diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,35033 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 20.0,
+  "eval_steps": 500,
+  "global_step": 5000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.004,
+      "grad_norm": 4.403661251068115,
+      "learning_rate": 5e-06,
+      "loss": 3.5443,
+      "step": 1
+    },
+    {
+      "epoch": 0.008,
+      "grad_norm": 3.535775661468506,
+      "learning_rate": 1e-05,
+      "loss": 1.8447,
+      "step": 2
+    },
+    {
+      "epoch": 0.012,
+      "grad_norm": 2.976018190383911,
+      "learning_rate": 1.5e-05,
+      "loss": 1.7894,
+      "step": 3
+    },
+    {
+      "epoch": 0.016,
+      "grad_norm": 3.0810327529907227,
+      "learning_rate": 2e-05,
+      "loss": 2.9688,
+      "step": 4
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.9574649333953857,
+      "learning_rate": 2.5e-05,
+      "loss": 1.6972,
+      "step": 5
+    },
+    {
+      "epoch": 0.024,
+      "grad_norm": 2.9108431339263916,
+      "learning_rate": 3e-05,
+      "loss": 1.6871,
+      "step": 6
+    },
+    {
+      "epoch": 0.028,
+      "grad_norm": 2.5679922103881836,
+      "learning_rate": 3.5e-05,
+      "loss": 1.6313,
+      "step": 7
+    },
+    {
+      "epoch": 0.032,
+      "grad_norm": 2.8970062732696533,
+      "learning_rate": 4e-05,
+      "loss": 1.5489,
+      "step": 8
+    },
+    {
+      "epoch": 0.036,
+      "grad_norm": 2.5108256340026855,
+      "learning_rate": 4.5e-05,
+      "loss": 1.3924,
+      "step": 9
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.4062409400939941,
+      "learning_rate": 5e-05,
+      "loss": 1.4039,
+      "step": 10
+    },
+    {
+      "epoch": 0.044,
+      "grad_norm": 1.4537838697433472,
+      "learning_rate": 4.9989979959919844e-05,
+      "loss": 1.3405,
+      "step": 11
+    },
+    {
+      "epoch": 0.048,
+      "grad_norm": 1.1395865678787231,
+      "learning_rate": 4.997995991983968e-05,
+      "loss": 1.2754,
+      "step": 12
+    },
+    {
+      "epoch": 0.052,
+      "grad_norm": 1.3618974685668945,
+      "learning_rate": 4.996993987975952e-05,
+      "loss": 1.2058,
+      "step": 13
+    },
+    {
+      "epoch": 0.056,
+      "grad_norm": 1.0007574558258057,
+      "learning_rate": 4.995991983967936e-05,
+      "loss": 1.1149,
+      "step": 14
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0973924398422241,
+      "learning_rate": 4.99498997995992e-05,
+      "loss": 1.0393,
+      "step": 15
+    },
+    {
+      "epoch": 0.064,
+      "grad_norm": 0.9514708518981934,
+      "learning_rate": 4.993987975951904e-05,
+      "loss": 1.1058,
+      "step": 16
+    },
+    {
+      "epoch": 0.068,
+      "grad_norm": 0.8538547158241272,
+      "learning_rate": 4.9929859719438885e-05,
+      "loss": 1.0185,
+      "step": 17
+    },
+    {
+      "epoch": 0.072,
+      "grad_norm": 0.7519181966781616,
+      "learning_rate": 4.991983967935872e-05,
+      "loss": 0.9962,
+      "step": 18
+    },
+    {
+      "epoch": 0.076,
+      "grad_norm": 0.9052465558052063,
+      "learning_rate": 4.990981963927856e-05,
+      "loss": 1.034,
+      "step": 19
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9109005928039551,
+      "learning_rate": 4.98997995991984e-05,
+      "loss": 1.0563,
+      "step": 20
+    },
+    {
+      "epoch": 0.084,
+      "grad_norm": 1.0023589134216309,
+      "learning_rate": 4.9889779559118236e-05,
+      "loss": 0.8603,
+      "step": 21
+    },
+    {
+      "epoch": 0.088,
+      "grad_norm": 0.8098154664039612,
+      "learning_rate": 4.987975951903808e-05,
+      "loss": 0.9769,
+      "step": 22
+    },
+    {
+      "epoch": 0.092,
+      "grad_norm": 0.7375625967979431,
+      "learning_rate": 4.986973947895792e-05,
+      "loss": 0.8612,
+      "step": 23
+    },
+    {
+      "epoch": 0.096,
+      "grad_norm": 0.6169605851173401,
+      "learning_rate": 4.985971943887775e-05,
+      "loss": 0.8061,
+      "step": 24
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.5663151144981384,
+      "learning_rate": 4.98496993987976e-05,
+      "loss": 0.7382,
+      "step": 25
+    },
+    {
+      "epoch": 0.104,
+      "grad_norm": 0.5792577266693115,
+      "learning_rate": 4.983967935871744e-05,
+      "loss": 0.8212,
+      "step": 26
+    },
+    {
+      "epoch": 0.108,
+      "grad_norm": 0.5520902276039124,
+      "learning_rate": 4.982965931863728e-05,
+      "loss": 0.7566,
+      "step": 27
+    },
+    {
+      "epoch": 0.112,
+      "grad_norm": 0.6001718044281006,
+      "learning_rate": 4.981963927855712e-05,
+      "loss": 0.7855,
+      "step": 28
+    },
+    {
+      "epoch": 0.116,
+      "grad_norm": 0.9115943312644958,
+      "learning_rate": 4.980961923847696e-05,
+      "loss": 0.7621,
+      "step": 29
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.47501370310783386,
+      "learning_rate": 4.9799599198396794e-05,
+      "loss": 0.6687,
+      "step": 30
+    },
+    {
+      "epoch": 0.124,
+      "grad_norm": 0.48663514852523804,
+      "learning_rate": 4.9789579158316635e-05,
+      "loss": 0.7291,
+      "step": 31
+    },
+    {
+      "epoch": 0.128,
+      "grad_norm": 0.49639806151390076,
+      "learning_rate": 4.977955911823648e-05,
+      "loss": 0.7049,
+      "step": 32
+    },
+    {
+      "epoch": 0.132,
+      "grad_norm": 0.49986907839775085,
+      "learning_rate": 4.976953907815631e-05,
+      "loss": 0.7347,
+      "step": 33
+    },
+    {
+      "epoch": 0.136,
+      "grad_norm": 0.5302708745002747,
+      "learning_rate": 4.975951903807615e-05,
+      "loss": 0.771,
+      "step": 34
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.5710265040397644,
+      "learning_rate": 4.9749498997995994e-05,
+      "loss": 0.6717,
+      "step": 35
+    },
+    {
+      "epoch": 0.144,
+      "grad_norm": 0.4576522707939148,
+      "learning_rate": 4.9739478957915835e-05,
+      "loss": 0.6038,
+      "step": 36
+    },
+    {
+      "epoch": 0.148,
+      "grad_norm": 0.509702205657959,
+      "learning_rate": 4.9729458917835676e-05,
+      "loss": 0.681,
+      "step": 37
+    },
+    {
+      "epoch": 0.152,
+      "grad_norm": 0.5716866254806519,
+      "learning_rate": 4.971943887775551e-05,
+      "loss": 0.7039,
+      "step": 38
+    },
+    {
+      "epoch": 0.156,
+      "grad_norm": 0.4261268079280853,
+      "learning_rate": 4.970941883767535e-05,
+      "loss": 0.6841,
+      "step": 39
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.5015498399734497,
+      "learning_rate": 4.9699398797595193e-05,
+      "loss": 0.6062,
+      "step": 40
+    },
+    {
+      "epoch": 0.164,
+      "grad_norm": 0.4798198640346527,
+      "learning_rate": 4.9689378757515035e-05,
+      "loss": 0.6174,
+      "step": 41
+    },
+    {
+      "epoch": 0.168,
+      "grad_norm": 0.47841379046440125,
+      "learning_rate": 4.967935871743487e-05,
+      "loss": 0.71,
+      "step": 42
+    },
+    {
+      "epoch": 0.172,
+      "grad_norm": 0.38819581270217896,
+      "learning_rate": 4.966933867735471e-05,
+      "loss": 0.6236,
+      "step": 43
+    },
+    {
+      "epoch": 0.176,
+      "grad_norm": 0.3802018165588379,
+      "learning_rate": 4.965931863727455e-05,
+      "loss": 0.5662,
+      "step": 44
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.4659491777420044,
+      "learning_rate": 4.964929859719439e-05,
+      "loss": 0.6553,
+      "step": 45
+    },
+    {
+      "epoch": 0.184,
+      "grad_norm": 0.4309292733669281,
+      "learning_rate": 4.9639278557114234e-05,
+      "loss": 0.6317,
+      "step": 46
+    },
+    {
+      "epoch": 0.188,
+      "grad_norm": 0.36458510160446167,
+      "learning_rate": 4.962925851703407e-05,
+      "loss": 0.5853,
+      "step": 47
+    },
+    {
+      "epoch": 0.192,
+      "grad_norm": 0.3789045214653015,
+      "learning_rate": 4.961923847695391e-05,
+      "loss": 0.6565,
+      "step": 48
+    },
+    {
+      "epoch": 0.196,
+      "grad_norm": 0.43124228715896606,
+      "learning_rate": 4.960921843687375e-05,
+      "loss": 0.6484,
+      "step": 49
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": Infinity,
+      "learning_rate": 4.960921843687375e-05,
+      "loss": 3.1127,
+      "step": 50
+    },
+    {
+      "epoch": 0.204,
+      "grad_norm": 0.40404313802719116,
+      "learning_rate": 4.9599198396793586e-05,
+      "loss": 0.6541,
+      "step": 51
+    },
+    {
+      "epoch": 0.208,
+      "grad_norm": 0.329449862241745,
+      "learning_rate": 4.958917835671343e-05,
+      "loss": 0.5814,
+      "step": 52
+    },
+    {
+      "epoch": 0.212,
+      "grad_norm": 0.3724288046360016,
+      "learning_rate": 4.957915831663327e-05,
+      "loss": 0.6095,
+      "step": 53
+    },
+    {
+      "epoch": 0.216,
+      "grad_norm": 0.3768303096294403,
+      "learning_rate": 4.956913827655311e-05,
+      "loss": 0.6467,
+      "step": 54
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.3847925066947937,
+      "learning_rate": 4.9559118236472944e-05,
+      "loss": 0.6584,
+      "step": 55
+    },
+    {
+      "epoch": 0.224,
+      "grad_norm": 0.38740023970603943,
+      "learning_rate": 4.954909819639279e-05,
+      "loss": 0.6108,
+      "step": 56
+    },
+    {
+      "epoch": 0.228,
+      "grad_norm": 0.4133378863334656,
+      "learning_rate": 4.953907815631263e-05,
+      "loss": 0.569,
+      "step": 57
+    },
+    {
+      "epoch": 0.232,
+      "grad_norm": 0.3934808373451233,
+      "learning_rate": 4.952905811623247e-05,
+      "loss": 0.5979,
+      "step": 58
+    },
+    {
+      "epoch": 0.236,
+      "grad_norm": 0.39415234327316284,
+      "learning_rate": 4.951903807615231e-05,
+      "loss": 0.5732,
+      "step": 59
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.41425755620002747,
+      "learning_rate": 4.9509018036072144e-05,
+      "loss": 0.6131,
+      "step": 60
+    },
+    {
+      "epoch": 0.244,
+      "grad_norm": 0.3662201166152954,
+      "learning_rate": 4.9498997995991985e-05,
+      "loss": 0.582,
+      "step": 61
+    },
+    {
+      "epoch": 0.248,
+      "grad_norm": 0.6343560218811035,
+      "learning_rate": 4.9488977955911826e-05,
+      "loss": 0.7075,
+      "step": 62
+    },
+    {
+      "epoch": 0.252,
+      "grad_norm": 0.362642765045166,
+      "learning_rate": 4.947895791583166e-05,
+      "loss": 0.5534,
+      "step": 63
+    },
+    {
+      "epoch": 0.256,
+      "grad_norm": 0.3604782819747925,
+      "learning_rate": 4.94689378757515e-05,
+      "loss": 0.5873,
+      "step": 64
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.3327144384384155,
+      "learning_rate": 4.9458917835671344e-05,
+      "loss": 0.5309,
+      "step": 65
+    },
+    {
+      "epoch": 0.264,
+      "grad_norm": 0.35557445883750916,
+      "learning_rate": 4.9448897795591185e-05,
+      "loss": 0.5307,
+      "step": 66
+    },
+    {
+      "epoch": 0.268,
+      "grad_norm": 0.5837879180908203,
+      "learning_rate": 4.9438877755511026e-05,
+      "loss": 0.6416,
+      "step": 67
+    },
+    {
+      "epoch": 0.272,
+      "grad_norm": 0.7987128496170044,
+      "learning_rate": 4.942885771543087e-05,
+      "loss": 0.5083,
+      "step": 68
+    },
+    {
+      "epoch": 0.276,
+      "grad_norm": 0.3976365923881531,
+      "learning_rate": 4.94188376753507e-05,
+      "loss": 0.6014,
+      "step": 69
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.3430960774421692,
+      "learning_rate": 4.940881763527054e-05,
+      "loss": 0.5841,
+      "step": 70
+    },
+    {
+      "epoch": 0.284,
+      "grad_norm": 0.3691798746585846,
+      "learning_rate": 4.9398797595190384e-05,
+      "loss": 0.5346,
+      "step": 71
+    },
+    {
+      "epoch": 0.288,
+      "grad_norm": 0.3781915307044983,
+      "learning_rate": 4.938877755511022e-05,
+      "loss": 0.5438,
+      "step": 72
+    },
+    {
+      "epoch": 0.292,
+      "grad_norm": 0.43587324023246765,
+      "learning_rate": 4.937875751503006e-05,
+      "loss": 0.5575,
+      "step": 73
+    },
+    {
+      "epoch": 0.296,
+      "grad_norm": 0.5425245761871338,
+      "learning_rate": 4.93687374749499e-05,
+      "loss": 0.5755,
+      "step": 74
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.3869353234767914,
+      "learning_rate": 4.935871743486974e-05,
+      "loss": 0.5267,
+      "step": 75
+    },
+    {
+      "epoch": 0.304,
+      "grad_norm": 1.0698070526123047,
+      "learning_rate": 4.9348697394789584e-05,
+      "loss": 0.5967,
+      "step": 76
+    },
+    {
+      "epoch": 0.308,
+      "grad_norm": 0.3409326672554016,
+      "learning_rate": 4.9338677354709425e-05,
+      "loss": 0.5764,
+      "step": 77
+    },
+    {
+      "epoch": 0.312,
+      "grad_norm": 0.42487403750419617,
+      "learning_rate": 4.932865731462926e-05,
+      "loss": 0.5788,
+      "step": 78
+    },
+    {
+      "epoch": 0.316,
+      "grad_norm": 0.355347216129303,
+      "learning_rate": 4.93186372745491e-05,
+      "loss": 0.5232,
+      "step": 79
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.3655643165111542,
+      "learning_rate": 4.930861723446894e-05,
+      "loss": 0.5693,
+      "step": 80
+    },
+    {
+      "epoch": 0.324,
+      "grad_norm": 0.40408679842948914,
+      "learning_rate": 4.929859719438878e-05,
+      "loss": 0.5206,
+      "step": 81
+    },
+    {
+      "epoch": 0.328,
+      "grad_norm": 0.358632355928421,
+      "learning_rate": 4.928857715430862e-05,
+      "loss": 0.5344,
+      "step": 82
+    },
+    {
+      "epoch": 0.332,
+      "grad_norm": 0.3510683476924896,
+      "learning_rate": 4.927855711422846e-05,
+      "loss": 0.5673,
+      "step": 83
+    },
+    {
+      "epoch": 0.336,
+      "grad_norm": 0.3240058422088623,
+      "learning_rate": 4.9268537074148294e-05,
+      "loss": 0.5064,
+      "step": 84
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.39716836810112,
+      "learning_rate": 4.925851703406814e-05,
+      "loss": 0.4974,
+      "step": 85
+    },
+    {
+      "epoch": 0.344,
+      "grad_norm": 0.3803369998931885,
+      "learning_rate": 4.9248496993987983e-05,
+      "loss": 0.5433,
+      "step": 86
+    },
+    {
+      "epoch": 0.348,
+      "grad_norm": 0.345559298992157,
+      "learning_rate": 4.923847695390782e-05,
+      "loss": 0.5194,
+      "step": 87
+    },
+    {
+      "epoch": 0.352,
+      "grad_norm": 0.3797976076602936,
+      "learning_rate": 4.922845691382766e-05,
+      "loss": 0.5668,
+      "step": 88
+    },
+    {
+      "epoch": 0.356,
+      "grad_norm": 0.4122374653816223,
+      "learning_rate": 4.92184368737475e-05,
+      "loss": 0.5709,
+      "step": 89
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.3925560414791107,
+      "learning_rate": 4.9208416833667335e-05,
+      "loss": 0.5528,
+      "step": 90
+    },
+    {
+      "epoch": 0.364,
+      "grad_norm": 0.35586780309677124,
+      "learning_rate": 4.9198396793587176e-05,
+      "loss": 0.5123,
+      "step": 91
+    },
+    {
+      "epoch": 0.368,
+      "grad_norm": 0.5167198181152344,
+      "learning_rate": 4.918837675350702e-05,
+      "loss": 0.5667,
+      "step": 92
+    },
+    {
+      "epoch": 0.372,
+      "grad_norm": 0.6427658200263977,
+      "learning_rate": 4.917835671342685e-05,
+      "loss": 0.5046,
+      "step": 93
+    },
+    {
+      "epoch": 0.376,
+      "grad_norm": 0.3925032615661621,
+      "learning_rate": 4.916833667334669e-05,
+      "loss": 0.5729,
+      "step": 94
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.4006875157356262,
+      "learning_rate": 4.9158316633266535e-05,
+      "loss": 0.5932,
+      "step": 95
+    },
+    {
+      "epoch": 0.384,
+      "grad_norm": 0.3980340361595154,
+      "learning_rate": 4.9148296593186376e-05,
+      "loss": 0.5277,
+      "step": 96
+    },
+    {
+      "epoch": 0.388,
+      "grad_norm": 0.35809895396232605,
+      "learning_rate": 4.913827655310622e-05,
+      "loss": 0.5806,
+      "step": 97
+    },
+    {
+      "epoch": 0.392,
+      "grad_norm": 0.3831680417060852,
+      "learning_rate": 4.912825651302606e-05,
+      "loss": 0.5285,
+      "step": 98
+    },
+    {
+      "epoch": 0.396,
+      "grad_norm": 8.12513256072998,
+      "learning_rate": 4.911823647294589e-05,
+      "loss": 1.9624,
+      "step": 99
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.38112860918045044,
+      "learning_rate": 4.9108216432865734e-05,
+      "loss": 0.5226,
+      "step": 100
+    },
+    {
+      "epoch": 0.404,
+      "grad_norm": 0.37594351172447205,
+      "learning_rate": 4.9098196392785576e-05,
+      "loss": 0.521,
+      "step": 101
+    },
+    {
+      "epoch": 0.408,
+      "grad_norm": 0.35807371139526367,
+      "learning_rate": 4.908817635270541e-05,
+      "loss": 0.574,
+      "step": 102
+    },
+    {
+      "epoch": 0.412,
+      "grad_norm": 0.42019104957580566,
+      "learning_rate": 4.907815631262525e-05,
+      "loss": 0.5854,
+      "step": 103
+    },
+    {
+      "epoch": 0.416,
+      "grad_norm": 4.040385723114014,
+      "learning_rate": 4.906813627254509e-05,
+      "loss": 2.0486,
+      "step": 104
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.40117359161376953,
+      "learning_rate": 4.9058116232464934e-05,
+      "loss": 0.4933,
+      "step": 105
+    },
+    {
+      "epoch": 0.424,
+      "grad_norm": 0.3785370886325836,
+      "learning_rate": 4.9048096192384775e-05,
+      "loss": 0.5409,
+      "step": 106
+    },
+    {
+      "epoch": 0.428,
+      "grad_norm": 0.5202479958534241,
+      "learning_rate": 4.903807615230461e-05,
+      "loss": 0.6319,
+      "step": 107
+    },
+    {
+      "epoch": 0.432,
+      "grad_norm": 0.3808040916919708,
+      "learning_rate": 4.902805611222445e-05,
+      "loss": 0.5222,
+      "step": 108
+    },
+    {
+      "epoch": 0.436,
+      "grad_norm": 0.33765068650245667,
+      "learning_rate": 4.901803607214429e-05,
+      "loss": 0.5191,
+      "step": 109
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.37931010127067566,
+      "learning_rate": 4.9008016032064134e-05,
+      "loss": 0.509,
+      "step": 110
+    },
+    {
+      "epoch": 0.444,
+      "grad_norm": 0.46715351939201355,
+      "learning_rate": 4.899799599198397e-05,
+      "loss": 0.5343,
+      "step": 111
+    },
+    {
+      "epoch": 0.448,
+      "grad_norm": 0.372823029756546,
+      "learning_rate": 4.898797595190381e-05,
+      "loss": 0.542,
+      "step": 112
+    },
+    {
+      "epoch": 0.452,
+      "grad_norm": 0.3744858205318451,
+      "learning_rate": 4.897795591182365e-05,
+      "loss": 0.4599,
+      "step": 113
+    },
+    {
+      "epoch": 0.456,
+      "grad_norm": 0.4025891423225403,
+      "learning_rate": 4.8967935871743485e-05,
+      "loss": 0.5455,
+      "step": 114
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.33246585726737976,
+      "learning_rate": 4.895791583166333e-05,
+      "loss": 0.5026,
+      "step": 115
+    },
+    {
+      "epoch": 0.464,
+      "grad_norm": 0.4333687126636505,
+      "learning_rate": 4.894789579158317e-05,
+      "loss": 0.5372,
+      "step": 116
+    },
+    {
+      "epoch": 0.468,
+      "grad_norm": 0.36540961265563965,
+      "learning_rate": 4.893787575150301e-05,
+      "loss": 0.4877,
+      "step": 117
+    },
+    {
+      "epoch": 0.472,
+      "grad_norm": 0.39382699131965637,
+      "learning_rate": 4.892785571142285e-05,
+      "loss": 0.4705,
+      "step": 118
+    },
+    {
+      "epoch": 0.476,
+      "grad_norm": 0.37407657504081726,
+      "learning_rate": 4.8917835671342685e-05,
+      "loss": 0.5003,
+      "step": 119
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.36774665117263794,
+      "learning_rate": 4.8907815631262526e-05,
+      "loss": 0.5304,
+      "step": 120
+    },
+    {
+      "epoch": 0.484,
+      "grad_norm": 0.37522393465042114,
+      "learning_rate": 4.889779559118237e-05,
+      "loss": 0.5437,
+      "step": 121
+    },
+    {
+      "epoch": 0.488,
+      "grad_norm": 0.5063856244087219,
+      "learning_rate": 4.88877755511022e-05,
+      "loss": 0.558,
+      "step": 122
+    },
+    {
+      "epoch": 0.492,
+      "grad_norm": 5.0222487449646,
+      "learning_rate": 4.887775551102204e-05,
+      "loss": 1.7772,
+      "step": 123
+    },
+    {
+      "epoch": 0.496,
+      "grad_norm": 0.4300785958766937,
+      "learning_rate": 4.886773547094189e-05,
+      "loss": 0.5218,
+      "step": 124
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.5471500754356384,
+      "learning_rate": 4.8857715430861726e-05,
+      "loss": 0.5194,
+      "step": 125
+    },
+    {
+      "epoch": 0.504,
+      "grad_norm": 0.40047940611839294,
+      "learning_rate": 4.884769539078157e-05,
+      "loss": 0.47,
+      "step": 126
+    },
+    {
+      "epoch": 0.508,
+      "grad_norm": 0.39165449142456055,
+      "learning_rate": 4.883767535070141e-05,
+      "loss": 0.5365,
+      "step": 127
+    },
+    {
+      "epoch": 0.512,
+      "grad_norm": 0.4630558490753174,
+      "learning_rate": 4.882765531062124e-05,
+      "loss": 0.5676,
+      "step": 128
+    },
+    {
+      "epoch": 0.516,
+      "grad_norm": 0.3909466862678528,
+      "learning_rate": 4.8817635270541084e-05,
+      "loss": 0.565,
+      "step": 129
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.49433305859565735,
+      "learning_rate": 4.8807615230460925e-05,
+      "loss": 0.4734,
+      "step": 130
+    },
+    {
+      "epoch": 0.524,
+      "grad_norm": 0.40937337279319763,
+      "learning_rate": 4.879759519038076e-05,
+      "loss": 0.4959,
+      "step": 131
+    },
+    {
+      "epoch": 0.528,
+      "grad_norm": 0.4436751902103424,
+      "learning_rate": 4.87875751503006e-05,
+      "loss": 0.5009,
+      "step": 132
+    },
+    {
+      "epoch": 0.532,
+      "grad_norm": 0.3952745199203491,
+      "learning_rate": 4.877755511022044e-05,
+      "loss": 0.5088,
+      "step": 133
+    },
+    {
+      "epoch": 0.536,
+      "grad_norm": 0.4065265357494354,
+      "learning_rate": 4.8767535070140284e-05,
+      "loss": 0.5368,
+      "step": 134
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.351837694644928,
+      "learning_rate": 4.8757515030060125e-05,
+      "loss": 0.4719,
+      "step": 135
+    },
+    {
+      "epoch": 0.544,
+      "grad_norm": 0.38240760564804077,
+      "learning_rate": 4.8747494989979966e-05,
+      "loss": 0.5045,
+      "step": 136
+    },
+    {
+      "epoch": 0.548,
+      "grad_norm": 1.0215145349502563,
+      "learning_rate": 4.87374749498998e-05,
+      "loss": 0.4706,
+      "step": 137
+    },
+    {
+      "epoch": 0.552,
+      "grad_norm": 0.44947350025177,
+      "learning_rate": 4.872745490981964e-05,
+      "loss": 0.4955,
+      "step": 138
+    },
+    {
+      "epoch": 0.556,
+      "grad_norm": 0.4560631811618805,
+      "learning_rate": 4.871743486973948e-05,
+      "loss": 0.4635,
+      "step": 139
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.44674062728881836,
+      "learning_rate": 4.870741482965932e-05,
+      "loss": 0.5357,
+      "step": 140
+    },
+    {
+      "epoch": 0.564,
+      "grad_norm": 0.3947732150554657,
+      "learning_rate": 4.869739478957916e-05,
+      "loss": 0.5075,
+      "step": 141
+    },
+    {
+      "epoch": 0.568,
+      "grad_norm": 0.42195793986320496,
+      "learning_rate": 4.8687374749499e-05,
+      "loss": 0.5268,
+      "step": 142
+    },
+    {
+      "epoch": 0.572,
+      "grad_norm": 0.350239634513855,
+      "learning_rate": 4.8677354709418835e-05,
+      "loss": 0.4759,
+      "step": 143
+    },
+    {
+      "epoch": 0.576,
+      "grad_norm": 0.4829655885696411,
+      "learning_rate": 4.866733466933868e-05,
+      "loss": 0.4905,
+      "step": 144
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.37556272745132446,
+      "learning_rate": 4.8657314629258524e-05,
+      "loss": 0.4722,
+      "step": 145
+    },
+    {
+      "epoch": 0.584,
+      "grad_norm": 0.3916873335838318,
+      "learning_rate": 4.864729458917836e-05,
+      "loss": 0.4706,
+      "step": 146
+    },
+    {
+      "epoch": 0.588,
+      "grad_norm": 0.4552304148674011,
+      "learning_rate": 4.86372745490982e-05,
+      "loss": 0.5148,
+      "step": 147
+    },
+    {
+      "epoch": 0.592,
+      "grad_norm": 0.43419304490089417,
+      "learning_rate": 4.862725450901804e-05,
+      "loss": 0.5318,
+      "step": 148
+    },
+    {
+      "epoch": 0.596,
+      "grad_norm": 0.4283067584037781,
+      "learning_rate": 4.8617234468937876e-05,
+      "loss": 0.5103,
+      "step": 149
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.39657995104789734,
+      "learning_rate": 4.860721442885772e-05,
+      "loss": 0.4643,
+      "step": 150
+    },
+    {
+      "epoch": 0.604,
+      "grad_norm": 0.43963828682899475,
+      "learning_rate": 4.859719438877756e-05,
+      "loss": 0.5201,
+      "step": 151
+    },
+    {
+      "epoch": 0.608,
+      "grad_norm": 0.4126660227775574,
+      "learning_rate": 4.858717434869739e-05,
+      "loss": 0.5069,
+      "step": 152
+    },
+    {
+      "epoch": 0.612,
+      "grad_norm": 0.42871612310409546,
+      "learning_rate": 4.8577154308617234e-05,
+      "loss": 0.527,
+      "step": 153
+    },
+    {
+      "epoch": 0.616,
+      "grad_norm": 0.44571855664253235,
+      "learning_rate": 4.856713426853708e-05,
+      "loss": 0.5771,
+      "step": 154
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3940712809562683,
+      "learning_rate": 4.855711422845692e-05,
+      "loss": 0.5059,
+      "step": 155
+    },
+    {
+      "epoch": 0.624,
+      "grad_norm": 0.39730727672576904,
+      "learning_rate": 4.854709418837676e-05,
+      "loss": 0.4702,
+      "step": 156
+    },
+    {
+      "epoch": 0.628,
+      "grad_norm": 0.4266124367713928,
+      "learning_rate": 4.85370741482966e-05,
+      "loss": 0.4936,
+      "step": 157
+    },
+    {
+      "epoch": 0.632,
+      "grad_norm": 0.49682965874671936,
+      "learning_rate": 4.8527054108216434e-05,
+      "loss": 0.5417,
+      "step": 158
+    },
+    {
+      "epoch": 0.636,
+      "grad_norm": 0.43528032302856445,
+      "learning_rate": 4.8517034068136275e-05,
+      "loss": 0.5538,
+      "step": 159
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.4789420962333679,
+      "learning_rate": 4.8507014028056116e-05,
+      "loss": 0.4866,
+      "step": 160
+    },
+    {
+      "epoch": 0.644,
+      "grad_norm": 0.4257548153400421,
+      "learning_rate": 4.849699398797595e-05,
+      "loss": 0.4858,
+      "step": 161
+    },
+    {
+      "epoch": 0.648,
+      "grad_norm": 0.38716921210289,
+      "learning_rate": 4.848697394789579e-05,
+      "loss": 0.4976,
+      "step": 162
+    },
+    {
+      "epoch": 0.652,
+      "grad_norm": 10.990571975708008,
+      "learning_rate": 4.8476953907815633e-05,
+      "loss": 2.1143,
+      "step": 163
+    },
+    {
+      "epoch": 0.656,
+      "grad_norm": 0.42327937483787537,
+      "learning_rate": 4.8466933867735475e-05,
+      "loss": 0.5064,
+      "step": 164
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.35529690980911255,
+      "learning_rate": 4.8456913827655316e-05,
+      "loss": 0.4584,
+      "step": 165
+    },
+    {
+      "epoch": 0.664,
+      "grad_norm": 0.39093905687332153,
+      "learning_rate": 4.844689378757515e-05,
+      "loss": 0.4638,
+      "step": 166
+    },
+    {
+      "epoch": 0.668,
+      "grad_norm": 0.4272507429122925,
+      "learning_rate": 4.843687374749499e-05,
+      "loss": 0.5097,
+      "step": 167
+    },
+    {
+      "epoch": 0.672,
+      "grad_norm": 0.4245583117008209,
+      "learning_rate": 4.842685370741483e-05,
+      "loss": 0.5568,
+      "step": 168
+    },
+    {
+      "epoch": 0.676,
+      "grad_norm": 3.0908520221710205,
+      "learning_rate": 4.8416833667334674e-05,
+      "loss": 1.8422,
+      "step": 169
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.4836314916610718,
+      "learning_rate": 4.840681362725451e-05,
+      "loss": 0.5187,
+      "step": 170
+    },
+    {
+      "epoch": 0.684,
+      "grad_norm": 0.4002329409122467,
+      "learning_rate": 4.839679358717435e-05,
+      "loss": 0.4804,
+      "step": 171
+    },
+    {
+      "epoch": 0.688,
+      "grad_norm": 2.849741220474243,
+      "learning_rate": 4.838677354709419e-05,
+      "loss": 1.5929,
+      "step": 172
+    },
+    {
+      "epoch": 0.692,
+      "grad_norm": 0.3997988700866699,
+      "learning_rate": 4.8376753507014026e-05,
+      "loss": 0.4551,
+      "step": 173
+    },
+    {
+      "epoch": 0.696,
+      "grad_norm": 0.4090086817741394,
+      "learning_rate": 4.8366733466933874e-05,
+      "loss": 0.5154,
+      "step": 174
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.4361508786678314,
+      "learning_rate": 4.835671342685371e-05,
+      "loss": 0.4584,
+      "step": 175
+    },
+    {
+      "epoch": 0.704,
+      "grad_norm": 0.41879937052726746,
+      "learning_rate": 4.834669338677355e-05,
+      "loss": 0.5122,
+      "step": 176
+    },
+    {
+      "epoch": 0.708,
+      "grad_norm": 0.43722084164619446,
+      "learning_rate": 4.833667334669339e-05,
+      "loss": 0.503,
+      "step": 177
+    },
+    {
+      "epoch": 0.712,
+      "grad_norm": 0.3997744917869568,
+      "learning_rate": 4.8326653306613226e-05,
+      "loss": 0.5121,
+      "step": 178
+    },
+    {
+      "epoch": 0.716,
+      "grad_norm": 0.4667453169822693,
+      "learning_rate": 4.831663326653307e-05,
+      "loss": 0.4888,
+      "step": 179
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.47008612751960754,
+      "learning_rate": 4.830661322645291e-05,
+      "loss": 0.5683,
+      "step": 180
+    },
+    {
+      "epoch": 0.724,
+      "grad_norm": 0.4221080541610718,
+      "learning_rate": 4.829659318637275e-05,
+      "loss": 0.4872,
+      "step": 181
+    },
+    {
+      "epoch": 0.728,
+      "grad_norm": 0.4071381688117981,
+      "learning_rate": 4.8286573146292584e-05,
+      "loss": 0.4833,
+      "step": 182
+    },
+    {
+      "epoch": 0.732,
+      "grad_norm": 0.427738219499588,
+      "learning_rate": 4.827655310621243e-05,
+      "loss": 0.4781,
+      "step": 183
+    },
+    {
+      "epoch": 0.736,
+      "grad_norm": 0.3943522274494171,
+      "learning_rate": 4.8266533066132266e-05,
+      "loss": 0.5011,
+      "step": 184
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.4425494968891144,
+      "learning_rate": 4.825651302605211e-05,
+      "loss": 0.461,
+      "step": 185
+    },
+    {
+      "epoch": 0.744,
+      "grad_norm": 0.4339994192123413,
+      "learning_rate": 4.824649298597195e-05,
+      "loss": 0.4756,
+      "step": 186
+    },
+    {
+      "epoch": 0.748,
+      "grad_norm": 0.38859328627586365,
+      "learning_rate": 4.8236472945891784e-05,
+      "loss": 0.451,
+      "step": 187
+    },
+    {
+      "epoch": 0.752,
+      "grad_norm": 0.4013572931289673,
+      "learning_rate": 4.8226452905811625e-05,
+      "loss": 0.5298,
+      "step": 188
+    },
+    {
+      "epoch": 0.756,
+      "grad_norm": 9.899824142456055,
+      "learning_rate": 4.8216432865731466e-05,
+      "loss": 1.542,
+      "step": 189
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.41241151094436646,
+      "learning_rate": 4.82064128256513e-05,
+      "loss": 0.5358,
+      "step": 190
+    },
+    {
+      "epoch": 0.764,
+      "grad_norm": 0.408150315284729,
+      "learning_rate": 4.819639278557114e-05,
+      "loss": 0.4647,
+      "step": 191
+    },
+    {
+      "epoch": 0.768,
+      "grad_norm": 0.40550801157951355,
+      "learning_rate": 4.818637274549098e-05,
+      "loss": 0.4881,
+      "step": 192
+    },
+    {
+      "epoch": 0.772,
+      "grad_norm": 0.35784292221069336,
+      "learning_rate": 4.8176352705410824e-05,
+      "loss": 0.4493,
+      "step": 193
+    },
+    {
+      "epoch": 0.776,
+      "grad_norm": 0.3900321424007416,
+      "learning_rate": 4.8166332665330666e-05,
+      "loss": 0.4015,
+      "step": 194
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.38830649852752686,
+      "learning_rate": 4.815631262525051e-05,
+      "loss": 0.4702,
+      "step": 195
+    },
+    {
+      "epoch": 0.784,
+      "grad_norm": 0.4276293218135834,
+      "learning_rate": 4.814629258517034e-05,
+      "loss": 0.4875,
+      "step": 196
+    },
+    {
+      "epoch": 0.788,
+      "grad_norm": 0.3857548236846924,
+      "learning_rate": 4.813627254509018e-05,
+      "loss": 0.4491,
+      "step": 197
+    },
+    {
+      "epoch": 0.792,
+      "grad_norm": 0.4131658375263214,
+      "learning_rate": 4.8126252505010024e-05,
+      "loss": 0.495,
+      "step": 198
+    },
+    {
+      "epoch": 0.796,
+      "grad_norm": 0.41396304965019226,
+      "learning_rate": 4.811623246492986e-05,
+      "loss": 0.5112,
+      "step": 199
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.4702424108982086,
+      "learning_rate": 4.81062124248497e-05,
+      "loss": 0.548,
+      "step": 200
+    },
+    {
+      "epoch": 0.804,
+      "grad_norm": 0.41151168942451477,
+      "learning_rate": 4.809619238476954e-05,
+      "loss": 0.4484,
+      "step": 201
+    },
+    {
+      "epoch": 0.808,
+      "grad_norm": 0.4377236068248749,
+      "learning_rate": 4.8086172344689376e-05,
+      "loss": 0.4626,
+      "step": 202
+    },
+    {
+      "epoch": 0.812,
+      "grad_norm": 0.47114676237106323,
+      "learning_rate": 4.8076152304609224e-05,
+      "loss": 0.505,
+      "step": 203
+    },
+    {
+      "epoch": 0.816,
+      "grad_norm": 0.44927552342414856,
+      "learning_rate": 4.8066132264529065e-05,
+      "loss": 0.4697,
+      "step": 204
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.4451066255569458,
+      "learning_rate": 4.80561122244489e-05,
+      "loss": 0.5453,
+      "step": 205
+    },
+    {
+      "epoch": 0.824,
+      "grad_norm": 0.7564288973808289,
+      "learning_rate": 4.804609218436874e-05,
+      "loss": 0.521,
+      "step": 206
+    },
+    {
+      "epoch": 0.828,
+      "grad_norm": 0.4182622730731964,
+      "learning_rate": 4.803607214428858e-05,
+      "loss": 0.4806,
+      "step": 207
+    },
+    {
+      "epoch": 0.832,
+      "grad_norm": 0.3740067780017853,
+      "learning_rate": 4.8026052104208417e-05,
+      "loss": 0.4551,
+      "step": 208
+    },
+    {
+      "epoch": 0.836,
+      "grad_norm": 0.4166138768196106,
+      "learning_rate": 4.801603206412826e-05,
+      "loss": 0.5156,
+      "step": 209
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.4035174548625946,
+      "learning_rate": 4.80060120240481e-05,
+      "loss": 0.4798,
+      "step": 210
+    },
+    {
+      "epoch": 0.844,
+      "grad_norm": 0.46257126331329346,
+      "learning_rate": 4.7995991983967934e-05,
+      "loss": 0.527,
+      "step": 211
+    },
+    {
+      "epoch": 0.848,
+      "grad_norm": 0.4034459590911865,
+      "learning_rate": 4.7985971943887775e-05,
+      "loss": 0.4595,
+      "step": 212
+    },
+    {
+      "epoch": 0.852,
+      "grad_norm": 0.37898552417755127,
+      "learning_rate": 4.797595190380762e-05,
+      "loss": 0.4495,
+      "step": 213
+    },
+    {
+      "epoch": 0.856,
+      "grad_norm": 0.4459609389305115,
+      "learning_rate": 4.796593186372746e-05,
+      "loss": 0.5279,
+      "step": 214
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.436012864112854,
+      "learning_rate": 4.79559118236473e-05,
+      "loss": 0.5159,
+      "step": 215
+    },
+    {
+      "epoch": 0.864,
+      "grad_norm": 0.42498722672462463,
+      "learning_rate": 4.794589178356714e-05,
+      "loss": 0.4828,
+      "step": 216
+    },
+    {
+      "epoch": 0.868,
+      "grad_norm": 0.4183887541294098,
+      "learning_rate": 4.7935871743486975e-05,
+      "loss": 0.4886,
+      "step": 217
+    },
+    {
+      "epoch": 0.872,
+      "grad_norm": 0.4127841889858246,
+      "learning_rate": 4.7925851703406816e-05,
+      "loss": 0.4601,
+      "step": 218
+    },
+    {
+      "epoch": 0.876,
+      "grad_norm": 0.42419275641441345,
+      "learning_rate": 4.791583166332666e-05,
+      "loss": 0.5311,
+      "step": 219
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.3534486293792725,
+      "learning_rate": 4.790581162324649e-05,
+      "loss": 1.7676,
+      "step": 220
+    },
+    {
+      "epoch": 0.884,
+      "grad_norm": 0.44655194878578186,
+      "learning_rate": 4.789579158316633e-05,
+      "loss": 0.4378,
+      "step": 221
+    },
+    {
+      "epoch": 0.888,
+      "grad_norm": 0.3902299702167511,
+      "learning_rate": 4.7885771543086174e-05,
+      "loss": 0.4638,
+      "step": 222
+    },
+    {
+      "epoch": 0.892,
+      "grad_norm": 0.700932502746582,
+      "learning_rate": 4.7875751503006016e-05,
+      "loss": 0.5176,
+      "step": 223
+    },
+    {
+      "epoch": 0.896,
+      "grad_norm": 0.4358803927898407,
+      "learning_rate": 4.786573146292586e-05,
+      "loss": 0.4604,
+      "step": 224
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.4845912754535675,
+      "learning_rate": 4.78557114228457e-05,
+      "loss": 0.5406,
+      "step": 225
+    },
+    {
+      "epoch": 0.904,
+      "grad_norm": 0.38343381881713867,
+      "learning_rate": 4.784569138276553e-05,
+      "loss": 0.4693,
+      "step": 226
+    },
+    {
+      "epoch": 0.908,
+      "grad_norm": 0.48064181208610535,
+      "learning_rate": 4.7835671342685374e-05,
+      "loss": 0.4616,
+      "step": 227
+    },
+    {
+      "epoch": 0.912,
+      "grad_norm": 0.5941019654273987,
+      "learning_rate": 4.7825651302605215e-05,
+      "loss": 0.6027,
+      "step": 228
+    },
+    {
+      "epoch": 0.916,
+      "grad_norm": 0.4344301223754883,
+      "learning_rate": 4.781563126252505e-05,
+      "loss": 0.4404,
+      "step": 229
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.4292423129081726,
+      "learning_rate": 4.780561122244489e-05,
+      "loss": 0.5116,
+      "step": 230
+    },
+    {
+      "epoch": 0.924,
+      "grad_norm": 6.355248928070068,
+      "learning_rate": 4.779559118236473e-05,
+      "loss": 1.3833,
+      "step": 231
+    },
+    {
+      "epoch": 0.928,
+      "grad_norm": 6.541921138763428,
+      "learning_rate": 4.778557114228457e-05,
+      "loss": 1.459,
+      "step": 232
+    },
+    {
+      "epoch": 0.932,
+      "grad_norm": 0.42501968145370483,
+      "learning_rate": 4.7775551102204415e-05,
+      "loss": 0.5074,
+      "step": 233
+    },
+    {
+      "epoch": 0.936,
+      "grad_norm": 0.4112103581428528,
+      "learning_rate": 4.776553106212425e-05,
+      "loss": 0.4512,
+      "step": 234
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.40065306425094604,
+      "learning_rate": 4.775551102204409e-05,
+      "loss": 0.4646,
+      "step": 235
+    },
+    {
+      "epoch": 0.944,
+      "grad_norm": 0.43869850039482117,
+      "learning_rate": 4.774549098196393e-05,
+      "loss": 0.5059,
+      "step": 236
+    },
+    {
+      "epoch": 0.948,
+      "grad_norm": 0.4382988214492798,
+      "learning_rate": 4.773547094188377e-05,
+      "loss": 0.4654,
+      "step": 237
+    },
+    {
+      "epoch": 0.952,
+      "grad_norm": 0.437419056892395,
+      "learning_rate": 4.772545090180361e-05,
+      "loss": 0.4978,
+      "step": 238
+    },
+    {
+      "epoch": 0.956,
+      "grad_norm": 0.3500848412513733,
+      "learning_rate": 4.771543086172345e-05,
+      "loss": 0.4537,
+      "step": 239
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.39310508966445923,
+      "learning_rate": 4.770541082164329e-05,
+      "loss": 0.4443,
+      "step": 240
+    },
+    {
+      "epoch": 0.964,
+      "grad_norm": 0.4204409420490265,
+      "learning_rate": 4.7695390781563125e-05,
+      "loss": 0.4745,
+      "step": 241
+    },
+    {
+      "epoch": 0.968,
+      "grad_norm": 0.42494168877601624,
+      "learning_rate": 4.768537074148297e-05,
+      "loss": 0.4443,
+      "step": 242
+    },
+    {
+      "epoch": 0.972,
+      "grad_norm": 0.406387060880661,
+      "learning_rate": 4.767535070140281e-05,
+      "loss": 0.491,
+      "step": 243
+    },
+    {
+      "epoch": 0.976,
+      "grad_norm": 0.4731022119522095,
+      "learning_rate": 4.766533066132265e-05,
+      "loss": 0.5007,
+      "step": 244
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.4447345733642578,
+      "learning_rate": 4.765531062124249e-05,
+      "loss": 0.5206,
+      "step": 245
+    },
+    {
+      "epoch": 0.984,
+      "grad_norm": 0.47526809573173523,
+      "learning_rate": 4.7645290581162324e-05,
+      "loss": 0.4576,
+      "step": 246
+    },
+    {
+      "epoch": 0.988,
+      "grad_norm": 0.43649783730506897,
+      "learning_rate": 4.7635270541082166e-05,
+      "loss": 0.5029,
+      "step": 247
+    },
+    {
+      "epoch": 0.992,
+      "grad_norm": 0.4308774173259735,
+      "learning_rate": 4.762525050100201e-05,
+      "loss": 0.4237,
+      "step": 248
+    },
+    {
+      "epoch": 0.996,
+      "grad_norm": 0.3801085948944092,
+      "learning_rate": 4.761523046092184e-05,
+      "loss": 0.4387,
+      "step": 249
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.4443725645542145,
+      "learning_rate": 4.760521042084168e-05,
+      "loss": 0.4304,
+      "step": 250
+    },
+    {
+      "epoch": 1.004,
+      "grad_norm": 0.3666495680809021,
+      "learning_rate": 4.7595190380761524e-05,
+      "loss": 0.4053,
+      "step": 251
+    },
+    {
+      "epoch": 1.008,
+      "grad_norm": 0.4456152617931366,
+      "learning_rate": 4.7585170340681365e-05,
+      "loss": 0.4301,
+      "step": 252
+    },
+    {
+      "epoch": 1.012,
+      "grad_norm": 0.4105600118637085,
+      "learning_rate": 4.7575150300601207e-05,
+      "loss": 0.4437,
+      "step": 253
+    },
+    {
+      "epoch": 1.016,
+      "grad_norm": 0.42938780784606934,
+      "learning_rate": 4.756513026052105e-05,
+      "loss": 0.4685,
+      "step": 254
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.4617982506752014,
+      "learning_rate": 4.755511022044088e-05,
+      "loss": 0.5046,
+      "step": 255
+    },
+    {
+      "epoch": 1.024,
+      "grad_norm": 0.3736300468444824,
+      "learning_rate": 4.7545090180360724e-05,
+      "loss": 0.4222,
+      "step": 256
+    },
+    {
+      "epoch": 1.028,
+      "grad_norm": 0.40071138739585876,
+      "learning_rate": 4.7535070140280565e-05,
+      "loss": 0.4113,
+      "step": 257
+    },
+    {
+      "epoch": 1.032,
+      "grad_norm": 0.43367400765419006,
+      "learning_rate": 4.75250501002004e-05,
+      "loss": 0.4743,
+      "step": 258
+    },
+    {
+      "epoch": 1.036,
+      "grad_norm": 0.5123701691627502,
+      "learning_rate": 4.751503006012024e-05,
+      "loss": 0.4289,
+      "step": 259
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.4922557771205902,
+      "learning_rate": 4.750501002004008e-05,
+      "loss": 0.4976,
+      "step": 260
+    },
+    {
+      "epoch": 1.044,
+      "grad_norm": 0.4395969808101654,
+      "learning_rate": 4.7494989979959916e-05,
+      "loss": 0.4344,
+      "step": 261
+    },
+    {
+      "epoch": 1.048,
+      "grad_norm": 0.4711666405200958,
+      "learning_rate": 4.7484969939879765e-05,
+      "loss": 0.4616,
+      "step": 262
+    },
+    {
+      "epoch": 1.052,
+      "grad_norm": 0.4415505528450012,
+      "learning_rate": 4.7474949899799606e-05,
+      "loss": 0.3927,
+      "step": 263
+    },
+    {
+      "epoch": 1.056,
+      "grad_norm": 0.47160524129867554,
+      "learning_rate": 4.746492985971944e-05,
+      "loss": 0.4112,
+      "step": 264
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.4496021568775177,
+      "learning_rate": 4.745490981963928e-05,
+      "loss": 0.4275,
+      "step": 265
+    },
+    {
+      "epoch": 1.064,
+      "grad_norm": 0.42134368419647217,
+      "learning_rate": 4.744488977955912e-05,
+      "loss": 0.4191,
+      "step": 266
+    },
+    {
+      "epoch": 1.068,
+      "grad_norm": 0.4793007969856262,
+      "learning_rate": 4.743486973947896e-05,
+      "loss": 0.4858,
+      "step": 267
+    },
+    {
+      "epoch": 1.072,
+      "grad_norm": 0.4894791841506958,
+      "learning_rate": 4.74248496993988e-05,
+      "loss": 0.4462,
+      "step": 268
+    },
+    {
+      "epoch": 1.076,
+      "grad_norm": 0.45502549409866333,
+      "learning_rate": 4.741482965931864e-05,
+      "loss": 0.4883,
+      "step": 269
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.4357253313064575,
+      "learning_rate": 4.7404809619238474e-05,
+      "loss": 0.4329,
+      "step": 270
+    },
+    {
+      "epoch": 1.084,
+      "grad_norm": 0.4542175233364105,
+      "learning_rate": 4.7394789579158316e-05,
+      "loss": 0.4501,
+      "step": 271
+    },
+    {
+      "epoch": 1.088,
+      "grad_norm": 0.4260950982570648,
+      "learning_rate": 4.7384769539078164e-05,
+      "loss": 0.4186,
+      "step": 272
+    },
+    {
+      "epoch": 1.092,
+      "grad_norm": 0.5126277804374695,
+      "learning_rate": 4.7374749498998e-05,
+      "loss": 0.4815,
+      "step": 273
+    },
+    {
+      "epoch": 1.096,
+      "grad_norm": 0.4585254490375519,
+      "learning_rate": 4.736472945891784e-05,
+      "loss": 0.4461,
+      "step": 274
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.4200245440006256,
+      "learning_rate": 4.735470941883768e-05,
+      "loss": 0.4348,
+      "step": 275
+    },
+    {
+      "epoch": 1.104,
+      "grad_norm": 0.4564264118671417,
+      "learning_rate": 4.7344689378757515e-05,
+      "loss": 0.4483,
+      "step": 276
+    },
+    {
+      "epoch": 1.108,
+      "grad_norm": 0.38692179322242737,
+      "learning_rate": 4.733466933867736e-05,
+      "loss": 0.4139,
+      "step": 277
+    },
+    {
+      "epoch": 1.112,
+      "grad_norm": 0.5220334529876709,
+      "learning_rate": 4.73246492985972e-05,
+      "loss": 0.4688,
+      "step": 278
+    },
+    {
+      "epoch": 1.116,
+      "grad_norm": 0.47125041484832764,
+      "learning_rate": 4.731462925851703e-05,
+      "loss": 0.4965,
+      "step": 279
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.41665714979171753,
+      "learning_rate": 4.7304609218436874e-05,
+      "loss": 0.4569,
+      "step": 280
+    },
+    {
+      "epoch": 1.124,
+      "grad_norm": 0.45704329013824463,
+      "learning_rate": 4.7294589178356715e-05,
+      "loss": 0.4441,
+      "step": 281
+    },
+    {
+      "epoch": 1.1280000000000001,
+      "grad_norm": 0.5076858401298523,
+      "learning_rate": 4.7284569138276556e-05,
+      "loss": 0.4441,
+      "step": 282
+    },
+    {
+      "epoch": 1.1320000000000001,
+      "grad_norm": 0.4565269649028778,
+      "learning_rate": 4.72745490981964e-05,
+      "loss": 0.4001,
+      "step": 283
+    },
+    {
+      "epoch": 1.1360000000000001,
+      "grad_norm": 0.4239833652973175,
+      "learning_rate": 4.726452905811624e-05,
+      "loss": 0.3982,
+      "step": 284
+    },
+    {
+      "epoch": 1.1400000000000001,
+      "grad_norm": 0.4458796977996826,
+      "learning_rate": 4.725450901803607e-05,
+      "loss": 0.4151,
+      "step": 285
+    },
+    {
+      "epoch": 1.144,
+      "grad_norm": 0.49094000458717346,
+      "learning_rate": 4.7244488977955915e-05,
+      "loss": 0.4491,
+      "step": 286
+    },
+    {
+      "epoch": 1.148,
+      "grad_norm": 11.27839183807373,
+      "learning_rate": 4.7234468937875756e-05,
+      "loss": 0.934,
+      "step": 287
+    },
+    {
+      "epoch": 1.152,
+      "grad_norm": 0.5622511506080627,
+      "learning_rate": 4.722444889779559e-05,
+      "loss": 0.4862,
+      "step": 288
+    },
+    {
+      "epoch": 1.156,
+      "grad_norm": 0.5482255220413208,
+      "learning_rate": 4.721442885771543e-05,
+      "loss": 0.4668,
+      "step": 289
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 7.071150779724121,
+      "learning_rate": 4.720440881763527e-05,
+      "loss": 0.8774,
+      "step": 290
+    },
+    {
+      "epoch": 1.164,
+      "grad_norm": 1.3377376794815063,
+      "learning_rate": 4.7194388777555114e-05,
+      "loss": 0.4714,
+      "step": 291
+    },
+    {
+      "epoch": 1.168,
+      "grad_norm": 0.5460695028305054,
+      "learning_rate": 4.7184368737474956e-05,
+      "loss": 0.4769,
+      "step": 292
+    },
+    {
+      "epoch": 1.172,
+      "grad_norm": 0.4908643364906311,
+      "learning_rate": 4.717434869739479e-05,
+      "loss": 0.4904,
+      "step": 293
+    },
+    {
+      "epoch": 1.176,
+      "grad_norm": 0.42432764172554016,
+      "learning_rate": 4.716432865731463e-05,
+      "loss": 0.4213,
+      "step": 294
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.5326458811759949,
+      "learning_rate": 4.715430861723447e-05,
+      "loss": 0.4568,
+      "step": 295
+    },
+    {
+      "epoch": 1.184,
+      "grad_norm": 0.4412928521633148,
+      "learning_rate": 4.7144288577154314e-05,
+      "loss": 0.4776,
+      "step": 296
+    },
+    {
+      "epoch": 1.188,
+      "grad_norm": 0.4530280828475952,
+      "learning_rate": 4.713426853707415e-05,
+      "loss": 0.4139,
+      "step": 297
+    },
+    {
+      "epoch": 1.192,
+      "grad_norm": 0.4728831946849823,
+      "learning_rate": 4.712424849699399e-05,
+      "loss": 0.4586,
+      "step": 298
+    },
+    {
+      "epoch": 1.196,
+      "grad_norm": 0.4735589325428009,
+      "learning_rate": 4.711422845691383e-05,
+      "loss": 0.4707,
+      "step": 299
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.5155274271965027,
+      "learning_rate": 4.7104208416833666e-05,
+      "loss": 0.425,
+      "step": 300
+    },
+    {
+      "epoch": 1.204,
+      "grad_norm": 0.5296701788902283,
+      "learning_rate": 4.7094188376753514e-05,
+      "loss": 0.4858,
+      "step": 301
+    },
+    {
+      "epoch": 1.208,
+      "grad_norm": 0.478127121925354,
+      "learning_rate": 4.708416833667335e-05,
+      "loss": 0.4522,
+      "step": 302
+    },
+    {
+      "epoch": 1.212,
+      "grad_norm": 0.4879436194896698,
+      "learning_rate": 4.707414829659319e-05,
+      "loss": 0.45,
+      "step": 303
+    },
+    {
+      "epoch": 1.216,
+      "grad_norm": 0.4468042850494385,
+      "learning_rate": 4.706412825651303e-05,
+      "loss": 0.4119,
+      "step": 304
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.47193828225135803,
+      "learning_rate": 4.7054108216432865e-05,
+      "loss": 0.4911,
+      "step": 305
+    },
+    {
+      "epoch": 1.224,
+      "grad_norm": 0.46041783690452576,
+      "learning_rate": 4.7044088176352706e-05,
+      "loss": 0.4087,
+      "step": 306
+    },
+    {
+      "epoch": 1.228,
+      "grad_norm": 0.5108659267425537,
+      "learning_rate": 4.703406813627255e-05,
+      "loss": 0.4601,
+      "step": 307
+    },
+    {
+      "epoch": 1.232,
+      "grad_norm": 0.4466314911842346,
+      "learning_rate": 4.702404809619239e-05,
+      "loss": 0.4084,
+      "step": 308
+    },
+    {
+      "epoch": 1.236,
+      "grad_norm": 0.5967437624931335,
+      "learning_rate": 4.7014028056112224e-05,
+      "loss": 0.4609,
+      "step": 309
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.4608320891857147,
+      "learning_rate": 4.7004008016032065e-05,
+      "loss": 0.3945,
+      "step": 310
+    },
+    {
+      "epoch": 1.244,
+      "grad_norm": 0.47840774059295654,
+      "learning_rate": 4.6993987975951906e-05,
+      "loss": 0.4506,
+      "step": 311
+    },
+    {
+      "epoch": 1.248,
+      "grad_norm": 0.49409082531929016,
+      "learning_rate": 4.698396793587175e-05,
+      "loss": 0.4387,
+      "step": 312
+    },
+    {
+      "epoch": 1.252,
+      "grad_norm": 0.46387237310409546,
+      "learning_rate": 4.697394789579159e-05,
+      "loss": 0.436,
+      "step": 313
+    },
+    {
+      "epoch": 1.256,
+      "grad_norm": 1.0678813457489014,
+      "learning_rate": 4.696392785571142e-05,
+      "loss": 0.4139,
+      "step": 314
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.5538118481636047,
+      "learning_rate": 4.6953907815631264e-05,
+      "loss": 0.4666,
+      "step": 315
+    },
+    {
+      "epoch": 1.264,
+      "grad_norm": 7.208223819732666,
+      "learning_rate": 4.6943887775551106e-05,
+      "loss": 0.6192,
+      "step": 316
+    },
+    {
+      "epoch": 1.268,
+      "grad_norm": 0.483923077583313,
+      "learning_rate": 4.693386773547094e-05,
+      "loss": 0.4103,
+      "step": 317
+    },
+    {
+      "epoch": 1.272,
+      "grad_norm": 0.4743936061859131,
+      "learning_rate": 4.692384769539078e-05,
+      "loss": 0.4514,
+      "step": 318
+    },
+    {
+      "epoch": 1.276,
+      "grad_norm": 0.5241031646728516,
+      "learning_rate": 4.691382765531062e-05,
+      "loss": 0.4661,
+      "step": 319
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.44694361090660095,
+      "learning_rate": 4.690380761523046e-05,
+      "loss": 0.4173,
+      "step": 320
+    },
+    {
+      "epoch": 1.284,
+      "grad_norm": 0.4751509726047516,
+      "learning_rate": 4.6893787575150305e-05,
+      "loss": 0.3872,
+      "step": 321
+    },
+    {
+      "epoch": 1.288,
+      "grad_norm": 0.548673152923584,
+      "learning_rate": 4.688376753507015e-05,
+      "loss": 0.474,
+      "step": 322
+    },
+    {
+      "epoch": 1.292,
+      "grad_norm": 0.46685653924942017,
+      "learning_rate": 4.687374749498998e-05,
+      "loss": 0.4433,
+      "step": 323
+    },
+    {
+      "epoch": 1.296,
+      "grad_norm": 3.28513240814209,
+      "learning_rate": 4.686372745490982e-05,
+      "loss": 0.4851,
+      "step": 324
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.5524126887321472,
+      "learning_rate": 4.6853707414829664e-05,
+      "loss": 0.4318,
+      "step": 325
+    },
+    {
+      "epoch": 1.304,
+      "grad_norm": 0.48285192251205444,
+      "learning_rate": 4.68436873747495e-05,
+      "loss": 0.4203,
+      "step": 326
+    },
+    {
+      "epoch": 1.308,
+      "grad_norm": 2.8871777057647705,
+      "learning_rate": 4.683366733466934e-05,
+      "loss": 0.4773,
+      "step": 327
+    },
+    {
+      "epoch": 1.312,
+      "grad_norm": 0.5716524720191956,
+      "learning_rate": 4.682364729458918e-05,
+      "loss": 0.4876,
+      "step": 328
+    },
+    {
+      "epoch": 1.316,
+      "grad_norm": 0.5107681155204773,
+      "learning_rate": 4.6813627254509015e-05,
+      "loss": 0.4194,
+      "step": 329
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.5048828125,
+      "learning_rate": 4.6803607214428857e-05,
+      "loss": 0.4635,
+      "step": 330
+    },
+    {
+      "epoch": 1.324,
+      "grad_norm": 0.4749213755130768,
+      "learning_rate": 4.6793587174348705e-05,
+      "loss": 0.4625,
+      "step": 331
+    },
+    {
+      "epoch": 1.328,
+      "grad_norm": 0.5069287419319153,
+      "learning_rate": 4.678356713426854e-05,
+      "loss": 0.3894,
+      "step": 332
+    },
+    {
+      "epoch": 1.332,
+      "grad_norm": 0.5243629217147827,
+      "learning_rate": 4.677354709418838e-05,
+      "loss": 0.4179,
+      "step": 333
+    },
+    {
+      "epoch": 1.336,
+      "grad_norm": 0.4602661728858948,
+      "learning_rate": 4.676352705410822e-05,
+      "loss": 0.4378,
+      "step": 334
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.591300904750824,
+      "learning_rate": 4.6753507014028056e-05,
+      "loss": 0.4615,
+      "step": 335
+    },
+    {
+      "epoch": 1.3439999999999999,
+      "grad_norm": 0.5004926919937134,
+      "learning_rate": 4.67434869739479e-05,
+      "loss": 0.4199,
+      "step": 336
+    },
+    {
+      "epoch": 1.3479999999999999,
+      "grad_norm": 0.5834985971450806,
+      "learning_rate": 4.673346693386774e-05,
+      "loss": 0.499,
+      "step": 337
+    },
+    {
+      "epoch": 1.3519999999999999,
+      "grad_norm": 0.4432782530784607,
+      "learning_rate": 4.672344689378757e-05,
+      "loss": 0.3882,
+      "step": 338
+    },
+    {
+      "epoch": 1.3559999999999999,
+      "grad_norm": 0.5421860814094543,
+      "learning_rate": 4.6713426853707415e-05,
+      "loss": 0.479,
+      "step": 339
+    },
+    {
+      "epoch": 1.3599999999999999,
+      "grad_norm": 0.5701932907104492,
+      "learning_rate": 4.6703406813627256e-05,
+      "loss": 0.4337,
+      "step": 340
+    },
+    {
+      "epoch": 1.3639999999999999,
+      "grad_norm": 0.5272790193557739,
+      "learning_rate": 4.66933867735471e-05,
+      "loss": 0.4419,
+      "step": 341
+    },
+    {
+      "epoch": 1.3679999999999999,
+      "grad_norm": 0.5008136034011841,
+      "learning_rate": 4.668336673346694e-05,
+      "loss": 0.4162,
+      "step": 342
+    },
+    {
+      "epoch": 1.3719999999999999,
+      "grad_norm": 0.47261759638786316,
+      "learning_rate": 4.667334669338678e-05,
+      "loss": 0.4037,
+      "step": 343
+    },
+    {
+      "epoch": 1.376,
+      "grad_norm": 0.49282076954841614,
+      "learning_rate": 4.6663326653306614e-05,
+      "loss": 0.4203,
+      "step": 344
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.5760653614997864,
+      "learning_rate": 4.6653306613226455e-05,
+      "loss": 0.4935,
+      "step": 345
+    },
+    {
+      "epoch": 1.384,
+      "grad_norm": 0.4914257824420929,
+      "learning_rate": 4.66432865731463e-05,
+      "loss": 0.3675,
+      "step": 346
+    },
+    {
+      "epoch": 1.388,
+      "grad_norm": 0.6651628017425537,
+      "learning_rate": 4.663326653306613e-05,
+      "loss": 0.4484,
+      "step": 347
+    },
+    {
+      "epoch": 1.392,
+      "grad_norm": 1.1650779247283936,
+      "learning_rate": 4.662324649298597e-05,
+      "loss": 0.496,
+      "step": 348
+    },
+    {
+      "epoch": 1.396,
+      "grad_norm": 0.4959772825241089,
+      "learning_rate": 4.6613226452905814e-05,
+      "loss": 0.3795,
+      "step": 349
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 11.361075401306152,
+      "learning_rate": 4.6603206412825655e-05,
+      "loss": 0.4368,
+      "step": 350
+    },
+    {
+      "epoch": 1.404,
+      "grad_norm": 0.5089369416236877,
+      "learning_rate": 4.6593186372745496e-05,
+      "loss": 0.3721,
+      "step": 351
+    },
+    {
+      "epoch": 1.408,
+      "grad_norm": 0.4523348808288574,
+      "learning_rate": 4.658316633266534e-05,
+      "loss": 0.4223,
+      "step": 352
+    },
+    {
+      "epoch": 1.412,
+      "grad_norm": 0.5115591287612915,
+      "learning_rate": 4.657314629258517e-05,
+      "loss": 0.4383,
+      "step": 353
+    },
+    {
+      "epoch": 1.416,
+      "grad_norm": 0.4922471344470978,
+      "learning_rate": 4.6563126252505013e-05,
+      "loss": 0.4566,
+      "step": 354
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.49494296312332153,
+      "learning_rate": 4.6553106212424855e-05,
+      "loss": 0.4328,
+      "step": 355
+    },
+    {
+      "epoch": 1.424,
+      "grad_norm": 0.6130332946777344,
+      "learning_rate": 4.654308617234469e-05,
+      "loss": 0.4885,
+      "step": 356
+    },
+    {
+      "epoch": 1.428,
+      "grad_norm": 0.5084357857704163,
+      "learning_rate": 4.653306613226453e-05,
+      "loss": 0.4918,
+      "step": 357
+    },
+    {
+      "epoch": 1.432,
+      "grad_norm": 0.5193467140197754,
+      "learning_rate": 4.652304609218437e-05,
+      "loss": 0.4379,
+      "step": 358
+    },
+    {
+      "epoch": 1.436,
+      "grad_norm": 0.5419803261756897,
+      "learning_rate": 4.6513026052104206e-05,
+      "loss": 0.444,
+      "step": 359
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.5299074649810791,
+      "learning_rate": 4.6503006012024054e-05,
+      "loss": 0.4716,
+      "step": 360
+    },
+    {
+      "epoch": 1.444,
+      "grad_norm": 1.1689682006835938,
+      "learning_rate": 4.649298597194389e-05,
+      "loss": 0.2592,
+      "step": 361
+    },
+    {
+      "epoch": 1.448,
+      "grad_norm": 0.532464861869812,
+      "learning_rate": 4.648296593186373e-05,
+      "loss": 0.4572,
+      "step": 362
+    },
+    {
+      "epoch": 1.452,
+      "grad_norm": 0.4852958023548126,
+      "learning_rate": 4.647294589178357e-05,
+      "loss": 0.4132,
+      "step": 363
+    },
+    {
+      "epoch": 1.456,
+      "grad_norm": 0.4912414848804474,
+      "learning_rate": 4.646292585170341e-05,
+      "loss": 0.5046,
+      "step": 364
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.5567641854286194,
+      "learning_rate": 4.645290581162325e-05,
+      "loss": 0.4648,
+      "step": 365
+    },
+    {
+      "epoch": 1.464,
+      "grad_norm": 0.5151598453521729,
+      "learning_rate": 4.644288577154309e-05,
+      "loss": 0.4677,
+      "step": 366
+    },
+    {
+      "epoch": 1.468,
+      "grad_norm": 0.5214430093765259,
+      "learning_rate": 4.643286573146293e-05,
+      "loss": 0.4396,
+      "step": 367
+    },
+    {
+      "epoch": 1.472,
+      "grad_norm": 0.5087822079658508,
+      "learning_rate": 4.6422845691382764e-05,
+      "loss": 0.4185,
+      "step": 368
+    },
+    {
+      "epoch": 1.476,
+      "grad_norm": 0.49884939193725586,
+      "learning_rate": 4.6412825651302606e-05,
+      "loss": 0.427,
+      "step": 369
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.5047423243522644,
+      "learning_rate": 4.640280561122245e-05,
+      "loss": 0.4682,
+      "step": 370
+    },
+    {
+      "epoch": 1.484,
+      "grad_norm": 0.5239180326461792,
+      "learning_rate": 4.639278557114229e-05,
+      "loss": 0.4701,
+      "step": 371
+    },
+    {
+      "epoch": 1.488,
+      "grad_norm": 0.48768556118011475,
+      "learning_rate": 4.638276553106213e-05,
+      "loss": 0.4158,
+      "step": 372
+    },
+    {
+      "epoch": 1.492,
+      "grad_norm": 0.5886668562889099,
+      "learning_rate": 4.6372745490981964e-05,
+      "loss": 0.4679,
+      "step": 373
+    },
+    {
+      "epoch": 1.496,
+      "grad_norm": 0.5226956009864807,
+      "learning_rate": 4.6362725450901805e-05,
+      "loss": 0.4523,
+      "step": 374
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.5257872939109802,
+      "learning_rate": 4.6352705410821647e-05,
+      "loss": 0.4399,
+      "step": 375
+    },
+    {
+      "epoch": 1.504,
+      "grad_norm": 0.6213669180870056,
+      "learning_rate": 4.634268537074148e-05,
+      "loss": 0.4535,
+      "step": 376
+    },
+    {
+      "epoch": 1.508,
+      "grad_norm": 0.49080657958984375,
+      "learning_rate": 4.633266533066132e-05,
+      "loss": 0.4379,
+      "step": 377
+    },
+    {
+      "epoch": 1.512,
+      "grad_norm": 0.6423617005348206,
+      "learning_rate": 4.6322645290581164e-05,
+      "loss": 0.5559,
+      "step": 378
+    },
+    {
+      "epoch": 1.516,
+      "grad_norm": 0.6420733332633972,
+      "learning_rate": 4.6312625250501005e-05,
+      "loss": 0.4543,
+      "step": 379
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.5223842859268188,
+      "learning_rate": 4.6302605210420846e-05,
+      "loss": 0.466,
+      "step": 380
+    },
+    {
+      "epoch": 1.524,
+      "grad_norm": 0.4663429856300354,
+      "learning_rate": 4.629258517034069e-05,
+      "loss": 0.4153,
+      "step": 381
+    },
+    {
+      "epoch": 1.528,
+      "grad_norm": 0.5308319926261902,
+      "learning_rate": 4.628256513026052e-05,
+      "loss": 0.4136,
+      "step": 382
+    },
+    {
+      "epoch": 1.532,
+      "grad_norm": 0.48709505796432495,
+      "learning_rate": 4.627254509018036e-05,
+      "loss": 0.4478,
+      "step": 383
+    },
+    {
+      "epoch": 1.536,
+      "grad_norm": 0.4852888584136963,
+      "learning_rate": 4.6262525050100205e-05,
+      "loss": 0.3736,
+      "step": 384
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.510391116142273,
+      "learning_rate": 4.625250501002004e-05,
+      "loss": 0.4505,
+      "step": 385
+    },
+    {
+      "epoch": 1.544,
+      "grad_norm": 0.48822277784347534,
+      "learning_rate": 4.624248496993988e-05,
+      "loss": 0.384,
+      "step": 386
+    },
+    {
+      "epoch": 1.548,
+      "grad_norm": 0.5103605389595032,
+      "learning_rate": 4.623246492985972e-05,
+      "loss": 0.3747,
+      "step": 387
+    },
+    {
+      "epoch": 1.552,
+      "grad_norm": 0.5321424603462219,
+      "learning_rate": 4.6222444889779556e-05,
+      "loss": 0.4732,
+      "step": 388
+    },
+    {
+      "epoch": 1.556,
+      "grad_norm": 0.5756722092628479,
+      "learning_rate": 4.62124248496994e-05,
+      "loss": 0.4735,
+      "step": 389
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.5002274513244629,
+      "learning_rate": 4.6202404809619245e-05,
+      "loss": 0.4149,
+      "step": 390
+    },
+    {
+      "epoch": 1.564,
+      "grad_norm": 0.5540143847465515,
+      "learning_rate": 4.619238476953908e-05,
+      "loss": 0.4388,
+      "step": 391
+    },
+    {
+      "epoch": 1.568,
+      "grad_norm": 0.5765349864959717,
+      "learning_rate": 4.618236472945892e-05,
+      "loss": 0.5177,
+      "step": 392
+    },
+    {
+      "epoch": 1.572,
+      "grad_norm": 0.5519753694534302,
+      "learning_rate": 4.617234468937876e-05,
+      "loss": 0.4497,
+      "step": 393
+    },
+    {
+      "epoch": 1.576,
+      "grad_norm": 0.5800144672393799,
+      "learning_rate": 4.61623246492986e-05,
+      "loss": 0.448,
+      "step": 394
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.5450941920280457,
+      "learning_rate": 4.615230460921844e-05,
+      "loss": 0.4368,
+      "step": 395
+    },
+    {
+      "epoch": 1.584,
+      "grad_norm": 0.50678551197052,
+      "learning_rate": 4.614228456913828e-05,
+      "loss": 0.4496,
+      "step": 396
+    },
+    {
+      "epoch": 1.588,
+      "grad_norm": 0.48308175802230835,
+      "learning_rate": 4.6132264529058114e-05,
+      "loss": 0.3646,
+      "step": 397
+    },
+    {
+      "epoch": 1.592,
+      "grad_norm": 0.5025294423103333,
+      "learning_rate": 4.6122244488977955e-05,
+      "loss": 0.4331,
+      "step": 398
+    },
+    {
+      "epoch": 1.596,
+      "grad_norm": 0.4999145567417145,
+      "learning_rate": 4.61122244488978e-05,
+      "loss": 0.3968,
+      "step": 399
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.5278040170669556,
+      "learning_rate": 4.610220440881764e-05,
+      "loss": 0.3938,
+      "step": 400
+    },
+    {
+      "epoch": 1.604,
+      "grad_norm": 0.5218545198440552,
+      "learning_rate": 4.609218436873748e-05,
+      "loss": 0.4465,
+      "step": 401
+    },
+    {
+      "epoch": 1.608,
+      "grad_norm": 0.5310875177383423,
+      "learning_rate": 4.608216432865732e-05,
+      "loss": 0.4094,
+      "step": 402
+    },
+    {
+      "epoch": 1.612,
+      "grad_norm": 0.47267839312553406,
+      "learning_rate": 4.6072144288577155e-05,
+      "loss": 0.4233,
+      "step": 403
+    },
+    {
+      "epoch": 1.616,
+      "grad_norm": 0.9498124122619629,
+      "learning_rate": 4.6062124248496996e-05,
+      "loss": 0.4357,
+      "step": 404
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.46646052598953247,
+      "learning_rate": 4.605210420841684e-05,
+      "loss": 0.4146,
+      "step": 405
+    },
+    {
+      "epoch": 1.624,
+      "grad_norm": 1.2247719764709473,
+      "learning_rate": 4.604208416833667e-05,
+      "loss": 0.3475,
+      "step": 406
+    },
+    {
+      "epoch": 1.6280000000000001,
+      "grad_norm": 0.5586639642715454,
+      "learning_rate": 4.603206412825651e-05,
+      "loss": 0.444,
+      "step": 407
+    },
+    {
+      "epoch": 1.6320000000000001,
+      "grad_norm": 0.552677571773529,
+      "learning_rate": 4.6022044088176355e-05,
+      "loss": 0.4234,
+      "step": 408
+    },
+    {
+      "epoch": 1.6360000000000001,
+      "grad_norm": 0.5379041433334351,
+      "learning_rate": 4.6012024048096196e-05,
+      "loss": 0.4755,
+      "step": 409
+    },
+    {
+      "epoch": 1.6400000000000001,
+      "grad_norm": 0.5361511707305908,
+      "learning_rate": 4.600200400801604e-05,
+      "loss": 0.3844,
+      "step": 410
+    },
+    {
+      "epoch": 1.6440000000000001,
+      "grad_norm": 0.5484291315078735,
+      "learning_rate": 4.599198396793588e-05,
+      "loss": 0.4237,
+      "step": 411
+    },
+    {
+      "epoch": 1.6480000000000001,
+      "grad_norm": 0.47265616059303284,
+      "learning_rate": 4.598196392785571e-05,
+      "loss": 0.4055,
+      "step": 412
+    },
+    {
+      "epoch": 1.6520000000000001,
+      "grad_norm": 0.586786150932312,
+      "learning_rate": 4.5971943887775554e-05,
+      "loss": 0.5109,
+      "step": 413
+    },
+    {
+      "epoch": 1.6560000000000001,
+      "grad_norm": 0.4930340051651001,
+      "learning_rate": 4.5961923847695396e-05,
+      "loss": 0.3967,
+      "step": 414
+    },
+    {
+      "epoch": 1.6600000000000001,
+      "grad_norm": 0.5312355160713196,
+      "learning_rate": 4.595190380761523e-05,
+      "loss": 0.4881,
+      "step": 415
+    },
+    {
+      "epoch": 1.6640000000000001,
+      "grad_norm": 0.5358534455299377,
+      "learning_rate": 4.594188376753507e-05,
+      "loss": 0.4785,
+      "step": 416
+    },
+    {
+      "epoch": 1.6680000000000001,
+      "grad_norm": 0.5466772317886353,
+      "learning_rate": 4.593186372745491e-05,
+      "loss": 0.4695,
+      "step": 417
+    },
+    {
+      "epoch": 1.6720000000000002,
+      "grad_norm": 0.5252442359924316,
+      "learning_rate": 4.592184368737475e-05,
+      "loss": 0.4647,
+      "step": 418
+    },
+    {
+      "epoch": 1.6760000000000002,
+      "grad_norm": 0.6336548924446106,
+      "learning_rate": 4.5911823647294595e-05,
+      "loss": 0.4535,
+      "step": 419
+    },
+    {
+      "epoch": 1.6800000000000002,
+      "grad_norm": 0.5073897838592529,
+      "learning_rate": 4.590180360721443e-05,
+      "loss": 0.4622,
+      "step": 420
+    },
+    {
+      "epoch": 1.6840000000000002,
+      "grad_norm": 0.5029900074005127,
+      "learning_rate": 4.589178356713427e-05,
+      "loss": 0.4221,
+      "step": 421
+    },
+    {
+      "epoch": 1.688,
+      "grad_norm": 0.590262234210968,
+      "learning_rate": 4.588176352705411e-05,
+      "loss": 0.4097,
+      "step": 422
+    },
+    {
+      "epoch": 1.692,
+      "grad_norm": 0.5236296653747559,
+      "learning_rate": 4.5871743486973954e-05,
+      "loss": 0.4445,
+      "step": 423
+    },
+    {
+      "epoch": 1.696,
+      "grad_norm": 0.5591893792152405,
+      "learning_rate": 4.586172344689379e-05,
+      "loss": 0.4016,
+      "step": 424
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.48660945892333984,
+      "learning_rate": 4.585170340681363e-05,
+      "loss": 0.3968,
+      "step": 425
+    },
+    {
+      "epoch": 1.704,
+      "grad_norm": 1.2732222080230713,
+      "learning_rate": 4.584168336673347e-05,
+      "loss": 0.2928,
+      "step": 426
+    },
+    {
+      "epoch": 1.708,
+      "grad_norm": 0.5137063264846802,
+      "learning_rate": 4.5831663326653305e-05,
+      "loss": 0.4274,
+      "step": 427
+    },
+    {
+      "epoch": 1.712,
+      "grad_norm": 0.532532811164856,
+      "learning_rate": 4.5821643286573146e-05,
+      "loss": 0.443,
+      "step": 428
+    },
+    {
+      "epoch": 1.716,
+      "grad_norm": 0.7271252870559692,
+      "learning_rate": 4.581162324649299e-05,
+      "loss": 0.5061,
+      "step": 429
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.49782320857048035,
+      "learning_rate": 4.580160320641283e-05,
+      "loss": 0.4603,
+      "step": 430
+    },
+    {
+      "epoch": 1.724,
+      "grad_norm": 0.49443551898002625,
+      "learning_rate": 4.579158316633267e-05,
+      "loss": 0.3841,
+      "step": 431
+    },
+    {
+      "epoch": 1.728,
+      "grad_norm": 0.5045324563980103,
+      "learning_rate": 4.5781563126252505e-05,
+      "loss": 0.3932,
+      "step": 432
+    },
+    {
+      "epoch": 1.732,
+      "grad_norm": 0.5934986472129822,
+      "learning_rate": 4.5771543086172346e-05,
+      "loss": 0.5067,
+      "step": 433
+    },
+    {
+      "epoch": 1.736,
+      "grad_norm": 0.5429759621620178,
+      "learning_rate": 4.576152304609219e-05,
+      "loss": 0.3985,
+      "step": 434
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.6646963357925415,
+      "learning_rate": 4.575150300601203e-05,
+      "loss": 0.4736,
+      "step": 435
+    },
+    {
+      "epoch": 1.744,
+      "grad_norm": 0.5723690986633301,
+      "learning_rate": 4.574148296593186e-05,
+      "loss": 0.4137,
+      "step": 436
+    },
+    {
+      "epoch": 1.748,
+      "grad_norm": 0.5194367170333862,
+      "learning_rate": 4.5731462925851704e-05,
+      "loss": 0.3921,
+      "step": 437
+    },
+    {
+      "epoch": 1.752,
+      "grad_norm": 0.510465145111084,
+      "learning_rate": 4.5721442885771546e-05,
+      "loss": 0.4055,
+      "step": 438
+    },
+    {
+      "epoch": 1.756,
+      "grad_norm": 0.6061939597129822,
+      "learning_rate": 4.571142284569139e-05,
+      "loss": 0.4404,
+      "step": 439
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.5550720691680908,
+      "learning_rate": 4.570140280561123e-05,
+      "loss": 0.4366,
+      "step": 440
+    },
+    {
+      "epoch": 1.764,
+      "grad_norm": 0.5832468867301941,
+      "learning_rate": 4.569138276553106e-05,
+      "loss": 0.4176,
+      "step": 441
+    },
+    {
+      "epoch": 1.768,
+      "grad_norm": 0.5731306672096252,
+      "learning_rate": 4.5681362725450904e-05,
+      "loss": 0.3685,
+      "step": 442
+    },
+    {
+      "epoch": 1.772,
+      "grad_norm": 0.6895073056221008,
+      "learning_rate": 4.5671342685370745e-05,
+      "loss": 0.4906,
+      "step": 443
+    },
+    {
+      "epoch": 1.776,
+      "grad_norm": 0.6335456967353821,
+      "learning_rate": 4.566132264529058e-05,
+      "loss": 0.4004,
+      "step": 444
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.5286179780960083,
+      "learning_rate": 4.565130260521042e-05,
+      "loss": 0.409,
+      "step": 445
+    },
+    {
+      "epoch": 1.784,
+      "grad_norm": 0.626899003982544,
+      "learning_rate": 4.564128256513026e-05,
+      "loss": 0.4522,
+      "step": 446
+    },
+    {
+      "epoch": 1.788,
+      "grad_norm": 0.5576806664466858,
+      "learning_rate": 4.56312625250501e-05,
+      "loss": 0.4494,
+      "step": 447
+    },
+    {
+      "epoch": 1.792,
+      "grad_norm": 0.5164839625358582,
+      "learning_rate": 4.562124248496994e-05,
+      "loss": 0.4463,
+      "step": 448
+    },
+    {
+      "epoch": 1.796,
+      "grad_norm": 0.5332900881767273,
+      "learning_rate": 4.5611222444889786e-05,
+      "loss": 0.4454,
+      "step": 449
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.6575366258621216,
+      "learning_rate": 4.560120240480962e-05,
+      "loss": 0.4384,
+      "step": 450
+    },
+    {
+      "epoch": 1.804,
+      "grad_norm": 0.571654200553894,
+      "learning_rate": 4.559118236472946e-05,
+      "loss": 0.4229,
+      "step": 451
+    },
+    {
+      "epoch": 1.808,
+      "grad_norm": 0.5586163997650146,
+      "learning_rate": 4.55811623246493e-05,
+      "loss": 0.4484,
+      "step": 452
+    },
+    {
+      "epoch": 1.812,
+      "grad_norm": 0.5708151459693909,
+      "learning_rate": 4.557114228456914e-05,
+      "loss": 0.4247,
+      "step": 453
+    },
+    {
+      "epoch": 1.8159999999999998,
+      "grad_norm": 0.5310369729995728,
+      "learning_rate": 4.556112224448898e-05,
+      "loss": 0.4246,
+      "step": 454
+    },
+    {
+      "epoch": 1.8199999999999998,
+      "grad_norm": 0.5038295984268188,
+      "learning_rate": 4.555110220440882e-05,
+      "loss": 0.3868,
+      "step": 455
+    },
+    {
+      "epoch": 1.8239999999999998,
+      "grad_norm": 0.6038540005683899,
+      "learning_rate": 4.5541082164328655e-05,
+      "loss": 0.5129,
+      "step": 456
+    },
+    {
+      "epoch": 1.8279999999999998,
+      "grad_norm": 0.6133531928062439,
+      "learning_rate": 4.5531062124248496e-05,
+      "loss": 0.4609,
+      "step": 457
+    },
+    {
+      "epoch": 1.8319999999999999,
+      "grad_norm": 0.6467635035514832,
+      "learning_rate": 4.5521042084168344e-05,
+      "loss": 0.4598,
+      "step": 458
+    },
+    {
+      "epoch": 1.8359999999999999,
+      "grad_norm": 0.5676400661468506,
+      "learning_rate": 4.551102204408818e-05,
+      "loss": 0.3724,
+      "step": 459
+    },
+    {
+      "epoch": 1.8399999999999999,
+      "grad_norm": 3.835160732269287,
+      "learning_rate": 4.550100200400802e-05,
+      "loss": 0.3293,
+      "step": 460
+    },
+    {
+      "epoch": 1.8439999999999999,
+      "grad_norm": 0.5771289467811584,
+      "learning_rate": 4.549098196392786e-05,
+      "loss": 0.4299,
+      "step": 461
+    },
+    {
+      "epoch": 1.8479999999999999,
+      "grad_norm": 0.5256178379058838,
+      "learning_rate": 4.5480961923847696e-05,
+      "loss": 0.3774,
+      "step": 462
+    },
+    {
+      "epoch": 1.8519999999999999,
+      "grad_norm": 0.5819401741027832,
+      "learning_rate": 4.547094188376754e-05,
+      "loss": 0.4542,
+      "step": 463
+    },
+    {
+      "epoch": 1.8559999999999999,
+      "grad_norm": 0.5631145238876343,
+      "learning_rate": 4.546092184368738e-05,
+      "loss": 0.4326,
+      "step": 464
+    },
+    {
+      "epoch": 1.8599999999999999,
+      "grad_norm": 0.5169239044189453,
+      "learning_rate": 4.545090180360721e-05,
+      "loss": 0.3941,
+      "step": 465
+    },
+    {
+      "epoch": 1.8639999999999999,
+      "grad_norm": 0.5691187381744385,
+      "learning_rate": 4.5440881763527054e-05,
+      "loss": 0.483,
+      "step": 466
+    },
+    {
+      "epoch": 1.8679999999999999,
+      "grad_norm": 0.5725812911987305,
+      "learning_rate": 4.5430861723446895e-05,
+      "loss": 0.448,
+      "step": 467
+    },
+    {
+      "epoch": 1.8719999999999999,
+      "grad_norm": 0.5266627669334412,
+      "learning_rate": 4.542084168336674e-05,
+      "loss": 0.4355,
+      "step": 468
+    },
+    {
+      "epoch": 1.876,
+      "grad_norm": 0.4844062924385071,
+      "learning_rate": 4.541082164328658e-05,
+      "loss": 0.4057,
+      "step": 469
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.5010110139846802,
+      "learning_rate": 4.540080160320642e-05,
+      "loss": 0.4314,
+      "step": 470
+    },
+    {
+      "epoch": 1.884,
+      "grad_norm": 0.5683146715164185,
+      "learning_rate": 4.5390781563126254e-05,
+      "loss": 0.4376,
+      "step": 471
+    },
+    {
+      "epoch": 1.888,
+      "grad_norm": 0.535389244556427,
+      "learning_rate": 4.5380761523046095e-05,
+      "loss": 0.4047,
+      "step": 472
+    },
+    {
+      "epoch": 1.892,
+      "grad_norm": 0.49679070711135864,
+      "learning_rate": 4.5370741482965936e-05,
+      "loss": 0.4082,
+      "step": 473
+    },
+    {
+      "epoch": 1.896,
+      "grad_norm": 0.527479887008667,
+      "learning_rate": 4.536072144288577e-05,
+      "loss": 0.4829,
+      "step": 474
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.5545944571495056,
+      "learning_rate": 4.535070140280561e-05,
+      "loss": 0.4503,
+      "step": 475
+    },
+    {
+      "epoch": 1.904,
+      "grad_norm": 0.5213431119918823,
+      "learning_rate": 4.5340681362725453e-05,
+      "loss": 0.3898,
+      "step": 476
+    },
+    {
+      "epoch": 1.908,
+      "grad_norm": 0.5331659317016602,
+      "learning_rate": 4.533066132264529e-05,
+      "loss": 0.4274,
+      "step": 477
+    },
+    {
+      "epoch": 1.912,
+      "grad_norm": 0.4619419574737549,
+      "learning_rate": 4.5320641282565136e-05,
+      "loss": 0.3848,
+      "step": 478
+    },
+    {
+      "epoch": 1.916,
+      "grad_norm": 0.5063828825950623,
+      "learning_rate": 4.531062124248498e-05,
+      "loss": 0.4625,
+      "step": 479
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.4973587095737457,
+      "learning_rate": 4.530060120240481e-05,
+      "loss": 0.4274,
+      "step": 480
+    },
+    {
+      "epoch": 1.924,
+      "grad_norm": 0.49490657448768616,
+      "learning_rate": 4.529058116232465e-05,
+      "loss": 0.4322,
+      "step": 481
+    },
+    {
+      "epoch": 1.928,
+      "grad_norm": 4.3649067878723145,
+      "learning_rate": 4.5280561122244494e-05,
+      "loss": 0.4085,
+      "step": 482
+    },
+    {
+      "epoch": 1.932,
+      "grad_norm": 0.5393351316452026,
+      "learning_rate": 4.527054108216433e-05,
+      "loss": 0.4309,
+      "step": 483
+    },
+    {
+      "epoch": 1.936,
+      "grad_norm": 0.6917522549629211,
+      "learning_rate": 4.526052104208417e-05,
+      "loss": 0.268,
+      "step": 484
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.5249350666999817,
+      "learning_rate": 4.525050100200401e-05,
+      "loss": 0.4548,
+      "step": 485
+    },
+    {
+      "epoch": 1.944,
+      "grad_norm": 0.5986626148223877,
+      "learning_rate": 4.5240480961923846e-05,
+      "loss": 0.4347,
+      "step": 486
+    },
+    {
+      "epoch": 1.948,
+      "grad_norm": 0.6276787519454956,
+      "learning_rate": 4.523046092184369e-05,
+      "loss": 0.5127,
+      "step": 487
+    },
+    {
+      "epoch": 1.952,
+      "grad_norm": 0.5993108749389648,
+      "learning_rate": 4.522044088176353e-05,
+      "loss": 0.4559,
+      "step": 488
+    },
+    {
+      "epoch": 1.956,
+      "grad_norm": 0.4931926131248474,
+      "learning_rate": 4.521042084168337e-05,
+      "loss": 0.4144,
+      "step": 489
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.482890784740448,
+      "learning_rate": 4.520040080160321e-05,
+      "loss": 0.2603,
+      "step": 490
+    },
+    {
+      "epoch": 1.964,
+      "grad_norm": 0.6805306077003479,
+      "learning_rate": 4.519038076152305e-05,
+      "loss": 0.5272,
+      "step": 491
+    },
+    {
+      "epoch": 1.968,
+      "grad_norm": 0.6288460493087769,
+      "learning_rate": 4.518036072144289e-05,
+      "loss": 0.4455,
+      "step": 492
+    },
+    {
+      "epoch": 1.972,
+      "grad_norm": 0.5547501444816589,
+      "learning_rate": 4.517034068136273e-05,
+      "loss": 0.4432,
+      "step": 493
+    },
+    {
+      "epoch": 1.976,
+      "grad_norm": 0.545608639717102,
+      "learning_rate": 4.516032064128257e-05,
+      "loss": 0.4219,
+      "step": 494
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.5442771911621094,
+      "learning_rate": 4.5150300601202404e-05,
+      "loss": 0.4113,
+      "step": 495
+    },
+    {
+      "epoch": 1.984,
+      "grad_norm": 0.5810039043426514,
+      "learning_rate": 4.5140280561122245e-05,
+      "loss": 0.4304,
+      "step": 496
+    },
+    {
+      "epoch": 1.988,
+      "grad_norm": 0.5679133534431458,
+      "learning_rate": 4.5130260521042086e-05,
+      "loss": 0.3819,
+      "step": 497
+    },
+    {
+      "epoch": 1.992,
+      "grad_norm": 0.5476593375205994,
+      "learning_rate": 4.512024048096193e-05,
+      "loss": 0.4566,
+      "step": 498
+    },
+    {
+      "epoch": 1.996,
+      "grad_norm": 0.516356348991394,
+      "learning_rate": 4.511022044088177e-05,
+      "loss": 0.427,
+      "step": 499
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.5612138509750366,
+      "learning_rate": 4.5100200400801604e-05,
+      "loss": 0.4089,
+      "step": 500
+    },
+    {
+      "epoch": 2.004,
+      "grad_norm": 0.5504969358444214,
+      "learning_rate": 4.5090180360721445e-05,
+      "loss": 0.3998,
+      "step": 501
+    },
+    {
+      "epoch": 2.008,
+      "grad_norm": 0.5878968238830566,
+      "learning_rate": 4.5080160320641286e-05,
+      "loss": 0.3539,
+      "step": 502
+    },
+    {
+      "epoch": 2.012,
+      "grad_norm": 0.5411523580551147,
+      "learning_rate": 4.507014028056112e-05,
+      "loss": 0.3457,
+      "step": 503
+    },
+    {
+      "epoch": 2.016,
+      "grad_norm": 0.4899858236312866,
+      "learning_rate": 4.506012024048096e-05,
+      "loss": 0.3602,
+      "step": 504
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 0.5289257764816284,
+      "learning_rate": 4.50501002004008e-05,
+      "loss": 0.4093,
+      "step": 505
+    },
+    {
+      "epoch": 2.024,
+      "grad_norm": 0.5221056938171387,
+      "learning_rate": 4.5040080160320644e-05,
+      "loss": 0.3818,
+      "step": 506
+    },
+    {
+      "epoch": 2.028,
+      "grad_norm": 0.5114055871963501,
+      "learning_rate": 4.503006012024048e-05,
+      "loss": 0.3801,
+      "step": 507
+    },
+    {
+      "epoch": 2.032,
+      "grad_norm": 0.5133730173110962,
+      "learning_rate": 4.502004008016033e-05,
+      "loss": 0.3916,
+      "step": 508
+    },
+    {
+      "epoch": 2.036,
+      "grad_norm": 0.6223206520080566,
+      "learning_rate": 4.501002004008016e-05,
+      "loss": 0.3904,
+      "step": 509
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 0.597932755947113,
+      "learning_rate": 4.5e-05,
+      "loss": 0.3352,
+      "step": 510
+    },
+    {
+      "epoch": 2.044,
+      "grad_norm": 0.6192795038223267,
+      "learning_rate": 4.4989979959919844e-05,
+      "loss": 0.3788,
+      "step": 511
+    },
+    {
+      "epoch": 2.048,
+      "grad_norm": 0.5938431024551392,
+      "learning_rate": 4.497995991983968e-05,
+      "loss": 0.3946,
+      "step": 512
+    },
+    {
+      "epoch": 2.052,
+      "grad_norm": 0.5885916352272034,
+      "learning_rate": 4.496993987975952e-05,
+      "loss": 0.3706,
+      "step": 513
+    },
+    {
+      "epoch": 2.056,
+      "grad_norm": 0.6078701615333557,
+      "learning_rate": 4.495991983967936e-05,
+      "loss": 0.3561,
+      "step": 514
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 0.6931913495063782,
+      "learning_rate": 4.4949899799599196e-05,
+      "loss": 0.3697,
+      "step": 515
+    },
+    {
+      "epoch": 2.064,
+      "grad_norm": 0.6855788230895996,
+      "learning_rate": 4.493987975951904e-05,
+      "loss": 0.4235,
+      "step": 516
+    },
+    {
+      "epoch": 2.068,
+      "grad_norm": 0.5608630180358887,
+      "learning_rate": 4.4929859719438885e-05,
+      "loss": 0.3561,
+      "step": 517
+    },
+    {
+      "epoch": 2.072,
+      "grad_norm": 0.6403824687004089,
+      "learning_rate": 4.491983967935872e-05,
+      "loss": 0.3701,
+      "step": 518
+    },
+    {
+      "epoch": 2.076,
+      "grad_norm": 0.5921077728271484,
+      "learning_rate": 4.490981963927856e-05,
+      "loss": 0.36,
+      "step": 519
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 0.6234927773475647,
+      "learning_rate": 4.48997995991984e-05,
+      "loss": 0.4041,
+      "step": 520
+    },
+    {
+      "epoch": 2.084,
+      "grad_norm": 0.6670336723327637,
+      "learning_rate": 4.488977955911824e-05,
+      "loss": 0.3733,
+      "step": 521
+    },
+    {
+      "epoch": 2.088,
+      "grad_norm": 0.6524705290794373,
+      "learning_rate": 4.487975951903808e-05,
+      "loss": 0.344,
+      "step": 522
+    },
+    {
+      "epoch": 2.092,
+      "grad_norm": 0.6538939476013184,
+      "learning_rate": 4.486973947895792e-05,
+      "loss": 0.4209,
+      "step": 523
+    },
+    {
+      "epoch": 2.096,
+      "grad_norm": 0.6096150875091553,
+      "learning_rate": 4.4859719438877754e-05,
+      "loss": 0.3739,
+      "step": 524
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.5730370879173279,
+      "learning_rate": 4.4849699398797595e-05,
+      "loss": 0.3912,
+      "step": 525
+    },
+    {
+      "epoch": 2.104,
+      "grad_norm": 0.6088821887969971,
+      "learning_rate": 4.4839679358717436e-05,
+      "loss": 0.3633,
+      "step": 526
+    },
+    {
+      "epoch": 2.108,
+      "grad_norm": 0.9573196768760681,
+      "learning_rate": 4.482965931863728e-05,
+      "loss": 0.2372,
+      "step": 527
+    },
+    {
+      "epoch": 2.112,
+      "grad_norm": 0.6470226049423218,
+      "learning_rate": 4.481963927855712e-05,
+      "loss": 0.369,
+      "step": 528
+    },
+    {
+      "epoch": 2.116,
+      "grad_norm": 0.6785104274749756,
+      "learning_rate": 4.480961923847696e-05,
+      "loss": 0.4002,
+      "step": 529
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.46951213479042053,
+      "learning_rate": 4.4799599198396795e-05,
+      "loss": 0.2299,
+      "step": 530
+    },
+    {
+      "epoch": 2.124,
+      "grad_norm": 0.6221892237663269,
+      "learning_rate": 4.4789579158316636e-05,
+      "loss": 0.3324,
+      "step": 531
+    },
+    {
+      "epoch": 2.128,
+      "grad_norm": 0.6120625138282776,
+      "learning_rate": 4.477955911823648e-05,
+      "loss": 0.3473,
+      "step": 532
+    },
+    {
+      "epoch": 2.132,
+      "grad_norm": 0.6631773710250854,
+      "learning_rate": 4.476953907815631e-05,
+      "loss": 0.3919,
+      "step": 533
+    },
+    {
+      "epoch": 2.136,
+      "grad_norm": 0.6568106412887573,
+      "learning_rate": 4.475951903807615e-05,
+      "loss": 0.223,
+      "step": 534
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.6405401825904846,
+      "learning_rate": 4.4749498997995994e-05,
+      "loss": 0.3438,
+      "step": 535
+    },
+    {
+      "epoch": 2.144,
+      "grad_norm": 0.7879334092140198,
+      "learning_rate": 4.473947895791583e-05,
+      "loss": 0.3801,
+      "step": 536
+    },
+    {
+      "epoch": 2.148,
+      "grad_norm": 0.6276340484619141,
+      "learning_rate": 4.472945891783568e-05,
+      "loss": 0.365,
+      "step": 537
+    },
+    {
+      "epoch": 2.152,
+      "grad_norm": 0.7515552043914795,
+      "learning_rate": 4.471943887775552e-05,
+      "loss": 0.3993,
+      "step": 538
+    },
+    {
+      "epoch": 2.156,
+      "grad_norm": 0.6806215643882751,
+      "learning_rate": 4.470941883767535e-05,
+      "loss": 0.3654,
+      "step": 539
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.6749470829963684,
+      "learning_rate": 4.4699398797595194e-05,
+      "loss": 0.3565,
+      "step": 540
+    },
+    {
+      "epoch": 2.164,
+      "grad_norm": 0.8084760308265686,
+      "learning_rate": 4.4689378757515035e-05,
+      "loss": 0.4149,
+      "step": 541
+    },
+    {
+      "epoch": 2.168,
+      "grad_norm": 0.6896063089370728,
+      "learning_rate": 4.467935871743487e-05,
+      "loss": 0.3676,
+      "step": 542
+    },
+    {
+      "epoch": 2.172,
+      "grad_norm": 0.665913999080658,
+      "learning_rate": 4.466933867735471e-05,
+      "loss": 0.3514,
+      "step": 543
+    },
+    {
+      "epoch": 2.176,
+      "grad_norm": 0.6971928477287292,
+      "learning_rate": 4.465931863727455e-05,
+      "loss": 0.4026,
+      "step": 544
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 0.752924919128418,
+      "learning_rate": 4.464929859719439e-05,
+      "loss": 0.3677,
+      "step": 545
+    },
+    {
+      "epoch": 2.184,
+      "grad_norm": 0.6077944040298462,
+      "learning_rate": 4.463927855711423e-05,
+      "loss": 0.3712,
+      "step": 546
+    },
+    {
+      "epoch": 2.188,
+      "grad_norm": 0.6650569438934326,
+      "learning_rate": 4.462925851703407e-05,
+      "loss": 0.3999,
+      "step": 547
+    },
+    {
+      "epoch": 2.192,
+      "grad_norm": 0.6830900311470032,
+      "learning_rate": 4.461923847695391e-05,
+      "loss": 0.3617,
+      "step": 548
+    },
+    {
+      "epoch": 2.196,
+      "grad_norm": 0.6619176268577576,
+      "learning_rate": 4.460921843687375e-05,
+      "loss": 0.3707,
+      "step": 549
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 0.6596035361289978,
+      "learning_rate": 4.459919839679359e-05,
+      "loss": 0.4169,
+      "step": 550
+    },
+    {
+      "epoch": 2.204,
+      "grad_norm": 0.636611819267273,
+      "learning_rate": 4.458917835671343e-05,
+      "loss": 0.3806,
+      "step": 551
+    },
+    {
+      "epoch": 2.208,
+      "grad_norm": 0.6402906179428101,
+      "learning_rate": 4.457915831663327e-05,
+      "loss": 0.3406,
+      "step": 552
+    },
+    {
+      "epoch": 2.212,
+      "grad_norm": 0.6241163015365601,
+      "learning_rate": 4.456913827655311e-05,
+      "loss": 0.3814,
+      "step": 553
+    },
+    {
+      "epoch": 2.216,
+      "grad_norm": 0.73136305809021,
+      "learning_rate": 4.4559118236472945e-05,
+      "loss": 0.3599,
+      "step": 554
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 0.9286080002784729,
+      "learning_rate": 4.4549098196392786e-05,
+      "loss": 0.3629,
+      "step": 555
+    },
+    {
+      "epoch": 2.224,
+      "grad_norm": 0.6749750375747681,
+      "learning_rate": 4.453907815631263e-05,
+      "loss": 0.3829,
+      "step": 556
+    },
+    {
+      "epoch": 2.228,
+      "grad_norm": 0.6232475638389587,
+      "learning_rate": 4.452905811623247e-05,
+      "loss": 0.3816,
+      "step": 557
+    },
+    {
+      "epoch": 2.232,
+      "grad_norm": 0.6914759874343872,
+      "learning_rate": 4.451903807615231e-05,
+      "loss": 0.383,
+      "step": 558
+    },
+    {
+      "epoch": 2.2359999999999998,
+      "grad_norm": 0.786461591720581,
+      "learning_rate": 4.4509018036072144e-05,
+      "loss": 0.3644,
+      "step": 559
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 0.7495032548904419,
+      "learning_rate": 4.4498997995991986e-05,
+      "loss": 0.3818,
+      "step": 560
+    },
+    {
+      "epoch": 2.2439999999999998,
+      "grad_norm": 0.7083197236061096,
+      "learning_rate": 4.448897795591183e-05,
+      "loss": 0.3574,
+      "step": 561
+    },
+    {
+      "epoch": 2.248,
+      "grad_norm": 0.6997913718223572,
+      "learning_rate": 4.447895791583167e-05,
+      "loss": 0.3473,
+      "step": 562
+    },
+    {
+      "epoch": 2.252,
+      "grad_norm": 0.7174849510192871,
+      "learning_rate": 4.44689378757515e-05,
+      "loss": 0.331,
+      "step": 563
+    },
+    {
+      "epoch": 2.2560000000000002,
+      "grad_norm": 0.6463573575019836,
+      "learning_rate": 4.4458917835671344e-05,
+      "loss": 0.3907,
+      "step": 564
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.6492186784744263,
+      "learning_rate": 4.4448897795591185e-05,
+      "loss": 0.373,
+      "step": 565
+    },
+    {
+      "epoch": 2.2640000000000002,
+      "grad_norm": 0.7330689430236816,
+      "learning_rate": 4.4438877755511027e-05,
+      "loss": 0.3704,
+      "step": 566
+    },
+    {
+      "epoch": 2.268,
+      "grad_norm": 0.7165907621383667,
+      "learning_rate": 4.442885771543087e-05,
+      "loss": 0.4114,
+      "step": 567
+    },
+    {
+      "epoch": 2.2720000000000002,
+      "grad_norm": 0.7889212369918823,
+      "learning_rate": 4.44188376753507e-05,
+      "loss": 0.3837,
+      "step": 568
+    },
+    {
+      "epoch": 2.276,
+      "grad_norm": 0.7672508358955383,
+      "learning_rate": 4.4408817635270544e-05,
+      "loss": 0.3647,
+      "step": 569
+    },
+    {
+      "epoch": 2.2800000000000002,
+      "grad_norm": 0.7716801762580872,
+      "learning_rate": 4.4398797595190385e-05,
+      "loss": 0.4272,
+      "step": 570
+    },
+    {
+      "epoch": 2.284,
+      "grad_norm": 0.6814711689949036,
+      "learning_rate": 4.438877755511022e-05,
+      "loss": 0.4071,
+      "step": 571
+    },
+    {
+      "epoch": 2.288,
+      "grad_norm": 0.7873872518539429,
+      "learning_rate": 4.437875751503006e-05,
+      "loss": 0.3599,
+      "step": 572
+    },
+    {
+      "epoch": 2.292,
+      "grad_norm": 0.6841912865638733,
+      "learning_rate": 4.43687374749499e-05,
+      "loss": 0.3494,
+      "step": 573
+    },
+    {
+      "epoch": 2.296,
+      "grad_norm": 1.1426913738250732,
+      "learning_rate": 4.4358717434869737e-05,
+      "loss": 0.2038,
+      "step": 574
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.7427853345870972,
+      "learning_rate": 4.434869739478958e-05,
+      "loss": 0.4356,
+      "step": 575
+    },
+    {
+      "epoch": 2.304,
+      "grad_norm": 0.6541914939880371,
+      "learning_rate": 4.4338677354709426e-05,
+      "loss": 0.3992,
+      "step": 576
+    },
+    {
+      "epoch": 2.308,
+      "grad_norm": 0.6165962219238281,
+      "learning_rate": 4.432865731462926e-05,
+      "loss": 0.3442,
+      "step": 577
+    },
+    {
+      "epoch": 2.312,
+      "grad_norm": 0.6729843616485596,
+      "learning_rate": 4.43186372745491e-05,
+      "loss": 0.3985,
+      "step": 578
+    },
+    {
+      "epoch": 2.316,
+      "grad_norm": 0.735693097114563,
+      "learning_rate": 4.430861723446894e-05,
+      "loss": 0.3458,
+      "step": 579
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.6572169065475464,
+      "learning_rate": 4.429859719438878e-05,
+      "loss": 0.3533,
+      "step": 580
+    },
+    {
+      "epoch": 2.324,
+      "grad_norm": 0.7412426471710205,
+      "learning_rate": 4.428857715430862e-05,
+      "loss": 0.3494,
+      "step": 581
+    },
+    {
+      "epoch": 2.328,
+      "grad_norm": 0.7071847319602966,
+      "learning_rate": 4.427855711422846e-05,
+      "loss": 0.3647,
+      "step": 582
+    },
+    {
+      "epoch": 2.332,
+      "grad_norm": 0.8219819068908691,
+      "learning_rate": 4.4268537074148294e-05,
+      "loss": 0.4073,
+      "step": 583
+    },
+    {
+      "epoch": 2.336,
+      "grad_norm": 0.8309367895126343,
+      "learning_rate": 4.4258517034068136e-05,
+      "loss": 0.3826,
+      "step": 584
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 0.7580353617668152,
+      "learning_rate": 4.424849699398798e-05,
+      "loss": 0.3815,
+      "step": 585
+    },
+    {
+      "epoch": 2.344,
+      "grad_norm": 0.6513029336929321,
+      "learning_rate": 4.423847695390782e-05,
+      "loss": 0.3712,
+      "step": 586
+    },
+    {
+      "epoch": 2.348,
+      "grad_norm": 0.7023140788078308,
+      "learning_rate": 4.422845691382766e-05,
+      "loss": 0.3733,
+      "step": 587
+    },
+    {
+      "epoch": 2.352,
+      "grad_norm": 0.7600807547569275,
+      "learning_rate": 4.42184368737475e-05,
+      "loss": 0.3862,
+      "step": 588
+    },
+    {
+      "epoch": 2.356,
+      "grad_norm": 0.7764167189598083,
+      "learning_rate": 4.4208416833667335e-05,
+      "loss": 0.415,
+      "step": 589
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 0.8243778347969055,
+      "learning_rate": 4.419839679358718e-05,
+      "loss": 0.3776,
+      "step": 590
+    },
+    {
+      "epoch": 2.364,
+      "grad_norm": 0.7651914954185486,
+      "learning_rate": 4.418837675350702e-05,
+      "loss": 0.3871,
+      "step": 591
+    },
+    {
+      "epoch": 2.368,
+      "grad_norm": 0.7026492357254028,
+      "learning_rate": 4.417835671342685e-05,
+      "loss": 0.3352,
+      "step": 592
+    },
+    {
+      "epoch": 2.372,
+      "grad_norm": 0.6480612754821777,
+      "learning_rate": 4.4168336673346694e-05,
+      "loss": 0.3574,
+      "step": 593
+    },
+    {
+      "epoch": 2.376,
+      "grad_norm": 0.6376757621765137,
+      "learning_rate": 4.4158316633266535e-05,
+      "loss": 0.3865,
+      "step": 594
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 0.7384195327758789,
+      "learning_rate": 4.414829659318637e-05,
+      "loss": 0.346,
+      "step": 595
+    },
+    {
+      "epoch": 2.384,
+      "grad_norm": 0.7906630635261536,
+      "learning_rate": 4.413827655310622e-05,
+      "loss": 0.409,
+      "step": 596
+    },
+    {
+      "epoch": 2.388,
+      "grad_norm": 0.7145755290985107,
+      "learning_rate": 4.412825651302606e-05,
+      "loss": 0.3461,
+      "step": 597
+    },
+    {
+      "epoch": 2.392,
+      "grad_norm": 0.8087967038154602,
+      "learning_rate": 4.4118236472945893e-05,
+      "loss": 0.215,
+      "step": 598
+    },
+    {
+      "epoch": 2.396,
+      "grad_norm": 0.7960618138313293,
+      "learning_rate": 4.4108216432865735e-05,
+      "loss": 0.4122,
+      "step": 599
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 0.6827014684677124,
+      "learning_rate": 4.4098196392785576e-05,
+      "loss": 0.3964,
+      "step": 600
+    },
+    {
+      "epoch": 2.404,
+      "grad_norm": 0.7268480062484741,
+      "learning_rate": 4.408817635270541e-05,
+      "loss": 0.3557,
+      "step": 601
+    },
+    {
+      "epoch": 2.408,
+      "grad_norm": 0.7457350492477417,
+      "learning_rate": 4.407815631262525e-05,
+      "loss": 0.4288,
+      "step": 602
+    },
+    {
+      "epoch": 2.412,
+      "grad_norm": 0.7236207127571106,
+      "learning_rate": 4.406813627254509e-05,
+      "loss": 0.4079,
+      "step": 603
+    },
+    {
+      "epoch": 2.416,
+      "grad_norm": 0.6497883200645447,
+      "learning_rate": 4.405811623246493e-05,
+      "loss": 0.4348,
+      "step": 604
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 0.7279455065727234,
+      "learning_rate": 4.404809619238477e-05,
+      "loss": 0.3534,
+      "step": 605
+    },
+    {
+      "epoch": 2.424,
+      "grad_norm": 0.7461408972740173,
+      "learning_rate": 4.403807615230462e-05,
+      "loss": 0.3619,
+      "step": 606
+    },
+    {
+      "epoch": 2.428,
+      "grad_norm": 0.7696019411087036,
+      "learning_rate": 4.402805611222445e-05,
+      "loss": 0.3546,
+      "step": 607
+    },
+    {
+      "epoch": 2.432,
+      "grad_norm": 0.7570071816444397,
+      "learning_rate": 4.401803607214429e-05,
+      "loss": 0.3496,
+      "step": 608
+    },
+    {
+      "epoch": 2.436,
+      "grad_norm": 0.6898276805877686,
+      "learning_rate": 4.4008016032064134e-05,
+      "loss": 0.3789,
+      "step": 609
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.7434217929840088,
+      "learning_rate": 4.399799599198397e-05,
+      "loss": 0.3736,
+      "step": 610
+    },
+    {
+      "epoch": 2.444,
+      "grad_norm": 0.8145261406898499,
+      "learning_rate": 4.398797595190381e-05,
+      "loss": 0.3516,
+      "step": 611
+    },
+    {
+      "epoch": 2.448,
+      "grad_norm": 0.7646612524986267,
+      "learning_rate": 4.397795591182365e-05,
+      "loss": 0.3867,
+      "step": 612
+    },
+    {
+      "epoch": 2.452,
+      "grad_norm": 0.9545761942863464,
+      "learning_rate": 4.3967935871743486e-05,
+      "loss": 0.3563,
+      "step": 613
+    },
+    {
+      "epoch": 2.456,
+      "grad_norm": 0.8131189346313477,
+      "learning_rate": 4.395791583166333e-05,
+      "loss": 0.389,
+      "step": 614
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 0.7146084308624268,
+      "learning_rate": 4.394789579158317e-05,
+      "loss": 0.3727,
+      "step": 615
+    },
+    {
+      "epoch": 2.464,
+      "grad_norm": 0.7887970209121704,
+      "learning_rate": 4.393787575150301e-05,
+      "loss": 0.3596,
+      "step": 616
+    },
+    {
+      "epoch": 2.468,
+      "grad_norm": 0.7648036479949951,
+      "learning_rate": 4.392785571142285e-05,
+      "loss": 0.3388,
+      "step": 617
+    },
+    {
+      "epoch": 2.472,
+      "grad_norm": 0.80208420753479,
+      "learning_rate": 4.391783567134269e-05,
+      "loss": 0.3498,
+      "step": 618
+    },
+    {
+      "epoch": 2.476,
+      "grad_norm": 0.7335792183876038,
+      "learning_rate": 4.3907815631262526e-05,
+      "loss": 0.3678,
+      "step": 619
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 0.7437960505485535,
+      "learning_rate": 4.389779559118237e-05,
+      "loss": 0.3697,
+      "step": 620
+    },
+    {
+      "epoch": 2.484,
+      "grad_norm": 0.687762975692749,
+      "learning_rate": 4.388777555110221e-05,
+      "loss": 0.3292,
+      "step": 621
+    },
+    {
+      "epoch": 2.488,
+      "grad_norm": 0.7212290167808533,
+      "learning_rate": 4.3877755511022044e-05,
+      "loss": 0.3707,
+      "step": 622
+    },
+    {
+      "epoch": 2.492,
+      "grad_norm": 0.6740947365760803,
+      "learning_rate": 4.3867735470941885e-05,
+      "loss": 0.33,
+      "step": 623
+    },
+    {
+      "epoch": 2.496,
+      "grad_norm": 0.7523700594902039,
+      "learning_rate": 4.3857715430861726e-05,
+      "loss": 0.348,
+      "step": 624
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 0.8317469954490662,
+      "learning_rate": 4.384769539078157e-05,
+      "loss": 0.4042,
+      "step": 625
+    },
+    {
+      "epoch": 2.504,
+      "grad_norm": 0.8676090836524963,
+      "learning_rate": 4.383767535070141e-05,
+      "loss": 0.3682,
+      "step": 626
+    },
+    {
+      "epoch": 2.508,
+      "grad_norm": 0.7623687386512756,
+      "learning_rate": 4.382765531062124e-05,
+      "loss": 0.4103,
+      "step": 627
+    },
+    {
+      "epoch": 2.512,
+      "grad_norm": 0.756071925163269,
+      "learning_rate": 4.3817635270541084e-05,
+      "loss": 0.3673,
+      "step": 628
+    },
+    {
+      "epoch": 2.516,
+      "grad_norm": 0.7560201287269592,
+      "learning_rate": 4.3807615230460926e-05,
+      "loss": 0.3961,
+      "step": 629
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 0.8405265212059021,
+      "learning_rate": 4.379759519038076e-05,
+      "loss": 0.3891,
+      "step": 630
+    },
+    {
+      "epoch": 2.524,
+      "grad_norm": 0.7683942317962646,
+      "learning_rate": 4.37875751503006e-05,
+      "loss": 0.4056,
+      "step": 631
+    },
+    {
+      "epoch": 2.528,
+      "grad_norm": 0.7256227135658264,
+      "learning_rate": 4.377755511022044e-05,
+      "loss": 0.3555,
+      "step": 632
+    },
+    {
+      "epoch": 2.532,
+      "grad_norm": 0.8724226355552673,
+      "learning_rate": 4.3767535070140284e-05,
+      "loss": 0.4297,
+      "step": 633
+    },
+    {
+      "epoch": 2.536,
+      "grad_norm": 0.8041679263114929,
+      "learning_rate": 4.375751503006012e-05,
+      "loss": 0.3846,
+      "step": 634
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 0.7637845873832703,
+      "learning_rate": 4.374749498997997e-05,
+      "loss": 0.3578,
+      "step": 635
+    },
+    {
+      "epoch": 2.544,
+      "grad_norm": 0.7199541330337524,
+      "learning_rate": 4.37374749498998e-05,
+      "loss": 0.3484,
+      "step": 636
+    },
+    {
+      "epoch": 2.548,
+      "grad_norm": 0.7177748680114746,
+      "learning_rate": 4.372745490981964e-05,
+      "loss": 0.341,
+      "step": 637
+    },
+    {
+      "epoch": 2.552,
+      "grad_norm": 0.683948814868927,
+      "learning_rate": 4.3717434869739484e-05,
+      "loss": 0.3593,
+      "step": 638
+    },
+    {
+      "epoch": 2.556,
+      "grad_norm": 0.7331079840660095,
+      "learning_rate": 4.370741482965932e-05,
+      "loss": 0.3521,
+      "step": 639
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 0.7923868298530579,
+      "learning_rate": 4.369739478957916e-05,
+      "loss": 0.3775,
+      "step": 640
+    },
+    {
+      "epoch": 2.564,
+      "grad_norm": 0.8066992163658142,
+      "learning_rate": 4.3687374749499e-05,
+      "loss": 0.362,
+      "step": 641
+    },
+    {
+      "epoch": 2.568,
+      "grad_norm": 0.8344401121139526,
+      "learning_rate": 4.3677354709418835e-05,
+      "loss": 0.3877,
+      "step": 642
+    },
+    {
+      "epoch": 2.572,
+      "grad_norm": 0.7773652076721191,
+      "learning_rate": 4.3667334669338677e-05,
+      "loss": 0.3878,
+      "step": 643
+    },
+    {
+      "epoch": 2.576,
+      "grad_norm": 0.9228508472442627,
+      "learning_rate": 4.365731462925852e-05,
+      "loss": 0.3737,
+      "step": 644
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 0.7461867928504944,
+      "learning_rate": 4.364729458917836e-05,
+      "loss": 0.3422,
+      "step": 645
+    },
+    {
+      "epoch": 2.584,
+      "grad_norm": 0.7550526261329651,
+      "learning_rate": 4.36372745490982e-05,
+      "loss": 0.3748,
+      "step": 646
+    },
+    {
+      "epoch": 2.588,
+      "grad_norm": 0.7776896953582764,
+      "learning_rate": 4.362725450901804e-05,
+      "loss": 0.3629,
+      "step": 647
+    },
+    {
+      "epoch": 2.592,
+      "grad_norm": 0.8079218864440918,
+      "learning_rate": 4.3617234468937876e-05,
+      "loss": 0.352,
+      "step": 648
+    },
+    {
+      "epoch": 2.596,
+      "grad_norm": 0.7805841565132141,
+      "learning_rate": 4.360721442885772e-05,
+      "loss": 0.4112,
+      "step": 649
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 0.6444045305252075,
+      "learning_rate": 4.359719438877756e-05,
+      "loss": 0.2286,
+      "step": 650
+    },
+    {
+      "epoch": 2.604,
+      "grad_norm": 0.7929733395576477,
+      "learning_rate": 4.358717434869739e-05,
+      "loss": 0.368,
+      "step": 651
+    },
+    {
+      "epoch": 2.608,
+      "grad_norm": 0.7788984775543213,
+      "learning_rate": 4.3577154308617235e-05,
+      "loss": 0.3658,
+      "step": 652
+    },
+    {
+      "epoch": 2.612,
+      "grad_norm": 0.7166882753372192,
+      "learning_rate": 4.3567134268537076e-05,
+      "loss": 0.3472,
+      "step": 653
+    },
+    {
+      "epoch": 2.616,
+      "grad_norm": 0.7755756974220276,
+      "learning_rate": 4.355711422845691e-05,
+      "loss": 0.3748,
+      "step": 654
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 0.8172996640205383,
+      "learning_rate": 4.354709418837676e-05,
+      "loss": 0.3924,
+      "step": 655
+    },
+    {
+      "epoch": 2.624,
+      "grad_norm": 0.739886462688446,
+      "learning_rate": 4.35370741482966e-05,
+      "loss": 0.4263,
+      "step": 656
+    },
+    {
+      "epoch": 2.628,
+      "grad_norm": 0.8171461224555969,
+      "learning_rate": 4.3527054108216434e-05,
+      "loss": 0.3717,
+      "step": 657
+    },
+    {
+      "epoch": 2.632,
+      "grad_norm": 0.7296169400215149,
+      "learning_rate": 4.3517034068136275e-05,
+      "loss": 0.3753,
+      "step": 658
+    },
+    {
+      "epoch": 2.636,
+      "grad_norm": 0.7279816269874573,
+      "learning_rate": 4.350701402805612e-05,
+      "loss": 0.4018,
+      "step": 659
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 0.6948480606079102,
+      "learning_rate": 4.349699398797595e-05,
+      "loss": 0.3866,
+      "step": 660
+    },
+    {
+      "epoch": 2.644,
+      "grad_norm": 0.7570900321006775,
+      "learning_rate": 4.348697394789579e-05,
+      "loss": 0.3787,
+      "step": 661
+    },
+    {
+      "epoch": 2.648,
+      "grad_norm": 0.7140201330184937,
+      "learning_rate": 4.3476953907815634e-05,
+      "loss": 0.3161,
+      "step": 662
+    },
+    {
+      "epoch": 2.652,
+      "grad_norm": 0.7074480056762695,
+      "learning_rate": 4.346693386773547e-05,
+      "loss": 0.4135,
+      "step": 663
+    },
+    {
+      "epoch": 2.656,
+      "grad_norm": 0.7118886113166809,
+      "learning_rate": 4.345691382765531e-05,
+      "loss": 0.3456,
+      "step": 664
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 0.7889545559883118,
+      "learning_rate": 4.344689378757516e-05,
+      "loss": 0.3727,
+      "step": 665
+    },
+    {
+      "epoch": 2.664,
+      "grad_norm": 0.7217925786972046,
+      "learning_rate": 4.343687374749499e-05,
+      "loss": 0.3396,
+      "step": 666
+    },
+    {
+      "epoch": 2.668,
+      "grad_norm": 0.5060538053512573,
+      "learning_rate": 4.3426853707414833e-05,
+      "loss": 0.2113,
+      "step": 667
+    },
+    {
+      "epoch": 2.672,
+      "grad_norm": 0.8558411002159119,
+      "learning_rate": 4.3416833667334675e-05,
+      "loss": 0.4217,
+      "step": 668
+    },
+    {
+      "epoch": 2.676,
+      "grad_norm": 0.7091007828712463,
+      "learning_rate": 4.340681362725451e-05,
+      "loss": 0.3889,
+      "step": 669
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 0.8151986598968506,
+      "learning_rate": 4.339679358717435e-05,
+      "loss": 0.3808,
+      "step": 670
+    },
+    {
+      "epoch": 2.684,
+      "grad_norm": 0.8028424978256226,
+      "learning_rate": 4.338677354709419e-05,
+      "loss": 0.376,
+      "step": 671
+    },
+    {
+      "epoch": 2.6879999999999997,
+      "grad_norm": 0.7256678938865662,
+      "learning_rate": 4.3376753507014026e-05,
+      "loss": 0.3924,
+      "step": 672
+    },
+    {
+      "epoch": 2.692,
+      "grad_norm": 0.7175251245498657,
+      "learning_rate": 4.336673346693387e-05,
+      "loss": 0.3795,
+      "step": 673
+    },
+    {
+      "epoch": 2.6959999999999997,
+      "grad_norm": 0.8073795437812805,
+      "learning_rate": 4.335671342685371e-05,
+      "loss": 0.3812,
+      "step": 674
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 0.7120643258094788,
+      "learning_rate": 4.334669338677355e-05,
+      "loss": 0.3856,
+      "step": 675
+    },
+    {
+      "epoch": 2.7039999999999997,
+      "grad_norm": 0.6971157193183899,
+      "learning_rate": 4.333667334669339e-05,
+      "loss": 0.3223,
+      "step": 676
+    },
+    {
+      "epoch": 2.708,
+      "grad_norm": 0.7540895938873291,
+      "learning_rate": 4.332665330661323e-05,
+      "loss": 0.3582,
+      "step": 677
+    },
+    {
+      "epoch": 2.7119999999999997,
+      "grad_norm": 0.7137064337730408,
+      "learning_rate": 4.331663326653307e-05,
+      "loss": 0.3178,
+      "step": 678
+    },
+    {
+      "epoch": 2.716,
+      "grad_norm": 0.7786818742752075,
+      "learning_rate": 4.330661322645291e-05,
+      "loss": 0.38,
+      "step": 679
+    },
+    {
+      "epoch": 2.7199999999999998,
+      "grad_norm": 0.7963460087776184,
+      "learning_rate": 4.329659318637275e-05,
+      "loss": 0.3811,
+      "step": 680
+    },
+    {
+      "epoch": 2.724,
+      "grad_norm": 0.9932142496109009,
+      "learning_rate": 4.3286573146292584e-05,
+      "loss": 0.4158,
+      "step": 681
+    },
+    {
+      "epoch": 2.7279999999999998,
+      "grad_norm": 0.7535082101821899,
+      "learning_rate": 4.3276553106212426e-05,
+      "loss": 0.4385,
+      "step": 682
+    },
+    {
+      "epoch": 2.732,
+      "grad_norm": 0.6805514097213745,
+      "learning_rate": 4.326653306613227e-05,
+      "loss": 0.3881,
+      "step": 683
+    },
+    {
+      "epoch": 2.7359999999999998,
+      "grad_norm": 0.569618284702301,
+      "learning_rate": 4.325651302605211e-05,
+      "loss": 0.2244,
+      "step": 684
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 0.6559450030326843,
+      "learning_rate": 4.324649298597195e-05,
+      "loss": 0.3176,
+      "step": 685
+    },
+    {
+      "epoch": 2.7439999999999998,
+      "grad_norm": 0.7690601944923401,
+      "learning_rate": 4.3236472945891784e-05,
+      "loss": 0.3672,
+      "step": 686
+    },
+    {
+      "epoch": 2.748,
+      "grad_norm": 0.662957489490509,
+      "learning_rate": 4.3226452905811625e-05,
+      "loss": 0.3441,
+      "step": 687
+    },
+    {
+      "epoch": 2.752,
+      "grad_norm": 0.5383203029632568,
+      "learning_rate": 4.3216432865731467e-05,
+      "loss": 0.2406,
+      "step": 688
+    },
+    {
+      "epoch": 2.7560000000000002,
+      "grad_norm": 0.5156013369560242,
+      "learning_rate": 4.320641282565131e-05,
+      "loss": 0.1923,
+      "step": 689
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 0.798517644405365,
+      "learning_rate": 4.319639278557114e-05,
+      "loss": 0.3142,
+      "step": 690
+    },
+    {
+      "epoch": 2.7640000000000002,
+      "grad_norm": 0.7717877626419067,
+      "learning_rate": 4.3186372745490984e-05,
+      "loss": 0.3537,
+      "step": 691
+    },
+    {
+      "epoch": 2.768,
+      "grad_norm": 0.8068984150886536,
+      "learning_rate": 4.3176352705410825e-05,
+      "loss": 0.3579,
+      "step": 692
+    },
+    {
+      "epoch": 2.7720000000000002,
+      "grad_norm": 0.7683663368225098,
+      "learning_rate": 4.316633266533066e-05,
+      "loss": 0.3463,
+      "step": 693
+    },
+    {
+      "epoch": 2.776,
+      "grad_norm": 0.7648158669471741,
+      "learning_rate": 4.315631262525051e-05,
+      "loss": 0.3613,
+      "step": 694
+    },
+    {
+      "epoch": 2.7800000000000002,
+      "grad_norm": 0.6192529201507568,
+      "learning_rate": 4.314629258517034e-05,
+      "loss": 0.2317,
+      "step": 695
+    },
+    {
+      "epoch": 2.784,
+      "grad_norm": 0.7966995239257812,
+      "learning_rate": 4.313627254509018e-05,
+      "loss": 0.3571,
+      "step": 696
+    },
+    {
+      "epoch": 2.7880000000000003,
+      "grad_norm": 0.7972453236579895,
+      "learning_rate": 4.3126252505010025e-05,
+      "loss": 0.3334,
+      "step": 697
+    },
+    {
+      "epoch": 2.792,
+      "grad_norm": 0.7164406180381775,
+      "learning_rate": 4.311623246492986e-05,
+      "loss": 0.3394,
+      "step": 698
+    },
+    {
+      "epoch": 2.7960000000000003,
+      "grad_norm": 0.8824707865715027,
+      "learning_rate": 4.31062124248497e-05,
+      "loss": 0.4308,
+      "step": 699
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.7947747111320496,
+      "learning_rate": 4.309619238476954e-05,
+      "loss": 0.3586,
+      "step": 700
+    },
+    {
+      "epoch": 2.8040000000000003,
+      "grad_norm": 0.8112667202949524,
+      "learning_rate": 4.3086172344689376e-05,
+      "loss": 0.424,
+      "step": 701
+    },
+    {
+      "epoch": 2.808,
+      "grad_norm": 0.7361273765563965,
+      "learning_rate": 4.307615230460922e-05,
+      "loss": 0.357,
+      "step": 702
+    },
+    {
+      "epoch": 2.8120000000000003,
+      "grad_norm": 0.7715202569961548,
+      "learning_rate": 4.306613226452906e-05,
+      "loss": 0.4038,
+      "step": 703
+    },
+    {
+      "epoch": 2.816,
+      "grad_norm": 0.7197621464729309,
+      "learning_rate": 4.30561122244489e-05,
+      "loss": 0.3378,
+      "step": 704
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 0.7480655312538147,
+      "learning_rate": 4.304609218436874e-05,
+      "loss": 0.4022,
+      "step": 705
+    },
+    {
+      "epoch": 2.824,
+      "grad_norm": 0.6959013938903809,
+      "learning_rate": 4.303607214428858e-05,
+      "loss": 0.3704,
+      "step": 706
+    },
+    {
+      "epoch": 2.828,
+      "grad_norm": 0.7385444641113281,
+      "learning_rate": 4.302605210420842e-05,
+      "loss": 0.3868,
+      "step": 707
+    },
+    {
+      "epoch": 2.832,
+      "grad_norm": 0.7400097846984863,
+      "learning_rate": 4.301603206412826e-05,
+      "loss": 0.3851,
+      "step": 708
+    },
+    {
+      "epoch": 2.836,
+      "grad_norm": 0.7167990207672119,
+      "learning_rate": 4.30060120240481e-05,
+      "loss": 0.3516,
+      "step": 709
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 0.7471911907196045,
+      "learning_rate": 4.2995991983967934e-05,
+      "loss": 0.3891,
+      "step": 710
+    },
+    {
+      "epoch": 2.844,
+      "grad_norm": 0.670911431312561,
+      "learning_rate": 4.2985971943887775e-05,
+      "loss": 0.1749,
+      "step": 711
+    },
+    {
+      "epoch": 2.848,
+      "grad_norm": 0.7623646259307861,
+      "learning_rate": 4.297595190380762e-05,
+      "loss": 0.3528,
+      "step": 712
+    },
+    {
+      "epoch": 2.852,
+      "grad_norm": 0.7303867936134338,
+      "learning_rate": 4.296593186372745e-05,
+      "loss": 0.3443,
+      "step": 713
+    },
+    {
+      "epoch": 2.856,
+      "grad_norm": 0.8334202766418457,
+      "learning_rate": 4.29559118236473e-05,
+      "loss": 0.3797,
+      "step": 714
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 0.8006396889686584,
+      "learning_rate": 4.294589178356714e-05,
+      "loss": 0.3702,
+      "step": 715
+    },
+    {
+      "epoch": 2.864,
+      "grad_norm": 5.877392768859863,
+      "learning_rate": 4.2935871743486975e-05,
+      "loss": 0.3651,
+      "step": 716
+    },
+    {
+      "epoch": 2.868,
+      "grad_norm": 0.7039988040924072,
+      "learning_rate": 4.2925851703406816e-05,
+      "loss": 0.3728,
+      "step": 717
+    },
+    {
+      "epoch": 2.872,
+      "grad_norm": 0.777847409248352,
+      "learning_rate": 4.291583166332666e-05,
+      "loss": 0.3986,
+      "step": 718
+    },
+    {
+      "epoch": 2.876,
+      "grad_norm": 0.7246628999710083,
+      "learning_rate": 4.290581162324649e-05,
+      "loss": 0.3821,
+      "step": 719
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 0.773358941078186,
+      "learning_rate": 4.289579158316633e-05,
+      "loss": 0.3758,
+      "step": 720
+    },
+    {
+      "epoch": 2.884,
+      "grad_norm": 0.7800974249839783,
+      "learning_rate": 4.2885771543086175e-05,
+      "loss": 0.364,
+      "step": 721
+    },
+    {
+      "epoch": 2.888,
+      "grad_norm": 0.8116373419761658,
+      "learning_rate": 4.287575150300601e-05,
+      "loss": 0.457,
+      "step": 722
+    },
+    {
+      "epoch": 2.892,
+      "grad_norm": 0.6917723417282104,
+      "learning_rate": 4.286573146292585e-05,
+      "loss": 0.3549,
+      "step": 723
+    },
+    {
+      "epoch": 2.896,
+      "grad_norm": 0.8000931739807129,
+      "learning_rate": 4.28557114228457e-05,
+      "loss": 0.3616,
+      "step": 724
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 0.7841587066650391,
+      "learning_rate": 4.284569138276553e-05,
+      "loss": 0.4008,
+      "step": 725
+    },
+    {
+      "epoch": 2.904,
+      "grad_norm": 0.723857581615448,
+      "learning_rate": 4.2835671342685374e-05,
+      "loss": 0.3367,
+      "step": 726
+    },
+    {
+      "epoch": 2.908,
+      "grad_norm": 0.7573955059051514,
+      "learning_rate": 4.2825651302605216e-05,
+      "loss": 0.3473,
+      "step": 727
+    },
+    {
+      "epoch": 2.912,
+      "grad_norm": 0.6830270290374756,
+      "learning_rate": 4.281563126252505e-05,
+      "loss": 0.4192,
+      "step": 728
+    },
+    {
+      "epoch": 2.916,
+      "grad_norm": 0.7317824363708496,
+      "learning_rate": 4.280561122244489e-05,
+      "loss": 0.3519,
+      "step": 729
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 0.8307793736457825,
+      "learning_rate": 4.279559118236473e-05,
+      "loss": 0.385,
+      "step": 730
+    },
+    {
+      "epoch": 2.924,
+      "grad_norm": 0.5436768531799316,
+      "learning_rate": 4.278557114228457e-05,
+      "loss": 0.225,
+      "step": 731
+    },
+    {
+      "epoch": 2.928,
+      "grad_norm": 0.798338770866394,
+      "learning_rate": 4.277555110220441e-05,
+      "loss": 0.4136,
+      "step": 732
+    },
+    {
+      "epoch": 2.932,
+      "grad_norm": 0.7588498592376709,
+      "learning_rate": 4.2765531062124256e-05,
+      "loss": 0.3705,
+      "step": 733
+    },
+    {
+      "epoch": 2.936,
+      "grad_norm": 0.8589876890182495,
+      "learning_rate": 4.275551102204409e-05,
+      "loss": 0.3989,
+      "step": 734
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 0.6937852501869202,
+      "learning_rate": 4.274549098196393e-05,
+      "loss": 0.3804,
+      "step": 735
+    },
+    {
+      "epoch": 2.944,
+      "grad_norm": 0.8076114058494568,
+      "learning_rate": 4.2735470941883774e-05,
+      "loss": 0.3755,
+      "step": 736
+    },
+    {
+      "epoch": 2.948,
+      "grad_norm": 0.7671177983283997,
+      "learning_rate": 4.272545090180361e-05,
+      "loss": 0.3414,
+      "step": 737
+    },
+    {
+      "epoch": 2.952,
+      "grad_norm": 0.7929058074951172,
+      "learning_rate": 4.271543086172345e-05,
+      "loss": 0.4099,
+      "step": 738
+    },
+    {
+      "epoch": 2.956,
+      "grad_norm": 0.8027397394180298,
+      "learning_rate": 4.270541082164329e-05,
+      "loss": 0.3635,
+      "step": 739
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 0.7325429916381836,
+      "learning_rate": 4.2695390781563125e-05,
+      "loss": 0.3457,
+      "step": 740
+    },
+    {
+      "epoch": 2.964,
+      "grad_norm": 0.7555444240570068,
+      "learning_rate": 4.2685370741482966e-05,
+      "loss": 0.4271,
+      "step": 741
+    },
+    {
+      "epoch": 2.968,
+      "grad_norm": 0.6851345300674438,
+      "learning_rate": 4.267535070140281e-05,
+      "loss": 0.374,
+      "step": 742
+    },
+    {
+      "epoch": 2.972,
+      "grad_norm": 0.7643485069274902,
+      "learning_rate": 4.266533066132265e-05,
+      "loss": 0.3338,
+      "step": 743
+    },
+    {
+      "epoch": 2.976,
+      "grad_norm": 0.8755035996437073,
+      "learning_rate": 4.265531062124249e-05,
+      "loss": 0.3444,
+      "step": 744
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 0.6367295980453491,
+      "learning_rate": 4.264529058116233e-05,
+      "loss": 0.3667,
+      "step": 745
+    },
+    {
+      "epoch": 2.984,
+      "grad_norm": 0.8797886967658997,
+      "learning_rate": 4.2635270541082166e-05,
+      "loss": 0.4266,
+      "step": 746
+    },
+    {
+      "epoch": 2.988,
+      "grad_norm": 0.6971140503883362,
+      "learning_rate": 4.262525050100201e-05,
+      "loss": 0.337,
+      "step": 747
+    },
+    {
+      "epoch": 2.992,
+      "grad_norm": 0.751207709312439,
+      "learning_rate": 4.261523046092185e-05,
+      "loss": 0.3907,
+      "step": 748
+    },
+    {
+      "epoch": 2.996,
+      "grad_norm": 0.7211257219314575,
+      "learning_rate": 4.260521042084168e-05,
+      "loss": 0.396,
+      "step": 749
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.9239338636398315,
+      "learning_rate": 4.2595190380761524e-05,
+      "loss": 0.4502,
+      "step": 750
+    },
+    {
+      "epoch": 3.004,
+      "grad_norm": 0.779204249382019,
+      "learning_rate": 4.2585170340681366e-05,
+      "loss": 0.2879,
+      "step": 751
+    },
+    {
+      "epoch": 3.008,
+      "grad_norm": 0.6356750726699829,
+      "learning_rate": 4.25751503006012e-05,
+      "loss": 0.2986,
+      "step": 752
+    },
+    {
+      "epoch": 3.012,
+      "grad_norm": 0.8113148212432861,
+      "learning_rate": 4.256513026052105e-05,
+      "loss": 0.3308,
+      "step": 753
+    },
+    {
+      "epoch": 3.016,
+      "grad_norm": 0.8562130331993103,
+      "learning_rate": 4.255511022044088e-05,
+      "loss": 0.3309,
+      "step": 754
+    },
+    {
+      "epoch": 3.02,
+      "grad_norm": 0.7852639555931091,
+      "learning_rate": 4.2545090180360724e-05,
+      "loss": 0.253,
+      "step": 755
+    },
+    {
+      "epoch": 3.024,
+      "grad_norm": 0.83427894115448,
+      "learning_rate": 4.2535070140280565e-05,
+      "loss": 0.2877,
+      "step": 756
+    },
+    {
+      "epoch": 3.028,
+      "grad_norm": 0.8282837867736816,
+      "learning_rate": 4.25250501002004e-05,
+      "loss": 0.304,
+      "step": 757
+    },
+    {
+      "epoch": 3.032,
+      "grad_norm": 0.895320475101471,
+      "learning_rate": 4.251503006012024e-05,
+      "loss": 0.3018,
+      "step": 758
+    },
+    {
+      "epoch": 3.036,
+      "grad_norm": 0.9096167683601379,
+      "learning_rate": 4.250501002004008e-05,
+      "loss": 0.2782,
+      "step": 759
+    },
+    {
+      "epoch": 3.04,
+      "grad_norm": 0.9886460304260254,
+      "learning_rate": 4.2494989979959924e-05,
+      "loss": 0.2964,
+      "step": 760
+    },
+    {
+      "epoch": 3.044,
+      "grad_norm": 1.022255778312683,
+      "learning_rate": 4.248496993987976e-05,
+      "loss": 0.3164,
+      "step": 761
+    },
+    {
+      "epoch": 3.048,
+      "grad_norm": 1.0864485502243042,
+      "learning_rate": 4.24749498997996e-05,
+      "loss": 0.2808,
+      "step": 762
+    },
+    {
+      "epoch": 3.052,
+      "grad_norm": 1.1385424137115479,
+      "learning_rate": 4.246492985971944e-05,
+      "loss": 0.3147,
+      "step": 763
+    },
+    {
+      "epoch": 3.056,
+      "grad_norm": 1.4085267782211304,
+      "learning_rate": 4.245490981963928e-05,
+      "loss": 0.3169,
+      "step": 764
+    },
+    {
+      "epoch": 3.06,
+      "grad_norm": 0.9816059470176697,
+      "learning_rate": 4.244488977955912e-05,
+      "loss": 0.2724,
+      "step": 765
+    },
+    {
+      "epoch": 3.064,
+      "grad_norm": 0.9588344693183899,
+      "learning_rate": 4.243486973947896e-05,
+      "loss": 0.2695,
+      "step": 766
+    },
+    {
+      "epoch": 3.068,
+      "grad_norm": 0.9727541208267212,
+      "learning_rate": 4.24248496993988e-05,
+      "loss": 0.307,
+      "step": 767
+    },
+    {
+      "epoch": 3.072,
+      "grad_norm": 0.9045794010162354,
+      "learning_rate": 4.241482965931864e-05,
+      "loss": 0.281,
+      "step": 768
+    },
+    {
+      "epoch": 3.076,
+      "grad_norm": 0.8429794907569885,
+      "learning_rate": 4.2404809619238475e-05,
+      "loss": 0.2693,
+      "step": 769
+    },
+    {
+      "epoch": 3.08,
+      "grad_norm": 0.9454277157783508,
+      "learning_rate": 4.2394789579158316e-05,
+      "loss": 0.2927,
+      "step": 770
+    },
+    {
+      "epoch": 3.084,
+      "grad_norm": 0.8746525645256042,
+      "learning_rate": 4.238476953907816e-05,
+      "loss": 0.3275,
+      "step": 771
+    },
+    {
+      "epoch": 3.088,
+      "grad_norm": 1.0423359870910645,
+      "learning_rate": 4.2374749498998e-05,
+      "loss": 0.2617,
+      "step": 772
+    },
+    {
+      "epoch": 3.092,
+      "grad_norm": 0.9890905022621155,
+      "learning_rate": 4.236472945891784e-05,
+      "loss": 0.268,
+      "step": 773
+    },
+    {
+      "epoch": 3.096,
+      "grad_norm": 0.9057809114456177,
+      "learning_rate": 4.235470941883768e-05,
+      "loss": 0.2951,
+      "step": 774
+    },
+    {
+      "epoch": 3.1,
+      "grad_norm": 0.841378927230835,
+      "learning_rate": 4.2344689378757516e-05,
+      "loss": 0.251,
+      "step": 775
+    },
+    {
+      "epoch": 3.104,
+      "grad_norm": 0.926152229309082,
+      "learning_rate": 4.233466933867736e-05,
+      "loss": 0.2939,
+      "step": 776
+    },
+    {
+      "epoch": 3.108,
+      "grad_norm": 1.010184407234192,
+      "learning_rate": 4.23246492985972e-05,
+      "loss": 0.2795,
+      "step": 777
+    },
+    {
+      "epoch": 3.112,
+      "grad_norm": 1.0386322736740112,
+      "learning_rate": 4.231462925851703e-05,
+      "loss": 0.3257,
+      "step": 778
+    },
+    {
+      "epoch": 3.116,
+      "grad_norm": 0.9416765570640564,
+      "learning_rate": 4.2304609218436874e-05,
+      "loss": 0.2857,
+      "step": 779
+    },
+    {
+      "epoch": 3.12,
+      "grad_norm": 0.6299401521682739,
+      "learning_rate": 4.2294589178356715e-05,
+      "loss": 0.1485,
+      "step": 780
+    },
+    {
+      "epoch": 3.124,
+      "grad_norm": 0.8813037872314453,
+      "learning_rate": 4.228456913827655e-05,
+      "loss": 0.2762,
+      "step": 781
+    },
+    {
+      "epoch": 3.128,
+      "grad_norm": 0.9580820798873901,
+      "learning_rate": 4.227454909819639e-05,
+      "loss": 0.3148,
+      "step": 782
+    },
+    {
+      "epoch": 3.132,
+      "grad_norm": 0.9623346924781799,
+      "learning_rate": 4.226452905811624e-05,
+      "loss": 0.3172,
+      "step": 783
+    },
+    {
+      "epoch": 3.136,
+      "grad_norm": 0.9738514423370361,
+      "learning_rate": 4.2254509018036074e-05,
+      "loss": 0.2936,
+      "step": 784
+    },
+    {
+      "epoch": 3.14,
+      "grad_norm": 0.9638810753822327,
+      "learning_rate": 4.2244488977955915e-05,
+      "loss": 0.2686,
+      "step": 785
+    },
+    {
+      "epoch": 3.144,
+      "grad_norm": 1.004857063293457,
+      "learning_rate": 4.2234468937875756e-05,
+      "loss": 0.3125,
+      "step": 786
+    },
+    {
+      "epoch": 3.148,
+      "grad_norm": 0.9582427740097046,
+      "learning_rate": 4.222444889779559e-05,
+      "loss": 0.2873,
+      "step": 787
+    },
+    {
+      "epoch": 3.152,
+      "grad_norm": 0.9714578986167908,
+      "learning_rate": 4.221442885771543e-05,
+      "loss": 0.3185,
+      "step": 788
+    },
+    {
+      "epoch": 3.156,
+      "grad_norm": 0.9590511322021484,
+      "learning_rate": 4.2204408817635273e-05,
+      "loss": 0.2962,
+      "step": 789
+    },
+    {
+      "epoch": 3.16,
+      "grad_norm": 0.8241173624992371,
+      "learning_rate": 4.219438877755511e-05,
+      "loss": 0.273,
+      "step": 790
+    },
+    {
+      "epoch": 3.164,
+      "grad_norm": 0.9807001352310181,
+      "learning_rate": 4.218436873747495e-05,
+      "loss": 0.2778,
+      "step": 791
+    },
+    {
+      "epoch": 3.168,
+      "grad_norm": 1.012277364730835,
+      "learning_rate": 4.21743486973948e-05,
+      "loss": 0.2907,
+      "step": 792
+    },
+    {
+      "epoch": 3.172,
+      "grad_norm": 0.7485466003417969,
+      "learning_rate": 4.216432865731463e-05,
+      "loss": 0.1743,
+      "step": 793
+    },
+    {
+      "epoch": 3.176,
+      "grad_norm": 1.0279030799865723,
+      "learning_rate": 4.215430861723447e-05,
+      "loss": 0.2856,
+      "step": 794
+    },
+    {
+      "epoch": 3.18,
+      "grad_norm": 1.0178649425506592,
+      "learning_rate": 4.2144288577154314e-05,
+      "loss": 0.2993,
+      "step": 795
+    },
+    {
+      "epoch": 3.184,
+      "grad_norm": 0.8605381846427917,
+      "learning_rate": 4.213426853707415e-05,
+      "loss": 0.2982,
+      "step": 796
+    },
+    {
+      "epoch": 3.188,
+      "grad_norm": 1.0019651651382446,
+      "learning_rate": 4.212424849699399e-05,
+      "loss": 0.2708,
+      "step": 797
+    },
+    {
+      "epoch": 3.192,
+      "grad_norm": 0.8550509810447693,
+      "learning_rate": 4.211422845691383e-05,
+      "loss": 0.2795,
+      "step": 798
+    },
+    {
+      "epoch": 3.196,
+      "grad_norm": 0.9066852927207947,
+      "learning_rate": 4.2104208416833666e-05,
+      "loss": 0.294,
+      "step": 799
+    },
+    {
+      "epoch": 3.2,
+      "grad_norm": 1.0385713577270508,
+      "learning_rate": 4.209418837675351e-05,
+      "loss": 0.3046,
+      "step": 800
+    },
+    {
+      "epoch": 3.204,
+      "grad_norm": 0.958522379398346,
+      "learning_rate": 4.208416833667335e-05,
+      "loss": 0.3016,
+      "step": 801
+    },
+    {
+      "epoch": 3.208,
+      "grad_norm": 1.0640708208084106,
+      "learning_rate": 4.207414829659319e-05,
+      "loss": 0.2636,
+      "step": 802
+    },
+    {
+      "epoch": 3.212,
+      "grad_norm": 0.9335046410560608,
+      "learning_rate": 4.206412825651303e-05,
+      "loss": 0.2668,
+      "step": 803
+    },
+    {
+      "epoch": 3.216,
+      "grad_norm": 1.0041309595108032,
+      "learning_rate": 4.205410821643287e-05,
+      "loss": 0.2774,
+      "step": 804
+    },
+    {
+      "epoch": 3.22,
+      "grad_norm": 0.8523897528648376,
+      "learning_rate": 4.204408817635271e-05,
+      "loss": 0.2638,
+      "step": 805
+    },
+    {
+      "epoch": 3.224,
+      "grad_norm": 0.9941080212593079,
+      "learning_rate": 4.203406813627255e-05,
+      "loss": 0.2988,
+      "step": 806
+    },
+    {
+      "epoch": 3.228,
+      "grad_norm": 1.0812398195266724,
+      "learning_rate": 4.202404809619239e-05,
+      "loss": 0.277,
+      "step": 807
+    },
+    {
+      "epoch": 3.232,
+      "grad_norm": 0.9684290289878845,
+      "learning_rate": 4.2014028056112224e-05,
+      "loss": 0.2944,
+      "step": 808
+    },
+    {
+      "epoch": 3.2359999999999998,
+      "grad_norm": 1.021715521812439,
+      "learning_rate": 4.2004008016032065e-05,
+      "loss": 0.2548,
+      "step": 809
+    },
+    {
+      "epoch": 3.24,
+      "grad_norm": 1.052801489830017,
+      "learning_rate": 4.1993987975951907e-05,
+      "loss": 0.2885,
+      "step": 810
+    },
+    {
+      "epoch": 3.2439999999999998,
+      "grad_norm": 1.0573465824127197,
+      "learning_rate": 4.198396793587174e-05,
+      "loss": 0.3444,
+      "step": 811
+    },
+    {
+      "epoch": 3.248,
+      "grad_norm": 0.7046388387680054,
+      "learning_rate": 4.197394789579159e-05,
+      "loss": 0.134,
+      "step": 812
+    },
+    {
+      "epoch": 3.252,
+      "grad_norm": 1.030924677848816,
+      "learning_rate": 4.1963927855711424e-05,
+      "loss": 0.3304,
+      "step": 813
+    },
+    {
+      "epoch": 3.2560000000000002,
+      "grad_norm": 0.9746088981628418,
+      "learning_rate": 4.1953907815631265e-05,
+      "loss": 0.286,
+      "step": 814
+    },
+    {
+      "epoch": 3.26,
+      "grad_norm": 0.9044510722160339,
+      "learning_rate": 4.1943887775551106e-05,
+      "loss": 0.2653,
+      "step": 815
+    },
+    {
+      "epoch": 3.2640000000000002,
+      "grad_norm": 1.0177682638168335,
+      "learning_rate": 4.193386773547095e-05,
+      "loss": 0.3103,
+      "step": 816
+    },
+    {
+      "epoch": 3.268,
+      "grad_norm": 1.0040210485458374,
+      "learning_rate": 4.192384769539078e-05,
+      "loss": 0.2804,
+      "step": 817
+    },
+    {
+      "epoch": 3.2720000000000002,
+      "grad_norm": 0.9195308685302734,
+      "learning_rate": 4.191382765531062e-05,
+      "loss": 0.3041,
+      "step": 818
+    },
+    {
+      "epoch": 3.276,
+      "grad_norm": 1.0614639520645142,
+      "learning_rate": 4.1903807615230465e-05,
+      "loss": 0.2694,
+      "step": 819
+    },
+    {
+      "epoch": 3.2800000000000002,
+      "grad_norm": 0.9703024625778198,
+      "learning_rate": 4.18937875751503e-05,
+      "loss": 0.3042,
+      "step": 820
+    },
+    {
+      "epoch": 3.284,
+      "grad_norm": 0.7372400760650635,
+      "learning_rate": 4.188376753507014e-05,
+      "loss": 0.1776,
+      "step": 821
+    },
+    {
+      "epoch": 3.288,
+      "grad_norm": 0.9801708459854126,
+      "learning_rate": 4.187374749498998e-05,
+      "loss": 0.3445,
+      "step": 822
+    },
+    {
+      "epoch": 3.292,
+      "grad_norm": 1.0228462219238281,
+      "learning_rate": 4.186372745490982e-05,
+      "loss": 0.2965,
+      "step": 823
+    },
+    {
+      "epoch": 3.296,
+      "grad_norm": 0.9982706904411316,
+      "learning_rate": 4.1853707414829664e-05,
+      "loss": 0.2613,
+      "step": 824
+    },
+    {
+      "epoch": 3.3,
+      "grad_norm": 1.1089848279953003,
+      "learning_rate": 4.18436873747495e-05,
+      "loss": 0.3156,
+      "step": 825
+    },
+    {
+      "epoch": 3.304,
+      "grad_norm": 0.9240087866783142,
+      "learning_rate": 4.183366733466934e-05,
+      "loss": 0.2621,
+      "step": 826
+    },
+    {
+      "epoch": 3.308,
+      "grad_norm": 0.9882985353469849,
+      "learning_rate": 4.182364729458918e-05,
+      "loss": 0.2766,
+      "step": 827
+    },
+    {
+      "epoch": 3.312,
+      "grad_norm": 1.092639446258545,
+      "learning_rate": 4.1813627254509016e-05,
+      "loss": 0.3167,
+      "step": 828
+    },
+    {
+      "epoch": 3.316,
+      "grad_norm": 1.0920966863632202,
+      "learning_rate": 4.180360721442886e-05,
+      "loss": 0.3217,
+      "step": 829
+    },
+    {
+      "epoch": 3.32,
+      "grad_norm": 1.0740662813186646,
+      "learning_rate": 4.17935871743487e-05,
+      "loss": 0.2861,
+      "step": 830
+    },
+    {
+      "epoch": 3.324,
+      "grad_norm": 0.969103991985321,
+      "learning_rate": 4.178356713426854e-05,
+      "loss": 0.3083,
+      "step": 831
+    },
+    {
+      "epoch": 3.328,
+      "grad_norm": 1.1013503074645996,
+      "learning_rate": 4.177354709418838e-05,
+      "loss": 0.3094,
+      "step": 832
+    },
+    {
+      "epoch": 3.332,
+      "grad_norm": 0.9775644540786743,
+      "learning_rate": 4.176352705410822e-05,
+      "loss": 0.2919,
+      "step": 833
+    },
+    {
+      "epoch": 3.336,
+      "grad_norm": 0.9920254349708557,
+      "learning_rate": 4.175350701402806e-05,
+      "loss": 0.2904,
+      "step": 834
+    },
+    {
+      "epoch": 3.34,
+      "grad_norm": 1.0171326398849487,
+      "learning_rate": 4.17434869739479e-05,
+      "loss": 0.2888,
+      "step": 835
+    },
+    {
+      "epoch": 3.344,
+      "grad_norm": 0.8502336144447327,
+      "learning_rate": 4.173346693386774e-05,
+      "loss": 0.3193,
+      "step": 836
+    },
+    {
+      "epoch": 3.348,
+      "grad_norm": 0.9221425652503967,
+      "learning_rate": 4.1723446893787574e-05,
+      "loss": 0.3163,
+      "step": 837
+    },
+    {
+      "epoch": 3.352,
+      "grad_norm": 0.7386908531188965,
+      "learning_rate": 4.1713426853707415e-05,
+      "loss": 0.1987,
+      "step": 838
+    },
+    {
+      "epoch": 3.356,
+      "grad_norm": 0.9897122383117676,
+      "learning_rate": 4.1703406813627256e-05,
+      "loss": 0.3081,
+      "step": 839
+    },
+    {
+      "epoch": 3.36,
+      "grad_norm": 1.0372047424316406,
+      "learning_rate": 4.169338677354709e-05,
+      "loss": 0.3217,
+      "step": 840
+    },
+    {
+      "epoch": 3.364,
+      "grad_norm": 1.108982801437378,
+      "learning_rate": 4.168336673346693e-05,
+      "loss": 0.3282,
+      "step": 841
+    },
+    {
+      "epoch": 3.368,
+      "grad_norm": 0.9839752316474915,
+      "learning_rate": 4.167334669338678e-05,
+      "loss": 0.2829,
+      "step": 842
+    },
+    {
+      "epoch": 3.372,
+      "grad_norm": 0.9582000970840454,
+      "learning_rate": 4.1663326653306615e-05,
+      "loss": 0.3538,
+      "step": 843
+    },
+    {
+      "epoch": 3.376,
+      "grad_norm": 1.0338131189346313,
+      "learning_rate": 4.1653306613226456e-05,
+      "loss": 0.3183,
+      "step": 844
+    },
+    {
+      "epoch": 3.38,
+      "grad_norm": 0.969628632068634,
+      "learning_rate": 4.16432865731463e-05,
+      "loss": 0.2753,
+      "step": 845
+    },
+    {
+      "epoch": 3.384,
+      "grad_norm": 0.9594615697860718,
+      "learning_rate": 4.163326653306613e-05,
+      "loss": 0.3348,
+      "step": 846
+    },
+    {
+      "epoch": 3.388,
+      "grad_norm": 1.0317904949188232,
+      "learning_rate": 4.162324649298597e-05,
+      "loss": 0.3075,
+      "step": 847
+    },
+    {
+      "epoch": 3.392,
+      "grad_norm": 1.00275719165802,
+      "learning_rate": 4.1613226452905814e-05,
+      "loss": 0.2883,
+      "step": 848
+    },
+    {
+      "epoch": 3.396,
+      "grad_norm": 0.902736246585846,
+      "learning_rate": 4.160320641282565e-05,
+      "loss": 0.293,
+      "step": 849
+    },
+    {
+      "epoch": 3.4,
+      "grad_norm": 0.9426193833351135,
+      "learning_rate": 4.159318637274549e-05,
+      "loss": 0.3869,
+      "step": 850
+    },
+    {
+      "epoch": 3.404,
+      "grad_norm": 0.8548994660377502,
+      "learning_rate": 4.158316633266534e-05,
+      "loss": 0.2635,
+      "step": 851
+    },
+    {
+      "epoch": 3.408,
+      "grad_norm": 0.9464280605316162,
+      "learning_rate": 4.157314629258517e-05,
+      "loss": 0.3005,
+      "step": 852
+    },
+    {
+      "epoch": 3.412,
+      "grad_norm": 0.5138210654258728,
+      "learning_rate": 4.1563126252505014e-05,
+      "loss": 0.101,
+      "step": 853
+    },
+    {
+      "epoch": 3.416,
+      "grad_norm": 0.8847005367279053,
+      "learning_rate": 4.1553106212424855e-05,
+      "loss": 0.2996,
+      "step": 854
+    },
+    {
+      "epoch": 3.42,
+      "grad_norm": 1.042490005493164,
+      "learning_rate": 4.154308617234469e-05,
+      "loss": 0.2794,
+      "step": 855
+    },
+    {
+      "epoch": 3.424,
+      "grad_norm": 1.0797277688980103,
+      "learning_rate": 4.153306613226453e-05,
+      "loss": 0.2884,
+      "step": 856
+    },
+    {
+      "epoch": 3.428,
+      "grad_norm": 1.097922921180725,
+      "learning_rate": 4.152304609218437e-05,
+      "loss": 0.2846,
+      "step": 857
+    },
+    {
+      "epoch": 3.432,
+      "grad_norm": 1.0516830682754517,
+      "learning_rate": 4.151302605210421e-05,
+      "loss": 0.3018,
+      "step": 858
+    },
+    {
+      "epoch": 3.436,
+      "grad_norm": 1.0133203268051147,
+      "learning_rate": 4.150300601202405e-05,
+      "loss": 0.2871,
+      "step": 859
+    },
+    {
+      "epoch": 3.44,
+      "grad_norm": 1.0574767589569092,
+      "learning_rate": 4.149298597194389e-05,
+      "loss": 0.3106,
+      "step": 860
+    },
+    {
+      "epoch": 3.444,
+      "grad_norm": 0.9849914908409119,
+      "learning_rate": 4.148296593186373e-05,
+      "loss": 0.3729,
+      "step": 861
+    },
+    {
+      "epoch": 3.448,
+      "grad_norm": 0.8353570103645325,
+      "learning_rate": 4.147294589178357e-05,
+      "loss": 0.2873,
+      "step": 862
+    },
+    {
+      "epoch": 3.452,
+      "grad_norm": 1.0005288124084473,
+      "learning_rate": 4.146292585170341e-05,
+      "loss": 0.309,
+      "step": 863
+    },
+    {
+      "epoch": 3.456,
+      "grad_norm": 1.139251708984375,
+      "learning_rate": 4.145290581162325e-05,
+      "loss": 0.3101,
+      "step": 864
+    },
+    {
+      "epoch": 3.46,
+      "grad_norm": 0.9400432109832764,
+      "learning_rate": 4.144288577154309e-05,
+      "loss": 0.3201,
+      "step": 865
+    },
+    {
+      "epoch": 3.464,
+      "grad_norm": 1.0056957006454468,
+      "learning_rate": 4.143286573146293e-05,
+      "loss": 0.2908,
+      "step": 866
+    },
+    {
+      "epoch": 3.468,
+      "grad_norm": 1.0107009410858154,
+      "learning_rate": 4.1422845691382765e-05,
+      "loss": 0.2781,
+      "step": 867
+    },
+    {
+      "epoch": 3.472,
+      "grad_norm": 0.8538976311683655,
+      "learning_rate": 4.1412825651302606e-05,
+      "loss": 0.271,
+      "step": 868
+    },
+    {
+      "epoch": 3.476,
+      "grad_norm": 0.8999854922294617,
+      "learning_rate": 4.140280561122245e-05,
+      "loss": 0.3038,
+      "step": 869
+    },
+    {
+      "epoch": 3.48,
+      "grad_norm": 0.8919850587844849,
+      "learning_rate": 4.139278557114228e-05,
+      "loss": 0.2953,
+      "step": 870
+    },
+    {
+      "epoch": 3.484,
+      "grad_norm": 0.9376264810562134,
+      "learning_rate": 4.138276553106213e-05,
+      "loss": 0.309,
+      "step": 871
+    },
+    {
+      "epoch": 3.488,
+      "grad_norm": 1.0939548015594482,
+      "learning_rate": 4.137274549098197e-05,
+      "loss": 0.3156,
+      "step": 872
+    },
+    {
+      "epoch": 3.492,
+      "grad_norm": 0.9430792927742004,
+      "learning_rate": 4.1362725450901806e-05,
+      "loss": 0.3003,
+      "step": 873
+    },
+    {
+      "epoch": 3.496,
+      "grad_norm": 0.9949555397033691,
+      "learning_rate": 4.135270541082165e-05,
+      "loss": 0.2998,
+      "step": 874
+    },
+    {
+      "epoch": 3.5,
+      "grad_norm": 0.9819671511650085,
+      "learning_rate": 4.134268537074149e-05,
+      "loss": 0.318,
+      "step": 875
+    },
+    {
+      "epoch": 3.504,
+      "grad_norm": 0.9571810960769653,
+      "learning_rate": 4.133266533066132e-05,
+      "loss": 0.3308,
+      "step": 876
+    },
+    {
+      "epoch": 3.508,
+      "grad_norm": 0.9577171206474304,
+      "learning_rate": 4.1322645290581164e-05,
+      "loss": 0.3261,
+      "step": 877
+    },
+    {
+      "epoch": 3.512,
+      "grad_norm": 0.9510439038276672,
+      "learning_rate": 4.1312625250501005e-05,
+      "loss": 0.2901,
+      "step": 878
+    },
+    {
+      "epoch": 3.516,
+      "grad_norm": 0.9860422015190125,
+      "learning_rate": 4.130260521042084e-05,
+      "loss": 0.2799,
+      "step": 879
+    },
+    {
+      "epoch": 3.52,
+      "grad_norm": 1.0223884582519531,
+      "learning_rate": 4.129258517034068e-05,
+      "loss": 0.3264,
+      "step": 880
+    },
+    {
+      "epoch": 3.524,
+      "grad_norm": 1.0917876958847046,
+      "learning_rate": 4.128256513026052e-05,
+      "loss": 0.2869,
+      "step": 881
+    },
+    {
+      "epoch": 3.528,
+      "grad_norm": 1.0941041707992554,
+      "learning_rate": 4.1272545090180364e-05,
+      "loss": 0.3296,
+      "step": 882
+    },
+    {
+      "epoch": 3.532,
+      "grad_norm": 0.9982560276985168,
+      "learning_rate": 4.1262525050100205e-05,
+      "loss": 0.3123,
+      "step": 883
+    },
+    {
+      "epoch": 3.536,
+      "grad_norm": 1.1089354753494263,
+      "learning_rate": 4.125250501002004e-05,
+      "loss": 0.3245,
+      "step": 884
+    },
+    {
+      "epoch": 3.54,
+      "grad_norm": 0.9130812883377075,
+      "learning_rate": 4.124248496993988e-05,
+      "loss": 0.265,
+      "step": 885
+    },
+    {
+      "epoch": 3.544,
+      "grad_norm": 0.9193147420883179,
+      "learning_rate": 4.123246492985972e-05,
+      "loss": 0.2947,
+      "step": 886
+    },
+    {
+      "epoch": 3.548,
+      "grad_norm": 0.908898651599884,
+      "learning_rate": 4.122244488977956e-05,
+      "loss": 0.3173,
+      "step": 887
+    },
+    {
+      "epoch": 3.552,
+      "grad_norm": 1.038451910018921,
+      "learning_rate": 4.12124248496994e-05,
+      "loss": 0.3317,
+      "step": 888
+    },
+    {
+      "epoch": 3.556,
+      "grad_norm": 1.0614579916000366,
+      "learning_rate": 4.120240480961924e-05,
+      "loss": 0.3206,
+      "step": 889
+    },
+    {
+      "epoch": 3.56,
+      "grad_norm": 1.0630053281784058,
+      "learning_rate": 4.119238476953908e-05,
+      "loss": 0.3096,
+      "step": 890
+    },
+    {
+      "epoch": 3.564,
+      "grad_norm": 0.9061043858528137,
+      "learning_rate": 4.118236472945892e-05,
+      "loss": 0.3105,
+      "step": 891
+    },
+    {
+      "epoch": 3.568,
+      "grad_norm": 0.9253777265548706,
+      "learning_rate": 4.117234468937876e-05,
+      "loss": 0.2853,
+      "step": 892
+    },
+    {
+      "epoch": 3.572,
+      "grad_norm": 1.0427800416946411,
+      "learning_rate": 4.11623246492986e-05,
+      "loss": 0.3032,
+      "step": 893
+    },
+    {
+      "epoch": 3.576,
+      "grad_norm": 0.9730731844902039,
+      "learning_rate": 4.115230460921844e-05,
+      "loss": 0.315,
+      "step": 894
+    },
+    {
+      "epoch": 3.58,
+      "grad_norm": 0.9691907167434692,
+      "learning_rate": 4.114228456913828e-05,
+      "loss": 0.3382,
+      "step": 895
+    },
+    {
+      "epoch": 3.584,
+      "grad_norm": 0.905629575252533,
+      "learning_rate": 4.1132264529058115e-05,
+      "loss": 0.2965,
+      "step": 896
+    },
+    {
+      "epoch": 3.588,
+      "grad_norm": 1.1402881145477295,
+      "learning_rate": 4.1122244488977956e-05,
+      "loss": 0.3044,
+      "step": 897
+    },
+    {
+      "epoch": 3.592,
+      "grad_norm": 1.0247617959976196,
+      "learning_rate": 4.11122244488978e-05,
+      "loss": 0.3063,
+      "step": 898
+    },
+    {
+      "epoch": 3.596,
+      "grad_norm": 1.0826162099838257,
+      "learning_rate": 4.110220440881764e-05,
+      "loss": 0.2926,
+      "step": 899
+    },
+    {
+      "epoch": 3.6,
+      "grad_norm": 1.0896751880645752,
+      "learning_rate": 4.109218436873748e-05,
+      "loss": 0.3232,
+      "step": 900
+    },
+    {
+      "epoch": 3.604,
+      "grad_norm": 1.0806505680084229,
+      "learning_rate": 4.108216432865732e-05,
+      "loss": 0.2913,
+      "step": 901
+    },
+    {
+      "epoch": 3.608,
+      "grad_norm": 1.0015305280685425,
+      "learning_rate": 4.1072144288577155e-05,
+      "loss": 0.2998,
+      "step": 902
+    },
+    {
+      "epoch": 3.612,
+      "grad_norm": 1.0075455904006958,
+      "learning_rate": 4.1062124248497e-05,
+      "loss": 0.2716,
+      "step": 903
+    },
+    {
+      "epoch": 3.616,
+      "grad_norm": 1.06728196144104,
+      "learning_rate": 4.105210420841684e-05,
+      "loss": 0.3269,
+      "step": 904
+    },
+    {
+      "epoch": 3.62,
+      "grad_norm": 0.8644483089447021,
+      "learning_rate": 4.104208416833667e-05,
+      "loss": 0.2804,
+      "step": 905
+    },
+    {
+      "epoch": 3.624,
+      "grad_norm": 0.9951996803283691,
+      "learning_rate": 4.1032064128256514e-05,
+      "loss": 0.3223,
+      "step": 906
+    },
+    {
+      "epoch": 3.628,
+      "grad_norm": 0.9931237697601318,
+      "learning_rate": 4.1022044088176355e-05,
+      "loss": 0.2715,
+      "step": 907
+    },
+    {
+      "epoch": 3.632,
+      "grad_norm": 1.0275764465332031,
+      "learning_rate": 4.101202404809619e-05,
+      "loss": 0.3444,
+      "step": 908
+    },
+    {
+      "epoch": 3.636,
+      "grad_norm": 0.9816417098045349,
+      "learning_rate": 4.100200400801603e-05,
+      "loss": 0.2724,
+      "step": 909
+    },
+    {
+      "epoch": 3.64,
+      "grad_norm": 1.1031150817871094,
+      "learning_rate": 4.099198396793588e-05,
+      "loss": 0.3406,
+      "step": 910
+    },
+    {
+      "epoch": 3.644,
+      "grad_norm": 0.7077162265777588,
+      "learning_rate": 4.0981963927855713e-05,
+      "loss": 0.1978,
+      "step": 911
+    },
+    {
+      "epoch": 3.648,
+      "grad_norm": 1.0858787298202515,
+      "learning_rate": 4.0971943887775555e-05,
+      "loss": 0.2975,
+      "step": 912
+    },
+    {
+      "epoch": 3.652,
+      "grad_norm": 0.9886104464530945,
+      "learning_rate": 4.0961923847695396e-05,
+      "loss": 0.3237,
+      "step": 913
+    },
+    {
+      "epoch": 3.656,
+      "grad_norm": 1.1059352159500122,
+      "learning_rate": 4.095190380761523e-05,
+      "loss": 0.3181,
+      "step": 914
+    },
+    {
+      "epoch": 3.66,
+      "grad_norm": 0.9329758882522583,
+      "learning_rate": 4.094188376753507e-05,
+      "loss": 0.2751,
+      "step": 915
+    },
+    {
+      "epoch": 3.664,
+      "grad_norm": 1.0664880275726318,
+      "learning_rate": 4.093186372745491e-05,
+      "loss": 0.3348,
+      "step": 916
+    },
+    {
+      "epoch": 3.668,
+      "grad_norm": 0.9057939052581787,
+      "learning_rate": 4.092184368737475e-05,
+      "loss": 0.3088,
+      "step": 917
+    },
+    {
+      "epoch": 3.672,
+      "grad_norm": 0.9240928888320923,
+      "learning_rate": 4.091182364729459e-05,
+      "loss": 0.1825,
+      "step": 918
+    },
+    {
+      "epoch": 3.676,
+      "grad_norm": 1.0612246990203857,
+      "learning_rate": 4.090180360721443e-05,
+      "loss": 0.2613,
+      "step": 919
+    },
+    {
+      "epoch": 3.68,
+      "grad_norm": 0.8932524919509888,
+      "learning_rate": 4.089178356713427e-05,
+      "loss": 0.3157,
+      "step": 920
+    },
+    {
+      "epoch": 3.684,
+      "grad_norm": 0.9782359600067139,
+      "learning_rate": 4.088176352705411e-05,
+      "loss": 0.2984,
+      "step": 921
+    },
+    {
+      "epoch": 3.6879999999999997,
+      "grad_norm": 1.0117908716201782,
+      "learning_rate": 4.0871743486973954e-05,
+      "loss": 0.3072,
+      "step": 922
+    },
+    {
+      "epoch": 3.692,
+      "grad_norm": 1.1856647729873657,
+      "learning_rate": 4.086172344689379e-05,
+      "loss": 0.3281,
+      "step": 923
+    },
+    {
+      "epoch": 3.6959999999999997,
+      "grad_norm": 0.767086386680603,
+      "learning_rate": 4.085170340681363e-05,
+      "loss": 0.1969,
+      "step": 924
+    },
+    {
+      "epoch": 3.7,
+      "grad_norm": 1.2814222574234009,
+      "learning_rate": 4.084168336673347e-05,
+      "loss": 0.3409,
+      "step": 925
+    },
+    {
+      "epoch": 3.7039999999999997,
+      "grad_norm": 1.0232386589050293,
+      "learning_rate": 4.0831663326653306e-05,
+      "loss": 0.2866,
+      "step": 926
+    },
+    {
+      "epoch": 3.708,
+      "grad_norm": 1.0636742115020752,
+      "learning_rate": 4.082164328657315e-05,
+      "loss": 0.3536,
+      "step": 927
+    },
+    {
+      "epoch": 3.7119999999999997,
+      "grad_norm": 0.9983400106430054,
+      "learning_rate": 4.081162324649299e-05,
+      "loss": 0.3254,
+      "step": 928
+    },
+    {
+      "epoch": 3.716,
+      "grad_norm": 0.9435644745826721,
+      "learning_rate": 4.080160320641282e-05,
+      "loss": 0.3213,
+      "step": 929
+    },
+    {
+      "epoch": 3.7199999999999998,
+      "grad_norm": 0.9191159009933472,
+      "learning_rate": 4.079158316633267e-05,
+      "loss": 0.2703,
+      "step": 930
+    },
+    {
+      "epoch": 3.724,
+      "grad_norm": 0.6620362401008606,
+      "learning_rate": 4.078156312625251e-05,
+      "loss": 0.1913,
+      "step": 931
+    },
+    {
+      "epoch": 3.7279999999999998,
+      "grad_norm": 0.7755175828933716,
+      "learning_rate": 4.0771543086172346e-05,
+      "loss": 0.2269,
+      "step": 932
+    },
+    {
+      "epoch": 3.732,
+      "grad_norm": 1.0141972303390503,
+      "learning_rate": 4.076152304609219e-05,
+      "loss": 0.3042,
+      "step": 933
+    },
+    {
+      "epoch": 3.7359999999999998,
+      "grad_norm": 1.0013058185577393,
+      "learning_rate": 4.075150300601203e-05,
+      "loss": 0.3252,
+      "step": 934
+    },
+    {
+      "epoch": 3.74,
+      "grad_norm": 1.1181739568710327,
+      "learning_rate": 4.0741482965931864e-05,
+      "loss": 0.3186,
+      "step": 935
+    },
+    {
+      "epoch": 3.7439999999999998,
+      "grad_norm": 1.0314396619796753,
+      "learning_rate": 4.0731462925851705e-05,
+      "loss": 0.2691,
+      "step": 936
+    },
+    {
+      "epoch": 3.748,
+      "grad_norm": 0.9695926308631897,
+      "learning_rate": 4.0721442885771546e-05,
+      "loss": 0.3565,
+      "step": 937
+    },
+    {
+      "epoch": 3.752,
+      "grad_norm": 0.9737102389335632,
+      "learning_rate": 4.071142284569138e-05,
+      "loss": 0.2772,
+      "step": 938
+    },
+    {
+      "epoch": 3.7560000000000002,
+      "grad_norm": 1.114227056503296,
+      "learning_rate": 4.070140280561122e-05,
+      "loss": 0.316,
+      "step": 939
+    },
+    {
+      "epoch": 3.76,
+      "grad_norm": 1.2355971336364746,
+      "learning_rate": 4.069138276553106e-05,
+      "loss": 0.331,
+      "step": 940
+    },
+    {
+      "epoch": 3.7640000000000002,
+      "grad_norm": 0.8689810037612915,
+      "learning_rate": 4.0681362725450904e-05,
+      "loss": 0.2914,
+      "step": 941
+    },
+    {
+      "epoch": 3.768,
+      "grad_norm": 1.04837167263031,
+      "learning_rate": 4.0671342685370746e-05,
+      "loss": 0.3081,
+      "step": 942
+    },
+    {
+      "epoch": 3.7720000000000002,
+      "grad_norm": 1.047351598739624,
+      "learning_rate": 4.066132264529059e-05,
+      "loss": 0.3239,
+      "step": 943
+    },
+    {
+      "epoch": 3.776,
+      "grad_norm": 0.9564910531044006,
+      "learning_rate": 4.065130260521042e-05,
+      "loss": 0.3193,
+      "step": 944
+    },
+    {
+      "epoch": 3.7800000000000002,
+      "grad_norm": 1.0410816669464111,
+      "learning_rate": 4.064128256513026e-05,
+      "loss": 0.3311,
+      "step": 945
+    },
+    {
+      "epoch": 3.784,
+      "grad_norm": 0.9697766304016113,
+      "learning_rate": 4.0631262525050104e-05,
+      "loss": 0.3043,
+      "step": 946
+    },
+    {
+      "epoch": 3.7880000000000003,
+      "grad_norm": 1.0359306335449219,
+      "learning_rate": 4.062124248496994e-05,
+      "loss": 0.3646,
+      "step": 947
+    },
+    {
+      "epoch": 3.792,
+      "grad_norm": 0.8712097406387329,
+      "learning_rate": 4.061122244488978e-05,
+      "loss": 0.2917,
+      "step": 948
+    },
+    {
+      "epoch": 3.7960000000000003,
+      "grad_norm": 0.9690885543823242,
+      "learning_rate": 4.060120240480962e-05,
+      "loss": 0.3265,
+      "step": 949
+    },
+    {
+      "epoch": 3.8,
+      "grad_norm": 0.9399465322494507,
+      "learning_rate": 4.059118236472946e-05,
+      "loss": 0.3144,
+      "step": 950
+    },
+    {
+      "epoch": 3.8040000000000003,
+      "grad_norm": 1.0360187292099,
+      "learning_rate": 4.0581162324649304e-05,
+      "loss": 0.3777,
+      "step": 951
+    },
+    {
+      "epoch": 3.808,
+      "grad_norm": 1.233900785446167,
+      "learning_rate": 4.057114228456914e-05,
+      "loss": 0.3032,
+      "step": 952
+    },
+    {
+      "epoch": 3.8120000000000003,
+      "grad_norm": 1.0116194486618042,
+      "learning_rate": 4.056112224448898e-05,
+      "loss": 0.2787,
+      "step": 953
+    },
+    {
+      "epoch": 3.816,
+      "grad_norm": 0.9884048700332642,
+      "learning_rate": 4.055110220440882e-05,
+      "loss": 0.3147,
+      "step": 954
+    },
+    {
+      "epoch": 3.82,
+      "grad_norm": 0.9382745623588562,
+      "learning_rate": 4.0541082164328655e-05,
+      "loss": 0.3232,
+      "step": 955
+    },
+    {
+      "epoch": 3.824,
+      "grad_norm": 0.9225025177001953,
+      "learning_rate": 4.0531062124248497e-05,
+      "loss": 0.3334,
+      "step": 956
+    },
+    {
+      "epoch": 3.828,
+      "grad_norm": 0.8968762159347534,
+      "learning_rate": 4.052104208416834e-05,
+      "loss": 0.3146,
+      "step": 957
+    },
+    {
+      "epoch": 3.832,
+      "grad_norm": 0.9080137610435486,
+      "learning_rate": 4.051102204408818e-05,
+      "loss": 0.2877,
+      "step": 958
+    },
+    {
+      "epoch": 3.836,
+      "grad_norm": 0.9056142568588257,
+      "learning_rate": 4.050100200400802e-05,
+      "loss": 0.2728,
+      "step": 959
+    },
+    {
+      "epoch": 3.84,
+      "grad_norm": 0.976733922958374,
+      "learning_rate": 4.049098196392786e-05,
+      "loss": 0.3077,
+      "step": 960
+    },
+    {
+      "epoch": 3.844,
+      "grad_norm": 1.0319982767105103,
+      "learning_rate": 4.0480961923847696e-05,
+      "loss": 0.3124,
+      "step": 961
+    },
+    {
+      "epoch": 3.848,
+      "grad_norm": 0.8904579281806946,
+      "learning_rate": 4.047094188376754e-05,
+      "loss": 0.3197,
+      "step": 962
+    },
+    {
+      "epoch": 3.852,
+      "grad_norm": 0.9294513463973999,
+      "learning_rate": 4.046092184368738e-05,
+      "loss": 0.318,
+      "step": 963
+    },
+    {
+      "epoch": 3.856,
+      "grad_norm": 0.9072182178497314,
+      "learning_rate": 4.045090180360721e-05,
+      "loss": 0.2971,
+      "step": 964
+    },
+    {
+      "epoch": 3.86,
+      "grad_norm": 1.0785537958145142,
+      "learning_rate": 4.0440881763527055e-05,
+      "loss": 0.3362,
+      "step": 965
+    },
+    {
+      "epoch": 3.864,
+      "grad_norm": 1.1097102165222168,
+      "learning_rate": 4.0430861723446896e-05,
+      "loss": 0.3528,
+      "step": 966
+    },
+    {
+      "epoch": 3.868,
+      "grad_norm": 1.0673046112060547,
+      "learning_rate": 4.042084168336673e-05,
+      "loss": 0.3085,
+      "step": 967
+    },
+    {
+      "epoch": 3.872,
+      "grad_norm": 0.8683971762657166,
+      "learning_rate": 4.041082164328657e-05,
+      "loss": 0.2852,
+      "step": 968
+    },
+    {
+      "epoch": 3.876,
+      "grad_norm": 0.9497571587562561,
+      "learning_rate": 4.040080160320642e-05,
+      "loss": 0.2755,
+      "step": 969
+    },
+    {
+      "epoch": 3.88,
+      "grad_norm": 1.0285412073135376,
+      "learning_rate": 4.0390781563126254e-05,
+      "loss": 0.3037,
+      "step": 970
+    },
+    {
+      "epoch": 3.884,
+      "grad_norm": 1.1578757762908936,
+      "learning_rate": 4.0380761523046096e-05,
+      "loss": 0.3113,
+      "step": 971
+    },
+    {
+      "epoch": 3.888,
+      "grad_norm": 1.017016887664795,
+      "learning_rate": 4.037074148296594e-05,
+      "loss": 0.3655,
+      "step": 972
+    },
+    {
+      "epoch": 3.892,
+      "grad_norm": 1.044767141342163,
+      "learning_rate": 4.036072144288577e-05,
+      "loss": 0.3088,
+      "step": 973
+    },
+    {
+      "epoch": 3.896,
+      "grad_norm": 1.098922848701477,
+      "learning_rate": 4.035070140280561e-05,
+      "loss": 0.3294,
+      "step": 974
+    },
+    {
+      "epoch": 3.9,
+      "grad_norm": 0.9794385433197021,
+      "learning_rate": 4.0340681362725454e-05,
+      "loss": 0.302,
+      "step": 975
+    },
+    {
+      "epoch": 3.904,
+      "grad_norm": 0.9461687207221985,
+      "learning_rate": 4.033066132264529e-05,
+      "loss": 0.3199,
+      "step": 976
+    },
+    {
+      "epoch": 3.908,
+      "grad_norm": 0.971648633480072,
+      "learning_rate": 4.032064128256513e-05,
+      "loss": 0.3065,
+      "step": 977
+    },
+    {
+      "epoch": 3.912,
+      "grad_norm": 0.8333442807197571,
+      "learning_rate": 4.031062124248497e-05,
+      "loss": 0.2888,
+      "step": 978
+    },
+    {
+      "epoch": 3.916,
+      "grad_norm": 0.8855047225952148,
+      "learning_rate": 4.030060120240481e-05,
+      "loss": 0.3273,
+      "step": 979
+    },
+    {
+      "epoch": 3.92,
+      "grad_norm": 0.7457961440086365,
+      "learning_rate": 4.0290581162324654e-05,
+      "loss": 0.2061,
+      "step": 980
+    },
+    {
+      "epoch": 3.924,
+      "grad_norm": 0.9684552550315857,
+      "learning_rate": 4.0280561122244495e-05,
+      "loss": 0.3369,
+      "step": 981
+    },
+    {
+      "epoch": 3.928,
+      "grad_norm": 1.0121392011642456,
+      "learning_rate": 4.027054108216433e-05,
+      "loss": 0.3324,
+      "step": 982
+    },
+    {
+      "epoch": 3.932,
+      "grad_norm": 0.9265434145927429,
+      "learning_rate": 4.026052104208417e-05,
+      "loss": 0.3017,
+      "step": 983
+    },
+    {
+      "epoch": 3.936,
+      "grad_norm": 1.0135233402252197,
+      "learning_rate": 4.025050100200401e-05,
+      "loss": 0.2969,
+      "step": 984
+    },
+    {
+      "epoch": 3.94,
+      "grad_norm": 1.0143775939941406,
+      "learning_rate": 4.0240480961923846e-05,
+      "loss": 0.2992,
+      "step": 985
+    },
+    {
+      "epoch": 3.944,
+      "grad_norm": 0.9987703561782837,
+      "learning_rate": 4.023046092184369e-05,
+      "loss": 0.316,
+      "step": 986
+    },
+    {
+      "epoch": 3.948,
+      "grad_norm": 0.9264063239097595,
+      "learning_rate": 4.022044088176353e-05,
+      "loss": 0.3123,
+      "step": 987
+    },
+    {
+      "epoch": 3.952,
+      "grad_norm": 1.1002973318099976,
+      "learning_rate": 4.0210420841683363e-05,
+      "loss": 0.3324,
+      "step": 988
+    },
+    {
+      "epoch": 3.956,
+      "grad_norm": 1.0875300168991089,
+      "learning_rate": 4.020040080160321e-05,
+      "loss": 0.3193,
+      "step": 989
+    },
+    {
+      "epoch": 3.96,
+      "grad_norm": 1.137242078781128,
+      "learning_rate": 4.019038076152305e-05,
+      "loss": 0.2993,
+      "step": 990
+    },
+    {
+      "epoch": 3.964,
+      "grad_norm": 1.010221242904663,
+      "learning_rate": 4.018036072144289e-05,
+      "loss": 0.2902,
+      "step": 991
+    },
+    {
+      "epoch": 3.968,
+      "grad_norm": 1.0185948610305786,
+      "learning_rate": 4.017034068136273e-05,
+      "loss": 0.2858,
+      "step": 992
+    },
+    {
+      "epoch": 3.972,
+      "grad_norm": 1.1539437770843506,
+      "learning_rate": 4.016032064128257e-05,
+      "loss": 0.3179,
+      "step": 993
+    },
+    {
+      "epoch": 3.976,
+      "grad_norm": 1.0386922359466553,
+      "learning_rate": 4.0150300601202404e-05,
+      "loss": 0.3271,
+      "step": 994
+    },
+    {
+      "epoch": 3.98,
+      "grad_norm": 1.041878581047058,
+      "learning_rate": 4.0140280561122246e-05,
+      "loss": 0.3355,
+      "step": 995
+    },
+    {
+      "epoch": 3.984,
+      "grad_norm": 1.1333723068237305,
+      "learning_rate": 4.013026052104209e-05,
+      "loss": 0.3385,
+      "step": 996
+    },
+    {
+      "epoch": 3.988,
+      "grad_norm": 1.0429155826568604,
+      "learning_rate": 4.012024048096192e-05,
+      "loss": 0.2962,
+      "step": 997
+    },
+    {
+      "epoch": 3.992,
+      "grad_norm": 1.0766241550445557,
+      "learning_rate": 4.011022044088176e-05,
+      "loss": 0.3384,
+      "step": 998
+    },
+    {
+      "epoch": 3.996,
+      "grad_norm": 0.979697048664093,
+      "learning_rate": 4.010020040080161e-05,
+      "loss": 0.2805,
+      "step": 999
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 1.0180047750473022,
+      "learning_rate": 4.0090180360721445e-05,
+      "loss": 0.3099,
+      "step": 1000
+    },
+    {
+      "epoch": 4.004,
+      "grad_norm": 1.014128565788269,
+      "learning_rate": 4.0080160320641287e-05,
+      "loss": 0.2158,
+      "step": 1001
+    },
+    {
+      "epoch": 4.008,
+      "grad_norm": 0.9923162460327148,
+      "learning_rate": 4.007014028056113e-05,
+      "loss": 0.2208,
+      "step": 1002
+    },
+    {
+      "epoch": 4.012,
+      "grad_norm": 1.0028576850891113,
+      "learning_rate": 4.006012024048096e-05,
+      "loss": 0.2285,
+      "step": 1003
+    },
+    {
+      "epoch": 4.016,
+      "grad_norm": 0.962719738483429,
+      "learning_rate": 4.0050100200400804e-05,
+      "loss": 0.2034,
+      "step": 1004
+    },
+    {
+      "epoch": 4.02,
+      "grad_norm": 1.013881802558899,
+      "learning_rate": 4.0040080160320645e-05,
+      "loss": 0.1846,
+      "step": 1005
+    },
+    {
+      "epoch": 4.024,
+      "grad_norm": 0.8782246708869934,
+      "learning_rate": 4.003006012024048e-05,
+      "loss": 0.2139,
+      "step": 1006
+    },
+    {
+      "epoch": 4.028,
+      "grad_norm": 1.26191246509552,
+      "learning_rate": 4.002004008016032e-05,
+      "loss": 0.2172,
+      "step": 1007
+    },
+    {
+      "epoch": 4.032,
+      "grad_norm": 1.2841931581497192,
+      "learning_rate": 4.001002004008016e-05,
+      "loss": 0.2612,
+      "step": 1008
+    },
+    {
+      "epoch": 4.036,
+      "grad_norm": 1.2775436639785767,
+      "learning_rate": 4e-05,
+      "loss": 0.1986,
+      "step": 1009
+    },
+    {
+      "epoch": 4.04,
+      "grad_norm": 1.3614190816879272,
+      "learning_rate": 3.9989979959919845e-05,
+      "loss": 0.2709,
+      "step": 1010
+    },
+    {
+      "epoch": 4.044,
+      "grad_norm": 1.4308106899261475,
+      "learning_rate": 3.997995991983968e-05,
+      "loss": 0.2196,
+      "step": 1011
+    },
+    {
+      "epoch": 4.048,
+      "grad_norm": 1.4295969009399414,
+      "learning_rate": 3.996993987975952e-05,
+      "loss": 0.2131,
+      "step": 1012
+    },
+    {
+      "epoch": 4.052,
+      "grad_norm": 1.4375779628753662,
+      "learning_rate": 3.995991983967936e-05,
+      "loss": 0.203,
+      "step": 1013
+    },
+    {
+      "epoch": 4.056,
+      "grad_norm": 1.3883522748947144,
+      "learning_rate": 3.99498997995992e-05,
+      "loss": 0.2139,
+      "step": 1014
+    },
+    {
+      "epoch": 4.06,
+      "grad_norm": 1.211121678352356,
+      "learning_rate": 3.993987975951904e-05,
+      "loss": 0.2068,
+      "step": 1015
+    },
+    {
+      "epoch": 4.064,
+      "grad_norm": 1.3009765148162842,
+      "learning_rate": 3.992985971943888e-05,
+      "loss": 0.2104,
+      "step": 1016
+    },
+    {
+      "epoch": 4.068,
+      "grad_norm": 1.137965202331543,
+      "learning_rate": 3.991983967935872e-05,
+      "loss": 0.1963,
+      "step": 1017
+    },
+    {
+      "epoch": 4.072,
+      "grad_norm": 1.2894974946975708,
+      "learning_rate": 3.990981963927856e-05,
+      "loss": 0.2235,
+      "step": 1018
+    },
+    {
+      "epoch": 4.076,
+      "grad_norm": 0.706529974937439,
+      "learning_rate": 3.98997995991984e-05,
+      "loss": 0.098,
+      "step": 1019
+    },
+    {
+      "epoch": 4.08,
+      "grad_norm": 1.130771279335022,
+      "learning_rate": 3.988977955911824e-05,
+      "loss": 0.2235,
+      "step": 1020
+    },
+    {
+      "epoch": 4.084,
+      "grad_norm": 1.171426773071289,
+      "learning_rate": 3.987975951903808e-05,
+      "loss": 0.1852,
+      "step": 1021
+    },
+    {
+      "epoch": 4.088,
+      "grad_norm": 1.0851662158966064,
+      "learning_rate": 3.986973947895792e-05,
+      "loss": 0.1879,
+      "step": 1022
+    },
+    {
+      "epoch": 4.092,
+      "grad_norm": 1.3458068370819092,
+      "learning_rate": 3.9859719438877754e-05,
+      "loss": 0.2321,
+      "step": 1023
+    },
+    {
+      "epoch": 4.096,
+      "grad_norm": 1.1501377820968628,
+      "learning_rate": 3.9849699398797595e-05,
+      "loss": 0.2157,
+      "step": 1024
+    },
+    {
+      "epoch": 4.1,
+      "grad_norm": 1.2266032695770264,
+      "learning_rate": 3.983967935871744e-05,
+      "loss": 0.22,
+      "step": 1025
+    },
+    {
+      "epoch": 4.104,
+      "grad_norm": 1.1825183629989624,
+      "learning_rate": 3.982965931863728e-05,
+      "loss": 0.1887,
+      "step": 1026
+    },
+    {
+      "epoch": 4.108,
+      "grad_norm": 1.1460607051849365,
+      "learning_rate": 3.981963927855711e-05,
+      "loss": 0.1834,
+      "step": 1027
+    },
+    {
+      "epoch": 4.112,
+      "grad_norm": 1.2845118045806885,
+      "learning_rate": 3.980961923847696e-05,
+      "loss": 0.223,
+      "step": 1028
+    },
+    {
+      "epoch": 4.116,
+      "grad_norm": 1.3526803255081177,
+      "learning_rate": 3.9799599198396795e-05,
+      "loss": 0.1907,
+      "step": 1029
+    },
+    {
+      "epoch": 4.12,
+      "grad_norm": 1.2375893592834473,
+      "learning_rate": 3.9789579158316636e-05,
+      "loss": 0.188,
+      "step": 1030
+    },
+    {
+      "epoch": 4.124,
+      "grad_norm": 1.2433942556381226,
+      "learning_rate": 3.977955911823648e-05,
+      "loss": 0.2173,
+      "step": 1031
+    },
+    {
+      "epoch": 4.128,
+      "grad_norm": 1.2910653352737427,
+      "learning_rate": 3.976953907815631e-05,
+      "loss": 0.192,
+      "step": 1032
+    },
+    {
+      "epoch": 4.132,
+      "grad_norm": 1.0230827331542969,
+      "learning_rate": 3.9759519038076153e-05,
+      "loss": 0.1544,
+      "step": 1033
+    },
+    {
+      "epoch": 4.136,
+      "grad_norm": 0.9551684856414795,
+      "learning_rate": 3.9749498997995995e-05,
+      "loss": 0.1518,
+      "step": 1034
+    },
+    {
+      "epoch": 4.14,
+      "grad_norm": 1.2070391178131104,
+      "learning_rate": 3.973947895791583e-05,
+      "loss": 0.1783,
+      "step": 1035
+    },
+    {
+      "epoch": 4.144,
+      "grad_norm": 1.4213783740997314,
+      "learning_rate": 3.972945891783567e-05,
+      "loss": 0.2237,
+      "step": 1036
+    },
+    {
+      "epoch": 4.148,
+      "grad_norm": 1.2974165678024292,
+      "learning_rate": 3.971943887775551e-05,
+      "loss": 0.1768,
+      "step": 1037
+    },
+    {
+      "epoch": 4.152,
+      "grad_norm": 0.8103154301643372,
+      "learning_rate": 3.970941883767535e-05,
+      "loss": 0.0919,
+      "step": 1038
+    },
+    {
+      "epoch": 4.156,
+      "grad_norm": 1.279154896736145,
+      "learning_rate": 3.9699398797595194e-05,
+      "loss": 0.2302,
+      "step": 1039
+    },
+    {
+      "epoch": 4.16,
+      "grad_norm": 1.3680810928344727,
+      "learning_rate": 3.9689378757515036e-05,
+      "loss": 0.2242,
+      "step": 1040
+    },
+    {
+      "epoch": 4.164,
+      "grad_norm": 1.184191107749939,
+      "learning_rate": 3.967935871743487e-05,
+      "loss": 0.2504,
+      "step": 1041
+    },
+    {
+      "epoch": 4.168,
+      "grad_norm": 1.2547346353530884,
+      "learning_rate": 3.966933867735471e-05,
+      "loss": 0.2051,
+      "step": 1042
+    },
+    {
+      "epoch": 4.172,
+      "grad_norm": 1.1803265810012817,
+      "learning_rate": 3.965931863727455e-05,
+      "loss": 0.2276,
+      "step": 1043
+    },
+    {
+      "epoch": 4.176,
+      "grad_norm": 1.4127157926559448,
+      "learning_rate": 3.964929859719439e-05,
+      "loss": 0.2378,
+      "step": 1044
+    },
+    {
+      "epoch": 4.18,
+      "grad_norm": 1.2966582775115967,
+      "learning_rate": 3.963927855711423e-05,
+      "loss": 0.1672,
+      "step": 1045
+    },
+    {
+      "epoch": 4.184,
+      "grad_norm": 1.1143783330917358,
+      "learning_rate": 3.962925851703407e-05,
+      "loss": 0.2011,
+      "step": 1046
+    },
+    {
+      "epoch": 4.188,
+      "grad_norm": 1.2296042442321777,
+      "learning_rate": 3.9619238476953904e-05,
+      "loss": 0.1928,
+      "step": 1047
+    },
+    {
+      "epoch": 4.192,
+      "grad_norm": 1.2343521118164062,
+      "learning_rate": 3.960921843687375e-05,
+      "loss": 0.2026,
+      "step": 1048
+    },
+    {
+      "epoch": 4.196,
+      "grad_norm": 1.2687957286834717,
+      "learning_rate": 3.9599198396793594e-05,
+      "loss": 0.2408,
+      "step": 1049
+    },
+    {
+      "epoch": 4.2,
+      "grad_norm": 1.172644019126892,
+      "learning_rate": 3.958917835671343e-05,
+      "loss": 0.2206,
+      "step": 1050
+    },
+    {
+      "epoch": 4.204,
+      "grad_norm": 1.220900297164917,
+      "learning_rate": 3.957915831663327e-05,
+      "loss": 0.2039,
+      "step": 1051
+    },
+    {
+      "epoch": 4.208,
+      "grad_norm": 1.3167065382003784,
+      "learning_rate": 3.956913827655311e-05,
+      "loss": 0.1992,
+      "step": 1052
+    },
+    {
+      "epoch": 4.212,
+      "grad_norm": 1.2286665439605713,
+      "learning_rate": 3.9559118236472945e-05,
+      "loss": 0.1897,
+      "step": 1053
+    },
+    {
+      "epoch": 4.216,
+      "grad_norm": 1.2382471561431885,
+      "learning_rate": 3.9549098196392786e-05,
+      "loss": 0.2299,
+      "step": 1054
+    },
+    {
+      "epoch": 4.22,
+      "grad_norm": 1.2562413215637207,
+      "learning_rate": 3.953907815631263e-05,
+      "loss": 0.252,
+      "step": 1055
+    },
+    {
+      "epoch": 4.224,
+      "grad_norm": 1.2567752599716187,
+      "learning_rate": 3.952905811623246e-05,
+      "loss": 0.2026,
+      "step": 1056
+    },
+    {
+      "epoch": 4.228,
+      "grad_norm": 1.1952612400054932,
+      "learning_rate": 3.9519038076152304e-05,
+      "loss": 0.2166,
+      "step": 1057
+    },
+    {
+      "epoch": 4.232,
+      "grad_norm": 1.23219633102417,
+      "learning_rate": 3.950901803607215e-05,
+      "loss": 0.2183,
+      "step": 1058
+    },
+    {
+      "epoch": 4.236,
+      "grad_norm": 1.3385809659957886,
+      "learning_rate": 3.9498997995991986e-05,
+      "loss": 0.2031,
+      "step": 1059
+    },
+    {
+      "epoch": 4.24,
+      "grad_norm": 1.0677396059036255,
+      "learning_rate": 3.948897795591183e-05,
+      "loss": 0.2011,
+      "step": 1060
+    },
+    {
+      "epoch": 4.244,
+      "grad_norm": 1.4090005159378052,
+      "learning_rate": 3.947895791583167e-05,
+      "loss": 0.2227,
+      "step": 1061
+    },
+    {
+      "epoch": 4.248,
+      "grad_norm": 1.1532093286514282,
+      "learning_rate": 3.94689378757515e-05,
+      "loss": 0.1974,
+      "step": 1062
+    },
+    {
+      "epoch": 4.252,
+      "grad_norm": 1.1573532819747925,
+      "learning_rate": 3.9458917835671344e-05,
+      "loss": 0.2248,
+      "step": 1063
+    },
+    {
+      "epoch": 4.256,
+      "grad_norm": 1.2687122821807861,
+      "learning_rate": 3.9448897795591186e-05,
+      "loss": 0.2144,
+      "step": 1064
+    },
+    {
+      "epoch": 4.26,
+      "grad_norm": 1.2371104955673218,
+      "learning_rate": 3.943887775551102e-05,
+      "loss": 0.1946,
+      "step": 1065
+    },
+    {
+      "epoch": 4.264,
+      "grad_norm": 0.7894278764724731,
+      "learning_rate": 3.942885771543086e-05,
+      "loss": 0.1066,
+      "step": 1066
+    },
+    {
+      "epoch": 4.268,
+      "grad_norm": 1.357580304145813,
+      "learning_rate": 3.94188376753507e-05,
+      "loss": 0.2478,
+      "step": 1067
+    },
+    {
+      "epoch": 4.272,
+      "grad_norm": 1.174869418144226,
+      "learning_rate": 3.9408817635270544e-05,
+      "loss": 0.1946,
+      "step": 1068
+    },
+    {
+      "epoch": 4.276,
+      "grad_norm": 1.242233157157898,
+      "learning_rate": 3.9398797595190385e-05,
+      "loss": 0.2277,
+      "step": 1069
+    },
+    {
+      "epoch": 4.28,
+      "grad_norm": 1.31795334815979,
+      "learning_rate": 3.938877755511023e-05,
+      "loss": 0.2155,
+      "step": 1070
+    },
+    {
+      "epoch": 4.284,
+      "grad_norm": 0.8449123501777649,
+      "learning_rate": 3.937875751503006e-05,
+      "loss": 0.1393,
+      "step": 1071
+    },
+    {
+      "epoch": 4.288,
+      "grad_norm": 1.4304672479629517,
+      "learning_rate": 3.93687374749499e-05,
+      "loss": 0.2227,
+      "step": 1072
+    },
+    {
+      "epoch": 4.292,
+      "grad_norm": 1.1797378063201904,
+      "learning_rate": 3.9358717434869744e-05,
+      "loss": 0.2414,
+      "step": 1073
+    },
+    {
+      "epoch": 4.296,
+      "grad_norm": 1.1290992498397827,
+      "learning_rate": 3.934869739478958e-05,
+      "loss": 0.2528,
+      "step": 1074
+    },
+    {
+      "epoch": 4.3,
+      "grad_norm": 1.2640211582183838,
+      "learning_rate": 3.933867735470942e-05,
+      "loss": 0.2229,
+      "step": 1075
+    },
+    {
+      "epoch": 4.304,
+      "grad_norm": 1.117639422416687,
+      "learning_rate": 3.932865731462926e-05,
+      "loss": 0.2176,
+      "step": 1076
+    },
+    {
+      "epoch": 4.308,
+      "grad_norm": 1.35614013671875,
+      "learning_rate": 3.93186372745491e-05,
+      "loss": 0.1896,
+      "step": 1077
+    },
+    {
+      "epoch": 4.312,
+      "grad_norm": 1.2274762392044067,
+      "learning_rate": 3.930861723446894e-05,
+      "loss": 0.2213,
+      "step": 1078
+    },
+    {
+      "epoch": 4.316,
+      "grad_norm": 1.1732356548309326,
+      "learning_rate": 3.929859719438878e-05,
+      "loss": 0.2233,
+      "step": 1079
+    },
+    {
+      "epoch": 4.32,
+      "grad_norm": 1.1939340829849243,
+      "learning_rate": 3.928857715430862e-05,
+      "loss": 0.1964,
+      "step": 1080
+    },
+    {
+      "epoch": 4.324,
+      "grad_norm": 1.3930495977401733,
+      "learning_rate": 3.927855711422846e-05,
+      "loss": 0.1965,
+      "step": 1081
+    },
+    {
+      "epoch": 4.328,
+      "grad_norm": 1.371573567390442,
+      "learning_rate": 3.9268537074148295e-05,
+      "loss": 0.2191,
+      "step": 1082
+    },
+    {
+      "epoch": 4.332,
+      "grad_norm": 1.4507895708084106,
+      "learning_rate": 3.9258517034068136e-05,
+      "loss": 0.2927,
+      "step": 1083
+    },
+    {
+      "epoch": 4.336,
+      "grad_norm": 1.3624083995819092,
+      "learning_rate": 3.924849699398798e-05,
+      "loss": 0.2211,
+      "step": 1084
+    },
+    {
+      "epoch": 4.34,
+      "grad_norm": 1.1890075206756592,
+      "learning_rate": 3.923847695390782e-05,
+      "loss": 0.1766,
+      "step": 1085
+    },
+    {
+      "epoch": 4.344,
+      "grad_norm": 1.4425567388534546,
+      "learning_rate": 3.922845691382765e-05,
+      "loss": 0.205,
+      "step": 1086
+    },
+    {
+      "epoch": 4.348,
+      "grad_norm": 1.2859835624694824,
+      "learning_rate": 3.92184368737475e-05,
+      "loss": 0.2194,
+      "step": 1087
+    },
+    {
+      "epoch": 4.352,
+      "grad_norm": 1.2856872081756592,
+      "learning_rate": 3.9208416833667336e-05,
+      "loss": 0.197,
+      "step": 1088
+    },
+    {
+      "epoch": 4.356,
+      "grad_norm": 1.2462953329086304,
+      "learning_rate": 3.919839679358718e-05,
+      "loss": 0.2121,
+      "step": 1089
+    },
+    {
+      "epoch": 4.36,
+      "grad_norm": 1.364990472793579,
+      "learning_rate": 3.918837675350702e-05,
+      "loss": 0.2367,
+      "step": 1090
+    },
+    {
+      "epoch": 4.364,
+      "grad_norm": 1.3470253944396973,
+      "learning_rate": 3.917835671342685e-05,
+      "loss": 0.2263,
+      "step": 1091
+    },
+    {
+      "epoch": 4.368,
+      "grad_norm": 1.2777583599090576,
+      "learning_rate": 3.9168336673346694e-05,
+      "loss": 0.2483,
+      "step": 1092
+    },
+    {
+      "epoch": 4.372,
+      "grad_norm": 1.1814221143722534,
+      "learning_rate": 3.9158316633266535e-05,
+      "loss": 0.2304,
+      "step": 1093
+    },
+    {
+      "epoch": 4.376,
+      "grad_norm": 1.329244613647461,
+      "learning_rate": 3.914829659318637e-05,
+      "loss": 0.258,
+      "step": 1094
+    },
+    {
+      "epoch": 4.38,
+      "grad_norm": 1.2682360410690308,
+      "learning_rate": 3.913827655310621e-05,
+      "loss": 0.2278,
+      "step": 1095
+    },
+    {
+      "epoch": 4.384,
+      "grad_norm": 1.2741080522537231,
+      "learning_rate": 3.912825651302605e-05,
+      "loss": 0.2437,
+      "step": 1096
+    },
+    {
+      "epoch": 4.388,
+      "grad_norm": 1.1235476732254028,
+      "learning_rate": 3.9118236472945894e-05,
+      "loss": 0.2209,
+      "step": 1097
+    },
+    {
+      "epoch": 4.392,
+      "grad_norm": 1.1556679010391235,
+      "learning_rate": 3.9108216432865735e-05,
+      "loss": 0.2668,
+      "step": 1098
+    },
+    {
+      "epoch": 4.396,
+      "grad_norm": 1.2366710901260376,
+      "learning_rate": 3.9098196392785576e-05,
+      "loss": 0.2822,
+      "step": 1099
+    },
+    {
+      "epoch": 4.4,
+      "grad_norm": 1.325208067893982,
+      "learning_rate": 3.908817635270541e-05,
+      "loss": 0.2203,
+      "step": 1100
+    },
+    {
+      "epoch": 4.404,
+      "grad_norm": 1.166352391242981,
+      "learning_rate": 3.907815631262525e-05,
+      "loss": 0.2441,
+      "step": 1101
+    },
+    {
+      "epoch": 4.408,
+      "grad_norm": 1.2230168581008911,
+      "learning_rate": 3.9068136272545093e-05,
+      "loss": 0.2217,
+      "step": 1102
+    },
+    {
+      "epoch": 4.412,
+      "grad_norm": 1.2550815343856812,
+      "learning_rate": 3.905811623246493e-05,
+      "loss": 0.2299,
+      "step": 1103
+    },
+    {
+      "epoch": 4.416,
+      "grad_norm": 1.3714336156845093,
+      "learning_rate": 3.904809619238477e-05,
+      "loss": 0.2247,
+      "step": 1104
+    },
+    {
+      "epoch": 4.42,
+      "grad_norm": 1.2269859313964844,
+      "learning_rate": 3.903807615230461e-05,
+      "loss": 0.21,
+      "step": 1105
+    },
+    {
+      "epoch": 4.424,
+      "grad_norm": 1.420852780342102,
+      "learning_rate": 3.9028056112224445e-05,
+      "loss": 0.2248,
+      "step": 1106
+    },
+    {
+      "epoch": 4.428,
+      "grad_norm": 1.2329579591751099,
+      "learning_rate": 3.901803607214429e-05,
+      "loss": 0.2506,
+      "step": 1107
+    },
+    {
+      "epoch": 4.432,
+      "grad_norm": 1.2145705223083496,
+      "learning_rate": 3.9008016032064134e-05,
+      "loss": 0.2244,
+      "step": 1108
+    },
+    {
+      "epoch": 4.436,
+      "grad_norm": 1.3034549951553345,
+      "learning_rate": 3.899799599198397e-05,
+      "loss": 0.1866,
+      "step": 1109
+    },
+    {
+      "epoch": 4.44,
+      "grad_norm": 1.1892132759094238,
+      "learning_rate": 3.898797595190381e-05,
+      "loss": 0.2283,
+      "step": 1110
+    },
+    {
+      "epoch": 4.444,
+      "grad_norm": 1.279307246208191,
+      "learning_rate": 3.897795591182365e-05,
+      "loss": 0.2039,
+      "step": 1111
+    },
+    {
+      "epoch": 4.448,
+      "grad_norm": 0.917590320110321,
+      "learning_rate": 3.8967935871743486e-05,
+      "loss": 0.1489,
+      "step": 1112
+    },
+    {
+      "epoch": 4.452,
+      "grad_norm": 1.2451698780059814,
+      "learning_rate": 3.895791583166333e-05,
+      "loss": 0.1824,
+      "step": 1113
+    },
+    {
+      "epoch": 4.456,
+      "grad_norm": 1.3629359006881714,
+      "learning_rate": 3.894789579158317e-05,
+      "loss": 0.2174,
+      "step": 1114
+    },
+    {
+      "epoch": 4.46,
+      "grad_norm": 1.4630622863769531,
+      "learning_rate": 3.8937875751503e-05,
+      "loss": 0.2627,
+      "step": 1115
+    },
+    {
+      "epoch": 4.464,
+      "grad_norm": 1.2522321939468384,
+      "learning_rate": 3.8927855711422844e-05,
+      "loss": 0.2285,
+      "step": 1116
+    },
+    {
+      "epoch": 4.468,
+      "grad_norm": 1.2080490589141846,
+      "learning_rate": 3.891783567134269e-05,
+      "loss": 0.2458,
+      "step": 1117
+    },
+    {
+      "epoch": 4.4719999999999995,
+      "grad_norm": 1.2194185256958008,
+      "learning_rate": 3.890781563126253e-05,
+      "loss": 0.2394,
+      "step": 1118
+    },
+    {
+      "epoch": 4.476,
+      "grad_norm": 1.1908085346221924,
+      "learning_rate": 3.889779559118237e-05,
+      "loss": 0.2071,
+      "step": 1119
+    },
+    {
+      "epoch": 4.48,
+      "grad_norm": 1.3185300827026367,
+      "learning_rate": 3.888777555110221e-05,
+      "loss": 0.1841,
+      "step": 1120
+    },
+    {
+      "epoch": 4.484,
+      "grad_norm": 1.2330673933029175,
+      "learning_rate": 3.8877755511022044e-05,
+      "loss": 0.2216,
+      "step": 1121
+    },
+    {
+      "epoch": 4.4879999999999995,
+      "grad_norm": 1.1211673021316528,
+      "learning_rate": 3.8867735470941885e-05,
+      "loss": 0.2207,
+      "step": 1122
+    },
+    {
+      "epoch": 4.492,
+      "grad_norm": 1.3571367263793945,
+      "learning_rate": 3.8857715430861727e-05,
+      "loss": 0.3118,
+      "step": 1123
+    },
+    {
+      "epoch": 4.496,
+      "grad_norm": 1.3315199613571167,
+      "learning_rate": 3.884769539078156e-05,
+      "loss": 0.2559,
+      "step": 1124
+    },
+    {
+      "epoch": 4.5,
+      "grad_norm": 1.2943772077560425,
+      "learning_rate": 3.88376753507014e-05,
+      "loss": 0.2258,
+      "step": 1125
+    },
+    {
+      "epoch": 4.504,
+      "grad_norm": 1.3622092008590698,
+      "learning_rate": 3.882765531062125e-05,
+      "loss": 0.2172,
+      "step": 1126
+    },
+    {
+      "epoch": 4.508,
+      "grad_norm": 1.2951812744140625,
+      "learning_rate": 3.8817635270541085e-05,
+      "loss": 0.2419,
+      "step": 1127
+    },
+    {
+      "epoch": 4.5120000000000005,
+      "grad_norm": 1.2247873544692993,
+      "learning_rate": 3.8807615230460926e-05,
+      "loss": 0.1889,
+      "step": 1128
+    },
+    {
+      "epoch": 4.516,
+      "grad_norm": 1.29974365234375,
+      "learning_rate": 3.879759519038077e-05,
+      "loss": 0.2432,
+      "step": 1129
+    },
+    {
+      "epoch": 4.52,
+      "grad_norm": 1.2597442865371704,
+      "learning_rate": 3.87875751503006e-05,
+      "loss": 0.25,
+      "step": 1130
+    },
+    {
+      "epoch": 4.524,
+      "grad_norm": 1.388176679611206,
+      "learning_rate": 3.877755511022044e-05,
+      "loss": 0.2447,
+      "step": 1131
+    },
+    {
+      "epoch": 4.5280000000000005,
+      "grad_norm": 1.1391148567199707,
+      "learning_rate": 3.8767535070140285e-05,
+      "loss": 0.273,
+      "step": 1132
+    },
+    {
+      "epoch": 4.532,
+      "grad_norm": 1.2334431409835815,
+      "learning_rate": 3.875751503006012e-05,
+      "loss": 0.2268,
+      "step": 1133
+    },
+    {
+      "epoch": 4.536,
+      "grad_norm": 1.2976422309875488,
+      "learning_rate": 3.874749498997996e-05,
+      "loss": 0.2243,
+      "step": 1134
+    },
+    {
+      "epoch": 4.54,
+      "grad_norm": 1.2898563146591187,
+      "learning_rate": 3.87374749498998e-05,
+      "loss": 0.226,
+      "step": 1135
+    },
+    {
+      "epoch": 4.5440000000000005,
+      "grad_norm": 1.309234380722046,
+      "learning_rate": 3.872745490981964e-05,
+      "loss": 0.193,
+      "step": 1136
+    },
+    {
+      "epoch": 4.548,
+      "grad_norm": 0.9056435227394104,
+      "learning_rate": 3.8717434869739484e-05,
+      "loss": 0.1361,
+      "step": 1137
+    },
+    {
+      "epoch": 4.552,
+      "grad_norm": 1.2060729265213013,
+      "learning_rate": 3.870741482965932e-05,
+      "loss": 0.2172,
+      "step": 1138
+    },
+    {
+      "epoch": 4.556,
+      "grad_norm": 1.299154281616211,
+      "learning_rate": 3.869739478957916e-05,
+      "loss": 0.2291,
+      "step": 1139
+    },
+    {
+      "epoch": 4.5600000000000005,
+      "grad_norm": 1.1153815984725952,
+      "learning_rate": 3.8687374749499e-05,
+      "loss": 0.1993,
+      "step": 1140
+    },
+    {
+      "epoch": 4.564,
+      "grad_norm": 1.402708649635315,
+      "learning_rate": 3.867735470941884e-05,
+      "loss": 0.2377,
+      "step": 1141
+    },
+    {
+      "epoch": 4.568,
+      "grad_norm": 1.3014065027236938,
+      "learning_rate": 3.866733466933868e-05,
+      "loss": 0.2178,
+      "step": 1142
+    },
+    {
+      "epoch": 4.572,
+      "grad_norm": 1.4406603574752808,
+      "learning_rate": 3.865731462925852e-05,
+      "loss": 0.2547,
+      "step": 1143
+    },
+    {
+      "epoch": 4.576,
+      "grad_norm": 1.3197580575942993,
+      "learning_rate": 3.864729458917836e-05,
+      "loss": 0.234,
+      "step": 1144
+    },
+    {
+      "epoch": 4.58,
+      "grad_norm": 1.2924617528915405,
+      "learning_rate": 3.8637274549098194e-05,
+      "loss": 0.2327,
+      "step": 1145
+    },
+    {
+      "epoch": 4.584,
+      "grad_norm": 1.121708631515503,
+      "learning_rate": 3.862725450901804e-05,
+      "loss": 0.2183,
+      "step": 1146
+    },
+    {
+      "epoch": 4.588,
+      "grad_norm": 1.1940151453018188,
+      "learning_rate": 3.861723446893788e-05,
+      "loss": 0.2617,
+      "step": 1147
+    },
+    {
+      "epoch": 4.592,
+      "grad_norm": 1.2300734519958496,
+      "learning_rate": 3.860721442885772e-05,
+      "loss": 0.1954,
+      "step": 1148
+    },
+    {
+      "epoch": 4.596,
+      "grad_norm": 1.2809008359909058,
+      "learning_rate": 3.859719438877756e-05,
+      "loss": 0.2435,
+      "step": 1149
+    },
+    {
+      "epoch": 4.6,
+      "grad_norm": 1.3110355138778687,
+      "learning_rate": 3.8587174348697394e-05,
+      "loss": 0.2235,
+      "step": 1150
+    },
+    {
+      "epoch": 4.604,
+      "grad_norm": 1.3215099573135376,
+      "learning_rate": 3.8577154308617235e-05,
+      "loss": 0.2544,
+      "step": 1151
+    },
+    {
+      "epoch": 4.608,
+      "grad_norm": 1.1234500408172607,
+      "learning_rate": 3.8567134268537076e-05,
+      "loss": 0.275,
+      "step": 1152
+    },
+    {
+      "epoch": 4.612,
+      "grad_norm": 1.1621915102005005,
+      "learning_rate": 3.855711422845692e-05,
+      "loss": 0.2289,
+      "step": 1153
+    },
+    {
+      "epoch": 4.616,
+      "grad_norm": 1.163620114326477,
+      "learning_rate": 3.854709418837675e-05,
+      "loss": 0.25,
+      "step": 1154
+    },
+    {
+      "epoch": 4.62,
+      "grad_norm": 1.219595193862915,
+      "learning_rate": 3.853707414829659e-05,
+      "loss": 0.2388,
+      "step": 1155
+    },
+    {
+      "epoch": 4.624,
+      "grad_norm": 1.2607448101043701,
+      "learning_rate": 3.8527054108216435e-05,
+      "loss": 0.1885,
+      "step": 1156
+    },
+    {
+      "epoch": 4.628,
+      "grad_norm": 0.5879749655723572,
+      "learning_rate": 3.8517034068136276e-05,
+      "loss": 0.068,
+      "step": 1157
+    },
+    {
+      "epoch": 4.632,
+      "grad_norm": 1.3314592838287354,
+      "learning_rate": 3.850701402805612e-05,
+      "loss": 0.2302,
+      "step": 1158
+    },
+    {
+      "epoch": 4.636,
+      "grad_norm": 1.4725823402404785,
+      "learning_rate": 3.849699398797595e-05,
+      "loss": 0.2359,
+      "step": 1159
+    },
+    {
+      "epoch": 4.64,
+      "grad_norm": 1.294355034828186,
+      "learning_rate": 3.848697394789579e-05,
+      "loss": 0.2359,
+      "step": 1160
+    },
+    {
+      "epoch": 4.644,
+      "grad_norm": 1.3472228050231934,
+      "learning_rate": 3.8476953907815634e-05,
+      "loss": 0.2347,
+      "step": 1161
+    },
+    {
+      "epoch": 4.648,
+      "grad_norm": 1.2284510135650635,
+      "learning_rate": 3.846693386773547e-05,
+      "loss": 0.2115,
+      "step": 1162
+    },
+    {
+      "epoch": 4.652,
+      "grad_norm": 1.3090240955352783,
+      "learning_rate": 3.845691382765531e-05,
+      "loss": 0.2571,
+      "step": 1163
+    },
+    {
+      "epoch": 4.656,
+      "grad_norm": 1.103739857673645,
+      "learning_rate": 3.844689378757515e-05,
+      "loss": 0.2399,
+      "step": 1164
+    },
+    {
+      "epoch": 4.66,
+      "grad_norm": 1.189477801322937,
+      "learning_rate": 3.8436873747494986e-05,
+      "loss": 0.2247,
+      "step": 1165
+    },
+    {
+      "epoch": 4.664,
+      "grad_norm": 1.3449199199676514,
+      "learning_rate": 3.8426853707414834e-05,
+      "loss": 0.2112,
+      "step": 1166
+    },
+    {
+      "epoch": 4.668,
+      "grad_norm": 1.3756353855133057,
+      "learning_rate": 3.8416833667334675e-05,
+      "loss": 0.2268,
+      "step": 1167
+    },
+    {
+      "epoch": 4.672,
+      "grad_norm": 1.3317773342132568,
+      "learning_rate": 3.840681362725451e-05,
+      "loss": 0.2692,
+      "step": 1168
+    },
+    {
+      "epoch": 4.676,
+      "grad_norm": 1.2678979635238647,
+      "learning_rate": 3.839679358717435e-05,
+      "loss": 0.2411,
+      "step": 1169
+    },
+    {
+      "epoch": 4.68,
+      "grad_norm": 0.8288043141365051,
+      "learning_rate": 3.838677354709419e-05,
+      "loss": 0.1286,
+      "step": 1170
+    },
+    {
+      "epoch": 4.684,
+      "grad_norm": 1.362588882446289,
+      "learning_rate": 3.837675350701403e-05,
+      "loss": 0.2238,
+      "step": 1171
+    },
+    {
+      "epoch": 4.688,
+      "grad_norm": 1.2444343566894531,
+      "learning_rate": 3.836673346693387e-05,
+      "loss": 0.2223,
+      "step": 1172
+    },
+    {
+      "epoch": 4.692,
+      "grad_norm": 1.3691511154174805,
+      "learning_rate": 3.835671342685371e-05,
+      "loss": 0.2527,
+      "step": 1173
+    },
+    {
+      "epoch": 4.696,
+      "grad_norm": 1.233352780342102,
+      "learning_rate": 3.8346693386773544e-05,
+      "loss": 0.2361,
+      "step": 1174
+    },
+    {
+      "epoch": 4.7,
+      "grad_norm": 1.1726874113082886,
+      "learning_rate": 3.8336673346693385e-05,
+      "loss": 0.2026,
+      "step": 1175
+    },
+    {
+      "epoch": 4.704,
+      "grad_norm": 1.3368510007858276,
+      "learning_rate": 3.832665330661323e-05,
+      "loss": 0.2683,
+      "step": 1176
+    },
+    {
+      "epoch": 4.708,
+      "grad_norm": 1.355893611907959,
+      "learning_rate": 3.831663326653307e-05,
+      "loss": 0.2503,
+      "step": 1177
+    },
+    {
+      "epoch": 4.712,
+      "grad_norm": 1.2774181365966797,
+      "learning_rate": 3.830661322645291e-05,
+      "loss": 0.2655,
+      "step": 1178
+    },
+    {
+      "epoch": 4.716,
+      "grad_norm": 1.2439703941345215,
+      "learning_rate": 3.829659318637275e-05,
+      "loss": 0.2311,
+      "step": 1179
+    },
+    {
+      "epoch": 4.72,
+      "grad_norm": 1.2443056106567383,
+      "learning_rate": 3.8286573146292585e-05,
+      "loss": 0.2011,
+      "step": 1180
+    },
+    {
+      "epoch": 4.724,
+      "grad_norm": 1.3085763454437256,
+      "learning_rate": 3.8276553106212426e-05,
+      "loss": 0.2363,
+      "step": 1181
+    },
+    {
+      "epoch": 4.728,
+      "grad_norm": 1.3546162843704224,
+      "learning_rate": 3.826653306613227e-05,
+      "loss": 0.2611,
+      "step": 1182
+    },
+    {
+      "epoch": 4.732,
+      "grad_norm": 1.2781968116760254,
+      "learning_rate": 3.82565130260521e-05,
+      "loss": 0.207,
+      "step": 1183
+    },
+    {
+      "epoch": 4.736,
+      "grad_norm": 3.8161380290985107,
+      "learning_rate": 3.824649298597194e-05,
+      "loss": 0.2138,
+      "step": 1184
+    },
+    {
+      "epoch": 4.74,
+      "grad_norm": 1.2593066692352295,
+      "learning_rate": 3.823647294589179e-05,
+      "loss": 0.2364,
+      "step": 1185
+    },
+    {
+      "epoch": 4.744,
+      "grad_norm": 1.3970239162445068,
+      "learning_rate": 3.8226452905811626e-05,
+      "loss": 0.209,
+      "step": 1186
+    },
+    {
+      "epoch": 4.748,
+      "grad_norm": 1.306069016456604,
+      "learning_rate": 3.821643286573147e-05,
+      "loss": 0.2443,
+      "step": 1187
+    },
+    {
+      "epoch": 4.752,
+      "grad_norm": 1.1042245626449585,
+      "learning_rate": 3.820641282565131e-05,
+      "loss": 0.2157,
+      "step": 1188
+    },
+    {
+      "epoch": 4.756,
+      "grad_norm": 1.157153606414795,
+      "learning_rate": 3.819639278557114e-05,
+      "loss": 0.2211,
+      "step": 1189
+    },
+    {
+      "epoch": 4.76,
+      "grad_norm": 1.1273726224899292,
+      "learning_rate": 3.8186372745490984e-05,
+      "loss": 0.2221,
+      "step": 1190
+    },
+    {
+      "epoch": 4.764,
+      "grad_norm": 1.3015018701553345,
+      "learning_rate": 3.8176352705410825e-05,
+      "loss": 0.2381,
+      "step": 1191
+    },
+    {
+      "epoch": 4.768,
+      "grad_norm": 1.152994990348816,
+      "learning_rate": 3.816633266533066e-05,
+      "loss": 0.2523,
+      "step": 1192
+    },
+    {
+      "epoch": 4.772,
+      "grad_norm": 1.5288457870483398,
+      "learning_rate": 3.81563126252505e-05,
+      "loss": 0.2474,
+      "step": 1193
+    },
+    {
+      "epoch": 4.776,
+      "grad_norm": 1.488034963607788,
+      "learning_rate": 3.814629258517034e-05,
+      "loss": 0.2602,
+      "step": 1194
+    },
+    {
+      "epoch": 4.78,
+      "grad_norm": 1.257806420326233,
+      "learning_rate": 3.8136272545090184e-05,
+      "loss": 0.2165,
+      "step": 1195
+    },
+    {
+      "epoch": 4.784,
+      "grad_norm": 1.1476024389266968,
+      "learning_rate": 3.8126252505010025e-05,
+      "loss": 0.2374,
+      "step": 1196
+    },
+    {
+      "epoch": 4.788,
+      "grad_norm": 1.3641656637191772,
+      "learning_rate": 3.8116232464929866e-05,
+      "loss": 0.2371,
+      "step": 1197
+    },
+    {
+      "epoch": 4.792,
+      "grad_norm": 1.5159633159637451,
+      "learning_rate": 3.81062124248497e-05,
+      "loss": 0.2511,
+      "step": 1198
+    },
+    {
+      "epoch": 4.796,
+      "grad_norm": 1.186079740524292,
+      "learning_rate": 3.809619238476954e-05,
+      "loss": 0.2566,
+      "step": 1199
+    },
+    {
+      "epoch": 4.8,
+      "grad_norm": 1.5716686248779297,
+      "learning_rate": 3.808617234468938e-05,
+      "loss": 0.2318,
+      "step": 1200
+    },
+    {
+      "epoch": 4.804,
+      "grad_norm": 1.3118908405303955,
+      "learning_rate": 3.807615230460922e-05,
+      "loss": 0.2452,
+      "step": 1201
+    },
+    {
+      "epoch": 4.808,
+      "grad_norm": 1.3244889974594116,
+      "learning_rate": 3.806613226452906e-05,
+      "loss": 0.2362,
+      "step": 1202
+    },
+    {
+      "epoch": 4.812,
+      "grad_norm": 1.1287165880203247,
+      "learning_rate": 3.80561122244489e-05,
+      "loss": 0.1915,
+      "step": 1203
+    },
+    {
+      "epoch": 4.816,
+      "grad_norm": 1.2781370878219604,
+      "learning_rate": 3.8046092184368735e-05,
+      "loss": 0.2024,
+      "step": 1204
+    },
+    {
+      "epoch": 4.82,
+      "grad_norm": 1.2718026638031006,
+      "learning_rate": 3.803607214428858e-05,
+      "loss": 0.2099,
+      "step": 1205
+    },
+    {
+      "epoch": 4.824,
+      "grad_norm": 1.2175809144973755,
+      "learning_rate": 3.802605210420842e-05,
+      "loss": 0.2387,
+      "step": 1206
+    },
+    {
+      "epoch": 4.828,
+      "grad_norm": 1.1810656785964966,
+      "learning_rate": 3.801603206412826e-05,
+      "loss": 0.2773,
+      "step": 1207
+    },
+    {
+      "epoch": 4.832,
+      "grad_norm": 1.0714348554611206,
+      "learning_rate": 3.80060120240481e-05,
+      "loss": 0.1103,
+      "step": 1208
+    },
+    {
+      "epoch": 4.836,
+      "grad_norm": 1.2614023685455322,
+      "learning_rate": 3.7995991983967935e-05,
+      "loss": 0.2011,
+      "step": 1209
+    },
+    {
+      "epoch": 4.84,
+      "grad_norm": 1.2104132175445557,
+      "learning_rate": 3.7985971943887776e-05,
+      "loss": 0.2036,
+      "step": 1210
+    },
+    {
+      "epoch": 4.844,
+      "grad_norm": 1.322792887687683,
+      "learning_rate": 3.797595190380762e-05,
+      "loss": 0.246,
+      "step": 1211
+    },
+    {
+      "epoch": 4.848,
+      "grad_norm": 1.1660313606262207,
+      "learning_rate": 3.796593186372746e-05,
+      "loss": 0.2658,
+      "step": 1212
+    },
+    {
+      "epoch": 4.852,
+      "grad_norm": 1.3373457193374634,
+      "learning_rate": 3.795591182364729e-05,
+      "loss": 0.2282,
+      "step": 1213
+    },
+    {
+      "epoch": 4.856,
+      "grad_norm": 1.284891963005066,
+      "learning_rate": 3.7945891783567134e-05,
+      "loss": 0.2356,
+      "step": 1214
+    },
+    {
+      "epoch": 4.86,
+      "grad_norm": 1.2078474760055542,
+      "learning_rate": 3.7935871743486975e-05,
+      "loss": 0.2295,
+      "step": 1215
+    },
+    {
+      "epoch": 4.864,
+      "grad_norm": 1.158579707145691,
+      "learning_rate": 3.792585170340682e-05,
+      "loss": 0.2112,
+      "step": 1216
+    },
+    {
+      "epoch": 4.868,
+      "grad_norm": 1.3074941635131836,
+      "learning_rate": 3.791583166332666e-05,
+      "loss": 0.2361,
+      "step": 1217
+    },
+    {
+      "epoch": 4.872,
+      "grad_norm": 0.9688441157341003,
+      "learning_rate": 3.790581162324649e-05,
+      "loss": 0.1319,
+      "step": 1218
+    },
+    {
+      "epoch": 4.876,
+      "grad_norm": 1.5012363195419312,
+      "learning_rate": 3.7895791583166334e-05,
+      "loss": 0.2426,
+      "step": 1219
+    },
+    {
+      "epoch": 4.88,
+      "grad_norm": 1.236053228378296,
+      "learning_rate": 3.7885771543086175e-05,
+      "loss": 0.2136,
+      "step": 1220
+    },
+    {
+      "epoch": 4.884,
+      "grad_norm": 1.2395013570785522,
+      "learning_rate": 3.787575150300601e-05,
+      "loss": 0.2054,
+      "step": 1221
+    },
+    {
+      "epoch": 4.888,
+      "grad_norm": 1.3079355955123901,
+      "learning_rate": 3.786573146292585e-05,
+      "loss": 0.2243,
+      "step": 1222
+    },
+    {
+      "epoch": 4.892,
+      "grad_norm": 1.2592490911483765,
+      "learning_rate": 3.785571142284569e-05,
+      "loss": 0.2462,
+      "step": 1223
+    },
+    {
+      "epoch": 4.896,
+      "grad_norm": 1.615578055381775,
+      "learning_rate": 3.7845691382765533e-05,
+      "loss": 0.3059,
+      "step": 1224
+    },
+    {
+      "epoch": 4.9,
+      "grad_norm": 1.2457841634750366,
+      "learning_rate": 3.7835671342685375e-05,
+      "loss": 0.1957,
+      "step": 1225
+    },
+    {
+      "epoch": 4.904,
+      "grad_norm": 1.286840558052063,
+      "learning_rate": 3.7825651302605216e-05,
+      "loss": 0.2372,
+      "step": 1226
+    },
+    {
+      "epoch": 4.908,
+      "grad_norm": 1.2358272075653076,
+      "learning_rate": 3.781563126252505e-05,
+      "loss": 0.2342,
+      "step": 1227
+    },
+    {
+      "epoch": 4.912,
+      "grad_norm": 1.3874037265777588,
+      "learning_rate": 3.780561122244489e-05,
+      "loss": 0.2249,
+      "step": 1228
+    },
+    {
+      "epoch": 4.916,
+      "grad_norm": 1.2254422903060913,
+      "learning_rate": 3.779559118236473e-05,
+      "loss": 0.2138,
+      "step": 1229
+    },
+    {
+      "epoch": 4.92,
+      "grad_norm": 1.2836542129516602,
+      "learning_rate": 3.778557114228457e-05,
+      "loss": 0.2212,
+      "step": 1230
+    },
+    {
+      "epoch": 4.924,
+      "grad_norm": 1.3436830043792725,
+      "learning_rate": 3.777555110220441e-05,
+      "loss": 0.1943,
+      "step": 1231
+    },
+    {
+      "epoch": 4.928,
+      "grad_norm": 1.2510602474212646,
+      "learning_rate": 3.776553106212425e-05,
+      "loss": 0.2079,
+      "step": 1232
+    },
+    {
+      "epoch": 4.932,
+      "grad_norm": 1.2235398292541504,
+      "learning_rate": 3.7755511022044085e-05,
+      "loss": 0.245,
+      "step": 1233
+    },
+    {
+      "epoch": 4.936,
+      "grad_norm": 1.2405413389205933,
+      "learning_rate": 3.774549098196393e-05,
+      "loss": 0.2368,
+      "step": 1234
+    },
+    {
+      "epoch": 4.9399999999999995,
+      "grad_norm": 1.2697187662124634,
+      "learning_rate": 3.7735470941883774e-05,
+      "loss": 0.2472,
+      "step": 1235
+    },
+    {
+      "epoch": 4.944,
+      "grad_norm": 1.250165343284607,
+      "learning_rate": 3.772545090180361e-05,
+      "loss": 0.245,
+      "step": 1236
+    },
+    {
+      "epoch": 4.948,
+      "grad_norm": 1.3872276544570923,
+      "learning_rate": 3.771543086172345e-05,
+      "loss": 0.2394,
+      "step": 1237
+    },
+    {
+      "epoch": 4.952,
+      "grad_norm": 1.4038046598434448,
+      "learning_rate": 3.770541082164329e-05,
+      "loss": 0.2684,
+      "step": 1238
+    },
+    {
+      "epoch": 4.9559999999999995,
+      "grad_norm": 1.1734498739242554,
+      "learning_rate": 3.7695390781563126e-05,
+      "loss": 0.2401,
+      "step": 1239
+    },
+    {
+      "epoch": 4.96,
+      "grad_norm": 1.181815266609192,
+      "learning_rate": 3.768537074148297e-05,
+      "loss": 0.2575,
+      "step": 1240
+    },
+    {
+      "epoch": 4.964,
+      "grad_norm": 1.4323796033859253,
+      "learning_rate": 3.767535070140281e-05,
+      "loss": 0.2368,
+      "step": 1241
+    },
+    {
+      "epoch": 4.968,
+      "grad_norm": 1.2690467834472656,
+      "learning_rate": 3.766533066132264e-05,
+      "loss": 0.2726,
+      "step": 1242
+    },
+    {
+      "epoch": 4.9719999999999995,
+      "grad_norm": 1.060066819190979,
+      "learning_rate": 3.7655310621242484e-05,
+      "loss": 0.2176,
+      "step": 1243
+    },
+    {
+      "epoch": 4.976,
+      "grad_norm": 1.293988823890686,
+      "learning_rate": 3.764529058116233e-05,
+      "loss": 0.2382,
+      "step": 1244
+    },
+    {
+      "epoch": 4.98,
+      "grad_norm": 1.1977914571762085,
+      "learning_rate": 3.7635270541082167e-05,
+      "loss": 0.2414,
+      "step": 1245
+    },
+    {
+      "epoch": 4.984,
+      "grad_norm": 1.373465895652771,
+      "learning_rate": 3.762525050100201e-05,
+      "loss": 0.2524,
+      "step": 1246
+    },
+    {
+      "epoch": 4.9879999999999995,
+      "grad_norm": 1.2298009395599365,
+      "learning_rate": 3.761523046092185e-05,
+      "loss": 0.2156,
+      "step": 1247
+    },
+    {
+      "epoch": 4.992,
+      "grad_norm": 1.460239052772522,
+      "learning_rate": 3.7605210420841684e-05,
+      "loss": 0.2943,
+      "step": 1248
+    },
+    {
+      "epoch": 4.996,
+      "grad_norm": 1.2077012062072754,
+      "learning_rate": 3.7595190380761525e-05,
+      "loss": 0.2488,
+      "step": 1249
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 1.2159690856933594,
+      "learning_rate": 3.7585170340681366e-05,
+      "loss": 0.237,
+      "step": 1250
+    },
+    {
+      "epoch": 5.004,
+      "grad_norm": 1.1079336404800415,
+      "learning_rate": 3.75751503006012e-05,
+      "loss": 0.1713,
+      "step": 1251
+    },
+    {
+      "epoch": 5.008,
+      "grad_norm": 1.1447618007659912,
+      "learning_rate": 3.756513026052104e-05,
+      "loss": 0.1614,
+      "step": 1252
+    },
+    {
+      "epoch": 5.012,
+      "grad_norm": 1.0622740983963013,
+      "learning_rate": 3.755511022044088e-05,
+      "loss": 0.1568,
+      "step": 1253
+    },
+    {
+      "epoch": 5.016,
+      "grad_norm": 1.1202747821807861,
+      "learning_rate": 3.7545090180360724e-05,
+      "loss": 0.1313,
+      "step": 1254
+    },
+    {
+      "epoch": 5.02,
+      "grad_norm": 1.1181975603103638,
+      "learning_rate": 3.7535070140280566e-05,
+      "loss": 0.166,
+      "step": 1255
+    },
+    {
+      "epoch": 5.024,
+      "grad_norm": 10.162293434143066,
+      "learning_rate": 3.752505010020041e-05,
+      "loss": 0.0938,
+      "step": 1256
+    },
+    {
+      "epoch": 5.028,
+      "grad_norm": 1.2769560813903809,
+      "learning_rate": 3.751503006012024e-05,
+      "loss": 0.1378,
+      "step": 1257
+    },
+    {
+      "epoch": 5.032,
+      "grad_norm": 1.2073560953140259,
+      "learning_rate": 3.750501002004008e-05,
+      "loss": 0.1715,
+      "step": 1258
+    },
+    {
+      "epoch": 5.036,
+      "grad_norm": 1.229649543762207,
+      "learning_rate": 3.7494989979959924e-05,
+      "loss": 0.14,
+      "step": 1259
+    },
+    {
+      "epoch": 5.04,
+      "grad_norm": 1.3796989917755127,
+      "learning_rate": 3.748496993987976e-05,
+      "loss": 0.189,
+      "step": 1260
+    },
+    {
+      "epoch": 5.044,
+      "grad_norm": 1.5101726055145264,
+      "learning_rate": 3.74749498997996e-05,
+      "loss": 0.1487,
+      "step": 1261
+    },
+    {
+      "epoch": 5.048,
+      "grad_norm": 1.451194167137146,
+      "learning_rate": 3.746492985971944e-05,
+      "loss": 0.1871,
+      "step": 1262
+    },
+    {
+      "epoch": 5.052,
+      "grad_norm": 1.608769416809082,
+      "learning_rate": 3.7454909819639276e-05,
+      "loss": 0.1676,
+      "step": 1263
+    },
+    {
+      "epoch": 5.056,
+      "grad_norm": 1.3911021947860718,
+      "learning_rate": 3.7444889779559124e-05,
+      "loss": 0.1217,
+      "step": 1264
+    },
+    {
+      "epoch": 5.06,
+      "grad_norm": 1.4321279525756836,
+      "learning_rate": 3.743486973947896e-05,
+      "loss": 0.1725,
+      "step": 1265
+    },
+    {
+      "epoch": 5.064,
+      "grad_norm": 1.49961519241333,
+      "learning_rate": 3.74248496993988e-05,
+      "loss": 0.1564,
+      "step": 1266
+    },
+    {
+      "epoch": 5.068,
+      "grad_norm": 1.374585509300232,
+      "learning_rate": 3.741482965931864e-05,
+      "loss": 0.1183,
+      "step": 1267
+    },
+    {
+      "epoch": 5.072,
+      "grad_norm": 1.5012727975845337,
+      "learning_rate": 3.740480961923848e-05,
+      "loss": 0.1466,
+      "step": 1268
+    },
+    {
+      "epoch": 5.076,
+      "grad_norm": Infinity,
+      "learning_rate": 3.740480961923848e-05,
+      "loss": 0.3217,
+      "step": 1269
+    },
+    {
+      "epoch": 5.08,
+      "grad_norm": 1.511431336402893,
+      "learning_rate": 3.739478957915832e-05,
+      "loss": 0.1452,
+      "step": 1270
+    },
+    {
+      "epoch": 5.084,
+      "grad_norm": 1.2745331525802612,
+      "learning_rate": 3.738476953907816e-05,
+      "loss": 0.1128,
+      "step": 1271
+    },
+    {
+      "epoch": 5.088,
+      "grad_norm": 1.2193913459777832,
+      "learning_rate": 3.7374749498998e-05,
+      "loss": 0.1502,
+      "step": 1272
+    },
+    {
+      "epoch": 5.092,
+      "grad_norm": 1.190537929534912,
+      "learning_rate": 3.7364729458917834e-05,
+      "loss": 0.117,
+      "step": 1273
+    },
+    {
+      "epoch": 5.096,
+      "grad_norm": 1.1268037557601929,
+      "learning_rate": 3.7354709418837675e-05,
+      "loss": 0.1432,
+      "step": 1274
+    },
+    {
+      "epoch": 5.1,
+      "grad_norm": 1.1972516775131226,
+      "learning_rate": 3.7344689378757516e-05,
+      "loss": 0.1263,
+      "step": 1275
+    },
+    {
+      "epoch": 5.104,
+      "grad_norm": 1.2640272378921509,
+      "learning_rate": 3.733466933867736e-05,
+      "loss": 0.1393,
+      "step": 1276
+    },
+    {
+      "epoch": 5.108,
+      "grad_norm": 1.2229938507080078,
+      "learning_rate": 3.73246492985972e-05,
+      "loss": 0.1254,
+      "step": 1277
+    },
+    {
+      "epoch": 5.112,
+      "grad_norm": 1.4548741579055786,
+      "learning_rate": 3.731462925851703e-05,
+      "loss": 0.129,
+      "step": 1278
+    },
+    {
+      "epoch": 5.116,
+      "grad_norm": 1.5529509782791138,
+      "learning_rate": 3.7304609218436875e-05,
+      "loss": 0.1918,
+      "step": 1279
+    },
+    {
+      "epoch": 5.12,
+      "grad_norm": 1.473236322402954,
+      "learning_rate": 3.7294589178356716e-05,
+      "loss": 0.129,
+      "step": 1280
+    },
+    {
+      "epoch": 5.124,
+      "grad_norm": 1.4448833465576172,
+      "learning_rate": 3.728456913827656e-05,
+      "loss": 0.1673,
+      "step": 1281
+    },
+    {
+      "epoch": 5.128,
+      "grad_norm": 1.455904245376587,
+      "learning_rate": 3.727454909819639e-05,
+      "loss": 0.1614,
+      "step": 1282
+    },
+    {
+      "epoch": 5.132,
+      "grad_norm": 47.209869384765625,
+      "learning_rate": 3.726452905811623e-05,
+      "loss": 0.6446,
+      "step": 1283
+    },
+    {
+      "epoch": 5.136,
+      "grad_norm": 1.2592720985412598,
+      "learning_rate": 3.7254509018036074e-05,
+      "loss": 0.1518,
+      "step": 1284
+    },
+    {
+      "epoch": 5.14,
+      "grad_norm": 1.3965915441513062,
+      "learning_rate": 3.7244488977955916e-05,
+      "loss": 0.1286,
+      "step": 1285
+    },
+    {
+      "epoch": 5.144,
+      "grad_norm": 1.0066782236099243,
+      "learning_rate": 3.723446893787576e-05,
+      "loss": 0.0742,
+      "step": 1286
+    },
+    {
+      "epoch": 5.148,
+      "grad_norm": 1.2181978225708008,
+      "learning_rate": 3.722444889779559e-05,
+      "loss": 0.1358,
+      "step": 1287
+    },
+    {
+      "epoch": 5.152,
+      "grad_norm": 1.3564952611923218,
+      "learning_rate": 3.721442885771543e-05,
+      "loss": 0.1212,
+      "step": 1288
+    },
+    {
+      "epoch": 5.156,
+      "grad_norm": 1.4054635763168335,
+      "learning_rate": 3.7204408817635274e-05,
+      "loss": 0.1534,
+      "step": 1289
+    },
+    {
+      "epoch": 5.16,
+      "grad_norm": 1.1692203283309937,
+      "learning_rate": 3.719438877755511e-05,
+      "loss": 0.1135,
+      "step": 1290
+    },
+    {
+      "epoch": 5.164,
+      "grad_norm": 1.528304934501648,
+      "learning_rate": 3.718436873747495e-05,
+      "loss": 0.1787,
+      "step": 1291
+    },
+    {
+      "epoch": 5.168,
+      "grad_norm": 1.4810398817062378,
+      "learning_rate": 3.717434869739479e-05,
+      "loss": 0.1711,
+      "step": 1292
+    },
+    {
+      "epoch": 5.172,
+      "grad_norm": 1.4008511304855347,
+      "learning_rate": 3.7164328657314625e-05,
+      "loss": 0.107,
+      "step": 1293
+    },
+    {
+      "epoch": 5.176,
+      "grad_norm": 1.5215219259262085,
+      "learning_rate": 3.7154308617234474e-05,
+      "loss": 0.1714,
+      "step": 1294
+    },
+    {
+      "epoch": 5.18,
+      "grad_norm": 1.3255380392074585,
+      "learning_rate": 3.7144288577154315e-05,
+      "loss": 0.1437,
+      "step": 1295
+    },
+    {
+      "epoch": 5.184,
+      "grad_norm": 1.540545105934143,
+      "learning_rate": 3.713426853707415e-05,
+      "loss": 0.1374,
+      "step": 1296
+    },
+    {
+      "epoch": 5.188,
+      "grad_norm": 1.2621541023254395,
+      "learning_rate": 3.712424849699399e-05,
+      "loss": 0.1114,
+      "step": 1297
+    },
+    {
+      "epoch": 5.192,
+      "grad_norm": 1.355939507484436,
+      "learning_rate": 3.711422845691383e-05,
+      "loss": 0.135,
+      "step": 1298
+    },
+    {
+      "epoch": 5.196,
+      "grad_norm": 1.5436731576919556,
+      "learning_rate": 3.7104208416833666e-05,
+      "loss": 0.1471,
+      "step": 1299
+    },
+    {
+      "epoch": 5.2,
+      "grad_norm": 1.4287269115447998,
+      "learning_rate": 3.709418837675351e-05,
+      "loss": 0.147,
+      "step": 1300
+    },
+    {
+      "epoch": 5.204,
+      "grad_norm": 1.37587308883667,
+      "learning_rate": 3.708416833667335e-05,
+      "loss": 0.1434,
+      "step": 1301
+    },
+    {
+      "epoch": 5.208,
+      "grad_norm": 1.3519721031188965,
+      "learning_rate": 3.7074148296593183e-05,
+      "loss": 0.1418,
+      "step": 1302
+    },
+    {
+      "epoch": 5.212,
+      "grad_norm": 1.3758410215377808,
+      "learning_rate": 3.7064128256513025e-05,
+      "loss": 0.1535,
+      "step": 1303
+    },
+    {
+      "epoch": 5.216,
+      "grad_norm": 1.3432583808898926,
+      "learning_rate": 3.705410821643287e-05,
+      "loss": 0.1477,
+      "step": 1304
+    },
+    {
+      "epoch": 5.22,
+      "grad_norm": 1.2263681888580322,
+      "learning_rate": 3.704408817635271e-05,
+      "loss": 0.1374,
+      "step": 1305
+    },
+    {
+      "epoch": 5.224,
+      "grad_norm": 1.2747381925582886,
+      "learning_rate": 3.703406813627255e-05,
+      "loss": 0.1498,
+      "step": 1306
+    },
+    {
+      "epoch": 5.228,
+      "grad_norm": 1.4037673473358154,
+      "learning_rate": 3.702404809619239e-05,
+      "loss": 0.1539,
+      "step": 1307
+    },
+    {
+      "epoch": 5.232,
+      "grad_norm": 1.4913922548294067,
+      "learning_rate": 3.7014028056112224e-05,
+      "loss": 0.1381,
+      "step": 1308
+    },
+    {
+      "epoch": 5.236,
+      "grad_norm": 1.3626859188079834,
+      "learning_rate": 3.7004008016032066e-05,
+      "loss": 0.1257,
+      "step": 1309
+    },
+    {
+      "epoch": 5.24,
+      "grad_norm": 1.508866548538208,
+      "learning_rate": 3.699398797595191e-05,
+      "loss": 0.1854,
+      "step": 1310
+    },
+    {
+      "epoch": 5.244,
+      "grad_norm": 1.3557127714157104,
+      "learning_rate": 3.698396793587174e-05,
+      "loss": 0.1417,
+      "step": 1311
+    },
+    {
+      "epoch": 5.248,
+      "grad_norm": 1.4805433750152588,
+      "learning_rate": 3.697394789579158e-05,
+      "loss": 0.1596,
+      "step": 1312
+    },
+    {
+      "epoch": 5.252,
+      "grad_norm": 1.4268604516983032,
+      "learning_rate": 3.6963927855711424e-05,
+      "loss": 0.1612,
+      "step": 1313
+    },
+    {
+      "epoch": 5.256,
+      "grad_norm": 1.4338222742080688,
+      "learning_rate": 3.6953907815631265e-05,
+      "loss": 0.1668,
+      "step": 1314
+    },
+    {
+      "epoch": 5.26,
+      "grad_norm": 1.3227003812789917,
+      "learning_rate": 3.6943887775551107e-05,
+      "loss": 0.1342,
+      "step": 1315
+    },
+    {
+      "epoch": 5.264,
+      "grad_norm": 1.3900845050811768,
+      "learning_rate": 3.693386773547095e-05,
+      "loss": 0.1342,
+      "step": 1316
+    },
+    {
+      "epoch": 5.268,
+      "grad_norm": 1.3115166425704956,
+      "learning_rate": 3.692384769539078e-05,
+      "loss": 0.1544,
+      "step": 1317
+    },
+    {
+      "epoch": 5.272,
+      "grad_norm": 1.6373382806777954,
+      "learning_rate": 3.6913827655310624e-05,
+      "loss": 0.1649,
+      "step": 1318
+    },
+    {
+      "epoch": 5.276,
+      "grad_norm": 1.2519011497497559,
+      "learning_rate": 3.6903807615230465e-05,
+      "loss": 0.1448,
+      "step": 1319
+    },
+    {
+      "epoch": 5.28,
+      "grad_norm": 1.4340311288833618,
+      "learning_rate": 3.68937875751503e-05,
+      "loss": 0.1744,
+      "step": 1320
+    },
+    {
+      "epoch": 5.284,
+      "grad_norm": 1.7507641315460205,
+      "learning_rate": 3.688376753507014e-05,
+      "loss": 0.1714,
+      "step": 1321
+    },
+    {
+      "epoch": 5.288,
+      "grad_norm": 1.4718471765518188,
+      "learning_rate": 3.687374749498998e-05,
+      "loss": 0.1769,
+      "step": 1322
+    },
+    {
+      "epoch": 5.292,
+      "grad_norm": 1.3668906688690186,
+      "learning_rate": 3.6863727454909817e-05,
+      "loss": 0.1593,
+      "step": 1323
+    },
+    {
+      "epoch": 5.296,
+      "grad_norm": 1.5843260288238525,
+      "learning_rate": 3.6853707414829665e-05,
+      "loss": 0.1379,
+      "step": 1324
+    },
+    {
+      "epoch": 5.3,
+      "grad_norm": 1.3435697555541992,
+      "learning_rate": 3.6843687374749506e-05,
+      "loss": 0.1084,
+      "step": 1325
+    },
+    {
+      "epoch": 5.304,
+      "grad_norm": 1.4321434497833252,
+      "learning_rate": 3.683366733466934e-05,
+      "loss": 0.1472,
+      "step": 1326
+    },
+    {
+      "epoch": 5.308,
+      "grad_norm": 1.3599714040756226,
+      "learning_rate": 3.682364729458918e-05,
+      "loss": 0.1392,
+      "step": 1327
+    },
+    {
+      "epoch": 5.312,
+      "grad_norm": 1.3491992950439453,
+      "learning_rate": 3.681362725450902e-05,
+      "loss": 0.1574,
+      "step": 1328
+    },
+    {
+      "epoch": 5.316,
+      "grad_norm": 1.4255551099777222,
+      "learning_rate": 3.680360721442886e-05,
+      "loss": 0.1349,
+      "step": 1329
+    },
+    {
+      "epoch": 5.32,
+      "grad_norm": 1.34247624874115,
+      "learning_rate": 3.67935871743487e-05,
+      "loss": 0.1745,
+      "step": 1330
+    },
+    {
+      "epoch": 5.324,
+      "grad_norm": 1.4770880937576294,
+      "learning_rate": 3.678356713426854e-05,
+      "loss": 0.1558,
+      "step": 1331
+    },
+    {
+      "epoch": 5.328,
+      "grad_norm": 1.468400001525879,
+      "learning_rate": 3.6773547094188375e-05,
+      "loss": 0.1388,
+      "step": 1332
+    },
+    {
+      "epoch": 5.332,
+      "grad_norm": 1.588028907775879,
+      "learning_rate": 3.6763527054108216e-05,
+      "loss": 0.1722,
+      "step": 1333
+    },
+    {
+      "epoch": 5.336,
+      "grad_norm": 1.3578060865402222,
+      "learning_rate": 3.675350701402806e-05,
+      "loss": 0.1314,
+      "step": 1334
+    },
+    {
+      "epoch": 5.34,
+      "grad_norm": 1.7125457525253296,
+      "learning_rate": 3.67434869739479e-05,
+      "loss": 0.1409,
+      "step": 1335
+    },
+    {
+      "epoch": 5.344,
+      "grad_norm": 1.482620358467102,
+      "learning_rate": 3.673346693386774e-05,
+      "loss": 0.1242,
+      "step": 1336
+    },
+    {
+      "epoch": 5.348,
+      "grad_norm": 1.468563437461853,
+      "learning_rate": 3.6723446893787574e-05,
+      "loss": 0.1509,
+      "step": 1337
+    },
+    {
+      "epoch": 5.352,
+      "grad_norm": 1.4916741847991943,
+      "learning_rate": 3.6713426853707415e-05,
+      "loss": 0.1477,
+      "step": 1338
+    },
+    {
+      "epoch": 5.356,
+      "grad_norm": 1.3862581253051758,
+      "learning_rate": 3.670340681362726e-05,
+      "loss": 0.1625,
+      "step": 1339
+    },
+    {
+      "epoch": 5.36,
+      "grad_norm": 1.3407968282699585,
+      "learning_rate": 3.66933867735471e-05,
+      "loss": 0.1499,
+      "step": 1340
+    },
+    {
+      "epoch": 5.364,
+      "grad_norm": 1.010241985321045,
+      "learning_rate": 3.668336673346693e-05,
+      "loss": 0.0938,
+      "step": 1341
+    },
+    {
+      "epoch": 5.368,
+      "grad_norm": 1.3833948373794556,
+      "learning_rate": 3.6673346693386774e-05,
+      "loss": 0.1439,
+      "step": 1342
+    },
+    {
+      "epoch": 5.372,
+      "grad_norm": 1.3611929416656494,
+      "learning_rate": 3.6663326653306615e-05,
+      "loss": 0.1722,
+      "step": 1343
+    },
+    {
+      "epoch": 5.376,
+      "grad_norm": 1.1714794635772705,
+      "learning_rate": 3.6653306613226456e-05,
+      "loss": 0.093,
+      "step": 1344
+    },
+    {
+      "epoch": 5.38,
+      "grad_norm": 1.0513854026794434,
+      "learning_rate": 3.66432865731463e-05,
+      "loss": 0.1028,
+      "step": 1345
+    },
+    {
+      "epoch": 5.384,
+      "grad_norm": 1.432370662689209,
+      "learning_rate": 3.663326653306613e-05,
+      "loss": 0.1621,
+      "step": 1346
+    },
+    {
+      "epoch": 5.388,
+      "grad_norm": 1.3324103355407715,
+      "learning_rate": 3.6623246492985973e-05,
+      "loss": 0.1261,
+      "step": 1347
+    },
+    {
+      "epoch": 5.392,
+      "grad_norm": 1.4653232097625732,
+      "learning_rate": 3.6613226452905815e-05,
+      "loss": 0.1342,
+      "step": 1348
+    },
+    {
+      "epoch": 5.396,
+      "grad_norm": 1.4007515907287598,
+      "learning_rate": 3.660320641282565e-05,
+      "loss": 0.1296,
+      "step": 1349
+    },
+    {
+      "epoch": 5.4,
+      "grad_norm": 1.5396267175674438,
+      "learning_rate": 3.659318637274549e-05,
+      "loss": 0.1519,
+      "step": 1350
+    },
+    {
+      "epoch": 5.404,
+      "grad_norm": 1.4199961423873901,
+      "learning_rate": 3.658316633266533e-05,
+      "loss": 0.1516,
+      "step": 1351
+    },
+    {
+      "epoch": 5.408,
+      "grad_norm": 1.455402135848999,
+      "learning_rate": 3.657314629258517e-05,
+      "loss": 0.19,
+      "step": 1352
+    },
+    {
+      "epoch": 5.412,
+      "grad_norm": 1.4118759632110596,
+      "learning_rate": 3.6563126252505014e-05,
+      "loss": 0.1687,
+      "step": 1353
+    },
+    {
+      "epoch": 5.416,
+      "grad_norm": 1.4946292638778687,
+      "learning_rate": 3.6553106212424856e-05,
+      "loss": 0.1587,
+      "step": 1354
+    },
+    {
+      "epoch": 5.42,
+      "grad_norm": 1.4371452331542969,
+      "learning_rate": 3.654308617234469e-05,
+      "loss": 0.1722,
+      "step": 1355
+    },
+    {
+      "epoch": 5.424,
+      "grad_norm": 1.4597325325012207,
+      "learning_rate": 3.653306613226453e-05,
+      "loss": 0.1954,
+      "step": 1356
+    },
+    {
+      "epoch": 5.428,
+      "grad_norm": 1.463151216506958,
+      "learning_rate": 3.652304609218437e-05,
+      "loss": 0.2056,
+      "step": 1357
+    },
+    {
+      "epoch": 5.432,
+      "grad_norm": 1.568338394165039,
+      "learning_rate": 3.651302605210421e-05,
+      "loss": 0.1665,
+      "step": 1358
+    },
+    {
+      "epoch": 5.436,
+      "grad_norm": 1.4030911922454834,
+      "learning_rate": 3.650300601202405e-05,
+      "loss": 0.1463,
+      "step": 1359
+    },
+    {
+      "epoch": 5.44,
+      "grad_norm": 1.3122279644012451,
+      "learning_rate": 3.649298597194389e-05,
+      "loss": 0.1324,
+      "step": 1360
+    },
+    {
+      "epoch": 5.444,
+      "grad_norm": 1.398829698562622,
+      "learning_rate": 3.6482965931863724e-05,
+      "loss": 0.182,
+      "step": 1361
+    },
+    {
+      "epoch": 5.448,
+      "grad_norm": 1.5118191242218018,
+      "learning_rate": 3.6472945891783566e-05,
+      "loss": 0.1485,
+      "step": 1362
+    },
+    {
+      "epoch": 5.452,
+      "grad_norm": 1.4859402179718018,
+      "learning_rate": 3.6462925851703414e-05,
+      "loss": 0.1429,
+      "step": 1363
+    },
+    {
+      "epoch": 5.456,
+      "grad_norm": 1.5563009977340698,
+      "learning_rate": 3.645290581162325e-05,
+      "loss": 0.1834,
+      "step": 1364
+    },
+    {
+      "epoch": 5.46,
+      "grad_norm": 1.403515338897705,
+      "learning_rate": 3.644288577154309e-05,
+      "loss": 0.1533,
+      "step": 1365
+    },
+    {
+      "epoch": 5.464,
+      "grad_norm": 1.3895883560180664,
+      "learning_rate": 3.643286573146293e-05,
+      "loss": 0.1668,
+      "step": 1366
+    },
+    {
+      "epoch": 5.468,
+      "grad_norm": 1.454662561416626,
+      "learning_rate": 3.6422845691382765e-05,
+      "loss": 0.1376,
+      "step": 1367
+    },
+    {
+      "epoch": 5.4719999999999995,
+      "grad_norm": 1.4135189056396484,
+      "learning_rate": 3.6412825651302606e-05,
+      "loss": 0.132,
+      "step": 1368
+    },
+    {
+      "epoch": 5.476,
+      "grad_norm": 1.3551510572433472,
+      "learning_rate": 3.640280561122245e-05,
+      "loss": 0.2015,
+      "step": 1369
+    },
+    {
+      "epoch": 5.48,
+      "grad_norm": 1.4458680152893066,
+      "learning_rate": 3.639278557114228e-05,
+      "loss": 0.1309,
+      "step": 1370
+    },
+    {
+      "epoch": 5.484,
+      "grad_norm": 1.508864402770996,
+      "learning_rate": 3.6382765531062124e-05,
+      "loss": 0.1945,
+      "step": 1371
+    },
+    {
+      "epoch": 5.4879999999999995,
+      "grad_norm": 1.439699411392212,
+      "learning_rate": 3.6372745490981965e-05,
+      "loss": 0.1537,
+      "step": 1372
+    },
+    {
+      "epoch": 5.492,
+      "grad_norm": 1.516624093055725,
+      "learning_rate": 3.6362725450901806e-05,
+      "loss": 0.1503,
+      "step": 1373
+    },
+    {
+      "epoch": 5.496,
+      "grad_norm": 1.4190101623535156,
+      "learning_rate": 3.635270541082165e-05,
+      "loss": 0.1483,
+      "step": 1374
+    },
+    {
+      "epoch": 5.5,
+      "grad_norm": 1.4877848625183105,
+      "learning_rate": 3.634268537074149e-05,
+      "loss": 0.1409,
+      "step": 1375
+    },
+    {
+      "epoch": 5.504,
+      "grad_norm": 1.448904037475586,
+      "learning_rate": 3.633266533066132e-05,
+      "loss": 0.1681,
+      "step": 1376
+    },
+    {
+      "epoch": 5.508,
+      "grad_norm": 1.4470781087875366,
+      "learning_rate": 3.6322645290581164e-05,
+      "loss": 0.1599,
+      "step": 1377
+    },
+    {
+      "epoch": 5.5120000000000005,
+      "grad_norm": 1.418031930923462,
+      "learning_rate": 3.6312625250501006e-05,
+      "loss": 0.1441,
+      "step": 1378
+    },
+    {
+      "epoch": 5.516,
+      "grad_norm": 1.4554909467697144,
+      "learning_rate": 3.630260521042084e-05,
+      "loss": 0.1463,
+      "step": 1379
+    },
+    {
+      "epoch": 5.52,
+      "grad_norm": 1.379330039024353,
+      "learning_rate": 3.629258517034068e-05,
+      "loss": 0.1285,
+      "step": 1380
+    },
+    {
+      "epoch": 5.524,
+      "grad_norm": 1.4990249872207642,
+      "learning_rate": 3.628256513026052e-05,
+      "loss": 0.2017,
+      "step": 1381
+    },
+    {
+      "epoch": 5.5280000000000005,
+      "grad_norm": 1.2808997631072998,
+      "learning_rate": 3.627254509018036e-05,
+      "loss": 0.1664,
+      "step": 1382
+    },
+    {
+      "epoch": 5.532,
+      "grad_norm": 1.6173251867294312,
+      "learning_rate": 3.6262525050100205e-05,
+      "loss": 0.174,
+      "step": 1383
+    },
+    {
+      "epoch": 5.536,
+      "grad_norm": 1.396598219871521,
+      "learning_rate": 3.625250501002005e-05,
+      "loss": 0.1454,
+      "step": 1384
+    },
+    {
+      "epoch": 5.54,
+      "grad_norm": 1.0782043933868408,
+      "learning_rate": 3.624248496993988e-05,
+      "loss": 0.119,
+      "step": 1385
+    },
+    {
+      "epoch": 5.5440000000000005,
+      "grad_norm": 1.3274977207183838,
+      "learning_rate": 3.623246492985972e-05,
+      "loss": 0.1369,
+      "step": 1386
+    },
+    {
+      "epoch": 5.548,
+      "grad_norm": 1.3669708967208862,
+      "learning_rate": 3.6222444889779564e-05,
+      "loss": 0.1481,
+      "step": 1387
+    },
+    {
+      "epoch": 5.552,
+      "grad_norm": 1.3589403629302979,
+      "learning_rate": 3.62124248496994e-05,
+      "loss": 0.2093,
+      "step": 1388
+    },
+    {
+      "epoch": 5.556,
+      "grad_norm": 1.4127261638641357,
+      "learning_rate": 3.620240480961924e-05,
+      "loss": 0.1664,
+      "step": 1389
+    },
+    {
+      "epoch": 5.5600000000000005,
+      "grad_norm": 1.4620351791381836,
+      "learning_rate": 3.619238476953908e-05,
+      "loss": 0.1434,
+      "step": 1390
+    },
+    {
+      "epoch": 5.564,
+      "grad_norm": 1.4212642908096313,
+      "learning_rate": 3.6182364729458915e-05,
+      "loss": 0.161,
+      "step": 1391
+    },
+    {
+      "epoch": 5.568,
+      "grad_norm": 1.4831411838531494,
+      "learning_rate": 3.6172344689378757e-05,
+      "loss": 0.17,
+      "step": 1392
+    },
+    {
+      "epoch": 5.572,
+      "grad_norm": 1.560911774635315,
+      "learning_rate": 3.61623246492986e-05,
+      "loss": 0.1737,
+      "step": 1393
+    },
+    {
+      "epoch": 5.576,
+      "grad_norm": 1.5256128311157227,
+      "learning_rate": 3.615230460921844e-05,
+      "loss": 0.1609,
+      "step": 1394
+    },
+    {
+      "epoch": 5.58,
+      "grad_norm": 1.407828688621521,
+      "learning_rate": 3.614228456913828e-05,
+      "loss": 0.1925,
+      "step": 1395
+    },
+    {
+      "epoch": 5.584,
+      "grad_norm": 1.3980634212493896,
+      "learning_rate": 3.613226452905812e-05,
+      "loss": 0.1535,
+      "step": 1396
+    },
+    {
+      "epoch": 5.588,
+      "grad_norm": 1.4387320280075073,
+      "learning_rate": 3.6122244488977956e-05,
+      "loss": 0.1489,
+      "step": 1397
+    },
+    {
+      "epoch": 5.592,
+      "grad_norm": 1.5303181409835815,
+      "learning_rate": 3.61122244488978e-05,
+      "loss": 0.1925,
+      "step": 1398
+    },
+    {
+      "epoch": 5.596,
+      "grad_norm": 1.5387216806411743,
+      "learning_rate": 3.610220440881764e-05,
+      "loss": 0.1709,
+      "step": 1399
+    },
+    {
+      "epoch": 5.6,
+      "grad_norm": 0.9596248865127563,
+      "learning_rate": 3.609218436873747e-05,
+      "loss": 0.0974,
+      "step": 1400
+    },
+    {
+      "epoch": 5.604,
+      "grad_norm": 1.4933253526687622,
+      "learning_rate": 3.6082164328657315e-05,
+      "loss": 0.2206,
+      "step": 1401
+    },
+    {
+      "epoch": 5.608,
+      "grad_norm": 1.3961461782455444,
+      "learning_rate": 3.6072144288577156e-05,
+      "loss": 0.1859,
+      "step": 1402
+    },
+    {
+      "epoch": 5.612,
+      "grad_norm": 1.2029403448104858,
+      "learning_rate": 3.6062124248497e-05,
+      "loss": 0.1289,
+      "step": 1403
+    },
+    {
+      "epoch": 5.616,
+      "grad_norm": 1.345209002494812,
+      "learning_rate": 3.605210420841684e-05,
+      "loss": 0.1448,
+      "step": 1404
+    },
+    {
+      "epoch": 5.62,
+      "grad_norm": 1.1782201528549194,
+      "learning_rate": 3.604208416833667e-05,
+      "loss": 0.1294,
+      "step": 1405
+    },
+    {
+      "epoch": 5.624,
+      "grad_norm": 1.461083173751831,
+      "learning_rate": 3.6032064128256514e-05,
+      "loss": 0.1534,
+      "step": 1406
+    },
+    {
+      "epoch": 5.628,
+      "grad_norm": 1.2814429998397827,
+      "learning_rate": 3.6022044088176356e-05,
+      "loss": 0.1259,
+      "step": 1407
+    },
+    {
+      "epoch": 5.632,
+      "grad_norm": 1.565420150756836,
+      "learning_rate": 3.60120240480962e-05,
+      "loss": 0.1861,
+      "step": 1408
+    },
+    {
+      "epoch": 5.636,
+      "grad_norm": 1.0360498428344727,
+      "learning_rate": 3.600200400801603e-05,
+      "loss": 0.0836,
+      "step": 1409
+    },
+    {
+      "epoch": 5.64,
+      "grad_norm": 1.4248356819152832,
+      "learning_rate": 3.599198396793587e-05,
+      "loss": 0.1988,
+      "step": 1410
+    },
+    {
+      "epoch": 5.644,
+      "grad_norm": 1.5098896026611328,
+      "learning_rate": 3.5981963927855714e-05,
+      "loss": 0.1708,
+      "step": 1411
+    },
+    {
+      "epoch": 5.648,
+      "grad_norm": 1.4749072790145874,
+      "learning_rate": 3.5971943887775555e-05,
+      "loss": 0.1791,
+      "step": 1412
+    },
+    {
+      "epoch": 5.652,
+      "grad_norm": 1.316433072090149,
+      "learning_rate": 3.5961923847695396e-05,
+      "loss": 0.1425,
+      "step": 1413
+    },
+    {
+      "epoch": 5.656,
+      "grad_norm": 1.5112534761428833,
+      "learning_rate": 3.595190380761523e-05,
+      "loss": 0.1417,
+      "step": 1414
+    },
+    {
+      "epoch": 5.66,
+      "grad_norm": 1.4549270868301392,
+      "learning_rate": 3.594188376753507e-05,
+      "loss": 0.1756,
+      "step": 1415
+    },
+    {
+      "epoch": 5.664,
+      "grad_norm": 1.5447784662246704,
+      "learning_rate": 3.5931863727454914e-05,
+      "loss": 0.1416,
+      "step": 1416
+    },
+    {
+      "epoch": 5.668,
+      "grad_norm": 1.4175710678100586,
+      "learning_rate": 3.592184368737475e-05,
+      "loss": 0.1437,
+      "step": 1417
+    },
+    {
+      "epoch": 5.672,
+      "grad_norm": 1.4404096603393555,
+      "learning_rate": 3.591182364729459e-05,
+      "loss": 0.1456,
+      "step": 1418
+    },
+    {
+      "epoch": 5.676,
+      "grad_norm": 1.4998631477355957,
+      "learning_rate": 3.590180360721443e-05,
+      "loss": 0.1666,
+      "step": 1419
+    },
+    {
+      "epoch": 5.68,
+      "grad_norm": 1.4109253883361816,
+      "learning_rate": 3.5891783567134265e-05,
+      "loss": 0.1846,
+      "step": 1420
+    },
+    {
+      "epoch": 5.684,
+      "grad_norm": 1.2808420658111572,
+      "learning_rate": 3.5881763527054106e-05,
+      "loss": 0.1424,
+      "step": 1421
+    },
+    {
+      "epoch": 5.688,
+      "grad_norm": 1.439148187637329,
+      "learning_rate": 3.5871743486973954e-05,
+      "loss": 0.153,
+      "step": 1422
+    },
+    {
+      "epoch": 5.692,
+      "grad_norm": 1.5616912841796875,
+      "learning_rate": 3.586172344689379e-05,
+      "loss": 0.1863,
+      "step": 1423
+    },
+    {
+      "epoch": 5.696,
+      "grad_norm": 1.353434681892395,
+      "learning_rate": 3.585170340681363e-05,
+      "loss": 0.158,
+      "step": 1424
+    },
+    {
+      "epoch": 5.7,
+      "grad_norm": 1.3511037826538086,
+      "learning_rate": 3.584168336673347e-05,
+      "loss": 0.162,
+      "step": 1425
+    },
+    {
+      "epoch": 5.704,
+      "grad_norm": 1.4532557725906372,
+      "learning_rate": 3.5831663326653306e-05,
+      "loss": 0.1671,
+      "step": 1426
+    },
+    {
+      "epoch": 5.708,
+      "grad_norm": 1.4470570087432861,
+      "learning_rate": 3.582164328657315e-05,
+      "loss": 0.1564,
+      "step": 1427
+    },
+    {
+      "epoch": 5.712,
+      "grad_norm": 1.4751670360565186,
+      "learning_rate": 3.581162324649299e-05,
+      "loss": 0.1445,
+      "step": 1428
+    },
+    {
+      "epoch": 5.716,
+      "grad_norm": 1.3708182573318481,
+      "learning_rate": 3.580160320641282e-05,
+      "loss": 0.1708,
+      "step": 1429
+    },
+    {
+      "epoch": 5.72,
+      "grad_norm": 1.4434928894042969,
+      "learning_rate": 3.5791583166332664e-05,
+      "loss": 0.1691,
+      "step": 1430
+    },
+    {
+      "epoch": 5.724,
+      "grad_norm": 1.0687934160232544,
+      "learning_rate": 3.5781563126252506e-05,
+      "loss": 0.1061,
+      "step": 1431
+    },
+    {
+      "epoch": 5.728,
+      "grad_norm": 1.4179203510284424,
+      "learning_rate": 3.577154308617235e-05,
+      "loss": 0.174,
+      "step": 1432
+    },
+    {
+      "epoch": 5.732,
+      "grad_norm": 1.318701148033142,
+      "learning_rate": 3.576152304609219e-05,
+      "loss": 0.1496,
+      "step": 1433
+    },
+    {
+      "epoch": 5.736,
+      "grad_norm": 1.4118852615356445,
+      "learning_rate": 3.575150300601203e-05,
+      "loss": 0.1978,
+      "step": 1434
+    },
+    {
+      "epoch": 5.74,
+      "grad_norm": 1.2677303552627563,
+      "learning_rate": 3.5741482965931864e-05,
+      "loss": 0.1672,
+      "step": 1435
+    },
+    {
+      "epoch": 5.744,
+      "grad_norm": 1.8141050338745117,
+      "learning_rate": 3.5731462925851705e-05,
+      "loss": 0.1801,
+      "step": 1436
+    },
+    {
+      "epoch": 5.748,
+      "grad_norm": 1.5993258953094482,
+      "learning_rate": 3.5721442885771547e-05,
+      "loss": 0.1583,
+      "step": 1437
+    },
+    {
+      "epoch": 5.752,
+      "grad_norm": 1.4515212774276733,
+      "learning_rate": 3.571142284569138e-05,
+      "loss": 0.2014,
+      "step": 1438
+    },
+    {
+      "epoch": 5.756,
+      "grad_norm": 1.3378204107284546,
+      "learning_rate": 3.570140280561122e-05,
+      "loss": 0.1366,
+      "step": 1439
+    },
+    {
+      "epoch": 5.76,
+      "grad_norm": 1.3408321142196655,
+      "learning_rate": 3.5691382765531064e-05,
+      "loss": 0.1301,
+      "step": 1440
+    },
+    {
+      "epoch": 5.764,
+      "grad_norm": 1.4962408542633057,
+      "learning_rate": 3.56813627254509e-05,
+      "loss": 0.1385,
+      "step": 1441
+    },
+    {
+      "epoch": 5.768,
+      "grad_norm": 1.589575171470642,
+      "learning_rate": 3.5671342685370746e-05,
+      "loss": 0.1777,
+      "step": 1442
+    },
+    {
+      "epoch": 5.772,
+      "grad_norm": 1.5674686431884766,
+      "learning_rate": 3.566132264529059e-05,
+      "loss": 0.166,
+      "step": 1443
+    },
+    {
+      "epoch": 5.776,
+      "grad_norm": 1.5545917749404907,
+      "learning_rate": 3.565130260521042e-05,
+      "loss": 0.1775,
+      "step": 1444
+    },
+    {
+      "epoch": 5.78,
+      "grad_norm": 1.4211684465408325,
+      "learning_rate": 3.564128256513026e-05,
+      "loss": 0.149,
+      "step": 1445
+    },
+    {
+      "epoch": 5.784,
+      "grad_norm": 1.6268999576568604,
+      "learning_rate": 3.5631262525050105e-05,
+      "loss": 0.1587,
+      "step": 1446
+    },
+    {
+      "epoch": 5.788,
+      "grad_norm": 1.6301075220108032,
+      "learning_rate": 3.562124248496994e-05,
+      "loss": 0.1824,
+      "step": 1447
+    },
+    {
+      "epoch": 5.792,
+      "grad_norm": 1.3886116743087769,
+      "learning_rate": 3.561122244488978e-05,
+      "loss": 0.1717,
+      "step": 1448
+    },
+    {
+      "epoch": 5.796,
+      "grad_norm": 1.5212382078170776,
+      "learning_rate": 3.560120240480962e-05,
+      "loss": 0.1772,
+      "step": 1449
+    },
+    {
+      "epoch": 5.8,
+      "grad_norm": 1.4963605403900146,
+      "learning_rate": 3.5591182364729456e-05,
+      "loss": 0.1917,
+      "step": 1450
+    },
+    {
+      "epoch": 5.804,
+      "grad_norm": 1.4401335716247559,
+      "learning_rate": 3.55811623246493e-05,
+      "loss": 0.1619,
+      "step": 1451
+    },
+    {
+      "epoch": 5.808,
+      "grad_norm": 1.2470486164093018,
+      "learning_rate": 3.5571142284569145e-05,
+      "loss": 0.1606,
+      "step": 1452
+    },
+    {
+      "epoch": 5.812,
+      "grad_norm": 1.4720518589019775,
+      "learning_rate": 3.556112224448898e-05,
+      "loss": 0.1735,
+      "step": 1453
+    },
+    {
+      "epoch": 5.816,
+      "grad_norm": 1.3243999481201172,
+      "learning_rate": 3.555110220440882e-05,
+      "loss": 0.172,
+      "step": 1454
+    },
+    {
+      "epoch": 5.82,
+      "grad_norm": 1.3949337005615234,
+      "learning_rate": 3.554108216432866e-05,
+      "loss": 0.1698,
+      "step": 1455
+    },
+    {
+      "epoch": 5.824,
+      "grad_norm": 1.554047703742981,
+      "learning_rate": 3.55310621242485e-05,
+      "loss": 0.1883,
+      "step": 1456
+    },
+    {
+      "epoch": 5.828,
+      "grad_norm": 1.4399428367614746,
+      "learning_rate": 3.552104208416834e-05,
+      "loss": 0.1341,
+      "step": 1457
+    },
+    {
+      "epoch": 5.832,
+      "grad_norm": 1.5571656227111816,
+      "learning_rate": 3.551102204408818e-05,
+      "loss": 0.1765,
+      "step": 1458
+    },
+    {
+      "epoch": 5.836,
+      "grad_norm": 1.5305334329605103,
+      "learning_rate": 3.5501002004008014e-05,
+      "loss": 0.1873,
+      "step": 1459
+    },
+    {
+      "epoch": 5.84,
+      "grad_norm": 1.5004547834396362,
+      "learning_rate": 3.5490981963927855e-05,
+      "loss": 0.1567,
+      "step": 1460
+    },
+    {
+      "epoch": 5.844,
+      "grad_norm": 1.336523413658142,
+      "learning_rate": 3.54809619238477e-05,
+      "loss": 0.1511,
+      "step": 1461
+    },
+    {
+      "epoch": 5.848,
+      "grad_norm": 1.7299772500991821,
+      "learning_rate": 3.547094188376754e-05,
+      "loss": 0.229,
+      "step": 1462
+    },
+    {
+      "epoch": 5.852,
+      "grad_norm": 1.3854844570159912,
+      "learning_rate": 3.546092184368738e-05,
+      "loss": 0.1588,
+      "step": 1463
+    },
+    {
+      "epoch": 5.856,
+      "grad_norm": 1.587310552597046,
+      "learning_rate": 3.5450901803607214e-05,
+      "loss": 0.1463,
+      "step": 1464
+    },
+    {
+      "epoch": 5.86,
+      "grad_norm": 1.3696925640106201,
+      "learning_rate": 3.5440881763527055e-05,
+      "loss": 0.1679,
+      "step": 1465
+    },
+    {
+      "epoch": 5.864,
+      "grad_norm": 0.932451605796814,
+      "learning_rate": 3.5430861723446896e-05,
+      "loss": 0.1001,
+      "step": 1466
+    },
+    {
+      "epoch": 5.868,
+      "grad_norm": 1.4562475681304932,
+      "learning_rate": 3.542084168336674e-05,
+      "loss": 0.1691,
+      "step": 1467
+    },
+    {
+      "epoch": 5.872,
+      "grad_norm": 1.3682665824890137,
+      "learning_rate": 3.541082164328657e-05,
+      "loss": 0.1666,
+      "step": 1468
+    },
+    {
+      "epoch": 5.876,
+      "grad_norm": 1.3185116052627563,
+      "learning_rate": 3.5400801603206413e-05,
+      "loss": 0.1698,
+      "step": 1469
+    },
+    {
+      "epoch": 5.88,
+      "grad_norm": 1.5660794973373413,
+      "learning_rate": 3.5390781563126255e-05,
+      "loss": 0.1482,
+      "step": 1470
+    },
+    {
+      "epoch": 5.884,
+      "grad_norm": 1.449788212776184,
+      "learning_rate": 3.5380761523046096e-05,
+      "loss": 0.1522,
+      "step": 1471
+    },
+    {
+      "epoch": 5.888,
+      "grad_norm": 1.434978723526001,
+      "learning_rate": 3.537074148296594e-05,
+      "loss": 0.1738,
+      "step": 1472
+    },
+    {
+      "epoch": 5.892,
+      "grad_norm": 1.4906777143478394,
+      "learning_rate": 3.536072144288577e-05,
+      "loss": 0.1801,
+      "step": 1473
+    },
+    {
+      "epoch": 5.896,
+      "grad_norm": 1.4387478828430176,
+      "learning_rate": 3.535070140280561e-05,
+      "loss": 0.1582,
+      "step": 1474
+    },
+    {
+      "epoch": 5.9,
+      "grad_norm": 1.5130301713943481,
+      "learning_rate": 3.5340681362725454e-05,
+      "loss": 0.1876,
+      "step": 1475
+    },
+    {
+      "epoch": 5.904,
+      "grad_norm": 1.44967782497406,
+      "learning_rate": 3.533066132264529e-05,
+      "loss": 0.1782,
+      "step": 1476
+    },
+    {
+      "epoch": 5.908,
+      "grad_norm": 1.5330886840820312,
+      "learning_rate": 3.532064128256513e-05,
+      "loss": 0.1584,
+      "step": 1477
+    },
+    {
+      "epoch": 5.912,
+      "grad_norm": 1.5080928802490234,
+      "learning_rate": 3.531062124248497e-05,
+      "loss": 0.1575,
+      "step": 1478
+    },
+    {
+      "epoch": 5.916,
+      "grad_norm": 1.455888271331787,
+      "learning_rate": 3.530060120240481e-05,
+      "loss": 0.1837,
+      "step": 1479
+    },
+    {
+      "epoch": 5.92,
+      "grad_norm": 1.5638760328292847,
+      "learning_rate": 3.529058116232465e-05,
+      "loss": 0.1553,
+      "step": 1480
+    },
+    {
+      "epoch": 5.924,
+      "grad_norm": 1.487268328666687,
+      "learning_rate": 3.5280561122244495e-05,
+      "loss": 0.1817,
+      "step": 1481
+    },
+    {
+      "epoch": 5.928,
+      "grad_norm": 1.4699187278747559,
+      "learning_rate": 3.527054108216433e-05,
+      "loss": 0.1571,
+      "step": 1482
+    },
+    {
+      "epoch": 5.932,
+      "grad_norm": 1.4532089233398438,
+      "learning_rate": 3.526052104208417e-05,
+      "loss": 0.1434,
+      "step": 1483
+    },
+    {
+      "epoch": 5.936,
+      "grad_norm": 1.591044306755066,
+      "learning_rate": 3.525050100200401e-05,
+      "loss": 0.1538,
+      "step": 1484
+    },
+    {
+      "epoch": 5.9399999999999995,
+      "grad_norm": 1.4258755445480347,
+      "learning_rate": 3.524048096192385e-05,
+      "loss": 0.1606,
+      "step": 1485
+    },
+    {
+      "epoch": 5.944,
+      "grad_norm": 1.4429696798324585,
+      "learning_rate": 3.523046092184369e-05,
+      "loss": 0.2038,
+      "step": 1486
+    },
+    {
+      "epoch": 5.948,
+      "grad_norm": 1.4724392890930176,
+      "learning_rate": 3.522044088176353e-05,
+      "loss": 0.1644,
+      "step": 1487
+    },
+    {
+      "epoch": 5.952,
+      "grad_norm": 1.488022804260254,
+      "learning_rate": 3.5210420841683364e-05,
+      "loss": 0.1528,
+      "step": 1488
+    },
+    {
+      "epoch": 5.9559999999999995,
+      "grad_norm": 1.5905804634094238,
+      "learning_rate": 3.5200400801603205e-05,
+      "loss": 0.2294,
+      "step": 1489
+    },
+    {
+      "epoch": 5.96,
+      "grad_norm": 1.4152902364730835,
+      "learning_rate": 3.5190380761523046e-05,
+      "loss": 0.2096,
+      "step": 1490
+    },
+    {
+      "epoch": 5.964,
+      "grad_norm": 1.3682752847671509,
+      "learning_rate": 3.518036072144289e-05,
+      "loss": 0.1684,
+      "step": 1491
+    },
+    {
+      "epoch": 5.968,
+      "grad_norm": 1.4440537691116333,
+      "learning_rate": 3.517034068136273e-05,
+      "loss": 0.1714,
+      "step": 1492
+    },
+    {
+      "epoch": 5.9719999999999995,
+      "grad_norm": 1.4367694854736328,
+      "learning_rate": 3.516032064128257e-05,
+      "loss": 0.1558,
+      "step": 1493
+    },
+    {
+      "epoch": 5.976,
+      "grad_norm": 1.351462483406067,
+      "learning_rate": 3.5150300601202405e-05,
+      "loss": 0.167,
+      "step": 1494
+    },
+    {
+      "epoch": 5.98,
+      "grad_norm": 1.5862767696380615,
+      "learning_rate": 3.5140280561122246e-05,
+      "loss": 0.2089,
+      "step": 1495
+    },
+    {
+      "epoch": 5.984,
+      "grad_norm": 1.0578093528747559,
+      "learning_rate": 3.513026052104209e-05,
+      "loss": 0.1061,
+      "step": 1496
+    },
+    {
+      "epoch": 5.9879999999999995,
+      "grad_norm": 1.7430568933486938,
+      "learning_rate": 3.512024048096192e-05,
+      "loss": 0.1775,
+      "step": 1497
+    },
+    {
+      "epoch": 5.992,
+      "grad_norm": 1.3868354558944702,
+      "learning_rate": 3.511022044088176e-05,
+      "loss": 0.1627,
+      "step": 1498
+    },
+    {
+      "epoch": 5.996,
+      "grad_norm": 1.4972561597824097,
+      "learning_rate": 3.5100200400801604e-05,
+      "loss": 0.1617,
+      "step": 1499
+    },
+    {
+      "epoch": 6.0,
+      "grad_norm": 1.5004100799560547,
+      "learning_rate": 3.509018036072144e-05,
+      "loss": 0.1657,
+      "step": 1500
+    },
+    {
+      "epoch": 6.004,
+      "grad_norm": 1.2470020055770874,
+      "learning_rate": 3.508016032064129e-05,
+      "loss": 0.1026,
+      "step": 1501
+    },
+    {
+      "epoch": 6.008,
+      "grad_norm": 1.0558257102966309,
+      "learning_rate": 3.507014028056113e-05,
+      "loss": 0.0933,
+      "step": 1502
+    },
+    {
+      "epoch": 6.012,
+      "grad_norm": 1.027756690979004,
+      "learning_rate": 3.506012024048096e-05,
+      "loss": 0.0819,
+      "step": 1503
+    },
+    {
+      "epoch": 6.016,
+      "grad_norm": 1.1106997728347778,
+      "learning_rate": 3.5050100200400804e-05,
+      "loss": 0.0955,
+      "step": 1504
+    },
+    {
+      "epoch": 6.02,
+      "grad_norm": 1.2579761743545532,
+      "learning_rate": 3.5040080160320645e-05,
+      "loss": 0.1063,
+      "step": 1505
+    },
+    {
+      "epoch": 6.024,
+      "grad_norm": 1.3087098598480225,
+      "learning_rate": 3.503006012024048e-05,
+      "loss": 0.0956,
+      "step": 1506
+    },
+    {
+      "epoch": 6.028,
+      "grad_norm": 1.2753757238388062,
+      "learning_rate": 3.502004008016032e-05,
+      "loss": 0.0847,
+      "step": 1507
+    },
+    {
+      "epoch": 6.032,
+      "grad_norm": 0.8829823732376099,
+      "learning_rate": 3.501002004008016e-05,
+      "loss": 0.0567,
+      "step": 1508
+    },
+    {
+      "epoch": 6.036,
+      "grad_norm": 1.3138492107391357,
+      "learning_rate": 3.5e-05,
+      "loss": 0.1009,
+      "step": 1509
+    },
+    {
+      "epoch": 6.04,
+      "grad_norm": 1.252717137336731,
+      "learning_rate": 3.4989979959919845e-05,
+      "loss": 0.0815,
+      "step": 1510
+    },
+    {
+      "epoch": 6.044,
+      "grad_norm": 1.5361855030059814,
+      "learning_rate": 3.4979959919839686e-05,
+      "loss": 0.0982,
+      "step": 1511
+    },
+    {
+      "epoch": 6.048,
+      "grad_norm": 1.8114335536956787,
+      "learning_rate": 3.496993987975952e-05,
+      "loss": 0.107,
+      "step": 1512
+    },
+    {
+      "epoch": 6.052,
+      "grad_norm": 1.430020809173584,
+      "learning_rate": 3.495991983967936e-05,
+      "loss": 0.0926,
+      "step": 1513
+    },
+    {
+      "epoch": 6.056,
+      "grad_norm": 1.3625584840774536,
+      "learning_rate": 3.49498997995992e-05,
+      "loss": 0.0849,
+      "step": 1514
+    },
+    {
+      "epoch": 6.06,
+      "grad_norm": 1.577380895614624,
+      "learning_rate": 3.493987975951904e-05,
+      "loss": 0.1411,
+      "step": 1515
+    },
+    {
+      "epoch": 6.064,
+      "grad_norm": 1.2233928442001343,
+      "learning_rate": 3.492985971943888e-05,
+      "loss": 0.0837,
+      "step": 1516
+    },
+    {
+      "epoch": 6.068,
+      "grad_norm": 1.4746198654174805,
+      "learning_rate": 3.491983967935872e-05,
+      "loss": 0.091,
+      "step": 1517
+    },
+    {
+      "epoch": 6.072,
+      "grad_norm": 1.5226060152053833,
+      "learning_rate": 3.4909819639278555e-05,
+      "loss": 0.1285,
+      "step": 1518
+    },
+    {
+      "epoch": 6.076,
+      "grad_norm": 1.3142298460006714,
+      "learning_rate": 3.4899799599198396e-05,
+      "loss": 0.105,
+      "step": 1519
+    },
+    {
+      "epoch": 6.08,
+      "grad_norm": 1.3603146076202393,
+      "learning_rate": 3.488977955911824e-05,
+      "loss": 0.0849,
+      "step": 1520
+    },
+    {
+      "epoch": 6.084,
+      "grad_norm": 1.2942161560058594,
+      "learning_rate": 3.487975951903808e-05,
+      "loss": 0.0896,
+      "step": 1521
+    },
+    {
+      "epoch": 6.088,
+      "grad_norm": 1.6908429861068726,
+      "learning_rate": 3.486973947895792e-05,
+      "loss": 0.1167,
+      "step": 1522
+    },
+    {
+      "epoch": 6.092,
+      "grad_norm": 1.1075533628463745,
+      "learning_rate": 3.485971943887776e-05,
+      "loss": 0.0823,
+      "step": 1523
+    },
+    {
+      "epoch": 6.096,
+      "grad_norm": 1.4377602338790894,
+      "learning_rate": 3.4849699398797596e-05,
+      "loss": 0.0825,
+      "step": 1524
+    },
+    {
+      "epoch": 6.1,
+      "grad_norm": 1.474347472190857,
+      "learning_rate": 3.483967935871744e-05,
+      "loss": 0.1007,
+      "step": 1525
+    },
+    {
+      "epoch": 6.104,
+      "grad_norm": 1.4804532527923584,
+      "learning_rate": 3.482965931863728e-05,
+      "loss": 0.0888,
+      "step": 1526
+    },
+    {
+      "epoch": 6.108,
+      "grad_norm": 1.413860559463501,
+      "learning_rate": 3.481963927855711e-05,
+      "loss": 0.0858,
+      "step": 1527
+    },
+    {
+      "epoch": 6.112,
+      "grad_norm": 1.3580888509750366,
+      "learning_rate": 3.4809619238476954e-05,
+      "loss": 0.0751,
+      "step": 1528
+    },
+    {
+      "epoch": 6.116,
+      "grad_norm": 1.015170693397522,
+      "learning_rate": 3.4799599198396795e-05,
+      "loss": 0.0622,
+      "step": 1529
+    },
+    {
+      "epoch": 6.12,
+      "grad_norm": 1.4240541458129883,
+      "learning_rate": 3.478957915831664e-05,
+      "loss": 0.0968,
+      "step": 1530
+    },
+    {
+      "epoch": 6.124,
+      "grad_norm": 1.4235905408859253,
+      "learning_rate": 3.477955911823648e-05,
+      "loss": 0.0997,
+      "step": 1531
+    },
+    {
+      "epoch": 6.128,
+      "grad_norm": 1.6935840845108032,
+      "learning_rate": 3.476953907815631e-05,
+      "loss": 0.1425,
+      "step": 1532
+    },
+    {
+      "epoch": 6.132,
+      "grad_norm": 1.4013699293136597,
+      "learning_rate": 3.4759519038076154e-05,
+      "loss": 0.0898,
+      "step": 1533
+    },
+    {
+      "epoch": 6.136,
+      "grad_norm": 1.4652191400527954,
+      "learning_rate": 3.4749498997995995e-05,
+      "loss": 0.0968,
+      "step": 1534
+    },
+    {
+      "epoch": 6.14,
+      "grad_norm": 1.4197686910629272,
+      "learning_rate": 3.4739478957915836e-05,
+      "loss": 0.1328,
+      "step": 1535
+    },
+    {
+      "epoch": 6.144,
+      "grad_norm": 1.2468719482421875,
+      "learning_rate": 3.472945891783567e-05,
+      "loss": 0.0874,
+      "step": 1536
+    },
+    {
+      "epoch": 6.148,
+      "grad_norm": 1.5248289108276367,
+      "learning_rate": 3.471943887775551e-05,
+      "loss": 0.0975,
+      "step": 1537
+    },
+    {
+      "epoch": 6.152,
+      "grad_norm": 1.3224518299102783,
+      "learning_rate": 3.4709418837675353e-05,
+      "loss": 0.0761,
+      "step": 1538
+    },
+    {
+      "epoch": 6.156,
+      "grad_norm": 1.4312621355056763,
+      "learning_rate": 3.469939879759519e-05,
+      "loss": 0.0833,
+      "step": 1539
+    },
+    {
+      "epoch": 6.16,
+      "grad_norm": 1.3523186445236206,
+      "learning_rate": 3.4689378757515036e-05,
+      "loss": 0.0912,
+      "step": 1540
+    },
+    {
+      "epoch": 6.164,
+      "grad_norm": 1.4507619142532349,
+      "learning_rate": 3.467935871743487e-05,
+      "loss": 0.0977,
+      "step": 1541
+    },
+    {
+      "epoch": 6.168,
+      "grad_norm": 1.0727993249893188,
+      "learning_rate": 3.466933867735471e-05,
+      "loss": 0.0608,
+      "step": 1542
+    },
+    {
+      "epoch": 6.172,
+      "grad_norm": 1.7440407276153564,
+      "learning_rate": 3.465931863727455e-05,
+      "loss": 0.1061,
+      "step": 1543
+    },
+    {
+      "epoch": 6.176,
+      "grad_norm": 1.238020420074463,
+      "learning_rate": 3.464929859719439e-05,
+      "loss": 0.0703,
+      "step": 1544
+    },
+    {
+      "epoch": 6.18,
+      "grad_norm": 1.3099385499954224,
+      "learning_rate": 3.463927855711423e-05,
+      "loss": 0.1048,
+      "step": 1545
+    },
+    {
+      "epoch": 6.184,
+      "grad_norm": 1.4899845123291016,
+      "learning_rate": 3.462925851703407e-05,
+      "loss": 0.1037,
+      "step": 1546
+    },
+    {
+      "epoch": 6.188,
+      "grad_norm": 1.4137481451034546,
+      "learning_rate": 3.4619238476953905e-05,
+      "loss": 0.1138,
+      "step": 1547
+    },
+    {
+      "epoch": 6.192,
+      "grad_norm": 1.402271032333374,
+      "learning_rate": 3.4609218436873746e-05,
+      "loss": 0.1069,
+      "step": 1548
+    },
+    {
+      "epoch": 6.196,
+      "grad_norm": 1.4417635202407837,
+      "learning_rate": 3.459919839679359e-05,
+      "loss": 0.1024,
+      "step": 1549
+    },
+    {
+      "epoch": 6.2,
+      "grad_norm": 1.2592614889144897,
+      "learning_rate": 3.458917835671343e-05,
+      "loss": 0.1002,
+      "step": 1550
+    },
+    {
+      "epoch": 6.204,
+      "grad_norm": 1.6195659637451172,
+      "learning_rate": 3.457915831663327e-05,
+      "loss": 0.1135,
+      "step": 1551
+    },
+    {
+      "epoch": 6.208,
+      "grad_norm": 1.3179293870925903,
+      "learning_rate": 3.456913827655311e-05,
+      "loss": 0.1026,
+      "step": 1552
+    },
+    {
+      "epoch": 6.212,
+      "grad_norm": 1.3877010345458984,
+      "learning_rate": 3.4559118236472946e-05,
+      "loss": 0.0814,
+      "step": 1553
+    },
+    {
+      "epoch": 6.216,
+      "grad_norm": 1.2459914684295654,
+      "learning_rate": 3.454909819639279e-05,
+      "loss": 0.0978,
+      "step": 1554
+    },
+    {
+      "epoch": 6.22,
+      "grad_norm": 1.368811011314392,
+      "learning_rate": 3.453907815631263e-05,
+      "loss": 0.0927,
+      "step": 1555
+    },
+    {
+      "epoch": 6.224,
+      "grad_norm": 1.629564642906189,
+      "learning_rate": 3.452905811623246e-05,
+      "loss": 0.1302,
+      "step": 1556
+    },
+    {
+      "epoch": 6.228,
+      "grad_norm": 1.299174189567566,
+      "learning_rate": 3.4519038076152304e-05,
+      "loss": 0.0974,
+      "step": 1557
+    },
+    {
+      "epoch": 6.232,
+      "grad_norm": 1.39913809299469,
+      "learning_rate": 3.4509018036072145e-05,
+      "loss": 0.1,
+      "step": 1558
+    },
+    {
+      "epoch": 6.236,
+      "grad_norm": 1.5463366508483887,
+      "learning_rate": 3.449899799599198e-05,
+      "loss": 0.1092,
+      "step": 1559
+    },
+    {
+      "epoch": 6.24,
+      "grad_norm": 1.416552186012268,
+      "learning_rate": 3.448897795591183e-05,
+      "loss": 0.1247,
+      "step": 1560
+    },
+    {
+      "epoch": 6.244,
+      "grad_norm": 1.3216407299041748,
+      "learning_rate": 3.447895791583167e-05,
+      "loss": 0.1219,
+      "step": 1561
+    },
+    {
+      "epoch": 6.248,
+      "grad_norm": 1.2443658113479614,
+      "learning_rate": 3.4468937875751504e-05,
+      "loss": 0.085,
+      "step": 1562
+    },
+    {
+      "epoch": 6.252,
+      "grad_norm": 1.3868099451065063,
+      "learning_rate": 3.4458917835671345e-05,
+      "loss": 0.0941,
+      "step": 1563
+    },
+    {
+      "epoch": 6.256,
+      "grad_norm": 1.5421968698501587,
+      "learning_rate": 3.4448897795591186e-05,
+      "loss": 0.1034,
+      "step": 1564
+    },
+    {
+      "epoch": 6.26,
+      "grad_norm": 1.3406178951263428,
+      "learning_rate": 3.443887775551102e-05,
+      "loss": 0.0902,
+      "step": 1565
+    },
+    {
+      "epoch": 6.264,
+      "grad_norm": 1.5938392877578735,
+      "learning_rate": 3.442885771543086e-05,
+      "loss": 0.1159,
+      "step": 1566
+    },
+    {
+      "epoch": 6.268,
+      "grad_norm": 1.5965449810028076,
+      "learning_rate": 3.44188376753507e-05,
+      "loss": 0.1174,
+      "step": 1567
+    },
+    {
+      "epoch": 6.272,
+      "grad_norm": 1.4595623016357422,
+      "learning_rate": 3.440881763527054e-05,
+      "loss": 0.1078,
+      "step": 1568
+    },
+    {
+      "epoch": 6.276,
+      "grad_norm": 1.6079479455947876,
+      "learning_rate": 3.4398797595190386e-05,
+      "loss": 0.1158,
+      "step": 1569
+    },
+    {
+      "epoch": 6.28,
+      "grad_norm": 1.078923225402832,
+      "learning_rate": 3.438877755511023e-05,
+      "loss": 0.0722,
+      "step": 1570
+    },
+    {
+      "epoch": 6.284,
+      "grad_norm": 1.4052515029907227,
+      "learning_rate": 3.437875751503006e-05,
+      "loss": 0.0988,
+      "step": 1571
+    },
+    {
+      "epoch": 6.288,
+      "grad_norm": 1.262578010559082,
+      "learning_rate": 3.43687374749499e-05,
+      "loss": 0.0981,
+      "step": 1572
+    },
+    {
+      "epoch": 6.292,
+      "grad_norm": 1.333280324935913,
+      "learning_rate": 3.4358717434869744e-05,
+      "loss": 0.11,
+      "step": 1573
+    },
+    {
+      "epoch": 6.296,
+      "grad_norm": 1.4792897701263428,
+      "learning_rate": 3.434869739478958e-05,
+      "loss": 0.0854,
+      "step": 1574
+    },
+    {
+      "epoch": 6.3,
+      "grad_norm": 1.2670187950134277,
+      "learning_rate": 3.433867735470942e-05,
+      "loss": 0.0929,
+      "step": 1575
+    },
+    {
+      "epoch": 6.304,
+      "grad_norm": 1.4768058061599731,
+      "learning_rate": 3.432865731462926e-05,
+      "loss": 0.0961,
+      "step": 1576
+    },
+    {
+      "epoch": 6.308,
+      "grad_norm": 1.6261510848999023,
+      "learning_rate": 3.4318637274549096e-05,
+      "loss": 0.1069,
+      "step": 1577
+    },
+    {
+      "epoch": 6.312,
+      "grad_norm": 1.4705723524093628,
+      "learning_rate": 3.430861723446894e-05,
+      "loss": 0.0841,
+      "step": 1578
+    },
+    {
+      "epoch": 6.316,
+      "grad_norm": 1.6650081872940063,
+      "learning_rate": 3.4298597194388785e-05,
+      "loss": 0.1196,
+      "step": 1579
+    },
+    {
+      "epoch": 6.32,
+      "grad_norm": 1.4160841703414917,
+      "learning_rate": 3.428857715430862e-05,
+      "loss": 0.1067,
+      "step": 1580
+    },
+    {
+      "epoch": 6.324,
+      "grad_norm": 0.9616687297821045,
+      "learning_rate": 3.427855711422846e-05,
+      "loss": 0.055,
+      "step": 1581
+    },
+    {
+      "epoch": 6.328,
+      "grad_norm": 1.5261203050613403,
+      "learning_rate": 3.42685370741483e-05,
+      "loss": 0.115,
+      "step": 1582
+    },
+    {
+      "epoch": 6.332,
+      "grad_norm": 1.568793773651123,
+      "learning_rate": 3.425851703406814e-05,
+      "loss": 0.1151,
+      "step": 1583
+    },
+    {
+      "epoch": 6.336,
+      "grad_norm": 1.351989507675171,
+      "learning_rate": 3.424849699398798e-05,
+      "loss": 0.1155,
+      "step": 1584
+    },
+    {
+      "epoch": 6.34,
+      "grad_norm": 1.3044042587280273,
+      "learning_rate": 3.423847695390782e-05,
+      "loss": 0.0911,
+      "step": 1585
+    },
+    {
+      "epoch": 6.344,
+      "grad_norm": 1.4144433736801147,
+      "learning_rate": 3.4228456913827654e-05,
+      "loss": 0.1016,
+      "step": 1586
+    },
+    {
+      "epoch": 6.348,
+      "grad_norm": 2.237967014312744,
+      "learning_rate": 3.4218436873747495e-05,
+      "loss": 0.0831,
+      "step": 1587
+    },
+    {
+      "epoch": 6.352,
+      "grad_norm": 1.522283911705017,
+      "learning_rate": 3.4208416833667336e-05,
+      "loss": 0.1045,
+      "step": 1588
+    },
+    {
+      "epoch": 6.356,
+      "grad_norm": 1.3800389766693115,
+      "learning_rate": 3.419839679358718e-05,
+      "loss": 0.1121,
+      "step": 1589
+    },
+    {
+      "epoch": 6.36,
+      "grad_norm": 0.9566818475723267,
+      "learning_rate": 3.418837675350702e-05,
+      "loss": 0.0592,
+      "step": 1590
+    },
+    {
+      "epoch": 6.364,
+      "grad_norm": 1.6034566164016724,
+      "learning_rate": 3.417835671342685e-05,
+      "loss": 0.097,
+      "step": 1591
+    },
+    {
+      "epoch": 6.368,
+      "grad_norm": 1.4005167484283447,
+      "learning_rate": 3.4168336673346695e-05,
+      "loss": 0.0934,
+      "step": 1592
+    },
+    {
+      "epoch": 6.372,
+      "grad_norm": 1.4644639492034912,
+      "learning_rate": 3.4158316633266536e-05,
+      "loss": 0.0891,
+      "step": 1593
+    },
+    {
+      "epoch": 6.376,
+      "grad_norm": 1.265230417251587,
+      "learning_rate": 3.414829659318638e-05,
+      "loss": 0.1015,
+      "step": 1594
+    },
+    {
+      "epoch": 6.38,
+      "grad_norm": 1.375074863433838,
+      "learning_rate": 3.413827655310621e-05,
+      "loss": 0.1105,
+      "step": 1595
+    },
+    {
+      "epoch": 6.384,
+      "grad_norm": 1.3185374736785889,
+      "learning_rate": 3.412825651302605e-05,
+      "loss": 0.0965,
+      "step": 1596
+    },
+    {
+      "epoch": 6.388,
+      "grad_norm": 1.3281744718551636,
+      "learning_rate": 3.4118236472945894e-05,
+      "loss": 0.0881,
+      "step": 1597
+    },
+    {
+      "epoch": 6.392,
+      "grad_norm": 1.5773831605911255,
+      "learning_rate": 3.410821643286573e-05,
+      "loss": 0.1217,
+      "step": 1598
+    },
+    {
+      "epoch": 6.396,
+      "grad_norm": 1.4592585563659668,
+      "learning_rate": 3.409819639278558e-05,
+      "loss": 0.0941,
+      "step": 1599
+    },
+    {
+      "epoch": 6.4,
+      "grad_norm": 1.4319229125976562,
+      "learning_rate": 3.408817635270541e-05,
+      "loss": 0.1064,
+      "step": 1600
+    },
+    {
+      "epoch": 6.404,
+      "grad_norm": 1.5549246072769165,
+      "learning_rate": 3.407815631262525e-05,
+      "loss": 0.1062,
+      "step": 1601
+    },
+    {
+      "epoch": 6.408,
+      "grad_norm": 1.5339570045471191,
+      "learning_rate": 3.4068136272545094e-05,
+      "loss": 0.1068,
+      "step": 1602
+    },
+    {
+      "epoch": 6.412,
+      "grad_norm": 1.4197008609771729,
+      "learning_rate": 3.405811623246493e-05,
+      "loss": 0.093,
+      "step": 1603
+    },
+    {
+      "epoch": 6.416,
+      "grad_norm": 1.3577507734298706,
+      "learning_rate": 3.404809619238477e-05,
+      "loss": 0.1006,
+      "step": 1604
+    },
+    {
+      "epoch": 6.42,
+      "grad_norm": 0.8569574356079102,
+      "learning_rate": 3.403807615230461e-05,
+      "loss": 0.0439,
+      "step": 1605
+    },
+    {
+      "epoch": 6.424,
+      "grad_norm": 1.6528804302215576,
+      "learning_rate": 3.402805611222445e-05,
+      "loss": 0.1212,
+      "step": 1606
+    },
+    {
+      "epoch": 6.428,
+      "grad_norm": 1.4956883192062378,
+      "learning_rate": 3.401803607214429e-05,
+      "loss": 0.0922,
+      "step": 1607
+    },
+    {
+      "epoch": 6.432,
+      "grad_norm": 1.6123335361480713,
+      "learning_rate": 3.400801603206413e-05,
+      "loss": 0.1055,
+      "step": 1608
+    },
+    {
+      "epoch": 6.436,
+      "grad_norm": 1.539689302444458,
+      "learning_rate": 3.399799599198397e-05,
+      "loss": 0.0866,
+      "step": 1609
+    },
+    {
+      "epoch": 6.44,
+      "grad_norm": 1.570948600769043,
+      "learning_rate": 3.398797595190381e-05,
+      "loss": 0.1343,
+      "step": 1610
+    },
+    {
+      "epoch": 6.444,
+      "grad_norm": 1.6961643695831299,
+      "learning_rate": 3.397795591182365e-05,
+      "loss": 0.1115,
+      "step": 1611
+    },
+    {
+      "epoch": 6.448,
+      "grad_norm": 1.4908407926559448,
+      "learning_rate": 3.3967935871743486e-05,
+      "loss": 0.0983,
+      "step": 1612
+    },
+    {
+      "epoch": 6.452,
+      "grad_norm": 1.48050057888031,
+      "learning_rate": 3.395791583166333e-05,
+      "loss": 0.0982,
+      "step": 1613
+    },
+    {
+      "epoch": 6.456,
+      "grad_norm": 1.5200427770614624,
+      "learning_rate": 3.394789579158317e-05,
+      "loss": 0.095,
+      "step": 1614
+    },
+    {
+      "epoch": 6.46,
+      "grad_norm": 1.004806399345398,
+      "learning_rate": 3.3937875751503003e-05,
+      "loss": 0.0564,
+      "step": 1615
+    },
+    {
+      "epoch": 6.464,
+      "grad_norm": 1.4128756523132324,
+      "learning_rate": 3.3927855711422845e-05,
+      "loss": 0.1113,
+      "step": 1616
+    },
+    {
+      "epoch": 6.468,
+      "grad_norm": 1.1182926893234253,
+      "learning_rate": 3.3917835671342686e-05,
+      "loss": 0.0857,
+      "step": 1617
+    },
+    {
+      "epoch": 6.4719999999999995,
+      "grad_norm": 1.3846279382705688,
+      "learning_rate": 3.390781563126252e-05,
+      "loss": 0.0842,
+      "step": 1618
+    },
+    {
+      "epoch": 6.476,
+      "grad_norm": 1.4602874517440796,
+      "learning_rate": 3.389779559118237e-05,
+      "loss": 0.0941,
+      "step": 1619
+    },
+    {
+      "epoch": 6.48,
+      "grad_norm": 1.6256506443023682,
+      "learning_rate": 3.388777555110221e-05,
+      "loss": 0.1055,
+      "step": 1620
+    },
+    {
+      "epoch": 6.484,
+      "grad_norm": 0.9726806282997131,
+      "learning_rate": 3.3877755511022044e-05,
+      "loss": 0.0599,
+      "step": 1621
+    },
+    {
+      "epoch": 6.4879999999999995,
+      "grad_norm": 2.609361410140991,
+      "learning_rate": 3.3867735470941886e-05,
+      "loss": 0.1239,
+      "step": 1622
+    },
+    {
+      "epoch": 6.492,
+      "grad_norm": 1.5318516492843628,
+      "learning_rate": 3.385771543086173e-05,
+      "loss": 0.1655,
+      "step": 1623
+    },
+    {
+      "epoch": 6.496,
+      "grad_norm": 1.4403146505355835,
+      "learning_rate": 3.384769539078156e-05,
+      "loss": 0.1011,
+      "step": 1624
+    },
+    {
+      "epoch": 6.5,
+      "grad_norm": 1.5402027368545532,
+      "learning_rate": 3.38376753507014e-05,
+      "loss": 0.1085,
+      "step": 1625
+    },
+    {
+      "epoch": 6.504,
+      "grad_norm": 1.6094437837600708,
+      "learning_rate": 3.3827655310621244e-05,
+      "loss": 0.1067,
+      "step": 1626
+    },
+    {
+      "epoch": 6.508,
+      "grad_norm": 1.4926447868347168,
+      "learning_rate": 3.381763527054108e-05,
+      "loss": 0.0976,
+      "step": 1627
+    },
+    {
+      "epoch": 6.5120000000000005,
+      "grad_norm": 1.4129664897918701,
+      "learning_rate": 3.3807615230460927e-05,
+      "loss": 0.0947,
+      "step": 1628
+    },
+    {
+      "epoch": 6.516,
+      "grad_norm": 1.3668807744979858,
+      "learning_rate": 3.379759519038077e-05,
+      "loss": 0.1382,
+      "step": 1629
+    },
+    {
+      "epoch": 6.52,
+      "grad_norm": 1.3848040103912354,
+      "learning_rate": 3.37875751503006e-05,
+      "loss": 0.1103,
+      "step": 1630
+    },
+    {
+      "epoch": 6.524,
+      "grad_norm": 1.2794206142425537,
+      "learning_rate": 3.3777555110220444e-05,
+      "loss": 0.0964,
+      "step": 1631
+    },
+    {
+      "epoch": 6.5280000000000005,
+      "grad_norm": 1.6137124300003052,
+      "learning_rate": 3.3767535070140285e-05,
+      "loss": 0.1081,
+      "step": 1632
+    },
+    {
+      "epoch": 6.532,
+      "grad_norm": 1.6709508895874023,
+      "learning_rate": 3.375751503006012e-05,
+      "loss": 0.1414,
+      "step": 1633
+    },
+    {
+      "epoch": 6.536,
+      "grad_norm": 1.5419001579284668,
+      "learning_rate": 3.374749498997996e-05,
+      "loss": 0.1198,
+      "step": 1634
+    },
+    {
+      "epoch": 6.54,
+      "grad_norm": 1.83645761013031,
+      "learning_rate": 3.37374749498998e-05,
+      "loss": 0.1293,
+      "step": 1635
+    },
+    {
+      "epoch": 6.5440000000000005,
+      "grad_norm": 1.4904134273529053,
+      "learning_rate": 3.3727454909819637e-05,
+      "loss": 0.1015,
+      "step": 1636
+    },
+    {
+      "epoch": 6.548,
+      "grad_norm": 1.4517120122909546,
+      "learning_rate": 3.371743486973948e-05,
+      "loss": 0.0854,
+      "step": 1637
+    },
+    {
+      "epoch": 6.552,
+      "grad_norm": 1.6505271196365356,
+      "learning_rate": 3.3707414829659326e-05,
+      "loss": 0.1111,
+      "step": 1638
+    },
+    {
+      "epoch": 6.556,
+      "grad_norm": 2.092888116836548,
+      "learning_rate": 3.369739478957916e-05,
+      "loss": 0.0974,
+      "step": 1639
+    },
+    {
+      "epoch": 6.5600000000000005,
+      "grad_norm": 1.2148265838623047,
+      "learning_rate": 3.3687374749499e-05,
+      "loss": 0.1113,
+      "step": 1640
+    },
+    {
+      "epoch": 6.564,
+      "grad_norm": 1.4497480392456055,
+      "learning_rate": 3.367735470941884e-05,
+      "loss": 0.1082,
+      "step": 1641
+    },
+    {
+      "epoch": 6.568,
+      "grad_norm": 1.4580659866333008,
+      "learning_rate": 3.366733466933868e-05,
+      "loss": 0.1145,
+      "step": 1642
+    },
+    {
+      "epoch": 6.572,
+      "grad_norm": 1.3077856302261353,
+      "learning_rate": 3.365731462925852e-05,
+      "loss": 0.1039,
+      "step": 1643
+    },
+    {
+      "epoch": 6.576,
+      "grad_norm": 1.3982523679733276,
+      "learning_rate": 3.364729458917836e-05,
+      "loss": 0.1041,
+      "step": 1644
+    },
+    {
+      "epoch": 6.58,
+      "grad_norm": 1.5847138166427612,
+      "learning_rate": 3.3637274549098195e-05,
+      "loss": 0.1368,
+      "step": 1645
+    },
+    {
+      "epoch": 6.584,
+      "grad_norm": 1.542724847793579,
+      "learning_rate": 3.3627254509018036e-05,
+      "loss": 0.1215,
+      "step": 1646
+    },
+    {
+      "epoch": 6.588,
+      "grad_norm": 1.2111552953720093,
+      "learning_rate": 3.361723446893788e-05,
+      "loss": 0.0983,
+      "step": 1647
+    },
+    {
+      "epoch": 6.592,
+      "grad_norm": 1.5661391019821167,
+      "learning_rate": 3.360721442885772e-05,
+      "loss": 0.1315,
+      "step": 1648
+    },
+    {
+      "epoch": 6.596,
+      "grad_norm": 1.6418293714523315,
+      "learning_rate": 3.359719438877756e-05,
+      "loss": 0.1096,
+      "step": 1649
+    },
+    {
+      "epoch": 6.6,
+      "grad_norm": 1.4757168292999268,
+      "learning_rate": 3.35871743486974e-05,
+      "loss": 0.1102,
+      "step": 1650
+    },
+    {
+      "epoch": 6.604,
+      "grad_norm": 1.40822434425354,
+      "learning_rate": 3.3577154308617235e-05,
+      "loss": 0.0912,
+      "step": 1651
+    },
+    {
+      "epoch": 6.608,
+      "grad_norm": 1.426364541053772,
+      "learning_rate": 3.356713426853708e-05,
+      "loss": 0.0913,
+      "step": 1652
+    },
+    {
+      "epoch": 6.612,
+      "grad_norm": 1.2223780155181885,
+      "learning_rate": 3.355711422845692e-05,
+      "loss": 0.0945,
+      "step": 1653
+    },
+    {
+      "epoch": 6.616,
+      "grad_norm": 1.4718832969665527,
+      "learning_rate": 3.354709418837675e-05,
+      "loss": 0.1092,
+      "step": 1654
+    },
+    {
+      "epoch": 6.62,
+      "grad_norm": 1.6908650398254395,
+      "learning_rate": 3.3537074148296594e-05,
+      "loss": 0.1143,
+      "step": 1655
+    },
+    {
+      "epoch": 6.624,
+      "grad_norm": 1.5453354120254517,
+      "learning_rate": 3.3527054108216435e-05,
+      "loss": 0.101,
+      "step": 1656
+    },
+    {
+      "epoch": 6.628,
+      "grad_norm": 1.1836323738098145,
+      "learning_rate": 3.351703406813627e-05,
+      "loss": 0.0705,
+      "step": 1657
+    },
+    {
+      "epoch": 6.632,
+      "grad_norm": 1.5613477230072021,
+      "learning_rate": 3.350701402805612e-05,
+      "loss": 0.109,
+      "step": 1658
+    },
+    {
+      "epoch": 6.636,
+      "grad_norm": 1.5052834749221802,
+      "learning_rate": 3.349699398797595e-05,
+      "loss": 0.1057,
+      "step": 1659
+    },
+    {
+      "epoch": 6.64,
+      "grad_norm": 1.1341853141784668,
+      "learning_rate": 3.3486973947895793e-05,
+      "loss": 0.0539,
+      "step": 1660
+    },
+    {
+      "epoch": 6.644,
+      "grad_norm": 1.6815075874328613,
+      "learning_rate": 3.3476953907815635e-05,
+      "loss": 0.1237,
+      "step": 1661
+    },
+    {
+      "epoch": 6.648,
+      "grad_norm": 1.5323514938354492,
+      "learning_rate": 3.3466933867735476e-05,
+      "loss": 0.0888,
+      "step": 1662
+    },
+    {
+      "epoch": 6.652,
+      "grad_norm": 1.7266521453857422,
+      "learning_rate": 3.345691382765531e-05,
+      "loss": 0.1308,
+      "step": 1663
+    },
+    {
+      "epoch": 6.656,
+      "grad_norm": 1.6895675659179688,
+      "learning_rate": 3.344689378757515e-05,
+      "loss": 0.1054,
+      "step": 1664
+    },
+    {
+      "epoch": 6.66,
+      "grad_norm": 1.456880807876587,
+      "learning_rate": 3.343687374749499e-05,
+      "loss": 0.1082,
+      "step": 1665
+    },
+    {
+      "epoch": 6.664,
+      "grad_norm": 1.5108088254928589,
+      "learning_rate": 3.342685370741483e-05,
+      "loss": 0.1111,
+      "step": 1666
+    },
+    {
+      "epoch": 6.668,
+      "grad_norm": 1.3145025968551636,
+      "learning_rate": 3.341683366733467e-05,
+      "loss": 0.096,
+      "step": 1667
+    },
+    {
+      "epoch": 6.672,
+      "grad_norm": 1.4500359296798706,
+      "learning_rate": 3.340681362725451e-05,
+      "loss": 0.0917,
+      "step": 1668
+    },
+    {
+      "epoch": 6.676,
+      "grad_norm": 1.5362236499786377,
+      "learning_rate": 3.339679358717435e-05,
+      "loss": 0.1375,
+      "step": 1669
+    },
+    {
+      "epoch": 6.68,
+      "grad_norm": 1.490013599395752,
+      "learning_rate": 3.338677354709419e-05,
+      "loss": 0.1124,
+      "step": 1670
+    },
+    {
+      "epoch": 6.684,
+      "grad_norm": 1.6905912160873413,
+      "learning_rate": 3.337675350701403e-05,
+      "loss": 0.1238,
+      "step": 1671
+    },
+    {
+      "epoch": 6.688,
+      "grad_norm": 1.4641087055206299,
+      "learning_rate": 3.336673346693387e-05,
+      "loss": 0.1398,
+      "step": 1672
+    },
+    {
+      "epoch": 6.692,
+      "grad_norm": 1.4538987874984741,
+      "learning_rate": 3.335671342685371e-05,
+      "loss": 0.1147,
+      "step": 1673
+    },
+    {
+      "epoch": 6.696,
+      "grad_norm": 1.6322211027145386,
+      "learning_rate": 3.3346693386773544e-05,
+      "loss": 0.1176,
+      "step": 1674
+    },
+    {
+      "epoch": 6.7,
+      "grad_norm": 1.3581397533416748,
+      "learning_rate": 3.3336673346693386e-05,
+      "loss": 0.0936,
+      "step": 1675
+    },
+    {
+      "epoch": 6.704,
+      "grad_norm": 1.356339693069458,
+      "learning_rate": 3.332665330661323e-05,
+      "loss": 0.0987,
+      "step": 1676
+    },
+    {
+      "epoch": 6.708,
+      "grad_norm": 1.5190908908843994,
+      "learning_rate": 3.331663326653307e-05,
+      "loss": 0.1148,
+      "step": 1677
+    },
+    {
+      "epoch": 6.712,
+      "grad_norm": 1.6022852659225464,
+      "learning_rate": 3.330661322645291e-05,
+      "loss": 0.1146,
+      "step": 1678
+    },
+    {
+      "epoch": 6.716,
+      "grad_norm": 1.7773820161819458,
+      "learning_rate": 3.329659318637275e-05,
+      "loss": 0.1239,
+      "step": 1679
+    },
+    {
+      "epoch": 6.72,
+      "grad_norm": 1.3916137218475342,
+      "learning_rate": 3.3286573146292585e-05,
+      "loss": 0.0931,
+      "step": 1680
+    },
+    {
+      "epoch": 6.724,
+      "grad_norm": 1.4522221088409424,
+      "learning_rate": 3.3276553106212426e-05,
+      "loss": 0.1173,
+      "step": 1681
+    },
+    {
+      "epoch": 6.728,
+      "grad_norm": 1.3560315370559692,
+      "learning_rate": 3.326653306613227e-05,
+      "loss": 0.1113,
+      "step": 1682
+    },
+    {
+      "epoch": 6.732,
+      "grad_norm": 1.4816689491271973,
+      "learning_rate": 3.32565130260521e-05,
+      "loss": 0.1027,
+      "step": 1683
+    },
+    {
+      "epoch": 6.736,
+      "grad_norm": 1.453627586364746,
+      "learning_rate": 3.3246492985971944e-05,
+      "loss": 0.1071,
+      "step": 1684
+    },
+    {
+      "epoch": 6.74,
+      "grad_norm": 1.4733647108078003,
+      "learning_rate": 3.3236472945891785e-05,
+      "loss": 0.1208,
+      "step": 1685
+    },
+    {
+      "epoch": 6.744,
+      "grad_norm": 1.3408794403076172,
+      "learning_rate": 3.322645290581162e-05,
+      "loss": 0.1031,
+      "step": 1686
+    },
+    {
+      "epoch": 6.748,
+      "grad_norm": 1.5712541341781616,
+      "learning_rate": 3.321643286573147e-05,
+      "loss": 0.1169,
+      "step": 1687
+    },
+    {
+      "epoch": 6.752,
+      "grad_norm": 1.2861677408218384,
+      "learning_rate": 3.320641282565131e-05,
+      "loss": 0.0946,
+      "step": 1688
+    },
+    {
+      "epoch": 6.756,
+      "grad_norm": 1.5849751234054565,
+      "learning_rate": 3.319639278557114e-05,
+      "loss": 0.0961,
+      "step": 1689
+    },
+    {
+      "epoch": 6.76,
+      "grad_norm": 1.468711018562317,
+      "learning_rate": 3.3186372745490984e-05,
+      "loss": 0.1496,
+      "step": 1690
+    },
+    {
+      "epoch": 6.764,
+      "grad_norm": 1.6362906694412231,
+      "learning_rate": 3.3176352705410826e-05,
+      "loss": 0.1169,
+      "step": 1691
+    },
+    {
+      "epoch": 6.768,
+      "grad_norm": 1.5838176012039185,
+      "learning_rate": 3.316633266533066e-05,
+      "loss": 0.1395,
+      "step": 1692
+    },
+    {
+      "epoch": 6.772,
+      "grad_norm": 1.5610008239746094,
+      "learning_rate": 3.31563126252505e-05,
+      "loss": 0.0939,
+      "step": 1693
+    },
+    {
+      "epoch": 6.776,
+      "grad_norm": 1.5154725313186646,
+      "learning_rate": 3.314629258517034e-05,
+      "loss": 0.1003,
+      "step": 1694
+    },
+    {
+      "epoch": 6.78,
+      "grad_norm": 1.5459731817245483,
+      "learning_rate": 3.313627254509018e-05,
+      "loss": 0.1314,
+      "step": 1695
+    },
+    {
+      "epoch": 6.784,
+      "grad_norm": 1.492156744003296,
+      "learning_rate": 3.312625250501002e-05,
+      "loss": 0.1215,
+      "step": 1696
+    },
+    {
+      "epoch": 6.788,
+      "grad_norm": 1.5393439531326294,
+      "learning_rate": 3.311623246492987e-05,
+      "loss": 0.1102,
+      "step": 1697
+    },
+    {
+      "epoch": 6.792,
+      "grad_norm": 1.3706822395324707,
+      "learning_rate": 3.31062124248497e-05,
+      "loss": 0.1246,
+      "step": 1698
+    },
+    {
+      "epoch": 6.796,
+      "grad_norm": 1.3498224020004272,
+      "learning_rate": 3.309619238476954e-05,
+      "loss": 0.1176,
+      "step": 1699
+    },
+    {
+      "epoch": 6.8,
+      "grad_norm": 1.5214121341705322,
+      "learning_rate": 3.3086172344689384e-05,
+      "loss": 0.1092,
+      "step": 1700
+    },
+    {
+      "epoch": 6.804,
+      "grad_norm": 1.5812162160873413,
+      "learning_rate": 3.307615230460922e-05,
+      "loss": 0.1313,
+      "step": 1701
+    },
+    {
+      "epoch": 6.808,
+      "grad_norm": 1.4328696727752686,
+      "learning_rate": 3.306613226452906e-05,
+      "loss": 0.1188,
+      "step": 1702
+    },
+    {
+      "epoch": 6.812,
+      "grad_norm": 1.542195200920105,
+      "learning_rate": 3.30561122244489e-05,
+      "loss": 0.1328,
+      "step": 1703
+    },
+    {
+      "epoch": 6.816,
+      "grad_norm": 1.2997112274169922,
+      "learning_rate": 3.3046092184368735e-05,
+      "loss": 0.1103,
+      "step": 1704
+    },
+    {
+      "epoch": 6.82,
+      "grad_norm": 1.746984839439392,
+      "learning_rate": 3.303607214428858e-05,
+      "loss": 0.1134,
+      "step": 1705
+    },
+    {
+      "epoch": 6.824,
+      "grad_norm": 1.5972540378570557,
+      "learning_rate": 3.302605210420842e-05,
+      "loss": 0.1525,
+      "step": 1706
+    },
+    {
+      "epoch": 6.828,
+      "grad_norm": 1.4174765348434448,
+      "learning_rate": 3.301603206412826e-05,
+      "loss": 0.0846,
+      "step": 1707
+    },
+    {
+      "epoch": 6.832,
+      "grad_norm": 1.3693246841430664,
+      "learning_rate": 3.30060120240481e-05,
+      "loss": 0.1022,
+      "step": 1708
+    },
+    {
+      "epoch": 6.836,
+      "grad_norm": 1.5433459281921387,
+      "learning_rate": 3.299599198396794e-05,
+      "loss": 0.1151,
+      "step": 1709
+    },
+    {
+      "epoch": 6.84,
+      "grad_norm": 1.515580415725708,
+      "learning_rate": 3.2985971943887776e-05,
+      "loss": 0.1157,
+      "step": 1710
+    },
+    {
+      "epoch": 6.844,
+      "grad_norm": 1.4604809284210205,
+      "learning_rate": 3.297595190380762e-05,
+      "loss": 0.1203,
+      "step": 1711
+    },
+    {
+      "epoch": 6.848,
+      "grad_norm": 1.8675997257232666,
+      "learning_rate": 3.296593186372746e-05,
+      "loss": 0.1357,
+      "step": 1712
+    },
+    {
+      "epoch": 6.852,
+      "grad_norm": 1.4611036777496338,
+      "learning_rate": 3.295591182364729e-05,
+      "loss": 0.1203,
+      "step": 1713
+    },
+    {
+      "epoch": 6.856,
+      "grad_norm": 1.5014901161193848,
+      "learning_rate": 3.2945891783567135e-05,
+      "loss": 0.1151,
+      "step": 1714
+    },
+    {
+      "epoch": 6.86,
+      "grad_norm": 1.5301997661590576,
+      "learning_rate": 3.2935871743486976e-05,
+      "loss": 0.1121,
+      "step": 1715
+    },
+    {
+      "epoch": 6.864,
+      "grad_norm": 1.5152416229248047,
+      "learning_rate": 3.292585170340681e-05,
+      "loss": 0.0901,
+      "step": 1716
+    },
+    {
+      "epoch": 6.868,
+      "grad_norm": 1.5249059200286865,
+      "learning_rate": 3.291583166332666e-05,
+      "loss": 0.1061,
+      "step": 1717
+    },
+    {
+      "epoch": 6.872,
+      "grad_norm": 1.729848027229309,
+      "learning_rate": 3.290581162324649e-05,
+      "loss": 0.1259,
+      "step": 1718
+    },
+    {
+      "epoch": 6.876,
+      "grad_norm": 1.510529637336731,
+      "learning_rate": 3.2895791583166334e-05,
+      "loss": 0.11,
+      "step": 1719
+    },
+    {
+      "epoch": 6.88,
+      "grad_norm": 1.3543599843978882,
+      "learning_rate": 3.2885771543086176e-05,
+      "loss": 0.1085,
+      "step": 1720
+    },
+    {
+      "epoch": 6.884,
+      "grad_norm": 1.5983351469039917,
+      "learning_rate": 3.287575150300602e-05,
+      "loss": 0.144,
+      "step": 1721
+    },
+    {
+      "epoch": 6.888,
+      "grad_norm": 1.4449596405029297,
+      "learning_rate": 3.286573146292585e-05,
+      "loss": 0.1132,
+      "step": 1722
+    },
+    {
+      "epoch": 6.892,
+      "grad_norm": 1.5098458528518677,
+      "learning_rate": 3.285571142284569e-05,
+      "loss": 0.122,
+      "step": 1723
+    },
+    {
+      "epoch": 6.896,
+      "grad_norm": 1.5662916898727417,
+      "learning_rate": 3.2845691382765534e-05,
+      "loss": 0.1112,
+      "step": 1724
+    },
+    {
+      "epoch": 6.9,
+      "grad_norm": 1.4259101152420044,
+      "learning_rate": 3.283567134268537e-05,
+      "loss": 0.0982,
+      "step": 1725
+    },
+    {
+      "epoch": 6.904,
+      "grad_norm": 1.3410980701446533,
+      "learning_rate": 3.282565130260521e-05,
+      "loss": 0.0893,
+      "step": 1726
+    },
+    {
+      "epoch": 6.908,
+      "grad_norm": 1.5725864171981812,
+      "learning_rate": 3.281563126252505e-05,
+      "loss": 0.126,
+      "step": 1727
+    },
+    {
+      "epoch": 6.912,
+      "grad_norm": 1.5315643548965454,
+      "learning_rate": 3.280561122244489e-05,
+      "loss": 0.1013,
+      "step": 1728
+    },
+    {
+      "epoch": 6.916,
+      "grad_norm": 1.2779252529144287,
+      "learning_rate": 3.2795591182364734e-05,
+      "loss": 0.0934,
+      "step": 1729
+    },
+    {
+      "epoch": 6.92,
+      "grad_norm": 1.4510380029678345,
+      "learning_rate": 3.278557114228457e-05,
+      "loss": 0.0971,
+      "step": 1730
+    },
+    {
+      "epoch": 6.924,
+      "grad_norm": 1.527717113494873,
+      "learning_rate": 3.277555110220441e-05,
+      "loss": 0.1138,
+      "step": 1731
+    },
+    {
+      "epoch": 6.928,
+      "grad_norm": 1.5672757625579834,
+      "learning_rate": 3.276553106212425e-05,
+      "loss": 0.1001,
+      "step": 1732
+    },
+    {
+      "epoch": 6.932,
+      "grad_norm": 1.557324767112732,
+      "learning_rate": 3.275551102204409e-05,
+      "loss": 0.1195,
+      "step": 1733
+    },
+    {
+      "epoch": 6.936,
+      "grad_norm": 1.4140926599502563,
+      "learning_rate": 3.2745490981963926e-05,
+      "loss": 0.1113,
+      "step": 1734
+    },
+    {
+      "epoch": 6.9399999999999995,
+      "grad_norm": 1.4385108947753906,
+      "learning_rate": 3.273547094188377e-05,
+      "loss": 0.1139,
+      "step": 1735
+    },
+    {
+      "epoch": 6.944,
+      "grad_norm": 1.5246384143829346,
+      "learning_rate": 3.272545090180361e-05,
+      "loss": 0.1125,
+      "step": 1736
+    },
+    {
+      "epoch": 6.948,
+      "grad_norm": 1.0488942861557007,
+      "learning_rate": 3.271543086172345e-05,
+      "loss": 0.0691,
+      "step": 1737
+    },
+    {
+      "epoch": 6.952,
+      "grad_norm": 1.495911717414856,
+      "learning_rate": 3.270541082164329e-05,
+      "loss": 0.0927,
+      "step": 1738
+    },
+    {
+      "epoch": 6.9559999999999995,
+      "grad_norm": 1.6264322996139526,
+      "learning_rate": 3.2695390781563126e-05,
+      "loss": 0.1127,
+      "step": 1739
+    },
+    {
+      "epoch": 6.96,
+      "grad_norm": 1.3778818845748901,
+      "learning_rate": 3.268537074148297e-05,
+      "loss": 0.1266,
+      "step": 1740
+    },
+    {
+      "epoch": 6.964,
+      "grad_norm": 1.5300289392471313,
+      "learning_rate": 3.267535070140281e-05,
+      "loss": 0.1133,
+      "step": 1741
+    },
+    {
+      "epoch": 6.968,
+      "grad_norm": 1.5947483777999878,
+      "learning_rate": 3.266533066132264e-05,
+      "loss": 0.1196,
+      "step": 1742
+    },
+    {
+      "epoch": 6.9719999999999995,
+      "grad_norm": 1.4825366735458374,
+      "learning_rate": 3.2655310621242484e-05,
+      "loss": 0.1068,
+      "step": 1743
+    },
+    {
+      "epoch": 6.976,
+      "grad_norm": 1.7026481628417969,
+      "learning_rate": 3.2645290581162326e-05,
+      "loss": 0.0977,
+      "step": 1744
+    },
+    {
+      "epoch": 6.98,
+      "grad_norm": 1.6448338031768799,
+      "learning_rate": 3.263527054108216e-05,
+      "loss": 0.1112,
+      "step": 1745
+    },
+    {
+      "epoch": 6.984,
+      "grad_norm": 1.6659256219863892,
+      "learning_rate": 3.262525050100201e-05,
+      "loss": 0.1179,
+      "step": 1746
+    },
+    {
+      "epoch": 6.9879999999999995,
+      "grad_norm": 1.652055025100708,
+      "learning_rate": 3.261523046092185e-05,
+      "loss": 0.1535,
+      "step": 1747
+    },
+    {
+      "epoch": 6.992,
+      "grad_norm": 1.4880528450012207,
+      "learning_rate": 3.2605210420841684e-05,
+      "loss": 0.067,
+      "step": 1748
+    },
+    {
+      "epoch": 6.996,
+      "grad_norm": 1.5849748849868774,
+      "learning_rate": 3.2595190380761525e-05,
+      "loss": 0.1162,
+      "step": 1749
+    },
+    {
+      "epoch": 7.0,
+      "grad_norm": 1.6205350160598755,
+      "learning_rate": 3.2585170340681367e-05,
+      "loss": 0.127,
+      "step": 1750
+    },
+    {
+      "epoch": 7.004,
+      "grad_norm": 1.1027536392211914,
+      "learning_rate": 3.25751503006012e-05,
+      "loss": 0.0632,
+      "step": 1751
+    },
+    {
+      "epoch": 7.008,
+      "grad_norm": 1.131488561630249,
+      "learning_rate": 3.256513026052104e-05,
+      "loss": 0.0832,
+      "step": 1752
+    },
+    {
+      "epoch": 7.012,
+      "grad_norm": 1.1157985925674438,
+      "learning_rate": 3.2555110220440884e-05,
+      "loss": 0.0597,
+      "step": 1753
+    },
+    {
+      "epoch": 7.016,
+      "grad_norm": 1.2950737476348877,
+      "learning_rate": 3.254509018036072e-05,
+      "loss": 0.07,
+      "step": 1754
+    },
+    {
+      "epoch": 7.02,
+      "grad_norm": 1.0984954833984375,
+      "learning_rate": 3.253507014028056e-05,
+      "loss": 0.0584,
+      "step": 1755
+    },
+    {
+      "epoch": 7.024,
+      "grad_norm": 1.1092177629470825,
+      "learning_rate": 3.252505010020041e-05,
+      "loss": 0.0605,
+      "step": 1756
+    },
+    {
+      "epoch": 7.028,
+      "grad_norm": 0.8248789310455322,
+      "learning_rate": 3.251503006012024e-05,
+      "loss": 0.029,
+      "step": 1757
+    },
+    {
+      "epoch": 7.032,
+      "grad_norm": 1.0940172672271729,
+      "learning_rate": 3.250501002004008e-05,
+      "loss": 0.05,
+      "step": 1758
+    },
+    {
+      "epoch": 7.036,
+      "grad_norm": 1.4330635070800781,
+      "learning_rate": 3.2494989979959925e-05,
+      "loss": 0.0788,
+      "step": 1759
+    },
+    {
+      "epoch": 7.04,
+      "grad_norm": 1.5703409910202026,
+      "learning_rate": 3.248496993987976e-05,
+      "loss": 0.0811,
+      "step": 1760
+    },
+    {
+      "epoch": 7.044,
+      "grad_norm": 1.1306759119033813,
+      "learning_rate": 3.24749498997996e-05,
+      "loss": 0.0555,
+      "step": 1761
+    },
+    {
+      "epoch": 7.048,
+      "grad_norm": 1.2541637420654297,
+      "learning_rate": 3.246492985971944e-05,
+      "loss": 0.0572,
+      "step": 1762
+    },
+    {
+      "epoch": 7.052,
+      "grad_norm": 1.333361268043518,
+      "learning_rate": 3.2454909819639276e-05,
+      "loss": 0.0642,
+      "step": 1763
+    },
+    {
+      "epoch": 7.056,
+      "grad_norm": 1.1626932621002197,
+      "learning_rate": 3.244488977955912e-05,
+      "loss": 0.0489,
+      "step": 1764
+    },
+    {
+      "epoch": 7.06,
+      "grad_norm": 1.2295629978179932,
+      "learning_rate": 3.243486973947896e-05,
+      "loss": 0.0544,
+      "step": 1765
+    },
+    {
+      "epoch": 7.064,
+      "grad_norm": 1.3224353790283203,
+      "learning_rate": 3.24248496993988e-05,
+      "loss": 0.0667,
+      "step": 1766
+    },
+    {
+      "epoch": 7.068,
+      "grad_norm": 1.2137049436569214,
+      "learning_rate": 3.241482965931864e-05,
+      "loss": 0.0581,
+      "step": 1767
+    },
+    {
+      "epoch": 7.072,
+      "grad_norm": 1.1840201616287231,
+      "learning_rate": 3.240480961923848e-05,
+      "loss": 0.0649,
+      "step": 1768
+    },
+    {
+      "epoch": 7.076,
+      "grad_norm": 1.3883408308029175,
+      "learning_rate": 3.239478957915832e-05,
+      "loss": 0.0767,
+      "step": 1769
+    },
+    {
+      "epoch": 7.08,
+      "grad_norm": 0.8142508268356323,
+      "learning_rate": 3.238476953907816e-05,
+      "loss": 0.0378,
+      "step": 1770
+    },
+    {
+      "epoch": 7.084,
+      "grad_norm": 1.2708982229232788,
+      "learning_rate": 3.2374749498998e-05,
+      "loss": 0.0578,
+      "step": 1771
+    },
+    {
+      "epoch": 7.088,
+      "grad_norm": 1.1672301292419434,
+      "learning_rate": 3.2364729458917834e-05,
+      "loss": 0.0538,
+      "step": 1772
+    },
+    {
+      "epoch": 7.092,
+      "grad_norm": 1.0839089155197144,
+      "learning_rate": 3.2354709418837675e-05,
+      "loss": 0.0452,
+      "step": 1773
+    },
+    {
+      "epoch": 7.096,
+      "grad_norm": 1.2967957258224487,
+      "learning_rate": 3.234468937875752e-05,
+      "loss": 0.0676,
+      "step": 1774
+    },
+    {
+      "epoch": 7.1,
+      "grad_norm": 1.3709466457366943,
+      "learning_rate": 3.233466933867735e-05,
+      "loss": 0.0607,
+      "step": 1775
+    },
+    {
+      "epoch": 7.104,
+      "grad_norm": 1.3387433290481567,
+      "learning_rate": 3.23246492985972e-05,
+      "loss": 0.0577,
+      "step": 1776
+    },
+    {
+      "epoch": 7.108,
+      "grad_norm": 1.308927297592163,
+      "learning_rate": 3.231462925851704e-05,
+      "loss": 0.0827,
+      "step": 1777
+    },
+    {
+      "epoch": 7.112,
+      "grad_norm": 1.114564061164856,
+      "learning_rate": 3.2304609218436875e-05,
+      "loss": 0.0473,
+      "step": 1778
+    },
+    {
+      "epoch": 7.116,
+      "grad_norm": 1.2246047258377075,
+      "learning_rate": 3.2294589178356716e-05,
+      "loss": 0.0585,
+      "step": 1779
+    },
+    {
+      "epoch": 7.12,
+      "grad_norm": 1.3446717262268066,
+      "learning_rate": 3.228456913827656e-05,
+      "loss": 0.0696,
+      "step": 1780
+    },
+    {
+      "epoch": 7.124,
+      "grad_norm": 1.2419359683990479,
+      "learning_rate": 3.227454909819639e-05,
+      "loss": 0.0715,
+      "step": 1781
+    },
+    {
+      "epoch": 7.128,
+      "grad_norm": 1.3638228178024292,
+      "learning_rate": 3.2264529058116233e-05,
+      "loss": 0.0566,
+      "step": 1782
+    },
+    {
+      "epoch": 7.132,
+      "grad_norm": 1.1495225429534912,
+      "learning_rate": 3.2254509018036075e-05,
+      "loss": 0.057,
+      "step": 1783
+    },
+    {
+      "epoch": 7.136,
+      "grad_norm": 1.3355066776275635,
+      "learning_rate": 3.224448897795591e-05,
+      "loss": 0.067,
+      "step": 1784
+    },
+    {
+      "epoch": 7.14,
+      "grad_norm": 1.2129656076431274,
+      "learning_rate": 3.223446893787575e-05,
+      "loss": 0.052,
+      "step": 1785
+    },
+    {
+      "epoch": 7.144,
+      "grad_norm": 1.1906895637512207,
+      "learning_rate": 3.222444889779559e-05,
+      "loss": 0.0627,
+      "step": 1786
+    },
+    {
+      "epoch": 7.148,
+      "grad_norm": 1.0992094278335571,
+      "learning_rate": 3.221442885771543e-05,
+      "loss": 0.0473,
+      "step": 1787
+    },
+    {
+      "epoch": 7.152,
+      "grad_norm": 1.235259771347046,
+      "learning_rate": 3.2204408817635274e-05,
+      "loss": 0.0644,
+      "step": 1788
+    },
+    {
+      "epoch": 7.156,
+      "grad_norm": 1.3004990816116333,
+      "learning_rate": 3.2194388777555116e-05,
+      "loss": 0.0675,
+      "step": 1789
+    },
+    {
+      "epoch": 7.16,
+      "grad_norm": 1.2756885290145874,
+      "learning_rate": 3.218436873747495e-05,
+      "loss": 0.067,
+      "step": 1790
+    },
+    {
+      "epoch": 7.164,
+      "grad_norm": 1.1897791624069214,
+      "learning_rate": 3.217434869739479e-05,
+      "loss": 0.0483,
+      "step": 1791
+    },
+    {
+      "epoch": 7.168,
+      "grad_norm": 1.2008836269378662,
+      "learning_rate": 3.216432865731463e-05,
+      "loss": 0.0651,
+      "step": 1792
+    },
+    {
+      "epoch": 7.172,
+      "grad_norm": 1.4409757852554321,
+      "learning_rate": 3.215430861723447e-05,
+      "loss": 0.0662,
+      "step": 1793
+    },
+    {
+      "epoch": 7.176,
+      "grad_norm": 1.4043469429016113,
+      "learning_rate": 3.214428857715431e-05,
+      "loss": 0.0937,
+      "step": 1794
+    },
+    {
+      "epoch": 7.18,
+      "grad_norm": 1.6467530727386475,
+      "learning_rate": 3.213426853707415e-05,
+      "loss": 0.0913,
+      "step": 1795
+    },
+    {
+      "epoch": 7.184,
+      "grad_norm": 1.4349619150161743,
+      "learning_rate": 3.212424849699399e-05,
+      "loss": 0.0623,
+      "step": 1796
+    },
+    {
+      "epoch": 7.188,
+      "grad_norm": 1.2570191621780396,
+      "learning_rate": 3.211422845691383e-05,
+      "loss": 0.0597,
+      "step": 1797
+    },
+    {
+      "epoch": 7.192,
+      "grad_norm": 1.2177025079727173,
+      "learning_rate": 3.210420841683367e-05,
+      "loss": 0.0568,
+      "step": 1798
+    },
+    {
+      "epoch": 7.196,
+      "grad_norm": 1.1672996282577515,
+      "learning_rate": 3.209418837675351e-05,
+      "loss": 0.0621,
+      "step": 1799
+    },
+    {
+      "epoch": 7.2,
+      "grad_norm": 1.3348926305770874,
+      "learning_rate": 3.208416833667335e-05,
+      "loss": 0.0656,
+      "step": 1800
+    },
+    {
+      "epoch": 7.204,
+      "grad_norm": 1.202222466468811,
+      "learning_rate": 3.2074148296593184e-05,
+      "loss": 0.0549,
+      "step": 1801
+    },
+    {
+      "epoch": 7.208,
+      "grad_norm": 1.333136796951294,
+      "learning_rate": 3.2064128256513025e-05,
+      "loss": 0.0645,
+      "step": 1802
+    },
+    {
+      "epoch": 7.212,
+      "grad_norm": 1.3489561080932617,
+      "learning_rate": 3.2054108216432866e-05,
+      "loss": 0.0583,
+      "step": 1803
+    },
+    {
+      "epoch": 7.216,
+      "grad_norm": 1.4054896831512451,
+      "learning_rate": 3.204408817635271e-05,
+      "loss": 0.0581,
+      "step": 1804
+    },
+    {
+      "epoch": 7.22,
+      "grad_norm": 1.1050463914871216,
+      "learning_rate": 3.203406813627255e-05,
+      "loss": 0.0552,
+      "step": 1805
+    },
+    {
+      "epoch": 7.224,
+      "grad_norm": 0.9208634495735168,
+      "learning_rate": 3.202404809619239e-05,
+      "loss": 0.0349,
+      "step": 1806
+    },
+    {
+      "epoch": 7.228,
+      "grad_norm": 1.2044702768325806,
+      "learning_rate": 3.2014028056112225e-05,
+      "loss": 0.0567,
+      "step": 1807
+    },
+    {
+      "epoch": 7.232,
+      "grad_norm": 1.2432702779769897,
+      "learning_rate": 3.2004008016032066e-05,
+      "loss": 0.0576,
+      "step": 1808
+    },
+    {
+      "epoch": 7.236,
+      "grad_norm": 1.2744909524917603,
+      "learning_rate": 3.199398797595191e-05,
+      "loss": 0.0682,
+      "step": 1809
+    },
+    {
+      "epoch": 7.24,
+      "grad_norm": 1.3941128253936768,
+      "learning_rate": 3.198396793587174e-05,
+      "loss": 0.0596,
+      "step": 1810
+    },
+    {
+      "epoch": 7.244,
+      "grad_norm": 1.7390588521957397,
+      "learning_rate": 3.197394789579158e-05,
+      "loss": 0.0646,
+      "step": 1811
+    },
+    {
+      "epoch": 7.248,
+      "grad_norm": 1.6309089660644531,
+      "learning_rate": 3.1963927855711424e-05,
+      "loss": 0.0645,
+      "step": 1812
+    },
+    {
+      "epoch": 7.252,
+      "grad_norm": 1.3789124488830566,
+      "learning_rate": 3.195390781563126e-05,
+      "loss": 0.0684,
+      "step": 1813
+    },
+    {
+      "epoch": 7.256,
+      "grad_norm": 1.2757648229599,
+      "learning_rate": 3.19438877755511e-05,
+      "loss": 0.0706,
+      "step": 1814
+    },
+    {
+      "epoch": 7.26,
+      "grad_norm": 1.0513287782669067,
+      "learning_rate": 3.193386773547095e-05,
+      "loss": 0.0525,
+      "step": 1815
+    },
+    {
+      "epoch": 7.264,
+      "grad_norm": 1.4275181293487549,
+      "learning_rate": 3.192384769539078e-05,
+      "loss": 0.0563,
+      "step": 1816
+    },
+    {
+      "epoch": 7.268,
+      "grad_norm": 1.3130202293395996,
+      "learning_rate": 3.1913827655310624e-05,
+      "loss": 0.0626,
+      "step": 1817
+    },
+    {
+      "epoch": 7.272,
+      "grad_norm": 1.3455928564071655,
+      "learning_rate": 3.1903807615230465e-05,
+      "loss": 0.0688,
+      "step": 1818
+    },
+    {
+      "epoch": 7.276,
+      "grad_norm": 1.3660778999328613,
+      "learning_rate": 3.18937875751503e-05,
+      "loss": 0.0666,
+      "step": 1819
+    },
+    {
+      "epoch": 7.28,
+      "grad_norm": 1.1925787925720215,
+      "learning_rate": 3.188376753507014e-05,
+      "loss": 0.0685,
+      "step": 1820
+    },
+    {
+      "epoch": 7.284,
+      "grad_norm": 1.4987269639968872,
+      "learning_rate": 3.187374749498998e-05,
+      "loss": 0.0747,
+      "step": 1821
+    },
+    {
+      "epoch": 7.288,
+      "grad_norm": 1.1474624872207642,
+      "learning_rate": 3.186372745490982e-05,
+      "loss": 0.0558,
+      "step": 1822
+    },
+    {
+      "epoch": 7.292,
+      "grad_norm": 1.4372210502624512,
+      "learning_rate": 3.185370741482966e-05,
+      "loss": 0.068,
+      "step": 1823
+    },
+    {
+      "epoch": 7.296,
+      "grad_norm": 1.5136364698410034,
+      "learning_rate": 3.18436873747495e-05,
+      "loss": 0.0852,
+      "step": 1824
+    },
+    {
+      "epoch": 7.3,
+      "grad_norm": 0.8756834268569946,
+      "learning_rate": 3.183366733466934e-05,
+      "loss": 0.0306,
+      "step": 1825
+    },
+    {
+      "epoch": 7.304,
+      "grad_norm": 1.6666139364242554,
+      "learning_rate": 3.182364729458918e-05,
+      "loss": 0.0818,
+      "step": 1826
+    },
+    {
+      "epoch": 7.308,
+      "grad_norm": 1.2882213592529297,
+      "learning_rate": 3.181362725450902e-05,
+      "loss": 0.0663,
+      "step": 1827
+    },
+    {
+      "epoch": 7.312,
+      "grad_norm": 1.2303308248519897,
+      "learning_rate": 3.180360721442886e-05,
+      "loss": 0.0696,
+      "step": 1828
+    },
+    {
+      "epoch": 7.316,
+      "grad_norm": 1.3530824184417725,
+      "learning_rate": 3.17935871743487e-05,
+      "loss": 0.0665,
+      "step": 1829
+    },
+    {
+      "epoch": 7.32,
+      "grad_norm": 1.375704288482666,
+      "learning_rate": 3.178356713426854e-05,
+      "loss": 0.0606,
+      "step": 1830
+    },
+    {
+      "epoch": 7.324,
+      "grad_norm": 1.5587910413742065,
+      "learning_rate": 3.1773547094188375e-05,
+      "loss": 0.0797,
+      "step": 1831
+    },
+    {
+      "epoch": 7.328,
+      "grad_norm": 1.2583500146865845,
+      "learning_rate": 3.1763527054108216e-05,
+      "loss": 0.0551,
+      "step": 1832
+    },
+    {
+      "epoch": 7.332,
+      "grad_norm": 1.2027583122253418,
+      "learning_rate": 3.175350701402806e-05,
+      "loss": 0.0578,
+      "step": 1833
+    },
+    {
+      "epoch": 7.336,
+      "grad_norm": 0.9741259813308716,
+      "learning_rate": 3.174348697394789e-05,
+      "loss": 0.0362,
+      "step": 1834
+    },
+    {
+      "epoch": 7.34,
+      "grad_norm": 1.3720448017120361,
+      "learning_rate": 3.173346693386774e-05,
+      "loss": 0.0693,
+      "step": 1835
+    },
+    {
+      "epoch": 7.344,
+      "grad_norm": 1.2734616994857788,
+      "learning_rate": 3.172344689378758e-05,
+      "loss": 0.0569,
+      "step": 1836
+    },
+    {
+      "epoch": 7.348,
+      "grad_norm": 1.4324922561645508,
+      "learning_rate": 3.1713426853707416e-05,
+      "loss": 0.0765,
+      "step": 1837
+    },
+    {
+      "epoch": 7.352,
+      "grad_norm": 1.2667415142059326,
+      "learning_rate": 3.170340681362726e-05,
+      "loss": 0.0605,
+      "step": 1838
+    },
+    {
+      "epoch": 7.356,
+      "grad_norm": 1.3741072416305542,
+      "learning_rate": 3.16933867735471e-05,
+      "loss": 0.0634,
+      "step": 1839
+    },
+    {
+      "epoch": 7.36,
+      "grad_norm": 1.2768627405166626,
+      "learning_rate": 3.168336673346693e-05,
+      "loss": 0.0575,
+      "step": 1840
+    },
+    {
+      "epoch": 7.364,
+      "grad_norm": 1.393515944480896,
+      "learning_rate": 3.1673346693386774e-05,
+      "loss": 0.0642,
+      "step": 1841
+    },
+    {
+      "epoch": 7.368,
+      "grad_norm": 1.6649253368377686,
+      "learning_rate": 3.1663326653306616e-05,
+      "loss": 0.0837,
+      "step": 1842
+    },
+    {
+      "epoch": 7.372,
+      "grad_norm": 1.5157742500305176,
+      "learning_rate": 3.165330661322645e-05,
+      "loss": 0.07,
+      "step": 1843
+    },
+    {
+      "epoch": 7.376,
+      "grad_norm": 1.4178087711334229,
+      "learning_rate": 3.16432865731463e-05,
+      "loss": 0.0717,
+      "step": 1844
+    },
+    {
+      "epoch": 7.38,
+      "grad_norm": 1.1105268001556396,
+      "learning_rate": 3.163326653306614e-05,
+      "loss": 0.0495,
+      "step": 1845
+    },
+    {
+      "epoch": 7.384,
+      "grad_norm": 1.3235563039779663,
+      "learning_rate": 3.1623246492985974e-05,
+      "loss": 0.0687,
+      "step": 1846
+    },
+    {
+      "epoch": 7.388,
+      "grad_norm": 1.3781366348266602,
+      "learning_rate": 3.1613226452905815e-05,
+      "loss": 0.0741,
+      "step": 1847
+    },
+    {
+      "epoch": 7.392,
+      "grad_norm": 1.330330491065979,
+      "learning_rate": 3.1603206412825656e-05,
+      "loss": 0.061,
+      "step": 1848
+    },
+    {
+      "epoch": 7.396,
+      "grad_norm": 1.3802903890609741,
+      "learning_rate": 3.159318637274549e-05,
+      "loss": 0.068,
+      "step": 1849
+    },
+    {
+      "epoch": 7.4,
+      "grad_norm": 0.8295315504074097,
+      "learning_rate": 3.158316633266533e-05,
+      "loss": 0.0321,
+      "step": 1850
+    },
+    {
+      "epoch": 7.404,
+      "grad_norm": 1.4891018867492676,
+      "learning_rate": 3.1573146292585173e-05,
+      "loss": 0.0585,
+      "step": 1851
+    },
+    {
+      "epoch": 7.408,
+      "grad_norm": 1.496936321258545,
+      "learning_rate": 3.156312625250501e-05,
+      "loss": 0.0689,
+      "step": 1852
+    },
+    {
+      "epoch": 7.412,
+      "grad_norm": 1.4590027332305908,
+      "learning_rate": 3.155310621242485e-05,
+      "loss": 0.0722,
+      "step": 1853
+    },
+    {
+      "epoch": 7.416,
+      "grad_norm": 1.5951083898544312,
+      "learning_rate": 3.154308617234469e-05,
+      "loss": 0.0666,
+      "step": 1854
+    },
+    {
+      "epoch": 7.42,
+      "grad_norm": 1.2343343496322632,
+      "learning_rate": 3.153306613226453e-05,
+      "loss": 0.07,
+      "step": 1855
+    },
+    {
+      "epoch": 7.424,
+      "grad_norm": 1.3191311359405518,
+      "learning_rate": 3.152304609218437e-05,
+      "loss": 0.0684,
+      "step": 1856
+    },
+    {
+      "epoch": 7.428,
+      "grad_norm": 1.3911070823669434,
+      "learning_rate": 3.151302605210421e-05,
+      "loss": 0.0812,
+      "step": 1857
+    },
+    {
+      "epoch": 7.432,
+      "grad_norm": 1.2717417478561401,
+      "learning_rate": 3.150300601202405e-05,
+      "loss": 0.0693,
+      "step": 1858
+    },
+    {
+      "epoch": 7.436,
+      "grad_norm": 1.1313704252243042,
+      "learning_rate": 3.149298597194389e-05,
+      "loss": 0.0705,
+      "step": 1859
+    },
+    {
+      "epoch": 7.44,
+      "grad_norm": 1.0588295459747314,
+      "learning_rate": 3.148296593186373e-05,
+      "loss": 0.0578,
+      "step": 1860
+    },
+    {
+      "epoch": 7.444,
+      "grad_norm": 1.2012568712234497,
+      "learning_rate": 3.1472945891783566e-05,
+      "loss": 0.0607,
+      "step": 1861
+    },
+    {
+      "epoch": 7.448,
+      "grad_norm": 2.303619861602783,
+      "learning_rate": 3.146292585170341e-05,
+      "loss": 0.0796,
+      "step": 1862
+    },
+    {
+      "epoch": 7.452,
+      "grad_norm": 1.1877150535583496,
+      "learning_rate": 3.145290581162325e-05,
+      "loss": 0.0563,
+      "step": 1863
+    },
+    {
+      "epoch": 7.456,
+      "grad_norm": 1.2776094675064087,
+      "learning_rate": 3.144288577154309e-05,
+      "loss": 0.0591,
+      "step": 1864
+    },
+    {
+      "epoch": 7.46,
+      "grad_norm": 2.1104819774627686,
+      "learning_rate": 3.143286573146293e-05,
+      "loss": 0.0901,
+      "step": 1865
+    },
+    {
+      "epoch": 7.464,
+      "grad_norm": 1.0128964185714722,
+      "learning_rate": 3.1422845691382766e-05,
+      "loss": 0.0475,
+      "step": 1866
+    },
+    {
+      "epoch": 7.468,
+      "grad_norm": 1.3107222318649292,
+      "learning_rate": 3.141282565130261e-05,
+      "loss": 0.0709,
+      "step": 1867
+    },
+    {
+      "epoch": 7.4719999999999995,
+      "grad_norm": 1.4184036254882812,
+      "learning_rate": 3.140280561122245e-05,
+      "loss": 0.0801,
+      "step": 1868
+    },
+    {
+      "epoch": 7.476,
+      "grad_norm": 1.6373494863510132,
+      "learning_rate": 3.139278557114228e-05,
+      "loss": 0.0716,
+      "step": 1869
+    },
+    {
+      "epoch": 7.48,
+      "grad_norm": 1.555827260017395,
+      "learning_rate": 3.1382765531062124e-05,
+      "loss": 0.0653,
+      "step": 1870
+    },
+    {
+      "epoch": 7.484,
+      "grad_norm": 1.7851310968399048,
+      "learning_rate": 3.1372745490981965e-05,
+      "loss": 0.0892,
+      "step": 1871
+    },
+    {
+      "epoch": 7.4879999999999995,
+      "grad_norm": 1.5268523693084717,
+      "learning_rate": 3.13627254509018e-05,
+      "loss": 0.0688,
+      "step": 1872
+    },
+    {
+      "epoch": 7.492,
+      "grad_norm": 1.477304220199585,
+      "learning_rate": 3.135270541082164e-05,
+      "loss": 0.0733,
+      "step": 1873
+    },
+    {
+      "epoch": 7.496,
+      "grad_norm": 1.2932136058807373,
+      "learning_rate": 3.134268537074149e-05,
+      "loss": 0.0636,
+      "step": 1874
+    },
+    {
+      "epoch": 7.5,
+      "grad_norm": 1.4812061786651611,
+      "learning_rate": 3.1332665330661324e-05,
+      "loss": 0.0747,
+      "step": 1875
+    },
+    {
+      "epoch": 7.504,
+      "grad_norm": 1.303823471069336,
+      "learning_rate": 3.1322645290581165e-05,
+      "loss": 0.0616,
+      "step": 1876
+    },
+    {
+      "epoch": 7.508,
+      "grad_norm": 1.4868488311767578,
+      "learning_rate": 3.1312625250501006e-05,
+      "loss": 0.0898,
+      "step": 1877
+    },
+    {
+      "epoch": 7.5120000000000005,
+      "grad_norm": 1.6528791189193726,
+      "learning_rate": 3.130260521042084e-05,
+      "loss": 0.0661,
+      "step": 1878
+    },
+    {
+      "epoch": 7.516,
+      "grad_norm": 1.552954077720642,
+      "learning_rate": 3.129258517034068e-05,
+      "loss": 0.0716,
+      "step": 1879
+    },
+    {
+      "epoch": 7.52,
+      "grad_norm": 1.2300714254379272,
+      "learning_rate": 3.128256513026052e-05,
+      "loss": 0.0625,
+      "step": 1880
+    },
+    {
+      "epoch": 7.524,
+      "grad_norm": 1.54698646068573,
+      "learning_rate": 3.127254509018036e-05,
+      "loss": 0.0765,
+      "step": 1881
+    },
+    {
+      "epoch": 7.5280000000000005,
+      "grad_norm": 1.5242317914962769,
+      "learning_rate": 3.12625250501002e-05,
+      "loss": 0.0922,
+      "step": 1882
+    },
+    {
+      "epoch": 7.532,
+      "grad_norm": 1.4363588094711304,
+      "learning_rate": 3.125250501002004e-05,
+      "loss": 0.0784,
+      "step": 1883
+    },
+    {
+      "epoch": 7.536,
+      "grad_norm": 1.3856273889541626,
+      "learning_rate": 3.124248496993988e-05,
+      "loss": 0.0669,
+      "step": 1884
+    },
+    {
+      "epoch": 7.54,
+      "grad_norm": 1.296464443206787,
+      "learning_rate": 3.123246492985972e-05,
+      "loss": 0.0713,
+      "step": 1885
+    },
+    {
+      "epoch": 7.5440000000000005,
+      "grad_norm": 1.3681565523147583,
+      "learning_rate": 3.1222444889779564e-05,
+      "loss": 0.0678,
+      "step": 1886
+    },
+    {
+      "epoch": 7.548,
+      "grad_norm": 1.34380304813385,
+      "learning_rate": 3.12124248496994e-05,
+      "loss": 0.0552,
+      "step": 1887
+    },
+    {
+      "epoch": 7.552,
+      "grad_norm": 1.2859309911727905,
+      "learning_rate": 3.120240480961924e-05,
+      "loss": 0.0602,
+      "step": 1888
+    },
+    {
+      "epoch": 7.556,
+      "grad_norm": 1.4414995908737183,
+      "learning_rate": 3.119238476953908e-05,
+      "loss": 0.0646,
+      "step": 1889
+    },
+    {
+      "epoch": 7.5600000000000005,
+      "grad_norm": 1.549656629562378,
+      "learning_rate": 3.1182364729458916e-05,
+      "loss": 0.0848,
+      "step": 1890
+    },
+    {
+      "epoch": 7.564,
+      "grad_norm": 1.274708867073059,
+      "learning_rate": 3.117234468937876e-05,
+      "loss": 0.0634,
+      "step": 1891
+    },
+    {
+      "epoch": 7.568,
+      "grad_norm": 2.1878645420074463,
+      "learning_rate": 3.11623246492986e-05,
+      "loss": 0.0593,
+      "step": 1892
+    },
+    {
+      "epoch": 7.572,
+      "grad_norm": 1.367893099784851,
+      "learning_rate": 3.115230460921843e-05,
+      "loss": 0.0637,
+      "step": 1893
+    },
+    {
+      "epoch": 7.576,
+      "grad_norm": 1.5797450542449951,
+      "learning_rate": 3.114228456913828e-05,
+      "loss": 0.0853,
+      "step": 1894
+    },
+    {
+      "epoch": 7.58,
+      "grad_norm": 1.5319100618362427,
+      "learning_rate": 3.113226452905812e-05,
+      "loss": 0.0795,
+      "step": 1895
+    },
+    {
+      "epoch": 7.584,
+      "grad_norm": 1.3238308429718018,
+      "learning_rate": 3.112224448897796e-05,
+      "loss": 0.0622,
+      "step": 1896
+    },
+    {
+      "epoch": 7.588,
+      "grad_norm": 1.4970932006835938,
+      "learning_rate": 3.11122244488978e-05,
+      "loss": 0.0776,
+      "step": 1897
+    },
+    {
+      "epoch": 7.592,
+      "grad_norm": 0.7695464491844177,
+      "learning_rate": 3.110220440881764e-05,
+      "loss": 0.0338,
+      "step": 1898
+    },
+    {
+      "epoch": 7.596,
+      "grad_norm": 1.3819239139556885,
+      "learning_rate": 3.1092184368737474e-05,
+      "loss": 0.0804,
+      "step": 1899
+    },
+    {
+      "epoch": 7.6,
+      "grad_norm": 1.5732004642486572,
+      "learning_rate": 3.1082164328657315e-05,
+      "loss": 0.0869,
+      "step": 1900
+    },
+    {
+      "epoch": 7.604,
+      "grad_norm": 1.3113768100738525,
+      "learning_rate": 3.1072144288577156e-05,
+      "loss": 0.0659,
+      "step": 1901
+    },
+    {
+      "epoch": 7.608,
+      "grad_norm": 1.4455841779708862,
+      "learning_rate": 3.106212424849699e-05,
+      "loss": 0.0847,
+      "step": 1902
+    },
+    {
+      "epoch": 7.612,
+      "grad_norm": 0.8127650618553162,
+      "learning_rate": 3.105210420841684e-05,
+      "loss": 0.044,
+      "step": 1903
+    },
+    {
+      "epoch": 7.616,
+      "grad_norm": 1.613440990447998,
+      "learning_rate": 3.104208416833668e-05,
+      "loss": 0.0779,
+      "step": 1904
+    },
+    {
+      "epoch": 7.62,
+      "grad_norm": 1.3210656642913818,
+      "learning_rate": 3.1032064128256515e-05,
+      "loss": 0.058,
+      "step": 1905
+    },
+    {
+      "epoch": 7.624,
+      "grad_norm": 1.4061453342437744,
+      "learning_rate": 3.1022044088176356e-05,
+      "loss": 0.0759,
+      "step": 1906
+    },
+    {
+      "epoch": 7.628,
+      "grad_norm": 1.5052586793899536,
+      "learning_rate": 3.10120240480962e-05,
+      "loss": 0.0785,
+      "step": 1907
+    },
+    {
+      "epoch": 7.632,
+      "grad_norm": 1.3261641263961792,
+      "learning_rate": 3.100200400801603e-05,
+      "loss": 0.0922,
+      "step": 1908
+    },
+    {
+      "epoch": 7.636,
+      "grad_norm": 1.5588467121124268,
+      "learning_rate": 3.099198396793587e-05,
+      "loss": 0.0703,
+      "step": 1909
+    },
+    {
+      "epoch": 7.64,
+      "grad_norm": 1.2506775856018066,
+      "learning_rate": 3.0981963927855714e-05,
+      "loss": 0.0537,
+      "step": 1910
+    },
+    {
+      "epoch": 7.644,
+      "grad_norm": 1.6546599864959717,
+      "learning_rate": 3.097194388777555e-05,
+      "loss": 0.0913,
+      "step": 1911
+    },
+    {
+      "epoch": 7.648,
+      "grad_norm": 1.600527286529541,
+      "learning_rate": 3.096192384769539e-05,
+      "loss": 0.0788,
+      "step": 1912
+    },
+    {
+      "epoch": 7.652,
+      "grad_norm": 1.5613360404968262,
+      "learning_rate": 3.095190380761523e-05,
+      "loss": 0.068,
+      "step": 1913
+    },
+    {
+      "epoch": 7.656,
+      "grad_norm": 1.5077072381973267,
+      "learning_rate": 3.094188376753507e-05,
+      "loss": 0.0729,
+      "step": 1914
+    },
+    {
+      "epoch": 7.66,
+      "grad_norm": 1.6048588752746582,
+      "learning_rate": 3.0931863727454914e-05,
+      "loss": 0.0876,
+      "step": 1915
+    },
+    {
+      "epoch": 7.664,
+      "grad_norm": 1.4152404069900513,
+      "learning_rate": 3.0921843687374755e-05,
+      "loss": 0.0876,
+      "step": 1916
+    },
+    {
+      "epoch": 7.668,
+      "grad_norm": 1.5331817865371704,
+      "learning_rate": 3.091182364729459e-05,
+      "loss": 0.0765,
+      "step": 1917
+    },
+    {
+      "epoch": 7.672,
+      "grad_norm": 1.5034297704696655,
+      "learning_rate": 3.090180360721443e-05,
+      "loss": 0.0618,
+      "step": 1918
+    },
+    {
+      "epoch": 7.676,
+      "grad_norm": 1.4812408685684204,
+      "learning_rate": 3.089178356713427e-05,
+      "loss": 0.0791,
+      "step": 1919
+    },
+    {
+      "epoch": 7.68,
+      "grad_norm": 1.4071189165115356,
+      "learning_rate": 3.088176352705411e-05,
+      "loss": 0.0773,
+      "step": 1920
+    },
+    {
+      "epoch": 7.684,
+      "grad_norm": 1.3819324970245361,
+      "learning_rate": 3.087174348697395e-05,
+      "loss": 0.0734,
+      "step": 1921
+    },
+    {
+      "epoch": 7.688,
+      "grad_norm": 1.3420048952102661,
+      "learning_rate": 3.086172344689379e-05,
+      "loss": 0.0659,
+      "step": 1922
+    },
+    {
+      "epoch": 7.692,
+      "grad_norm": 1.4877381324768066,
+      "learning_rate": 3.085170340681363e-05,
+      "loss": 0.0729,
+      "step": 1923
+    },
+    {
+      "epoch": 7.696,
+      "grad_norm": 1.2913259267807007,
+      "learning_rate": 3.084168336673347e-05,
+      "loss": 0.0593,
+      "step": 1924
+    },
+    {
+      "epoch": 7.7,
+      "grad_norm": 1.3754169940948486,
+      "learning_rate": 3.0831663326653306e-05,
+      "loss": 0.0668,
+      "step": 1925
+    },
+    {
+      "epoch": 7.704,
+      "grad_norm": 1.3683477640151978,
+      "learning_rate": 3.082164328657315e-05,
+      "loss": 0.0736,
+      "step": 1926
+    },
+    {
+      "epoch": 7.708,
+      "grad_norm": 1.151803970336914,
+      "learning_rate": 3.081162324649299e-05,
+      "loss": 0.0581,
+      "step": 1927
+    },
+    {
+      "epoch": 7.712,
+      "grad_norm": 1.4740242958068848,
+      "learning_rate": 3.0801603206412824e-05,
+      "loss": 0.0903,
+      "step": 1928
+    },
+    {
+      "epoch": 7.716,
+      "grad_norm": 1.588667869567871,
+      "learning_rate": 3.0791583166332665e-05,
+      "loss": 0.083,
+      "step": 1929
+    },
+    {
+      "epoch": 7.72,
+      "grad_norm": 1.4795126914978027,
+      "learning_rate": 3.0781563126252506e-05,
+      "loss": 0.0743,
+      "step": 1930
+    },
+    {
+      "epoch": 7.724,
+      "grad_norm": 1.3423376083374023,
+      "learning_rate": 3.077154308617235e-05,
+      "loss": 0.0701,
+      "step": 1931
+    },
+    {
+      "epoch": 7.728,
+      "grad_norm": 1.1936395168304443,
+      "learning_rate": 3.076152304609218e-05,
+      "loss": 0.0645,
+      "step": 1932
+    },
+    {
+      "epoch": 7.732,
+      "grad_norm": 1.1713727712631226,
+      "learning_rate": 3.075150300601203e-05,
+      "loss": 0.0536,
+      "step": 1933
+    },
+    {
+      "epoch": 7.736,
+      "grad_norm": 1.3141846656799316,
+      "learning_rate": 3.0741482965931864e-05,
+      "loss": 0.0723,
+      "step": 1934
+    },
+    {
+      "epoch": 7.74,
+      "grad_norm": 1.4886362552642822,
+      "learning_rate": 3.0731462925851706e-05,
+      "loss": 0.0902,
+      "step": 1935
+    },
+    {
+      "epoch": 7.744,
+      "grad_norm": 1.4828537702560425,
+      "learning_rate": 3.072144288577155e-05,
+      "loss": 0.0807,
+      "step": 1936
+    },
+    {
+      "epoch": 7.748,
+      "grad_norm": 1.341299295425415,
+      "learning_rate": 3.071142284569138e-05,
+      "loss": 0.0692,
+      "step": 1937
+    },
+    {
+      "epoch": 7.752,
+      "grad_norm": 1.3996098041534424,
+      "learning_rate": 3.070140280561122e-05,
+      "loss": 0.0658,
+      "step": 1938
+    },
+    {
+      "epoch": 7.756,
+      "grad_norm": 1.3277019262313843,
+      "learning_rate": 3.0691382765531064e-05,
+      "loss": 0.0825,
+      "step": 1939
+    },
+    {
+      "epoch": 7.76,
+      "grad_norm": 1.4799069166183472,
+      "learning_rate": 3.06813627254509e-05,
+      "loss": 0.0631,
+      "step": 1940
+    },
+    {
+      "epoch": 7.764,
+      "grad_norm": 1.4274985790252686,
+      "learning_rate": 3.067134268537074e-05,
+      "loss": 0.0756,
+      "step": 1941
+    },
+    {
+      "epoch": 7.768,
+      "grad_norm": 1.5825673341751099,
+      "learning_rate": 3.066132264529058e-05,
+      "loss": 0.0992,
+      "step": 1942
+    },
+    {
+      "epoch": 7.772,
+      "grad_norm": 1.4629828929901123,
+      "learning_rate": 3.065130260521042e-05,
+      "loss": 0.0893,
+      "step": 1943
+    },
+    {
+      "epoch": 7.776,
+      "grad_norm": 1.2774440050125122,
+      "learning_rate": 3.0641282565130264e-05,
+      "loss": 0.059,
+      "step": 1944
+    },
+    {
+      "epoch": 7.78,
+      "grad_norm": 1.3957229852676392,
+      "learning_rate": 3.0631262525050105e-05,
+      "loss": 0.0699,
+      "step": 1945
+    },
+    {
+      "epoch": 7.784,
+      "grad_norm": 1.5899467468261719,
+      "learning_rate": 3.062124248496994e-05,
+      "loss": 0.0865,
+      "step": 1946
+    },
+    {
+      "epoch": 7.788,
+      "grad_norm": 1.4467995166778564,
+      "learning_rate": 3.061122244488978e-05,
+      "loss": 0.0765,
+      "step": 1947
+    },
+    {
+      "epoch": 7.792,
+      "grad_norm": 1.4088798761367798,
+      "learning_rate": 3.060120240480962e-05,
+      "loss": 0.07,
+      "step": 1948
+    },
+    {
+      "epoch": 7.796,
+      "grad_norm": 1.3148294687271118,
+      "learning_rate": 3.0591182364729457e-05,
+      "loss": 0.0845,
+      "step": 1949
+    },
+    {
+      "epoch": 7.8,
+      "grad_norm": 1.445083737373352,
+      "learning_rate": 3.05811623246493e-05,
+      "loss": 0.0739,
+      "step": 1950
+    },
+    {
+      "epoch": 7.804,
+      "grad_norm": 1.4596608877182007,
+      "learning_rate": 3.057114228456914e-05,
+      "loss": 0.0684,
+      "step": 1951
+    },
+    {
+      "epoch": 7.808,
+      "grad_norm": 1.3999098539352417,
+      "learning_rate": 3.0561122244488974e-05,
+      "loss": 0.0638,
+      "step": 1952
+    },
+    {
+      "epoch": 7.812,
+      "grad_norm": 1.6097480058670044,
+      "learning_rate": 3.055110220440882e-05,
+      "loss": 0.0812,
+      "step": 1953
+    },
+    {
+      "epoch": 7.816,
+      "grad_norm": 1.6041979789733887,
+      "learning_rate": 3.054108216432866e-05,
+      "loss": 0.0989,
+      "step": 1954
+    },
+    {
+      "epoch": 7.82,
+      "grad_norm": 1.4978128671646118,
+      "learning_rate": 3.05310621242485e-05,
+      "loss": 0.0799,
+      "step": 1955
+    },
+    {
+      "epoch": 7.824,
+      "grad_norm": 1.465773582458496,
+      "learning_rate": 3.052104208416834e-05,
+      "loss": 0.0827,
+      "step": 1956
+    },
+    {
+      "epoch": 7.828,
+      "grad_norm": 1.273937463760376,
+      "learning_rate": 3.0511022044088177e-05,
+      "loss": 0.0664,
+      "step": 1957
+    },
+    {
+      "epoch": 7.832,
+      "grad_norm": 1.212803602218628,
+      "learning_rate": 3.0501002004008018e-05,
+      "loss": 0.0612,
+      "step": 1958
+    },
+    {
+      "epoch": 7.836,
+      "grad_norm": 1.3836429119110107,
+      "learning_rate": 3.0490981963927856e-05,
+      "loss": 0.0711,
+      "step": 1959
+    },
+    {
+      "epoch": 7.84,
+      "grad_norm": 1.5588130950927734,
+      "learning_rate": 3.0480961923847694e-05,
+      "loss": 0.0792,
+      "step": 1960
+    },
+    {
+      "epoch": 7.844,
+      "grad_norm": 1.5219345092773438,
+      "learning_rate": 3.0470941883767535e-05,
+      "loss": 0.0826,
+      "step": 1961
+    },
+    {
+      "epoch": 7.848,
+      "grad_norm": 1.6055107116699219,
+      "learning_rate": 3.046092184368738e-05,
+      "loss": 0.0766,
+      "step": 1962
+    },
+    {
+      "epoch": 7.852,
+      "grad_norm": 1.2326933145523071,
+      "learning_rate": 3.0450901803607218e-05,
+      "loss": 0.0622,
+      "step": 1963
+    },
+    {
+      "epoch": 7.856,
+      "grad_norm": 1.6281651258468628,
+      "learning_rate": 3.0440881763527055e-05,
+      "loss": 0.0789,
+      "step": 1964
+    },
+    {
+      "epoch": 7.86,
+      "grad_norm": 1.5395461320877075,
+      "learning_rate": 3.0430861723446897e-05,
+      "loss": 0.0793,
+      "step": 1965
+    },
+    {
+      "epoch": 7.864,
+      "grad_norm": 1.4532355070114136,
+      "learning_rate": 3.0420841683366735e-05,
+      "loss": 0.0667,
+      "step": 1966
+    },
+    {
+      "epoch": 7.868,
+      "grad_norm": 1.4249470233917236,
+      "learning_rate": 3.0410821643286576e-05,
+      "loss": 0.0675,
+      "step": 1967
+    },
+    {
+      "epoch": 7.872,
+      "grad_norm": 1.4712321758270264,
+      "learning_rate": 3.0400801603206414e-05,
+      "loss": 0.0563,
+      "step": 1968
+    },
+    {
+      "epoch": 7.876,
+      "grad_norm": 1.6727486848831177,
+      "learning_rate": 3.0390781563126252e-05,
+      "loss": 0.0915,
+      "step": 1969
+    },
+    {
+      "epoch": 7.88,
+      "grad_norm": 1.4988911151885986,
+      "learning_rate": 3.0380761523046093e-05,
+      "loss": 0.0754,
+      "step": 1970
+    },
+    {
+      "epoch": 7.884,
+      "grad_norm": 1.4844058752059937,
+      "learning_rate": 3.037074148296593e-05,
+      "loss": 0.0777,
+      "step": 1971
+    },
+    {
+      "epoch": 7.888,
+      "grad_norm": 1.4595845937728882,
+      "learning_rate": 3.0360721442885776e-05,
+      "loss": 0.0714,
+      "step": 1972
+    },
+    {
+      "epoch": 7.892,
+      "grad_norm": 1.6251410245895386,
+      "learning_rate": 3.0350701402805613e-05,
+      "loss": 0.0891,
+      "step": 1973
+    },
+    {
+      "epoch": 7.896,
+      "grad_norm": 1.4736934900283813,
+      "learning_rate": 3.0340681362725455e-05,
+      "loss": 0.0776,
+      "step": 1974
+    },
+    {
+      "epoch": 7.9,
+      "grad_norm": 1.6098906993865967,
+      "learning_rate": 3.0330661322645293e-05,
+      "loss": 0.0738,
+      "step": 1975
+    },
+    {
+      "epoch": 7.904,
+      "grad_norm": 1.4233413934707642,
+      "learning_rate": 3.032064128256513e-05,
+      "loss": 0.0681,
+      "step": 1976
+    },
+    {
+      "epoch": 7.908,
+      "grad_norm": 1.4190319776535034,
+      "learning_rate": 3.0310621242484972e-05,
+      "loss": 0.0941,
+      "step": 1977
+    },
+    {
+      "epoch": 7.912,
+      "grad_norm": 1.4623279571533203,
+      "learning_rate": 3.030060120240481e-05,
+      "loss": 0.0736,
+      "step": 1978
+    },
+    {
+      "epoch": 7.916,
+      "grad_norm": 1.6991535425186157,
+      "learning_rate": 3.029058116232465e-05,
+      "loss": 0.0896,
+      "step": 1979
+    },
+    {
+      "epoch": 7.92,
+      "grad_norm": 1.6135594844818115,
+      "learning_rate": 3.028056112224449e-05,
+      "loss": 0.0841,
+      "step": 1980
+    },
+    {
+      "epoch": 7.924,
+      "grad_norm": 1.5118712186813354,
+      "learning_rate": 3.0270541082164327e-05,
+      "loss": 0.0966,
+      "step": 1981
+    },
+    {
+      "epoch": 7.928,
+      "grad_norm": 1.3589533567428589,
+      "learning_rate": 3.026052104208417e-05,
+      "loss": 0.0725,
+      "step": 1982
+    },
+    {
+      "epoch": 7.932,
+      "grad_norm": 1.5694752931594849,
+      "learning_rate": 3.0250501002004013e-05,
+      "loss": 0.0785,
+      "step": 1983
+    },
+    {
+      "epoch": 7.936,
+      "grad_norm": 1.42135751247406,
+      "learning_rate": 3.024048096192385e-05,
+      "loss": 0.0802,
+      "step": 1984
+    },
+    {
+      "epoch": 7.9399999999999995,
+      "grad_norm": 1.4186774492263794,
+      "learning_rate": 3.023046092184369e-05,
+      "loss": 0.071,
+      "step": 1985
+    },
+    {
+      "epoch": 7.944,
+      "grad_norm": 0.7611780166625977,
+      "learning_rate": 3.022044088176353e-05,
+      "loss": 0.0184,
+      "step": 1986
+    },
+    {
+      "epoch": 7.948,
+      "grad_norm": 1.5581532716751099,
+      "learning_rate": 3.0210420841683368e-05,
+      "loss": 0.0853,
+      "step": 1987
+    },
+    {
+      "epoch": 7.952,
+      "grad_norm": 1.394500494003296,
+      "learning_rate": 3.0200400801603206e-05,
+      "loss": 0.0915,
+      "step": 1988
+    },
+    {
+      "epoch": 7.9559999999999995,
+      "grad_norm": 1.33413565158844,
+      "learning_rate": 3.0190380761523047e-05,
+      "loss": 0.0689,
+      "step": 1989
+    },
+    {
+      "epoch": 7.96,
+      "grad_norm": 1.417820692062378,
+      "learning_rate": 3.0180360721442885e-05,
+      "loss": 0.0861,
+      "step": 1990
+    },
+    {
+      "epoch": 7.964,
+      "grad_norm": 1.309669852256775,
+      "learning_rate": 3.0170340681362723e-05,
+      "loss": 0.0703,
+      "step": 1991
+    },
+    {
+      "epoch": 7.968,
+      "grad_norm": 1.4722129106521606,
+      "learning_rate": 3.0160320641282567e-05,
+      "loss": 0.0636,
+      "step": 1992
+    },
+    {
+      "epoch": 7.9719999999999995,
+      "grad_norm": 1.4907193183898926,
+      "learning_rate": 3.015030060120241e-05,
+      "loss": 0.0699,
+      "step": 1993
+    },
+    {
+      "epoch": 7.976,
+      "grad_norm": 1.5335925817489624,
+      "learning_rate": 3.0140280561122247e-05,
+      "loss": 0.0712,
+      "step": 1994
+    },
+    {
+      "epoch": 7.98,
+      "grad_norm": 1.2255603075027466,
+      "learning_rate": 3.0130260521042088e-05,
+      "loss": 0.0602,
+      "step": 1995
+    },
+    {
+      "epoch": 7.984,
+      "grad_norm": 1.4105703830718994,
+      "learning_rate": 3.0120240480961926e-05,
+      "loss": 0.0731,
+      "step": 1996
+    },
+    {
+      "epoch": 7.9879999999999995,
+      "grad_norm": 1.5889180898666382,
+      "learning_rate": 3.0110220440881764e-05,
+      "loss": 0.0752,
+      "step": 1997
+    },
+    {
+      "epoch": 7.992,
+      "grad_norm": 1.6140540838241577,
+      "learning_rate": 3.0100200400801605e-05,
+      "loss": 0.0753,
+      "step": 1998
+    },
+    {
+      "epoch": 7.996,
+      "grad_norm": 1.826938509941101,
+      "learning_rate": 3.0090180360721443e-05,
+      "loss": 0.0852,
+      "step": 1999
+    },
+    {
+      "epoch": 8.0,
+      "grad_norm": 1.0301682949066162,
+      "learning_rate": 3.008016032064128e-05,
+      "loss": 0.045,
+      "step": 2000
+    },
+    {
+      "epoch": 8.004,
+      "grad_norm": 0.9636511206626892,
+      "learning_rate": 3.0070140280561122e-05,
+      "loss": 0.0422,
+      "step": 2001
+    },
+    {
+      "epoch": 8.008,
+      "grad_norm": 0.9918342232704163,
+      "learning_rate": 3.0060120240480967e-05,
+      "loss": 0.0413,
+      "step": 2002
+    },
+    {
+      "epoch": 8.012,
+      "grad_norm": 0.8967450857162476,
+      "learning_rate": 3.0050100200400805e-05,
+      "loss": 0.0343,
+      "step": 2003
+    },
+    {
+      "epoch": 8.016,
+      "grad_norm": 1.0246479511260986,
+      "learning_rate": 3.0040080160320642e-05,
+      "loss": 0.0399,
+      "step": 2004
+    },
+    {
+      "epoch": 8.02,
+      "grad_norm": 1.0252851247787476,
+      "learning_rate": 3.0030060120240484e-05,
+      "loss": 0.0458,
+      "step": 2005
+    },
+    {
+      "epoch": 8.024,
+      "grad_norm": 0.974763810634613,
+      "learning_rate": 3.002004008016032e-05,
+      "loss": 0.0437,
+      "step": 2006
+    },
+    {
+      "epoch": 8.028,
+      "grad_norm": 0.9569265842437744,
+      "learning_rate": 3.0010020040080163e-05,
+      "loss": 0.0364,
+      "step": 2007
+    },
+    {
+      "epoch": 8.032,
+      "grad_norm": 1.104304313659668,
+      "learning_rate": 3e-05,
+      "loss": 0.0397,
+      "step": 2008
+    },
+    {
+      "epoch": 8.036,
+      "grad_norm": 1.042075753211975,
+      "learning_rate": 2.998997995991984e-05,
+      "loss": 0.0319,
+      "step": 2009
+    },
+    {
+      "epoch": 8.04,
+      "grad_norm": 1.41454017162323,
+      "learning_rate": 2.997995991983968e-05,
+      "loss": 0.0519,
+      "step": 2010
+    },
+    {
+      "epoch": 8.044,
+      "grad_norm": 1.2228070497512817,
+      "learning_rate": 2.9969939879759525e-05,
+      "loss": 0.0471,
+      "step": 2011
+    },
+    {
+      "epoch": 8.048,
+      "grad_norm": 1.3753093481063843,
+      "learning_rate": 2.9959919839679363e-05,
+      "loss": 0.0426,
+      "step": 2012
+    },
+    {
+      "epoch": 8.052,
+      "grad_norm": 1.2249881029129028,
+      "learning_rate": 2.99498997995992e-05,
+      "loss": 0.0375,
+      "step": 2013
+    },
+    {
+      "epoch": 8.056,
+      "grad_norm": 1.1894644498825073,
+      "learning_rate": 2.993987975951904e-05,
+      "loss": 0.0388,
+      "step": 2014
+    },
+    {
+      "epoch": 8.06,
+      "grad_norm": 1.2309092283248901,
+      "learning_rate": 2.992985971943888e-05,
+      "loss": 0.0403,
+      "step": 2015
+    },
+    {
+      "epoch": 8.064,
+      "grad_norm": 1.237890362739563,
+      "learning_rate": 2.9919839679358717e-05,
+      "loss": 0.0414,
+      "step": 2016
+    },
+    {
+      "epoch": 8.068,
+      "grad_norm": 1.3843622207641602,
+      "learning_rate": 2.990981963927856e-05,
+      "loss": 0.0462,
+      "step": 2017
+    },
+    {
+      "epoch": 8.072,
+      "grad_norm": 1.204715609550476,
+      "learning_rate": 2.9899799599198397e-05,
+      "loss": 0.0455,
+      "step": 2018
+    },
+    {
+      "epoch": 8.076,
+      "grad_norm": 1.1877174377441406,
+      "learning_rate": 2.9889779559118235e-05,
+      "loss": 0.0429,
+      "step": 2019
+    },
+    {
+      "epoch": 8.08,
+      "grad_norm": 0.9619832634925842,
+      "learning_rate": 2.9879759519038076e-05,
+      "loss": 0.0344,
+      "step": 2020
+    },
+    {
+      "epoch": 8.084,
+      "grad_norm": 1.2770909070968628,
+      "learning_rate": 2.986973947895792e-05,
+      "loss": 0.0434,
+      "step": 2021
+    },
+    {
+      "epoch": 8.088,
+      "grad_norm": 2.0295021533966064,
+      "learning_rate": 2.985971943887776e-05,
+      "loss": 0.0478,
+      "step": 2022
+    },
+    {
+      "epoch": 8.092,
+      "grad_norm": 1.0458402633666992,
+      "learning_rate": 2.98496993987976e-05,
+      "loss": 0.0407,
+      "step": 2023
+    },
+    {
+      "epoch": 8.096,
+      "grad_norm": 1.3169829845428467,
+      "learning_rate": 2.9839679358717438e-05,
+      "loss": 0.0383,
+      "step": 2024
+    },
+    {
+      "epoch": 8.1,
+      "grad_norm": 1.2271572351455688,
+      "learning_rate": 2.9829659318637275e-05,
+      "loss": 0.0426,
+      "step": 2025
+    },
+    {
+      "epoch": 8.104,
+      "grad_norm": 0.6691854596138,
+      "learning_rate": 2.9819639278557117e-05,
+      "loss": 0.0178,
+      "step": 2026
+    },
+    {
+      "epoch": 8.108,
+      "grad_norm": 1.2203044891357422,
+      "learning_rate": 2.9809619238476955e-05,
+      "loss": 0.049,
+      "step": 2027
+    },
+    {
+      "epoch": 8.112,
+      "grad_norm": 1.1004842519760132,
+      "learning_rate": 2.9799599198396793e-05,
+      "loss": 0.0413,
+      "step": 2028
+    },
+    {
+      "epoch": 8.116,
+      "grad_norm": 1.1500310897827148,
+      "learning_rate": 2.9789579158316634e-05,
+      "loss": 0.052,
+      "step": 2029
+    },
+    {
+      "epoch": 8.12,
+      "grad_norm": 1.0582787990570068,
+      "learning_rate": 2.9779559118236472e-05,
+      "loss": 0.0387,
+      "step": 2030
+    },
+    {
+      "epoch": 8.124,
+      "grad_norm": 1.1282780170440674,
+      "learning_rate": 2.9769539078156316e-05,
+      "loss": 0.0463,
+      "step": 2031
+    },
+    {
+      "epoch": 8.128,
+      "grad_norm": 1.2757964134216309,
+      "learning_rate": 2.9759519038076154e-05,
+      "loss": 0.0458,
+      "step": 2032
+    },
+    {
+      "epoch": 8.132,
+      "grad_norm": 1.3332687616348267,
+      "learning_rate": 2.9749498997995996e-05,
+      "loss": 0.0527,
+      "step": 2033
+    },
+    {
+      "epoch": 8.136,
+      "grad_norm": 1.269371747970581,
+      "learning_rate": 2.9739478957915833e-05,
+      "loss": 0.0555,
+      "step": 2034
+    },
+    {
+      "epoch": 8.14,
+      "grad_norm": 1.179306149482727,
+      "learning_rate": 2.9729458917835675e-05,
+      "loss": 0.0379,
+      "step": 2035
+    },
+    {
+      "epoch": 8.144,
+      "grad_norm": 1.3321940898895264,
+      "learning_rate": 2.9719438877755513e-05,
+      "loss": 0.0554,
+      "step": 2036
+    },
+    {
+      "epoch": 8.148,
+      "grad_norm": 0.9157149791717529,
+      "learning_rate": 2.970941883767535e-05,
+      "loss": 0.033,
+      "step": 2037
+    },
+    {
+      "epoch": 8.152,
+      "grad_norm": 1.1154097318649292,
+      "learning_rate": 2.9699398797595192e-05,
+      "loss": 0.0384,
+      "step": 2038
+    },
+    {
+      "epoch": 8.156,
+      "grad_norm": 0.9774281978607178,
+      "learning_rate": 2.968937875751503e-05,
+      "loss": 0.0333,
+      "step": 2039
+    },
+    {
+      "epoch": 8.16,
+      "grad_norm": 1.5343128442764282,
+      "learning_rate": 2.9679358717434868e-05,
+      "loss": 0.0353,
+      "step": 2040
+    },
+    {
+      "epoch": 8.164,
+      "grad_norm": 1.0981296300888062,
+      "learning_rate": 2.9669338677354712e-05,
+      "loss": 0.0372,
+      "step": 2041
+    },
+    {
+      "epoch": 8.168,
+      "grad_norm": 1.17277193069458,
+      "learning_rate": 2.9659318637274554e-05,
+      "loss": 0.0456,
+      "step": 2042
+    },
+    {
+      "epoch": 8.172,
+      "grad_norm": 1.0405935049057007,
+      "learning_rate": 2.964929859719439e-05,
+      "loss": 0.0431,
+      "step": 2043
+    },
+    {
+      "epoch": 8.176,
+      "grad_norm": 1.213351845741272,
+      "learning_rate": 2.963927855711423e-05,
+      "loss": 0.0469,
+      "step": 2044
+    },
+    {
+      "epoch": 8.18,
+      "grad_norm": 1.091701865196228,
+      "learning_rate": 2.962925851703407e-05,
+      "loss": 0.0409,
+      "step": 2045
+    },
+    {
+      "epoch": 8.184,
+      "grad_norm": 1.140825629234314,
+      "learning_rate": 2.961923847695391e-05,
+      "loss": 0.0399,
+      "step": 2046
+    },
+    {
+      "epoch": 8.188,
+      "grad_norm": 1.3966275453567505,
+      "learning_rate": 2.9609218436873746e-05,
+      "loss": 0.047,
+      "step": 2047
+    },
+    {
+      "epoch": 8.192,
+      "grad_norm": 1.0607054233551025,
+      "learning_rate": 2.9599198396793588e-05,
+      "loss": 0.0403,
+      "step": 2048
+    },
+    {
+      "epoch": 8.196,
+      "grad_norm": 1.3309967517852783,
+      "learning_rate": 2.9589178356713426e-05,
+      "loss": 0.0424,
+      "step": 2049
+    },
+    {
+      "epoch": 8.2,
+      "grad_norm": 1.007638931274414,
+      "learning_rate": 2.9579158316633267e-05,
+      "loss": 0.0319,
+      "step": 2050
+    },
+    {
+      "epoch": 8.204,
+      "grad_norm": 1.2654892206192017,
+      "learning_rate": 2.956913827655311e-05,
+      "loss": 0.0453,
+      "step": 2051
+    },
+    {
+      "epoch": 8.208,
+      "grad_norm": 1.064241647720337,
+      "learning_rate": 2.955911823647295e-05,
+      "loss": 0.0436,
+      "step": 2052
+    },
+    {
+      "epoch": 8.212,
+      "grad_norm": 1.1412560939788818,
+      "learning_rate": 2.9549098196392787e-05,
+      "loss": 0.0434,
+      "step": 2053
+    },
+    {
+      "epoch": 8.216,
+      "grad_norm": 1.2644426822662354,
+      "learning_rate": 2.953907815631263e-05,
+      "loss": 0.0443,
+      "step": 2054
+    },
+    {
+      "epoch": 8.22,
+      "grad_norm": 1.153268814086914,
+      "learning_rate": 2.9529058116232467e-05,
+      "loss": 0.0446,
+      "step": 2055
+    },
+    {
+      "epoch": 8.224,
+      "grad_norm": 1.1821683645248413,
+      "learning_rate": 2.9519038076152304e-05,
+      "loss": 0.0403,
+      "step": 2056
+    },
+    {
+      "epoch": 8.228,
+      "grad_norm": 1.1083087921142578,
+      "learning_rate": 2.9509018036072146e-05,
+      "loss": 0.0376,
+      "step": 2057
+    },
+    {
+      "epoch": 8.232,
+      "grad_norm": 1.251597285270691,
+      "learning_rate": 2.9498997995991984e-05,
+      "loss": 0.0458,
+      "step": 2058
+    },
+    {
+      "epoch": 8.236,
+      "grad_norm": 1.3240916728973389,
+      "learning_rate": 2.948897795591182e-05,
+      "loss": 0.0505,
+      "step": 2059
+    },
+    {
+      "epoch": 8.24,
+      "grad_norm": 1.3797112703323364,
+      "learning_rate": 2.9478957915831663e-05,
+      "loss": 0.0468,
+      "step": 2060
+    },
+    {
+      "epoch": 8.244,
+      "grad_norm": 1.4539932012557983,
+      "learning_rate": 2.9468937875751507e-05,
+      "loss": 0.0459,
+      "step": 2061
+    },
+    {
+      "epoch": 8.248,
+      "grad_norm": 1.0563139915466309,
+      "learning_rate": 2.9458917835671345e-05,
+      "loss": 0.0375,
+      "step": 2062
+    },
+    {
+      "epoch": 8.252,
+      "grad_norm": 1.1842182874679565,
+      "learning_rate": 2.9448897795591183e-05,
+      "loss": 0.0441,
+      "step": 2063
+    },
+    {
+      "epoch": 8.256,
+      "grad_norm": 1.2775763273239136,
+      "learning_rate": 2.9438877755511024e-05,
+      "loss": 0.0487,
+      "step": 2064
+    },
+    {
+      "epoch": 8.26,
+      "grad_norm": 1.2534736394882202,
+      "learning_rate": 2.9428857715430862e-05,
+      "loss": 0.0389,
+      "step": 2065
+    },
+    {
+      "epoch": 8.264,
+      "grad_norm": 1.3463037014007568,
+      "learning_rate": 2.9418837675350704e-05,
+      "loss": 0.053,
+      "step": 2066
+    },
+    {
+      "epoch": 8.268,
+      "grad_norm": 0.9820384383201599,
+      "learning_rate": 2.940881763527054e-05,
+      "loss": 0.0467,
+      "step": 2067
+    },
+    {
+      "epoch": 8.272,
+      "grad_norm": 1.1850944757461548,
+      "learning_rate": 2.939879759519038e-05,
+      "loss": 0.0398,
+      "step": 2068
+    },
+    {
+      "epoch": 8.276,
+      "grad_norm": 0.9997429847717285,
+      "learning_rate": 2.938877755511022e-05,
+      "loss": 0.0343,
+      "step": 2069
+    },
+    {
+      "epoch": 8.28,
+      "grad_norm": 1.2655606269836426,
+      "learning_rate": 2.9378757515030065e-05,
+      "loss": 0.0437,
+      "step": 2070
+    },
+    {
+      "epoch": 8.284,
+      "grad_norm": 1.0781662464141846,
+      "learning_rate": 2.9368737474949903e-05,
+      "loss": 0.0349,
+      "step": 2071
+    },
+    {
+      "epoch": 8.288,
+      "grad_norm": 1.158593773841858,
+      "learning_rate": 2.935871743486974e-05,
+      "loss": 0.046,
+      "step": 2072
+    },
+    {
+      "epoch": 8.292,
+      "grad_norm": 1.2248380184173584,
+      "learning_rate": 2.9348697394789582e-05,
+      "loss": 0.0406,
+      "step": 2073
+    },
+    {
+      "epoch": 8.296,
+      "grad_norm": 1.362268090248108,
+      "learning_rate": 2.933867735470942e-05,
+      "loss": 0.0516,
+      "step": 2074
+    },
+    {
+      "epoch": 8.3,
+      "grad_norm": 1.1527179479599,
+      "learning_rate": 2.9328657314629258e-05,
+      "loss": 0.0396,
+      "step": 2075
+    },
+    {
+      "epoch": 8.304,
+      "grad_norm": 1.1694166660308838,
+      "learning_rate": 2.93186372745491e-05,
+      "loss": 0.0374,
+      "step": 2076
+    },
+    {
+      "epoch": 8.308,
+      "grad_norm": 1.28037428855896,
+      "learning_rate": 2.9308617234468937e-05,
+      "loss": 0.0416,
+      "step": 2077
+    },
+    {
+      "epoch": 8.312,
+      "grad_norm": 1.2145733833312988,
+      "learning_rate": 2.929859719438878e-05,
+      "loss": 0.0433,
+      "step": 2078
+    },
+    {
+      "epoch": 8.316,
+      "grad_norm": 1.0807759761810303,
+      "learning_rate": 2.9288577154308617e-05,
+      "loss": 0.0364,
+      "step": 2079
+    },
+    {
+      "epoch": 8.32,
+      "grad_norm": 1.364870548248291,
+      "learning_rate": 2.927855711422846e-05,
+      "loss": 0.0582,
+      "step": 2080
+    },
+    {
+      "epoch": 8.324,
+      "grad_norm": 1.2568074464797974,
+      "learning_rate": 2.92685370741483e-05,
+      "loss": 0.042,
+      "step": 2081
+    },
+    {
+      "epoch": 8.328,
+      "grad_norm": 1.4024349451065063,
+      "learning_rate": 2.925851703406814e-05,
+      "loss": 0.0465,
+      "step": 2082
+    },
+    {
+      "epoch": 8.332,
+      "grad_norm": 1.2387995719909668,
+      "learning_rate": 2.924849699398798e-05,
+      "loss": 0.0375,
+      "step": 2083
+    },
+    {
+      "epoch": 8.336,
+      "grad_norm": 1.2565348148345947,
+      "learning_rate": 2.9238476953907816e-05,
+      "loss": 0.04,
+      "step": 2084
+    },
+    {
+      "epoch": 8.34,
+      "grad_norm": 1.384647011756897,
+      "learning_rate": 2.9228456913827658e-05,
+      "loss": 0.0525,
+      "step": 2085
+    },
+    {
+      "epoch": 8.344,
+      "grad_norm": 0.9669448137283325,
+      "learning_rate": 2.9218436873747495e-05,
+      "loss": 0.0398,
+      "step": 2086
+    },
+    {
+      "epoch": 8.348,
+      "grad_norm": 1.022433876991272,
+      "learning_rate": 2.9208416833667333e-05,
+      "loss": 0.037,
+      "step": 2087
+    },
+    {
+      "epoch": 8.352,
+      "grad_norm": 1.2381759881973267,
+      "learning_rate": 2.9198396793587175e-05,
+      "loss": 0.0512,
+      "step": 2088
+    },
+    {
+      "epoch": 8.356,
+      "grad_norm": 0.967339813709259,
+      "learning_rate": 2.9188376753507013e-05,
+      "loss": 0.0392,
+      "step": 2089
+    },
+    {
+      "epoch": 8.36,
+      "grad_norm": 1.4872891902923584,
+      "learning_rate": 2.9178356713426857e-05,
+      "loss": 0.0491,
+      "step": 2090
+    },
+    {
+      "epoch": 8.364,
+      "grad_norm": 1.1258385181427002,
+      "learning_rate": 2.9168336673346695e-05,
+      "loss": 0.0395,
+      "step": 2091
+    },
+    {
+      "epoch": 8.368,
+      "grad_norm": 1.4340835809707642,
+      "learning_rate": 2.9158316633266536e-05,
+      "loss": 0.0561,
+      "step": 2092
+    },
+    {
+      "epoch": 8.372,
+      "grad_norm": 1.2354909181594849,
+      "learning_rate": 2.9148296593186374e-05,
+      "loss": 0.0415,
+      "step": 2093
+    },
+    {
+      "epoch": 8.376,
+      "grad_norm": 1.414834976196289,
+      "learning_rate": 2.9138276553106216e-05,
+      "loss": 0.0563,
+      "step": 2094
+    },
+    {
+      "epoch": 8.38,
+      "grad_norm": 1.3660783767700195,
+      "learning_rate": 2.9128256513026053e-05,
+      "loss": 0.049,
+      "step": 2095
+    },
+    {
+      "epoch": 8.384,
+      "grad_norm": 1.2326569557189941,
+      "learning_rate": 2.911823647294589e-05,
+      "loss": 0.0458,
+      "step": 2096
+    },
+    {
+      "epoch": 8.388,
+      "grad_norm": 1.343539834022522,
+      "learning_rate": 2.9108216432865733e-05,
+      "loss": 0.0498,
+      "step": 2097
+    },
+    {
+      "epoch": 8.392,
+      "grad_norm": 1.4126864671707153,
+      "learning_rate": 2.909819639278557e-05,
+      "loss": 0.0485,
+      "step": 2098
+    },
+    {
+      "epoch": 8.396,
+      "grad_norm": 1.0817058086395264,
+      "learning_rate": 2.908817635270541e-05,
+      "loss": 0.0342,
+      "step": 2099
+    },
+    {
+      "epoch": 8.4,
+      "grad_norm": 1.1074159145355225,
+      "learning_rate": 2.9078156312625253e-05,
+      "loss": 0.0371,
+      "step": 2100
+    },
+    {
+      "epoch": 8.404,
+      "grad_norm": 1.0859711170196533,
+      "learning_rate": 2.9068136272545094e-05,
+      "loss": 0.0408,
+      "step": 2101
+    },
+    {
+      "epoch": 8.408,
+      "grad_norm": 1.2138441801071167,
+      "learning_rate": 2.9058116232464932e-05,
+      "loss": 0.0497,
+      "step": 2102
+    },
+    {
+      "epoch": 8.412,
+      "grad_norm": 0.9061650037765503,
+      "learning_rate": 2.904809619238477e-05,
+      "loss": 0.0325,
+      "step": 2103
+    },
+    {
+      "epoch": 8.416,
+      "grad_norm": 1.2485976219177246,
+      "learning_rate": 2.903807615230461e-05,
+      "loss": 0.0525,
+      "step": 2104
+    },
+    {
+      "epoch": 8.42,
+      "grad_norm": 1.1576125621795654,
+      "learning_rate": 2.902805611222445e-05,
+      "loss": 0.036,
+      "step": 2105
+    },
+    {
+      "epoch": 8.424,
+      "grad_norm": 1.201346755027771,
+      "learning_rate": 2.901803607214429e-05,
+      "loss": 0.0455,
+      "step": 2106
+    },
+    {
+      "epoch": 8.428,
+      "grad_norm": 1.1513805389404297,
+      "learning_rate": 2.900801603206413e-05,
+      "loss": 0.0441,
+      "step": 2107
+    },
+    {
+      "epoch": 8.432,
+      "grad_norm": 1.2415015697479248,
+      "learning_rate": 2.8997995991983966e-05,
+      "loss": 0.0485,
+      "step": 2108
+    },
+    {
+      "epoch": 8.436,
+      "grad_norm": 1.4972500801086426,
+      "learning_rate": 2.8987975951903808e-05,
+      "loss": 0.0563,
+      "step": 2109
+    },
+    {
+      "epoch": 8.44,
+      "grad_norm": 1.346211314201355,
+      "learning_rate": 2.8977955911823652e-05,
+      "loss": 0.0537,
+      "step": 2110
+    },
+    {
+      "epoch": 8.444,
+      "grad_norm": 1.4355465173721313,
+      "learning_rate": 2.896793587174349e-05,
+      "loss": 0.0446,
+      "step": 2111
+    },
+    {
+      "epoch": 8.448,
+      "grad_norm": 1.4319212436676025,
+      "learning_rate": 2.8957915831663328e-05,
+      "loss": 0.0439,
+      "step": 2112
+    },
+    {
+      "epoch": 8.452,
+      "grad_norm": 1.0715012550354004,
+      "learning_rate": 2.894789579158317e-05,
+      "loss": 0.0377,
+      "step": 2113
+    },
+    {
+      "epoch": 8.456,
+      "grad_norm": 1.2712820768356323,
+      "learning_rate": 2.8937875751503007e-05,
+      "loss": 0.0433,
+      "step": 2114
+    },
+    {
+      "epoch": 8.46,
+      "grad_norm": 1.385491132736206,
+      "learning_rate": 2.8927855711422845e-05,
+      "loss": 0.0472,
+      "step": 2115
+    },
+    {
+      "epoch": 8.464,
+      "grad_norm": 1.0765126943588257,
+      "learning_rate": 2.8917835671342686e-05,
+      "loss": 0.0392,
+      "step": 2116
+    },
+    {
+      "epoch": 8.468,
+      "grad_norm": 0.964982807636261,
+      "learning_rate": 2.8907815631262524e-05,
+      "loss": 0.0364,
+      "step": 2117
+    },
+    {
+      "epoch": 8.472,
+      "grad_norm": 1.0207576751708984,
+      "learning_rate": 2.8897795591182362e-05,
+      "loss": 0.0353,
+      "step": 2118
+    },
+    {
+      "epoch": 8.475999999999999,
+      "grad_norm": 1.363625407218933,
+      "learning_rate": 2.8887775551102207e-05,
+      "loss": 0.0479,
+      "step": 2119
+    },
+    {
+      "epoch": 8.48,
+      "grad_norm": 1.3118845224380493,
+      "learning_rate": 2.8877755511022048e-05,
+      "loss": 0.05,
+      "step": 2120
+    },
+    {
+      "epoch": 8.484,
+      "grad_norm": 1.3515814542770386,
+      "learning_rate": 2.8867735470941886e-05,
+      "loss": 0.0449,
+      "step": 2121
+    },
+    {
+      "epoch": 8.488,
+      "grad_norm": 1.2568910121917725,
+      "learning_rate": 2.8857715430861727e-05,
+      "loss": 0.0421,
+      "step": 2122
+    },
+    {
+      "epoch": 8.492,
+      "grad_norm": 1.1728308200836182,
+      "learning_rate": 2.8847695390781565e-05,
+      "loss": 0.0402,
+      "step": 2123
+    },
+    {
+      "epoch": 8.496,
+      "grad_norm": 1.3032832145690918,
+      "learning_rate": 2.8837675350701403e-05,
+      "loss": 0.0431,
+      "step": 2124
+    },
+    {
+      "epoch": 8.5,
+      "grad_norm": 1.277818202972412,
+      "learning_rate": 2.8827655310621244e-05,
+      "loss": 0.0703,
+      "step": 2125
+    },
+    {
+      "epoch": 8.504,
+      "grad_norm": 1.165521264076233,
+      "learning_rate": 2.8817635270541082e-05,
+      "loss": 0.0411,
+      "step": 2126
+    },
+    {
+      "epoch": 8.508,
+      "grad_norm": 1.190546989440918,
+      "learning_rate": 2.880761523046092e-05,
+      "loss": 0.0426,
+      "step": 2127
+    },
+    {
+      "epoch": 8.512,
+      "grad_norm": 1.323736310005188,
+      "learning_rate": 2.879759519038076e-05,
+      "loss": 0.0498,
+      "step": 2128
+    },
+    {
+      "epoch": 8.516,
+      "grad_norm": 1.3947193622589111,
+      "learning_rate": 2.8787575150300606e-05,
+      "loss": 0.0562,
+      "step": 2129
+    },
+    {
+      "epoch": 8.52,
+      "grad_norm": 1.3394712209701538,
+      "learning_rate": 2.8777555110220444e-05,
+      "loss": 0.0568,
+      "step": 2130
+    },
+    {
+      "epoch": 8.524000000000001,
+      "grad_norm": 1.2725778818130493,
+      "learning_rate": 2.8767535070140282e-05,
+      "loss": 0.0453,
+      "step": 2131
+    },
+    {
+      "epoch": 8.528,
+      "grad_norm": 0.6503176093101501,
+      "learning_rate": 2.8757515030060123e-05,
+      "loss": 0.0155,
+      "step": 2132
+    },
+    {
+      "epoch": 8.532,
+      "grad_norm": 1.112267017364502,
+      "learning_rate": 2.874749498997996e-05,
+      "loss": 0.0351,
+      "step": 2133
+    },
+    {
+      "epoch": 8.536,
+      "grad_norm": 1.2222670316696167,
+      "learning_rate": 2.8737474949899802e-05,
+      "loss": 0.04,
+      "step": 2134
+    },
+    {
+      "epoch": 8.54,
+      "grad_norm": 1.2434343099594116,
+      "learning_rate": 2.872745490981964e-05,
+      "loss": 0.0432,
+      "step": 2135
+    },
+    {
+      "epoch": 8.544,
+      "grad_norm": 1.369353175163269,
+      "learning_rate": 2.8717434869739478e-05,
+      "loss": 0.0541,
+      "step": 2136
+    },
+    {
+      "epoch": 8.548,
+      "grad_norm": 1.3330413103103638,
+      "learning_rate": 2.870741482965932e-05,
+      "loss": 0.043,
+      "step": 2137
+    },
+    {
+      "epoch": 8.552,
+      "grad_norm": 1.0383684635162354,
+      "learning_rate": 2.8697394789579157e-05,
+      "loss": 0.0414,
+      "step": 2138
+    },
+    {
+      "epoch": 8.556000000000001,
+      "grad_norm": 1.3430683612823486,
+      "learning_rate": 2.8687374749499002e-05,
+      "loss": 0.0558,
+      "step": 2139
+    },
+    {
+      "epoch": 8.56,
+      "grad_norm": 1.6391087770462036,
+      "learning_rate": 2.867735470941884e-05,
+      "loss": 0.0523,
+      "step": 2140
+    },
+    {
+      "epoch": 8.564,
+      "grad_norm": 1.2342334985733032,
+      "learning_rate": 2.866733466933868e-05,
+      "loss": 0.0495,
+      "step": 2141
+    },
+    {
+      "epoch": 8.568,
+      "grad_norm": 1.2186678647994995,
+      "learning_rate": 2.865731462925852e-05,
+      "loss": 0.0468,
+      "step": 2142
+    },
+    {
+      "epoch": 8.572,
+      "grad_norm": 1.2047197818756104,
+      "learning_rate": 2.8647294589178357e-05,
+      "loss": 0.0485,
+      "step": 2143
+    },
+    {
+      "epoch": 8.576,
+      "grad_norm": 1.139602541923523,
+      "learning_rate": 2.86372745490982e-05,
+      "loss": 0.0398,
+      "step": 2144
+    },
+    {
+      "epoch": 8.58,
+      "grad_norm": 1.393538475036621,
+      "learning_rate": 2.8627254509018036e-05,
+      "loss": 0.0499,
+      "step": 2145
+    },
+    {
+      "epoch": 8.584,
+      "grad_norm": 0.7064828276634216,
+      "learning_rate": 2.8617234468937874e-05,
+      "loss": 0.0208,
+      "step": 2146
+    },
+    {
+      "epoch": 8.588,
+      "grad_norm": 1.0699363946914673,
+      "learning_rate": 2.8607214428857715e-05,
+      "loss": 0.0452,
+      "step": 2147
+    },
+    {
+      "epoch": 8.592,
+      "grad_norm": 0.9702959060668945,
+      "learning_rate": 2.8597194388777553e-05,
+      "loss": 0.0258,
+      "step": 2148
+    },
+    {
+      "epoch": 8.596,
+      "grad_norm": 1.0714532136917114,
+      "learning_rate": 2.8587174348697398e-05,
+      "loss": 0.0399,
+      "step": 2149
+    },
+    {
+      "epoch": 8.6,
+      "grad_norm": 0.7246587872505188,
+      "learning_rate": 2.857715430861724e-05,
+      "loss": 0.0249,
+      "step": 2150
+    },
+    {
+      "epoch": 8.604,
+      "grad_norm": 1.354813814163208,
+      "learning_rate": 2.8567134268537077e-05,
+      "loss": 0.0458,
+      "step": 2151
+    },
+    {
+      "epoch": 8.608,
+      "grad_norm": 1.3162986040115356,
+      "learning_rate": 2.8557114228456915e-05,
+      "loss": 0.0508,
+      "step": 2152
+    },
+    {
+      "epoch": 8.612,
+      "grad_norm": 1.2697739601135254,
+      "learning_rate": 2.8547094188376756e-05,
+      "loss": 0.0472,
+      "step": 2153
+    },
+    {
+      "epoch": 8.616,
+      "grad_norm": 1.3663972616195679,
+      "learning_rate": 2.8537074148296594e-05,
+      "loss": 0.0535,
+      "step": 2154
+    },
+    {
+      "epoch": 8.62,
+      "grad_norm": 1.2056630849838257,
+      "learning_rate": 2.8527054108216432e-05,
+      "loss": 0.0419,
+      "step": 2155
+    },
+    {
+      "epoch": 8.624,
+      "grad_norm": 1.1879265308380127,
+      "learning_rate": 2.8517034068136273e-05,
+      "loss": 0.042,
+      "step": 2156
+    },
+    {
+      "epoch": 8.628,
+      "grad_norm": 1.250275731086731,
+      "learning_rate": 2.850701402805611e-05,
+      "loss": 0.0428,
+      "step": 2157
+    },
+    {
+      "epoch": 8.632,
+      "grad_norm": 1.2584730386734009,
+      "learning_rate": 2.849699398797595e-05,
+      "loss": 0.0453,
+      "step": 2158
+    },
+    {
+      "epoch": 8.636,
+      "grad_norm": 1.3111329078674316,
+      "learning_rate": 2.8486973947895794e-05,
+      "loss": 0.0501,
+      "step": 2159
+    },
+    {
+      "epoch": 8.64,
+      "grad_norm": 1.344029426574707,
+      "learning_rate": 2.8476953907815635e-05,
+      "loss": 0.0479,
+      "step": 2160
+    },
+    {
+      "epoch": 8.644,
+      "grad_norm": 1.4813225269317627,
+      "learning_rate": 2.8466933867735473e-05,
+      "loss": 0.0489,
+      "step": 2161
+    },
+    {
+      "epoch": 8.648,
+      "grad_norm": 1.3338834047317505,
+      "learning_rate": 2.8456913827655314e-05,
+      "loss": 0.0481,
+      "step": 2162
+    },
+    {
+      "epoch": 8.652,
+      "grad_norm": 1.295943260192871,
+      "learning_rate": 2.8446893787575152e-05,
+      "loss": 0.0403,
+      "step": 2163
+    },
+    {
+      "epoch": 8.656,
+      "grad_norm": 1.1170644760131836,
+      "learning_rate": 2.843687374749499e-05,
+      "loss": 0.0453,
+      "step": 2164
+    },
+    {
+      "epoch": 8.66,
+      "grad_norm": 1.6115245819091797,
+      "learning_rate": 2.842685370741483e-05,
+      "loss": 0.0496,
+      "step": 2165
+    },
+    {
+      "epoch": 8.664,
+      "grad_norm": 1.2752050161361694,
+      "learning_rate": 2.841683366733467e-05,
+      "loss": 0.0441,
+      "step": 2166
+    },
+    {
+      "epoch": 8.668,
+      "grad_norm": 1.237499475479126,
+      "learning_rate": 2.8406813627254507e-05,
+      "loss": 0.0474,
+      "step": 2167
+    },
+    {
+      "epoch": 8.672,
+      "grad_norm": 1.2884820699691772,
+      "learning_rate": 2.839679358717435e-05,
+      "loss": 0.0489,
+      "step": 2168
+    },
+    {
+      "epoch": 8.676,
+      "grad_norm": 1.2853950262069702,
+      "learning_rate": 2.8386773547094193e-05,
+      "loss": 0.0514,
+      "step": 2169
+    },
+    {
+      "epoch": 8.68,
+      "grad_norm": 1.3021364212036133,
+      "learning_rate": 2.837675350701403e-05,
+      "loss": 0.0451,
+      "step": 2170
+    },
+    {
+      "epoch": 8.684,
+      "grad_norm": 1.1973280906677246,
+      "learning_rate": 2.836673346693387e-05,
+      "loss": 0.046,
+      "step": 2171
+    },
+    {
+      "epoch": 8.688,
+      "grad_norm": 0.8755884170532227,
+      "learning_rate": 2.835671342685371e-05,
+      "loss": 0.0345,
+      "step": 2172
+    },
+    {
+      "epoch": 8.692,
+      "grad_norm": 1.2425113916397095,
+      "learning_rate": 2.8346693386773548e-05,
+      "loss": 0.0407,
+      "step": 2173
+    },
+    {
+      "epoch": 8.696,
+      "grad_norm": 1.3100374937057495,
+      "learning_rate": 2.8336673346693386e-05,
+      "loss": 0.0459,
+      "step": 2174
+    },
+    {
+      "epoch": 8.7,
+      "grad_norm": 1.5196589231491089,
+      "learning_rate": 2.8326653306613227e-05,
+      "loss": 0.0492,
+      "step": 2175
+    },
+    {
+      "epoch": 8.704,
+      "grad_norm": 1.3412437438964844,
+      "learning_rate": 2.8316633266533065e-05,
+      "loss": 0.0522,
+      "step": 2176
+    },
+    {
+      "epoch": 8.708,
+      "grad_norm": 1.1661128997802734,
+      "learning_rate": 2.8306613226452906e-05,
+      "loss": 0.0533,
+      "step": 2177
+    },
+    {
+      "epoch": 8.712,
+      "grad_norm": 1.1783884763717651,
+      "learning_rate": 2.829659318637275e-05,
+      "loss": 0.0511,
+      "step": 2178
+    },
+    {
+      "epoch": 8.716,
+      "grad_norm": 1.162697434425354,
+      "learning_rate": 2.828657314629259e-05,
+      "loss": 0.0452,
+      "step": 2179
+    },
+    {
+      "epoch": 8.72,
+      "grad_norm": 1.5059949159622192,
+      "learning_rate": 2.8276553106212427e-05,
+      "loss": 0.0503,
+      "step": 2180
+    },
+    {
+      "epoch": 8.724,
+      "grad_norm": 1.33864426612854,
+      "learning_rate": 2.8266533066132268e-05,
+      "loss": 0.0519,
+      "step": 2181
+    },
+    {
+      "epoch": 8.728,
+      "grad_norm": 1.316199779510498,
+      "learning_rate": 2.8256513026052106e-05,
+      "loss": 0.0515,
+      "step": 2182
+    },
+    {
+      "epoch": 8.732,
+      "grad_norm": 1.2155145406723022,
+      "learning_rate": 2.8246492985971944e-05,
+      "loss": 0.0425,
+      "step": 2183
+    },
+    {
+      "epoch": 8.736,
+      "grad_norm": 1.4967896938323975,
+      "learning_rate": 2.8236472945891785e-05,
+      "loss": 0.0589,
+      "step": 2184
+    },
+    {
+      "epoch": 8.74,
+      "grad_norm": 1.188382863998413,
+      "learning_rate": 2.8226452905811623e-05,
+      "loss": 0.0446,
+      "step": 2185
+    },
+    {
+      "epoch": 8.744,
+      "grad_norm": 1.3887944221496582,
+      "learning_rate": 2.821643286573146e-05,
+      "loss": 0.0513,
+      "step": 2186
+    },
+    {
+      "epoch": 8.748,
+      "grad_norm": 1.4028888940811157,
+      "learning_rate": 2.8206412825651302e-05,
+      "loss": 0.0536,
+      "step": 2187
+    },
+    {
+      "epoch": 8.752,
+      "grad_norm": 1.4226263761520386,
+      "learning_rate": 2.8196392785571147e-05,
+      "loss": 0.0561,
+      "step": 2188
+    },
+    {
+      "epoch": 8.756,
+      "grad_norm": 1.1816654205322266,
+      "learning_rate": 2.8186372745490985e-05,
+      "loss": 0.039,
+      "step": 2189
+    },
+    {
+      "epoch": 8.76,
+      "grad_norm": 1.3053758144378662,
+      "learning_rate": 2.8176352705410823e-05,
+      "loss": 0.043,
+      "step": 2190
+    },
+    {
+      "epoch": 8.764,
+      "grad_norm": 1.021385669708252,
+      "learning_rate": 2.8166332665330664e-05,
+      "loss": 0.0425,
+      "step": 2191
+    },
+    {
+      "epoch": 8.768,
+      "grad_norm": 1.6020197868347168,
+      "learning_rate": 2.8156312625250502e-05,
+      "loss": 0.0488,
+      "step": 2192
+    },
+    {
+      "epoch": 8.772,
+      "grad_norm": 1.498152256011963,
+      "learning_rate": 2.8146292585170343e-05,
+      "loss": 0.0512,
+      "step": 2193
+    },
+    {
+      "epoch": 8.776,
+      "grad_norm": 1.3913947343826294,
+      "learning_rate": 2.813627254509018e-05,
+      "loss": 0.0509,
+      "step": 2194
+    },
+    {
+      "epoch": 8.78,
+      "grad_norm": 1.3088997602462769,
+      "learning_rate": 2.812625250501002e-05,
+      "loss": 0.0483,
+      "step": 2195
+    },
+    {
+      "epoch": 8.784,
+      "grad_norm": 0.9024881720542908,
+      "learning_rate": 2.811623246492986e-05,
+      "loss": 0.0297,
+      "step": 2196
+    },
+    {
+      "epoch": 8.788,
+      "grad_norm": 1.3125743865966797,
+      "learning_rate": 2.8106212424849698e-05,
+      "loss": 0.0552,
+      "step": 2197
+    },
+    {
+      "epoch": 8.792,
+      "grad_norm": 1.094606876373291,
+      "learning_rate": 2.8096192384769543e-05,
+      "loss": 0.0398,
+      "step": 2198
+    },
+    {
+      "epoch": 8.796,
+      "grad_norm": 1.1351479291915894,
+      "learning_rate": 2.808617234468938e-05,
+      "loss": 0.0518,
+      "step": 2199
+    },
+    {
+      "epoch": 8.8,
+      "grad_norm": 1.218656063079834,
+      "learning_rate": 2.8076152304609222e-05,
+      "loss": 0.0547,
+      "step": 2200
+    },
+    {
+      "epoch": 8.804,
+      "grad_norm": 1.1994880437850952,
+      "learning_rate": 2.806613226452906e-05,
+      "loss": 0.0405,
+      "step": 2201
+    },
+    {
+      "epoch": 8.808,
+      "grad_norm": 1.25562584400177,
+      "learning_rate": 2.8056112224448898e-05,
+      "loss": 0.0537,
+      "step": 2202
+    },
+    {
+      "epoch": 8.812,
+      "grad_norm": 1.5509350299835205,
+      "learning_rate": 2.804609218436874e-05,
+      "loss": 0.0624,
+      "step": 2203
+    },
+    {
+      "epoch": 8.816,
+      "grad_norm": 1.4089373350143433,
+      "learning_rate": 2.8036072144288577e-05,
+      "loss": 0.0627,
+      "step": 2204
+    },
+    {
+      "epoch": 8.82,
+      "grad_norm": 1.2322133779525757,
+      "learning_rate": 2.802605210420842e-05,
+      "loss": 0.0432,
+      "step": 2205
+    },
+    {
+      "epoch": 8.824,
+      "grad_norm": 1.5502153635025024,
+      "learning_rate": 2.8016032064128256e-05,
+      "loss": 0.0682,
+      "step": 2206
+    },
+    {
+      "epoch": 8.828,
+      "grad_norm": 1.2696598768234253,
+      "learning_rate": 2.8006012024048094e-05,
+      "loss": 0.046,
+      "step": 2207
+    },
+    {
+      "epoch": 8.832,
+      "grad_norm": 1.2160747051239014,
+      "learning_rate": 2.799599198396794e-05,
+      "loss": 0.0401,
+      "step": 2208
+    },
+    {
+      "epoch": 8.836,
+      "grad_norm": 1.38071608543396,
+      "learning_rate": 2.798597194388778e-05,
+      "loss": 0.0495,
+      "step": 2209
+    },
+    {
+      "epoch": 8.84,
+      "grad_norm": 1.6408637762069702,
+      "learning_rate": 2.7975951903807618e-05,
+      "loss": 0.0646,
+      "step": 2210
+    },
+    {
+      "epoch": 8.844,
+      "grad_norm": 1.340449333190918,
+      "learning_rate": 2.7965931863727456e-05,
+      "loss": 0.0475,
+      "step": 2211
+    },
+    {
+      "epoch": 8.848,
+      "grad_norm": 1.3540539741516113,
+      "learning_rate": 2.7955911823647297e-05,
+      "loss": 0.0526,
+      "step": 2212
+    },
+    {
+      "epoch": 8.852,
+      "grad_norm": 1.4355782270431519,
+      "learning_rate": 2.7945891783567135e-05,
+      "loss": 0.0478,
+      "step": 2213
+    },
+    {
+      "epoch": 8.856,
+      "grad_norm": 1.5981539487838745,
+      "learning_rate": 2.7935871743486973e-05,
+      "loss": 0.0546,
+      "step": 2214
+    },
+    {
+      "epoch": 8.86,
+      "grad_norm": 1.3910454511642456,
+      "learning_rate": 2.7925851703406814e-05,
+      "loss": 0.0513,
+      "step": 2215
+    },
+    {
+      "epoch": 8.864,
+      "grad_norm": 1.2546508312225342,
+      "learning_rate": 2.7915831663326652e-05,
+      "loss": 0.0506,
+      "step": 2216
+    },
+    {
+      "epoch": 8.868,
+      "grad_norm": 1.4212666749954224,
+      "learning_rate": 2.790581162324649e-05,
+      "loss": 0.05,
+      "step": 2217
+    },
+    {
+      "epoch": 8.872,
+      "grad_norm": 1.2646417617797852,
+      "learning_rate": 2.7895791583166335e-05,
+      "loss": 0.046,
+      "step": 2218
+    },
+    {
+      "epoch": 8.876,
+      "grad_norm": 1.1945741176605225,
+      "learning_rate": 2.7885771543086176e-05,
+      "loss": 0.0445,
+      "step": 2219
+    },
+    {
+      "epoch": 8.88,
+      "grad_norm": 1.351228952407837,
+      "learning_rate": 2.7875751503006014e-05,
+      "loss": 0.0574,
+      "step": 2220
+    },
+    {
+      "epoch": 8.884,
+      "grad_norm": 1.2459324598312378,
+      "learning_rate": 2.7865731462925855e-05,
+      "loss": 0.0479,
+      "step": 2221
+    },
+    {
+      "epoch": 8.888,
+      "grad_norm": 1.5551952123641968,
+      "learning_rate": 2.7855711422845693e-05,
+      "loss": 0.0703,
+      "step": 2222
+    },
+    {
+      "epoch": 8.892,
+      "grad_norm": 1.3713449239730835,
+      "learning_rate": 2.784569138276553e-05,
+      "loss": 0.049,
+      "step": 2223
+    },
+    {
+      "epoch": 8.896,
+      "grad_norm": 1.3457205295562744,
+      "learning_rate": 2.7835671342685372e-05,
+      "loss": 0.0417,
+      "step": 2224
+    },
+    {
+      "epoch": 8.9,
+      "grad_norm": 1.2651445865631104,
+      "learning_rate": 2.782565130260521e-05,
+      "loss": 0.0554,
+      "step": 2225
+    },
+    {
+      "epoch": 8.904,
+      "grad_norm": 1.3092762231826782,
+      "learning_rate": 2.7815631262525048e-05,
+      "loss": 0.05,
+      "step": 2226
+    },
+    {
+      "epoch": 8.908,
+      "grad_norm": 1.3287698030471802,
+      "learning_rate": 2.780561122244489e-05,
+      "loss": 0.0495,
+      "step": 2227
+    },
+    {
+      "epoch": 8.912,
+      "grad_norm": 1.1512635946273804,
+      "learning_rate": 2.7795591182364734e-05,
+      "loss": 0.0453,
+      "step": 2228
+    },
+    {
+      "epoch": 8.916,
+      "grad_norm": 1.3240880966186523,
+      "learning_rate": 2.7785571142284572e-05,
+      "loss": 0.0417,
+      "step": 2229
+    },
+    {
+      "epoch": 8.92,
+      "grad_norm": 1.21474289894104,
+      "learning_rate": 2.777555110220441e-05,
+      "loss": 0.047,
+      "step": 2230
+    },
+    {
+      "epoch": 8.924,
+      "grad_norm": 1.4137842655181885,
+      "learning_rate": 2.776553106212425e-05,
+      "loss": 0.0747,
+      "step": 2231
+    },
+    {
+      "epoch": 8.928,
+      "grad_norm": 1.2826186418533325,
+      "learning_rate": 2.775551102204409e-05,
+      "loss": 0.0541,
+      "step": 2232
+    },
+    {
+      "epoch": 8.932,
+      "grad_norm": 1.2921860218048096,
+      "learning_rate": 2.774549098196393e-05,
+      "loss": 0.0671,
+      "step": 2233
+    },
+    {
+      "epoch": 8.936,
+      "grad_norm": 1.2751491069793701,
+      "learning_rate": 2.7735470941883768e-05,
+      "loss": 0.0357,
+      "step": 2234
+    },
+    {
+      "epoch": 8.94,
+      "grad_norm": 1.2974969148635864,
+      "learning_rate": 2.7725450901803606e-05,
+      "loss": 0.05,
+      "step": 2235
+    },
+    {
+      "epoch": 8.943999999999999,
+      "grad_norm": 1.132696509361267,
+      "learning_rate": 2.7715430861723447e-05,
+      "loss": 0.0425,
+      "step": 2236
+    },
+    {
+      "epoch": 8.948,
+      "grad_norm": 0.6913595199584961,
+      "learning_rate": 2.7705410821643292e-05,
+      "loss": 0.0187,
+      "step": 2237
+    },
+    {
+      "epoch": 8.952,
+      "grad_norm": 1.1450154781341553,
+      "learning_rate": 2.769539078156313e-05,
+      "loss": 0.0421,
+      "step": 2238
+    },
+    {
+      "epoch": 8.956,
+      "grad_norm": 1.154909372329712,
+      "learning_rate": 2.7685370741482968e-05,
+      "loss": 0.0384,
+      "step": 2239
+    },
+    {
+      "epoch": 8.96,
+      "grad_norm": 1.3707146644592285,
+      "learning_rate": 2.767535070140281e-05,
+      "loss": 0.0517,
+      "step": 2240
+    },
+    {
+      "epoch": 8.964,
+      "grad_norm": 1.5236717462539673,
+      "learning_rate": 2.7665330661322647e-05,
+      "loss": 0.0684,
+      "step": 2241
+    },
+    {
+      "epoch": 8.968,
+      "grad_norm": 1.2843303680419922,
+      "learning_rate": 2.7655310621242485e-05,
+      "loss": 0.0427,
+      "step": 2242
+    },
+    {
+      "epoch": 8.972,
+      "grad_norm": 1.1353422403335571,
+      "learning_rate": 2.7645290581162326e-05,
+      "loss": 0.043,
+      "step": 2243
+    },
+    {
+      "epoch": 8.975999999999999,
+      "grad_norm": 1.2605583667755127,
+      "learning_rate": 2.7635270541082164e-05,
+      "loss": 0.0486,
+      "step": 2244
+    },
+    {
+      "epoch": 8.98,
+      "grad_norm": 1.3495458364486694,
+      "learning_rate": 2.7625250501002002e-05,
+      "loss": 0.0474,
+      "step": 2245
+    },
+    {
+      "epoch": 8.984,
+      "grad_norm": 1.3079636096954346,
+      "learning_rate": 2.7615230460921843e-05,
+      "loss": 0.0639,
+      "step": 2246
+    },
+    {
+      "epoch": 8.988,
+      "grad_norm": 1.171911358833313,
+      "learning_rate": 2.7605210420841688e-05,
+      "loss": 0.0417,
+      "step": 2247
+    },
+    {
+      "epoch": 8.992,
+      "grad_norm": 1.34932541847229,
+      "learning_rate": 2.7595190380761526e-05,
+      "loss": 0.0482,
+      "step": 2248
+    },
+    {
+      "epoch": 8.996,
+      "grad_norm": 1.653420329093933,
+      "learning_rate": 2.7585170340681367e-05,
+      "loss": 0.0602,
+      "step": 2249
+    },
+    {
+      "epoch": 9.0,
+      "grad_norm": 1.4358410835266113,
+      "learning_rate": 2.7575150300601205e-05,
+      "loss": 0.0582,
+      "step": 2250
+    },
+    {
+      "epoch": 9.004,
+      "grad_norm": 0.6799370050430298,
+      "learning_rate": 2.7565130260521043e-05,
+      "loss": 0.0219,
+      "step": 2251
+    },
+    {
+      "epoch": 9.008,
+      "grad_norm": 0.6610779762268066,
+      "learning_rate": 2.7555110220440884e-05,
+      "loss": 0.022,
+      "step": 2252
+    },
+    {
+      "epoch": 9.012,
+      "grad_norm": 0.9135136604309082,
+      "learning_rate": 2.7545090180360722e-05,
+      "loss": 0.0262,
+      "step": 2253
+    },
+    {
+      "epoch": 9.016,
+      "grad_norm": 0.8344765305519104,
+      "learning_rate": 2.753507014028056e-05,
+      "loss": 0.0254,
+      "step": 2254
+    },
+    {
+      "epoch": 9.02,
+      "grad_norm": 0.8494986891746521,
+      "learning_rate": 2.75250501002004e-05,
+      "loss": 0.0273,
+      "step": 2255
+    },
+    {
+      "epoch": 9.024,
+      "grad_norm": 0.9093011021614075,
+      "learning_rate": 2.751503006012024e-05,
+      "loss": 0.0281,
+      "step": 2256
+    },
+    {
+      "epoch": 9.028,
+      "grad_norm": 0.9227543473243713,
+      "learning_rate": 2.7505010020040084e-05,
+      "loss": 0.0294,
+      "step": 2257
+    },
+    {
+      "epoch": 9.032,
+      "grad_norm": 0.6513843536376953,
+      "learning_rate": 2.749498997995992e-05,
+      "loss": 0.0201,
+      "step": 2258
+    },
+    {
+      "epoch": 9.036,
+      "grad_norm": 0.8121753334999084,
+      "learning_rate": 2.7484969939879763e-05,
+      "loss": 0.0262,
+      "step": 2259
+    },
+    {
+      "epoch": 9.04,
+      "grad_norm": 1.054721713066101,
+      "learning_rate": 2.74749498997996e-05,
+      "loss": 0.0277,
+      "step": 2260
+    },
+    {
+      "epoch": 9.044,
+      "grad_norm": 0.8000840544700623,
+      "learning_rate": 2.7464929859719442e-05,
+      "loss": 0.0227,
+      "step": 2261
+    },
+    {
+      "epoch": 9.048,
+      "grad_norm": 0.8721778988838196,
+      "learning_rate": 2.745490981963928e-05,
+      "loss": 0.0236,
+      "step": 2262
+    },
+    {
+      "epoch": 9.052,
+      "grad_norm": 0.9745427370071411,
+      "learning_rate": 2.7444889779559118e-05,
+      "loss": 0.0236,
+      "step": 2263
+    },
+    {
+      "epoch": 9.056,
+      "grad_norm": 0.972322404384613,
+      "learning_rate": 2.743486973947896e-05,
+      "loss": 0.0278,
+      "step": 2264
+    },
+    {
+      "epoch": 9.06,
+      "grad_norm": 1.111842155456543,
+      "learning_rate": 2.7424849699398797e-05,
+      "loss": 0.0299,
+      "step": 2265
+    },
+    {
+      "epoch": 9.064,
+      "grad_norm": 0.7864890694618225,
+      "learning_rate": 2.7414829659318635e-05,
+      "loss": 0.0243,
+      "step": 2266
+    },
+    {
+      "epoch": 9.068,
+      "grad_norm": 1.2372549772262573,
+      "learning_rate": 2.740480961923848e-05,
+      "loss": 0.043,
+      "step": 2267
+    },
+    {
+      "epoch": 9.072,
+      "grad_norm": 0.7905459403991699,
+      "learning_rate": 2.739478957915832e-05,
+      "loss": 0.021,
+      "step": 2268
+    },
+    {
+      "epoch": 9.076,
+      "grad_norm": 1.100546956062317,
+      "learning_rate": 2.738476953907816e-05,
+      "loss": 0.0277,
+      "step": 2269
+    },
+    {
+      "epoch": 9.08,
+      "grad_norm": 0.8412137031555176,
+      "learning_rate": 2.7374749498997997e-05,
+      "loss": 0.025,
+      "step": 2270
+    },
+    {
+      "epoch": 9.084,
+      "grad_norm": 0.9313427805900574,
+      "learning_rate": 2.7364729458917838e-05,
+      "loss": 0.0231,
+      "step": 2271
+    },
+    {
+      "epoch": 9.088,
+      "grad_norm": 1.1528956890106201,
+      "learning_rate": 2.7354709418837676e-05,
+      "loss": 0.0338,
+      "step": 2272
+    },
+    {
+      "epoch": 9.092,
+      "grad_norm": 0.7774552702903748,
+      "learning_rate": 2.7344689378757514e-05,
+      "loss": 0.0247,
+      "step": 2273
+    },
+    {
+      "epoch": 9.096,
+      "grad_norm": 1.030387282371521,
+      "learning_rate": 2.7334669338677355e-05,
+      "loss": 0.0266,
+      "step": 2274
+    },
+    {
+      "epoch": 9.1,
+      "grad_norm": 0.8086778521537781,
+      "learning_rate": 2.7324649298597193e-05,
+      "loss": 0.0246,
+      "step": 2275
+    },
+    {
+      "epoch": 9.104,
+      "grad_norm": 0.8489255905151367,
+      "learning_rate": 2.7314629258517034e-05,
+      "loss": 0.0249,
+      "step": 2276
+    },
+    {
+      "epoch": 9.108,
+      "grad_norm": 0.7739505171775818,
+      "learning_rate": 2.730460921843688e-05,
+      "loss": 0.0257,
+      "step": 2277
+    },
+    {
+      "epoch": 9.112,
+      "grad_norm": 0.8986899852752686,
+      "learning_rate": 2.7294589178356717e-05,
+      "loss": 0.0258,
+      "step": 2278
+    },
+    {
+      "epoch": 9.116,
+      "grad_norm": 1.1886520385742188,
+      "learning_rate": 2.7284569138276555e-05,
+      "loss": 0.0349,
+      "step": 2279
+    },
+    {
+      "epoch": 9.12,
+      "grad_norm": 0.9789061546325684,
+      "learning_rate": 2.7274549098196396e-05,
+      "loss": 0.0277,
+      "step": 2280
+    },
+    {
+      "epoch": 9.124,
+      "grad_norm": 1.192771315574646,
+      "learning_rate": 2.7264529058116234e-05,
+      "loss": 0.0336,
+      "step": 2281
+    },
+    {
+      "epoch": 9.128,
+      "grad_norm": 1.048073649406433,
+      "learning_rate": 2.7254509018036072e-05,
+      "loss": 0.031,
+      "step": 2282
+    },
+    {
+      "epoch": 9.132,
+      "grad_norm": 1.163774013519287,
+      "learning_rate": 2.7244488977955913e-05,
+      "loss": 0.0309,
+      "step": 2283
+    },
+    {
+      "epoch": 9.136,
+      "grad_norm": 0.8994669914245605,
+      "learning_rate": 2.723446893787575e-05,
+      "loss": 0.0204,
+      "step": 2284
+    },
+    {
+      "epoch": 9.14,
+      "grad_norm": 1.123653769493103,
+      "learning_rate": 2.722444889779559e-05,
+      "loss": 0.0325,
+      "step": 2285
+    },
+    {
+      "epoch": 9.144,
+      "grad_norm": 0.9699426889419556,
+      "learning_rate": 2.7214428857715433e-05,
+      "loss": 0.0253,
+      "step": 2286
+    },
+    {
+      "epoch": 9.148,
+      "grad_norm": 0.8264506459236145,
+      "learning_rate": 2.7204408817635275e-05,
+      "loss": 0.0234,
+      "step": 2287
+    },
+    {
+      "epoch": 9.152,
+      "grad_norm": 1.0332759618759155,
+      "learning_rate": 2.7194388777555113e-05,
+      "loss": 0.0315,
+      "step": 2288
+    },
+    {
+      "epoch": 9.156,
+      "grad_norm": 1.0331981182098389,
+      "learning_rate": 2.7184368737474954e-05,
+      "loss": 0.0336,
+      "step": 2289
+    },
+    {
+      "epoch": 9.16,
+      "grad_norm": 2.753606081008911,
+      "learning_rate": 2.7174348697394792e-05,
+      "loss": 0.0322,
+      "step": 2290
+    },
+    {
+      "epoch": 9.164,
+      "grad_norm": 0.8838220834732056,
+      "learning_rate": 2.716432865731463e-05,
+      "loss": 0.0244,
+      "step": 2291
+    },
+    {
+      "epoch": 9.168,
+      "grad_norm": 0.9516180753707886,
+      "learning_rate": 2.715430861723447e-05,
+      "loss": 0.0325,
+      "step": 2292
+    },
+    {
+      "epoch": 9.172,
+      "grad_norm": 0.9610980153083801,
+      "learning_rate": 2.714428857715431e-05,
+      "loss": 0.0257,
+      "step": 2293
+    },
+    {
+      "epoch": 9.176,
+      "grad_norm": 1.0998841524124146,
+      "learning_rate": 2.7134268537074147e-05,
+      "loss": 0.0291,
+      "step": 2294
+    },
+    {
+      "epoch": 9.18,
+      "grad_norm": 1.0278939008712769,
+      "learning_rate": 2.7124248496993988e-05,
+      "loss": 0.0308,
+      "step": 2295
+    },
+    {
+      "epoch": 9.184,
+      "grad_norm": 1.050432562828064,
+      "learning_rate": 2.7114228456913833e-05,
+      "loss": 0.0362,
+      "step": 2296
+    },
+    {
+      "epoch": 9.188,
+      "grad_norm": 1.3083264827728271,
+      "learning_rate": 2.710420841683367e-05,
+      "loss": 0.0347,
+      "step": 2297
+    },
+    {
+      "epoch": 9.192,
+      "grad_norm": 1.0661804676055908,
+      "learning_rate": 2.709418837675351e-05,
+      "loss": 0.0317,
+      "step": 2298
+    },
+    {
+      "epoch": 9.196,
+      "grad_norm": 0.8684881925582886,
+      "learning_rate": 2.708416833667335e-05,
+      "loss": 0.028,
+      "step": 2299
+    },
+    {
+      "epoch": 9.2,
+      "grad_norm": 0.4513463079929352,
+      "learning_rate": 2.7074148296593188e-05,
+      "loss": 0.0072,
+      "step": 2300
+    },
+    {
+      "epoch": 9.204,
+      "grad_norm": 0.9160906076431274,
+      "learning_rate": 2.7064128256513026e-05,
+      "loss": 0.0268,
+      "step": 2301
+    },
+    {
+      "epoch": 9.208,
+      "grad_norm": 1.1022686958312988,
+      "learning_rate": 2.7054108216432867e-05,
+      "loss": 0.0261,
+      "step": 2302
+    },
+    {
+      "epoch": 9.212,
+      "grad_norm": 1.2009657621383667,
+      "learning_rate": 2.7044088176352705e-05,
+      "loss": 0.0388,
+      "step": 2303
+    },
+    {
+      "epoch": 9.216,
+      "grad_norm": 1.1833915710449219,
+      "learning_rate": 2.7034068136272546e-05,
+      "loss": 0.031,
+      "step": 2304
+    },
+    {
+      "epoch": 9.22,
+      "grad_norm": 1.0764391422271729,
+      "learning_rate": 2.7024048096192384e-05,
+      "loss": 0.0327,
+      "step": 2305
+    },
+    {
+      "epoch": 9.224,
+      "grad_norm": 0.9721907377243042,
+      "learning_rate": 2.701402805611223e-05,
+      "loss": 0.0326,
+      "step": 2306
+    },
+    {
+      "epoch": 9.228,
+      "grad_norm": 1.1473323106765747,
+      "learning_rate": 2.7004008016032067e-05,
+      "loss": 0.0345,
+      "step": 2307
+    },
+    {
+      "epoch": 9.232,
+      "grad_norm": 1.1332943439483643,
+      "learning_rate": 2.6993987975951908e-05,
+      "loss": 0.0313,
+      "step": 2308
+    },
+    {
+      "epoch": 9.236,
+      "grad_norm": 1.263730525970459,
+      "learning_rate": 2.6983967935871746e-05,
+      "loss": 0.0302,
+      "step": 2309
+    },
+    {
+      "epoch": 9.24,
+      "grad_norm": 1.0454695224761963,
+      "learning_rate": 2.6973947895791584e-05,
+      "loss": 0.028,
+      "step": 2310
+    },
+    {
+      "epoch": 9.244,
+      "grad_norm": 1.1046359539031982,
+      "learning_rate": 2.6963927855711425e-05,
+      "loss": 0.0343,
+      "step": 2311
+    },
+    {
+      "epoch": 9.248,
+      "grad_norm": 1.1158086061477661,
+      "learning_rate": 2.6953907815631263e-05,
+      "loss": 0.0298,
+      "step": 2312
+    },
+    {
+      "epoch": 9.252,
+      "grad_norm": 1.1569856405258179,
+      "learning_rate": 2.69438877755511e-05,
+      "loss": 0.0281,
+      "step": 2313
+    },
+    {
+      "epoch": 9.256,
+      "grad_norm": 0.6049909591674805,
+      "learning_rate": 2.6933867735470942e-05,
+      "loss": 0.0132,
+      "step": 2314
+    },
+    {
+      "epoch": 9.26,
+      "grad_norm": 1.2866177558898926,
+      "learning_rate": 2.692384769539078e-05,
+      "loss": 0.0361,
+      "step": 2315
+    },
+    {
+      "epoch": 9.264,
+      "grad_norm": 0.9961463809013367,
+      "learning_rate": 2.6913827655310625e-05,
+      "loss": 0.0267,
+      "step": 2316
+    },
+    {
+      "epoch": 9.268,
+      "grad_norm": 1.083455204963684,
+      "learning_rate": 2.6903807615230462e-05,
+      "loss": 0.0271,
+      "step": 2317
+    },
+    {
+      "epoch": 9.272,
+      "grad_norm": 0.8019450306892395,
+      "learning_rate": 2.6893787575150304e-05,
+      "loss": 0.0313,
+      "step": 2318
+    },
+    {
+      "epoch": 9.276,
+      "grad_norm": 1.0289325714111328,
+      "learning_rate": 2.688376753507014e-05,
+      "loss": 0.0338,
+      "step": 2319
+    },
+    {
+      "epoch": 9.28,
+      "grad_norm": 1.015210509300232,
+      "learning_rate": 2.6873747494989983e-05,
+      "loss": 0.028,
+      "step": 2320
+    },
+    {
+      "epoch": 9.284,
+      "grad_norm": 0.8635653257369995,
+      "learning_rate": 2.686372745490982e-05,
+      "loss": 0.0256,
+      "step": 2321
+    },
+    {
+      "epoch": 9.288,
+      "grad_norm": 0.9069684743881226,
+      "learning_rate": 2.685370741482966e-05,
+      "loss": 0.0324,
+      "step": 2322
+    },
+    {
+      "epoch": 9.292,
+      "grad_norm": 1.1683540344238281,
+      "learning_rate": 2.68436873747495e-05,
+      "loss": 0.0279,
+      "step": 2323
+    },
+    {
+      "epoch": 9.296,
+      "grad_norm": 1.084283471107483,
+      "learning_rate": 2.6833667334669338e-05,
+      "loss": 0.033,
+      "step": 2324
+    },
+    {
+      "epoch": 9.3,
+      "grad_norm": 1.0750457048416138,
+      "learning_rate": 2.6823647294589176e-05,
+      "loss": 0.026,
+      "step": 2325
+    },
+    {
+      "epoch": 9.304,
+      "grad_norm": 1.1331920623779297,
+      "learning_rate": 2.681362725450902e-05,
+      "loss": 0.0282,
+      "step": 2326
+    },
+    {
+      "epoch": 9.308,
+      "grad_norm": 0.7947516441345215,
+      "learning_rate": 2.6803607214428862e-05,
+      "loss": 0.0229,
+      "step": 2327
+    },
+    {
+      "epoch": 9.312,
+      "grad_norm": 1.2119550704956055,
+      "learning_rate": 2.67935871743487e-05,
+      "loss": 0.0352,
+      "step": 2328
+    },
+    {
+      "epoch": 9.316,
+      "grad_norm": 0.9553737640380859,
+      "learning_rate": 2.6783567134268537e-05,
+      "loss": 0.0291,
+      "step": 2329
+    },
+    {
+      "epoch": 9.32,
+      "grad_norm": 0.9401413202285767,
+      "learning_rate": 2.677354709418838e-05,
+      "loss": 0.0278,
+      "step": 2330
+    },
+    {
+      "epoch": 9.324,
+      "grad_norm": 0.9966760873794556,
+      "learning_rate": 2.6763527054108217e-05,
+      "loss": 0.0247,
+      "step": 2331
+    },
+    {
+      "epoch": 9.328,
+      "grad_norm": 1.0120066404342651,
+      "learning_rate": 2.6753507014028058e-05,
+      "loss": 0.0313,
+      "step": 2332
+    },
+    {
+      "epoch": 9.332,
+      "grad_norm": 1.1858829259872437,
+      "learning_rate": 2.6743486973947896e-05,
+      "loss": 0.0358,
+      "step": 2333
+    },
+    {
+      "epoch": 9.336,
+      "grad_norm": 1.082135796546936,
+      "learning_rate": 2.6733466933867734e-05,
+      "loss": 0.0311,
+      "step": 2334
+    },
+    {
+      "epoch": 9.34,
+      "grad_norm": 0.9590717554092407,
+      "learning_rate": 2.6723446893787575e-05,
+      "loss": 0.0316,
+      "step": 2335
+    },
+    {
+      "epoch": 9.344,
+      "grad_norm": 1.2413767576217651,
+      "learning_rate": 2.671342685370742e-05,
+      "loss": 0.0373,
+      "step": 2336
+    },
+    {
+      "epoch": 9.348,
+      "grad_norm": 1.1042641401290894,
+      "learning_rate": 2.6703406813627258e-05,
+      "loss": 0.0325,
+      "step": 2337
+    },
+    {
+      "epoch": 9.352,
+      "grad_norm": 0.944624125957489,
+      "learning_rate": 2.6693386773547095e-05,
+      "loss": 0.0273,
+      "step": 2338
+    },
+    {
+      "epoch": 9.356,
+      "grad_norm": 1.1185686588287354,
+      "learning_rate": 2.6683366733466937e-05,
+      "loss": 0.0377,
+      "step": 2339
+    },
+    {
+      "epoch": 9.36,
+      "grad_norm": 1.205579400062561,
+      "learning_rate": 2.6673346693386775e-05,
+      "loss": 0.0345,
+      "step": 2340
+    },
+    {
+      "epoch": 9.364,
+      "grad_norm": 1.2945799827575684,
+      "learning_rate": 2.6663326653306613e-05,
+      "loss": 0.0324,
+      "step": 2341
+    },
+    {
+      "epoch": 9.368,
+      "grad_norm": 0.9837092161178589,
+      "learning_rate": 2.6653306613226454e-05,
+      "loss": 0.0276,
+      "step": 2342
+    },
+    {
+      "epoch": 9.372,
+      "grad_norm": 1.1606407165527344,
+      "learning_rate": 2.6643286573146292e-05,
+      "loss": 0.0331,
+      "step": 2343
+    },
+    {
+      "epoch": 9.376,
+      "grad_norm": 1.0431427955627441,
+      "learning_rate": 2.663326653306613e-05,
+      "loss": 0.0295,
+      "step": 2344
+    },
+    {
+      "epoch": 9.38,
+      "grad_norm": 1.0889356136322021,
+      "learning_rate": 2.6623246492985974e-05,
+      "loss": 0.0291,
+      "step": 2345
+    },
+    {
+      "epoch": 9.384,
+      "grad_norm": 0.9746689200401306,
+      "learning_rate": 2.6613226452905816e-05,
+      "loss": 0.0281,
+      "step": 2346
+    },
+    {
+      "epoch": 9.388,
+      "grad_norm": 1.1333224773406982,
+      "learning_rate": 2.6603206412825653e-05,
+      "loss": 0.0306,
+      "step": 2347
+    },
+    {
+      "epoch": 9.392,
+      "grad_norm": 1.082342505455017,
+      "learning_rate": 2.6593186372745495e-05,
+      "loss": 0.032,
+      "step": 2348
+    },
+    {
+      "epoch": 9.396,
+      "grad_norm": 1.0573281049728394,
+      "learning_rate": 2.6583166332665333e-05,
+      "loss": 0.0261,
+      "step": 2349
+    },
+    {
+      "epoch": 9.4,
+      "grad_norm": 1.1344096660614014,
+      "learning_rate": 2.657314629258517e-05,
+      "loss": 0.03,
+      "step": 2350
+    },
+    {
+      "epoch": 9.404,
+      "grad_norm": 1.1539257764816284,
+      "learning_rate": 2.6563126252505012e-05,
+      "loss": 0.0325,
+      "step": 2351
+    },
+    {
+      "epoch": 9.408,
+      "grad_norm": 1.318395733833313,
+      "learning_rate": 2.655310621242485e-05,
+      "loss": 0.0383,
+      "step": 2352
+    },
+    {
+      "epoch": 9.412,
+      "grad_norm": 0.6710442900657654,
+      "learning_rate": 2.6543086172344688e-05,
+      "loss": 0.0202,
+      "step": 2353
+    },
+    {
+      "epoch": 9.416,
+      "grad_norm": 1.0864232778549194,
+      "learning_rate": 2.653306613226453e-05,
+      "loss": 0.0286,
+      "step": 2354
+    },
+    {
+      "epoch": 9.42,
+      "grad_norm": 1.046763300895691,
+      "learning_rate": 2.6523046092184374e-05,
+      "loss": 0.0292,
+      "step": 2355
+    },
+    {
+      "epoch": 9.424,
+      "grad_norm": 1.1092429161071777,
+      "learning_rate": 2.651302605210421e-05,
+      "loss": 0.0308,
+      "step": 2356
+    },
+    {
+      "epoch": 9.428,
+      "grad_norm": 1.0427887439727783,
+      "learning_rate": 2.650300601202405e-05,
+      "loss": 0.0282,
+      "step": 2357
+    },
+    {
+      "epoch": 9.432,
+      "grad_norm": 1.171705722808838,
+      "learning_rate": 2.649298597194389e-05,
+      "loss": 0.0326,
+      "step": 2358
+    },
+    {
+      "epoch": 9.436,
+      "grad_norm": 0.982445240020752,
+      "learning_rate": 2.648296593186373e-05,
+      "loss": 0.0275,
+      "step": 2359
+    },
+    {
+      "epoch": 9.44,
+      "grad_norm": 0.9876815676689148,
+      "learning_rate": 2.647294589178357e-05,
+      "loss": 0.0366,
+      "step": 2360
+    },
+    {
+      "epoch": 9.444,
+      "grad_norm": 0.9229361414909363,
+      "learning_rate": 2.6462925851703408e-05,
+      "loss": 0.0284,
+      "step": 2361
+    },
+    {
+      "epoch": 9.448,
+      "grad_norm": 0.9455786943435669,
+      "learning_rate": 2.6452905811623246e-05,
+      "loss": 0.0279,
+      "step": 2362
+    },
+    {
+      "epoch": 9.452,
+      "grad_norm": 0.5432707071304321,
+      "learning_rate": 2.6442885771543087e-05,
+      "loss": 0.0153,
+      "step": 2363
+    },
+    {
+      "epoch": 9.456,
+      "grad_norm": 1.1233261823654175,
+      "learning_rate": 2.6432865731462925e-05,
+      "loss": 0.0322,
+      "step": 2364
+    },
+    {
+      "epoch": 9.46,
+      "grad_norm": 1.0408973693847656,
+      "learning_rate": 2.642284569138277e-05,
+      "loss": 0.0334,
+      "step": 2365
+    },
+    {
+      "epoch": 9.464,
+      "grad_norm": 1.3444018363952637,
+      "learning_rate": 2.6412825651302607e-05,
+      "loss": 0.0354,
+      "step": 2366
+    },
+    {
+      "epoch": 9.468,
+      "grad_norm": 1.0790801048278809,
+      "learning_rate": 2.640280561122245e-05,
+      "loss": 0.0291,
+      "step": 2367
+    },
+    {
+      "epoch": 9.472,
+      "grad_norm": 1.0651098489761353,
+      "learning_rate": 2.6392785571142287e-05,
+      "loss": 0.0412,
+      "step": 2368
+    },
+    {
+      "epoch": 9.475999999999999,
+      "grad_norm": 1.0434693098068237,
+      "learning_rate": 2.6382765531062124e-05,
+      "loss": 0.0287,
+      "step": 2369
+    },
+    {
+      "epoch": 9.48,
+      "grad_norm": 1.1507903337478638,
+      "learning_rate": 2.6372745490981966e-05,
+      "loss": 0.032,
+      "step": 2370
+    },
+    {
+      "epoch": 9.484,
+      "grad_norm": 1.0209254026412964,
+      "learning_rate": 2.6362725450901804e-05,
+      "loss": 0.0312,
+      "step": 2371
+    },
+    {
+      "epoch": 9.488,
+      "grad_norm": 1.2612141370773315,
+      "learning_rate": 2.635270541082164e-05,
+      "loss": 0.038,
+      "step": 2372
+    },
+    {
+      "epoch": 9.492,
+      "grad_norm": 0.9201658368110657,
+      "learning_rate": 2.6342685370741483e-05,
+      "loss": 0.0302,
+      "step": 2373
+    },
+    {
+      "epoch": 9.496,
+      "grad_norm": 1.2548552751541138,
+      "learning_rate": 2.633266533066132e-05,
+      "loss": 0.0356,
+      "step": 2374
+    },
+    {
+      "epoch": 9.5,
+      "grad_norm": 0.9841686487197876,
+      "learning_rate": 2.6322645290581165e-05,
+      "loss": 0.0262,
+      "step": 2375
+    },
+    {
+      "epoch": 9.504,
+      "grad_norm": 1.228925108909607,
+      "learning_rate": 2.6312625250501007e-05,
+      "loss": 0.0354,
+      "step": 2376
+    },
+    {
+      "epoch": 9.508,
+      "grad_norm": 1.1918587684631348,
+      "learning_rate": 2.6302605210420845e-05,
+      "loss": 0.0334,
+      "step": 2377
+    },
+    {
+      "epoch": 9.512,
+      "grad_norm": 1.1744786500930786,
+      "learning_rate": 2.6292585170340682e-05,
+      "loss": 0.0391,
+      "step": 2378
+    },
+    {
+      "epoch": 9.516,
+      "grad_norm": 1.0026384592056274,
+      "learning_rate": 2.6282565130260524e-05,
+      "loss": 0.0361,
+      "step": 2379
+    },
+    {
+      "epoch": 9.52,
+      "grad_norm": 1.3792084455490112,
+      "learning_rate": 2.627254509018036e-05,
+      "loss": 0.0323,
+      "step": 2380
+    },
+    {
+      "epoch": 9.524000000000001,
+      "grad_norm": 0.7252622842788696,
+      "learning_rate": 2.62625250501002e-05,
+      "loss": 0.0176,
+      "step": 2381
+    },
+    {
+      "epoch": 9.528,
+      "grad_norm": 1.0033866167068481,
+      "learning_rate": 2.625250501002004e-05,
+      "loss": 0.0307,
+      "step": 2382
+    },
+    {
+      "epoch": 9.532,
+      "grad_norm": 1.1288398504257202,
+      "learning_rate": 2.624248496993988e-05,
+      "loss": 0.0363,
+      "step": 2383
+    },
+    {
+      "epoch": 9.536,
+      "grad_norm": 1.11295485496521,
+      "learning_rate": 2.6232464929859717e-05,
+      "loss": 0.0359,
+      "step": 2384
+    },
+    {
+      "epoch": 9.54,
+      "grad_norm": 0.9084358215332031,
+      "learning_rate": 2.622244488977956e-05,
+      "loss": 0.026,
+      "step": 2385
+    },
+    {
+      "epoch": 9.544,
+      "grad_norm": 1.091408371925354,
+      "learning_rate": 2.6212424849699403e-05,
+      "loss": 0.0328,
+      "step": 2386
+    },
+    {
+      "epoch": 9.548,
+      "grad_norm": 1.1012197732925415,
+      "learning_rate": 2.620240480961924e-05,
+      "loss": 0.0337,
+      "step": 2387
+    },
+    {
+      "epoch": 9.552,
+      "grad_norm": 1.2079898118972778,
+      "learning_rate": 2.619238476953908e-05,
+      "loss": 0.0318,
+      "step": 2388
+    },
+    {
+      "epoch": 9.556000000000001,
+      "grad_norm": 0.6078687310218811,
+      "learning_rate": 2.618236472945892e-05,
+      "loss": 0.0192,
+      "step": 2389
+    },
+    {
+      "epoch": 9.56,
+      "grad_norm": 0.9216514825820923,
+      "learning_rate": 2.6172344689378757e-05,
+      "loss": 0.0291,
+      "step": 2390
+    },
+    {
+      "epoch": 9.564,
+      "grad_norm": 0.9543266892433167,
+      "learning_rate": 2.61623246492986e-05,
+      "loss": 0.0283,
+      "step": 2391
+    },
+    {
+      "epoch": 9.568,
+      "grad_norm": 1.2358766794204712,
+      "learning_rate": 2.6152304609218437e-05,
+      "loss": 0.0338,
+      "step": 2392
+    },
+    {
+      "epoch": 9.572,
+      "grad_norm": 1.0046981573104858,
+      "learning_rate": 2.6142284569138275e-05,
+      "loss": 0.0234,
+      "step": 2393
+    },
+    {
+      "epoch": 9.576,
+      "grad_norm": 1.3283228874206543,
+      "learning_rate": 2.6132264529058116e-05,
+      "loss": 0.0488,
+      "step": 2394
+    },
+    {
+      "epoch": 9.58,
+      "grad_norm": 0.8663226962089539,
+      "learning_rate": 2.612224448897796e-05,
+      "loss": 0.0219,
+      "step": 2395
+    },
+    {
+      "epoch": 9.584,
+      "grad_norm": 1.2725272178649902,
+      "learning_rate": 2.61122244488978e-05,
+      "loss": 0.037,
+      "step": 2396
+    },
+    {
+      "epoch": 9.588,
+      "grad_norm": 1.0414756536483765,
+      "learning_rate": 2.6102204408817636e-05,
+      "loss": 0.0337,
+      "step": 2397
+    },
+    {
+      "epoch": 9.592,
+      "grad_norm": 1.0537278652191162,
+      "learning_rate": 2.6092184368737478e-05,
+      "loss": 0.0296,
+      "step": 2398
+    },
+    {
+      "epoch": 9.596,
+      "grad_norm": 1.2478158473968506,
+      "learning_rate": 2.6082164328657315e-05,
+      "loss": 0.0344,
+      "step": 2399
+    },
+    {
+      "epoch": 9.6,
+      "grad_norm": 1.1711362600326538,
+      "learning_rate": 2.6072144288577153e-05,
+      "loss": 0.036,
+      "step": 2400
+    },
+    {
+      "epoch": 9.604,
+      "grad_norm": 1.2118823528289795,
+      "learning_rate": 2.6062124248496995e-05,
+      "loss": 0.0284,
+      "step": 2401
+    },
+    {
+      "epoch": 9.608,
+      "grad_norm": 1.1261894702911377,
+      "learning_rate": 2.6052104208416833e-05,
+      "loss": 0.0339,
+      "step": 2402
+    },
+    {
+      "epoch": 9.612,
+      "grad_norm": 1.8169751167297363,
+      "learning_rate": 2.6042084168336674e-05,
+      "loss": 0.0326,
+      "step": 2403
+    },
+    {
+      "epoch": 9.616,
+      "grad_norm": 1.1546857357025146,
+      "learning_rate": 2.603206412825652e-05,
+      "loss": 0.0358,
+      "step": 2404
+    },
+    {
+      "epoch": 9.62,
+      "grad_norm": 1.1949758529663086,
+      "learning_rate": 2.6022044088176356e-05,
+      "loss": 0.0337,
+      "step": 2405
+    },
+    {
+      "epoch": 9.624,
+      "grad_norm": 0.9737666249275208,
+      "learning_rate": 2.6012024048096194e-05,
+      "loss": 0.0294,
+      "step": 2406
+    },
+    {
+      "epoch": 9.628,
+      "grad_norm": 1.1526328325271606,
+      "learning_rate": 2.6002004008016036e-05,
+      "loss": 0.0343,
+      "step": 2407
+    },
+    {
+      "epoch": 9.632,
+      "grad_norm": 1.266064167022705,
+      "learning_rate": 2.5991983967935873e-05,
+      "loss": 0.0309,
+      "step": 2408
+    },
+    {
+      "epoch": 9.636,
+      "grad_norm": 1.0405514240264893,
+      "learning_rate": 2.598196392785571e-05,
+      "loss": 0.0325,
+      "step": 2409
+    },
+    {
+      "epoch": 9.64,
+      "grad_norm": 1.1470423936843872,
+      "learning_rate": 2.5971943887775553e-05,
+      "loss": 0.0315,
+      "step": 2410
+    },
+    {
+      "epoch": 9.644,
+      "grad_norm": 1.1034044027328491,
+      "learning_rate": 2.596192384769539e-05,
+      "loss": 0.0325,
+      "step": 2411
+    },
+    {
+      "epoch": 9.648,
+      "grad_norm": 1.101983666419983,
+      "learning_rate": 2.595190380761523e-05,
+      "loss": 0.0281,
+      "step": 2412
+    },
+    {
+      "epoch": 9.652,
+      "grad_norm": 1.0074468851089478,
+      "learning_rate": 2.594188376753507e-05,
+      "loss": 0.0311,
+      "step": 2413
+    },
+    {
+      "epoch": 9.656,
+      "grad_norm": 1.2595995664596558,
+      "learning_rate": 2.5931863727454914e-05,
+      "loss": 0.0396,
+      "step": 2414
+    },
+    {
+      "epoch": 9.66,
+      "grad_norm": 1.195958137512207,
+      "learning_rate": 2.5921843687374752e-05,
+      "loss": 0.0329,
+      "step": 2415
+    },
+    {
+      "epoch": 9.664,
+      "grad_norm": 1.3391655683517456,
+      "learning_rate": 2.5911823647294594e-05,
+      "loss": 0.0402,
+      "step": 2416
+    },
+    {
+      "epoch": 9.668,
+      "grad_norm": 1.2114450931549072,
+      "learning_rate": 2.590180360721443e-05,
+      "loss": 0.0401,
+      "step": 2417
+    },
+    {
+      "epoch": 9.672,
+      "grad_norm": 1.2215373516082764,
+      "learning_rate": 2.589178356713427e-05,
+      "loss": 0.0325,
+      "step": 2418
+    },
+    {
+      "epoch": 9.676,
+      "grad_norm": 1.1048833131790161,
+      "learning_rate": 2.588176352705411e-05,
+      "loss": 0.0337,
+      "step": 2419
+    },
+    {
+      "epoch": 9.68,
+      "grad_norm": 0.823464035987854,
+      "learning_rate": 2.587174348697395e-05,
+      "loss": 0.0204,
+      "step": 2420
+    },
+    {
+      "epoch": 9.684,
+      "grad_norm": 1.2144451141357422,
+      "learning_rate": 2.5861723446893786e-05,
+      "loss": 0.0337,
+      "step": 2421
+    },
+    {
+      "epoch": 9.688,
+      "grad_norm": 0.928974986076355,
+      "learning_rate": 2.5851703406813628e-05,
+      "loss": 0.0276,
+      "step": 2422
+    },
+    {
+      "epoch": 9.692,
+      "grad_norm": 1.3975306749343872,
+      "learning_rate": 2.5841683366733466e-05,
+      "loss": 0.041,
+      "step": 2423
+    },
+    {
+      "epoch": 9.696,
+      "grad_norm": 1.0173219442367554,
+      "learning_rate": 2.583166332665331e-05,
+      "loss": 0.0308,
+      "step": 2424
+    },
+    {
+      "epoch": 9.7,
+      "grad_norm": 0.8563686013221741,
+      "learning_rate": 2.5821643286573148e-05,
+      "loss": 0.0308,
+      "step": 2425
+    },
+    {
+      "epoch": 9.704,
+      "grad_norm": 0.930245041847229,
+      "learning_rate": 2.581162324649299e-05,
+      "loss": 0.0298,
+      "step": 2426
+    },
+    {
+      "epoch": 9.708,
+      "grad_norm": 1.0278750658035278,
+      "learning_rate": 2.5801603206412827e-05,
+      "loss": 0.0298,
+      "step": 2427
+    },
+    {
+      "epoch": 9.712,
+      "grad_norm": 0.9554607272148132,
+      "learning_rate": 2.5791583166332665e-05,
+      "loss": 0.0321,
+      "step": 2428
+    },
+    {
+      "epoch": 9.716,
+      "grad_norm": 0.9845343828201294,
+      "learning_rate": 2.5781563126252507e-05,
+      "loss": 0.0365,
+      "step": 2429
+    },
+    {
+      "epoch": 9.72,
+      "grad_norm": 0.959926187992096,
+      "learning_rate": 2.5771543086172344e-05,
+      "loss": 0.0323,
+      "step": 2430
+    },
+    {
+      "epoch": 9.724,
+      "grad_norm": 1.3230634927749634,
+      "learning_rate": 2.5761523046092186e-05,
+      "loss": 0.0397,
+      "step": 2431
+    },
+    {
+      "epoch": 9.728,
+      "grad_norm": 1.1665371656417847,
+      "learning_rate": 2.5751503006012024e-05,
+      "loss": 0.0299,
+      "step": 2432
+    },
+    {
+      "epoch": 9.732,
+      "grad_norm": 1.1947264671325684,
+      "learning_rate": 2.574148296593186e-05,
+      "loss": 0.0344,
+      "step": 2433
+    },
+    {
+      "epoch": 9.736,
+      "grad_norm": 1.088255763053894,
+      "learning_rate": 2.5731462925851706e-05,
+      "loss": 0.0308,
+      "step": 2434
+    },
+    {
+      "epoch": 9.74,
+      "grad_norm": 1.1948275566101074,
+      "learning_rate": 2.5721442885771547e-05,
+      "loss": 0.0433,
+      "step": 2435
+    },
+    {
+      "epoch": 9.744,
+      "grad_norm": 0.96909099817276,
+      "learning_rate": 2.5711422845691385e-05,
+      "loss": 0.0311,
+      "step": 2436
+    },
+    {
+      "epoch": 9.748,
+      "grad_norm": 1.1078877449035645,
+      "learning_rate": 2.5701402805611223e-05,
+      "loss": 0.033,
+      "step": 2437
+    },
+    {
+      "epoch": 9.752,
+      "grad_norm": 1.0920991897583008,
+      "learning_rate": 2.5691382765531065e-05,
+      "loss": 0.0345,
+      "step": 2438
+    },
+    {
+      "epoch": 9.756,
+      "grad_norm": 1.28229558467865,
+      "learning_rate": 2.5681362725450902e-05,
+      "loss": 0.0386,
+      "step": 2439
+    },
+    {
+      "epoch": 9.76,
+      "grad_norm": 1.235954761505127,
+      "learning_rate": 2.567134268537074e-05,
+      "loss": 0.037,
+      "step": 2440
+    },
+    {
+      "epoch": 9.764,
+      "grad_norm": 1.089188575744629,
+      "learning_rate": 2.566132264529058e-05,
+      "loss": 0.0308,
+      "step": 2441
+    },
+    {
+      "epoch": 9.768,
+      "grad_norm": 1.021936058998108,
+      "learning_rate": 2.565130260521042e-05,
+      "loss": 0.0275,
+      "step": 2442
+    },
+    {
+      "epoch": 9.772,
+      "grad_norm": 1.178109049797058,
+      "learning_rate": 2.564128256513026e-05,
+      "loss": 0.0328,
+      "step": 2443
+    },
+    {
+      "epoch": 9.776,
+      "grad_norm": 1.2006083726882935,
+      "learning_rate": 2.5631262525050102e-05,
+      "loss": 0.0372,
+      "step": 2444
+    },
+    {
+      "epoch": 9.78,
+      "grad_norm": 1.1977381706237793,
+      "learning_rate": 2.5621242484969943e-05,
+      "loss": 0.0331,
+      "step": 2445
+    },
+    {
+      "epoch": 9.784,
+      "grad_norm": 1.240316390991211,
+      "learning_rate": 2.561122244488978e-05,
+      "loss": 0.0335,
+      "step": 2446
+    },
+    {
+      "epoch": 9.788,
+      "grad_norm": 1.1987797021865845,
+      "learning_rate": 2.5601202404809622e-05,
+      "loss": 0.0341,
+      "step": 2447
+    },
+    {
+      "epoch": 9.792,
+      "grad_norm": 1.0871502161026,
+      "learning_rate": 2.559118236472946e-05,
+      "loss": 0.0343,
+      "step": 2448
+    },
+    {
+      "epoch": 9.796,
+      "grad_norm": 0.9493705034255981,
+      "learning_rate": 2.5581162324649298e-05,
+      "loss": 0.0262,
+      "step": 2449
+    },
+    {
+      "epoch": 9.8,
+      "grad_norm": 1.1285532712936401,
+      "learning_rate": 2.557114228456914e-05,
+      "loss": 0.0334,
+      "step": 2450
+    },
+    {
+      "epoch": 9.804,
+      "grad_norm": 0.9567793011665344,
+      "learning_rate": 2.5561122244488977e-05,
+      "loss": 0.0291,
+      "step": 2451
+    },
+    {
+      "epoch": 9.808,
+      "grad_norm": 1.1705495119094849,
+      "learning_rate": 2.5551102204408815e-05,
+      "loss": 0.0334,
+      "step": 2452
+    },
+    {
+      "epoch": 9.812,
+      "grad_norm": 1.1043469905853271,
+      "learning_rate": 2.554108216432866e-05,
+      "loss": 0.0329,
+      "step": 2453
+    },
+    {
+      "epoch": 9.816,
+      "grad_norm": 1.5261973142623901,
+      "learning_rate": 2.55310621242485e-05,
+      "loss": 0.04,
+      "step": 2454
+    },
+    {
+      "epoch": 9.82,
+      "grad_norm": 1.164994716644287,
+      "learning_rate": 2.552104208416834e-05,
+      "loss": 0.0349,
+      "step": 2455
+    },
+    {
+      "epoch": 9.824,
+      "grad_norm": 0.960659384727478,
+      "learning_rate": 2.5511022044088177e-05,
+      "loss": 0.0251,
+      "step": 2456
+    },
+    {
+      "epoch": 9.828,
+      "grad_norm": 1.2397003173828125,
+      "learning_rate": 2.550100200400802e-05,
+      "loss": 0.0379,
+      "step": 2457
+    },
+    {
+      "epoch": 9.832,
+      "grad_norm": 1.0921655893325806,
+      "learning_rate": 2.5490981963927856e-05,
+      "loss": 0.0377,
+      "step": 2458
+    },
+    {
+      "epoch": 9.836,
+      "grad_norm": 1.165229320526123,
+      "learning_rate": 2.5480961923847698e-05,
+      "loss": 0.032,
+      "step": 2459
+    },
+    {
+      "epoch": 9.84,
+      "grad_norm": 1.2369738817214966,
+      "learning_rate": 2.5470941883767535e-05,
+      "loss": 0.0372,
+      "step": 2460
+    },
+    {
+      "epoch": 9.844,
+      "grad_norm": 1.1303399801254272,
+      "learning_rate": 2.5460921843687373e-05,
+      "loss": 0.031,
+      "step": 2461
+    },
+    {
+      "epoch": 9.848,
+      "grad_norm": 1.3156812191009521,
+      "learning_rate": 2.5450901803607215e-05,
+      "loss": 0.0334,
+      "step": 2462
+    },
+    {
+      "epoch": 9.852,
+      "grad_norm": 1.2975099086761475,
+      "learning_rate": 2.544088176352706e-05,
+      "loss": 0.0342,
+      "step": 2463
+    },
+    {
+      "epoch": 9.856,
+      "grad_norm": 1.2751260995864868,
+      "learning_rate": 2.5430861723446897e-05,
+      "loss": 0.0391,
+      "step": 2464
+    },
+    {
+      "epoch": 9.86,
+      "grad_norm": 1.1819888353347778,
+      "learning_rate": 2.5420841683366735e-05,
+      "loss": 0.0326,
+      "step": 2465
+    },
+    {
+      "epoch": 9.864,
+      "grad_norm": 1.6855754852294922,
+      "learning_rate": 2.5410821643286576e-05,
+      "loss": 0.0476,
+      "step": 2466
+    },
+    {
+      "epoch": 9.868,
+      "grad_norm": 1.1449778079986572,
+      "learning_rate": 2.5400801603206414e-05,
+      "loss": 0.0332,
+      "step": 2467
+    },
+    {
+      "epoch": 9.872,
+      "grad_norm": 1.15505850315094,
+      "learning_rate": 2.5390781563126252e-05,
+      "loss": 0.0334,
+      "step": 2468
+    },
+    {
+      "epoch": 9.876,
+      "grad_norm": 1.2710686922073364,
+      "learning_rate": 2.5380761523046093e-05,
+      "loss": 0.0317,
+      "step": 2469
+    },
+    {
+      "epoch": 9.88,
+      "grad_norm": 1.2653512954711914,
+      "learning_rate": 2.537074148296593e-05,
+      "loss": 0.0387,
+      "step": 2470
+    },
+    {
+      "epoch": 9.884,
+      "grad_norm": 1.1764731407165527,
+      "learning_rate": 2.536072144288577e-05,
+      "loss": 0.0375,
+      "step": 2471
+    },
+    {
+      "epoch": 9.888,
+      "grad_norm": 0.7444102764129639,
+      "learning_rate": 2.535070140280561e-05,
+      "loss": 0.0182,
+      "step": 2472
+    },
+    {
+      "epoch": 9.892,
+      "grad_norm": 1.2663006782531738,
+      "learning_rate": 2.5340681362725455e-05,
+      "loss": 0.034,
+      "step": 2473
+    },
+    {
+      "epoch": 9.896,
+      "grad_norm": 1.14084792137146,
+      "learning_rate": 2.5330661322645293e-05,
+      "loss": 0.0365,
+      "step": 2474
+    },
+    {
+      "epoch": 9.9,
+      "grad_norm": 1.0226420164108276,
+      "learning_rate": 2.5320641282565134e-05,
+      "loss": 0.0301,
+      "step": 2475
+    },
+    {
+      "epoch": 9.904,
+      "grad_norm": 1.0205984115600586,
+      "learning_rate": 2.5310621242484972e-05,
+      "loss": 0.0305,
+      "step": 2476
+    },
+    {
+      "epoch": 9.908,
+      "grad_norm": 1.1715893745422363,
+      "learning_rate": 2.530060120240481e-05,
+      "loss": 0.0322,
+      "step": 2477
+    },
+    {
+      "epoch": 9.912,
+      "grad_norm": 1.4050195217132568,
+      "learning_rate": 2.529058116232465e-05,
+      "loss": 0.0434,
+      "step": 2478
+    },
+    {
+      "epoch": 9.916,
+      "grad_norm": 1.1078695058822632,
+      "learning_rate": 2.528056112224449e-05,
+      "loss": 0.0311,
+      "step": 2479
+    },
+    {
+      "epoch": 9.92,
+      "grad_norm": 1.0815905332565308,
+      "learning_rate": 2.5270541082164327e-05,
+      "loss": 0.0372,
+      "step": 2480
+    },
+    {
+      "epoch": 9.924,
+      "grad_norm": 1.1205084323883057,
+      "learning_rate": 2.526052104208417e-05,
+      "loss": 0.0309,
+      "step": 2481
+    },
+    {
+      "epoch": 9.928,
+      "grad_norm": 1.123581051826477,
+      "learning_rate": 2.5250501002004006e-05,
+      "loss": 0.0332,
+      "step": 2482
+    },
+    {
+      "epoch": 9.932,
+      "grad_norm": 1.1913464069366455,
+      "learning_rate": 2.524048096192385e-05,
+      "loss": 0.0344,
+      "step": 2483
+    },
+    {
+      "epoch": 9.936,
+      "grad_norm": 1.033087134361267,
+      "learning_rate": 2.523046092184369e-05,
+      "loss": 0.0326,
+      "step": 2484
+    },
+    {
+      "epoch": 9.94,
+      "grad_norm": 1.0734940767288208,
+      "learning_rate": 2.522044088176353e-05,
+      "loss": 0.0364,
+      "step": 2485
+    },
+    {
+      "epoch": 9.943999999999999,
+      "grad_norm": 1.0637986660003662,
+      "learning_rate": 2.5210420841683368e-05,
+      "loss": 0.038,
+      "step": 2486
+    },
+    {
+      "epoch": 9.948,
+      "grad_norm": 1.1711645126342773,
+      "learning_rate": 2.520040080160321e-05,
+      "loss": 0.0343,
+      "step": 2487
+    },
+    {
+      "epoch": 9.952,
+      "grad_norm": 1.1081736087799072,
+      "learning_rate": 2.5190380761523047e-05,
+      "loss": 0.032,
+      "step": 2488
+    },
+    {
+      "epoch": 9.956,
+      "grad_norm": 1.0930702686309814,
+      "learning_rate": 2.5180360721442885e-05,
+      "loss": 0.0389,
+      "step": 2489
+    },
+    {
+      "epoch": 9.96,
+      "grad_norm": 1.1130561828613281,
+      "learning_rate": 2.5170340681362726e-05,
+      "loss": 0.0332,
+      "step": 2490
+    },
+    {
+      "epoch": 9.964,
+      "grad_norm": 1.0953967571258545,
+      "learning_rate": 2.5160320641282564e-05,
+      "loss": 0.0427,
+      "step": 2491
+    },
+    {
+      "epoch": 9.968,
+      "grad_norm": 1.0438133478164673,
+      "learning_rate": 2.5150300601202402e-05,
+      "loss": 0.0337,
+      "step": 2492
+    },
+    {
+      "epoch": 9.972,
+      "grad_norm": 1.1139825582504272,
+      "learning_rate": 2.5140280561122247e-05,
+      "loss": 0.0326,
+      "step": 2493
+    },
+    {
+      "epoch": 9.975999999999999,
+      "grad_norm": 1.1483579874038696,
+      "learning_rate": 2.5130260521042088e-05,
+      "loss": 0.0321,
+      "step": 2494
+    },
+    {
+      "epoch": 9.98,
+      "grad_norm": 0.9585171937942505,
+      "learning_rate": 2.5120240480961926e-05,
+      "loss": 0.0264,
+      "step": 2495
+    },
+    {
+      "epoch": 9.984,
+      "grad_norm": 1.2960076332092285,
+      "learning_rate": 2.5110220440881764e-05,
+      "loss": 0.0475,
+      "step": 2496
+    },
+    {
+      "epoch": 9.988,
+      "grad_norm": 1.278764009475708,
+      "learning_rate": 2.5100200400801605e-05,
+      "loss": 0.0376,
+      "step": 2497
+    },
+    {
+      "epoch": 9.992,
+      "grad_norm": 1.109755039215088,
+      "learning_rate": 2.5090180360721443e-05,
+      "loss": 0.0304,
+      "step": 2498
+    },
+    {
+      "epoch": 9.996,
+      "grad_norm": 1.2089651823043823,
+      "learning_rate": 2.508016032064128e-05,
+      "loss": 0.0345,
+      "step": 2499
+    },
+    {
+      "epoch": 10.0,
+      "grad_norm": 1.3059208393096924,
+      "learning_rate": 2.5070140280561122e-05,
+      "loss": 0.044,
+      "step": 2500
+    },
+    {
+      "epoch": 10.004,
+      "grad_norm": 0.838786780834198,
+      "learning_rate": 2.506012024048096e-05,
+      "loss": 0.0223,
+      "step": 2501
+    },
+    {
+      "epoch": 10.008,
+      "grad_norm": 0.9346208572387695,
+      "learning_rate": 2.50501002004008e-05,
+      "loss": 0.0308,
+      "step": 2502
+    },
+    {
+      "epoch": 10.012,
+      "grad_norm": 0.627692699432373,
+      "learning_rate": 2.5040080160320646e-05,
+      "loss": 0.0171,
+      "step": 2503
+    },
+    {
+      "epoch": 10.016,
+      "grad_norm": 0.789028525352478,
+      "learning_rate": 2.5030060120240484e-05,
+      "loss": 0.0211,
+      "step": 2504
+    },
+    {
+      "epoch": 10.02,
+      "grad_norm": 0.8333077430725098,
+      "learning_rate": 2.5020040080160322e-05,
+      "loss": 0.0253,
+      "step": 2505
+    },
+    {
+      "epoch": 10.024,
+      "grad_norm": 0.7909857630729675,
+      "learning_rate": 2.5010020040080163e-05,
+      "loss": 0.0231,
+      "step": 2506
+    },
+    {
+      "epoch": 10.028,
+      "grad_norm": 0.9159106016159058,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0219,
+      "step": 2507
+    },
+    {
+      "epoch": 10.032,
+      "grad_norm": 0.5722668766975403,
+      "learning_rate": 2.498997995991984e-05,
+      "loss": 0.0185,
+      "step": 2508
+    },
+    {
+      "epoch": 10.036,
+      "grad_norm": 0.8425304889678955,
+      "learning_rate": 2.497995991983968e-05,
+      "loss": 0.0223,
+      "step": 2509
+    },
+    {
+      "epoch": 10.04,
+      "grad_norm": 0.9075009226799011,
+      "learning_rate": 2.496993987975952e-05,
+      "loss": 0.0228,
+      "step": 2510
+    },
+    {
+      "epoch": 10.044,
+      "grad_norm": 0.7858025431632996,
+      "learning_rate": 2.495991983967936e-05,
+      "loss": 0.0196,
+      "step": 2511
+    },
+    {
+      "epoch": 10.048,
+      "grad_norm": 0.6345151662826538,
+      "learning_rate": 2.49498997995992e-05,
+      "loss": 0.0161,
+      "step": 2512
+    },
+    {
+      "epoch": 10.052,
+      "grad_norm": 0.6853556036949158,
+      "learning_rate": 2.493987975951904e-05,
+      "loss": 0.0174,
+      "step": 2513
+    },
+    {
+      "epoch": 10.056,
+      "grad_norm": 0.7381585240364075,
+      "learning_rate": 2.4929859719438877e-05,
+      "loss": 0.0182,
+      "step": 2514
+    },
+    {
+      "epoch": 10.06,
+      "grad_norm": 1.0127655267715454,
+      "learning_rate": 2.491983967935872e-05,
+      "loss": 0.0226,
+      "step": 2515
+    },
+    {
+      "epoch": 10.064,
+      "grad_norm": 1.03529691696167,
+      "learning_rate": 2.490981963927856e-05,
+      "loss": 0.0265,
+      "step": 2516
+    },
+    {
+      "epoch": 10.068,
+      "grad_norm": 1.1951006650924683,
+      "learning_rate": 2.4899799599198397e-05,
+      "loss": 0.0231,
+      "step": 2517
+    },
+    {
+      "epoch": 10.072,
+      "grad_norm": 0.7003726959228516,
+      "learning_rate": 2.488977955911824e-05,
+      "loss": 0.0186,
+      "step": 2518
+    },
+    {
+      "epoch": 10.076,
+      "grad_norm": 0.7388056516647339,
+      "learning_rate": 2.4879759519038076e-05,
+      "loss": 0.0185,
+      "step": 2519
+    },
+    {
+      "epoch": 10.08,
+      "grad_norm": 0.766671895980835,
+      "learning_rate": 2.4869739478957918e-05,
+      "loss": 0.0196,
+      "step": 2520
+    },
+    {
+      "epoch": 10.084,
+      "grad_norm": 0.6915393471717834,
+      "learning_rate": 2.4859719438877755e-05,
+      "loss": 0.0161,
+      "step": 2521
+    },
+    {
+      "epoch": 10.088,
+      "grad_norm": 0.831326961517334,
+      "learning_rate": 2.4849699398797597e-05,
+      "loss": 0.0222,
+      "step": 2522
+    },
+    {
+      "epoch": 10.092,
+      "grad_norm": 0.8071035146713257,
+      "learning_rate": 2.4839679358717435e-05,
+      "loss": 0.0198,
+      "step": 2523
+    },
+    {
+      "epoch": 10.096,
+      "grad_norm": 0.8890426754951477,
+      "learning_rate": 2.4829659318637276e-05,
+      "loss": 0.0222,
+      "step": 2524
+    },
+    {
+      "epoch": 10.1,
+      "grad_norm": 0.8120467066764832,
+      "learning_rate": 2.4819639278557117e-05,
+      "loss": 0.0193,
+      "step": 2525
+    },
+    {
+      "epoch": 10.104,
+      "grad_norm": 0.7911978960037231,
+      "learning_rate": 2.4809619238476955e-05,
+      "loss": 0.0228,
+      "step": 2526
+    },
+    {
+      "epoch": 10.108,
+      "grad_norm": 0.82595294713974,
+      "learning_rate": 2.4799599198396793e-05,
+      "loss": 0.0213,
+      "step": 2527
+    },
+    {
+      "epoch": 10.112,
+      "grad_norm": 0.40247252583503723,
+      "learning_rate": 2.4789579158316634e-05,
+      "loss": 0.0098,
+      "step": 2528
+    },
+    {
+      "epoch": 10.116,
+      "grad_norm": 1.0697221755981445,
+      "learning_rate": 2.4779559118236472e-05,
+      "loss": 0.0252,
+      "step": 2529
+    },
+    {
+      "epoch": 10.12,
+      "grad_norm": 0.8177196979522705,
+      "learning_rate": 2.4769539078156313e-05,
+      "loss": 0.0214,
+      "step": 2530
+    },
+    {
+      "epoch": 10.124,
+      "grad_norm": 0.6448612213134766,
+      "learning_rate": 2.4759519038076155e-05,
+      "loss": 0.0163,
+      "step": 2531
+    },
+    {
+      "epoch": 10.128,
+      "grad_norm": 0.7582941055297852,
+      "learning_rate": 2.4749498997995993e-05,
+      "loss": 0.0191,
+      "step": 2532
+    },
+    {
+      "epoch": 10.132,
+      "grad_norm": 0.6828077435493469,
+      "learning_rate": 2.473947895791583e-05,
+      "loss": 0.0174,
+      "step": 2533
+    },
+    {
+      "epoch": 10.136,
+      "grad_norm": 0.7914385199546814,
+      "learning_rate": 2.4729458917835672e-05,
+      "loss": 0.0198,
+      "step": 2534
+    },
+    {
+      "epoch": 10.14,
+      "grad_norm": 0.6168143153190613,
+      "learning_rate": 2.4719438877755513e-05,
+      "loss": 0.0171,
+      "step": 2535
+    },
+    {
+      "epoch": 10.144,
+      "grad_norm": 0.7246816158294678,
+      "learning_rate": 2.470941883767535e-05,
+      "loss": 0.0207,
+      "step": 2536
+    },
+    {
+      "epoch": 10.148,
+      "grad_norm": 1.0454907417297363,
+      "learning_rate": 2.4699398797595192e-05,
+      "loss": 0.025,
+      "step": 2537
+    },
+    {
+      "epoch": 10.152,
+      "grad_norm": 0.9220485091209412,
+      "learning_rate": 2.468937875751503e-05,
+      "loss": 0.0234,
+      "step": 2538
+    },
+    {
+      "epoch": 10.156,
+      "grad_norm": 0.8142296671867371,
+      "learning_rate": 2.467935871743487e-05,
+      "loss": 0.0197,
+      "step": 2539
+    },
+    {
+      "epoch": 10.16,
+      "grad_norm": 0.7292171716690063,
+      "learning_rate": 2.4669338677354713e-05,
+      "loss": 0.0197,
+      "step": 2540
+    },
+    {
+      "epoch": 10.164,
+      "grad_norm": 0.8474435210227966,
+      "learning_rate": 2.465931863727455e-05,
+      "loss": 0.0199,
+      "step": 2541
+    },
+    {
+      "epoch": 10.168,
+      "grad_norm": 0.677284836769104,
+      "learning_rate": 2.464929859719439e-05,
+      "loss": 0.0161,
+      "step": 2542
+    },
+    {
+      "epoch": 10.172,
+      "grad_norm": 0.8956500887870789,
+      "learning_rate": 2.463927855711423e-05,
+      "loss": 0.0233,
+      "step": 2543
+    },
+    {
+      "epoch": 10.176,
+      "grad_norm": 1.009035348892212,
+      "learning_rate": 2.462925851703407e-05,
+      "loss": 0.0243,
+      "step": 2544
+    },
+    {
+      "epoch": 10.18,
+      "grad_norm": 0.867087721824646,
+      "learning_rate": 2.461923847695391e-05,
+      "loss": 0.0201,
+      "step": 2545
+    },
+    {
+      "epoch": 10.184,
+      "grad_norm": 1.1653364896774292,
+      "learning_rate": 2.460921843687375e-05,
+      "loss": 0.0296,
+      "step": 2546
+    },
+    {
+      "epoch": 10.188,
+      "grad_norm": 0.7345436215400696,
+      "learning_rate": 2.4599198396793588e-05,
+      "loss": 0.0194,
+      "step": 2547
+    },
+    {
+      "epoch": 10.192,
+      "grad_norm": 0.9385821223258972,
+      "learning_rate": 2.4589178356713426e-05,
+      "loss": 0.0223,
+      "step": 2548
+    },
+    {
+      "epoch": 10.196,
+      "grad_norm": 0.6956146359443665,
+      "learning_rate": 2.4579158316633267e-05,
+      "loss": 0.0169,
+      "step": 2549
+    },
+    {
+      "epoch": 10.2,
+      "grad_norm": 0.7195176482200623,
+      "learning_rate": 2.456913827655311e-05,
+      "loss": 0.0189,
+      "step": 2550
+    },
+    {
+      "epoch": 10.204,
+      "grad_norm": 1.0395700931549072,
+      "learning_rate": 2.4559118236472946e-05,
+      "loss": 0.0214,
+      "step": 2551
+    },
+    {
+      "epoch": 10.208,
+      "grad_norm": 1.0134451389312744,
+      "learning_rate": 2.4549098196392788e-05,
+      "loss": 0.0218,
+      "step": 2552
+    },
+    {
+      "epoch": 10.212,
+      "grad_norm": 1.1786450147628784,
+      "learning_rate": 2.4539078156312626e-05,
+      "loss": 0.0177,
+      "step": 2553
+    },
+    {
+      "epoch": 10.216,
+      "grad_norm": 0.8216089606285095,
+      "learning_rate": 2.4529058116232467e-05,
+      "loss": 0.0197,
+      "step": 2554
+    },
+    {
+      "epoch": 10.22,
+      "grad_norm": 0.5676589608192444,
+      "learning_rate": 2.4519038076152305e-05,
+      "loss": 0.0171,
+      "step": 2555
+    },
+    {
+      "epoch": 10.224,
+      "grad_norm": 0.8638578653335571,
+      "learning_rate": 2.4509018036072146e-05,
+      "loss": 0.0225,
+      "step": 2556
+    },
+    {
+      "epoch": 10.228,
+      "grad_norm": 0.9760416746139526,
+      "learning_rate": 2.4498997995991984e-05,
+      "loss": 0.0222,
+      "step": 2557
+    },
+    {
+      "epoch": 10.232,
+      "grad_norm": 0.8012329339981079,
+      "learning_rate": 2.4488977955911825e-05,
+      "loss": 0.0164,
+      "step": 2558
+    },
+    {
+      "epoch": 10.236,
+      "grad_norm": 0.6655237078666687,
+      "learning_rate": 2.4478957915831667e-05,
+      "loss": 0.0189,
+      "step": 2559
+    },
+    {
+      "epoch": 10.24,
+      "grad_norm": 1.0602320432662964,
+      "learning_rate": 2.4468937875751504e-05,
+      "loss": 0.0271,
+      "step": 2560
+    },
+    {
+      "epoch": 10.244,
+      "grad_norm": 0.8225724697113037,
+      "learning_rate": 2.4458917835671342e-05,
+      "loss": 0.0207,
+      "step": 2561
+    },
+    {
+      "epoch": 10.248,
+      "grad_norm": 0.783803403377533,
+      "learning_rate": 2.4448897795591184e-05,
+      "loss": 0.0186,
+      "step": 2562
+    },
+    {
+      "epoch": 10.252,
+      "grad_norm": 0.704971969127655,
+      "learning_rate": 2.443887775551102e-05,
+      "loss": 0.0179,
+      "step": 2563
+    },
+    {
+      "epoch": 10.256,
+      "grad_norm": 0.8094855546951294,
+      "learning_rate": 2.4428857715430863e-05,
+      "loss": 0.0209,
+      "step": 2564
+    },
+    {
+      "epoch": 10.26,
+      "grad_norm": 0.8836872577667236,
+      "learning_rate": 2.4418837675350704e-05,
+      "loss": 0.0192,
+      "step": 2565
+    },
+    {
+      "epoch": 10.264,
+      "grad_norm": 0.901753842830658,
+      "learning_rate": 2.4408817635270542e-05,
+      "loss": 0.0241,
+      "step": 2566
+    },
+    {
+      "epoch": 10.268,
+      "grad_norm": 0.7555407881736755,
+      "learning_rate": 2.439879759519038e-05,
+      "loss": 0.0191,
+      "step": 2567
+    },
+    {
+      "epoch": 10.272,
+      "grad_norm": 0.8224442005157471,
+      "learning_rate": 2.438877755511022e-05,
+      "loss": 0.0209,
+      "step": 2568
+    },
+    {
+      "epoch": 10.276,
+      "grad_norm": 0.9603815078735352,
+      "learning_rate": 2.4378757515030062e-05,
+      "loss": 0.0205,
+      "step": 2569
+    },
+    {
+      "epoch": 10.28,
+      "grad_norm": 0.7587599158287048,
+      "learning_rate": 2.43687374749499e-05,
+      "loss": 0.0165,
+      "step": 2570
+    },
+    {
+      "epoch": 10.284,
+      "grad_norm": 0.8181880712509155,
+      "learning_rate": 2.435871743486974e-05,
+      "loss": 0.0217,
+      "step": 2571
+    },
+    {
+      "epoch": 10.288,
+      "grad_norm": 0.8437207341194153,
+      "learning_rate": 2.434869739478958e-05,
+      "loss": 0.0209,
+      "step": 2572
+    },
+    {
+      "epoch": 10.292,
+      "grad_norm": 0.9303336143493652,
+      "learning_rate": 2.4338677354709417e-05,
+      "loss": 0.0271,
+      "step": 2573
+    },
+    {
+      "epoch": 10.296,
+      "grad_norm": 1.0167421102523804,
+      "learning_rate": 2.4328657314629262e-05,
+      "loss": 0.0206,
+      "step": 2574
+    },
+    {
+      "epoch": 10.3,
+      "grad_norm": 0.9930081367492676,
+      "learning_rate": 2.43186372745491e-05,
+      "loss": 0.0221,
+      "step": 2575
+    },
+    {
+      "epoch": 10.304,
+      "grad_norm": 0.6778939962387085,
+      "learning_rate": 2.4308617234468938e-05,
+      "loss": 0.0205,
+      "step": 2576
+    },
+    {
+      "epoch": 10.308,
+      "grad_norm": 0.8133848309516907,
+      "learning_rate": 2.429859719438878e-05,
+      "loss": 0.0196,
+      "step": 2577
+    },
+    {
+      "epoch": 10.312,
+      "grad_norm": 0.7504951357841492,
+      "learning_rate": 2.4288577154308617e-05,
+      "loss": 0.0173,
+      "step": 2578
+    },
+    {
+      "epoch": 10.316,
+      "grad_norm": 1.1020333766937256,
+      "learning_rate": 2.427855711422846e-05,
+      "loss": 0.0258,
+      "step": 2579
+    },
+    {
+      "epoch": 10.32,
+      "grad_norm": 0.807967483997345,
+      "learning_rate": 2.42685370741483e-05,
+      "loss": 0.0199,
+      "step": 2580
+    },
+    {
+      "epoch": 10.324,
+      "grad_norm": 0.9082990884780884,
+      "learning_rate": 2.4258517034068138e-05,
+      "loss": 0.0199,
+      "step": 2581
+    },
+    {
+      "epoch": 10.328,
+      "grad_norm": 0.8345447182655334,
+      "learning_rate": 2.4248496993987975e-05,
+      "loss": 0.021,
+      "step": 2582
+    },
+    {
+      "epoch": 10.332,
+      "grad_norm": 0.9574979543685913,
+      "learning_rate": 2.4238476953907817e-05,
+      "loss": 0.0243,
+      "step": 2583
+    },
+    {
+      "epoch": 10.336,
+      "grad_norm": 0.8273745775222778,
+      "learning_rate": 2.4228456913827658e-05,
+      "loss": 0.02,
+      "step": 2584
+    },
+    {
+      "epoch": 10.34,
+      "grad_norm": 0.695869505405426,
+      "learning_rate": 2.4218436873747496e-05,
+      "loss": 0.0154,
+      "step": 2585
+    },
+    {
+      "epoch": 10.344,
+      "grad_norm": 0.8283644914627075,
+      "learning_rate": 2.4208416833667337e-05,
+      "loss": 0.0204,
+      "step": 2586
+    },
+    {
+      "epoch": 10.348,
+      "grad_norm": 1.0260083675384521,
+      "learning_rate": 2.4198396793587175e-05,
+      "loss": 0.0259,
+      "step": 2587
+    },
+    {
+      "epoch": 10.352,
+      "grad_norm": 0.9744479656219482,
+      "learning_rate": 2.4188376753507013e-05,
+      "loss": 0.0243,
+      "step": 2588
+    },
+    {
+      "epoch": 10.356,
+      "grad_norm": 0.937955379486084,
+      "learning_rate": 2.4178356713426854e-05,
+      "loss": 0.0223,
+      "step": 2589
+    },
+    {
+      "epoch": 10.36,
+      "grad_norm": 1.1364651918411255,
+      "learning_rate": 2.4168336673346696e-05,
+      "loss": 0.0244,
+      "step": 2590
+    },
+    {
+      "epoch": 10.364,
+      "grad_norm": 0.9114662408828735,
+      "learning_rate": 2.4158316633266533e-05,
+      "loss": 0.027,
+      "step": 2591
+    },
+    {
+      "epoch": 10.368,
+      "grad_norm": 0.7826032638549805,
+      "learning_rate": 2.4148296593186375e-05,
+      "loss": 0.0192,
+      "step": 2592
+    },
+    {
+      "epoch": 10.372,
+      "grad_norm": 0.45732957124710083,
+      "learning_rate": 2.4138276553106216e-05,
+      "loss": 0.0133,
+      "step": 2593
+    },
+    {
+      "epoch": 10.376,
+      "grad_norm": 0.6898213624954224,
+      "learning_rate": 2.4128256513026054e-05,
+      "loss": 0.0214,
+      "step": 2594
+    },
+    {
+      "epoch": 10.38,
+      "grad_norm": 0.7986359596252441,
+      "learning_rate": 2.4118236472945892e-05,
+      "loss": 0.0184,
+      "step": 2595
+    },
+    {
+      "epoch": 10.384,
+      "grad_norm": 0.9170551896095276,
+      "learning_rate": 2.4108216432865733e-05,
+      "loss": 0.0215,
+      "step": 2596
+    },
+    {
+      "epoch": 10.388,
+      "grad_norm": 0.8053649067878723,
+      "learning_rate": 2.409819639278557e-05,
+      "loss": 0.0213,
+      "step": 2597
+    },
+    {
+      "epoch": 10.392,
+      "grad_norm": 0.8867815136909485,
+      "learning_rate": 2.4088176352705412e-05,
+      "loss": 0.0226,
+      "step": 2598
+    },
+    {
+      "epoch": 10.396,
+      "grad_norm": 0.8279624581336975,
+      "learning_rate": 2.4078156312625254e-05,
+      "loss": 0.0209,
+      "step": 2599
+    },
+    {
+      "epoch": 10.4,
+      "grad_norm": 0.6529654264450073,
+      "learning_rate": 2.406813627254509e-05,
+      "loss": 0.0134,
+      "step": 2600
+    },
+    {
+      "epoch": 10.404,
+      "grad_norm": 0.8741607666015625,
+      "learning_rate": 2.405811623246493e-05,
+      "loss": 0.0199,
+      "step": 2601
+    },
+    {
+      "epoch": 10.408,
+      "grad_norm": 0.9696851968765259,
+      "learning_rate": 2.404809619238477e-05,
+      "loss": 0.0212,
+      "step": 2602
+    },
+    {
+      "epoch": 10.412,
+      "grad_norm": 0.7984656095504761,
+      "learning_rate": 2.4038076152304612e-05,
+      "loss": 0.0197,
+      "step": 2603
+    },
+    {
+      "epoch": 10.416,
+      "grad_norm": 0.7449184656143188,
+      "learning_rate": 2.402805611222445e-05,
+      "loss": 0.0203,
+      "step": 2604
+    },
+    {
+      "epoch": 10.42,
+      "grad_norm": 0.7866941094398499,
+      "learning_rate": 2.401803607214429e-05,
+      "loss": 0.0214,
+      "step": 2605
+    },
+    {
+      "epoch": 10.424,
+      "grad_norm": 0.802543580532074,
+      "learning_rate": 2.400801603206413e-05,
+      "loss": 0.0216,
+      "step": 2606
+    },
+    {
+      "epoch": 10.428,
+      "grad_norm": 0.7483913898468018,
+      "learning_rate": 2.3997995991983967e-05,
+      "loss": 0.0205,
+      "step": 2607
+    },
+    {
+      "epoch": 10.432,
+      "grad_norm": 0.8060435652732849,
+      "learning_rate": 2.398797595190381e-05,
+      "loss": 0.0199,
+      "step": 2608
+    },
+    {
+      "epoch": 10.436,
+      "grad_norm": 0.7773402333259583,
+      "learning_rate": 2.397795591182365e-05,
+      "loss": 0.0184,
+      "step": 2609
+    },
+    {
+      "epoch": 10.44,
+      "grad_norm": 0.7629860639572144,
+      "learning_rate": 2.3967935871743487e-05,
+      "loss": 0.0191,
+      "step": 2610
+    },
+    {
+      "epoch": 10.444,
+      "grad_norm": 1.0891491174697876,
+      "learning_rate": 2.395791583166333e-05,
+      "loss": 0.0267,
+      "step": 2611
+    },
+    {
+      "epoch": 10.448,
+      "grad_norm": 0.8873718976974487,
+      "learning_rate": 2.3947895791583166e-05,
+      "loss": 0.023,
+      "step": 2612
+    },
+    {
+      "epoch": 10.452,
+      "grad_norm": 0.9421393871307373,
+      "learning_rate": 2.3937875751503008e-05,
+      "loss": 0.0231,
+      "step": 2613
+    },
+    {
+      "epoch": 10.456,
+      "grad_norm": 0.7414140105247498,
+      "learning_rate": 2.392785571142285e-05,
+      "loss": 0.0202,
+      "step": 2614
+    },
+    {
+      "epoch": 10.46,
+      "grad_norm": 0.9896325469017029,
+      "learning_rate": 2.3917835671342687e-05,
+      "loss": 0.0234,
+      "step": 2615
+    },
+    {
+      "epoch": 10.464,
+      "grad_norm": 0.7930034399032593,
+      "learning_rate": 2.3907815631262525e-05,
+      "loss": 0.021,
+      "step": 2616
+    },
+    {
+      "epoch": 10.468,
+      "grad_norm": 0.9249135851860046,
+      "learning_rate": 2.3897795591182366e-05,
+      "loss": 0.0225,
+      "step": 2617
+    },
+    {
+      "epoch": 10.472,
+      "grad_norm": 0.837981641292572,
+      "learning_rate": 2.3887775551102207e-05,
+      "loss": 0.0232,
+      "step": 2618
+    },
+    {
+      "epoch": 10.475999999999999,
+      "grad_norm": 1.0759623050689697,
+      "learning_rate": 2.3877755511022045e-05,
+      "loss": 0.0241,
+      "step": 2619
+    },
+    {
+      "epoch": 10.48,
+      "grad_norm": 0.8368483781814575,
+      "learning_rate": 2.3867735470941887e-05,
+      "loss": 0.0222,
+      "step": 2620
+    },
+    {
+      "epoch": 10.484,
+      "grad_norm": 0.9433078169822693,
+      "learning_rate": 2.3857715430861724e-05,
+      "loss": 0.0259,
+      "step": 2621
+    },
+    {
+      "epoch": 10.488,
+      "grad_norm": 0.9696186780929565,
+      "learning_rate": 2.3847695390781562e-05,
+      "loss": 0.0223,
+      "step": 2622
+    },
+    {
+      "epoch": 10.492,
+      "grad_norm": 1.07587468624115,
+      "learning_rate": 2.3837675350701404e-05,
+      "loss": 0.0252,
+      "step": 2623
+    },
+    {
+      "epoch": 10.496,
+      "grad_norm": 0.7479248642921448,
+      "learning_rate": 2.3827655310621245e-05,
+      "loss": 0.0196,
+      "step": 2624
+    },
+    {
+      "epoch": 10.5,
+      "grad_norm": 0.678727924823761,
+      "learning_rate": 2.3817635270541083e-05,
+      "loss": 0.0215,
+      "step": 2625
+    },
+    {
+      "epoch": 10.504,
+      "grad_norm": 0.7351118326187134,
+      "learning_rate": 2.380761523046092e-05,
+      "loss": 0.0214,
+      "step": 2626
+    },
+    {
+      "epoch": 10.508,
+      "grad_norm": 1.21925687789917,
+      "learning_rate": 2.3797595190380762e-05,
+      "loss": 0.0252,
+      "step": 2627
+    },
+    {
+      "epoch": 10.512,
+      "grad_norm": 0.9814172983169556,
+      "learning_rate": 2.3787575150300603e-05,
+      "loss": 0.0235,
+      "step": 2628
+    },
+    {
+      "epoch": 10.516,
+      "grad_norm": 0.880654513835907,
+      "learning_rate": 2.377755511022044e-05,
+      "loss": 0.0215,
+      "step": 2629
+    },
+    {
+      "epoch": 10.52,
+      "grad_norm": 0.9024983644485474,
+      "learning_rate": 2.3767535070140282e-05,
+      "loss": 0.0229,
+      "step": 2630
+    },
+    {
+      "epoch": 10.524000000000001,
+      "grad_norm": 1.2271602153778076,
+      "learning_rate": 2.375751503006012e-05,
+      "loss": 0.0266,
+      "step": 2631
+    },
+    {
+      "epoch": 10.528,
+      "grad_norm": 1.0690782070159912,
+      "learning_rate": 2.3747494989979958e-05,
+      "loss": 0.0242,
+      "step": 2632
+    },
+    {
+      "epoch": 10.532,
+      "grad_norm": 0.753580629825592,
+      "learning_rate": 2.3737474949899803e-05,
+      "loss": 0.0183,
+      "step": 2633
+    },
+    {
+      "epoch": 10.536,
+      "grad_norm": 0.7326211333274841,
+      "learning_rate": 2.372745490981964e-05,
+      "loss": 0.0209,
+      "step": 2634
+    },
+    {
+      "epoch": 10.54,
+      "grad_norm": 0.9598518013954163,
+      "learning_rate": 2.371743486973948e-05,
+      "loss": 0.0246,
+      "step": 2635
+    },
+    {
+      "epoch": 10.544,
+      "grad_norm": 0.7582380771636963,
+      "learning_rate": 2.370741482965932e-05,
+      "loss": 0.0197,
+      "step": 2636
+    },
+    {
+      "epoch": 10.548,
+      "grad_norm": 0.7412502765655518,
+      "learning_rate": 2.3697394789579158e-05,
+      "loss": 0.0207,
+      "step": 2637
+    },
+    {
+      "epoch": 10.552,
+      "grad_norm": 0.8043134808540344,
+      "learning_rate": 2.3687374749499e-05,
+      "loss": 0.0207,
+      "step": 2638
+    },
+    {
+      "epoch": 10.556000000000001,
+      "grad_norm": 1.084338665008545,
+      "learning_rate": 2.367735470941884e-05,
+      "loss": 0.0292,
+      "step": 2639
+    },
+    {
+      "epoch": 10.56,
+      "grad_norm": 0.871437668800354,
+      "learning_rate": 2.366733466933868e-05,
+      "loss": 0.0261,
+      "step": 2640
+    },
+    {
+      "epoch": 10.564,
+      "grad_norm": 0.8200016617774963,
+      "learning_rate": 2.3657314629258516e-05,
+      "loss": 0.0198,
+      "step": 2641
+    },
+    {
+      "epoch": 10.568,
+      "grad_norm": 0.9645569920539856,
+      "learning_rate": 2.3647294589178358e-05,
+      "loss": 0.0223,
+      "step": 2642
+    },
+    {
+      "epoch": 10.572,
+      "grad_norm": 1.119446873664856,
+      "learning_rate": 2.36372745490982e-05,
+      "loss": 0.0252,
+      "step": 2643
+    },
+    {
+      "epoch": 10.576,
+      "grad_norm": 1.2717159986495972,
+      "learning_rate": 2.3627254509018037e-05,
+      "loss": 0.0264,
+      "step": 2644
+    },
+    {
+      "epoch": 10.58,
+      "grad_norm": 0.7650407552719116,
+      "learning_rate": 2.3617234468937878e-05,
+      "loss": 0.0204,
+      "step": 2645
+    },
+    {
+      "epoch": 10.584,
+      "grad_norm": 0.48650267720222473,
+      "learning_rate": 2.3607214428857716e-05,
+      "loss": 0.011,
+      "step": 2646
+    },
+    {
+      "epoch": 10.588,
+      "grad_norm": 1.0167272090911865,
+      "learning_rate": 2.3597194388777557e-05,
+      "loss": 0.0281,
+      "step": 2647
+    },
+    {
+      "epoch": 10.592,
+      "grad_norm": 1.0820997953414917,
+      "learning_rate": 2.3587174348697395e-05,
+      "loss": 0.029,
+      "step": 2648
+    },
+    {
+      "epoch": 10.596,
+      "grad_norm": 0.4999312460422516,
+      "learning_rate": 2.3577154308617236e-05,
+      "loss": 0.0094,
+      "step": 2649
+    },
+    {
+      "epoch": 10.6,
+      "grad_norm": 0.9790526628494263,
+      "learning_rate": 2.3567134268537074e-05,
+      "loss": 0.0226,
+      "step": 2650
+    },
+    {
+      "epoch": 10.604,
+      "grad_norm": 0.9920722842216492,
+      "learning_rate": 2.3557114228456916e-05,
+      "loss": 0.0232,
+      "step": 2651
+    },
+    {
+      "epoch": 10.608,
+      "grad_norm": 0.6009891033172607,
+      "learning_rate": 2.3547094188376757e-05,
+      "loss": 0.0203,
+      "step": 2652
+    },
+    {
+      "epoch": 10.612,
+      "grad_norm": 0.7871741056442261,
+      "learning_rate": 2.3537074148296595e-05,
+      "loss": 0.0215,
+      "step": 2653
+    },
+    {
+      "epoch": 10.616,
+      "grad_norm": 1.217743992805481,
+      "learning_rate": 2.3527054108216433e-05,
+      "loss": 0.0251,
+      "step": 2654
+    },
+    {
+      "epoch": 10.62,
+      "grad_norm": 1.0484158992767334,
+      "learning_rate": 2.3517034068136274e-05,
+      "loss": 0.0261,
+      "step": 2655
+    },
+    {
+      "epoch": 10.624,
+      "grad_norm": 1.014896273612976,
+      "learning_rate": 2.3507014028056112e-05,
+      "loss": 0.0233,
+      "step": 2656
+    },
+    {
+      "epoch": 10.628,
+      "grad_norm": 1.0810414552688599,
+      "learning_rate": 2.3496993987975953e-05,
+      "loss": 0.0216,
+      "step": 2657
+    },
+    {
+      "epoch": 10.632,
+      "grad_norm": 0.7048183083534241,
+      "learning_rate": 2.3486973947895794e-05,
+      "loss": 0.0145,
+      "step": 2658
+    },
+    {
+      "epoch": 10.636,
+      "grad_norm": 1.054413914680481,
+      "learning_rate": 2.3476953907815632e-05,
+      "loss": 0.0252,
+      "step": 2659
+    },
+    {
+      "epoch": 10.64,
+      "grad_norm": 0.8256332874298096,
+      "learning_rate": 2.346693386773547e-05,
+      "loss": 0.021,
+      "step": 2660
+    },
+    {
+      "epoch": 10.644,
+      "grad_norm": 0.9366419911384583,
+      "learning_rate": 2.345691382765531e-05,
+      "loss": 0.0245,
+      "step": 2661
+    },
+    {
+      "epoch": 10.648,
+      "grad_norm": 0.9146870970726013,
+      "learning_rate": 2.3446893787575153e-05,
+      "loss": 0.0218,
+      "step": 2662
+    },
+    {
+      "epoch": 10.652,
+      "grad_norm": 0.9589238166809082,
+      "learning_rate": 2.343687374749499e-05,
+      "loss": 0.0272,
+      "step": 2663
+    },
+    {
+      "epoch": 10.656,
+      "grad_norm": 0.7012674808502197,
+      "learning_rate": 2.3426853707414832e-05,
+      "loss": 0.0205,
+      "step": 2664
+    },
+    {
+      "epoch": 10.66,
+      "grad_norm": 0.872169554233551,
+      "learning_rate": 2.341683366733467e-05,
+      "loss": 0.0212,
+      "step": 2665
+    },
+    {
+      "epoch": 10.664,
+      "grad_norm": 1.1105194091796875,
+      "learning_rate": 2.3406813627254508e-05,
+      "loss": 0.0236,
+      "step": 2666
+    },
+    {
+      "epoch": 10.668,
+      "grad_norm": 0.9917757511138916,
+      "learning_rate": 2.3396793587174352e-05,
+      "loss": 0.0274,
+      "step": 2667
+    },
+    {
+      "epoch": 10.672,
+      "grad_norm": 0.7267541289329529,
+      "learning_rate": 2.338677354709419e-05,
+      "loss": 0.0186,
+      "step": 2668
+    },
+    {
+      "epoch": 10.676,
+      "grad_norm": 0.8156752586364746,
+      "learning_rate": 2.3376753507014028e-05,
+      "loss": 0.0192,
+      "step": 2669
+    },
+    {
+      "epoch": 10.68,
+      "grad_norm": 0.8577165603637695,
+      "learning_rate": 2.336673346693387e-05,
+      "loss": 0.0235,
+      "step": 2670
+    },
+    {
+      "epoch": 10.684,
+      "grad_norm": 0.5355784893035889,
+      "learning_rate": 2.3356713426853707e-05,
+      "loss": 0.0113,
+      "step": 2671
+    },
+    {
+      "epoch": 10.688,
+      "grad_norm": 1.0173070430755615,
+      "learning_rate": 2.334669338677355e-05,
+      "loss": 0.0263,
+      "step": 2672
+    },
+    {
+      "epoch": 10.692,
+      "grad_norm": 0.8700453639030457,
+      "learning_rate": 2.333667334669339e-05,
+      "loss": 0.0253,
+      "step": 2673
+    },
+    {
+      "epoch": 10.696,
+      "grad_norm": 0.8330980539321899,
+      "learning_rate": 2.3326653306613228e-05,
+      "loss": 0.0198,
+      "step": 2674
+    },
+    {
+      "epoch": 10.7,
+      "grad_norm": 0.7934946417808533,
+      "learning_rate": 2.3316633266533066e-05,
+      "loss": 0.0204,
+      "step": 2675
+    },
+    {
+      "epoch": 10.704,
+      "grad_norm": 0.7809090614318848,
+      "learning_rate": 2.3306613226452907e-05,
+      "loss": 0.026,
+      "step": 2676
+    },
+    {
+      "epoch": 10.708,
+      "grad_norm": 0.8526442646980286,
+      "learning_rate": 2.3296593186372748e-05,
+      "loss": 0.0208,
+      "step": 2677
+    },
+    {
+      "epoch": 10.712,
+      "grad_norm": 0.9328563213348389,
+      "learning_rate": 2.3286573146292586e-05,
+      "loss": 0.0229,
+      "step": 2678
+    },
+    {
+      "epoch": 10.716,
+      "grad_norm": 0.767686665058136,
+      "learning_rate": 2.3276553106212427e-05,
+      "loss": 0.021,
+      "step": 2679
+    },
+    {
+      "epoch": 10.72,
+      "grad_norm": 1.204643726348877,
+      "learning_rate": 2.3266533066132265e-05,
+      "loss": 0.0283,
+      "step": 2680
+    },
+    {
+      "epoch": 10.724,
+      "grad_norm": 0.8405128121376038,
+      "learning_rate": 2.3256513026052103e-05,
+      "loss": 0.0216,
+      "step": 2681
+    },
+    {
+      "epoch": 10.728,
+      "grad_norm": 0.9929158091545105,
+      "learning_rate": 2.3246492985971944e-05,
+      "loss": 0.0236,
+      "step": 2682
+    },
+    {
+      "epoch": 10.732,
+      "grad_norm": 1.1628848314285278,
+      "learning_rate": 2.3236472945891786e-05,
+      "loss": 0.0234,
+      "step": 2683
+    },
+    {
+      "epoch": 10.736,
+      "grad_norm": 1.0076568126678467,
+      "learning_rate": 2.3226452905811624e-05,
+      "loss": 0.0232,
+      "step": 2684
+    },
+    {
+      "epoch": 10.74,
+      "grad_norm": 0.8345778584480286,
+      "learning_rate": 2.3216432865731465e-05,
+      "loss": 0.0221,
+      "step": 2685
+    },
+    {
+      "epoch": 10.744,
+      "grad_norm": 0.8815706968307495,
+      "learning_rate": 2.3206412825651303e-05,
+      "loss": 0.0255,
+      "step": 2686
+    },
+    {
+      "epoch": 10.748,
+      "grad_norm": 1.1422778367996216,
+      "learning_rate": 2.3196392785571144e-05,
+      "loss": 0.0309,
+      "step": 2687
+    },
+    {
+      "epoch": 10.752,
+      "grad_norm": 0.9792990684509277,
+      "learning_rate": 2.3186372745490982e-05,
+      "loss": 0.0271,
+      "step": 2688
+    },
+    {
+      "epoch": 10.756,
+      "grad_norm": 0.972955048084259,
+      "learning_rate": 2.3176352705410823e-05,
+      "loss": 0.0263,
+      "step": 2689
+    },
+    {
+      "epoch": 10.76,
+      "grad_norm": 0.9153329133987427,
+      "learning_rate": 2.316633266533066e-05,
+      "loss": 0.0213,
+      "step": 2690
+    },
+    {
+      "epoch": 10.764,
+      "grad_norm": 1.1069750785827637,
+      "learning_rate": 2.3156312625250502e-05,
+      "loss": 0.028,
+      "step": 2691
+    },
+    {
+      "epoch": 10.768,
+      "grad_norm": 1.0217959880828857,
+      "learning_rate": 2.3146292585170344e-05,
+      "loss": 0.0308,
+      "step": 2692
+    },
+    {
+      "epoch": 10.772,
+      "grad_norm": 0.8818560838699341,
+      "learning_rate": 2.313627254509018e-05,
+      "loss": 0.0241,
+      "step": 2693
+    },
+    {
+      "epoch": 10.776,
+      "grad_norm": 1.0862679481506348,
+      "learning_rate": 2.312625250501002e-05,
+      "loss": 0.0253,
+      "step": 2694
+    },
+    {
+      "epoch": 10.78,
+      "grad_norm": 0.8923512697219849,
+      "learning_rate": 2.311623246492986e-05,
+      "loss": 0.0232,
+      "step": 2695
+    },
+    {
+      "epoch": 10.784,
+      "grad_norm": 0.8306083679199219,
+      "learning_rate": 2.31062124248497e-05,
+      "loss": 0.0169,
+      "step": 2696
+    },
+    {
+      "epoch": 10.788,
+      "grad_norm": 0.904339075088501,
+      "learning_rate": 2.309619238476954e-05,
+      "loss": 0.0221,
+      "step": 2697
+    },
+    {
+      "epoch": 10.792,
+      "grad_norm": 1.143362283706665,
+      "learning_rate": 2.308617234468938e-05,
+      "loss": 0.0272,
+      "step": 2698
+    },
+    {
+      "epoch": 10.796,
+      "grad_norm": 0.9651113152503967,
+      "learning_rate": 2.307615230460922e-05,
+      "loss": 0.0271,
+      "step": 2699
+    },
+    {
+      "epoch": 10.8,
+      "grad_norm": 1.1126807928085327,
+      "learning_rate": 2.3066132264529057e-05,
+      "loss": 0.0264,
+      "step": 2700
+    },
+    {
+      "epoch": 10.804,
+      "grad_norm": 1.1432498693466187,
+      "learning_rate": 2.30561122244489e-05,
+      "loss": 0.0294,
+      "step": 2701
+    },
+    {
+      "epoch": 10.808,
+      "grad_norm": 0.907027542591095,
+      "learning_rate": 2.304609218436874e-05,
+      "loss": 0.0257,
+      "step": 2702
+    },
+    {
+      "epoch": 10.812,
+      "grad_norm": 1.038845181465149,
+      "learning_rate": 2.3036072144288577e-05,
+      "loss": 0.0269,
+      "step": 2703
+    },
+    {
+      "epoch": 10.816,
+      "grad_norm": 0.6149145364761353,
+      "learning_rate": 2.302605210420842e-05,
+      "loss": 0.012,
+      "step": 2704
+    },
+    {
+      "epoch": 10.82,
+      "grad_norm": 0.7798912525177002,
+      "learning_rate": 2.3016032064128257e-05,
+      "loss": 0.0213,
+      "step": 2705
+    },
+    {
+      "epoch": 10.824,
+      "grad_norm": 0.8147614002227783,
+      "learning_rate": 2.3006012024048098e-05,
+      "loss": 0.0209,
+      "step": 2706
+    },
+    {
+      "epoch": 10.828,
+      "grad_norm": 1.100545048713684,
+      "learning_rate": 2.299599198396794e-05,
+      "loss": 0.0239,
+      "step": 2707
+    },
+    {
+      "epoch": 10.832,
+      "grad_norm": 0.9930369853973389,
+      "learning_rate": 2.2985971943887777e-05,
+      "loss": 0.0247,
+      "step": 2708
+    },
+    {
+      "epoch": 10.836,
+      "grad_norm": 0.8570621609687805,
+      "learning_rate": 2.2975951903807615e-05,
+      "loss": 0.0212,
+      "step": 2709
+    },
+    {
+      "epoch": 10.84,
+      "grad_norm": 0.8312192559242249,
+      "learning_rate": 2.2965931863727456e-05,
+      "loss": 0.0168,
+      "step": 2710
+    },
+    {
+      "epoch": 10.844,
+      "grad_norm": 1.1362755298614502,
+      "learning_rate": 2.2955911823647298e-05,
+      "loss": 0.0303,
+      "step": 2711
+    },
+    {
+      "epoch": 10.848,
+      "grad_norm": 1.1022303104400635,
+      "learning_rate": 2.2945891783567135e-05,
+      "loss": 0.0293,
+      "step": 2712
+    },
+    {
+      "epoch": 10.852,
+      "grad_norm": 1.1668660640716553,
+      "learning_rate": 2.2935871743486977e-05,
+      "loss": 0.0259,
+      "step": 2713
+    },
+    {
+      "epoch": 10.856,
+      "grad_norm": 1.326250672340393,
+      "learning_rate": 2.2925851703406815e-05,
+      "loss": 0.0259,
+      "step": 2714
+    },
+    {
+      "epoch": 10.86,
+      "grad_norm": 0.9431809782981873,
+      "learning_rate": 2.2915831663326653e-05,
+      "loss": 0.0223,
+      "step": 2715
+    },
+    {
+      "epoch": 10.864,
+      "grad_norm": 1.0126006603240967,
+      "learning_rate": 2.2905811623246494e-05,
+      "loss": 0.0231,
+      "step": 2716
+    },
+    {
+      "epoch": 10.868,
+      "grad_norm": 0.9708862900733948,
+      "learning_rate": 2.2895791583166335e-05,
+      "loss": 0.0237,
+      "step": 2717
+    },
+    {
+      "epoch": 10.872,
+      "grad_norm": 0.9651861786842346,
+      "learning_rate": 2.2885771543086173e-05,
+      "loss": 0.0229,
+      "step": 2718
+    },
+    {
+      "epoch": 10.876,
+      "grad_norm": 0.8365417718887329,
+      "learning_rate": 2.2875751503006014e-05,
+      "loss": 0.0233,
+      "step": 2719
+    },
+    {
+      "epoch": 10.88,
+      "grad_norm": 0.8717615008354187,
+      "learning_rate": 2.2865731462925852e-05,
+      "loss": 0.022,
+      "step": 2720
+    },
+    {
+      "epoch": 10.884,
+      "grad_norm": 1.1816632747650146,
+      "learning_rate": 2.2855711422845693e-05,
+      "loss": 0.0273,
+      "step": 2721
+    },
+    {
+      "epoch": 10.888,
+      "grad_norm": 0.9528033137321472,
+      "learning_rate": 2.284569138276553e-05,
+      "loss": 0.0229,
+      "step": 2722
+    },
+    {
+      "epoch": 10.892,
+      "grad_norm": 1.0653725862503052,
+      "learning_rate": 2.2835671342685373e-05,
+      "loss": 0.027,
+      "step": 2723
+    },
+    {
+      "epoch": 10.896,
+      "grad_norm": 0.5483238101005554,
+      "learning_rate": 2.282565130260521e-05,
+      "loss": 0.013,
+      "step": 2724
+    },
+    {
+      "epoch": 10.9,
+      "grad_norm": 1.0084025859832764,
+      "learning_rate": 2.281563126252505e-05,
+      "loss": 0.0233,
+      "step": 2725
+    },
+    {
+      "epoch": 10.904,
+      "grad_norm": 1.0322855710983276,
+      "learning_rate": 2.2805611222444893e-05,
+      "loss": 0.0251,
+      "step": 2726
+    },
+    {
+      "epoch": 10.908,
+      "grad_norm": 0.9313704967498779,
+      "learning_rate": 2.279559118236473e-05,
+      "loss": 0.0234,
+      "step": 2727
+    },
+    {
+      "epoch": 10.912,
+      "grad_norm": 1.1278380155563354,
+      "learning_rate": 2.278557114228457e-05,
+      "loss": 0.0299,
+      "step": 2728
+    },
+    {
+      "epoch": 10.916,
+      "grad_norm": 1.118483304977417,
+      "learning_rate": 2.277555110220441e-05,
+      "loss": 0.0252,
+      "step": 2729
+    },
+    {
+      "epoch": 10.92,
+      "grad_norm": 0.8251449465751648,
+      "learning_rate": 2.2765531062124248e-05,
+      "loss": 0.0198,
+      "step": 2730
+    },
+    {
+      "epoch": 10.924,
+      "grad_norm": 1.0784047842025757,
+      "learning_rate": 2.275551102204409e-05,
+      "loss": 0.0305,
+      "step": 2731
+    },
+    {
+      "epoch": 10.928,
+      "grad_norm": 1.0076247453689575,
+      "learning_rate": 2.274549098196393e-05,
+      "loss": 0.028,
+      "step": 2732
+    },
+    {
+      "epoch": 10.932,
+      "grad_norm": 1.17436683177948,
+      "learning_rate": 2.273547094188377e-05,
+      "loss": 0.0264,
+      "step": 2733
+    },
+    {
+      "epoch": 10.936,
+      "grad_norm": 1.0806264877319336,
+      "learning_rate": 2.2725450901803606e-05,
+      "loss": 0.025,
+      "step": 2734
+    },
+    {
+      "epoch": 10.94,
+      "grad_norm": 0.8272656202316284,
+      "learning_rate": 2.2715430861723448e-05,
+      "loss": 0.0243,
+      "step": 2735
+    },
+    {
+      "epoch": 10.943999999999999,
+      "grad_norm": 1.1387969255447388,
+      "learning_rate": 2.270541082164329e-05,
+      "loss": 0.0316,
+      "step": 2736
+    },
+    {
+      "epoch": 10.948,
+      "grad_norm": 0.8464391231536865,
+      "learning_rate": 2.2695390781563127e-05,
+      "loss": 0.0245,
+      "step": 2737
+    },
+    {
+      "epoch": 10.952,
+      "grad_norm": 0.8681789636611938,
+      "learning_rate": 2.2685370741482968e-05,
+      "loss": 0.0247,
+      "step": 2738
+    },
+    {
+      "epoch": 10.956,
+      "grad_norm": 0.7716442942619324,
+      "learning_rate": 2.2675350701402806e-05,
+      "loss": 0.0196,
+      "step": 2739
+    },
+    {
+      "epoch": 10.96,
+      "grad_norm": 0.8722916841506958,
+      "learning_rate": 2.2665330661322644e-05,
+      "loss": 0.0243,
+      "step": 2740
+    },
+    {
+      "epoch": 10.964,
+      "grad_norm": 0.9783482551574707,
+      "learning_rate": 2.265531062124249e-05,
+      "loss": 0.025,
+      "step": 2741
+    },
+    {
+      "epoch": 10.968,
+      "grad_norm": 0.7285762429237366,
+      "learning_rate": 2.2645290581162327e-05,
+      "loss": 0.023,
+      "step": 2742
+    },
+    {
+      "epoch": 10.972,
+      "grad_norm": 1.012169599533081,
+      "learning_rate": 2.2635270541082164e-05,
+      "loss": 0.0263,
+      "step": 2743
+    },
+    {
+      "epoch": 10.975999999999999,
+      "grad_norm": 1.017822265625,
+      "learning_rate": 2.2625250501002006e-05,
+      "loss": 0.0249,
+      "step": 2744
+    },
+    {
+      "epoch": 10.98,
+      "grad_norm": 0.8947110772132874,
+      "learning_rate": 2.2615230460921844e-05,
+      "loss": 0.0242,
+      "step": 2745
+    },
+    {
+      "epoch": 10.984,
+      "grad_norm": 0.8670669198036194,
+      "learning_rate": 2.2605210420841685e-05,
+      "loss": 0.0209,
+      "step": 2746
+    },
+    {
+      "epoch": 10.988,
+      "grad_norm": 1.1200064420700073,
+      "learning_rate": 2.2595190380761526e-05,
+      "loss": 0.0294,
+      "step": 2747
+    },
+    {
+      "epoch": 10.992,
+      "grad_norm": 0.7422370910644531,
+      "learning_rate": 2.2585170340681364e-05,
+      "loss": 0.0218,
+      "step": 2748
+    },
+    {
+      "epoch": 10.996,
+      "grad_norm": 1.087212324142456,
+      "learning_rate": 2.2575150300601202e-05,
+      "loss": 0.0252,
+      "step": 2749
+    },
+    {
+      "epoch": 11.0,
+      "grad_norm": 0.9888896942138672,
+      "learning_rate": 2.2565130260521043e-05,
+      "loss": 0.0232,
+      "step": 2750
+    },
+    {
+      "epoch": 11.004,
+      "grad_norm": 0.7371053695678711,
+      "learning_rate": 2.2555110220440885e-05,
+      "loss": 0.0197,
+      "step": 2751
+    },
+    {
+      "epoch": 11.008,
+      "grad_norm": 0.7480749487876892,
+      "learning_rate": 2.2545090180360722e-05,
+      "loss": 0.0151,
+      "step": 2752
+    },
+    {
+      "epoch": 11.012,
+      "grad_norm": 0.47101399302482605,
+      "learning_rate": 2.253507014028056e-05,
+      "loss": 0.0126,
+      "step": 2753
+    },
+    {
+      "epoch": 11.016,
+      "grad_norm": 0.5534325242042542,
+      "learning_rate": 2.25250501002004e-05,
+      "loss": 0.0152,
+      "step": 2754
+    },
+    {
+      "epoch": 11.02,
+      "grad_norm": 0.6473478078842163,
+      "learning_rate": 2.251503006012024e-05,
+      "loss": 0.0152,
+      "step": 2755
+    },
+    {
+      "epoch": 11.024,
+      "grad_norm": 0.6150280833244324,
+      "learning_rate": 2.250501002004008e-05,
+      "loss": 0.0164,
+      "step": 2756
+    },
+    {
+      "epoch": 11.028,
+      "grad_norm": 0.6889513731002808,
+      "learning_rate": 2.2494989979959922e-05,
+      "loss": 0.0138,
+      "step": 2757
+    },
+    {
+      "epoch": 11.032,
+      "grad_norm": 0.7769390344619751,
+      "learning_rate": 2.248496993987976e-05,
+      "loss": 0.0163,
+      "step": 2758
+    },
+    {
+      "epoch": 11.036,
+      "grad_norm": 0.7019205093383789,
+      "learning_rate": 2.2474949899799598e-05,
+      "loss": 0.0157,
+      "step": 2759
+    },
+    {
+      "epoch": 11.04,
+      "grad_norm": 0.966162383556366,
+      "learning_rate": 2.2464929859719443e-05,
+      "loss": 0.0205,
+      "step": 2760
+    },
+    {
+      "epoch": 11.044,
+      "grad_norm": 0.7418727278709412,
+      "learning_rate": 2.245490981963928e-05,
+      "loss": 0.0172,
+      "step": 2761
+    },
+    {
+      "epoch": 11.048,
+      "grad_norm": 0.5667153596878052,
+      "learning_rate": 2.244488977955912e-05,
+      "loss": 0.0099,
+      "step": 2762
+    },
+    {
+      "epoch": 11.052,
+      "grad_norm": 0.9118229150772095,
+      "learning_rate": 2.243486973947896e-05,
+      "loss": 0.0182,
+      "step": 2763
+    },
+    {
+      "epoch": 11.056,
+      "grad_norm": 0.8428515195846558,
+      "learning_rate": 2.2424849699398797e-05,
+      "loss": 0.018,
+      "step": 2764
+    },
+    {
+      "epoch": 11.06,
+      "grad_norm": 0.5862798094749451,
+      "learning_rate": 2.241482965931864e-05,
+      "loss": 0.015,
+      "step": 2765
+    },
+    {
+      "epoch": 11.064,
+      "grad_norm": 0.6175693273544312,
+      "learning_rate": 2.240480961923848e-05,
+      "loss": 0.0171,
+      "step": 2766
+    },
+    {
+      "epoch": 11.068,
+      "grad_norm": 0.7874165177345276,
+      "learning_rate": 2.2394789579158318e-05,
+      "loss": 0.0183,
+      "step": 2767
+    },
+    {
+      "epoch": 11.072,
+      "grad_norm": 0.9282909631729126,
+      "learning_rate": 2.2384769539078156e-05,
+      "loss": 0.0184,
+      "step": 2768
+    },
+    {
+      "epoch": 11.076,
+      "grad_norm": 0.7215555906295776,
+      "learning_rate": 2.2374749498997997e-05,
+      "loss": 0.0185,
+      "step": 2769
+    },
+    {
+      "epoch": 11.08,
+      "grad_norm": 0.8601921200752258,
+      "learning_rate": 2.236472945891784e-05,
+      "loss": 0.0185,
+      "step": 2770
+    },
+    {
+      "epoch": 11.084,
+      "grad_norm": 0.9709322452545166,
+      "learning_rate": 2.2354709418837676e-05,
+      "loss": 0.0185,
+      "step": 2771
+    },
+    {
+      "epoch": 11.088,
+      "grad_norm": 0.799983561038971,
+      "learning_rate": 2.2344689378757518e-05,
+      "loss": 0.0164,
+      "step": 2772
+    },
+    {
+      "epoch": 11.092,
+      "grad_norm": 0.47060427069664,
+      "learning_rate": 2.2334669338677355e-05,
+      "loss": 0.0136,
+      "step": 2773
+    },
+    {
+      "epoch": 11.096,
+      "grad_norm": 0.8248326182365417,
+      "learning_rate": 2.2324649298597193e-05,
+      "loss": 0.018,
+      "step": 2774
+    },
+    {
+      "epoch": 11.1,
+      "grad_norm": 0.7863957285881042,
+      "learning_rate": 2.2314629258517035e-05,
+      "loss": 0.0199,
+      "step": 2775
+    },
+    {
+      "epoch": 11.104,
+      "grad_norm": 0.584855318069458,
+      "learning_rate": 2.2304609218436876e-05,
+      "loss": 0.0125,
+      "step": 2776
+    },
+    {
+      "epoch": 11.108,
+      "grad_norm": 0.7124798893928528,
+      "learning_rate": 2.2294589178356714e-05,
+      "loss": 0.0128,
+      "step": 2777
+    },
+    {
+      "epoch": 11.112,
+      "grad_norm": 0.6670029759407043,
+      "learning_rate": 2.2284569138276555e-05,
+      "loss": 0.0167,
+      "step": 2778
+    },
+    {
+      "epoch": 11.116,
+      "grad_norm": 0.7364521026611328,
+      "learning_rate": 2.2274549098196393e-05,
+      "loss": 0.0155,
+      "step": 2779
+    },
+    {
+      "epoch": 11.12,
+      "grad_norm": 0.7299067974090576,
+      "learning_rate": 2.2264529058116234e-05,
+      "loss": 0.0162,
+      "step": 2780
+    },
+    {
+      "epoch": 11.124,
+      "grad_norm": 0.5108133554458618,
+      "learning_rate": 2.2254509018036072e-05,
+      "loss": 0.0162,
+      "step": 2781
+    },
+    {
+      "epoch": 11.128,
+      "grad_norm": 0.8171691298484802,
+      "learning_rate": 2.2244488977955913e-05,
+      "loss": 0.0185,
+      "step": 2782
+    },
+    {
+      "epoch": 11.132,
+      "grad_norm": 0.7556437253952026,
+      "learning_rate": 2.223446893787575e-05,
+      "loss": 0.0181,
+      "step": 2783
+    },
+    {
+      "epoch": 11.136,
+      "grad_norm": 0.8455727100372314,
+      "learning_rate": 2.2224448897795593e-05,
+      "loss": 0.0158,
+      "step": 2784
+    },
+    {
+      "epoch": 11.14,
+      "grad_norm": 0.7372032999992371,
+      "learning_rate": 2.2214428857715434e-05,
+      "loss": 0.0168,
+      "step": 2785
+    },
+    {
+      "epoch": 11.144,
+      "grad_norm": 0.9267023801803589,
+      "learning_rate": 2.2204408817635272e-05,
+      "loss": 0.0162,
+      "step": 2786
+    },
+    {
+      "epoch": 11.148,
+      "grad_norm": 0.7094591856002808,
+      "learning_rate": 2.219438877755511e-05,
+      "loss": 0.0167,
+      "step": 2787
+    },
+    {
+      "epoch": 11.152,
+      "grad_norm": 0.8599808812141418,
+      "learning_rate": 2.218436873747495e-05,
+      "loss": 0.0223,
+      "step": 2788
+    },
+    {
+      "epoch": 11.156,
+      "grad_norm": 0.6255855560302734,
+      "learning_rate": 2.217434869739479e-05,
+      "loss": 0.014,
+      "step": 2789
+    },
+    {
+      "epoch": 11.16,
+      "grad_norm": 0.5172903537750244,
+      "learning_rate": 2.216432865731463e-05,
+      "loss": 0.0137,
+      "step": 2790
+    },
+    {
+      "epoch": 11.164,
+      "grad_norm": 0.971419095993042,
+      "learning_rate": 2.215430861723447e-05,
+      "loss": 0.0165,
+      "step": 2791
+    },
+    {
+      "epoch": 11.168,
+      "grad_norm": 0.8272615671157837,
+      "learning_rate": 2.214428857715431e-05,
+      "loss": 0.0165,
+      "step": 2792
+    },
+    {
+      "epoch": 11.172,
+      "grad_norm": 0.8575184345245361,
+      "learning_rate": 2.2134268537074147e-05,
+      "loss": 0.0194,
+      "step": 2793
+    },
+    {
+      "epoch": 11.176,
+      "grad_norm": 0.54851895570755,
+      "learning_rate": 2.212424849699399e-05,
+      "loss": 0.0147,
+      "step": 2794
+    },
+    {
+      "epoch": 11.18,
+      "grad_norm": 0.6194243431091309,
+      "learning_rate": 2.211422845691383e-05,
+      "loss": 0.0155,
+      "step": 2795
+    },
+    {
+      "epoch": 11.184,
+      "grad_norm": 0.6905847191810608,
+      "learning_rate": 2.2104208416833668e-05,
+      "loss": 0.0166,
+      "step": 2796
+    },
+    {
+      "epoch": 11.188,
+      "grad_norm": 0.7952155470848083,
+      "learning_rate": 2.209418837675351e-05,
+      "loss": 0.0172,
+      "step": 2797
+    },
+    {
+      "epoch": 11.192,
+      "grad_norm": 0.5479825735092163,
+      "learning_rate": 2.2084168336673347e-05,
+      "loss": 0.0138,
+      "step": 2798
+    },
+    {
+      "epoch": 11.196,
+      "grad_norm": 0.7587847709655762,
+      "learning_rate": 2.2074148296593185e-05,
+      "loss": 0.0174,
+      "step": 2799
+    },
+    {
+      "epoch": 11.2,
+      "grad_norm": 0.6983550786972046,
+      "learning_rate": 2.206412825651303e-05,
+      "loss": 0.0158,
+      "step": 2800
+    },
+    {
+      "epoch": 11.204,
+      "grad_norm": 0.7654238939285278,
+      "learning_rate": 2.2054108216432867e-05,
+      "loss": 0.0162,
+      "step": 2801
+    },
+    {
+      "epoch": 11.208,
+      "grad_norm": 0.7754288911819458,
+      "learning_rate": 2.2044088176352705e-05,
+      "loss": 0.016,
+      "step": 2802
+    },
+    {
+      "epoch": 11.212,
+      "grad_norm": 0.8081237077713013,
+      "learning_rate": 2.2034068136272547e-05,
+      "loss": 0.0176,
+      "step": 2803
+    },
+    {
+      "epoch": 11.216,
+      "grad_norm": 0.6752600073814392,
+      "learning_rate": 2.2024048096192384e-05,
+      "loss": 0.0147,
+      "step": 2804
+    },
+    {
+      "epoch": 11.22,
+      "grad_norm": 0.6993342041969299,
+      "learning_rate": 2.2014028056112226e-05,
+      "loss": 0.0159,
+      "step": 2805
+    },
+    {
+      "epoch": 11.224,
+      "grad_norm": 0.6669374108314514,
+      "learning_rate": 2.2004008016032067e-05,
+      "loss": 0.0163,
+      "step": 2806
+    },
+    {
+      "epoch": 11.228,
+      "grad_norm": 0.6162493228912354,
+      "learning_rate": 2.1993987975951905e-05,
+      "loss": 0.0168,
+      "step": 2807
+    },
+    {
+      "epoch": 11.232,
+      "grad_norm": 1.0216128826141357,
+      "learning_rate": 2.1983967935871743e-05,
+      "loss": 0.0179,
+      "step": 2808
+    },
+    {
+      "epoch": 11.236,
+      "grad_norm": 0.8339212536811829,
+      "learning_rate": 2.1973947895791584e-05,
+      "loss": 0.0161,
+      "step": 2809
+    },
+    {
+      "epoch": 11.24,
+      "grad_norm": 0.6215983033180237,
+      "learning_rate": 2.1963927855711425e-05,
+      "loss": 0.0166,
+      "step": 2810
+    },
+    {
+      "epoch": 11.244,
+      "grad_norm": 0.5474804639816284,
+      "learning_rate": 2.1953907815631263e-05,
+      "loss": 0.0149,
+      "step": 2811
+    },
+    {
+      "epoch": 11.248,
+      "grad_norm": 0.5640460252761841,
+      "learning_rate": 2.1943887775551105e-05,
+      "loss": 0.0132,
+      "step": 2812
+    },
+    {
+      "epoch": 11.252,
+      "grad_norm": 0.8308565616607666,
+      "learning_rate": 2.1933867735470942e-05,
+      "loss": 0.0185,
+      "step": 2813
+    },
+    {
+      "epoch": 11.256,
+      "grad_norm": 1.0242302417755127,
+      "learning_rate": 2.1923847695390784e-05,
+      "loss": 0.0197,
+      "step": 2814
+    },
+    {
+      "epoch": 11.26,
+      "grad_norm": 0.8074948787689209,
+      "learning_rate": 2.191382765531062e-05,
+      "loss": 0.0146,
+      "step": 2815
+    },
+    {
+      "epoch": 11.264,
+      "grad_norm": 0.5657005906105042,
+      "learning_rate": 2.1903807615230463e-05,
+      "loss": 0.0135,
+      "step": 2816
+    },
+    {
+      "epoch": 11.268,
+      "grad_norm": 0.8721190690994263,
+      "learning_rate": 2.18937875751503e-05,
+      "loss": 0.0193,
+      "step": 2817
+    },
+    {
+      "epoch": 11.272,
+      "grad_norm": 0.5107552409172058,
+      "learning_rate": 2.1883767535070142e-05,
+      "loss": 0.0158,
+      "step": 2818
+    },
+    {
+      "epoch": 11.276,
+      "grad_norm": 0.747622549533844,
+      "learning_rate": 2.1873747494989983e-05,
+      "loss": 0.0152,
+      "step": 2819
+    },
+    {
+      "epoch": 11.28,
+      "grad_norm": 0.7449367046356201,
+      "learning_rate": 2.186372745490982e-05,
+      "loss": 0.0163,
+      "step": 2820
+    },
+    {
+      "epoch": 11.284,
+      "grad_norm": 0.47581028938293457,
+      "learning_rate": 2.185370741482966e-05,
+      "loss": 0.0142,
+      "step": 2821
+    },
+    {
+      "epoch": 11.288,
+      "grad_norm": 0.5532326698303223,
+      "learning_rate": 2.18436873747495e-05,
+      "loss": 0.014,
+      "step": 2822
+    },
+    {
+      "epoch": 11.292,
+      "grad_norm": 0.7616000175476074,
+      "learning_rate": 2.1833667334669338e-05,
+      "loss": 0.0169,
+      "step": 2823
+    },
+    {
+      "epoch": 11.296,
+      "grad_norm": 0.4382837414741516,
+      "learning_rate": 2.182364729458918e-05,
+      "loss": 0.0098,
+      "step": 2824
+    },
+    {
+      "epoch": 11.3,
+      "grad_norm": 0.6851035952568054,
+      "learning_rate": 2.181362725450902e-05,
+      "loss": 0.0172,
+      "step": 2825
+    },
+    {
+      "epoch": 11.304,
+      "grad_norm": 0.7380544543266296,
+      "learning_rate": 2.180360721442886e-05,
+      "loss": 0.0176,
+      "step": 2826
+    },
+    {
+      "epoch": 11.308,
+      "grad_norm": 0.4597725570201874,
+      "learning_rate": 2.1793587174348697e-05,
+      "loss": 0.0134,
+      "step": 2827
+    },
+    {
+      "epoch": 11.312,
+      "grad_norm": 0.40633803606033325,
+      "learning_rate": 2.1783567134268538e-05,
+      "loss": 0.0129,
+      "step": 2828
+    },
+    {
+      "epoch": 11.316,
+      "grad_norm": 0.9875346422195435,
+      "learning_rate": 2.177354709418838e-05,
+      "loss": 0.0103,
+      "step": 2829
+    },
+    {
+      "epoch": 11.32,
+      "grad_norm": 0.9050310254096985,
+      "learning_rate": 2.1763527054108217e-05,
+      "loss": 0.0154,
+      "step": 2830
+    },
+    {
+      "epoch": 11.324,
+      "grad_norm": 0.7824398279190063,
+      "learning_rate": 2.175350701402806e-05,
+      "loss": 0.0186,
+      "step": 2831
+    },
+    {
+      "epoch": 11.328,
+      "grad_norm": 0.9175853729248047,
+      "learning_rate": 2.1743486973947896e-05,
+      "loss": 0.018,
+      "step": 2832
+    },
+    {
+      "epoch": 11.332,
+      "grad_norm": 0.75944584608078,
+      "learning_rate": 2.1733466933867734e-05,
+      "loss": 0.0188,
+      "step": 2833
+    },
+    {
+      "epoch": 11.336,
+      "grad_norm": 0.7953121662139893,
+      "learning_rate": 2.172344689378758e-05,
+      "loss": 0.0164,
+      "step": 2834
+    },
+    {
+      "epoch": 11.34,
+      "grad_norm": 0.662932276725769,
+      "learning_rate": 2.1713426853707417e-05,
+      "loss": 0.016,
+      "step": 2835
+    },
+    {
+      "epoch": 11.344,
+      "grad_norm": 0.6539945602416992,
+      "learning_rate": 2.1703406813627255e-05,
+      "loss": 0.0145,
+      "step": 2836
+    },
+    {
+      "epoch": 11.348,
+      "grad_norm": 0.8975854516029358,
+      "learning_rate": 2.1693386773547096e-05,
+      "loss": 0.0186,
+      "step": 2837
+    },
+    {
+      "epoch": 11.352,
+      "grad_norm": 0.8039414882659912,
+      "learning_rate": 2.1683366733466934e-05,
+      "loss": 0.0185,
+      "step": 2838
+    },
+    {
+      "epoch": 11.356,
+      "grad_norm": 0.891394853591919,
+      "learning_rate": 2.1673346693386775e-05,
+      "loss": 0.018,
+      "step": 2839
+    },
+    {
+      "epoch": 11.36,
+      "grad_norm": 0.6643425226211548,
+      "learning_rate": 2.1663326653306616e-05,
+      "loss": 0.0151,
+      "step": 2840
+    },
+    {
+      "epoch": 11.364,
+      "grad_norm": 0.6176607012748718,
+      "learning_rate": 2.1653306613226454e-05,
+      "loss": 0.0145,
+      "step": 2841
+    },
+    {
+      "epoch": 11.368,
+      "grad_norm": 0.4472271203994751,
+      "learning_rate": 2.1643286573146292e-05,
+      "loss": 0.0092,
+      "step": 2842
+    },
+    {
+      "epoch": 11.372,
+      "grad_norm": 0.7300881743431091,
+      "learning_rate": 2.1633266533066133e-05,
+      "loss": 0.0182,
+      "step": 2843
+    },
+    {
+      "epoch": 11.376,
+      "grad_norm": 0.5038206577301025,
+      "learning_rate": 2.1623246492985975e-05,
+      "loss": 0.0171,
+      "step": 2844
+    },
+    {
+      "epoch": 11.38,
+      "grad_norm": 0.8443142175674438,
+      "learning_rate": 2.1613226452905813e-05,
+      "loss": 0.0155,
+      "step": 2845
+    },
+    {
+      "epoch": 11.384,
+      "grad_norm": 0.7423885464668274,
+      "learning_rate": 2.1603206412825654e-05,
+      "loss": 0.0155,
+      "step": 2846
+    },
+    {
+      "epoch": 11.388,
+      "grad_norm": 0.6411489248275757,
+      "learning_rate": 2.1593186372745492e-05,
+      "loss": 0.0157,
+      "step": 2847
+    },
+    {
+      "epoch": 11.392,
+      "grad_norm": 0.7509790062904358,
+      "learning_rate": 2.158316633266533e-05,
+      "loss": 0.0163,
+      "step": 2848
+    },
+    {
+      "epoch": 11.396,
+      "grad_norm": 0.7018547058105469,
+      "learning_rate": 2.157314629258517e-05,
+      "loss": 0.019,
+      "step": 2849
+    },
+    {
+      "epoch": 11.4,
+      "grad_norm": 0.8679090738296509,
+      "learning_rate": 2.1563126252505012e-05,
+      "loss": 0.0145,
+      "step": 2850
+    },
+    {
+      "epoch": 11.404,
+      "grad_norm": 0.8156054019927979,
+      "learning_rate": 2.155310621242485e-05,
+      "loss": 0.0157,
+      "step": 2851
+    },
+    {
+      "epoch": 11.408,
+      "grad_norm": 0.7699236869812012,
+      "learning_rate": 2.1543086172344688e-05,
+      "loss": 0.0169,
+      "step": 2852
+    },
+    {
+      "epoch": 11.412,
+      "grad_norm": 0.9097781777381897,
+      "learning_rate": 2.153306613226453e-05,
+      "loss": 0.014,
+      "step": 2853
+    },
+    {
+      "epoch": 11.416,
+      "grad_norm": 0.3984467387199402,
+      "learning_rate": 2.152304609218437e-05,
+      "loss": 0.0126,
+      "step": 2854
+    },
+    {
+      "epoch": 11.42,
+      "grad_norm": 0.8444249033927917,
+      "learning_rate": 2.151302605210421e-05,
+      "loss": 0.0182,
+      "step": 2855
+    },
+    {
+      "epoch": 11.424,
+      "grad_norm": 0.9840474724769592,
+      "learning_rate": 2.150300601202405e-05,
+      "loss": 0.0199,
+      "step": 2856
+    },
+    {
+      "epoch": 11.428,
+      "grad_norm": 0.778589129447937,
+      "learning_rate": 2.1492985971943888e-05,
+      "loss": 0.015,
+      "step": 2857
+    },
+    {
+      "epoch": 11.432,
+      "grad_norm": 0.5163719654083252,
+      "learning_rate": 2.1482965931863726e-05,
+      "loss": 0.0148,
+      "step": 2858
+    },
+    {
+      "epoch": 11.436,
+      "grad_norm": 0.6548240780830383,
+      "learning_rate": 2.147294589178357e-05,
+      "loss": 0.0159,
+      "step": 2859
+    },
+    {
+      "epoch": 11.44,
+      "grad_norm": 0.6455692648887634,
+      "learning_rate": 2.1462925851703408e-05,
+      "loss": 0.0169,
+      "step": 2860
+    },
+    {
+      "epoch": 11.444,
+      "grad_norm": 0.7446917295455933,
+      "learning_rate": 2.1452905811623246e-05,
+      "loss": 0.0189,
+      "step": 2861
+    },
+    {
+      "epoch": 11.448,
+      "grad_norm": 0.6257652640342712,
+      "learning_rate": 2.1442885771543087e-05,
+      "loss": 0.0153,
+      "step": 2862
+    },
+    {
+      "epoch": 11.452,
+      "grad_norm": 0.9588707685470581,
+      "learning_rate": 2.1432865731462925e-05,
+      "loss": 0.0209,
+      "step": 2863
+    },
+    {
+      "epoch": 11.456,
+      "grad_norm": 0.825225830078125,
+      "learning_rate": 2.1422845691382767e-05,
+      "loss": 0.0163,
+      "step": 2864
+    },
+    {
+      "epoch": 11.46,
+      "grad_norm": 0.8426507711410522,
+      "learning_rate": 2.1412825651302608e-05,
+      "loss": 0.0208,
+      "step": 2865
+    },
+    {
+      "epoch": 11.464,
+      "grad_norm": 0.6643405556678772,
+      "learning_rate": 2.1402805611222446e-05,
+      "loss": 0.0169,
+      "step": 2866
+    },
+    {
+      "epoch": 11.468,
+      "grad_norm": 0.604695737361908,
+      "learning_rate": 2.1392785571142284e-05,
+      "loss": 0.0145,
+      "step": 2867
+    },
+    {
+      "epoch": 11.472,
+      "grad_norm": 0.6502740383148193,
+      "learning_rate": 2.1382765531062128e-05,
+      "loss": 0.0152,
+      "step": 2868
+    },
+    {
+      "epoch": 11.475999999999999,
+      "grad_norm": 0.47215506434440613,
+      "learning_rate": 2.1372745490981966e-05,
+      "loss": 0.0089,
+      "step": 2869
+    },
+    {
+      "epoch": 11.48,
+      "grad_norm": 0.597215473651886,
+      "learning_rate": 2.1362725450901804e-05,
+      "loss": 0.0157,
+      "step": 2870
+    },
+    {
+      "epoch": 11.484,
+      "grad_norm": 0.5259137153625488,
+      "learning_rate": 2.1352705410821645e-05,
+      "loss": 0.0144,
+      "step": 2871
+    },
+    {
+      "epoch": 11.488,
+      "grad_norm": 0.680182695388794,
+      "learning_rate": 2.1342685370741483e-05,
+      "loss": 0.0183,
+      "step": 2872
+    },
+    {
+      "epoch": 11.492,
+      "grad_norm": 0.8447750210762024,
+      "learning_rate": 2.1332665330661324e-05,
+      "loss": 0.0199,
+      "step": 2873
+    },
+    {
+      "epoch": 11.496,
+      "grad_norm": 0.8049308657646179,
+      "learning_rate": 2.1322645290581166e-05,
+      "loss": 0.0175,
+      "step": 2874
+    },
+    {
+      "epoch": 11.5,
+      "grad_norm": 0.916851818561554,
+      "learning_rate": 2.1312625250501004e-05,
+      "loss": 0.0191,
+      "step": 2875
+    },
+    {
+      "epoch": 11.504,
+      "grad_norm": 0.7663381695747375,
+      "learning_rate": 2.130260521042084e-05,
+      "loss": 0.0162,
+      "step": 2876
+    },
+    {
+      "epoch": 11.508,
+      "grad_norm": 0.639718770980835,
+      "learning_rate": 2.1292585170340683e-05,
+      "loss": 0.0128,
+      "step": 2877
+    },
+    {
+      "epoch": 11.512,
+      "grad_norm": 0.7572237849235535,
+      "learning_rate": 2.1282565130260524e-05,
+      "loss": 0.0167,
+      "step": 2878
+    },
+    {
+      "epoch": 11.516,
+      "grad_norm": 0.8541759252548218,
+      "learning_rate": 2.1272545090180362e-05,
+      "loss": 0.0182,
+      "step": 2879
+    },
+    {
+      "epoch": 11.52,
+      "grad_norm": 0.8712204098701477,
+      "learning_rate": 2.12625250501002e-05,
+      "loss": 0.02,
+      "step": 2880
+    },
+    {
+      "epoch": 11.524000000000001,
+      "grad_norm": 0.7454221248626709,
+      "learning_rate": 2.125250501002004e-05,
+      "loss": 0.0155,
+      "step": 2881
+    },
+    {
+      "epoch": 11.528,
+      "grad_norm": 0.47309982776641846,
+      "learning_rate": 2.124248496993988e-05,
+      "loss": 0.0129,
+      "step": 2882
+    },
+    {
+      "epoch": 11.532,
+      "grad_norm": 1.0623902082443237,
+      "learning_rate": 2.123246492985972e-05,
+      "loss": 0.0201,
+      "step": 2883
+    },
+    {
+      "epoch": 11.536,
+      "grad_norm": 0.600472092628479,
+      "learning_rate": 2.122244488977956e-05,
+      "loss": 0.0149,
+      "step": 2884
+    },
+    {
+      "epoch": 11.54,
+      "grad_norm": 0.9830151200294495,
+      "learning_rate": 2.12124248496994e-05,
+      "loss": 0.0171,
+      "step": 2885
+    },
+    {
+      "epoch": 11.544,
+      "grad_norm": 0.49249526858329773,
+      "learning_rate": 2.1202404809619237e-05,
+      "loss": 0.0116,
+      "step": 2886
+    },
+    {
+      "epoch": 11.548,
+      "grad_norm": 0.7066675424575806,
+      "learning_rate": 2.119238476953908e-05,
+      "loss": 0.0167,
+      "step": 2887
+    },
+    {
+      "epoch": 11.552,
+      "grad_norm": 0.9759204983711243,
+      "learning_rate": 2.118236472945892e-05,
+      "loss": 0.0221,
+      "step": 2888
+    },
+    {
+      "epoch": 11.556000000000001,
+      "grad_norm": 0.6902356743812561,
+      "learning_rate": 2.1172344689378758e-05,
+      "loss": 0.0154,
+      "step": 2889
+    },
+    {
+      "epoch": 11.56,
+      "grad_norm": 0.7708073854446411,
+      "learning_rate": 2.11623246492986e-05,
+      "loss": 0.0142,
+      "step": 2890
+    },
+    {
+      "epoch": 11.564,
+      "grad_norm": 0.6646739840507507,
+      "learning_rate": 2.1152304609218437e-05,
+      "loss": 0.0132,
+      "step": 2891
+    },
+    {
+      "epoch": 11.568,
+      "grad_norm": 0.5692991614341736,
+      "learning_rate": 2.1142284569138275e-05,
+      "loss": 0.0158,
+      "step": 2892
+    },
+    {
+      "epoch": 11.572,
+      "grad_norm": 0.683018147945404,
+      "learning_rate": 2.113226452905812e-05,
+      "loss": 0.0158,
+      "step": 2893
+    },
+    {
+      "epoch": 11.576,
+      "grad_norm": 0.6154299974441528,
+      "learning_rate": 2.1122244488977958e-05,
+      "loss": 0.0185,
+      "step": 2894
+    },
+    {
+      "epoch": 11.58,
+      "grad_norm": 0.9322575926780701,
+      "learning_rate": 2.1112224448897795e-05,
+      "loss": 0.0197,
+      "step": 2895
+    },
+    {
+      "epoch": 11.584,
+      "grad_norm": 0.6337608098983765,
+      "learning_rate": 2.1102204408817637e-05,
+      "loss": 0.0163,
+      "step": 2896
+    },
+    {
+      "epoch": 11.588,
+      "grad_norm": 0.8087349534034729,
+      "learning_rate": 2.1092184368737475e-05,
+      "loss": 0.017,
+      "step": 2897
+    },
+    {
+      "epoch": 11.592,
+      "grad_norm": 0.5522025227546692,
+      "learning_rate": 2.1082164328657316e-05,
+      "loss": 0.0155,
+      "step": 2898
+    },
+    {
+      "epoch": 11.596,
+      "grad_norm": 0.5902281403541565,
+      "learning_rate": 2.1072144288577157e-05,
+      "loss": 0.017,
+      "step": 2899
+    },
+    {
+      "epoch": 11.6,
+      "grad_norm": 0.5093696713447571,
+      "learning_rate": 2.1062124248496995e-05,
+      "loss": 0.0154,
+      "step": 2900
+    },
+    {
+      "epoch": 11.604,
+      "grad_norm": 0.602093517780304,
+      "learning_rate": 2.1052104208416833e-05,
+      "loss": 0.0163,
+      "step": 2901
+    },
+    {
+      "epoch": 11.608,
+      "grad_norm": 0.4907231628894806,
+      "learning_rate": 2.1042084168336674e-05,
+      "loss": 0.0105,
+      "step": 2902
+    },
+    {
+      "epoch": 11.612,
+      "grad_norm": 0.6661096811294556,
+      "learning_rate": 2.1032064128256516e-05,
+      "loss": 0.0167,
+      "step": 2903
+    },
+    {
+      "epoch": 11.616,
+      "grad_norm": 1.0688780546188354,
+      "learning_rate": 2.1022044088176353e-05,
+      "loss": 0.021,
+      "step": 2904
+    },
+    {
+      "epoch": 11.62,
+      "grad_norm": 0.7456181645393372,
+      "learning_rate": 2.1012024048096195e-05,
+      "loss": 0.0168,
+      "step": 2905
+    },
+    {
+      "epoch": 11.624,
+      "grad_norm": 0.7473933696746826,
+      "learning_rate": 2.1002004008016033e-05,
+      "loss": 0.0162,
+      "step": 2906
+    },
+    {
+      "epoch": 11.628,
+      "grad_norm": 0.539610743522644,
+      "learning_rate": 2.099198396793587e-05,
+      "loss": 0.0148,
+      "step": 2907
+    },
+    {
+      "epoch": 11.632,
+      "grad_norm": 1.1146992444992065,
+      "learning_rate": 2.0981963927855712e-05,
+      "loss": 0.0239,
+      "step": 2908
+    },
+    {
+      "epoch": 11.636,
+      "grad_norm": 0.7290841341018677,
+      "learning_rate": 2.0971943887775553e-05,
+      "loss": 0.0145,
+      "step": 2909
+    },
+    {
+      "epoch": 11.64,
+      "grad_norm": 0.6614787578582764,
+      "learning_rate": 2.096192384769539e-05,
+      "loss": 0.0169,
+      "step": 2910
+    },
+    {
+      "epoch": 11.644,
+      "grad_norm": 0.8798813819885254,
+      "learning_rate": 2.0951903807615232e-05,
+      "loss": 0.0196,
+      "step": 2911
+    },
+    {
+      "epoch": 11.648,
+      "grad_norm": 0.5717126727104187,
+      "learning_rate": 2.094188376753507e-05,
+      "loss": 0.0155,
+      "step": 2912
+    },
+    {
+      "epoch": 11.652,
+      "grad_norm": 1.0631353855133057,
+      "learning_rate": 2.093186372745491e-05,
+      "loss": 0.0191,
+      "step": 2913
+    },
+    {
+      "epoch": 11.656,
+      "grad_norm": 0.7153030633926392,
+      "learning_rate": 2.092184368737475e-05,
+      "loss": 0.0173,
+      "step": 2914
+    },
+    {
+      "epoch": 11.66,
+      "grad_norm": 0.8468221426010132,
+      "learning_rate": 2.091182364729459e-05,
+      "loss": 0.0203,
+      "step": 2915
+    },
+    {
+      "epoch": 11.664,
+      "grad_norm": 0.6937319040298462,
+      "learning_rate": 2.090180360721443e-05,
+      "loss": 0.016,
+      "step": 2916
+    },
+    {
+      "epoch": 11.668,
+      "grad_norm": 0.7014555931091309,
+      "learning_rate": 2.089178356713427e-05,
+      "loss": 0.0207,
+      "step": 2917
+    },
+    {
+      "epoch": 11.672,
+      "grad_norm": 0.7516029477119446,
+      "learning_rate": 2.088176352705411e-05,
+      "loss": 0.016,
+      "step": 2918
+    },
+    {
+      "epoch": 11.676,
+      "grad_norm": 0.8219577074050903,
+      "learning_rate": 2.087174348697395e-05,
+      "loss": 0.0167,
+      "step": 2919
+    },
+    {
+      "epoch": 11.68,
+      "grad_norm": 0.5712000131607056,
+      "learning_rate": 2.0861723446893787e-05,
+      "loss": 0.0175,
+      "step": 2920
+    },
+    {
+      "epoch": 11.684,
+      "grad_norm": 0.9513866305351257,
+      "learning_rate": 2.0851703406813628e-05,
+      "loss": 0.0197,
+      "step": 2921
+    },
+    {
+      "epoch": 11.688,
+      "grad_norm": 0.7864239811897278,
+      "learning_rate": 2.0841683366733466e-05,
+      "loss": 0.0175,
+      "step": 2922
+    },
+    {
+      "epoch": 11.692,
+      "grad_norm": 0.7664267420768738,
+      "learning_rate": 2.0831663326653307e-05,
+      "loss": 0.0189,
+      "step": 2923
+    },
+    {
+      "epoch": 11.696,
+      "grad_norm": 0.6849053502082825,
+      "learning_rate": 2.082164328657315e-05,
+      "loss": 0.0175,
+      "step": 2924
+    },
+    {
+      "epoch": 11.7,
+      "grad_norm": 0.7375827431678772,
+      "learning_rate": 2.0811623246492986e-05,
+      "loss": 0.0177,
+      "step": 2925
+    },
+    {
+      "epoch": 11.704,
+      "grad_norm": 0.661365807056427,
+      "learning_rate": 2.0801603206412824e-05,
+      "loss": 0.0154,
+      "step": 2926
+    },
+    {
+      "epoch": 11.708,
+      "grad_norm": 0.5127745866775513,
+      "learning_rate": 2.079158316633267e-05,
+      "loss": 0.014,
+      "step": 2927
+    },
+    {
+      "epoch": 11.712,
+      "grad_norm": 0.6667503714561462,
+      "learning_rate": 2.0781563126252507e-05,
+      "loss": 0.0158,
+      "step": 2928
+    },
+    {
+      "epoch": 11.716,
+      "grad_norm": 0.6455088257789612,
+      "learning_rate": 2.0771543086172345e-05,
+      "loss": 0.0156,
+      "step": 2929
+    },
+    {
+      "epoch": 11.72,
+      "grad_norm": 0.683720588684082,
+      "learning_rate": 2.0761523046092186e-05,
+      "loss": 0.015,
+      "step": 2930
+    },
+    {
+      "epoch": 11.724,
+      "grad_norm": 0.6581544280052185,
+      "learning_rate": 2.0751503006012024e-05,
+      "loss": 0.0158,
+      "step": 2931
+    },
+    {
+      "epoch": 11.728,
+      "grad_norm": 0.5899677276611328,
+      "learning_rate": 2.0741482965931865e-05,
+      "loss": 0.0159,
+      "step": 2932
+    },
+    {
+      "epoch": 11.732,
+      "grad_norm": 0.6744598746299744,
+      "learning_rate": 2.0731462925851707e-05,
+      "loss": 0.0157,
+      "step": 2933
+    },
+    {
+      "epoch": 11.736,
+      "grad_norm": 0.8028687834739685,
+      "learning_rate": 2.0721442885771544e-05,
+      "loss": 0.0172,
+      "step": 2934
+    },
+    {
+      "epoch": 11.74,
+      "grad_norm": 0.8576624393463135,
+      "learning_rate": 2.0711422845691382e-05,
+      "loss": 0.0201,
+      "step": 2935
+    },
+    {
+      "epoch": 11.744,
+      "grad_norm": 0.8122572302818298,
+      "learning_rate": 2.0701402805611224e-05,
+      "loss": 0.0187,
+      "step": 2936
+    },
+    {
+      "epoch": 11.748,
+      "grad_norm": 0.5539607405662537,
+      "learning_rate": 2.0691382765531065e-05,
+      "loss": 0.0155,
+      "step": 2937
+    },
+    {
+      "epoch": 11.752,
+      "grad_norm": 0.6936144232749939,
+      "learning_rate": 2.0681362725450903e-05,
+      "loss": 0.017,
+      "step": 2938
+    },
+    {
+      "epoch": 11.756,
+      "grad_norm": 0.8226979970932007,
+      "learning_rate": 2.0671342685370744e-05,
+      "loss": 0.0196,
+      "step": 2939
+    },
+    {
+      "epoch": 11.76,
+      "grad_norm": 0.49609795212745667,
+      "learning_rate": 2.0661322645290582e-05,
+      "loss": 0.014,
+      "step": 2940
+    },
+    {
+      "epoch": 11.764,
+      "grad_norm": 0.8211017847061157,
+      "learning_rate": 2.065130260521042e-05,
+      "loss": 0.0176,
+      "step": 2941
+    },
+    {
+      "epoch": 11.768,
+      "grad_norm": 0.6408782601356506,
+      "learning_rate": 2.064128256513026e-05,
+      "loss": 0.0166,
+      "step": 2942
+    },
+    {
+      "epoch": 11.772,
+      "grad_norm": 0.721869945526123,
+      "learning_rate": 2.0631262525050102e-05,
+      "loss": 0.0168,
+      "step": 2943
+    },
+    {
+      "epoch": 11.776,
+      "grad_norm": 0.7406426668167114,
+      "learning_rate": 2.062124248496994e-05,
+      "loss": 0.0172,
+      "step": 2944
+    },
+    {
+      "epoch": 11.78,
+      "grad_norm": 0.9095814228057861,
+      "learning_rate": 2.061122244488978e-05,
+      "loss": 0.0199,
+      "step": 2945
+    },
+    {
+      "epoch": 11.784,
+      "grad_norm": 0.6619588136672974,
+      "learning_rate": 2.060120240480962e-05,
+      "loss": 0.0193,
+      "step": 2946
+    },
+    {
+      "epoch": 11.788,
+      "grad_norm": 0.8828310966491699,
+      "learning_rate": 2.059118236472946e-05,
+      "loss": 0.0172,
+      "step": 2947
+    },
+    {
+      "epoch": 11.792,
+      "grad_norm": 0.6653797626495361,
+      "learning_rate": 2.05811623246493e-05,
+      "loss": 0.0158,
+      "step": 2948
+    },
+    {
+      "epoch": 11.796,
+      "grad_norm": 0.78447425365448,
+      "learning_rate": 2.057114228456914e-05,
+      "loss": 0.016,
+      "step": 2949
+    },
+    {
+      "epoch": 11.8,
+      "grad_norm": 0.7116568088531494,
+      "learning_rate": 2.0561122244488978e-05,
+      "loss": 0.0183,
+      "step": 2950
+    },
+    {
+      "epoch": 11.804,
+      "grad_norm": 0.5957525372505188,
+      "learning_rate": 2.055110220440882e-05,
+      "loss": 0.0165,
+      "step": 2951
+    },
+    {
+      "epoch": 11.808,
+      "grad_norm": 0.6143118143081665,
+      "learning_rate": 2.054108216432866e-05,
+      "loss": 0.0168,
+      "step": 2952
+    },
+    {
+      "epoch": 11.812,
+      "grad_norm": 0.8501677513122559,
+      "learning_rate": 2.05310621242485e-05,
+      "loss": 0.0147,
+      "step": 2953
+    },
+    {
+      "epoch": 11.816,
+      "grad_norm": 0.8333978056907654,
+      "learning_rate": 2.0521042084168336e-05,
+      "loss": 0.0178,
+      "step": 2954
+    },
+    {
+      "epoch": 11.82,
+      "grad_norm": 0.6845681667327881,
+      "learning_rate": 2.0511022044088178e-05,
+      "loss": 0.0186,
+      "step": 2955
+    },
+    {
+      "epoch": 11.824,
+      "grad_norm": 0.6612383127212524,
+      "learning_rate": 2.0501002004008015e-05,
+      "loss": 0.0161,
+      "step": 2956
+    },
+    {
+      "epoch": 11.828,
+      "grad_norm": 0.3920148015022278,
+      "learning_rate": 2.0490981963927857e-05,
+      "loss": 0.0082,
+      "step": 2957
+    },
+    {
+      "epoch": 11.832,
+      "grad_norm": 0.9177662134170532,
+      "learning_rate": 2.0480961923847698e-05,
+      "loss": 0.0208,
+      "step": 2958
+    },
+    {
+      "epoch": 11.836,
+      "grad_norm": 0.635342538356781,
+      "learning_rate": 2.0470941883767536e-05,
+      "loss": 0.0139,
+      "step": 2959
+    },
+    {
+      "epoch": 11.84,
+      "grad_norm": 0.9092418551445007,
+      "learning_rate": 2.0460921843687374e-05,
+      "loss": 0.0228,
+      "step": 2960
+    },
+    {
+      "epoch": 11.844,
+      "grad_norm": 0.7522537112236023,
+      "learning_rate": 2.0450901803607215e-05,
+      "loss": 0.0165,
+      "step": 2961
+    },
+    {
+      "epoch": 11.848,
+      "grad_norm": 0.8673317432403564,
+      "learning_rate": 2.0440881763527056e-05,
+      "loss": 0.0196,
+      "step": 2962
+    },
+    {
+      "epoch": 11.852,
+      "grad_norm": 0.5442277193069458,
+      "learning_rate": 2.0430861723446894e-05,
+      "loss": 0.015,
+      "step": 2963
+    },
+    {
+      "epoch": 11.856,
+      "grad_norm": 0.8119293451309204,
+      "learning_rate": 2.0420841683366736e-05,
+      "loss": 0.0189,
+      "step": 2964
+    },
+    {
+      "epoch": 11.86,
+      "grad_norm": 0.8945901989936829,
+      "learning_rate": 2.0410821643286573e-05,
+      "loss": 0.0165,
+      "step": 2965
+    },
+    {
+      "epoch": 11.864,
+      "grad_norm": 0.6969497203826904,
+      "learning_rate": 2.040080160320641e-05,
+      "loss": 0.0157,
+      "step": 2966
+    },
+    {
+      "epoch": 11.868,
+      "grad_norm": 0.700520396232605,
+      "learning_rate": 2.0390781563126256e-05,
+      "loss": 0.0168,
+      "step": 2967
+    },
+    {
+      "epoch": 11.872,
+      "grad_norm": 0.679799497127533,
+      "learning_rate": 2.0380761523046094e-05,
+      "loss": 0.016,
+      "step": 2968
+    },
+    {
+      "epoch": 11.876,
+      "grad_norm": 0.6355128884315491,
+      "learning_rate": 2.0370741482965932e-05,
+      "loss": 0.0156,
+      "step": 2969
+    },
+    {
+      "epoch": 11.88,
+      "grad_norm": 0.7955642342567444,
+      "learning_rate": 2.0360721442885773e-05,
+      "loss": 0.0174,
+      "step": 2970
+    },
+    {
+      "epoch": 11.884,
+      "grad_norm": 0.8534349203109741,
+      "learning_rate": 2.035070140280561e-05,
+      "loss": 0.0172,
+      "step": 2971
+    },
+    {
+      "epoch": 11.888,
+      "grad_norm": 0.9360595941543579,
+      "learning_rate": 2.0340681362725452e-05,
+      "loss": 0.0177,
+      "step": 2972
+    },
+    {
+      "epoch": 11.892,
+      "grad_norm": 0.7169991731643677,
+      "learning_rate": 2.0330661322645294e-05,
+      "loss": 0.0197,
+      "step": 2973
+    },
+    {
+      "epoch": 11.896,
+      "grad_norm": 0.853161096572876,
+      "learning_rate": 2.032064128256513e-05,
+      "loss": 0.0197,
+      "step": 2974
+    },
+    {
+      "epoch": 11.9,
+      "grad_norm": 0.714497447013855,
+      "learning_rate": 2.031062124248497e-05,
+      "loss": 0.0158,
+      "step": 2975
+    },
+    {
+      "epoch": 11.904,
+      "grad_norm": 0.8602288365364075,
+      "learning_rate": 2.030060120240481e-05,
+      "loss": 0.0176,
+      "step": 2976
+    },
+    {
+      "epoch": 11.908,
+      "grad_norm": 0.7017549276351929,
+      "learning_rate": 2.0290581162324652e-05,
+      "loss": 0.0171,
+      "step": 2977
+    },
+    {
+      "epoch": 11.912,
+      "grad_norm": 0.7314296960830688,
+      "learning_rate": 2.028056112224449e-05,
+      "loss": 0.0177,
+      "step": 2978
+    },
+    {
+      "epoch": 11.916,
+      "grad_norm": 0.5251822471618652,
+      "learning_rate": 2.0270541082164328e-05,
+      "loss": 0.0145,
+      "step": 2979
+    },
+    {
+      "epoch": 11.92,
+      "grad_norm": 0.5692286491394043,
+      "learning_rate": 2.026052104208417e-05,
+      "loss": 0.0152,
+      "step": 2980
+    },
+    {
+      "epoch": 11.924,
+      "grad_norm": 0.9542722105979919,
+      "learning_rate": 2.025050100200401e-05,
+      "loss": 0.019,
+      "step": 2981
+    },
+    {
+      "epoch": 11.928,
+      "grad_norm": 0.4988907277584076,
+      "learning_rate": 2.0240480961923848e-05,
+      "loss": 0.0148,
+      "step": 2982
+    },
+    {
+      "epoch": 11.932,
+      "grad_norm": 0.3393643796443939,
+      "learning_rate": 2.023046092184369e-05,
+      "loss": 0.0062,
+      "step": 2983
+    },
+    {
+      "epoch": 11.936,
+      "grad_norm": 0.6072190999984741,
+      "learning_rate": 2.0220440881763527e-05,
+      "loss": 0.0157,
+      "step": 2984
+    },
+    {
+      "epoch": 11.94,
+      "grad_norm": 0.7978100776672363,
+      "learning_rate": 2.0210420841683365e-05,
+      "loss": 0.0168,
+      "step": 2985
+    },
+    {
+      "epoch": 11.943999999999999,
+      "grad_norm": 0.6525113582611084,
+      "learning_rate": 2.020040080160321e-05,
+      "loss": 0.0162,
+      "step": 2986
+    },
+    {
+      "epoch": 11.948,
+      "grad_norm": 0.7721410989761353,
+      "learning_rate": 2.0190380761523048e-05,
+      "loss": 0.0168,
+      "step": 2987
+    },
+    {
+      "epoch": 11.952,
+      "grad_norm": 0.9068525433540344,
+      "learning_rate": 2.0180360721442886e-05,
+      "loss": 0.019,
+      "step": 2988
+    },
+    {
+      "epoch": 11.956,
+      "grad_norm": 0.6718014478683472,
+      "learning_rate": 2.0170340681362727e-05,
+      "loss": 0.0163,
+      "step": 2989
+    },
+    {
+      "epoch": 11.96,
+      "grad_norm": 0.8195679187774658,
+      "learning_rate": 2.0160320641282565e-05,
+      "loss": 0.0176,
+      "step": 2990
+    },
+    {
+      "epoch": 11.964,
+      "grad_norm": 0.6604552865028381,
+      "learning_rate": 2.0150300601202406e-05,
+      "loss": 0.0153,
+      "step": 2991
+    },
+    {
+      "epoch": 11.968,
+      "grad_norm": 0.7295999526977539,
+      "learning_rate": 2.0140280561122247e-05,
+      "loss": 0.0154,
+      "step": 2992
+    },
+    {
+      "epoch": 11.972,
+      "grad_norm": 0.9136268496513367,
+      "learning_rate": 2.0130260521042085e-05,
+      "loss": 0.0192,
+      "step": 2993
+    },
+    {
+      "epoch": 11.975999999999999,
+      "grad_norm": 0.8510957956314087,
+      "learning_rate": 2.0120240480961923e-05,
+      "loss": 0.0182,
+      "step": 2994
+    },
+    {
+      "epoch": 11.98,
+      "grad_norm": 0.9759835600852966,
+      "learning_rate": 2.0110220440881764e-05,
+      "loss": 0.0199,
+      "step": 2995
+    },
+    {
+      "epoch": 11.984,
+      "grad_norm": 0.5948774814605713,
+      "learning_rate": 2.0100200400801606e-05,
+      "loss": 0.0156,
+      "step": 2996
+    },
+    {
+      "epoch": 11.988,
+      "grad_norm": 0.6279386281967163,
+      "learning_rate": 2.0090180360721444e-05,
+      "loss": 0.0171,
+      "step": 2997
+    },
+    {
+      "epoch": 11.992,
+      "grad_norm": 0.4541473686695099,
+      "learning_rate": 2.0080160320641285e-05,
+      "loss": 0.0098,
+      "step": 2998
+    },
+    {
+      "epoch": 11.996,
+      "grad_norm": 0.7771435379981995,
+      "learning_rate": 2.0070140280561123e-05,
+      "loss": 0.0174,
+      "step": 2999
+    },
+    {
+      "epoch": 12.0,
+      "grad_norm": 0.8217807412147522,
+      "learning_rate": 2.006012024048096e-05,
+      "loss": 0.021,
+      "step": 3000
+    },
+    {
+      "epoch": 12.004,
+      "grad_norm": 0.6948625445365906,
+      "learning_rate": 2.0050100200400805e-05,
+      "loss": 0.0124,
+      "step": 3001
+    },
+    {
+      "epoch": 12.008,
+      "grad_norm": 0.4800888001918793,
+      "learning_rate": 2.0040080160320643e-05,
+      "loss": 0.0112,
+      "step": 3002
+    },
+    {
+      "epoch": 12.012,
+      "grad_norm": 0.5231291055679321,
+      "learning_rate": 2.003006012024048e-05,
+      "loss": 0.0113,
+      "step": 3003
+    },
+    {
+      "epoch": 12.016,
+      "grad_norm": 0.35470184683799744,
+      "learning_rate": 2.0020040080160322e-05,
+      "loss": 0.0113,
+      "step": 3004
+    },
+    {
+      "epoch": 12.02,
+      "grad_norm": 0.5849162936210632,
+      "learning_rate": 2.001002004008016e-05,
+      "loss": 0.0122,
+      "step": 3005
+    },
+    {
+      "epoch": 12.024,
+      "grad_norm": 0.5772752165794373,
+      "learning_rate": 2e-05,
+      "loss": 0.0126,
+      "step": 3006
+    },
+    {
+      "epoch": 12.028,
+      "grad_norm": 0.39507535099983215,
+      "learning_rate": 1.998997995991984e-05,
+      "loss": 0.0104,
+      "step": 3007
+    },
+    {
+      "epoch": 12.032,
+      "grad_norm": 0.5409855842590332,
+      "learning_rate": 1.997995991983968e-05,
+      "loss": 0.0125,
+      "step": 3008
+    },
+    {
+      "epoch": 12.036,
+      "grad_norm": 0.5285557508468628,
+      "learning_rate": 1.996993987975952e-05,
+      "loss": 0.014,
+      "step": 3009
+    },
+    {
+      "epoch": 12.04,
+      "grad_norm": 0.520379364490509,
+      "learning_rate": 1.995991983967936e-05,
+      "loss": 0.0131,
+      "step": 3010
+    },
+    {
+      "epoch": 12.044,
+      "grad_norm": 0.5537810325622559,
+      "learning_rate": 1.99498997995992e-05,
+      "loss": 0.0127,
+      "step": 3011
+    },
+    {
+      "epoch": 12.048,
+      "grad_norm": 0.4161956310272217,
+      "learning_rate": 1.993987975951904e-05,
+      "loss": 0.0123,
+      "step": 3012
+    },
+    {
+      "epoch": 12.052,
+      "grad_norm": 0.2786479890346527,
+      "learning_rate": 1.9929859719438877e-05,
+      "loss": 0.0063,
+      "step": 3013
+    },
+    {
+      "epoch": 12.056,
+      "grad_norm": 0.6354433298110962,
+      "learning_rate": 1.991983967935872e-05,
+      "loss": 0.0134,
+      "step": 3014
+    },
+    {
+      "epoch": 12.06,
+      "grad_norm": 0.4777648150920868,
+      "learning_rate": 1.9909819639278556e-05,
+      "loss": 0.0113,
+      "step": 3015
+    },
+    {
+      "epoch": 12.064,
+      "grad_norm": 0.622520923614502,
+      "learning_rate": 1.9899799599198398e-05,
+      "loss": 0.0121,
+      "step": 3016
+    },
+    {
+      "epoch": 12.068,
+      "grad_norm": 0.4313918650150299,
+      "learning_rate": 1.988977955911824e-05,
+      "loss": 0.0111,
+      "step": 3017
+    },
+    {
+      "epoch": 12.072,
+      "grad_norm": 0.35469359159469604,
+      "learning_rate": 1.9879759519038077e-05,
+      "loss": 0.0076,
+      "step": 3018
+    },
+    {
+      "epoch": 12.076,
+      "grad_norm": 0.6942519545555115,
+      "learning_rate": 1.9869739478957915e-05,
+      "loss": 0.0121,
+      "step": 3019
+    },
+    {
+      "epoch": 12.08,
+      "grad_norm": 0.5710741281509399,
+      "learning_rate": 1.9859719438877756e-05,
+      "loss": 0.0127,
+      "step": 3020
+    },
+    {
+      "epoch": 12.084,
+      "grad_norm": 0.3416895270347595,
+      "learning_rate": 1.9849699398797597e-05,
+      "loss": 0.0116,
+      "step": 3021
+    },
+    {
+      "epoch": 12.088,
+      "grad_norm": 0.5317337512969971,
+      "learning_rate": 1.9839679358717435e-05,
+      "loss": 0.0127,
+      "step": 3022
+    },
+    {
+      "epoch": 12.092,
+      "grad_norm": 0.45818451046943665,
+      "learning_rate": 1.9829659318637276e-05,
+      "loss": 0.012,
+      "step": 3023
+    },
+    {
+      "epoch": 12.096,
+      "grad_norm": 0.564237654209137,
+      "learning_rate": 1.9819639278557114e-05,
+      "loss": 0.0112,
+      "step": 3024
+    },
+    {
+      "epoch": 12.1,
+      "grad_norm": 0.5866971611976624,
+      "learning_rate": 1.9809619238476952e-05,
+      "loss": 0.0153,
+      "step": 3025
+    },
+    {
+      "epoch": 12.104,
+      "grad_norm": 0.44592806696891785,
+      "learning_rate": 1.9799599198396797e-05,
+      "loss": 0.0105,
+      "step": 3026
+    },
+    {
+      "epoch": 12.108,
+      "grad_norm": 0.3792635500431061,
+      "learning_rate": 1.9789579158316635e-05,
+      "loss": 0.0102,
+      "step": 3027
+    },
+    {
+      "epoch": 12.112,
+      "grad_norm": 0.5635868906974792,
+      "learning_rate": 1.9779559118236473e-05,
+      "loss": 0.0113,
+      "step": 3028
+    },
+    {
+      "epoch": 12.116,
+      "grad_norm": 0.6451796293258667,
+      "learning_rate": 1.9769539078156314e-05,
+      "loss": 0.012,
+      "step": 3029
+    },
+    {
+      "epoch": 12.12,
+      "grad_norm": 0.35301733016967773,
+      "learning_rate": 1.9759519038076152e-05,
+      "loss": 0.0103,
+      "step": 3030
+    },
+    {
+      "epoch": 12.124,
+      "grad_norm": 0.4701129198074341,
+      "learning_rate": 1.9749498997995993e-05,
+      "loss": 0.0113,
+      "step": 3031
+    },
+    {
+      "epoch": 12.128,
+      "grad_norm": 0.39844104647636414,
+      "learning_rate": 1.9739478957915834e-05,
+      "loss": 0.011,
+      "step": 3032
+    },
+    {
+      "epoch": 12.132,
+      "grad_norm": 0.5066357254981995,
+      "learning_rate": 1.9729458917835672e-05,
+      "loss": 0.0114,
+      "step": 3033
+    },
+    {
+      "epoch": 12.136,
+      "grad_norm": 0.3592749536037445,
+      "learning_rate": 1.971943887775551e-05,
+      "loss": 0.0109,
+      "step": 3034
+    },
+    {
+      "epoch": 12.14,
+      "grad_norm": 0.5839434266090393,
+      "learning_rate": 1.970941883767535e-05,
+      "loss": 0.0137,
+      "step": 3035
+    },
+    {
+      "epoch": 12.144,
+      "grad_norm": 0.35629791021347046,
+      "learning_rate": 1.9699398797595193e-05,
+      "loss": 0.0108,
+      "step": 3036
+    },
+    {
+      "epoch": 12.148,
+      "grad_norm": 0.474680095911026,
+      "learning_rate": 1.968937875751503e-05,
+      "loss": 0.0099,
+      "step": 3037
+    },
+    {
+      "epoch": 12.152,
+      "grad_norm": 0.4389362335205078,
+      "learning_rate": 1.9679358717434872e-05,
+      "loss": 0.0121,
+      "step": 3038
+    },
+    {
+      "epoch": 12.156,
+      "grad_norm": 0.5454397201538086,
+      "learning_rate": 1.966933867735471e-05,
+      "loss": 0.012,
+      "step": 3039
+    },
+    {
+      "epoch": 12.16,
+      "grad_norm": 0.4310266375541687,
+      "learning_rate": 1.965931863727455e-05,
+      "loss": 0.0124,
+      "step": 3040
+    },
+    {
+      "epoch": 12.164,
+      "grad_norm": 0.38319042325019836,
+      "learning_rate": 1.964929859719439e-05,
+      "loss": 0.0115,
+      "step": 3041
+    },
+    {
+      "epoch": 12.168,
+      "grad_norm": 0.5554765462875366,
+      "learning_rate": 1.963927855711423e-05,
+      "loss": 0.0123,
+      "step": 3042
+    },
+    {
+      "epoch": 12.172,
+      "grad_norm": 0.43762677907943726,
+      "learning_rate": 1.9629258517034068e-05,
+      "loss": 0.0126,
+      "step": 3043
+    },
+    {
+      "epoch": 12.176,
+      "grad_norm": 0.6405286192893982,
+      "learning_rate": 1.961923847695391e-05,
+      "loss": 0.0109,
+      "step": 3044
+    },
+    {
+      "epoch": 12.18,
+      "grad_norm": 0.4233160614967346,
+      "learning_rate": 1.960921843687375e-05,
+      "loss": 0.0131,
+      "step": 3045
+    },
+    {
+      "epoch": 12.184,
+      "grad_norm": 0.4447900652885437,
+      "learning_rate": 1.959919839679359e-05,
+      "loss": 0.012,
+      "step": 3046
+    },
+    {
+      "epoch": 12.188,
+      "grad_norm": 0.44814789295196533,
+      "learning_rate": 1.9589178356713426e-05,
+      "loss": 0.0134,
+      "step": 3047
+    },
+    {
+      "epoch": 12.192,
+      "grad_norm": 0.3883339762687683,
+      "learning_rate": 1.9579158316633268e-05,
+      "loss": 0.0127,
+      "step": 3048
+    },
+    {
+      "epoch": 12.196,
+      "grad_norm": 0.476978063583374,
+      "learning_rate": 1.9569138276553106e-05,
+      "loss": 0.0114,
+      "step": 3049
+    },
+    {
+      "epoch": 12.2,
+      "grad_norm": 0.41597089171409607,
+      "learning_rate": 1.9559118236472947e-05,
+      "loss": 0.0108,
+      "step": 3050
+    },
+    {
+      "epoch": 12.204,
+      "grad_norm": 0.3592725992202759,
+      "learning_rate": 1.9549098196392788e-05,
+      "loss": 0.0114,
+      "step": 3051
+    },
+    {
+      "epoch": 12.208,
+      "grad_norm": 0.5016531348228455,
+      "learning_rate": 1.9539078156312626e-05,
+      "loss": 0.0128,
+      "step": 3052
+    },
+    {
+      "epoch": 12.212,
+      "grad_norm": 0.46499642729759216,
+      "learning_rate": 1.9529058116232464e-05,
+      "loss": 0.01,
+      "step": 3053
+    },
+    {
+      "epoch": 12.216,
+      "grad_norm": 0.6043004989624023,
+      "learning_rate": 1.9519038076152305e-05,
+      "loss": 0.0102,
+      "step": 3054
+    },
+    {
+      "epoch": 12.22,
+      "grad_norm": 0.48706644773483276,
+      "learning_rate": 1.9509018036072147e-05,
+      "loss": 0.0126,
+      "step": 3055
+    },
+    {
+      "epoch": 12.224,
+      "grad_norm": 0.7386260032653809,
+      "learning_rate": 1.9498997995991984e-05,
+      "loss": 0.0117,
+      "step": 3056
+    },
+    {
+      "epoch": 12.228,
+      "grad_norm": 0.6791260838508606,
+      "learning_rate": 1.9488977955911826e-05,
+      "loss": 0.0133,
+      "step": 3057
+    },
+    {
+      "epoch": 12.232,
+      "grad_norm": 0.38665708899497986,
+      "learning_rate": 1.9478957915831664e-05,
+      "loss": 0.012,
+      "step": 3058
+    },
+    {
+      "epoch": 12.236,
+      "grad_norm": 0.6571320295333862,
+      "learning_rate": 1.94689378757515e-05,
+      "loss": 0.0113,
+      "step": 3059
+    },
+    {
+      "epoch": 12.24,
+      "grad_norm": 0.5152968168258667,
+      "learning_rate": 1.9458917835671346e-05,
+      "loss": 0.0121,
+      "step": 3060
+    },
+    {
+      "epoch": 12.244,
+      "grad_norm": 1.3934143781661987,
+      "learning_rate": 1.9448897795591184e-05,
+      "loss": 0.0139,
+      "step": 3061
+    },
+    {
+      "epoch": 12.248,
+      "grad_norm": 0.47328341007232666,
+      "learning_rate": 1.9438877755511022e-05,
+      "loss": 0.0149,
+      "step": 3062
+    },
+    {
+      "epoch": 12.252,
+      "grad_norm": 0.24610672891139984,
+      "learning_rate": 1.9428857715430863e-05,
+      "loss": 0.0089,
+      "step": 3063
+    },
+    {
+      "epoch": 12.256,
+      "grad_norm": 0.6528359055519104,
+      "learning_rate": 1.94188376753507e-05,
+      "loss": 0.0132,
+      "step": 3064
+    },
+    {
+      "epoch": 12.26,
+      "grad_norm": 0.5463185906410217,
+      "learning_rate": 1.9408817635270542e-05,
+      "loss": 0.0133,
+      "step": 3065
+    },
+    {
+      "epoch": 12.264,
+      "grad_norm": 0.40765008330345154,
+      "learning_rate": 1.9398797595190384e-05,
+      "loss": 0.0109,
+      "step": 3066
+    },
+    {
+      "epoch": 12.268,
+      "grad_norm": 0.6481624841690063,
+      "learning_rate": 1.938877755511022e-05,
+      "loss": 0.0168,
+      "step": 3067
+    },
+    {
+      "epoch": 12.272,
+      "grad_norm": 0.4352094829082489,
+      "learning_rate": 1.937875751503006e-05,
+      "loss": 0.0118,
+      "step": 3068
+    },
+    {
+      "epoch": 12.276,
+      "grad_norm": 0.3659978210926056,
+      "learning_rate": 1.93687374749499e-05,
+      "loss": 0.0123,
+      "step": 3069
+    },
+    {
+      "epoch": 12.28,
+      "grad_norm": 0.5505935549736023,
+      "learning_rate": 1.9358717434869742e-05,
+      "loss": 0.0125,
+      "step": 3070
+    },
+    {
+      "epoch": 12.284,
+      "grad_norm": 0.5281729102134705,
+      "learning_rate": 1.934869739478958e-05,
+      "loss": 0.0158,
+      "step": 3071
+    },
+    {
+      "epoch": 12.288,
+      "grad_norm": 0.42845743894577026,
+      "learning_rate": 1.933867735470942e-05,
+      "loss": 0.0114,
+      "step": 3072
+    },
+    {
+      "epoch": 12.292,
+      "grad_norm": 0.3950668275356293,
+      "learning_rate": 1.932865731462926e-05,
+      "loss": 0.0115,
+      "step": 3073
+    },
+    {
+      "epoch": 12.296,
+      "grad_norm": 0.6833834052085876,
+      "learning_rate": 1.9318637274549097e-05,
+      "loss": 0.0145,
+      "step": 3074
+    },
+    {
+      "epoch": 12.3,
+      "grad_norm": 0.3856029808521271,
+      "learning_rate": 1.930861723446894e-05,
+      "loss": 0.0132,
+      "step": 3075
+    },
+    {
+      "epoch": 12.304,
+      "grad_norm": 0.4963999092578888,
+      "learning_rate": 1.929859719438878e-05,
+      "loss": 0.0123,
+      "step": 3076
+    },
+    {
+      "epoch": 12.308,
+      "grad_norm": 0.46567070484161377,
+      "learning_rate": 1.9288577154308618e-05,
+      "loss": 0.0106,
+      "step": 3077
+    },
+    {
+      "epoch": 12.312,
+      "grad_norm": 0.5482111573219299,
+      "learning_rate": 1.927855711422846e-05,
+      "loss": 0.0145,
+      "step": 3078
+    },
+    {
+      "epoch": 12.316,
+      "grad_norm": 0.5668717622756958,
+      "learning_rate": 1.9268537074148297e-05,
+      "loss": 0.0119,
+      "step": 3079
+    },
+    {
+      "epoch": 12.32,
+      "grad_norm": 0.4780943691730499,
+      "learning_rate": 1.9258517034068138e-05,
+      "loss": 0.0131,
+      "step": 3080
+    },
+    {
+      "epoch": 12.324,
+      "grad_norm": 0.5395445227622986,
+      "learning_rate": 1.9248496993987976e-05,
+      "loss": 0.0119,
+      "step": 3081
+    },
+    {
+      "epoch": 12.328,
+      "grad_norm": 0.346057265996933,
+      "learning_rate": 1.9238476953907817e-05,
+      "loss": 0.011,
+      "step": 3082
+    },
+    {
+      "epoch": 12.332,
+      "grad_norm": 0.37034234404563904,
+      "learning_rate": 1.9228456913827655e-05,
+      "loss": 0.0119,
+      "step": 3083
+    },
+    {
+      "epoch": 12.336,
+      "grad_norm": 0.3500770032405853,
+      "learning_rate": 1.9218436873747493e-05,
+      "loss": 0.0125,
+      "step": 3084
+    },
+    {
+      "epoch": 12.34,
+      "grad_norm": 0.6100295782089233,
+      "learning_rate": 1.9208416833667338e-05,
+      "loss": 0.0154,
+      "step": 3085
+    },
+    {
+      "epoch": 12.344,
+      "grad_norm": 0.5396653413772583,
+      "learning_rate": 1.9198396793587175e-05,
+      "loss": 0.0139,
+      "step": 3086
+    },
+    {
+      "epoch": 12.348,
+      "grad_norm": 0.5847299695014954,
+      "learning_rate": 1.9188376753507013e-05,
+      "loss": 0.0119,
+      "step": 3087
+    },
+    {
+      "epoch": 12.352,
+      "grad_norm": 0.38033461570739746,
+      "learning_rate": 1.9178356713426855e-05,
+      "loss": 0.012,
+      "step": 3088
+    },
+    {
+      "epoch": 12.356,
+      "grad_norm": 0.4676550030708313,
+      "learning_rate": 1.9168336673346693e-05,
+      "loss": 0.0131,
+      "step": 3089
+    },
+    {
+      "epoch": 12.36,
+      "grad_norm": 0.40676653385162354,
+      "learning_rate": 1.9158316633266534e-05,
+      "loss": 0.0128,
+      "step": 3090
+    },
+    {
+      "epoch": 12.364,
+      "grad_norm": 0.4821133017539978,
+      "learning_rate": 1.9148296593186375e-05,
+      "loss": 0.0127,
+      "step": 3091
+    },
+    {
+      "epoch": 12.368,
+      "grad_norm": 0.3716961145401001,
+      "learning_rate": 1.9138276553106213e-05,
+      "loss": 0.0123,
+      "step": 3092
+    },
+    {
+      "epoch": 12.372,
+      "grad_norm": 0.4660569131374359,
+      "learning_rate": 1.912825651302605e-05,
+      "loss": 0.0112,
+      "step": 3093
+    },
+    {
+      "epoch": 12.376,
+      "grad_norm": 0.579636812210083,
+      "learning_rate": 1.9118236472945896e-05,
+      "loss": 0.0129,
+      "step": 3094
+    },
+    {
+      "epoch": 12.38,
+      "grad_norm": 0.373367577791214,
+      "learning_rate": 1.9108216432865733e-05,
+      "loss": 0.0112,
+      "step": 3095
+    },
+    {
+      "epoch": 12.384,
+      "grad_norm": 0.41835731267929077,
+      "learning_rate": 1.909819639278557e-05,
+      "loss": 0.0079,
+      "step": 3096
+    },
+    {
+      "epoch": 12.388,
+      "grad_norm": 0.43698638677597046,
+      "learning_rate": 1.9088176352705413e-05,
+      "loss": 0.0112,
+      "step": 3097
+    },
+    {
+      "epoch": 12.392,
+      "grad_norm": 0.7125802636146545,
+      "learning_rate": 1.907815631262525e-05,
+      "loss": 0.0122,
+      "step": 3098
+    },
+    {
+      "epoch": 12.396,
+      "grad_norm": 0.6462778449058533,
+      "learning_rate": 1.9068136272545092e-05,
+      "loss": 0.0124,
+      "step": 3099
+    },
+    {
+      "epoch": 12.4,
+      "grad_norm": 0.3642807900905609,
+      "learning_rate": 1.9058116232464933e-05,
+      "loss": 0.0109,
+      "step": 3100
+    },
+    {
+      "epoch": 12.404,
+      "grad_norm": 0.44956448674201965,
+      "learning_rate": 1.904809619238477e-05,
+      "loss": 0.0127,
+      "step": 3101
+    },
+    {
+      "epoch": 12.408,
+      "grad_norm": 0.398598849773407,
+      "learning_rate": 1.903807615230461e-05,
+      "loss": 0.0076,
+      "step": 3102
+    },
+    {
+      "epoch": 12.412,
+      "grad_norm": 0.8257546424865723,
+      "learning_rate": 1.902805611222445e-05,
+      "loss": 0.0131,
+      "step": 3103
+    },
+    {
+      "epoch": 12.416,
+      "grad_norm": 0.5510148406028748,
+      "learning_rate": 1.901803607214429e-05,
+      "loss": 0.0134,
+      "step": 3104
+    },
+    {
+      "epoch": 12.42,
+      "grad_norm": 0.415303111076355,
+      "learning_rate": 1.900801603206413e-05,
+      "loss": 0.0125,
+      "step": 3105
+    },
+    {
+      "epoch": 12.424,
+      "grad_norm": 0.36105599999427795,
+      "learning_rate": 1.8997995991983967e-05,
+      "loss": 0.0126,
+      "step": 3106
+    },
+    {
+      "epoch": 12.428,
+      "grad_norm": 0.45794200897216797,
+      "learning_rate": 1.898797595190381e-05,
+      "loss": 0.0139,
+      "step": 3107
+    },
+    {
+      "epoch": 12.432,
+      "grad_norm": 0.5848199129104614,
+      "learning_rate": 1.8977955911823646e-05,
+      "loss": 0.0119,
+      "step": 3108
+    },
+    {
+      "epoch": 12.436,
+      "grad_norm": 0.528476893901825,
+      "learning_rate": 1.8967935871743488e-05,
+      "loss": 0.013,
+      "step": 3109
+    },
+    {
+      "epoch": 12.44,
+      "grad_norm": 0.43130603432655334,
+      "learning_rate": 1.895791583166333e-05,
+      "loss": 0.0125,
+      "step": 3110
+    },
+    {
+      "epoch": 12.444,
+      "grad_norm": 0.5508957505226135,
+      "learning_rate": 1.8947895791583167e-05,
+      "loss": 0.014,
+      "step": 3111
+    },
+    {
+      "epoch": 12.448,
+      "grad_norm": 0.5716979503631592,
+      "learning_rate": 1.8937875751503005e-05,
+      "loss": 0.0123,
+      "step": 3112
+    },
+    {
+      "epoch": 12.452,
+      "grad_norm": 0.5015970468521118,
+      "learning_rate": 1.8927855711422846e-05,
+      "loss": 0.0122,
+      "step": 3113
+    },
+    {
+      "epoch": 12.456,
+      "grad_norm": 0.379427045583725,
+      "learning_rate": 1.8917835671342687e-05,
+      "loss": 0.0117,
+      "step": 3114
+    },
+    {
+      "epoch": 12.46,
+      "grad_norm": 0.3971169888973236,
+      "learning_rate": 1.8907815631262525e-05,
+      "loss": 0.0131,
+      "step": 3115
+    },
+    {
+      "epoch": 12.464,
+      "grad_norm": 0.540484607219696,
+      "learning_rate": 1.8897795591182367e-05,
+      "loss": 0.012,
+      "step": 3116
+    },
+    {
+      "epoch": 12.468,
+      "grad_norm": 0.6403137445449829,
+      "learning_rate": 1.8887775551102204e-05,
+      "loss": 0.0119,
+      "step": 3117
+    },
+    {
+      "epoch": 12.472,
+      "grad_norm": 0.4189258813858032,
+      "learning_rate": 1.8877755511022042e-05,
+      "loss": 0.0121,
+      "step": 3118
+    },
+    {
+      "epoch": 12.475999999999999,
+      "grad_norm": 0.731610119342804,
+      "learning_rate": 1.8867735470941887e-05,
+      "loss": 0.0126,
+      "step": 3119
+    },
+    {
+      "epoch": 12.48,
+      "grad_norm": 0.5180526375770569,
+      "learning_rate": 1.8857715430861725e-05,
+      "loss": 0.0113,
+      "step": 3120
+    },
+    {
+      "epoch": 12.484,
+      "grad_norm": 0.45408037304878235,
+      "learning_rate": 1.8847695390781563e-05,
+      "loss": 0.0114,
+      "step": 3121
+    },
+    {
+      "epoch": 12.488,
+      "grad_norm": 0.5250622034072876,
+      "learning_rate": 1.8837675350701404e-05,
+      "loss": 0.0125,
+      "step": 3122
+    },
+    {
+      "epoch": 12.492,
+      "grad_norm": 0.41708803176879883,
+      "learning_rate": 1.8827655310621242e-05,
+      "loss": 0.0108,
+      "step": 3123
+    },
+    {
+      "epoch": 12.496,
+      "grad_norm": 0.4630829989910126,
+      "learning_rate": 1.8817635270541083e-05,
+      "loss": 0.014,
+      "step": 3124
+    },
+    {
+      "epoch": 12.5,
+      "grad_norm": 0.5149716138839722,
+      "learning_rate": 1.8807615230460925e-05,
+      "loss": 0.0145,
+      "step": 3125
+    },
+    {
+      "epoch": 12.504,
+      "grad_norm": 0.4337029457092285,
+      "learning_rate": 1.8797595190380762e-05,
+      "loss": 0.0129,
+      "step": 3126
+    },
+    {
+      "epoch": 12.508,
+      "grad_norm": 0.5135932564735413,
+      "learning_rate": 1.87875751503006e-05,
+      "loss": 0.0123,
+      "step": 3127
+    },
+    {
+      "epoch": 12.512,
+      "grad_norm": 0.5922970771789551,
+      "learning_rate": 1.877755511022044e-05,
+      "loss": 0.0142,
+      "step": 3128
+    },
+    {
+      "epoch": 12.516,
+      "grad_norm": 0.6139187216758728,
+      "learning_rate": 1.8767535070140283e-05,
+      "loss": 0.0155,
+      "step": 3129
+    },
+    {
+      "epoch": 12.52,
+      "grad_norm": 0.34099677205085754,
+      "learning_rate": 1.875751503006012e-05,
+      "loss": 0.0114,
+      "step": 3130
+    },
+    {
+      "epoch": 12.524000000000001,
+      "grad_norm": 0.35690101981163025,
+      "learning_rate": 1.8747494989979962e-05,
+      "loss": 0.0115,
+      "step": 3131
+    },
+    {
+      "epoch": 12.528,
+      "grad_norm": 0.4974209666252136,
+      "learning_rate": 1.87374749498998e-05,
+      "loss": 0.0166,
+      "step": 3132
+    },
+    {
+      "epoch": 12.532,
+      "grad_norm": 0.47665518522262573,
+      "learning_rate": 1.8727454909819638e-05,
+      "loss": 0.0135,
+      "step": 3133
+    },
+    {
+      "epoch": 12.536,
+      "grad_norm": 0.45617246627807617,
+      "learning_rate": 1.871743486973948e-05,
+      "loss": 0.0135,
+      "step": 3134
+    },
+    {
+      "epoch": 12.54,
+      "grad_norm": 0.4416828453540802,
+      "learning_rate": 1.870741482965932e-05,
+      "loss": 0.0118,
+      "step": 3135
+    },
+    {
+      "epoch": 12.544,
+      "grad_norm": 0.4680028259754181,
+      "learning_rate": 1.869739478957916e-05,
+      "loss": 0.0128,
+      "step": 3136
+    },
+    {
+      "epoch": 12.548,
+      "grad_norm": 0.4106837809085846,
+      "learning_rate": 1.8687374749499e-05,
+      "loss": 0.0121,
+      "step": 3137
+    },
+    {
+      "epoch": 12.552,
+      "grad_norm": 0.53117835521698,
+      "learning_rate": 1.8677354709418837e-05,
+      "loss": 0.0121,
+      "step": 3138
+    },
+    {
+      "epoch": 12.556000000000001,
+      "grad_norm": 0.43182119727134705,
+      "learning_rate": 1.866733466933868e-05,
+      "loss": 0.0123,
+      "step": 3139
+    },
+    {
+      "epoch": 12.56,
+      "grad_norm": 0.7537881731987,
+      "learning_rate": 1.8657314629258517e-05,
+      "loss": 0.0145,
+      "step": 3140
+    },
+    {
+      "epoch": 12.564,
+      "grad_norm": 0.3404240906238556,
+      "learning_rate": 1.8647294589178358e-05,
+      "loss": 0.0115,
+      "step": 3141
+    },
+    {
+      "epoch": 12.568,
+      "grad_norm": 0.6614292860031128,
+      "learning_rate": 1.8637274549098196e-05,
+      "loss": 0.0117,
+      "step": 3142
+    },
+    {
+      "epoch": 12.572,
+      "grad_norm": 0.46803048253059387,
+      "learning_rate": 1.8627254509018037e-05,
+      "loss": 0.0117,
+      "step": 3143
+    },
+    {
+      "epoch": 12.576,
+      "grad_norm": 0.3678813874721527,
+      "learning_rate": 1.861723446893788e-05,
+      "loss": 0.0115,
+      "step": 3144
+    },
+    {
+      "epoch": 12.58,
+      "grad_norm": 0.30099377036094666,
+      "learning_rate": 1.8607214428857716e-05,
+      "loss": 0.0103,
+      "step": 3145
+    },
+    {
+      "epoch": 12.584,
+      "grad_norm": 0.47530147433280945,
+      "learning_rate": 1.8597194388777554e-05,
+      "loss": 0.013,
+      "step": 3146
+    },
+    {
+      "epoch": 12.588,
+      "grad_norm": 1.0010900497436523,
+      "learning_rate": 1.8587174348697395e-05,
+      "loss": 0.0121,
+      "step": 3147
+    },
+    {
+      "epoch": 12.592,
+      "grad_norm": 0.3169981837272644,
+      "learning_rate": 1.8577154308617237e-05,
+      "loss": 0.0113,
+      "step": 3148
+    },
+    {
+      "epoch": 12.596,
+      "grad_norm": 0.4307844936847687,
+      "learning_rate": 1.8567134268537075e-05,
+      "loss": 0.0123,
+      "step": 3149
+    },
+    {
+      "epoch": 12.6,
+      "grad_norm": 0.5654677748680115,
+      "learning_rate": 1.8557114228456916e-05,
+      "loss": 0.0135,
+      "step": 3150
+    },
+    {
+      "epoch": 12.604,
+      "grad_norm": 0.432781457901001,
+      "learning_rate": 1.8547094188376754e-05,
+      "loss": 0.0112,
+      "step": 3151
+    },
+    {
+      "epoch": 12.608,
+      "grad_norm": 0.6548603177070618,
+      "learning_rate": 1.8537074148296592e-05,
+      "loss": 0.013,
+      "step": 3152
+    },
+    {
+      "epoch": 12.612,
+      "grad_norm": 0.5078880786895752,
+      "learning_rate": 1.8527054108216436e-05,
+      "loss": 0.0114,
+      "step": 3153
+    },
+    {
+      "epoch": 12.616,
+      "grad_norm": 0.6445592641830444,
+      "learning_rate": 1.8517034068136274e-05,
+      "loss": 0.0132,
+      "step": 3154
+    },
+    {
+      "epoch": 12.62,
+      "grad_norm": 0.29496875405311584,
+      "learning_rate": 1.8507014028056112e-05,
+      "loss": 0.0109,
+      "step": 3155
+    },
+    {
+      "epoch": 12.624,
+      "grad_norm": 1.0399456024169922,
+      "learning_rate": 1.8496993987975953e-05,
+      "loss": 0.0156,
+      "step": 3156
+    },
+    {
+      "epoch": 12.628,
+      "grad_norm": 0.5184885263442993,
+      "learning_rate": 1.848697394789579e-05,
+      "loss": 0.011,
+      "step": 3157
+    },
+    {
+      "epoch": 12.632,
+      "grad_norm": 0.251708984375,
+      "learning_rate": 1.8476953907815633e-05,
+      "loss": 0.0064,
+      "step": 3158
+    },
+    {
+      "epoch": 12.636,
+      "grad_norm": 0.5945011377334595,
+      "learning_rate": 1.8466933867735474e-05,
+      "loss": 0.0151,
+      "step": 3159
+    },
+    {
+      "epoch": 12.64,
+      "grad_norm": 0.3488098382949829,
+      "learning_rate": 1.8456913827655312e-05,
+      "loss": 0.0112,
+      "step": 3160
+    },
+    {
+      "epoch": 12.644,
+      "grad_norm": 0.46110039949417114,
+      "learning_rate": 1.844689378757515e-05,
+      "loss": 0.0123,
+      "step": 3161
+    },
+    {
+      "epoch": 12.648,
+      "grad_norm": 0.5689302682876587,
+      "learning_rate": 1.843687374749499e-05,
+      "loss": 0.012,
+      "step": 3162
+    },
+    {
+      "epoch": 12.652,
+      "grad_norm": 0.4283682703971863,
+      "learning_rate": 1.8426853707414832e-05,
+      "loss": 0.0115,
+      "step": 3163
+    },
+    {
+      "epoch": 12.656,
+      "grad_norm": 0.5536277890205383,
+      "learning_rate": 1.841683366733467e-05,
+      "loss": 0.0108,
+      "step": 3164
+    },
+    {
+      "epoch": 12.66,
+      "grad_norm": 0.5680063366889954,
+      "learning_rate": 1.840681362725451e-05,
+      "loss": 0.014,
+      "step": 3165
+    },
+    {
+      "epoch": 12.664,
+      "grad_norm": 0.42657792568206787,
+      "learning_rate": 1.839679358717435e-05,
+      "loss": 0.013,
+      "step": 3166
+    },
+    {
+      "epoch": 12.668,
+      "grad_norm": 0.33542749285697937,
+      "learning_rate": 1.8386773547094187e-05,
+      "loss": 0.0123,
+      "step": 3167
+    },
+    {
+      "epoch": 12.672,
+      "grad_norm": 0.7784429788589478,
+      "learning_rate": 1.837675350701403e-05,
+      "loss": 0.0186,
+      "step": 3168
+    },
+    {
+      "epoch": 12.676,
+      "grad_norm": 0.35441887378692627,
+      "learning_rate": 1.836673346693387e-05,
+      "loss": 0.0113,
+      "step": 3169
+    },
+    {
+      "epoch": 12.68,
+      "grad_norm": 0.6247470378875732,
+      "learning_rate": 1.8356713426853708e-05,
+      "loss": 0.0117,
+      "step": 3170
+    },
+    {
+      "epoch": 12.684,
+      "grad_norm": 0.7267860770225525,
+      "learning_rate": 1.834669338677355e-05,
+      "loss": 0.0126,
+      "step": 3171
+    },
+    {
+      "epoch": 12.688,
+      "grad_norm": 0.40373706817626953,
+      "learning_rate": 1.8336673346693387e-05,
+      "loss": 0.0124,
+      "step": 3172
+    },
+    {
+      "epoch": 12.692,
+      "grad_norm": 0.38257259130477905,
+      "learning_rate": 1.8326653306613228e-05,
+      "loss": 0.0122,
+      "step": 3173
+    },
+    {
+      "epoch": 12.696,
+      "grad_norm": 0.3000936210155487,
+      "learning_rate": 1.8316633266533066e-05,
+      "loss": 0.0107,
+      "step": 3174
+    },
+    {
+      "epoch": 12.7,
+      "grad_norm": 0.5243247747421265,
+      "learning_rate": 1.8306613226452907e-05,
+      "loss": 0.012,
+      "step": 3175
+    },
+    {
+      "epoch": 12.704,
+      "grad_norm": 0.6114549040794373,
+      "learning_rate": 1.8296593186372745e-05,
+      "loss": 0.0106,
+      "step": 3176
+    },
+    {
+      "epoch": 12.708,
+      "grad_norm": 0.9472494721412659,
+      "learning_rate": 1.8286573146292587e-05,
+      "loss": 0.0164,
+      "step": 3177
+    },
+    {
+      "epoch": 12.712,
+      "grad_norm": 0.5240547060966492,
+      "learning_rate": 1.8276553106212428e-05,
+      "loss": 0.012,
+      "step": 3178
+    },
+    {
+      "epoch": 12.716,
+      "grad_norm": 0.6829943656921387,
+      "learning_rate": 1.8266533066132266e-05,
+      "loss": 0.0123,
+      "step": 3179
+    },
+    {
+      "epoch": 12.72,
+      "grad_norm": 0.4161919951438904,
+      "learning_rate": 1.8256513026052104e-05,
+      "loss": 0.0115,
+      "step": 3180
+    },
+    {
+      "epoch": 12.724,
+      "grad_norm": 0.2771739959716797,
+      "learning_rate": 1.8246492985971945e-05,
+      "loss": 0.0073,
+      "step": 3181
+    },
+    {
+      "epoch": 12.728,
+      "grad_norm": 0.40933287143707275,
+      "learning_rate": 1.8236472945891783e-05,
+      "loss": 0.0119,
+      "step": 3182
+    },
+    {
+      "epoch": 12.732,
+      "grad_norm": 0.44479766488075256,
+      "learning_rate": 1.8226452905811624e-05,
+      "loss": 0.0143,
+      "step": 3183
+    },
+    {
+      "epoch": 12.736,
+      "grad_norm": 0.32802703976631165,
+      "learning_rate": 1.8216432865731465e-05,
+      "loss": 0.0102,
+      "step": 3184
+    },
+    {
+      "epoch": 12.74,
+      "grad_norm": 0.43562448024749756,
+      "learning_rate": 1.8206412825651303e-05,
+      "loss": 0.0125,
+      "step": 3185
+    },
+    {
+      "epoch": 12.744,
+      "grad_norm": 0.2993567883968353,
+      "learning_rate": 1.819639278557114e-05,
+      "loss": 0.0064,
+      "step": 3186
+    },
+    {
+      "epoch": 12.748,
+      "grad_norm": 0.45686689019203186,
+      "learning_rate": 1.8186372745490982e-05,
+      "loss": 0.0137,
+      "step": 3187
+    },
+    {
+      "epoch": 12.752,
+      "grad_norm": 0.6637890934944153,
+      "learning_rate": 1.8176352705410824e-05,
+      "loss": 0.0117,
+      "step": 3188
+    },
+    {
+      "epoch": 12.756,
+      "grad_norm": 0.4745820462703705,
+      "learning_rate": 1.816633266533066e-05,
+      "loss": 0.0127,
+      "step": 3189
+    },
+    {
+      "epoch": 12.76,
+      "grad_norm": 0.5020652413368225,
+      "learning_rate": 1.8156312625250503e-05,
+      "loss": 0.0141,
+      "step": 3190
+    },
+    {
+      "epoch": 12.764,
+      "grad_norm": 0.49237433075904846,
+      "learning_rate": 1.814629258517034e-05,
+      "loss": 0.0151,
+      "step": 3191
+    },
+    {
+      "epoch": 12.768,
+      "grad_norm": 0.6075673699378967,
+      "learning_rate": 1.813627254509018e-05,
+      "loss": 0.0133,
+      "step": 3192
+    },
+    {
+      "epoch": 12.772,
+      "grad_norm": 0.5600119233131409,
+      "learning_rate": 1.8126252505010023e-05,
+      "loss": 0.0158,
+      "step": 3193
+    },
+    {
+      "epoch": 12.776,
+      "grad_norm": 0.4790187478065491,
+      "learning_rate": 1.811623246492986e-05,
+      "loss": 0.0146,
+      "step": 3194
+    },
+    {
+      "epoch": 12.78,
+      "grad_norm": 0.43495261669158936,
+      "learning_rate": 1.81062124248497e-05,
+      "loss": 0.0118,
+      "step": 3195
+    },
+    {
+      "epoch": 12.784,
+      "grad_norm": 0.36963024735450745,
+      "learning_rate": 1.809619238476954e-05,
+      "loss": 0.0116,
+      "step": 3196
+    },
+    {
+      "epoch": 12.788,
+      "grad_norm": 0.49386468529701233,
+      "learning_rate": 1.8086172344689378e-05,
+      "loss": 0.0144,
+      "step": 3197
+    },
+    {
+      "epoch": 12.792,
+      "grad_norm": 0.6671590805053711,
+      "learning_rate": 1.807615230460922e-05,
+      "loss": 0.013,
+      "step": 3198
+    },
+    {
+      "epoch": 12.796,
+      "grad_norm": 0.39627164602279663,
+      "learning_rate": 1.806613226452906e-05,
+      "loss": 0.0122,
+      "step": 3199
+    },
+    {
+      "epoch": 12.8,
+      "grad_norm": 0.5293060541152954,
+      "learning_rate": 1.80561122244489e-05,
+      "loss": 0.0139,
+      "step": 3200
+    },
+    {
+      "epoch": 12.804,
+      "grad_norm": 0.4009303152561188,
+      "learning_rate": 1.8046092184368737e-05,
+      "loss": 0.0131,
+      "step": 3201
+    },
+    {
+      "epoch": 12.808,
+      "grad_norm": 0.4425906240940094,
+      "learning_rate": 1.8036072144288578e-05,
+      "loss": 0.0124,
+      "step": 3202
+    },
+    {
+      "epoch": 12.812,
+      "grad_norm": 0.38386768102645874,
+      "learning_rate": 1.802605210420842e-05,
+      "loss": 0.0126,
+      "step": 3203
+    },
+    {
+      "epoch": 12.816,
+      "grad_norm": 0.48892340064048767,
+      "learning_rate": 1.8016032064128257e-05,
+      "loss": 0.0118,
+      "step": 3204
+    },
+    {
+      "epoch": 12.82,
+      "grad_norm": 0.44380101561546326,
+      "learning_rate": 1.80060120240481e-05,
+      "loss": 0.0129,
+      "step": 3205
+    },
+    {
+      "epoch": 12.824,
+      "grad_norm": 0.3459862172603607,
+      "learning_rate": 1.7995991983967936e-05,
+      "loss": 0.0119,
+      "step": 3206
+    },
+    {
+      "epoch": 12.828,
+      "grad_norm": 0.41011375188827515,
+      "learning_rate": 1.7985971943887778e-05,
+      "loss": 0.0119,
+      "step": 3207
+    },
+    {
+      "epoch": 12.832,
+      "grad_norm": 0.5159206986427307,
+      "learning_rate": 1.7975951903807615e-05,
+      "loss": 0.0167,
+      "step": 3208
+    },
+    {
+      "epoch": 12.836,
+      "grad_norm": 0.22304032742977142,
+      "learning_rate": 1.7965931863727457e-05,
+      "loss": 0.0066,
+      "step": 3209
+    },
+    {
+      "epoch": 12.84,
+      "grad_norm": 0.3644813597202301,
+      "learning_rate": 1.7955911823647295e-05,
+      "loss": 0.0084,
+      "step": 3210
+    },
+    {
+      "epoch": 12.844,
+      "grad_norm": 0.5323387384414673,
+      "learning_rate": 1.7945891783567133e-05,
+      "loss": 0.0146,
+      "step": 3211
+    },
+    {
+      "epoch": 12.848,
+      "grad_norm": 0.647765576839447,
+      "learning_rate": 1.7935871743486977e-05,
+      "loss": 0.0127,
+      "step": 3212
+    },
+    {
+      "epoch": 12.852,
+      "grad_norm": 0.4638051688671112,
+      "learning_rate": 1.7925851703406815e-05,
+      "loss": 0.0083,
+      "step": 3213
+    },
+    {
+      "epoch": 12.856,
+      "grad_norm": 0.7238399982452393,
+      "learning_rate": 1.7915831663326653e-05,
+      "loss": 0.0136,
+      "step": 3214
+    },
+    {
+      "epoch": 12.86,
+      "grad_norm": 0.7715800404548645,
+      "learning_rate": 1.7905811623246494e-05,
+      "loss": 0.0149,
+      "step": 3215
+    },
+    {
+      "epoch": 12.864,
+      "grad_norm": 0.7044159173965454,
+      "learning_rate": 1.7895791583166332e-05,
+      "loss": 0.0154,
+      "step": 3216
+    },
+    {
+      "epoch": 12.868,
+      "grad_norm": 0.45285147428512573,
+      "learning_rate": 1.7885771543086173e-05,
+      "loss": 0.0113,
+      "step": 3217
+    },
+    {
+      "epoch": 12.872,
+      "grad_norm": 0.42862364649772644,
+      "learning_rate": 1.7875751503006015e-05,
+      "loss": 0.0122,
+      "step": 3218
+    },
+    {
+      "epoch": 12.876,
+      "grad_norm": 0.4750185012817383,
+      "learning_rate": 1.7865731462925853e-05,
+      "loss": 0.0129,
+      "step": 3219
+    },
+    {
+      "epoch": 12.88,
+      "grad_norm": 0.4365493059158325,
+      "learning_rate": 1.785571142284569e-05,
+      "loss": 0.0143,
+      "step": 3220
+    },
+    {
+      "epoch": 12.884,
+      "grad_norm": 0.7560842633247375,
+      "learning_rate": 1.7845691382765532e-05,
+      "loss": 0.0162,
+      "step": 3221
+    },
+    {
+      "epoch": 12.888,
+      "grad_norm": 0.21909433603286743,
+      "learning_rate": 1.7835671342685373e-05,
+      "loss": 0.0064,
+      "step": 3222
+    },
+    {
+      "epoch": 12.892,
+      "grad_norm": 0.6783998608589172,
+      "learning_rate": 1.782565130260521e-05,
+      "loss": 0.0135,
+      "step": 3223
+    },
+    {
+      "epoch": 12.896,
+      "grad_norm": 0.3435244560241699,
+      "learning_rate": 1.7815631262525052e-05,
+      "loss": 0.0119,
+      "step": 3224
+    },
+    {
+      "epoch": 12.9,
+      "grad_norm": 0.20075704157352448,
+      "learning_rate": 1.780561122244489e-05,
+      "loss": 0.0071,
+      "step": 3225
+    },
+    {
+      "epoch": 12.904,
+      "grad_norm": 0.44066622853279114,
+      "learning_rate": 1.7795591182364728e-05,
+      "loss": 0.0124,
+      "step": 3226
+    },
+    {
+      "epoch": 12.908,
+      "grad_norm": 0.3443538248538971,
+      "learning_rate": 1.7785571142284573e-05,
+      "loss": 0.0116,
+      "step": 3227
+    },
+    {
+      "epoch": 12.912,
+      "grad_norm": 0.3867914378643036,
+      "learning_rate": 1.777555110220441e-05,
+      "loss": 0.0131,
+      "step": 3228
+    },
+    {
+      "epoch": 12.916,
+      "grad_norm": 0.44781914353370667,
+      "learning_rate": 1.776553106212425e-05,
+      "loss": 0.0138,
+      "step": 3229
+    },
+    {
+      "epoch": 12.92,
+      "grad_norm": 0.4673486649990082,
+      "learning_rate": 1.775551102204409e-05,
+      "loss": 0.0128,
+      "step": 3230
+    },
+    {
+      "epoch": 12.924,
+      "grad_norm": 0.8357963562011719,
+      "learning_rate": 1.7745490981963928e-05,
+      "loss": 0.0163,
+      "step": 3231
+    },
+    {
+      "epoch": 12.928,
+      "grad_norm": 0.52073073387146,
+      "learning_rate": 1.773547094188377e-05,
+      "loss": 0.0121,
+      "step": 3232
+    },
+    {
+      "epoch": 12.932,
+      "grad_norm": 0.620764970779419,
+      "learning_rate": 1.7725450901803607e-05,
+      "loss": 0.0132,
+      "step": 3233
+    },
+    {
+      "epoch": 12.936,
+      "grad_norm": 0.5402158498764038,
+      "learning_rate": 1.7715430861723448e-05,
+      "loss": 0.0147,
+      "step": 3234
+    },
+    {
+      "epoch": 12.94,
+      "grad_norm": 0.6234936118125916,
+      "learning_rate": 1.7705410821643286e-05,
+      "loss": 0.0142,
+      "step": 3235
+    },
+    {
+      "epoch": 12.943999999999999,
+      "grad_norm": 0.7595784664154053,
+      "learning_rate": 1.7695390781563127e-05,
+      "loss": 0.017,
+      "step": 3236
+    },
+    {
+      "epoch": 12.948,
+      "grad_norm": 0.7245892882347107,
+      "learning_rate": 1.768537074148297e-05,
+      "loss": 0.014,
+      "step": 3237
+    },
+    {
+      "epoch": 12.952,
+      "grad_norm": 0.6920550465583801,
+      "learning_rate": 1.7675350701402807e-05,
+      "loss": 0.0126,
+      "step": 3238
+    },
+    {
+      "epoch": 12.956,
+      "grad_norm": 0.6069607734680176,
+      "learning_rate": 1.7665330661322644e-05,
+      "loss": 0.014,
+      "step": 3239
+    },
+    {
+      "epoch": 12.96,
+      "grad_norm": 0.40507733821868896,
+      "learning_rate": 1.7655310621242486e-05,
+      "loss": 0.0138,
+      "step": 3240
+    },
+    {
+      "epoch": 12.964,
+      "grad_norm": 0.40632104873657227,
+      "learning_rate": 1.7645290581162324e-05,
+      "loss": 0.0138,
+      "step": 3241
+    },
+    {
+      "epoch": 12.968,
+      "grad_norm": 0.6928650140762329,
+      "learning_rate": 1.7635270541082165e-05,
+      "loss": 0.016,
+      "step": 3242
+    },
+    {
+      "epoch": 12.972,
+      "grad_norm": 0.5469399094581604,
+      "learning_rate": 1.7625250501002006e-05,
+      "loss": 0.0149,
+      "step": 3243
+    },
+    {
+      "epoch": 12.975999999999999,
+      "grad_norm": 0.44268378615379333,
+      "learning_rate": 1.7615230460921844e-05,
+      "loss": 0.013,
+      "step": 3244
+    },
+    {
+      "epoch": 12.98,
+      "grad_norm": 0.40536338090896606,
+      "learning_rate": 1.7605210420841682e-05,
+      "loss": 0.012,
+      "step": 3245
+    },
+    {
+      "epoch": 12.984,
+      "grad_norm": 0.4465670585632324,
+      "learning_rate": 1.7595190380761523e-05,
+      "loss": 0.0125,
+      "step": 3246
+    },
+    {
+      "epoch": 12.988,
+      "grad_norm": 0.9920943975448608,
+      "learning_rate": 1.7585170340681365e-05,
+      "loss": 0.014,
+      "step": 3247
+    },
+    {
+      "epoch": 12.992,
+      "grad_norm": 0.26608768105506897,
+      "learning_rate": 1.7575150300601202e-05,
+      "loss": 0.0072,
+      "step": 3248
+    },
+    {
+      "epoch": 12.996,
+      "grad_norm": 0.48054268956184387,
+      "learning_rate": 1.7565130260521044e-05,
+      "loss": 0.0135,
+      "step": 3249
+    },
+    {
+      "epoch": 13.0,
+      "grad_norm": 0.7327283024787903,
+      "learning_rate": 1.755511022044088e-05,
+      "loss": 0.0247,
+      "step": 3250
+    },
+    {
+      "epoch": 13.004,
+      "grad_norm": 0.27949610352516174,
+      "learning_rate": 1.754509018036072e-05,
+      "loss": 0.0092,
+      "step": 3251
+    },
+    {
+      "epoch": 13.008,
+      "grad_norm": 0.5924502015113831,
+      "learning_rate": 1.7535070140280564e-05,
+      "loss": 0.0124,
+      "step": 3252
+    },
+    {
+      "epoch": 13.012,
+      "grad_norm": 0.2856701612472534,
+      "learning_rate": 1.7525050100200402e-05,
+      "loss": 0.0088,
+      "step": 3253
+    },
+    {
+      "epoch": 13.016,
+      "grad_norm": 0.6892845034599304,
+      "learning_rate": 1.751503006012024e-05,
+      "loss": 0.0101,
+      "step": 3254
+    },
+    {
+      "epoch": 13.02,
+      "grad_norm": 0.28304752707481384,
+      "learning_rate": 1.750501002004008e-05,
+      "loss": 0.0093,
+      "step": 3255
+    },
+    {
+      "epoch": 13.024,
+      "grad_norm": 0.44429293274879456,
+      "learning_rate": 1.7494989979959922e-05,
+      "loss": 0.0106,
+      "step": 3256
+    },
+    {
+      "epoch": 13.028,
+      "grad_norm": 0.4208962917327881,
+      "learning_rate": 1.748496993987976e-05,
+      "loss": 0.011,
+      "step": 3257
+    },
+    {
+      "epoch": 13.032,
+      "grad_norm": 0.24816329777240753,
+      "learning_rate": 1.74749498997996e-05,
+      "loss": 0.0091,
+      "step": 3258
+    },
+    {
+      "epoch": 13.036,
+      "grad_norm": 0.3193182647228241,
+      "learning_rate": 1.746492985971944e-05,
+      "loss": 0.0109,
+      "step": 3259
+    },
+    {
+      "epoch": 13.04,
+      "grad_norm": 0.8636166453361511,
+      "learning_rate": 1.7454909819639277e-05,
+      "loss": 0.0097,
+      "step": 3260
+    },
+    {
+      "epoch": 13.044,
+      "grad_norm": 0.26683780550956726,
+      "learning_rate": 1.744488977955912e-05,
+      "loss": 0.0095,
+      "step": 3261
+    },
+    {
+      "epoch": 13.048,
+      "grad_norm": 0.5071774125099182,
+      "learning_rate": 1.743486973947896e-05,
+      "loss": 0.0126,
+      "step": 3262
+    },
+    {
+      "epoch": 13.052,
+      "grad_norm": 0.21876634657382965,
+      "learning_rate": 1.7424849699398798e-05,
+      "loss": 0.0094,
+      "step": 3263
+    },
+    {
+      "epoch": 13.056,
+      "grad_norm": 0.18555937707424164,
+      "learning_rate": 1.741482965931864e-05,
+      "loss": 0.0059,
+      "step": 3264
+    },
+    {
+      "epoch": 13.06,
+      "grad_norm": 0.2228597104549408,
+      "learning_rate": 1.7404809619238477e-05,
+      "loss": 0.0094,
+      "step": 3265
+    },
+    {
+      "epoch": 13.064,
+      "grad_norm": 0.21233904361724854,
+      "learning_rate": 1.739478957915832e-05,
+      "loss": 0.0082,
+      "step": 3266
+    },
+    {
+      "epoch": 13.068,
+      "grad_norm": 0.2386617809534073,
+      "learning_rate": 1.7384769539078156e-05,
+      "loss": 0.0094,
+      "step": 3267
+    },
+    {
+      "epoch": 13.072,
+      "grad_norm": 0.37460488080978394,
+      "learning_rate": 1.7374749498997998e-05,
+      "loss": 0.0108,
+      "step": 3268
+    },
+    {
+      "epoch": 13.076,
+      "grad_norm": 0.2160538136959076,
+      "learning_rate": 1.7364729458917835e-05,
+      "loss": 0.0094,
+      "step": 3269
+    },
+    {
+      "epoch": 13.08,
+      "grad_norm": 0.2701886296272278,
+      "learning_rate": 1.7354709418837677e-05,
+      "loss": 0.0065,
+      "step": 3270
+    },
+    {
+      "epoch": 13.084,
+      "grad_norm": 0.3855418264865875,
+      "learning_rate": 1.7344689378757518e-05,
+      "loss": 0.0096,
+      "step": 3271
+    },
+    {
+      "epoch": 13.088,
+      "grad_norm": 0.31014111638069153,
+      "learning_rate": 1.7334669338677356e-05,
+      "loss": 0.0095,
+      "step": 3272
+    },
+    {
+      "epoch": 13.092,
+      "grad_norm": 0.24090217053890228,
+      "learning_rate": 1.7324649298597194e-05,
+      "loss": 0.0095,
+      "step": 3273
+    },
+    {
+      "epoch": 13.096,
+      "grad_norm": 0.2191203087568283,
+      "learning_rate": 1.7314629258517035e-05,
+      "loss": 0.0093,
+      "step": 3274
+    },
+    {
+      "epoch": 13.1,
+      "grad_norm": 0.4989674389362335,
+      "learning_rate": 1.7304609218436873e-05,
+      "loss": 0.0121,
+      "step": 3275
+    },
+    {
+      "epoch": 13.104,
+      "grad_norm": 0.3217163383960724,
+      "learning_rate": 1.7294589178356714e-05,
+      "loss": 0.011,
+      "step": 3276
+    },
+    {
+      "epoch": 13.108,
+      "grad_norm": 0.3589065968990326,
+      "learning_rate": 1.7284569138276556e-05,
+      "loss": 0.0101,
+      "step": 3277
+    },
+    {
+      "epoch": 13.112,
+      "grad_norm": 0.3767361640930176,
+      "learning_rate": 1.7274549098196393e-05,
+      "loss": 0.0098,
+      "step": 3278
+    },
+    {
+      "epoch": 13.116,
+      "grad_norm": 0.3067415952682495,
+      "learning_rate": 1.726452905811623e-05,
+      "loss": 0.0095,
+      "step": 3279
+    },
+    {
+      "epoch": 13.12,
+      "grad_norm": 0.32393768429756165,
+      "learning_rate": 1.7254509018036073e-05,
+      "loss": 0.0106,
+      "step": 3280
+    },
+    {
+      "epoch": 13.124,
+      "grad_norm": 0.579896867275238,
+      "learning_rate": 1.7244488977955914e-05,
+      "loss": 0.0111,
+      "step": 3281
+    },
+    {
+      "epoch": 13.128,
+      "grad_norm": 0.16686947643756866,
+      "learning_rate": 1.7234468937875752e-05,
+      "loss": 0.006,
+      "step": 3282
+    },
+    {
+      "epoch": 13.132,
+      "grad_norm": 0.44065243005752563,
+      "learning_rate": 1.7224448897795593e-05,
+      "loss": 0.0106,
+      "step": 3283
+    },
+    {
+      "epoch": 13.136,
+      "grad_norm": 0.3439742922782898,
+      "learning_rate": 1.721442885771543e-05,
+      "loss": 0.0118,
+      "step": 3284
+    },
+    {
+      "epoch": 13.14,
+      "grad_norm": 0.35956379771232605,
+      "learning_rate": 1.720440881763527e-05,
+      "loss": 0.0101,
+      "step": 3285
+    },
+    {
+      "epoch": 13.144,
+      "grad_norm": 0.33086642622947693,
+      "learning_rate": 1.7194388777555114e-05,
+      "loss": 0.0103,
+      "step": 3286
+    },
+    {
+      "epoch": 13.148,
+      "grad_norm": 0.3616533577442169,
+      "learning_rate": 1.718436873747495e-05,
+      "loss": 0.0094,
+      "step": 3287
+    },
+    {
+      "epoch": 13.152,
+      "grad_norm": 0.3333515524864197,
+      "learning_rate": 1.717434869739479e-05,
+      "loss": 0.0093,
+      "step": 3288
+    },
+    {
+      "epoch": 13.156,
+      "grad_norm": 0.3492039144039154,
+      "learning_rate": 1.716432865731463e-05,
+      "loss": 0.011,
+      "step": 3289
+    },
+    {
+      "epoch": 13.16,
+      "grad_norm": 0.29600805044174194,
+      "learning_rate": 1.715430861723447e-05,
+      "loss": 0.0117,
+      "step": 3290
+    },
+    {
+      "epoch": 13.164,
+      "grad_norm": 0.3409759998321533,
+      "learning_rate": 1.714428857715431e-05,
+      "loss": 0.01,
+      "step": 3291
+    },
+    {
+      "epoch": 13.168,
+      "grad_norm": 0.237446591258049,
+      "learning_rate": 1.713426853707415e-05,
+      "loss": 0.0088,
+      "step": 3292
+    },
+    {
+      "epoch": 13.172,
+      "grad_norm": 0.3831733763217926,
+      "learning_rate": 1.712424849699399e-05,
+      "loss": 0.0099,
+      "step": 3293
+    },
+    {
+      "epoch": 13.176,
+      "grad_norm": 0.3086282014846802,
+      "learning_rate": 1.7114228456913827e-05,
+      "loss": 0.0101,
+      "step": 3294
+    },
+    {
+      "epoch": 13.18,
+      "grad_norm": 0.2323417216539383,
+      "learning_rate": 1.7104208416833668e-05,
+      "loss": 0.01,
+      "step": 3295
+    },
+    {
+      "epoch": 13.184,
+      "grad_norm": 0.2264842838048935,
+      "learning_rate": 1.709418837675351e-05,
+      "loss": 0.0089,
+      "step": 3296
+    },
+    {
+      "epoch": 13.188,
+      "grad_norm": 0.18145760893821716,
+      "learning_rate": 1.7084168336673347e-05,
+      "loss": 0.0087,
+      "step": 3297
+    },
+    {
+      "epoch": 13.192,
+      "grad_norm": 0.21976445615291595,
+      "learning_rate": 1.707414829659319e-05,
+      "loss": 0.0085,
+      "step": 3298
+    },
+    {
+      "epoch": 13.196,
+      "grad_norm": 0.2721193730831146,
+      "learning_rate": 1.7064128256513026e-05,
+      "loss": 0.0092,
+      "step": 3299
+    },
+    {
+      "epoch": 13.2,
+      "grad_norm": 0.2802695035934448,
+      "learning_rate": 1.7054108216432864e-05,
+      "loss": 0.0104,
+      "step": 3300
+    },
+    {
+      "epoch": 13.204,
+      "grad_norm": 0.48233532905578613,
+      "learning_rate": 1.7044088176352706e-05,
+      "loss": 0.0097,
+      "step": 3301
+    },
+    {
+      "epoch": 13.208,
+      "grad_norm": 0.2295309603214264,
+      "learning_rate": 1.7034068136272547e-05,
+      "loss": 0.0094,
+      "step": 3302
+    },
+    {
+      "epoch": 13.212,
+      "grad_norm": 0.32852253317832947,
+      "learning_rate": 1.7024048096192385e-05,
+      "loss": 0.0102,
+      "step": 3303
+    },
+    {
+      "epoch": 13.216,
+      "grad_norm": 0.30892303586006165,
+      "learning_rate": 1.7014028056112226e-05,
+      "loss": 0.01,
+      "step": 3304
+    },
+    {
+      "epoch": 13.22,
+      "grad_norm": 0.46433284878730774,
+      "learning_rate": 1.7004008016032064e-05,
+      "loss": 0.0097,
+      "step": 3305
+    },
+    {
+      "epoch": 13.224,
+      "grad_norm": 0.2719483971595764,
+      "learning_rate": 1.6993987975951905e-05,
+      "loss": 0.0094,
+      "step": 3306
+    },
+    {
+      "epoch": 13.228,
+      "grad_norm": 0.1855124831199646,
+      "learning_rate": 1.6983967935871743e-05,
+      "loss": 0.0085,
+      "step": 3307
+    },
+    {
+      "epoch": 13.232,
+      "grad_norm": 0.3346885144710541,
+      "learning_rate": 1.6973947895791584e-05,
+      "loss": 0.0109,
+      "step": 3308
+    },
+    {
+      "epoch": 13.236,
+      "grad_norm": 0.4800672233104706,
+      "learning_rate": 1.6963927855711422e-05,
+      "loss": 0.0123,
+      "step": 3309
+    },
+    {
+      "epoch": 13.24,
+      "grad_norm": 0.3444778025150299,
+      "learning_rate": 1.695390781563126e-05,
+      "loss": 0.0103,
+      "step": 3310
+    },
+    {
+      "epoch": 13.244,
+      "grad_norm": 0.2996631860733032,
+      "learning_rate": 1.6943887775551105e-05,
+      "loss": 0.0092,
+      "step": 3311
+    },
+    {
+      "epoch": 13.248,
+      "grad_norm": 0.2334967851638794,
+      "learning_rate": 1.6933867735470943e-05,
+      "loss": 0.0093,
+      "step": 3312
+    },
+    {
+      "epoch": 13.252,
+      "grad_norm": 0.20295573770999908,
+      "learning_rate": 1.692384769539078e-05,
+      "loss": 0.0086,
+      "step": 3313
+    },
+    {
+      "epoch": 13.256,
+      "grad_norm": 0.242875874042511,
+      "learning_rate": 1.6913827655310622e-05,
+      "loss": 0.0092,
+      "step": 3314
+    },
+    {
+      "epoch": 13.26,
+      "grad_norm": 0.41469478607177734,
+      "learning_rate": 1.6903807615230463e-05,
+      "loss": 0.0112,
+      "step": 3315
+    },
+    {
+      "epoch": 13.264,
+      "grad_norm": 0.2873092591762543,
+      "learning_rate": 1.68937875751503e-05,
+      "loss": 0.01,
+      "step": 3316
+    },
+    {
+      "epoch": 13.268,
+      "grad_norm": 0.16174815595149994,
+      "learning_rate": 1.6883767535070142e-05,
+      "loss": 0.005,
+      "step": 3317
+    },
+    {
+      "epoch": 13.272,
+      "grad_norm": 0.5048874020576477,
+      "learning_rate": 1.687374749498998e-05,
+      "loss": 0.012,
+      "step": 3318
+    },
+    {
+      "epoch": 13.276,
+      "grad_norm": 0.3545554578304291,
+      "learning_rate": 1.6863727454909818e-05,
+      "loss": 0.0095,
+      "step": 3319
+    },
+    {
+      "epoch": 13.28,
+      "grad_norm": 0.40919992327690125,
+      "learning_rate": 1.6853707414829663e-05,
+      "loss": 0.0111,
+      "step": 3320
+    },
+    {
+      "epoch": 13.284,
+      "grad_norm": 0.2165553867816925,
+      "learning_rate": 1.68436873747495e-05,
+      "loss": 0.0084,
+      "step": 3321
+    },
+    {
+      "epoch": 13.288,
+      "grad_norm": 0.369000107049942,
+      "learning_rate": 1.683366733466934e-05,
+      "loss": 0.0117,
+      "step": 3322
+    },
+    {
+      "epoch": 13.292,
+      "grad_norm": 0.3298990726470947,
+      "learning_rate": 1.682364729458918e-05,
+      "loss": 0.0087,
+      "step": 3323
+    },
+    {
+      "epoch": 13.296,
+      "grad_norm": 0.37005317211151123,
+      "learning_rate": 1.6813627254509018e-05,
+      "loss": 0.0103,
+      "step": 3324
+    },
+    {
+      "epoch": 13.3,
+      "grad_norm": 0.5654655694961548,
+      "learning_rate": 1.680360721442886e-05,
+      "loss": 0.0102,
+      "step": 3325
+    },
+    {
+      "epoch": 13.304,
+      "grad_norm": 0.7227338552474976,
+      "learning_rate": 1.67935871743487e-05,
+      "loss": 0.0166,
+      "step": 3326
+    },
+    {
+      "epoch": 13.308,
+      "grad_norm": 0.443639874458313,
+      "learning_rate": 1.678356713426854e-05,
+      "loss": 0.0111,
+      "step": 3327
+    },
+    {
+      "epoch": 13.312,
+      "grad_norm": 0.25988972187042236,
+      "learning_rate": 1.6773547094188376e-05,
+      "loss": 0.0093,
+      "step": 3328
+    },
+    {
+      "epoch": 13.316,
+      "grad_norm": 0.3140908181667328,
+      "learning_rate": 1.6763527054108218e-05,
+      "loss": 0.0094,
+      "step": 3329
+    },
+    {
+      "epoch": 13.32,
+      "grad_norm": 0.23378266394138336,
+      "learning_rate": 1.675350701402806e-05,
+      "loss": 0.0096,
+      "step": 3330
+    },
+    {
+      "epoch": 13.324,
+      "grad_norm": 0.31349173188209534,
+      "learning_rate": 1.6743486973947897e-05,
+      "loss": 0.0071,
+      "step": 3331
+    },
+    {
+      "epoch": 13.328,
+      "grad_norm": 0.3660728633403778,
+      "learning_rate": 1.6733466933867738e-05,
+      "loss": 0.0111,
+      "step": 3332
+    },
+    {
+      "epoch": 13.332,
+      "grad_norm": 0.4420227110385895,
+      "learning_rate": 1.6723446893787576e-05,
+      "loss": 0.0137,
+      "step": 3333
+    },
+    {
+      "epoch": 13.336,
+      "grad_norm": 0.2265954315662384,
+      "learning_rate": 1.6713426853707414e-05,
+      "loss": 0.0093,
+      "step": 3334
+    },
+    {
+      "epoch": 13.34,
+      "grad_norm": 0.31898754835128784,
+      "learning_rate": 1.6703406813627255e-05,
+      "loss": 0.0103,
+      "step": 3335
+    },
+    {
+      "epoch": 13.344,
+      "grad_norm": 0.34343066811561584,
+      "learning_rate": 1.6693386773547096e-05,
+      "loss": 0.0112,
+      "step": 3336
+    },
+    {
+      "epoch": 13.348,
+      "grad_norm": 0.27078312635421753,
+      "learning_rate": 1.6683366733466934e-05,
+      "loss": 0.01,
+      "step": 3337
+    },
+    {
+      "epoch": 13.352,
+      "grad_norm": 0.2199242115020752,
+      "learning_rate": 1.6673346693386772e-05,
+      "loss": 0.011,
+      "step": 3338
+    },
+    {
+      "epoch": 13.356,
+      "grad_norm": 0.25304946303367615,
+      "learning_rate": 1.6663326653306613e-05,
+      "loss": 0.0101,
+      "step": 3339
+    },
+    {
+      "epoch": 13.36,
+      "grad_norm": 0.3057776689529419,
+      "learning_rate": 1.6653306613226455e-05,
+      "loss": 0.0099,
+      "step": 3340
+    },
+    {
+      "epoch": 13.364,
+      "grad_norm": 0.17325368523597717,
+      "learning_rate": 1.6643286573146293e-05,
+      "loss": 0.0087,
+      "step": 3341
+    },
+    {
+      "epoch": 13.368,
+      "grad_norm": 0.2971237599849701,
+      "learning_rate": 1.6633266533066134e-05,
+      "loss": 0.0109,
+      "step": 3342
+    },
+    {
+      "epoch": 13.372,
+      "grad_norm": 0.2874193787574768,
+      "learning_rate": 1.6623246492985972e-05,
+      "loss": 0.0099,
+      "step": 3343
+    },
+    {
+      "epoch": 13.376,
+      "grad_norm": 0.3510386347770691,
+      "learning_rate": 1.661322645290581e-05,
+      "loss": 0.0119,
+      "step": 3344
+    },
+    {
+      "epoch": 13.38,
+      "grad_norm": 0.4247153401374817,
+      "learning_rate": 1.6603206412825654e-05,
+      "loss": 0.0138,
+      "step": 3345
+    },
+    {
+      "epoch": 13.384,
+      "grad_norm": 0.2908627986907959,
+      "learning_rate": 1.6593186372745492e-05,
+      "loss": 0.0103,
+      "step": 3346
+    },
+    {
+      "epoch": 13.388,
+      "grad_norm": 0.2727614939212799,
+      "learning_rate": 1.658316633266533e-05,
+      "loss": 0.0101,
+      "step": 3347
+    },
+    {
+      "epoch": 13.392,
+      "grad_norm": 0.3280937969684601,
+      "learning_rate": 1.657314629258517e-05,
+      "loss": 0.0103,
+      "step": 3348
+    },
+    {
+      "epoch": 13.396,
+      "grad_norm": 0.2449714094400406,
+      "learning_rate": 1.656312625250501e-05,
+      "loss": 0.0104,
+      "step": 3349
+    },
+    {
+      "epoch": 13.4,
+      "grad_norm": 0.24496877193450928,
+      "learning_rate": 1.655310621242485e-05,
+      "loss": 0.0099,
+      "step": 3350
+    },
+    {
+      "epoch": 13.404,
+      "grad_norm": 0.4488162398338318,
+      "learning_rate": 1.6543086172344692e-05,
+      "loss": 0.0127,
+      "step": 3351
+    },
+    {
+      "epoch": 13.408,
+      "grad_norm": 0.3072626292705536,
+      "learning_rate": 1.653306613226453e-05,
+      "loss": 0.0105,
+      "step": 3352
+    },
+    {
+      "epoch": 13.412,
+      "grad_norm": 0.27603739500045776,
+      "learning_rate": 1.6523046092184368e-05,
+      "loss": 0.0103,
+      "step": 3353
+    },
+    {
+      "epoch": 13.416,
+      "grad_norm": 0.3227679431438446,
+      "learning_rate": 1.651302605210421e-05,
+      "loss": 0.0104,
+      "step": 3354
+    },
+    {
+      "epoch": 13.42,
+      "grad_norm": 0.23910482227802277,
+      "learning_rate": 1.650300601202405e-05,
+      "loss": 0.0101,
+      "step": 3355
+    },
+    {
+      "epoch": 13.424,
+      "grad_norm": 0.4852541387081146,
+      "learning_rate": 1.6492985971943888e-05,
+      "loss": 0.0118,
+      "step": 3356
+    },
+    {
+      "epoch": 13.428,
+      "grad_norm": 0.3126225471496582,
+      "learning_rate": 1.648296593186373e-05,
+      "loss": 0.0103,
+      "step": 3357
+    },
+    {
+      "epoch": 13.432,
+      "grad_norm": 0.2821721136569977,
+      "learning_rate": 1.6472945891783567e-05,
+      "loss": 0.0098,
+      "step": 3358
+    },
+    {
+      "epoch": 13.436,
+      "grad_norm": 0.3140818774700165,
+      "learning_rate": 1.6462925851703405e-05,
+      "loss": 0.0103,
+      "step": 3359
+    },
+    {
+      "epoch": 13.44,
+      "grad_norm": 0.3383825719356537,
+      "learning_rate": 1.6452905811623246e-05,
+      "loss": 0.0112,
+      "step": 3360
+    },
+    {
+      "epoch": 13.444,
+      "grad_norm": 0.3161516487598419,
+      "learning_rate": 1.6442885771543088e-05,
+      "loss": 0.0109,
+      "step": 3361
+    },
+    {
+      "epoch": 13.448,
+      "grad_norm": 0.24821779131889343,
+      "learning_rate": 1.6432865731462926e-05,
+      "loss": 0.0093,
+      "step": 3362
+    },
+    {
+      "epoch": 13.452,
+      "grad_norm": 0.3279327154159546,
+      "learning_rate": 1.6422845691382767e-05,
+      "loss": 0.0098,
+      "step": 3363
+    },
+    {
+      "epoch": 13.456,
+      "grad_norm": 0.3004770874977112,
+      "learning_rate": 1.6412825651302605e-05,
+      "loss": 0.0101,
+      "step": 3364
+    },
+    {
+      "epoch": 13.46,
+      "grad_norm": 0.26334452629089355,
+      "learning_rate": 1.6402805611222446e-05,
+      "loss": 0.0094,
+      "step": 3365
+    },
+    {
+      "epoch": 13.464,
+      "grad_norm": 0.2865053713321686,
+      "learning_rate": 1.6392785571142284e-05,
+      "loss": 0.0107,
+      "step": 3366
+    },
+    {
+      "epoch": 13.468,
+      "grad_norm": 0.3839758038520813,
+      "learning_rate": 1.6382765531062125e-05,
+      "loss": 0.0103,
+      "step": 3367
+    },
+    {
+      "epoch": 13.472,
+      "grad_norm": 0.48596179485321045,
+      "learning_rate": 1.6372745490981963e-05,
+      "loss": 0.0097,
+      "step": 3368
+    },
+    {
+      "epoch": 13.475999999999999,
+      "grad_norm": 0.22549477219581604,
+      "learning_rate": 1.6362725450901804e-05,
+      "loss": 0.0085,
+      "step": 3369
+    },
+    {
+      "epoch": 13.48,
+      "grad_norm": 0.3293830156326294,
+      "learning_rate": 1.6352705410821646e-05,
+      "loss": 0.0096,
+      "step": 3370
+    },
+    {
+      "epoch": 13.484,
+      "grad_norm": 0.2493169605731964,
+      "learning_rate": 1.6342685370741484e-05,
+      "loss": 0.0093,
+      "step": 3371
+    },
+    {
+      "epoch": 13.488,
+      "grad_norm": 0.4540705978870392,
+      "learning_rate": 1.633266533066132e-05,
+      "loss": 0.0115,
+      "step": 3372
+    },
+    {
+      "epoch": 13.492,
+      "grad_norm": 0.2564983665943146,
+      "learning_rate": 1.6322645290581163e-05,
+      "loss": 0.0096,
+      "step": 3373
+    },
+    {
+      "epoch": 13.496,
+      "grad_norm": 0.26169297099113464,
+      "learning_rate": 1.6312625250501004e-05,
+      "loss": 0.0099,
+      "step": 3374
+    },
+    {
+      "epoch": 13.5,
+      "grad_norm": 0.31739065051078796,
+      "learning_rate": 1.6302605210420842e-05,
+      "loss": 0.0097,
+      "step": 3375
+    },
+    {
+      "epoch": 13.504,
+      "grad_norm": 0.2799374759197235,
+      "learning_rate": 1.6292585170340683e-05,
+      "loss": 0.0099,
+      "step": 3376
+    },
+    {
+      "epoch": 13.508,
+      "grad_norm": 0.22275328636169434,
+      "learning_rate": 1.628256513026052e-05,
+      "loss": 0.0095,
+      "step": 3377
+    },
+    {
+      "epoch": 13.512,
+      "grad_norm": 0.37261059880256653,
+      "learning_rate": 1.627254509018036e-05,
+      "loss": 0.0096,
+      "step": 3378
+    },
+    {
+      "epoch": 13.516,
+      "grad_norm": 0.23295381665229797,
+      "learning_rate": 1.6262525050100204e-05,
+      "loss": 0.0089,
+      "step": 3379
+    },
+    {
+      "epoch": 13.52,
+      "grad_norm": 0.23815542459487915,
+      "learning_rate": 1.625250501002004e-05,
+      "loss": 0.0101,
+      "step": 3380
+    },
+    {
+      "epoch": 13.524000000000001,
+      "grad_norm": 0.27734702825546265,
+      "learning_rate": 1.624248496993988e-05,
+      "loss": 0.01,
+      "step": 3381
+    },
+    {
+      "epoch": 13.528,
+      "grad_norm": 0.2713322937488556,
+      "learning_rate": 1.623246492985972e-05,
+      "loss": 0.0092,
+      "step": 3382
+    },
+    {
+      "epoch": 13.532,
+      "grad_norm": 0.4412878751754761,
+      "learning_rate": 1.622244488977956e-05,
+      "loss": 0.0113,
+      "step": 3383
+    },
+    {
+      "epoch": 13.536,
+      "grad_norm": 0.3151780068874359,
+      "learning_rate": 1.62124248496994e-05,
+      "loss": 0.0106,
+      "step": 3384
+    },
+    {
+      "epoch": 13.54,
+      "grad_norm": 0.3254270851612091,
+      "learning_rate": 1.620240480961924e-05,
+      "loss": 0.0102,
+      "step": 3385
+    },
+    {
+      "epoch": 13.544,
+      "grad_norm": 0.3681427240371704,
+      "learning_rate": 1.619238476953908e-05,
+      "loss": 0.0116,
+      "step": 3386
+    },
+    {
+      "epoch": 13.548,
+      "grad_norm": 0.3805301785469055,
+      "learning_rate": 1.6182364729458917e-05,
+      "loss": 0.009,
+      "step": 3387
+    },
+    {
+      "epoch": 13.552,
+      "grad_norm": 0.2876235246658325,
+      "learning_rate": 1.617234468937876e-05,
+      "loss": 0.0103,
+      "step": 3388
+    },
+    {
+      "epoch": 13.556000000000001,
+      "grad_norm": 0.3005918264389038,
+      "learning_rate": 1.61623246492986e-05,
+      "loss": 0.0099,
+      "step": 3389
+    },
+    {
+      "epoch": 13.56,
+      "grad_norm": 0.382587730884552,
+      "learning_rate": 1.6152304609218438e-05,
+      "loss": 0.0108,
+      "step": 3390
+    },
+    {
+      "epoch": 13.564,
+      "grad_norm": 0.4565657377243042,
+      "learning_rate": 1.614228456913828e-05,
+      "loss": 0.01,
+      "step": 3391
+    },
+    {
+      "epoch": 13.568,
+      "grad_norm": 0.14987795054912567,
+      "learning_rate": 1.6132264529058117e-05,
+      "loss": 0.0059,
+      "step": 3392
+    },
+    {
+      "epoch": 13.572,
+      "grad_norm": 0.21870999038219452,
+      "learning_rate": 1.6122244488977955e-05,
+      "loss": 0.009,
+      "step": 3393
+    },
+    {
+      "epoch": 13.576,
+      "grad_norm": 0.5497450232505798,
+      "learning_rate": 1.6112224448897796e-05,
+      "loss": 0.0117,
+      "step": 3394
+    },
+    {
+      "epoch": 13.58,
+      "grad_norm": 0.35143905878067017,
+      "learning_rate": 1.6102204408817637e-05,
+      "loss": 0.0111,
+      "step": 3395
+    },
+    {
+      "epoch": 13.584,
+      "grad_norm": 0.27477556467056274,
+      "learning_rate": 1.6092184368737475e-05,
+      "loss": 0.0105,
+      "step": 3396
+    },
+    {
+      "epoch": 13.588,
+      "grad_norm": 0.29407718777656555,
+      "learning_rate": 1.6082164328657316e-05,
+      "loss": 0.0095,
+      "step": 3397
+    },
+    {
+      "epoch": 13.592,
+      "grad_norm": 0.1857583224773407,
+      "learning_rate": 1.6072144288577154e-05,
+      "loss": 0.0092,
+      "step": 3398
+    },
+    {
+      "epoch": 13.596,
+      "grad_norm": 0.4153088927268982,
+      "learning_rate": 1.6062124248496996e-05,
+      "loss": 0.0096,
+      "step": 3399
+    },
+    {
+      "epoch": 13.6,
+      "grad_norm": 0.28967979550361633,
+      "learning_rate": 1.6052104208416833e-05,
+      "loss": 0.0107,
+      "step": 3400
+    },
+    {
+      "epoch": 13.604,
+      "grad_norm": 0.24863949418067932,
+      "learning_rate": 1.6042084168336675e-05,
+      "loss": 0.0103,
+      "step": 3401
+    },
+    {
+      "epoch": 13.608,
+      "grad_norm": 0.3102586269378662,
+      "learning_rate": 1.6032064128256513e-05,
+      "loss": 0.0103,
+      "step": 3402
+    },
+    {
+      "epoch": 13.612,
+      "grad_norm": 0.361162394285202,
+      "learning_rate": 1.6022044088176354e-05,
+      "loss": 0.0125,
+      "step": 3403
+    },
+    {
+      "epoch": 13.616,
+      "grad_norm": 0.40168410539627075,
+      "learning_rate": 1.6012024048096195e-05,
+      "loss": 0.0114,
+      "step": 3404
+    },
+    {
+      "epoch": 13.62,
+      "grad_norm": 0.29243239760398865,
+      "learning_rate": 1.6002004008016033e-05,
+      "loss": 0.0055,
+      "step": 3405
+    },
+    {
+      "epoch": 13.624,
+      "grad_norm": 0.26829758286476135,
+      "learning_rate": 1.599198396793587e-05,
+      "loss": 0.0112,
+      "step": 3406
+    },
+    {
+      "epoch": 13.628,
+      "grad_norm": 0.40442490577697754,
+      "learning_rate": 1.5981963927855712e-05,
+      "loss": 0.0115,
+      "step": 3407
+    },
+    {
+      "epoch": 13.632,
+      "grad_norm": 0.31696653366088867,
+      "learning_rate": 1.597194388777555e-05,
+      "loss": 0.0109,
+      "step": 3408
+    },
+    {
+      "epoch": 13.636,
+      "grad_norm": 0.44148167967796326,
+      "learning_rate": 1.596192384769539e-05,
+      "loss": 0.0107,
+      "step": 3409
+    },
+    {
+      "epoch": 13.64,
+      "grad_norm": 0.2826448082923889,
+      "learning_rate": 1.5951903807615233e-05,
+      "loss": 0.0107,
+      "step": 3410
+    },
+    {
+      "epoch": 13.644,
+      "grad_norm": 0.524307906627655,
+      "learning_rate": 1.594188376753507e-05,
+      "loss": 0.0111,
+      "step": 3411
+    },
+    {
+      "epoch": 13.648,
+      "grad_norm": 0.30534476041793823,
+      "learning_rate": 1.593186372745491e-05,
+      "loss": 0.0106,
+      "step": 3412
+    },
+    {
+      "epoch": 13.652,
+      "grad_norm": 0.3145725727081299,
+      "learning_rate": 1.592184368737475e-05,
+      "loss": 0.0098,
+      "step": 3413
+    },
+    {
+      "epoch": 13.656,
+      "grad_norm": 0.21220719814300537,
+      "learning_rate": 1.591182364729459e-05,
+      "loss": 0.0111,
+      "step": 3414
+    },
+    {
+      "epoch": 13.66,
+      "grad_norm": 0.37001675367355347,
+      "learning_rate": 1.590180360721443e-05,
+      "loss": 0.0102,
+      "step": 3415
+    },
+    {
+      "epoch": 13.664,
+      "grad_norm": 0.2675144672393799,
+      "learning_rate": 1.589178356713427e-05,
+      "loss": 0.0104,
+      "step": 3416
+    },
+    {
+      "epoch": 13.668,
+      "grad_norm": 0.4971553683280945,
+      "learning_rate": 1.5881763527054108e-05,
+      "loss": 0.0144,
+      "step": 3417
+    },
+    {
+      "epoch": 13.672,
+      "grad_norm": 0.32960245013237,
+      "learning_rate": 1.5871743486973946e-05,
+      "loss": 0.0096,
+      "step": 3418
+    },
+    {
+      "epoch": 13.676,
+      "grad_norm": 0.1954725980758667,
+      "learning_rate": 1.586172344689379e-05,
+      "loss": 0.0084,
+      "step": 3419
+    },
+    {
+      "epoch": 13.68,
+      "grad_norm": 0.338614821434021,
+      "learning_rate": 1.585170340681363e-05,
+      "loss": 0.012,
+      "step": 3420
+    },
+    {
+      "epoch": 13.684,
+      "grad_norm": 0.27286550402641296,
+      "learning_rate": 1.5841683366733466e-05,
+      "loss": 0.0105,
+      "step": 3421
+    },
+    {
+      "epoch": 13.688,
+      "grad_norm": 0.3043253421783447,
+      "learning_rate": 1.5831663326653308e-05,
+      "loss": 0.0111,
+      "step": 3422
+    },
+    {
+      "epoch": 13.692,
+      "grad_norm": 0.608754575252533,
+      "learning_rate": 1.582164328657315e-05,
+      "loss": 0.0122,
+      "step": 3423
+    },
+    {
+      "epoch": 13.696,
+      "grad_norm": 0.44034209847450256,
+      "learning_rate": 1.5811623246492987e-05,
+      "loss": 0.0103,
+      "step": 3424
+    },
+    {
+      "epoch": 13.7,
+      "grad_norm": 0.3596172630786896,
+      "learning_rate": 1.5801603206412828e-05,
+      "loss": 0.0101,
+      "step": 3425
+    },
+    {
+      "epoch": 13.704,
+      "grad_norm": 0.1947639137506485,
+      "learning_rate": 1.5791583166332666e-05,
+      "loss": 0.0086,
+      "step": 3426
+    },
+    {
+      "epoch": 13.708,
+      "grad_norm": 0.595519483089447,
+      "learning_rate": 1.5781563126252504e-05,
+      "loss": 0.0113,
+      "step": 3427
+    },
+    {
+      "epoch": 13.712,
+      "grad_norm": 0.30522915720939636,
+      "learning_rate": 1.5771543086172345e-05,
+      "loss": 0.011,
+      "step": 3428
+    },
+    {
+      "epoch": 13.716,
+      "grad_norm": 0.2296917587518692,
+      "learning_rate": 1.5761523046092187e-05,
+      "loss": 0.01,
+      "step": 3429
+    },
+    {
+      "epoch": 13.72,
+      "grad_norm": 0.5215935707092285,
+      "learning_rate": 1.5751503006012024e-05,
+      "loss": 0.0099,
+      "step": 3430
+    },
+    {
+      "epoch": 13.724,
+      "grad_norm": 0.22468657791614532,
+      "learning_rate": 1.5741482965931866e-05,
+      "loss": 0.0108,
+      "step": 3431
+    },
+    {
+      "epoch": 13.728,
+      "grad_norm": 0.25664982199668884,
+      "learning_rate": 1.5731462925851704e-05,
+      "loss": 0.0109,
+      "step": 3432
+    },
+    {
+      "epoch": 13.732,
+      "grad_norm": 0.2053966224193573,
+      "learning_rate": 1.5721442885771545e-05,
+      "loss": 0.0096,
+      "step": 3433
+    },
+    {
+      "epoch": 13.736,
+      "grad_norm": 0.2271842062473297,
+      "learning_rate": 1.5711422845691383e-05,
+      "loss": 0.0092,
+      "step": 3434
+    },
+    {
+      "epoch": 13.74,
+      "grad_norm": 0.41068193316459656,
+      "learning_rate": 1.5701402805611224e-05,
+      "loss": 0.0119,
+      "step": 3435
+    },
+    {
+      "epoch": 13.744,
+      "grad_norm": 0.2704119086265564,
+      "learning_rate": 1.5691382765531062e-05,
+      "loss": 0.0096,
+      "step": 3436
+    },
+    {
+      "epoch": 13.748,
+      "grad_norm": 0.37335240840911865,
+      "learning_rate": 1.56813627254509e-05,
+      "loss": 0.0106,
+      "step": 3437
+    },
+    {
+      "epoch": 13.752,
+      "grad_norm": 0.6146287322044373,
+      "learning_rate": 1.5671342685370745e-05,
+      "loss": 0.0117,
+      "step": 3438
+    },
+    {
+      "epoch": 13.756,
+      "grad_norm": 0.22417746484279633,
+      "learning_rate": 1.5661322645290582e-05,
+      "loss": 0.0061,
+      "step": 3439
+    },
+    {
+      "epoch": 13.76,
+      "grad_norm": 0.38575106859207153,
+      "learning_rate": 1.565130260521042e-05,
+      "loss": 0.0108,
+      "step": 3440
+    },
+    {
+      "epoch": 13.764,
+      "grad_norm": 0.2831723690032959,
+      "learning_rate": 1.564128256513026e-05,
+      "loss": 0.0103,
+      "step": 3441
+    },
+    {
+      "epoch": 13.768,
+      "grad_norm": 0.39680731296539307,
+      "learning_rate": 1.56312625250501e-05,
+      "loss": 0.0117,
+      "step": 3442
+    },
+    {
+      "epoch": 13.772,
+      "grad_norm": 0.5540252327919006,
+      "learning_rate": 1.562124248496994e-05,
+      "loss": 0.0123,
+      "step": 3443
+    },
+    {
+      "epoch": 13.776,
+      "grad_norm": 0.34284451603889465,
+      "learning_rate": 1.5611222444889782e-05,
+      "loss": 0.011,
+      "step": 3444
+    },
+    {
+      "epoch": 13.78,
+      "grad_norm": 0.23013213276863098,
+      "learning_rate": 1.560120240480962e-05,
+      "loss": 0.011,
+      "step": 3445
+    },
+    {
+      "epoch": 13.784,
+      "grad_norm": 0.583663284778595,
+      "learning_rate": 1.5591182364729458e-05,
+      "loss": 0.0121,
+      "step": 3446
+    },
+    {
+      "epoch": 13.788,
+      "grad_norm": 0.2305268794298172,
+      "learning_rate": 1.55811623246493e-05,
+      "loss": 0.0098,
+      "step": 3447
+    },
+    {
+      "epoch": 13.792,
+      "grad_norm": 0.2994031608104706,
+      "learning_rate": 1.557114228456914e-05,
+      "loss": 0.0102,
+      "step": 3448
+    },
+    {
+      "epoch": 13.796,
+      "grad_norm": 0.26790478825569153,
+      "learning_rate": 1.556112224448898e-05,
+      "loss": 0.009,
+      "step": 3449
+    },
+    {
+      "epoch": 13.8,
+      "grad_norm": 0.2465956211090088,
+      "learning_rate": 1.555110220440882e-05,
+      "loss": 0.0106,
+      "step": 3450
+    },
+    {
+      "epoch": 13.804,
+      "grad_norm": 0.40356552600860596,
+      "learning_rate": 1.5541082164328658e-05,
+      "loss": 0.0105,
+      "step": 3451
+    },
+    {
+      "epoch": 13.808,
+      "grad_norm": 0.2311774492263794,
+      "learning_rate": 1.5531062124248495e-05,
+      "loss": 0.0094,
+      "step": 3452
+    },
+    {
+      "epoch": 13.812,
+      "grad_norm": 0.32201912999153137,
+      "learning_rate": 1.552104208416834e-05,
+      "loss": 0.012,
+      "step": 3453
+    },
+    {
+      "epoch": 13.816,
+      "grad_norm": 0.35122230648994446,
+      "learning_rate": 1.5511022044088178e-05,
+      "loss": 0.0101,
+      "step": 3454
+    },
+    {
+      "epoch": 13.82,
+      "grad_norm": 0.3138692378997803,
+      "learning_rate": 1.5501002004008016e-05,
+      "loss": 0.0109,
+      "step": 3455
+    },
+    {
+      "epoch": 13.824,
+      "grad_norm": 0.3369031846523285,
+      "learning_rate": 1.5490981963927857e-05,
+      "loss": 0.0101,
+      "step": 3456
+    },
+    {
+      "epoch": 13.828,
+      "grad_norm": 0.31963375210762024,
+      "learning_rate": 1.5480961923847695e-05,
+      "loss": 0.011,
+      "step": 3457
+    },
+    {
+      "epoch": 13.832,
+      "grad_norm": 0.42217960953712463,
+      "learning_rate": 1.5470941883767536e-05,
+      "loss": 0.0111,
+      "step": 3458
+    },
+    {
+      "epoch": 13.836,
+      "grad_norm": 0.39974090456962585,
+      "learning_rate": 1.5460921843687378e-05,
+      "loss": 0.0114,
+      "step": 3459
+    },
+    {
+      "epoch": 13.84,
+      "grad_norm": 0.540280818939209,
+      "learning_rate": 1.5450901803607216e-05,
+      "loss": 0.0129,
+      "step": 3460
+    },
+    {
+      "epoch": 13.844,
+      "grad_norm": 0.2044568657875061,
+      "learning_rate": 1.5440881763527053e-05,
+      "loss": 0.0095,
+      "step": 3461
+    },
+    {
+      "epoch": 13.848,
+      "grad_norm": 0.376753032207489,
+      "learning_rate": 1.5430861723446895e-05,
+      "loss": 0.0115,
+      "step": 3462
+    },
+    {
+      "epoch": 13.852,
+      "grad_norm": 0.40467560291290283,
+      "learning_rate": 1.5420841683366736e-05,
+      "loss": 0.0122,
+      "step": 3463
+    },
+    {
+      "epoch": 13.856,
+      "grad_norm": 0.2836839556694031,
+      "learning_rate": 1.5410821643286574e-05,
+      "loss": 0.0098,
+      "step": 3464
+    },
+    {
+      "epoch": 13.86,
+      "grad_norm": 0.5280314683914185,
+      "learning_rate": 1.5400801603206412e-05,
+      "loss": 0.0105,
+      "step": 3465
+    },
+    {
+      "epoch": 13.864,
+      "grad_norm": 0.24809741973876953,
+      "learning_rate": 1.5390781563126253e-05,
+      "loss": 0.0101,
+      "step": 3466
+    },
+    {
+      "epoch": 13.868,
+      "grad_norm": 0.3241751790046692,
+      "learning_rate": 1.538076152304609e-05,
+      "loss": 0.0111,
+      "step": 3467
+    },
+    {
+      "epoch": 13.872,
+      "grad_norm": 0.2702431380748749,
+      "learning_rate": 1.5370741482965932e-05,
+      "loss": 0.0112,
+      "step": 3468
+    },
+    {
+      "epoch": 13.876,
+      "grad_norm": 0.4810154139995575,
+      "learning_rate": 1.5360721442885773e-05,
+      "loss": 0.0118,
+      "step": 3469
+    },
+    {
+      "epoch": 13.88,
+      "grad_norm": 0.3676247000694275,
+      "learning_rate": 1.535070140280561e-05,
+      "loss": 0.0104,
+      "step": 3470
+    },
+    {
+      "epoch": 13.884,
+      "grad_norm": 0.27984535694122314,
+      "learning_rate": 1.534068136272545e-05,
+      "loss": 0.0099,
+      "step": 3471
+    },
+    {
+      "epoch": 13.888,
+      "grad_norm": 0.20753216743469238,
+      "learning_rate": 1.533066132264529e-05,
+      "loss": 0.01,
+      "step": 3472
+    },
+    {
+      "epoch": 13.892,
+      "grad_norm": 0.22736266255378723,
+      "learning_rate": 1.5320641282565132e-05,
+      "loss": 0.0094,
+      "step": 3473
+    },
+    {
+      "epoch": 13.896,
+      "grad_norm": 0.2514065206050873,
+      "learning_rate": 1.531062124248497e-05,
+      "loss": 0.0095,
+      "step": 3474
+    },
+    {
+      "epoch": 13.9,
+      "grad_norm": 0.39425936341285706,
+      "learning_rate": 1.530060120240481e-05,
+      "loss": 0.011,
+      "step": 3475
+    },
+    {
+      "epoch": 13.904,
+      "grad_norm": 0.3547126054763794,
+      "learning_rate": 1.529058116232465e-05,
+      "loss": 0.01,
+      "step": 3476
+    },
+    {
+      "epoch": 13.908,
+      "grad_norm": 0.20616790652275085,
+      "learning_rate": 1.5280561122244487e-05,
+      "loss": 0.0067,
+      "step": 3477
+    },
+    {
+      "epoch": 13.912,
+      "grad_norm": 0.2141866683959961,
+      "learning_rate": 1.527054108216433e-05,
+      "loss": 0.0094,
+      "step": 3478
+    },
+    {
+      "epoch": 13.916,
+      "grad_norm": 0.3642483055591583,
+      "learning_rate": 1.526052104208417e-05,
+      "loss": 0.01,
+      "step": 3479
+    },
+    {
+      "epoch": 13.92,
+      "grad_norm": 0.24967001378536224,
+      "learning_rate": 1.5250501002004009e-05,
+      "loss": 0.0115,
+      "step": 3480
+    },
+    {
+      "epoch": 13.924,
+      "grad_norm": 0.31511619687080383,
+      "learning_rate": 1.5240480961923847e-05,
+      "loss": 0.009,
+      "step": 3481
+    },
+    {
+      "epoch": 13.928,
+      "grad_norm": 0.5050347447395325,
+      "learning_rate": 1.523046092184369e-05,
+      "loss": 0.0128,
+      "step": 3482
+    },
+    {
+      "epoch": 13.932,
+      "grad_norm": 0.3271845579147339,
+      "learning_rate": 1.5220440881763528e-05,
+      "loss": 0.0102,
+      "step": 3483
+    },
+    {
+      "epoch": 13.936,
+      "grad_norm": 0.49033862352371216,
+      "learning_rate": 1.5210420841683367e-05,
+      "loss": 0.0095,
+      "step": 3484
+    },
+    {
+      "epoch": 13.94,
+      "grad_norm": 0.34133315086364746,
+      "learning_rate": 1.5200400801603207e-05,
+      "loss": 0.0106,
+      "step": 3485
+    },
+    {
+      "epoch": 13.943999999999999,
+      "grad_norm": 0.29839426279067993,
+      "learning_rate": 1.5190380761523047e-05,
+      "loss": 0.0096,
+      "step": 3486
+    },
+    {
+      "epoch": 13.948,
+      "grad_norm": 0.21854914724826813,
+      "learning_rate": 1.5180360721442888e-05,
+      "loss": 0.0096,
+      "step": 3487
+    },
+    {
+      "epoch": 13.952,
+      "grad_norm": 0.41724851727485657,
+      "learning_rate": 1.5170340681362727e-05,
+      "loss": 0.012,
+      "step": 3488
+    },
+    {
+      "epoch": 13.956,
+      "grad_norm": 0.09979083389043808,
+      "learning_rate": 1.5160320641282565e-05,
+      "loss": 0.0041,
+      "step": 3489
+    },
+    {
+      "epoch": 13.96,
+      "grad_norm": 0.1923419088125229,
+      "learning_rate": 1.5150300601202405e-05,
+      "loss": 0.0062,
+      "step": 3490
+    },
+    {
+      "epoch": 13.964,
+      "grad_norm": 0.2782224416732788,
+      "learning_rate": 1.5140280561122244e-05,
+      "loss": 0.0104,
+      "step": 3491
+    },
+    {
+      "epoch": 13.968,
+      "grad_norm": 0.3307152986526489,
+      "learning_rate": 1.5130260521042086e-05,
+      "loss": 0.0108,
+      "step": 3492
+    },
+    {
+      "epoch": 13.972,
+      "grad_norm": 0.247494637966156,
+      "learning_rate": 1.5120240480961925e-05,
+      "loss": 0.006,
+      "step": 3493
+    },
+    {
+      "epoch": 13.975999999999999,
+      "grad_norm": 0.27174144983291626,
+      "learning_rate": 1.5110220440881765e-05,
+      "loss": 0.0081,
+      "step": 3494
+    },
+    {
+      "epoch": 13.98,
+      "grad_norm": 0.26648882031440735,
+      "learning_rate": 1.5100200400801603e-05,
+      "loss": 0.0101,
+      "step": 3495
+    },
+    {
+      "epoch": 13.984,
+      "grad_norm": 0.2658537030220032,
+      "learning_rate": 1.5090180360721442e-05,
+      "loss": 0.0098,
+      "step": 3496
+    },
+    {
+      "epoch": 13.988,
+      "grad_norm": 0.2758139371871948,
+      "learning_rate": 1.5080160320641284e-05,
+      "loss": 0.0095,
+      "step": 3497
+    },
+    {
+      "epoch": 13.992,
+      "grad_norm": 0.3168604075908661,
+      "learning_rate": 1.5070140280561123e-05,
+      "loss": 0.0106,
+      "step": 3498
+    },
+    {
+      "epoch": 13.996,
+      "grad_norm": 0.3637627065181732,
+      "learning_rate": 1.5060120240480963e-05,
+      "loss": 0.012,
+      "step": 3499
+    },
+    {
+      "epoch": 14.0,
+      "grad_norm": 0.3541504144668579,
+      "learning_rate": 1.5050100200400802e-05,
+      "loss": 0.0065,
+      "step": 3500
+    },
+    {
+      "epoch": 14.004,
+      "grad_norm": 0.15899381041526794,
+      "learning_rate": 1.504008016032064e-05,
+      "loss": 0.008,
+      "step": 3501
+    },
+    {
+      "epoch": 14.008,
+      "grad_norm": 0.18751265108585358,
+      "learning_rate": 1.5030060120240483e-05,
+      "loss": 0.0079,
+      "step": 3502
+    },
+    {
+      "epoch": 14.012,
+      "grad_norm": 0.1956561952829361,
+      "learning_rate": 1.5020040080160321e-05,
+      "loss": 0.0084,
+      "step": 3503
+    },
+    {
+      "epoch": 14.016,
+      "grad_norm": 0.17938251793384552,
+      "learning_rate": 1.501002004008016e-05,
+      "loss": 0.0093,
+      "step": 3504
+    },
+    {
+      "epoch": 14.02,
+      "grad_norm": 0.2153725028038025,
+      "learning_rate": 1.5e-05,
+      "loss": 0.0086,
+      "step": 3505
+    },
+    {
+      "epoch": 14.024,
+      "grad_norm": 0.08399572223424911,
+      "learning_rate": 1.498997995991984e-05,
+      "loss": 0.0041,
+      "step": 3506
+    },
+    {
+      "epoch": 14.028,
+      "grad_norm": 0.20708952844142914,
+      "learning_rate": 1.4979959919839681e-05,
+      "loss": 0.0078,
+      "step": 3507
+    },
+    {
+      "epoch": 14.032,
+      "grad_norm": 0.18916714191436768,
+      "learning_rate": 1.496993987975952e-05,
+      "loss": 0.0089,
+      "step": 3508
+    },
+    {
+      "epoch": 14.036,
+      "grad_norm": 0.14934112131595612,
+      "learning_rate": 1.4959919839679359e-05,
+      "loss": 0.0076,
+      "step": 3509
+    },
+    {
+      "epoch": 14.04,
+      "grad_norm": 0.2596082389354706,
+      "learning_rate": 1.4949899799599198e-05,
+      "loss": 0.0092,
+      "step": 3510
+    },
+    {
+      "epoch": 14.044,
+      "grad_norm": 0.16445767879486084,
+      "learning_rate": 1.4939879759519038e-05,
+      "loss": 0.0054,
+      "step": 3511
+    },
+    {
+      "epoch": 14.048,
+      "grad_norm": 0.2873588800430298,
+      "learning_rate": 1.492985971943888e-05,
+      "loss": 0.0098,
+      "step": 3512
+    },
+    {
+      "epoch": 14.052,
+      "grad_norm": 0.19016903638839722,
+      "learning_rate": 1.4919839679358719e-05,
+      "loss": 0.0082,
+      "step": 3513
+    },
+    {
+      "epoch": 14.056,
+      "grad_norm": 0.1945473998785019,
+      "learning_rate": 1.4909819639278558e-05,
+      "loss": 0.0084,
+      "step": 3514
+    },
+    {
+      "epoch": 14.06,
+      "grad_norm": 0.5994365215301514,
+      "learning_rate": 1.4899799599198396e-05,
+      "loss": 0.0127,
+      "step": 3515
+    },
+    {
+      "epoch": 14.064,
+      "grad_norm": 0.2284667044878006,
+      "learning_rate": 1.4889779559118236e-05,
+      "loss": 0.0081,
+      "step": 3516
+    },
+    {
+      "epoch": 14.068,
+      "grad_norm": 0.1868845373392105,
+      "learning_rate": 1.4879759519038077e-05,
+      "loss": 0.0087,
+      "step": 3517
+    },
+    {
+      "epoch": 14.072,
+      "grad_norm": 0.25140196084976196,
+      "learning_rate": 1.4869739478957917e-05,
+      "loss": 0.0087,
+      "step": 3518
+    },
+    {
+      "epoch": 14.076,
+      "grad_norm": 0.30023810267448425,
+      "learning_rate": 1.4859719438877756e-05,
+      "loss": 0.0098,
+      "step": 3519
+    },
+    {
+      "epoch": 14.08,
+      "grad_norm": 0.3479897379875183,
+      "learning_rate": 1.4849699398797596e-05,
+      "loss": 0.0087,
+      "step": 3520
+    },
+    {
+      "epoch": 14.084,
+      "grad_norm": 0.33798903226852417,
+      "learning_rate": 1.4839679358717434e-05,
+      "loss": 0.0078,
+      "step": 3521
+    },
+    {
+      "epoch": 14.088,
+      "grad_norm": 0.21001707017421722,
+      "learning_rate": 1.4829659318637277e-05,
+      "loss": 0.0056,
+      "step": 3522
+    },
+    {
+      "epoch": 14.092,
+      "grad_norm": 0.3976731598377228,
+      "learning_rate": 1.4819639278557115e-05,
+      "loss": 0.0098,
+      "step": 3523
+    },
+    {
+      "epoch": 14.096,
+      "grad_norm": 0.26884251832962036,
+      "learning_rate": 1.4809619238476954e-05,
+      "loss": 0.008,
+      "step": 3524
+    },
+    {
+      "epoch": 14.1,
+      "grad_norm": 0.16603799164295197,
+      "learning_rate": 1.4799599198396794e-05,
+      "loss": 0.0081,
+      "step": 3525
+    },
+    {
+      "epoch": 14.104,
+      "grad_norm": 0.25354495644569397,
+      "learning_rate": 1.4789579158316633e-05,
+      "loss": 0.0085,
+      "step": 3526
+    },
+    {
+      "epoch": 14.108,
+      "grad_norm": 0.49945899844169617,
+      "learning_rate": 1.4779559118236475e-05,
+      "loss": 0.0098,
+      "step": 3527
+    },
+    {
+      "epoch": 14.112,
+      "grad_norm": 0.15931391716003418,
+      "learning_rate": 1.4769539078156314e-05,
+      "loss": 0.0051,
+      "step": 3528
+    },
+    {
+      "epoch": 14.116,
+      "grad_norm": 0.2964676022529602,
+      "learning_rate": 1.4759519038076152e-05,
+      "loss": 0.0096,
+      "step": 3529
+    },
+    {
+      "epoch": 14.12,
+      "grad_norm": 0.2884165048599243,
+      "learning_rate": 1.4749498997995992e-05,
+      "loss": 0.008,
+      "step": 3530
+    },
+    {
+      "epoch": 14.124,
+      "grad_norm": 1.5515947341918945,
+      "learning_rate": 1.4739478957915831e-05,
+      "loss": 0.0073,
+      "step": 3531
+    },
+    {
+      "epoch": 14.128,
+      "grad_norm": 0.2911422550678253,
+      "learning_rate": 1.4729458917835673e-05,
+      "loss": 0.0107,
+      "step": 3532
+    },
+    {
+      "epoch": 14.132,
+      "grad_norm": 0.23672667145729065,
+      "learning_rate": 1.4719438877755512e-05,
+      "loss": 0.009,
+      "step": 3533
+    },
+    {
+      "epoch": 14.136,
+      "grad_norm": 0.4554106593132019,
+      "learning_rate": 1.4709418837675352e-05,
+      "loss": 0.0095,
+      "step": 3534
+    },
+    {
+      "epoch": 14.14,
+      "grad_norm": 0.27175405621528625,
+      "learning_rate": 1.469939879759519e-05,
+      "loss": 0.0049,
+      "step": 3535
+    },
+    {
+      "epoch": 14.144,
+      "grad_norm": 0.3456001281738281,
+      "learning_rate": 1.4689378757515033e-05,
+      "loss": 0.0107,
+      "step": 3536
+    },
+    {
+      "epoch": 14.148,
+      "grad_norm": 0.2605019509792328,
+      "learning_rate": 1.467935871743487e-05,
+      "loss": 0.0064,
+      "step": 3537
+    },
+    {
+      "epoch": 14.152,
+      "grad_norm": 0.26379692554473877,
+      "learning_rate": 1.466933867735471e-05,
+      "loss": 0.0084,
+      "step": 3538
+    },
+    {
+      "epoch": 14.156,
+      "grad_norm": 0.23636199533939362,
+      "learning_rate": 1.465931863727455e-05,
+      "loss": 0.0082,
+      "step": 3539
+    },
+    {
+      "epoch": 14.16,
+      "grad_norm": 0.1825660616159439,
+      "learning_rate": 1.464929859719439e-05,
+      "loss": 0.008,
+      "step": 3540
+    },
+    {
+      "epoch": 14.164,
+      "grad_norm": 0.21577121317386627,
+      "learning_rate": 1.463927855711423e-05,
+      "loss": 0.0088,
+      "step": 3541
+    },
+    {
+      "epoch": 14.168,
+      "grad_norm": 0.18342304229736328,
+      "learning_rate": 1.462925851703407e-05,
+      "loss": 0.008,
+      "step": 3542
+    },
+    {
+      "epoch": 14.172,
+      "grad_norm": 0.16573582589626312,
+      "learning_rate": 1.4619238476953908e-05,
+      "loss": 0.0082,
+      "step": 3543
+    },
+    {
+      "epoch": 14.176,
+      "grad_norm": 0.20480182766914368,
+      "learning_rate": 1.4609218436873748e-05,
+      "loss": 0.0081,
+      "step": 3544
+    },
+    {
+      "epoch": 14.18,
+      "grad_norm": 0.19577351212501526,
+      "learning_rate": 1.4599198396793587e-05,
+      "loss": 0.0093,
+      "step": 3545
+    },
+    {
+      "epoch": 14.184,
+      "grad_norm": 0.23374129831790924,
+      "learning_rate": 1.4589178356713429e-05,
+      "loss": 0.0082,
+      "step": 3546
+    },
+    {
+      "epoch": 14.188,
+      "grad_norm": 0.22724345326423645,
+      "learning_rate": 1.4579158316633268e-05,
+      "loss": 0.0089,
+      "step": 3547
+    },
+    {
+      "epoch": 14.192,
+      "grad_norm": 0.31487712264060974,
+      "learning_rate": 1.4569138276553108e-05,
+      "loss": 0.0082,
+      "step": 3548
+    },
+    {
+      "epoch": 14.196,
+      "grad_norm": 0.2387700378894806,
+      "learning_rate": 1.4559118236472946e-05,
+      "loss": 0.0093,
+      "step": 3549
+    },
+    {
+      "epoch": 14.2,
+      "grad_norm": 0.24899888038635254,
+      "learning_rate": 1.4549098196392785e-05,
+      "loss": 0.0086,
+      "step": 3550
+    },
+    {
+      "epoch": 14.204,
+      "grad_norm": 0.21508151292800903,
+      "learning_rate": 1.4539078156312627e-05,
+      "loss": 0.0093,
+      "step": 3551
+    },
+    {
+      "epoch": 14.208,
+      "grad_norm": 0.2187836617231369,
+      "learning_rate": 1.4529058116232466e-05,
+      "loss": 0.0084,
+      "step": 3552
+    },
+    {
+      "epoch": 14.212,
+      "grad_norm": 0.2252088487148285,
+      "learning_rate": 1.4519038076152306e-05,
+      "loss": 0.0078,
+      "step": 3553
+    },
+    {
+      "epoch": 14.216,
+      "grad_norm": 0.22216878831386566,
+      "learning_rate": 1.4509018036072145e-05,
+      "loss": 0.0088,
+      "step": 3554
+    },
+    {
+      "epoch": 14.22,
+      "grad_norm": 0.27992987632751465,
+      "learning_rate": 1.4498997995991983e-05,
+      "loss": 0.0082,
+      "step": 3555
+    },
+    {
+      "epoch": 14.224,
+      "grad_norm": 0.21498125791549683,
+      "learning_rate": 1.4488977955911826e-05,
+      "loss": 0.0088,
+      "step": 3556
+    },
+    {
+      "epoch": 14.228,
+      "grad_norm": 0.15717551112174988,
+      "learning_rate": 1.4478957915831664e-05,
+      "loss": 0.0074,
+      "step": 3557
+    },
+    {
+      "epoch": 14.232,
+      "grad_norm": 0.22314034402370453,
+      "learning_rate": 1.4468937875751504e-05,
+      "loss": 0.0082,
+      "step": 3558
+    },
+    {
+      "epoch": 14.236,
+      "grad_norm": 0.28885912895202637,
+      "learning_rate": 1.4458917835671343e-05,
+      "loss": 0.0101,
+      "step": 3559
+    },
+    {
+      "epoch": 14.24,
+      "grad_norm": 0.22253185510635376,
+      "learning_rate": 1.4448897795591181e-05,
+      "loss": 0.0084,
+      "step": 3560
+    },
+    {
+      "epoch": 14.244,
+      "grad_norm": 0.22679786384105682,
+      "learning_rate": 1.4438877755511024e-05,
+      "loss": 0.0097,
+      "step": 3561
+    },
+    {
+      "epoch": 14.248,
+      "grad_norm": 0.22354552149772644,
+      "learning_rate": 1.4428857715430864e-05,
+      "loss": 0.0085,
+      "step": 3562
+    },
+    {
+      "epoch": 14.252,
+      "grad_norm": 0.24927063286304474,
+      "learning_rate": 1.4418837675350702e-05,
+      "loss": 0.0096,
+      "step": 3563
+    },
+    {
+      "epoch": 14.256,
+      "grad_norm": 0.20839311182498932,
+      "learning_rate": 1.4408817635270541e-05,
+      "loss": 0.0084,
+      "step": 3564
+    },
+    {
+      "epoch": 14.26,
+      "grad_norm": 0.15880200266838074,
+      "learning_rate": 1.439879759519038e-05,
+      "loss": 0.0084,
+      "step": 3565
+    },
+    {
+      "epoch": 14.264,
+      "grad_norm": 0.1647965908050537,
+      "learning_rate": 1.4388777555110222e-05,
+      "loss": 0.0077,
+      "step": 3566
+    },
+    {
+      "epoch": 14.268,
+      "grad_norm": 0.20082905888557434,
+      "learning_rate": 1.4378757515030062e-05,
+      "loss": 0.0092,
+      "step": 3567
+    },
+    {
+      "epoch": 14.272,
+      "grad_norm": 0.49212315678596497,
+      "learning_rate": 1.4368737474949901e-05,
+      "loss": 0.012,
+      "step": 3568
+    },
+    {
+      "epoch": 14.276,
+      "grad_norm": 0.20353491604328156,
+      "learning_rate": 1.4358717434869739e-05,
+      "loss": 0.0085,
+      "step": 3569
+    },
+    {
+      "epoch": 14.28,
+      "grad_norm": 0.18611657619476318,
+      "learning_rate": 1.4348697394789579e-05,
+      "loss": 0.0086,
+      "step": 3570
+    },
+    {
+      "epoch": 14.284,
+      "grad_norm": 0.2258971929550171,
+      "learning_rate": 1.433867735470942e-05,
+      "loss": 0.0081,
+      "step": 3571
+    },
+    {
+      "epoch": 14.288,
+      "grad_norm": 0.21564999222755432,
+      "learning_rate": 1.432865731462926e-05,
+      "loss": 0.0076,
+      "step": 3572
+    },
+    {
+      "epoch": 14.292,
+      "grad_norm": 0.17343463003635406,
+      "learning_rate": 1.43186372745491e-05,
+      "loss": 0.0085,
+      "step": 3573
+    },
+    {
+      "epoch": 14.296,
+      "grad_norm": 0.367103636264801,
+      "learning_rate": 1.4308617234468937e-05,
+      "loss": 0.01,
+      "step": 3574
+    },
+    {
+      "epoch": 14.3,
+      "grad_norm": 0.21690668165683746,
+      "learning_rate": 1.4298597194388777e-05,
+      "loss": 0.0094,
+      "step": 3575
+    },
+    {
+      "epoch": 14.304,
+      "grad_norm": 0.15850843489170074,
+      "learning_rate": 1.428857715430862e-05,
+      "loss": 0.0075,
+      "step": 3576
+    },
+    {
+      "epoch": 14.308,
+      "grad_norm": 0.31984981894493103,
+      "learning_rate": 1.4278557114228458e-05,
+      "loss": 0.0101,
+      "step": 3577
+    },
+    {
+      "epoch": 14.312,
+      "grad_norm": 0.45315784215927124,
+      "learning_rate": 1.4268537074148297e-05,
+      "loss": 0.0115,
+      "step": 3578
+    },
+    {
+      "epoch": 14.316,
+      "grad_norm": 0.207364022731781,
+      "learning_rate": 1.4258517034068137e-05,
+      "loss": 0.0086,
+      "step": 3579
+    },
+    {
+      "epoch": 14.32,
+      "grad_norm": 0.18384985625743866,
+      "learning_rate": 1.4248496993987975e-05,
+      "loss": 0.0086,
+      "step": 3580
+    },
+    {
+      "epoch": 14.324,
+      "grad_norm": 0.23308269679546356,
+      "learning_rate": 1.4238476953907818e-05,
+      "loss": 0.0089,
+      "step": 3581
+    },
+    {
+      "epoch": 14.328,
+      "grad_norm": 0.3302237093448639,
+      "learning_rate": 1.4228456913827657e-05,
+      "loss": 0.0093,
+      "step": 3582
+    },
+    {
+      "epoch": 14.332,
+      "grad_norm": 0.23539598286151886,
+      "learning_rate": 1.4218436873747495e-05,
+      "loss": 0.0089,
+      "step": 3583
+    },
+    {
+      "epoch": 14.336,
+      "grad_norm": 0.16788506507873535,
+      "learning_rate": 1.4208416833667335e-05,
+      "loss": 0.0072,
+      "step": 3584
+    },
+    {
+      "epoch": 14.34,
+      "grad_norm": 0.20347057282924652,
+      "learning_rate": 1.4198396793587174e-05,
+      "loss": 0.0087,
+      "step": 3585
+    },
+    {
+      "epoch": 14.344,
+      "grad_norm": 0.2108476608991623,
+      "learning_rate": 1.4188376753507016e-05,
+      "loss": 0.0092,
+      "step": 3586
+    },
+    {
+      "epoch": 14.348,
+      "grad_norm": 0.24422381818294525,
+      "learning_rate": 1.4178356713426855e-05,
+      "loss": 0.0096,
+      "step": 3587
+    },
+    {
+      "epoch": 14.352,
+      "grad_norm": 0.16197939217090607,
+      "learning_rate": 1.4168336673346693e-05,
+      "loss": 0.0081,
+      "step": 3588
+    },
+    {
+      "epoch": 14.356,
+      "grad_norm": 0.20140737295150757,
+      "learning_rate": 1.4158316633266533e-05,
+      "loss": 0.0085,
+      "step": 3589
+    },
+    {
+      "epoch": 14.36,
+      "grad_norm": 0.27182310819625854,
+      "learning_rate": 1.4148296593186376e-05,
+      "loss": 0.0097,
+      "step": 3590
+    },
+    {
+      "epoch": 14.364,
+      "grad_norm": 0.1505798101425171,
+      "learning_rate": 1.4138276553106213e-05,
+      "loss": 0.0081,
+      "step": 3591
+    },
+    {
+      "epoch": 14.368,
+      "grad_norm": 0.17416727542877197,
+      "learning_rate": 1.4128256513026053e-05,
+      "loss": 0.0081,
+      "step": 3592
+    },
+    {
+      "epoch": 14.372,
+      "grad_norm": 0.21804209053516388,
+      "learning_rate": 1.4118236472945893e-05,
+      "loss": 0.0083,
+      "step": 3593
+    },
+    {
+      "epoch": 14.376,
+      "grad_norm": 0.36001965403556824,
+      "learning_rate": 1.410821643286573e-05,
+      "loss": 0.0104,
+      "step": 3594
+    },
+    {
+      "epoch": 14.38,
+      "grad_norm": 0.20571143925189972,
+      "learning_rate": 1.4098196392785574e-05,
+      "loss": 0.0082,
+      "step": 3595
+    },
+    {
+      "epoch": 14.384,
+      "grad_norm": 0.19949229061603546,
+      "learning_rate": 1.4088176352705411e-05,
+      "loss": 0.0089,
+      "step": 3596
+    },
+    {
+      "epoch": 14.388,
+      "grad_norm": 0.667675256729126,
+      "learning_rate": 1.4078156312625251e-05,
+      "loss": 0.0088,
+      "step": 3597
+    },
+    {
+      "epoch": 14.392,
+      "grad_norm": 0.167070671916008,
+      "learning_rate": 1.406813627254509e-05,
+      "loss": 0.0087,
+      "step": 3598
+    },
+    {
+      "epoch": 14.396,
+      "grad_norm": 0.3734051585197449,
+      "learning_rate": 1.405811623246493e-05,
+      "loss": 0.0095,
+      "step": 3599
+    },
+    {
+      "epoch": 14.4,
+      "grad_norm": 0.28370603919029236,
+      "learning_rate": 1.4048096192384771e-05,
+      "loss": 0.0088,
+      "step": 3600
+    },
+    {
+      "epoch": 14.404,
+      "grad_norm": 0.30198901891708374,
+      "learning_rate": 1.4038076152304611e-05,
+      "loss": 0.0098,
+      "step": 3601
+    },
+    {
+      "epoch": 14.408,
+      "grad_norm": 0.23166725039482117,
+      "learning_rate": 1.4028056112224449e-05,
+      "loss": 0.008,
+      "step": 3602
+    },
+    {
+      "epoch": 14.412,
+      "grad_norm": 0.19708774983882904,
+      "learning_rate": 1.4018036072144289e-05,
+      "loss": 0.009,
+      "step": 3603
+    },
+    {
+      "epoch": 14.416,
+      "grad_norm": 0.30307242274284363,
+      "learning_rate": 1.4008016032064128e-05,
+      "loss": 0.0101,
+      "step": 3604
+    },
+    {
+      "epoch": 14.42,
+      "grad_norm": 0.16705302894115448,
+      "learning_rate": 1.399799599198397e-05,
+      "loss": 0.0076,
+      "step": 3605
+    },
+    {
+      "epoch": 14.424,
+      "grad_norm": 0.1781936138868332,
+      "learning_rate": 1.3987975951903809e-05,
+      "loss": 0.0087,
+      "step": 3606
+    },
+    {
+      "epoch": 14.428,
+      "grad_norm": 0.2449142336845398,
+      "learning_rate": 1.3977955911823649e-05,
+      "loss": 0.0104,
+      "step": 3607
+    },
+    {
+      "epoch": 14.432,
+      "grad_norm": 0.16610880196094513,
+      "learning_rate": 1.3967935871743486e-05,
+      "loss": 0.0082,
+      "step": 3608
+    },
+    {
+      "epoch": 14.436,
+      "grad_norm": 0.26270821690559387,
+      "learning_rate": 1.3957915831663326e-05,
+      "loss": 0.0096,
+      "step": 3609
+    },
+    {
+      "epoch": 14.44,
+      "grad_norm": 0.18053866922855377,
+      "learning_rate": 1.3947895791583167e-05,
+      "loss": 0.0055,
+      "step": 3610
+    },
+    {
+      "epoch": 14.444,
+      "grad_norm": 0.17106251418590546,
+      "learning_rate": 1.3937875751503007e-05,
+      "loss": 0.0085,
+      "step": 3611
+    },
+    {
+      "epoch": 14.448,
+      "grad_norm": 1.2041096687316895,
+      "learning_rate": 1.3927855711422847e-05,
+      "loss": 0.0099,
+      "step": 3612
+    },
+    {
+      "epoch": 14.452,
+      "grad_norm": 0.2078244984149933,
+      "learning_rate": 1.3917835671342686e-05,
+      "loss": 0.0093,
+      "step": 3613
+    },
+    {
+      "epoch": 14.456,
+      "grad_norm": 0.2774932384490967,
+      "learning_rate": 1.3907815631262524e-05,
+      "loss": 0.0095,
+      "step": 3614
+    },
+    {
+      "epoch": 14.46,
+      "grad_norm": 0.17771194875240326,
+      "learning_rate": 1.3897795591182367e-05,
+      "loss": 0.0075,
+      "step": 3615
+    },
+    {
+      "epoch": 14.464,
+      "grad_norm": 0.17751413583755493,
+      "learning_rate": 1.3887775551102205e-05,
+      "loss": 0.0086,
+      "step": 3616
+    },
+    {
+      "epoch": 14.468,
+      "grad_norm": 0.289604514837265,
+      "learning_rate": 1.3877755511022044e-05,
+      "loss": 0.0094,
+      "step": 3617
+    },
+    {
+      "epoch": 14.472,
+      "grad_norm": 0.1762779951095581,
+      "learning_rate": 1.3867735470941884e-05,
+      "loss": 0.0083,
+      "step": 3618
+    },
+    {
+      "epoch": 14.475999999999999,
+      "grad_norm": 0.2782403528690338,
+      "learning_rate": 1.3857715430861724e-05,
+      "loss": 0.0064,
+      "step": 3619
+    },
+    {
+      "epoch": 14.48,
+      "grad_norm": 0.22804078459739685,
+      "learning_rate": 1.3847695390781565e-05,
+      "loss": 0.0094,
+      "step": 3620
+    },
+    {
+      "epoch": 14.484,
+      "grad_norm": 0.29062676429748535,
+      "learning_rate": 1.3837675350701405e-05,
+      "loss": 0.0109,
+      "step": 3621
+    },
+    {
+      "epoch": 14.488,
+      "grad_norm": 0.2231641560792923,
+      "learning_rate": 1.3827655310621242e-05,
+      "loss": 0.0086,
+      "step": 3622
+    },
+    {
+      "epoch": 14.492,
+      "grad_norm": 0.1812341958284378,
+      "learning_rate": 1.3817635270541082e-05,
+      "loss": 0.0088,
+      "step": 3623
+    },
+    {
+      "epoch": 14.496,
+      "grad_norm": 0.1771288812160492,
+      "learning_rate": 1.3807615230460922e-05,
+      "loss": 0.009,
+      "step": 3624
+    },
+    {
+      "epoch": 14.5,
+      "grad_norm": 0.23491171002388,
+      "learning_rate": 1.3797595190380763e-05,
+      "loss": 0.0091,
+      "step": 3625
+    },
+    {
+      "epoch": 14.504,
+      "grad_norm": 0.20434904098510742,
+      "learning_rate": 1.3787575150300602e-05,
+      "loss": 0.0086,
+      "step": 3626
+    },
+    {
+      "epoch": 14.508,
+      "grad_norm": 0.16007202863693237,
+      "learning_rate": 1.3777555110220442e-05,
+      "loss": 0.0075,
+      "step": 3627
+    },
+    {
+      "epoch": 14.512,
+      "grad_norm": 0.21671073138713837,
+      "learning_rate": 1.376753507014028e-05,
+      "loss": 0.0085,
+      "step": 3628
+    },
+    {
+      "epoch": 14.516,
+      "grad_norm": 0.23178523778915405,
+      "learning_rate": 1.375751503006012e-05,
+      "loss": 0.009,
+      "step": 3629
+    },
+    {
+      "epoch": 14.52,
+      "grad_norm": 0.2952883541584015,
+      "learning_rate": 1.374749498997996e-05,
+      "loss": 0.0104,
+      "step": 3630
+    },
+    {
+      "epoch": 14.524000000000001,
+      "grad_norm": 0.22766168415546417,
+      "learning_rate": 1.37374749498998e-05,
+      "loss": 0.0082,
+      "step": 3631
+    },
+    {
+      "epoch": 14.528,
+      "grad_norm": 0.17687074840068817,
+      "learning_rate": 1.372745490981964e-05,
+      "loss": 0.0077,
+      "step": 3632
+    },
+    {
+      "epoch": 14.532,
+      "grad_norm": 0.2766241133213043,
+      "learning_rate": 1.371743486973948e-05,
+      "loss": 0.0097,
+      "step": 3633
+    },
+    {
+      "epoch": 14.536,
+      "grad_norm": 0.2916935980319977,
+      "learning_rate": 1.3707414829659317e-05,
+      "loss": 0.011,
+      "step": 3634
+    },
+    {
+      "epoch": 14.54,
+      "grad_norm": 0.4812479615211487,
+      "learning_rate": 1.369739478957916e-05,
+      "loss": 0.0104,
+      "step": 3635
+    },
+    {
+      "epoch": 14.544,
+      "grad_norm": 0.20633623003959656,
+      "learning_rate": 1.3687374749498998e-05,
+      "loss": 0.008,
+      "step": 3636
+    },
+    {
+      "epoch": 14.548,
+      "grad_norm": 0.20415335893630981,
+      "learning_rate": 1.3677354709418838e-05,
+      "loss": 0.0083,
+      "step": 3637
+    },
+    {
+      "epoch": 14.552,
+      "grad_norm": 0.23814812302589417,
+      "learning_rate": 1.3667334669338678e-05,
+      "loss": 0.0095,
+      "step": 3638
+    },
+    {
+      "epoch": 14.556000000000001,
+      "grad_norm": 0.33633625507354736,
+      "learning_rate": 1.3657314629258517e-05,
+      "loss": 0.0099,
+      "step": 3639
+    },
+    {
+      "epoch": 14.56,
+      "grad_norm": 0.3303033113479614,
+      "learning_rate": 1.3647294589178358e-05,
+      "loss": 0.0094,
+      "step": 3640
+    },
+    {
+      "epoch": 14.564,
+      "grad_norm": 0.2839348316192627,
+      "learning_rate": 1.3637274549098198e-05,
+      "loss": 0.0103,
+      "step": 3641
+    },
+    {
+      "epoch": 14.568,
+      "grad_norm": 0.2589602470397949,
+      "learning_rate": 1.3627254509018036e-05,
+      "loss": 0.0086,
+      "step": 3642
+    },
+    {
+      "epoch": 14.572,
+      "grad_norm": 0.17141884565353394,
+      "learning_rate": 1.3617234468937875e-05,
+      "loss": 0.0085,
+      "step": 3643
+    },
+    {
+      "epoch": 14.576,
+      "grad_norm": 0.20422735810279846,
+      "learning_rate": 1.3607214428857717e-05,
+      "loss": 0.0087,
+      "step": 3644
+    },
+    {
+      "epoch": 14.58,
+      "grad_norm": 0.36146315932273865,
+      "learning_rate": 1.3597194388777556e-05,
+      "loss": 0.0103,
+      "step": 3645
+    },
+    {
+      "epoch": 14.584,
+      "grad_norm": 0.3737517297267914,
+      "learning_rate": 1.3587174348697396e-05,
+      "loss": 0.0085,
+      "step": 3646
+    },
+    {
+      "epoch": 14.588,
+      "grad_norm": 0.20324692130088806,
+      "learning_rate": 1.3577154308617236e-05,
+      "loss": 0.0083,
+      "step": 3647
+    },
+    {
+      "epoch": 14.592,
+      "grad_norm": 0.21344539523124695,
+      "learning_rate": 1.3567134268537073e-05,
+      "loss": 0.009,
+      "step": 3648
+    },
+    {
+      "epoch": 14.596,
+      "grad_norm": 0.3655132055282593,
+      "learning_rate": 1.3557114228456916e-05,
+      "loss": 0.011,
+      "step": 3649
+    },
+    {
+      "epoch": 14.6,
+      "grad_norm": 0.20782031118869781,
+      "learning_rate": 1.3547094188376754e-05,
+      "loss": 0.0089,
+      "step": 3650
+    },
+    {
+      "epoch": 14.604,
+      "grad_norm": 0.28657761216163635,
+      "learning_rate": 1.3537074148296594e-05,
+      "loss": 0.0093,
+      "step": 3651
+    },
+    {
+      "epoch": 14.608,
+      "grad_norm": 0.25337931513786316,
+      "learning_rate": 1.3527054108216433e-05,
+      "loss": 0.0093,
+      "step": 3652
+    },
+    {
+      "epoch": 14.612,
+      "grad_norm": 0.36946016550064087,
+      "learning_rate": 1.3517034068136273e-05,
+      "loss": 0.0113,
+      "step": 3653
+    },
+    {
+      "epoch": 14.616,
+      "grad_norm": 0.18227268755435944,
+      "learning_rate": 1.3507014028056114e-05,
+      "loss": 0.0084,
+      "step": 3654
+    },
+    {
+      "epoch": 14.62,
+      "grad_norm": 0.1899784952402115,
+      "learning_rate": 1.3496993987975954e-05,
+      "loss": 0.0085,
+      "step": 3655
+    },
+    {
+      "epoch": 14.624,
+      "grad_norm": 0.16006070375442505,
+      "learning_rate": 1.3486973947895792e-05,
+      "loss": 0.0086,
+      "step": 3656
+    },
+    {
+      "epoch": 14.628,
+      "grad_norm": 0.2663494348526001,
+      "learning_rate": 1.3476953907815631e-05,
+      "loss": 0.0104,
+      "step": 3657
+    },
+    {
+      "epoch": 14.632,
+      "grad_norm": 0.19320881366729736,
+      "learning_rate": 1.3466933867735471e-05,
+      "loss": 0.0092,
+      "step": 3658
+    },
+    {
+      "epoch": 14.636,
+      "grad_norm": 0.5768705010414124,
+      "learning_rate": 1.3456913827655312e-05,
+      "loss": 0.0127,
+      "step": 3659
+    },
+    {
+      "epoch": 14.64,
+      "grad_norm": 0.15998607873916626,
+      "learning_rate": 1.3446893787575152e-05,
+      "loss": 0.0089,
+      "step": 3660
+    },
+    {
+      "epoch": 14.644,
+      "grad_norm": 0.21522219479084015,
+      "learning_rate": 1.3436873747494991e-05,
+      "loss": 0.0079,
+      "step": 3661
+    },
+    {
+      "epoch": 14.648,
+      "grad_norm": 0.2248634248971939,
+      "learning_rate": 1.342685370741483e-05,
+      "loss": 0.0098,
+      "step": 3662
+    },
+    {
+      "epoch": 14.652,
+      "grad_norm": 0.23990559577941895,
+      "learning_rate": 1.3416833667334669e-05,
+      "loss": 0.0096,
+      "step": 3663
+    },
+    {
+      "epoch": 14.656,
+      "grad_norm": 0.18346965312957764,
+      "learning_rate": 1.340681362725451e-05,
+      "loss": 0.0081,
+      "step": 3664
+    },
+    {
+      "epoch": 14.66,
+      "grad_norm": 0.46722784638404846,
+      "learning_rate": 1.339679358717435e-05,
+      "loss": 0.0119,
+      "step": 3665
+    },
+    {
+      "epoch": 14.664,
+      "grad_norm": 0.19062618911266327,
+      "learning_rate": 1.338677354709419e-05,
+      "loss": 0.0082,
+      "step": 3666
+    },
+    {
+      "epoch": 14.668,
+      "grad_norm": 0.40659964084625244,
+      "learning_rate": 1.3376753507014029e-05,
+      "loss": 0.0097,
+      "step": 3667
+    },
+    {
+      "epoch": 14.672,
+      "grad_norm": 0.22992847859859467,
+      "learning_rate": 1.3366733466933867e-05,
+      "loss": 0.0094,
+      "step": 3668
+    },
+    {
+      "epoch": 14.676,
+      "grad_norm": 0.3991800844669342,
+      "learning_rate": 1.335671342685371e-05,
+      "loss": 0.0129,
+      "step": 3669
+    },
+    {
+      "epoch": 14.68,
+      "grad_norm": 0.1878584921360016,
+      "learning_rate": 1.3346693386773548e-05,
+      "loss": 0.0085,
+      "step": 3670
+    },
+    {
+      "epoch": 14.684,
+      "grad_norm": 0.1736784130334854,
+      "learning_rate": 1.3336673346693387e-05,
+      "loss": 0.0094,
+      "step": 3671
+    },
+    {
+      "epoch": 14.688,
+      "grad_norm": 0.18842560052871704,
+      "learning_rate": 1.3326653306613227e-05,
+      "loss": 0.0085,
+      "step": 3672
+    },
+    {
+      "epoch": 14.692,
+      "grad_norm": 0.2932375967502594,
+      "learning_rate": 1.3316633266533065e-05,
+      "loss": 0.0094,
+      "step": 3673
+    },
+    {
+      "epoch": 14.696,
+      "grad_norm": 0.5152665376663208,
+      "learning_rate": 1.3306613226452908e-05,
+      "loss": 0.0096,
+      "step": 3674
+    },
+    {
+      "epoch": 14.7,
+      "grad_norm": 0.34712889790534973,
+      "learning_rate": 1.3296593186372747e-05,
+      "loss": 0.0092,
+      "step": 3675
+    },
+    {
+      "epoch": 14.704,
+      "grad_norm": 0.18002943694591522,
+      "learning_rate": 1.3286573146292585e-05,
+      "loss": 0.0089,
+      "step": 3676
+    },
+    {
+      "epoch": 14.708,
+      "grad_norm": 0.26756277680397034,
+      "learning_rate": 1.3276553106212425e-05,
+      "loss": 0.0092,
+      "step": 3677
+    },
+    {
+      "epoch": 14.712,
+      "grad_norm": 0.24038799107074738,
+      "learning_rate": 1.3266533066132264e-05,
+      "loss": 0.01,
+      "step": 3678
+    },
+    {
+      "epoch": 14.716,
+      "grad_norm": 0.526962161064148,
+      "learning_rate": 1.3256513026052106e-05,
+      "loss": 0.0112,
+      "step": 3679
+    },
+    {
+      "epoch": 14.72,
+      "grad_norm": 0.184894397854805,
+      "learning_rate": 1.3246492985971945e-05,
+      "loss": 0.008,
+      "step": 3680
+    },
+    {
+      "epoch": 14.724,
+      "grad_norm": 0.2070516049861908,
+      "learning_rate": 1.3236472945891785e-05,
+      "loss": 0.0087,
+      "step": 3681
+    },
+    {
+      "epoch": 14.728,
+      "grad_norm": 0.23932699859142303,
+      "learning_rate": 1.3226452905811623e-05,
+      "loss": 0.0099,
+      "step": 3682
+    },
+    {
+      "epoch": 14.732,
+      "grad_norm": 0.24801121652126312,
+      "learning_rate": 1.3216432865731462e-05,
+      "loss": 0.0094,
+      "step": 3683
+    },
+    {
+      "epoch": 14.736,
+      "grad_norm": 0.21665263175964355,
+      "learning_rate": 1.3206412825651304e-05,
+      "loss": 0.0094,
+      "step": 3684
+    },
+    {
+      "epoch": 14.74,
+      "grad_norm": 0.19703127443790436,
+      "learning_rate": 1.3196392785571143e-05,
+      "loss": 0.008,
+      "step": 3685
+    },
+    {
+      "epoch": 14.744,
+      "grad_norm": 0.40190309286117554,
+      "learning_rate": 1.3186372745490983e-05,
+      "loss": 0.0089,
+      "step": 3686
+    },
+    {
+      "epoch": 14.748,
+      "grad_norm": 0.2351587414741516,
+      "learning_rate": 1.317635270541082e-05,
+      "loss": 0.0086,
+      "step": 3687
+    },
+    {
+      "epoch": 14.752,
+      "grad_norm": 0.26315975189208984,
+      "learning_rate": 1.316633266533066e-05,
+      "loss": 0.01,
+      "step": 3688
+    },
+    {
+      "epoch": 14.756,
+      "grad_norm": 0.33228597044944763,
+      "learning_rate": 1.3156312625250503e-05,
+      "loss": 0.0115,
+      "step": 3689
+    },
+    {
+      "epoch": 14.76,
+      "grad_norm": 0.2187896966934204,
+      "learning_rate": 1.3146292585170341e-05,
+      "loss": 0.0086,
+      "step": 3690
+    },
+    {
+      "epoch": 14.764,
+      "grad_norm": 0.2327001541852951,
+      "learning_rate": 1.313627254509018e-05,
+      "loss": 0.0096,
+      "step": 3691
+    },
+    {
+      "epoch": 14.768,
+      "grad_norm": 0.1458553969860077,
+      "learning_rate": 1.312625250501002e-05,
+      "loss": 0.0083,
+      "step": 3692
+    },
+    {
+      "epoch": 14.772,
+      "grad_norm": 0.2837975323200226,
+      "learning_rate": 1.3116232464929858e-05,
+      "loss": 0.0096,
+      "step": 3693
+    },
+    {
+      "epoch": 14.776,
+      "grad_norm": 0.2698477804660797,
+      "learning_rate": 1.3106212424849701e-05,
+      "loss": 0.0095,
+      "step": 3694
+    },
+    {
+      "epoch": 14.78,
+      "grad_norm": 0.6040169596672058,
+      "learning_rate": 1.309619238476954e-05,
+      "loss": 0.0119,
+      "step": 3695
+    },
+    {
+      "epoch": 14.784,
+      "grad_norm": 0.14232851564884186,
+      "learning_rate": 1.3086172344689379e-05,
+      "loss": 0.0085,
+      "step": 3696
+    },
+    {
+      "epoch": 14.788,
+      "grad_norm": 0.24106432497501373,
+      "learning_rate": 1.3076152304609218e-05,
+      "loss": 0.0088,
+      "step": 3697
+    },
+    {
+      "epoch": 14.792,
+      "grad_norm": 0.2642269432544708,
+      "learning_rate": 1.3066132264529058e-05,
+      "loss": 0.0098,
+      "step": 3698
+    },
+    {
+      "epoch": 14.796,
+      "grad_norm": 0.23594169318675995,
+      "learning_rate": 1.30561122244489e-05,
+      "loss": 0.0092,
+      "step": 3699
+    },
+    {
+      "epoch": 14.8,
+      "grad_norm": 0.24775654077529907,
+      "learning_rate": 1.3046092184368739e-05,
+      "loss": 0.0092,
+      "step": 3700
+    },
+    {
+      "epoch": 14.804,
+      "grad_norm": 0.2097250074148178,
+      "learning_rate": 1.3036072144288577e-05,
+      "loss": 0.0096,
+      "step": 3701
+    },
+    {
+      "epoch": 14.808,
+      "grad_norm": 0.2012004554271698,
+      "learning_rate": 1.3026052104208416e-05,
+      "loss": 0.0092,
+      "step": 3702
+    },
+    {
+      "epoch": 14.812,
+      "grad_norm": 0.20894677937030792,
+      "learning_rate": 1.301603206412826e-05,
+      "loss": 0.0093,
+      "step": 3703
+    },
+    {
+      "epoch": 14.816,
+      "grad_norm": 0.23382696509361267,
+      "learning_rate": 1.3006012024048097e-05,
+      "loss": 0.009,
+      "step": 3704
+    },
+    {
+      "epoch": 14.82,
+      "grad_norm": 0.19932790100574493,
+      "learning_rate": 1.2995991983967937e-05,
+      "loss": 0.0103,
+      "step": 3705
+    },
+    {
+      "epoch": 14.824,
+      "grad_norm": 0.28219786286354065,
+      "learning_rate": 1.2985971943887776e-05,
+      "loss": 0.01,
+      "step": 3706
+    },
+    {
+      "epoch": 14.828,
+      "grad_norm": 0.2530341148376465,
+      "learning_rate": 1.2975951903807614e-05,
+      "loss": 0.0092,
+      "step": 3707
+    },
+    {
+      "epoch": 14.832,
+      "grad_norm": 0.26421231031417847,
+      "learning_rate": 1.2965931863727457e-05,
+      "loss": 0.0097,
+      "step": 3708
+    },
+    {
+      "epoch": 14.836,
+      "grad_norm": 0.1763448268175125,
+      "learning_rate": 1.2955911823647297e-05,
+      "loss": 0.0084,
+      "step": 3709
+    },
+    {
+      "epoch": 14.84,
+      "grad_norm": 0.26100030541419983,
+      "learning_rate": 1.2945891783567135e-05,
+      "loss": 0.0102,
+      "step": 3710
+    },
+    {
+      "epoch": 14.844,
+      "grad_norm": 0.22259047627449036,
+      "learning_rate": 1.2935871743486974e-05,
+      "loss": 0.0092,
+      "step": 3711
+    },
+    {
+      "epoch": 14.848,
+      "grad_norm": 0.19816388189792633,
+      "learning_rate": 1.2925851703406814e-05,
+      "loss": 0.0059,
+      "step": 3712
+    },
+    {
+      "epoch": 14.852,
+      "grad_norm": 0.270816832780838,
+      "learning_rate": 1.2915831663326655e-05,
+      "loss": 0.01,
+      "step": 3713
+    },
+    {
+      "epoch": 14.856,
+      "grad_norm": 0.25951826572418213,
+      "learning_rate": 1.2905811623246495e-05,
+      "loss": 0.0098,
+      "step": 3714
+    },
+    {
+      "epoch": 14.86,
+      "grad_norm": 0.28285837173461914,
+      "learning_rate": 1.2895791583166333e-05,
+      "loss": 0.0093,
+      "step": 3715
+    },
+    {
+      "epoch": 14.864,
+      "grad_norm": 0.42073768377304077,
+      "learning_rate": 1.2885771543086172e-05,
+      "loss": 0.0108,
+      "step": 3716
+    },
+    {
+      "epoch": 14.868,
+      "grad_norm": 0.506669282913208,
+      "learning_rate": 1.2875751503006012e-05,
+      "loss": 0.0094,
+      "step": 3717
+    },
+    {
+      "epoch": 14.872,
+      "grad_norm": 0.18972696363925934,
+      "learning_rate": 1.2865731462925853e-05,
+      "loss": 0.009,
+      "step": 3718
+    },
+    {
+      "epoch": 14.876,
+      "grad_norm": 0.17668847739696503,
+      "learning_rate": 1.2855711422845693e-05,
+      "loss": 0.0082,
+      "step": 3719
+    },
+    {
+      "epoch": 14.88,
+      "grad_norm": 0.30280017852783203,
+      "learning_rate": 1.2845691382765532e-05,
+      "loss": 0.0109,
+      "step": 3720
+    },
+    {
+      "epoch": 14.884,
+      "grad_norm": 0.21172870695590973,
+      "learning_rate": 1.283567134268537e-05,
+      "loss": 0.0094,
+      "step": 3721
+    },
+    {
+      "epoch": 14.888,
+      "grad_norm": 0.17289844155311584,
+      "learning_rate": 1.282565130260521e-05,
+      "loss": 0.0074,
+      "step": 3722
+    },
+    {
+      "epoch": 14.892,
+      "grad_norm": 0.17328234016895294,
+      "learning_rate": 1.2815631262525051e-05,
+      "loss": 0.0082,
+      "step": 3723
+    },
+    {
+      "epoch": 14.896,
+      "grad_norm": 0.2376927137374878,
+      "learning_rate": 1.280561122244489e-05,
+      "loss": 0.01,
+      "step": 3724
+    },
+    {
+      "epoch": 14.9,
+      "grad_norm": 0.21705985069274902,
+      "learning_rate": 1.279559118236473e-05,
+      "loss": 0.0086,
+      "step": 3725
+    },
+    {
+      "epoch": 14.904,
+      "grad_norm": 0.272956520318985,
+      "learning_rate": 1.278557114228457e-05,
+      "loss": 0.0092,
+      "step": 3726
+    },
+    {
+      "epoch": 14.908,
+      "grad_norm": 0.18439574539661407,
+      "learning_rate": 1.2775551102204408e-05,
+      "loss": 0.0084,
+      "step": 3727
+    },
+    {
+      "epoch": 14.912,
+      "grad_norm": 0.19920729100704193,
+      "learning_rate": 1.276553106212425e-05,
+      "loss": 0.0091,
+      "step": 3728
+    },
+    {
+      "epoch": 14.916,
+      "grad_norm": 0.2559821605682373,
+      "learning_rate": 1.2755511022044089e-05,
+      "loss": 0.0091,
+      "step": 3729
+    },
+    {
+      "epoch": 14.92,
+      "grad_norm": 0.23575206100940704,
+      "learning_rate": 1.2745490981963928e-05,
+      "loss": 0.0084,
+      "step": 3730
+    },
+    {
+      "epoch": 14.924,
+      "grad_norm": 0.2531851828098297,
+      "learning_rate": 1.2735470941883768e-05,
+      "loss": 0.0089,
+      "step": 3731
+    },
+    {
+      "epoch": 14.928,
+      "grad_norm": 0.21447163820266724,
+      "learning_rate": 1.2725450901803607e-05,
+      "loss": 0.0099,
+      "step": 3732
+    },
+    {
+      "epoch": 14.932,
+      "grad_norm": 0.23144927620887756,
+      "learning_rate": 1.2715430861723449e-05,
+      "loss": 0.0098,
+      "step": 3733
+    },
+    {
+      "epoch": 14.936,
+      "grad_norm": 0.16847439110279083,
+      "learning_rate": 1.2705410821643288e-05,
+      "loss": 0.0083,
+      "step": 3734
+    },
+    {
+      "epoch": 14.94,
+      "grad_norm": 0.25669029355049133,
+      "learning_rate": 1.2695390781563126e-05,
+      "loss": 0.0097,
+      "step": 3735
+    },
+    {
+      "epoch": 14.943999999999999,
+      "grad_norm": 0.3748454451560974,
+      "learning_rate": 1.2685370741482966e-05,
+      "loss": 0.0097,
+      "step": 3736
+    },
+    {
+      "epoch": 14.948,
+      "grad_norm": 0.2604009509086609,
+      "learning_rate": 1.2675350701402805e-05,
+      "loss": 0.0094,
+      "step": 3737
+    },
+    {
+      "epoch": 14.952,
+      "grad_norm": 0.09197622537612915,
+      "learning_rate": 1.2665330661322647e-05,
+      "loss": 0.004,
+      "step": 3738
+    },
+    {
+      "epoch": 14.956,
+      "grad_norm": 0.2926792800426483,
+      "learning_rate": 1.2655310621242486e-05,
+      "loss": 0.0103,
+      "step": 3739
+    },
+    {
+      "epoch": 14.96,
+      "grad_norm": 0.19025073945522308,
+      "learning_rate": 1.2645290581162326e-05,
+      "loss": 0.009,
+      "step": 3740
+    },
+    {
+      "epoch": 14.964,
+      "grad_norm": 0.17044049501419067,
+      "learning_rate": 1.2635270541082164e-05,
+      "loss": 0.006,
+      "step": 3741
+    },
+    {
+      "epoch": 14.968,
+      "grad_norm": 0.16821922361850739,
+      "learning_rate": 1.2625250501002003e-05,
+      "loss": 0.006,
+      "step": 3742
+    },
+    {
+      "epoch": 14.972,
+      "grad_norm": 0.2913142740726471,
+      "learning_rate": 1.2615230460921844e-05,
+      "loss": 0.0097,
+      "step": 3743
+    },
+    {
+      "epoch": 14.975999999999999,
+      "grad_norm": 0.48879778385162354,
+      "learning_rate": 1.2605210420841684e-05,
+      "loss": 0.0101,
+      "step": 3744
+    },
+    {
+      "epoch": 14.98,
+      "grad_norm": 0.2455979734659195,
+      "learning_rate": 1.2595190380761524e-05,
+      "loss": 0.009,
+      "step": 3745
+    },
+    {
+      "epoch": 14.984,
+      "grad_norm": 0.20019696652889252,
+      "learning_rate": 1.2585170340681363e-05,
+      "loss": 0.009,
+      "step": 3746
+    },
+    {
+      "epoch": 14.988,
+      "grad_norm": 0.30625882744789124,
+      "learning_rate": 1.2575150300601201e-05,
+      "loss": 0.0101,
+      "step": 3747
+    },
+    {
+      "epoch": 14.992,
+      "grad_norm": 0.18524271249771118,
+      "learning_rate": 1.2565130260521044e-05,
+      "loss": 0.0093,
+      "step": 3748
+    },
+    {
+      "epoch": 14.996,
+      "grad_norm": 0.33599841594696045,
+      "learning_rate": 1.2555110220440882e-05,
+      "loss": 0.0118,
+      "step": 3749
+    },
+    {
+      "epoch": 15.0,
+      "grad_norm": 0.17473024129867554,
+      "learning_rate": 1.2545090180360722e-05,
+      "loss": 0.006,
+      "step": 3750
+    },
+    {
+      "epoch": 15.004,
+      "grad_norm": 0.19427764415740967,
+      "learning_rate": 1.2535070140280561e-05,
+      "loss": 0.0093,
+      "step": 3751
+    },
+    {
+      "epoch": 15.008,
+      "grad_norm": 0.1349514275789261,
+      "learning_rate": 1.25250501002004e-05,
+      "loss": 0.0072,
+      "step": 3752
+    },
+    {
+      "epoch": 15.012,
+      "grad_norm": 0.11184310913085938,
+      "learning_rate": 1.2515030060120242e-05,
+      "loss": 0.0056,
+      "step": 3753
+    },
+    {
+      "epoch": 15.016,
+      "grad_norm": 0.22410862147808075,
+      "learning_rate": 1.2505010020040082e-05,
+      "loss": 0.0085,
+      "step": 3754
+    },
+    {
+      "epoch": 15.02,
+      "grad_norm": 0.20344886183738708,
+      "learning_rate": 1.249498997995992e-05,
+      "loss": 0.0077,
+      "step": 3755
+    },
+    {
+      "epoch": 15.024,
+      "grad_norm": 0.20372788608074188,
+      "learning_rate": 1.248496993987976e-05,
+      "loss": 0.0083,
+      "step": 3756
+    },
+    {
+      "epoch": 15.028,
+      "grad_norm": 0.31820449233055115,
+      "learning_rate": 1.24749498997996e-05,
+      "loss": 0.0081,
+      "step": 3757
+    },
+    {
+      "epoch": 15.032,
+      "grad_norm": 0.2090875804424286,
+      "learning_rate": 1.2464929859719438e-05,
+      "loss": 0.0087,
+      "step": 3758
+    },
+    {
+      "epoch": 15.036,
+      "grad_norm": 0.15436923503875732,
+      "learning_rate": 1.245490981963928e-05,
+      "loss": 0.0069,
+      "step": 3759
+    },
+    {
+      "epoch": 15.04,
+      "grad_norm": 0.17816013097763062,
+      "learning_rate": 1.244488977955912e-05,
+      "loss": 0.0087,
+      "step": 3760
+    },
+    {
+      "epoch": 15.044,
+      "grad_norm": 0.14754058420658112,
+      "learning_rate": 1.2434869739478959e-05,
+      "loss": 0.0072,
+      "step": 3761
+    },
+    {
+      "epoch": 15.048,
+      "grad_norm": 0.29081836342811584,
+      "learning_rate": 1.2424849699398798e-05,
+      "loss": 0.0085,
+      "step": 3762
+    },
+    {
+      "epoch": 15.052,
+      "grad_norm": 0.14590558409690857,
+      "learning_rate": 1.2414829659318638e-05,
+      "loss": 0.0067,
+      "step": 3763
+    },
+    {
+      "epoch": 15.056,
+      "grad_norm": 0.1646532565355301,
+      "learning_rate": 1.2404809619238478e-05,
+      "loss": 0.0072,
+      "step": 3764
+    },
+    {
+      "epoch": 15.06,
+      "grad_norm": 0.3623543977737427,
+      "learning_rate": 1.2394789579158317e-05,
+      "loss": 0.0097,
+      "step": 3765
+    },
+    {
+      "epoch": 15.064,
+      "grad_norm": 0.074567049741745,
+      "learning_rate": 1.2384769539078157e-05,
+      "loss": 0.0032,
+      "step": 3766
+    },
+    {
+      "epoch": 15.068,
+      "grad_norm": 0.25378715991973877,
+      "learning_rate": 1.2374749498997996e-05,
+      "loss": 0.0085,
+      "step": 3767
+    },
+    {
+      "epoch": 15.072,
+      "grad_norm": 0.14567115902900696,
+      "learning_rate": 1.2364729458917836e-05,
+      "loss": 0.0076,
+      "step": 3768
+    },
+    {
+      "epoch": 15.076,
+      "grad_norm": 0.13616079092025757,
+      "learning_rate": 1.2354709418837675e-05,
+      "loss": 0.0077,
+      "step": 3769
+    },
+    {
+      "epoch": 15.08,
+      "grad_norm": 0.13720935583114624,
+      "learning_rate": 1.2344689378757515e-05,
+      "loss": 0.0076,
+      "step": 3770
+    },
+    {
+      "epoch": 15.084,
+      "grad_norm": 0.13663603365421295,
+      "learning_rate": 1.2334669338677356e-05,
+      "loss": 0.0045,
+      "step": 3771
+    },
+    {
+      "epoch": 15.088,
+      "grad_norm": 0.16399826109409332,
+      "learning_rate": 1.2324649298597194e-05,
+      "loss": 0.0079,
+      "step": 3772
+    },
+    {
+      "epoch": 15.092,
+      "grad_norm": 0.25261420011520386,
+      "learning_rate": 1.2314629258517036e-05,
+      "loss": 0.0087,
+      "step": 3773
+    },
+    {
+      "epoch": 15.096,
+      "grad_norm": 0.17582829296588898,
+      "learning_rate": 1.2304609218436875e-05,
+      "loss": 0.0068,
+      "step": 3774
+    },
+    {
+      "epoch": 15.1,
+      "grad_norm": 0.18607257306575775,
+      "learning_rate": 1.2294589178356713e-05,
+      "loss": 0.0079,
+      "step": 3775
+    },
+    {
+      "epoch": 15.104,
+      "grad_norm": 0.19465355575084686,
+      "learning_rate": 1.2284569138276554e-05,
+      "loss": 0.0067,
+      "step": 3776
+    },
+    {
+      "epoch": 15.108,
+      "grad_norm": 0.28906580805778503,
+      "learning_rate": 1.2274549098196394e-05,
+      "loss": 0.0092,
+      "step": 3777
+    },
+    {
+      "epoch": 15.112,
+      "grad_norm": 0.2137903869152069,
+      "learning_rate": 1.2264529058116233e-05,
+      "loss": 0.0084,
+      "step": 3778
+    },
+    {
+      "epoch": 15.116,
+      "grad_norm": 0.18388208746910095,
+      "learning_rate": 1.2254509018036073e-05,
+      "loss": 0.0075,
+      "step": 3779
+    },
+    {
+      "epoch": 15.12,
+      "grad_norm": 0.237356498837471,
+      "learning_rate": 1.2244488977955913e-05,
+      "loss": 0.008,
+      "step": 3780
+    },
+    {
+      "epoch": 15.124,
+      "grad_norm": 0.22582365572452545,
+      "learning_rate": 1.2234468937875752e-05,
+      "loss": 0.0094,
+      "step": 3781
+    },
+    {
+      "epoch": 15.128,
+      "grad_norm": 0.150029718875885,
+      "learning_rate": 1.2224448897795592e-05,
+      "loss": 0.008,
+      "step": 3782
+    },
+    {
+      "epoch": 15.132,
+      "grad_norm": 0.2120167315006256,
+      "learning_rate": 1.2214428857715431e-05,
+      "loss": 0.0091,
+      "step": 3783
+    },
+    {
+      "epoch": 15.136,
+      "grad_norm": 0.18410436809062958,
+      "learning_rate": 1.2204408817635271e-05,
+      "loss": 0.0073,
+      "step": 3784
+    },
+    {
+      "epoch": 15.14,
+      "grad_norm": 0.18777211010456085,
+      "learning_rate": 1.219438877755511e-05,
+      "loss": 0.0084,
+      "step": 3785
+    },
+    {
+      "epoch": 15.144,
+      "grad_norm": 0.1506580412387848,
+      "learning_rate": 1.218436873747495e-05,
+      "loss": 0.0077,
+      "step": 3786
+    },
+    {
+      "epoch": 15.148,
+      "grad_norm": 0.1882610023021698,
+      "learning_rate": 1.217434869739479e-05,
+      "loss": 0.0072,
+      "step": 3787
+    },
+    {
+      "epoch": 15.152,
+      "grad_norm": 0.14596189558506012,
+      "learning_rate": 1.2164328657314631e-05,
+      "loss": 0.0074,
+      "step": 3788
+    },
+    {
+      "epoch": 15.156,
+      "grad_norm": 0.14775827527046204,
+      "learning_rate": 1.2154308617234469e-05,
+      "loss": 0.0074,
+      "step": 3789
+    },
+    {
+      "epoch": 15.16,
+      "grad_norm": 0.1532842367887497,
+      "learning_rate": 1.2144288577154309e-05,
+      "loss": 0.0071,
+      "step": 3790
+    },
+    {
+      "epoch": 15.164,
+      "grad_norm": 0.18244397640228271,
+      "learning_rate": 1.213426853707415e-05,
+      "loss": 0.0085,
+      "step": 3791
+    },
+    {
+      "epoch": 15.168,
+      "grad_norm": 0.17389196157455444,
+      "learning_rate": 1.2124248496993988e-05,
+      "loss": 0.0081,
+      "step": 3792
+    },
+    {
+      "epoch": 15.172,
+      "grad_norm": 0.24255871772766113,
+      "learning_rate": 1.2114228456913829e-05,
+      "loss": 0.0095,
+      "step": 3793
+    },
+    {
+      "epoch": 15.176,
+      "grad_norm": 0.13241663575172424,
+      "learning_rate": 1.2104208416833669e-05,
+      "loss": 0.0066,
+      "step": 3794
+    },
+    {
+      "epoch": 15.18,
+      "grad_norm": 0.2261979728937149,
+      "learning_rate": 1.2094188376753506e-05,
+      "loss": 0.0105,
+      "step": 3795
+    },
+    {
+      "epoch": 15.184,
+      "grad_norm": 0.1342063695192337,
+      "learning_rate": 1.2084168336673348e-05,
+      "loss": 0.0073,
+      "step": 3796
+    },
+    {
+      "epoch": 15.188,
+      "grad_norm": 0.18595094978809357,
+      "learning_rate": 1.2074148296593187e-05,
+      "loss": 0.0078,
+      "step": 3797
+    },
+    {
+      "epoch": 15.192,
+      "grad_norm": 0.21512697637081146,
+      "learning_rate": 1.2064128256513027e-05,
+      "loss": 0.0083,
+      "step": 3798
+    },
+    {
+      "epoch": 15.196,
+      "grad_norm": 0.19803418219089508,
+      "learning_rate": 1.2054108216432867e-05,
+      "loss": 0.0079,
+      "step": 3799
+    },
+    {
+      "epoch": 15.2,
+      "grad_norm": 0.2116500288248062,
+      "learning_rate": 1.2044088176352706e-05,
+      "loss": 0.0083,
+      "step": 3800
+    },
+    {
+      "epoch": 15.204,
+      "grad_norm": 0.18894079327583313,
+      "learning_rate": 1.2034068136272546e-05,
+      "loss": 0.0075,
+      "step": 3801
+    },
+    {
+      "epoch": 15.208,
+      "grad_norm": 0.30808115005493164,
+      "learning_rate": 1.2024048096192385e-05,
+      "loss": 0.0085,
+      "step": 3802
+    },
+    {
+      "epoch": 15.212,
+      "grad_norm": 0.17777995765209198,
+      "learning_rate": 1.2014028056112225e-05,
+      "loss": 0.0078,
+      "step": 3803
+    },
+    {
+      "epoch": 15.216,
+      "grad_norm": 0.28307095170021057,
+      "learning_rate": 1.2004008016032064e-05,
+      "loss": 0.0089,
+      "step": 3804
+    },
+    {
+      "epoch": 15.22,
+      "grad_norm": 0.2219671756029129,
+      "learning_rate": 1.1993987975951906e-05,
+      "loss": 0.009,
+      "step": 3805
+    },
+    {
+      "epoch": 15.224,
+      "grad_norm": 0.1308339238166809,
+      "learning_rate": 1.1983967935871744e-05,
+      "loss": 0.008,
+      "step": 3806
+    },
+    {
+      "epoch": 15.228,
+      "grad_norm": 0.1603306233882904,
+      "learning_rate": 1.1973947895791583e-05,
+      "loss": 0.0073,
+      "step": 3807
+    },
+    {
+      "epoch": 15.232,
+      "grad_norm": 0.21934445202350616,
+      "learning_rate": 1.1963927855711425e-05,
+      "loss": 0.0089,
+      "step": 3808
+    },
+    {
+      "epoch": 15.236,
+      "grad_norm": 0.16111411154270172,
+      "learning_rate": 1.1953907815631262e-05,
+      "loss": 0.0081,
+      "step": 3809
+    },
+    {
+      "epoch": 15.24,
+      "grad_norm": 0.1615634709596634,
+      "learning_rate": 1.1943887775551104e-05,
+      "loss": 0.0078,
+      "step": 3810
+    },
+    {
+      "epoch": 15.244,
+      "grad_norm": 0.1142813116312027,
+      "learning_rate": 1.1933867735470943e-05,
+      "loss": 0.0042,
+      "step": 3811
+    },
+    {
+      "epoch": 15.248,
+      "grad_norm": 0.20265451073646545,
+      "learning_rate": 1.1923847695390781e-05,
+      "loss": 0.0075,
+      "step": 3812
+    },
+    {
+      "epoch": 15.252,
+      "grad_norm": 0.17693500220775604,
+      "learning_rate": 1.1913827655310622e-05,
+      "loss": 0.0091,
+      "step": 3813
+    },
+    {
+      "epoch": 15.256,
+      "grad_norm": 0.16070616245269775,
+      "learning_rate": 1.190380761523046e-05,
+      "loss": 0.0084,
+      "step": 3814
+    },
+    {
+      "epoch": 15.26,
+      "grad_norm": 0.2477542757987976,
+      "learning_rate": 1.1893787575150302e-05,
+      "loss": 0.0087,
+      "step": 3815
+    },
+    {
+      "epoch": 15.264,
+      "grad_norm": 0.2083113044500351,
+      "learning_rate": 1.1883767535070141e-05,
+      "loss": 0.0086,
+      "step": 3816
+    },
+    {
+      "epoch": 15.268,
+      "grad_norm": 0.20015212893486023,
+      "learning_rate": 1.1873747494989979e-05,
+      "loss": 0.0087,
+      "step": 3817
+    },
+    {
+      "epoch": 15.272,
+      "grad_norm": 0.1395539939403534,
+      "learning_rate": 1.186372745490982e-05,
+      "loss": 0.0073,
+      "step": 3818
+    },
+    {
+      "epoch": 15.276,
+      "grad_norm": 0.1408277451992035,
+      "learning_rate": 1.185370741482966e-05,
+      "loss": 0.0081,
+      "step": 3819
+    },
+    {
+      "epoch": 15.28,
+      "grad_norm": 0.20688875019550323,
+      "learning_rate": 1.18436873747495e-05,
+      "loss": 0.0082,
+      "step": 3820
+    },
+    {
+      "epoch": 15.284,
+      "grad_norm": 0.21537630259990692,
+      "learning_rate": 1.183366733466934e-05,
+      "loss": 0.0085,
+      "step": 3821
+    },
+    {
+      "epoch": 15.288,
+      "grad_norm": 0.22216440737247467,
+      "learning_rate": 1.1823647294589179e-05,
+      "loss": 0.0074,
+      "step": 3822
+    },
+    {
+      "epoch": 15.292,
+      "grad_norm": 0.21135936677455902,
+      "learning_rate": 1.1813627254509018e-05,
+      "loss": 0.0091,
+      "step": 3823
+    },
+    {
+      "epoch": 15.296,
+      "grad_norm": 0.23614716529846191,
+      "learning_rate": 1.1803607214428858e-05,
+      "loss": 0.0082,
+      "step": 3824
+    },
+    {
+      "epoch": 15.3,
+      "grad_norm": 0.16992735862731934,
+      "learning_rate": 1.1793587174348698e-05,
+      "loss": 0.0074,
+      "step": 3825
+    },
+    {
+      "epoch": 15.304,
+      "grad_norm": 0.21491000056266785,
+      "learning_rate": 1.1783567134268537e-05,
+      "loss": 0.0073,
+      "step": 3826
+    },
+    {
+      "epoch": 15.308,
+      "grad_norm": 0.16719304025173187,
+      "learning_rate": 1.1773547094188378e-05,
+      "loss": 0.0069,
+      "step": 3827
+    },
+    {
+      "epoch": 15.312,
+      "grad_norm": 0.2545108497142792,
+      "learning_rate": 1.1763527054108216e-05,
+      "loss": 0.0094,
+      "step": 3828
+    },
+    {
+      "epoch": 15.316,
+      "grad_norm": 0.16995342075824738,
+      "learning_rate": 1.1753507014028056e-05,
+      "loss": 0.0072,
+      "step": 3829
+    },
+    {
+      "epoch": 15.32,
+      "grad_norm": 0.16017405688762665,
+      "learning_rate": 1.1743486973947897e-05,
+      "loss": 0.007,
+      "step": 3830
+    },
+    {
+      "epoch": 15.324,
+      "grad_norm": 0.2081645429134369,
+      "learning_rate": 1.1733466933867735e-05,
+      "loss": 0.009,
+      "step": 3831
+    },
+    {
+      "epoch": 15.328,
+      "grad_norm": 0.1570480614900589,
+      "learning_rate": 1.1723446893787576e-05,
+      "loss": 0.0075,
+      "step": 3832
+    },
+    {
+      "epoch": 15.332,
+      "grad_norm": 0.23571652173995972,
+      "learning_rate": 1.1713426853707416e-05,
+      "loss": 0.0095,
+      "step": 3833
+    },
+    {
+      "epoch": 15.336,
+      "grad_norm": 0.2799575626850128,
+      "learning_rate": 1.1703406813627254e-05,
+      "loss": 0.0107,
+      "step": 3834
+    },
+    {
+      "epoch": 15.34,
+      "grad_norm": 0.22484329342842102,
+      "learning_rate": 1.1693386773547095e-05,
+      "loss": 0.0087,
+      "step": 3835
+    },
+    {
+      "epoch": 15.344,
+      "grad_norm": 0.2471844106912613,
+      "learning_rate": 1.1683366733466935e-05,
+      "loss": 0.009,
+      "step": 3836
+    },
+    {
+      "epoch": 15.348,
+      "grad_norm": 0.21544387936592102,
+      "learning_rate": 1.1673346693386774e-05,
+      "loss": 0.0083,
+      "step": 3837
+    },
+    {
+      "epoch": 15.352,
+      "grad_norm": 0.21488280594348907,
+      "learning_rate": 1.1663326653306614e-05,
+      "loss": 0.0088,
+      "step": 3838
+    },
+    {
+      "epoch": 15.356,
+      "grad_norm": 0.12035142630338669,
+      "learning_rate": 1.1653306613226453e-05,
+      "loss": 0.0068,
+      "step": 3839
+    },
+    {
+      "epoch": 15.36,
+      "grad_norm": 0.20783907175064087,
+      "learning_rate": 1.1643286573146293e-05,
+      "loss": 0.0079,
+      "step": 3840
+    },
+    {
+      "epoch": 15.364,
+      "grad_norm": 0.15776130557060242,
+      "learning_rate": 1.1633266533066133e-05,
+      "loss": 0.0078,
+      "step": 3841
+    },
+    {
+      "epoch": 15.368,
+      "grad_norm": 0.1693926751613617,
+      "learning_rate": 1.1623246492985972e-05,
+      "loss": 0.0085,
+      "step": 3842
+    },
+    {
+      "epoch": 15.372,
+      "grad_norm": 0.20709924399852753,
+      "learning_rate": 1.1613226452905812e-05,
+      "loss": 0.0088,
+      "step": 3843
+    },
+    {
+      "epoch": 15.376,
+      "grad_norm": 0.33956825733184814,
+      "learning_rate": 1.1603206412825651e-05,
+      "loss": 0.0091,
+      "step": 3844
+    },
+    {
+      "epoch": 15.38,
+      "grad_norm": 0.20944558084011078,
+      "learning_rate": 1.1593186372745491e-05,
+      "loss": 0.0087,
+      "step": 3845
+    },
+    {
+      "epoch": 15.384,
+      "grad_norm": 0.16297529637813568,
+      "learning_rate": 1.158316633266533e-05,
+      "loss": 0.0078,
+      "step": 3846
+    },
+    {
+      "epoch": 15.388,
+      "grad_norm": 0.2716394364833832,
+      "learning_rate": 1.1573146292585172e-05,
+      "loss": 0.0089,
+      "step": 3847
+    },
+    {
+      "epoch": 15.392,
+      "grad_norm": 0.15041834115982056,
+      "learning_rate": 1.156312625250501e-05,
+      "loss": 0.0076,
+      "step": 3848
+    },
+    {
+      "epoch": 15.396,
+      "grad_norm": 0.16877737641334534,
+      "learning_rate": 1.155310621242485e-05,
+      "loss": 0.0074,
+      "step": 3849
+    },
+    {
+      "epoch": 15.4,
+      "grad_norm": 0.2795097231864929,
+      "learning_rate": 1.154308617234469e-05,
+      "loss": 0.0096,
+      "step": 3850
+    },
+    {
+      "epoch": 15.404,
+      "grad_norm": 0.13509052991867065,
+      "learning_rate": 1.1533066132264529e-05,
+      "loss": 0.0078,
+      "step": 3851
+    },
+    {
+      "epoch": 15.408,
+      "grad_norm": 0.2292792648077011,
+      "learning_rate": 1.152304609218437e-05,
+      "loss": 0.0084,
+      "step": 3852
+    },
+    {
+      "epoch": 15.412,
+      "grad_norm": 0.19881927967071533,
+      "learning_rate": 1.151302605210421e-05,
+      "loss": 0.0076,
+      "step": 3853
+    },
+    {
+      "epoch": 15.416,
+      "grad_norm": 0.2702074646949768,
+      "learning_rate": 1.1503006012024049e-05,
+      "loss": 0.0095,
+      "step": 3854
+    },
+    {
+      "epoch": 15.42,
+      "grad_norm": 0.2394787222146988,
+      "learning_rate": 1.1492985971943889e-05,
+      "loss": 0.0099,
+      "step": 3855
+    },
+    {
+      "epoch": 15.424,
+      "grad_norm": 0.13445042073726654,
+      "learning_rate": 1.1482965931863728e-05,
+      "loss": 0.0073,
+      "step": 3856
+    },
+    {
+      "epoch": 15.428,
+      "grad_norm": 0.2362317591905594,
+      "learning_rate": 1.1472945891783568e-05,
+      "loss": 0.0074,
+      "step": 3857
+    },
+    {
+      "epoch": 15.432,
+      "grad_norm": 0.14894196391105652,
+      "learning_rate": 1.1462925851703407e-05,
+      "loss": 0.0086,
+      "step": 3858
+    },
+    {
+      "epoch": 15.436,
+      "grad_norm": 0.17250634729862213,
+      "learning_rate": 1.1452905811623247e-05,
+      "loss": 0.009,
+      "step": 3859
+    },
+    {
+      "epoch": 15.44,
+      "grad_norm": 0.197704017162323,
+      "learning_rate": 1.1442885771543087e-05,
+      "loss": 0.0094,
+      "step": 3860
+    },
+    {
+      "epoch": 15.444,
+      "grad_norm": 0.23201380670070648,
+      "learning_rate": 1.1432865731462926e-05,
+      "loss": 0.009,
+      "step": 3861
+    },
+    {
+      "epoch": 15.448,
+      "grad_norm": 0.26006773114204407,
+      "learning_rate": 1.1422845691382766e-05,
+      "loss": 0.009,
+      "step": 3862
+    },
+    {
+      "epoch": 15.452,
+      "grad_norm": 0.21038317680358887,
+      "learning_rate": 1.1412825651302605e-05,
+      "loss": 0.0085,
+      "step": 3863
+    },
+    {
+      "epoch": 15.456,
+      "grad_norm": 0.2623082399368286,
+      "learning_rate": 1.1402805611222447e-05,
+      "loss": 0.0097,
+      "step": 3864
+    },
+    {
+      "epoch": 15.46,
+      "grad_norm": 0.17351366579532623,
+      "learning_rate": 1.1392785571142284e-05,
+      "loss": 0.0081,
+      "step": 3865
+    },
+    {
+      "epoch": 15.464,
+      "grad_norm": 0.1778712272644043,
+      "learning_rate": 1.1382765531062124e-05,
+      "loss": 0.0085,
+      "step": 3866
+    },
+    {
+      "epoch": 15.468,
+      "grad_norm": 0.16661031544208527,
+      "learning_rate": 1.1372745490981965e-05,
+      "loss": 0.008,
+      "step": 3867
+    },
+    {
+      "epoch": 15.472,
+      "grad_norm": 0.1771150380373001,
+      "learning_rate": 1.1362725450901803e-05,
+      "loss": 0.0084,
+      "step": 3868
+    },
+    {
+      "epoch": 15.475999999999999,
+      "grad_norm": 0.2267809361219406,
+      "learning_rate": 1.1352705410821645e-05,
+      "loss": 0.0086,
+      "step": 3869
+    },
+    {
+      "epoch": 15.48,
+      "grad_norm": 0.26965567469596863,
+      "learning_rate": 1.1342685370741484e-05,
+      "loss": 0.0091,
+      "step": 3870
+    },
+    {
+      "epoch": 15.484,
+      "grad_norm": 0.1582583636045456,
+      "learning_rate": 1.1332665330661322e-05,
+      "loss": 0.0056,
+      "step": 3871
+    },
+    {
+      "epoch": 15.488,
+      "grad_norm": 0.24081505835056305,
+      "learning_rate": 1.1322645290581163e-05,
+      "loss": 0.0089,
+      "step": 3872
+    },
+    {
+      "epoch": 15.492,
+      "grad_norm": 0.12798239290714264,
+      "learning_rate": 1.1312625250501003e-05,
+      "loss": 0.0049,
+      "step": 3873
+    },
+    {
+      "epoch": 15.496,
+      "grad_norm": 0.14275750517845154,
+      "learning_rate": 1.1302605210420842e-05,
+      "loss": 0.0083,
+      "step": 3874
+    },
+    {
+      "epoch": 15.5,
+      "grad_norm": 0.22652091085910797,
+      "learning_rate": 1.1292585170340682e-05,
+      "loss": 0.0089,
+      "step": 3875
+    },
+    {
+      "epoch": 15.504,
+      "grad_norm": 0.18880586326122284,
+      "learning_rate": 1.1282565130260522e-05,
+      "loss": 0.0083,
+      "step": 3876
+    },
+    {
+      "epoch": 15.508,
+      "grad_norm": 0.23349466919898987,
+      "learning_rate": 1.1272545090180361e-05,
+      "loss": 0.0088,
+      "step": 3877
+    },
+    {
+      "epoch": 15.512,
+      "grad_norm": 0.2626950740814209,
+      "learning_rate": 1.12625250501002e-05,
+      "loss": 0.0096,
+      "step": 3878
+    },
+    {
+      "epoch": 15.516,
+      "grad_norm": 0.1874002069234848,
+      "learning_rate": 1.125250501002004e-05,
+      "loss": 0.0085,
+      "step": 3879
+    },
+    {
+      "epoch": 15.52,
+      "grad_norm": 0.2023652046918869,
+      "learning_rate": 1.124248496993988e-05,
+      "loss": 0.0085,
+      "step": 3880
+    },
+    {
+      "epoch": 15.524000000000001,
+      "grad_norm": 0.3214608132839203,
+      "learning_rate": 1.1232464929859721e-05,
+      "loss": 0.0097,
+      "step": 3881
+    },
+    {
+      "epoch": 15.528,
+      "grad_norm": 0.13317032158374786,
+      "learning_rate": 1.122244488977956e-05,
+      "loss": 0.0071,
+      "step": 3882
+    },
+    {
+      "epoch": 15.532,
+      "grad_norm": 0.16016507148742676,
+      "learning_rate": 1.1212424849699399e-05,
+      "loss": 0.0081,
+      "step": 3883
+    },
+    {
+      "epoch": 15.536,
+      "grad_norm": 0.2174006998538971,
+      "learning_rate": 1.120240480961924e-05,
+      "loss": 0.008,
+      "step": 3884
+    },
+    {
+      "epoch": 15.54,
+      "grad_norm": 0.28932812809944153,
+      "learning_rate": 1.1192384769539078e-05,
+      "loss": 0.0095,
+      "step": 3885
+    },
+    {
+      "epoch": 15.544,
+      "grad_norm": 0.20081889629364014,
+      "learning_rate": 1.118236472945892e-05,
+      "loss": 0.0083,
+      "step": 3886
+    },
+    {
+      "epoch": 15.548,
+      "grad_norm": 0.2505418658256531,
+      "learning_rate": 1.1172344689378759e-05,
+      "loss": 0.0095,
+      "step": 3887
+    },
+    {
+      "epoch": 15.552,
+      "grad_norm": 0.19087974727153778,
+      "learning_rate": 1.1162324649298597e-05,
+      "loss": 0.0074,
+      "step": 3888
+    },
+    {
+      "epoch": 15.556000000000001,
+      "grad_norm": 0.2052469551563263,
+      "learning_rate": 1.1152304609218438e-05,
+      "loss": 0.0082,
+      "step": 3889
+    },
+    {
+      "epoch": 15.56,
+      "grad_norm": 0.24596913158893585,
+      "learning_rate": 1.1142284569138278e-05,
+      "loss": 0.0084,
+      "step": 3890
+    },
+    {
+      "epoch": 15.564,
+      "grad_norm": 0.46544262766838074,
+      "learning_rate": 1.1132264529058117e-05,
+      "loss": 0.0084,
+      "step": 3891
+    },
+    {
+      "epoch": 15.568,
+      "grad_norm": 0.20327705144882202,
+      "learning_rate": 1.1122244488977957e-05,
+      "loss": 0.0079,
+      "step": 3892
+    },
+    {
+      "epoch": 15.572,
+      "grad_norm": 0.21792593598365784,
+      "learning_rate": 1.1112224448897796e-05,
+      "loss": 0.0094,
+      "step": 3893
+    },
+    {
+      "epoch": 15.576,
+      "grad_norm": 0.1825850009918213,
+      "learning_rate": 1.1102204408817636e-05,
+      "loss": 0.0079,
+      "step": 3894
+    },
+    {
+      "epoch": 15.58,
+      "grad_norm": 0.18869619071483612,
+      "learning_rate": 1.1092184368737475e-05,
+      "loss": 0.0078,
+      "step": 3895
+    },
+    {
+      "epoch": 15.584,
+      "grad_norm": 0.26426199078559875,
+      "learning_rate": 1.1082164328657315e-05,
+      "loss": 0.0112,
+      "step": 3896
+    },
+    {
+      "epoch": 15.588,
+      "grad_norm": 0.20740513503551483,
+      "learning_rate": 1.1072144288577155e-05,
+      "loss": 0.0094,
+      "step": 3897
+    },
+    {
+      "epoch": 15.592,
+      "grad_norm": 0.16762283444404602,
+      "learning_rate": 1.1062124248496994e-05,
+      "loss": 0.0079,
+      "step": 3898
+    },
+    {
+      "epoch": 15.596,
+      "grad_norm": 0.28626206517219543,
+      "learning_rate": 1.1052104208416834e-05,
+      "loss": 0.0091,
+      "step": 3899
+    },
+    {
+      "epoch": 15.6,
+      "grad_norm": 0.1966615617275238,
+      "learning_rate": 1.1042084168336673e-05,
+      "loss": 0.0076,
+      "step": 3900
+    },
+    {
+      "epoch": 15.604,
+      "grad_norm": 0.2372523993253708,
+      "learning_rate": 1.1032064128256515e-05,
+      "loss": 0.0089,
+      "step": 3901
+    },
+    {
+      "epoch": 15.608,
+      "grad_norm": 0.11428368836641312,
+      "learning_rate": 1.1022044088176353e-05,
+      "loss": 0.0048,
+      "step": 3902
+    },
+    {
+      "epoch": 15.612,
+      "grad_norm": 0.20153571665287018,
+      "learning_rate": 1.1012024048096192e-05,
+      "loss": 0.0087,
+      "step": 3903
+    },
+    {
+      "epoch": 15.616,
+      "grad_norm": 0.2549794614315033,
+      "learning_rate": 1.1002004008016033e-05,
+      "loss": 0.0087,
+      "step": 3904
+    },
+    {
+      "epoch": 15.62,
+      "grad_norm": 0.11948023736476898,
+      "learning_rate": 1.0991983967935871e-05,
+      "loss": 0.0052,
+      "step": 3905
+    },
+    {
+      "epoch": 15.624,
+      "grad_norm": 0.1888275444507599,
+      "learning_rate": 1.0981963927855713e-05,
+      "loss": 0.0085,
+      "step": 3906
+    },
+    {
+      "epoch": 15.628,
+      "grad_norm": 0.38151490688323975,
+      "learning_rate": 1.0971943887775552e-05,
+      "loss": 0.009,
+      "step": 3907
+    },
+    {
+      "epoch": 15.632,
+      "grad_norm": 0.12330213189125061,
+      "learning_rate": 1.0961923847695392e-05,
+      "loss": 0.0047,
+      "step": 3908
+    },
+    {
+      "epoch": 15.636,
+      "grad_norm": 0.20336763560771942,
+      "learning_rate": 1.0951903807615231e-05,
+      "loss": 0.0083,
+      "step": 3909
+    },
+    {
+      "epoch": 15.64,
+      "grad_norm": 0.26230552792549133,
+      "learning_rate": 1.0941883767535071e-05,
+      "loss": 0.0104,
+      "step": 3910
+    },
+    {
+      "epoch": 15.644,
+      "grad_norm": 0.2377796471118927,
+      "learning_rate": 1.093186372745491e-05,
+      "loss": 0.0086,
+      "step": 3911
+    },
+    {
+      "epoch": 15.648,
+      "grad_norm": 0.17421042919158936,
+      "learning_rate": 1.092184368737475e-05,
+      "loss": 0.0084,
+      "step": 3912
+    },
+    {
+      "epoch": 15.652,
+      "grad_norm": 0.21408884227275848,
+      "learning_rate": 1.091182364729459e-05,
+      "loss": 0.0089,
+      "step": 3913
+    },
+    {
+      "epoch": 15.656,
+      "grad_norm": 0.15429489314556122,
+      "learning_rate": 1.090180360721443e-05,
+      "loss": 0.0085,
+      "step": 3914
+    },
+    {
+      "epoch": 15.66,
+      "grad_norm": 0.15924520790576935,
+      "learning_rate": 1.0891783567134269e-05,
+      "loss": 0.0055,
+      "step": 3915
+    },
+    {
+      "epoch": 15.664,
+      "grad_norm": 0.3604001998901367,
+      "learning_rate": 1.0881763527054109e-05,
+      "loss": 0.0098,
+      "step": 3916
+    },
+    {
+      "epoch": 15.668,
+      "grad_norm": 0.1900898814201355,
+      "learning_rate": 1.0871743486973948e-05,
+      "loss": 0.0084,
+      "step": 3917
+    },
+    {
+      "epoch": 15.672,
+      "grad_norm": 0.2091568261384964,
+      "learning_rate": 1.086172344689379e-05,
+      "loss": 0.0091,
+      "step": 3918
+    },
+    {
+      "epoch": 15.676,
+      "grad_norm": 0.32461073994636536,
+      "learning_rate": 1.0851703406813627e-05,
+      "loss": 0.0093,
+      "step": 3919
+    },
+    {
+      "epoch": 15.68,
+      "grad_norm": 0.34765294194221497,
+      "learning_rate": 1.0841683366733467e-05,
+      "loss": 0.0097,
+      "step": 3920
+    },
+    {
+      "epoch": 15.684,
+      "grad_norm": 0.1439502090215683,
+      "learning_rate": 1.0831663326653308e-05,
+      "loss": 0.0073,
+      "step": 3921
+    },
+    {
+      "epoch": 15.688,
+      "grad_norm": 0.11668162047863007,
+      "learning_rate": 1.0821643286573146e-05,
+      "loss": 0.0078,
+      "step": 3922
+    },
+    {
+      "epoch": 15.692,
+      "grad_norm": 0.16795441508293152,
+      "learning_rate": 1.0811623246492987e-05,
+      "loss": 0.0083,
+      "step": 3923
+    },
+    {
+      "epoch": 15.696,
+      "grad_norm": 0.32226186990737915,
+      "learning_rate": 1.0801603206412827e-05,
+      "loss": 0.0103,
+      "step": 3924
+    },
+    {
+      "epoch": 15.7,
+      "grad_norm": 0.13408587872982025,
+      "learning_rate": 1.0791583166332665e-05,
+      "loss": 0.005,
+      "step": 3925
+    },
+    {
+      "epoch": 15.704,
+      "grad_norm": 0.12357212603092194,
+      "learning_rate": 1.0781563126252506e-05,
+      "loss": 0.0058,
+      "step": 3926
+    },
+    {
+      "epoch": 15.708,
+      "grad_norm": 0.11612733453512192,
+      "learning_rate": 1.0771543086172344e-05,
+      "loss": 0.0045,
+      "step": 3927
+    },
+    {
+      "epoch": 15.712,
+      "grad_norm": 0.23471392691135406,
+      "learning_rate": 1.0761523046092185e-05,
+      "loss": 0.0087,
+      "step": 3928
+    },
+    {
+      "epoch": 15.716,
+      "grad_norm": 0.1405946910381317,
+      "learning_rate": 1.0751503006012025e-05,
+      "loss": 0.008,
+      "step": 3929
+    },
+    {
+      "epoch": 15.72,
+      "grad_norm": 0.168296217918396,
+      "learning_rate": 1.0741482965931863e-05,
+      "loss": 0.0079,
+      "step": 3930
+    },
+    {
+      "epoch": 15.724,
+      "grad_norm": 0.3210127353668213,
+      "learning_rate": 1.0731462925851704e-05,
+      "loss": 0.0106,
+      "step": 3931
+    },
+    {
+      "epoch": 15.728,
+      "grad_norm": 0.17187538743019104,
+      "learning_rate": 1.0721442885771544e-05,
+      "loss": 0.0085,
+      "step": 3932
+    },
+    {
+      "epoch": 15.732,
+      "grad_norm": 0.15461215376853943,
+      "learning_rate": 1.0711422845691383e-05,
+      "loss": 0.009,
+      "step": 3933
+    },
+    {
+      "epoch": 15.736,
+      "grad_norm": 0.15376779437065125,
+      "learning_rate": 1.0701402805611223e-05,
+      "loss": 0.0082,
+      "step": 3934
+    },
+    {
+      "epoch": 15.74,
+      "grad_norm": 0.3120775818824768,
+      "learning_rate": 1.0691382765531064e-05,
+      "loss": 0.0087,
+      "step": 3935
+    },
+    {
+      "epoch": 15.744,
+      "grad_norm": 0.16850018501281738,
+      "learning_rate": 1.0681362725450902e-05,
+      "loss": 0.0085,
+      "step": 3936
+    },
+    {
+      "epoch": 15.748,
+      "grad_norm": 0.21848592162132263,
+      "learning_rate": 1.0671342685370742e-05,
+      "loss": 0.0083,
+      "step": 3937
+    },
+    {
+      "epoch": 15.752,
+      "grad_norm": 0.16288349032402039,
+      "learning_rate": 1.0661322645290583e-05,
+      "loss": 0.0073,
+      "step": 3938
+    },
+    {
+      "epoch": 15.756,
+      "grad_norm": 0.24063876271247864,
+      "learning_rate": 1.065130260521042e-05,
+      "loss": 0.008,
+      "step": 3939
+    },
+    {
+      "epoch": 15.76,
+      "grad_norm": 0.20195181667804718,
+      "learning_rate": 1.0641282565130262e-05,
+      "loss": 0.0079,
+      "step": 3940
+    },
+    {
+      "epoch": 15.764,
+      "grad_norm": 0.16952313482761383,
+      "learning_rate": 1.06312625250501e-05,
+      "loss": 0.0082,
+      "step": 3941
+    },
+    {
+      "epoch": 15.768,
+      "grad_norm": 0.23509082198143005,
+      "learning_rate": 1.062124248496994e-05,
+      "loss": 0.009,
+      "step": 3942
+    },
+    {
+      "epoch": 15.772,
+      "grad_norm": 0.1740461140871048,
+      "learning_rate": 1.061122244488978e-05,
+      "loss": 0.0081,
+      "step": 3943
+    },
+    {
+      "epoch": 15.776,
+      "grad_norm": 0.23396913707256317,
+      "learning_rate": 1.0601202404809619e-05,
+      "loss": 0.0093,
+      "step": 3944
+    },
+    {
+      "epoch": 15.78,
+      "grad_norm": 0.1290149837732315,
+      "learning_rate": 1.059118236472946e-05,
+      "loss": 0.0071,
+      "step": 3945
+    },
+    {
+      "epoch": 15.784,
+      "grad_norm": 0.19442108273506165,
+      "learning_rate": 1.05811623246493e-05,
+      "loss": 0.0086,
+      "step": 3946
+    },
+    {
+      "epoch": 15.788,
+      "grad_norm": 0.24171869456768036,
+      "learning_rate": 1.0571142284569137e-05,
+      "loss": 0.0085,
+      "step": 3947
+    },
+    {
+      "epoch": 15.792,
+      "grad_norm": 0.17903035879135132,
+      "learning_rate": 1.0561122244488979e-05,
+      "loss": 0.0082,
+      "step": 3948
+    },
+    {
+      "epoch": 15.796,
+      "grad_norm": 0.23411180078983307,
+      "learning_rate": 1.0551102204408818e-05,
+      "loss": 0.0091,
+      "step": 3949
+    },
+    {
+      "epoch": 15.8,
+      "grad_norm": 0.6385198831558228,
+      "learning_rate": 1.0541082164328658e-05,
+      "loss": 0.0077,
+      "step": 3950
+    },
+    {
+      "epoch": 15.804,
+      "grad_norm": 0.201161727309227,
+      "learning_rate": 1.0531062124248498e-05,
+      "loss": 0.0102,
+      "step": 3951
+    },
+    {
+      "epoch": 15.808,
+      "grad_norm": 0.17722225189208984,
+      "learning_rate": 1.0521042084168337e-05,
+      "loss": 0.0083,
+      "step": 3952
+    },
+    {
+      "epoch": 15.812,
+      "grad_norm": 0.16090677678585052,
+      "learning_rate": 1.0511022044088177e-05,
+      "loss": 0.0084,
+      "step": 3953
+    },
+    {
+      "epoch": 15.816,
+      "grad_norm": 0.1804235279560089,
+      "learning_rate": 1.0501002004008016e-05,
+      "loss": 0.0088,
+      "step": 3954
+    },
+    {
+      "epoch": 15.82,
+      "grad_norm": 0.18935038149356842,
+      "learning_rate": 1.0490981963927856e-05,
+      "loss": 0.0093,
+      "step": 3955
+    },
+    {
+      "epoch": 15.824,
+      "grad_norm": 0.1825006753206253,
+      "learning_rate": 1.0480961923847695e-05,
+      "loss": 0.0093,
+      "step": 3956
+    },
+    {
+      "epoch": 15.828,
+      "grad_norm": 0.24779827892780304,
+      "learning_rate": 1.0470941883767535e-05,
+      "loss": 0.01,
+      "step": 3957
+    },
+    {
+      "epoch": 15.832,
+      "grad_norm": 0.26754239201545715,
+      "learning_rate": 1.0460921843687375e-05,
+      "loss": 0.0092,
+      "step": 3958
+    },
+    {
+      "epoch": 15.836,
+      "grad_norm": 0.20227602124214172,
+      "learning_rate": 1.0450901803607214e-05,
+      "loss": 0.0086,
+      "step": 3959
+    },
+    {
+      "epoch": 15.84,
+      "grad_norm": 0.28584498167037964,
+      "learning_rate": 1.0440881763527056e-05,
+      "loss": 0.0108,
+      "step": 3960
+    },
+    {
+      "epoch": 15.844,
+      "grad_norm": 0.23494118452072144,
+      "learning_rate": 1.0430861723446893e-05,
+      "loss": 0.0086,
+      "step": 3961
+    },
+    {
+      "epoch": 15.848,
+      "grad_norm": 0.23064729571342468,
+      "learning_rate": 1.0420841683366733e-05,
+      "loss": 0.0093,
+      "step": 3962
+    },
+    {
+      "epoch": 15.852,
+      "grad_norm": 0.3154512643814087,
+      "learning_rate": 1.0410821643286574e-05,
+      "loss": 0.0097,
+      "step": 3963
+    },
+    {
+      "epoch": 15.856,
+      "grad_norm": 0.18013764917850494,
+      "learning_rate": 1.0400801603206412e-05,
+      "loss": 0.0083,
+      "step": 3964
+    },
+    {
+      "epoch": 15.86,
+      "grad_norm": 0.1942710429430008,
+      "learning_rate": 1.0390781563126253e-05,
+      "loss": 0.0089,
+      "step": 3965
+    },
+    {
+      "epoch": 15.864,
+      "grad_norm": 0.24880248308181763,
+      "learning_rate": 1.0380761523046093e-05,
+      "loss": 0.0087,
+      "step": 3966
+    },
+    {
+      "epoch": 15.868,
+      "grad_norm": 0.20355558395385742,
+      "learning_rate": 1.0370741482965933e-05,
+      "loss": 0.008,
+      "step": 3967
+    },
+    {
+      "epoch": 15.872,
+      "grad_norm": 0.2536620795726776,
+      "learning_rate": 1.0360721442885772e-05,
+      "loss": 0.0085,
+      "step": 3968
+    },
+    {
+      "epoch": 15.876,
+      "grad_norm": 0.17376026511192322,
+      "learning_rate": 1.0350701402805612e-05,
+      "loss": 0.008,
+      "step": 3969
+    },
+    {
+      "epoch": 15.88,
+      "grad_norm": 0.2233988642692566,
+      "learning_rate": 1.0340681362725451e-05,
+      "loss": 0.0091,
+      "step": 3970
+    },
+    {
+      "epoch": 15.884,
+      "grad_norm": 0.1942223310470581,
+      "learning_rate": 1.0330661322645291e-05,
+      "loss": 0.0092,
+      "step": 3971
+    },
+    {
+      "epoch": 15.888,
+      "grad_norm": 0.1970679759979248,
+      "learning_rate": 1.032064128256513e-05,
+      "loss": 0.0083,
+      "step": 3972
+    },
+    {
+      "epoch": 15.892,
+      "grad_norm": 0.14123409986495972,
+      "learning_rate": 1.031062124248497e-05,
+      "loss": 0.0077,
+      "step": 3973
+    },
+    {
+      "epoch": 15.896,
+      "grad_norm": 0.1591396927833557,
+      "learning_rate": 1.030060120240481e-05,
+      "loss": 0.008,
+      "step": 3974
+    },
+    {
+      "epoch": 15.9,
+      "grad_norm": 0.1460416465997696,
+      "learning_rate": 1.029058116232465e-05,
+      "loss": 0.0079,
+      "step": 3975
+    },
+    {
+      "epoch": 15.904,
+      "grad_norm": 0.1960773915052414,
+      "learning_rate": 1.0280561122244489e-05,
+      "loss": 0.0085,
+      "step": 3976
+    },
+    {
+      "epoch": 15.908,
+      "grad_norm": 0.16314558684825897,
+      "learning_rate": 1.027054108216433e-05,
+      "loss": 0.0077,
+      "step": 3977
+    },
+    {
+      "epoch": 15.912,
+      "grad_norm": 0.3076308071613312,
+      "learning_rate": 1.0260521042084168e-05,
+      "loss": 0.0098,
+      "step": 3978
+    },
+    {
+      "epoch": 15.916,
+      "grad_norm": 0.8374438285827637,
+      "learning_rate": 1.0250501002004008e-05,
+      "loss": 0.0114,
+      "step": 3979
+    },
+    {
+      "epoch": 15.92,
+      "grad_norm": 0.31852617859840393,
+      "learning_rate": 1.0240480961923849e-05,
+      "loss": 0.0107,
+      "step": 3980
+    },
+    {
+      "epoch": 15.924,
+      "grad_norm": 0.19266337156295776,
+      "learning_rate": 1.0230460921843687e-05,
+      "loss": 0.0084,
+      "step": 3981
+    },
+    {
+      "epoch": 15.928,
+      "grad_norm": 0.20687268674373627,
+      "learning_rate": 1.0220440881763528e-05,
+      "loss": 0.0089,
+      "step": 3982
+    },
+    {
+      "epoch": 15.932,
+      "grad_norm": 0.17380601167678833,
+      "learning_rate": 1.0210420841683368e-05,
+      "loss": 0.0083,
+      "step": 3983
+    },
+    {
+      "epoch": 15.936,
+      "grad_norm": 0.2257053554058075,
+      "learning_rate": 1.0200400801603206e-05,
+      "loss": 0.0084,
+      "step": 3984
+    },
+    {
+      "epoch": 15.94,
+      "grad_norm": 0.23071831464767456,
+      "learning_rate": 1.0190380761523047e-05,
+      "loss": 0.0102,
+      "step": 3985
+    },
+    {
+      "epoch": 15.943999999999999,
+      "grad_norm": 0.21640700101852417,
+      "learning_rate": 1.0180360721442887e-05,
+      "loss": 0.0085,
+      "step": 3986
+    },
+    {
+      "epoch": 15.948,
+      "grad_norm": 0.17279183864593506,
+      "learning_rate": 1.0170340681362726e-05,
+      "loss": 0.0082,
+      "step": 3987
+    },
+    {
+      "epoch": 15.952,
+      "grad_norm": 0.15583649277687073,
+      "learning_rate": 1.0160320641282566e-05,
+      "loss": 0.0082,
+      "step": 3988
+    },
+    {
+      "epoch": 15.956,
+      "grad_norm": 0.30920568108558655,
+      "learning_rate": 1.0150300601202405e-05,
+      "loss": 0.0101,
+      "step": 3989
+    },
+    {
+      "epoch": 15.96,
+      "grad_norm": 0.23646380007266998,
+      "learning_rate": 1.0140280561122245e-05,
+      "loss": 0.0092,
+      "step": 3990
+    },
+    {
+      "epoch": 15.964,
+      "grad_norm": 0.16509555280208588,
+      "learning_rate": 1.0130260521042084e-05,
+      "loss": 0.0079,
+      "step": 3991
+    },
+    {
+      "epoch": 15.968,
+      "grad_norm": 0.23459596931934357,
+      "learning_rate": 1.0120240480961924e-05,
+      "loss": 0.0088,
+      "step": 3992
+    },
+    {
+      "epoch": 15.972,
+      "grad_norm": 0.1768113523721695,
+      "learning_rate": 1.0110220440881764e-05,
+      "loss": 0.0088,
+      "step": 3993
+    },
+    {
+      "epoch": 15.975999999999999,
+      "grad_norm": 0.22669987380504608,
+      "learning_rate": 1.0100200400801605e-05,
+      "loss": 0.0087,
+      "step": 3994
+    },
+    {
+      "epoch": 15.98,
+      "grad_norm": 0.1484929621219635,
+      "learning_rate": 1.0090180360721443e-05,
+      "loss": 0.0078,
+      "step": 3995
+    },
+    {
+      "epoch": 15.984,
+      "grad_norm": 0.16619373857975006,
+      "learning_rate": 1.0080160320641282e-05,
+      "loss": 0.0083,
+      "step": 3996
+    },
+    {
+      "epoch": 15.988,
+      "grad_norm": 0.14768655598163605,
+      "learning_rate": 1.0070140280561124e-05,
+      "loss": 0.007,
+      "step": 3997
+    },
+    {
+      "epoch": 15.992,
+      "grad_norm": 0.169325053691864,
+      "learning_rate": 1.0060120240480962e-05,
+      "loss": 0.0081,
+      "step": 3998
+    },
+    {
+      "epoch": 15.996,
+      "grad_norm": 0.22410063445568085,
+      "learning_rate": 1.0050100200400803e-05,
+      "loss": 0.0086,
+      "step": 3999
+    },
+    {
+      "epoch": 16.0,
+      "grad_norm": 0.25347474217414856,
+      "learning_rate": 1.0040080160320642e-05,
+      "loss": 0.0092,
+      "step": 4000
+    },
+    {
+      "epoch": 16.004,
+      "grad_norm": 0.1537349820137024,
+      "learning_rate": 1.003006012024048e-05,
+      "loss": 0.0079,
+      "step": 4001
+    },
+    {
+      "epoch": 16.008,
+      "grad_norm": 0.16942495107650757,
+      "learning_rate": 1.0020040080160322e-05,
+      "loss": 0.0078,
+      "step": 4002
+    },
+    {
+      "epoch": 16.012,
+      "grad_norm": 0.15320508182048798,
+      "learning_rate": 1.0010020040080161e-05,
+      "loss": 0.0078,
+      "step": 4003
+    },
+    {
+      "epoch": 16.016,
+      "grad_norm": 0.2386036515235901,
+      "learning_rate": 1e-05,
+      "loss": 0.0079,
+      "step": 4004
+    },
+    {
+      "epoch": 16.02,
+      "grad_norm": 0.14939050376415253,
+      "learning_rate": 9.98997995991984e-06,
+      "loss": 0.008,
+      "step": 4005
+    },
+    {
+      "epoch": 16.024,
+      "grad_norm": 0.14771635830402374,
+      "learning_rate": 9.97995991983968e-06,
+      "loss": 0.0074,
+      "step": 4006
+    },
+    {
+      "epoch": 16.028,
+      "grad_norm": 0.12847012281417847,
+      "learning_rate": 9.96993987975952e-06,
+      "loss": 0.0073,
+      "step": 4007
+    },
+    {
+      "epoch": 16.032,
+      "grad_norm": 0.2023051530122757,
+      "learning_rate": 9.95991983967936e-06,
+      "loss": 0.0075,
+      "step": 4008
+    },
+    {
+      "epoch": 16.036,
+      "grad_norm": 0.1175389364361763,
+      "learning_rate": 9.949899799599199e-06,
+      "loss": 0.006,
+      "step": 4009
+    },
+    {
+      "epoch": 16.04,
+      "grad_norm": 0.2609016001224518,
+      "learning_rate": 9.939879759519038e-06,
+      "loss": 0.0071,
+      "step": 4010
+    },
+    {
+      "epoch": 16.044,
+      "grad_norm": 0.1463293582201004,
+      "learning_rate": 9.929859719438878e-06,
+      "loss": 0.0071,
+      "step": 4011
+    },
+    {
+      "epoch": 16.048,
+      "grad_norm": 0.15348109602928162,
+      "learning_rate": 9.919839679358718e-06,
+      "loss": 0.0076,
+      "step": 4012
+    },
+    {
+      "epoch": 16.052,
+      "grad_norm": 0.15261180698871613,
+      "learning_rate": 9.909819639278557e-06,
+      "loss": 0.0072,
+      "step": 4013
+    },
+    {
+      "epoch": 16.056,
+      "grad_norm": 0.21835850179195404,
+      "learning_rate": 9.899799599198398e-06,
+      "loss": 0.007,
+      "step": 4014
+    },
+    {
+      "epoch": 16.06,
+      "grad_norm": 0.1864071786403656,
+      "learning_rate": 9.889779559118236e-06,
+      "loss": 0.0075,
+      "step": 4015
+    },
+    {
+      "epoch": 16.064,
+      "grad_norm": 0.23007088899612427,
+      "learning_rate": 9.879759519038076e-06,
+      "loss": 0.0098,
+      "step": 4016
+    },
+    {
+      "epoch": 16.068,
+      "grad_norm": 0.12705695629119873,
+      "learning_rate": 9.869739478957917e-06,
+      "loss": 0.0073,
+      "step": 4017
+    },
+    {
+      "epoch": 16.072,
+      "grad_norm": 0.16708585619926453,
+      "learning_rate": 9.859719438877755e-06,
+      "loss": 0.0071,
+      "step": 4018
+    },
+    {
+      "epoch": 16.076,
+      "grad_norm": 0.1379920393228531,
+      "learning_rate": 9.849699398797596e-06,
+      "loss": 0.0075,
+      "step": 4019
+    },
+    {
+      "epoch": 16.08,
+      "grad_norm": 0.16059952974319458,
+      "learning_rate": 9.839679358717436e-06,
+      "loss": 0.0077,
+      "step": 4020
+    },
+    {
+      "epoch": 16.084,
+      "grad_norm": 0.14642521739006042,
+      "learning_rate": 9.829659318637276e-06,
+      "loss": 0.0081,
+      "step": 4021
+    },
+    {
+      "epoch": 16.088,
+      "grad_norm": 0.15344499051570892,
+      "learning_rate": 9.819639278557115e-06,
+      "loss": 0.0079,
+      "step": 4022
+    },
+    {
+      "epoch": 16.092,
+      "grad_norm": 0.17185145616531372,
+      "learning_rate": 9.809619238476955e-06,
+      "loss": 0.0071,
+      "step": 4023
+    },
+    {
+      "epoch": 16.096,
+      "grad_norm": 0.14292530715465546,
+      "learning_rate": 9.799599198396794e-06,
+      "loss": 0.0074,
+      "step": 4024
+    },
+    {
+      "epoch": 16.1,
+      "grad_norm": 0.15338358283042908,
+      "learning_rate": 9.789579158316634e-06,
+      "loss": 0.007,
+      "step": 4025
+    },
+    {
+      "epoch": 16.104,
+      "grad_norm": 0.1363472044467926,
+      "learning_rate": 9.779559118236473e-06,
+      "loss": 0.0065,
+      "step": 4026
+    },
+    {
+      "epoch": 16.108,
+      "grad_norm": 0.18110625445842743,
+      "learning_rate": 9.769539078156313e-06,
+      "loss": 0.0075,
+      "step": 4027
+    },
+    {
+      "epoch": 16.112,
+      "grad_norm": 0.24270915985107422,
+      "learning_rate": 9.759519038076153e-06,
+      "loss": 0.0073,
+      "step": 4028
+    },
+    {
+      "epoch": 16.116,
+      "grad_norm": 0.11325805634260178,
+      "learning_rate": 9.749498997995992e-06,
+      "loss": 0.0045,
+      "step": 4029
+    },
+    {
+      "epoch": 16.12,
+      "grad_norm": 0.6353485584259033,
+      "learning_rate": 9.739478957915832e-06,
+      "loss": 0.0083,
+      "step": 4030
+    },
+    {
+      "epoch": 16.124,
+      "grad_norm": 0.20861081779003143,
+      "learning_rate": 9.729458917835673e-06,
+      "loss": 0.008,
+      "step": 4031
+    },
+    {
+      "epoch": 16.128,
+      "grad_norm": 0.1678023636341095,
+      "learning_rate": 9.719438877755511e-06,
+      "loss": 0.0075,
+      "step": 4032
+    },
+    {
+      "epoch": 16.132,
+      "grad_norm": 0.22754140198230743,
+      "learning_rate": 9.70941883767535e-06,
+      "loss": 0.0077,
+      "step": 4033
+    },
+    {
+      "epoch": 16.136,
+      "grad_norm": 0.16630183160305023,
+      "learning_rate": 9.699398797595192e-06,
+      "loss": 0.0082,
+      "step": 4034
+    },
+    {
+      "epoch": 16.14,
+      "grad_norm": 0.1556076854467392,
+      "learning_rate": 9.68937875751503e-06,
+      "loss": 0.0075,
+      "step": 4035
+    },
+    {
+      "epoch": 16.144,
+      "grad_norm": 0.12683729827404022,
+      "learning_rate": 9.679358717434871e-06,
+      "loss": 0.0068,
+      "step": 4036
+    },
+    {
+      "epoch": 16.148,
+      "grad_norm": 0.17109414935112,
+      "learning_rate": 9.66933867735471e-06,
+      "loss": 0.0085,
+      "step": 4037
+    },
+    {
+      "epoch": 16.152,
+      "grad_norm": 0.20201754570007324,
+      "learning_rate": 9.659318637274549e-06,
+      "loss": 0.0071,
+      "step": 4038
+    },
+    {
+      "epoch": 16.156,
+      "grad_norm": 0.30220654606819153,
+      "learning_rate": 9.64929859719439e-06,
+      "loss": 0.0078,
+      "step": 4039
+    },
+    {
+      "epoch": 16.16,
+      "grad_norm": 0.17886076867580414,
+      "learning_rate": 9.63927855711423e-06,
+      "loss": 0.0081,
+      "step": 4040
+    },
+    {
+      "epoch": 16.164,
+      "grad_norm": 0.14879950881004333,
+      "learning_rate": 9.629258517034069e-06,
+      "loss": 0.0073,
+      "step": 4041
+    },
+    {
+      "epoch": 16.168,
+      "grad_norm": 0.21824462711811066,
+      "learning_rate": 9.619238476953909e-06,
+      "loss": 0.0055,
+      "step": 4042
+    },
+    {
+      "epoch": 16.172,
+      "grad_norm": 0.2450965940952301,
+      "learning_rate": 9.609218436873746e-06,
+      "loss": 0.0088,
+      "step": 4043
+    },
+    {
+      "epoch": 16.176,
+      "grad_norm": 0.16798727214336395,
+      "learning_rate": 9.599198396793588e-06,
+      "loss": 0.008,
+      "step": 4044
+    },
+    {
+      "epoch": 16.18,
+      "grad_norm": 0.21162693202495575,
+      "learning_rate": 9.589178356713427e-06,
+      "loss": 0.0072,
+      "step": 4045
+    },
+    {
+      "epoch": 16.184,
+      "grad_norm": 0.1760333925485611,
+      "learning_rate": 9.579158316633267e-06,
+      "loss": 0.0073,
+      "step": 4046
+    },
+    {
+      "epoch": 16.188,
+      "grad_norm": 0.2047117054462433,
+      "learning_rate": 9.569138276553107e-06,
+      "loss": 0.008,
+      "step": 4047
+    },
+    {
+      "epoch": 16.192,
+      "grad_norm": 0.18758143484592438,
+      "learning_rate": 9.559118236472948e-06,
+      "loss": 0.0069,
+      "step": 4048
+    },
+    {
+      "epoch": 16.196,
+      "grad_norm": 0.2109622359275818,
+      "learning_rate": 9.549098196392786e-06,
+      "loss": 0.0079,
+      "step": 4049
+    },
+    {
+      "epoch": 16.2,
+      "grad_norm": 0.16541019082069397,
+      "learning_rate": 9.539078156312625e-06,
+      "loss": 0.0065,
+      "step": 4050
+    },
+    {
+      "epoch": 16.204,
+      "grad_norm": 0.12194262444972992,
+      "learning_rate": 9.529058116232467e-06,
+      "loss": 0.0067,
+      "step": 4051
+    },
+    {
+      "epoch": 16.208,
+      "grad_norm": 0.5753403306007385,
+      "learning_rate": 9.519038076152304e-06,
+      "loss": 0.0117,
+      "step": 4052
+    },
+    {
+      "epoch": 16.212,
+      "grad_norm": 0.2605521082878113,
+      "learning_rate": 9.509018036072146e-06,
+      "loss": 0.0094,
+      "step": 4053
+    },
+    {
+      "epoch": 16.216,
+      "grad_norm": 0.2662070393562317,
+      "learning_rate": 9.498997995991984e-06,
+      "loss": 0.0089,
+      "step": 4054
+    },
+    {
+      "epoch": 16.22,
+      "grad_norm": 0.17253991961479187,
+      "learning_rate": 9.488977955911823e-06,
+      "loss": 0.0064,
+      "step": 4055
+    },
+    {
+      "epoch": 16.224,
+      "grad_norm": 0.1860446333885193,
+      "learning_rate": 9.478957915831665e-06,
+      "loss": 0.0077,
+      "step": 4056
+    },
+    {
+      "epoch": 16.228,
+      "grad_norm": 0.1807558834552765,
+      "learning_rate": 9.468937875751502e-06,
+      "loss": 0.0084,
+      "step": 4057
+    },
+    {
+      "epoch": 16.232,
+      "grad_norm": 0.2518766224384308,
+      "learning_rate": 9.458917835671344e-06,
+      "loss": 0.0085,
+      "step": 4058
+    },
+    {
+      "epoch": 16.236,
+      "grad_norm": 0.18722839653491974,
+      "learning_rate": 9.448897795591183e-06,
+      "loss": 0.0073,
+      "step": 4059
+    },
+    {
+      "epoch": 16.24,
+      "grad_norm": 0.16822689771652222,
+      "learning_rate": 9.438877755511021e-06,
+      "loss": 0.0072,
+      "step": 4060
+    },
+    {
+      "epoch": 16.244,
+      "grad_norm": 0.26538798213005066,
+      "learning_rate": 9.428857715430862e-06,
+      "loss": 0.0082,
+      "step": 4061
+    },
+    {
+      "epoch": 16.248,
+      "grad_norm": 0.24949723482131958,
+      "learning_rate": 9.418837675350702e-06,
+      "loss": 0.009,
+      "step": 4062
+    },
+    {
+      "epoch": 16.252,
+      "grad_norm": 0.1420658975839615,
+      "learning_rate": 9.408817635270542e-06,
+      "loss": 0.0066,
+      "step": 4063
+    },
+    {
+      "epoch": 16.256,
+      "grad_norm": 0.1801360845565796,
+      "learning_rate": 9.398797595190381e-06,
+      "loss": 0.0076,
+      "step": 4064
+    },
+    {
+      "epoch": 16.26,
+      "grad_norm": 0.16072385013103485,
+      "learning_rate": 9.38877755511022e-06,
+      "loss": 0.0074,
+      "step": 4065
+    },
+    {
+      "epoch": 16.264,
+      "grad_norm": 0.1723981499671936,
+      "learning_rate": 9.37875751503006e-06,
+      "loss": 0.0079,
+      "step": 4066
+    },
+    {
+      "epoch": 16.268,
+      "grad_norm": 0.20122207701206207,
+      "learning_rate": 9.3687374749499e-06,
+      "loss": 0.0085,
+      "step": 4067
+    },
+    {
+      "epoch": 16.272,
+      "grad_norm": 0.16807159781455994,
+      "learning_rate": 9.35871743486974e-06,
+      "loss": 0.0081,
+      "step": 4068
+    },
+    {
+      "epoch": 16.276,
+      "grad_norm": 0.21395786106586456,
+      "learning_rate": 9.34869739478958e-06,
+      "loss": 0.0084,
+      "step": 4069
+    },
+    {
+      "epoch": 16.28,
+      "grad_norm": 0.1668098419904709,
+      "learning_rate": 9.338677354709419e-06,
+      "loss": 0.0071,
+      "step": 4070
+    },
+    {
+      "epoch": 16.284,
+      "grad_norm": 0.1805601865053177,
+      "learning_rate": 9.328657314629258e-06,
+      "loss": 0.0071,
+      "step": 4071
+    },
+    {
+      "epoch": 16.288,
+      "grad_norm": 0.2028336524963379,
+      "learning_rate": 9.318637274549098e-06,
+      "loss": 0.0074,
+      "step": 4072
+    },
+    {
+      "epoch": 16.292,
+      "grad_norm": 0.20740242302417755,
+      "learning_rate": 9.30861723446894e-06,
+      "loss": 0.0076,
+      "step": 4073
+    },
+    {
+      "epoch": 16.296,
+      "grad_norm": 0.21578271687030792,
+      "learning_rate": 9.298597194388777e-06,
+      "loss": 0.0082,
+      "step": 4074
+    },
+    {
+      "epoch": 16.3,
+      "grad_norm": 0.16717369854450226,
+      "learning_rate": 9.288577154308618e-06,
+      "loss": 0.0076,
+      "step": 4075
+    },
+    {
+      "epoch": 16.304,
+      "grad_norm": 0.1616104692220688,
+      "learning_rate": 9.278557114228458e-06,
+      "loss": 0.008,
+      "step": 4076
+    },
+    {
+      "epoch": 16.308,
+      "grad_norm": 0.16851823031902313,
+      "learning_rate": 9.268537074148296e-06,
+      "loss": 0.0072,
+      "step": 4077
+    },
+    {
+      "epoch": 16.312,
+      "grad_norm": 0.18806764483451843,
+      "learning_rate": 9.258517034068137e-06,
+      "loss": 0.008,
+      "step": 4078
+    },
+    {
+      "epoch": 16.316,
+      "grad_norm": 0.21248526871204376,
+      "learning_rate": 9.248496993987977e-06,
+      "loss": 0.0076,
+      "step": 4079
+    },
+    {
+      "epoch": 16.32,
+      "grad_norm": 0.2105952352285385,
+      "learning_rate": 9.238476953907816e-06,
+      "loss": 0.009,
+      "step": 4080
+    },
+    {
+      "epoch": 16.324,
+      "grad_norm": 0.1982223242521286,
+      "learning_rate": 9.228456913827656e-06,
+      "loss": 0.0071,
+      "step": 4081
+    },
+    {
+      "epoch": 16.328,
+      "grad_norm": 0.13595516979694366,
+      "learning_rate": 9.218436873747496e-06,
+      "loss": 0.0075,
+      "step": 4082
+    },
+    {
+      "epoch": 16.332,
+      "grad_norm": 0.16603784263134003,
+      "learning_rate": 9.208416833667335e-06,
+      "loss": 0.0079,
+      "step": 4083
+    },
+    {
+      "epoch": 16.336,
+      "grad_norm": 0.12831614911556244,
+      "learning_rate": 9.198396793587175e-06,
+      "loss": 0.0063,
+      "step": 4084
+    },
+    {
+      "epoch": 16.34,
+      "grad_norm": 0.1970067024230957,
+      "learning_rate": 9.188376753507014e-06,
+      "loss": 0.0067,
+      "step": 4085
+    },
+    {
+      "epoch": 16.344,
+      "grad_norm": 0.1089383140206337,
+      "learning_rate": 9.178356713426854e-06,
+      "loss": 0.0045,
+      "step": 4086
+    },
+    {
+      "epoch": 16.348,
+      "grad_norm": 0.2165975272655487,
+      "learning_rate": 9.168336673346693e-06,
+      "loss": 0.008,
+      "step": 4087
+    },
+    {
+      "epoch": 16.352,
+      "grad_norm": 0.1895694136619568,
+      "learning_rate": 9.158316633266533e-06,
+      "loss": 0.007,
+      "step": 4088
+    },
+    {
+      "epoch": 16.356,
+      "grad_norm": 0.2537732720375061,
+      "learning_rate": 9.148296593186373e-06,
+      "loss": 0.0093,
+      "step": 4089
+    },
+    {
+      "epoch": 16.36,
+      "grad_norm": 0.16854915022850037,
+      "learning_rate": 9.138276553106214e-06,
+      "loss": 0.0087,
+      "step": 4090
+    },
+    {
+      "epoch": 16.364,
+      "grad_norm": 0.1767825335264206,
+      "learning_rate": 9.128256513026052e-06,
+      "loss": 0.0075,
+      "step": 4091
+    },
+    {
+      "epoch": 16.368,
+      "grad_norm": 0.19032791256904602,
+      "learning_rate": 9.118236472945891e-06,
+      "loss": 0.0078,
+      "step": 4092
+    },
+    {
+      "epoch": 16.372,
+      "grad_norm": 0.21050649881362915,
+      "learning_rate": 9.108216432865733e-06,
+      "loss": 0.0083,
+      "step": 4093
+    },
+    {
+      "epoch": 16.376,
+      "grad_norm": 0.1695554405450821,
+      "learning_rate": 9.09819639278557e-06,
+      "loss": 0.0079,
+      "step": 4094
+    },
+    {
+      "epoch": 16.38,
+      "grad_norm": 0.15526901185512543,
+      "learning_rate": 9.088176352705412e-06,
+      "loss": 0.0084,
+      "step": 4095
+    },
+    {
+      "epoch": 16.384,
+      "grad_norm": 0.18327704071998596,
+      "learning_rate": 9.078156312625251e-06,
+      "loss": 0.0076,
+      "step": 4096
+    },
+    {
+      "epoch": 16.388,
+      "grad_norm": 0.1327105313539505,
+      "learning_rate": 9.06813627254509e-06,
+      "loss": 0.007,
+      "step": 4097
+    },
+    {
+      "epoch": 16.392,
+      "grad_norm": 0.17368905246257782,
+      "learning_rate": 9.05811623246493e-06,
+      "loss": 0.0078,
+      "step": 4098
+    },
+    {
+      "epoch": 16.396,
+      "grad_norm": 0.2445645034313202,
+      "learning_rate": 9.04809619238477e-06,
+      "loss": 0.0084,
+      "step": 4099
+    },
+    {
+      "epoch": 16.4,
+      "grad_norm": 0.1971755027770996,
+      "learning_rate": 9.03807615230461e-06,
+      "loss": 0.0084,
+      "step": 4100
+    },
+    {
+      "epoch": 16.404,
+      "grad_norm": 0.22356528043746948,
+      "learning_rate": 9.02805611222445e-06,
+      "loss": 0.0085,
+      "step": 4101
+    },
+    {
+      "epoch": 16.408,
+      "grad_norm": 0.2354351133108139,
+      "learning_rate": 9.018036072144289e-06,
+      "loss": 0.0088,
+      "step": 4102
+    },
+    {
+      "epoch": 16.412,
+      "grad_norm": 0.12401222437620163,
+      "learning_rate": 9.008016032064129e-06,
+      "loss": 0.0047,
+      "step": 4103
+    },
+    {
+      "epoch": 16.416,
+      "grad_norm": 0.10349331051111221,
+      "learning_rate": 8.997995991983968e-06,
+      "loss": 0.005,
+      "step": 4104
+    },
+    {
+      "epoch": 16.42,
+      "grad_norm": 0.16208389401435852,
+      "learning_rate": 8.987975951903808e-06,
+      "loss": 0.0084,
+      "step": 4105
+    },
+    {
+      "epoch": 16.424,
+      "grad_norm": 0.243452250957489,
+      "learning_rate": 8.977955911823647e-06,
+      "loss": 0.0094,
+      "step": 4106
+    },
+    {
+      "epoch": 16.428,
+      "grad_norm": 0.142629936337471,
+      "learning_rate": 8.967935871743489e-06,
+      "loss": 0.0075,
+      "step": 4107
+    },
+    {
+      "epoch": 16.432,
+      "grad_norm": 0.15112538635730743,
+      "learning_rate": 8.957915831663327e-06,
+      "loss": 0.0075,
+      "step": 4108
+    },
+    {
+      "epoch": 16.436,
+      "grad_norm": 0.15232206881046295,
+      "learning_rate": 8.947895791583166e-06,
+      "loss": 0.004,
+      "step": 4109
+    },
+    {
+      "epoch": 16.44,
+      "grad_norm": 0.20458529889583588,
+      "learning_rate": 8.937875751503007e-06,
+      "loss": 0.0081,
+      "step": 4110
+    },
+    {
+      "epoch": 16.444,
+      "grad_norm": 0.1483834683895111,
+      "learning_rate": 8.927855711422845e-06,
+      "loss": 0.0086,
+      "step": 4111
+    },
+    {
+      "epoch": 16.448,
+      "grad_norm": 0.13278226554393768,
+      "learning_rate": 8.917835671342687e-06,
+      "loss": 0.0067,
+      "step": 4112
+    },
+    {
+      "epoch": 16.452,
+      "grad_norm": 0.16911619901657104,
+      "learning_rate": 8.907815631262526e-06,
+      "loss": 0.0074,
+      "step": 4113
+    },
+    {
+      "epoch": 16.456,
+      "grad_norm": 0.2102874517440796,
+      "learning_rate": 8.897795591182364e-06,
+      "loss": 0.0089,
+      "step": 4114
+    },
+    {
+      "epoch": 16.46,
+      "grad_norm": 0.2927187383174896,
+      "learning_rate": 8.887775551102205e-06,
+      "loss": 0.0083,
+      "step": 4115
+    },
+    {
+      "epoch": 16.464,
+      "grad_norm": 0.13784025609493256,
+      "learning_rate": 8.877755511022045e-06,
+      "loss": 0.0079,
+      "step": 4116
+    },
+    {
+      "epoch": 16.468,
+      "grad_norm": 0.15802910923957825,
+      "learning_rate": 8.867735470941884e-06,
+      "loss": 0.0077,
+      "step": 4117
+    },
+    {
+      "epoch": 16.472,
+      "grad_norm": 0.15576040744781494,
+      "learning_rate": 8.857715430861724e-06,
+      "loss": 0.007,
+      "step": 4118
+    },
+    {
+      "epoch": 16.476,
+      "grad_norm": 0.1388392150402069,
+      "learning_rate": 8.847695390781564e-06,
+      "loss": 0.0073,
+      "step": 4119
+    },
+    {
+      "epoch": 16.48,
+      "grad_norm": 0.14177972078323364,
+      "learning_rate": 8.837675350701403e-06,
+      "loss": 0.0075,
+      "step": 4120
+    },
+    {
+      "epoch": 16.484,
+      "grad_norm": 0.19317297637462616,
+      "learning_rate": 8.827655310621243e-06,
+      "loss": 0.0072,
+      "step": 4121
+    },
+    {
+      "epoch": 16.488,
+      "grad_norm": 0.15589751303195953,
+      "learning_rate": 8.817635270541082e-06,
+      "loss": 0.0079,
+      "step": 4122
+    },
+    {
+      "epoch": 16.492,
+      "grad_norm": 0.23068901896476746,
+      "learning_rate": 8.807615230460922e-06,
+      "loss": 0.0078,
+      "step": 4123
+    },
+    {
+      "epoch": 16.496,
+      "grad_norm": 0.2934131920337677,
+      "learning_rate": 8.797595190380762e-06,
+      "loss": 0.008,
+      "step": 4124
+    },
+    {
+      "epoch": 16.5,
+      "grad_norm": 0.20778466761112213,
+      "learning_rate": 8.787575150300601e-06,
+      "loss": 0.0087,
+      "step": 4125
+    },
+    {
+      "epoch": 16.504,
+      "grad_norm": 0.1556091457605362,
+      "learning_rate": 8.77755511022044e-06,
+      "loss": 0.0071,
+      "step": 4126
+    },
+    {
+      "epoch": 16.508,
+      "grad_norm": 0.1435934156179428,
+      "learning_rate": 8.767535070140282e-06,
+      "loss": 0.0078,
+      "step": 4127
+    },
+    {
+      "epoch": 16.512,
+      "grad_norm": 0.18272621929645538,
+      "learning_rate": 8.75751503006012e-06,
+      "loss": 0.0082,
+      "step": 4128
+    },
+    {
+      "epoch": 16.516,
+      "grad_norm": 0.18199694156646729,
+      "learning_rate": 8.747494989979961e-06,
+      "loss": 0.0073,
+      "step": 4129
+    },
+    {
+      "epoch": 16.52,
+      "grad_norm": 0.18646559119224548,
+      "learning_rate": 8.7374749498998e-06,
+      "loss": 0.0071,
+      "step": 4130
+    },
+    {
+      "epoch": 16.524,
+      "grad_norm": 0.20238129794597626,
+      "learning_rate": 8.727454909819639e-06,
+      "loss": 0.0087,
+      "step": 4131
+    },
+    {
+      "epoch": 16.528,
+      "grad_norm": 0.17325668036937714,
+      "learning_rate": 8.71743486973948e-06,
+      "loss": 0.0049,
+      "step": 4132
+    },
+    {
+      "epoch": 16.532,
+      "grad_norm": 0.21818231046199799,
+      "learning_rate": 8.70741482965932e-06,
+      "loss": 0.008,
+      "step": 4133
+    },
+    {
+      "epoch": 16.536,
+      "grad_norm": 0.12345243245363235,
+      "learning_rate": 8.69739478957916e-06,
+      "loss": 0.0066,
+      "step": 4134
+    },
+    {
+      "epoch": 16.54,
+      "grad_norm": 0.21801672875881195,
+      "learning_rate": 8.687374749498999e-06,
+      "loss": 0.009,
+      "step": 4135
+    },
+    {
+      "epoch": 16.544,
+      "grad_norm": 0.22668403387069702,
+      "learning_rate": 8.677354709418838e-06,
+      "loss": 0.0087,
+      "step": 4136
+    },
+    {
+      "epoch": 16.548000000000002,
+      "grad_norm": 0.11883547157049179,
+      "learning_rate": 8.667334669338678e-06,
+      "loss": 0.0069,
+      "step": 4137
+    },
+    {
+      "epoch": 16.552,
+      "grad_norm": 0.3186958134174347,
+      "learning_rate": 8.657314629258518e-06,
+      "loss": 0.0092,
+      "step": 4138
+    },
+    {
+      "epoch": 16.556,
+      "grad_norm": 0.1984918713569641,
+      "learning_rate": 8.647294589178357e-06,
+      "loss": 0.0085,
+      "step": 4139
+    },
+    {
+      "epoch": 16.56,
+      "grad_norm": 0.2220780849456787,
+      "learning_rate": 8.637274549098197e-06,
+      "loss": 0.0093,
+      "step": 4140
+    },
+    {
+      "epoch": 16.564,
+      "grad_norm": 0.11941434442996979,
+      "learning_rate": 8.627254509018036e-06,
+      "loss": 0.0067,
+      "step": 4141
+    },
+    {
+      "epoch": 16.568,
+      "grad_norm": 0.1803411990404129,
+      "learning_rate": 8.617234468937876e-06,
+      "loss": 0.0081,
+      "step": 4142
+    },
+    {
+      "epoch": 16.572,
+      "grad_norm": 0.2058972716331482,
+      "learning_rate": 8.607214428857715e-06,
+      "loss": 0.0075,
+      "step": 4143
+    },
+    {
+      "epoch": 16.576,
+      "grad_norm": 0.23917993903160095,
+      "learning_rate": 8.597194388777557e-06,
+      "loss": 0.0091,
+      "step": 4144
+    },
+    {
+      "epoch": 16.58,
+      "grad_norm": 0.1594621241092682,
+      "learning_rate": 8.587174348697395e-06,
+      "loss": 0.0082,
+      "step": 4145
+    },
+    {
+      "epoch": 16.584,
+      "grad_norm": 0.08639764040708542,
+      "learning_rate": 8.577154308617234e-06,
+      "loss": 0.004,
+      "step": 4146
+    },
+    {
+      "epoch": 16.588,
+      "grad_norm": 0.2543061673641205,
+      "learning_rate": 8.567134268537076e-06,
+      "loss": 0.009,
+      "step": 4147
+    },
+    {
+      "epoch": 16.592,
+      "grad_norm": 0.2126331478357315,
+      "learning_rate": 8.557114228456913e-06,
+      "loss": 0.0082,
+      "step": 4148
+    },
+    {
+      "epoch": 16.596,
+      "grad_norm": 0.2556276023387909,
+      "learning_rate": 8.547094188376755e-06,
+      "loss": 0.0088,
+      "step": 4149
+    },
+    {
+      "epoch": 16.6,
+      "grad_norm": 0.14997954666614532,
+      "learning_rate": 8.537074148296594e-06,
+      "loss": 0.008,
+      "step": 4150
+    },
+    {
+      "epoch": 16.604,
+      "grad_norm": 0.18085747957229614,
+      "learning_rate": 8.527054108216432e-06,
+      "loss": 0.008,
+      "step": 4151
+    },
+    {
+      "epoch": 16.608,
+      "grad_norm": 0.22281257808208466,
+      "learning_rate": 8.517034068136273e-06,
+      "loss": 0.0076,
+      "step": 4152
+    },
+    {
+      "epoch": 16.612,
+      "grad_norm": 0.11172794550657272,
+      "learning_rate": 8.507014028056113e-06,
+      "loss": 0.0052,
+      "step": 4153
+    },
+    {
+      "epoch": 16.616,
+      "grad_norm": 0.2109166383743286,
+      "learning_rate": 8.496993987975953e-06,
+      "loss": 0.0081,
+      "step": 4154
+    },
+    {
+      "epoch": 16.62,
+      "grad_norm": 0.19735772907733917,
+      "learning_rate": 8.486973947895792e-06,
+      "loss": 0.0084,
+      "step": 4155
+    },
+    {
+      "epoch": 16.624,
+      "grad_norm": 0.1670539826154709,
+      "learning_rate": 8.47695390781563e-06,
+      "loss": 0.0076,
+      "step": 4156
+    },
+    {
+      "epoch": 16.628,
+      "grad_norm": 0.26279884576797485,
+      "learning_rate": 8.466933867735471e-06,
+      "loss": 0.0094,
+      "step": 4157
+    },
+    {
+      "epoch": 16.632,
+      "grad_norm": 0.24270443618297577,
+      "learning_rate": 8.456913827655311e-06,
+      "loss": 0.0088,
+      "step": 4158
+    },
+    {
+      "epoch": 16.636,
+      "grad_norm": 0.2702789604663849,
+      "learning_rate": 8.44689378757515e-06,
+      "loss": 0.0094,
+      "step": 4159
+    },
+    {
+      "epoch": 16.64,
+      "grad_norm": 0.13629254698753357,
+      "learning_rate": 8.43687374749499e-06,
+      "loss": 0.007,
+      "step": 4160
+    },
+    {
+      "epoch": 16.644,
+      "grad_norm": 0.1675589680671692,
+      "learning_rate": 8.426853707414831e-06,
+      "loss": 0.0093,
+      "step": 4161
+    },
+    {
+      "epoch": 16.648,
+      "grad_norm": 0.10212656110525131,
+      "learning_rate": 8.41683366733467e-06,
+      "loss": 0.0048,
+      "step": 4162
+    },
+    {
+      "epoch": 16.652,
+      "grad_norm": 0.39032217860221863,
+      "learning_rate": 8.406813627254509e-06,
+      "loss": 0.0074,
+      "step": 4163
+    },
+    {
+      "epoch": 16.656,
+      "grad_norm": 0.20752616226673126,
+      "learning_rate": 8.39679358717435e-06,
+      "loss": 0.0095,
+      "step": 4164
+    },
+    {
+      "epoch": 16.66,
+      "grad_norm": 0.15666551887989044,
+      "learning_rate": 8.386773547094188e-06,
+      "loss": 0.0083,
+      "step": 4165
+    },
+    {
+      "epoch": 16.664,
+      "grad_norm": 0.17233043909072876,
+      "learning_rate": 8.37675350701403e-06,
+      "loss": 0.0076,
+      "step": 4166
+    },
+    {
+      "epoch": 16.668,
+      "grad_norm": 0.17786544561386108,
+      "learning_rate": 8.366733466933869e-06,
+      "loss": 0.008,
+      "step": 4167
+    },
+    {
+      "epoch": 16.672,
+      "grad_norm": 0.14327113330364227,
+      "learning_rate": 8.356713426853707e-06,
+      "loss": 0.0077,
+      "step": 4168
+    },
+    {
+      "epoch": 16.676,
+      "grad_norm": 0.16869743168354034,
+      "learning_rate": 8.346693386773548e-06,
+      "loss": 0.0074,
+      "step": 4169
+    },
+    {
+      "epoch": 16.68,
+      "grad_norm": 0.15478970110416412,
+      "learning_rate": 8.336673346693386e-06,
+      "loss": 0.0077,
+      "step": 4170
+    },
+    {
+      "epoch": 16.684,
+      "grad_norm": 0.29708924889564514,
+      "learning_rate": 8.326653306613227e-06,
+      "loss": 0.009,
+      "step": 4171
+    },
+    {
+      "epoch": 16.688,
+      "grad_norm": 0.1700715571641922,
+      "learning_rate": 8.316633266533067e-06,
+      "loss": 0.0086,
+      "step": 4172
+    },
+    {
+      "epoch": 16.692,
+      "grad_norm": 0.16658703982830048,
+      "learning_rate": 8.306613226452905e-06,
+      "loss": 0.0086,
+      "step": 4173
+    },
+    {
+      "epoch": 16.696,
+      "grad_norm": 0.18442490696907043,
+      "learning_rate": 8.296593186372746e-06,
+      "loss": 0.0084,
+      "step": 4174
+    },
+    {
+      "epoch": 16.7,
+      "grad_norm": 0.2534857392311096,
+      "learning_rate": 8.286573146292586e-06,
+      "loss": 0.0094,
+      "step": 4175
+    },
+    {
+      "epoch": 16.704,
+      "grad_norm": 0.193229541182518,
+      "learning_rate": 8.276553106212425e-06,
+      "loss": 0.0084,
+      "step": 4176
+    },
+    {
+      "epoch": 16.708,
+      "grad_norm": 0.14113031327724457,
+      "learning_rate": 8.266533066132265e-06,
+      "loss": 0.0072,
+      "step": 4177
+    },
+    {
+      "epoch": 16.712,
+      "grad_norm": 0.1453218162059784,
+      "learning_rate": 8.256513026052104e-06,
+      "loss": 0.0083,
+      "step": 4178
+    },
+    {
+      "epoch": 16.716,
+      "grad_norm": 0.17128872871398926,
+      "learning_rate": 8.246492985971944e-06,
+      "loss": 0.0084,
+      "step": 4179
+    },
+    {
+      "epoch": 16.72,
+      "grad_norm": 0.17818258702754974,
+      "learning_rate": 8.236472945891784e-06,
+      "loss": 0.0067,
+      "step": 4180
+    },
+    {
+      "epoch": 16.724,
+      "grad_norm": 0.1808307021856308,
+      "learning_rate": 8.226452905811623e-06,
+      "loss": 0.0084,
+      "step": 4181
+    },
+    {
+      "epoch": 16.728,
+      "grad_norm": 0.11610844731330872,
+      "learning_rate": 8.216432865731463e-06,
+      "loss": 0.0048,
+      "step": 4182
+    },
+    {
+      "epoch": 16.732,
+      "grad_norm": 0.25724536180496216,
+      "learning_rate": 8.206412825651302e-06,
+      "loss": 0.0097,
+      "step": 4183
+    },
+    {
+      "epoch": 16.736,
+      "grad_norm": 0.26306334137916565,
+      "learning_rate": 8.196392785571142e-06,
+      "loss": 0.0093,
+      "step": 4184
+    },
+    {
+      "epoch": 16.74,
+      "grad_norm": 0.17018313705921173,
+      "learning_rate": 8.186372745490982e-06,
+      "loss": 0.0073,
+      "step": 4185
+    },
+    {
+      "epoch": 16.744,
+      "grad_norm": 0.20504972338676453,
+      "learning_rate": 8.176352705410823e-06,
+      "loss": 0.0085,
+      "step": 4186
+    },
+    {
+      "epoch": 16.748,
+      "grad_norm": 0.19006863236427307,
+      "learning_rate": 8.16633266533066e-06,
+      "loss": 0.0079,
+      "step": 4187
+    },
+    {
+      "epoch": 16.752,
+      "grad_norm": 0.22111523151397705,
+      "learning_rate": 8.156312625250502e-06,
+      "loss": 0.009,
+      "step": 4188
+    },
+    {
+      "epoch": 16.756,
+      "grad_norm": 0.17327627539634705,
+      "learning_rate": 8.146292585170342e-06,
+      "loss": 0.0079,
+      "step": 4189
+    },
+    {
+      "epoch": 16.76,
+      "grad_norm": 0.20672562718391418,
+      "learning_rate": 8.13627254509018e-06,
+      "loss": 0.0095,
+      "step": 4190
+    },
+    {
+      "epoch": 16.764,
+      "grad_norm": 0.3867163062095642,
+      "learning_rate": 8.12625250501002e-06,
+      "loss": 0.0091,
+      "step": 4191
+    },
+    {
+      "epoch": 16.768,
+      "grad_norm": 0.22634045779705048,
+      "learning_rate": 8.11623246492986e-06,
+      "loss": 0.0076,
+      "step": 4192
+    },
+    {
+      "epoch": 16.772,
+      "grad_norm": 0.27298519015312195,
+      "learning_rate": 8.1062124248497e-06,
+      "loss": 0.0097,
+      "step": 4193
+    },
+    {
+      "epoch": 16.776,
+      "grad_norm": 0.14704295992851257,
+      "learning_rate": 8.09619238476954e-06,
+      "loss": 0.0082,
+      "step": 4194
+    },
+    {
+      "epoch": 16.78,
+      "grad_norm": 0.12649138271808624,
+      "learning_rate": 8.08617234468938e-06,
+      "loss": 0.007,
+      "step": 4195
+    },
+    {
+      "epoch": 16.784,
+      "grad_norm": 0.2581845223903656,
+      "learning_rate": 8.076152304609219e-06,
+      "loss": 0.0093,
+      "step": 4196
+    },
+    {
+      "epoch": 16.788,
+      "grad_norm": 0.19466523826122284,
+      "learning_rate": 8.066132264529058e-06,
+      "loss": 0.008,
+      "step": 4197
+    },
+    {
+      "epoch": 16.792,
+      "grad_norm": 0.2284821718931198,
+      "learning_rate": 8.056112224448898e-06,
+      "loss": 0.009,
+      "step": 4198
+    },
+    {
+      "epoch": 16.796,
+      "grad_norm": 0.18178433179855347,
+      "learning_rate": 8.046092184368738e-06,
+      "loss": 0.0096,
+      "step": 4199
+    },
+    {
+      "epoch": 16.8,
+      "grad_norm": 0.143863707780838,
+      "learning_rate": 8.036072144288577e-06,
+      "loss": 0.005,
+      "step": 4200
+    },
+    {
+      "epoch": 16.804,
+      "grad_norm": 0.14635920524597168,
+      "learning_rate": 8.026052104208417e-06,
+      "loss": 0.0085,
+      "step": 4201
+    },
+    {
+      "epoch": 16.808,
+      "grad_norm": 0.11327727884054184,
+      "learning_rate": 8.016032064128256e-06,
+      "loss": 0.0046,
+      "step": 4202
+    },
+    {
+      "epoch": 16.812,
+      "grad_norm": 0.19041374325752258,
+      "learning_rate": 8.006012024048098e-06,
+      "loss": 0.0089,
+      "step": 4203
+    },
+    {
+      "epoch": 16.816,
+      "grad_norm": 0.13083264231681824,
+      "learning_rate": 7.995991983967935e-06,
+      "loss": 0.0072,
+      "step": 4204
+    },
+    {
+      "epoch": 16.82,
+      "grad_norm": 0.21465055644512177,
+      "learning_rate": 7.985971943887775e-06,
+      "loss": 0.0083,
+      "step": 4205
+    },
+    {
+      "epoch": 16.824,
+      "grad_norm": 0.20523211359977722,
+      "learning_rate": 7.975951903807616e-06,
+      "loss": 0.008,
+      "step": 4206
+    },
+    {
+      "epoch": 16.828,
+      "grad_norm": 0.14747971296310425,
+      "learning_rate": 7.965931863727454e-06,
+      "loss": 0.0076,
+      "step": 4207
+    },
+    {
+      "epoch": 16.832,
+      "grad_norm": 0.15081685781478882,
+      "learning_rate": 7.955911823647296e-06,
+      "loss": 0.0072,
+      "step": 4208
+    },
+    {
+      "epoch": 16.836,
+      "grad_norm": 0.18735671043395996,
+      "learning_rate": 7.945891783567135e-06,
+      "loss": 0.0078,
+      "step": 4209
+    },
+    {
+      "epoch": 16.84,
+      "grad_norm": 0.2500723898410797,
+      "learning_rate": 7.935871743486973e-06,
+      "loss": 0.0099,
+      "step": 4210
+    },
+    {
+      "epoch": 16.844,
+      "grad_norm": 0.44487854838371277,
+      "learning_rate": 7.925851703406814e-06,
+      "loss": 0.0097,
+      "step": 4211
+    },
+    {
+      "epoch": 16.848,
+      "grad_norm": 0.18985402584075928,
+      "learning_rate": 7.915831663326654e-06,
+      "loss": 0.0082,
+      "step": 4212
+    },
+    {
+      "epoch": 16.852,
+      "grad_norm": 0.16304108500480652,
+      "learning_rate": 7.905811623246493e-06,
+      "loss": 0.0079,
+      "step": 4213
+    },
+    {
+      "epoch": 16.856,
+      "grad_norm": 0.14347808063030243,
+      "learning_rate": 7.895791583166333e-06,
+      "loss": 0.008,
+      "step": 4214
+    },
+    {
+      "epoch": 16.86,
+      "grad_norm": 0.18546195328235626,
+      "learning_rate": 7.885771543086173e-06,
+      "loss": 0.0085,
+      "step": 4215
+    },
+    {
+      "epoch": 16.864,
+      "grad_norm": 0.16561359167099,
+      "learning_rate": 7.875751503006012e-06,
+      "loss": 0.0084,
+      "step": 4216
+    },
+    {
+      "epoch": 16.868,
+      "grad_norm": 0.1679631471633911,
+      "learning_rate": 7.865731462925852e-06,
+      "loss": 0.0083,
+      "step": 4217
+    },
+    {
+      "epoch": 16.872,
+      "grad_norm": 0.1606558859348297,
+      "learning_rate": 7.855711422845691e-06,
+      "loss": 0.0077,
+      "step": 4218
+    },
+    {
+      "epoch": 16.876,
+      "grad_norm": 0.20817658305168152,
+      "learning_rate": 7.845691382765531e-06,
+      "loss": 0.0089,
+      "step": 4219
+    },
+    {
+      "epoch": 16.88,
+      "grad_norm": 0.14732685685157776,
+      "learning_rate": 7.835671342685372e-06,
+      "loss": 0.0077,
+      "step": 4220
+    },
+    {
+      "epoch": 16.884,
+      "grad_norm": 0.20664165914058685,
+      "learning_rate": 7.82565130260521e-06,
+      "loss": 0.0081,
+      "step": 4221
+    },
+    {
+      "epoch": 16.888,
+      "grad_norm": 0.20258696377277374,
+      "learning_rate": 7.81563126252505e-06,
+      "loss": 0.0093,
+      "step": 4222
+    },
+    {
+      "epoch": 16.892,
+      "grad_norm": 0.19178465008735657,
+      "learning_rate": 7.805611222444891e-06,
+      "loss": 0.0082,
+      "step": 4223
+    },
+    {
+      "epoch": 16.896,
+      "grad_norm": 0.30760034918785095,
+      "learning_rate": 7.795591182364729e-06,
+      "loss": 0.0096,
+      "step": 4224
+    },
+    {
+      "epoch": 16.9,
+      "grad_norm": 0.17052769660949707,
+      "learning_rate": 7.78557114228457e-06,
+      "loss": 0.0086,
+      "step": 4225
+    },
+    {
+      "epoch": 16.904,
+      "grad_norm": 0.1625903993844986,
+      "learning_rate": 7.77555110220441e-06,
+      "loss": 0.0082,
+      "step": 4226
+    },
+    {
+      "epoch": 16.908,
+      "grad_norm": 0.17967870831489563,
+      "learning_rate": 7.765531062124248e-06,
+      "loss": 0.0087,
+      "step": 4227
+    },
+    {
+      "epoch": 16.912,
+      "grad_norm": 0.2507075071334839,
+      "learning_rate": 7.755511022044089e-06,
+      "loss": 0.0091,
+      "step": 4228
+    },
+    {
+      "epoch": 16.916,
+      "grad_norm": 0.33597537875175476,
+      "learning_rate": 7.745490981963929e-06,
+      "loss": 0.0086,
+      "step": 4229
+    },
+    {
+      "epoch": 16.92,
+      "grad_norm": 0.15713933110237122,
+      "learning_rate": 7.735470941883768e-06,
+      "loss": 0.0075,
+      "step": 4230
+    },
+    {
+      "epoch": 16.924,
+      "grad_norm": 0.23833885788917542,
+      "learning_rate": 7.725450901803608e-06,
+      "loss": 0.0087,
+      "step": 4231
+    },
+    {
+      "epoch": 16.928,
+      "grad_norm": 0.18948806822299957,
+      "learning_rate": 7.715430861723447e-06,
+      "loss": 0.0083,
+      "step": 4232
+    },
+    {
+      "epoch": 16.932,
+      "grad_norm": 0.23685553669929504,
+      "learning_rate": 7.705410821643287e-06,
+      "loss": 0.0086,
+      "step": 4233
+    },
+    {
+      "epoch": 16.936,
+      "grad_norm": 0.22197188436985016,
+      "learning_rate": 7.695390781563127e-06,
+      "loss": 0.0092,
+      "step": 4234
+    },
+    {
+      "epoch": 16.94,
+      "grad_norm": 0.14752846956253052,
+      "learning_rate": 7.685370741482966e-06,
+      "loss": 0.0075,
+      "step": 4235
+    },
+    {
+      "epoch": 16.944,
+      "grad_norm": 0.16525785624980927,
+      "learning_rate": 7.675350701402806e-06,
+      "loss": 0.0082,
+      "step": 4236
+    },
+    {
+      "epoch": 16.948,
+      "grad_norm": 0.18567903339862823,
+      "learning_rate": 7.665330661322645e-06,
+      "loss": 0.0079,
+      "step": 4237
+    },
+    {
+      "epoch": 16.951999999999998,
+      "grad_norm": 0.16079868376255035,
+      "learning_rate": 7.655310621242485e-06,
+      "loss": 0.0079,
+      "step": 4238
+    },
+    {
+      "epoch": 16.956,
+      "grad_norm": 0.17427866160869598,
+      "learning_rate": 7.645290581162324e-06,
+      "loss": 0.0088,
+      "step": 4239
+    },
+    {
+      "epoch": 16.96,
+      "grad_norm": 0.18866820633411407,
+      "learning_rate": 7.635270541082166e-06,
+      "loss": 0.0085,
+      "step": 4240
+    },
+    {
+      "epoch": 16.964,
+      "grad_norm": 0.15932509303092957,
+      "learning_rate": 7.6252505010020045e-06,
+      "loss": 0.0074,
+      "step": 4241
+    },
+    {
+      "epoch": 16.968,
+      "grad_norm": 0.2091926783323288,
+      "learning_rate": 7.615230460921845e-06,
+      "loss": 0.0093,
+      "step": 4242
+    },
+    {
+      "epoch": 16.972,
+      "grad_norm": 0.2609868049621582,
+      "learning_rate": 7.605210420841684e-06,
+      "loss": 0.0103,
+      "step": 4243
+    },
+    {
+      "epoch": 16.976,
+      "grad_norm": 0.1595424860715866,
+      "learning_rate": 7.595190380761523e-06,
+      "loss": 0.0079,
+      "step": 4244
+    },
+    {
+      "epoch": 16.98,
+      "grad_norm": 0.1985984891653061,
+      "learning_rate": 7.585170340681364e-06,
+      "loss": 0.0089,
+      "step": 4245
+    },
+    {
+      "epoch": 16.984,
+      "grad_norm": 0.17337600886821747,
+      "learning_rate": 7.5751503006012024e-06,
+      "loss": 0.0074,
+      "step": 4246
+    },
+    {
+      "epoch": 16.988,
+      "grad_norm": 0.18920986354351044,
+      "learning_rate": 7.565130260521043e-06,
+      "loss": 0.0083,
+      "step": 4247
+    },
+    {
+      "epoch": 16.992,
+      "grad_norm": 0.17387863993644714,
+      "learning_rate": 7.5551102204408825e-06,
+      "loss": 0.0085,
+      "step": 4248
+    },
+    {
+      "epoch": 16.996,
+      "grad_norm": 0.20932914316654205,
+      "learning_rate": 7.545090180360721e-06,
+      "loss": 0.0081,
+      "step": 4249
+    },
+    {
+      "epoch": 17.0,
+      "grad_norm": 0.18405738472938538,
+      "learning_rate": 7.535070140280562e-06,
+      "loss": 0.0078,
+      "step": 4250
+    },
+    {
+      "epoch": 17.004,
+      "grad_norm": 0.12063789367675781,
+      "learning_rate": 7.525050100200401e-06,
+      "loss": 0.0067,
+      "step": 4251
+    },
+    {
+      "epoch": 17.008,
+      "grad_norm": 0.13446062803268433,
+      "learning_rate": 7.515030060120242e-06,
+      "loss": 0.0067,
+      "step": 4252
+    },
+    {
+      "epoch": 17.012,
+      "grad_norm": 0.13746428489685059,
+      "learning_rate": 7.50501002004008e-06,
+      "loss": 0.0073,
+      "step": 4253
+    },
+    {
+      "epoch": 17.016,
+      "grad_norm": 0.17416371405124664,
+      "learning_rate": 7.49498997995992e-06,
+      "loss": 0.0079,
+      "step": 4254
+    },
+    {
+      "epoch": 17.02,
+      "grad_norm": 0.19402289390563965,
+      "learning_rate": 7.48496993987976e-06,
+      "loss": 0.0073,
+      "step": 4255
+    },
+    {
+      "epoch": 17.024,
+      "grad_norm": 0.14378122985363007,
+      "learning_rate": 7.474949899799599e-06,
+      "loss": 0.0065,
+      "step": 4256
+    },
+    {
+      "epoch": 17.028,
+      "grad_norm": 0.14554685354232788,
+      "learning_rate": 7.46492985971944e-06,
+      "loss": 0.0069,
+      "step": 4257
+    },
+    {
+      "epoch": 17.032,
+      "grad_norm": 0.14952172338962555,
+      "learning_rate": 7.454909819639279e-06,
+      "loss": 0.0075,
+      "step": 4258
+    },
+    {
+      "epoch": 17.036,
+      "grad_norm": 0.08101173490285873,
+      "learning_rate": 7.444889779559118e-06,
+      "loss": 0.0038,
+      "step": 4259
+    },
+    {
+      "epoch": 17.04,
+      "grad_norm": 0.13694711029529572,
+      "learning_rate": 7.434869739478958e-06,
+      "loss": 0.0065,
+      "step": 4260
+    },
+    {
+      "epoch": 17.044,
+      "grad_norm": 0.1457206904888153,
+      "learning_rate": 7.424849699398798e-06,
+      "loss": 0.0075,
+      "step": 4261
+    },
+    {
+      "epoch": 17.048,
+      "grad_norm": 0.1786690503358841,
+      "learning_rate": 7.414829659318638e-06,
+      "loss": 0.0067,
+      "step": 4262
+    },
+    {
+      "epoch": 17.052,
+      "grad_norm": 0.16678613424301147,
+      "learning_rate": 7.404809619238477e-06,
+      "loss": 0.007,
+      "step": 4263
+    },
+    {
+      "epoch": 17.056,
+      "grad_norm": 0.12600558996200562,
+      "learning_rate": 7.394789579158317e-06,
+      "loss": 0.0074,
+      "step": 4264
+    },
+    {
+      "epoch": 17.06,
+      "grad_norm": 0.1651485711336136,
+      "learning_rate": 7.384769539078157e-06,
+      "loss": 0.0077,
+      "step": 4265
+    },
+    {
+      "epoch": 17.064,
+      "grad_norm": 0.12365109473466873,
+      "learning_rate": 7.374749498997996e-06,
+      "loss": 0.0065,
+      "step": 4266
+    },
+    {
+      "epoch": 17.068,
+      "grad_norm": 0.16177420318126678,
+      "learning_rate": 7.364729458917836e-06,
+      "loss": 0.007,
+      "step": 4267
+    },
+    {
+      "epoch": 17.072,
+      "grad_norm": 0.1336231380701065,
+      "learning_rate": 7.354709418837676e-06,
+      "loss": 0.0067,
+      "step": 4268
+    },
+    {
+      "epoch": 17.076,
+      "grad_norm": 0.12120088934898376,
+      "learning_rate": 7.344689378757516e-06,
+      "loss": 0.0044,
+      "step": 4269
+    },
+    {
+      "epoch": 17.08,
+      "grad_norm": 0.1796593815088272,
+      "learning_rate": 7.334669338677355e-06,
+      "loss": 0.0076,
+      "step": 4270
+    },
+    {
+      "epoch": 17.084,
+      "grad_norm": 0.158540278673172,
+      "learning_rate": 7.324649298597195e-06,
+      "loss": 0.0069,
+      "step": 4271
+    },
+    {
+      "epoch": 17.088,
+      "grad_norm": 0.19492676854133606,
+      "learning_rate": 7.314629258517035e-06,
+      "loss": 0.0073,
+      "step": 4272
+    },
+    {
+      "epoch": 17.092,
+      "grad_norm": 0.19501705467700958,
+      "learning_rate": 7.304609218436874e-06,
+      "loss": 0.008,
+      "step": 4273
+    },
+    {
+      "epoch": 17.096,
+      "grad_norm": 0.16361992061138153,
+      "learning_rate": 7.294589178356714e-06,
+      "loss": 0.0068,
+      "step": 4274
+    },
+    {
+      "epoch": 17.1,
+      "grad_norm": 0.14692838490009308,
+      "learning_rate": 7.284569138276554e-06,
+      "loss": 0.007,
+      "step": 4275
+    },
+    {
+      "epoch": 17.104,
+      "grad_norm": 0.1860724836587906,
+      "learning_rate": 7.274549098196393e-06,
+      "loss": 0.0079,
+      "step": 4276
+    },
+    {
+      "epoch": 17.108,
+      "grad_norm": 0.2329706847667694,
+      "learning_rate": 7.264529058116233e-06,
+      "loss": 0.0071,
+      "step": 4277
+    },
+    {
+      "epoch": 17.112,
+      "grad_norm": 0.16117851436138153,
+      "learning_rate": 7.254509018036073e-06,
+      "loss": 0.0077,
+      "step": 4278
+    },
+    {
+      "epoch": 17.116,
+      "grad_norm": 0.15216809511184692,
+      "learning_rate": 7.244488977955913e-06,
+      "loss": 0.0071,
+      "step": 4279
+    },
+    {
+      "epoch": 17.12,
+      "grad_norm": 0.19427397847175598,
+      "learning_rate": 7.234468937875752e-06,
+      "loss": 0.0074,
+      "step": 4280
+    },
+    {
+      "epoch": 17.124,
+      "grad_norm": 0.23492863774299622,
+      "learning_rate": 7.2244488977955906e-06,
+      "loss": 0.0078,
+      "step": 4281
+    },
+    {
+      "epoch": 17.128,
+      "grad_norm": 0.1491718888282776,
+      "learning_rate": 7.214428857715432e-06,
+      "loss": 0.0064,
+      "step": 4282
+    },
+    {
+      "epoch": 17.132,
+      "grad_norm": 0.14911611378192902,
+      "learning_rate": 7.204408817635271e-06,
+      "loss": 0.0069,
+      "step": 4283
+    },
+    {
+      "epoch": 17.136,
+      "grad_norm": 0.17563241720199585,
+      "learning_rate": 7.194388777555111e-06,
+      "loss": 0.0076,
+      "step": 4284
+    },
+    {
+      "epoch": 17.14,
+      "grad_norm": 0.1660158634185791,
+      "learning_rate": 7.184368737474951e-06,
+      "loss": 0.007,
+      "step": 4285
+    },
+    {
+      "epoch": 17.144,
+      "grad_norm": 0.1971948891878128,
+      "learning_rate": 7.174348697394789e-06,
+      "loss": 0.0067,
+      "step": 4286
+    },
+    {
+      "epoch": 17.148,
+      "grad_norm": 0.15479083359241486,
+      "learning_rate": 7.16432865731463e-06,
+      "loss": 0.0066,
+      "step": 4287
+    },
+    {
+      "epoch": 17.152,
+      "grad_norm": 0.20346534252166748,
+      "learning_rate": 7.1543086172344685e-06,
+      "loss": 0.0068,
+      "step": 4288
+    },
+    {
+      "epoch": 17.156,
+      "grad_norm": 0.22112035751342773,
+      "learning_rate": 7.14428857715431e-06,
+      "loss": 0.0071,
+      "step": 4289
+    },
+    {
+      "epoch": 17.16,
+      "grad_norm": 0.17985723912715912,
+      "learning_rate": 7.1342685370741486e-06,
+      "loss": 0.0081,
+      "step": 4290
+    },
+    {
+      "epoch": 17.164,
+      "grad_norm": 0.20944520831108093,
+      "learning_rate": 7.124248496993987e-06,
+      "loss": 0.007,
+      "step": 4291
+    },
+    {
+      "epoch": 17.168,
+      "grad_norm": 0.1437358558177948,
+      "learning_rate": 7.114228456913829e-06,
+      "loss": 0.0084,
+      "step": 4292
+    },
+    {
+      "epoch": 17.172,
+      "grad_norm": 0.14544442296028137,
+      "learning_rate": 7.104208416833667e-06,
+      "loss": 0.0047,
+      "step": 4293
+    },
+    {
+      "epoch": 17.176,
+      "grad_norm": 0.19183100759983063,
+      "learning_rate": 7.094188376753508e-06,
+      "loss": 0.0074,
+      "step": 4294
+    },
+    {
+      "epoch": 17.18,
+      "grad_norm": 0.16251499950885773,
+      "learning_rate": 7.0841683366733465e-06,
+      "loss": 0.0068,
+      "step": 4295
+    },
+    {
+      "epoch": 17.184,
+      "grad_norm": 0.16117817163467407,
+      "learning_rate": 7.074148296593188e-06,
+      "loss": 0.0069,
+      "step": 4296
+    },
+    {
+      "epoch": 17.188,
+      "grad_norm": 0.15662899613380432,
+      "learning_rate": 7.0641282565130265e-06,
+      "loss": 0.0076,
+      "step": 4297
+    },
+    {
+      "epoch": 17.192,
+      "grad_norm": 0.21028119325637817,
+      "learning_rate": 7.054108216432865e-06,
+      "loss": 0.0078,
+      "step": 4298
+    },
+    {
+      "epoch": 17.196,
+      "grad_norm": 0.19951249659061432,
+      "learning_rate": 7.044088176352706e-06,
+      "loss": 0.0075,
+      "step": 4299
+    },
+    {
+      "epoch": 17.2,
+      "grad_norm": 0.17216528952121735,
+      "learning_rate": 7.034068136272545e-06,
+      "loss": 0.0069,
+      "step": 4300
+    },
+    {
+      "epoch": 17.204,
+      "grad_norm": 0.14699934422969818,
+      "learning_rate": 7.024048096192386e-06,
+      "loss": 0.0045,
+      "step": 4301
+    },
+    {
+      "epoch": 17.208,
+      "grad_norm": 0.15212714672088623,
+      "learning_rate": 7.0140280561122245e-06,
+      "loss": 0.0079,
+      "step": 4302
+    },
+    {
+      "epoch": 17.212,
+      "grad_norm": 0.1797187179327011,
+      "learning_rate": 7.004008016032064e-06,
+      "loss": 0.0071,
+      "step": 4303
+    },
+    {
+      "epoch": 17.216,
+      "grad_norm": 0.13188792765140533,
+      "learning_rate": 6.9939879759519045e-06,
+      "loss": 0.0068,
+      "step": 4304
+    },
+    {
+      "epoch": 17.22,
+      "grad_norm": 0.1433761864900589,
+      "learning_rate": 6.983967935871743e-06,
+      "loss": 0.0067,
+      "step": 4305
+    },
+    {
+      "epoch": 17.224,
+      "grad_norm": 0.22716593742370605,
+      "learning_rate": 6.973947895791584e-06,
+      "loss": 0.0087,
+      "step": 4306
+    },
+    {
+      "epoch": 17.228,
+      "grad_norm": 0.09545940160751343,
+      "learning_rate": 6.963927855711423e-06,
+      "loss": 0.0042,
+      "step": 4307
+    },
+    {
+      "epoch": 17.232,
+      "grad_norm": 0.15111321210861206,
+      "learning_rate": 6.953907815631262e-06,
+      "loss": 0.0072,
+      "step": 4308
+    },
+    {
+      "epoch": 17.236,
+      "grad_norm": 0.14687678217887878,
+      "learning_rate": 6.9438877755511024e-06,
+      "loss": 0.0069,
+      "step": 4309
+    },
+    {
+      "epoch": 17.24,
+      "grad_norm": 0.148050919175148,
+      "learning_rate": 6.933867735470942e-06,
+      "loss": 0.0068,
+      "step": 4310
+    },
+    {
+      "epoch": 17.244,
+      "grad_norm": 0.12253053486347198,
+      "learning_rate": 6.9238476953907825e-06,
+      "loss": 0.0066,
+      "step": 4311
+    },
+    {
+      "epoch": 17.248,
+      "grad_norm": 0.19216535985469818,
+      "learning_rate": 6.913827655310621e-06,
+      "loss": 0.0082,
+      "step": 4312
+    },
+    {
+      "epoch": 17.252,
+      "grad_norm": 0.22438839077949524,
+      "learning_rate": 6.903807615230461e-06,
+      "loss": 0.0075,
+      "step": 4313
+    },
+    {
+      "epoch": 17.256,
+      "grad_norm": 0.2512076497077942,
+      "learning_rate": 6.893787575150301e-06,
+      "loss": 0.0082,
+      "step": 4314
+    },
+    {
+      "epoch": 17.26,
+      "grad_norm": 0.15971721708774567,
+      "learning_rate": 6.88376753507014e-06,
+      "loss": 0.0068,
+      "step": 4315
+    },
+    {
+      "epoch": 17.264,
+      "grad_norm": 0.14771561324596405,
+      "learning_rate": 6.87374749498998e-06,
+      "loss": 0.0078,
+      "step": 4316
+    },
+    {
+      "epoch": 17.268,
+      "grad_norm": 0.1611126810312271,
+      "learning_rate": 6.86372745490982e-06,
+      "loss": 0.0083,
+      "step": 4317
+    },
+    {
+      "epoch": 17.272,
+      "grad_norm": 0.2246340811252594,
+      "learning_rate": 6.853707414829659e-06,
+      "loss": 0.0079,
+      "step": 4318
+    },
+    {
+      "epoch": 17.276,
+      "grad_norm": 0.13150276243686676,
+      "learning_rate": 6.843687374749499e-06,
+      "loss": 0.0048,
+      "step": 4319
+    },
+    {
+      "epoch": 17.28,
+      "grad_norm": 0.12884242832660675,
+      "learning_rate": 6.833667334669339e-06,
+      "loss": 0.0062,
+      "step": 4320
+    },
+    {
+      "epoch": 17.284,
+      "grad_norm": 0.16102537512779236,
+      "learning_rate": 6.823647294589179e-06,
+      "loss": 0.0063,
+      "step": 4321
+    },
+    {
+      "epoch": 17.288,
+      "grad_norm": 0.21450555324554443,
+      "learning_rate": 6.813627254509018e-06,
+      "loss": 0.0075,
+      "step": 4322
+    },
+    {
+      "epoch": 17.292,
+      "grad_norm": 0.21209336817264557,
+      "learning_rate": 6.803607214428858e-06,
+      "loss": 0.0076,
+      "step": 4323
+    },
+    {
+      "epoch": 17.296,
+      "grad_norm": 0.1917915791273117,
+      "learning_rate": 6.793587174348698e-06,
+      "loss": 0.0074,
+      "step": 4324
+    },
+    {
+      "epoch": 17.3,
+      "grad_norm": 0.14592121541500092,
+      "learning_rate": 6.783567134268537e-06,
+      "loss": 0.0071,
+      "step": 4325
+    },
+    {
+      "epoch": 17.304,
+      "grad_norm": 0.1582149863243103,
+      "learning_rate": 6.773547094188377e-06,
+      "loss": 0.007,
+      "step": 4326
+    },
+    {
+      "epoch": 17.308,
+      "grad_norm": 0.21334248781204224,
+      "learning_rate": 6.763527054108217e-06,
+      "loss": 0.0081,
+      "step": 4327
+    },
+    {
+      "epoch": 17.312,
+      "grad_norm": 0.15731792151927948,
+      "learning_rate": 6.753507014028057e-06,
+      "loss": 0.0068,
+      "step": 4328
+    },
+    {
+      "epoch": 17.316,
+      "grad_norm": 0.15384608507156372,
+      "learning_rate": 6.743486973947896e-06,
+      "loss": 0.0079,
+      "step": 4329
+    },
+    {
+      "epoch": 17.32,
+      "grad_norm": 0.23401755094528198,
+      "learning_rate": 6.7334669338677355e-06,
+      "loss": 0.008,
+      "step": 4330
+    },
+    {
+      "epoch": 17.324,
+      "grad_norm": 0.16333438456058502,
+      "learning_rate": 6.723446893787576e-06,
+      "loss": 0.0076,
+      "step": 4331
+    },
+    {
+      "epoch": 17.328,
+      "grad_norm": 0.25040316581726074,
+      "learning_rate": 6.713426853707415e-06,
+      "loss": 0.0085,
+      "step": 4332
+    },
+    {
+      "epoch": 17.332,
+      "grad_norm": 0.22485579550266266,
+      "learning_rate": 6.703406813627255e-06,
+      "loss": 0.0087,
+      "step": 4333
+    },
+    {
+      "epoch": 17.336,
+      "grad_norm": 0.16370800137519836,
+      "learning_rate": 6.693386773547095e-06,
+      "loss": 0.0076,
+      "step": 4334
+    },
+    {
+      "epoch": 17.34,
+      "grad_norm": 0.18383802473545074,
+      "learning_rate": 6.6833667334669334e-06,
+      "loss": 0.0074,
+      "step": 4335
+    },
+    {
+      "epoch": 17.344,
+      "grad_norm": 0.13991597294807434,
+      "learning_rate": 6.673346693386774e-06,
+      "loss": 0.0071,
+      "step": 4336
+    },
+    {
+      "epoch": 17.348,
+      "grad_norm": 0.24963268637657166,
+      "learning_rate": 6.6633266533066135e-06,
+      "loss": 0.0084,
+      "step": 4337
+    },
+    {
+      "epoch": 17.352,
+      "grad_norm": 0.22435139119625092,
+      "learning_rate": 6.653306613226454e-06,
+      "loss": 0.0079,
+      "step": 4338
+    },
+    {
+      "epoch": 17.356,
+      "grad_norm": 0.2332351952791214,
+      "learning_rate": 6.643286573146293e-06,
+      "loss": 0.0082,
+      "step": 4339
+    },
+    {
+      "epoch": 17.36,
+      "grad_norm": 0.1963132917881012,
+      "learning_rate": 6.633266533066132e-06,
+      "loss": 0.0069,
+      "step": 4340
+    },
+    {
+      "epoch": 17.364,
+      "grad_norm": 0.21712207794189453,
+      "learning_rate": 6.623246492985973e-06,
+      "loss": 0.0086,
+      "step": 4341
+    },
+    {
+      "epoch": 17.368,
+      "grad_norm": 0.12130995839834213,
+      "learning_rate": 6.613226452905811e-06,
+      "loss": 0.0038,
+      "step": 4342
+    },
+    {
+      "epoch": 17.372,
+      "grad_norm": 0.2439410537481308,
+      "learning_rate": 6.603206412825652e-06,
+      "loss": 0.0081,
+      "step": 4343
+    },
+    {
+      "epoch": 17.376,
+      "grad_norm": 0.17727676033973694,
+      "learning_rate": 6.5931863727454914e-06,
+      "loss": 0.0071,
+      "step": 4344
+    },
+    {
+      "epoch": 17.38,
+      "grad_norm": 0.15301989018917084,
+      "learning_rate": 6.58316633266533e-06,
+      "loss": 0.0065,
+      "step": 4345
+    },
+    {
+      "epoch": 17.384,
+      "grad_norm": 0.17886261641979218,
+      "learning_rate": 6.573146292585171e-06,
+      "loss": 0.0079,
+      "step": 4346
+    },
+    {
+      "epoch": 17.388,
+      "grad_norm": 0.21048425137996674,
+      "learning_rate": 6.56312625250501e-06,
+      "loss": 0.0077,
+      "step": 4347
+    },
+    {
+      "epoch": 17.392,
+      "grad_norm": 0.23739394545555115,
+      "learning_rate": 6.553106212424851e-06,
+      "loss": 0.0077,
+      "step": 4348
+    },
+    {
+      "epoch": 17.396,
+      "grad_norm": 0.22822335362434387,
+      "learning_rate": 6.543086172344689e-06,
+      "loss": 0.0078,
+      "step": 4349
+    },
+    {
+      "epoch": 17.4,
+      "grad_norm": 0.24672290682792664,
+      "learning_rate": 6.533066132264529e-06,
+      "loss": 0.0077,
+      "step": 4350
+    },
+    {
+      "epoch": 17.404,
+      "grad_norm": 0.22589755058288574,
+      "learning_rate": 6.523046092184369e-06,
+      "loss": 0.008,
+      "step": 4351
+    },
+    {
+      "epoch": 17.408,
+      "grad_norm": 0.1355731040239334,
+      "learning_rate": 6.513026052104208e-06,
+      "loss": 0.0076,
+      "step": 4352
+    },
+    {
+      "epoch": 17.412,
+      "grad_norm": 0.15726511180400848,
+      "learning_rate": 6.5030060120240486e-06,
+      "loss": 0.007,
+      "step": 4353
+    },
+    {
+      "epoch": 17.416,
+      "grad_norm": 0.14194290339946747,
+      "learning_rate": 6.492985971943888e-06,
+      "loss": 0.0069,
+      "step": 4354
+    },
+    {
+      "epoch": 17.42,
+      "grad_norm": 0.15629169344902039,
+      "learning_rate": 6.482965931863729e-06,
+      "loss": 0.0077,
+      "step": 4355
+    },
+    {
+      "epoch": 17.424,
+      "grad_norm": 0.20658808946609497,
+      "learning_rate": 6.472945891783567e-06,
+      "loss": 0.0071,
+      "step": 4356
+    },
+    {
+      "epoch": 17.428,
+      "grad_norm": 0.13976025581359863,
+      "learning_rate": 6.462925851703407e-06,
+      "loss": 0.0074,
+      "step": 4357
+    },
+    {
+      "epoch": 17.432,
+      "grad_norm": 0.14922749996185303,
+      "learning_rate": 6.452905811623247e-06,
+      "loss": 0.0075,
+      "step": 4358
+    },
+    {
+      "epoch": 17.436,
+      "grad_norm": 0.31609266996383667,
+      "learning_rate": 6.442885771543086e-06,
+      "loss": 0.0083,
+      "step": 4359
+    },
+    {
+      "epoch": 17.44,
+      "grad_norm": 0.17123810946941376,
+      "learning_rate": 6.4328657314629265e-06,
+      "loss": 0.0068,
+      "step": 4360
+    },
+    {
+      "epoch": 17.444,
+      "grad_norm": 0.2661895155906677,
+      "learning_rate": 6.422845691382766e-06,
+      "loss": 0.0085,
+      "step": 4361
+    },
+    {
+      "epoch": 17.448,
+      "grad_norm": 0.20732519030570984,
+      "learning_rate": 6.412825651302605e-06,
+      "loss": 0.0082,
+      "step": 4362
+    },
+    {
+      "epoch": 17.452,
+      "grad_norm": 0.20721253752708435,
+      "learning_rate": 6.402805611222445e-06,
+      "loss": 0.0087,
+      "step": 4363
+    },
+    {
+      "epoch": 17.456,
+      "grad_norm": 0.15347276628017426,
+      "learning_rate": 6.392785571142285e-06,
+      "loss": 0.0063,
+      "step": 4364
+    },
+    {
+      "epoch": 17.46,
+      "grad_norm": 0.19956089556217194,
+      "learning_rate": 6.382765531062125e-06,
+      "loss": 0.0082,
+      "step": 4365
+    },
+    {
+      "epoch": 17.464,
+      "grad_norm": 0.15364766120910645,
+      "learning_rate": 6.372745490981964e-06,
+      "loss": 0.0075,
+      "step": 4366
+    },
+    {
+      "epoch": 17.468,
+      "grad_norm": 0.48688241839408875,
+      "learning_rate": 6.362725450901804e-06,
+      "loss": 0.0109,
+      "step": 4367
+    },
+    {
+      "epoch": 17.472,
+      "grad_norm": 0.19626305997371674,
+      "learning_rate": 6.352705410821644e-06,
+      "loss": 0.0085,
+      "step": 4368
+    },
+    {
+      "epoch": 17.476,
+      "grad_norm": 0.21328666806221008,
+      "learning_rate": 6.342685370741483e-06,
+      "loss": 0.0087,
+      "step": 4369
+    },
+    {
+      "epoch": 17.48,
+      "grad_norm": 0.2773035764694214,
+      "learning_rate": 6.332665330661323e-06,
+      "loss": 0.0107,
+      "step": 4370
+    },
+    {
+      "epoch": 17.484,
+      "grad_norm": 0.15668006241321564,
+      "learning_rate": 6.322645290581163e-06,
+      "loss": 0.0075,
+      "step": 4371
+    },
+    {
+      "epoch": 17.488,
+      "grad_norm": 0.20465034246444702,
+      "learning_rate": 6.312625250501002e-06,
+      "loss": 0.0079,
+      "step": 4372
+    },
+    {
+      "epoch": 17.492,
+      "grad_norm": 0.1748172491788864,
+      "learning_rate": 6.302605210420842e-06,
+      "loss": 0.0074,
+      "step": 4373
+    },
+    {
+      "epoch": 17.496,
+      "grad_norm": 0.13092002272605896,
+      "learning_rate": 6.292585170340682e-06,
+      "loss": 0.0077,
+      "step": 4374
+    },
+    {
+      "epoch": 17.5,
+      "grad_norm": 0.13474947214126587,
+      "learning_rate": 6.282565130260522e-06,
+      "loss": 0.0064,
+      "step": 4375
+    },
+    {
+      "epoch": 17.504,
+      "grad_norm": 0.17254269123077393,
+      "learning_rate": 6.272545090180361e-06,
+      "loss": 0.0076,
+      "step": 4376
+    },
+    {
+      "epoch": 17.508,
+      "grad_norm": 0.1626187562942505,
+      "learning_rate": 6.2625250501002e-06,
+      "loss": 0.008,
+      "step": 4377
+    },
+    {
+      "epoch": 17.512,
+      "grad_norm": 0.12236899137496948,
+      "learning_rate": 6.252505010020041e-06,
+      "loss": 0.0073,
+      "step": 4378
+    },
+    {
+      "epoch": 17.516,
+      "grad_norm": 0.1585134118795395,
+      "learning_rate": 6.24248496993988e-06,
+      "loss": 0.0067,
+      "step": 4379
+    },
+    {
+      "epoch": 17.52,
+      "grad_norm": 0.2114783227443695,
+      "learning_rate": 6.232464929859719e-06,
+      "loss": 0.009,
+      "step": 4380
+    },
+    {
+      "epoch": 17.524,
+      "grad_norm": 0.17043016850948334,
+      "learning_rate": 6.22244488977956e-06,
+      "loss": 0.0076,
+      "step": 4381
+    },
+    {
+      "epoch": 17.528,
+      "grad_norm": 0.2268737554550171,
+      "learning_rate": 6.212424849699399e-06,
+      "loss": 0.0084,
+      "step": 4382
+    },
+    {
+      "epoch": 17.532,
+      "grad_norm": 0.18764908611774445,
+      "learning_rate": 6.202404809619239e-06,
+      "loss": 0.0078,
+      "step": 4383
+    },
+    {
+      "epoch": 17.536,
+      "grad_norm": 0.1297231912612915,
+      "learning_rate": 6.192384769539078e-06,
+      "loss": 0.0069,
+      "step": 4384
+    },
+    {
+      "epoch": 17.54,
+      "grad_norm": 0.134876549243927,
+      "learning_rate": 6.182364729458918e-06,
+      "loss": 0.0074,
+      "step": 4385
+    },
+    {
+      "epoch": 17.544,
+      "grad_norm": 0.1665986031293869,
+      "learning_rate": 6.1723446893787575e-06,
+      "loss": 0.008,
+      "step": 4386
+    },
+    {
+      "epoch": 17.548000000000002,
+      "grad_norm": 0.17875370383262634,
+      "learning_rate": 6.162324649298597e-06,
+      "loss": 0.0077,
+      "step": 4387
+    },
+    {
+      "epoch": 17.552,
+      "grad_norm": 0.11975561082363129,
+      "learning_rate": 6.1523046092184376e-06,
+      "loss": 0.0048,
+      "step": 4388
+    },
+    {
+      "epoch": 17.556,
+      "grad_norm": 0.15035223960876465,
+      "learning_rate": 6.142284569138277e-06,
+      "loss": 0.0087,
+      "step": 4389
+    },
+    {
+      "epoch": 17.56,
+      "grad_norm": 0.1503482609987259,
+      "learning_rate": 6.132264529058117e-06,
+      "loss": 0.0077,
+      "step": 4390
+    },
+    {
+      "epoch": 17.564,
+      "grad_norm": 0.12905697524547577,
+      "learning_rate": 6.122244488977956e-06,
+      "loss": 0.0065,
+      "step": 4391
+    },
+    {
+      "epoch": 17.568,
+      "grad_norm": 0.13570000231266022,
+      "learning_rate": 6.112224448897796e-06,
+      "loss": 0.0078,
+      "step": 4392
+    },
+    {
+      "epoch": 17.572,
+      "grad_norm": 0.12492106854915619,
+      "learning_rate": 6.1022044088176355e-06,
+      "loss": 0.0068,
+      "step": 4393
+    },
+    {
+      "epoch": 17.576,
+      "grad_norm": 0.23991592228412628,
+      "learning_rate": 6.092184368737475e-06,
+      "loss": 0.0086,
+      "step": 4394
+    },
+    {
+      "epoch": 17.58,
+      "grad_norm": 0.15174011886119843,
+      "learning_rate": 6.0821643286573155e-06,
+      "loss": 0.0077,
+      "step": 4395
+    },
+    {
+      "epoch": 17.584,
+      "grad_norm": 0.16359587013721466,
+      "learning_rate": 6.072144288577154e-06,
+      "loss": 0.0077,
+      "step": 4396
+    },
+    {
+      "epoch": 17.588,
+      "grad_norm": 0.2310338318347931,
+      "learning_rate": 6.062124248496994e-06,
+      "loss": 0.0084,
+      "step": 4397
+    },
+    {
+      "epoch": 17.592,
+      "grad_norm": 0.1341230720281601,
+      "learning_rate": 6.052104208416834e-06,
+      "loss": 0.0083,
+      "step": 4398
+    },
+    {
+      "epoch": 17.596,
+      "grad_norm": 0.21559903025627136,
+      "learning_rate": 6.042084168336674e-06,
+      "loss": 0.0089,
+      "step": 4399
+    },
+    {
+      "epoch": 17.6,
+      "grad_norm": 0.15158264338970184,
+      "learning_rate": 6.0320641282565135e-06,
+      "loss": 0.0073,
+      "step": 4400
+    },
+    {
+      "epoch": 17.604,
+      "grad_norm": 0.2421329915523529,
+      "learning_rate": 6.022044088176353e-06,
+      "loss": 0.0081,
+      "step": 4401
+    },
+    {
+      "epoch": 17.608,
+      "grad_norm": 0.2801038920879364,
+      "learning_rate": 6.012024048096193e-06,
+      "loss": 0.0089,
+      "step": 4402
+    },
+    {
+      "epoch": 17.612,
+      "grad_norm": 0.1329454481601715,
+      "learning_rate": 6.002004008016032e-06,
+      "loss": 0.0078,
+      "step": 4403
+    },
+    {
+      "epoch": 17.616,
+      "grad_norm": 0.23923999071121216,
+      "learning_rate": 5.991983967935872e-06,
+      "loss": 0.0086,
+      "step": 4404
+    },
+    {
+      "epoch": 17.62,
+      "grad_norm": 0.15022501349449158,
+      "learning_rate": 5.981963927855712e-06,
+      "loss": 0.0067,
+      "step": 4405
+    },
+    {
+      "epoch": 17.624,
+      "grad_norm": 0.1838655024766922,
+      "learning_rate": 5.971943887775552e-06,
+      "loss": 0.0073,
+      "step": 4406
+    },
+    {
+      "epoch": 17.628,
+      "grad_norm": 0.15019722282886505,
+      "learning_rate": 5.961923847695391e-06,
+      "loss": 0.0072,
+      "step": 4407
+    },
+    {
+      "epoch": 17.632,
+      "grad_norm": 0.2248644381761551,
+      "learning_rate": 5.95190380761523e-06,
+      "loss": 0.0085,
+      "step": 4408
+    },
+    {
+      "epoch": 17.636,
+      "grad_norm": 0.25571855902671814,
+      "learning_rate": 5.941883767535071e-06,
+      "loss": 0.0098,
+      "step": 4409
+    },
+    {
+      "epoch": 17.64,
+      "grad_norm": 0.14798125624656677,
+      "learning_rate": 5.93186372745491e-06,
+      "loss": 0.0076,
+      "step": 4410
+    },
+    {
+      "epoch": 17.644,
+      "grad_norm": 0.199660062789917,
+      "learning_rate": 5.92184368737475e-06,
+      "loss": 0.0085,
+      "step": 4411
+    },
+    {
+      "epoch": 17.648,
+      "grad_norm": 0.15380799770355225,
+      "learning_rate": 5.911823647294589e-06,
+      "loss": 0.008,
+      "step": 4412
+    },
+    {
+      "epoch": 17.652,
+      "grad_norm": 0.20436356961727142,
+      "learning_rate": 5.901803607214429e-06,
+      "loss": 0.0054,
+      "step": 4413
+    },
+    {
+      "epoch": 17.656,
+      "grad_norm": 0.22345702350139618,
+      "learning_rate": 5.8917835671342686e-06,
+      "loss": 0.0082,
+      "step": 4414
+    },
+    {
+      "epoch": 17.66,
+      "grad_norm": 0.17874562740325928,
+      "learning_rate": 5.881763527054108e-06,
+      "loss": 0.0084,
+      "step": 4415
+    },
+    {
+      "epoch": 17.664,
+      "grad_norm": 0.14801347255706787,
+      "learning_rate": 5.871743486973949e-06,
+      "loss": 0.0072,
+      "step": 4416
+    },
+    {
+      "epoch": 17.668,
+      "grad_norm": 0.20537561178207397,
+      "learning_rate": 5.861723446893788e-06,
+      "loss": 0.0079,
+      "step": 4417
+    },
+    {
+      "epoch": 17.672,
+      "grad_norm": 0.1717330515384674,
+      "learning_rate": 5.851703406813627e-06,
+      "loss": 0.0077,
+      "step": 4418
+    },
+    {
+      "epoch": 17.676,
+      "grad_norm": 0.13762032985687256,
+      "learning_rate": 5.841683366733467e-06,
+      "loss": 0.0069,
+      "step": 4419
+    },
+    {
+      "epoch": 17.68,
+      "grad_norm": 0.16380003094673157,
+      "learning_rate": 5.831663326653307e-06,
+      "loss": 0.0077,
+      "step": 4420
+    },
+    {
+      "epoch": 17.684,
+      "grad_norm": 0.20809531211853027,
+      "learning_rate": 5.8216432865731465e-06,
+      "loss": 0.0084,
+      "step": 4421
+    },
+    {
+      "epoch": 17.688,
+      "grad_norm": 0.24906416237354279,
+      "learning_rate": 5.811623246492986e-06,
+      "loss": 0.0077,
+      "step": 4422
+    },
+    {
+      "epoch": 17.692,
+      "grad_norm": 0.15348020195960999,
+      "learning_rate": 5.801603206412826e-06,
+      "loss": 0.0081,
+      "step": 4423
+    },
+    {
+      "epoch": 17.696,
+      "grad_norm": 0.15915721654891968,
+      "learning_rate": 5.791583166332665e-06,
+      "loss": 0.0071,
+      "step": 4424
+    },
+    {
+      "epoch": 17.7,
+      "grad_norm": 0.1927608698606491,
+      "learning_rate": 5.781563126252505e-06,
+      "loss": 0.0071,
+      "step": 4425
+    },
+    {
+      "epoch": 17.704,
+      "grad_norm": 0.23460881412029266,
+      "learning_rate": 5.771543086172345e-06,
+      "loss": 0.0077,
+      "step": 4426
+    },
+    {
+      "epoch": 17.708,
+      "grad_norm": 0.1741648018360138,
+      "learning_rate": 5.761523046092185e-06,
+      "loss": 0.008,
+      "step": 4427
+    },
+    {
+      "epoch": 17.712,
+      "grad_norm": 0.13403233885765076,
+      "learning_rate": 5.7515030060120245e-06,
+      "loss": 0.005,
+      "step": 4428
+    },
+    {
+      "epoch": 17.716,
+      "grad_norm": 0.0899837389588356,
+      "learning_rate": 5.741482965931864e-06,
+      "loss": 0.0045,
+      "step": 4429
+    },
+    {
+      "epoch": 17.72,
+      "grad_norm": 0.15737628936767578,
+      "learning_rate": 5.731462925851704e-06,
+      "loss": 0.0073,
+      "step": 4430
+    },
+    {
+      "epoch": 17.724,
+      "grad_norm": 0.14937400817871094,
+      "learning_rate": 5.721442885771543e-06,
+      "loss": 0.0081,
+      "step": 4431
+    },
+    {
+      "epoch": 17.728,
+      "grad_norm": 0.12557938694953918,
+      "learning_rate": 5.711422845691383e-06,
+      "loss": 0.0067,
+      "step": 4432
+    },
+    {
+      "epoch": 17.732,
+      "grad_norm": 0.20801903307437897,
+      "learning_rate": 5.701402805611223e-06,
+      "loss": 0.0083,
+      "step": 4433
+    },
+    {
+      "epoch": 17.736,
+      "grad_norm": 0.17545443773269653,
+      "learning_rate": 5.691382765531062e-06,
+      "loss": 0.0076,
+      "step": 4434
+    },
+    {
+      "epoch": 17.74,
+      "grad_norm": 0.16065756976604462,
+      "learning_rate": 5.681362725450902e-06,
+      "loss": 0.0078,
+      "step": 4435
+    },
+    {
+      "epoch": 17.744,
+      "grad_norm": 0.15100617706775665,
+      "learning_rate": 5.671342685370742e-06,
+      "loss": 0.0067,
+      "step": 4436
+    },
+    {
+      "epoch": 17.748,
+      "grad_norm": 0.17027676105499268,
+      "learning_rate": 5.661322645290582e-06,
+      "loss": 0.0078,
+      "step": 4437
+    },
+    {
+      "epoch": 17.752,
+      "grad_norm": 0.19848494231700897,
+      "learning_rate": 5.651302605210421e-06,
+      "loss": 0.0075,
+      "step": 4438
+    },
+    {
+      "epoch": 17.756,
+      "grad_norm": 0.15401864051818848,
+      "learning_rate": 5.641282565130261e-06,
+      "loss": 0.0073,
+      "step": 4439
+    },
+    {
+      "epoch": 17.76,
+      "grad_norm": 0.21219518780708313,
+      "learning_rate": 5.6312625250501e-06,
+      "loss": 0.0081,
+      "step": 4440
+    },
+    {
+      "epoch": 17.764,
+      "grad_norm": 0.09641958028078079,
+      "learning_rate": 5.62124248496994e-06,
+      "loss": 0.0047,
+      "step": 4441
+    },
+    {
+      "epoch": 17.768,
+      "grad_norm": 0.15153850615024567,
+      "learning_rate": 5.61122244488978e-06,
+      "loss": 0.0073,
+      "step": 4442
+    },
+    {
+      "epoch": 17.772,
+      "grad_norm": 0.20264235138893127,
+      "learning_rate": 5.60120240480962e-06,
+      "loss": 0.0078,
+      "step": 4443
+    },
+    {
+      "epoch": 17.776,
+      "grad_norm": 0.14838512241840363,
+      "learning_rate": 5.59118236472946e-06,
+      "loss": 0.0071,
+      "step": 4444
+    },
+    {
+      "epoch": 17.78,
+      "grad_norm": 0.1825137883424759,
+      "learning_rate": 5.581162324649298e-06,
+      "loss": 0.0068,
+      "step": 4445
+    },
+    {
+      "epoch": 17.784,
+      "grad_norm": 0.14169864356517792,
+      "learning_rate": 5.571142284569139e-06,
+      "loss": 0.0074,
+      "step": 4446
+    },
+    {
+      "epoch": 17.788,
+      "grad_norm": 0.20076483488082886,
+      "learning_rate": 5.561122244488978e-06,
+      "loss": 0.0085,
+      "step": 4447
+    },
+    {
+      "epoch": 17.792,
+      "grad_norm": 0.23179060220718384,
+      "learning_rate": 5.551102204408818e-06,
+      "loss": 0.0081,
+      "step": 4448
+    },
+    {
+      "epoch": 17.796,
+      "grad_norm": 0.15632283687591553,
+      "learning_rate": 5.5410821643286575e-06,
+      "loss": 0.0075,
+      "step": 4449
+    },
+    {
+      "epoch": 17.8,
+      "grad_norm": 0.16295547783374786,
+      "learning_rate": 5.531062124248497e-06,
+      "loss": 0.0078,
+      "step": 4450
+    },
+    {
+      "epoch": 17.804,
+      "grad_norm": 0.15852996706962585,
+      "learning_rate": 5.521042084168337e-06,
+      "loss": 0.0076,
+      "step": 4451
+    },
+    {
+      "epoch": 17.808,
+      "grad_norm": 0.24899086356163025,
+      "learning_rate": 5.511022044088176e-06,
+      "loss": 0.008,
+      "step": 4452
+    },
+    {
+      "epoch": 17.812,
+      "grad_norm": 0.26219606399536133,
+      "learning_rate": 5.501002004008017e-06,
+      "loss": 0.0079,
+      "step": 4453
+    },
+    {
+      "epoch": 17.816,
+      "grad_norm": 0.15716837346553802,
+      "learning_rate": 5.490981963927856e-06,
+      "loss": 0.0068,
+      "step": 4454
+    },
+    {
+      "epoch": 17.82,
+      "grad_norm": 0.21653732657432556,
+      "learning_rate": 5.480961923847696e-06,
+      "loss": 0.008,
+      "step": 4455
+    },
+    {
+      "epoch": 17.824,
+      "grad_norm": 0.1806323230266571,
+      "learning_rate": 5.4709418837675355e-06,
+      "loss": 0.0086,
+      "step": 4456
+    },
+    {
+      "epoch": 17.828,
+      "grad_norm": 0.2788422703742981,
+      "learning_rate": 5.460921843687375e-06,
+      "loss": 0.009,
+      "step": 4457
+    },
+    {
+      "epoch": 17.832,
+      "grad_norm": 0.16361011564731598,
+      "learning_rate": 5.450901803607215e-06,
+      "loss": 0.0073,
+      "step": 4458
+    },
+    {
+      "epoch": 17.836,
+      "grad_norm": 0.24502210319042206,
+      "learning_rate": 5.440881763527054e-06,
+      "loss": 0.0088,
+      "step": 4459
+    },
+    {
+      "epoch": 17.84,
+      "grad_norm": 0.18323510885238647,
+      "learning_rate": 5.430861723446895e-06,
+      "loss": 0.0073,
+      "step": 4460
+    },
+    {
+      "epoch": 17.844,
+      "grad_norm": 0.1967632919549942,
+      "learning_rate": 5.4208416833667335e-06,
+      "loss": 0.0085,
+      "step": 4461
+    },
+    {
+      "epoch": 17.848,
+      "grad_norm": 0.19177305698394775,
+      "learning_rate": 5.410821643286573e-06,
+      "loss": 0.0079,
+      "step": 4462
+    },
+    {
+      "epoch": 17.852,
+      "grad_norm": 0.1529979109764099,
+      "learning_rate": 5.4008016032064135e-06,
+      "loss": 0.0077,
+      "step": 4463
+    },
+    {
+      "epoch": 17.856,
+      "grad_norm": 0.16155672073364258,
+      "learning_rate": 5.390781563126253e-06,
+      "loss": 0.0083,
+      "step": 4464
+    },
+    {
+      "epoch": 17.86,
+      "grad_norm": 0.1222570464015007,
+      "learning_rate": 5.380761523046093e-06,
+      "loss": 0.0075,
+      "step": 4465
+    },
+    {
+      "epoch": 17.864,
+      "grad_norm": 0.17447544634342194,
+      "learning_rate": 5.370741482965931e-06,
+      "loss": 0.0074,
+      "step": 4466
+    },
+    {
+      "epoch": 17.868,
+      "grad_norm": 0.18004770576953888,
+      "learning_rate": 5.360721442885772e-06,
+      "loss": 0.008,
+      "step": 4467
+    },
+    {
+      "epoch": 17.872,
+      "grad_norm": 0.12705212831497192,
+      "learning_rate": 5.350701402805611e-06,
+      "loss": 0.0065,
+      "step": 4468
+    },
+    {
+      "epoch": 17.876,
+      "grad_norm": 0.13531920313835144,
+      "learning_rate": 5.340681362725451e-06,
+      "loss": 0.0077,
+      "step": 4469
+    },
+    {
+      "epoch": 17.88,
+      "grad_norm": 0.11091826111078262,
+      "learning_rate": 5.3306613226452914e-06,
+      "loss": 0.0038,
+      "step": 4470
+    },
+    {
+      "epoch": 17.884,
+      "grad_norm": 0.17355629801750183,
+      "learning_rate": 5.320641282565131e-06,
+      "loss": 0.0073,
+      "step": 4471
+    },
+    {
+      "epoch": 17.888,
+      "grad_norm": 0.2770105302333832,
+      "learning_rate": 5.31062124248497e-06,
+      "loss": 0.0092,
+      "step": 4472
+    },
+    {
+      "epoch": 17.892,
+      "grad_norm": 0.13555404543876648,
+      "learning_rate": 5.300601202404809e-06,
+      "loss": 0.007,
+      "step": 4473
+    },
+    {
+      "epoch": 17.896,
+      "grad_norm": 0.13651050627231598,
+      "learning_rate": 5.29058116232465e-06,
+      "loss": 0.0071,
+      "step": 4474
+    },
+    {
+      "epoch": 17.9,
+      "grad_norm": 0.22955292463302612,
+      "learning_rate": 5.280561122244489e-06,
+      "loss": 0.0086,
+      "step": 4475
+    },
+    {
+      "epoch": 17.904,
+      "grad_norm": 0.13148556649684906,
+      "learning_rate": 5.270541082164329e-06,
+      "loss": 0.007,
+      "step": 4476
+    },
+    {
+      "epoch": 17.908,
+      "grad_norm": 0.1831294298171997,
+      "learning_rate": 5.2605210420841686e-06,
+      "loss": 0.0084,
+      "step": 4477
+    },
+    {
+      "epoch": 17.912,
+      "grad_norm": 0.15592306852340698,
+      "learning_rate": 5.250501002004008e-06,
+      "loss": 0.0076,
+      "step": 4478
+    },
+    {
+      "epoch": 17.916,
+      "grad_norm": 0.20427504181861877,
+      "learning_rate": 5.240480961923848e-06,
+      "loss": 0.0083,
+      "step": 4479
+    },
+    {
+      "epoch": 17.92,
+      "grad_norm": 0.1998140513896942,
+      "learning_rate": 5.230460921843687e-06,
+      "loss": 0.0089,
+      "step": 4480
+    },
+    {
+      "epoch": 17.924,
+      "grad_norm": 0.1798437237739563,
+      "learning_rate": 5.220440881763528e-06,
+      "loss": 0.0085,
+      "step": 4481
+    },
+    {
+      "epoch": 17.928,
+      "grad_norm": 0.1546415388584137,
+      "learning_rate": 5.2104208416833665e-06,
+      "loss": 0.0076,
+      "step": 4482
+    },
+    {
+      "epoch": 17.932,
+      "grad_norm": 0.17189155519008636,
+      "learning_rate": 5.200400801603206e-06,
+      "loss": 0.0074,
+      "step": 4483
+    },
+    {
+      "epoch": 17.936,
+      "grad_norm": 0.26644057035446167,
+      "learning_rate": 5.1903807615230465e-06,
+      "loss": 0.0084,
+      "step": 4484
+    },
+    {
+      "epoch": 17.94,
+      "grad_norm": 0.12923316657543182,
+      "learning_rate": 5.180360721442886e-06,
+      "loss": 0.0067,
+      "step": 4485
+    },
+    {
+      "epoch": 17.944,
+      "grad_norm": 0.20447267591953278,
+      "learning_rate": 5.170340681362726e-06,
+      "loss": 0.0075,
+      "step": 4486
+    },
+    {
+      "epoch": 17.948,
+      "grad_norm": 0.17053300142288208,
+      "learning_rate": 5.160320641282565e-06,
+      "loss": 0.0076,
+      "step": 4487
+    },
+    {
+      "epoch": 17.951999999999998,
+      "grad_norm": 0.1922687590122223,
+      "learning_rate": 5.150300601202405e-06,
+      "loss": 0.0082,
+      "step": 4488
+    },
+    {
+      "epoch": 17.956,
+      "grad_norm": 0.17536620795726776,
+      "learning_rate": 5.1402805611222445e-06,
+      "loss": 0.0081,
+      "step": 4489
+    },
+    {
+      "epoch": 17.96,
+      "grad_norm": 0.1868142932653427,
+      "learning_rate": 5.130260521042084e-06,
+      "loss": 0.0079,
+      "step": 4490
+    },
+    {
+      "epoch": 17.964,
+      "grad_norm": 0.21813417971134186,
+      "learning_rate": 5.1202404809619245e-06,
+      "loss": 0.0082,
+      "step": 4491
+    },
+    {
+      "epoch": 17.968,
+      "grad_norm": 0.1625450849533081,
+      "learning_rate": 5.110220440881764e-06,
+      "loss": 0.0083,
+      "step": 4492
+    },
+    {
+      "epoch": 17.972,
+      "grad_norm": 0.1685769259929657,
+      "learning_rate": 5.100200400801603e-06,
+      "loss": 0.0083,
+      "step": 4493
+    },
+    {
+      "epoch": 17.976,
+      "grad_norm": 0.16878901422023773,
+      "learning_rate": 5.090180360721443e-06,
+      "loss": 0.008,
+      "step": 4494
+    },
+    {
+      "epoch": 17.98,
+      "grad_norm": 0.26042139530181885,
+      "learning_rate": 5.080160320641283e-06,
+      "loss": 0.009,
+      "step": 4495
+    },
+    {
+      "epoch": 17.984,
+      "grad_norm": 0.22069434821605682,
+      "learning_rate": 5.0701402805611224e-06,
+      "loss": 0.0072,
+      "step": 4496
+    },
+    {
+      "epoch": 17.988,
+      "grad_norm": 0.1699599176645279,
+      "learning_rate": 5.060120240480962e-06,
+      "loss": 0.0089,
+      "step": 4497
+    },
+    {
+      "epoch": 17.992,
+      "grad_norm": 0.1470976173877716,
+      "learning_rate": 5.0501002004008025e-06,
+      "loss": 0.0074,
+      "step": 4498
+    },
+    {
+      "epoch": 17.996,
+      "grad_norm": 0.22268114984035492,
+      "learning_rate": 5.040080160320641e-06,
+      "loss": 0.0084,
+      "step": 4499
+    },
+    {
+      "epoch": 18.0,
+      "grad_norm": 0.1696414351463318,
+      "learning_rate": 5.030060120240481e-06,
+      "loss": 0.0081,
+      "step": 4500
+    },
+    {
+      "epoch": 18.004,
+      "grad_norm": 0.18585962057113647,
+      "learning_rate": 5.020040080160321e-06,
+      "loss": 0.0072,
+      "step": 4501
+    },
+    {
+      "epoch": 18.008,
+      "grad_norm": 0.1416616141796112,
+      "learning_rate": 5.010020040080161e-06,
+      "loss": 0.0064,
+      "step": 4502
+    },
+    {
+      "epoch": 18.012,
+      "grad_norm": 0.1328905075788498,
+      "learning_rate": 5e-06,
+      "loss": 0.0064,
+      "step": 4503
+    },
+    {
+      "epoch": 18.016,
+      "grad_norm": 0.16968417167663574,
+      "learning_rate": 4.98997995991984e-06,
+      "loss": 0.0073,
+      "step": 4504
+    },
+    {
+      "epoch": 18.02,
+      "grad_norm": 0.15814343094825745,
+      "learning_rate": 4.97995991983968e-06,
+      "loss": 0.0073,
+      "step": 4505
+    },
+    {
+      "epoch": 18.024,
+      "grad_norm": 0.17646895349025726,
+      "learning_rate": 4.969939879759519e-06,
+      "loss": 0.0076,
+      "step": 4506
+    },
+    {
+      "epoch": 18.028,
+      "grad_norm": 0.14789192378520966,
+      "learning_rate": 4.959919839679359e-06,
+      "loss": 0.0064,
+      "step": 4507
+    },
+    {
+      "epoch": 18.032,
+      "grad_norm": 0.15149499475955963,
+      "learning_rate": 4.949899799599199e-06,
+      "loss": 0.0075,
+      "step": 4508
+    },
+    {
+      "epoch": 18.036,
+      "grad_norm": 0.12191706895828247,
+      "learning_rate": 4.939879759519038e-06,
+      "loss": 0.0065,
+      "step": 4509
+    },
+    {
+      "epoch": 18.04,
+      "grad_norm": 0.1561209261417389,
+      "learning_rate": 4.9298597194388775e-06,
+      "loss": 0.0079,
+      "step": 4510
+    },
+    {
+      "epoch": 18.044,
+      "grad_norm": 0.15725496411323547,
+      "learning_rate": 4.919839679358718e-06,
+      "loss": 0.0072,
+      "step": 4511
+    },
+    {
+      "epoch": 18.048,
+      "grad_norm": 0.14337027072906494,
+      "learning_rate": 4.9098196392785576e-06,
+      "loss": 0.0065,
+      "step": 4512
+    },
+    {
+      "epoch": 18.052,
+      "grad_norm": 0.16354084014892578,
+      "learning_rate": 4.899799599198397e-06,
+      "loss": 0.0068,
+      "step": 4513
+    },
+    {
+      "epoch": 18.056,
+      "grad_norm": 0.16241081058979034,
+      "learning_rate": 4.889779559118237e-06,
+      "loss": 0.0066,
+      "step": 4514
+    },
+    {
+      "epoch": 18.06,
+      "grad_norm": 0.1941567361354828,
+      "learning_rate": 4.879759519038076e-06,
+      "loss": 0.0062,
+      "step": 4515
+    },
+    {
+      "epoch": 18.064,
+      "grad_norm": 0.1292446106672287,
+      "learning_rate": 4.869739478957916e-06,
+      "loss": 0.0066,
+      "step": 4516
+    },
+    {
+      "epoch": 18.068,
+      "grad_norm": 0.09367750585079193,
+      "learning_rate": 4.8597194388777555e-06,
+      "loss": 0.0047,
+      "step": 4517
+    },
+    {
+      "epoch": 18.072,
+      "grad_norm": 0.14819225668907166,
+      "learning_rate": 4.849699398797596e-06,
+      "loss": 0.0074,
+      "step": 4518
+    },
+    {
+      "epoch": 18.076,
+      "grad_norm": 0.17116355895996094,
+      "learning_rate": 4.8396793587174355e-06,
+      "loss": 0.0064,
+      "step": 4519
+    },
+    {
+      "epoch": 18.08,
+      "grad_norm": 0.19013841450214386,
+      "learning_rate": 4.829659318637274e-06,
+      "loss": 0.0075,
+      "step": 4520
+    },
+    {
+      "epoch": 18.084,
+      "grad_norm": 0.17657625675201416,
+      "learning_rate": 4.819639278557115e-06,
+      "loss": 0.0077,
+      "step": 4521
+    },
+    {
+      "epoch": 18.088,
+      "grad_norm": 0.14406649768352509,
+      "learning_rate": 4.809619238476954e-06,
+      "loss": 0.0074,
+      "step": 4522
+    },
+    {
+      "epoch": 18.092,
+      "grad_norm": 0.17921116948127747,
+      "learning_rate": 4.799599198396794e-06,
+      "loss": 0.0071,
+      "step": 4523
+    },
+    {
+      "epoch": 18.096,
+      "grad_norm": 0.22926273941993713,
+      "learning_rate": 4.7895791583166335e-06,
+      "loss": 0.0083,
+      "step": 4524
+    },
+    {
+      "epoch": 18.1,
+      "grad_norm": 0.1705189198255539,
+      "learning_rate": 4.779559118236474e-06,
+      "loss": 0.0074,
+      "step": 4525
+    },
+    {
+      "epoch": 18.104,
+      "grad_norm": 0.15004006028175354,
+      "learning_rate": 4.769539078156313e-06,
+      "loss": 0.0063,
+      "step": 4526
+    },
+    {
+      "epoch": 18.108,
+      "grad_norm": 0.1770288497209549,
+      "learning_rate": 4.759519038076152e-06,
+      "loss": 0.0077,
+      "step": 4527
+    },
+    {
+      "epoch": 18.112,
+      "grad_norm": 0.14755572378635406,
+      "learning_rate": 4.749498997995992e-06,
+      "loss": 0.0065,
+      "step": 4528
+    },
+    {
+      "epoch": 18.116,
+      "grad_norm": 0.12540753185749054,
+      "learning_rate": 4.739478957915832e-06,
+      "loss": 0.0067,
+      "step": 4529
+    },
+    {
+      "epoch": 18.12,
+      "grad_norm": 0.1722092181444168,
+      "learning_rate": 4.729458917835672e-06,
+      "loss": 0.0076,
+      "step": 4530
+    },
+    {
+      "epoch": 18.124,
+      "grad_norm": 0.15818531811237335,
+      "learning_rate": 4.719438877755511e-06,
+      "loss": 0.0065,
+      "step": 4531
+    },
+    {
+      "epoch": 18.128,
+      "grad_norm": 0.17607039213180542,
+      "learning_rate": 4.709418837675351e-06,
+      "loss": 0.0058,
+      "step": 4532
+    },
+    {
+      "epoch": 18.132,
+      "grad_norm": 0.15239058434963226,
+      "learning_rate": 4.699398797595191e-06,
+      "loss": 0.0072,
+      "step": 4533
+    },
+    {
+      "epoch": 18.136,
+      "grad_norm": 0.1865680068731308,
+      "learning_rate": 4.68937875751503e-06,
+      "loss": 0.0074,
+      "step": 4534
+    },
+    {
+      "epoch": 18.14,
+      "grad_norm": 0.13207893073558807,
+      "learning_rate": 4.67935871743487e-06,
+      "loss": 0.0069,
+      "step": 4535
+    },
+    {
+      "epoch": 18.144,
+      "grad_norm": 0.1481158584356308,
+      "learning_rate": 4.669338677354709e-06,
+      "loss": 0.0068,
+      "step": 4536
+    },
+    {
+      "epoch": 18.148,
+      "grad_norm": 0.1650175005197525,
+      "learning_rate": 4.659318637274549e-06,
+      "loss": 0.0084,
+      "step": 4537
+    },
+    {
+      "epoch": 18.152,
+      "grad_norm": 0.2189430296421051,
+      "learning_rate": 4.6492985971943886e-06,
+      "loss": 0.0068,
+      "step": 4538
+    },
+    {
+      "epoch": 18.156,
+      "grad_norm": 0.17136210203170776,
+      "learning_rate": 4.639278557114229e-06,
+      "loss": 0.0071,
+      "step": 4539
+    },
+    {
+      "epoch": 18.16,
+      "grad_norm": 0.1906738430261612,
+      "learning_rate": 4.6292585170340686e-06,
+      "loss": 0.007,
+      "step": 4540
+    },
+    {
+      "epoch": 18.164,
+      "grad_norm": 0.1356390118598938,
+      "learning_rate": 4.619238476953908e-06,
+      "loss": 0.0063,
+      "step": 4541
+    },
+    {
+      "epoch": 18.168,
+      "grad_norm": 0.1401359885931015,
+      "learning_rate": 4.609218436873748e-06,
+      "loss": 0.0073,
+      "step": 4542
+    },
+    {
+      "epoch": 18.172,
+      "grad_norm": 0.22201108932495117,
+      "learning_rate": 4.599198396793587e-06,
+      "loss": 0.0082,
+      "step": 4543
+    },
+    {
+      "epoch": 18.176,
+      "grad_norm": 0.17944709956645966,
+      "learning_rate": 4.589178356713427e-06,
+      "loss": 0.0065,
+      "step": 4544
+    },
+    {
+      "epoch": 18.18,
+      "grad_norm": 0.19789981842041016,
+      "learning_rate": 4.5791583166332665e-06,
+      "loss": 0.0071,
+      "step": 4545
+    },
+    {
+      "epoch": 18.184,
+      "grad_norm": 0.2103254646062851,
+      "learning_rate": 4.569138276553107e-06,
+      "loss": 0.0074,
+      "step": 4546
+    },
+    {
+      "epoch": 18.188,
+      "grad_norm": 0.15887366235256195,
+      "learning_rate": 4.559118236472946e-06,
+      "loss": 0.0064,
+      "step": 4547
+    },
+    {
+      "epoch": 18.192,
+      "grad_norm": 0.12371140718460083,
+      "learning_rate": 4.549098196392785e-06,
+      "loss": 0.0064,
+      "step": 4548
+    },
+    {
+      "epoch": 18.196,
+      "grad_norm": 0.20762166380882263,
+      "learning_rate": 4.539078156312626e-06,
+      "loss": 0.0078,
+      "step": 4549
+    },
+    {
+      "epoch": 18.2,
+      "grad_norm": 0.12717363238334656,
+      "learning_rate": 4.529058116232465e-06,
+      "loss": 0.0064,
+      "step": 4550
+    },
+    {
+      "epoch": 18.204,
+      "grad_norm": 0.2185417264699936,
+      "learning_rate": 4.519038076152305e-06,
+      "loss": 0.0068,
+      "step": 4551
+    },
+    {
+      "epoch": 18.208,
+      "grad_norm": 0.1922868937253952,
+      "learning_rate": 4.5090180360721445e-06,
+      "loss": 0.0049,
+      "step": 4552
+    },
+    {
+      "epoch": 18.212,
+      "grad_norm": 0.20411092042922974,
+      "learning_rate": 4.498997995991984e-06,
+      "loss": 0.0076,
+      "step": 4553
+    },
+    {
+      "epoch": 18.216,
+      "grad_norm": 0.18740029633045197,
+      "learning_rate": 4.488977955911824e-06,
+      "loss": 0.008,
+      "step": 4554
+    },
+    {
+      "epoch": 18.22,
+      "grad_norm": 0.14314641058444977,
+      "learning_rate": 4.478957915831663e-06,
+      "loss": 0.0068,
+      "step": 4555
+    },
+    {
+      "epoch": 18.224,
+      "grad_norm": 0.15303204953670502,
+      "learning_rate": 4.468937875751504e-06,
+      "loss": 0.0069,
+      "step": 4556
+    },
+    {
+      "epoch": 18.228,
+      "grad_norm": 0.21865060925483704,
+      "learning_rate": 4.458917835671343e-06,
+      "loss": 0.0068,
+      "step": 4557
+    },
+    {
+      "epoch": 18.232,
+      "grad_norm": 0.14031033217906952,
+      "learning_rate": 4.448897795591182e-06,
+      "loss": 0.0071,
+      "step": 4558
+    },
+    {
+      "epoch": 18.236,
+      "grad_norm": 0.18095235526561737,
+      "learning_rate": 4.4388777555110225e-06,
+      "loss": 0.0071,
+      "step": 4559
+    },
+    {
+      "epoch": 18.24,
+      "grad_norm": 0.16418349742889404,
+      "learning_rate": 4.428857715430862e-06,
+      "loss": 0.0077,
+      "step": 4560
+    },
+    {
+      "epoch": 18.244,
+      "grad_norm": 0.14331473410129547,
+      "learning_rate": 4.418837675350702e-06,
+      "loss": 0.0067,
+      "step": 4561
+    },
+    {
+      "epoch": 18.248,
+      "grad_norm": 0.15830665826797485,
+      "learning_rate": 4.408817635270541e-06,
+      "loss": 0.0074,
+      "step": 4562
+    },
+    {
+      "epoch": 18.252,
+      "grad_norm": 0.1355806142091751,
+      "learning_rate": 4.398797595190381e-06,
+      "loss": 0.0069,
+      "step": 4563
+    },
+    {
+      "epoch": 18.256,
+      "grad_norm": 0.05929170548915863,
+      "learning_rate": 4.38877755511022e-06,
+      "loss": 0.0023,
+      "step": 4564
+    },
+    {
+      "epoch": 18.26,
+      "grad_norm": 0.15484470129013062,
+      "learning_rate": 4.37875751503006e-06,
+      "loss": 0.0061,
+      "step": 4565
+    },
+    {
+      "epoch": 18.264,
+      "grad_norm": 0.17139868438243866,
+      "learning_rate": 4.3687374749499e-06,
+      "loss": 0.006,
+      "step": 4566
+    },
+    {
+      "epoch": 18.268,
+      "grad_norm": 0.23054589331150055,
+      "learning_rate": 4.35871743486974e-06,
+      "loss": 0.008,
+      "step": 4567
+    },
+    {
+      "epoch": 18.272,
+      "grad_norm": 0.1985149085521698,
+      "learning_rate": 4.34869739478958e-06,
+      "loss": 0.0072,
+      "step": 4568
+    },
+    {
+      "epoch": 18.276,
+      "grad_norm": 0.17223723232746124,
+      "learning_rate": 4.338677354709419e-06,
+      "loss": 0.0072,
+      "step": 4569
+    },
+    {
+      "epoch": 18.28,
+      "grad_norm": 0.14117112755775452,
+      "learning_rate": 4.328657314629259e-06,
+      "loss": 0.0065,
+      "step": 4570
+    },
+    {
+      "epoch": 18.284,
+      "grad_norm": 0.22822153568267822,
+      "learning_rate": 4.318637274549098e-06,
+      "loss": 0.0074,
+      "step": 4571
+    },
+    {
+      "epoch": 18.288,
+      "grad_norm": 0.21039901673793793,
+      "learning_rate": 4.308617234468938e-06,
+      "loss": 0.0079,
+      "step": 4572
+    },
+    {
+      "epoch": 18.292,
+      "grad_norm": 0.14674365520477295,
+      "learning_rate": 4.298597194388778e-06,
+      "loss": 0.0065,
+      "step": 4573
+    },
+    {
+      "epoch": 18.296,
+      "grad_norm": 0.1451415717601776,
+      "learning_rate": 4.288577154308617e-06,
+      "loss": 0.0065,
+      "step": 4574
+    },
+    {
+      "epoch": 18.3,
+      "grad_norm": 0.15624050796031952,
+      "learning_rate": 4.278557114228457e-06,
+      "loss": 0.0068,
+      "step": 4575
+    },
+    {
+      "epoch": 18.304,
+      "grad_norm": 0.1852012276649475,
+      "learning_rate": 4.268537074148297e-06,
+      "loss": 0.0077,
+      "step": 4576
+    },
+    {
+      "epoch": 18.308,
+      "grad_norm": 0.22112013399600983,
+      "learning_rate": 4.258517034068137e-06,
+      "loss": 0.0074,
+      "step": 4577
+    },
+    {
+      "epoch": 18.312,
+      "grad_norm": 0.1755540817975998,
+      "learning_rate": 4.248496993987976e-06,
+      "loss": 0.0078,
+      "step": 4578
+    },
+    {
+      "epoch": 18.316,
+      "grad_norm": 0.16024050116539001,
+      "learning_rate": 4.238476953907815e-06,
+      "loss": 0.007,
+      "step": 4579
+    },
+    {
+      "epoch": 18.32,
+      "grad_norm": 0.18745197355747223,
+      "learning_rate": 4.2284569138276555e-06,
+      "loss": 0.0077,
+      "step": 4580
+    },
+    {
+      "epoch": 18.324,
+      "grad_norm": 0.18656589090824127,
+      "learning_rate": 4.218436873747495e-06,
+      "loss": 0.0081,
+      "step": 4581
+    },
+    {
+      "epoch": 18.328,
+      "grad_norm": 0.16219571232795715,
+      "learning_rate": 4.208416833667335e-06,
+      "loss": 0.0072,
+      "step": 4582
+    },
+    {
+      "epoch": 18.332,
+      "grad_norm": 0.23128695785999298,
+      "learning_rate": 4.198396793587175e-06,
+      "loss": 0.0079,
+      "step": 4583
+    },
+    {
+      "epoch": 18.336,
+      "grad_norm": 0.21390704810619354,
+      "learning_rate": 4.188376753507015e-06,
+      "loss": 0.0088,
+      "step": 4584
+    },
+    {
+      "epoch": 18.34,
+      "grad_norm": 0.23539131879806519,
+      "learning_rate": 4.1783567134268534e-06,
+      "loss": 0.0081,
+      "step": 4585
+    },
+    {
+      "epoch": 18.344,
+      "grad_norm": 0.19893336296081543,
+      "learning_rate": 4.168336673346693e-06,
+      "loss": 0.0068,
+      "step": 4586
+    },
+    {
+      "epoch": 18.348,
+      "grad_norm": 0.44295942783355713,
+      "learning_rate": 4.1583166332665335e-06,
+      "loss": 0.0096,
+      "step": 4587
+    },
+    {
+      "epoch": 18.352,
+      "grad_norm": 0.17842140793800354,
+      "learning_rate": 4.148296593186373e-06,
+      "loss": 0.0077,
+      "step": 4588
+    },
+    {
+      "epoch": 18.356,
+      "grad_norm": 0.22559908032417297,
+      "learning_rate": 4.138276553106213e-06,
+      "loss": 0.0067,
+      "step": 4589
+    },
+    {
+      "epoch": 18.36,
+      "grad_norm": 0.19061769545078278,
+      "learning_rate": 4.128256513026052e-06,
+      "loss": 0.0076,
+      "step": 4590
+    },
+    {
+      "epoch": 18.364,
+      "grad_norm": 0.14413903653621674,
+      "learning_rate": 4.118236472945892e-06,
+      "loss": 0.0074,
+      "step": 4591
+    },
+    {
+      "epoch": 18.368,
+      "grad_norm": 0.1512213498353958,
+      "learning_rate": 4.108216432865731e-06,
+      "loss": 0.0082,
+      "step": 4592
+    },
+    {
+      "epoch": 18.372,
+      "grad_norm": 0.1925954520702362,
+      "learning_rate": 4.098196392785571e-06,
+      "loss": 0.0071,
+      "step": 4593
+    },
+    {
+      "epoch": 18.376,
+      "grad_norm": 0.18109238147735596,
+      "learning_rate": 4.0881763527054114e-06,
+      "loss": 0.0075,
+      "step": 4594
+    },
+    {
+      "epoch": 18.38,
+      "grad_norm": 0.1389477699995041,
+      "learning_rate": 4.078156312625251e-06,
+      "loss": 0.0066,
+      "step": 4595
+    },
+    {
+      "epoch": 18.384,
+      "grad_norm": 0.24592560529708862,
+      "learning_rate": 4.06813627254509e-06,
+      "loss": 0.0075,
+      "step": 4596
+    },
+    {
+      "epoch": 18.388,
+      "grad_norm": 0.18193602561950684,
+      "learning_rate": 4.05811623246493e-06,
+      "loss": 0.0081,
+      "step": 4597
+    },
+    {
+      "epoch": 18.392,
+      "grad_norm": 0.26493698358535767,
+      "learning_rate": 4.04809619238477e-06,
+      "loss": 0.0081,
+      "step": 4598
+    },
+    {
+      "epoch": 18.396,
+      "grad_norm": 0.16297456622123718,
+      "learning_rate": 4.038076152304609e-06,
+      "loss": 0.0072,
+      "step": 4599
+    },
+    {
+      "epoch": 18.4,
+      "grad_norm": 0.27221083641052246,
+      "learning_rate": 4.028056112224449e-06,
+      "loss": 0.0088,
+      "step": 4600
+    },
+    {
+      "epoch": 18.404,
+      "grad_norm": 0.19877958297729492,
+      "learning_rate": 4.0180360721442886e-06,
+      "loss": 0.0083,
+      "step": 4601
+    },
+    {
+      "epoch": 18.408,
+      "grad_norm": 0.19414174556732178,
+      "learning_rate": 4.008016032064128e-06,
+      "loss": 0.0071,
+      "step": 4602
+    },
+    {
+      "epoch": 18.412,
+      "grad_norm": 0.1475425511598587,
+      "learning_rate": 3.997995991983968e-06,
+      "loss": 0.0051,
+      "step": 4603
+    },
+    {
+      "epoch": 18.416,
+      "grad_norm": 0.13370120525360107,
+      "learning_rate": 3.987975951903808e-06,
+      "loss": 0.0073,
+      "step": 4604
+    },
+    {
+      "epoch": 18.42,
+      "grad_norm": 0.17306354641914368,
+      "learning_rate": 3.977955911823648e-06,
+      "loss": 0.0077,
+      "step": 4605
+    },
+    {
+      "epoch": 18.424,
+      "grad_norm": 0.21375983953475952,
+      "learning_rate": 3.9679358717434865e-06,
+      "loss": 0.0068,
+      "step": 4606
+    },
+    {
+      "epoch": 18.428,
+      "grad_norm": 0.17107589542865753,
+      "learning_rate": 3.957915831663327e-06,
+      "loss": 0.0076,
+      "step": 4607
+    },
+    {
+      "epoch": 18.432,
+      "grad_norm": 0.15007686614990234,
+      "learning_rate": 3.9478957915831665e-06,
+      "loss": 0.008,
+      "step": 4608
+    },
+    {
+      "epoch": 18.436,
+      "grad_norm": 0.13160204887390137,
+      "learning_rate": 3.937875751503006e-06,
+      "loss": 0.0064,
+      "step": 4609
+    },
+    {
+      "epoch": 18.44,
+      "grad_norm": 0.2253139317035675,
+      "learning_rate": 3.927855711422846e-06,
+      "loss": 0.0077,
+      "step": 4610
+    },
+    {
+      "epoch": 18.444,
+      "grad_norm": 0.12682516872882843,
+      "learning_rate": 3.917835671342686e-06,
+      "loss": 0.0035,
+      "step": 4611
+    },
+    {
+      "epoch": 18.448,
+      "grad_norm": 0.17346949875354767,
+      "learning_rate": 3.907815631262525e-06,
+      "loss": 0.007,
+      "step": 4612
+    },
+    {
+      "epoch": 18.452,
+      "grad_norm": 0.24276018142700195,
+      "learning_rate": 3.8977955911823645e-06,
+      "loss": 0.0075,
+      "step": 4613
+    },
+    {
+      "epoch": 18.456,
+      "grad_norm": 0.13902723789215088,
+      "learning_rate": 3.887775551102205e-06,
+      "loss": 0.0067,
+      "step": 4614
+    },
+    {
+      "epoch": 18.46,
+      "grad_norm": 0.23678214848041534,
+      "learning_rate": 3.8777555110220445e-06,
+      "loss": 0.0079,
+      "step": 4615
+    },
+    {
+      "epoch": 18.464,
+      "grad_norm": 0.11600477248430252,
+      "learning_rate": 3.867735470941884e-06,
+      "loss": 0.0069,
+      "step": 4616
+    },
+    {
+      "epoch": 18.468,
+      "grad_norm": 0.17199131846427917,
+      "learning_rate": 3.857715430861724e-06,
+      "loss": 0.0079,
+      "step": 4617
+    },
+    {
+      "epoch": 18.472,
+      "grad_norm": 0.13820935785770416,
+      "learning_rate": 3.847695390781563e-06,
+      "loss": 0.0065,
+      "step": 4618
+    },
+    {
+      "epoch": 18.476,
+      "grad_norm": 0.2043878436088562,
+      "learning_rate": 3.837675350701403e-06,
+      "loss": 0.008,
+      "step": 4619
+    },
+    {
+      "epoch": 18.48,
+      "grad_norm": 0.1664237231016159,
+      "learning_rate": 3.8276553106212424e-06,
+      "loss": 0.0073,
+      "step": 4620
+    },
+    {
+      "epoch": 18.484,
+      "grad_norm": 0.14281480014324188,
+      "learning_rate": 3.817635270541083e-06,
+      "loss": 0.0077,
+      "step": 4621
+    },
+    {
+      "epoch": 18.488,
+      "grad_norm": 0.15422894060611725,
+      "learning_rate": 3.8076152304609225e-06,
+      "loss": 0.0071,
+      "step": 4622
+    },
+    {
+      "epoch": 18.492,
+      "grad_norm": 0.18872609734535217,
+      "learning_rate": 3.7975951903807616e-06,
+      "loss": 0.0075,
+      "step": 4623
+    },
+    {
+      "epoch": 18.496,
+      "grad_norm": 0.2351578027009964,
+      "learning_rate": 3.7875751503006012e-06,
+      "loss": 0.0079,
+      "step": 4624
+    },
+    {
+      "epoch": 18.5,
+      "grad_norm": 0.15533076226711273,
+      "learning_rate": 3.7775551102204412e-06,
+      "loss": 0.007,
+      "step": 4625
+    },
+    {
+      "epoch": 18.504,
+      "grad_norm": 0.19388796389102936,
+      "learning_rate": 3.767535070140281e-06,
+      "loss": 0.0069,
+      "step": 4626
+    },
+    {
+      "epoch": 18.508,
+      "grad_norm": 0.17046746611595154,
+      "learning_rate": 3.757515030060121e-06,
+      "loss": 0.0067,
+      "step": 4627
+    },
+    {
+      "epoch": 18.512,
+      "grad_norm": 0.15375731885433197,
+      "learning_rate": 3.74749498997996e-06,
+      "loss": 0.0074,
+      "step": 4628
+    },
+    {
+      "epoch": 18.516,
+      "grad_norm": 0.2605665922164917,
+      "learning_rate": 3.7374749498997996e-06,
+      "loss": 0.0073,
+      "step": 4629
+    },
+    {
+      "epoch": 18.52,
+      "grad_norm": 0.18455125391483307,
+      "learning_rate": 3.7274549098196396e-06,
+      "loss": 0.0081,
+      "step": 4630
+    },
+    {
+      "epoch": 18.524,
+      "grad_norm": 0.21785873174667358,
+      "learning_rate": 3.717434869739479e-06,
+      "loss": 0.0082,
+      "step": 4631
+    },
+    {
+      "epoch": 18.528,
+      "grad_norm": 0.14746534824371338,
+      "learning_rate": 3.707414829659319e-06,
+      "loss": 0.0074,
+      "step": 4632
+    },
+    {
+      "epoch": 18.532,
+      "grad_norm": 0.1808655709028244,
+      "learning_rate": 3.6973947895791584e-06,
+      "loss": 0.0073,
+      "step": 4633
+    },
+    {
+      "epoch": 18.536,
+      "grad_norm": 0.20626331865787506,
+      "learning_rate": 3.687374749498998e-06,
+      "loss": 0.0075,
+      "step": 4634
+    },
+    {
+      "epoch": 18.54,
+      "grad_norm": 0.19164638221263885,
+      "learning_rate": 3.677354709418838e-06,
+      "loss": 0.0073,
+      "step": 4635
+    },
+    {
+      "epoch": 18.544,
+      "grad_norm": 0.12534405291080475,
+      "learning_rate": 3.6673346693386775e-06,
+      "loss": 0.0072,
+      "step": 4636
+    },
+    {
+      "epoch": 18.548000000000002,
+      "grad_norm": 0.1504947692155838,
+      "learning_rate": 3.6573146292585176e-06,
+      "loss": 0.007,
+      "step": 4637
+    },
+    {
+      "epoch": 18.552,
+      "grad_norm": 0.2330176681280136,
+      "learning_rate": 3.647294589178357e-06,
+      "loss": 0.0077,
+      "step": 4638
+    },
+    {
+      "epoch": 18.556,
+      "grad_norm": 0.20846612751483917,
+      "learning_rate": 3.6372745490981963e-06,
+      "loss": 0.0087,
+      "step": 4639
+    },
+    {
+      "epoch": 18.56,
+      "grad_norm": 0.15900227427482605,
+      "learning_rate": 3.6272545090180363e-06,
+      "loss": 0.0062,
+      "step": 4640
+    },
+    {
+      "epoch": 18.564,
+      "grad_norm": 0.10888917744159698,
+      "learning_rate": 3.617234468937876e-06,
+      "loss": 0.0062,
+      "step": 4641
+    },
+    {
+      "epoch": 18.568,
+      "grad_norm": 0.18537414073944092,
+      "learning_rate": 3.607214428857716e-06,
+      "loss": 0.0078,
+      "step": 4642
+    },
+    {
+      "epoch": 18.572,
+      "grad_norm": 0.2435263842344284,
+      "learning_rate": 3.5971943887775555e-06,
+      "loss": 0.008,
+      "step": 4643
+    },
+    {
+      "epoch": 18.576,
+      "grad_norm": 0.1557953655719757,
+      "learning_rate": 3.5871743486973947e-06,
+      "loss": 0.0066,
+      "step": 4644
+    },
+    {
+      "epoch": 18.58,
+      "grad_norm": 0.22416846454143524,
+      "learning_rate": 3.5771543086172343e-06,
+      "loss": 0.0078,
+      "step": 4645
+    },
+    {
+      "epoch": 18.584,
+      "grad_norm": 0.1947488784790039,
+      "learning_rate": 3.5671342685370743e-06,
+      "loss": 0.0077,
+      "step": 4646
+    },
+    {
+      "epoch": 18.588,
+      "grad_norm": 0.1567157357931137,
+      "learning_rate": 3.5571142284569143e-06,
+      "loss": 0.0064,
+      "step": 4647
+    },
+    {
+      "epoch": 18.592,
+      "grad_norm": 0.19222493469715118,
+      "learning_rate": 3.547094188376754e-06,
+      "loss": 0.0074,
+      "step": 4648
+    },
+    {
+      "epoch": 18.596,
+      "grad_norm": 0.11136149615049362,
+      "learning_rate": 3.537074148296594e-06,
+      "loss": 0.0041,
+      "step": 4649
+    },
+    {
+      "epoch": 18.6,
+      "grad_norm": 0.22324241697788239,
+      "learning_rate": 3.5270541082164326e-06,
+      "loss": 0.0079,
+      "step": 4650
+    },
+    {
+      "epoch": 18.604,
+      "grad_norm": 0.19871820509433746,
+      "learning_rate": 3.5170340681362726e-06,
+      "loss": 0.008,
+      "step": 4651
+    },
+    {
+      "epoch": 18.608,
+      "grad_norm": 0.14418363571166992,
+      "learning_rate": 3.5070140280561122e-06,
+      "loss": 0.0074,
+      "step": 4652
+    },
+    {
+      "epoch": 18.612,
+      "grad_norm": 0.10854968428611755,
+      "learning_rate": 3.4969939879759522e-06,
+      "loss": 0.0045,
+      "step": 4653
+    },
+    {
+      "epoch": 18.616,
+      "grad_norm": 0.1685395985841751,
+      "learning_rate": 3.486973947895792e-06,
+      "loss": 0.0074,
+      "step": 4654
+    },
+    {
+      "epoch": 18.62,
+      "grad_norm": 0.1491825133562088,
+      "learning_rate": 3.476953907815631e-06,
+      "loss": 0.0081,
+      "step": 4655
+    },
+    {
+      "epoch": 18.624,
+      "grad_norm": 0.10171408206224442,
+      "learning_rate": 3.466933867735471e-06,
+      "loss": 0.0042,
+      "step": 4656
+    },
+    {
+      "epoch": 18.628,
+      "grad_norm": 0.18516850471496582,
+      "learning_rate": 3.4569138276553106e-06,
+      "loss": 0.0076,
+      "step": 4657
+    },
+    {
+      "epoch": 18.632,
+      "grad_norm": 0.192645862698555,
+      "learning_rate": 3.4468937875751506e-06,
+      "loss": 0.0067,
+      "step": 4658
+    },
+    {
+      "epoch": 18.636,
+      "grad_norm": 0.17298291623592377,
+      "learning_rate": 3.43687374749499e-06,
+      "loss": 0.0089,
+      "step": 4659
+    },
+    {
+      "epoch": 18.64,
+      "grad_norm": 0.20988403260707855,
+      "learning_rate": 3.4268537074148294e-06,
+      "loss": 0.0089,
+      "step": 4660
+    },
+    {
+      "epoch": 18.644,
+      "grad_norm": 0.13690654933452606,
+      "learning_rate": 3.4168336673346694e-06,
+      "loss": 0.0072,
+      "step": 4661
+    },
+    {
+      "epoch": 18.648,
+      "grad_norm": 0.29314905405044556,
+      "learning_rate": 3.406813627254509e-06,
+      "loss": 0.0084,
+      "step": 4662
+    },
+    {
+      "epoch": 18.652,
+      "grad_norm": 0.18612822890281677,
+      "learning_rate": 3.396793587174349e-06,
+      "loss": 0.0072,
+      "step": 4663
+    },
+    {
+      "epoch": 18.656,
+      "grad_norm": 0.15141285955905914,
+      "learning_rate": 3.3867735470941886e-06,
+      "loss": 0.0066,
+      "step": 4664
+    },
+    {
+      "epoch": 18.66,
+      "grad_norm": 0.13616497814655304,
+      "learning_rate": 3.3767535070140286e-06,
+      "loss": 0.0069,
+      "step": 4665
+    },
+    {
+      "epoch": 18.664,
+      "grad_norm": 0.2509894073009491,
+      "learning_rate": 3.3667334669338677e-06,
+      "loss": 0.008,
+      "step": 4666
+    },
+    {
+      "epoch": 18.668,
+      "grad_norm": 0.2138143926858902,
+      "learning_rate": 3.3567134268537073e-06,
+      "loss": 0.0073,
+      "step": 4667
+    },
+    {
+      "epoch": 18.672,
+      "grad_norm": 0.16789166629314423,
+      "learning_rate": 3.3466933867735473e-06,
+      "loss": 0.0077,
+      "step": 4668
+    },
+    {
+      "epoch": 18.676,
+      "grad_norm": 0.16509471833705902,
+      "learning_rate": 3.336673346693387e-06,
+      "loss": 0.0073,
+      "step": 4669
+    },
+    {
+      "epoch": 18.68,
+      "grad_norm": 0.2040925770998001,
+      "learning_rate": 3.326653306613227e-06,
+      "loss": 0.0071,
+      "step": 4670
+    },
+    {
+      "epoch": 18.684,
+      "grad_norm": 0.17239294946193695,
+      "learning_rate": 3.316633266533066e-06,
+      "loss": 0.0077,
+      "step": 4671
+    },
+    {
+      "epoch": 18.688,
+      "grad_norm": 0.17981763184070587,
+      "learning_rate": 3.3066132264529057e-06,
+      "loss": 0.0073,
+      "step": 4672
+    },
+    {
+      "epoch": 18.692,
+      "grad_norm": 0.15703114867210388,
+      "learning_rate": 3.2965931863727457e-06,
+      "loss": 0.008,
+      "step": 4673
+    },
+    {
+      "epoch": 18.696,
+      "grad_norm": 0.17290978133678436,
+      "learning_rate": 3.2865731462925853e-06,
+      "loss": 0.0085,
+      "step": 4674
+    },
+    {
+      "epoch": 18.7,
+      "grad_norm": 0.18817895650863647,
+      "learning_rate": 3.2765531062124253e-06,
+      "loss": 0.0083,
+      "step": 4675
+    },
+    {
+      "epoch": 18.704,
+      "grad_norm": 0.1855512112379074,
+      "learning_rate": 3.2665330661322645e-06,
+      "loss": 0.0075,
+      "step": 4676
+    },
+    {
+      "epoch": 18.708,
+      "grad_norm": 0.2292889952659607,
+      "learning_rate": 3.256513026052104e-06,
+      "loss": 0.0081,
+      "step": 4677
+    },
+    {
+      "epoch": 18.712,
+      "grad_norm": 0.15478086471557617,
+      "learning_rate": 3.246492985971944e-06,
+      "loss": 0.0072,
+      "step": 4678
+    },
+    {
+      "epoch": 18.716,
+      "grad_norm": 0.16357959806919098,
+      "learning_rate": 3.2364729458917837e-06,
+      "loss": 0.0077,
+      "step": 4679
+    },
+    {
+      "epoch": 18.72,
+      "grad_norm": 0.17677243053913116,
+      "learning_rate": 3.2264529058116237e-06,
+      "loss": 0.0072,
+      "step": 4680
+    },
+    {
+      "epoch": 18.724,
+      "grad_norm": 0.23427382111549377,
+      "learning_rate": 3.2164328657314633e-06,
+      "loss": 0.0089,
+      "step": 4681
+    },
+    {
+      "epoch": 18.728,
+      "grad_norm": 0.1855786293745041,
+      "learning_rate": 3.2064128256513024e-06,
+      "loss": 0.0069,
+      "step": 4682
+    },
+    {
+      "epoch": 18.732,
+      "grad_norm": 0.17159710824489594,
+      "learning_rate": 3.1963927855711424e-06,
+      "loss": 0.0071,
+      "step": 4683
+    },
+    {
+      "epoch": 18.736,
+      "grad_norm": 0.18755950033664703,
+      "learning_rate": 3.186372745490982e-06,
+      "loss": 0.0075,
+      "step": 4684
+    },
+    {
+      "epoch": 18.74,
+      "grad_norm": 0.16439513862133026,
+      "learning_rate": 3.176352705410822e-06,
+      "loss": 0.0072,
+      "step": 4685
+    },
+    {
+      "epoch": 18.744,
+      "grad_norm": 0.17373932898044586,
+      "learning_rate": 3.1663326653306616e-06,
+      "loss": 0.0072,
+      "step": 4686
+    },
+    {
+      "epoch": 18.748,
+      "grad_norm": 0.19658909738063812,
+      "learning_rate": 3.156312625250501e-06,
+      "loss": 0.008,
+      "step": 4687
+    },
+    {
+      "epoch": 18.752,
+      "grad_norm": 0.1813708245754242,
+      "learning_rate": 3.146292585170341e-06,
+      "loss": 0.0068,
+      "step": 4688
+    },
+    {
+      "epoch": 18.756,
+      "grad_norm": 0.15275795757770538,
+      "learning_rate": 3.1362725450901804e-06,
+      "loss": 0.0066,
+      "step": 4689
+    },
+    {
+      "epoch": 18.76,
+      "grad_norm": 0.20497362315654755,
+      "learning_rate": 3.1262525050100204e-06,
+      "loss": 0.0081,
+      "step": 4690
+    },
+    {
+      "epoch": 18.764,
+      "grad_norm": 0.1795947104692459,
+      "learning_rate": 3.1162324649298596e-06,
+      "loss": 0.0077,
+      "step": 4691
+    },
+    {
+      "epoch": 18.768,
+      "grad_norm": 0.14779050648212433,
+      "learning_rate": 3.1062124248496996e-06,
+      "loss": 0.0072,
+      "step": 4692
+    },
+    {
+      "epoch": 18.772,
+      "grad_norm": 0.15518996119499207,
+      "learning_rate": 3.096192384769539e-06,
+      "loss": 0.0048,
+      "step": 4693
+    },
+    {
+      "epoch": 18.776,
+      "grad_norm": 0.1515381932258606,
+      "learning_rate": 3.0861723446893788e-06,
+      "loss": 0.0068,
+      "step": 4694
+    },
+    {
+      "epoch": 18.78,
+      "grad_norm": 0.2062983363866806,
+      "learning_rate": 3.0761523046092188e-06,
+      "loss": 0.0085,
+      "step": 4695
+    },
+    {
+      "epoch": 18.784,
+      "grad_norm": 0.1976705640554428,
+      "learning_rate": 3.0661322645290584e-06,
+      "loss": 0.0072,
+      "step": 4696
+    },
+    {
+      "epoch": 18.788,
+      "grad_norm": 0.13696104288101196,
+      "learning_rate": 3.056112224448898e-06,
+      "loss": 0.0065,
+      "step": 4697
+    },
+    {
+      "epoch": 18.792,
+      "grad_norm": 0.19733761250972748,
+      "learning_rate": 3.0460921843687375e-06,
+      "loss": 0.0072,
+      "step": 4698
+    },
+    {
+      "epoch": 18.796,
+      "grad_norm": 0.17642058432102203,
+      "learning_rate": 3.036072144288577e-06,
+      "loss": 0.0078,
+      "step": 4699
+    },
+    {
+      "epoch": 18.8,
+      "grad_norm": 0.17116330564022064,
+      "learning_rate": 3.026052104208417e-06,
+      "loss": 0.0069,
+      "step": 4700
+    },
+    {
+      "epoch": 18.804,
+      "grad_norm": 0.2022354155778885,
+      "learning_rate": 3.0160320641282567e-06,
+      "loss": 0.0079,
+      "step": 4701
+    },
+    {
+      "epoch": 18.808,
+      "grad_norm": 0.147251158952713,
+      "learning_rate": 3.0060120240480963e-06,
+      "loss": 0.0062,
+      "step": 4702
+    },
+    {
+      "epoch": 18.812,
+      "grad_norm": 0.1599157601594925,
+      "learning_rate": 2.995991983967936e-06,
+      "loss": 0.008,
+      "step": 4703
+    },
+    {
+      "epoch": 18.816,
+      "grad_norm": 0.12859754264354706,
+      "learning_rate": 2.985971943887776e-06,
+      "loss": 0.0073,
+      "step": 4704
+    },
+    {
+      "epoch": 18.82,
+      "grad_norm": 0.2207632064819336,
+      "learning_rate": 2.975951903807615e-06,
+      "loss": 0.0078,
+      "step": 4705
+    },
+    {
+      "epoch": 18.824,
+      "grad_norm": 0.16606071591377258,
+      "learning_rate": 2.965931863727455e-06,
+      "loss": 0.0072,
+      "step": 4706
+    },
+    {
+      "epoch": 18.828,
+      "grad_norm": 0.1599171757698059,
+      "learning_rate": 2.9559118236472947e-06,
+      "loss": 0.008,
+      "step": 4707
+    },
+    {
+      "epoch": 18.832,
+      "grad_norm": 0.14735926687717438,
+      "learning_rate": 2.9458917835671343e-06,
+      "loss": 0.0071,
+      "step": 4708
+    },
+    {
+      "epoch": 18.836,
+      "grad_norm": 0.25691819190979004,
+      "learning_rate": 2.9358717434869743e-06,
+      "loss": 0.0079,
+      "step": 4709
+    },
+    {
+      "epoch": 18.84,
+      "grad_norm": 0.19598759710788727,
+      "learning_rate": 2.9258517034068135e-06,
+      "loss": 0.0076,
+      "step": 4710
+    },
+    {
+      "epoch": 18.844,
+      "grad_norm": 0.19333742558956146,
+      "learning_rate": 2.9158316633266535e-06,
+      "loss": 0.0076,
+      "step": 4711
+    },
+    {
+      "epoch": 18.848,
+      "grad_norm": 0.1969856470823288,
+      "learning_rate": 2.905811623246493e-06,
+      "loss": 0.0075,
+      "step": 4712
+    },
+    {
+      "epoch": 18.852,
+      "grad_norm": 0.1487724334001541,
+      "learning_rate": 2.8957915831663326e-06,
+      "loss": 0.0067,
+      "step": 4713
+    },
+    {
+      "epoch": 18.856,
+      "grad_norm": 0.1891922503709793,
+      "learning_rate": 2.8857715430861727e-06,
+      "loss": 0.007,
+      "step": 4714
+    },
+    {
+      "epoch": 18.86,
+      "grad_norm": 0.1953631490468979,
+      "learning_rate": 2.8757515030060122e-06,
+      "loss": 0.0077,
+      "step": 4715
+    },
+    {
+      "epoch": 18.864,
+      "grad_norm": 0.17311617732048035,
+      "learning_rate": 2.865731462925852e-06,
+      "loss": 0.0072,
+      "step": 4716
+    },
+    {
+      "epoch": 18.868,
+      "grad_norm": 0.10396530479192734,
+      "learning_rate": 2.8557114228456914e-06,
+      "loss": 0.0041,
+      "step": 4717
+    },
+    {
+      "epoch": 18.872,
+      "grad_norm": 0.1536823958158493,
+      "learning_rate": 2.845691382765531e-06,
+      "loss": 0.008,
+      "step": 4718
+    },
+    {
+      "epoch": 18.876,
+      "grad_norm": 0.1858183890581131,
+      "learning_rate": 2.835671342685371e-06,
+      "loss": 0.0076,
+      "step": 4719
+    },
+    {
+      "epoch": 18.88,
+      "grad_norm": 0.13414987921714783,
+      "learning_rate": 2.8256513026052106e-06,
+      "loss": 0.0068,
+      "step": 4720
+    },
+    {
+      "epoch": 18.884,
+      "grad_norm": 0.11735019087791443,
+      "learning_rate": 2.81563126252505e-06,
+      "loss": 0.0059,
+      "step": 4721
+    },
+    {
+      "epoch": 18.888,
+      "grad_norm": 0.17751845717430115,
+      "learning_rate": 2.80561122244489e-06,
+      "loss": 0.0073,
+      "step": 4722
+    },
+    {
+      "epoch": 18.892,
+      "grad_norm": 0.22168204188346863,
+      "learning_rate": 2.79559118236473e-06,
+      "loss": 0.0083,
+      "step": 4723
+    },
+    {
+      "epoch": 18.896,
+      "grad_norm": 0.18267209827899933,
+      "learning_rate": 2.7855711422845694e-06,
+      "loss": 0.0077,
+      "step": 4724
+    },
+    {
+      "epoch": 18.9,
+      "grad_norm": 0.1117076501250267,
+      "learning_rate": 2.775551102204409e-06,
+      "loss": 0.0069,
+      "step": 4725
+    },
+    {
+      "epoch": 18.904,
+      "grad_norm": 0.12114907801151276,
+      "learning_rate": 2.7655310621242486e-06,
+      "loss": 0.007,
+      "step": 4726
+    },
+    {
+      "epoch": 18.908,
+      "grad_norm": 0.2513674199581146,
+      "learning_rate": 2.755511022044088e-06,
+      "loss": 0.0084,
+      "step": 4727
+    },
+    {
+      "epoch": 18.912,
+      "grad_norm": 0.1959051787853241,
+      "learning_rate": 2.745490981963928e-06,
+      "loss": 0.0073,
+      "step": 4728
+    },
+    {
+      "epoch": 18.916,
+      "grad_norm": 0.2004287838935852,
+      "learning_rate": 2.7354709418837678e-06,
+      "loss": 0.0077,
+      "step": 4729
+    },
+    {
+      "epoch": 18.92,
+      "grad_norm": 0.17070572078227997,
+      "learning_rate": 2.7254509018036073e-06,
+      "loss": 0.0068,
+      "step": 4730
+    },
+    {
+      "epoch": 18.924,
+      "grad_norm": 0.14630602300167084,
+      "learning_rate": 2.7154308617234474e-06,
+      "loss": 0.0071,
+      "step": 4731
+    },
+    {
+      "epoch": 18.928,
+      "grad_norm": 0.16959349811077118,
+      "learning_rate": 2.7054108216432865e-06,
+      "loss": 0.0083,
+      "step": 4732
+    },
+    {
+      "epoch": 18.932,
+      "grad_norm": 0.15642155706882477,
+      "learning_rate": 2.6953907815631265e-06,
+      "loss": 0.0075,
+      "step": 4733
+    },
+    {
+      "epoch": 18.936,
+      "grad_norm": 0.17925404012203217,
+      "learning_rate": 2.6853707414829657e-06,
+      "loss": 0.0065,
+      "step": 4734
+    },
+    {
+      "epoch": 18.94,
+      "grad_norm": 0.19508227705955505,
+      "learning_rate": 2.6753507014028057e-06,
+      "loss": 0.008,
+      "step": 4735
+    },
+    {
+      "epoch": 18.944,
+      "grad_norm": 0.16290459036827087,
+      "learning_rate": 2.6653306613226457e-06,
+      "loss": 0.005,
+      "step": 4736
+    },
+    {
+      "epoch": 18.948,
+      "grad_norm": 0.15462210774421692,
+      "learning_rate": 2.655310621242485e-06,
+      "loss": 0.0078,
+      "step": 4737
+    },
+    {
+      "epoch": 18.951999999999998,
+      "grad_norm": 0.11808783560991287,
+      "learning_rate": 2.645290581162325e-06,
+      "loss": 0.0066,
+      "step": 4738
+    },
+    {
+      "epoch": 18.956,
+      "grad_norm": 0.20686569809913635,
+      "learning_rate": 2.6352705410821645e-06,
+      "loss": 0.0078,
+      "step": 4739
+    },
+    {
+      "epoch": 18.96,
+      "grad_norm": 0.17440447211265564,
+      "learning_rate": 2.625250501002004e-06,
+      "loss": 0.0081,
+      "step": 4740
+    },
+    {
+      "epoch": 18.964,
+      "grad_norm": 0.337189644575119,
+      "learning_rate": 2.6152304609218437e-06,
+      "loss": 0.0079,
+      "step": 4741
+    },
+    {
+      "epoch": 18.968,
+      "grad_norm": 0.18942351639270782,
+      "learning_rate": 2.6052104208416833e-06,
+      "loss": 0.0091,
+      "step": 4742
+    },
+    {
+      "epoch": 18.972,
+      "grad_norm": 0.10684970766305923,
+      "learning_rate": 2.5951903807615233e-06,
+      "loss": 0.0046,
+      "step": 4743
+    },
+    {
+      "epoch": 18.976,
+      "grad_norm": 0.22341260313987732,
+      "learning_rate": 2.585170340681363e-06,
+      "loss": 0.0077,
+      "step": 4744
+    },
+    {
+      "epoch": 18.98,
+      "grad_norm": 0.24727453291416168,
+      "learning_rate": 2.5751503006012024e-06,
+      "loss": 0.0088,
+      "step": 4745
+    },
+    {
+      "epoch": 18.984,
+      "grad_norm": 0.19121108949184418,
+      "learning_rate": 2.565130260521042e-06,
+      "loss": 0.0073,
+      "step": 4746
+    },
+    {
+      "epoch": 18.988,
+      "grad_norm": 0.2110673189163208,
+      "learning_rate": 2.555110220440882e-06,
+      "loss": 0.0077,
+      "step": 4747
+    },
+    {
+      "epoch": 18.992,
+      "grad_norm": 0.13829149305820465,
+      "learning_rate": 2.5450901803607216e-06,
+      "loss": 0.0069,
+      "step": 4748
+    },
+    {
+      "epoch": 18.996,
+      "grad_norm": 0.1673704981803894,
+      "learning_rate": 2.5350701402805612e-06,
+      "loss": 0.0073,
+      "step": 4749
+    },
+    {
+      "epoch": 19.0,
+      "grad_norm": 0.18628820776939392,
+      "learning_rate": 2.5250501002004012e-06,
+      "loss": 0.0076,
+      "step": 4750
+    },
+    {
+      "epoch": 19.004,
+      "grad_norm": 0.17194223403930664,
+      "learning_rate": 2.5150300601202404e-06,
+      "loss": 0.0065,
+      "step": 4751
+    },
+    {
+      "epoch": 19.008,
+      "grad_norm": 0.14391052722930908,
+      "learning_rate": 2.5050100200400804e-06,
+      "loss": 0.0066,
+      "step": 4752
+    },
+    {
+      "epoch": 19.012,
+      "grad_norm": 0.1288471221923828,
+      "learning_rate": 2.49498997995992e-06,
+      "loss": 0.0064,
+      "step": 4753
+    },
+    {
+      "epoch": 19.016,
+      "grad_norm": 0.158019557595253,
+      "learning_rate": 2.4849699398797596e-06,
+      "loss": 0.007,
+      "step": 4754
+    },
+    {
+      "epoch": 19.02,
+      "grad_norm": 0.19652923941612244,
+      "learning_rate": 2.4749498997995996e-06,
+      "loss": 0.0078,
+      "step": 4755
+    },
+    {
+      "epoch": 19.024,
+      "grad_norm": 0.1863459348678589,
+      "learning_rate": 2.4649298597194388e-06,
+      "loss": 0.0073,
+      "step": 4756
+    },
+    {
+      "epoch": 19.028,
+      "grad_norm": 0.16609062254428864,
+      "learning_rate": 2.4549098196392788e-06,
+      "loss": 0.0073,
+      "step": 4757
+    },
+    {
+      "epoch": 19.032,
+      "grad_norm": 0.1344534158706665,
+      "learning_rate": 2.4448897795591184e-06,
+      "loss": 0.0062,
+      "step": 4758
+    },
+    {
+      "epoch": 19.036,
+      "grad_norm": 0.11711183190345764,
+      "learning_rate": 2.434869739478958e-06,
+      "loss": 0.0058,
+      "step": 4759
+    },
+    {
+      "epoch": 19.04,
+      "grad_norm": 0.12532389163970947,
+      "learning_rate": 2.424849699398798e-06,
+      "loss": 0.0068,
+      "step": 4760
+    },
+    {
+      "epoch": 19.044,
+      "grad_norm": 0.14596407115459442,
+      "learning_rate": 2.414829659318637e-06,
+      "loss": 0.007,
+      "step": 4761
+    },
+    {
+      "epoch": 19.048,
+      "grad_norm": 0.13948693871498108,
+      "learning_rate": 2.404809619238477e-06,
+      "loss": 0.0067,
+      "step": 4762
+    },
+    {
+      "epoch": 19.052,
+      "grad_norm": 0.15014758706092834,
+      "learning_rate": 2.3947895791583167e-06,
+      "loss": 0.0067,
+      "step": 4763
+    },
+    {
+      "epoch": 19.056,
+      "grad_norm": 0.11399639397859573,
+      "learning_rate": 2.3847695390781563e-06,
+      "loss": 0.0039,
+      "step": 4764
+    },
+    {
+      "epoch": 19.06,
+      "grad_norm": 0.27429330348968506,
+      "learning_rate": 2.374749498997996e-06,
+      "loss": 0.0077,
+      "step": 4765
+    },
+    {
+      "epoch": 19.064,
+      "grad_norm": 0.1491885483264923,
+      "learning_rate": 2.364729458917836e-06,
+      "loss": 0.0069,
+      "step": 4766
+    },
+    {
+      "epoch": 19.068,
+      "grad_norm": 0.1384284943342209,
+      "learning_rate": 2.3547094188376755e-06,
+      "loss": 0.0063,
+      "step": 4767
+    },
+    {
+      "epoch": 19.072,
+      "grad_norm": 0.15902647376060486,
+      "learning_rate": 2.344689378757515e-06,
+      "loss": 0.0076,
+      "step": 4768
+    },
+    {
+      "epoch": 19.076,
+      "grad_norm": 0.17550063133239746,
+      "learning_rate": 2.3346693386773547e-06,
+      "loss": 0.0075,
+      "step": 4769
+    },
+    {
+      "epoch": 19.08,
+      "grad_norm": 0.11770374327898026,
+      "learning_rate": 2.3246492985971943e-06,
+      "loss": 0.0065,
+      "step": 4770
+    },
+    {
+      "epoch": 19.084,
+      "grad_norm": 0.13193221390247345,
+      "learning_rate": 2.3146292585170343e-06,
+      "loss": 0.0064,
+      "step": 4771
+    },
+    {
+      "epoch": 19.088,
+      "grad_norm": 0.20296931266784668,
+      "learning_rate": 2.304609218436874e-06,
+      "loss": 0.0082,
+      "step": 4772
+    },
+    {
+      "epoch": 19.092,
+      "grad_norm": 0.17178785800933838,
+      "learning_rate": 2.2945891783567135e-06,
+      "loss": 0.007,
+      "step": 4773
+    },
+    {
+      "epoch": 19.096,
+      "grad_norm": 0.18985581398010254,
+      "learning_rate": 2.2845691382765535e-06,
+      "loss": 0.0064,
+      "step": 4774
+    },
+    {
+      "epoch": 19.1,
+      "grad_norm": 0.180537611246109,
+      "learning_rate": 2.2745490981963926e-06,
+      "loss": 0.0069,
+      "step": 4775
+    },
+    {
+      "epoch": 19.104,
+      "grad_norm": 0.13604559004306793,
+      "learning_rate": 2.2645290581162327e-06,
+      "loss": 0.0045,
+      "step": 4776
+    },
+    {
+      "epoch": 19.108,
+      "grad_norm": 0.13354989886283875,
+      "learning_rate": 2.2545090180360722e-06,
+      "loss": 0.0061,
+      "step": 4777
+    },
+    {
+      "epoch": 19.112,
+      "grad_norm": 0.1494477540254593,
+      "learning_rate": 2.244488977955912e-06,
+      "loss": 0.0069,
+      "step": 4778
+    },
+    {
+      "epoch": 19.116,
+      "grad_norm": 0.22611655294895172,
+      "learning_rate": 2.234468937875752e-06,
+      "loss": 0.008,
+      "step": 4779
+    },
+    {
+      "epoch": 19.12,
+      "grad_norm": 0.11629709601402283,
+      "learning_rate": 2.224448897795591e-06,
+      "loss": 0.0045,
+      "step": 4780
+    },
+    {
+      "epoch": 19.124,
+      "grad_norm": 0.16238947212696075,
+      "learning_rate": 2.214428857715431e-06,
+      "loss": 0.007,
+      "step": 4781
+    },
+    {
+      "epoch": 19.128,
+      "grad_norm": 0.15134558081626892,
+      "learning_rate": 2.2044088176352706e-06,
+      "loss": 0.0072,
+      "step": 4782
+    },
+    {
+      "epoch": 19.132,
+      "grad_norm": 0.1289103478193283,
+      "learning_rate": 2.19438877755511e-06,
+      "loss": 0.0056,
+      "step": 4783
+    },
+    {
+      "epoch": 19.136,
+      "grad_norm": 0.13187222182750702,
+      "learning_rate": 2.18436873747495e-06,
+      "loss": 0.0062,
+      "step": 4784
+    },
+    {
+      "epoch": 19.14,
+      "grad_norm": 0.16803032159805298,
+      "learning_rate": 2.17434869739479e-06,
+      "loss": 0.0072,
+      "step": 4785
+    },
+    {
+      "epoch": 19.144,
+      "grad_norm": 0.16196125745773315,
+      "learning_rate": 2.1643286573146294e-06,
+      "loss": 0.0071,
+      "step": 4786
+    },
+    {
+      "epoch": 19.148,
+      "grad_norm": 0.14935888350009918,
+      "learning_rate": 2.154308617234469e-06,
+      "loss": 0.007,
+      "step": 4787
+    },
+    {
+      "epoch": 19.152,
+      "grad_norm": 0.17535895109176636,
+      "learning_rate": 2.1442885771543086e-06,
+      "loss": 0.007,
+      "step": 4788
+    },
+    {
+      "epoch": 19.156,
+      "grad_norm": 0.12928767502307892,
+      "learning_rate": 2.1342685370741486e-06,
+      "loss": 0.0065,
+      "step": 4789
+    },
+    {
+      "epoch": 19.16,
+      "grad_norm": 0.1758950650691986,
+      "learning_rate": 2.124248496993988e-06,
+      "loss": 0.0064,
+      "step": 4790
+    },
+    {
+      "epoch": 19.164,
+      "grad_norm": 0.1001003235578537,
+      "learning_rate": 2.1142284569138278e-06,
+      "loss": 0.0058,
+      "step": 4791
+    },
+    {
+      "epoch": 19.168,
+      "grad_norm": 0.17448782920837402,
+      "learning_rate": 2.1042084168336673e-06,
+      "loss": 0.0064,
+      "step": 4792
+    },
+    {
+      "epoch": 19.172,
+      "grad_norm": 0.16154441237449646,
+      "learning_rate": 2.0941883767535074e-06,
+      "loss": 0.0072,
+      "step": 4793
+    },
+    {
+      "epoch": 19.176,
+      "grad_norm": 0.16718144714832306,
+      "learning_rate": 2.0841683366733465e-06,
+      "loss": 0.007,
+      "step": 4794
+    },
+    {
+      "epoch": 19.18,
+      "grad_norm": 0.16893278062343597,
+      "learning_rate": 2.0741482965931865e-06,
+      "loss": 0.0077,
+      "step": 4795
+    },
+    {
+      "epoch": 19.184,
+      "grad_norm": 0.14809642732143402,
+      "learning_rate": 2.064128256513026e-06,
+      "loss": 0.0062,
+      "step": 4796
+    },
+    {
+      "epoch": 19.188,
+      "grad_norm": 0.16328303515911102,
+      "learning_rate": 2.0541082164328657e-06,
+      "loss": 0.0067,
+      "step": 4797
+    },
+    {
+      "epoch": 19.192,
+      "grad_norm": 0.17468783259391785,
+      "learning_rate": 2.0440881763527057e-06,
+      "loss": 0.0071,
+      "step": 4798
+    },
+    {
+      "epoch": 19.196,
+      "grad_norm": 0.1770590990781784,
+      "learning_rate": 2.034068136272545e-06,
+      "loss": 0.0073,
+      "step": 4799
+    },
+    {
+      "epoch": 19.2,
+      "grad_norm": 0.20840556919574738,
+      "learning_rate": 2.024048096192385e-06,
+      "loss": 0.0077,
+      "step": 4800
+    },
+    {
+      "epoch": 19.204,
+      "grad_norm": 0.16028790175914764,
+      "learning_rate": 2.0140280561122245e-06,
+      "loss": 0.0075,
+      "step": 4801
+    },
+    {
+      "epoch": 19.208,
+      "grad_norm": 0.10381535440683365,
+      "learning_rate": 2.004008016032064e-06,
+      "loss": 0.0042,
+      "step": 4802
+    },
+    {
+      "epoch": 19.212,
+      "grad_norm": 0.1724594235420227,
+      "learning_rate": 1.993987975951904e-06,
+      "loss": 0.0079,
+      "step": 4803
+    },
+    {
+      "epoch": 19.216,
+      "grad_norm": 0.17494012415409088,
+      "learning_rate": 1.9839679358717433e-06,
+      "loss": 0.0069,
+      "step": 4804
+    },
+    {
+      "epoch": 19.22,
+      "grad_norm": 0.16234153509140015,
+      "learning_rate": 1.9739478957915833e-06,
+      "loss": 0.0069,
+      "step": 4805
+    },
+    {
+      "epoch": 19.224,
+      "grad_norm": 0.15438240766525269,
+      "learning_rate": 1.963927855711423e-06,
+      "loss": 0.0073,
+      "step": 4806
+    },
+    {
+      "epoch": 19.228,
+      "grad_norm": 0.16121751070022583,
+      "learning_rate": 1.9539078156312624e-06,
+      "loss": 0.0074,
+      "step": 4807
+    },
+    {
+      "epoch": 19.232,
+      "grad_norm": 0.15929311513900757,
+      "learning_rate": 1.9438877755511025e-06,
+      "loss": 0.0073,
+      "step": 4808
+    },
+    {
+      "epoch": 19.236,
+      "grad_norm": 0.1370072364807129,
+      "learning_rate": 1.933867735470942e-06,
+      "loss": 0.007,
+      "step": 4809
+    },
+    {
+      "epoch": 19.24,
+      "grad_norm": 0.15893737971782684,
+      "learning_rate": 1.9238476953907816e-06,
+      "loss": 0.0063,
+      "step": 4810
+    },
+    {
+      "epoch": 19.244,
+      "grad_norm": 0.19493524730205536,
+      "learning_rate": 1.9138276553106212e-06,
+      "loss": 0.0072,
+      "step": 4811
+    },
+    {
+      "epoch": 19.248,
+      "grad_norm": 0.1659090220928192,
+      "learning_rate": 1.9038076152304612e-06,
+      "loss": 0.0069,
+      "step": 4812
+    },
+    {
+      "epoch": 19.252,
+      "grad_norm": 0.14420440793037415,
+      "learning_rate": 1.8937875751503006e-06,
+      "loss": 0.0065,
+      "step": 4813
+    },
+    {
+      "epoch": 19.256,
+      "grad_norm": 0.16298946738243103,
+      "learning_rate": 1.8837675350701404e-06,
+      "loss": 0.0065,
+      "step": 4814
+    },
+    {
+      "epoch": 19.26,
+      "grad_norm": 0.13460707664489746,
+      "learning_rate": 1.87374749498998e-06,
+      "loss": 0.0069,
+      "step": 4815
+    },
+    {
+      "epoch": 19.264,
+      "grad_norm": 0.17782555520534515,
+      "learning_rate": 1.8637274549098198e-06,
+      "loss": 0.0078,
+      "step": 4816
+    },
+    {
+      "epoch": 19.268,
+      "grad_norm": 0.21626275777816772,
+      "learning_rate": 1.8537074148296596e-06,
+      "loss": 0.0083,
+      "step": 4817
+    },
+    {
+      "epoch": 19.272,
+      "grad_norm": 0.07763174921274185,
+      "learning_rate": 1.843687374749499e-06,
+      "loss": 0.0036,
+      "step": 4818
+    },
+    {
+      "epoch": 19.276,
+      "grad_norm": 0.19351311028003693,
+      "learning_rate": 1.8336673346693388e-06,
+      "loss": 0.0074,
+      "step": 4819
+    },
+    {
+      "epoch": 19.28,
+      "grad_norm": 0.18536561727523804,
+      "learning_rate": 1.8236472945891786e-06,
+      "loss": 0.0074,
+      "step": 4820
+    },
+    {
+      "epoch": 19.284,
+      "grad_norm": 0.15970304608345032,
+      "learning_rate": 1.8136272545090182e-06,
+      "loss": 0.0066,
+      "step": 4821
+    },
+    {
+      "epoch": 19.288,
+      "grad_norm": 0.14813347160816193,
+      "learning_rate": 1.803607214428858e-06,
+      "loss": 0.0061,
+      "step": 4822
+    },
+    {
+      "epoch": 19.292,
+      "grad_norm": 0.1950247585773468,
+      "learning_rate": 1.7935871743486973e-06,
+      "loss": 0.0076,
+      "step": 4823
+    },
+    {
+      "epoch": 19.296,
+      "grad_norm": 0.19264726340770721,
+      "learning_rate": 1.7835671342685371e-06,
+      "loss": 0.0069,
+      "step": 4824
+    },
+    {
+      "epoch": 19.3,
+      "grad_norm": 0.19691282510757446,
+      "learning_rate": 1.773547094188377e-06,
+      "loss": 0.0067,
+      "step": 4825
+    },
+    {
+      "epoch": 19.304,
+      "grad_norm": 0.1652364283800125,
+      "learning_rate": 1.7635270541082163e-06,
+      "loss": 0.0065,
+      "step": 4826
+    },
+    {
+      "epoch": 19.308,
+      "grad_norm": 0.18058018386363983,
+      "learning_rate": 1.7535070140280561e-06,
+      "loss": 0.008,
+      "step": 4827
+    },
+    {
+      "epoch": 19.312,
+      "grad_norm": 0.1553662121295929,
+      "learning_rate": 1.743486973947896e-06,
+      "loss": 0.0067,
+      "step": 4828
+    },
+    {
+      "epoch": 19.316,
+      "grad_norm": 0.12730355560779572,
+      "learning_rate": 1.7334669338677355e-06,
+      "loss": 0.0068,
+      "step": 4829
+    },
+    {
+      "epoch": 19.32,
+      "grad_norm": 0.1950482428073883,
+      "learning_rate": 1.7234468937875753e-06,
+      "loss": 0.008,
+      "step": 4830
+    },
+    {
+      "epoch": 19.324,
+      "grad_norm": 0.17624934017658234,
+      "learning_rate": 1.7134268537074147e-06,
+      "loss": 0.0076,
+      "step": 4831
+    },
+    {
+      "epoch": 19.328,
+      "grad_norm": 0.1288178265094757,
+      "learning_rate": 1.7034068136272545e-06,
+      "loss": 0.0068,
+      "step": 4832
+    },
+    {
+      "epoch": 19.332,
+      "grad_norm": 0.1886533796787262,
+      "learning_rate": 1.6933867735470943e-06,
+      "loss": 0.0063,
+      "step": 4833
+    },
+    {
+      "epoch": 19.336,
+      "grad_norm": 0.16468536853790283,
+      "learning_rate": 1.6833667334669339e-06,
+      "loss": 0.0069,
+      "step": 4834
+    },
+    {
+      "epoch": 19.34,
+      "grad_norm": 0.17551806569099426,
+      "learning_rate": 1.6733466933867737e-06,
+      "loss": 0.0077,
+      "step": 4835
+    },
+    {
+      "epoch": 19.344,
+      "grad_norm": 0.1710132509469986,
+      "learning_rate": 1.6633266533066135e-06,
+      "loss": 0.0076,
+      "step": 4836
+    },
+    {
+      "epoch": 19.348,
+      "grad_norm": 0.17541347444057465,
+      "learning_rate": 1.6533066132264529e-06,
+      "loss": 0.0068,
+      "step": 4837
+    },
+    {
+      "epoch": 19.352,
+      "grad_norm": 0.1492847353219986,
+      "learning_rate": 1.6432865731462927e-06,
+      "loss": 0.0067,
+      "step": 4838
+    },
+    {
+      "epoch": 19.356,
+      "grad_norm": 0.17386400699615479,
+      "learning_rate": 1.6332665330661322e-06,
+      "loss": 0.0074,
+      "step": 4839
+    },
+    {
+      "epoch": 19.36,
+      "grad_norm": 0.15245819091796875,
+      "learning_rate": 1.623246492985972e-06,
+      "loss": 0.0061,
+      "step": 4840
+    },
+    {
+      "epoch": 19.364,
+      "grad_norm": 0.18433190882205963,
+      "learning_rate": 1.6132264529058118e-06,
+      "loss": 0.007,
+      "step": 4841
+    },
+    {
+      "epoch": 19.368,
+      "grad_norm": 0.16951580345630646,
+      "learning_rate": 1.6032064128256512e-06,
+      "loss": 0.0069,
+      "step": 4842
+    },
+    {
+      "epoch": 19.372,
+      "grad_norm": 0.1776171624660492,
+      "learning_rate": 1.593186372745491e-06,
+      "loss": 0.0068,
+      "step": 4843
+    },
+    {
+      "epoch": 19.376,
+      "grad_norm": 0.2017744779586792,
+      "learning_rate": 1.5831663326653308e-06,
+      "loss": 0.0069,
+      "step": 4844
+    },
+    {
+      "epoch": 19.38,
+      "grad_norm": 0.22619450092315674,
+      "learning_rate": 1.5731462925851704e-06,
+      "loss": 0.0071,
+      "step": 4845
+    },
+    {
+      "epoch": 19.384,
+      "grad_norm": 0.1772204488515854,
+      "learning_rate": 1.5631262525050102e-06,
+      "loss": 0.0069,
+      "step": 4846
+    },
+    {
+      "epoch": 19.388,
+      "grad_norm": 0.2213824987411499,
+      "learning_rate": 1.5531062124248498e-06,
+      "loss": 0.0075,
+      "step": 4847
+    },
+    {
+      "epoch": 19.392,
+      "grad_norm": 0.21730723977088928,
+      "learning_rate": 1.5430861723446894e-06,
+      "loss": 0.0078,
+      "step": 4848
+    },
+    {
+      "epoch": 19.396,
+      "grad_norm": 0.13313718140125275,
+      "learning_rate": 1.5330661322645292e-06,
+      "loss": 0.0063,
+      "step": 4849
+    },
+    {
+      "epoch": 19.4,
+      "grad_norm": 0.23904912173748016,
+      "learning_rate": 1.5230460921843688e-06,
+      "loss": 0.0074,
+      "step": 4850
+    },
+    {
+      "epoch": 19.404,
+      "grad_norm": 0.1711878627538681,
+      "learning_rate": 1.5130260521042086e-06,
+      "loss": 0.0072,
+      "step": 4851
+    },
+    {
+      "epoch": 19.408,
+      "grad_norm": 0.19314146041870117,
+      "learning_rate": 1.5030060120240482e-06,
+      "loss": 0.0066,
+      "step": 4852
+    },
+    {
+      "epoch": 19.412,
+      "grad_norm": 0.16302086412906647,
+      "learning_rate": 1.492985971943888e-06,
+      "loss": 0.0078,
+      "step": 4853
+    },
+    {
+      "epoch": 19.416,
+      "grad_norm": 0.15653206408023834,
+      "learning_rate": 1.4829659318637276e-06,
+      "loss": 0.0067,
+      "step": 4854
+    },
+    {
+      "epoch": 19.42,
+      "grad_norm": 0.12465237826108932,
+      "learning_rate": 1.4729458917835671e-06,
+      "loss": 0.0064,
+      "step": 4855
+    },
+    {
+      "epoch": 19.424,
+      "grad_norm": 0.19219818711280823,
+      "learning_rate": 1.4629258517034067e-06,
+      "loss": 0.0075,
+      "step": 4856
+    },
+    {
+      "epoch": 19.428,
+      "grad_norm": 0.16947104036808014,
+      "learning_rate": 1.4529058116232465e-06,
+      "loss": 0.0056,
+      "step": 4857
+    },
+    {
+      "epoch": 19.432,
+      "grad_norm": 0.16427640616893768,
+      "learning_rate": 1.4428857715430863e-06,
+      "loss": 0.0068,
+      "step": 4858
+    },
+    {
+      "epoch": 19.436,
+      "grad_norm": 0.24073843657970428,
+      "learning_rate": 1.432865731462926e-06,
+      "loss": 0.0071,
+      "step": 4859
+    },
+    {
+      "epoch": 19.44,
+      "grad_norm": 0.22962218523025513,
+      "learning_rate": 1.4228456913827655e-06,
+      "loss": 0.008,
+      "step": 4860
+    },
+    {
+      "epoch": 19.444,
+      "grad_norm": 0.1950879544019699,
+      "learning_rate": 1.4128256513026053e-06,
+      "loss": 0.0067,
+      "step": 4861
+    },
+    {
+      "epoch": 19.448,
+      "grad_norm": 0.16316533088684082,
+      "learning_rate": 1.402805611222445e-06,
+      "loss": 0.0074,
+      "step": 4862
+    },
+    {
+      "epoch": 19.452,
+      "grad_norm": 0.18425200879573822,
+      "learning_rate": 1.3927855711422847e-06,
+      "loss": 0.0075,
+      "step": 4863
+    },
+    {
+      "epoch": 19.456,
+      "grad_norm": 0.1857520490884781,
+      "learning_rate": 1.3827655310621243e-06,
+      "loss": 0.0071,
+      "step": 4864
+    },
+    {
+      "epoch": 19.46,
+      "grad_norm": 0.20833329856395721,
+      "learning_rate": 1.372745490981964e-06,
+      "loss": 0.007,
+      "step": 4865
+    },
+    {
+      "epoch": 19.464,
+      "grad_norm": 0.21447627246379852,
+      "learning_rate": 1.3627254509018037e-06,
+      "loss": 0.0074,
+      "step": 4866
+    },
+    {
+      "epoch": 19.468,
+      "grad_norm": 0.14376424252986908,
+      "learning_rate": 1.3527054108216433e-06,
+      "loss": 0.0063,
+      "step": 4867
+    },
+    {
+      "epoch": 19.472,
+      "grad_norm": 0.17494480311870575,
+      "learning_rate": 1.3426853707414828e-06,
+      "loss": 0.0075,
+      "step": 4868
+    },
+    {
+      "epoch": 19.476,
+      "grad_norm": 0.19328361749649048,
+      "learning_rate": 1.3326653306613229e-06,
+      "loss": 0.008,
+      "step": 4869
+    },
+    {
+      "epoch": 19.48,
+      "grad_norm": 0.15253637731075287,
+      "learning_rate": 1.3226452905811624e-06,
+      "loss": 0.0076,
+      "step": 4870
+    },
+    {
+      "epoch": 19.484,
+      "grad_norm": 0.15423879027366638,
+      "learning_rate": 1.312625250501002e-06,
+      "loss": 0.0066,
+      "step": 4871
+    },
+    {
+      "epoch": 19.488,
+      "grad_norm": 0.16223280131816864,
+      "learning_rate": 1.3026052104208416e-06,
+      "loss": 0.0071,
+      "step": 4872
+    },
+    {
+      "epoch": 19.492,
+      "grad_norm": 0.14479829370975494,
+      "learning_rate": 1.2925851703406814e-06,
+      "loss": 0.0065,
+      "step": 4873
+    },
+    {
+      "epoch": 19.496,
+      "grad_norm": 0.1785711795091629,
+      "learning_rate": 1.282565130260521e-06,
+      "loss": 0.0078,
+      "step": 4874
+    },
+    {
+      "epoch": 19.5,
+      "grad_norm": 0.14600437879562378,
+      "learning_rate": 1.2725450901803608e-06,
+      "loss": 0.0064,
+      "step": 4875
+    },
+    {
+      "epoch": 19.504,
+      "grad_norm": 0.19797396659851074,
+      "learning_rate": 1.2625250501002006e-06,
+      "loss": 0.0069,
+      "step": 4876
+    },
+    {
+      "epoch": 19.508,
+      "grad_norm": 0.19664421677589417,
+      "learning_rate": 1.2525050100200402e-06,
+      "loss": 0.0082,
+      "step": 4877
+    },
+    {
+      "epoch": 19.512,
+      "grad_norm": 0.20880340039730072,
+      "learning_rate": 1.2424849699398798e-06,
+      "loss": 0.0082,
+      "step": 4878
+    },
+    {
+      "epoch": 19.516,
+      "grad_norm": 0.14585073292255402,
+      "learning_rate": 1.2324649298597194e-06,
+      "loss": 0.0066,
+      "step": 4879
+    },
+    {
+      "epoch": 19.52,
+      "grad_norm": 0.2517703175544739,
+      "learning_rate": 1.2224448897795592e-06,
+      "loss": 0.0073,
+      "step": 4880
+    },
+    {
+      "epoch": 19.524,
+      "grad_norm": 0.18812379240989685,
+      "learning_rate": 1.212424849699399e-06,
+      "loss": 0.008,
+      "step": 4881
+    },
+    {
+      "epoch": 19.528,
+      "grad_norm": 0.15317322313785553,
+      "learning_rate": 1.2024048096192386e-06,
+      "loss": 0.0064,
+      "step": 4882
+    },
+    {
+      "epoch": 19.532,
+      "grad_norm": 0.22038224339485168,
+      "learning_rate": 1.1923847695390782e-06,
+      "loss": 0.0072,
+      "step": 4883
+    },
+    {
+      "epoch": 19.536,
+      "grad_norm": 0.16358880698680878,
+      "learning_rate": 1.182364729458918e-06,
+      "loss": 0.0067,
+      "step": 4884
+    },
+    {
+      "epoch": 19.54,
+      "grad_norm": 0.17958605289459229,
+      "learning_rate": 1.1723446893787575e-06,
+      "loss": 0.0065,
+      "step": 4885
+    },
+    {
+      "epoch": 19.544,
+      "grad_norm": 0.1586514711380005,
+      "learning_rate": 1.1623246492985971e-06,
+      "loss": 0.0061,
+      "step": 4886
+    },
+    {
+      "epoch": 19.548000000000002,
+      "grad_norm": 0.08892928808927536,
+      "learning_rate": 1.152304609218437e-06,
+      "loss": 0.0023,
+      "step": 4887
+    },
+    {
+      "epoch": 19.552,
+      "grad_norm": 0.13370400667190552,
+      "learning_rate": 1.1422845691382767e-06,
+      "loss": 0.0057,
+      "step": 4888
+    },
+    {
+      "epoch": 19.556,
+      "grad_norm": 0.17751452326774597,
+      "learning_rate": 1.1322645290581163e-06,
+      "loss": 0.0066,
+      "step": 4889
+    },
+    {
+      "epoch": 19.56,
+      "grad_norm": 0.20876137912273407,
+      "learning_rate": 1.122244488977956e-06,
+      "loss": 0.0066,
+      "step": 4890
+    },
+    {
+      "epoch": 19.564,
+      "grad_norm": 0.19675932824611664,
+      "learning_rate": 1.1122244488977955e-06,
+      "loss": 0.0083,
+      "step": 4891
+    },
+    {
+      "epoch": 19.568,
+      "grad_norm": 0.157494455575943,
+      "learning_rate": 1.1022044088176353e-06,
+      "loss": 0.0071,
+      "step": 4892
+    },
+    {
+      "epoch": 19.572,
+      "grad_norm": 0.1072336882352829,
+      "learning_rate": 1.092184368737475e-06,
+      "loss": 0.0032,
+      "step": 4893
+    },
+    {
+      "epoch": 19.576,
+      "grad_norm": 0.19063591957092285,
+      "learning_rate": 1.0821643286573147e-06,
+      "loss": 0.0073,
+      "step": 4894
+    },
+    {
+      "epoch": 19.58,
+      "grad_norm": 0.1833014339208603,
+      "learning_rate": 1.0721442885771543e-06,
+      "loss": 0.0076,
+      "step": 4895
+    },
+    {
+      "epoch": 19.584,
+      "grad_norm": 0.19544434547424316,
+      "learning_rate": 1.062124248496994e-06,
+      "loss": 0.0083,
+      "step": 4896
+    },
+    {
+      "epoch": 19.588,
+      "grad_norm": 0.19219093024730682,
+      "learning_rate": 1.0521042084168337e-06,
+      "loss": 0.0072,
+      "step": 4897
+    },
+    {
+      "epoch": 19.592,
+      "grad_norm": 0.18003998696804047,
+      "learning_rate": 1.0420841683366733e-06,
+      "loss": 0.0075,
+      "step": 4898
+    },
+    {
+      "epoch": 19.596,
+      "grad_norm": 0.14643754065036774,
+      "learning_rate": 1.032064128256513e-06,
+      "loss": 0.0074,
+      "step": 4899
+    },
+    {
+      "epoch": 19.6,
+      "grad_norm": 0.14837734401226044,
+      "learning_rate": 1.0220440881763529e-06,
+      "loss": 0.0069,
+      "step": 4900
+    },
+    {
+      "epoch": 19.604,
+      "grad_norm": 0.21540893614292145,
+      "learning_rate": 1.0120240480961924e-06,
+      "loss": 0.0069,
+      "step": 4901
+    },
+    {
+      "epoch": 19.608,
+      "grad_norm": 0.15125292539596558,
+      "learning_rate": 1.002004008016032e-06,
+      "loss": 0.0072,
+      "step": 4902
+    },
+    {
+      "epoch": 19.612,
+      "grad_norm": 0.2173326164484024,
+      "learning_rate": 9.919839679358716e-07,
+      "loss": 0.0074,
+      "step": 4903
+    },
+    {
+      "epoch": 19.616,
+      "grad_norm": 0.22422145307064056,
+      "learning_rate": 9.819639278557114e-07,
+      "loss": 0.0076,
+      "step": 4904
+    },
+    {
+      "epoch": 19.62,
+      "grad_norm": 0.21010397374629974,
+      "learning_rate": 9.719438877755512e-07,
+      "loss": 0.0074,
+      "step": 4905
+    },
+    {
+      "epoch": 19.624,
+      "grad_norm": 0.16973888874053955,
+      "learning_rate": 9.619238476953908e-07,
+      "loss": 0.0065,
+      "step": 4906
+    },
+    {
+      "epoch": 19.628,
+      "grad_norm": 0.19270877540111542,
+      "learning_rate": 9.519038076152306e-07,
+      "loss": 0.0068,
+      "step": 4907
+    },
+    {
+      "epoch": 19.632,
+      "grad_norm": 0.14899659156799316,
+      "learning_rate": 9.418837675350702e-07,
+      "loss": 0.0074,
+      "step": 4908
+    },
+    {
+      "epoch": 19.636,
+      "grad_norm": 0.18225519359111786,
+      "learning_rate": 9.318637274549099e-07,
+      "loss": 0.0065,
+      "step": 4909
+    },
+    {
+      "epoch": 19.64,
+      "grad_norm": 0.175700843334198,
+      "learning_rate": 9.218436873747495e-07,
+      "loss": 0.0075,
+      "step": 4910
+    },
+    {
+      "epoch": 19.644,
+      "grad_norm": 0.1648159772157669,
+      "learning_rate": 9.118236472945893e-07,
+      "loss": 0.0073,
+      "step": 4911
+    },
+    {
+      "epoch": 19.648,
+      "grad_norm": 0.22431142628192902,
+      "learning_rate": 9.01803607214429e-07,
+      "loss": 0.0073,
+      "step": 4912
+    },
+    {
+      "epoch": 19.652,
+      "grad_norm": 0.19765180349349976,
+      "learning_rate": 8.917835671342686e-07,
+      "loss": 0.0072,
+      "step": 4913
+    },
+    {
+      "epoch": 19.656,
+      "grad_norm": 0.16144563257694244,
+      "learning_rate": 8.817635270541082e-07,
+      "loss": 0.007,
+      "step": 4914
+    },
+    {
+      "epoch": 19.66,
+      "grad_norm": 0.16720423102378845,
+      "learning_rate": 8.71743486973948e-07,
+      "loss": 0.0068,
+      "step": 4915
+    },
+    {
+      "epoch": 19.664,
+      "grad_norm": 0.201465904712677,
+      "learning_rate": 8.617234468937877e-07,
+      "loss": 0.0077,
+      "step": 4916
+    },
+    {
+      "epoch": 19.668,
+      "grad_norm": 0.16798268258571625,
+      "learning_rate": 8.517034068136272e-07,
+      "loss": 0.0063,
+      "step": 4917
+    },
+    {
+      "epoch": 19.672,
+      "grad_norm": 0.1629098355770111,
+      "learning_rate": 8.416833667334669e-07,
+      "loss": 0.0075,
+      "step": 4918
+    },
+    {
+      "epoch": 19.676,
+      "grad_norm": 0.165368914604187,
+      "learning_rate": 8.316633266533067e-07,
+      "loss": 0.0068,
+      "step": 4919
+    },
+    {
+      "epoch": 19.68,
+      "grad_norm": 0.21303412318229675,
+      "learning_rate": 8.216432865731463e-07,
+      "loss": 0.0078,
+      "step": 4920
+    },
+    {
+      "epoch": 19.684,
+      "grad_norm": 0.16514278948307037,
+      "learning_rate": 8.11623246492986e-07,
+      "loss": 0.0066,
+      "step": 4921
+    },
+    {
+      "epoch": 19.688,
+      "grad_norm": 0.14475776255130768,
+      "learning_rate": 8.016032064128256e-07,
+      "loss": 0.007,
+      "step": 4922
+    },
+    {
+      "epoch": 19.692,
+      "grad_norm": 0.2859732508659363,
+      "learning_rate": 7.915831663326654e-07,
+      "loss": 0.008,
+      "step": 4923
+    },
+    {
+      "epoch": 19.696,
+      "grad_norm": 0.2173309475183487,
+      "learning_rate": 7.815631262525051e-07,
+      "loss": 0.0079,
+      "step": 4924
+    },
+    {
+      "epoch": 19.7,
+      "grad_norm": 0.14563588798046112,
+      "learning_rate": 7.715430861723447e-07,
+      "loss": 0.0072,
+      "step": 4925
+    },
+    {
+      "epoch": 19.704,
+      "grad_norm": 0.1795637160539627,
+      "learning_rate": 7.615230460921844e-07,
+      "loss": 0.007,
+      "step": 4926
+    },
+    {
+      "epoch": 19.708,
+      "grad_norm": 0.1522742211818695,
+      "learning_rate": 7.515030060120241e-07,
+      "loss": 0.0062,
+      "step": 4927
+    },
+    {
+      "epoch": 19.712,
+      "grad_norm": 0.23236194252967834,
+      "learning_rate": 7.414829659318638e-07,
+      "loss": 0.0078,
+      "step": 4928
+    },
+    {
+      "epoch": 19.716,
+      "grad_norm": 0.15131531655788422,
+      "learning_rate": 7.314629258517034e-07,
+      "loss": 0.0063,
+      "step": 4929
+    },
+    {
+      "epoch": 19.72,
+      "grad_norm": 0.15103590488433838,
+      "learning_rate": 7.214428857715432e-07,
+      "loss": 0.0069,
+      "step": 4930
+    },
+    {
+      "epoch": 19.724,
+      "grad_norm": 0.15178599953651428,
+      "learning_rate": 7.114228456913828e-07,
+      "loss": 0.0066,
+      "step": 4931
+    },
+    {
+      "epoch": 19.728,
+      "grad_norm": 0.22268012166023254,
+      "learning_rate": 7.014028056112224e-07,
+      "loss": 0.0079,
+      "step": 4932
+    },
+    {
+      "epoch": 19.732,
+      "grad_norm": 0.1779235601425171,
+      "learning_rate": 6.913827655310621e-07,
+      "loss": 0.0064,
+      "step": 4933
+    },
+    {
+      "epoch": 19.736,
+      "grad_norm": 0.17940625548362732,
+      "learning_rate": 6.813627254509018e-07,
+      "loss": 0.007,
+      "step": 4934
+    },
+    {
+      "epoch": 19.74,
+      "grad_norm": 0.17642860114574432,
+      "learning_rate": 6.713426853707414e-07,
+      "loss": 0.0062,
+      "step": 4935
+    },
+    {
+      "epoch": 19.744,
+      "grad_norm": 0.18925072252750397,
+      "learning_rate": 6.613226452905812e-07,
+      "loss": 0.0079,
+      "step": 4936
+    },
+    {
+      "epoch": 19.748,
+      "grad_norm": 0.1245235726237297,
+      "learning_rate": 6.513026052104208e-07,
+      "loss": 0.0063,
+      "step": 4937
+    },
+    {
+      "epoch": 19.752,
+      "grad_norm": 0.21628788113594055,
+      "learning_rate": 6.412825651302605e-07,
+      "loss": 0.0081,
+      "step": 4938
+    },
+    {
+      "epoch": 19.756,
+      "grad_norm": 0.13739117980003357,
+      "learning_rate": 6.312625250501003e-07,
+      "loss": 0.0041,
+      "step": 4939
+    },
+    {
+      "epoch": 19.76,
+      "grad_norm": 0.18183039128780365,
+      "learning_rate": 6.212424849699399e-07,
+      "loss": 0.0064,
+      "step": 4940
+    },
+    {
+      "epoch": 19.764,
+      "grad_norm": 0.18417534232139587,
+      "learning_rate": 6.112224448897796e-07,
+      "loss": 0.0076,
+      "step": 4941
+    },
+    {
+      "epoch": 19.768,
+      "grad_norm": 0.18779101967811584,
+      "learning_rate": 6.012024048096193e-07,
+      "loss": 0.0076,
+      "step": 4942
+    },
+    {
+      "epoch": 19.772,
+      "grad_norm": 0.1617640256881714,
+      "learning_rate": 5.91182364729459e-07,
+      "loss": 0.0065,
+      "step": 4943
+    },
+    {
+      "epoch": 19.776,
+      "grad_norm": 0.15822680294513702,
+      "learning_rate": 5.811623246492986e-07,
+      "loss": 0.006,
+      "step": 4944
+    },
+    {
+      "epoch": 19.78,
+      "grad_norm": 0.17007651925086975,
+      "learning_rate": 5.711422845691384e-07,
+      "loss": 0.0069,
+      "step": 4945
+    },
+    {
+      "epoch": 19.784,
+      "grad_norm": 0.1721976101398468,
+      "learning_rate": 5.61122244488978e-07,
+      "loss": 0.0071,
+      "step": 4946
+    },
+    {
+      "epoch": 19.788,
+      "grad_norm": 0.1397552639245987,
+      "learning_rate": 5.511022044088177e-07,
+      "loss": 0.0069,
+      "step": 4947
+    },
+    {
+      "epoch": 19.792,
+      "grad_norm": 0.18421714007854462,
+      "learning_rate": 5.410821643286573e-07,
+      "loss": 0.0048,
+      "step": 4948
+    },
+    {
+      "epoch": 19.796,
+      "grad_norm": 0.16504515707492828,
+      "learning_rate": 5.31062124248497e-07,
+      "loss": 0.0066,
+      "step": 4949
+    },
+    {
+      "epoch": 19.8,
+      "grad_norm": 0.18747548758983612,
+      "learning_rate": 5.210420841683366e-07,
+      "loss": 0.0081,
+      "step": 4950
+    },
+    {
+      "epoch": 19.804,
+      "grad_norm": 0.17722636461257935,
+      "learning_rate": 5.110220440881764e-07,
+      "loss": 0.0071,
+      "step": 4951
+    },
+    {
+      "epoch": 19.808,
+      "grad_norm": 0.14402155578136444,
+      "learning_rate": 5.01002004008016e-07,
+      "loss": 0.0061,
+      "step": 4952
+    },
+    {
+      "epoch": 19.812,
+      "grad_norm": 0.2421988993883133,
+      "learning_rate": 4.909819639278557e-07,
+      "loss": 0.0075,
+      "step": 4953
+    },
+    {
+      "epoch": 19.816,
+      "grad_norm": 0.1611681580543518,
+      "learning_rate": 4.809619238476954e-07,
+      "loss": 0.0063,
+      "step": 4954
+    },
+    {
+      "epoch": 19.82,
+      "grad_norm": 0.17111510038375854,
+      "learning_rate": 4.709418837675351e-07,
+      "loss": 0.0064,
+      "step": 4955
+    },
+    {
+      "epoch": 19.824,
+      "grad_norm": 0.2002701759338379,
+      "learning_rate": 4.6092184368737474e-07,
+      "loss": 0.0066,
+      "step": 4956
+    },
+    {
+      "epoch": 19.828,
+      "grad_norm": 0.23351456224918365,
+      "learning_rate": 4.509018036072145e-07,
+      "loss": 0.0073,
+      "step": 4957
+    },
+    {
+      "epoch": 19.832,
+      "grad_norm": 0.12465520948171616,
+      "learning_rate": 4.408817635270541e-07,
+      "loss": 0.0057,
+      "step": 4958
+    },
+    {
+      "epoch": 19.836,
+      "grad_norm": 0.2019532173871994,
+      "learning_rate": 4.3086172344689383e-07,
+      "loss": 0.0068,
+      "step": 4959
+    },
+    {
+      "epoch": 19.84,
+      "grad_norm": 0.1522103101015091,
+      "learning_rate": 4.2084168336673347e-07,
+      "loss": 0.0062,
+      "step": 4960
+    },
+    {
+      "epoch": 19.844,
+      "grad_norm": 0.22643589973449707,
+      "learning_rate": 4.1082164328657316e-07,
+      "loss": 0.0079,
+      "step": 4961
+    },
+    {
+      "epoch": 19.848,
+      "grad_norm": 0.1775081604719162,
+      "learning_rate": 4.008016032064128e-07,
+      "loss": 0.0065,
+      "step": 4962
+    },
+    {
+      "epoch": 19.852,
+      "grad_norm": 0.17318783700466156,
+      "learning_rate": 3.9078156312625255e-07,
+      "loss": 0.007,
+      "step": 4963
+    },
+    {
+      "epoch": 19.856,
+      "grad_norm": 0.20053575932979584,
+      "learning_rate": 3.807615230460922e-07,
+      "loss": 0.0074,
+      "step": 4964
+    },
+    {
+      "epoch": 19.86,
+      "grad_norm": 0.2197975367307663,
+      "learning_rate": 3.707414829659319e-07,
+      "loss": 0.0077,
+      "step": 4965
+    },
+    {
+      "epoch": 19.864,
+      "grad_norm": 0.11314887553453445,
+      "learning_rate": 3.607214428857716e-07,
+      "loss": 0.0039,
+      "step": 4966
+    },
+    {
+      "epoch": 19.868,
+      "grad_norm": 0.23044568300247192,
+      "learning_rate": 3.507014028056112e-07,
+      "loss": 0.0081,
+      "step": 4967
+    },
+    {
+      "epoch": 19.872,
+      "grad_norm": 0.15248127281665802,
+      "learning_rate": 3.406813627254509e-07,
+      "loss": 0.0068,
+      "step": 4968
+    },
+    {
+      "epoch": 19.876,
+      "grad_norm": 0.16394023597240448,
+      "learning_rate": 3.306613226452906e-07,
+      "loss": 0.0066,
+      "step": 4969
+    },
+    {
+      "epoch": 19.88,
+      "grad_norm": 0.1749032884836197,
+      "learning_rate": 3.2064128256513025e-07,
+      "loss": 0.0073,
+      "step": 4970
+    },
+    {
+      "epoch": 19.884,
+      "grad_norm": 0.17276360094547272,
+      "learning_rate": 3.1062124248496995e-07,
+      "loss": 0.007,
+      "step": 4971
+    },
+    {
+      "epoch": 19.888,
+      "grad_norm": 0.13797195255756378,
+      "learning_rate": 3.0060120240480964e-07,
+      "loss": 0.007,
+      "step": 4972
+    },
+    {
+      "epoch": 19.892,
+      "grad_norm": 0.25307130813598633,
+      "learning_rate": 2.905811623246493e-07,
+      "loss": 0.0085,
+      "step": 4973
+    },
+    {
+      "epoch": 19.896,
+      "grad_norm": 0.17503662407398224,
+      "learning_rate": 2.80561122244489e-07,
+      "loss": 0.0076,
+      "step": 4974
+    },
+    {
+      "epoch": 19.9,
+      "grad_norm": 0.14293161034584045,
+      "learning_rate": 2.7054108216432867e-07,
+      "loss": 0.007,
+      "step": 4975
+    },
+    {
+      "epoch": 19.904,
+      "grad_norm": 0.1660107970237732,
+      "learning_rate": 2.605210420841683e-07,
+      "loss": 0.0065,
+      "step": 4976
+    },
+    {
+      "epoch": 19.908,
+      "grad_norm": 0.245314821600914,
+      "learning_rate": 2.50501002004008e-07,
+      "loss": 0.0092,
+      "step": 4977
+    },
+    {
+      "epoch": 19.912,
+      "grad_norm": 0.21174436807632446,
+      "learning_rate": 2.404809619238477e-07,
+      "loss": 0.007,
+      "step": 4978
+    },
+    {
+      "epoch": 19.916,
+      "grad_norm": 0.170380100607872,
+      "learning_rate": 2.3046092184368737e-07,
+      "loss": 0.0061,
+      "step": 4979
+    },
+    {
+      "epoch": 19.92,
+      "grad_norm": 0.2893669307231903,
+      "learning_rate": 2.2044088176352704e-07,
+      "loss": 0.0086,
+      "step": 4980
+    },
+    {
+      "epoch": 19.924,
+      "grad_norm": 0.13519108295440674,
+      "learning_rate": 2.1042084168336673e-07,
+      "loss": 0.006,
+      "step": 4981
+    },
+    {
+      "epoch": 19.928,
+      "grad_norm": 0.17455951869487762,
+      "learning_rate": 2.004008016032064e-07,
+      "loss": 0.0076,
+      "step": 4982
+    },
+    {
+      "epoch": 19.932,
+      "grad_norm": 0.2173195630311966,
+      "learning_rate": 1.903807615230461e-07,
+      "loss": 0.0086,
+      "step": 4983
+    },
+    {
+      "epoch": 19.936,
+      "grad_norm": 0.16455821692943573,
+      "learning_rate": 1.803607214428858e-07,
+      "loss": 0.0066,
+      "step": 4984
+    },
+    {
+      "epoch": 19.94,
+      "grad_norm": 0.16139689087867737,
+      "learning_rate": 1.7034068136272546e-07,
+      "loss": 0.0075,
+      "step": 4985
+    },
+    {
+      "epoch": 19.944,
+      "grad_norm": 0.18227294087409973,
+      "learning_rate": 1.6032064128256513e-07,
+      "loss": 0.007,
+      "step": 4986
+    },
+    {
+      "epoch": 19.948,
+      "grad_norm": 0.13132084906101227,
+      "learning_rate": 1.5030060120240482e-07,
+      "loss": 0.0043,
+      "step": 4987
+    },
+    {
+      "epoch": 19.951999999999998,
+      "grad_norm": 0.1532142460346222,
+      "learning_rate": 1.402805611222445e-07,
+      "loss": 0.0063,
+      "step": 4988
+    },
+    {
+      "epoch": 19.956,
+      "grad_norm": 0.18583548069000244,
+      "learning_rate": 1.3026052104208416e-07,
+      "loss": 0.0074,
+      "step": 4989
+    },
+    {
+      "epoch": 19.96,
+      "grad_norm": 0.16315986216068268,
+      "learning_rate": 1.2024048096192385e-07,
+      "loss": 0.0066,
+      "step": 4990
+    },
+    {
+      "epoch": 19.964,
+      "grad_norm": 0.19338533282279968,
+      "learning_rate": 1.1022044088176352e-07,
+      "loss": 0.0062,
+      "step": 4991
+    },
+    {
+      "epoch": 19.968,
+      "grad_norm": 0.1874208003282547,
+      "learning_rate": 1.002004008016032e-07,
+      "loss": 0.0072,
+      "step": 4992
+    },
+    {
+      "epoch": 19.972,
+      "grad_norm": 0.19708944857120514,
+      "learning_rate": 9.01803607214429e-08,
+      "loss": 0.0075,
+      "step": 4993
+    },
+    {
+      "epoch": 19.976,
+      "grad_norm": 0.13537871837615967,
+      "learning_rate": 8.016032064128256e-08,
+      "loss": 0.0072,
+      "step": 4994
+    },
+    {
+      "epoch": 19.98,
+      "grad_norm": 0.200593501329422,
+      "learning_rate": 7.014028056112224e-08,
+      "loss": 0.0077,
+      "step": 4995
+    },
+    {
+      "epoch": 19.984,
+      "grad_norm": 0.146713986992836,
+      "learning_rate": 6.012024048096193e-08,
+      "loss": 0.0074,
+      "step": 4996
+    },
+    {
+      "epoch": 19.988,
+      "grad_norm": 0.20530632138252258,
+      "learning_rate": 5.01002004008016e-08,
+      "loss": 0.0066,
+      "step": 4997
+    },
+    {
+      "epoch": 19.992,
+      "grad_norm": 0.1384311020374298,
+      "learning_rate": 4.008016032064128e-08,
+      "loss": 0.0072,
+      "step": 4998
+    },
+    {
+      "epoch": 19.996,
+      "grad_norm": 0.1999179571866989,
+      "learning_rate": 3.006012024048096e-08,
+      "loss": 0.0075,
+      "step": 4999
+    },
+    {
+      "epoch": 20.0,
+      "grad_norm": 0.22525227069854736,
+      "learning_rate": 2.004008016032064e-08,
+      "loss": 0.0047,
+      "step": 5000
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 5000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 20,
+  "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": 2.07226802884608e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}