      "epoch": 0.9044918476509635,
+      "grad_norm": 2.9273438306325014e-05,
+      "learning_rate": 1.0683852488363767e-06,
+      "loss": 0.0,
+      "step": 14035
+    },
+    {
+      "epoch": 0.9045562930978926,
+      "grad_norm": 0.001225609437622927,
+      "learning_rate": 1.0676691729323308e-06,
+      "loss": 0.0,
+      "step": 14036
+    },
+    {
+      "epoch": 0.9046207385448218,
+      "grad_norm": 0.000198674450168846,
+      "learning_rate": 1.066953097028285e-06,
+      "loss": 0.0,
+      "step": 14037
+    },
+    {
+      "epoch": 0.904685183991751,
+      "grad_norm": 0.0008732649309423083,
+      "learning_rate": 1.0662370211242392e-06,
+      "loss": 0.0,
+      "step": 14038
+    },
+    {
+      "epoch": 0.9047496294386802,
+      "grad_norm": 8.002756539824641e-05,
+      "learning_rate": 1.0655209452201935e-06,
+      "loss": 0.0,
+      "step": 14039
+    },
+    {
+      "epoch": 0.9048140748856093,
+      "grad_norm": 0.01401901919825385,
+      "learning_rate": 1.0648048693161476e-06,
+      "loss": 0.0,
+      "step": 14040
+    },
+    {
+      "epoch": 0.9048785203325385,
+      "grad_norm": 0.014786355131894208,
+      "learning_rate": 1.0640887934121019e-06,
+      "loss": 0.0,
+      "step": 14041
+    },
+    {
+      "epoch": 0.9049429657794676,
+      "grad_norm": 0.004576884247605246,
+      "learning_rate": 1.063372717508056e-06,
+      "loss": 0.0,
+      "step": 14042
+    },
+    {
+      "epoch": 0.9050074112263968,
+      "grad_norm": 0.05576671370867512,
+      "learning_rate": 1.0626566416040103e-06,
+      "loss": 0.0016,
+      "step": 14043
+    },
+    {
+      "epoch": 0.905071856673326,
+      "grad_norm": 0.49583190856322845,
+      "learning_rate": 1.0619405656999643e-06,
+      "loss": 0.0008,
+      "step": 14044
+    },
+    {
+      "epoch": 0.9051363021202552,
+      "grad_norm": 0.003267573240087532,
+      "learning_rate": 1.0612244897959184e-06,
+      "loss": 0.0,
+      "step": 14045
+    },
+    {
+      "epoch": 0.9052007475671844,
+      "grad_norm": 0.000656877303059541,
+      "learning_rate": 1.0605084138918727e-06,
+      "loss": 0.0,
+      "step": 14046
+    },
+    {
+      "epoch": 0.9052651930141136,
+      "grad_norm": 0.39314117668237164,
+      "learning_rate": 1.0597923379878268e-06,
+      "loss": 0.0009,
+      "step": 14047
+    },
+    {
+      "epoch": 0.9053296384610428,
+      "grad_norm": 0.000852336283270037,
+      "learning_rate": 1.059076262083781e-06,
+      "loss": 0.0,
+      "step": 14048
+    },
+    {
+      "epoch": 0.9053940839079719,
+      "grad_norm": 0.00025226461633294403,
+      "learning_rate": 1.0583601861797352e-06,
+      "loss": 0.0,
+      "step": 14049
+    },
+    {
+      "epoch": 0.9054585293549011,
+      "grad_norm": 9.980164147839169e-05,
+      "learning_rate": 1.0576441102756893e-06,
+      "loss": 0.0,
+      "step": 14050
+    },
+    {
+      "epoch": 0.9055229748018303,
+      "grad_norm": 0.0008063489059367108,
+      "learning_rate": 1.0569280343716434e-06,
+      "loss": 0.0,
+      "step": 14051
+    },
+    {
+      "epoch": 0.9055874202487594,
+      "grad_norm": 0.006412100452888351,
+      "learning_rate": 1.0562119584675977e-06,
+      "loss": 0.0001,
+      "step": 14052
+    },
+    {
+      "epoch": 0.9056518656956886,
+      "grad_norm": 0.0010209184892873981,
+      "learning_rate": 1.0554958825635518e-06,
+      "loss": 0.0,
+      "step": 14053
+    },
+    {
+      "epoch": 0.9057163111426177,
+      "grad_norm": 1.1038661163874564,
+      "learning_rate": 1.0547798066595058e-06,
+      "loss": 0.0033,
+      "step": 14054
+    },
+    {
+      "epoch": 0.9057807565895469,
+      "grad_norm": 0.005365736820451497,
+      "learning_rate": 1.0540637307554601e-06,
+      "loss": 0.0,
+      "step": 14055
+    },
+    {
+      "epoch": 0.9058452020364761,
+      "grad_norm": 0.0005929740637682495,
+      "learning_rate": 1.0533476548514142e-06,
+      "loss": 0.0,
+      "step": 14056
+    },
+    {
+      "epoch": 0.9059096474834053,
+      "grad_norm": 0.0018838975597551804,
+      "learning_rate": 1.0526315789473685e-06,
+      "loss": 0.0,
+      "step": 14057
+    },
+    {
+      "epoch": 0.9059740929303345,
+      "grad_norm": 0.00012933569279299136,
+      "learning_rate": 1.0519155030433226e-06,
+      "loss": 0.0,
+      "step": 14058
+    },
+    {
+      "epoch": 0.9060385383772637,
+      "grad_norm": 0.0006639789205916543,
+      "learning_rate": 1.051199427139277e-06,
+      "loss": 0.0,
+      "step": 14059
+    },
+    {
+      "epoch": 0.9061029838241929,
+      "grad_norm": 0.0002403311090842733,
+      "learning_rate": 1.050483351235231e-06,
+      "loss": 0.0,
+      "step": 14060
+    },
+    {
+      "epoch": 0.906167429271122,
+      "grad_norm": 0.0005676848754295945,
+      "learning_rate": 1.0497672753311853e-06,
+      "loss": 0.0,
+      "step": 14061
+    },
+    {
+      "epoch": 0.9062318747180512,
+      "grad_norm": 0.0001549777960638785,
+      "learning_rate": 1.0490511994271394e-06,
+      "loss": 0.0,
+      "step": 14062
+    },
+    {
+      "epoch": 0.9062963201649803,
+      "grad_norm": 0.008584136048833318,
+      "learning_rate": 1.0483351235230935e-06,
+      "loss": 0.0001,
+      "step": 14063
+    },
+    {
+      "epoch": 0.9063607656119095,
+      "grad_norm": 0.05943516934136947,
+      "learning_rate": 1.0476190476190478e-06,
+      "loss": 0.0001,
+      "step": 14064
+    },
+    {
+      "epoch": 0.9064252110588387,
+      "grad_norm": 0.003992473211187499,
+      "learning_rate": 1.0469029717150019e-06,
+      "loss": 0.0,
+      "step": 14065
+    },
+    {
+      "epoch": 0.9064896565057678,
+      "grad_norm": 0.0011531323035489564,
+      "learning_rate": 1.0461868958109562e-06,
+      "loss": 0.0,
+      "step": 14066
+    },
+    {
+      "epoch": 0.906554101952697,
+      "grad_norm": 0.15580797256069312,
+      "learning_rate": 1.0454708199069102e-06,
+      "loss": 0.0003,
+      "step": 14067
+    },
+    {
+      "epoch": 0.9066185473996262,
+      "grad_norm": 0.000671488923623265,
+      "learning_rate": 1.0447547440028643e-06,
+      "loss": 0.0,
+      "step": 14068
+    },
+    {
+      "epoch": 0.9066829928465554,
+      "grad_norm": 0.144156905549645,
+      "learning_rate": 1.0440386680988186e-06,
+      "loss": 0.0005,
+      "step": 14069
+    },
+    {
+      "epoch": 0.9067474382934846,
+      "grad_norm": 0.018324477936150364,
+      "learning_rate": 1.0433225921947727e-06,
+      "loss": 0.0001,
+      "step": 14070
+    },
+    {
+      "epoch": 0.9068118837404138,
+      "grad_norm": 0.004034000423565145,
+      "learning_rate": 1.0426065162907268e-06,
+      "loss": 0.0,
+      "step": 14071
+    },
+    {
+      "epoch": 0.906876329187343,
+      "grad_norm": 0.0016369275846041295,
+      "learning_rate": 1.041890440386681e-06,
+      "loss": 0.0,
+      "step": 14072
+    },
+    {
+      "epoch": 0.9069407746342721,
+      "grad_norm": 0.02112292707345774,
+      "learning_rate": 1.0411743644826352e-06,
+      "loss": 0.0001,
+      "step": 14073
+    },
+    {
+      "epoch": 0.9070052200812012,
+      "grad_norm": 0.0006955568561556969,
+      "learning_rate": 1.0404582885785893e-06,
+      "loss": 0.0,
+      "step": 14074
+    },
+    {
+      "epoch": 0.9070696655281304,
+      "grad_norm": 0.4678265376205017,
+      "learning_rate": 1.0397422126745436e-06,
+      "loss": 0.0008,
+      "step": 14075
+    },
+    {
+      "epoch": 0.9071341109750596,
+      "grad_norm": 1.7344620141548626,
+      "learning_rate": 1.0390261367704977e-06,
+      "loss": 0.0136,
+      "step": 14076
+    },
+    {
+      "epoch": 0.9071985564219888,
+      "grad_norm": 0.020784053798210578,
+      "learning_rate": 1.038310060866452e-06,
+      "loss": 0.0002,
+      "step": 14077
+    },
+    {
+      "epoch": 0.907263001868918,
+      "grad_norm": 0.3764581147815863,
+      "learning_rate": 1.037593984962406e-06,
+      "loss": 0.0031,
+      "step": 14078
+    },
+    {
+      "epoch": 0.9073274473158471,
+      "grad_norm": 0.015363784519826553,
+      "learning_rate": 1.0368779090583603e-06,
+      "loss": 0.0,
+      "step": 14079
+    },
+    {
+      "epoch": 0.9073918927627763,
+      "grad_norm": 9.048272690266486e-05,
+      "learning_rate": 1.0361618331543144e-06,
+      "loss": 0.0,
+      "step": 14080
+    },
+    {
+      "epoch": 0.9074563382097055,
+      "grad_norm": 0.4679583858813428,
+      "learning_rate": 1.0354457572502687e-06,
+      "loss": 0.0033,
+      "step": 14081
+    },
+    {
+      "epoch": 0.9075207836566347,
+      "grad_norm": 0.6642780402944453,
+      "learning_rate": 1.0347296813462228e-06,
+      "loss": 0.0019,
+      "step": 14082
+    },
+    {
+      "epoch": 0.9075852291035639,
+      "grad_norm": 0.0001042014709199235,
+      "learning_rate": 1.034013605442177e-06,
+      "loss": 0.0,
+      "step": 14083
+    },
+    {
+      "epoch": 0.9076496745504931,
+      "grad_norm": 0.001710960297502053,
+      "learning_rate": 1.0332975295381312e-06,
+      "loss": 0.0,
+      "step": 14084
+    },
+    {
+      "epoch": 0.9077141199974221,
+      "grad_norm": 0.008412103343040999,
+      "learning_rate": 1.0325814536340853e-06,
+      "loss": 0.0,
+      "step": 14085
+    },
+    {
+      "epoch": 0.9077785654443513,
+      "grad_norm": 0.0004960663945197895,
+      "learning_rate": 1.0318653777300394e-06,
+      "loss": 0.0,
+      "step": 14086
+    },
+    {
+      "epoch": 0.9078430108912805,
+      "grad_norm": 0.0005570506348725557,
+      "learning_rate": 1.0311493018259937e-06,
+      "loss": 0.0,
+      "step": 14087
+    },
+    {
+      "epoch": 0.9079074563382097,
+      "grad_norm": 6.0842231279498473e-05,
+      "learning_rate": 1.0304332259219478e-06,
+      "loss": 0.0,
+      "step": 14088
+    },
+    {
+      "epoch": 0.9079719017851389,
+      "grad_norm": 0.005005108687200161,
+      "learning_rate": 1.0297171500179019e-06,
+      "loss": 0.0,
+      "step": 14089
+    },
+    {
+      "epoch": 0.908036347232068,
+      "grad_norm": 0.006442428525242369,
+      "learning_rate": 1.0290010741138562e-06,
+      "loss": 0.0,
+      "step": 14090
+    },
+    {
+      "epoch": 0.9081007926789972,
+      "grad_norm": 0.1994985317478551,
+      "learning_rate": 1.0282849982098102e-06,
+      "loss": 0.0003,
+      "step": 14091
+    },
+    {
+      "epoch": 0.9081652381259264,
+      "grad_norm": 0.000303360637107461,
+      "learning_rate": 1.0275689223057645e-06,
+      "loss": 0.0,
+      "step": 14092
+    },
+    {
+      "epoch": 0.9082296835728556,
+      "grad_norm": 0.08736249878582208,
+      "learning_rate": 1.0268528464017186e-06,
+      "loss": 0.0003,
+      "step": 14093
+    },
+    {
+      "epoch": 0.9082941290197848,
+      "grad_norm": 0.0004485341202632126,
+      "learning_rate": 1.026136770497673e-06,
+      "loss": 0.0,
+      "step": 14094
+    },
+    {
+      "epoch": 0.908358574466714,
+      "grad_norm": 0.008703583972100429,
+      "learning_rate": 1.025420694593627e-06,
+      "loss": 0.0,
+      "step": 14095
+    },
+    {
+      "epoch": 0.908423019913643,
+      "grad_norm": 0.0042350023156911215,
+      "learning_rate": 1.0247046186895813e-06,
+      "loss": 0.0,
+      "step": 14096
+    },
+    {
+      "epoch": 0.9084874653605722,
+      "grad_norm": 0.0005834744678831924,
+      "learning_rate": 1.0239885427855354e-06,
+      "loss": 0.0,
+      "step": 14097
+    },
+    {
+      "epoch": 0.9085519108075014,
+      "grad_norm": 0.00024838361865375634,
+      "learning_rate": 1.0232724668814895e-06,
+      "loss": 0.0,
+      "step": 14098
+    },
+    {
+      "epoch": 0.9086163562544306,
+      "grad_norm": 0.0010090785760577745,
+      "learning_rate": 1.0225563909774438e-06,
+      "loss": 0.0,
+      "step": 14099
+    },
+    {
+      "epoch": 0.9086808017013598,
+      "grad_norm": 0.00846511578975414,
+      "learning_rate": 1.0218403150733979e-06,
+      "loss": 0.0,
+      "step": 14100
+    },
+    {
+      "epoch": 0.908745247148289,
+      "grad_norm": 0.003092826092203375,
+      "learning_rate": 1.021124239169352e-06,
+      "loss": 0.0,
+      "step": 14101
+    },
+    {
+      "epoch": 0.9088096925952182,
+      "grad_norm": 0.004059035298200513,
+      "learning_rate": 1.0204081632653063e-06,
+      "loss": 0.0,
+      "step": 14102
+    },
+    {
+      "epoch": 0.9088741380421473,
+      "grad_norm": 0.031057954187540666,
+      "learning_rate": 1.0196920873612603e-06,
+      "loss": 0.0002,
+      "step": 14103
+    },
+    {
+      "epoch": 0.9089385834890765,
+      "grad_norm": 0.007398061002103076,
+      "learning_rate": 1.0189760114572146e-06,
+      "loss": 0.0,
+      "step": 14104
+    },
+    {
+      "epoch": 0.9090030289360057,
+      "grad_norm": 0.0037488336929485094,
+      "learning_rate": 1.0182599355531687e-06,
+      "loss": 0.0,
+      "step": 14105
+    },
+    {
+      "epoch": 0.9090674743829349,
+      "grad_norm": 0.0013617477461019352,
+      "learning_rate": 1.0175438596491228e-06,
+      "loss": 0.0,
+      "step": 14106
+    },
+    {
+      "epoch": 0.909131919829864,
+      "grad_norm": 0.0011576876094899014,
+      "learning_rate": 1.0168277837450771e-06,
+      "loss": 0.0,
+      "step": 14107
+    },
+    {
+      "epoch": 0.9091963652767932,
+      "grad_norm": 0.012976751798772565,
+      "learning_rate": 1.0161117078410312e-06,
+      "loss": 0.0,
+      "step": 14108
+    },
+    {
+      "epoch": 0.9092608107237223,
+      "grad_norm": 0.0016032709037401551,
+      "learning_rate": 1.0153956319369853e-06,
+      "loss": 0.0,
+      "step": 14109
+    },
+    {
+      "epoch": 0.9093252561706515,
+      "grad_norm": 0.0405984077396907,
+      "learning_rate": 1.0146795560329396e-06,
+      "loss": 0.0,
+      "step": 14110
+    },
+    {
+      "epoch": 0.9093897016175807,
+      "grad_norm": 9.815901939114581e-05,
+      "learning_rate": 1.0139634801288937e-06,
+      "loss": 0.0,
+      "step": 14111
+    },
+    {
+      "epoch": 0.9094541470645099,
+      "grad_norm": 0.0007421945859217288,
+      "learning_rate": 1.013247404224848e-06,
+      "loss": 0.0,
+      "step": 14112
+    },
+    {
+      "epoch": 0.9095185925114391,
+      "grad_norm": 0.0044882259964135664,
+      "learning_rate": 1.012531328320802e-06,
+      "loss": 0.0,
+      "step": 14113
+    },
+    {
+      "epoch": 0.9095830379583683,
+      "grad_norm": 0.37177406992972634,
+      "learning_rate": 1.0118152524167564e-06,
+      "loss": 0.0012,
+      "step": 14114
+    },
+    {
+      "epoch": 0.9096474834052974,
+      "grad_norm": 0.0072924378929102186,
+      "learning_rate": 1.0110991765127104e-06,
+      "loss": 0.0,
+      "step": 14115
+    },
+    {
+      "epoch": 0.9097119288522266,
+      "grad_norm": 0.0006270749704708487,
+      "learning_rate": 1.0103831006086647e-06,
+      "loss": 0.0,
+      "step": 14116
+    },
+    {
+      "epoch": 0.9097763742991558,
+      "grad_norm": 0.006042327240725415,
+      "learning_rate": 1.0096670247046188e-06,
+      "loss": 0.0,
+      "step": 14117
+    },
+    {
+      "epoch": 0.909840819746085,
+      "grad_norm": 0.00024216543431723143,
+      "learning_rate": 1.008950948800573e-06,
+      "loss": 0.0,
+      "step": 14118
+    },
+    {
+      "epoch": 0.9099052651930141,
+      "grad_norm": 0.11918004743883662,
+      "learning_rate": 1.0082348728965272e-06,
+      "loss": 0.0003,
+      "step": 14119
+    },
+    {
+      "epoch": 0.9099697106399433,
+      "grad_norm": 0.007700771472075251,
+      "learning_rate": 1.0075187969924813e-06,
+      "loss": 0.0,
+      "step": 14120
+    },
+    {
+      "epoch": 0.9100341560868724,
+      "grad_norm": 0.001777663954785876,
+      "learning_rate": 1.0068027210884354e-06,
+      "loss": 0.0,
+      "step": 14121
+    },
+    {
+      "epoch": 0.9100986015338016,
+      "grad_norm": 0.004718985009600268,
+      "learning_rate": 1.0060866451843897e-06,
+      "loss": 0.0,
+      "step": 14122
+    },
+    {
+      "epoch": 0.9101630469807308,
+      "grad_norm": 0.006394187944082488,
+      "learning_rate": 1.0053705692803438e-06,
+      "loss": 0.0,
+      "step": 14123
+    },
+    {
+      "epoch": 0.91022749242766,
+      "grad_norm": 0.04192140346511436,
+      "learning_rate": 1.0046544933762979e-06,
+      "loss": 0.0001,
+      "step": 14124
+    },
+    {
+      "epoch": 0.9102919378745892,
+      "grad_norm": 0.020396504587853657,
+      "learning_rate": 1.0039384174722522e-06,
+      "loss": 0.0001,
+      "step": 14125
+    },
+    {
+      "epoch": 0.9103563833215184,
+      "grad_norm": 9.335500306033411e-05,
+      "learning_rate": 1.0032223415682062e-06,
+      "loss": 0.0,
+      "step": 14126
+    },
+    {
+      "epoch": 0.9104208287684475,
+      "grad_norm": 0.001052229329982989,
+      "learning_rate": 1.0025062656641603e-06,
+      "loss": 0.0,
+      "step": 14127
+    },
+    {
+      "epoch": 0.9104852742153767,
+      "grad_norm": 0.00044013625982259485,
+      "learning_rate": 1.0017901897601146e-06,
+      "loss": 0.0,
+      "step": 14128
+    },
+    {
+      "epoch": 0.9105497196623059,
+      "grad_norm": 0.07613235756810091,
+      "learning_rate": 1.0010741138560687e-06,
+      "loss": 0.0008,
+      "step": 14129
+    },
+    {
+      "epoch": 0.910614165109235,
+      "grad_norm": 0.0015174921341569256,
+      "learning_rate": 1.000358037952023e-06,
+      "loss": 0.0,
+      "step": 14130
+    },
+    {
+      "epoch": 0.9106786105561642,
+      "grad_norm": 8.869528142380125e-05,
+      "learning_rate": 9.99641962047977e-07,
+      "loss": 0.0,
+      "step": 14131
+    },
+    {
+      "epoch": 0.9107430560030934,
+      "grad_norm": 0.0018528149454641407,
+      "learning_rate": 9.989258861439314e-07,
+      "loss": 0.0,
+      "step": 14132
+    },
+    {
+      "epoch": 0.9108075014500225,
+      "grad_norm": 0.0013100437281134071,
+      "learning_rate": 9.982098102398855e-07,
+      "loss": 0.0,
+      "step": 14133
+    },
+    {
+      "epoch": 0.9108719468969517,
+      "grad_norm": 0.0038834947242926132,
+      "learning_rate": 9.974937343358398e-07,
+      "loss": 0.0,
+      "step": 14134
+    },
+    {
+      "epoch": 0.9109363923438809,
+      "grad_norm": 0.003056712747435566,
+      "learning_rate": 9.967776584317939e-07,
+      "loss": 0.0,
+      "step": 14135
+    },
+    {
+      "epoch": 0.9110008377908101,
+      "grad_norm": 0.022727158280140002,
+      "learning_rate": 9.96061582527748e-07,
+      "loss": 0.0001,
+      "step": 14136
+    },
+    {
+      "epoch": 0.9110652832377393,
+      "grad_norm": 0.000252373752242693,
+      "learning_rate": 9.953455066237023e-07,
+      "loss": 0.0,
+      "step": 14137
+    },
+    {
+      "epoch": 0.9111297286846685,
+      "grad_norm": 0.005239760950976765,
+      "learning_rate": 9.946294307196563e-07,
+      "loss": 0.0,
+      "step": 14138
+    },
+    {
+      "epoch": 0.9111941741315976,
+      "grad_norm": 0.07205721082201468,
+      "learning_rate": 9.939133548156106e-07,
+      "loss": 0.0001,
+      "step": 14139
+    },
+    {
+      "epoch": 0.9112586195785268,
+      "grad_norm": 0.001012468350364244,
+      "learning_rate": 9.931972789115647e-07,
+      "loss": 0.0,
+      "step": 14140
+    },
+    {
+      "epoch": 0.9113230650254559,
+      "grad_norm": 0.020371513483527344,
+      "learning_rate": 9.924812030075188e-07,
+      "loss": 0.0001,
+      "step": 14141
+    },
+    {
+      "epoch": 0.9113875104723851,
+      "grad_norm": 0.5710255593894316,
+      "learning_rate": 9.917651271034731e-07,
+      "loss": 0.0038,
+      "step": 14142
+    },
+    {
+      "epoch": 0.9114519559193143,
+      "grad_norm": 0.003678925592998401,
+      "learning_rate": 9.910490511994272e-07,
+      "loss": 0.0,
+      "step": 14143
+    },
+    {
+      "epoch": 0.9115164013662435,
+      "grad_norm": 0.0010209924279294914,
+      "learning_rate": 9.903329752953813e-07,
+      "loss": 0.0,
+      "step": 14144
+    },
+    {
+      "epoch": 0.9115808468131726,
+      "grad_norm": 0.002210604517047804,
+      "learning_rate": 9.896168993913356e-07,
+      "loss": 0.0,
+      "step": 14145
+    },
+    {
+      "epoch": 0.9116452922601018,
+      "grad_norm": 0.004980856990466322,
+      "learning_rate": 9.889008234872897e-07,
+      "loss": 0.0,
+      "step": 14146
+    },
+    {
+      "epoch": 0.911709737707031,
+      "grad_norm": 0.005197732728780156,
+      "learning_rate": 9.881847475832438e-07,
+      "loss": 0.0,
+      "step": 14147
+    },
+    {
+      "epoch": 0.9117741831539602,
+      "grad_norm": 0.0006288855310315486,
+      "learning_rate": 9.87468671679198e-07,
+      "loss": 0.0,
+      "step": 14148
+    },
+    {
+      "epoch": 0.9118386286008894,
+      "grad_norm": 0.2870331747440102,
+      "learning_rate": 9.867525957751522e-07,
+      "loss": 0.0006,
+      "step": 14149
+    },
+    {
+      "epoch": 0.9119030740478186,
+      "grad_norm": 0.009226923949112864,
+      "learning_rate": 9.860365198711065e-07,
+      "loss": 0.0,
+      "step": 14150
+    },
+    {
+      "epoch": 0.9119675194947477,
+      "grad_norm": 0.00021664613314001927,
+      "learning_rate": 9.853204439670605e-07,
+      "loss": 0.0,
+      "step": 14151
+    },
+    {
+      "epoch": 0.9120319649416768,
+      "grad_norm": 0.07574482256106105,
+      "learning_rate": 9.846043680630148e-07,
+      "loss": 0.0001,
+      "step": 14152
+    },
+    {
+      "epoch": 0.912096410388606,
+      "grad_norm": 0.0005642558123564372,
+      "learning_rate": 9.83888292158969e-07,
+      "loss": 0.0,
+      "step": 14153
+    },
+    {
+      "epoch": 0.9121608558355352,
+      "grad_norm": 0.00046067530354494283,
+      "learning_rate": 9.831722162549232e-07,
+      "loss": 0.0,
+      "step": 14154
+    },
+    {
+      "epoch": 0.9122253012824644,
+      "grad_norm": 0.0019799729249998076,
+      "learning_rate": 9.824561403508773e-07,
+      "loss": 0.0,
+      "step": 14155
+    },
+    {
+      "epoch": 0.9122897467293936,
+      "grad_norm": 0.0037062705460312938,
+      "learning_rate": 9.817400644468314e-07,
+      "loss": 0.0,
+      "step": 14156
+    },
+    {
+      "epoch": 0.9123541921763227,
+      "grad_norm": 0.21694178506613537,
+      "learning_rate": 9.810239885427857e-07,
+      "loss": 0.0012,
+      "step": 14157
+    },
+    {
+      "epoch": 0.9124186376232519,
+      "grad_norm": 2.7329962563106694e-05,
+      "learning_rate": 9.803079126387398e-07,
+      "loss": 0.0,
+      "step": 14158
+    },
+    {
+      "epoch": 0.9124830830701811,
+      "grad_norm": 0.017096042360411064,
+      "learning_rate": 9.795918367346939e-07,
+      "loss": 0.0001,
+      "step": 14159
+    },
+    {
+      "epoch": 0.9125475285171103,
+      "grad_norm": 0.0022227801850019,
+      "learning_rate": 9.788757608306482e-07,
+      "loss": 0.0,
+      "step": 14160
+    },
+    {
+      "epoch": 0.9126119739640395,
+      "grad_norm": 0.00022310574290482324,
+      "learning_rate": 9.781596849266023e-07,
+      "loss": 0.0,
+      "step": 14161
+    },
+    {
+      "epoch": 0.9126764194109687,
+      "grad_norm": 0.0001928504205817919,
+      "learning_rate": 9.774436090225563e-07,
+      "loss": 0.0,
+      "step": 14162
+    },
+    {
+      "epoch": 0.9127408648578977,
+      "grad_norm": 0.0003620854373529105,
+      "learning_rate": 9.767275331185106e-07,
+      "loss": 0.0,
+      "step": 14163
+    },
+    {
+      "epoch": 0.9128053103048269,
+      "grad_norm": 0.0003246124005665155,
+      "learning_rate": 9.760114572144647e-07,
+      "loss": 0.0,
+      "step": 14164
+    },
+    {
+      "epoch": 0.9128697557517561,
+      "grad_norm": 0.00037746760821441995,
+      "learning_rate": 9.75295381310419e-07,
+      "loss": 0.0,
+      "step": 14165
+    },
+    {
+      "epoch": 0.9129342011986853,
+      "grad_norm": 0.0021465206942596557,
+      "learning_rate": 9.745793054063731e-07,
+      "loss": 0.0,
+      "step": 14166
+    },
+    {
+      "epoch": 0.9129986466456145,
+      "grad_norm": 0.009832061962162885,
+      "learning_rate": 9.738632295023274e-07,
+      "loss": 0.0,
+      "step": 14167
+    },
+    {
+      "epoch": 0.9130630920925437,
+      "grad_norm": 0.011101602739312185,
+      "learning_rate": 9.731471535982815e-07,
+      "loss": 0.0001,
+      "step": 14168
+    },
+    {
+      "epoch": 0.9131275375394728,
+      "grad_norm": 0.002187157704356501,
+      "learning_rate": 9.724310776942358e-07,
+      "loss": 0.0,
+      "step": 14169
+    },
+    {
+      "epoch": 0.913191982986402,
+      "grad_norm": 0.01536422494149394,
+      "learning_rate": 9.717150017901899e-07,
+      "loss": 0.0001,
+      "step": 14170
+    },
+    {
+      "epoch": 0.9132564284333312,
+      "grad_norm": 0.14876547142920768,
+      "learning_rate": 9.70998925886144e-07,
+      "loss": 0.0003,
+      "step": 14171
+    },
+    {
+      "epoch": 0.9133208738802604,
+      "grad_norm": 0.0063150968108142535,
+      "learning_rate": 9.702828499820983e-07,
+      "loss": 0.0001,
+      "step": 14172
+    },
+    {
+      "epoch": 0.9133853193271896,
+      "grad_norm": 0.0011484692990875677,
+      "learning_rate": 9.695667740780524e-07,
+      "loss": 0.0,
+      "step": 14173
+    },
+    {
+      "epoch": 0.9134497647741187,
+      "grad_norm": 0.00021408494086622607,
+      "learning_rate": 9.688506981740067e-07,
+      "loss": 0.0,
+      "step": 14174
+    },
+    {
+      "epoch": 0.9135142102210478,
+      "grad_norm": 0.005474376006443933,
+      "learning_rate": 9.681346222699607e-07,
+      "loss": 0.0,
+      "step": 14175
+    },
+    {
+      "epoch": 0.913578655667977,
+      "grad_norm": 0.0009916894050175195,
+      "learning_rate": 9.674185463659148e-07,
+      "loss": 0.0,
+      "step": 14176
+    },
+    {
+      "epoch": 0.9136431011149062,
+      "grad_norm": 0.004285597191006188,
+      "learning_rate": 9.667024704618691e-07,
+      "loss": 0.0,
+      "step": 14177
+    },
+    {
+      "epoch": 0.9137075465618354,
+      "grad_norm": 6.016757867062615e-05,
+      "learning_rate": 9.659863945578232e-07,
+      "loss": 0.0,
+      "step": 14178
+    },
+    {
+      "epoch": 0.9137719920087646,
+      "grad_norm": 0.00013869091499201224,
+      "learning_rate": 9.652703186537773e-07,
+      "loss": 0.0,
+      "step": 14179
+    },
+    {
+      "epoch": 0.9138364374556938,
+      "grad_norm": 0.008156751396045343,
+      "learning_rate": 9.645542427497316e-07,
+      "loss": 0.0001,
+      "step": 14180
+    },
+    {
+      "epoch": 0.913900882902623,
+      "grad_norm": 0.0010459882012239957,
+      "learning_rate": 9.638381668456857e-07,
+      "loss": 0.0,
+      "step": 14181
+    },
+    {
+      "epoch": 0.9139653283495521,
+      "grad_norm": 0.05334048614408453,
+      "learning_rate": 9.631220909416398e-07,
+      "loss": 0.0001,
+      "step": 14182
+    },
+    {
+      "epoch": 0.9140297737964813,
+      "grad_norm": 0.0015840045943365014,
+      "learning_rate": 9.62406015037594e-07,
+      "loss": 0.0,
+      "step": 14183
+    },
+    {
+      "epoch": 0.9140942192434105,
+      "grad_norm": 0.7855814830833372,
+      "learning_rate": 9.616899391335482e-07,
+      "loss": 0.0021,
+      "step": 14184
+    },
+    {
+      "epoch": 0.9141586646903396,
+      "grad_norm": 0.00020580164141785254,
+      "learning_rate": 9.609738632295025e-07,
+      "loss": 0.0,
+      "step": 14185
+    },
+    {
+      "epoch": 0.9142231101372688,
+      "grad_norm": 0.0011450973071098205,
+      "learning_rate": 9.602577873254565e-07,
+      "loss": 0.0,
+      "step": 14186
+    },
+    {
+      "epoch": 0.9142875555841979,
+      "grad_norm": 0.008593922428411569,
+      "learning_rate": 9.595417114214108e-07,
+      "loss": 0.0,
+      "step": 14187
+    },
+    {
+      "epoch": 0.9143520010311271,
+      "grad_norm": 0.0016867673517670938,
+      "learning_rate": 9.58825635517365e-07,
+      "loss": 0.0,
+      "step": 14188
+    },
+    {
+      "epoch": 0.9144164464780563,
+      "grad_norm": 0.0015443399622426112,
+      "learning_rate": 9.581095596133192e-07,
+      "loss": 0.0,
+      "step": 14189
+    },
+    {
+      "epoch": 0.9144808919249855,
+      "grad_norm": 0.02063164477752474,
+      "learning_rate": 9.573934837092733e-07,
+      "loss": 0.0001,
+      "step": 14190
+    },
+    {
+      "epoch": 0.9145453373719147,
+      "grad_norm": 0.0024049854891011677,
+      "learning_rate": 9.566774078052274e-07,
+      "loss": 0.0,
+      "step": 14191
+    },
+    {
+      "epoch": 0.9146097828188439,
+      "grad_norm": 0.01716273295228107,
+      "learning_rate": 9.559613319011817e-07,
+      "loss": 0.0002,
+      "step": 14192
+    },
+    {
+      "epoch": 0.914674228265773,
+      "grad_norm": 0.12328758990372396,
+      "learning_rate": 9.552452559971358e-07,
+      "loss": 0.0012,
+      "step": 14193
+    },
+    {
+      "epoch": 0.9147386737127022,
+      "grad_norm": 0.04222494037999026,
+      "learning_rate": 9.545291800930899e-07,
+      "loss": 0.0,
+      "step": 14194
+    },
+    {
+      "epoch": 0.9148031191596314,
+      "grad_norm": 0.0006967093892120433,
+      "learning_rate": 9.538131041890442e-07,
+      "loss": 0.0,
+      "step": 14195
+    },
+    {
+      "epoch": 0.9148675646065606,
+      "grad_norm": 0.10183633970320935,
+      "learning_rate": 9.530970282849983e-07,
+      "loss": 0.0014,
+      "step": 14196
+    },
+    {
+      "epoch": 0.9149320100534897,
+      "grad_norm": 0.04194055060462203,
+      "learning_rate": 9.523809523809525e-07,
+      "loss": 0.0017,
+      "step": 14197
+    },
+    {
+      "epoch": 0.9149964555004189,
+      "grad_norm": 0.2690406148430233,
+      "learning_rate": 9.516648764769066e-07,
+      "loss": 0.0005,
+      "step": 14198
+    },
+    {
+      "epoch": 0.915060900947348,
+      "grad_norm": 0.00021367249271819308,
+      "learning_rate": 9.509488005728608e-07,
+      "loss": 0.0,
+      "step": 14199
+    },
+    {
+      "epoch": 0.9151253463942772,
+      "grad_norm": 0.0014708756146450935,
+      "learning_rate": 9.502327246688149e-07,
+      "loss": 0.0,
+      "step": 14200
+    },
+    {
+      "epoch": 0.9151897918412064,
+      "grad_norm": 0.0002496282942443766,
+      "learning_rate": 9.495166487647692e-07,
+      "loss": 0.0,
+      "step": 14201
+    },
+    {
+      "epoch": 0.9152542372881356,
+      "grad_norm": 0.016152459797736374,
+      "learning_rate": 9.488005728607233e-07,
+      "loss": 0.0,
+      "step": 14202
+    },
+    {
+      "epoch": 0.9153186827350648,
+      "grad_norm": 0.010979395241874716,
+      "learning_rate": 9.480844969566774e-07,
+      "loss": 0.0,
+      "step": 14203
+    },
+    {
+      "epoch": 0.915383128181994,
+      "grad_norm": 0.0009375031145412428,
+      "learning_rate": 9.473684210526317e-07,
+      "loss": 0.0,
+      "step": 14204
+    },
+    {
+      "epoch": 0.9154475736289232,
+      "grad_norm": 0.01797304484362447,
+      "learning_rate": 9.466523451485858e-07,
+      "loss": 0.0001,
+      "step": 14205
+    },
+    {
+      "epoch": 0.9155120190758523,
+      "grad_norm": 0.0002960165775091309,
+      "learning_rate": 9.4593626924454e-07,
+      "loss": 0.0,
+      "step": 14206
+    },
+    {
+      "epoch": 0.9155764645227815,
+      "grad_norm": 0.00013216728878975067,
+      "learning_rate": 9.452201933404942e-07,
+      "loss": 0.0,
+      "step": 14207
+    },
+    {
+      "epoch": 0.9156409099697106,
+      "grad_norm": 0.0016359036208272025,
+      "learning_rate": 9.445041174364484e-07,
+      "loss": 0.0,
+      "step": 14208
+    },
+    {
+      "epoch": 0.9157053554166398,
+      "grad_norm": 0.002744799936497066,
+      "learning_rate": 9.437880415324025e-07,
+      "loss": 0.0,
+      "step": 14209
+    },
+    {
+      "epoch": 0.915769800863569,
+      "grad_norm": 0.00023160867892980746,
+      "learning_rate": 9.430719656283567e-07,
+      "loss": 0.0,
+      "step": 14210
+    },
+    {
+      "epoch": 0.9158342463104981,
+      "grad_norm": 0.009411741817506578,
+      "learning_rate": 9.423558897243108e-07,
+      "loss": 0.0,
+      "step": 14211
+    },
+    {
+      "epoch": 0.9158986917574273,
+      "grad_norm": 0.001378941707229188,
+      "learning_rate": 9.416398138202651e-07,
+      "loss": 0.0,
+      "step": 14212
+    },
+    {
+      "epoch": 0.9159631372043565,
+      "grad_norm": 0.025205632894264283,
+      "learning_rate": 9.409237379162192e-07,
+      "loss": 0.0001,
+      "step": 14213
+    },
+    {
+      "epoch": 0.9160275826512857,
+      "grad_norm": 0.39096405573360754,
+      "learning_rate": 9.402076620121733e-07,
+      "loss": 0.001,
+      "step": 14214
+    },
+    {
+      "epoch": 0.9160920280982149,
+      "grad_norm": 0.00014380302746555912,
+      "learning_rate": 9.394915861081276e-07,
+      "loss": 0.0,
+      "step": 14215
+    },
+    {
+      "epoch": 0.9161564735451441,
+      "grad_norm": 0.002670300688032658,
+      "learning_rate": 9.387755102040817e-07,
+      "loss": 0.0,
+      "step": 14216
+    },
+    {
+      "epoch": 0.9162209189920733,
+      "grad_norm": 0.11628580086246942,
+      "learning_rate": 9.380594343000359e-07,
+      "loss": 0.0004,
+      "step": 14217
+    },
+    {
+      "epoch": 0.9162853644390024,
+      "grad_norm": 0.0002583038235821294,
+      "learning_rate": 9.373433583959901e-07,
+      "loss": 0.0,
+      "step": 14218
+    },
+    {
+      "epoch": 0.9163498098859315,
+      "grad_norm": 0.022131033830474392,
+      "learning_rate": 9.366272824919443e-07,
+      "loss": 0.0,
+      "step": 14219
+    },
+    {
+      "epoch": 0.9164142553328607,
+      "grad_norm": 0.0015146300558742055,
+      "learning_rate": 9.359112065878984e-07,
+      "loss": 0.0,
+      "step": 14220
+    },
+    {
+      "epoch": 0.9164787007797899,
+      "grad_norm": 4.8640373473175795e-05,
+      "learning_rate": 9.351951306838527e-07,
+      "loss": 0.0,
+      "step": 14221
+    },
+    {
+      "epoch": 0.9165431462267191,
+      "grad_norm": 0.4697828516869718,
+      "learning_rate": 9.344790547798067e-07,
+      "loss": 0.003,
+      "step": 14222
+    },
+    {
+      "epoch": 0.9166075916736482,
+      "grad_norm": 0.057128720813026404,
+      "learning_rate": 9.337629788757608e-07,
+      "loss": 0.0016,
+      "step": 14223
+    },
+    {
+      "epoch": 0.9166720371205774,
+      "grad_norm": 0.03367085797727787,
+      "learning_rate": 9.330469029717151e-07,
+      "loss": 0.0001,
+      "step": 14224
+    },
+    {
+      "epoch": 0.9167364825675066,
+      "grad_norm": 0.0030701623211639023,
+      "learning_rate": 9.323308270676692e-07,
+      "loss": 0.0,
+      "step": 14225
+    },
+    {
+      "epoch": 0.9168009280144358,
+      "grad_norm": 0.0019137412471503506,
+      "learning_rate": 9.316147511636234e-07,
+      "loss": 0.0,
+      "step": 14226
+    },
+    {
+      "epoch": 0.916865373461365,
+      "grad_norm": 0.0006687030619872437,
+      "learning_rate": 9.308986752595776e-07,
+      "loss": 0.0,
+      "step": 14227
+    },
+    {
+      "epoch": 0.9169298189082942,
+      "grad_norm": 1.7368992487315262,
+      "learning_rate": 9.301825993555318e-07,
+      "loss": 0.0119,
+      "step": 14228
+    },
+    {
+      "epoch": 0.9169942643552234,
+      "grad_norm": 0.0003233422494308578,
+      "learning_rate": 9.294665234514859e-07,
+      "loss": 0.0,
+      "step": 14229
+    },
+    {
+      "epoch": 0.9170587098021524,
+      "grad_norm": 0.0069097439182549385,
+      "learning_rate": 9.287504475474402e-07,
+      "loss": 0.0,
+      "step": 14230
+    },
+    {
+      "epoch": 0.9171231552490816,
+      "grad_norm": 0.0029392214582604566,
+      "learning_rate": 9.280343716433943e-07,
+      "loss": 0.0,
+      "step": 14231
+    },
+    {
+      "epoch": 0.9171876006960108,
+      "grad_norm": 0.0067482743978787415,
+      "learning_rate": 9.273182957393484e-07,
+      "loss": 0.0,
+      "step": 14232
+    },
+    {
+      "epoch": 0.91725204614294,
+      "grad_norm": 0.0020425040118576547,
+      "learning_rate": 9.266022198353027e-07,
+      "loss": 0.0,
+      "step": 14233
+    },
+    {
+      "epoch": 0.9173164915898692,
+      "grad_norm": 0.043513331215841396,
+      "learning_rate": 9.258861439312567e-07,
+      "loss": 0.0001,
+      "step": 14234
+    },
+    {
+      "epoch": 0.9173809370367983,
+      "grad_norm": 0.0005674751602963122,
+      "learning_rate": 9.251700680272109e-07,
+      "loss": 0.0,
+      "step": 14235
+    },
+    {
+      "epoch": 0.9174453824837275,
+      "grad_norm": 0.00017620281295303855,
+      "learning_rate": 9.244539921231651e-07,
+      "loss": 0.0,
+      "step": 14236
+    },
+    {
+      "epoch": 0.9175098279306567,
+      "grad_norm": 0.001052507181908841,
+      "learning_rate": 9.237379162191193e-07,
+      "loss": 0.0,
+      "step": 14237
+    },
+    {
+      "epoch": 0.9175742733775859,
+      "grad_norm": 0.14251254807613536,
+      "learning_rate": 9.230218403150734e-07,
+      "loss": 0.0019,
+      "step": 14238
+    },
+    {
+      "epoch": 0.9176387188245151,
+      "grad_norm": 0.0016286015558987282,
+      "learning_rate": 9.223057644110277e-07,
+      "loss": 0.0,
+      "step": 14239
+    },
+    {
+      "epoch": 0.9177031642714443,
+      "grad_norm": 0.00010581374894291006,
+      "learning_rate": 9.215896885069818e-07,
+      "loss": 0.0,
+      "step": 14240
+    },
+    {
+      "epoch": 0.9177676097183733,
+      "grad_norm": 0.0015797488393140531,
+      "learning_rate": 9.208736126029359e-07,
+      "loss": 0.0,
+      "step": 14241
+    },
+    {
+      "epoch": 0.9178320551653025,
+      "grad_norm": 0.0008123762794186068,
+      "learning_rate": 9.201575366988902e-07,
+      "loss": 0.0,
+      "step": 14242
+    },
+    {
+      "epoch": 0.9178965006122317,
+      "grad_norm": 0.008421636058342962,
+      "learning_rate": 9.194414607948443e-07,
+      "loss": 0.0,
+      "step": 14243
+    },
+    {
+      "epoch": 0.9179609460591609,
+      "grad_norm": 0.000672241911371534,
+      "learning_rate": 9.187253848907985e-07,
+      "loss": 0.0,
+      "step": 14244
+    },
+    {
+      "epoch": 0.9180253915060901,
+      "grad_norm": 0.0005130752034765722,
+      "learning_rate": 9.180093089867527e-07,
+      "loss": 0.0,
+      "step": 14245
+    },
+    {
+      "epoch": 0.9180898369530193,
+      "grad_norm": 6.957164338496054e-05,
+      "learning_rate": 9.172932330827068e-07,
+      "loss": 0.0,
+      "step": 14246
+    },
+    {
+      "epoch": 0.9181542823999485,
+      "grad_norm": 0.004817899785118754,
+      "learning_rate": 9.16577157178661e-07,
+      "loss": 0.0,
+      "step": 14247
+    },
+    {
+      "epoch": 0.9182187278468776,
+      "grad_norm": 0.0004698135349077357,
+      "learning_rate": 9.158610812746152e-07,
+      "loss": 0.0,
+      "step": 14248
+    },
+    {
+      "epoch": 0.9182831732938068,
+      "grad_norm": 0.022730519540926336,
+      "learning_rate": 9.151450053705693e-07,
+      "loss": 0.0002,
+      "step": 14249
+    },
+    {
+      "epoch": 0.918347618740736,
+      "grad_norm": 0.0039667025926992784,
+      "learning_rate": 9.144289294665236e-07,
+      "loss": 0.0,
+      "step": 14250
+    },
+    {
+      "epoch": 0.9184120641876652,
+      "grad_norm": 0.002694283624502539,
+      "learning_rate": 9.137128535624777e-07,
+      "loss": 0.0,
+      "step": 14251
+    },
+    {
+      "epoch": 0.9184765096345943,
+      "grad_norm": 0.012668603875843228,
+      "learning_rate": 9.129967776584318e-07,
+      "loss": 0.0001,
+      "step": 14252
+    },
+    {
+      "epoch": 0.9185409550815234,
+      "grad_norm": 0.00016267248176073663,
+      "learning_rate": 9.122807017543861e-07,
+      "loss": 0.0,
+      "step": 14253
+    },
+    {
+      "epoch": 0.9186054005284526,
+      "grad_norm": 0.16982769822718005,
+      "learning_rate": 9.115646258503402e-07,
+      "loss": 0.0005,
+      "step": 14254
+    },
+    {
+      "epoch": 0.9186698459753818,
+      "grad_norm": 0.0004709426465504047,
+      "learning_rate": 9.108485499462944e-07,
+      "loss": 0.0,
+      "step": 14255
+    },
+    {
+      "epoch": 0.918734291422311,
+      "grad_norm": 0.0028317174475590155,
+      "learning_rate": 9.101324740422486e-07,
+      "loss": 0.0,
+      "step": 14256
+    },
+    {
+      "epoch": 0.9187987368692402,
+      "grad_norm": 0.00039515144590099115,
+      "learning_rate": 9.094163981382028e-07,
+      "loss": 0.0,
+      "step": 14257
+    },
+    {
+      "epoch": 0.9188631823161694,
+      "grad_norm": 0.034502345692150146,
+      "learning_rate": 9.087003222341568e-07,
+      "loss": 0.0,
+      "step": 14258
+    },
+    {
+      "epoch": 0.9189276277630986,
+      "grad_norm": 0.013624407437962447,
+      "learning_rate": 9.079842463301111e-07,
+      "loss": 0.0001,
+      "step": 14259
+    },
+    {
+      "epoch": 0.9189920732100277,
+      "grad_norm": 0.31261428888863013,
+      "learning_rate": 9.072681704260652e-07,
+      "loss": 0.0011,
+      "step": 14260
+    },
+    {
+      "epoch": 0.9190565186569569,
+      "grad_norm": 0.27199733765961076,
+      "learning_rate": 9.065520945220194e-07,
+      "loss": 0.0005,
+      "step": 14261
+    },
+    {
+      "epoch": 0.9191209641038861,
+      "grad_norm": 0.04721813969246109,
+      "learning_rate": 9.058360186179736e-07,
+      "loss": 0.0016,
+      "step": 14262
+    },
+    {
+      "epoch": 0.9191854095508152,
+      "grad_norm": 0.0005272377698462465,
+      "learning_rate": 9.051199427139278e-07,
+      "loss": 0.0,
+      "step": 14263
+    },
+    {
+      "epoch": 0.9192498549977444,
+      "grad_norm": 0.003632628093650156,
+      "learning_rate": 9.044038668098819e-07,
+      "loss": 0.0,
+      "step": 14264
+    },
+    {
+      "epoch": 0.9193143004446735,
+      "grad_norm": 0.08249185312389225,
+      "learning_rate": 9.036877909058362e-07,
+      "loss": 0.0001,
+      "step": 14265
+    },
+    {
+      "epoch": 0.9193787458916027,
+      "grad_norm": 0.01769209298130543,
+      "learning_rate": 9.029717150017903e-07,
+      "loss": 0.0001,
+      "step": 14266
+    },
+    {
+      "epoch": 0.9194431913385319,
+      "grad_norm": 0.007572932169365242,
+      "learning_rate": 9.022556390977444e-07,
+      "loss": 0.0,
+      "step": 14267
+    },
+    {
+      "epoch": 0.9195076367854611,
+      "grad_norm": 0.0006129075108329983,
+      "learning_rate": 9.015395631936987e-07,
+      "loss": 0.0,
+      "step": 14268
+    },
+    {
+      "epoch": 0.9195720822323903,
+      "grad_norm": 0.15367907565903965,
+      "learning_rate": 9.008234872896527e-07,
+      "loss": 0.0019,
+      "step": 14269
+    },
+    {
+      "epoch": 0.9196365276793195,
+      "grad_norm": 0.00012896780691753785,
+      "learning_rate": 9.001074113856069e-07,
+      "loss": 0.0,
+      "step": 14270
+    },
+    {
+      "epoch": 0.9197009731262487,
+      "grad_norm": 0.04012686418183218,
+      "learning_rate": 8.993913354815611e-07,
+      "loss": 0.0001,
+      "step": 14271
+    },
+    {
+      "epoch": 0.9197654185731778,
+      "grad_norm": 0.009376539846461099,
+      "learning_rate": 8.986752595775153e-07,
+      "loss": 0.0,
+      "step": 14272
+    },
+    {
+      "epoch": 0.919829864020107,
+      "grad_norm": 0.10376941411654185,
+      "learning_rate": 8.979591836734694e-07,
+      "loss": 0.0001,
+      "step": 14273
+    },
+    {
+      "epoch": 0.9198943094670362,
+      "grad_norm": 0.004417011550164127,
+      "learning_rate": 8.972431077694237e-07,
+      "loss": 0.0,
+      "step": 14274
+    },
+    {
+      "epoch": 0.9199587549139653,
+      "grad_norm": 0.0010578230684077303,
+      "learning_rate": 8.965270318653778e-07,
+      "loss": 0.0,
+      "step": 14275
+    },
+    {
+      "epoch": 0.9200232003608945,
+      "grad_norm": 0.005991646645697451,
+      "learning_rate": 8.958109559613319e-07,
+      "loss": 0.0,
+      "step": 14276
+    },
+    {
+      "epoch": 0.9200876458078237,
+      "grad_norm": 0.0013921188764913148,
+      "learning_rate": 8.950948800572862e-07,
+      "loss": 0.0,
+      "step": 14277
+    },
+    {
+      "epoch": 0.9201520912547528,
+      "grad_norm": 0.003431294612354495,
+      "learning_rate": 8.943788041532403e-07,
+      "loss": 0.0,
+      "step": 14278
+    },
+    {
+      "epoch": 0.920216536701682,
+      "grad_norm": 0.0009395657251499541,
+      "learning_rate": 8.936627282491945e-07,
+      "loss": 0.0,
+      "step": 14279
+    },
+    {
+      "epoch": 0.9202809821486112,
+      "grad_norm": 0.21383604058028396,
+      "learning_rate": 8.929466523451487e-07,
+      "loss": 0.0007,
+      "step": 14280
+    },
+    {
+      "epoch": 0.9203454275955404,
+      "grad_norm": 0.07478550921576256,
+      "learning_rate": 8.922305764411029e-07,
+      "loss": 0.0002,
+      "step": 14281
+    },
+    {
+      "epoch": 0.9204098730424696,
+      "grad_norm": 0.0018806258634827169,
+      "learning_rate": 8.915145005370569e-07,
+      "loss": 0.0,
+      "step": 14282
+    },
+    {
+      "epoch": 0.9204743184893988,
+      "grad_norm": 0.0006642417777200569,
+      "learning_rate": 8.907984246330112e-07,
+      "loss": 0.0,
+      "step": 14283
+    },
+    {
+      "epoch": 0.9205387639363279,
+      "grad_norm": 0.033905118569600774,
+      "learning_rate": 8.900823487289653e-07,
+      "loss": 0.0001,
+      "step": 14284
+    },
+    {
+      "epoch": 0.9206032093832571,
+      "grad_norm": 0.00029521834576314433,
+      "learning_rate": 8.893662728249196e-07,
+      "loss": 0.0,
+      "step": 14285
+    },
+    {
+      "epoch": 0.9206676548301862,
+      "grad_norm": 0.0026862544176196293,
+      "learning_rate": 8.886501969208737e-07,
+      "loss": 0.0,
+      "step": 14286
+    },
+    {
+      "epoch": 0.9207321002771154,
+      "grad_norm": 0.029072616489818775,
+      "learning_rate": 8.879341210168278e-07,
+      "loss": 0.0015,
+      "step": 14287
+    },
+    {
+      "epoch": 0.9207965457240446,
+      "grad_norm": 0.005206056469858584,
+      "learning_rate": 8.872180451127821e-07,
+      "loss": 0.0,
+      "step": 14288
+    },
+    {
+      "epoch": 0.9208609911709738,
+      "grad_norm": 0.004492865880830228,
+      "learning_rate": 8.865019692087362e-07,
+      "loss": 0.0,
+      "step": 14289
+    },
+    {
+      "epoch": 0.9209254366179029,
+      "grad_norm": 0.04546572249122386,
+      "learning_rate": 8.857858933046904e-07,
+      "loss": 0.0,
+      "step": 14290
+    },
+    {
+      "epoch": 0.9209898820648321,
+      "grad_norm": 0.005094889517667441,
+      "learning_rate": 8.850698174006446e-07,
+      "loss": 0.0001,
+      "step": 14291
+    },
+    {
+      "epoch": 0.9210543275117613,
+      "grad_norm": 0.6241890175700587,
+      "learning_rate": 8.843537414965988e-07,
+      "loss": 0.0136,
+      "step": 14292
+    },
+    {
+      "epoch": 0.9211187729586905,
+      "grad_norm": 0.489125871665835,
+      "learning_rate": 8.836376655925528e-07,
+      "loss": 0.0015,
+      "step": 14293
+    },
+    {
+      "epoch": 0.9211832184056197,
+      "grad_norm": 0.09846547066195413,
+      "learning_rate": 8.829215896885071e-07,
+      "loss": 0.0002,
+      "step": 14294
+    },
+    {
+      "epoch": 0.9212476638525489,
+      "grad_norm": 0.0005087791308881729,
+      "learning_rate": 8.822055137844612e-07,
+      "loss": 0.0,
+      "step": 14295
+    },
+    {
+      "epoch": 0.921312109299478,
+      "grad_norm": 0.0017957632932818469,
+      "learning_rate": 8.814894378804153e-07,
+      "loss": 0.0,
+      "step": 14296
+    },
+    {
+      "epoch": 0.9213765547464071,
+      "grad_norm": 0.04921608342410101,
+      "learning_rate": 8.807733619763696e-07,
+      "loss": 0.0004,
+      "step": 14297
+    },
+    {
+      "epoch": 0.9214410001933363,
+      "grad_norm": 0.0911446236348484,
+      "learning_rate": 8.800572860723237e-07,
+      "loss": 0.0016,
+      "step": 14298
+    },
+    {
+      "epoch": 0.9215054456402655,
+      "grad_norm": 0.0002452915588548585,
+      "learning_rate": 8.793412101682779e-07,
+      "loss": 0.0,
+      "step": 14299
+    },
+    {
+      "epoch": 0.9215698910871947,
+      "grad_norm": 0.0007380774980053137,
+      "learning_rate": 8.786251342642321e-07,
+      "loss": 0.0,
+      "step": 14300
+    },
+    {
+      "epoch": 0.9216343365341239,
+      "grad_norm": 0.000311741694295826,
+      "learning_rate": 8.779090583601863e-07,
+      "loss": 0.0,
+      "step": 14301
+    },
+    {
+      "epoch": 0.921698781981053,
+      "grad_norm": 0.0879874067442999,
+      "learning_rate": 8.771929824561404e-07,
+      "loss": 0.001,
+      "step": 14302
+    },
+    {
+      "epoch": 0.9217632274279822,
+      "grad_norm": 0.005910103099318323,
+      "learning_rate": 8.764769065520947e-07,
+      "loss": 0.0,
+      "step": 14303
+    },
+    {
+      "epoch": 0.9218276728749114,
+      "grad_norm": 0.00033256577001777654,
+      "learning_rate": 8.757608306480488e-07,
+      "loss": 0.0,
+      "step": 14304
+    },
+    {
+      "epoch": 0.9218921183218406,
+      "grad_norm": 0.009101280903348382,
+      "learning_rate": 8.750447547440028e-07,
+      "loss": 0.0,
+      "step": 14305
+    },
+    {
+      "epoch": 0.9219565637687698,
+      "grad_norm": 0.0008341859623354918,
+      "learning_rate": 8.743286788399571e-07,
+      "loss": 0.0,
+      "step": 14306
+    },
+    {
+      "epoch": 0.922021009215699,
+      "grad_norm": 0.0005060169378157814,
+      "learning_rate": 8.736126029359112e-07,
+      "loss": 0.0,
+      "step": 14307
+    },
+    {
+      "epoch": 0.922085454662628,
+      "grad_norm": 0.2939196223695395,
+      "learning_rate": 8.728965270318654e-07,
+      "loss": 0.001,
+      "step": 14308
+    },
+    {
+      "epoch": 0.9221499001095572,
+      "grad_norm": 0.00019054155313982495,
+      "learning_rate": 8.721804511278196e-07,
+      "loss": 0.0,
+      "step": 14309
+    },
+    {
+      "epoch": 0.9222143455564864,
+      "grad_norm": 0.011363912084689533,
+      "learning_rate": 8.714643752237738e-07,
+      "loss": 0.0001,
+      "step": 14310
+    },
+    {
+      "epoch": 0.9222787910034156,
+      "grad_norm": 0.0007240625442157176,
+      "learning_rate": 8.707482993197279e-07,
+      "loss": 0.0,
+      "step": 14311
+    },
+    {
+      "epoch": 0.9223432364503448,
+      "grad_norm": 0.22208887725504373,
+      "learning_rate": 8.700322234156822e-07,
+      "loss": 0.0003,
+      "step": 14312
+    },
+    {
+      "epoch": 0.922407681897274,
+      "grad_norm": 0.013504279937124145,
+      "learning_rate": 8.693161475116363e-07,
+      "loss": 0.0,
+      "step": 14313
+    },
+    {
+      "epoch": 0.9224721273442031,
+      "grad_norm": 0.0020501837505711128,
+      "learning_rate": 8.686000716075904e-07,
+      "loss": 0.0,
+      "step": 14314
+    },
+    {
+      "epoch": 0.9225365727911323,
+      "grad_norm": 0.28737357044486633,
+      "learning_rate": 8.678839957035447e-07,
+      "loss": 0.0003,
+      "step": 14315
+    },
+    {
+      "epoch": 0.9226010182380615,
+      "grad_norm": 0.0010953903883233731,
+      "learning_rate": 8.671679197994988e-07,
+      "loss": 0.0,
+      "step": 14316
+    },
+    {
+      "epoch": 0.9226654636849907,
+      "grad_norm": 0.009895230589149193,
+      "learning_rate": 8.664518438954529e-07,
+      "loss": 0.0001,
+      "step": 14317
+    },
+    {
+      "epoch": 0.9227299091319199,
+      "grad_norm": 0.04075174082050103,
+      "learning_rate": 8.657357679914071e-07,
+      "loss": 0.0,
+      "step": 14318
+    },
+    {
+      "epoch": 0.922794354578849,
+      "grad_norm": 0.0002692212550240471,
+      "learning_rate": 8.650196920873613e-07,
+      "loss": 0.0,
+      "step": 14319
+    },
+    {
+      "epoch": 0.9228588000257781,
+      "grad_norm": 0.00048252134601037153,
+      "learning_rate": 8.643036161833155e-07,
+      "loss": 0.0,
+      "step": 14320
+    },
+    {
+      "epoch": 0.9229232454727073,
+      "grad_norm": 3.825583155027952e-05,
+      "learning_rate": 8.635875402792697e-07,
+      "loss": 0.0,
+      "step": 14321
+    },
+    {
+      "epoch": 0.9229876909196365,
+      "grad_norm": 0.0018811696074475584,
+      "learning_rate": 8.628714643752238e-07,
+      "loss": 0.0,
+      "step": 14322
+    },
+    {
+      "epoch": 0.9230521363665657,
+      "grad_norm": 0.07775464787976358,
+      "learning_rate": 8.621553884711781e-07,
+      "loss": 0.0051,
+      "step": 14323
+    },
+    {
+      "epoch": 0.9231165818134949,
+      "grad_norm": 0.058226357454030764,
+      "learning_rate": 8.614393125671322e-07,
+      "loss": 0.0001,
+      "step": 14324
+    },
+    {
+      "epoch": 0.9231810272604241,
+      "grad_norm": 0.012174789048439801,
+      "learning_rate": 8.607232366630863e-07,
+      "loss": 0.0,
+      "step": 14325
+    },
+    {
+      "epoch": 0.9232454727073532,
+      "grad_norm": 0.0017584853190584743,
+      "learning_rate": 8.600071607590406e-07,
+      "loss": 0.0,
+      "step": 14326
+    },
+    {
+      "epoch": 0.9233099181542824,
+      "grad_norm": 0.000461639247452941,
+      "learning_rate": 8.592910848549947e-07,
+      "loss": 0.0,
+      "step": 14327
+    },
+    {
+      "epoch": 0.9233743636012116,
+      "grad_norm": 0.07013433453961926,
+      "learning_rate": 8.585750089509489e-07,
+      "loss": 0.0001,
+      "step": 14328
+    },
+    {
+      "epoch": 0.9234388090481408,
+      "grad_norm": 0.00018168184062102873,
+      "learning_rate": 8.57858933046903e-07,
+      "loss": 0.0,
+      "step": 14329
+    },
+    {
+      "epoch": 0.9235032544950699,
+      "grad_norm": 0.009711667844548244,
+      "learning_rate": 8.571428571428572e-07,
+      "loss": 0.0,
+      "step": 14330
+    },
+    {
+      "epoch": 0.923567699941999,
+      "grad_norm": 0.0006406341603395895,
+      "learning_rate": 8.564267812388113e-07,
+      "loss": 0.0,
+      "step": 14331
+    },
+    {
+      "epoch": 0.9236321453889282,
+      "grad_norm": 8.876367947043951e-05,
+      "learning_rate": 8.557107053347656e-07,
+      "loss": 0.0,
+      "step": 14332
+    },
+    {
+      "epoch": 0.9236965908358574,
+      "grad_norm": 0.6359557512368327,
+      "learning_rate": 8.549946294307197e-07,
+      "loss": 0.0067,
+      "step": 14333
+    },
+    {
+      "epoch": 0.9237610362827866,
+      "grad_norm": 0.00493237461485779,
+      "learning_rate": 8.542785535266739e-07,
+      "loss": 0.0,
+      "step": 14334
+    },
+    {
+      "epoch": 0.9238254817297158,
+      "grad_norm": 0.0052610262877635415,
+      "learning_rate": 8.535624776226281e-07,
+      "loss": 0.0,
+      "step": 14335
+    },
+    {
+      "epoch": 0.923889927176645,
+      "grad_norm": 0.00011238967086745099,
+      "learning_rate": 8.528464017185823e-07,
+      "loss": 0.0,
+      "step": 14336
+    },
+    {
+      "epoch": 0.9239543726235742,
+      "grad_norm": 2.3785792728021552e-05,
+      "learning_rate": 8.521303258145364e-07,
+      "loss": 0.0,
+      "step": 14337
+    },
+    {
+      "epoch": 0.9240188180705033,
+      "grad_norm": 0.000555459237647623,
+      "learning_rate": 8.514142499104907e-07,
+      "loss": 0.0,
+      "step": 14338
+    },
+    {
+      "epoch": 0.9240832635174325,
+      "grad_norm": 0.010765654901809371,
+      "learning_rate": 8.506981740064448e-07,
+      "loss": 0.0,
+      "step": 14339
+    },
+    {
+      "epoch": 0.9241477089643617,
+      "grad_norm": 0.004778807092981911,
+      "learning_rate": 8.499820981023989e-07,
+      "loss": 0.0,
+      "step": 14340
+    },
+    {
+      "epoch": 0.9242121544112908,
+      "grad_norm": 0.005520193338135902,
+      "learning_rate": 8.492660221983531e-07,
+      "loss": 0.0,
+      "step": 14341
+    },
+    {
+      "epoch": 0.92427659985822,
+      "grad_norm": 0.3407595779824848,
+      "learning_rate": 8.485499462943072e-07,
+      "loss": 0.0006,
+      "step": 14342
+    },
+    {
+      "epoch": 0.9243410453051492,
+      "grad_norm": 4.109653675093666e-05,
+      "learning_rate": 8.478338703902614e-07,
+      "loss": 0.0,
+      "step": 14343
+    },
+    {
+      "epoch": 0.9244054907520783,
+      "grad_norm": 0.05114493215860188,
+      "learning_rate": 8.471177944862156e-07,
+      "loss": 0.0001,
+      "step": 14344
+    },
+    {
+      "epoch": 0.9244699361990075,
+      "grad_norm": 0.004551564714074149,
+      "learning_rate": 8.464017185821698e-07,
+      "loss": 0.0,
+      "step": 14345
+    },
+    {
+      "epoch": 0.9245343816459367,
+      "grad_norm": 0.0021971180571567555,
+      "learning_rate": 8.456856426781239e-07,
+      "loss": 0.0,
+      "step": 14346
+    },
+    {
+      "epoch": 0.9245988270928659,
+      "grad_norm": 0.0005614679583030613,
+      "learning_rate": 8.449695667740782e-07,
+      "loss": 0.0,
+      "step": 14347
+    },
+    {
+      "epoch": 0.9246632725397951,
+      "grad_norm": 0.3467760408850226,
+      "learning_rate": 8.442534908700323e-07,
+      "loss": 0.0017,
+      "step": 14348
+    },
+    {
+      "epoch": 0.9247277179867243,
+      "grad_norm": 0.08151521610210832,
+      "learning_rate": 8.435374149659864e-07,
+      "loss": 0.0008,
+      "step": 14349
+    },
+    {
+      "epoch": 0.9247921634336534,
+      "grad_norm": 0.004019518333800433,
+      "learning_rate": 8.428213390619407e-07,
+      "loss": 0.0,
+      "step": 14350
+    },
+    {
+      "epoch": 0.9248566088805826,
+      "grad_norm": 0.07447157949814238,
+      "learning_rate": 8.421052631578948e-07,
+      "loss": 0.0001,
+      "step": 14351
+    },
+    {
+      "epoch": 0.9249210543275118,
+      "grad_norm": 0.014641115226128458,
+      "learning_rate": 8.41389187253849e-07,
+      "loss": 0.0,
+      "step": 14352
+    },
+    {
+      "epoch": 0.9249854997744409,
+      "grad_norm": 0.3754926862321877,
+      "learning_rate": 8.406731113498031e-07,
+      "loss": 0.0031,
+      "step": 14353
+    },
+    {
+      "epoch": 0.9250499452213701,
+      "grad_norm": 0.0006381091658297029,
+      "learning_rate": 8.399570354457573e-07,
+      "loss": 0.0,
+      "step": 14354
+    },
+    {
+      "epoch": 0.9251143906682993,
+      "grad_norm": 0.0007651448783783858,
+      "learning_rate": 8.392409595417115e-07,
+      "loss": 0.0,
+      "step": 14355
+    },
+    {
+      "epoch": 0.9251788361152284,
+      "grad_norm": 0.010261725668583554,
+      "learning_rate": 8.385248836376657e-07,
+      "loss": 0.0,
+      "step": 14356
+    },
+    {
+      "epoch": 0.9252432815621576,
+      "grad_norm": 0.004215161724131631,
+      "learning_rate": 8.378088077336198e-07,
+      "loss": 0.0,
+      "step": 14357
+    },
+    {
+      "epoch": 0.9253077270090868,
+      "grad_norm": 0.010846641072357954,
+      "learning_rate": 8.370927318295741e-07,
+      "loss": 0.0,
+      "step": 14358
+    },
+    {
+      "epoch": 0.925372172456016,
+      "grad_norm": 0.0036661152239055466,
+      "learning_rate": 8.363766559255282e-07,
+      "loss": 0.0,
+      "step": 14359
+    },
+    {
+      "epoch": 0.9254366179029452,
+      "grad_norm": 0.2835533081727429,
+      "learning_rate": 8.356605800214823e-07,
+      "loss": 0.0023,
+      "step": 14360
+    },
+    {
+      "epoch": 0.9255010633498744,
+      "grad_norm": 0.0005097153361995352,
+      "learning_rate": 8.349445041174366e-07,
+      "loss": 0.0,
+      "step": 14361
+    },
+    {
+      "epoch": 0.9255655087968035,
+      "grad_norm": 0.0035745185335236363,
+      "learning_rate": 8.342284282133907e-07,
+      "loss": 0.0,
+      "step": 14362
+    },
+    {
+      "epoch": 0.9256299542437327,
+      "grad_norm": 0.0022912225414757144,
+      "learning_rate": 8.335123523093449e-07,
+      "loss": 0.0,
+      "step": 14363
+    },
+    {
+      "epoch": 0.9256943996906618,
+      "grad_norm": 0.006479848061025811,
+      "learning_rate": 8.327962764052991e-07,
+      "loss": 0.0,
+      "step": 14364
+    },
+    {
+      "epoch": 0.925758845137591,
+      "grad_norm": 0.04159580182992835,
+      "learning_rate": 8.320802005012532e-07,
+      "loss": 0.0001,
+      "step": 14365
+    },
+    {
+      "epoch": 0.9258232905845202,
+      "grad_norm": 0.003863711422317911,
+      "learning_rate": 8.313641245972073e-07,
+      "loss": 0.0,
+      "step": 14366
+    },
+    {
+      "epoch": 0.9258877360314494,
+      "grad_norm": 0.00010527621767379967,
+      "learning_rate": 8.306480486931616e-07,
+      "loss": 0.0,
+      "step": 14367
+    },
+    {
+      "epoch": 0.9259521814783785,
+      "grad_norm": 0.0008107817368530498,
+      "learning_rate": 8.299319727891157e-07,
+      "loss": 0.0,
+      "step": 14368
+    },
+    {
+      "epoch": 0.9260166269253077,
+      "grad_norm": 0.39974754607215024,
+      "learning_rate": 8.292158968850698e-07,
+      "loss": 0.0026,
+      "step": 14369
+    },
+    {
+      "epoch": 0.9260810723722369,
+      "grad_norm": 0.0003916170677837437,
+      "learning_rate": 8.284998209810241e-07,
+      "loss": 0.0,
+      "step": 14370
+    },
+    {
+      "epoch": 0.9261455178191661,
+      "grad_norm": 0.00032176926961688867,
+      "learning_rate": 8.277837450769782e-07,
+      "loss": 0.0,
+      "step": 14371
+    },
+    {
+      "epoch": 0.9262099632660953,
+      "grad_norm": 0.004748618602019491,
+      "learning_rate": 8.270676691729324e-07,
+      "loss": 0.0,
+      "step": 14372
+    },
+    {
+      "epoch": 0.9262744087130245,
+      "grad_norm": 0.0005727854899369785,
+      "learning_rate": 8.263515932688866e-07,
+      "loss": 0.0,
+      "step": 14373
+    },
+    {
+      "epoch": 0.9263388541599537,
+      "grad_norm": 9.207616447138975e-05,
+      "learning_rate": 8.256355173648408e-07,
+      "loss": 0.0,
+      "step": 14374
+    },
+    {
+      "epoch": 0.9264032996068827,
+      "grad_norm": 0.00040550139835061086,
+      "learning_rate": 8.249194414607949e-07,
+      "loss": 0.0,
+      "step": 14375
+    },
+    {
+      "epoch": 0.9264677450538119,
+      "grad_norm": 0.0011182196861263627,
+      "learning_rate": 8.242033655567492e-07,
+      "loss": 0.0,
+      "step": 14376
+    },
+    {
+      "epoch": 0.9265321905007411,
+      "grad_norm": 0.002358272684080987,
+      "learning_rate": 8.234872896527032e-07,
+      "loss": 0.0,
+      "step": 14377
+    },
+    {
+      "epoch": 0.9265966359476703,
+      "grad_norm": 0.12104582607336277,
+      "learning_rate": 8.227712137486573e-07,
+      "loss": 0.0001,
+      "step": 14378
+    },
+    {
+      "epoch": 0.9266610813945995,
+      "grad_norm": 0.007726797377316758,
+      "learning_rate": 8.220551378446116e-07,
+      "loss": 0.0,
+      "step": 14379
+    },
+    {
+      "epoch": 0.9267255268415286,
+      "grad_norm": 0.008248464069637097,
+      "learning_rate": 8.213390619405657e-07,
+      "loss": 0.0,
+      "step": 14380
+    },
+    {
+      "epoch": 0.9267899722884578,
+      "grad_norm": 0.0001702704616484478,
+      "learning_rate": 8.206229860365199e-07,
+      "loss": 0.0,
+      "step": 14381
+    },
+    {
+      "epoch": 0.926854417735387,
+      "grad_norm": 0.000260795053944009,
+      "learning_rate": 8.199069101324741e-07,
+      "loss": 0.0,
+      "step": 14382
+    },
+    {
+      "epoch": 0.9269188631823162,
+      "grad_norm": 0.0007138246285888987,
+      "learning_rate": 8.191908342284283e-07,
+      "loss": 0.0,
+      "step": 14383
+    },
+    {
+      "epoch": 0.9269833086292454,
+      "grad_norm": 0.00015103320664588784,
+      "learning_rate": 8.184747583243824e-07,
+      "loss": 0.0,
+      "step": 14384
+    },
+    {
+      "epoch": 0.9270477540761746,
+      "grad_norm": 0.6980205264079195,
+      "learning_rate": 8.177586824203367e-07,
+      "loss": 0.0055,
+      "step": 14385
+    },
+    {
+      "epoch": 0.9271121995231036,
+      "grad_norm": 0.06006724016241198,
+      "learning_rate": 8.170426065162908e-07,
+      "loss": 0.0001,
+      "step": 14386
+    },
+    {
+      "epoch": 0.9271766449700328,
+      "grad_norm": 0.1601703742242303,
+      "learning_rate": 8.163265306122449e-07,
+      "loss": 0.0037,
+      "step": 14387
+    },
+    {
+      "epoch": 0.927241090416962,
+      "grad_norm": 0.0006902069417974614,
+      "learning_rate": 8.156104547081992e-07,
+      "loss": 0.0,
+      "step": 14388
+    },
+    {
+      "epoch": 0.9273055358638912,
+      "grad_norm": 0.001034304148302098,
+      "learning_rate": 8.148943788041532e-07,
+      "loss": 0.0,
+      "step": 14389
+    },
+    {
+      "epoch": 0.9273699813108204,
+      "grad_norm": 0.0006792645530657716,
+      "learning_rate": 8.141783029001074e-07,
+      "loss": 0.0,
+      "step": 14390
+    },
+    {
+      "epoch": 0.9274344267577496,
+      "grad_norm": 0.006263223817885295,
+      "learning_rate": 8.134622269960616e-07,
+      "loss": 0.0,
+      "step": 14391
+    },
+    {
+      "epoch": 0.9274988722046787,
+      "grad_norm": 0.0018722316210985034,
+      "learning_rate": 8.127461510920158e-07,
+      "loss": 0.0,
+      "step": 14392
+    },
+    {
+      "epoch": 0.9275633176516079,
+      "grad_norm": 0.000974320736434331,
+      "learning_rate": 8.1203007518797e-07,
+      "loss": 0.0,
+      "step": 14393
+    },
+    {
+      "epoch": 0.9276277630985371,
+      "grad_norm": 0.0003239978115912819,
+      "learning_rate": 8.113139992839242e-07,
+      "loss": 0.0,
+      "step": 14394
+    },
+    {
+      "epoch": 0.9276922085454663,
+      "grad_norm": 0.0013150495822469358,
+      "learning_rate": 8.105979233798783e-07,
+      "loss": 0.0,
+      "step": 14395
+    },
+    {
+      "epoch": 0.9277566539923955,
+      "grad_norm": 0.015568682946537292,
+      "learning_rate": 8.098818474758326e-07,
+      "loss": 0.0,
+      "step": 14396
+    },
+    {
+      "epoch": 0.9278210994393246,
+      "grad_norm": 0.059350822962951,
+      "learning_rate": 8.091657715717867e-07,
+      "loss": 0.0,
+      "step": 14397
+    },
+    {
+      "epoch": 0.9278855448862537,
+      "grad_norm": 0.019161179937591782,
+      "learning_rate": 8.084496956677408e-07,
+      "loss": 0.0,
+      "step": 14398
+    },
+    {
+      "epoch": 0.9279499903331829,
+      "grad_norm": 0.0008154856476937402,
+      "learning_rate": 8.077336197636951e-07,
+      "loss": 0.0,
+      "step": 14399
+    },
+    {
+      "epoch": 0.9280144357801121,
+      "grad_norm": 0.0017092046185970635,
+      "learning_rate": 8.070175438596491e-07,
+      "loss": 0.0,
+      "step": 14400
+    },
+    {
+      "epoch": 0.9280788812270413,
+      "grad_norm": 0.0223845070364754,
+      "learning_rate": 8.063014679556033e-07,
+      "loss": 0.0001,
+      "step": 14401
+    },
+    {
+      "epoch": 0.9281433266739705,
+      "grad_norm": 0.08140399362746532,
+      "learning_rate": 8.055853920515575e-07,
+      "loss": 0.0007,
+      "step": 14402
+    },
+    {
+      "epoch": 0.9282077721208997,
+      "grad_norm": 0.007940254101373572,
+      "learning_rate": 8.048693161475117e-07,
+      "loss": 0.0,
+      "step": 14403
+    },
+    {
+      "epoch": 0.9282722175678289,
+      "grad_norm": 0.03842277464554137,
+      "learning_rate": 8.041532402434658e-07,
+      "loss": 0.0,
+      "step": 14404
+    },
+    {
+      "epoch": 0.928336663014758,
+      "grad_norm": 0.0015349968048687594,
+      "learning_rate": 8.034371643394201e-07,
+      "loss": 0.0,
+      "step": 14405
+    },
+    {
+      "epoch": 0.9284011084616872,
+      "grad_norm": 0.7215145680998309,
+      "learning_rate": 8.027210884353742e-07,
+      "loss": 0.0047,
+      "step": 14406
+    },
+    {
+      "epoch": 0.9284655539086164,
+      "grad_norm": 0.14605131986601827,
+      "learning_rate": 8.020050125313284e-07,
+      "loss": 0.0,
+      "step": 14407
+    },
+    {
+      "epoch": 0.9285299993555455,
+      "grad_norm": 0.045908688354199964,
+      "learning_rate": 8.012889366272826e-07,
+      "loss": 0.0001,
+      "step": 14408
+    },
+    {
+      "epoch": 0.9285944448024747,
+      "grad_norm": 0.002024083815106119,
+      "learning_rate": 8.005728607232368e-07,
+      "loss": 0.0,
+      "step": 14409
+    },
+    {
+      "epoch": 0.9286588902494038,
+      "grad_norm": 0.002112488711979832,
+      "learning_rate": 7.998567848191909e-07,
+      "loss": 0.0,
+      "step": 14410
+    },
+    {
+      "epoch": 0.928723335696333,
+      "grad_norm": 0.008628272653972625,
+      "learning_rate": 7.991407089151452e-07,
+      "loss": 0.0,
+      "step": 14411
+    },
+    {
+      "epoch": 0.9287877811432622,
+      "grad_norm": 0.09060448977613139,
+      "learning_rate": 7.984246330110993e-07,
+      "loss": 0.0002,
+      "step": 14412
+    },
+    {
+      "epoch": 0.9288522265901914,
+      "grad_norm": 0.0005132101275217304,
+      "learning_rate": 7.977085571070533e-07,
+      "loss": 0.0,
+      "step": 14413
+    },
+    {
+      "epoch": 0.9289166720371206,
+      "grad_norm": 0.1643300826197386,
+      "learning_rate": 7.969924812030076e-07,
+      "loss": 0.0003,
+      "step": 14414
+    },
+    {
+      "epoch": 0.9289811174840498,
+      "grad_norm": 0.0016093713965732953,
+      "learning_rate": 7.962764052989617e-07,
+      "loss": 0.0015,
+      "step": 14415
+    },
+    {
+      "epoch": 0.929045562930979,
+      "grad_norm": 0.21922135659460637,
+      "learning_rate": 7.955603293949159e-07,
+      "loss": 0.0008,
+      "step": 14416
+    },
+    {
+      "epoch": 0.9291100083779081,
+      "grad_norm": 0.0008585809892495993,
+      "learning_rate": 7.948442534908701e-07,
+      "loss": 0.0,
+      "step": 14417
+    },
+    {
+      "epoch": 0.9291744538248373,
+      "grad_norm": 0.009192170655711953,
+      "learning_rate": 7.941281775868243e-07,
+      "loss": 0.0,
+      "step": 14418
+    },
+    {
+      "epoch": 0.9292388992717664,
+      "grad_norm": 0.25046316098969706,
+      "learning_rate": 7.934121016827784e-07,
+      "loss": 0.0013,
+      "step": 14419
+    },
+    {
+      "epoch": 0.9293033447186956,
+      "grad_norm": 0.0051350095650668295,
+      "learning_rate": 7.926960257787327e-07,
+      "loss": 0.0,
+      "step": 14420
+    },
+    {
+      "epoch": 0.9293677901656248,
+      "grad_norm": 0.0026750066483972193,
+      "learning_rate": 7.919799498746868e-07,
+      "loss": 0.0,
+      "step": 14421
+    },
+    {
+      "epoch": 0.929432235612554,
+      "grad_norm": 0.004276538716577033,
+      "learning_rate": 7.912638739706409e-07,
+      "loss": 0.0,
+      "step": 14422
+    },
+    {
+      "epoch": 0.9294966810594831,
+      "grad_norm": 0.3635859288520741,
+      "learning_rate": 7.905477980665952e-07,
+      "loss": 0.0028,
+      "step": 14423
+    },
+    {
+      "epoch": 0.9295611265064123,
+      "grad_norm": 0.01937078856667039,
+      "learning_rate": 7.898317221625492e-07,
+      "loss": 0.0,
+      "step": 14424
+    },
+    {
+      "epoch": 0.9296255719533415,
+      "grad_norm": 0.001685193181962873,
+      "learning_rate": 7.891156462585034e-07,
+      "loss": 0.0,
+      "step": 14425
+    },
+    {
+      "epoch": 0.9296900174002707,
+      "grad_norm": 0.0005174008097750863,
+      "learning_rate": 7.883995703544576e-07,
+      "loss": 0.0,
+      "step": 14426
+    },
+    {
+      "epoch": 0.9297544628471999,
+      "grad_norm": 3.9323004013979845e-05,
+      "learning_rate": 7.876834944504118e-07,
+      "loss": 0.0,
+      "step": 14427
+    },
+    {
+      "epoch": 0.929818908294129,
+      "grad_norm": 0.23817010538682193,
+      "learning_rate": 7.86967418546366e-07,
+      "loss": 0.0003,
+      "step": 14428
+    },
+    {
+      "epoch": 0.9298833537410582,
+      "grad_norm": 0.014149205828395605,
+      "learning_rate": 7.862513426423202e-07,
+      "loss": 0.0,
+      "step": 14429
+    },
+    {
+      "epoch": 0.9299477991879874,
+      "grad_norm": 0.0005149379627849445,
+      "learning_rate": 7.855352667382743e-07,
+      "loss": 0.0,
+      "step": 14430
+    },
+    {
+      "epoch": 0.9300122446349165,
+      "grad_norm": 0.31164079805275147,
+      "learning_rate": 7.848191908342286e-07,
+      "loss": 0.001,
+      "step": 14431
+    },
+    {
+      "epoch": 0.9300766900818457,
+      "grad_norm": 0.027360878841657936,
+      "learning_rate": 7.841031149301827e-07,
+      "loss": 0.0,
+      "step": 14432
+    },
+    {
+      "epoch": 0.9301411355287749,
+      "grad_norm": 0.0019418651000683484,
+      "learning_rate": 7.833870390261368e-07,
+      "loss": 0.0,
+      "step": 14433
+    },
+    {
+      "epoch": 0.930205580975704,
+      "grad_norm": 0.00013151979960604563,
+      "learning_rate": 7.826709631220911e-07,
+      "loss": 0.0,
+      "step": 14434
+    },
+    {
+      "epoch": 0.9302700264226332,
+      "grad_norm": 0.0005116016903126348,
+      "learning_rate": 7.819548872180452e-07,
+      "loss": 0.0,
+      "step": 14435
+    },
+    {
+      "epoch": 0.9303344718695624,
+      "grad_norm": 0.003360318785380491,
+      "learning_rate": 7.812388113139993e-07,
+      "loss": 0.0,
+      "step": 14436
+    },
+    {
+      "epoch": 0.9303989173164916,
+      "grad_norm": 0.005443713674886183,
+      "learning_rate": 7.805227354099535e-07,
+      "loss": 0.0,
+      "step": 14437
+    },
+    {
+      "epoch": 0.9304633627634208,
+      "grad_norm": 0.006804981826999122,
+      "learning_rate": 7.798066595059077e-07,
+      "loss": 0.0,
+      "step": 14438
+    },
+    {
+      "epoch": 0.93052780821035,
+      "grad_norm": 0.004543146095431098,
+      "learning_rate": 7.790905836018618e-07,
+      "loss": 0.0,
+      "step": 14439
+    },
+    {
+      "epoch": 0.9305922536572792,
+      "grad_norm": 0.004907974032332147,
+      "learning_rate": 7.783745076978161e-07,
+      "loss": 0.0,
+      "step": 14440
+    },
+    {
+      "epoch": 0.9306566991042083,
+      "grad_norm": 0.007826499670387036,
+      "learning_rate": 7.776584317937702e-07,
+      "loss": 0.0,
+      "step": 14441
+    },
+    {
+      "epoch": 0.9307211445511374,
+      "grad_norm": 0.0004948724082326526,
+      "learning_rate": 7.769423558897243e-07,
+      "loss": 0.0,
+      "step": 14442
+    },
+    {
+      "epoch": 0.9307855899980666,
+      "grad_norm": 0.008689859857442889,
+      "learning_rate": 7.762262799856786e-07,
+      "loss": 0.0,
+      "step": 14443
+    },
+    {
+      "epoch": 0.9308500354449958,
+      "grad_norm": 0.003751375962738023,
+      "learning_rate": 7.755102040816327e-07,
+      "loss": 0.0,
+      "step": 14444
+    },
+    {
+      "epoch": 0.930914480891925,
+      "grad_norm": 0.00013363970816859745,
+      "learning_rate": 7.747941281775869e-07,
+      "loss": 0.0,
+      "step": 14445
+    },
+    {
+      "epoch": 0.9309789263388542,
+      "grad_norm": 0.0018097887452774059,
+      "learning_rate": 7.740780522735411e-07,
+      "loss": 0.0,
+      "step": 14446
+    },
+    {
+      "epoch": 0.9310433717857833,
+      "grad_norm": 0.0018097887452774059,
+      "learning_rate": 7.740780522735411e-07,
+      "loss": 0.0306,
+      "step": 14447
+    },
+    {
+      "epoch": 0.9311078172327125,
+      "grad_norm": 0.743557799372613,
+      "learning_rate": 7.733619763694953e-07,
+      "loss": 0.0037,
+      "step": 14448
+    },
+    {
+      "epoch": 0.9311722626796417,
+      "grad_norm": 0.003228825487809219,
+      "learning_rate": 7.726459004654493e-07,
+      "loss": 0.0,
+      "step": 14449
+    },
+    {
+      "epoch": 0.9312367081265709,
+      "grad_norm": 0.0332350349660026,
+      "learning_rate": 7.719298245614036e-07,
+      "loss": 0.0001,
+      "step": 14450
+    },
+    {
+      "epoch": 0.9313011535735001,
+      "grad_norm": 0.004896196076213735,
+      "learning_rate": 7.712137486573577e-07,
+      "loss": 0.0,
+      "step": 14451
+    },
+    {
+      "epoch": 0.9313655990204293,
+      "grad_norm": 6.021259437749865e-05,
+      "learning_rate": 7.704976727533118e-07,
+      "loss": 0.0,
+      "step": 14452
+    },
+    {
+      "epoch": 0.9314300444673583,
+      "grad_norm": 0.003628188229278082,
+      "learning_rate": 7.697815968492661e-07,
+      "loss": 0.0,
+      "step": 14453
+    },
+    {
+      "epoch": 0.9314944899142875,
+      "grad_norm": 0.0017509759768460378,
+      "learning_rate": 7.690655209452202e-07,
+      "loss": 0.0,
+      "step": 14454
+    },
+    {
+      "epoch": 0.9315589353612167,
+      "grad_norm": 0.6015600476896165,
+      "learning_rate": 7.683494450411744e-07,
+      "loss": 0.0004,
+      "step": 14455
+    },
+    {
+      "epoch": 0.9316233808081459,
+      "grad_norm": 0.0020188174805083393,
+      "learning_rate": 7.676333691371286e-07,
+      "loss": 0.0,
+      "step": 14456
+    },
+    {
+      "epoch": 0.9316878262550751,
+      "grad_norm": 0.014562413827033343,
+      "learning_rate": 7.669172932330828e-07,
+      "loss": 0.0,
+      "step": 14457
+    },
+    {
+      "epoch": 0.9317522717020043,
+      "grad_norm": 0.0032687377442210427,
+      "learning_rate": 7.662012173290369e-07,
+      "loss": 0.0,
+      "step": 14458
+    },
+    {
+      "epoch": 0.9318167171489334,
+      "grad_norm": 3.5950292092738695e-05,
+      "learning_rate": 7.654851414249912e-07,
+      "loss": 0.0,
+      "step": 14459
+    },
+    {
+      "epoch": 0.9318811625958626,
+      "grad_norm": 0.007673081489013223,
+      "learning_rate": 7.647690655209453e-07,
+      "loss": 0.0,
+      "step": 14460
+    },
+    {
+      "epoch": 0.9319456080427918,
+      "grad_norm": 0.2343050216474507,
+      "learning_rate": 7.640529896168993e-07,
+      "loss": 0.0024,
+      "step": 14461
+    },
+    {
+      "epoch": 0.932010053489721,
+      "grad_norm": 0.0435925417565156,
+      "learning_rate": 7.633369137128536e-07,
+      "loss": 0.0005,
+      "step": 14462
+    },
+    {
+      "epoch": 0.9320744989366502,
+      "grad_norm": 0.20216371414031317,
+      "learning_rate": 7.626208378088077e-07,
+      "loss": 0.0003,
+      "step": 14463
+    },
+    {
+      "epoch": 0.9321389443835792,
+      "grad_norm": 0.0002489122222914264,
+      "learning_rate": 7.61904761904762e-07,
+      "loss": 0.0,
+      "step": 14464
+    },
+    {
+      "epoch": 0.9322033898305084,
+      "grad_norm": 0.00018979941206031793,
+      "learning_rate": 7.611886860007161e-07,
+      "loss": 0.0,
+      "step": 14465
+    },
+    {
+      "epoch": 0.9322678352774376,
+      "grad_norm": 0.0036174353999676827,
+      "learning_rate": 7.604726100966703e-07,
+      "loss": 0.0,
+      "step": 14466
+    },
+    {
+      "epoch": 0.9323322807243668,
+      "grad_norm": 0.00010448488517973264,
+      "learning_rate": 7.597565341926245e-07,
+      "loss": 0.0,
+      "step": 14467
+    },
+    {
+      "epoch": 0.932396726171296,
+      "grad_norm": 0.06316045030658364,
+      "learning_rate": 7.590404582885787e-07,
+      "loss": 0.0004,
+      "step": 14468
+    },
+    {
+      "epoch": 0.9324611716182252,
+      "grad_norm": 0.5535720570292619,
+      "learning_rate": 7.583243823845328e-07,
+      "loss": 0.0008,
+      "step": 14469
+    },
+    {
+      "epoch": 0.9325256170651544,
+      "grad_norm": 0.009089972948693839,
+      "learning_rate": 7.576083064804871e-07,
+      "loss": 0.0,
+      "step": 14470
+    },
+    {
+      "epoch": 0.9325900625120835,
+      "grad_norm": 0.009528753366977356,
+      "learning_rate": 7.568922305764412e-07,
+      "loss": 0.0,
+      "step": 14471
+    },
+    {
+      "epoch": 0.9326545079590127,
+      "grad_norm": 0.08273048072921844,
+      "learning_rate": 7.561761546723953e-07,
+      "loss": 0.0001,
+      "step": 14472
+    },
+    {
+      "epoch": 0.9327189534059419,
+      "grad_norm": 0.0024652483846048165,
+      "learning_rate": 7.554600787683496e-07,
+      "loss": 0.0,
+      "step": 14473
+    },
+    {
+      "epoch": 0.9327833988528711,
+      "grad_norm": 0.04121828109839269,
+      "learning_rate": 7.547440028643036e-07,
+      "loss": 0.0,
+      "step": 14474
+    },
+    {
+      "epoch": 0.9328478442998002,
+      "grad_norm": 0.14005019012037867,
+      "learning_rate": 7.540279269602578e-07,
+      "loss": 0.0002,
+      "step": 14475
+    },
+    {
+      "epoch": 0.9329122897467294,
+      "grad_norm": 0.020674897633846747,
+      "learning_rate": 7.53311851056212e-07,
+      "loss": 0.0,
+      "step": 14476
+    },
+    {
+      "epoch": 0.9329767351936585,
+      "grad_norm": 0.39202941200795505,
+      "learning_rate": 7.525957751521662e-07,
+      "loss": 0.0017,
+      "step": 14477
+    },
+    {
+      "epoch": 0.9330411806405877,
+      "grad_norm": 0.0019212015864760914,
+      "learning_rate": 7.518796992481203e-07,
+      "loss": 0.0,
+      "step": 14478
+    },
+    {
+      "epoch": 0.9331056260875169,
+      "grad_norm": 0.058421093156673494,
+      "learning_rate": 7.511636233440746e-07,
+      "loss": 0.0001,
+      "step": 14479
+    },
+    {
+      "epoch": 0.9331700715344461,
+      "grad_norm": 2.9589337738041364,
+      "learning_rate": 7.504475474400287e-07,
+      "loss": 0.019,
+      "step": 14480
+    },
+    {
+      "epoch": 0.9332345169813753,
+      "grad_norm": 0.0026238409074990647,
+      "learning_rate": 7.497314715359829e-07,
+      "loss": 0.0,
+      "step": 14481
+    },
+    {
+      "epoch": 0.9332989624283045,
+      "grad_norm": 0.00039909083865957056,
+      "learning_rate": 7.490153956319371e-07,
+      "loss": 0.0,
+      "step": 14482
+    },
+    {
+      "epoch": 0.9333634078752336,
+      "grad_norm": 0.061526487668434125,
+      "learning_rate": 7.482993197278913e-07,
+      "loss": 0.0002,
+      "step": 14483
+    },
+    {
+      "epoch": 0.9334278533221628,
+      "grad_norm": 0.0009454930574369729,
+      "learning_rate": 7.475832438238454e-07,
+      "loss": 0.0,
+      "step": 14484
+    },
+    {
+      "epoch": 0.933492298769092,
+      "grad_norm": 0.008536048096211434,
+      "learning_rate": 7.468671679197997e-07,
+      "loss": 0.0001,
+      "step": 14485
+    },
+    {
+      "epoch": 0.9335567442160211,
+      "grad_norm": 0.01352383029196929,
+      "learning_rate": 7.461510920157537e-07,
+      "loss": 0.0,
+      "step": 14486
+    },
+    {
+      "epoch": 0.9336211896629503,
+      "grad_norm": 5.9316654650585815e-05,
+      "learning_rate": 7.454350161117078e-07,
+      "loss": 0.0,
+      "step": 14487
+    },
+    {
+      "epoch": 0.9336856351098795,
+      "grad_norm": 0.00803302243530583,
+      "learning_rate": 7.447189402076621e-07,
+      "loss": 0.0,
+      "step": 14488
+    },
+    {
+      "epoch": 0.9337500805568086,
+      "grad_norm": 0.0003945896452044587,
+      "learning_rate": 7.440028643036162e-07,
+      "loss": 0.0,
+      "step": 14489
+    },
+    {
+      "epoch": 0.9338145260037378,
+      "grad_norm": 0.030083509488064174,
+      "learning_rate": 7.432867883995704e-07,
+      "loss": 0.0,
+      "step": 14490
+    },
+    {
+      "epoch": 0.933878971450667,
+      "grad_norm": 0.00690686336390923,
+      "learning_rate": 7.425707124955246e-07,
+      "loss": 0.0,
+      "step": 14491
+    },
+    {
+      "epoch": 0.9339434168975962,
+      "grad_norm": 0.005939841642474873,
+      "learning_rate": 7.418546365914788e-07,
+      "loss": 0.0,
+      "step": 14492
+    },
+    {
+      "epoch": 0.9340078623445254,
+      "grad_norm": 0.007773213962894371,
+      "learning_rate": 7.411385606874329e-07,
+      "loss": 0.0,
+      "step": 14493
+    },
+    {
+      "epoch": 0.9340723077914546,
+      "grad_norm": 0.015595200460516845,
+      "learning_rate": 7.404224847833872e-07,
+      "loss": 0.0,
+      "step": 14494
+    },
+    {
+      "epoch": 0.9341367532383837,
+      "grad_norm": 0.0009196707970691353,
+      "learning_rate": 7.397064088793413e-07,
+      "loss": 0.0,
+      "step": 14495
+    },
+    {
+      "epoch": 0.9342011986853129,
+      "grad_norm": 0.25542377927708176,
+      "learning_rate": 7.389903329752953e-07,
+      "loss": 0.0011,
+      "step": 14496
+    },
+    {
+      "epoch": 0.9342656441322421,
+      "grad_norm": 0.006353587293436875,
+      "learning_rate": 7.382742570712496e-07,
+      "loss": 0.0,
+      "step": 14497
+    },
+    {
+      "epoch": 0.9343300895791712,
+      "grad_norm": 0.13289643888858305,
+      "learning_rate": 7.375581811672037e-07,
+      "loss": 0.0002,
+      "step": 14498
+    },
+    {
+      "epoch": 0.9343945350261004,
+      "grad_norm": 0.002705438309291519,
+      "learning_rate": 7.368421052631579e-07,
+      "loss": 0.0,
+      "step": 14499
+    },
+    {
+      "epoch": 0.9344589804730296,
+      "grad_norm": 0.009522138093246023,
+      "learning_rate": 7.361260293591121e-07,
+      "loss": 0.0,
+      "step": 14500
+    },
+    {
+      "epoch": 0.9345234259199587,
+      "grad_norm": 9.089710282497509e-05,
+      "learning_rate": 7.354099534550663e-07,
+      "loss": 0.0,
+      "step": 14501
+    },
+    {
+      "epoch": 0.9345878713668879,
+      "grad_norm": 0.0021077175463139405,
+      "learning_rate": 7.346938775510205e-07,
+      "loss": 0.0,
+      "step": 14502
+    },
+    {
+      "epoch": 0.9346523168138171,
+      "grad_norm": 0.0025995208788616095,
+      "learning_rate": 7.339778016469747e-07,
+      "loss": 0.0,
+      "step": 14503
+    },
+    {
+      "epoch": 0.9347167622607463,
+      "grad_norm": 0.004934568402119453,
+      "learning_rate": 7.332617257429288e-07,
+      "loss": 0.0001,
+      "step": 14504
+    },
+    {
+      "epoch": 0.9347812077076755,
+      "grad_norm": 0.0359308569018603,
+      "learning_rate": 7.325456498388831e-07,
+      "loss": 0.0001,
+      "step": 14505
+    },
+    {
+      "epoch": 0.9348456531546047,
+      "grad_norm": 0.00332544182917945,
+      "learning_rate": 7.318295739348372e-07,
+      "loss": 0.0,
+      "step": 14506
+    },
+    {
+      "epoch": 0.9349100986015338,
+      "grad_norm": 0.02133744018917271,
+      "learning_rate": 7.311134980307913e-07,
+      "loss": 0.0002,
+      "step": 14507
+    },
+    {
+      "epoch": 0.934974544048463,
+      "grad_norm": 0.0002502341843614646,
+      "learning_rate": 7.303974221267456e-07,
+      "loss": 0.0,
+      "step": 14508
+    },
+    {
+      "epoch": 0.9350389894953921,
+      "grad_norm": 0.14950297708202429,
+      "learning_rate": 7.296813462226996e-07,
+      "loss": 0.0018,
+      "step": 14509
+    },
+    {
+      "epoch": 0.9351034349423213,
+      "grad_norm": 0.009830572551912385,
+      "learning_rate": 7.289652703186538e-07,
+      "loss": 0.0,
+      "step": 14510
+    },
+    {
+      "epoch": 0.9351678803892505,
+      "grad_norm": 0.006458090662713216,
+      "learning_rate": 7.28249194414608e-07,
+      "loss": 0.0,
+      "step": 14511
+    },
+    {
+      "epoch": 0.9352323258361797,
+      "grad_norm": 0.03246659776362317,
+      "learning_rate": 7.275331185105622e-07,
+      "loss": 0.0,
+      "step": 14512
+    },
+    {
+      "epoch": 0.9352967712831088,
+      "grad_norm": 0.0006368826106888853,
+      "learning_rate": 7.268170426065163e-07,
+      "loss": 0.0,
+      "step": 14513
+    },
+    {
+      "epoch": 0.935361216730038,
+      "grad_norm": 0.19962360346299066,
+      "learning_rate": 7.261009667024706e-07,
+      "loss": 0.0023,
+      "step": 14514
+    },
+    {
+      "epoch": 0.9354256621769672,
+      "grad_norm": 0.0033006617114257246,
+      "learning_rate": 7.253848907984247e-07,
+      "loss": 0.0,
+      "step": 14515
+    },
+    {
+      "epoch": 0.9354901076238964,
+      "grad_norm": 1.7566595243507924e-05,
+      "learning_rate": 7.246688148943788e-07,
+      "loss": 0.0,
+      "step": 14516
+    },
+    {
+      "epoch": 0.9355545530708256,
+      "grad_norm": 0.00010102185496019302,
+      "learning_rate": 7.239527389903331e-07,
+      "loss": 0.0,
+      "step": 14517
+    },
+    {
+      "epoch": 0.9356189985177548,
+      "grad_norm": 0.012268373618383493,
+      "learning_rate": 7.232366630862872e-07,
+      "loss": 0.0,
+      "step": 14518
+    },
+    {
+      "epoch": 0.935683443964684,
+      "grad_norm": 0.005793788340712795,
+      "learning_rate": 7.225205871822414e-07,
+      "loss": 0.0,
+      "step": 14519
+    },
+    {
+      "epoch": 0.935747889411613,
+      "grad_norm": 0.0018801939476720873,
+      "learning_rate": 7.218045112781956e-07,
+      "loss": 0.0,
+      "step": 14520
+    },
+    {
+      "epoch": 0.9358123348585422,
+      "grad_norm": 0.0018678091905745316,
+      "learning_rate": 7.210884353741497e-07,
+      "loss": 0.0,
+      "step": 14521
+    },
+    {
+      "epoch": 0.9358767803054714,
+      "grad_norm": 0.03396311333269752,
+      "learning_rate": 7.203723594701038e-07,
+      "loss": 0.0,
+      "step": 14522
+    },
+    {
+      "epoch": 0.9359412257524006,
+      "grad_norm": 0.003039178851842719,
+      "learning_rate": 7.196562835660581e-07,
+      "loss": 0.0,
+      "step": 14523
+    },
+    {
+      "epoch": 0.9360056711993298,
+      "grad_norm": 0.0005585103175504581,
+      "learning_rate": 7.189402076620122e-07,
+      "loss": 0.0,
+      "step": 14524
+    },
+    {
+      "epoch": 0.9360701166462589,
+      "grad_norm": 0.001817373108679384,
+      "learning_rate": 7.182241317579663e-07,
+      "loss": 0.0,
+      "step": 14525
+    },
+    {
+      "epoch": 0.9361345620931881,
+      "grad_norm": 0.0021193474744381205,
+      "learning_rate": 7.175080558539206e-07,
+      "loss": 0.0,
+      "step": 14526
+    },
+    {
+      "epoch": 0.9361990075401173,
+      "grad_norm": 0.002530882383435495,
+      "learning_rate": 7.167919799498747e-07,
+      "loss": 0.0,
+      "step": 14527
+    },
+    {
+      "epoch": 0.9362634529870465,
+      "grad_norm": 0.000818128099790117,
+      "learning_rate": 7.160759040458289e-07,
+      "loss": 0.0,
+      "step": 14528
+    },
+    {
+      "epoch": 0.9363278984339757,
+      "grad_norm": 0.009652635970240627,
+      "learning_rate": 7.153598281417831e-07,
+      "loss": 0.0,
+      "step": 14529
+    },
+    {
+      "epoch": 0.9363923438809049,
+      "grad_norm": 3.3612020853590016e-05,
+      "learning_rate": 7.146437522377373e-07,
+      "loss": 0.0,
+      "step": 14530
+    },
+    {
+      "epoch": 0.9364567893278339,
+      "grad_norm": 0.0013493935115245926,
+      "learning_rate": 7.139276763336914e-07,
+      "loss": 0.0,
+      "step": 14531
+    },
+    {
+      "epoch": 0.9365212347747631,
+      "grad_norm": 0.0024142204573161376,
+      "learning_rate": 7.132116004296457e-07,
+      "loss": 0.0,
+      "step": 14532
+    },
+    {
+      "epoch": 0.9365856802216923,
+      "grad_norm": 0.2993254916879797,
+      "learning_rate": 7.124955245255997e-07,
+      "loss": 0.001,
+      "step": 14533
+    },
+    {
+      "epoch": 0.9366501256686215,
+      "grad_norm": 0.009092237905382923,
+      "learning_rate": 7.117794486215538e-07,
+      "loss": 0.0,
+      "step": 14534
+    },
+    {
+      "epoch": 0.9367145711155507,
+      "grad_norm": 0.006500430507406041,
+      "learning_rate": 7.110633727175081e-07,
+      "loss": 0.0,
+      "step": 14535
+    },
+    {
+      "epoch": 0.9367790165624799,
+      "grad_norm": 0.0011474112037967268,
+      "learning_rate": 7.103472968134622e-07,
+      "loss": 0.0,
+      "step": 14536
+    },
+    {
+      "epoch": 0.936843462009409,
+      "grad_norm": 0.0018082425170928157,
+      "learning_rate": 7.096312209094165e-07,
+      "loss": 0.0,
+      "step": 14537
+    },
+    {
+      "epoch": 0.9369079074563382,
+      "grad_norm": 0.04529145066339704,
+      "learning_rate": 7.089151450053706e-07,
+      "loss": 0.0001,
+      "step": 14538
+    },
+    {
+      "epoch": 0.9369723529032674,
+      "grad_norm": 0.00024786370618296017,
+      "learning_rate": 7.081990691013248e-07,
+      "loss": 0.0,
+      "step": 14539
+    },
+    {
+      "epoch": 0.9370367983501966,
+      "grad_norm": 0.0014020936697642379,
+      "learning_rate": 7.07482993197279e-07,
+      "loss": 0.0,
+      "step": 14540
+    },
+    {
+      "epoch": 0.9371012437971258,
+      "grad_norm": 0.0005209691458706825,
+      "learning_rate": 7.067669172932332e-07,
+      "loss": 0.0,
+      "step": 14541
+    },
+    {
+      "epoch": 0.9371656892440549,
+      "grad_norm": 0.0019354748003275726,
+      "learning_rate": 7.060508413891873e-07,
+      "loss": 0.0,
+      "step": 14542
+    },
+    {
+      "epoch": 0.937230134690984,
+      "grad_norm": 0.0009135498107519807,
+      "learning_rate": 7.053347654851416e-07,
+      "loss": 0.0,
+      "step": 14543
+    },
+    {
+      "epoch": 0.9372945801379132,
+      "grad_norm": 0.005087708128836168,
+      "learning_rate": 7.046186895810957e-07,
+      "loss": 0.0,
+      "step": 14544
+    },
+    {
+      "epoch": 0.9373590255848424,
+      "grad_norm": 0.002904169890841453,
+      "learning_rate": 7.039026136770497e-07,
+      "loss": 0.0,
+      "step": 14545
+    },
+    {
+      "epoch": 0.9374234710317716,
+      "grad_norm": 0.0005245864079457482,
+      "learning_rate": 7.03186537773004e-07,
+      "loss": 0.0,
+      "step": 14546
+    },
+    {
+      "epoch": 0.9374879164787008,
+      "grad_norm": 0.00027773400644070097,
+      "learning_rate": 7.024704618689581e-07,
+      "loss": 0.0,
+      "step": 14547
+    },
+    {
+      "epoch": 0.93755236192563,
+      "grad_norm": 0.01755280790804899,
+      "learning_rate": 7.017543859649123e-07,
+      "loss": 0.0,
+      "step": 14548
+    },
+    {
+      "epoch": 0.9376168073725591,
+      "grad_norm": 0.00025611269007827297,
+      "learning_rate": 7.010383100608665e-07,
+      "loss": 0.0,
+      "step": 14549
+    },
+    {
+      "epoch": 0.9376812528194883,
+      "grad_norm": 0.6315640272738314,
+      "learning_rate": 7.003222341568207e-07,
+      "loss": 0.0037,
+      "step": 14550
+    },
+    {
+      "epoch": 0.9377456982664175,
+      "grad_norm": 0.021922899573825504,
+      "learning_rate": 6.996061582527748e-07,
+      "loss": 0.0002,
+      "step": 14551
+    },
+    {
+      "epoch": 0.9378101437133467,
+      "grad_norm": 0.09865185582099384,
+      "learning_rate": 6.988900823487291e-07,
+      "loss": 0.0001,
+      "step": 14552
+    },
+    {
+      "epoch": 0.9378745891602758,
+      "grad_norm": 0.012986664406985956,
+      "learning_rate": 6.981740064446832e-07,
+      "loss": 0.0,
+      "step": 14553
+    },
+    {
+      "epoch": 0.937939034607205,
+      "grad_norm": 0.011523680815714164,
+      "learning_rate": 6.974579305406374e-07,
+      "loss": 0.0,
+      "step": 14554
+    },
+    {
+      "epoch": 0.9380034800541341,
+      "grad_norm": 0.0031393640358416347,
+      "learning_rate": 6.967418546365916e-07,
+      "loss": 0.0,
+      "step": 14555
+    },
+    {
+      "epoch": 0.9380679255010633,
+      "grad_norm": 0.019008252551270352,
+      "learning_rate": 6.960257787325458e-07,
+      "loss": 0.0001,
+      "step": 14556
+    },
+    {
+      "epoch": 0.9381323709479925,
+      "grad_norm": 0.0023275455744178716,
+      "learning_rate": 6.953097028284998e-07,
+      "loss": 0.0,
+      "step": 14557
+    },
+    {
+      "epoch": 0.9381968163949217,
+      "grad_norm": 0.005539228100833343,
+      "learning_rate": 6.945936269244541e-07,
+      "loss": 0.0,
+      "step": 14558
+    },
+    {
+      "epoch": 0.9382612618418509,
+      "grad_norm": 0.03702531724548152,
+      "learning_rate": 6.938775510204082e-07,
+      "loss": 0.0001,
+      "step": 14559
+    },
+    {
+      "epoch": 0.9383257072887801,
+      "grad_norm": 0.0015204518635009605,
+      "learning_rate": 6.931614751163623e-07,
+      "loss": 0.0,
+      "step": 14560
+    },
+    {
+      "epoch": 0.9383901527357092,
+      "grad_norm": 0.02489419878730623,
+      "learning_rate": 6.924453992123166e-07,
+      "loss": 0.0001,
+      "step": 14561
+    },
+    {
+      "epoch": 0.9384545981826384,
+      "grad_norm": 0.001340772269739388,
+      "learning_rate": 6.917293233082707e-07,
+      "loss": 0.0,
+      "step": 14562
+    },
+    {
+      "epoch": 0.9385190436295676,
+      "grad_norm": 0.00010016378777309323,
+      "learning_rate": 6.910132474042249e-07,
+      "loss": 0.0,
+      "step": 14563
+    },
+    {
+      "epoch": 0.9385834890764967,
+      "grad_norm": 0.0013275941617353507,
+      "learning_rate": 6.902971715001791e-07,
+      "loss": 0.0,
+      "step": 14564
+    },
+    {
+      "epoch": 0.9386479345234259,
+      "grad_norm": 0.22654219240400406,
+      "learning_rate": 6.895810955961333e-07,
+      "loss": 0.001,
+      "step": 14565
+    },
+    {
+      "epoch": 0.9387123799703551,
+      "grad_norm": 0.0022122313042215727,
+      "learning_rate": 6.888650196920874e-07,
+      "loss": 0.0,
+      "step": 14566
+    },
+    {
+      "epoch": 0.9387768254172842,
+      "grad_norm": 0.004470017713398727,
+      "learning_rate": 6.881489437880417e-07,
+      "loss": 0.0,
+      "step": 14567
+    },
+    {
+      "epoch": 0.9388412708642134,
+      "grad_norm": 0.0007068921499948373,
+      "learning_rate": 6.874328678839958e-07,
+      "loss": 0.0,
+      "step": 14568
+    },
+    {
+      "epoch": 0.9389057163111426,
+      "grad_norm": 0.12361918294424723,
+      "learning_rate": 6.867167919799498e-07,
+      "loss": 0.0,
+      "step": 14569
+    },
+    {
+      "epoch": 0.9389701617580718,
+      "grad_norm": 0.0011432207910996807,
+      "learning_rate": 6.860007160759041e-07,
+      "loss": 0.0,
+      "step": 14570
+    },
+    {
+      "epoch": 0.939034607205001,
+      "grad_norm": 9.127357167783924e-05,
+      "learning_rate": 6.852846401718582e-07,
+      "loss": 0.0,
+      "step": 14571
+    },
+    {
+      "epoch": 0.9390990526519302,
+      "grad_norm": 0.02928841687336939,
+      "learning_rate": 6.845685642678125e-07,
+      "loss": 0.0001,
+      "step": 14572
+    },
+    {
+      "epoch": 0.9391634980988594,
+      "grad_norm": 0.002151597050746372,
+      "learning_rate": 6.838524883637666e-07,
+      "loss": 0.0,
+      "step": 14573
+    },
+    {
+      "epoch": 0.9392279435457885,
+      "grad_norm": 0.0027989000811481177,
+      "learning_rate": 6.831364124597208e-07,
+      "loss": 0.0,
+      "step": 14574
+    },
+    {
+      "epoch": 0.9392923889927177,
+      "grad_norm": 0.000564584986700068,
+      "learning_rate": 6.82420336555675e-07,
+      "loss": 0.0,
+      "step": 14575
+    },
+    {
+      "epoch": 0.9393568344396468,
+      "grad_norm": 0.001180073223414386,
+      "learning_rate": 6.817042606516292e-07,
+      "loss": 0.0,
+      "step": 14576
+    },
+    {
+      "epoch": 0.939421279886576,
+      "grad_norm": 0.005913909066126716,
+      "learning_rate": 6.809881847475833e-07,
+      "loss": 0.0,
+      "step": 14577
+    },
+    {
+      "epoch": 0.9394857253335052,
+      "grad_norm": 0.0006904970453639949,
+      "learning_rate": 6.802721088435376e-07,
+      "loss": 0.0,
+      "step": 14578
+    },
+    {
+      "epoch": 0.9395501707804343,
+      "grad_norm": 0.0004785256442391764,
+      "learning_rate": 6.795560329394917e-07,
+      "loss": 0.0,
+      "step": 14579
+    },
+    {
+      "epoch": 0.9396146162273635,
+      "grad_norm": 0.0065965709683497,
+      "learning_rate": 6.788399570354457e-07,
+      "loss": 0.0,
+      "step": 14580
+    },
+    {
+      "epoch": 0.9396790616742927,
+      "grad_norm": 0.03483474602814386,
+      "learning_rate": 6.781238811314e-07,
+      "loss": 0.0001,
+      "step": 14581
+    },
+    {
+      "epoch": 0.9397435071212219,
+      "grad_norm": 0.0013426242906063755,
+      "learning_rate": 6.774078052273541e-07,
+      "loss": 0.0,
+      "step": 14582
+    },
+    {
+      "epoch": 0.9398079525681511,
+      "grad_norm": 0.11016212508584611,
+      "learning_rate": 6.766917293233083e-07,
+      "loss": 0.0003,
+      "step": 14583
+    },
+    {
+      "epoch": 0.9398723980150803,
+      "grad_norm": 0.0025457227936089575,
+      "learning_rate": 6.759756534192625e-07,
+      "loss": 0.0,
+      "step": 14584
+    },
+    {
+      "epoch": 0.9399368434620095,
+      "grad_norm": 0.0024229772528888424,
+      "learning_rate": 6.752595775152167e-07,
+      "loss": 0.0,
+      "step": 14585
+    },
+    {
+      "epoch": 0.9400012889089386,
+      "grad_norm": 0.030600970769373386,
+      "learning_rate": 6.745435016111708e-07,
+      "loss": 0.0,
+      "step": 14586
+    },
+    {
+      "epoch": 0.9400657343558677,
+      "grad_norm": 0.0005166095727222362,
+      "learning_rate": 6.738274257071251e-07,
+      "loss": 0.0,
+      "step": 14587
+    },
+    {
+      "epoch": 0.9401301798027969,
+      "grad_norm": 0.003762972511217016,
+      "learning_rate": 6.731113498030792e-07,
+      "loss": 0.0,
+      "step": 14588
+    },
+    {
+      "epoch": 0.9401946252497261,
+      "grad_norm": 0.09675340539245941,
+      "learning_rate": 6.723952738990333e-07,
+      "loss": 0.0001,
+      "step": 14589
+    },
+    {
+      "epoch": 0.9402590706966553,
+      "grad_norm": 0.4073272873244558,
+      "learning_rate": 6.716791979949876e-07,
+      "loss": 0.0027,
+      "step": 14590
+    },
+    {
+      "epoch": 0.9403235161435844,
+      "grad_norm": 0.00020267804633695348,
+      "learning_rate": 6.709631220909417e-07,
+      "loss": 0.0,
+      "step": 14591
+    },
+    {
+      "epoch": 0.9403879615905136,
+      "grad_norm": 0.004281621504404462,
+      "learning_rate": 6.702470461868958e-07,
+      "loss": 0.0,
+      "step": 14592
+    },
+    {
+      "epoch": 0.9404524070374428,
+      "grad_norm": 0.0028956385487686137,
+      "learning_rate": 6.6953097028285e-07,
+      "loss": 0.0,
+      "step": 14593
+    },
+    {
+      "epoch": 0.940516852484372,
+      "grad_norm": 0.2428985202888916,
+      "learning_rate": 6.688148943788042e-07,
+      "loss": 0.0008,
+      "step": 14594
+    },
+    {
+      "epoch": 0.9405812979313012,
+      "grad_norm": 0.008918584360090404,
+      "learning_rate": 6.680988184747583e-07,
+      "loss": 0.0,
+      "step": 14595
+    },
+    {
+      "epoch": 0.9406457433782304,
+      "grad_norm": 0.0010786200917876955,
+      "learning_rate": 6.673827425707126e-07,
+      "loss": 0.0,
+      "step": 14596
+    },
+    {
+      "epoch": 0.9407101888251596,
+      "grad_norm": 0.001239924009437483,
+      "learning_rate": 6.666666666666667e-07,
+      "loss": 0.0,
+      "step": 14597
+    },
+    {
+      "epoch": 0.9407746342720886,
+      "grad_norm": 0.17303675628424575,
+      "learning_rate": 6.659505907626208e-07,
+      "loss": 0.0006,
+      "step": 14598
+    },
+    {
+      "epoch": 0.9408390797190178,
+      "grad_norm": 0.0022172739171691714,
+      "learning_rate": 6.652345148585751e-07,
+      "loss": 0.0,
+      "step": 14599
+    },
+    {
+      "epoch": 0.940903525165947,
+      "grad_norm": 0.0009180732993219411,
+      "learning_rate": 6.645184389545292e-07,
+      "loss": 0.0,
+      "step": 14600
+    },
+    {
+      "epoch": 0.9409679706128762,
+      "grad_norm": 0.218649577182511,
+      "learning_rate": 6.638023630504834e-07,
+      "loss": 0.0006,
+      "step": 14601
+    },
+    {
+      "epoch": 0.9410324160598054,
+      "grad_norm": 0.001048798207710388,
+      "learning_rate": 6.630862871464376e-07,
+      "loss": 0.0,
+      "step": 14602
+    },
+    {
+      "epoch": 0.9410968615067345,
+      "grad_norm": 0.0015205476430884322,
+      "learning_rate": 6.623702112423918e-07,
+      "loss": 0.0,
+      "step": 14603
+    },
+    {
+      "epoch": 0.9411613069536637,
+      "grad_norm": 0.022077528704297078,
+      "learning_rate": 6.616541353383458e-07,
+      "loss": 0.0002,
+      "step": 14604
+    },
+    {
+      "epoch": 0.9412257524005929,
+      "grad_norm": 0.0004027494800950605,
+      "learning_rate": 6.609380594343001e-07,
+      "loss": 0.0,
+      "step": 14605
+    },
+    {
+      "epoch": 0.9412901978475221,
+      "grad_norm": 0.2511364171702047,
+      "learning_rate": 6.602219835302542e-07,
+      "loss": 0.0009,
+      "step": 14606
+    },
+    {
+      "epoch": 0.9413546432944513,
+      "grad_norm": 0.7542887295599734,
+      "learning_rate": 6.595059076262083e-07,
+      "loss": 0.0064,
+      "step": 14607
+    },
+    {
+      "epoch": 0.9414190887413805,
+      "grad_norm": 0.004123371457966273,
+      "learning_rate": 6.587898317221626e-07,
+      "loss": 0.0,
+      "step": 14608
+    },
+    {
+      "epoch": 0.9414835341883095,
+      "grad_norm": 0.002666736267158426,
+      "learning_rate": 6.580737558181167e-07,
+      "loss": 0.0,
+      "step": 14609
+    },
+    {
+      "epoch": 0.9415479796352387,
+      "grad_norm": 0.3278604758169207,
+      "learning_rate": 6.57357679914071e-07,
+      "loss": 0.0023,
+      "step": 14610
+    },
+    {
+      "epoch": 0.9416124250821679,
+      "grad_norm": 0.0035760959578561863,
+      "learning_rate": 6.566416040100251e-07,
+      "loss": 0.0,
+      "step": 14611
+    },
+    {
+      "epoch": 0.9416768705290971,
+      "grad_norm": 0.11170839315038376,
+      "learning_rate": 6.559255281059793e-07,
+      "loss": 0.0017,
+      "step": 14612
+    },
+    {
+      "epoch": 0.9417413159760263,
+      "grad_norm": 0.15507088651507142,
+      "learning_rate": 6.552094522019335e-07,
+      "loss": 0.0003,
+      "step": 14613
+    },
+    {
+      "epoch": 0.9418057614229555,
+      "grad_norm": 0.016109021664184563,
+      "learning_rate": 6.544933762978877e-07,
+      "loss": 0.0001,
+      "step": 14614
+    },
+    {
+      "epoch": 0.9418702068698847,
+      "grad_norm": 0.027611918240110323,
+      "learning_rate": 6.537773003938418e-07,
+      "loss": 0.0002,
+      "step": 14615
+    },
+    {
+      "epoch": 0.9419346523168138,
+      "grad_norm": 0.047448985239203184,
+      "learning_rate": 6.530612244897961e-07,
+      "loss": 0.0005,
+      "step": 14616
+    },
+    {
+      "epoch": 0.941999097763743,
+      "grad_norm": 0.0013488067911044182,
+      "learning_rate": 6.523451485857501e-07,
+      "loss": 0.0,
+      "step": 14617
+    },
+    {
+      "epoch": 0.9420635432106722,
+      "grad_norm": 0.011817455589975698,
+      "learning_rate": 6.516290726817042e-07,
+      "loss": 0.0,
+      "step": 14618
+    },
+    {
+      "epoch": 0.9421279886576014,
+      "grad_norm": 0.0021414563002524058,
+      "learning_rate": 6.509129967776585e-07,
+      "loss": 0.0,
+      "step": 14619
+    },
+    {
+      "epoch": 0.9421924341045305,
+      "grad_norm": 0.021960025706933405,
+      "learning_rate": 6.501969208736126e-07,
+      "loss": 0.0,
+      "step": 14620
+    },
+    {
+      "epoch": 0.9422568795514596,
+      "grad_norm": 0.03452706750454334,
+      "learning_rate": 6.494808449695668e-07,
+      "loss": 0.0001,
+      "step": 14621
+    },
+    {
+      "epoch": 0.9423213249983888,
+      "grad_norm": 0.012501871096559888,
+      "learning_rate": 6.48764769065521e-07,
+      "loss": 0.0,
+      "step": 14622
+    },
+    {
+      "epoch": 0.942385770445318,
+      "grad_norm": 0.0062451286698885266,
+      "learning_rate": 6.480486931614752e-07,
+      "loss": 0.0,
+      "step": 14623
+    },
+    {
+      "epoch": 0.9424502158922472,
+      "grad_norm": 0.052520878296810417,
+      "learning_rate": 6.473326172574293e-07,
+      "loss": 0.0001,
+      "step": 14624
+    },
+    {
+      "epoch": 0.9425146613391764,
+      "grad_norm": 0.002352564833617126,
+      "learning_rate": 6.466165413533836e-07,
+      "loss": 0.0,
+      "step": 14625
+    },
+    {
+      "epoch": 0.9425791067861056,
+      "grad_norm": 0.0006536874626899899,
+      "learning_rate": 6.459004654493377e-07,
+      "loss": 0.0,
+      "step": 14626
+    },
+    {
+      "epoch": 0.9426435522330348,
+      "grad_norm": 0.0061842458013730735,
+      "learning_rate": 6.451843895452919e-07,
+      "loss": 0.0,
+      "step": 14627
+    },
+    {
+      "epoch": 0.9427079976799639,
+      "grad_norm": 0.16732236779005036,
+      "learning_rate": 6.44468313641246e-07,
+      "loss": 0.0007,
+      "step": 14628
+    },
+    {
+      "epoch": 0.9427724431268931,
+      "grad_norm": 0.0007157079141550446,
+      "learning_rate": 6.437522377372002e-07,
+      "loss": 0.0,
+      "step": 14629
+    },
+    {
+      "epoch": 0.9428368885738223,
+      "grad_norm": 0.0006728257843449776,
+      "learning_rate": 6.430361618331543e-07,
+      "loss": 0.0,
+      "step": 14630
+    },
+    {
+      "epoch": 0.9429013340207514,
+      "grad_norm": 0.0019405160432858116,
+      "learning_rate": 6.423200859291086e-07,
+      "loss": 0.0,
+      "step": 14631
+    },
+    {
+      "epoch": 0.9429657794676806,
+      "grad_norm": 0.016156358466427594,
+      "learning_rate": 6.416040100250627e-07,
+      "loss": 0.0001,
+      "step": 14632
+    },
+    {
+      "epoch": 0.9430302249146097,
+      "grad_norm": 0.0034454949082682366,
+      "learning_rate": 6.408879341210168e-07,
+      "loss": 0.0,
+      "step": 14633
+    },
+    {
+      "epoch": 0.9430946703615389,
+      "grad_norm": 0.0012120870879107008,
+      "learning_rate": 6.401718582169711e-07,
+      "loss": 0.0,
+      "step": 14634
+    },
+    {
+      "epoch": 0.9431591158084681,
+      "grad_norm": 0.0002194363515000686,
+      "learning_rate": 6.394557823129252e-07,
+      "loss": 0.0,
+      "step": 14635
+    },
+    {
+      "epoch": 0.9432235612553973,
+      "grad_norm": 0.29232680880760875,
+      "learning_rate": 6.387397064088794e-07,
+      "loss": 0.0026,
+      "step": 14636
+    },
+    {
+      "epoch": 0.9432880067023265,
+      "grad_norm": 0.0003396572970585941,
+      "learning_rate": 6.380236305048336e-07,
+      "loss": 0.0,
+      "step": 14637
+    },
+    {
+      "epoch": 0.9433524521492557,
+      "grad_norm": 1.5725421922903065,
+      "learning_rate": 6.373075546007878e-07,
+      "loss": 0.0148,
+      "step": 14638
+    },
+    {
+      "epoch": 0.9434168975961849,
+      "grad_norm": 0.006618061930661491,
+      "learning_rate": 6.365914786967419e-07,
+      "loss": 0.0,
+      "step": 14639
+    },
+    {
+      "epoch": 0.943481343043114,
+      "grad_norm": 1.0823258091984502,
+      "learning_rate": 6.358754027926962e-07,
+      "loss": 0.0037,
+      "step": 14640
+    },
+    {
+      "epoch": 0.9435457884900432,
+      "grad_norm": 0.037367184014392985,
+      "learning_rate": 6.351593268886502e-07,
+      "loss": 0.0003,
+      "step": 14641
+    },
+    {
+      "epoch": 0.9436102339369723,
+      "grad_norm": 0.00023618925437843773,
+      "learning_rate": 6.344432509846043e-07,
+      "loss": 0.0,
+      "step": 14642
+    },
+    {
+      "epoch": 0.9436746793839015,
+      "grad_norm": 0.019535478491815153,
+      "learning_rate": 6.337271750805586e-07,
+      "loss": 0.0001,
+      "step": 14643
+    },
+    {
+      "epoch": 0.9437391248308307,
+      "grad_norm": 0.23517341519437462,
+      "learning_rate": 6.330110991765127e-07,
+      "loss": 0.0003,
+      "step": 14644
+    },
+    {
+      "epoch": 0.9438035702777599,
+      "grad_norm": 0.0012437920009850324,
+      "learning_rate": 6.32295023272467e-07,
+      "loss": 0.0,
+      "step": 14645
+    },
+    {
+      "epoch": 0.943868015724689,
+      "grad_norm": 0.005780556663777061,
+      "learning_rate": 6.315789473684211e-07,
+      "loss": 0.0,
+      "step": 14646
+    },
+    {
+      "epoch": 0.9439324611716182,
+      "grad_norm": 0.027663328196446874,
+      "learning_rate": 6.308628714643753e-07,
+      "loss": 0.0001,
+      "step": 14647
+    },
+    {
+      "epoch": 0.9439969066185474,
+      "grad_norm": 0.03173487718373573,
+      "learning_rate": 6.301467955603295e-07,
+      "loss": 0.0003,
+      "step": 14648
+    },
+    {
+      "epoch": 0.9440613520654766,
+      "grad_norm": 0.0038316525115960707,
+      "learning_rate": 6.294307196562837e-07,
+      "loss": 0.0,
+      "step": 14649
+    },
+    {
+      "epoch": 0.9441257975124058,
+      "grad_norm": 0.007783604901538253,
+      "learning_rate": 6.287146437522378e-07,
+      "loss": 0.0,
+      "step": 14650
+    },
+    {
+      "epoch": 0.944190242959335,
+      "grad_norm": 0.2807085202444718,
+      "learning_rate": 6.279985678481921e-07,
+      "loss": 0.0013,
+      "step": 14651
+    },
+    {
+      "epoch": 0.9442546884062641,
+      "grad_norm": 3.8764283243383565e-05,
+      "learning_rate": 6.272824919441461e-07,
+      "loss": 0.0,
+      "step": 14652
+    },
+    {
+      "epoch": 0.9443191338531933,
+      "grad_norm": 0.33581027127862345,
+      "learning_rate": 6.265664160401002e-07,
+      "loss": 0.0022,
+      "step": 14653
+    },
+    {
+      "epoch": 0.9443835793001224,
+      "grad_norm": 0.011455286908056026,
+      "learning_rate": 6.258503401360545e-07,
+      "loss": 0.0,
+      "step": 14654
+    },
+    {
+      "epoch": 0.9444480247470516,
+      "grad_norm": 0.01142539022155588,
+      "learning_rate": 6.251342642320086e-07,
+      "loss": 0.0,
+      "step": 14655
+    },
+    {
+      "epoch": 0.9445124701939808,
+      "grad_norm": 0.004852478793097276,
+      "learning_rate": 6.244181883279628e-07,
+      "loss": 0.0,
+      "step": 14656
+    },
+    {
+      "epoch": 0.94457691564091,
+      "grad_norm": 0.0019989295768203233,
+      "learning_rate": 6.23702112423917e-07,
+      "loss": 0.0,
+      "step": 14657
+    },
+    {
+      "epoch": 0.9446413610878391,
+      "grad_norm": 0.000900061966794239,
+      "learning_rate": 6.229860365198712e-07,
+      "loss": 0.0,
+      "step": 14658
+    },
+    {
+      "epoch": 0.9447058065347683,
+      "grad_norm": 0.0002101536116015678,
+      "learning_rate": 6.222699606158254e-07,
+      "loss": 0.0,
+      "step": 14659
+    },
+    {
+      "epoch": 0.9447702519816975,
+      "grad_norm": 0.008588863535994664,
+      "learning_rate": 6.215538847117795e-07,
+      "loss": 0.0,
+      "step": 14660
+    },
+    {
+      "epoch": 0.9448346974286267,
+      "grad_norm": 0.04302587423745892,
+      "learning_rate": 6.208378088077337e-07,
+      "loss": 0.0001,
+      "step": 14661
+    },
+    {
+      "epoch": 0.9448991428755559,
+      "grad_norm": 0.005371981311745458,
+      "learning_rate": 6.201217329036879e-07,
+      "loss": 0.0001,
+      "step": 14662
+    },
+    {
+      "epoch": 0.9449635883224851,
+      "grad_norm": 0.018151983512062528,
+      "learning_rate": 6.19405656999642e-07,
+      "loss": 0.0,
+      "step": 14663
+    },
+    {
+      "epoch": 0.9450280337694142,
+      "grad_norm": 0.0012138588332065569,
+      "learning_rate": 6.186895810955961e-07,
+      "loss": 0.0,
+      "step": 14664
+    },
+    {
+      "epoch": 0.9450924792163433,
+      "grad_norm": 0.001995413386031814,
+      "learning_rate": 6.179735051915503e-07,
+      "loss": 0.0,
+      "step": 14665
+    },
+    {
+      "epoch": 0.9451569246632725,
+      "grad_norm": 0.0007131550059482363,
+      "learning_rate": 6.172574292875045e-07,
+      "loss": 0.0,
+      "step": 14666
+    },
+    {
+      "epoch": 0.9452213701102017,
+      "grad_norm": 0.44967737421083515,
+      "learning_rate": 6.165413533834587e-07,
+      "loss": 0.0004,
+      "step": 14667
+    },
+    {
+      "epoch": 0.9452858155571309,
+      "grad_norm": 0.007368867849144508,
+      "learning_rate": 6.158252774794129e-07,
+      "loss": 0.0,
+      "step": 14668
+    },
+    {
+      "epoch": 0.94535026100406,
+      "grad_norm": 0.0074857147827597816,
+      "learning_rate": 6.15109201575367e-07,
+      "loss": 0.0,
+      "step": 14669
+    },
+    {
+      "epoch": 0.9454147064509892,
+      "grad_norm": 0.00023285440752826664,
+      "learning_rate": 6.143931256713212e-07,
+      "loss": 0.0,
+      "step": 14670
+    },
+    {
+      "epoch": 0.9454791518979184,
+      "grad_norm": 0.000904988311691927,
+      "learning_rate": 6.136770497672754e-07,
+      "loss": 0.0,
+      "step": 14671
+    },
+    {
+      "epoch": 0.9455435973448476,
+      "grad_norm": 0.12923123113742113,
+      "learning_rate": 6.129609738632296e-07,
+      "loss": 0.0003,
+      "step": 14672
+    },
+    {
+      "epoch": 0.9456080427917768,
+      "grad_norm": 0.0025449127476049856,
+      "learning_rate": 6.122448979591837e-07,
+      "loss": 0.0,
+      "step": 14673
+    },
+    {
+      "epoch": 0.945672488238706,
+      "grad_norm": 0.0006296391375895605,
+      "learning_rate": 6.115288220551379e-07,
+      "loss": 0.0,
+      "step": 14674
+    },
+    {
+      "epoch": 0.9457369336856352,
+      "grad_norm": 0.1292198729783402,
+      "learning_rate": 6.108127461510921e-07,
+      "loss": 0.0004,
+      "step": 14675
+    },
+    {
+      "epoch": 0.9458013791325642,
+      "grad_norm": 0.001215412270731588,
+      "learning_rate": 6.100966702470462e-07,
+      "loss": 0.0,
+      "step": 14676
+    },
+    {
+      "epoch": 0.9458658245794934,
+      "grad_norm": 0.001039974299197542,
+      "learning_rate": 6.093805943430004e-07,
+      "loss": 0.0,
+      "step": 14677
+    },
+    {
+      "epoch": 0.9459302700264226,
+      "grad_norm": 0.6429765156199316,
+      "learning_rate": 6.086645184389546e-07,
+      "loss": 0.0003,
+      "step": 14678
+    },
+    {
+      "epoch": 0.9459947154733518,
+      "grad_norm": 0.0008230507368414166,
+      "learning_rate": 6.079484425349087e-07,
+      "loss": 0.0,
+      "step": 14679
+    },
+    {
+      "epoch": 0.946059160920281,
+      "grad_norm": 0.00024397614750560574,
+      "learning_rate": 6.072323666308629e-07,
+      "loss": 0.0,
+      "step": 14680
+    },
+    {
+      "epoch": 0.9461236063672102,
+      "grad_norm": 0.007197154311858535,
+      "learning_rate": 6.065162907268171e-07,
+      "loss": 0.0,
+      "step": 14681
+    },
+    {
+      "epoch": 0.9461880518141393,
+      "grad_norm": 0.1378045088749925,
+      "learning_rate": 6.058002148227712e-07,
+      "loss": 0.0002,
+      "step": 14682
+    },
+    {
+      "epoch": 0.9462524972610685,
+      "grad_norm": 0.0013014266775007563,
+      "learning_rate": 6.050841389187254e-07,
+      "loss": 0.0,
+      "step": 14683
+    },
+    {
+      "epoch": 0.9463169427079977,
+      "grad_norm": 0.000253774893202994,
+      "learning_rate": 6.043680630146796e-07,
+      "loss": 0.0,
+      "step": 14684
+    },
+    {
+      "epoch": 0.9463813881549269,
+      "grad_norm": 0.0038829678671018237,
+      "learning_rate": 6.036519871106338e-07,
+      "loss": 0.0,
+      "step": 14685
+    },
+    {
+      "epoch": 0.9464458336018561,
+      "grad_norm": 0.00037332670641363337,
+      "learning_rate": 6.02935911206588e-07,
+      "loss": 0.0,
+      "step": 14686
+    },
+    {
+      "epoch": 0.9465102790487852,
+      "grad_norm": 0.00011434537095249858,
+      "learning_rate": 6.022198353025422e-07,
+      "loss": 0.0,
+      "step": 14687
+    },
+    {
+      "epoch": 0.9465747244957143,
+      "grad_norm": 0.022522121263546153,
+      "learning_rate": 6.015037593984962e-07,
+      "loss": 0.0001,
+      "step": 14688
+    },
+    {
+      "epoch": 0.9466391699426435,
+      "grad_norm": 0.08534634801181562,
+      "learning_rate": 6.007876834944504e-07,
+      "loss": 0.0002,
+      "step": 14689
+    },
+    {
+      "epoch": 0.9467036153895727,
+      "grad_norm": 0.008271676467209783,
+      "learning_rate": 6.000716075904046e-07,
+      "loss": 0.0,
+      "step": 14690
+    },
+    {
+      "epoch": 0.9467680608365019,
+      "grad_norm": 0.01073189738361365,
+      "learning_rate": 5.993555316863588e-07,
+      "loss": 0.0,
+      "step": 14691
+    },
+    {
+      "epoch": 0.9468325062834311,
+      "grad_norm": 0.004863051197559781,
+      "learning_rate": 5.986394557823129e-07,
+      "loss": 0.0,
+      "step": 14692
+    },
+    {
+      "epoch": 0.9468969517303603,
+      "grad_norm": 0.010054763393330567,
+      "learning_rate": 5.979233798782671e-07,
+      "loss": 0.0,
+      "step": 14693
+    },
+    {
+      "epoch": 0.9469613971772894,
+      "grad_norm": 0.002437897019058872,
+      "learning_rate": 5.972073039742213e-07,
+      "loss": 0.0,
+      "step": 14694
+    },
+    {
+      "epoch": 0.9470258426242186,
+      "grad_norm": 0.013852156443718828,
+      "learning_rate": 5.964912280701755e-07,
+      "loss": 0.0001,
+      "step": 14695
+    },
+    {
+      "epoch": 0.9470902880711478,
+      "grad_norm": 0.0003029941248519201,
+      "learning_rate": 5.957751521661297e-07,
+      "loss": 0.0,
+      "step": 14696
+    },
+    {
+      "epoch": 0.947154733518077,
+      "grad_norm": 0.002594353409038197,
+      "learning_rate": 5.950590762620839e-07,
+      "loss": 0.0,
+      "step": 14697
+    },
+    {
+      "epoch": 0.9472191789650061,
+      "grad_norm": 0.2996835992179366,
+      "learning_rate": 5.94343000358038e-07,
+      "loss": 0.0019,
+      "step": 14698
+    },
+    {
+      "epoch": 0.9472836244119353,
+      "grad_norm": 0.011269491880908408,
+      "learning_rate": 5.936269244539922e-07,
+      "loss": 0.0,
+      "step": 14699
+    },
+    {
+      "epoch": 0.9473480698588644,
+      "grad_norm": 0.0009420079780451361,
+      "learning_rate": 5.929108485499463e-07,
+      "loss": 0.0,
+      "step": 14700
+    },
+    {
+      "epoch": 0.9474125153057936,
+      "grad_norm": 0.0017608986153713367,
+      "learning_rate": 5.921947726459005e-07,
+      "loss": 0.0,
+      "step": 14701
+    },
+    {
+      "epoch": 0.9474769607527228,
+      "grad_norm": 0.0013094455840437949,
+      "learning_rate": 5.914786967418547e-07,
+      "loss": 0.0,
+      "step": 14702
+    },
+    {
+      "epoch": 0.947541406199652,
+      "grad_norm": 0.002251141911154188,
+      "learning_rate": 5.907626208378089e-07,
+      "loss": 0.0,
+      "step": 14703
+    },
+    {
+      "epoch": 0.9476058516465812,
+      "grad_norm": 0.011790262267177658,
+      "learning_rate": 5.90046544933763e-07,
+      "loss": 0.0,
+      "step": 14704
+    },
+    {
+      "epoch": 0.9476702970935104,
+      "grad_norm": 0.002285259579575094,
+      "learning_rate": 5.893304690297172e-07,
+      "loss": 0.0,
+      "step": 14705
+    },
+    {
+      "epoch": 0.9477347425404395,
+      "grad_norm": 0.0006253211658536153,
+      "learning_rate": 5.886143931256714e-07,
+      "loss": 0.0,
+      "step": 14706
+    },
+    {
+      "epoch": 0.9477991879873687,
+      "grad_norm": 0.005398306974137369,
+      "learning_rate": 5.878983172216255e-07,
+      "loss": 0.0,
+      "step": 14707
+    },
+    {
+      "epoch": 0.9478636334342979,
+      "grad_norm": 0.0002294010027379783,
+      "learning_rate": 5.871822413175797e-07,
+      "loss": 0.0,
+      "step": 14708
+    },
+    {
+      "epoch": 0.947928078881227,
+      "grad_norm": 0.00025866017801789347,
+      "learning_rate": 5.864661654135339e-07,
+      "loss": 0.0,
+      "step": 14709
+    },
+    {
+      "epoch": 0.9479925243281562,
+      "grad_norm": 0.0005619206255567925,
+      "learning_rate": 5.857500895094881e-07,
+      "loss": 0.0,
+      "step": 14710
+    },
+    {
+      "epoch": 0.9480569697750854,
+      "grad_norm": 0.19647921181904326,
+      "learning_rate": 5.850340136054423e-07,
+      "loss": 0.0021,
+      "step": 14711
+    },
+    {
+      "epoch": 0.9481214152220145,
+      "grad_norm": 0.08082751242960239,
+      "learning_rate": 5.843179377013964e-07,
+      "loss": 0.0001,
+      "step": 14712
+    },
+    {
+      "epoch": 0.9481858606689437,
+      "grad_norm": 0.0014319050676462684,
+      "learning_rate": 5.836018617973506e-07,
+      "loss": 0.0,
+      "step": 14713
+    },
+    {
+      "epoch": 0.9482503061158729,
+      "grad_norm": 0.0021555817053756725,
+      "learning_rate": 5.828857858933047e-07,
+      "loss": 0.0,
+      "step": 14714
+    },
+    {
+      "epoch": 0.9483147515628021,
+      "grad_norm": 0.0019083600342925365,
+      "learning_rate": 5.821697099892589e-07,
+      "loss": 0.0,
+      "step": 14715
+    },
+    {
+      "epoch": 0.9483791970097313,
+      "grad_norm": 0.41414290673102844,
+      "learning_rate": 5.814536340852131e-07,
+      "loss": 0.003,
+      "step": 14716
+    },
+    {
+      "epoch": 0.9484436424566605,
+      "grad_norm": 0.008211531723705947,
+      "learning_rate": 5.807375581811672e-07,
+      "loss": 0.0,
+      "step": 14717
+    },
+    {
+      "epoch": 0.9485080879035896,
+      "grad_norm": 0.00016811693723598264,
+      "learning_rate": 5.800214822771214e-07,
+      "loss": 0.0,
+      "step": 14718
+    },
+    {
+      "epoch": 0.9485725333505188,
+      "grad_norm": 0.0008967912893646619,
+      "learning_rate": 5.793054063730756e-07,
+      "loss": 0.0,
+      "step": 14719
+    },
+    {
+      "epoch": 0.9486369787974479,
+      "grad_norm": 0.006830654593558417,
+      "learning_rate": 5.785893304690298e-07,
+      "loss": 0.0,
+      "step": 14720
+    },
+    {
+      "epoch": 0.9487014242443771,
+      "grad_norm": 0.02299987371644241,
+      "learning_rate": 5.77873254564984e-07,
+      "loss": 0.0001,
+      "step": 14721
+    },
+    {
+      "epoch": 0.9487658696913063,
+      "grad_norm": 0.023379648106420153,
+      "learning_rate": 5.771571786609382e-07,
+      "loss": 0.0002,
+      "step": 14722
+    },
+    {
+      "epoch": 0.9488303151382355,
+      "grad_norm": 0.011996984555252728,
+      "learning_rate": 5.764411027568922e-07,
+      "loss": 0.0001,
+      "step": 14723
+    },
+    {
+      "epoch": 0.9488947605851646,
+      "grad_norm": 0.0010316172728567184,
+      "learning_rate": 5.757250268528464e-07,
+      "loss": 0.0,
+      "step": 14724
+    },
+    {
+      "epoch": 0.9489592060320938,
+      "grad_norm": 0.0006875478425085163,
+      "learning_rate": 5.750089509488006e-07,
+      "loss": 0.0,
+      "step": 14725
+    },
+    {
+      "epoch": 0.949023651479023,
+      "grad_norm": 0.0007850438860483339,
+      "learning_rate": 5.742928750447548e-07,
+      "loss": 0.0,
+      "step": 14726
+    },
+    {
+      "epoch": 0.9490880969259522,
+      "grad_norm": 0.1185001488226447,
+      "learning_rate": 5.735767991407089e-07,
+      "loss": 0.0001,
+      "step": 14727
+    },
+    {
+      "epoch": 0.9491525423728814,
+      "grad_norm": 0.00042052385974219097,
+      "learning_rate": 5.728607232366631e-07,
+      "loss": 0.0,
+      "step": 14728
+    },
+    {
+      "epoch": 0.9492169878198106,
+      "grad_norm": 0.023423001652901322,
+      "learning_rate": 5.721446473326173e-07,
+      "loss": 0.0002,
+      "step": 14729
+    },
+    {
+      "epoch": 0.9492814332667397,
+      "grad_norm": 0.00019013561235615476,
+      "learning_rate": 5.714285714285715e-07,
+      "loss": 0.0,
+      "step": 14730
+    },
+    {
+      "epoch": 0.9493458787136689,
+      "grad_norm": 0.008218830548370379,
+      "learning_rate": 5.707124955245257e-07,
+      "loss": 0.0,
+      "step": 14731
+    },
+    {
+      "epoch": 0.949410324160598,
+      "grad_norm": 0.006414075984919605,
+      "learning_rate": 5.699964196204799e-07,
+      "loss": 0.0,
+      "step": 14732
+    },
+    {
+      "epoch": 0.9494747696075272,
+      "grad_norm": 0.0007078383412863962,
+      "learning_rate": 5.69280343716434e-07,
+      "loss": 0.0,
+      "step": 14733
+    },
+    {
+      "epoch": 0.9495392150544564,
+      "grad_norm": 0.030856944468306394,
+      "learning_rate": 5.685642678123882e-07,
+      "loss": 0.0,
+      "step": 14734
+    },
+    {
+      "epoch": 0.9496036605013856,
+      "grad_norm": 0.0069689116245061295,
+      "learning_rate": 5.678481919083424e-07,
+      "loss": 0.0,
+      "step": 14735
+    },
+    {
+      "epoch": 0.9496681059483147,
+      "grad_norm": 0.005919522472079522,
+      "learning_rate": 5.671321160042964e-07,
+      "loss": 0.0,
+      "step": 14736
+    },
+    {
+      "epoch": 0.9497325513952439,
+      "grad_norm": 0.0028393379796273445,
+      "learning_rate": 5.664160401002506e-07,
+      "loss": 0.0,
+      "step": 14737
+    },
+    {
+      "epoch": 0.9497969968421731,
+      "grad_norm": 0.036288034329051345,
+      "learning_rate": 5.656999641962048e-07,
+      "loss": 0.0001,
+      "step": 14738
+    },
+    {
+      "epoch": 0.9498614422891023,
+      "grad_norm": 0.0012907005085677104,
+      "learning_rate": 5.64983888292159e-07,
+      "loss": 0.0,
+      "step": 14739
+    },
+    {
+      "epoch": 0.9499258877360315,
+      "grad_norm": 0.01761313683527187,
+      "learning_rate": 5.642678123881132e-07,
+      "loss": 0.0001,
+      "step": 14740
+    },
+    {
+      "epoch": 0.9499903331829607,
+      "grad_norm": 0.009967518907129018,
+      "learning_rate": 5.635517364840674e-07,
+      "loss": 0.0,
+      "step": 14741
+    },
+    {
+      "epoch": 0.9500547786298899,
+      "grad_norm": 0.0018596338611846243,
+      "learning_rate": 5.628356605800215e-07,
+      "loss": 0.0,
+      "step": 14742
+    },
+    {
+      "epoch": 0.9501192240768189,
+      "grad_norm": 0.004828228626011528,
+      "learning_rate": 5.621195846759757e-07,
+      "loss": 0.0,
+      "step": 14743
+    },
+    {
+      "epoch": 0.9501836695237481,
+      "grad_norm": 0.006116660387896934,
+      "learning_rate": 5.614035087719299e-07,
+      "loss": 0.0,
+      "step": 14744
+    },
+    {
+      "epoch": 0.9502481149706773,
+      "grad_norm": 0.00060880213351492,
+      "learning_rate": 5.606874328678841e-07,
+      "loss": 0.0,
+      "step": 14745
+    },
+    {
+      "epoch": 0.9503125604176065,
+      "grad_norm": 0.14036646230097022,
+      "learning_rate": 5.599713569638382e-07,
+      "loss": 0.0002,
+      "step": 14746
+    },
+    {
+      "epoch": 0.9503770058645357,
+      "grad_norm": 0.3058433494413237,
+      "learning_rate": 5.592552810597923e-07,
+      "loss": 0.0008,
+      "step": 14747
+    },
+    {
+      "epoch": 0.9504414513114648,
+      "grad_norm": 0.024473252072277532,
+      "learning_rate": 5.585392051557465e-07,
+      "loss": 0.0001,
+      "step": 14748
+    },
+    {
+      "epoch": 0.950505896758394,
+      "grad_norm": 0.0001964074795710888,
+      "learning_rate": 5.578231292517007e-07,
+      "loss": 0.0,
+      "step": 14749
+    },
+    {
+      "epoch": 0.9505703422053232,
+      "grad_norm": 0.0008308660348116496,
+      "learning_rate": 5.571070533476549e-07,
+      "loss": 0.0,
+      "step": 14750
+    },
+    {
+      "epoch": 0.9506347876522524,
+      "grad_norm": 0.0006085072018791508,
+      "learning_rate": 5.563909774436091e-07,
+      "loss": 0.0,
+      "step": 14751
+    },
+    {
+      "epoch": 0.9506992330991816,
+      "grad_norm": 0.0014048571143467556,
+      "learning_rate": 5.556749015395632e-07,
+      "loss": 0.0,
+      "step": 14752
+    },
+    {
+      "epoch": 0.9507636785461108,
+      "grad_norm": 0.000805794097092772,
+      "learning_rate": 5.549588256355174e-07,
+      "loss": 0.0,
+      "step": 14753
+    },
+    {
+      "epoch": 0.9508281239930398,
+      "grad_norm": 0.00016851595053399371,
+      "learning_rate": 5.542427497314716e-07,
+      "loss": 0.0,
+      "step": 14754
+    },
+    {
+      "epoch": 0.950892569439969,
+      "grad_norm": 0.30476000730045005,
+      "learning_rate": 5.535266738274257e-07,
+      "loss": 0.001,
+      "step": 14755
+    },
+    {
+      "epoch": 0.9509570148868982,
+      "grad_norm": 0.017287191174224937,
+      "learning_rate": 5.528105979233799e-07,
+      "loss": 0.0,
+      "step": 14756
+    },
+    {
+      "epoch": 0.9510214603338274,
+      "grad_norm": 0.0008960561301208798,
+      "learning_rate": 5.520945220193341e-07,
+      "loss": 0.0,
+      "step": 14757
+    },
+    {
+      "epoch": 0.9510859057807566,
+      "grad_norm": 0.022833389816133105,
+      "learning_rate": 5.513784461152883e-07,
+      "loss": 0.0001,
+      "step": 14758
+    },
+    {
+      "epoch": 0.9511503512276858,
+      "grad_norm": 0.0007505371049380001,
+      "learning_rate": 5.506623702112424e-07,
+      "loss": 0.0,
+      "step": 14759
+    },
+    {
+      "epoch": 0.951214796674615,
+      "grad_norm": 0.004165736964287361,
+      "learning_rate": 5.499462943071966e-07,
+      "loss": 0.0,
+      "step": 14760
+    },
+    {
+      "epoch": 0.9512792421215441,
+      "grad_norm": 0.01478945535523023,
+      "learning_rate": 5.492302184031507e-07,
+      "loss": 0.0,
+      "step": 14761
+    },
+    {
+      "epoch": 0.9513436875684733,
+      "grad_norm": 0.0018664110528120968,
+      "learning_rate": 5.485141424991049e-07,
+      "loss": 0.0,
+      "step": 14762
+    },
+    {
+      "epoch": 0.9514081330154025,
+      "grad_norm": 0.5059164891348179,
+      "learning_rate": 5.477980665950591e-07,
+      "loss": 0.0014,
+      "step": 14763
+    },
+    {
+      "epoch": 0.9514725784623317,
+      "grad_norm": 0.0008552185764550522,
+      "learning_rate": 5.470819906910133e-07,
+      "loss": 0.0,
+      "step": 14764
+    },
+    {
+      "epoch": 0.9515370239092608,
+      "grad_norm": 0.003477670911935671,
+      "learning_rate": 5.463659147869674e-07,
+      "loss": 0.0,
+      "step": 14765
+    },
+    {
+      "epoch": 0.9516014693561899,
+      "grad_norm": 0.5282372710846787,
+      "learning_rate": 5.456498388829216e-07,
+      "loss": 0.0007,
+      "step": 14766
+    },
+    {
+      "epoch": 0.9516659148031191,
+      "grad_norm": 0.0669501936225196,
+      "learning_rate": 5.449337629788758e-07,
+      "loss": 0.0001,
+      "step": 14767
+    },
+    {
+      "epoch": 0.9517303602500483,
+      "grad_norm": 0.001082596470456628,
+      "learning_rate": 5.4421768707483e-07,
+      "loss": 0.0,
+      "step": 14768
+    },
+    {
+      "epoch": 0.9517948056969775,
+      "grad_norm": 0.013068741403885711,
+      "learning_rate": 5.435016111707842e-07,
+      "loss": 0.0,
+      "step": 14769
+    },
+    {
+      "epoch": 0.9518592511439067,
+      "grad_norm": 0.0013497502223646179,
+      "learning_rate": 5.427855352667384e-07,
+      "loss": 0.0,
+      "step": 14770
+    },
+    {
+      "epoch": 0.9519236965908359,
+      "grad_norm": 0.06839671137127833,
+      "learning_rate": 5.420694593626924e-07,
+      "loss": 0.0001,
+      "step": 14771
+    },
+    {
+      "epoch": 0.951988142037765,
+      "grad_norm": 0.00470835857793575,
+      "learning_rate": 5.413533834586466e-07,
+      "loss": 0.0,
+      "step": 14772
+    },
+    {
+      "epoch": 0.9520525874846942,
+      "grad_norm": 0.008123316991707919,
+      "learning_rate": 5.406373075546008e-07,
+      "loss": 0.0,
+      "step": 14773
+    },
+    {
+      "epoch": 0.9521170329316234,
+      "grad_norm": 0.1347828798719565,
+      "learning_rate": 5.39921231650555e-07,
+      "loss": 0.0002,
+      "step": 14774
+    },
+    {
+      "epoch": 0.9521814783785526,
+      "grad_norm": 0.0013437443289455824,
+      "learning_rate": 5.392051557465092e-07,
+      "loss": 0.0,
+      "step": 14775
+    },
+    {
+      "epoch": 0.9522459238254817,
+      "grad_norm": 0.03295035955820533,
+      "learning_rate": 5.384890798424634e-07,
+      "loss": 0.0,
+      "step": 14776
+    },
+    {
+      "epoch": 0.9523103692724109,
+      "grad_norm": 0.002494679766104263,
+      "learning_rate": 5.377730039384175e-07,
+      "loss": 0.0,
+      "step": 14777
+    },
+    {
+      "epoch": 0.95237481471934,
+      "grad_norm": 0.0004908145695510208,
+      "learning_rate": 5.370569280343717e-07,
+      "loss": 0.0,
+      "step": 14778
+    },
+    {
+      "epoch": 0.9524392601662692,
+      "grad_norm": 0.07464586197099465,
+      "learning_rate": 5.363408521303259e-07,
+      "loss": 0.0001,
+      "step": 14779
+    },
+    {
+      "epoch": 0.9525037056131984,
+      "grad_norm": 9.668897209007286e-05,
+      "learning_rate": 5.356247762262801e-07,
+      "loss": 0.0,
+      "step": 14780
+    },
+    {
+      "epoch": 0.9525681510601276,
+      "grad_norm": 0.0045286485902072664,
+      "learning_rate": 5.349087003222342e-07,
+      "loss": 0.0,
+      "step": 14781
+    },
+    {
+      "epoch": 0.9526325965070568,
+      "grad_norm": 0.0027660902273609613,
+      "learning_rate": 5.341926244181884e-07,
+      "loss": 0.0,
+      "step": 14782
+    },
+    {
+      "epoch": 0.952697041953986,
+      "grad_norm": 0.005766246153932424,
+      "learning_rate": 5.334765485141425e-07,
+      "loss": 0.0,
+      "step": 14783
+    },
+    {
+      "epoch": 0.9527614874009152,
+      "grad_norm": 0.001195186415794037,
+      "learning_rate": 5.327604726100967e-07,
+      "loss": 0.0,
+      "step": 14784
+    },
+    {
+      "epoch": 0.9528259328478443,
+      "grad_norm": 0.0010986515422721234,
+      "learning_rate": 5.320443967060509e-07,
+      "loss": 0.0,
+      "step": 14785
+    },
+    {
+      "epoch": 0.9528903782947735,
+      "grad_norm": 0.02101568137284521,
+      "learning_rate": 5.313283208020051e-07,
+      "loss": 0.0,
+      "step": 14786
+    },
+    {
+      "epoch": 0.9529548237417026,
+      "grad_norm": 0.004531570145919255,
+      "learning_rate": 5.306122448979592e-07,
+      "loss": 0.0,
+      "step": 14787
+    },
+    {
+      "epoch": 0.9530192691886318,
+      "grad_norm": 0.00046395947667538586,
+      "learning_rate": 5.298961689939134e-07,
+      "loss": 0.0,
+      "step": 14788
+    },
+    {
+      "epoch": 0.953083714635561,
+      "grad_norm": 0.07514254155018177,
+      "learning_rate": 5.291800930898676e-07,
+      "loss": 0.0017,
+      "step": 14789
+    },
+    {
+      "epoch": 0.9531481600824901,
+      "grad_norm": 0.00022843025780618325,
+      "learning_rate": 5.284640171858217e-07,
+      "loss": 0.0,
+      "step": 14790
+    },
+    {
+      "epoch": 0.9532126055294193,
+      "grad_norm": 0.004491149951619666,
+      "learning_rate": 5.277479412817759e-07,
+      "loss": 0.0015,
+      "step": 14791
+    },
+    {
+      "epoch": 0.9532770509763485,
+      "grad_norm": 0.0005985131719744827,
+      "learning_rate": 5.270318653777301e-07,
+      "loss": 0.0,
+      "step": 14792
+    },
+    {
+      "epoch": 0.9533414964232777,
+      "grad_norm": 0.0005078170219091677,
+      "learning_rate": 5.263157894736843e-07,
+      "loss": 0.0,
+      "step": 14793
+    },
+    {
+      "epoch": 0.9534059418702069,
+      "grad_norm": 0.002439488648787511,
+      "learning_rate": 5.255997135696385e-07,
+      "loss": 0.0015,
+      "step": 14794
+    },
+    {
+      "epoch": 0.9534703873171361,
+      "grad_norm": 0.001817523298432707,
+      "learning_rate": 5.248836376655927e-07,
+      "loss": 0.0,
+      "step": 14795
+    },
+    {
+      "epoch": 0.9535348327640653,
+      "grad_norm": 0.004825207351911492,
+      "learning_rate": 5.241675617615467e-07,
+      "loss": 0.0,
+      "step": 14796
+    },
+    {
+      "epoch": 0.9535992782109944,
+      "grad_norm": 0.0004920440792501377,
+      "learning_rate": 5.234514858575009e-07,
+      "loss": 0.0,
+      "step": 14797
+    },
+    {
+      "epoch": 0.9536637236579236,
+      "grad_norm": 0.003268970344869564,
+      "learning_rate": 5.227354099534551e-07,
+      "loss": 0.0,
+      "step": 14798
+    },
+    {
+      "epoch": 0.9537281691048527,
+      "grad_norm": 6.151258179910864e-05,
+      "learning_rate": 5.220193340494093e-07,
+      "loss": 0.0,
+      "step": 14799
+    },
+    {
+      "epoch": 0.9537926145517819,
+      "grad_norm": 0.0668628749985501,
+      "learning_rate": 5.213032581453634e-07,
+      "loss": 0.0002,
+      "step": 14800
+    },
+    {
+      "epoch": 0.9538570599987111,
+      "grad_norm": 0.000984831570237332,
+      "learning_rate": 5.205871822413176e-07,
+      "loss": 0.0,
+      "step": 14801
+    },
+    {
+      "epoch": 0.9539215054456402,
+      "grad_norm": 0.060623120937525234,
+      "learning_rate": 5.198711063372718e-07,
+      "loss": 0.0001,
+      "step": 14802
+    },
+    {
+      "epoch": 0.9539859508925694,
+      "grad_norm": 0.0007038617180894285,
+      "learning_rate": 5.19155030433226e-07,
+      "loss": 0.0,
+      "step": 14803
+    },
+    {
+      "epoch": 0.9540503963394986,
+      "grad_norm": 0.0028088622921580444,
+      "learning_rate": 5.184389545291802e-07,
+      "loss": 0.0,
+      "step": 14804
+    },
+    {
+      "epoch": 0.9541148417864278,
+      "grad_norm": 0.010322705780409896,
+      "learning_rate": 5.177228786251344e-07,
+      "loss": 0.0,
+      "step": 14805
+    },
+    {
+      "epoch": 0.954179287233357,
+      "grad_norm": 0.004806938039548968,
+      "learning_rate": 5.170068027210885e-07,
+      "loss": 0.0,
+      "step": 14806
+    },
+    {
+      "epoch": 0.9542437326802862,
+      "grad_norm": 0.00021441404646725413,
+      "learning_rate": 5.162907268170426e-07,
+      "loss": 0.0,
+      "step": 14807
+    },
+    {
+      "epoch": 0.9543081781272154,
+      "grad_norm": 0.04337507862455308,
+      "learning_rate": 5.155746509129968e-07,
+      "loss": 0.0001,
+      "step": 14808
+    },
+    {
+      "epoch": 0.9543726235741445,
+      "grad_norm": 0.16570335639962636,
+      "learning_rate": 5.148585750089509e-07,
+      "loss": 0.0002,
+      "step": 14809
+    },
+    {
+      "epoch": 0.9544370690210736,
+      "grad_norm": 0.006217551071894887,
+      "learning_rate": 5.141424991049051e-07,
+      "loss": 0.0,
+      "step": 14810
+    },
+    {
+      "epoch": 0.9545015144680028,
+      "grad_norm": 0.0012870383012455599,
+      "learning_rate": 5.134264232008593e-07,
+      "loss": 0.0,
+      "step": 14811
+    },
+    {
+      "epoch": 0.954565959914932,
+      "grad_norm": 0.016881525437275658,
+      "learning_rate": 5.127103472968135e-07,
+      "loss": 0.0,
+      "step": 14812
+    },
+    {
+      "epoch": 0.9546304053618612,
+      "grad_norm": 0.001555288436966095,
+      "learning_rate": 5.119942713927677e-07,
+      "loss": 0.0,
+      "step": 14813
+    },
+    {
+      "epoch": 0.9546948508087904,
+      "grad_norm": 0.008399473534627302,
+      "learning_rate": 5.112781954887219e-07,
+      "loss": 0.0,
+      "step": 14814
+    },
+    {
+      "epoch": 0.9547592962557195,
+      "grad_norm": 0.0026212772132492594,
+      "learning_rate": 5.10562119584676e-07,
+      "loss": 0.0,
+      "step": 14815
+    },
+    {
+      "epoch": 0.9548237417026487,
+      "grad_norm": 0.005733009960411951,
+      "learning_rate": 5.098460436806302e-07,
+      "loss": 0.0,
+      "step": 14816
+    },
+    {
+      "epoch": 0.9548881871495779,
+      "grad_norm": 0.030370706949806717,
+      "learning_rate": 5.091299677765844e-07,
+      "loss": 0.0001,
+      "step": 14817
+    },
+    {
+      "epoch": 0.9549526325965071,
+      "grad_norm": 0.0005157755746085372,
+      "learning_rate": 5.084138918725386e-07,
+      "loss": 0.0,
+      "step": 14818
+    },
+    {
+      "epoch": 0.9550170780434363,
+      "grad_norm": 2.917667148725611e-05,
+      "learning_rate": 5.076978159684926e-07,
+      "loss": 0.0,
+      "step": 14819
+    },
+    {
+      "epoch": 0.9550815234903655,
+      "grad_norm": 0.09262644516760862,
+      "learning_rate": 5.069817400644468e-07,
+      "loss": 0.0008,
+      "step": 14820
+    },
+    {
+      "epoch": 0.9551459689372945,
+      "grad_norm": 0.003610028407824278,
+      "learning_rate": 5.06265664160401e-07,
+      "loss": 0.0,
+      "step": 14821
+    },
+    {
+      "epoch": 0.9552104143842237,
+      "grad_norm": 0.0028029187425091767,
+      "learning_rate": 5.055495882563552e-07,
+      "loss": 0.0,
+      "step": 14822
+    },
+    {
+      "epoch": 0.9552748598311529,
+      "grad_norm": 0.00040773365753766214,
+      "learning_rate": 5.048335123523094e-07,
+      "loss": 0.0,
+      "step": 14823
+    },
+    {
+      "epoch": 0.9553393052780821,
+      "grad_norm": 0.00031257184924321485,
+      "learning_rate": 5.041174364482636e-07,
+      "loss": 0.0,
+      "step": 14824
+    },
+    {
+      "epoch": 0.9554037507250113,
+      "grad_norm": 0.00040114875345770496,
+      "learning_rate": 5.034013605442177e-07,
+      "loss": 0.0,
+      "step": 14825
+    },
+    {
+      "epoch": 0.9554681961719405,
+      "grad_norm": 0.008520543711840797,
+      "learning_rate": 5.026852846401719e-07,
+      "loss": 0.0,
+      "step": 14826
+    },
+    {
+      "epoch": 0.9555326416188696,
+      "grad_norm": 0.00017712328294400463,
+      "learning_rate": 5.019692087361261e-07,
+      "loss": 0.0,
+      "step": 14827
+    },
+    {
+      "epoch": 0.9555970870657988,
+      "grad_norm": 0.00448170861329047,
+      "learning_rate": 5.012531328320802e-07,
+      "loss": 0.0,
+      "step": 14828
+    },
+    {
+      "epoch": 0.955661532512728,
+      "grad_norm": 0.002184216611653189,
+      "learning_rate": 5.005370569280344e-07,
+      "loss": 0.0,
+      "step": 14829
+    },
+    {
+      "epoch": 0.9557259779596572,
+      "grad_norm": 0.2183928895592596,
+      "learning_rate": 4.998209810239886e-07,
+      "loss": 0.001,
+      "step": 14830
+    },
+    {
+      "epoch": 0.9557904234065864,
+      "grad_norm": 0.012881565613289223,
+      "learning_rate": 4.991049051199427e-07,
+      "loss": 0.0001,
+      "step": 14831
+    },
+    {
+      "epoch": 0.9558548688535154,
+      "grad_norm": 0.09858547151000312,
+      "learning_rate": 4.983888292158969e-07,
+      "loss": 0.0007,
+      "step": 14832
+    },
+    {
+      "epoch": 0.9559193143004446,
+      "grad_norm": 0.001604564122585394,
+      "learning_rate": 4.976727533118511e-07,
+      "loss": 0.0,
+      "step": 14833
+    },
+    {
+      "epoch": 0.9559837597473738,
+      "grad_norm": 0.002502730599271993,
+      "learning_rate": 4.969566774078053e-07,
+      "loss": 0.0,
+      "step": 14834
+    },
+    {
+      "epoch": 0.956048205194303,
+      "grad_norm": 0.039794989158713906,
+      "learning_rate": 4.962406015037594e-07,
+      "loss": 0.0,
+      "step": 14835
+    },
+    {
+      "epoch": 0.9561126506412322,
+      "grad_norm": 0.05801081561332805,
+      "learning_rate": 4.955245255997136e-07,
+      "loss": 0.0004,
+      "step": 14836
+    },
+    {
+      "epoch": 0.9561770960881614,
+      "grad_norm": 0.5109591488232742,
+      "learning_rate": 4.948084496956678e-07,
+      "loss": 0.0038,
+      "step": 14837
+    },
+    {
+      "epoch": 0.9562415415350906,
+      "grad_norm": 0.0008859893281252492,
+      "learning_rate": 4.940923737916219e-07,
+      "loss": 0.0,
+      "step": 14838
+    },
+    {
+      "epoch": 0.9563059869820197,
+      "grad_norm": 0.0004626148219854151,
+      "learning_rate": 4.933762978875761e-07,
+      "loss": 0.0,
+      "step": 14839
+    },
+    {
+      "epoch": 0.9563704324289489,
+      "grad_norm": 0.13976812235699776,
+      "learning_rate": 4.926602219835303e-07,
+      "loss": 0.0006,
+      "step": 14840
+    },
+    {
+      "epoch": 0.9564348778758781,
+      "grad_norm": 0.0003939018396175427,
+      "learning_rate": 4.919441460794845e-07,
+      "loss": 0.0,
+      "step": 14841
+    },
+    {
+      "epoch": 0.9564993233228073,
+      "grad_norm": 0.004048120224935769,
+      "learning_rate": 4.912280701754387e-07,
+      "loss": 0.0,
+      "step": 14842
+    },
+    {
+      "epoch": 0.9565637687697364,
+      "grad_norm": 0.2624320373657076,
+      "learning_rate": 4.905119942713928e-07,
+      "loss": 0.0012,
+      "step": 14843
+    },
+    {
+      "epoch": 0.9566282142166656,
+      "grad_norm": 0.014094358110339713,
+      "learning_rate": 4.897959183673469e-07,
+      "loss": 0.0,
+      "step": 14844
+    },
+    {
+      "epoch": 0.9566926596635947,
+      "grad_norm": 0.0006931803598242944,
+      "learning_rate": 4.890798424633011e-07,
+      "loss": 0.0,
+      "step": 14845
+    },
+    {
+      "epoch": 0.9567571051105239,
+      "grad_norm": 0.026155114545255122,
+      "learning_rate": 4.883637665592553e-07,
+      "loss": 0.0001,
+      "step": 14846
+    },
+    {
+      "epoch": 0.9568215505574531,
+      "grad_norm": 0.0004005979354409368,
+      "learning_rate": 4.876476906552095e-07,
+      "loss": 0.0,
+      "step": 14847
+    },
+    {
+      "epoch": 0.9568859960043823,
+      "grad_norm": 0.05163703402803826,
+      "learning_rate": 4.869316147511637e-07,
+      "loss": 0.0002,
+      "step": 14848
+    },
+    {
+      "epoch": 0.9569504414513115,
+      "grad_norm": 0.00026671949687704445,
+      "learning_rate": 4.862155388471179e-07,
+      "loss": 0.0,
+      "step": 14849
+    },
+    {
+      "epoch": 0.9570148868982407,
+      "grad_norm": 0.17085980114842259,
+      "learning_rate": 4.85499462943072e-07,
+      "loss": 0.0001,
+      "step": 14850
+    },
+    {
+      "epoch": 0.9570793323451698,
+      "grad_norm": 0.023131071071852638,
+      "learning_rate": 4.847833870390262e-07,
+      "loss": 0.0,
+      "step": 14851
+    },
+    {
+      "epoch": 0.957143777792099,
+      "grad_norm": 0.0025030747163234074,
+      "learning_rate": 4.840673111349804e-07,
+      "loss": 0.0,
+      "step": 14852
+    },
+    {
+      "epoch": 0.9572082232390282,
+      "grad_norm": 0.0014296599078724075,
+      "learning_rate": 4.833512352309346e-07,
+      "loss": 0.0,
+      "step": 14853
+    },
+    {
+      "epoch": 0.9572726686859573,
+      "grad_norm": 0.00019132692676034923,
+      "learning_rate": 4.826351593268887e-07,
+      "loss": 0.0,
+      "step": 14854
+    },
+    {
+      "epoch": 0.9573371141328865,
+      "grad_norm": 0.000852706506334407,
+      "learning_rate": 4.819190834228428e-07,
+      "loss": 0.0,
+      "step": 14855
+    },
+    {
+      "epoch": 0.9574015595798157,
+      "grad_norm": 0.002841819849834939,
+      "learning_rate": 4.81203007518797e-07,
+      "loss": 0.0,
+      "step": 14856
+    },
+    {
+      "epoch": 0.9574660050267448,
+      "grad_norm": 0.2582747621177174,
+      "learning_rate": 4.804869316147512e-07,
+      "loss": 0.0062,
+      "step": 14857
+    },
+    {
+      "epoch": 0.957530450473674,
+      "grad_norm": 0.0001664016880520021,
+      "learning_rate": 4.797708557107054e-07,
+      "loss": 0.0,
+      "step": 14858
+    },
+    {
+      "epoch": 0.9575948959206032,
+      "grad_norm": 0.03976250129414592,
+      "learning_rate": 4.790547798066596e-07,
+      "loss": 0.0,
+      "step": 14859
+    },
+    {
+      "epoch": 0.9576593413675324,
+      "grad_norm": 0.007408347848849625,
+      "learning_rate": 4.783387039026137e-07,
+      "loss": 0.0,
+      "step": 14860
+    },
+    {
+      "epoch": 0.9577237868144616,
+      "grad_norm": 0.0006574807862907933,
+      "learning_rate": 4.776226279985679e-07,
+      "loss": 0.0,
+      "step": 14861
+    },
+    {
+      "epoch": 0.9577882322613908,
+      "grad_norm": 0.0028955546851931594,
+      "learning_rate": 4.769065520945221e-07,
+      "loss": 0.0,
+      "step": 14862
+    },
+    {
+      "epoch": 0.9578526777083199,
+      "grad_norm": 0.007108678630948101,
+      "learning_rate": 4.7619047619047623e-07,
+      "loss": 0.0001,
+      "step": 14863
+    },
+    {
+      "epoch": 0.9579171231552491,
+      "grad_norm": 0.27302020792699677,
+      "learning_rate": 4.754744002864304e-07,
+      "loss": 0.002,
+      "step": 14864
+    },
+    {
+      "epoch": 0.9579815686021782,
+      "grad_norm": 0.00322635562029426,
+      "learning_rate": 4.747583243823846e-07,
+      "loss": 0.0,
+      "step": 14865
+    },
+    {
+      "epoch": 0.9580460140491074,
+      "grad_norm": 0.001582561611280091,
+      "learning_rate": 4.740422484783387e-07,
+      "loss": 0.0,
+      "step": 14866
+    },
+    {
+      "epoch": 0.9581104594960366,
+      "grad_norm": 0.03473567228204857,
+      "learning_rate": 4.733261725742929e-07,
+      "loss": 0.0001,
+      "step": 14867
+    },
+    {
+      "epoch": 0.9581749049429658,
+      "grad_norm": 0.004032275081067416,
+      "learning_rate": 4.726100966702471e-07,
+      "loss": 0.0,
+      "step": 14868
+    },
+    {
+      "epoch": 0.9582393503898949,
+      "grad_norm": 0.005672604264508217,
+      "learning_rate": 4.718940207662012e-07,
+      "loss": 0.0,
+      "step": 14869
+    },
+    {
+      "epoch": 0.9583037958368241,
+      "grad_norm": 0.0024886796876244506,
+      "learning_rate": 4.711779448621554e-07,
+      "loss": 0.0,
+      "step": 14870
+    },
+    {
+      "epoch": 0.9583682412837533,
+      "grad_norm": 0.38062694010373216,
+      "learning_rate": 4.704618689581096e-07,
+      "loss": 0.0015,
+      "step": 14871
+    },
+    {
+      "epoch": 0.9584326867306825,
+      "grad_norm": 0.0013407293218750376,
+      "learning_rate": 4.697457930540638e-07,
+      "loss": 0.0,
+      "step": 14872
+    },
+    {
+      "epoch": 0.9584971321776117,
+      "grad_norm": 0.12025046514185207,
+      "learning_rate": 4.6902971715001794e-07,
+      "loss": 0.0003,
+      "step": 14873
+    },
+    {
+      "epoch": 0.9585615776245409,
+      "grad_norm": 0.013953009693871022,
+      "learning_rate": 4.6831364124597214e-07,
+      "loss": 0.0,
+      "step": 14874
+    },
+    {
+      "epoch": 0.95862602307147,
+      "grad_norm": 0.000177093544672877,
+      "learning_rate": 4.6759756534192633e-07,
+      "loss": 0.0,
+      "step": 14875
+    },
+    {
+      "epoch": 0.9586904685183992,
+      "grad_norm": 0.011330444616391867,
+      "learning_rate": 4.668814894378804e-07,
+      "loss": 0.0,
+      "step": 14876
+    },
+    {
+      "epoch": 0.9587549139653283,
+      "grad_norm": 0.0004049873983944743,
+      "learning_rate": 4.661654135338346e-07,
+      "loss": 0.0,
+      "step": 14877
+    },
+    {
+      "epoch": 0.9588193594122575,
+      "grad_norm": 0.0850805202870303,
+      "learning_rate": 4.654493376297888e-07,
+      "loss": 0.0016,
+      "step": 14878
+    },
+    {
+      "epoch": 0.9588838048591867,
+      "grad_norm": 0.0004410744841034204,
+      "learning_rate": 4.6473326172574294e-07,
+      "loss": 0.0,
+      "step": 14879
+    },
+    {
+      "epoch": 0.9589482503061159,
+      "grad_norm": 0.011639213360409064,
+      "learning_rate": 4.6401718582169713e-07,
+      "loss": 0.0,
+      "step": 14880
+    },
+    {
+      "epoch": 0.959012695753045,
+      "grad_norm": 0.008735612177839278,
+      "learning_rate": 4.6330110991765133e-07,
+      "loss": 0.0,
+      "step": 14881
+    },
+    {
+      "epoch": 0.9590771411999742,
+      "grad_norm": 0.020911397296875626,
+      "learning_rate": 4.6258503401360547e-07,
+      "loss": 0.0,
+      "step": 14882
+    },
+    {
+      "epoch": 0.9591415866469034,
+      "grad_norm": 0.001564329687710305,
+      "learning_rate": 4.6186895810955966e-07,
+      "loss": 0.0,
+      "step": 14883
+    },
+    {
+      "epoch": 0.9592060320938326,
+      "grad_norm": 0.0016710818587629625,
+      "learning_rate": 4.6115288220551385e-07,
+      "loss": 0.0,
+      "step": 14884
+    },
+    {
+      "epoch": 0.9592704775407618,
+      "grad_norm": 0.0039500020524210425,
+      "learning_rate": 4.6043680630146794e-07,
+      "loss": 0.0,
+      "step": 14885
+    },
+    {
+      "epoch": 0.959334922987691,
+      "grad_norm": 0.00036652025819384343,
+      "learning_rate": 4.5972073039742213e-07,
+      "loss": 0.0,
+      "step": 14886
+    },
+    {
+      "epoch": 0.9593993684346201,
+      "grad_norm": 0.06222436145498517,
+      "learning_rate": 4.590046544933763e-07,
+      "loss": 0.0001,
+      "step": 14887
+    },
+    {
+      "epoch": 0.9594638138815492,
+      "grad_norm": 0.0022996133312288224,
+      "learning_rate": 4.582885785893305e-07,
+      "loss": 0.0,
+      "step": 14888
+    },
+    {
+      "epoch": 0.9595282593284784,
+      "grad_norm": 0.018717318945592407,
+      "learning_rate": 4.5757250268528466e-07,
+      "loss": 0.0001,
+      "step": 14889
+    },
+    {
+      "epoch": 0.9595927047754076,
+      "grad_norm": 0.0005461212142547301,
+      "learning_rate": 4.5685642678123885e-07,
+      "loss": 0.0,
+      "step": 14890
+    },
+    {
+      "epoch": 0.9596571502223368,
+      "grad_norm": 0.0004114048163026691,
+      "learning_rate": 4.5614035087719304e-07,
+      "loss": 0.0,
+      "step": 14891
+    },
+    {
+      "epoch": 0.959721595669266,
+      "grad_norm": 0.0038923848919269173,
+      "learning_rate": 4.554242749731472e-07,
+      "loss": 0.0,
+      "step": 14892
+    },
+    {
+      "epoch": 0.9597860411161951,
+      "grad_norm": 0.005338129096024907,
+      "learning_rate": 4.547081990691014e-07,
+      "loss": 0.0,
+      "step": 14893
+    },
+    {
+      "epoch": 0.9598504865631243,
+      "grad_norm": 0.04265827342489956,
+      "learning_rate": 4.5399212316505557e-07,
+      "loss": 0.0,
+      "step": 14894
+    },
+    {
+      "epoch": 0.9599149320100535,
+      "grad_norm": 0.00013671547659063613,
+      "learning_rate": 4.532760472610097e-07,
+      "loss": 0.0,
+      "step": 14895
+    },
+    {
+      "epoch": 0.9599793774569827,
+      "grad_norm": 0.00960232425371068,
+      "learning_rate": 4.525599713569639e-07,
+      "loss": 0.0,
+      "step": 14896
+    },
+    {
+      "epoch": 0.9600438229039119,
+      "grad_norm": 0.00965865226960607,
+      "learning_rate": 4.518438954529181e-07,
+      "loss": 0.0,
+      "step": 14897
+    },
+    {
+      "epoch": 0.9601082683508411,
+      "grad_norm": 9.192379356709644e-05,
+      "learning_rate": 4.511278195488722e-07,
+      "loss": 0.0,
+      "step": 14898
+    },
+    {
+      "epoch": 0.9601727137977701,
+      "grad_norm": 0.00015305449739958055,
+      "learning_rate": 4.504117436448264e-07,
+      "loss": 0.0,
+      "step": 14899
+    },
+    {
+      "epoch": 0.9602371592446993,
+      "grad_norm": 0.14304743566147973,
+      "learning_rate": 4.4969566774078057e-07,
+      "loss": 0.0003,
+      "step": 14900
+    },
+    {
+      "epoch": 0.9603016046916285,
+      "grad_norm": 0.009295935886398593,
+      "learning_rate": 4.489795918367347e-07,
+      "loss": 0.0,
+      "step": 14901
+    },
+    {
+      "epoch": 0.9603660501385577,
+      "grad_norm": 0.0002162139983364177,
+      "learning_rate": 4.482635159326889e-07,
+      "loss": 0.0,
+      "step": 14902
+    },
+    {
+      "epoch": 0.9604304955854869,
+      "grad_norm": 0.03242333980491538,
+      "learning_rate": 4.475474400286431e-07,
+      "loss": 0.0001,
+      "step": 14903
+    },
+    {
+      "epoch": 0.9604949410324161,
+      "grad_norm": 0.002406562227087017,
+      "learning_rate": 4.4683136412459723e-07,
+      "loss": 0.0,
+      "step": 14904
+    },
+    {
+      "epoch": 0.9605593864793452,
+      "grad_norm": 0.0056715677897925445,
+      "learning_rate": 4.461152882205514e-07,
+      "loss": 0.0,
+      "step": 14905
+    },
+    {
+      "epoch": 0.9606238319262744,
+      "grad_norm": 0.00045505831911909236,
+      "learning_rate": 4.453992123165056e-07,
+      "loss": 0.0,
+      "step": 14906
+    },
+    {
+      "epoch": 0.9606882773732036,
+      "grad_norm": 0.0007268449933700111,
+      "learning_rate": 4.446831364124598e-07,
+      "loss": 0.0,
+      "step": 14907
+    },
+    {
+      "epoch": 0.9607527228201328,
+      "grad_norm": 0.00022912322387547547,
+      "learning_rate": 4.439670605084139e-07,
+      "loss": 0.0,
+      "step": 14908
+    },
+    {
+      "epoch": 0.960817168267062,
+      "grad_norm": 0.0022316612740019702,
+      "learning_rate": 4.432509846043681e-07,
+      "loss": 0.0,
+      "step": 14909
+    },
+    {
+      "epoch": 0.9608816137139911,
+      "grad_norm": 0.004242630686752434,
+      "learning_rate": 4.425349087003223e-07,
+      "loss": 0.0,
+      "step": 14910
+    },
+    {
+      "epoch": 0.9609460591609202,
+      "grad_norm": 0.04213162492549318,
+      "learning_rate": 4.418188327962764e-07,
+      "loss": 0.0001,
+      "step": 14911
+    },
+    {
+      "epoch": 0.9610105046078494,
+      "grad_norm": 0.0008370413518882776,
+      "learning_rate": 4.411027568922306e-07,
+      "loss": 0.0,
+      "step": 14912
+    },
+    {
+      "epoch": 0.9610749500547786,
+      "grad_norm": 0.0009282456515276538,
+      "learning_rate": 4.403866809881848e-07,
+      "loss": 0.0,
+      "step": 14913
+    },
+    {
+      "epoch": 0.9611393955017078,
+      "grad_norm": 0.0004216024907836894,
+      "learning_rate": 4.3967060508413895e-07,
+      "loss": 0.0,
+      "step": 14914
+    },
+    {
+      "epoch": 0.961203840948637,
+      "grad_norm": 0.004501349239021442,
+      "learning_rate": 4.3895452918009314e-07,
+      "loss": 0.0,
+      "step": 14915
+    },
+    {
+      "epoch": 0.9612682863955662,
+      "grad_norm": 0.007865435385491804,
+      "learning_rate": 4.3823845327604733e-07,
+      "loss": 0.0001,
+      "step": 14916
+    },
+    {
+      "epoch": 0.9613327318424953,
+      "grad_norm": 0.0001713778914629011,
+      "learning_rate": 4.375223773720014e-07,
+      "loss": 0.0,
+      "step": 14917
+    },
+    {
+      "epoch": 0.9613971772894245,
+      "grad_norm": 0.1708792157943735,
+      "learning_rate": 4.368063014679556e-07,
+      "loss": 0.0007,
+      "step": 14918
+    },
+    {
+      "epoch": 0.9614616227363537,
+      "grad_norm": 0.000479988725056364,
+      "learning_rate": 4.360902255639098e-07,
+      "loss": 0.0,
+      "step": 14919
+    },
+    {
+      "epoch": 0.9615260681832829,
+      "grad_norm": 0.006377414893212848,
+      "learning_rate": 4.3537414965986395e-07,
+      "loss": 0.0,
+      "step": 14920
+    },
+    {
+      "epoch": 0.961590513630212,
+      "grad_norm": 0.0036946523398398512,
+      "learning_rate": 4.3465807375581814e-07,
+      "loss": 0.0,
+      "step": 14921
+    },
+    {
+      "epoch": 0.9616549590771412,
+      "grad_norm": 0.014419668187101007,
+      "learning_rate": 4.3394199785177233e-07,
+      "loss": 0.0,
+      "step": 14922
+    },
+    {
+      "epoch": 0.9617194045240703,
+      "grad_norm": 0.011962387624208598,
+      "learning_rate": 4.3322592194772647e-07,
+      "loss": 0.0001,
+      "step": 14923
+    },
+    {
+      "epoch": 0.9617838499709995,
+      "grad_norm": 0.012241861601313967,
+      "learning_rate": 4.3250984604368067e-07,
+      "loss": 0.0,
+      "step": 14924
+    },
+    {
+      "epoch": 0.9618482954179287,
+      "grad_norm": 0.0005817396474641864,
+      "learning_rate": 4.3179377013963486e-07,
+      "loss": 0.0,
+      "step": 14925
+    },
+    {
+      "epoch": 0.9619127408648579,
+      "grad_norm": 0.0004082905114981401,
+      "learning_rate": 4.3107769423558905e-07,
+      "loss": 0.0,
+      "step": 14926
+    },
+    {
+      "epoch": 0.9619771863117871,
+      "grad_norm": 0.0023784330583258473,
+      "learning_rate": 4.3036161833154314e-07,
+      "loss": 0.0,
+      "step": 14927
+    },
+    {
+      "epoch": 0.9620416317587163,
+      "grad_norm": 0.002489868312757046,
+      "learning_rate": 4.2964554242749733e-07,
+      "loss": 0.0,
+      "step": 14928
+    },
+    {
+      "epoch": 0.9621060772056454,
+      "grad_norm": 0.003762379342939597,
+      "learning_rate": 4.289294665234515e-07,
+      "loss": 0.0,
+      "step": 14929
+    },
+    {
+      "epoch": 0.9621705226525746,
+      "grad_norm": 0.0017895349142890476,
+      "learning_rate": 4.2821339061940566e-07,
+      "loss": 0.0,
+      "step": 14930
+    },
+    {
+      "epoch": 0.9622349680995038,
+      "grad_norm": 0.0004923169026729873,
+      "learning_rate": 4.2749731471535986e-07,
+      "loss": 0.0,
+      "step": 14931
+    },
+    {
+      "epoch": 0.9622994135464329,
+      "grad_norm": 0.00017590920193244695,
+      "learning_rate": 4.2678123881131405e-07,
+      "loss": 0.0,
+      "step": 14932
+    },
+    {
+      "epoch": 0.9623638589933621,
+      "grad_norm": 0.00048500241951370985,
+      "learning_rate": 4.260651629072682e-07,
+      "loss": 0.0,
+      "step": 14933
+    },
+    {
+      "epoch": 0.9624283044402913,
+      "grad_norm": 0.0010236614349840823,
+      "learning_rate": 4.253490870032224e-07,
+      "loss": 0.0,
+      "step": 14934
+    },
+    {
+      "epoch": 0.9624927498872204,
+      "grad_norm": 0.0016053604449928272,
+      "learning_rate": 4.246330110991766e-07,
+      "loss": 0.0,
+      "step": 14935
+    },
+    {
+      "epoch": 0.9625571953341496,
+      "grad_norm": 0.0012853831294313406,
+      "learning_rate": 4.239169351951307e-07,
+      "loss": 0.0,
+      "step": 14936
+    },
+    {
+      "epoch": 0.9626216407810788,
+      "grad_norm": 0.0011035914182909066,
+      "learning_rate": 4.232008592910849e-07,
+      "loss": 0.0,
+      "step": 14937
+    },
+    {
+      "epoch": 0.962686086228008,
+      "grad_norm": 0.07527333398147694,
+      "learning_rate": 4.224847833870391e-07,
+      "loss": 0.0008,
+      "step": 14938
+    },
+    {
+      "epoch": 0.9627505316749372,
+      "grad_norm": 0.03563464857298093,
+      "learning_rate": 4.217687074829932e-07,
+      "loss": 0.0001,
+      "step": 14939
+    },
+    {
+      "epoch": 0.9628149771218664,
+      "grad_norm": 0.02769180787150801,
+      "learning_rate": 4.210526315789474e-07,
+      "loss": 0.0,
+      "step": 14940
+    },
+    {
+      "epoch": 0.9628794225687956,
+      "grad_norm": 0.0017228653770633396,
+      "learning_rate": 4.2033655567490157e-07,
+      "loss": 0.0,
+      "step": 14941
+    },
+    {
+      "epoch": 0.9629438680157247,
+      "grad_norm": 0.013080418369412865,
+      "learning_rate": 4.1962047977085577e-07,
+      "loss": 0.0001,
+      "step": 14942
+    },
+    {
+      "epoch": 0.9630083134626538,
+      "grad_norm": 0.33033698275444734,
+      "learning_rate": 4.189044038668099e-07,
+      "loss": 0.0008,
+      "step": 14943
+    },
+    {
+      "epoch": 0.963072758909583,
+      "grad_norm": 0.004267734442137806,
+      "learning_rate": 4.181883279627641e-07,
+      "loss": 0.0,
+      "step": 14944
+    },
+    {
+      "epoch": 0.9631372043565122,
+      "grad_norm": 0.01560012273070935,
+      "learning_rate": 4.174722520587183e-07,
+      "loss": 0.0,
+      "step": 14945
+    },
+    {
+      "epoch": 0.9632016498034414,
+      "grad_norm": 0.0043316884787335036,
+      "learning_rate": 4.1675617615467243e-07,
+      "loss": 0.0,
+      "step": 14946
+    },
+    {
+      "epoch": 0.9632660952503705,
+      "grad_norm": 0.00013166375289761843,
+      "learning_rate": 4.160401002506266e-07,
+      "loss": 0.0,
+      "step": 14947
+    },
+    {
+      "epoch": 0.9633305406972997,
+      "grad_norm": 0.0010617943515079224,
+      "learning_rate": 4.153240243465808e-07,
+      "loss": 0.0,
+      "step": 14948
+    },
+    {
+      "epoch": 0.9633949861442289,
+      "grad_norm": 0.06773688748923617,
+      "learning_rate": 4.146079484425349e-07,
+      "loss": 0.0001,
+      "step": 14949
+    },
+    {
+      "epoch": 0.9634594315911581,
+      "grad_norm": 0.03567411390503899,
+      "learning_rate": 4.138918725384891e-07,
+      "loss": 0.0004,
+      "step": 14950
+    },
+    {
+      "epoch": 0.9635238770380873,
+      "grad_norm": 6.775566679261235e-05,
+      "learning_rate": 4.131757966344433e-07,
+      "loss": 0.0,
+      "step": 14951
+    },
+    {
+      "epoch": 0.9635883224850165,
+      "grad_norm": 0.04405645647140258,
+      "learning_rate": 4.1245972073039743e-07,
+      "loss": 0.0001,
+      "step": 14952
+    },
+    {
+      "epoch": 0.9636527679319457,
+      "grad_norm": 0.00023435635260378457,
+      "learning_rate": 4.117436448263516e-07,
+      "loss": 0.0,
+      "step": 14953
+    },
+    {
+      "epoch": 0.9637172133788748,
+      "grad_norm": 0.1547491531137055,
+      "learning_rate": 4.110275689223058e-07,
+      "loss": 0.0002,
+      "step": 14954
+    },
+    {
+      "epoch": 0.9637816588258039,
+      "grad_norm": 0.561563241625908,
+      "learning_rate": 4.1031149301825995e-07,
+      "loss": 0.0035,
+      "step": 14955
+    },
+    {
+      "epoch": 0.9638461042727331,
+      "grad_norm": 0.005156861571168343,
+      "learning_rate": 4.0959541711421415e-07,
+      "loss": 0.0001,
+      "step": 14956
+    },
+    {
+      "epoch": 0.9639105497196623,
+      "grad_norm": 0.19693480029990224,
+      "learning_rate": 4.0887934121016834e-07,
+      "loss": 0.0006,
+      "step": 14957
+    },
+    {
+      "epoch": 0.9639749951665915,
+      "grad_norm": 0.0005532097400886752,
+      "learning_rate": 4.0816326530612243e-07,
+      "loss": 0.0,
+      "step": 14958
+    },
+    {
+      "epoch": 0.9640394406135206,
+      "grad_norm": 0.00238901745451577,
+      "learning_rate": 4.074471894020766e-07,
+      "loss": 0.0,
+      "step": 14959
+    },
+    {
+      "epoch": 0.9641038860604498,
+      "grad_norm": 0.0026651981924149973,
+      "learning_rate": 4.067311134980308e-07,
+      "loss": 0.0,
+      "step": 14960
+    },
+    {
+      "epoch": 0.964168331507379,
+      "grad_norm": 0.003920529480750357,
+      "learning_rate": 4.06015037593985e-07,
+      "loss": 0.0,
+      "step": 14961
+    },
+    {
+      "epoch": 0.9642327769543082,
+      "grad_norm": 0.003843617604241777,
+      "learning_rate": 4.0529896168993915e-07,
+      "loss": 0.0,
+      "step": 14962
+    },
+    {
+      "epoch": 0.9642972224012374,
+      "grad_norm": 0.0025857098032880323,
+      "learning_rate": 4.0458288578589334e-07,
+      "loss": 0.0,
+      "step": 14963
+    },
+    {
+      "epoch": 0.9643616678481666,
+      "grad_norm": 0.018335899444566185,
+      "learning_rate": 4.0386680988184753e-07,
+      "loss": 0.0,
+      "step": 14964
+    },
+    {
+      "epoch": 0.9644261132950958,
+      "grad_norm": 0.0009981560033705578,
+      "learning_rate": 4.0315073397780167e-07,
+      "loss": 0.0,
+      "step": 14965
+    },
+    {
+      "epoch": 0.9644905587420248,
+      "grad_norm": 0.0009387569199021039,
+      "learning_rate": 4.0243465807375586e-07,
+      "loss": 0.0,
+      "step": 14966
+    },
+    {
+      "epoch": 0.964555004188954,
+      "grad_norm": 0.019251289181688305,
+      "learning_rate": 4.0171858216971006e-07,
+      "loss": 0.0001,
+      "step": 14967
+    },
+    {
+      "epoch": 0.9646194496358832,
+      "grad_norm": 0.00010738621659829706,
+      "learning_rate": 4.010025062656642e-07,
+      "loss": 0.0,
+      "step": 14968
+    },
+    {
+      "epoch": 0.9646838950828124,
+      "grad_norm": 0.03203928420734971,
+      "learning_rate": 4.002864303616184e-07,
+      "loss": 0.0001,
+      "step": 14969
+    },
+    {
+      "epoch": 0.9647483405297416,
+      "grad_norm": 0.0003362348993916054,
+      "learning_rate": 3.995703544575726e-07,
+      "loss": 0.0,
+      "step": 14970
+    },
+    {
+      "epoch": 0.9648127859766708,
+      "grad_norm": 0.000265084003407136,
+      "learning_rate": 3.9885427855352667e-07,
+      "loss": 0.0,
+      "step": 14971
+    },
+    {
+      "epoch": 0.9648772314235999,
+      "grad_norm": 0.004495177856111423,
+      "learning_rate": 3.9813820264948086e-07,
+      "loss": 0.0,
+      "step": 14972
+    },
+    {
+      "epoch": 0.9649416768705291,
+      "grad_norm": 0.04050054439674393,
+      "learning_rate": 3.9742212674543505e-07,
+      "loss": 0.0001,
+      "step": 14973
+    },
+    {
+      "epoch": 0.9650061223174583,
+      "grad_norm": 6.170983398313063e-05,
+      "learning_rate": 3.967060508413892e-07,
+      "loss": 0.0,
+      "step": 14974
+    },
+    {
+      "epoch": 0.9650705677643875,
+      "grad_norm": 0.00017663978023309493,
+      "learning_rate": 3.959899749373434e-07,
+      "loss": 0.0,
+      "step": 14975
+    },
+    {
+      "epoch": 0.9651350132113167,
+      "grad_norm": 5.699082884079711e-05,
+      "learning_rate": 3.952738990332976e-07,
+      "loss": 0.0,
+      "step": 14976
+    },
+    {
+      "epoch": 0.9651994586582457,
+      "grad_norm": 0.000602340558573276,
+      "learning_rate": 3.945578231292517e-07,
+      "loss": 0.0,
+      "step": 14977
+    },
+    {
+      "epoch": 0.9652639041051749,
+      "grad_norm": 0.0006447693375527919,
+      "learning_rate": 3.938417472252059e-07,
+      "loss": 0.0,
+      "step": 14978
+    },
+    {
+      "epoch": 0.9653283495521041,
+      "grad_norm": 0.005086575180647127,
+      "learning_rate": 3.931256713211601e-07,
+      "loss": 0.0,
+      "step": 14979
+    },
+    {
+      "epoch": 0.9653927949990333,
+      "grad_norm": 0.18404470009542506,
+      "learning_rate": 3.924095954171143e-07,
+      "loss": 0.0021,
+      "step": 14980
+    },
+    {
+      "epoch": 0.9654572404459625,
+      "grad_norm": 0.0012513134685006979,
+      "learning_rate": 3.916935195130684e-07,
+      "loss": 0.0,
+      "step": 14981
+    },
+    {
+      "epoch": 0.9655216858928917,
+      "grad_norm": 0.0026647520580636358,
+      "learning_rate": 3.909774436090226e-07,
+      "loss": 0.0,
+      "step": 14982
+    },
+    {
+      "epoch": 0.9655861313398209,
+      "grad_norm": 0.00042825462339338795,
+      "learning_rate": 3.9026136770497677e-07,
+      "loss": 0.0,
+      "step": 14983
+    },
+    {
+      "epoch": 0.96565057678675,
+      "grad_norm": 0.07452899937460482,
+      "learning_rate": 3.895452918009309e-07,
+      "loss": 0.0001,
+      "step": 14984
+    },
+    {
+      "epoch": 0.9657150222336792,
+      "grad_norm": 0.0003942303834534447,
+      "learning_rate": 3.888292158968851e-07,
+      "loss": 0.0,
+      "step": 14985
+    },
+    {
+      "epoch": 0.9657794676806084,
+      "grad_norm": 0.000612557581035135,
+      "learning_rate": 3.881131399928393e-07,
+      "loss": 0.0,
+      "step": 14986
+    },
+    {
+      "epoch": 0.9658439131275376,
+      "grad_norm": 0.015846453551998426,
+      "learning_rate": 3.8739706408879344e-07,
+      "loss": 0.0,
+      "step": 14987
+    },
+    {
+      "epoch": 0.9659083585744667,
+      "grad_norm": 0.10003049039179873,
+      "learning_rate": 3.8668098818474763e-07,
+      "loss": 0.0012,
+      "step": 14988
+    },
+    {
+      "epoch": 0.9659728040213958,
+      "grad_norm": 0.00026728753770736686,
+      "learning_rate": 3.859649122807018e-07,
+      "loss": 0.0,
+      "step": 14989
+    },
+    {
+      "epoch": 0.966037249468325,
+      "grad_norm": 0.0002459140433178353,
+      "learning_rate": 3.852488363766559e-07,
+      "loss": 0.0,
+      "step": 14990
+    },
+    {
+      "epoch": 0.9661016949152542,
+      "grad_norm": 0.0005397851917570067,
+      "learning_rate": 3.845327604726101e-07,
+      "loss": 0.0,
+      "step": 14991
+    },
+    {
+      "epoch": 0.9661661403621834,
+      "grad_norm": 0.00021344605552674962,
+      "learning_rate": 3.838166845685643e-07,
+      "loss": 0.0,
+      "step": 14992
+    },
+    {
+      "epoch": 0.9662305858091126,
+      "grad_norm": 0.0004351620778108871,
+      "learning_rate": 3.8310060866451843e-07,
+      "loss": 0.0,
+      "step": 14993
+    },
+    {
+      "epoch": 0.9662950312560418,
+      "grad_norm": 0.000685698942079955,
+      "learning_rate": 3.8238453276047263e-07,
+      "loss": 0.0,
+      "step": 14994
+    },
+    {
+      "epoch": 0.966359476702971,
+      "grad_norm": 0.04826283823719522,
+      "learning_rate": 3.816684568564268e-07,
+      "loss": 0.0003,
+      "step": 14995
+    },
+    {
+      "epoch": 0.9664239221499001,
+      "grad_norm": 0.0009347700493750752,
+      "learning_rate": 3.80952380952381e-07,
+      "loss": 0.0,
+      "step": 14996
+    },
+    {
+      "epoch": 0.9664883675968293,
+      "grad_norm": 6.683883255573541e-05,
+      "learning_rate": 3.8023630504833515e-07,
+      "loss": 0.0,
+      "step": 14997
+    },
+    {
+      "epoch": 0.9665528130437585,
+      "grad_norm": 0.014872162732202817,
+      "learning_rate": 3.7952022914428935e-07,
+      "loss": 0.0,
+      "step": 14998
+    },
+    {
+      "epoch": 0.9666172584906876,
+      "grad_norm": 0.002071540981744842,
+      "learning_rate": 3.7880415324024354e-07,
+      "loss": 0.0,
+      "step": 14999
+    },
+    {
+      "epoch": 0.9666817039376168,
+      "grad_norm": 0.0568428519555771,
+      "learning_rate": 3.780880773361976e-07,
+      "loss": 0.0004,
+      "step": 15000
+    },
+    {
+      "epoch": 0.966746149384546,
+      "grad_norm": 0.008794983717003983,
+      "learning_rate": 3.773720014321518e-07,
+      "loss": 0.0,
+      "step": 15001
+    },
+    {
+      "epoch": 0.9668105948314751,
+      "grad_norm": 0.00263075135347942,
+      "learning_rate": 3.76655925528106e-07,
+      "loss": 0.0,
+      "step": 15002
+    },
+    {
+      "epoch": 0.9668750402784043,
+      "grad_norm": 0.00011607515633291526,
+      "learning_rate": 3.7593984962406015e-07,
+      "loss": 0.0,
+      "step": 15003
+    },
+    {
+      "epoch": 0.9669394857253335,
+      "grad_norm": 0.00035570977344336345,
+      "learning_rate": 3.7522377372001434e-07,
+      "loss": 0.0,
+      "step": 15004
+    },
+    {
+      "epoch": 0.9670039311722627,
+      "grad_norm": 0.0003867004197191581,
+      "learning_rate": 3.7450769781596854e-07,
+      "loss": 0.0,
+      "step": 15005
+    },
+    {
+      "epoch": 0.9670683766191919,
+      "grad_norm": 0.00025640017192041177,
+      "learning_rate": 3.737916219119227e-07,
+      "loss": 0.0,
+      "step": 15006
+    },
+    {
+      "epoch": 0.9671328220661211,
+      "grad_norm": 0.001456672819982795,
+      "learning_rate": 3.7307554600787687e-07,
+      "loss": 0.0,
+      "step": 15007
+    },
+    {
+      "epoch": 0.9671972675130502,
+      "grad_norm": 0.0014484291111371464,
+      "learning_rate": 3.7235947010383106e-07,
+      "loss": 0.0,
+      "step": 15008
+    },
+    {
+      "epoch": 0.9672617129599794,
+      "grad_norm": 0.0015187991078433156,
+      "learning_rate": 3.716433941997852e-07,
+      "loss": 0.0,
+      "step": 15009
+    },
+    {
+      "epoch": 0.9673261584069085,
+      "grad_norm": 0.0004086356212154145,
+      "learning_rate": 3.709273182957394e-07,
+      "loss": 0.0,
+      "step": 15010
+    },
+    {
+      "epoch": 0.9673906038538377,
+      "grad_norm": 0.0012860003776702935,
+      "learning_rate": 3.702112423916936e-07,
+      "loss": 0.0,
+      "step": 15011
+    },
+    {
+      "epoch": 0.9674550493007669,
+      "grad_norm": 0.0003662492167377044,
+      "learning_rate": 3.694951664876477e-07,
+      "loss": 0.0,
+      "step": 15012
+    },
+    {
+      "epoch": 0.967519494747696,
+      "grad_norm": 0.0007289835883782952,
+      "learning_rate": 3.6877909058360187e-07,
+      "loss": 0.0,
+      "step": 15013
+    },
+    {
+      "epoch": 0.9675839401946252,
+      "grad_norm": 0.312234252945699,
+      "learning_rate": 3.6806301467955606e-07,
+      "loss": 0.0006,
+      "step": 15014
+    },
+    {
+      "epoch": 0.9676483856415544,
+      "grad_norm": 0.00048179295287462684,
+      "learning_rate": 3.6734693877551025e-07,
+      "loss": 0.0,
+      "step": 15015
+    },
+    {
+      "epoch": 0.9677128310884836,
+      "grad_norm": 0.001208616655439765,
+      "learning_rate": 3.666308628714644e-07,
+      "loss": 0.0,
+      "step": 15016
+    },
+    {
+      "epoch": 0.9677772765354128,
+      "grad_norm": 0.3449243621045279,
+      "learning_rate": 3.659147869674186e-07,
+      "loss": 0.0002,
+      "step": 15017
+    },
+    {
+      "epoch": 0.967841721982342,
+      "grad_norm": 0.015200695074966152,
+      "learning_rate": 3.651987110633728e-07,
+      "loss": 0.0,
+      "step": 15018
+    },
+    {
+      "epoch": 0.9679061674292712,
+      "grad_norm": 0.0014013163488261668,
+      "learning_rate": 3.644826351593269e-07,
+      "loss": 0.0,
+      "step": 15019
+    },
+    {
+      "epoch": 0.9679706128762003,
+      "grad_norm": 0.0005596241722062707,
+      "learning_rate": 3.637665592552811e-07,
+      "loss": 0.0,
+      "step": 15020
+    },
+    {
+      "epoch": 0.9680350583231294,
+      "grad_norm": 0.0005016758906052871,
+      "learning_rate": 3.630504833512353e-07,
+      "loss": 0.0,
+      "step": 15021
+    },
+    {
+      "epoch": 0.9680995037700586,
+      "grad_norm": 0.0002737052037732511,
+      "learning_rate": 3.623344074471894e-07,
+      "loss": 0.0,
+      "step": 15022
+    },
+    {
+      "epoch": 0.9681639492169878,
+      "grad_norm": 0.004652262607627751,
+      "learning_rate": 3.616183315431436e-07,
+      "loss": 0.0,
+      "step": 15023
+    },
+    {
+      "epoch": 0.968228394663917,
+      "grad_norm": 0.0029658107489418965,
+      "learning_rate": 3.609022556390978e-07,
+      "loss": 0.0,
+      "step": 15024
+    },
+    {
+      "epoch": 0.9682928401108462,
+      "grad_norm": 0.0018419730604352967,
+      "learning_rate": 3.601861797350519e-07,
+      "loss": 0.0,
+      "step": 15025
+    },
+    {
+      "epoch": 0.9683572855577753,
+      "grad_norm": 0.0028526369216188116,
+      "learning_rate": 3.594701038310061e-07,
+      "loss": 0.0,
+      "step": 15026
+    },
+    {
+      "epoch": 0.9684217310047045,
+      "grad_norm": 0.0008088329827487265,
+      "learning_rate": 3.587540279269603e-07,
+      "loss": 0.0,
+      "step": 15027
+    },
+    {
+      "epoch": 0.9684861764516337,
+      "grad_norm": 0.10358512542605385,
+      "learning_rate": 3.5803795202291444e-07,
+      "loss": 0.0001,
+      "step": 15028
+    },
+    {
+      "epoch": 0.9685506218985629,
+      "grad_norm": 0.0004286461885383949,
+      "learning_rate": 3.5732187611886864e-07,
+      "loss": 0.0,
+      "step": 15029
+    },
+    {
+      "epoch": 0.9686150673454921,
+      "grad_norm": 0.0017663911047688466,
+      "learning_rate": 3.5660580021482283e-07,
+      "loss": 0.0,
+      "step": 15030
+    },
+    {
+      "epoch": 0.9686795127924213,
+      "grad_norm": 0.0016889736383674534,
+      "learning_rate": 3.558897243107769e-07,
+      "loss": 0.0,
+      "step": 15031
+    },
+    {
+      "epoch": 0.9687439582393504,
+      "grad_norm": 0.0320417512712144,
+      "learning_rate": 3.551736484067311e-07,
+      "loss": 0.0001,
+      "step": 15032
+    },
+    {
+      "epoch": 0.9688084036862795,
+      "grad_norm": 0.014156902284652106,
+      "learning_rate": 3.544575725026853e-07,
+      "loss": 0.0,
+      "step": 15033
+    },
+    {
+      "epoch": 0.9688728491332087,
+      "grad_norm": 0.0018575665054532513,
+      "learning_rate": 3.537414965986395e-07,
+      "loss": 0.0,
+      "step": 15034
+    },
+    {
+      "epoch": 0.9689372945801379,
+      "grad_norm": 0.3058661502755169,
+      "learning_rate": 3.5302542069459363e-07,
+      "loss": 0.0011,
+      "step": 15035
+    },
+    {
+      "epoch": 0.9690017400270671,
+      "grad_norm": 0.05986802368608069,
+      "learning_rate": 3.523093447905478e-07,
+      "loss": 0.0002,
+      "step": 15036
+    },
+    {
+      "epoch": 0.9690661854739963,
+      "grad_norm": 0.0004834545781512201,
+      "learning_rate": 3.51593268886502e-07,
+      "loss": 0.0,
+      "step": 15037
+    },
+    {
+      "epoch": 0.9691306309209254,
+      "grad_norm": 0.012819738966230294,
+      "learning_rate": 3.5087719298245616e-07,
+      "loss": 0.0,
+      "step": 15038
+    },
+    {
+      "epoch": 0.9691950763678546,
+      "grad_norm": 6.932870897949043e-05,
+      "learning_rate": 3.5016111707841035e-07,
+      "loss": 0.0,
+      "step": 15039
+    },
+    {
+      "epoch": 0.9692595218147838,
+      "grad_norm": 0.03340801142225263,
+      "learning_rate": 3.4944504117436454e-07,
+      "loss": 0.0001,
+      "step": 15040
+    },
+    {
+      "epoch": 0.969323967261713,
+      "grad_norm": 0.0023576116642467257,
+      "learning_rate": 3.487289652703187e-07,
+      "loss": 0.0,
+      "step": 15041
+    },
+    {
+      "epoch": 0.9693884127086422,
+      "grad_norm": 0.001895887451075661,
+      "learning_rate": 3.480128893662729e-07,
+      "loss": 0.0,
+      "step": 15042
+    },
+    {
+      "epoch": 0.9694528581555714,
+      "grad_norm": 0.0004433743816133635,
+      "learning_rate": 3.4729681346222707e-07,
+      "loss": 0.0,
+      "step": 15043
+    },
+    {
+      "epoch": 0.9695173036025004,
+      "grad_norm": 0.19629095086095616,
+      "learning_rate": 3.4658073755818116e-07,
+      "loss": 0.0004,
+      "step": 15044
+    },
+    {
+      "epoch": 0.9695817490494296,
+      "grad_norm": 0.00023726558542802232,
+      "learning_rate": 3.4586466165413535e-07,
+      "loss": 0.0,
+      "step": 15045
+    },
+    {
+      "epoch": 0.9696461944963588,
+      "grad_norm": 0.018993789490638787,
+      "learning_rate": 3.4514858575008954e-07,
+      "loss": 0.0,
+      "step": 15046
+    },
+    {
+      "epoch": 0.969710639943288,
+      "grad_norm": 0.0002520063939503362,
+      "learning_rate": 3.444325098460437e-07,
+      "loss": 0.0,
+      "step": 15047
+    },
+    {
+      "epoch": 0.9697750853902172,
+      "grad_norm": 0.005074428359961007,
+      "learning_rate": 3.437164339419979e-07,
+      "loss": 0.0,
+      "step": 15048
+    },
+    {
+      "epoch": 0.9698395308371464,
+      "grad_norm": 0.0015976258024932811,
+      "learning_rate": 3.4300035803795207e-07,
+      "loss": 0.0,
+      "step": 15049
+    },
+    {
+      "epoch": 0.9699039762840755,
+      "grad_norm": 0.0011570189577347902,
+      "learning_rate": 3.4228428213390626e-07,
+      "loss": 0.0,
+      "step": 15050
+    },
+    {
+      "epoch": 0.9699684217310047,
+      "grad_norm": 0.0467625252062478,
+      "learning_rate": 3.415682062298604e-07,
+      "loss": 0.0,
+      "step": 15051
+    },
+    {
+      "epoch": 0.9700328671779339,
+      "grad_norm": 0.10716094057074699,
+      "learning_rate": 3.408521303258146e-07,
+      "loss": 0.0001,
+      "step": 15052
+    },
+    {
+      "epoch": 0.9700973126248631,
+      "grad_norm": 0.0010008353049710968,
+      "learning_rate": 3.401360544217688e-07,
+      "loss": 0.0,
+      "step": 15053
+    },
+    {
+      "epoch": 0.9701617580717923,
+      "grad_norm": 0.05783591134718608,
+      "learning_rate": 3.3941997851772287e-07,
+      "loss": 0.0,
+      "step": 15054
+    },
+    {
+      "epoch": 0.9702262035187214,
+      "grad_norm": 0.01778828143835162,
+      "learning_rate": 3.3870390261367707e-07,
+      "loss": 0.0001,
+      "step": 15055
+    },
+    {
+      "epoch": 0.9702906489656505,
+      "grad_norm": 0.011812855447108965,
+      "learning_rate": 3.3798782670963126e-07,
+      "loss": 0.0,
+      "step": 15056
+    },
+    {
+      "epoch": 0.9703550944125797,
+      "grad_norm": 0.0007705646428616686,
+      "learning_rate": 3.372717508055854e-07,
+      "loss": 0.0,
+      "step": 15057
+    },
+    {
+      "epoch": 0.9704195398595089,
+      "grad_norm": 0.0002786041604228193,
+      "learning_rate": 3.365556749015396e-07,
+      "loss": 0.0,
+      "step": 15058
+    },
+    {
+      "epoch": 0.9704839853064381,
+      "grad_norm": 0.03147547745520999,
+      "learning_rate": 3.358395989974938e-07,
+      "loss": 0.0003,
+      "step": 15059
+    },
+    {
+      "epoch": 0.9705484307533673,
+      "grad_norm": 0.0025172349651695577,
+      "learning_rate": 3.351235230934479e-07,
+      "loss": 0.0,
+      "step": 15060
+    },
+    {
+      "epoch": 0.9706128762002965,
+      "grad_norm": 0.0004264914551491804,
+      "learning_rate": 3.344074471894021e-07,
+      "loss": 0.0,
+      "step": 15061
+    },
+    {
+      "epoch": 0.9706773216472256,
+      "grad_norm": 0.37015364172979875,
+      "learning_rate": 3.336913712853563e-07,
+      "loss": 0.001,
+      "step": 15062
+    },
+    {
+      "epoch": 0.9707417670941548,
+      "grad_norm": 0.0026231783325515613,
+      "learning_rate": 3.329752953813104e-07,
+      "loss": 0.0,
+      "step": 15063
+    },
+    {
+      "epoch": 0.970806212541084,
+      "grad_norm": 2.1080350788028017,
+      "learning_rate": 3.322592194772646e-07,
+      "loss": 0.0038,
+      "step": 15064
+    },
+    {
+      "epoch": 0.9708706579880132,
+      "grad_norm": 0.001236758915207383,
+      "learning_rate": 3.315431435732188e-07,
+      "loss": 0.0,
+      "step": 15065
+    },
+    {
+      "epoch": 0.9709351034349423,
+      "grad_norm": 0.0008858550028966706,
+      "learning_rate": 3.308270676691729e-07,
+      "loss": 0.0,
+      "step": 15066
+    },
+    {
+      "epoch": 0.9709995488818715,
+      "grad_norm": 0.012957684924999985,
+      "learning_rate": 3.301109917651271e-07,
+      "loss": 0.0001,
+      "step": 15067
+    },
+    {
+      "epoch": 0.9710639943288006,
+      "grad_norm": 0.0035890522764365973,
+      "learning_rate": 3.293949158610813e-07,
+      "loss": 0.0,
+      "step": 15068
+    },
+    {
+      "epoch": 0.9711284397757298,
+      "grad_norm": 0.0075972949912963775,
+      "learning_rate": 3.286788399570355e-07,
+      "loss": 0.0001,
+      "step": 15069
+    },
+    {
+      "epoch": 0.971192885222659,
+      "grad_norm": 0.00017337491251541505,
+      "learning_rate": 3.2796276405298964e-07,
+      "loss": 0.0,
+      "step": 15070
+    },
+    {
+      "epoch": 0.9712573306695882,
+      "grad_norm": 0.00840977936296432,
+      "learning_rate": 3.2724668814894383e-07,
+      "loss": 0.0,
+      "step": 15071
+    },
+    {
+      "epoch": 0.9713217761165174,
+      "grad_norm": 0.02215838745153162,
+      "learning_rate": 3.2653061224489803e-07,
+      "loss": 0.0002,
+      "step": 15072
+    },
+    {
+      "epoch": 0.9713862215634466,
+      "grad_norm": 8.793780417752341,
+      "learning_rate": 3.258145363408521e-07,
+      "loss": 0.023,
+      "step": 15073
+    },
+    {
+      "epoch": 0.9714506670103757,
+      "grad_norm": 0.006577569953616285,
+      "learning_rate": 3.250984604368063e-07,
+      "loss": 0.0,
+      "step": 15074
+    },
+    {
+      "epoch": 0.9715151124573049,
+      "grad_norm": 0.11361466455953624,
+      "learning_rate": 3.243823845327605e-07,
+      "loss": 0.0003,
+      "step": 15075
+    },
+    {
+      "epoch": 0.9715795579042341,
+      "grad_norm": 0.0046748844567726456,
+      "learning_rate": 3.2366630862871464e-07,
+      "loss": 0.0,
+      "step": 15076
+    },
+    {
+      "epoch": 0.9716440033511632,
+      "grad_norm": 0.015705236972287314,
+      "learning_rate": 3.2295023272466883e-07,
+      "loss": 0.0001,
+      "step": 15077
+    },
+    {
+      "epoch": 0.9717084487980924,
+      "grad_norm": 0.27528340031077064,
+      "learning_rate": 3.22234156820623e-07,
+      "loss": 0.0022,
+      "step": 15078
+    },
+    {
+      "epoch": 0.9717728942450216,
+      "grad_norm": 0.016288971732021627,
+      "learning_rate": 3.2151808091657716e-07,
+      "loss": 0.0001,
+      "step": 15079
+    },
+    {
+      "epoch": 0.9718373396919507,
+      "grad_norm": 0.47387322685274225,
+      "learning_rate": 3.2080200501253136e-07,
+      "loss": 0.0044,
+      "step": 15080
+    },
+    {
+      "epoch": 0.9719017851388799,
+      "grad_norm": 0.0011477208401962785,
+      "learning_rate": 3.2008592910848555e-07,
+      "loss": 0.0,
+      "step": 15081
+    },
+    {
+      "epoch": 0.9719662305858091,
+      "grad_norm": 0.0034684748119078452,
+      "learning_rate": 3.193698532044397e-07,
+      "loss": 0.0,
+      "step": 15082
+    },
+    {
+      "epoch": 0.9720306760327383,
+      "grad_norm": 0.004370211072131715,
+      "learning_rate": 3.186537773003939e-07,
+      "loss": 0.0,
+      "step": 15083
+    },
+    {
+      "epoch": 0.9720951214796675,
+      "grad_norm": 0.004952676162059169,
+      "learning_rate": 3.179377013963481e-07,
+      "loss": 0.0,
+      "step": 15084
+    },
+    {
+      "epoch": 0.9721595669265967,
+      "grad_norm": 0.0369411296736311,
+      "learning_rate": 3.1722162549230216e-07,
+      "loss": 0.0001,
+      "step": 15085
+    },
+    {
+      "epoch": 0.9722240123735258,
+      "grad_norm": 0.0003230311543377619,
+      "learning_rate": 3.1650554958825636e-07,
+      "loss": 0.0,
+      "step": 15086
+    },
+    {
+      "epoch": 0.972288457820455,
+      "grad_norm": 0.0008222280461801855,
+      "learning_rate": 3.1578947368421055e-07,
+      "loss": 0.0,
+      "step": 15087
+    },
+    {
+      "epoch": 0.9723529032673841,
+      "grad_norm": 0.00021895085382019427,
+      "learning_rate": 3.1507339778016474e-07,
+      "loss": 0.0,
+      "step": 15088
+    },
+    {
+      "epoch": 0.9724173487143133,
+      "grad_norm": 0.0011354382390777846,
+      "learning_rate": 3.143573218761189e-07,
+      "loss": 0.0,
+      "step": 15089
+    },
+    {
+      "epoch": 0.9724817941612425,
+      "grad_norm": 0.0003584275642686638,
+      "learning_rate": 3.136412459720731e-07,
+      "loss": 0.0,
+      "step": 15090
+    },
+    {
+      "epoch": 0.9725462396081717,
+      "grad_norm": 0.00023074791790306088,
+      "learning_rate": 3.1292517006802727e-07,
+      "loss": 0.0,
+      "step": 15091
+    },
+    {
+      "epoch": 0.9726106850551008,
+      "grad_norm": 0.02096019492573941,
+      "learning_rate": 3.122090941639814e-07,
+      "loss": 0.0,
+      "step": 15092
+    },
+    {
+      "epoch": 0.97267513050203,
+      "grad_norm": 0.0014797492472057203,
+      "learning_rate": 3.114930182599356e-07,
+      "loss": 0.0,
+      "step": 15093
+    },
+    {
+      "epoch": 0.9727395759489592,
+      "grad_norm": 0.0010520218654664925,
+      "learning_rate": 3.1077694235588974e-07,
+      "loss": 0.0,
+      "step": 15094
+    },
+    {
+      "epoch": 0.9728040213958884,
+      "grad_norm": 0.0008259018489257181,
+      "learning_rate": 3.1006086645184393e-07,
+      "loss": 0.0,
+      "step": 15095
+    },
+    {
+      "epoch": 0.9728684668428176,
+      "grad_norm": 0.000960079676407256,
+      "learning_rate": 3.0934479054779807e-07,
+      "loss": 0.0,
+      "step": 15096
+    },
+    {
+      "epoch": 0.9729329122897468,
+      "grad_norm": 0.000389977335042752,
+      "learning_rate": 3.0862871464375226e-07,
+      "loss": 0.0,
+      "step": 15097
+    },
+    {
+      "epoch": 0.972997357736676,
+      "grad_norm": 0.008095213189473769,
+      "learning_rate": 3.0791263873970646e-07,
+      "loss": 0.0,
+      "step": 15098
+    },
+    {
+      "epoch": 0.973061803183605,
+      "grad_norm": 0.0006163576107396637,
+      "learning_rate": 3.071965628356606e-07,
+      "loss": 0.0,
+      "step": 15099
+    },
+    {
+      "epoch": 0.9731262486305342,
+      "grad_norm": 0.03904227537872416,
+      "learning_rate": 3.064804869316148e-07,
+      "loss": 0.0002,
+      "step": 15100
+    },
+    {
+      "epoch": 0.9731906940774634,
+      "grad_norm": 0.09723703527236527,
+      "learning_rate": 3.0576441102756893e-07,
+      "loss": 0.0002,
+      "step": 15101
+    },
+    {
+      "epoch": 0.9732551395243926,
+      "grad_norm": 0.1627526732828032,
+      "learning_rate": 3.050483351235231e-07,
+      "loss": 0.0006,
+      "step": 15102
+    },
+    {
+      "epoch": 0.9733195849713218,
+      "grad_norm": 0.055487841104563056,
+      "learning_rate": 3.043322592194773e-07,
+      "loss": 0.0001,
+      "step": 15103
+    },
+    {
+      "epoch": 0.973384030418251,
+      "grad_norm": 0.00046731365863853907,
+      "learning_rate": 3.0361618331543146e-07,
+      "loss": 0.0,
+      "step": 15104
+    },
+    {
+      "epoch": 0.9734484758651801,
+      "grad_norm": 0.0034985155925200406,
+      "learning_rate": 3.029001074113856e-07,
+      "loss": 0.0,
+      "step": 15105
+    },
+    {
+      "epoch": 0.9735129213121093,
+      "grad_norm": 0.2584352197511904,
+      "learning_rate": 3.021840315073398e-07,
+      "loss": 0.0013,
+      "step": 15106
+    },
+    {
+      "epoch": 0.9735773667590385,
+      "grad_norm": 0.017777880831103005,
+      "learning_rate": 3.01467955603294e-07,
+      "loss": 0.0001,
+      "step": 15107
+    },
+    {
+      "epoch": 0.9736418122059677,
+      "grad_norm": 0.155115485075132,
+      "learning_rate": 3.007518796992481e-07,
+      "loss": 0.0006,
+      "step": 15108
+    },
+    {
+      "epoch": 0.9737062576528969,
+      "grad_norm": 0.003783563882959519,
+      "learning_rate": 3.000358037952023e-07,
+      "loss": 0.0,
+      "step": 15109
+    },
+    {
+      "epoch": 0.973770703099826,
+      "grad_norm": 0.01880250699719194,
+      "learning_rate": 2.9931972789115645e-07,
+      "loss": 0.0,
+      "step": 15110
+    },
+    {
+      "epoch": 0.9738351485467551,
+      "grad_norm": 0.0008284128947969417,
+      "learning_rate": 2.9860365198711065e-07,
+      "loss": 0.0,
+      "step": 15111
+    },
+    {
+      "epoch": 0.9738995939936843,
+      "grad_norm": 0.00040061634608030576,
+      "learning_rate": 2.9788757608306484e-07,
+      "loss": 0.0,
+      "step": 15112
+    },
+    {
+      "epoch": 0.9739640394406135,
+      "grad_norm": 0.0015559103676428346,
+      "learning_rate": 2.97171500179019e-07,
+      "loss": 0.0,
+      "step": 15113
+    },
+    {
+      "epoch": 0.9740284848875427,
+      "grad_norm": 4.241696346927127e-05,
+      "learning_rate": 2.9645542427497317e-07,
+      "loss": 0.0,
+      "step": 15114
+    },
+    {
+      "epoch": 0.9740929303344719,
+      "grad_norm": 0.0075211743685994975,
+      "learning_rate": 2.9573934837092736e-07,
+      "loss": 0.0,
+      "step": 15115
+    },
+    {
+      "epoch": 0.974157375781401,
+      "grad_norm": 0.0007922172603788567,
+      "learning_rate": 2.950232724668815e-07,
+      "loss": 0.0,
+      "step": 15116
+    },
+    {
+      "epoch": 0.9742218212283302,
+      "grad_norm": 8.615947529956862e-05,
+      "learning_rate": 2.943071965628357e-07,
+      "loss": 0.0,
+      "step": 15117
+    },
+    {
+      "epoch": 0.9742862666752594,
+      "grad_norm": 0.0011283102264602943,
+      "learning_rate": 2.9359112065878984e-07,
+      "loss": 0.0,
+      "step": 15118
+    },
+    {
+      "epoch": 0.9743507121221886,
+      "grad_norm": 0.0003551449614521658,
+      "learning_rate": 2.9287504475474403e-07,
+      "loss": 0.0,
+      "step": 15119
+    },
+    {
+      "epoch": 0.9744151575691178,
+      "grad_norm": 0.011122411754986475,
+      "learning_rate": 2.921589688506982e-07,
+      "loss": 0.0,
+      "step": 15120
+    },
+    {
+      "epoch": 0.974479603016047,
+      "grad_norm": 0.002033701591427555,
+      "learning_rate": 2.9144289294665236e-07,
+      "loss": 0.0,
+      "step": 15121
+    },
+    {
+      "epoch": 0.974544048462976,
+      "grad_norm": 0.0004772722821963182,
+      "learning_rate": 2.9072681704260656e-07,
+      "loss": 0.0,
+      "step": 15122
+    },
+    {
+      "epoch": 0.9746084939099052,
+      "grad_norm": 0.014922254632288611,
+      "learning_rate": 2.900107411385607e-07,
+      "loss": 0.0,
+      "step": 15123
+    },
+    {
+      "epoch": 0.9746729393568344,
+      "grad_norm": 0.14024032856409033,
+      "learning_rate": 2.892946652345149e-07,
+      "loss": 0.0005,
+      "step": 15124
+    },
+    {
+      "epoch": 0.9747373848037636,
+      "grad_norm": 0.005984868862872008,
+      "learning_rate": 2.885785893304691e-07,
+      "loss": 0.0,
+      "step": 15125
+    },
+    {
+      "epoch": 0.9748018302506928,
+      "grad_norm": 0.004334931194733932,
+      "learning_rate": 2.878625134264232e-07,
+      "loss": 0.0,
+      "step": 15126
+    },
+    {
+      "epoch": 0.974866275697622,
+      "grad_norm": 0.002894429516225133,
+      "learning_rate": 2.871464375223774e-07,
+      "loss": 0.0,
+      "step": 15127
+    },
+    {
+      "epoch": 0.9749307211445511,
+      "grad_norm": 0.0003413323669872801,
+      "learning_rate": 2.8643036161833155e-07,
+      "loss": 0.0,
+      "step": 15128
+    },
+    {
+      "epoch": 0.9749951665914803,
+      "grad_norm": 0.0010935772408200211,
+      "learning_rate": 2.8571428571428575e-07,
+      "loss": 0.0,
+      "step": 15129
+    },
+    {
+      "epoch": 0.9750596120384095,
+      "grad_norm": 0.11490509704268698,
+      "learning_rate": 2.8499820981023994e-07,
+      "loss": 0.0001,
+      "step": 15130
+    },
+    {
+      "epoch": 0.9751240574853387,
+      "grad_norm": 0.24854126418726763,
+      "learning_rate": 2.842821339061941e-07,
+      "loss": 0.0005,
+      "step": 15131
+    },
+    {
+      "epoch": 0.9751885029322679,
+      "grad_norm": 0.008830847027775436,
+      "learning_rate": 2.835660580021482e-07,
+      "loss": 0.0,
+      "step": 15132
+    },
+    {
+      "epoch": 0.975252948379197,
+      "grad_norm": 0.0008734777265217501,
+      "learning_rate": 2.828499820981024e-07,
+      "loss": 0.0,
+      "step": 15133
+    },
+    {
+      "epoch": 0.9753173938261261,
+      "grad_norm": 0.36864601786661755,
+      "learning_rate": 2.821339061940566e-07,
+      "loss": 0.0007,
+      "step": 15134
+    },
+    {
+      "epoch": 0.9753818392730553,
+      "grad_norm": 0.0022641853155302765,
+      "learning_rate": 2.8141783029001074e-07,
+      "loss": 0.0,
+      "step": 15135
+    },
+    {
+      "epoch": 0.9754462847199845,
+      "grad_norm": 0.06988027629866522,
+      "learning_rate": 2.8070175438596494e-07,
+      "loss": 0.0002,
+      "step": 15136
+    },
+    {
+      "epoch": 0.9755107301669137,
+      "grad_norm": 0.0006696937469466642,
+      "learning_rate": 2.799856784819191e-07,
+      "loss": 0.0,
+      "step": 15137
+    },
+    {
+      "epoch": 0.9755751756138429,
+      "grad_norm": 0.0013796730065024532,
+      "learning_rate": 2.7926960257787327e-07,
+      "loss": 0.0,
+      "step": 15138
+    },
+    {
+      "epoch": 0.9756396210607721,
+      "grad_norm": 0.0004004937699149055,
+      "learning_rate": 2.7855352667382746e-07,
+      "loss": 0.0,
+      "step": 15139
+    },
+    {
+      "epoch": 0.9757040665077013,
+      "grad_norm": 0.005334612036549248,
+      "learning_rate": 2.778374507697816e-07,
+      "loss": 0.0,
+      "step": 15140
+    },
+    {
+      "epoch": 0.9757685119546304,
+      "grad_norm": 0.002374710997573885,
+      "learning_rate": 2.771213748657358e-07,
+      "loss": 0.0,
+      "step": 15141
+    },
+    {
+      "epoch": 0.9758329574015596,
+      "grad_norm": 0.0010283480037241186,
+      "learning_rate": 2.7640529896168994e-07,
+      "loss": 0.0,
+      "step": 15142
+    },
+    {
+      "epoch": 0.9758974028484888,
+      "grad_norm": 0.0018373973709965015,
+      "learning_rate": 2.7568922305764413e-07,
+      "loss": 0.0,
+      "step": 15143
+    },
+    {
+      "epoch": 0.9759618482954179,
+      "grad_norm": 0.00855921588567582,
+      "learning_rate": 2.749731471535983e-07,
+      "loss": 0.0,
+      "step": 15144
+    },
+    {
+      "epoch": 0.9760262937423471,
+      "grad_norm": 0.0020050944102561717,
+      "learning_rate": 2.7425707124955246e-07,
+      "loss": 0.0,
+      "step": 15145
+    },
+    {
+      "epoch": 0.9760907391892762,
+      "grad_norm": 0.022109847137461454,
+      "learning_rate": 2.7354099534550665e-07,
+      "loss": 0.0,
+      "step": 15146
+    },
+    {
+      "epoch": 0.9761551846362054,
+      "grad_norm": 0.0011914119683068853,
+      "learning_rate": 2.728249194414608e-07,
+      "loss": 0.0,
+      "step": 15147
+    },
+    {
+      "epoch": 0.9762196300831346,
+      "grad_norm": 0.0011568151027143419,
+      "learning_rate": 2.72108843537415e-07,
+      "loss": 0.0,
+      "step": 15148
+    },
+    {
+      "epoch": 0.9762840755300638,
+      "grad_norm": 0.0003769420355886379,
+      "learning_rate": 2.713927676333692e-07,
+      "loss": 0.0,
+      "step": 15149
+    },
+    {
+      "epoch": 0.976348520976993,
+      "grad_norm": 0.004740660637754758,
+      "learning_rate": 2.706766917293233e-07,
+      "loss": 0.0,
+      "step": 15150
+    },
+    {
+      "epoch": 0.9764129664239222,
+      "grad_norm": 0.01738213764846972,
+      "learning_rate": 2.699606158252775e-07,
+      "loss": 0.0002,
+      "step": 15151
+    },
+    {
+      "epoch": 0.9764774118708514,
+      "grad_norm": 0.0019549345364176865,
+      "learning_rate": 2.692445399212317e-07,
+      "loss": 0.0,
+      "step": 15152
+    },
+    {
+      "epoch": 0.9765418573177805,
+      "grad_norm": 0.0007493480390459141,
+      "learning_rate": 2.6852846401718585e-07,
+      "loss": 0.0,
+      "step": 15153
+    },
+    {
+      "epoch": 0.9766063027647097,
+      "grad_norm": 0.0026944598274480157,
+      "learning_rate": 2.6781238811314004e-07,
+      "loss": 0.0,
+      "step": 15154
+    },
+    {
+      "epoch": 0.9766707482116388,
+      "grad_norm": 0.011713198623444991,
+      "learning_rate": 2.670963122090942e-07,
+      "loss": 0.0,
+      "step": 15155
+    },
+    {
+      "epoch": 0.976735193658568,
+      "grad_norm": 0.022796453401214445,
+      "learning_rate": 2.6638023630504837e-07,
+      "loss": 0.0,
+      "step": 15156
+    },
+    {
+      "epoch": 0.9767996391054972,
+      "grad_norm": 0.0003065624087929334,
+      "learning_rate": 2.6566416040100256e-07,
+      "loss": 0.0,
+      "step": 15157
+    },
+    {
+      "epoch": 0.9768640845524263,
+      "grad_norm": 0.00849296321003075,
+      "learning_rate": 2.649480844969567e-07,
+      "loss": 0.0,
+      "step": 15158
+    },
+    {
+      "epoch": 0.9769285299993555,
+      "grad_norm": 0.0007528658127822991,
+      "learning_rate": 2.6423200859291084e-07,
+      "loss": 0.0,
+      "step": 15159
+    },
+    {
+      "epoch": 0.9769929754462847,
+      "grad_norm": 0.013537214203229,
+      "learning_rate": 2.6351593268886504e-07,
+      "loss": 0.0,
+      "step": 15160
+    },
+    {
+      "epoch": 0.9770574208932139,
+      "grad_norm": 0.008491264624011879,
+      "learning_rate": 2.6279985678481923e-07,
+      "loss": 0.0,
+      "step": 15161
+    },
+    {
+      "epoch": 0.9771218663401431,
+      "grad_norm": 0.040971711185676433,
+      "learning_rate": 2.6208378088077337e-07,
+      "loss": 0.0001,
+      "step": 15162
+    },
+    {
+      "epoch": 0.9771863117870723,
+      "grad_norm": 0.0021006739403628245,
+      "learning_rate": 2.6136770497672756e-07,
+      "loss": 0.0,
+      "step": 15163
+    },
+    {
+      "epoch": 0.9772507572340015,
+      "grad_norm": 0.0033201136895484383,
+      "learning_rate": 2.606516290726817e-07,
+      "loss": 0.0,
+      "step": 15164
+    },
+    {
+      "epoch": 0.9773152026809306,
+      "grad_norm": 0.0009516754660645897,
+      "learning_rate": 2.599355531686359e-07,
+      "loss": 0.0,
+      "step": 15165
+    },
+    {
+      "epoch": 0.9773796481278597,
+      "grad_norm": 0.0014256401567068185,
+      "learning_rate": 2.592194772645901e-07,
+      "loss": 0.0,
+      "step": 15166
+    },
+    {
+      "epoch": 0.9774440935747889,
+      "grad_norm": 0.007363190876471923,
+      "learning_rate": 2.5850340136054423e-07,
+      "loss": 0.0001,
+      "step": 15167
+    },
+    {
+      "epoch": 0.9775085390217181,
+      "grad_norm": 0.0001785714074357449,
+      "learning_rate": 2.577873254564984e-07,
+      "loss": 0.0,
+      "step": 15168
+    },
+    {
+      "epoch": 0.9775729844686473,
+      "grad_norm": 0.0010834485266816886,
+      "learning_rate": 2.5707124955245256e-07,
+      "loss": 0.0,
+      "step": 15169
+    },
+    {
+      "epoch": 0.9776374299155765,
+      "grad_norm": 0.14266291148414587,
+      "learning_rate": 2.5635517364840675e-07,
+      "loss": 0.0011,
+      "step": 15170
+    },
+    {
+      "epoch": 0.9777018753625056,
+      "grad_norm": 0.10230791742266519,
+      "learning_rate": 2.5563909774436095e-07,
+      "loss": 0.0003,
+      "step": 15171
+    },
+    {
+      "epoch": 0.9777663208094348,
+      "grad_norm": 0.18012416689572205,
+      "learning_rate": 2.549230218403151e-07,
+      "loss": 0.0011,
+      "step": 15172
+    },
+    {
+      "epoch": 0.977830766256364,
+      "grad_norm": 0.018825811985009936,
+      "learning_rate": 2.542069459362693e-07,
+      "loss": 0.0,
+      "step": 15173
+    },
+    {
+      "epoch": 0.9778952117032932,
+      "grad_norm": 0.0003657376167657997,
+      "learning_rate": 2.534908700322234e-07,
+      "loss": 0.0,
+      "step": 15174
+    },
+    {
+      "epoch": 0.9779596571502224,
+      "grad_norm": 0.0016266523017274357,
+      "learning_rate": 2.527747941281776e-07,
+      "loss": 0.0,
+      "step": 15175
+    },
+    {
+      "epoch": 0.9780241025971516,
+      "grad_norm": 0.031643844370067425,
+      "learning_rate": 2.520587182241318e-07,
+      "loss": 0.0002,
+      "step": 15176
+    },
+    {
+      "epoch": 0.9780885480440807,
+      "grad_norm": 0.0001272113728231804,
+      "learning_rate": 2.5134264232008594e-07,
+      "loss": 0.0,
+      "step": 15177
+    },
+    {
+      "epoch": 0.9781529934910098,
+      "grad_norm": 0.051368001198941485,
+      "learning_rate": 2.506265664160401e-07,
+      "loss": 0.0001,
+      "step": 15178
+    },
+    {
+      "epoch": 0.978217438937939,
+      "grad_norm": 0.11006070518329952,
+      "learning_rate": 2.499104905119943e-07,
+      "loss": 0.0017,
+      "step": 15179
+    },
+    {
+      "epoch": 0.9782818843848682,
+      "grad_norm": 0.007755887701504084,
+      "learning_rate": 2.4919441460794847e-07,
+      "loss": 0.0,
+      "step": 15180
+    },
+    {
+      "epoch": 0.9783463298317974,
+      "grad_norm": 0.0017534684004093396,
+      "learning_rate": 2.4847833870390266e-07,
+      "loss": 0.0,
+      "step": 15181
+    },
+    {
+      "epoch": 0.9784107752787266,
+      "grad_norm": 0.07081709706644274,
+      "learning_rate": 2.477622627998568e-07,
+      "loss": 0.0001,
+      "step": 15182
+    },
+    {
+      "epoch": 0.9784752207256557,
+      "grad_norm": 0.0018758709562702654,
+      "learning_rate": 2.4704618689581094e-07,
+      "loss": 0.0,
+      "step": 15183
+    },
+    {
+      "epoch": 0.9785396661725849,
+      "grad_norm": 0.001043274861493202,
+      "learning_rate": 2.4633011099176513e-07,
+      "loss": 0.0,
+      "step": 15184
+    },
+    {
+      "epoch": 0.9786041116195141,
+      "grad_norm": 0.0011120056715451592,
+      "learning_rate": 2.4561403508771933e-07,
+      "loss": 0.0,
+      "step": 15185
+    },
+    {
+      "epoch": 0.9786685570664433,
+      "grad_norm": 0.07570947081657063,
+      "learning_rate": 2.4489795918367347e-07,
+      "loss": 0.0002,
+      "step": 15186
+    },
+    {
+      "epoch": 0.9787330025133725,
+      "grad_norm": 0.0002145057054067127,
+      "learning_rate": 2.4418188327962766e-07,
+      "loss": 0.0,
+      "step": 15187
+    },
+    {
+      "epoch": 0.9787974479603017,
+      "grad_norm": 0.003613192592037386,
+      "learning_rate": 2.4346580737558185e-07,
+      "loss": 0.0,
+      "step": 15188
+    },
+    {
+      "epoch": 0.9788618934072307,
+      "grad_norm": 0.01221347519544524,
+      "learning_rate": 2.42749731471536e-07,
+      "loss": 0.0,
+      "step": 15189
+    },
+    {
+      "epoch": 0.9789263388541599,
+      "grad_norm": 0.0004138912713609249,
+      "learning_rate": 2.420336555674902e-07,
+      "loss": 0.0,
+      "step": 15190
+    },
+    {
+      "epoch": 0.9789907843010891,
+      "grad_norm": 0.013110504602359498,
+      "learning_rate": 2.413175796634443e-07,
+      "loss": 0.0,
+      "step": 15191
+    },
+    {
+      "epoch": 0.9790552297480183,
+      "grad_norm": 0.07918108957232092,
+      "learning_rate": 2.406015037593985e-07,
+      "loss": 0.0001,
+      "step": 15192
+    },
+    {
+      "epoch": 0.9791196751949475,
+      "grad_norm": 0.0009919039297520671,
+      "learning_rate": 2.398854278553527e-07,
+      "loss": 0.0,
+      "step": 15193
+    },
+    {
+      "epoch": 0.9791841206418767,
+      "grad_norm": 0.0015389498764722693,
+      "learning_rate": 2.3916935195130685e-07,
+      "loss": 0.0,
+      "step": 15194
+    },
+    {
+      "epoch": 0.9792485660888058,
+      "grad_norm": 0.0007932537619711912,
+      "learning_rate": 2.3845327604726104e-07,
+      "loss": 0.0,
+      "step": 15195
+    },
+    {
+      "epoch": 0.979313011535735,
+      "grad_norm": 0.3223723494933655,
+      "learning_rate": 2.377372001432152e-07,
+      "loss": 0.0007,
+      "step": 15196
+    },
+    {
+      "epoch": 0.9793774569826642,
+      "grad_norm": 0.0037370869206710114,
+      "learning_rate": 2.3702112423916935e-07,
+      "loss": 0.0,
+      "step": 15197
+    },
+    {
+      "epoch": 0.9794419024295934,
+      "grad_norm": 0.01667958658079532,
+      "learning_rate": 2.3630504833512354e-07,
+      "loss": 0.0,
+      "step": 15198
+    },
+    {
+      "epoch": 0.9795063478765226,
+      "grad_norm": 0.045919618893321044,
+      "learning_rate": 2.355889724310777e-07,
+      "loss": 0.0001,
+      "step": 15199
+    },
+    {
+      "epoch": 0.9795707933234516,
+      "grad_norm": 0.00020583061499714295,
+      "learning_rate": 2.348728965270319e-07,
+      "loss": 0.0,
+      "step": 15200
+    },
+    {
+      "epoch": 0.9796352387703808,
+      "grad_norm": 0.0035573688295585603,
+      "learning_rate": 2.3415682062298607e-07,
+      "loss": 0.0,
+      "step": 15201
+    },
+    {
+      "epoch": 0.97969968421731,
+      "grad_norm": 0.0003429043040178539,
+      "learning_rate": 2.334407447189402e-07,
+      "loss": 0.0,
+      "step": 15202
+    },
+    {
+      "epoch": 0.9797641296642392,
+      "grad_norm": 0.00013085403944323605,
+      "learning_rate": 2.327246688148944e-07,
+      "loss": 0.0,
+      "step": 15203
+    },
+    {
+      "epoch": 0.9798285751111684,
+      "grad_norm": 0.005921728071037252,
+      "learning_rate": 2.3200859291084857e-07,
+      "loss": 0.0,
+      "step": 15204
+    },
+    {
+      "epoch": 0.9798930205580976,
+      "grad_norm": 0.000992523744680652,
+      "learning_rate": 2.3129251700680273e-07,
+      "loss": 0.0,
+      "step": 15205
+    },
+    {
+      "epoch": 0.9799574660050268,
+      "grad_norm": 0.0006133625288807913,
+      "learning_rate": 2.3057644110275693e-07,
+      "loss": 0.0,
+      "step": 15206
+    },
+    {
+      "epoch": 0.9800219114519559,
+      "grad_norm": 0.0008549871418847644,
+      "learning_rate": 2.2986036519871107e-07,
+      "loss": 0.0,
+      "step": 15207
+    },
+    {
+      "epoch": 0.9800863568988851,
+      "grad_norm": 0.00020320387837958342,
+      "learning_rate": 2.2914428929466526e-07,
+      "loss": 0.0,
+      "step": 15208
+    },
+    {
+      "epoch": 0.9801508023458143,
+      "grad_norm": 0.22708419225833618,
+      "learning_rate": 2.2842821339061943e-07,
+      "loss": 0.0029,
+      "step": 15209
+    },
+    {
+      "epoch": 0.9802152477927435,
+      "grad_norm": 0.0004024974515455393,
+      "learning_rate": 2.277121374865736e-07,
+      "loss": 0.0,
+      "step": 15210
+    },
+    {
+      "epoch": 0.9802796932396726,
+      "grad_norm": 0.04274219695899788,
+      "learning_rate": 2.2699606158252778e-07,
+      "loss": 0.0001,
+      "step": 15211
+    },
+    {
+      "epoch": 0.9803441386866018,
+      "grad_norm": 0.034374677117413543,
+      "learning_rate": 2.2627998567848195e-07,
+      "loss": 0.0001,
+      "step": 15212
+    },
+    {
+      "epoch": 0.9804085841335309,
+      "grad_norm": 0.000535652454101118,
+      "learning_rate": 2.255639097744361e-07,
+      "loss": 0.0,
+      "step": 15213
+    },
+    {
+      "epoch": 0.9804730295804601,
+      "grad_norm": 0.14314235455022237,
+      "learning_rate": 2.2484783387039028e-07,
+      "loss": 0.0002,
+      "step": 15214
+    },
+    {
+      "epoch": 0.9805374750273893,
+      "grad_norm": 0.0018527310044761708,
+      "learning_rate": 2.2413175796634445e-07,
+      "loss": 0.0,
+      "step": 15215
+    },
+    {
+      "epoch": 0.9806019204743185,
+      "grad_norm": 0.013785798169173939,
+      "learning_rate": 2.2341568206229862e-07,
+      "loss": 0.0,
+      "step": 15216
+    },
+    {
+      "epoch": 0.9806663659212477,
+      "grad_norm": 0.008295814609581176,
+      "learning_rate": 2.226996061582528e-07,
+      "loss": 0.0,
+      "step": 15217
+    },
+    {
+      "epoch": 0.9807308113681769,
+      "grad_norm": 0.00018404890305874304,
+      "learning_rate": 2.2198353025420695e-07,
+      "loss": 0.0,
+      "step": 15218
+    },
+    {
+      "epoch": 0.980795256815106,
+      "grad_norm": 0.00012482961636409894,
+      "learning_rate": 2.2126745435016114e-07,
+      "loss": 0.0,
+      "step": 15219
+    },
+    {
+      "epoch": 0.9808597022620352,
+      "grad_norm": 0.0049240704985918085,
+      "learning_rate": 2.205513784461153e-07,
+      "loss": 0.0,
+      "step": 15220
+    },
+    {
+      "epoch": 0.9809241477089644,
+      "grad_norm": 0.0031281767796909107,
+      "learning_rate": 2.1983530254206947e-07,
+      "loss": 0.0,
+      "step": 15221
+    },
+    {
+      "epoch": 0.9809885931558935,
+      "grad_norm": 0.0008334030137153192,
+      "learning_rate": 2.1911922663802367e-07,
+      "loss": 0.0,
+      "step": 15222
+    },
+    {
+      "epoch": 0.9810530386028227,
+      "grad_norm": 0.00019850462293857934,
+      "learning_rate": 2.184031507339778e-07,
+      "loss": 0.0,
+      "step": 15223
+    },
+    {
+      "epoch": 0.9811174840497519,
+      "grad_norm": 0.00036782715442014083,
+      "learning_rate": 2.1768707482993197e-07,
+      "loss": 0.0,
+      "step": 15224
+    },
+    {
+      "epoch": 0.981181929496681,
+      "grad_norm": 0.0003710061416936885,
+      "learning_rate": 2.1697099892588617e-07,
+      "loss": 0.0,
+      "step": 15225
+    },
+    {
+      "epoch": 0.9812463749436102,
+      "grad_norm": 0.012523587796355857,
+      "learning_rate": 2.1625492302184033e-07,
+      "loss": 0.0,
+      "step": 15226
+    },
+    {
+      "epoch": 0.9813108203905394,
+      "grad_norm": 0.002729799152419526,
+      "learning_rate": 2.1553884711779453e-07,
+      "loss": 0.0,
+      "step": 15227
+    },
+    {
+      "epoch": 0.9813752658374686,
+      "grad_norm": 0.0001641298882696401,
+      "learning_rate": 2.1482277121374867e-07,
+      "loss": 0.0,
+      "step": 15228
+    },
+    {
+      "epoch": 0.9814397112843978,
+      "grad_norm": 1.309516649618846,
+      "learning_rate": 2.1410669530970283e-07,
+      "loss": 0.0006,
+      "step": 15229
+    },
+    {
+      "epoch": 0.981504156731327,
+      "grad_norm": 0.014743855267792875,
+      "learning_rate": 2.1339061940565702e-07,
+      "loss": 0.0,
+      "step": 15230
+    },
+    {
+      "epoch": 0.9815686021782561,
+      "grad_norm": 0.0023293280917756737,
+      "learning_rate": 2.126745435016112e-07,
+      "loss": 0.0,
+      "step": 15231
+    },
+    {
+      "epoch": 0.9816330476251853,
+      "grad_norm": 9.553283079941483e-05,
+      "learning_rate": 2.1195846759756536e-07,
+      "loss": 0.0,
+      "step": 15232
+    },
+    {
+      "epoch": 0.9816974930721144,
+      "grad_norm": 4.284839848487141e-05,
+      "learning_rate": 2.1124239169351955e-07,
+      "loss": 0.0,
+      "step": 15233
+    },
+    {
+      "epoch": 0.9817619385190436,
+      "grad_norm": 0.0060593230029311156,
+      "learning_rate": 2.105263157894737e-07,
+      "loss": 0.0001,
+      "step": 15234
+    },
+    {
+      "epoch": 0.9818263839659728,
+      "grad_norm": 0.0011964518485110813,
+      "learning_rate": 2.0981023988542788e-07,
+      "loss": 0.0,
+      "step": 15235
+    },
+    {
+      "epoch": 0.981890829412902,
+      "grad_norm": 0.32554900492517685,
+      "learning_rate": 2.0909416398138205e-07,
+      "loss": 0.0022,
+      "step": 15236
+    },
+    {
+      "epoch": 0.9819552748598311,
+      "grad_norm": 0.0006848094652021027,
+      "learning_rate": 2.0837808807733622e-07,
+      "loss": 0.0,
+      "step": 15237
+    },
+    {
+      "epoch": 0.9820197203067603,
+      "grad_norm": 0.001575442477279625,
+      "learning_rate": 2.076620121732904e-07,
+      "loss": 0.0,
+      "step": 15238
+    },
+    {
+      "epoch": 0.9820841657536895,
+      "grad_norm": 0.07122442175884686,
+      "learning_rate": 2.0694593626924455e-07,
+      "loss": 0.0003,
+      "step": 15239
+    },
+    {
+      "epoch": 0.9821486112006187,
+      "grad_norm": 0.047471250288807096,
+      "learning_rate": 2.0622986036519871e-07,
+      "loss": 0.0004,
+      "step": 15240
+    },
+    {
+      "epoch": 0.9822130566475479,
+      "grad_norm": 0.0006905600191781044,
+      "learning_rate": 2.055137844611529e-07,
+      "loss": 0.0,
+      "step": 15241
+    },
+    {
+      "epoch": 0.9822775020944771,
+      "grad_norm": 3.866391618307458,
+      "learning_rate": 2.0479770855710707e-07,
+      "loss": 0.0589,
+      "step": 15242
+    },
+    {
+      "epoch": 0.9823419475414062,
+      "grad_norm": 0.0015586490202091803,
+      "learning_rate": 2.0408163265306121e-07,
+      "loss": 0.0,
+      "step": 15243
+    },
+    {
+      "epoch": 0.9824063929883353,
+      "grad_norm": 0.0002299173824988018,
+      "learning_rate": 2.033655567490154e-07,
+      "loss": 0.0,
+      "step": 15244
+    },
+    {
+      "epoch": 0.9824708384352645,
+      "grad_norm": 0.05903546807777156,
+      "learning_rate": 2.0264948084496957e-07,
+      "loss": 0.0003,
+      "step": 15245
+    },
+    {
+      "epoch": 0.9825352838821937,
+      "grad_norm": 0.05004071891032356,
+      "learning_rate": 2.0193340494092377e-07,
+      "loss": 0.0001,
+      "step": 15246
+    },
+    {
+      "epoch": 0.9825997293291229,
+      "grad_norm": 0.00033025603537803906,
+      "learning_rate": 2.0121732903687793e-07,
+      "loss": 0.0,
+      "step": 15247
+    },
+    {
+      "epoch": 0.9826641747760521,
+      "grad_norm": 0.0006552670394381289,
+      "learning_rate": 2.005012531328321e-07,
+      "loss": 0.0,
+      "step": 15248
+    },
+    {
+      "epoch": 0.9827286202229812,
+      "grad_norm": 0.0016977273912929997,
+      "learning_rate": 1.997851772287863e-07,
+      "loss": 0.0,
+      "step": 15249
+    },
+    {
+      "epoch": 0.9827930656699104,
+      "grad_norm": 0.18565537141618688,
+      "learning_rate": 1.9906910132474043e-07,
+      "loss": 0.0005,
+      "step": 15250
+    },
+    {
+      "epoch": 0.9828575111168396,
+      "grad_norm": 0.07746788176045784,
+      "learning_rate": 1.983530254206946e-07,
+      "loss": 0.0017,
+      "step": 15251
+    },
+    {
+      "epoch": 0.9829219565637688,
+      "grad_norm": 0.001825236225272376,
+      "learning_rate": 1.976369495166488e-07,
+      "loss": 0.0,
+      "step": 15252
+    },
+    {
+      "epoch": 0.982986402010698,
+      "grad_norm": 0.02276023707093468,
+      "learning_rate": 1.9692087361260296e-07,
+      "loss": 0.0002,
+      "step": 15253
+    },
+    {
+      "epoch": 0.9830508474576272,
+      "grad_norm": 0.0002736884330044127,
+      "learning_rate": 1.9620479770855715e-07,
+      "loss": 0.0,
+      "step": 15254
+    },
+    {
+      "epoch": 0.9831152929045563,
+      "grad_norm": 0.003627516108369626,
+      "learning_rate": 1.954887218045113e-07,
+      "loss": 0.0,
+      "step": 15255
+    },
+    {
+      "epoch": 0.9831797383514854,
+      "grad_norm": 0.0022958784688874136,
+      "learning_rate": 1.9477264590046546e-07,
+      "loss": 0.0,
+      "step": 15256
+    },
+    {
+      "epoch": 0.9832441837984146,
+      "grad_norm": 0.00014915189100592606,
+      "learning_rate": 1.9405656999641965e-07,
+      "loss": 0.0,
+      "step": 15257
+    },
+    {
+      "epoch": 0.9833086292453438,
+      "grad_norm": 6.148843509926524e-05,
+      "learning_rate": 1.9334049409237381e-07,
+      "loss": 0.0,
+      "step": 15258
+    },
+    {
+      "epoch": 0.983373074692273,
+      "grad_norm": 0.00010684416463237988,
+      "learning_rate": 1.9262441818832795e-07,
+      "loss": 0.0,
+      "step": 15259
+    },
+    {
+      "epoch": 0.9834375201392022,
+      "grad_norm": 0.8740741054404098,
+      "learning_rate": 1.9190834228428215e-07,
+      "loss": 0.0013,
+      "step": 15260
+    },
+    {
+      "epoch": 0.9835019655861313,
+      "grad_norm": 0.001053662257371922,
+      "learning_rate": 1.9119226638023631e-07,
+      "loss": 0.0,
+      "step": 15261
+    },
+    {
+      "epoch": 0.9835664110330605,
+      "grad_norm": 0.2909590903034731,
+      "learning_rate": 1.904761904761905e-07,
+      "loss": 0.0005,
+      "step": 15262
+    },
+    {
+      "epoch": 0.9836308564799897,
+      "grad_norm": 0.001149833658102217,
+      "learning_rate": 1.8976011457214467e-07,
+      "loss": 0.0,
+      "step": 15263
+    },
+    {
+      "epoch": 0.9836953019269189,
+      "grad_norm": 0.0009560136118597889,
+      "learning_rate": 1.890440386680988e-07,
+      "loss": 0.0,
+      "step": 15264
+    },
+    {
+      "epoch": 0.9837597473738481,
+      "grad_norm": 0.04803480853553378,
+      "learning_rate": 1.88327962764053e-07,
+      "loss": 0.0002,
+      "step": 15265
+    },
+    {
+      "epoch": 0.9838241928207773,
+      "grad_norm": 0.0016538616173160513,
+      "learning_rate": 1.8761188686000717e-07,
+      "loss": 0.0,
+      "step": 15266
+    },
+    {
+      "epoch": 0.9838886382677063,
+      "grad_norm": 0.01374418114037808,
+      "learning_rate": 1.8689581095596134e-07,
+      "loss": 0.0001,
+      "step": 15267
+    },
+    {
+      "epoch": 0.9839530837146355,
+      "grad_norm": 0.19178463067401658,
+      "learning_rate": 1.8617973505191553e-07,
+      "loss": 0.0003,
+      "step": 15268
+    },
+    {
+      "epoch": 0.9840175291615647,
+      "grad_norm": 0.0006956476860202471,
+      "learning_rate": 1.854636591478697e-07,
+      "loss": 0.0,
+      "step": 15269
+    },
+    {
+      "epoch": 0.9840819746084939,
+      "grad_norm": 0.00035165631147443776,
+      "learning_rate": 1.8474758324382384e-07,
+      "loss": 0.0,
+      "step": 15270
+    },
+    {
+      "epoch": 0.9841464200554231,
+      "grad_norm": 0.002953816493226346,
+      "learning_rate": 1.8403150733977803e-07,
+      "loss": 0.0,
+      "step": 15271
+    },
+    {
+      "epoch": 0.9842108655023523,
+      "grad_norm": 0.0003641283605604771,
+      "learning_rate": 1.833154314357322e-07,
+      "loss": 0.0,
+      "step": 15272
+    },
+    {
+      "epoch": 0.9842753109492814,
+      "grad_norm": 0.0033303717674017914,
+      "learning_rate": 1.825993555316864e-07,
+      "loss": 0.0,
+      "step": 15273
+    },
+    {
+      "epoch": 0.9843397563962106,
+      "grad_norm": 0.0006977955450955413,
+      "learning_rate": 1.8188327962764056e-07,
+      "loss": 0.0,
+      "step": 15274
+    },
+    {
+      "epoch": 0.9844042018431398,
+      "grad_norm": 0.006914318477005871,
+      "learning_rate": 1.811672037235947e-07,
+      "loss": 0.0,
+      "step": 15275
+    },
+    {
+      "epoch": 0.984468647290069,
+      "grad_norm": 0.0029855712369581877,
+      "learning_rate": 1.804511278195489e-07,
+      "loss": 0.0,
+      "step": 15276
+    },
+    {
+      "epoch": 0.9845330927369982,
+      "grad_norm": 0.0007799178281663518,
+      "learning_rate": 1.7973505191550305e-07,
+      "loss": 0.0,
+      "step": 15277
+    },
+    {
+      "epoch": 0.9845975381839273,
+      "grad_norm": 0.010909400934901978,
+      "learning_rate": 1.7901897601145722e-07,
+      "loss": 0.0,
+      "step": 15278
+    },
+    {
+      "epoch": 0.9846619836308564,
+      "grad_norm": 0.0044053793582521416,
+      "learning_rate": 1.7830290010741141e-07,
+      "loss": 0.0,
+      "step": 15279
+    },
+    {
+      "epoch": 0.9847264290777856,
+      "grad_norm": 0.9564209529679507,
+      "learning_rate": 1.7758682420336555e-07,
+      "loss": 0.0091,
+      "step": 15280
+    },
+    {
+      "epoch": 0.9847908745247148,
+      "grad_norm": 0.40579912033949256,
+      "learning_rate": 1.7687074829931975e-07,
+      "loss": 0.0035,
+      "step": 15281
+    },
+    {
+      "epoch": 0.984855319971644,
+      "grad_norm": 0.09994860926754734,
+      "learning_rate": 1.761546723952739e-07,
+      "loss": 0.0001,
+      "step": 15282
+    },
+    {
+      "epoch": 0.9849197654185732,
+      "grad_norm": 0.04675263088519276,
+      "learning_rate": 1.7543859649122808e-07,
+      "loss": 0.0003,
+      "step": 15283
+    },
+    {
+      "epoch": 0.9849842108655024,
+      "grad_norm": 0.00011445797154276652,
+      "learning_rate": 1.7472252058718227e-07,
+      "loss": 0.0,
+      "step": 15284
+    },
+    {
+      "epoch": 0.9850486563124315,
+      "grad_norm": 0.012145525836946542,
+      "learning_rate": 1.7400644468313644e-07,
+      "loss": 0.0,
+      "step": 15285
+    },
+    {
+      "epoch": 0.9851131017593607,
+      "grad_norm": 0.0003475919052068295,
+      "learning_rate": 1.7329036877909058e-07,
+      "loss": 0.0,
+      "step": 15286
+    },
+    {
+      "epoch": 0.9851775472062899,
+      "grad_norm": 0.02558688960961057,
+      "learning_rate": 1.7257429287504477e-07,
+      "loss": 0.0,
+      "step": 15287
+    },
+    {
+      "epoch": 0.9852419926532191,
+      "grad_norm": 0.00010669744202209233,
+      "learning_rate": 1.7185821697099894e-07,
+      "loss": 0.0,
+      "step": 15288
+    },
+    {
+      "epoch": 0.9853064381001482,
+      "grad_norm": 0.0006361258570100312,
+      "learning_rate": 1.7114214106695313e-07,
+      "loss": 0.0,
+      "step": 15289
+    },
+    {
+      "epoch": 0.9853708835470774,
+      "grad_norm": 0.046747478735065066,
+      "learning_rate": 1.704260651629073e-07,
+      "loss": 0.0001,
+      "step": 15290
+    },
+    {
+      "epoch": 0.9854353289940065,
+      "grad_norm": 0.00012749033638977258,
+      "learning_rate": 1.6970998925886144e-07,
+      "loss": 0.0,
+      "step": 15291
+    },
+    {
+      "epoch": 0.9854997744409357,
+      "grad_norm": 0.0007303949571343691,
+      "learning_rate": 1.6899391335481563e-07,
+      "loss": 0.0,
+      "step": 15292
+    },
+    {
+      "epoch": 0.9855642198878649,
+      "grad_norm": 0.00033894960956004134,
+      "learning_rate": 1.682778374507698e-07,
+      "loss": 0.0,
+      "step": 15293
+    },
+    {
+      "epoch": 0.9856286653347941,
+      "grad_norm": 0.12599162817477413,
+      "learning_rate": 1.6756176154672396e-07,
+      "loss": 0.0014,
+      "step": 15294
+    },
+    {
+      "epoch": 0.9856931107817233,
+      "grad_norm": 0.0004907227780169364,
+      "learning_rate": 1.6684568564267816e-07,
+      "loss": 0.0,
+      "step": 15295
+    },
+    {
+      "epoch": 0.9857575562286525,
+      "grad_norm": 0.3005311471082716,
+      "learning_rate": 1.661296097386323e-07,
+      "loss": 0.0012,
+      "step": 15296
+    },
+    {
+      "epoch": 0.9858220016755816,
+      "grad_norm": 0.001568842005944896,
+      "learning_rate": 1.6541353383458646e-07,
+      "loss": 0.0,
+      "step": 15297
+    },
+    {
+      "epoch": 0.9858864471225108,
+      "grad_norm": 0.004407247136844806,
+      "learning_rate": 1.6469745793054065e-07,
+      "loss": 0.0,
+      "step": 15298
+    },
+    {
+      "epoch": 0.98595089256944,
+      "grad_norm": 0.0003473983895130904,
+      "learning_rate": 1.6398138202649482e-07,
+      "loss": 0.0,
+      "step": 15299
+    },
+    {
+      "epoch": 0.9860153380163691,
+      "grad_norm": 0.00863692447582535,
+      "learning_rate": 1.6326530612244901e-07,
+      "loss": 0.0,
+      "step": 15300
+    },
+    {
+      "epoch": 0.9860797834632983,
+      "grad_norm": 0.00211103398049022,
+      "learning_rate": 1.6254923021840315e-07,
+      "loss": 0.0,
+      "step": 15301
+    },
+    {
+      "epoch": 0.9861442289102275,
+      "grad_norm": 0.0018230068901888296,
+      "learning_rate": 1.6183315431435732e-07,
+      "loss": 0.0,
+      "step": 15302
+    },
+    {
+      "epoch": 0.9862086743571566,
+      "grad_norm": 0.015832869636882328,
+      "learning_rate": 1.611170784103115e-07,
+      "loss": 0.0,
+      "step": 15303
+    },
+    {
+      "epoch": 0.9862731198040858,
+      "grad_norm": 0.023791212077183488,
+      "learning_rate": 1.6040100250626568e-07,
+      "loss": 0.0015,
+      "step": 15304
+    },
+    {
+      "epoch": 0.986337565251015,
+      "grad_norm": 0.0038339565004358363,
+      "learning_rate": 1.5968492660221985e-07,
+      "loss": 0.0,
+      "step": 15305
+    },
+    {
+      "epoch": 0.9864020106979442,
+      "grad_norm": 0.043639507155779036,
+      "learning_rate": 1.5896885069817404e-07,
+      "loss": 0.0001,
+      "step": 15306
+    },
+    {
+      "epoch": 0.9864664561448734,
+      "grad_norm": 0.006059066329226813,
+      "learning_rate": 1.5825277479412818e-07,
+      "loss": 0.0,
+      "step": 15307
+    },
+    {
+      "epoch": 0.9865309015918026,
+      "grad_norm": 0.0073793358972736945,
+      "learning_rate": 1.5753669889008237e-07,
+      "loss": 0.0001,
+      "step": 15308
+    },
+    {
+      "epoch": 0.9865953470387318,
+      "grad_norm": 0.11897962667166657,
+      "learning_rate": 1.5682062298603654e-07,
+      "loss": 0.0004,
+      "step": 15309
+    },
+    {
+      "epoch": 0.9866597924856609,
+      "grad_norm": 0.0016394853548579568,
+      "learning_rate": 1.561045470819907e-07,
+      "loss": 0.0,
+      "step": 15310
+    },
+    {
+      "epoch": 0.98672423793259,
+      "grad_norm": 0.0023086441209033705,
+      "learning_rate": 1.5538847117794487e-07,
+      "loss": 0.0,
+      "step": 15311
+    },
+    {
+      "epoch": 0.9867886833795192,
+      "grad_norm": 0.000473328226471948,
+      "learning_rate": 1.5467239527389904e-07,
+      "loss": 0.0,
+      "step": 15312
+    },
+    {
+      "epoch": 0.9868531288264484,
+      "grad_norm": 0.003990274313165149,
+      "learning_rate": 1.5395631936985323e-07,
+      "loss": 0.0,
+      "step": 15313
+    },
+    {
+      "epoch": 0.9869175742733776,
+      "grad_norm": 0.17149084682895488,
+      "learning_rate": 1.532402434658074e-07,
+      "loss": 0.0006,
+      "step": 15314
+    },
+    {
+      "epoch": 0.9869820197203067,
+      "grad_norm": 0.009257975158263423,
+      "learning_rate": 1.5252416756176156e-07,
+      "loss": 0.0,
+      "step": 15315
+    },
+    {
+      "epoch": 0.9870464651672359,
+      "grad_norm": 0.014955583267460219,
+      "learning_rate": 1.5180809165771573e-07,
+      "loss": 0.0,
+      "step": 15316
+    },
+    {
+      "epoch": 0.9871109106141651,
+      "grad_norm": 0.10271010438525041,
+      "learning_rate": 1.510920157536699e-07,
+      "loss": 0.0001,
+      "step": 15317
+    },
+    {
+      "epoch": 0.9871753560610943,
+      "grad_norm": 0.001647549115393784,
+      "learning_rate": 1.5037593984962406e-07,
+      "loss": 0.0,
+      "step": 15318
+    },
+    {
+      "epoch": 0.9872398015080235,
+      "grad_norm": 0.00025357547641687695,
+      "learning_rate": 1.4965986394557823e-07,
+      "loss": 0.0,
+      "step": 15319
+    },
+    {
+      "epoch": 0.9873042469549527,
+      "grad_norm": 0.019612992119780556,
+      "learning_rate": 1.4894378804153242e-07,
+      "loss": 0.0,
+      "step": 15320
+    },
+    {
+      "epoch": 0.9873686924018819,
+      "grad_norm": 0.001171164674540023,
+      "learning_rate": 1.4822771213748659e-07,
+      "loss": 0.0,
+      "step": 15321
+    },
+    {
+      "epoch": 0.9874331378488109,
+      "grad_norm": 0.0034558749272922897,
+      "learning_rate": 1.4751163623344075e-07,
+      "loss": 0.0,
+      "step": 15322
+    },
+    {
+      "epoch": 0.9874975832957401,
+      "grad_norm": 0.00045139744625150767,
+      "learning_rate": 1.4679556032939492e-07,
+      "loss": 0.0,
+      "step": 15323
+    },
+    {
+      "epoch": 0.9875620287426693,
+      "grad_norm": 0.0007023775731221126,
+      "learning_rate": 1.460794844253491e-07,
+      "loss": 0.0,
+      "step": 15324
+    },
+    {
+      "epoch": 0.9876264741895985,
+      "grad_norm": 0.0009513016630730917,
+      "learning_rate": 1.4536340852130328e-07,
+      "loss": 0.0,
+      "step": 15325
+    },
+    {
+      "epoch": 0.9876909196365277,
+      "grad_norm": 0.05982552834564737,
+      "learning_rate": 1.4464733261725744e-07,
+      "loss": 0.0002,
+      "step": 15326
+    },
+    {
+      "epoch": 0.9877553650834568,
+      "grad_norm": 0.003501219232904495,
+      "learning_rate": 1.439312567132116e-07,
+      "loss": 0.0,
+      "step": 15327
+    },
+    {
+      "epoch": 0.987819810530386,
+      "grad_norm": 0.007539849747465022,
+      "learning_rate": 1.4321518080916578e-07,
+      "loss": 0.0,
+      "step": 15328
+    },
+    {
+      "epoch": 0.9878842559773152,
+      "grad_norm": 0.0007982695888039108,
+      "learning_rate": 1.4249910490511997e-07,
+      "loss": 0.0,
+      "step": 15329
+    },
+    {
+      "epoch": 0.9879487014242444,
+      "grad_norm": 0.0011078097149875995,
+      "learning_rate": 1.417830290010741e-07,
+      "loss": 0.0,
+      "step": 15330
+    },
+    {
+      "epoch": 0.9880131468711736,
+      "grad_norm": 0.0003443949979325242,
+      "learning_rate": 1.410669530970283e-07,
+      "loss": 0.0,
+      "step": 15331
+    },
+    {
+      "epoch": 0.9880775923181028,
+      "grad_norm": 0.1633003675412003,
+      "learning_rate": 1.4035087719298247e-07,
+      "loss": 0.0005,
+      "step": 15332
+    },
+    {
+      "epoch": 0.988142037765032,
+      "grad_norm": 0.18057904691791732,
+      "learning_rate": 1.3963480128893664e-07,
+      "loss": 0.0018,
+      "step": 15333
+    },
+    {
+      "epoch": 0.988206483211961,
+      "grad_norm": 0.0007753720428181999,
+      "learning_rate": 1.389187253848908e-07,
+      "loss": 0.0,
+      "step": 15334
+    },
+    {
+      "epoch": 0.9882709286588902,
+      "grad_norm": 0.0017123565601338646,
+      "learning_rate": 1.3820264948084497e-07,
+      "loss": 0.0,
+      "step": 15335
+    },
+    {
+      "epoch": 0.9883353741058194,
+      "grad_norm": 0.0344613932183989,
+      "learning_rate": 1.3748657357679916e-07,
+      "loss": 0.0,
+      "step": 15336
+    },
+    {
+      "epoch": 0.9883998195527486,
+      "grad_norm": 0.19325097253192008,
+      "learning_rate": 1.3677049767275333e-07,
+      "loss": 0.0002,
+      "step": 15337
+    },
+    {
+      "epoch": 0.9884642649996778,
+      "grad_norm": 0.0009819431197656918,
+      "learning_rate": 1.360544217687075e-07,
+      "loss": 0.0,
+      "step": 15338
+    },
+    {
+      "epoch": 0.988528710446607,
+      "grad_norm": 0.1516750895755982,
+      "learning_rate": 1.3533834586466166e-07,
+      "loss": 0.0023,
+      "step": 15339
+    },
+    {
+      "epoch": 0.9885931558935361,
+      "grad_norm": 0.001658402824212663,
+      "learning_rate": 1.3462226996061585e-07,
+      "loss": 0.0,
+      "step": 15340
+    },
+    {
+      "epoch": 0.9886576013404653,
+      "grad_norm": 0.0015698162912195304,
+      "learning_rate": 1.3390619405657002e-07,
+      "loss": 0.0,
+      "step": 15341
+    },
+    {
+      "epoch": 0.9887220467873945,
+      "grad_norm": 0.008429947141429642,
+      "learning_rate": 1.3319011815252419e-07,
+      "loss": 0.0001,
+      "step": 15342
+    },
+    {
+      "epoch": 0.9887864922343237,
+      "grad_norm": 0.0024822454325503923,
+      "learning_rate": 1.3247404224847835e-07,
+      "loss": 0.0,
+      "step": 15343
+    },
+    {
+      "epoch": 0.9888509376812529,
+      "grad_norm": 4.2695759973628696e-05,
+      "learning_rate": 1.3175796634443252e-07,
+      "loss": 0.0,
+      "step": 15344
+    },
+    {
+      "epoch": 0.988915383128182,
+      "grad_norm": 0.03248854376485151,
+      "learning_rate": 1.3104189044038668e-07,
+      "loss": 0.0001,
+      "step": 15345
+    },
+    {
+      "epoch": 0.9889798285751111,
+      "grad_norm": 0.009666005164565861,
+      "learning_rate": 1.3032581453634085e-07,
+      "loss": 0.0,
+      "step": 15346
+    },
+    {
+      "epoch": 0.9890442740220403,
+      "grad_norm": 0.012461908937499057,
+      "learning_rate": 1.2960973863229504e-07,
+      "loss": 0.0,
+      "step": 15347
+    },
+    {
+      "epoch": 0.9891087194689695,
+      "grad_norm": 0.009181846757203724,
+      "learning_rate": 1.288936627282492e-07,
+      "loss": 0.0,
+      "step": 15348
+    },
+    {
+      "epoch": 0.9891731649158987,
+      "grad_norm": 0.3086505185798492,
+      "learning_rate": 1.2817758682420338e-07,
+      "loss": 0.0005,
+      "step": 15349
+    },
+    {
+      "epoch": 0.9892376103628279,
+      "grad_norm": 9.000946232606643e-05,
+      "learning_rate": 1.2746151092015754e-07,
+      "loss": 0.0,
+      "step": 15350
+    },
+    {
+      "epoch": 0.989302055809757,
+      "grad_norm": 0.00017109891452988802,
+      "learning_rate": 1.267454350161117e-07,
+      "loss": 0.0,
+      "step": 15351
+    },
+    {
+      "epoch": 0.9893665012566862,
+      "grad_norm": 0.0014436826725710914,
+      "learning_rate": 1.260293591120659e-07,
+      "loss": 0.0,
+      "step": 15352
+    },
+    {
+      "epoch": 0.9894309467036154,
+      "grad_norm": 0.010794710318647652,
+      "learning_rate": 1.2531328320802004e-07,
+      "loss": 0.0,
+      "step": 15353
+    },
+    {
+      "epoch": 0.9894953921505446,
+      "grad_norm": 0.00487863410298011,
+      "learning_rate": 1.2459720730397423e-07,
+      "loss": 0.0,
+      "step": 15354
+    },
+    {
+      "epoch": 0.9895598375974738,
+      "grad_norm": 0.004251305964631465,
+      "learning_rate": 1.238811313999284e-07,
+      "loss": 0.0,
+      "step": 15355
+    },
+    {
+      "epoch": 0.9896242830444029,
+      "grad_norm": 0.0009995787571245016,
+      "learning_rate": 1.2316505549588257e-07,
+      "loss": 0.0,
+      "step": 15356
+    },
+    {
+      "epoch": 0.989688728491332,
+      "grad_norm": 0.0002976666463085584,
+      "learning_rate": 1.2244897959183673e-07,
+      "loss": 0.0,
+      "step": 15357
+    },
+    {
+      "epoch": 0.9897531739382612,
+      "grad_norm": 0.00029859447566665933,
+      "learning_rate": 1.2173290368779093e-07,
+      "loss": 0.0,
+      "step": 15358
+    },
+    {
+      "epoch": 0.9898176193851904,
+      "grad_norm": 0.017260082460229826,
+      "learning_rate": 1.210168277837451e-07,
+      "loss": 0.0002,
+      "step": 15359
+    },
+    {
+      "epoch": 0.9898820648321196,
+      "grad_norm": 0.03121110798224628,
+      "learning_rate": 1.2030075187969926e-07,
+      "loss": 0.0001,
+      "step": 15360
+    },
+    {
+      "epoch": 0.9899465102790488,
+      "grad_norm": 0.0005941675613315212,
+      "learning_rate": 1.1958467597565343e-07,
+      "loss": 0.0,
+      "step": 15361
+    },
+    {
+      "epoch": 0.990010955725978,
+      "grad_norm": 9.114690847879131e-05,
+      "learning_rate": 1.188686000716076e-07,
+      "loss": 0.0,
+      "step": 15362
+    },
+    {
+      "epoch": 0.9900754011729072,
+      "grad_norm": 1.7429828966388234,
+      "learning_rate": 1.1815252416756177e-07,
+      "loss": 0.0055,
+      "step": 15363
+    },
+    {
+      "epoch": 0.9901398466198363,
+      "grad_norm": 0.518626867181079,
+      "learning_rate": 1.1743644826351595e-07,
+      "loss": 0.0028,
+      "step": 15364
+    },
+    {
+      "epoch": 0.9902042920667655,
+      "grad_norm": 0.3350462068176509,
+      "learning_rate": 1.167203723594701e-07,
+      "loss": 0.0003,
+      "step": 15365
+    },
+    {
+      "epoch": 0.9902687375136947,
+      "grad_norm": 0.05405229337421042,
+      "learning_rate": 1.1600429645542428e-07,
+      "loss": 0.0001,
+      "step": 15366
+    },
+    {
+      "epoch": 0.9903331829606238,
+      "grad_norm": 0.06269417714193161,
+      "learning_rate": 1.1528822055137846e-07,
+      "loss": 0.0001,
+      "step": 15367
+    },
+    {
+      "epoch": 0.990397628407553,
+      "grad_norm": 0.00020441138102923534,
+      "learning_rate": 1.1457214464733263e-07,
+      "loss": 0.0,
+      "step": 15368
+    },
+    {
+      "epoch": 0.9904620738544822,
+      "grad_norm": 0.3037918203904865,
+      "learning_rate": 1.138560687432868e-07,
+      "loss": 0.0008,
+      "step": 15369
+    },
+    {
+      "epoch": 0.9905265193014113,
+      "grad_norm": 0.02238230572787554,
+      "learning_rate": 1.1313999283924098e-07,
+      "loss": 0.0002,
+      "step": 15370
+    },
+    {
+      "epoch": 0.9905909647483405,
+      "grad_norm": 0.1309645785006673,
+      "learning_rate": 1.1242391693519514e-07,
+      "loss": 0.0003,
+      "step": 15371
+    },
+    {
+      "epoch": 0.9906554101952697,
+      "grad_norm": 0.001562903964244142,
+      "learning_rate": 1.1170784103114931e-07,
+      "loss": 0.0,
+      "step": 15372
+    },
+    {
+      "epoch": 0.9907198556421989,
+      "grad_norm": 0.45439102627328637,
+      "learning_rate": 1.1099176512710347e-07,
+      "loss": 0.0021,
+      "step": 15373
+    },
+    {
+      "epoch": 0.9907843010891281,
+      "grad_norm": 0.4411138224181444,
+      "learning_rate": 1.1027568922305765e-07,
+      "loss": 0.0053,
+      "step": 15374
+    },
+    {
+      "epoch": 0.9908487465360573,
+      "grad_norm": 0.0006388717337027706,
+      "learning_rate": 1.0955961331901183e-07,
+      "loss": 0.0,
+      "step": 15375
+    },
+    {
+      "epoch": 0.9909131919829864,
+      "grad_norm": 0.08775647044388801,
+      "learning_rate": 1.0884353741496599e-07,
+      "loss": 0.0009,
+      "step": 15376
+    },
+    {
+      "epoch": 0.9909776374299156,
+      "grad_norm": 0.01739325179771781,
+      "learning_rate": 1.0812746151092017e-07,
+      "loss": 0.0001,
+      "step": 15377
+    },
+    {
+      "epoch": 0.9910420828768447,
+      "grad_norm": 0.034971450178768425,
+      "learning_rate": 1.0741138560687433e-07,
+      "loss": 0.0001,
+      "step": 15378
+    },
+    {
+      "epoch": 0.9911065283237739,
+      "grad_norm": 0.21837122514810614,
+      "learning_rate": 1.0669530970282851e-07,
+      "loss": 0.0027,
+      "step": 15379
+    },
+    {
+      "epoch": 0.9911709737707031,
+      "grad_norm": 0.03615418008543626,
+      "learning_rate": 1.0597923379878268e-07,
+      "loss": 0.0,
+      "step": 15380
+    },
+    {
+      "epoch": 0.9912354192176323,
+      "grad_norm": 0.0013153067476646975,
+      "learning_rate": 1.0526315789473685e-07,
+      "loss": 0.0,
+      "step": 15381
+    },
+    {
+      "epoch": 0.9912998646645614,
+      "grad_norm": 0.013645522977196328,
+      "learning_rate": 1.0454708199069102e-07,
+      "loss": 0.0001,
+      "step": 15382
+    },
+    {
+      "epoch": 0.9913643101114906,
+      "grad_norm": 0.018869085279380988,
+      "learning_rate": 1.038310060866452e-07,
+      "loss": 0.0,
+      "step": 15383
+    },
+    {
+      "epoch": 0.9914287555584198,
+      "grad_norm": 0.00017469868199359676,
+      "learning_rate": 1.0311493018259936e-07,
+      "loss": 0.0,
+      "step": 15384
+    },
+    {
+      "epoch": 0.991493201005349,
+      "grad_norm": 0.004191734400330657,
+      "learning_rate": 1.0239885427855354e-07,
+      "loss": 0.0,
+      "step": 15385
+    },
+    {
+      "epoch": 0.9915576464522782,
+      "grad_norm": 0.00111540569401605,
+      "learning_rate": 1.016827783745077e-07,
+      "loss": 0.0,
+      "step": 15386
+    },
+    {
+      "epoch": 0.9916220918992074,
+      "grad_norm": 0.0028105942217102444,
+      "learning_rate": 1.0096670247046188e-07,
+      "loss": 0.0,
+      "step": 15387
+    },
+    {
+      "epoch": 0.9916865373461365,
+      "grad_norm": 0.0014986055018981306,
+      "learning_rate": 1.0025062656641605e-07,
+      "loss": 0.0,
+      "step": 15388
+    },
+    {
+      "epoch": 0.9917509827930656,
+      "grad_norm": 0.0698733557021849,
+      "learning_rate": 9.953455066237022e-08,
+      "loss": 0.0001,
+      "step": 15389
+    },
+    {
+      "epoch": 0.9918154282399948,
+      "grad_norm": 1.3593856109555784,
+      "learning_rate": 9.88184747583244e-08,
+      "loss": 0.0078,
+      "step": 15390
+    },
+    {
+      "epoch": 0.991879873686924,
+      "grad_norm": 1.6501662604056602,
+      "learning_rate": 9.810239885427857e-08,
+      "loss": 0.0138,
+      "step": 15391
+    },
+    {
+      "epoch": 0.9919443191338532,
+      "grad_norm": 0.17455723899428105,
+      "learning_rate": 9.738632295023273e-08,
+      "loss": 0.0006,
+      "step": 15392
+    },
+    {
+      "epoch": 0.9920087645807824,
+      "grad_norm": 0.011819212039522532,
+      "learning_rate": 9.667024704618691e-08,
+      "loss": 0.0001,
+      "step": 15393
+    },
+    {
+      "epoch": 0.9920732100277115,
+      "grad_norm": 0.00014493524179976452,
+      "learning_rate": 9.595417114214107e-08,
+      "loss": 0.0,
+      "step": 15394
+    },
+    {
+      "epoch": 0.9921376554746407,
+      "grad_norm": 0.0006374483992628114,
+      "learning_rate": 9.523809523809525e-08,
+      "loss": 0.0,
+      "step": 15395
+    },
+    {
+      "epoch": 0.9922021009215699,
+      "grad_norm": 0.16560081944202426,
+      "learning_rate": 9.45220193340494e-08,
+      "loss": 0.0011,
+      "step": 15396
+    },
+    {
+      "epoch": 0.9922665463684991,
+      "grad_norm": 0.00026379790258233425,
+      "learning_rate": 9.380594343000359e-08,
+      "loss": 0.0,
+      "step": 15397
+    },
+    {
+      "epoch": 0.9923309918154283,
+      "grad_norm": 0.02047838287760723,
+      "learning_rate": 9.308986752595777e-08,
+      "loss": 0.0001,
+      "step": 15398
+    },
+    {
+      "epoch": 0.9923954372623575,
+      "grad_norm": 0.2446784085500637,
+      "learning_rate": 9.237379162191192e-08,
+      "loss": 0.0004,
+      "step": 15399
+    },
+    {
+      "epoch": 0.9924598827092865,
+      "grad_norm": 0.003227860027021074,
+      "learning_rate": 9.16577157178661e-08,
+      "loss": 0.0,
+      "step": 15400
+    },
+    {
+      "epoch": 0.9925243281562157,
+      "grad_norm": 0.0002485238947423622,
+      "learning_rate": 9.094163981382028e-08,
+      "loss": 0.0,
+      "step": 15401
+    },
+    {
+      "epoch": 0.9925887736031449,
+      "grad_norm": 0.01720941809554806,
+      "learning_rate": 9.022556390977444e-08,
+      "loss": 0.0001,
+      "step": 15402
+    },
+    {
+      "epoch": 0.9926532190500741,
+      "grad_norm": 0.00032100960062109813,
+      "learning_rate": 8.950948800572861e-08,
+      "loss": 0.0,
+      "step": 15403
+    },
+    {
+      "epoch": 0.9927176644970033,
+      "grad_norm": 0.00013967251299634785,
+      "learning_rate": 8.879341210168278e-08,
+      "loss": 0.0,
+      "step": 15404
+    },
+    {
+      "epoch": 0.9927821099439325,
+      "grad_norm": 0.009221607936256614,
+      "learning_rate": 8.807733619763696e-08,
+      "loss": 0.0,
+      "step": 15405
+    },
+    {
+      "epoch": 0.9928465553908616,
+      "grad_norm": 0.02119800257858215,
+      "learning_rate": 8.736126029359114e-08,
+      "loss": 0.0015,
+      "step": 15406
+    },
+    {
+      "epoch": 0.9929110008377908,
+      "grad_norm": 0.006607428148178863,
+      "learning_rate": 8.664518438954529e-08,
+      "loss": 0.0001,
+      "step": 15407
+    },
+    {
+      "epoch": 0.99297544628472,
+      "grad_norm": 0.000993933285930776,
+      "learning_rate": 8.592910848549947e-08,
+      "loss": 0.0,
+      "step": 15408
+    },
+    {
+      "epoch": 0.9930398917316492,
+      "grad_norm": 0.04713009178308316,
+      "learning_rate": 8.521303258145365e-08,
+      "loss": 0.0001,
+      "step": 15409
+    },
+    {
+      "epoch": 0.9931043371785784,
+      "grad_norm": 0.00726139924226023,
+      "learning_rate": 8.449695667740781e-08,
+      "loss": 0.0001,
+      "step": 15410
+    },
+    {
+      "epoch": 0.9931687826255076,
+      "grad_norm": 0.0018708575708756142,
+      "learning_rate": 8.378088077336198e-08,
+      "loss": 0.0,
+      "step": 15411
+    },
+    {
+      "epoch": 0.9932332280724366,
+      "grad_norm": 0.0834809032921711,
+      "learning_rate": 8.306480486931615e-08,
+      "loss": 0.0002,
+      "step": 15412
+    },
+    {
+      "epoch": 0.9932976735193658,
+      "grad_norm": 0.00013640415194037607,
+      "learning_rate": 8.234872896527033e-08,
+      "loss": 0.0,
+      "step": 15413
+    },
+    {
+      "epoch": 0.993362118966295,
+      "grad_norm": 0.014602227139775253,
+      "learning_rate": 8.163265306122451e-08,
+      "loss": 0.0,
+      "step": 15414
+    },
+    {
+      "epoch": 0.9934265644132242,
+      "grad_norm": 0.01868237382937571,
+      "learning_rate": 8.091657715717866e-08,
+      "loss": 0.0001,
+      "step": 15415
+    },
+    {
+      "epoch": 0.9934910098601534,
+      "grad_norm": 0.6517121232399404,
+      "learning_rate": 8.020050125313284e-08,
+      "loss": 0.0011,
+      "step": 15416
+    },
+    {
+      "epoch": 0.9935554553070826,
+      "grad_norm": 0.0210594611327598,
+      "learning_rate": 7.948442534908702e-08,
+      "loss": 0.0,
+      "step": 15417
+    },
+    {
+      "epoch": 0.9936199007540117,
+      "grad_norm": 0.20600616643122802,
+      "learning_rate": 7.876834944504119e-08,
+      "loss": 0.0003,
+      "step": 15418
+    },
+    {
+      "epoch": 0.9936843462009409,
+      "grad_norm": 0.0006812269480493978,
+      "learning_rate": 7.805227354099535e-08,
+      "loss": 0.0,
+      "step": 15419
+    },
+    {
+      "epoch": 0.9937487916478701,
+      "grad_norm": 0.011687333493238744,
+      "learning_rate": 7.733619763694952e-08,
+      "loss": 0.0,
+      "step": 15420
+    },
+    {
+      "epoch": 0.9938132370947993,
+      "grad_norm": 0.00022207912377938065,
+      "learning_rate": 7.66201217329037e-08,
+      "loss": 0.0,
+      "step": 15421
+    },
+    {
+      "epoch": 0.9938776825417285,
+      "grad_norm": 0.006105163470982768,
+      "learning_rate": 7.590404582885786e-08,
+      "loss": 0.0,
+      "step": 15422
+    },
+    {
+      "epoch": 0.9939421279886576,
+      "grad_norm": 0.004071650145522754,
+      "learning_rate": 7.518796992481203e-08,
+      "loss": 0.0,
+      "step": 15423
+    },
+    {
+      "epoch": 0.9940065734355867,
+      "grad_norm": 0.0001294769158907121,
+      "learning_rate": 7.447189402076621e-08,
+      "loss": 0.0,
+      "step": 15424
+    },
+    {
+      "epoch": 0.9940710188825159,
+      "grad_norm": 0.007293988964356481,
+      "learning_rate": 7.375581811672038e-08,
+      "loss": 0.0,
+      "step": 15425
+    },
+    {
+      "epoch": 0.9941354643294451,
+      "grad_norm": 0.5071361732953923,
+      "learning_rate": 7.303974221267456e-08,
+      "loss": 0.0024,
+      "step": 15426
+    },
+    {
+      "epoch": 0.9941999097763743,
+      "grad_norm": 0.004333334066827409,
+      "learning_rate": 7.232366630862872e-08,
+      "loss": 0.0,
+      "step": 15427
+    },
+    {
+      "epoch": 0.9942643552233035,
+      "grad_norm": 0.017363078925581327,
+      "learning_rate": 7.160759040458289e-08,
+      "loss": 0.0,
+      "step": 15428
+    },
+    {
+      "epoch": 0.9943288006702327,
+      "grad_norm": 0.0011615992960153968,
+      "learning_rate": 7.089151450053705e-08,
+      "loss": 0.0,
+      "step": 15429
+    },
+    {
+      "epoch": 0.9943932461171618,
+      "grad_norm": 0.0003038760653883744,
+      "learning_rate": 7.017543859649123e-08,
+      "loss": 0.0,
+      "step": 15430
+    },
+    {
+      "epoch": 0.994457691564091,
+      "grad_norm": 0.000252489032190593,
+      "learning_rate": 6.94593626924454e-08,
+      "loss": 0.0,
+      "step": 15431
+    },
+    {
+      "epoch": 0.9945221370110202,
+      "grad_norm": 0.0002281339806080607,
+      "learning_rate": 6.874328678839958e-08,
+      "loss": 0.0,
+      "step": 15432
+    },
+    {
+      "epoch": 0.9945865824579494,
+      "grad_norm": 0.0021669236006270874,
+      "learning_rate": 6.802721088435375e-08,
+      "loss": 0.0,
+      "step": 15433
+    },
+    {
+      "epoch": 0.9946510279048785,
+      "grad_norm": 0.0035040173439457962,
+      "learning_rate": 6.731113498030793e-08,
+      "loss": 0.0,
+      "step": 15434
+    },
+    {
+      "epoch": 0.9947154733518077,
+      "grad_norm": 4.2521529407209614e-05,
+      "learning_rate": 6.659505907626209e-08,
+      "loss": 0.0,
+      "step": 15435
+    },
+    {
+      "epoch": 0.9947799187987368,
+      "grad_norm": 0.0026709807723655685,
+      "learning_rate": 6.587898317221626e-08,
+      "loss": 0.0,
+      "step": 15436
+    },
+    {
+      "epoch": 0.994844364245666,
+      "grad_norm": 0.010820683146318837,
+      "learning_rate": 6.516290726817043e-08,
+      "loss": 0.0,
+      "step": 15437
+    },
+    {
+      "epoch": 0.9949088096925952,
+      "grad_norm": 0.003205757880112345,
+      "learning_rate": 6.44468313641246e-08,
+      "loss": 0.0,
+      "step": 15438
+    },
+    {
+      "epoch": 0.9949732551395244,
+      "grad_norm": 0.0005007387307297139,
+      "learning_rate": 6.373075546007877e-08,
+      "loss": 0.0,
+      "step": 15439
+    },
+    {
+      "epoch": 0.9950377005864536,
+      "grad_norm": 0.046509660126946864,
+      "learning_rate": 6.301467955603295e-08,
+      "loss": 0.0003,
+      "step": 15440
+    },
+    {
+      "epoch": 0.9951021460333828,
+      "grad_norm": 0.013549052359340931,
+      "learning_rate": 6.229860365198712e-08,
+      "loss": 0.0,
+      "step": 15441
+    },
+    {
+      "epoch": 0.995166591480312,
+      "grad_norm": 0.009939784287012338,
+      "learning_rate": 6.158252774794128e-08,
+      "loss": 0.0001,
+      "step": 15442
+    },
+    {
+      "epoch": 0.9952310369272411,
+      "grad_norm": 0.002741087693011581,
+      "learning_rate": 6.086645184389546e-08,
+      "loss": 0.0,
+      "step": 15443
+    },
+    {
+      "epoch": 0.9952954823741703,
+      "grad_norm": 0.006260263699628107,
+      "learning_rate": 6.015037593984963e-08,
+      "loss": 0.0,
+      "step": 15444
+    },
+    {
+      "epoch": 0.9953599278210994,
+      "grad_norm": 0.00049759231216585,
+      "learning_rate": 5.94343000358038e-08,
+      "loss": 0.0,
+      "step": 15445
+    },
+    {
+      "epoch": 0.9954243732680286,
+      "grad_norm": 0.1380881394314596,
+      "learning_rate": 5.8718224131757975e-08,
+      "loss": 0.0003,
+      "step": 15446
+    },
+    {
+      "epoch": 0.9954888187149578,
+      "grad_norm": 0.006507125813773554,
+      "learning_rate": 5.800214822771214e-08,
+      "loss": 0.0,
+      "step": 15447
+    },
+    {
+      "epoch": 0.9955532641618869,
+      "grad_norm": 0.007913763476941546,
+      "learning_rate": 5.7286072323666315e-08,
+      "loss": 0.0,
+      "step": 15448
+    },
+    {
+      "epoch": 0.9956177096088161,
+      "grad_norm": 0.0058919833109365456,
+      "learning_rate": 5.656999641962049e-08,
+      "loss": 0.0,
+      "step": 15449
+    },
+    {
+      "epoch": 0.9956821550557453,
+      "grad_norm": 0.042774442493267895,
+      "learning_rate": 5.5853920515574654e-08,
+      "loss": 0.0002,
+      "step": 15450
+    },
+    {
+      "epoch": 0.9957466005026745,
+      "grad_norm": 0.0007127180380731889,
+      "learning_rate": 5.513784461152883e-08,
+      "loss": 0.0,
+      "step": 15451
+    },
+    {
+      "epoch": 0.9958110459496037,
+      "grad_norm": 0.1438407771996523,
+      "learning_rate": 5.4421768707482993e-08,
+      "loss": 0.0006,
+      "step": 15452
+    },
+    {
+      "epoch": 0.9958754913965329,
+      "grad_norm": 0.0007669071129745034,
+      "learning_rate": 5.3705692803437166e-08,
+      "loss": 0.0,
+      "step": 15453
+    },
+    {
+      "epoch": 0.995939936843462,
+      "grad_norm": 0.0002747317806881671,
+      "learning_rate": 5.298961689939134e-08,
+      "loss": 0.0,
+      "step": 15454
+    },
+    {
+      "epoch": 0.9960043822903912,
+      "grad_norm": 5.1187894086526105e-05,
+      "learning_rate": 5.227354099534551e-08,
+      "loss": 0.0,
+      "step": 15455
+    },
+    {
+      "epoch": 0.9960688277373203,
+      "grad_norm": 0.19142032104464268,
+      "learning_rate": 5.155746509129968e-08,
+      "loss": 0.0006,
+      "step": 15456
+    },
+    {
+      "epoch": 0.9961332731842495,
+      "grad_norm": 0.6594525034042554,
+      "learning_rate": 5.084138918725385e-08,
+      "loss": 0.0018,
+      "step": 15457
+    },
+    {
+      "epoch": 0.9961977186311787,
+      "grad_norm": 0.0003211975571060089,
+      "learning_rate": 5.0125313283208025e-08,
+      "loss": 0.0,
+      "step": 15458
+    },
+    {
+      "epoch": 0.9962621640781079,
+      "grad_norm": 0.0012403521668988536,
+      "learning_rate": 4.94092373791622e-08,
+      "loss": 0.0,
+      "step": 15459
+    },
+    {
+      "epoch": 0.996326609525037,
+      "grad_norm": 0.038828634005938584,
+      "learning_rate": 4.8693161475116364e-08,
+      "loss": 0.0001,
+      "step": 15460
+    },
+    {
+      "epoch": 0.9963910549719662,
+      "grad_norm": 0.0019163043650427354,
+      "learning_rate": 4.797708557107054e-08,
+      "loss": 0.0,
+      "step": 15461
+    },
+    {
+      "epoch": 0.9964555004188954,
+      "grad_norm": 0.006335604384163084,
+      "learning_rate": 4.72610096670247e-08,
+      "loss": 0.0,
+      "step": 15462
+    },
+    {
+      "epoch": 0.9965199458658246,
+      "grad_norm": 0.0015440385523529667,
+      "learning_rate": 4.654493376297888e-08,
+      "loss": 0.0,
+      "step": 15463
+    },
+    {
+      "epoch": 0.9965843913127538,
+      "grad_norm": 0.000717511672539573,
+      "learning_rate": 4.582885785893305e-08,
+      "loss": 0.0,
+      "step": 15464
+    },
+    {
+      "epoch": 0.996648836759683,
+      "grad_norm": 0.00018937198611685636,
+      "learning_rate": 4.511278195488722e-08,
+      "loss": 0.0,
+      "step": 15465
+    },
+    {
+      "epoch": 0.9967132822066122,
+      "grad_norm": 0.0010951535217054588,
+      "learning_rate": 4.439670605084139e-08,
+      "loss": 0.0,
+      "step": 15466
+    },
+    {
+      "epoch": 0.9967777276535412,
+      "grad_norm": 0.038216985921274704,
+      "learning_rate": 4.368063014679557e-08,
+      "loss": 0.0,
+      "step": 15467
+    },
+    {
+      "epoch": 0.9968421731004704,
+      "grad_norm": 0.001310616487340104,
+      "learning_rate": 4.2964554242749734e-08,
+      "loss": 0.0,
+      "step": 15468
+    },
+    {
+      "epoch": 0.9969066185473996,
+      "grad_norm": 0.006292524679154751,
+      "learning_rate": 4.224847833870391e-08,
+      "loss": 0.0001,
+      "step": 15469
+    },
+    {
+      "epoch": 0.9969710639943288,
+      "grad_norm": 0.000838499974695811,
+      "learning_rate": 4.1532402434658074e-08,
+      "loss": 0.0,
+      "step": 15470
+    },
+    {
+      "epoch": 0.997035509441258,
+      "grad_norm": 0.0010050876167897022,
+      "learning_rate": 4.0816326530612253e-08,
+      "loss": 0.0,
+      "step": 15471
+    },
+    {
+      "epoch": 0.9970999548881871,
+      "grad_norm": 0.005370776582966026,
+      "learning_rate": 4.010025062656642e-08,
+      "loss": 0.0,
+      "step": 15472
+    },
+    {
+      "epoch": 0.9971644003351163,
+      "grad_norm": 0.0003092987419902564,
+      "learning_rate": 3.938417472252059e-08,
+      "loss": 0.0,
+      "step": 15473
+    },
+    {
+      "epoch": 0.9972288457820455,
+      "grad_norm": 0.0013591428298022192,
+      "learning_rate": 3.866809881847476e-08,
+      "loss": 0.0,
+      "step": 15474
+    },
+    {
+      "epoch": 0.9972932912289747,
+      "grad_norm": 0.23138436589381467,
+      "learning_rate": 3.795202291442893e-08,
+      "loss": 0.0018,
+      "step": 15475
+    },
+    {
+      "epoch": 0.9973577366759039,
+      "grad_norm": 0.001873350951168154,
+      "learning_rate": 3.7235947010383105e-08,
+      "loss": 0.0,
+      "step": 15476
+    },
+    {
+      "epoch": 0.9974221821228331,
+      "grad_norm": 0.004598840045965651,
+      "learning_rate": 3.651987110633728e-08,
+      "loss": 0.0,
+      "step": 15477
+    },
+    {
+      "epoch": 0.9974866275697621,
+      "grad_norm": 0.005085088395541839,
+      "learning_rate": 3.5803795202291444e-08,
+      "loss": 0.0,
+      "step": 15478
+    },
+    {
+      "epoch": 0.9975510730166913,
+      "grad_norm": 0.001091900765744506,
+      "learning_rate": 3.508771929824562e-08,
+      "loss": 0.0,
+      "step": 15479
+    },
+    {
+      "epoch": 0.9976155184636205,
+      "grad_norm": 0.0009730691764594346,
+      "learning_rate": 3.437164339419979e-08,
+      "loss": 0.0,
+      "step": 15480
+    },
+    {
+      "epoch": 0.9976799639105497,
+      "grad_norm": 0.0017135795294502306,
+      "learning_rate": 3.365556749015396e-08,
+      "loss": 0.0,
+      "step": 15481
+    },
+    {
+      "epoch": 0.9977444093574789,
+      "grad_norm": 0.01811945222196252,
+      "learning_rate": 3.293949158610813e-08,
+      "loss": 0.0002,
+      "step": 15482
+    },
+    {
+      "epoch": 0.9978088548044081,
+      "grad_norm": 0.009926933709437718,
+      "learning_rate": 3.22234156820623e-08,
+      "loss": 0.0,
+      "step": 15483
+    },
+    {
+      "epoch": 0.9978733002513372,
+      "grad_norm": 0.0017775873375116566,
+      "learning_rate": 3.1507339778016475e-08,
+      "loss": 0.0,
+      "step": 15484
+    },
+    {
+      "epoch": 0.9979377456982664,
+      "grad_norm": 0.0006008194079299575,
+      "learning_rate": 3.079126387397064e-08,
+      "loss": 0.0,
+      "step": 15485
+    },
+    {
+      "epoch": 0.9980021911451956,
+      "grad_norm": 0.00021254665315161218,
+      "learning_rate": 3.0075187969924815e-08,
+      "loss": 0.0,
+      "step": 15486
+    },
+    {
+      "epoch": 0.9980666365921248,
+      "grad_norm": 0.00026559116747325334,
+      "learning_rate": 2.9359112065878988e-08,
+      "loss": 0.0,
+      "step": 15487
+    },
+    {
+      "epoch": 0.998131082039054,
+      "grad_norm": 0.7161213032538332,
+      "learning_rate": 2.8643036161833157e-08,
+      "loss": 0.0038,
+      "step": 15488
+    },
+    {
+      "epoch": 0.9981955274859832,
+      "grad_norm": 0.00048493262049720823,
+      "learning_rate": 2.7926960257787327e-08,
+      "loss": 0.0,
+      "step": 15489
+    },
+    {
+      "epoch": 0.9982599729329122,
+      "grad_norm": 0.012352251282173892,
+      "learning_rate": 2.7210884353741497e-08,
+      "loss": 0.0,
+      "step": 15490
+    },
+    {
+      "epoch": 0.9983244183798414,
+      "grad_norm": 0.00370486564736607,
+      "learning_rate": 2.649480844969567e-08,
+      "loss": 0.0,
+      "step": 15491
+    },
+    {
+      "epoch": 0.9983888638267706,
+      "grad_norm": 0.0036452617386337023,
+      "learning_rate": 2.577873254564984e-08,
+      "loss": 0.0,
+      "step": 15492
+    },
+    {
+      "epoch": 0.9984533092736998,
+      "grad_norm": 0.1567663838989723,
+      "learning_rate": 2.5062656641604012e-08,
+      "loss": 0.0002,
+      "step": 15493
+    },
+    {
+      "epoch": 0.998517754720629,
+      "grad_norm": 0.0005995450102191095,
+      "learning_rate": 2.4346580737558182e-08,
+      "loss": 0.0,
+      "step": 15494
+    },
+    {
+      "epoch": 0.9985822001675582,
+      "grad_norm": 0.000392582037022992,
+      "learning_rate": 2.363050483351235e-08,
+      "loss": 0.0,
+      "step": 15495
+    },
+    {
+      "epoch": 0.9986466456144873,
+      "grad_norm": 0.0009271351805534993,
+      "learning_rate": 2.2914428929466525e-08,
+      "loss": 0.0,
+      "step": 15496
+    },
+    {
+      "epoch": 0.9987110910614165,
+      "grad_norm": 0.00014936378110164797,
+      "learning_rate": 2.2198353025420694e-08,
+      "loss": 0.0,
+      "step": 15497
+    },
+    {
+      "epoch": 0.9987755365083457,
+      "grad_norm": 0.00043512063478467827,
+      "learning_rate": 2.1482277121374867e-08,
+      "loss": 0.0,
+      "step": 15498
+    },
+    {
+      "epoch": 0.9988399819552749,
+      "grad_norm": 0.09968455133142981,
+      "learning_rate": 2.0766201217329037e-08,
+      "loss": 0.0003,
+      "step": 15499
+    },
+    {
+      "epoch": 0.9989044274022041,
+      "grad_norm": 0.00020179967403577038,
+      "learning_rate": 2.005012531328321e-08,
+      "loss": 0.0,
+      "step": 15500
+    },
+    {
+      "epoch": 0.9989688728491332,
+      "grad_norm": 0.04932118147213376,
+      "learning_rate": 1.933404940923738e-08,
+      "loss": 0.0001,
+      "step": 15501
+    },
+    {
+      "epoch": 0.9990333182960623,
+      "grad_norm": 0.021431060225883328,
+      "learning_rate": 1.8617973505191552e-08,
+      "loss": 0.0001,
+      "step": 15502
+    },
+    {
+      "epoch": 0.9990977637429915,
+      "grad_norm": 2.3081529524238576e-05,
+      "learning_rate": 1.7901897601145722e-08,
+      "loss": 0.0,
+      "step": 15503
+    },
+    {
+      "epoch": 0.9991622091899207,
+      "grad_norm": 0.0006019946843029049,
+      "learning_rate": 1.7185821697099895e-08,
+      "loss": 0.0,
+      "step": 15504
+    },
+    {
+      "epoch": 0.9992266546368499,
+      "grad_norm": 0.00010010075275028686,
+      "learning_rate": 1.6469745793054065e-08,
+      "loss": 0.0,
+      "step": 15505
+    },
+    {
+      "epoch": 0.9992911000837791,
+      "grad_norm": 0.004928500843443072,
+      "learning_rate": 1.5753669889008238e-08,
+      "loss": 0.0,
+      "step": 15506
+    },
+    {
+      "epoch": 0.9993555455307083,
+      "grad_norm": 0.0028870445087702625,
+      "learning_rate": 1.5037593984962407e-08,
+      "loss": 0.0,
+      "step": 15507
+    },
+    {
+      "epoch": 0.9994199909776375,
+      "grad_norm": 1.111107454028999,
+      "learning_rate": 1.4321518080916579e-08,
+      "loss": 0.0037,
+      "step": 15508
+    },
+    {
+      "epoch": 0.9994844364245666,
+      "grad_norm": 0.006669760087007271,
+      "learning_rate": 1.3605442176870748e-08,
+      "loss": 0.0001,
+      "step": 15509
+    },
+    {
+      "epoch": 0.9995488818714958,
+      "grad_norm": 0.007184647860438429,
+      "learning_rate": 1.288936627282492e-08,
+      "loss": 0.0,
+      "step": 15510
+    },
+    {
+      "epoch": 0.999613327318425,
+      "grad_norm": 0.000263493157723586,
+      "learning_rate": 1.2173290368779091e-08,
+      "loss": 0.0,
+      "step": 15511
+    },
+    {
+      "epoch": 0.9996777727653541,
+      "grad_norm": 0.019085772562670834,
+      "learning_rate": 1.1457214464733262e-08,
+      "loss": 0.0,
+      "step": 15512
+    },
+    {
+      "epoch": 0.9997422182122833,
+      "grad_norm": 0.12087096979193869,
+      "learning_rate": 1.0741138560687434e-08,
+      "loss": 0.0003,
+      "step": 15513
+    },
+    {
+      "epoch": 0.9998066636592124,
+      "grad_norm": 0.04529828473437381,
+      "learning_rate": 1.0025062656641605e-08,
+      "loss": 0.0,
+      "step": 15514
+    },
+    {
+      "epoch": 0.9998711091061416,
+      "grad_norm": 8.925880633125442e-05,
+      "learning_rate": 9.308986752595776e-09,
+      "loss": 0.0,
+      "step": 15515
+    },
+    {
+      "epoch": 0.9999355545530708,
+      "grad_norm": 0.0003679149024459179,
+      "learning_rate": 8.592910848549948e-09,
+      "loss": 0.0,
+      "step": 15516
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.2631523948960553,
+      "learning_rate": 7.876834944504119e-09,
+      "loss": 0.0003,
+      "step": 15517
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 15517,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "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": 0.0,
+  "train_batch_size": 12,
+  "trial_name": null,
+  "trial_params": null
+}