diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,4522 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 5.0,
+  "eval_steps": 500,
+  "global_step": 640,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.4907612239995434,
+      "learning_rate": 1.09375e-07,
+      "loss": 0.7509,
+      "step": 1
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.1504421825230433,
+      "learning_rate": 2.1875e-07,
+      "loss": 0.6507,
+      "step": 2
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.105418020487931,
+      "learning_rate": 3.28125e-07,
+      "loss": 0.6342,
+      "step": 3
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.254821921473971,
+      "learning_rate": 4.375e-07,
+      "loss": 0.6813,
+      "step": 4
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.3389206590297786,
+      "learning_rate": 5.46875e-07,
+      "loss": 0.7041,
+      "step": 5
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.103212376255131,
+      "learning_rate": 6.5625e-07,
+      "loss": 0.6574,
+      "step": 6
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.4026075935403863,
+      "learning_rate": 7.65625e-07,
+      "loss": 0.7184,
+      "step": 7
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.144883715920057,
+      "learning_rate": 8.75e-07,
+      "loss": 0.6639,
+      "step": 8
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.058002010109806,
+      "learning_rate": 9.84375e-07,
+      "loss": 0.6932,
+      "step": 9
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.0365307674019246,
+      "learning_rate": 1.09375e-06,
+      "loss": 0.6667,
+      "step": 10
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.6573450994955512,
+      "learning_rate": 1.203125e-06,
+      "loss": 0.6364,
+      "step": 11
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.734125128574585,
+      "learning_rate": 1.3125e-06,
+      "loss": 0.7094,
+      "step": 12
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.5814881821639086,
+      "learning_rate": 1.421875e-06,
+      "loss": 0.6469,
+      "step": 13
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.143717763435021,
+      "learning_rate": 1.53125e-06,
+      "loss": 0.6392,
+      "step": 14
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.113314084373293,
+      "learning_rate": 1.640625e-06,
+      "loss": 0.6419,
+      "step": 15
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.1248682912950092,
+      "learning_rate": 1.75e-06,
+      "loss": 0.6257,
+      "step": 16
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.0523745421583506,
+      "learning_rate": 1.8593749999999999e-06,
+      "loss": 0.6387,
+      "step": 17
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9540773623803008,
+      "learning_rate": 1.96875e-06,
+      "loss": 0.5753,
+      "step": 18
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.0055838205407808,
+      "learning_rate": 2.078125e-06,
+      "loss": 0.6348,
+      "step": 19
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.2432091605190205,
+      "learning_rate": 2.1875e-06,
+      "loss": 0.6318,
+      "step": 20
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.3869607173028649,
+      "learning_rate": 2.296875e-06,
+      "loss": 0.6012,
+      "step": 21
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.240637379277943,
+      "learning_rate": 2.40625e-06,
+      "loss": 0.5629,
+      "step": 22
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.1314993062351444,
+      "learning_rate": 2.515625e-06,
+      "loss": 0.5683,
+      "step": 23
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8815352840655305,
+      "learning_rate": 2.625e-06,
+      "loss": 0.5399,
+      "step": 24
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.8120210446427198,
+      "learning_rate": 2.734375e-06,
+      "loss": 0.5661,
+      "step": 25
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7406111049433269,
+      "learning_rate": 2.84375e-06,
+      "loss": 0.5374,
+      "step": 26
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.9308313980688961,
+      "learning_rate": 2.9531249999999998e-06,
+      "loss": 0.5139,
+      "step": 27
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.1735905075611526,
+      "learning_rate": 3.0625e-06,
+      "loss": 0.549,
+      "step": 28
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.993673114117623,
+      "learning_rate": 3.171875e-06,
+      "loss": 0.5097,
+      "step": 29
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.912014516728298,
+      "learning_rate": 3.28125e-06,
+      "loss": 0.5496,
+      "step": 30
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.7795311370852275,
+      "learning_rate": 3.390625e-06,
+      "loss": 0.5672,
+      "step": 31
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.7360032804580571,
+      "learning_rate": 3.5e-06,
+      "loss": 0.5452,
+      "step": 32
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.9290691764870666,
+      "learning_rate": 3.609375e-06,
+      "loss": 0.5574,
+      "step": 33
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.8323834195359898,
+      "learning_rate": 3.7187499999999998e-06,
+      "loss": 0.5093,
+      "step": 34
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.7793193471539842,
+      "learning_rate": 3.828125e-06,
+      "loss": 0.5053,
+      "step": 35
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.694220694288892,
+      "learning_rate": 3.9375e-06,
+      "loss": 0.5364,
+      "step": 36
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.5692085599440716,
+      "learning_rate": 4.046875e-06,
+      "loss": 0.5141,
+      "step": 37
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.5870629876760964,
+      "learning_rate": 4.15625e-06,
+      "loss": 0.5303,
+      "step": 38
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.5804059716624939,
+      "learning_rate": 4.265625e-06,
+      "loss": 0.4715,
+      "step": 39
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.6036445201285046,
+      "learning_rate": 4.375e-06,
+      "loss": 0.4963,
+      "step": 40
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.4904238208746072,
+      "learning_rate": 4.484375e-06,
+      "loss": 0.4614,
+      "step": 41
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.46399381294856473,
+      "learning_rate": 4.59375e-06,
+      "loss": 0.4743,
+      "step": 42
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.4606123524196029,
+      "learning_rate": 4.703125e-06,
+      "loss": 0.4905,
+      "step": 43
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.49558583641444864,
+      "learning_rate": 4.8125e-06,
+      "loss": 0.5502,
+      "step": 44
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.4954617188222101,
+      "learning_rate": 4.921875e-06,
+      "loss": 0.4903,
+      "step": 45
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.49117419459298145,
+      "learning_rate": 5.03125e-06,
+      "loss": 0.4955,
+      "step": 46
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.47373321434706367,
+      "learning_rate": 5.140625e-06,
+      "loss": 0.5362,
+      "step": 47
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.42677909808697095,
+      "learning_rate": 5.25e-06,
+      "loss": 0.5186,
+      "step": 48
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.8847880576479226,
+      "learning_rate": 5.359375e-06,
+      "loss": 0.465,
+      "step": 49
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.38006660937319003,
+      "learning_rate": 5.46875e-06,
+      "loss": 0.4458,
+      "step": 50
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.42406636746037957,
+      "learning_rate": 5.578125e-06,
+      "loss": 0.4738,
+      "step": 51
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4497405637038939,
+      "learning_rate": 5.6875e-06,
+      "loss": 0.4717,
+      "step": 52
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.42588364899074693,
+      "learning_rate": 5.796875e-06,
+      "loss": 0.4839,
+      "step": 53
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.40971064168289695,
+      "learning_rate": 5.9062499999999996e-06,
+      "loss": 0.5087,
+      "step": 54
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.3832469222413672,
+      "learning_rate": 6.015625e-06,
+      "loss": 0.4917,
+      "step": 55
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.36300109542286413,
+      "learning_rate": 6.125e-06,
+      "loss": 0.454,
+      "step": 56
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3693983257133135,
+      "learning_rate": 6.234375e-06,
+      "loss": 0.4612,
+      "step": 57
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.34955603337351576,
+      "learning_rate": 6.34375e-06,
+      "loss": 0.4519,
+      "step": 58
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.366486649899712,
+      "learning_rate": 6.453125e-06,
+      "loss": 0.4801,
+      "step": 59
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.3851964587046632,
+      "learning_rate": 6.5625e-06,
+      "loss": 0.4746,
+      "step": 60
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3429078789254216,
+      "learning_rate": 6.6718749999999995e-06,
+      "loss": 0.4544,
+      "step": 61
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3679977159831168,
+      "learning_rate": 6.78125e-06,
+      "loss": 0.4803,
+      "step": 62
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3730323682839911,
+      "learning_rate": 6.890625e-06,
+      "loss": 0.4986,
+      "step": 63
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3445733696341208,
+      "learning_rate": 7e-06,
+      "loss": 0.4693,
+      "step": 64
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.35641969322397765,
+      "learning_rate": 6.9999479414879755e-06,
+      "loss": 0.4668,
+      "step": 65
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.33068121294839087,
+      "learning_rate": 6.999791767500524e-06,
+      "loss": 0.4505,
+      "step": 66
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.33728314549188104,
+      "learning_rate": 6.999531482683467e-06,
+      "loss": 0.4393,
+      "step": 67
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.33775640987020233,
+      "learning_rate": 6.999167094779681e-06,
+      "loss": 0.4611,
+      "step": 68
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.343304798662571,
+      "learning_rate": 6.998698614628881e-06,
+      "loss": 0.4691,
+      "step": 69
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3553019759244247,
+      "learning_rate": 6.99812605616728e-06,
+      "loss": 0.4754,
+      "step": 70
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3151337286948089,
+      "learning_rate": 6.997449436427188e-06,
+      "loss": 0.4655,
+      "step": 71
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3154508848797514,
+      "learning_rate": 6.996668775536502e-06,
+      "loss": 0.4551,
+      "step": 72
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.30673808027443833,
+      "learning_rate": 6.9957840967181034e-06,
+      "loss": 0.4335,
+      "step": 73
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3684762500038093,
+      "learning_rate": 6.994795426289171e-06,
+      "loss": 0.447,
+      "step": 74
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3245066578888837,
+      "learning_rate": 6.993702793660396e-06,
+      "loss": 0.4354,
+      "step": 75
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.33960008906806616,
+      "learning_rate": 6.992506231335112e-06,
+      "loss": 0.4411,
+      "step": 76
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3393061927704471,
+      "learning_rate": 6.991205774908319e-06,
+      "loss": 0.4449,
+      "step": 77
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3611077743448448,
+      "learning_rate": 6.9898014630656335e-06,
+      "loss": 0.446,
+      "step": 78
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3727255174354603,
+      "learning_rate": 6.988293337582131e-06,
+      "loss": 0.4401,
+      "step": 79
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3626035130302128,
+      "learning_rate": 6.9866814433211094e-06,
+      "loss": 0.4821,
+      "step": 80
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3376405406112302,
+      "learning_rate": 6.984965828232749e-06,
+      "loss": 0.4625,
+      "step": 81
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.2976230880823188,
+      "learning_rate": 6.983146543352689e-06,
+      "loss": 0.4362,
+      "step": 82
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3055669447324052,
+      "learning_rate": 6.981223642800509e-06,
+      "loss": 0.4321,
+      "step": 83
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.33059939115881387,
+      "learning_rate": 6.979197183778118e-06,
+      "loss": 0.4667,
+      "step": 84
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3233396642976923,
+      "learning_rate": 6.977067226568055e-06,
+      "loss": 0.4615,
+      "step": 85
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3320915668626651,
+      "learning_rate": 6.974833834531692e-06,
+      "loss": 0.4447,
+      "step": 86
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.35242585608089094,
+      "learning_rate": 6.972497074107354e-06,
+      "loss": 0.4717,
+      "step": 87
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.335342569378105,
+      "learning_rate": 6.970057014808337e-06,
+      "loss": 0.4457,
+      "step": 88
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.33281453048842513,
+      "learning_rate": 6.967513729220844e-06,
+      "loss": 0.4485,
+      "step": 89
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3170073581449777,
+      "learning_rate": 6.964867293001827e-06,
+      "loss": 0.4521,
+      "step": 90
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.34489609925729525,
+      "learning_rate": 6.962117784876734e-06,
+      "loss": 0.4592,
+      "step": 91
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3133849074705517,
+      "learning_rate": 6.959265286637163e-06,
+      "loss": 0.4209,
+      "step": 92
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.35497309527917403,
+      "learning_rate": 6.956309883138437e-06,
+      "loss": 0.492,
+      "step": 93
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3437625485559197,
+      "learning_rate": 6.953251662297077e-06,
+      "loss": 0.438,
+      "step": 94
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3384164661888183,
+      "learning_rate": 6.950090715088181e-06,
+      "loss": 0.4447,
+      "step": 95
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.39820938339107037,
+      "learning_rate": 6.946827135542729e-06,
+      "loss": 0.4686,
+      "step": 96
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.31137355275022705,
+      "learning_rate": 6.94346102074477e-06,
+      "loss": 0.4116,
+      "step": 97
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3374951130929495,
+      "learning_rate": 6.939992470828554e-06,
+      "loss": 0.4267,
+      "step": 98
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.385791851799853,
+      "learning_rate": 6.936421588975533e-06,
+      "loss": 0.4749,
+      "step": 99
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.4217868488199523,
+      "learning_rate": 6.932748481411306e-06,
+      "loss": 0.4819,
+      "step": 100
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.32286969859849285,
+      "learning_rate": 6.928973257402453e-06,
+      "loss": 0.4355,
+      "step": 101
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3283858360190172,
+      "learning_rate": 6.925096029253284e-06,
+      "loss": 0.4585,
+      "step": 102
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.33207664810442833,
+      "learning_rate": 6.921116912302502e-06,
+      "loss": 0.4423,
+      "step": 103
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.36777168702028123,
+      "learning_rate": 6.917036024919767e-06,
+      "loss": 0.4466,
+      "step": 104
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.34291748925162174,
+      "learning_rate": 6.912853488502181e-06,
+      "loss": 0.4659,
+      "step": 105
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3414087931045369,
+      "learning_rate": 6.908569427470668e-06,
+      "loss": 0.4718,
+      "step": 106
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3184110518552108,
+      "learning_rate": 6.904183969266283e-06,
+      "loss": 0.4395,
+      "step": 107
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.33317082829069683,
+      "learning_rate": 6.899697244346414e-06,
+      "loss": 0.4554,
+      "step": 108
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.37433602938769583,
+      "learning_rate": 6.8951093861809044e-06,
+      "loss": 0.4714,
+      "step": 109
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.3599952403662053,
+      "learning_rate": 6.890420531248076e-06,
+      "loss": 0.4563,
+      "step": 110
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3083422618801623,
+      "learning_rate": 6.885630819030679e-06,
+      "loss": 0.412,
+      "step": 111
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.3386336304018581,
+      "learning_rate": 6.880740392011738e-06,
+      "loss": 0.4465,
+      "step": 112
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.3475827421782862,
+      "learning_rate": 6.875749395670313e-06,
+      "loss": 0.4577,
+      "step": 113
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3966450413893103,
+      "learning_rate": 6.870657978477169e-06,
+      "loss": 0.454,
+      "step": 114
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3406821834010313,
+      "learning_rate": 6.865466291890367e-06,
+      "loss": 0.4379,
+      "step": 115
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.3587077375756719,
+      "learning_rate": 6.860174490350751e-06,
+      "loss": 0.4513,
+      "step": 116
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.33640117371342315,
+      "learning_rate": 6.854782731277357e-06,
+      "loss": 0.4202,
+      "step": 117
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.3428259358656843,
+      "learning_rate": 6.849291175062731e-06,
+      "loss": 0.4435,
+      "step": 118
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.42577201899051426,
+      "learning_rate": 6.843699985068156e-06,
+      "loss": 0.4824,
+      "step": 119
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3316256364391201,
+      "learning_rate": 6.838009327618794e-06,
+      "loss": 0.4165,
+      "step": 120
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.39342611861287163,
+      "learning_rate": 6.8322193719987345e-06,
+      "loss": 0.4658,
+      "step": 121
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3619051246676873,
+      "learning_rate": 6.8263302904459634e-06,
+      "loss": 0.4214,
+      "step": 122
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3421694783963853,
+      "learning_rate": 6.820342258147237e-06,
+      "loss": 0.4205,
+      "step": 123
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3520272042873268,
+      "learning_rate": 6.81425545323287e-06,
+      "loss": 0.4326,
+      "step": 124
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.37261582297065365,
+      "learning_rate": 6.808070056771437e-06,
+      "loss": 0.436,
+      "step": 125
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.3190843731641372,
+      "learning_rate": 6.801786252764388e-06,
+      "loss": 0.408,
+      "step": 126
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.45801591328905905,
+      "learning_rate": 6.795404228140573e-06,
+      "loss": 0.4262,
+      "step": 127
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.3375844074069623,
+      "learning_rate": 6.788924172750679e-06,
+      "loss": 0.454,
+      "step": 128
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.37385172623439816,
+      "learning_rate": 6.782346279361589e-06,
+      "loss": 0.4898,
+      "step": 129
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.3181793747206892,
+      "learning_rate": 6.77567074365064e-06,
+      "loss": 0.4232,
+      "step": 130
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.3422195996524943,
+      "learning_rate": 6.768897764199808e-06,
+      "loss": 0.4019,
+      "step": 131
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.36276549218726206,
+      "learning_rate": 6.762027542489795e-06,
+      "loss": 0.436,
+      "step": 132
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.38061371338027,
+      "learning_rate": 6.755060282894042e-06,
+      "loss": 0.4579,
+      "step": 133
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.32719635914528883,
+      "learning_rate": 6.747996192672646e-06,
+      "loss": 0.4233,
+      "step": 134
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.3810736630935037,
+      "learning_rate": 6.740835481966191e-06,
+      "loss": 0.4567,
+      "step": 135
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.37868571648829663,
+      "learning_rate": 6.733578363789503e-06,
+      "loss": 0.4231,
+      "step": 136
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.36004410394238534,
+      "learning_rate": 6.726225054025311e-06,
+      "loss": 0.451,
+      "step": 137
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.3908878371379872,
+      "learning_rate": 6.718775771417823e-06,
+      "loss": 0.4303,
+      "step": 138
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.3459717178074042,
+      "learning_rate": 6.711230737566219e-06,
+      "loss": 0.4196,
+      "step": 139
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.3476179794053542,
+      "learning_rate": 6.7035901769180656e-06,
+      "loss": 0.4715,
+      "step": 140
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.3586451898247301,
+      "learning_rate": 6.6958543167626265e-06,
+      "loss": 0.4315,
+      "step": 141
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.3361222789247428,
+      "learning_rate": 6.688023387224115e-06,
+      "loss": 0.44,
+      "step": 142
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.3408702408553507,
+      "learning_rate": 6.6800976212548396e-06,
+      "loss": 0.4427,
+      "step": 143
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.3412669995759075,
+      "learning_rate": 6.672077254628275e-06,
+      "loss": 0.4327,
+      "step": 144
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.35466921769883636,
+      "learning_rate": 6.663962525932052e-06,
+      "loss": 0.4439,
+      "step": 145
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.3169267316634476,
+      "learning_rate": 6.655753676560856e-06,
+      "loss": 0.3991,
+      "step": 146
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.33593441440927124,
+      "learning_rate": 6.647450950709251e-06,
+      "loss": 0.4577,
+      "step": 147
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.33950045221890385,
+      "learning_rate": 6.639054595364409e-06,
+      "loss": 0.4591,
+      "step": 148
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.3615774798815532,
+      "learning_rate": 6.630564860298768e-06,
+      "loss": 0.4344,
+      "step": 149
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.34631817223512895,
+      "learning_rate": 6.6219819980625995e-06,
+      "loss": 0.4119,
+      "step": 150
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.3332561539660665,
+      "learning_rate": 6.613306263976496e-06,
+      "loss": 0.4147,
+      "step": 151
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.34402131787716334,
+      "learning_rate": 6.604537916123775e-06,
+      "loss": 0.3931,
+      "step": 152
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.3690152266668397,
+      "learning_rate": 6.595677215342806e-06,
+      "loss": 0.4215,
+      "step": 153
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.322342729858813,
+      "learning_rate": 6.5867244252192426e-06,
+      "loss": 0.4017,
+      "step": 154
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.31344067021908767,
+      "learning_rate": 6.577679812078189e-06,
+      "loss": 0.3929,
+      "step": 155
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.3568397607324737,
+      "learning_rate": 6.568543644976277e-06,
+      "loss": 0.4195,
+      "step": 156
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.33749366055496693,
+      "learning_rate": 6.559316195693656e-06,
+      "loss": 0.3897,
+      "step": 157
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.3455835380149386,
+      "learning_rate": 6.549997738725915e-06,
+      "loss": 0.4264,
+      "step": 158
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.37228654836514496,
+      "learning_rate": 6.540588551275913e-06,
+      "loss": 0.4392,
+      "step": 159
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.33154975556916305,
+      "learning_rate": 6.531088913245536e-06,
+      "loss": 0.427,
+      "step": 160
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.3458031035321322,
+      "learning_rate": 6.5214991072273635e-06,
+      "loss": 0.4375,
+      "step": 161
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.31276873609868006,
+      "learning_rate": 6.511819418496276e-06,
+      "loss": 0.4031,
+      "step": 162
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.34409165822067467,
+      "learning_rate": 6.502050135000952e-06,
+      "loss": 0.4011,
+      "step": 163
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.36279847439403795,
+      "learning_rate": 6.492191547355313e-06,
+      "loss": 0.4237,
+      "step": 164
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.34564833445832377,
+      "learning_rate": 6.482243948829876e-06,
+      "loss": 0.4055,
+      "step": 165
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.38077908458682896,
+      "learning_rate": 6.472207635343026e-06,
+      "loss": 0.4209,
+      "step": 166
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.329913816797509,
+      "learning_rate": 6.46208290545222e-06,
+      "loss": 0.3818,
+      "step": 167
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.3111225383454026,
+      "learning_rate": 6.4518700603451e-06,
+      "loss": 0.397,
+      "step": 168
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.3432239018285694,
+      "learning_rate": 6.441569403830533e-06,
+      "loss": 0.3698,
+      "step": 169
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.35339874192273096,
+      "learning_rate": 6.431181242329578e-06,
+      "loss": 0.3878,
+      "step": 170
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.3323530722856522,
+      "learning_rate": 6.420705884866365e-06,
+      "loss": 0.4014,
+      "step": 171
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.3521691617675391,
+      "learning_rate": 6.4101436430589085e-06,
+      "loss": 0.4458,
+      "step": 172
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.3538003813466273,
+      "learning_rate": 6.399494831109832e-06,
+      "loss": 0.4014,
+      "step": 173
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.33082565177535356,
+      "learning_rate": 6.3887597657970235e-06,
+      "loss": 0.4061,
+      "step": 174
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.3585534863365245,
+      "learning_rate": 6.377938766464212e-06,
+      "loss": 0.4353,
+      "step": 175
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.3817049240831521,
+      "learning_rate": 6.367032155011471e-06,
+      "loss": 0.4272,
+      "step": 176
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.3688197288995215,
+      "learning_rate": 6.3560402558856354e-06,
+      "loss": 0.3783,
+      "step": 177
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.3291013719807746,
+      "learning_rate": 6.3449633960706536e-06,
+      "loss": 0.3683,
+      "step": 178
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.34668256480725407,
+      "learning_rate": 6.333801905077864e-06,
+      "loss": 0.3886,
+      "step": 179
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.3139656780215767,
+      "learning_rate": 6.322556114936189e-06,
+      "loss": 0.3864,
+      "step": 180
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.3479964266741095,
+      "learning_rate": 6.311226360182257e-06,
+      "loss": 0.4026,
+      "step": 181
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.3365002537739493,
+      "learning_rate": 6.2998129778504535e-06,
+      "loss": 0.4205,
+      "step": 182
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.3395485075789255,
+      "learning_rate": 6.288316307462895e-06,
+      "loss": 0.406,
+      "step": 183
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.31182460780702925,
+      "learning_rate": 6.276736691019323e-06,
+      "loss": 0.377,
+      "step": 184
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.31900712972961165,
+      "learning_rate": 6.265074472986942e-06,
+      "loss": 0.3876,
+      "step": 185
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.3234199942872983,
+      "learning_rate": 6.253330000290159e-06,
+      "loss": 0.3754,
+      "step": 186
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.34875728639110426,
+      "learning_rate": 6.241503622300277e-06,
+      "loss": 0.402,
+      "step": 187
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.3406636880839683,
+      "learning_rate": 6.229595690825086e-06,
+      "loss": 0.3955,
+      "step": 188
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.38204417456095835,
+      "learning_rate": 6.217606560098415e-06,
+      "loss": 0.3795,
+      "step": 189
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.40729559753298156,
+      "learning_rate": 6.205536586769579e-06,
+      "loss": 0.4006,
+      "step": 190
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.35939142710468447,
+      "learning_rate": 6.193386129892782e-06,
+      "loss": 0.4145,
+      "step": 191
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.3990064878239788,
+      "learning_rate": 6.181155550916423e-06,
+      "loss": 0.3964,
+      "step": 192
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.37789265888304424,
+      "learning_rate": 6.168845213672358e-06,
+      "loss": 0.395,
+      "step": 193
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.334681217049792,
+      "learning_rate": 6.156455484365066e-06,
+      "loss": 0.3825,
+      "step": 194
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.31563685389160046,
+      "learning_rate": 6.143986731560761e-06,
+      "loss": 0.3668,
+      "step": 195
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.3532696477038572,
+      "learning_rate": 6.131439326176421e-06,
+      "loss": 0.3881,
+      "step": 196
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.3591118501043564,
+      "learning_rate": 6.118813641468765e-06,
+      "loss": 0.3941,
+      "step": 197
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.3233571068183556,
+      "learning_rate": 6.1061100530231424e-06,
+      "loss": 0.4001,
+      "step": 198
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.33223865496179206,
+      "learning_rate": 6.093328938742357e-06,
+      "loss": 0.3981,
+      "step": 199
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.3012035216026795,
+      "learning_rate": 6.080470678835434e-06,
+      "loss": 0.3896,
+      "step": 200
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.30212477918803177,
+      "learning_rate": 6.067535655806304e-06,
+      "loss": 0.3725,
+      "step": 201
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.3432977226199173,
+      "learning_rate": 6.054524254442424e-06,
+      "loss": 0.3756,
+      "step": 202
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.3196624845591416,
+      "learning_rate": 6.0414368618033354e-06,
+      "loss": 0.3676,
+      "step": 203
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.3327501182697744,
+      "learning_rate": 6.028273867209144e-06,
+      "loss": 0.3722,
+      "step": 204
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.31984015375544,
+      "learning_rate": 6.015035662228943e-06,
+      "loss": 0.3765,
+      "step": 205
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.35134173352065806,
+      "learning_rate": 6.001722640669162e-06,
+      "loss": 0.3762,
+      "step": 206
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.3440468808364095,
+      "learning_rate": 5.988335198561855e-06,
+      "loss": 0.3794,
+      "step": 207
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.34844527663899655,
+      "learning_rate": 5.974873734152916e-06,
+      "loss": 0.41,
+      "step": 208
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.34304312268288334,
+      "learning_rate": 5.961338647890235e-06,
+      "loss": 0.3936,
+      "step": 209
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.30945801228280895,
+      "learning_rate": 5.947730342411785e-06,
+      "loss": 0.3766,
+      "step": 210
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.3216707019915462,
+      "learning_rate": 5.93404922253364e-06,
+      "loss": 0.3731,
+      "step": 211
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.3427933027019099,
+      "learning_rate": 5.9202956952379435e-06,
+      "loss": 0.4022,
+      "step": 212
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.3322640935210238,
+      "learning_rate": 5.9064701696607854e-06,
+      "loss": 0.3974,
+      "step": 213
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.351184635298803,
+      "learning_rate": 5.892573057080049e-06,
+      "loss": 0.3829,
+      "step": 214
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.357169284418552,
+      "learning_rate": 5.878604770903163e-06,
+      "loss": 0.4057,
+      "step": 215
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.3485603587904994,
+      "learning_rate": 5.864565726654811e-06,
+      "loss": 0.3837,
+      "step": 216
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.3299707884038329,
+      "learning_rate": 5.850456341964565e-06,
+      "loss": 0.3866,
+      "step": 217
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.3301963858451784,
+      "learning_rate": 5.836277036554466e-06,
+      "loss": 0.3883,
+      "step": 218
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.35937103442524587,
+      "learning_rate": 5.822028232226539e-06,
+      "loss": 0.3935,
+      "step": 219
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.3143923577131876,
+      "learning_rate": 5.807710352850241e-06,
+      "loss": 0.3672,
+      "step": 220
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.34204735597602554,
+      "learning_rate": 5.793323824349856e-06,
+      "loss": 0.4238,
+      "step": 221
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.350414261224835,
+      "learning_rate": 5.778869074691822e-06,
+      "loss": 0.3761,
+      "step": 222
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.3366168053878449,
+      "learning_rate": 5.764346533872001e-06,
+      "loss": 0.3868,
+      "step": 223
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.3765806993799636,
+      "learning_rate": 5.749756633902887e-06,
+      "loss": 0.4002,
+      "step": 224
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.30582158798586045,
+      "learning_rate": 5.735099808800758e-06,
+      "loss": 0.3551,
+      "step": 225
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.3333315124044687,
+      "learning_rate": 5.720376494572759e-06,
+      "loss": 0.3674,
+      "step": 226
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.37190863363237675,
+      "learning_rate": 5.705587129203936e-06,
+      "loss": 0.4117,
+      "step": 227
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.4030535384641014,
+      "learning_rate": 5.690732152644207e-06,
+      "loss": 0.4165,
+      "step": 228
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.31140781430679837,
+      "learning_rate": 5.675812006795271e-06,
+      "loss": 0.3765,
+      "step": 229
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.3334852992049489,
+      "learning_rate": 5.6608271354974675e-06,
+      "loss": 0.4011,
+      "step": 230
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.3333953874618333,
+      "learning_rate": 5.645777984516568e-06,
+      "loss": 0.3842,
+      "step": 231
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.3484910738096921,
+      "learning_rate": 5.630665001530522e-06,
+      "loss": 0.3883,
+      "step": 232
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.33587659933991937,
+      "learning_rate": 5.615488636116131e-06,
+      "loss": 0.408,
+      "step": 233
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.3481828026794521,
+      "learning_rate": 5.600249339735683e-06,
+      "loss": 0.412,
+      "step": 234
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.32627533208737675,
+      "learning_rate": 5.584947565723517e-06,
+      "loss": 0.3815,
+      "step": 235
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.32794121285389627,
+      "learning_rate": 5.569583769272539e-06,
+      "loss": 0.3944,
+      "step": 236
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.3610564505429431,
+      "learning_rate": 5.554158407420681e-06,
+      "loss": 0.4111,
+      "step": 237
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.3485408935089911,
+      "learning_rate": 5.5386719390373075e-06,
+      "loss": 0.3984,
+      "step": 238
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.3111131027556224,
+      "learning_rate": 5.523124824809562e-06,
+      "loss": 0.3611,
+      "step": 239
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.3212167065940285,
+      "learning_rate": 5.507517527228661e-06,
+      "loss": 0.3902,
+      "step": 240
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.3429494707502764,
+      "learning_rate": 5.4918505105761435e-06,
+      "loss": 0.3988,
+      "step": 241
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.37745564096378414,
+      "learning_rate": 5.476124240910052e-06,
+      "loss": 0.391,
+      "step": 242
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.3370908454585386,
+      "learning_rate": 5.460339186051069e-06,
+      "loss": 0.3832,
+      "step": 243
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.35699424319652345,
+      "learning_rate": 5.444495815568607e-06,
+      "loss": 0.3931,
+      "step": 244
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.32181963953057935,
+      "learning_rate": 5.428594600766834e-06,
+      "loss": 0.366,
+      "step": 245
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.33474503624294333,
+      "learning_rate": 5.412636014670652e-06,
+      "loss": 0.3873,
+      "step": 246
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.3868760538229008,
+      "learning_rate": 5.396620532011631e-06,
+      "loss": 0.4187,
+      "step": 247
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.3265049011735541,
+      "learning_rate": 5.380548629213884e-06,
+      "loss": 0.3632,
+      "step": 248
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.36248572247880945,
+      "learning_rate": 5.364420784379892e-06,
+      "loss": 0.4009,
+      "step": 249
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.34416933574029995,
+      "learning_rate": 5.348237477276288e-06,
+      "loss": 0.3711,
+      "step": 250
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.33413802546232907,
+      "learning_rate": 5.331999189319578e-06,
+      "loss": 0.3664,
+      "step": 251
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.3379009176453982,
+      "learning_rate": 5.315706403561825e-06,
+      "loss": 0.3791,
+      "step": 252
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.3400903622872459,
+      "learning_rate": 5.299359604676275e-06,
+      "loss": 0.383,
+      "step": 253
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.3201446015275036,
+      "learning_rate": 5.282959278942947e-06,
+      "loss": 0.358,
+      "step": 254
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.31420468207920393,
+      "learning_rate": 5.266505914234152e-06,
+      "loss": 0.3759,
+      "step": 255
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.33144165880196313,
+      "learning_rate": 5.25e-06,
+      "loss": 0.3987,
+      "step": 256
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 0.35730368664695833,
+      "learning_rate": 5.233442027253823e-06,
+      "loss": 0.4236,
+      "step": 257
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 0.3204129143253973,
+      "learning_rate": 5.216832488557577e-06,
+      "loss": 0.3715,
+      "step": 258
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 0.3470930852931772,
+      "learning_rate": 5.20017187800719e-06,
+      "loss": 0.3494,
+      "step": 259
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.35698216535228194,
+      "learning_rate": 5.183460691217857e-06,
+      "loss": 0.3806,
+      "step": 260
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 0.36940897857930255,
+      "learning_rate": 5.166699425309303e-06,
+      "loss": 0.3972,
+      "step": 261
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.3248249397484033,
+      "learning_rate": 5.1498885788909926e-06,
+      "loss": 0.367,
+      "step": 262
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.37853950376269563,
+      "learning_rate": 5.133028652047296e-06,
+      "loss": 0.3954,
+      "step": 263
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 0.3903953620153008,
+      "learning_rate": 5.11612014632262e-06,
+      "loss": 0.3683,
+      "step": 264
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.3655129284014845,
+      "learning_rate": 5.099163564706473e-06,
+      "loss": 0.3932,
+      "step": 265
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 0.38542545846265136,
+      "learning_rate": 5.082159411618519e-06,
+      "loss": 0.3716,
+      "step": 266
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 0.3563013091684777,
+      "learning_rate": 5.065108192893563e-06,
+      "loss": 0.3689,
+      "step": 267
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 0.34414564307032347,
+      "learning_rate": 5.048010415766505e-06,
+      "loss": 0.414,
+      "step": 268
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.34099408232504325,
+      "learning_rate": 5.030866588857251e-06,
+      "loss": 0.3825,
+      "step": 269
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 0.32991209225304574,
+      "learning_rate": 5.013677222155581e-06,
+      "loss": 0.3894,
+      "step": 270
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.3481866752789252,
+      "learning_rate": 4.996442827005987e-06,
+      "loss": 0.3921,
+      "step": 271
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.3408069698194307,
+      "learning_rate": 4.979163916092448e-06,
+      "loss": 0.3842,
+      "step": 272
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 0.3521646614948515,
+      "learning_rate": 4.961841003423187e-06,
+      "loss": 0.391,
+      "step": 273
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.3241237346030365,
+      "learning_rate": 4.944474604315381e-06,
+      "loss": 0.3552,
+      "step": 274
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 0.3369201964822902,
+      "learning_rate": 4.927065235379828e-06,
+      "loss": 0.4064,
+      "step": 275
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.33897966935178175,
+      "learning_rate": 4.9096134145055806e-06,
+      "loss": 0.4034,
+      "step": 276
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.34618663822712353,
+      "learning_rate": 4.892119660844538e-06,
+      "loss": 0.3824,
+      "step": 277
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 0.3447506295061284,
+      "learning_rate": 4.87458449479601e-06,
+      "loss": 0.3632,
+      "step": 278
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 0.33961753482937246,
+      "learning_rate": 4.857008437991222e-06,
+      "loss": 0.3628,
+      "step": 279
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.3346934313378748,
+      "learning_rate": 4.8393920132778144e-06,
+      "loss": 0.3386,
+      "step": 280
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 0.3742410827458272,
+      "learning_rate": 4.821735744704276e-06,
+      "loss": 0.3719,
+      "step": 281
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 0.3467492717364462,
+      "learning_rate": 4.804040157504361e-06,
+      "loss": 0.3519,
+      "step": 282
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 0.31898977746556895,
+      "learning_rate": 4.786305778081462e-06,
+      "loss": 0.3499,
+      "step": 283
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 0.3725664879305011,
+      "learning_rate": 4.7685331339929555e-06,
+      "loss": 0.3703,
+      "step": 284
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 0.35775844350818625,
+      "learning_rate": 4.750722753934501e-06,
+      "loss": 0.3452,
+      "step": 285
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 0.33251154393279203,
+      "learning_rate": 4.732875167724318e-06,
+      "loss": 0.3786,
+      "step": 286
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 0.3652428072641884,
+      "learning_rate": 4.714990906287423e-06,
+      "loss": 0.3886,
+      "step": 287
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 0.34153035774512613,
+      "learning_rate": 4.697070501639841e-06,
+      "loss": 0.3803,
+      "step": 288
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.33024253081562993,
+      "learning_rate": 4.67911448687277e-06,
+      "loss": 0.3841,
+      "step": 289
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 0.31352646672532924,
+      "learning_rate": 4.661123396136733e-06,
+      "loss": 0.3575,
+      "step": 290
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 0.3591115191111562,
+      "learning_rate": 4.643097764625678e-06,
+      "loss": 0.3569,
+      "step": 291
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 0.3727369743742984,
+      "learning_rate": 4.625038128561065e-06,
+      "loss": 0.3735,
+      "step": 292
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 0.35276922439582065,
+      "learning_rate": 4.606945025175914e-06,
+      "loss": 0.3558,
+      "step": 293
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.3769733317611369,
+      "learning_rate": 4.588818992698818e-06,
+      "loss": 0.3707,
+      "step": 294
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.33808066630991396,
+      "learning_rate": 4.570660570337937e-06,
+      "loss": 0.3436,
+      "step": 295
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 0.324817784328749,
+      "learning_rate": 4.552470298264956e-06,
+      "loss": 0.3479,
+      "step": 296
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.32662676785729616,
+      "learning_rate": 4.534248717599016e-06,
+      "loss": 0.3255,
+      "step": 297
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 0.326421759505535,
+      "learning_rate": 4.5159963703906175e-06,
+      "loss": 0.3471,
+      "step": 298
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 0.34011208528881004,
+      "learning_rate": 4.497713799605498e-06,
+      "loss": 0.356,
+      "step": 299
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 0.3794880482586398,
+      "learning_rate": 4.479401549108473e-06,
+      "loss": 0.3915,
+      "step": 300
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.3210491323588062,
+      "learning_rate": 4.4610601636472636e-06,
+      "loss": 0.3576,
+      "step": 301
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 0.3419966824370152,
+      "learning_rate": 4.442690188836292e-06,
+      "loss": 0.3596,
+      "step": 302
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.3689177192376415,
+      "learning_rate": 4.424292171140445e-06,
+      "loss": 0.3794,
+      "step": 303
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 0.37058353484814044,
+      "learning_rate": 4.405866657858823e-06,
+      "loss": 0.377,
+      "step": 304
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 0.6109797175104753,
+      "learning_rate": 4.387414197108459e-06,
+      "loss": 0.3364,
+      "step": 305
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.31458714468904536,
+      "learning_rate": 4.368935337808006e-06,
+      "loss": 0.327,
+      "step": 306
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 0.36105992330152553,
+      "learning_rate": 4.350430629661424e-06,
+      "loss": 0.3412,
+      "step": 307
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.3406357854301439,
+      "learning_rate": 4.3319006231416055e-06,
+      "loss": 0.3384,
+      "step": 308
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.3737736032761594,
+      "learning_rate": 4.313345869474022e-06,
+      "loss": 0.3584,
+      "step": 309
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 0.35537041074074627,
+      "learning_rate": 4.294766920620306e-06,
+      "loss": 0.3719,
+      "step": 310
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 0.3409493794346439,
+      "learning_rate": 4.276164329261853e-06,
+      "loss": 0.3573,
+      "step": 311
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.34029967350445056,
+      "learning_rate": 4.25753864878336e-06,
+      "loss": 0.3334,
+      "step": 312
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 0.3513848739355917,
+      "learning_rate": 4.238890433256378e-06,
+      "loss": 0.3486,
+      "step": 313
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 0.3336942534916315,
+      "learning_rate": 4.220220237422822e-06,
+      "loss": 0.3313,
+      "step": 314
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 0.33387764080245036,
+      "learning_rate": 4.201528616678472e-06,
+      "loss": 0.3557,
+      "step": 315
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 0.3881832942635055,
+      "learning_rate": 4.182816127056449e-06,
+      "loss": 0.3486,
+      "step": 316
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 0.32396387681049105,
+      "learning_rate": 4.1640833252106775e-06,
+      "loss": 0.3358,
+      "step": 317
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 0.3498684089335312,
+      "learning_rate": 4.1453307683993216e-06,
+      "loss": 0.3513,
+      "step": 318
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 0.3806936187217104,
+      "learning_rate": 4.1265590144682155e-06,
+      "loss": 0.3617,
+      "step": 319
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 0.33925242861989363,
+      "learning_rate": 4.107768621834257e-06,
+      "loss": 0.3534,
+      "step": 320
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 0.3300810784465674,
+      "learning_rate": 4.088960149468808e-06,
+      "loss": 0.352,
+      "step": 321
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 0.3404645723726401,
+      "learning_rate": 4.070134156881061e-06,
+      "loss": 0.3409,
+      "step": 322
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 0.3655966353045549,
+      "learning_rate": 4.051291204101393e-06,
+      "loss": 0.3196,
+      "step": 323
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 0.37730833103284855,
+      "learning_rate": 4.032431851664708e-06,
+      "loss": 0.3416,
+      "step": 324
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 0.3302002834245634,
+      "learning_rate": 4.013556660593766e-06,
+      "loss": 0.3456,
+      "step": 325
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 0.35855386728187316,
+      "learning_rate": 3.9946661923824864e-06,
+      "loss": 0.3523,
+      "step": 326
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 0.36564829262836096,
+      "learning_rate": 3.97576100897925e-06,
+      "loss": 0.3556,
+      "step": 327
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 0.3668383201799081,
+      "learning_rate": 3.956841672770181e-06,
+      "loss": 0.349,
+      "step": 328
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 0.3138509419345777,
+      "learning_rate": 3.937908746562417e-06,
+      "loss": 0.3345,
+      "step": 329
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 0.3887214100972992,
+      "learning_rate": 3.918962793567368e-06,
+      "loss": 0.3284,
+      "step": 330
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 0.3614318215285855,
+      "learning_rate": 3.900004377383963e-06,
+      "loss": 0.3223,
+      "step": 331
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 0.3357424147189796,
+      "learning_rate": 3.881034061981876e-06,
+      "loss": 0.3262,
+      "step": 332
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 0.3454698621783415,
+      "learning_rate": 3.862052411684763e-06,
+      "loss": 0.3313,
+      "step": 333
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 0.35334068407857966,
+      "learning_rate": 3.843059991153463e-06,
+      "loss": 0.3277,
+      "step": 334
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 0.3918670742282634,
+      "learning_rate": 3.824057365369205e-06,
+      "loss": 0.3399,
+      "step": 335
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 0.3739422718773499,
+      "learning_rate": 3.805045099616804e-06,
+      "loss": 0.3595,
+      "step": 336
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 0.34257555009987084,
+      "learning_rate": 3.786023759467839e-06,
+      "loss": 0.3465,
+      "step": 337
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 0.32802479405729623,
+      "learning_rate": 3.766993910763834e-06,
+      "loss": 0.3369,
+      "step": 338
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 0.3203688904452787,
+      "learning_rate": 3.7479561195994195e-06,
+      "loss": 0.3347,
+      "step": 339
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 0.3410536222408465,
+      "learning_rate": 3.728910952305501e-06,
+      "loss": 0.3582,
+      "step": 340
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 0.31478038592605895,
+      "learning_rate": 3.7098589754324037e-06,
+      "loss": 0.3532,
+      "step": 341
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 0.32415333055330847,
+      "learning_rate": 3.6908007557330225e-06,
+      "loss": 0.3401,
+      "step": 342
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 0.35602242055976896,
+      "learning_rate": 3.6717368601459635e-06,
+      "loss": 0.3601,
+      "step": 343
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 0.3497515154481127,
+      "learning_rate": 3.6526678557786763e-06,
+      "loss": 0.3413,
+      "step": 344
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 0.34412660600849004,
+      "learning_rate": 3.633594309890586e-06,
+      "loss": 0.3452,
+      "step": 345
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 0.33351558637954704,
+      "learning_rate": 3.6145167898762167e-06,
+      "loss": 0.3444,
+      "step": 346
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 0.3419766355994506,
+      "learning_rate": 3.595435863248315e-06,
+      "loss": 0.3471,
+      "step": 347
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 0.3251581136971839,
+      "learning_rate": 3.576352097620964e-06,
+      "loss": 0.3308,
+      "step": 348
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 0.381853253533654,
+      "learning_rate": 3.557266060692704e-06,
+      "loss": 0.3757,
+      "step": 349
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 0.34450659180881255,
+      "learning_rate": 3.5381783202296382e-06,
+      "loss": 0.3323,
+      "step": 350
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 0.3400017029410289,
+      "learning_rate": 3.5190894440485483e-06,
+      "loss": 0.3461,
+      "step": 351
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 0.36714525068941245,
+      "learning_rate": 3.5e-06,
+      "loss": 0.3507,
+      "step": 352
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 0.3113553611505031,
+      "learning_rate": 3.480910555951451e-06,
+      "loss": 0.3146,
+      "step": 353
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 0.3570421820064217,
+      "learning_rate": 3.461821679770362e-06,
+      "loss": 0.3264,
+      "step": 354
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 0.38077478251107827,
+      "learning_rate": 3.442733939307296e-06,
+      "loss": 0.3667,
+      "step": 355
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 0.3743349045152784,
+      "learning_rate": 3.4236479023790363e-06,
+      "loss": 0.3694,
+      "step": 356
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 0.3230112425756345,
+      "learning_rate": 3.4045641367516857e-06,
+      "loss": 0.3344,
+      "step": 357
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.3423897120742137,
+      "learning_rate": 3.3854832101237836e-06,
+      "loss": 0.3606,
+      "step": 358
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.3461281931644015,
+      "learning_rate": 3.366405690109414e-06,
+      "loss": 0.3425,
+      "step": 359
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 0.3432644044960223,
+      "learning_rate": 3.3473321442213245e-06,
+      "loss": 0.3466,
+      "step": 360
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 0.34805031939451947,
+      "learning_rate": 3.328263139854037e-06,
+      "loss": 0.3668,
+      "step": 361
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 0.34793326646142414,
+      "learning_rate": 3.3091992442669774e-06,
+      "loss": 0.3699,
+      "step": 362
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 0.35783347717306163,
+      "learning_rate": 3.290141024567597e-06,
+      "loss": 0.3401,
+      "step": 363
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 0.33765692223581617,
+      "learning_rate": 3.271089047694499e-06,
+      "loss": 0.3502,
+      "step": 364
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 0.358125416429268,
+      "learning_rate": 3.252043880400581e-06,
+      "loss": 0.3677,
+      "step": 365
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 0.3645605208958676,
+      "learning_rate": 3.2330060892361665e-06,
+      "loss": 0.3565,
+      "step": 366
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 0.35182249395160453,
+      "learning_rate": 3.2139762405321623e-06,
+      "loss": 0.3256,
+      "step": 367
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 0.36526279600710565,
+      "learning_rate": 3.1949549003831962e-06,
+      "loss": 0.3496,
+      "step": 368
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 0.3417604543854069,
+      "learning_rate": 3.1759426346307963e-06,
+      "loss": 0.3556,
+      "step": 369
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 0.3538576376640646,
+      "learning_rate": 3.1569400088465375e-06,
+      "loss": 0.3447,
+      "step": 370
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 0.34693771863093653,
+      "learning_rate": 3.137947588315237e-06,
+      "loss": 0.3426,
+      "step": 371
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 0.38481964319346323,
+      "learning_rate": 3.118965938018125e-06,
+      "loss": 0.3498,
+      "step": 372
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 0.34675521547578386,
+      "learning_rate": 3.099995622616037e-06,
+      "loss": 0.3268,
+      "step": 373
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 0.3372531716548693,
+      "learning_rate": 3.0810372064326317e-06,
+      "loss": 0.3461,
+      "step": 374
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 0.38104519894600813,
+      "learning_rate": 3.0620912534375834e-06,
+      "loss": 0.3728,
+      "step": 375
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 0.3466001656961185,
+      "learning_rate": 3.0431583272298204e-06,
+      "loss": 0.3244,
+      "step": 376
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 0.37591315716887963,
+      "learning_rate": 3.0242389910207505e-06,
+      "loss": 0.3521,
+      "step": 377
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 0.3265662120545945,
+      "learning_rate": 3.0053338076175147e-06,
+      "loss": 0.334,
+      "step": 378
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 0.3457254552647638,
+      "learning_rate": 2.986443339406234e-06,
+      "loss": 0.3264,
+      "step": 379
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 0.37047097912111926,
+      "learning_rate": 2.9675681483352915e-06,
+      "loss": 0.3401,
+      "step": 380
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 0.34065288744920735,
+      "learning_rate": 2.948708795898608e-06,
+      "loss": 0.3444,
+      "step": 381
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 0.33705540164600195,
+      "learning_rate": 2.9298658431189395e-06,
+      "loss": 0.3219,
+      "step": 382
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 0.33349856248341736,
+      "learning_rate": 2.9110398505311927e-06,
+      "loss": 0.3414,
+      "step": 383
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.35835514865052853,
+      "learning_rate": 2.8922313781657437e-06,
+      "loss": 0.3586,
+      "step": 384
+    },
+    {
+      "epoch": 3.01,
+      "grad_norm": 0.4112703391832232,
+      "learning_rate": 2.873440985531786e-06,
+      "loss": 0.3742,
+      "step": 385
+    },
+    {
+      "epoch": 3.02,
+      "grad_norm": 0.35218060775962634,
+      "learning_rate": 2.8546692316006783e-06,
+      "loss": 0.333,
+      "step": 386
+    },
+    {
+      "epoch": 3.02,
+      "grad_norm": 0.3272084779142217,
+      "learning_rate": 2.8359166747893237e-06,
+      "loss": 0.3106,
+      "step": 387
+    },
+    {
+      "epoch": 3.03,
+      "grad_norm": 0.35734611520518955,
+      "learning_rate": 2.8171838729435513e-06,
+      "loss": 0.3393,
+      "step": 388
+    },
+    {
+      "epoch": 3.04,
+      "grad_norm": 0.38149060855665573,
+      "learning_rate": 2.7984713833215282e-06,
+      "loss": 0.3512,
+      "step": 389
+    },
+    {
+      "epoch": 3.05,
+      "grad_norm": 0.37164588957773215,
+      "learning_rate": 2.779779762577178e-06,
+      "loss": 0.3247,
+      "step": 390
+    },
+    {
+      "epoch": 3.05,
+      "grad_norm": 0.3722544608089407,
+      "learning_rate": 2.761109566743622e-06,
+      "loss": 0.3492,
+      "step": 391
+    },
+    {
+      "epoch": 3.06,
+      "grad_norm": 0.3934397097005587,
+      "learning_rate": 2.7424613512166398e-06,
+      "loss": 0.327,
+      "step": 392
+    },
+    {
+      "epoch": 3.07,
+      "grad_norm": 0.3654030239595223,
+      "learning_rate": 2.7238356707381474e-06,
+      "loss": 0.3488,
+      "step": 393
+    },
+    {
+      "epoch": 3.08,
+      "grad_norm": 0.3503984335464161,
+      "learning_rate": 2.705233079379694e-06,
+      "loss": 0.3266,
+      "step": 394
+    },
+    {
+      "epoch": 3.09,
+      "grad_norm": 0.32818020407257076,
+      "learning_rate": 2.6866541305259794e-06,
+      "loss": 0.3309,
+      "step": 395
+    },
+    {
+      "epoch": 3.09,
+      "grad_norm": 0.377998199299507,
+      "learning_rate": 2.6680993768583944e-06,
+      "loss": 0.371,
+      "step": 396
+    },
+    {
+      "epoch": 3.1,
+      "grad_norm": 0.34713137652697956,
+      "learning_rate": 2.6495693703385765e-06,
+      "loss": 0.3464,
+      "step": 397
+    },
+    {
+      "epoch": 3.11,
+      "grad_norm": 0.3620196231959417,
+      "learning_rate": 2.6310646621919923e-06,
+      "loss": 0.3519,
+      "step": 398
+    },
+    {
+      "epoch": 3.12,
+      "grad_norm": 0.3652450097403223,
+      "learning_rate": 2.612585802891542e-06,
+      "loss": 0.355,
+      "step": 399
+    },
+    {
+      "epoch": 3.12,
+      "grad_norm": 0.3561392599782829,
+      "learning_rate": 2.594133342141177e-06,
+      "loss": 0.3489,
+      "step": 400
+    },
+    {
+      "epoch": 3.13,
+      "grad_norm": 0.3615565981799769,
+      "learning_rate": 2.575707828859556e-06,
+      "loss": 0.3523,
+      "step": 401
+    },
+    {
+      "epoch": 3.14,
+      "grad_norm": 0.3152156285999487,
+      "learning_rate": 2.5573098111637088e-06,
+      "loss": 0.3232,
+      "step": 402
+    },
+    {
+      "epoch": 3.15,
+      "grad_norm": 0.35211559538338943,
+      "learning_rate": 2.5389398363527376e-06,
+      "loss": 0.3689,
+      "step": 403
+    },
+    {
+      "epoch": 3.16,
+      "grad_norm": 0.3709163488846079,
+      "learning_rate": 2.5205984508915277e-06,
+      "loss": 0.3624,
+      "step": 404
+    },
+    {
+      "epoch": 3.16,
+      "grad_norm": 0.34807484716577025,
+      "learning_rate": 2.502286200394503e-06,
+      "loss": 0.3441,
+      "step": 405
+    },
+    {
+      "epoch": 3.17,
+      "grad_norm": 0.32986214092483723,
+      "learning_rate": 2.4840036296093824e-06,
+      "loss": 0.327,
+      "step": 406
+    },
+    {
+      "epoch": 3.18,
+      "grad_norm": 0.34365285959643166,
+      "learning_rate": 2.465751282400983e-06,
+      "loss": 0.3234,
+      "step": 407
+    },
+    {
+      "epoch": 3.19,
+      "grad_norm": 0.33477375466750364,
+      "learning_rate": 2.4475297017350445e-06,
+      "loss": 0.2972,
+      "step": 408
+    },
+    {
+      "epoch": 3.2,
+      "grad_norm": 0.3530313958823634,
+      "learning_rate": 2.4293394296620625e-06,
+      "loss": 0.3343,
+      "step": 409
+    },
+    {
+      "epoch": 3.2,
+      "grad_norm": 0.3390812879356889,
+      "learning_rate": 2.411181007301182e-06,
+      "loss": 0.3138,
+      "step": 410
+    },
+    {
+      "epoch": 3.21,
+      "grad_norm": 0.33512846317555633,
+      "learning_rate": 2.393054974824086e-06,
+      "loss": 0.3177,
+      "step": 411
+    },
+    {
+      "epoch": 3.22,
+      "grad_norm": 0.3580805399639594,
+      "learning_rate": 2.374961871438935e-06,
+      "loss": 0.3327,
+      "step": 412
+    },
+    {
+      "epoch": 3.23,
+      "grad_norm": 0.3372867599039067,
+      "learning_rate": 2.356902235374323e-06,
+      "loss": 0.3127,
+      "step": 413
+    },
+    {
+      "epoch": 3.23,
+      "grad_norm": 0.33276453513059345,
+      "learning_rate": 2.3388766038632686e-06,
+      "loss": 0.3436,
+      "step": 414
+    },
+    {
+      "epoch": 3.24,
+      "grad_norm": 0.3576415009985178,
+      "learning_rate": 2.32088551312723e-06,
+      "loss": 0.3515,
+      "step": 415
+    },
+    {
+      "epoch": 3.25,
+      "grad_norm": 0.3294166745203506,
+      "learning_rate": 2.3029294983601598e-06,
+      "loss": 0.3458,
+      "step": 416
+    },
+    {
+      "epoch": 3.26,
+      "grad_norm": 0.3419414807485379,
+      "learning_rate": 2.2850090937125775e-06,
+      "loss": 0.3443,
+      "step": 417
+    },
+    {
+      "epoch": 3.27,
+      "grad_norm": 0.32368745015623135,
+      "learning_rate": 2.2671248322756827e-06,
+      "loss": 0.3235,
+      "step": 418
+    },
+    {
+      "epoch": 3.27,
+      "grad_norm": 0.34897803123816196,
+      "learning_rate": 2.2492772460655e-06,
+      "loss": 0.3241,
+      "step": 419
+    },
+    {
+      "epoch": 3.28,
+      "grad_norm": 0.34347437595206803,
+      "learning_rate": 2.2314668660070444e-06,
+      "loss": 0.3363,
+      "step": 420
+    },
+    {
+      "epoch": 3.29,
+      "grad_norm": 0.3594982446723844,
+      "learning_rate": 2.213694221918538e-06,
+      "loss": 0.3187,
+      "step": 421
+    },
+    {
+      "epoch": 3.3,
+      "grad_norm": 0.3447190484440951,
+      "learning_rate": 2.1959598424956397e-06,
+      "loss": 0.3334,
+      "step": 422
+    },
+    {
+      "epoch": 3.3,
+      "grad_norm": 0.32223135788338164,
+      "learning_rate": 2.178264255295725e-06,
+      "loss": 0.3158,
+      "step": 423
+    },
+    {
+      "epoch": 3.31,
+      "grad_norm": 0.33566912281178546,
+      "learning_rate": 2.160607986722186e-06,
+      "loss": 0.3113,
+      "step": 424
+    },
+    {
+      "epoch": 3.32,
+      "grad_norm": 0.33595879664180733,
+      "learning_rate": 2.1429915620087775e-06,
+      "loss": 0.2925,
+      "step": 425
+    },
+    {
+      "epoch": 3.33,
+      "grad_norm": 0.313823398660823,
+      "learning_rate": 2.125415505203991e-06,
+      "loss": 0.3169,
+      "step": 426
+    },
+    {
+      "epoch": 3.34,
+      "grad_norm": 0.3353191261291372,
+      "learning_rate": 2.1078803391554616e-06,
+      "loss": 0.3228,
+      "step": 427
+    },
+    {
+      "epoch": 3.34,
+      "grad_norm": 0.3616603942250458,
+      "learning_rate": 2.09038658549442e-06,
+      "loss": 0.3513,
+      "step": 428
+    },
+    {
+      "epoch": 3.35,
+      "grad_norm": 0.31987262953616186,
+      "learning_rate": 2.0729347646201717e-06,
+      "loss": 0.3256,
+      "step": 429
+    },
+    {
+      "epoch": 3.36,
+      "grad_norm": 0.34523684765216606,
+      "learning_rate": 2.0555253956846186e-06,
+      "loss": 0.3251,
+      "step": 430
+    },
+    {
+      "epoch": 3.37,
+      "grad_norm": 0.3721898929848971,
+      "learning_rate": 2.0381589965768126e-06,
+      "loss": 0.3374,
+      "step": 431
+    },
+    {
+      "epoch": 3.38,
+      "grad_norm": 0.34400955797489746,
+      "learning_rate": 2.0208360839075525e-06,
+      "loss": 0.3399,
+      "step": 432
+    },
+    {
+      "epoch": 3.38,
+      "grad_norm": 0.4080412320465832,
+      "learning_rate": 2.0035571729940133e-06,
+      "loss": 0.3009,
+      "step": 433
+    },
+    {
+      "epoch": 3.39,
+      "grad_norm": 0.32577460503727845,
+      "learning_rate": 1.9863227778444176e-06,
+      "loss": 0.2963,
+      "step": 434
+    },
+    {
+      "epoch": 3.4,
+      "grad_norm": 0.33702369973839313,
+      "learning_rate": 1.9691334111427506e-06,
+      "loss": 0.3057,
+      "step": 435
+    },
+    {
+      "epoch": 3.41,
+      "grad_norm": 0.3327729761550235,
+      "learning_rate": 1.951989584233496e-06,
+      "loss": 0.3026,
+      "step": 436
+    },
+    {
+      "epoch": 3.41,
+      "grad_norm": 0.33776747658890194,
+      "learning_rate": 1.934891807106438e-06,
+      "loss": 0.3255,
+      "step": 437
+    },
+    {
+      "epoch": 3.42,
+      "grad_norm": 0.3376479216926523,
+      "learning_rate": 1.917840588381481e-06,
+      "loss": 0.3361,
+      "step": 438
+    },
+    {
+      "epoch": 3.43,
+      "grad_norm": 0.33956065490995724,
+      "learning_rate": 1.9008364352935276e-06,
+      "loss": 0.3213,
+      "step": 439
+    },
+    {
+      "epoch": 3.44,
+      "grad_norm": 0.3202103047195963,
+      "learning_rate": 1.8838798536773821e-06,
+      "loss": 0.3014,
+      "step": 440
+    },
+    {
+      "epoch": 3.45,
+      "grad_norm": 0.32958945998206785,
+      "learning_rate": 1.8669713479527048e-06,
+      "loss": 0.3201,
+      "step": 441
+    },
+    {
+      "epoch": 3.45,
+      "grad_norm": 0.3266771071234881,
+      "learning_rate": 1.850111421109008e-06,
+      "loss": 0.2984,
+      "step": 442
+    },
+    {
+      "epoch": 3.46,
+      "grad_norm": 0.33932931342987876,
+      "learning_rate": 1.8333005746906976e-06,
+      "loss": 0.3213,
+      "step": 443
+    },
+    {
+      "epoch": 3.47,
+      "grad_norm": 0.35764343040036173,
+      "learning_rate": 1.8165393087821438e-06,
+      "loss": 0.3142,
+      "step": 444
+    },
+    {
+      "epoch": 3.48,
+      "grad_norm": 0.3317934468561935,
+      "learning_rate": 1.7998281219928094e-06,
+      "loss": 0.304,
+      "step": 445
+    },
+    {
+      "epoch": 3.48,
+      "grad_norm": 0.32267754599110376,
+      "learning_rate": 1.783167511442422e-06,
+      "loss": 0.3147,
+      "step": 446
+    },
+    {
+      "epoch": 3.49,
+      "grad_norm": 0.3593732066521806,
+      "learning_rate": 1.7665579727461771e-06,
+      "loss": 0.3226,
+      "step": 447
+    },
+    {
+      "epoch": 3.5,
+      "grad_norm": 0.3399224799991492,
+      "learning_rate": 1.7500000000000008e-06,
+      "loss": 0.3217,
+      "step": 448
+    },
+    {
+      "epoch": 3.51,
+      "grad_norm": 0.33121444418447826,
+      "learning_rate": 1.7334940857658472e-06,
+      "loss": 0.3205,
+      "step": 449
+    },
+    {
+      "epoch": 3.52,
+      "grad_norm": 0.32955427160524653,
+      "learning_rate": 1.7170407210570539e-06,
+      "loss": 0.3102,
+      "step": 450
+    },
+    {
+      "epoch": 3.52,
+      "grad_norm": 0.3362729432822011,
+      "learning_rate": 1.700640395323724e-06,
+      "loss": 0.2848,
+      "step": 451
+    },
+    {
+      "epoch": 3.53,
+      "grad_norm": 0.3452937005495939,
+      "learning_rate": 1.6842935964381741e-06,
+      "loss": 0.3076,
+      "step": 452
+    },
+    {
+      "epoch": 3.54,
+      "grad_norm": 0.3259636359618341,
+      "learning_rate": 1.6680008106804213e-06,
+      "loss": 0.3099,
+      "step": 453
+    },
+    {
+      "epoch": 3.55,
+      "grad_norm": 0.3410314706263269,
+      "learning_rate": 1.651762522723712e-06,
+      "loss": 0.3175,
+      "step": 454
+    },
+    {
+      "epoch": 3.55,
+      "grad_norm": 0.3289205759105589,
+      "learning_rate": 1.6355792156201085e-06,
+      "loss": 0.3245,
+      "step": 455
+    },
+    {
+      "epoch": 3.56,
+      "grad_norm": 0.3494368432511646,
+      "learning_rate": 1.619451370786116e-06,
+      "loss": 0.3194,
+      "step": 456
+    },
+    {
+      "epoch": 3.57,
+      "grad_norm": 0.3373773656095024,
+      "learning_rate": 1.60337946798837e-06,
+      "loss": 0.307,
+      "step": 457
+    },
+    {
+      "epoch": 3.58,
+      "grad_norm": 0.34836673049287364,
+      "learning_rate": 1.5873639853293484e-06,
+      "loss": 0.2942,
+      "step": 458
+    },
+    {
+      "epoch": 3.59,
+      "grad_norm": 0.336243339029711,
+      "learning_rate": 1.5714053992331667e-06,
+      "loss": 0.2902,
+      "step": 459
+    },
+    {
+      "epoch": 3.59,
+      "grad_norm": 0.3336044821385557,
+      "learning_rate": 1.5555041844313931e-06,
+      "loss": 0.2935,
+      "step": 460
+    },
+    {
+      "epoch": 3.6,
+      "grad_norm": 0.3855406448411789,
+      "learning_rate": 1.5396608139489307e-06,
+      "loss": 0.2987,
+      "step": 461
+    },
+    {
+      "epoch": 3.61,
+      "grad_norm": 0.3786473105435691,
+      "learning_rate": 1.5238757590899485e-06,
+      "loss": 0.2932,
+      "step": 462
+    },
+    {
+      "epoch": 3.62,
+      "grad_norm": 0.3439557208763382,
+      "learning_rate": 1.5081494894238554e-06,
+      "loss": 0.3114,
+      "step": 463
+    },
+    {
+      "epoch": 3.62,
+      "grad_norm": 0.3610410823922369,
+      "learning_rate": 1.4924824727713396e-06,
+      "loss": 0.3222,
+      "step": 464
+    },
+    {
+      "epoch": 3.63,
+      "grad_norm": 0.3407026252586559,
+      "learning_rate": 1.4768751751904387e-06,
+      "loss": 0.3128,
+      "step": 465
+    },
+    {
+      "epoch": 3.64,
+      "grad_norm": 0.33809041981905197,
+      "learning_rate": 1.4613280609626928e-06,
+      "loss": 0.3087,
+      "step": 466
+    },
+    {
+      "epoch": 3.65,
+      "grad_norm": 0.35834168771366504,
+      "learning_rate": 1.4458415925793196e-06,
+      "loss": 0.3077,
+      "step": 467
+    },
+    {
+      "epoch": 3.66,
+      "grad_norm": 0.38871459853044665,
+      "learning_rate": 1.4304162307274625e-06,
+      "loss": 0.3272,
+      "step": 468
+    },
+    {
+      "epoch": 3.66,
+      "grad_norm": 0.3463091561520958,
+      "learning_rate": 1.4150524342764833e-06,
+      "loss": 0.3226,
+      "step": 469
+    },
+    {
+      "epoch": 3.67,
+      "grad_norm": 0.33988619370787143,
+      "learning_rate": 1.399750660264317e-06,
+      "loss": 0.3104,
+      "step": 470
+    },
+    {
+      "epoch": 3.68,
+      "grad_norm": 0.34938108700342674,
+      "learning_rate": 1.384511363883869e-06,
+      "loss": 0.3281,
+      "step": 471
+    },
+    {
+      "epoch": 3.69,
+      "grad_norm": 0.33376707095971697,
+      "learning_rate": 1.3693349984694776e-06,
+      "loss": 0.3116,
+      "step": 472
+    },
+    {
+      "epoch": 3.7,
+      "grad_norm": 0.3316263604682683,
+      "learning_rate": 1.3542220154834316e-06,
+      "loss": 0.316,
+      "step": 473
+    },
+    {
+      "epoch": 3.7,
+      "grad_norm": 0.3526534924986156,
+      "learning_rate": 1.339172864502533e-06,
+      "loss": 0.3136,
+      "step": 474
+    },
+    {
+      "epoch": 3.71,
+      "grad_norm": 0.34150892420283485,
+      "learning_rate": 1.32418799320473e-06,
+      "loss": 0.3143,
+      "step": 475
+    },
+    {
+      "epoch": 3.72,
+      "grad_norm": 0.3184132035524764,
+      "learning_rate": 1.3092678473557933e-06,
+      "loss": 0.3056,
+      "step": 476
+    },
+    {
+      "epoch": 3.73,
+      "grad_norm": 0.3507359963811024,
+      "learning_rate": 1.294412870796064e-06,
+      "loss": 0.3424,
+      "step": 477
+    },
+    {
+      "epoch": 3.73,
+      "grad_norm": 0.32578196789023506,
+      "learning_rate": 1.2796235054272411e-06,
+      "loss": 0.3022,
+      "step": 478
+    },
+    {
+      "epoch": 3.74,
+      "grad_norm": 0.32010441528418093,
+      "learning_rate": 1.2649001911992413e-06,
+      "loss": 0.3183,
+      "step": 479
+    },
+    {
+      "epoch": 3.75,
+      "grad_norm": 0.3708179399239553,
+      "learning_rate": 1.2502433660971122e-06,
+      "loss": 0.3167,
+      "step": 480
+    },
+    {
+      "epoch": 3.76,
+      "grad_norm": 0.3127471229910051,
+      "learning_rate": 1.2356534661279994e-06,
+      "loss": 0.2867,
+      "step": 481
+    },
+    {
+      "epoch": 3.77,
+      "grad_norm": 0.34260266839337344,
+      "learning_rate": 1.2211309253081786e-06,
+      "loss": 0.2982,
+      "step": 482
+    },
+    {
+      "epoch": 3.77,
+      "grad_norm": 0.3324510300552272,
+      "learning_rate": 1.2066761756501436e-06,
+      "loss": 0.3354,
+      "step": 483
+    },
+    {
+      "epoch": 3.78,
+      "grad_norm": 0.3596583818265582,
+      "learning_rate": 1.192289647149759e-06,
+      "loss": 0.337,
+      "step": 484
+    },
+    {
+      "epoch": 3.79,
+      "grad_norm": 0.3199547049033629,
+      "learning_rate": 1.1779717677734615e-06,
+      "loss": 0.3055,
+      "step": 485
+    },
+    {
+      "epoch": 3.8,
+      "grad_norm": 0.3263505155905279,
+      "learning_rate": 1.1637229634455348e-06,
+      "loss": 0.3327,
+      "step": 486
+    },
+    {
+      "epoch": 3.8,
+      "grad_norm": 0.340329924281869,
+      "learning_rate": 1.1495436580354353e-06,
+      "loss": 0.3142,
+      "step": 487
+    },
+    {
+      "epoch": 3.81,
+      "grad_norm": 0.3259441664576382,
+      "learning_rate": 1.135434273345189e-06,
+      "loss": 0.3181,
+      "step": 488
+    },
+    {
+      "epoch": 3.82,
+      "grad_norm": 0.3430436145582583,
+      "learning_rate": 1.1213952290968368e-06,
+      "loss": 0.339,
+      "step": 489
+    },
+    {
+      "epoch": 3.83,
+      "grad_norm": 0.3542173756029115,
+      "learning_rate": 1.1074269429199503e-06,
+      "loss": 0.3419,
+      "step": 490
+    },
+    {
+      "epoch": 3.84,
+      "grad_norm": 0.33473973725214246,
+      "learning_rate": 1.093529830339214e-06,
+      "loss": 0.3128,
+      "step": 491
+    },
+    {
+      "epoch": 3.84,
+      "grad_norm": 0.331525816801152,
+      "learning_rate": 1.0797043047620575e-06,
+      "loss": 0.3209,
+      "step": 492
+    },
+    {
+      "epoch": 3.85,
+      "grad_norm": 0.35046397496028414,
+      "learning_rate": 1.0659507774663595e-06,
+      "loss": 0.3387,
+      "step": 493
+    },
+    {
+      "epoch": 3.86,
+      "grad_norm": 0.34176605906283997,
+      "learning_rate": 1.0522696575882148e-06,
+      "loss": 0.328,
+      "step": 494
+    },
+    {
+      "epoch": 3.87,
+      "grad_norm": 0.31267583422913253,
+      "learning_rate": 1.0386613521097656e-06,
+      "loss": 0.3017,
+      "step": 495
+    },
+    {
+      "epoch": 3.88,
+      "grad_norm": 0.34018126548941485,
+      "learning_rate": 1.0251262658470838e-06,
+      "loss": 0.322,
+      "step": 496
+    },
+    {
+      "epoch": 3.88,
+      "grad_norm": 0.33129322774090264,
+      "learning_rate": 1.0116648014381442e-06,
+      "loss": 0.3271,
+      "step": 497
+    },
+    {
+      "epoch": 3.89,
+      "grad_norm": 0.36466718788465824,
+      "learning_rate": 9.982773593308383e-07,
+      "loss": 0.3145,
+      "step": 498
+    },
+    {
+      "epoch": 3.9,
+      "grad_norm": 0.3347289464190966,
+      "learning_rate": 9.849643377710566e-07,
+      "loss": 0.3156,
+      "step": 499
+    },
+    {
+      "epoch": 3.91,
+      "grad_norm": 0.35870097599007505,
+      "learning_rate": 9.717261327908557e-07,
+      "loss": 0.3212,
+      "step": 500
+    },
+    {
+      "epoch": 3.91,
+      "grad_norm": 0.3404942487890118,
+      "learning_rate": 9.585631381966645e-07,
+      "loss": 0.3006,
+      "step": 501
+    },
+    {
+      "epoch": 3.92,
+      "grad_norm": 0.3450349827459326,
+      "learning_rate": 9.454757455575762e-07,
+      "loss": 0.3188,
+      "step": 502
+    },
+    {
+      "epoch": 3.93,
+      "grad_norm": 0.3631808968816027,
+      "learning_rate": 9.324643441936959e-07,
+      "loss": 0.3424,
+      "step": 503
+    },
+    {
+      "epoch": 3.94,
+      "grad_norm": 0.32472675900072473,
+      "learning_rate": 9.195293211645661e-07,
+      "loss": 0.2989,
+      "step": 504
+    },
+    {
+      "epoch": 3.95,
+      "grad_norm": 0.3497374588970107,
+      "learning_rate": 9.066710612576439e-07,
+      "loss": 0.3203,
+      "step": 505
+    },
+    {
+      "epoch": 3.95,
+      "grad_norm": 0.32654566686001346,
+      "learning_rate": 8.938899469768581e-07,
+      "loss": 0.3103,
+      "step": 506
+    },
+    {
+      "epoch": 3.96,
+      "grad_norm": 0.3473781125149438,
+      "learning_rate": 8.811863585312348e-07,
+      "loss": 0.3002,
+      "step": 507
+    },
+    {
+      "epoch": 3.97,
+      "grad_norm": 0.3395838807842396,
+      "learning_rate": 8.685606738235796e-07,
+      "loss": 0.3152,
+      "step": 508
+    },
+    {
+      "epoch": 3.98,
+      "grad_norm": 0.3578724503519501,
+      "learning_rate": 8.560132684392404e-07,
+      "loss": 0.32,
+      "step": 509
+    },
+    {
+      "epoch": 3.98,
+      "grad_norm": 0.3177446511801698,
+      "learning_rate": 8.435445156349334e-07,
+      "loss": 0.2991,
+      "step": 510
+    },
+    {
+      "epoch": 3.99,
+      "grad_norm": 0.34017313516990605,
+      "learning_rate": 8.311547863276417e-07,
+      "loss": 0.3195,
+      "step": 511
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.3411873808454031,
+      "learning_rate": 8.188444490835774e-07,
+      "loss": 0.333,
+      "step": 512
+    },
+    {
+      "epoch": 4.01,
+      "grad_norm": 0.39461744589677294,
+      "learning_rate": 8.066138701072195e-07,
+      "loss": 0.343,
+      "step": 513
+    },
+    {
+      "epoch": 4.02,
+      "grad_norm": 0.3310920364932005,
+      "learning_rate": 7.944634132304205e-07,
+      "loss": 0.3086,
+      "step": 514
+    },
+    {
+      "epoch": 4.02,
+      "grad_norm": 0.3230577112664011,
+      "learning_rate": 7.823934399015856e-07,
+      "loss": 0.2861,
+      "step": 515
+    },
+    {
+      "epoch": 4.03,
+      "grad_norm": 0.33762915771622776,
+      "learning_rate": 7.704043091749143e-07,
+      "loss": 0.3134,
+      "step": 516
+    },
+    {
+      "epoch": 4.04,
+      "grad_norm": 0.3577729507142605,
+      "learning_rate": 7.584963776997237e-07,
+      "loss": 0.3221,
+      "step": 517
+    },
+    {
+      "epoch": 4.05,
+      "grad_norm": 0.33785584510404026,
+      "learning_rate": 7.466699997098405e-07,
+      "loss": 0.2977,
+      "step": 518
+    },
+    {
+      "epoch": 4.05,
+      "grad_norm": 0.3686247902420513,
+      "learning_rate": 7.349255270130589e-07,
+      "loss": 0.3205,
+      "step": 519
+    },
+    {
+      "epoch": 4.06,
+      "grad_norm": 0.3704478433650854,
+      "learning_rate": 7.232633089806773e-07,
+      "loss": 0.3012,
+      "step": 520
+    },
+    {
+      "epoch": 4.07,
+      "grad_norm": 0.3567295734464394,
+      "learning_rate": 7.116836925371055e-07,
+      "loss": 0.3219,
+      "step": 521
+    },
+    {
+      "epoch": 4.08,
+      "grad_norm": 0.3641001366574356,
+      "learning_rate": 7.001870221495463e-07,
+      "loss": 0.2991,
+      "step": 522
+    },
+    {
+      "epoch": 4.09,
+      "grad_norm": 0.3504305641012147,
+      "learning_rate": 6.88773639817743e-07,
+      "loss": 0.3082,
+      "step": 523
+    },
+    {
+      "epoch": 4.09,
+      "grad_norm": 0.3795835236074235,
+      "learning_rate": 6.774438850638107e-07,
+      "loss": 0.345,
+      "step": 524
+    },
+    {
+      "epoch": 4.1,
+      "grad_norm": 0.3466954727567642,
+      "learning_rate": 6.661980949221356e-07,
+      "loss": 0.3249,
+      "step": 525
+    },
+    {
+      "epoch": 4.11,
+      "grad_norm": 0.35305077141361857,
+      "learning_rate": 6.550366039293471e-07,
+      "loss": 0.329,
+      "step": 526
+    },
+    {
+      "epoch": 4.12,
+      "grad_norm": 0.3629881477132717,
+      "learning_rate": 6.439597441143655e-07,
+      "loss": 0.3319,
+      "step": 527
+    },
+    {
+      "epoch": 4.12,
+      "grad_norm": 0.34628496427804784,
+      "learning_rate": 6.329678449885283e-07,
+      "loss": 0.3271,
+      "step": 528
+    },
+    {
+      "epoch": 4.13,
+      "grad_norm": 0.3458963677412187,
+      "learning_rate": 6.22061233535788e-07,
+      "loss": 0.3287,
+      "step": 529
+    },
+    {
+      "epoch": 4.14,
+      "grad_norm": 0.3258984143176478,
+      "learning_rate": 6.112402342029767e-07,
+      "loss": 0.3045,
+      "step": 530
+    },
+    {
+      "epoch": 4.15,
+      "grad_norm": 0.35759231804803177,
+      "learning_rate": 6.005051688901686e-07,
+      "loss": 0.3466,
+      "step": 531
+    },
+    {
+      "epoch": 4.16,
+      "grad_norm": 0.3808260746862728,
+      "learning_rate": 5.898563569410913e-07,
+      "loss": 0.3381,
+      "step": 532
+    },
+    {
+      "epoch": 4.16,
+      "grad_norm": 0.3680045197463959,
+      "learning_rate": 5.79294115133635e-07,
+      "loss": 0.3224,
+      "step": 533
+    },
+    {
+      "epoch": 4.17,
+      "grad_norm": 0.33757707750018473,
+      "learning_rate": 5.688187576704227e-07,
+      "loss": 0.3063,
+      "step": 534
+    },
+    {
+      "epoch": 4.18,
+      "grad_norm": 0.34941820581500316,
+      "learning_rate": 5.584305961694664e-07,
+      "loss": 0.3008,
+      "step": 535
+    },
+    {
+      "epoch": 4.19,
+      "grad_norm": 0.36034623026164275,
+      "learning_rate": 5.481299396549007e-07,
+      "loss": 0.2738,
+      "step": 536
+    },
+    {
+      "epoch": 4.2,
+      "grad_norm": 0.3455392371123542,
+      "learning_rate": 5.379170945477797e-07,
+      "loss": 0.3125,
+      "step": 537
+    },
+    {
+      "epoch": 4.2,
+      "grad_norm": 0.3308793612340533,
+      "learning_rate": 5.277923646569743e-07,
+      "loss": 0.2914,
+      "step": 538
+    },
+    {
+      "epoch": 4.21,
+      "grad_norm": 0.31467610286094216,
+      "learning_rate": 5.177560511701249e-07,
+      "loss": 0.2987,
+      "step": 539
+    },
+    {
+      "epoch": 4.22,
+      "grad_norm": 0.34939004004336754,
+      "learning_rate": 5.078084526446877e-07,
+      "loss": 0.3101,
+      "step": 540
+    },
+    {
+      "epoch": 4.23,
+      "grad_norm": 0.3239983541697057,
+      "learning_rate": 4.97949864999048e-07,
+      "loss": 0.2932,
+      "step": 541
+    },
+    {
+      "epoch": 4.23,
+      "grad_norm": 0.3457088720493278,
+      "learning_rate": 4.881805815037239e-07,
+      "loss": 0.3231,
+      "step": 542
+    },
+    {
+      "epoch": 4.24,
+      "grad_norm": 0.3441443295170719,
+      "learning_rate": 4.785008927726359e-07,
+      "loss": 0.3302,
+      "step": 543
+    },
+    {
+      "epoch": 4.25,
+      "grad_norm": 0.3324111381679956,
+      "learning_rate": 4.6891108675446453e-07,
+      "loss": 0.3265,
+      "step": 544
+    },
+    {
+      "epoch": 4.26,
+      "grad_norm": 0.33951739514591006,
+      "learning_rate": 4.59411448724087e-07,
+      "loss": 0.3223,
+      "step": 545
+    },
+    {
+      "epoch": 4.27,
+      "grad_norm": 0.32817326444431716,
+      "learning_rate": 4.500022612740856e-07,
+      "loss": 0.3045,
+      "step": 546
+    },
+    {
+      "epoch": 4.27,
+      "grad_norm": 0.3250627297920548,
+      "learning_rate": 4.406838043063446e-07,
+      "loss": 0.3059,
+      "step": 547
+    },
+    {
+      "epoch": 4.28,
+      "grad_norm": 0.3378123787646438,
+      "learning_rate": 4.314563550237231e-07,
+      "loss": 0.3157,
+      "step": 548
+    },
+    {
+      "epoch": 4.29,
+      "grad_norm": 0.33855002776019455,
+      "learning_rate": 4.2232018792181037e-07,
+      "loss": 0.2979,
+      "step": 549
+    },
+    {
+      "epoch": 4.3,
+      "grad_norm": 0.3388034403657763,
+      "learning_rate": 4.132755747807577e-07,
+      "loss": 0.3126,
+      "step": 550
+    },
+    {
+      "epoch": 4.3,
+      "grad_norm": 0.31032554781257515,
+      "learning_rate": 4.0432278465719386e-07,
+      "loss": 0.3,
+      "step": 551
+    },
+    {
+      "epoch": 4.31,
+      "grad_norm": 0.31720176424364194,
+      "learning_rate": 3.95462083876224e-07,
+      "loss": 0.2909,
+      "step": 552
+    },
+    {
+      "epoch": 4.32,
+      "grad_norm": 0.3088378496487171,
+      "learning_rate": 3.8669373602350414e-07,
+      "loss": 0.2743,
+      "step": 553
+    },
+    {
+      "epoch": 4.33,
+      "grad_norm": 0.307490909821479,
+      "learning_rate": 3.7801800193740066e-07,
+      "loss": 0.3003,
+      "step": 554
+    },
+    {
+      "epoch": 4.34,
+      "grad_norm": 0.3123422646558358,
+      "learning_rate": 3.6943513970123184e-07,
+      "loss": 0.3044,
+      "step": 555
+    },
+    {
+      "epoch": 4.34,
+      "grad_norm": 0.3560275027298391,
+      "learning_rate": 3.609454046355911e-07,
+      "loss": 0.3296,
+      "step": 556
+    },
+    {
+      "epoch": 4.35,
+      "grad_norm": 0.30395635084459943,
+      "learning_rate": 3.525490492907494e-07,
+      "loss": 0.308,
+      "step": 557
+    },
+    {
+      "epoch": 4.36,
+      "grad_norm": 0.31537748262010745,
+      "learning_rate": 3.442463234391441e-07,
+      "loss": 0.3065,
+      "step": 558
+    },
+    {
+      "epoch": 4.37,
+      "grad_norm": 0.3569650951838116,
+      "learning_rate": 3.3603747406794833e-07,
+      "loss": 0.315,
+      "step": 559
+    },
+    {
+      "epoch": 4.38,
+      "grad_norm": 0.3396558121035842,
+      "learning_rate": 3.279227453717252e-07,
+      "loss": 0.3199,
+      "step": 560
+    },
+    {
+      "epoch": 4.38,
+      "grad_norm": 0.3270067914121413,
+      "learning_rate": 3.1990237874516066e-07,
+      "loss": 0.2812,
+      "step": 561
+    },
+    {
+      "epoch": 4.39,
+      "grad_norm": 0.29552163074015286,
+      "learning_rate": 3.1197661277588436e-07,
+      "loss": 0.2799,
+      "step": 562
+    },
+    {
+      "epoch": 4.4,
+      "grad_norm": 0.3243293757851992,
+      "learning_rate": 3.0414568323737346e-07,
+      "loss": 0.2873,
+      "step": 563
+    },
+    {
+      "epoch": 4.41,
+      "grad_norm": 0.3131415442010758,
+      "learning_rate": 2.964098230819351e-07,
+      "loss": 0.2843,
+      "step": 564
+    },
+    {
+      "epoch": 4.41,
+      "grad_norm": 0.30987006697829866,
+      "learning_rate": 2.887692624337806e-07,
+      "loss": 0.3082,
+      "step": 565
+    },
+    {
+      "epoch": 4.42,
+      "grad_norm": 0.32359530816615634,
+      "learning_rate": 2.812242285821771e-07,
+      "loss": 0.3175,
+      "step": 566
+    },
+    {
+      "epoch": 4.43,
+      "grad_norm": 0.3285661078243146,
+      "learning_rate": 2.7377494597468916e-07,
+      "loss": 0.3028,
+      "step": 567
+    },
+    {
+      "epoch": 4.44,
+      "grad_norm": 0.3120391175324319,
+      "learning_rate": 2.664216362104964e-07,
+      "loss": 0.2848,
+      "step": 568
+    },
+    {
+      "epoch": 4.45,
+      "grad_norm": 0.3254624468839883,
+      "learning_rate": 2.591645180338085e-07,
+      "loss": 0.3057,
+      "step": 569
+    },
+    {
+      "epoch": 4.45,
+      "grad_norm": 0.3066889634179814,
+      "learning_rate": 2.5200380732735444e-07,
+      "loss": 0.2815,
+      "step": 570
+    },
+    {
+      "epoch": 4.46,
+      "grad_norm": 0.3219858581758564,
+      "learning_rate": 2.4493971710595773e-07,
+      "loss": 0.3037,
+      "step": 571
+    },
+    {
+      "epoch": 4.47,
+      "grad_norm": 0.3405241698528944,
+      "learning_rate": 2.3797245751020545e-07,
+      "loss": 0.2966,
+      "step": 572
+    },
+    {
+      "epoch": 4.48,
+      "grad_norm": 0.313000854277894,
+      "learning_rate": 2.3110223580019317e-07,
+      "loss": 0.288,
+      "step": 573
+    },
+    {
+      "epoch": 4.48,
+      "grad_norm": 0.3277910051867206,
+      "learning_rate": 2.2432925634936062e-07,
+      "loss": 0.2967,
+      "step": 574
+    },
+    {
+      "epoch": 4.49,
+      "grad_norm": 0.3602978682407753,
+      "learning_rate": 2.176537206384112e-07,
+      "loss": 0.3031,
+      "step": 575
+    },
+    {
+      "epoch": 4.5,
+      "grad_norm": 0.3237851867770472,
+      "learning_rate": 2.1107582724932088e-07,
+      "loss": 0.3059,
+      "step": 576
+    },
+    {
+      "epoch": 4.51,
+      "grad_norm": 0.32037287639379797,
+      "learning_rate": 2.0459577185942756e-07,
+      "loss": 0.3049,
+      "step": 577
+    },
+    {
+      "epoch": 4.52,
+      "grad_norm": 0.313744879178858,
+      "learning_rate": 1.9821374723561168e-07,
+      "loss": 0.2948,
+      "step": 578
+    },
+    {
+      "epoch": 4.52,
+      "grad_norm": 0.3170159860467313,
+      "learning_rate": 1.9192994322856282e-07,
+      "loss": 0.2679,
+      "step": 579
+    },
+    {
+      "epoch": 4.53,
+      "grad_norm": 0.32920768720751875,
+      "learning_rate": 1.8574454676713047e-07,
+      "loss": 0.291,
+      "step": 580
+    },
+    {
+      "epoch": 4.54,
+      "grad_norm": 0.32724821046512204,
+      "learning_rate": 1.7965774185276317e-07,
+      "loss": 0.2929,
+      "step": 581
+    },
+    {
+      "epoch": 4.55,
+      "grad_norm": 0.33727509928787114,
+      "learning_rate": 1.736697095540361e-07,
+      "loss": 0.3008,
+      "step": 582
+    },
+    {
+      "epoch": 4.55,
+      "grad_norm": 0.32258880398881606,
+      "learning_rate": 1.6778062800126503e-07,
+      "loss": 0.3101,
+      "step": 583
+    },
+    {
+      "epoch": 4.56,
+      "grad_norm": 0.31631420707873364,
+      "learning_rate": 1.6199067238120613e-07,
+      "loss": 0.3056,
+      "step": 584
+    },
+    {
+      "epoch": 4.57,
+      "grad_norm": 0.3103455602146561,
+      "learning_rate": 1.563000149318439e-07,
+      "loss": 0.2941,
+      "step": 585
+    },
+    {
+      "epoch": 4.58,
+      "grad_norm": 0.3278465787187694,
+      "learning_rate": 1.5070882493726911e-07,
+      "loss": 0.2778,
+      "step": 586
+    },
+    {
+      "epoch": 4.59,
+      "grad_norm": 0.3161655577398492,
+      "learning_rate": 1.4521726872264334e-07,
+      "loss": 0.2749,
+      "step": 587
+    },
+    {
+      "epoch": 4.59,
+      "grad_norm": 0.3201062365976047,
+      "learning_rate": 1.398255096492499e-07,
+      "loss": 0.2784,
+      "step": 588
+    },
+    {
+      "epoch": 4.6,
+      "grad_norm": 0.32972141755861767,
+      "learning_rate": 1.3453370810963294e-07,
+      "loss": 0.2836,
+      "step": 589
+    },
+    {
+      "epoch": 4.61,
+      "grad_norm": 0.3328324371601979,
+      "learning_rate": 1.2934202152283052e-07,
+      "loss": 0.2772,
+      "step": 590
+    },
+    {
+      "epoch": 4.62,
+      "grad_norm": 0.3189014403287748,
+      "learning_rate": 1.242506043296871e-07,
+      "loss": 0.2982,
+      "step": 591
+    },
+    {
+      "epoch": 4.62,
+      "grad_norm": 0.34902552630822714,
+      "learning_rate": 1.192596079882613e-07,
+      "loss": 0.3056,
+      "step": 592
+    },
+    {
+      "epoch": 4.63,
+      "grad_norm": 0.33213859814343094,
+      "learning_rate": 1.1436918096932042e-07,
+      "loss": 0.2979,
+      "step": 593
+    },
+    {
+      "epoch": 4.64,
+      "grad_norm": 0.3041910425282244,
+      "learning_rate": 1.095794687519242e-07,
+      "loss": 0.2958,
+      "step": 594
+    },
+    {
+      "epoch": 4.65,
+      "grad_norm": 0.3083108489167821,
+      "learning_rate": 1.0489061381909609e-07,
+      "loss": 0.2949,
+      "step": 595
+    },
+    {
+      "epoch": 4.66,
+      "grad_norm": 0.329428102687414,
+      "learning_rate": 1.0030275565358499e-07,
+      "loss": 0.3127,
+      "step": 596
+    },
+    {
+      "epoch": 4.66,
+      "grad_norm": 0.32701334264131404,
+      "learning_rate": 9.581603073371642e-08,
+      "loss": 0.3087,
+      "step": 597
+    },
+    {
+      "epoch": 4.67,
+      "grad_norm": 0.327187678866886,
+      "learning_rate": 9.143057252933229e-08,
+      "loss": 0.297,
+      "step": 598
+    },
+    {
+      "epoch": 4.68,
+      "grad_norm": 0.3324055738492342,
+      "learning_rate": 8.714651149782038e-08,
+      "loss": 0.3142,
+      "step": 599
+    },
+    {
+      "epoch": 4.69,
+      "grad_norm": 0.3146633682745703,
+      "learning_rate": 8.296397508023323e-08,
+      "loss": 0.2989,
+      "step": 600
+    },
+    {
+      "epoch": 4.7,
+      "grad_norm": 0.3201792191905935,
+      "learning_rate": 7.888308769749875e-08,
+      "loss": 0.3037,
+      "step": 601
+    },
+    {
+      "epoch": 4.7,
+      "grad_norm": 0.3306081284329322,
+      "learning_rate": 7.490397074671583e-08,
+      "loss": 0.3011,
+      "step": 602
+    },
+    {
+      "epoch": 4.71,
+      "grad_norm": 0.3296222930326982,
+      "learning_rate": 7.102674259754693e-08,
+      "loss": 0.3008,
+      "step": 603
+    },
+    {
+      "epoch": 4.72,
+      "grad_norm": 0.31859276472772396,
+      "learning_rate": 6.72515185886935e-08,
+      "loss": 0.2949,
+      "step": 604
+    },
+    {
+      "epoch": 4.73,
+      "grad_norm": 0.3473157998924945,
+      "learning_rate": 6.357841102446649e-08,
+      "loss": 0.3284,
+      "step": 605
+    },
+    {
+      "epoch": 4.73,
+      "grad_norm": 0.325767117495,
+      "learning_rate": 6.000752917144614e-08,
+      "loss": 0.2895,
+      "step": 606
+    },
+    {
+      "epoch": 4.74,
+      "grad_norm": 0.317107385679555,
+      "learning_rate": 5.653897925522877e-08,
+      "loss": 0.3068,
+      "step": 607
+    },
+    {
+      "epoch": 4.75,
+      "grad_norm": 0.34501499728066537,
+      "learning_rate": 5.3172864457271926e-08,
+      "loss": 0.3025,
+      "step": 608
+    },
+    {
+      "epoch": 4.76,
+      "grad_norm": 0.30943095831999046,
+      "learning_rate": 4.990928491181839e-08,
+      "loss": 0.2754,
+      "step": 609
+    },
+    {
+      "epoch": 4.77,
+      "grad_norm": 0.3316466343580564,
+      "learning_rate": 4.674833770292358e-08,
+      "loss": 0.2867,
+      "step": 610
+    },
+    {
+      "epoch": 4.77,
+      "grad_norm": 0.32304558047721826,
+      "learning_rate": 4.369011686156293e-08,
+      "loss": 0.3232,
+      "step": 611
+    },
+    {
+      "epoch": 4.78,
+      "grad_norm": 0.3387512123056888,
+      "learning_rate": 4.073471336283768e-08,
+      "loss": 0.3242,
+      "step": 612
+    },
+    {
+      "epoch": 4.79,
+      "grad_norm": 0.32043866040079066,
+      "learning_rate": 3.788221512326645e-08,
+      "loss": 0.2945,
+      "step": 613
+    },
+    {
+      "epoch": 4.8,
+      "grad_norm": 0.3293121499354337,
+      "learning_rate": 3.5132706998172444e-08,
+      "loss": 0.3219,
+      "step": 614
+    },
+    {
+      "epoch": 4.8,
+      "grad_norm": 0.3264621427885196,
+      "learning_rate": 3.248627077915578e-08,
+      "loss": 0.3031,
+      "step": 615
+    },
+    {
+      "epoch": 4.81,
+      "grad_norm": 0.3293128611402775,
+      "learning_rate": 2.994298519166366e-08,
+      "loss": 0.3072,
+      "step": 616
+    },
+    {
+      "epoch": 4.82,
+      "grad_norm": 0.34414365396850366,
+      "learning_rate": 2.7502925892646135e-08,
+      "loss": 0.3281,
+      "step": 617
+    },
+    {
+      "epoch": 4.83,
+      "grad_norm": 0.33142162835066347,
+      "learning_rate": 2.5166165468307356e-08,
+      "loss": 0.331,
+      "step": 618
+    },
+    {
+      "epoch": 4.84,
+      "grad_norm": 0.3277738639629037,
+      "learning_rate": 2.293277343194472e-08,
+      "loss": 0.3027,
+      "step": 619
+    },
+    {
+      "epoch": 4.84,
+      "grad_norm": 0.3307126527318936,
+      "learning_rate": 2.0802816221881235e-08,
+      "loss": 0.31,
+      "step": 620
+    },
+    {
+      "epoch": 4.85,
+      "grad_norm": 0.3334818389052575,
+      "learning_rate": 1.8776357199490778e-08,
+      "loss": 0.328,
+      "step": 621
+    },
+    {
+      "epoch": 4.86,
+      "grad_norm": 0.33980982527208864,
+      "learning_rate": 1.6853456647311137e-08,
+      "loss": 0.3174,
+      "step": 622
+    },
+    {
+      "epoch": 4.87,
+      "grad_norm": 0.31490770170987176,
+      "learning_rate": 1.5034171767251135e-08,
+      "loss": 0.2928,
+      "step": 623
+    },
+    {
+      "epoch": 4.88,
+      "grad_norm": 0.3317129206040837,
+      "learning_rate": 1.3318556678890592e-08,
+      "loss": 0.3118,
+      "step": 624
+    },
+    {
+      "epoch": 4.88,
+      "grad_norm": 0.32525833347477956,
+      "learning_rate": 1.1706662417868885e-08,
+      "loss": 0.3169,
+      "step": 625
+    },
+    {
+      "epoch": 4.89,
+      "grad_norm": 0.34817461775119474,
+      "learning_rate": 1.0198536934366786e-08,
+      "loss": 0.3034,
+      "step": 626
+    },
+    {
+      "epoch": 4.9,
+      "grad_norm": 0.32179354557665907,
+      "learning_rate": 8.794225091680763e-09,
+      "loss": 0.3057,
+      "step": 627
+    },
+    {
+      "epoch": 4.91,
+      "grad_norm": 0.34168991698735435,
+      "learning_rate": 7.493768664887822e-09,
+      "loss": 0.3106,
+      "step": 628
+    },
+    {
+      "epoch": 4.91,
+      "grad_norm": 0.3143573304513848,
+      "learning_rate": 6.2972063396032336e-09,
+      "loss": 0.291,
+      "step": 629
+    },
+    {
+      "epoch": 4.92,
+      "grad_norm": 0.3282355451726847,
+      "learning_rate": 5.204573710829163e-09,
+      "loss": 0.3088,
+      "step": 630
+    },
+    {
+      "epoch": 4.93,
+      "grad_norm": 0.3518094962496798,
+      "learning_rate": 4.2159032818965825e-09,
+      "loss": 0.3318,
+      "step": 631
+    },
+    {
+      "epoch": 4.94,
+      "grad_norm": 0.3140683067025284,
+      "learning_rate": 3.331224463497706e-09,
+      "loss": 0.29,
+      "step": 632
+    },
+    {
+      "epoch": 4.95,
+      "grad_norm": 0.358966960549393,
+      "learning_rate": 2.5505635728116927e-09,
+      "loss": 0.3095,
+      "step": 633
+    },
+    {
+      "epoch": 4.95,
+      "grad_norm": 0.31663660996902965,
+      "learning_rate": 1.873943832720104e-09,
+      "loss": 0.3019,
+      "step": 634
+    },
+    {
+      "epoch": 4.96,
+      "grad_norm": 0.3270301154008955,
+      "learning_rate": 1.3013853711191237e-09,
+      "loss": 0.291,
+      "step": 635
+    },
+    {
+      "epoch": 4.97,
+      "grad_norm": 0.32925831746753514,
+      "learning_rate": 8.329052203180364e-10,
+      "loss": 0.3063,
+      "step": 636
+    },
+    {
+      "epoch": 4.98,
+      "grad_norm": 0.333316577240123,
+      "learning_rate": 4.685173165336897e-10,
+      "loss": 0.3115,
+      "step": 637
+    },
+    {
+      "epoch": 4.98,
+      "grad_norm": 0.3146179236474361,
+      "learning_rate": 2.0823249947587997e-10,
+      "loss": 0.2914,
+      "step": 638
+    },
+    {
+      "epoch": 4.99,
+      "grad_norm": 0.3399539528453878,
+      "learning_rate": 5.205851202444434e-11,
+      "loss": 0.3122,
+      "step": 639
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.34346878676292253,
+      "learning_rate": 0.0,
+      "loss": 0.3243,
+      "step": 640
+    },
+    {
+      "epoch": 5.0,
+      "step": 640,
+      "total_flos": 0.0,
+      "train_loss": 0.3771400361787528,
+      "train_runtime": 5460.2717,
+      "train_samples_per_second": 15.018,
+      "train_steps_per_second": 0.117
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 640,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 100,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": false,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 0.0,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}