diff --git "a/VPE2/trainer_state.json" "b/VPE2/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/VPE2/trainer_state.json"
@@ -0,0 +1,8625 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.03918367346938775,
+  "eval_steps": 50,
+  "global_step": 1200,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 3.265306122448979e-05,
+      "grad_norm": 12.553070068359375,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 2.3591,
+      "step": 1
+    },
+    {
+      "epoch": 6.530612244897959e-05,
+      "grad_norm": 12.028986930847168,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 2.3144,
+      "step": 2
+    },
+    {
+      "epoch": 9.79591836734694e-05,
+      "grad_norm": 12.335774421691895,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 2.3323,
+      "step": 3
+    },
+    {
+      "epoch": 0.00013061224489795917,
+      "grad_norm": 12.695511817932129,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 2.3651,
+      "step": 4
+    },
+    {
+      "epoch": 0.00016326530612244898,
+      "grad_norm": 12.702936172485352,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 2.3563,
+      "step": 5
+    },
+    {
+      "epoch": 0.0001959183673469388,
+      "grad_norm": 11.689526557922363,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 2.3263,
+      "step": 6
+    },
+    {
+      "epoch": 0.00022857142857142857,
+      "grad_norm": 11.332144737243652,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 2.2861,
+      "step": 7
+    },
+    {
+      "epoch": 0.00026122448979591835,
+      "grad_norm": 11.530237197875977,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 2.2867,
+      "step": 8
+    },
+    {
+      "epoch": 0.0002938775510204082,
+      "grad_norm": 10.193593978881836,
+      "learning_rate": 1.8000000000000001e-06,
+      "loss": 2.2521,
+      "step": 9
+    },
+    {
+      "epoch": 0.00032653061224489796,
+      "grad_norm": 9.98508071899414,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 2.249,
+      "step": 10
+    },
+    {
+      "epoch": 0.00035918367346938774,
+      "grad_norm": 9.584476470947266,
+      "learning_rate": 2.2e-06,
+      "loss": 2.2116,
+      "step": 11
+    },
+    {
+      "epoch": 0.0003918367346938776,
+      "grad_norm": 9.574864387512207,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 2.2343,
+      "step": 12
+    },
+    {
+      "epoch": 0.00042448979591836735,
+      "grad_norm": 8.929769515991211,
+      "learning_rate": 2.6e-06,
+      "loss": 2.1599,
+      "step": 13
+    },
+    {
+      "epoch": 0.00045714285714285713,
+      "grad_norm": 8.262470245361328,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 2.1152,
+      "step": 14
+    },
+    {
+      "epoch": 0.0004897959183673469,
+      "grad_norm": 8.28957748413086,
+      "learning_rate": 3e-06,
+      "loss": 2.1048,
+      "step": 15
+    },
+    {
+      "epoch": 0.0005224489795918367,
+      "grad_norm": 8.160731315612793,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 2.054,
+      "step": 16
+    },
+    {
+      "epoch": 0.0005551020408163266,
+      "grad_norm": 7.5022077560424805,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 2.0559,
+      "step": 17
+    },
+    {
+      "epoch": 0.0005877551020408164,
+      "grad_norm": 7.718921661376953,
+      "learning_rate": 3.6000000000000003e-06,
+      "loss": 2.003,
+      "step": 18
+    },
+    {
+      "epoch": 0.0006204081632653061,
+      "grad_norm": 7.6420392990112305,
+      "learning_rate": 3.8000000000000005e-06,
+      "loss": 1.9646,
+      "step": 19
+    },
+    {
+      "epoch": 0.0006530612244897959,
+      "grad_norm": 7.363818645477295,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 1.918,
+      "step": 20
+    },
+    {
+      "epoch": 0.0006857142857142857,
+      "grad_norm": 7.367091178894043,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 1.9065,
+      "step": 21
+    },
+    {
+      "epoch": 0.0007183673469387755,
+      "grad_norm": 7.201758861541748,
+      "learning_rate": 4.4e-06,
+      "loss": 1.8989,
+      "step": 22
+    },
+    {
+      "epoch": 0.0007510204081632653,
+      "grad_norm": 7.093530654907227,
+      "learning_rate": 4.600000000000001e-06,
+      "loss": 1.8381,
+      "step": 23
+    },
+    {
+      "epoch": 0.0007836734693877551,
+      "grad_norm": 7.22179651260376,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 1.7959,
+      "step": 24
+    },
+    {
+      "epoch": 0.0008163265306122449,
+      "grad_norm": 7.264500141143799,
+      "learning_rate": 5e-06,
+      "loss": 1.7654,
+      "step": 25
+    },
+    {
+      "epoch": 0.0008489795918367347,
+      "grad_norm": 6.931406021118164,
+      "learning_rate": 5.2e-06,
+      "loss": 1.6971,
+      "step": 26
+    },
+    {
+      "epoch": 0.0008816326530612245,
+      "grad_norm": 6.8040547370910645,
+      "learning_rate": 5.400000000000001e-06,
+      "loss": 1.6738,
+      "step": 27
+    },
+    {
+      "epoch": 0.0009142857142857143,
+      "grad_norm": 7.414416313171387,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 1.6131,
+      "step": 28
+    },
+    {
+      "epoch": 0.000946938775510204,
+      "grad_norm": 8.00181770324707,
+      "learning_rate": 5.8e-06,
+      "loss": 1.5743,
+      "step": 29
+    },
+    {
+      "epoch": 0.0009795918367346938,
+      "grad_norm": 8.280529022216797,
+      "learning_rate": 6e-06,
+      "loss": 1.5139,
+      "step": 30
+    },
+    {
+      "epoch": 0.0010122448979591836,
+      "grad_norm": 24.16852378845215,
+      "learning_rate": 6.200000000000001e-06,
+      "loss": 1.4998,
+      "step": 31
+    },
+    {
+      "epoch": 0.0010448979591836734,
+      "grad_norm": 25.735715866088867,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 1.4774,
+      "step": 32
+    },
+    {
+      "epoch": 0.0010775510204081632,
+      "grad_norm": 18.589033126831055,
+      "learning_rate": 6.600000000000001e-06,
+      "loss": 1.3795,
+      "step": 33
+    },
+    {
+      "epoch": 0.0011102040816326532,
+      "grad_norm": 7.6697096824646,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 1.3513,
+      "step": 34
+    },
+    {
+      "epoch": 0.001142857142857143,
+      "grad_norm": 12.835890769958496,
+      "learning_rate": 7e-06,
+      "loss": 1.2908,
+      "step": 35
+    },
+    {
+      "epoch": 0.0011755102040816327,
+      "grad_norm": 12.289154052734375,
+      "learning_rate": 7.2000000000000005e-06,
+      "loss": 1.2354,
+      "step": 36
+    },
+    {
+      "epoch": 0.0012081632653061225,
+      "grad_norm": 5.951991558074951,
+      "learning_rate": 7.4e-06,
+      "loss": 1.2242,
+      "step": 37
+    },
+    {
+      "epoch": 0.0012408163265306123,
+      "grad_norm": 5.6563825607299805,
+      "learning_rate": 7.600000000000001e-06,
+      "loss": 1.2448,
+      "step": 38
+    },
+    {
+      "epoch": 0.001273469387755102,
+      "grad_norm": 5.884244441986084,
+      "learning_rate": 7.800000000000002e-06,
+      "loss": 1.2011,
+      "step": 39
+    },
+    {
+      "epoch": 0.0013061224489795918,
+      "grad_norm": 4.279225826263428,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 1.1616,
+      "step": 40
+    },
+    {
+      "epoch": 0.0013387755102040816,
+      "grad_norm": 4.273413181304932,
+      "learning_rate": 8.2e-06,
+      "loss": 1.1361,
+      "step": 41
+    },
+    {
+      "epoch": 0.0013714285714285714,
+      "grad_norm": 4.514435768127441,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 1.1194,
+      "step": 42
+    },
+    {
+      "epoch": 0.0014040816326530612,
+      "grad_norm": 3.1108927726745605,
+      "learning_rate": 8.6e-06,
+      "loss": 1.1359,
+      "step": 43
+    },
+    {
+      "epoch": 0.001436734693877551,
+      "grad_norm": 3.187659740447998,
+      "learning_rate": 8.8e-06,
+      "loss": 1.1088,
+      "step": 44
+    },
+    {
+      "epoch": 0.0014693877551020407,
+      "grad_norm": 3.0588572025299072,
+      "learning_rate": 9e-06,
+      "loss": 1.0367,
+      "step": 45
+    },
+    {
+      "epoch": 0.0015020408163265305,
+      "grad_norm": 3.0041353702545166,
+      "learning_rate": 9.200000000000002e-06,
+      "loss": 1.065,
+      "step": 46
+    },
+    {
+      "epoch": 0.0015346938775510203,
+      "grad_norm": 3.0067152976989746,
+      "learning_rate": 9.4e-06,
+      "loss": 1.106,
+      "step": 47
+    },
+    {
+      "epoch": 0.0015673469387755103,
+      "grad_norm": 2.8196375370025635,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 1.0463,
+      "step": 48
+    },
+    {
+      "epoch": 0.0016,
+      "grad_norm": 2.7588930130004883,
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 1.0707,
+      "step": 49
+    },
+    {
+      "epoch": 0.0016326530612244899,
+      "grad_norm": 2.776578903198242,
+      "learning_rate": 1e-05,
+      "loss": 1.057,
+      "step": 50
+    },
+    {
+      "epoch": 0.0016326530612244899,
+      "eval_loss": 1.0555493831634521,
+      "eval_runtime": 80.9919,
+      "eval_samples_per_second": 1.235,
+      "eval_steps_per_second": 1.235,
+      "step": 50
+    },
+    {
+      "epoch": 0.0016653061224489796,
+      "grad_norm": 3.0007307529449463,
+      "learning_rate": 9.999967341606794e-06,
+      "loss": 1.0359,
+      "step": 51
+    },
+    {
+      "epoch": 0.0016979591836734694,
+      "grad_norm": 2.820908308029175,
+      "learning_rate": 9.999934683213587e-06,
+      "loss": 1.0659,
+      "step": 52
+    },
+    {
+      "epoch": 0.0017306122448979592,
+      "grad_norm": 3.070078134536743,
+      "learning_rate": 9.999902024820379e-06,
+      "loss": 1.0676,
+      "step": 53
+    },
+    {
+      "epoch": 0.001763265306122449,
+      "grad_norm": 2.826664686203003,
+      "learning_rate": 9.999869366427172e-06,
+      "loss": 1.0396,
+      "step": 54
+    },
+    {
+      "epoch": 0.0017959183673469388,
+      "grad_norm": 2.8928892612457275,
+      "learning_rate": 9.999836708033965e-06,
+      "loss": 1.0043,
+      "step": 55
+    },
+    {
+      "epoch": 0.0018285714285714285,
+      "grad_norm": 2.9964358806610107,
+      "learning_rate": 9.999804049640759e-06,
+      "loss": 1.0335,
+      "step": 56
+    },
+    {
+      "epoch": 0.0018612244897959183,
+      "grad_norm": 2.796006202697754,
+      "learning_rate": 9.999771391247552e-06,
+      "loss": 0.9906,
+      "step": 57
+    },
+    {
+      "epoch": 0.001893877551020408,
+      "grad_norm": 2.638707399368286,
+      "learning_rate": 9.999738732854345e-06,
+      "loss": 1.0347,
+      "step": 58
+    },
+    {
+      "epoch": 0.0019265306122448979,
+      "grad_norm": 2.6315219402313232,
+      "learning_rate": 9.999706074461137e-06,
+      "loss": 1.0213,
+      "step": 59
+    },
+    {
+      "epoch": 0.0019591836734693877,
+      "grad_norm": 2.963063955307007,
+      "learning_rate": 9.99967341606793e-06,
+      "loss": 0.9669,
+      "step": 60
+    },
+    {
+      "epoch": 0.0019918367346938777,
+      "grad_norm": 2.913827657699585,
+      "learning_rate": 9.999640757674723e-06,
+      "loss": 1.013,
+      "step": 61
+    },
+    {
+      "epoch": 0.002024489795918367,
+      "grad_norm": 2.830906391143799,
+      "learning_rate": 9.999608099281516e-06,
+      "loss": 1.0036,
+      "step": 62
+    },
+    {
+      "epoch": 0.002057142857142857,
+      "grad_norm": 2.856300115585327,
+      "learning_rate": 9.99957544088831e-06,
+      "loss": 0.9806,
+      "step": 63
+    },
+    {
+      "epoch": 0.0020897959183673468,
+      "grad_norm": 2.8459157943725586,
+      "learning_rate": 9.999542782495103e-06,
+      "loss": 1.0253,
+      "step": 64
+    },
+    {
+      "epoch": 0.0021224489795918368,
+      "grad_norm": 2.748532772064209,
+      "learning_rate": 9.999510124101896e-06,
+      "loss": 1.0017,
+      "step": 65
+    },
+    {
+      "epoch": 0.0021551020408163263,
+      "grad_norm": 2.910276412963867,
+      "learning_rate": 9.999477465708688e-06,
+      "loss": 0.978,
+      "step": 66
+    },
+    {
+      "epoch": 0.0021877551020408163,
+      "grad_norm": 2.8665144443511963,
+      "learning_rate": 9.99944480731548e-06,
+      "loss": 0.9914,
+      "step": 67
+    },
+    {
+      "epoch": 0.0022204081632653063,
+      "grad_norm": 2.6549315452575684,
+      "learning_rate": 9.999412148922274e-06,
+      "loss": 0.9735,
+      "step": 68
+    },
+    {
+      "epoch": 0.002253061224489796,
+      "grad_norm": 2.7277936935424805,
+      "learning_rate": 9.999379490529067e-06,
+      "loss": 0.9498,
+      "step": 69
+    },
+    {
+      "epoch": 0.002285714285714286,
+      "grad_norm": 2.849747657775879,
+      "learning_rate": 9.99934683213586e-06,
+      "loss": 0.9976,
+      "step": 70
+    },
+    {
+      "epoch": 0.0023183673469387754,
+      "grad_norm": 2.8592312335968018,
+      "learning_rate": 9.999314173742652e-06,
+      "loss": 0.9523,
+      "step": 71
+    },
+    {
+      "epoch": 0.0023510204081632654,
+      "grad_norm": 2.7041122913360596,
+      "learning_rate": 9.999281515349445e-06,
+      "loss": 0.9692,
+      "step": 72
+    },
+    {
+      "epoch": 0.002383673469387755,
+      "grad_norm": 2.717122793197632,
+      "learning_rate": 9.999248856956238e-06,
+      "loss": 0.9491,
+      "step": 73
+    },
+    {
+      "epoch": 0.002416326530612245,
+      "grad_norm": 2.516847610473633,
+      "learning_rate": 9.999216198563032e-06,
+      "loss": 0.9632,
+      "step": 74
+    },
+    {
+      "epoch": 0.0024489795918367346,
+      "grad_norm": 2.766266107559204,
+      "learning_rate": 9.999183540169825e-06,
+      "loss": 0.932,
+      "step": 75
+    },
+    {
+      "epoch": 0.0024816326530612246,
+      "grad_norm": 2.8596270084381104,
+      "learning_rate": 9.999150881776616e-06,
+      "loss": 0.9479,
+      "step": 76
+    },
+    {
+      "epoch": 0.002514285714285714,
+      "grad_norm": 2.6034138202667236,
+      "learning_rate": 9.99911822338341e-06,
+      "loss": 0.9712,
+      "step": 77
+    },
+    {
+      "epoch": 0.002546938775510204,
+      "grad_norm": 2.662513256072998,
+      "learning_rate": 9.999085564990203e-06,
+      "loss": 0.9882,
+      "step": 78
+    },
+    {
+      "epoch": 0.0025795918367346937,
+      "grad_norm": 2.900061845779419,
+      "learning_rate": 9.999052906596996e-06,
+      "loss": 0.9232,
+      "step": 79
+    },
+    {
+      "epoch": 0.0026122448979591837,
+      "grad_norm": 2.7503371238708496,
+      "learning_rate": 9.99902024820379e-06,
+      "loss": 0.9483,
+      "step": 80
+    },
+    {
+      "epoch": 0.0026448979591836737,
+      "grad_norm": 2.61838960647583,
+      "learning_rate": 9.998987589810583e-06,
+      "loss": 0.9104,
+      "step": 81
+    },
+    {
+      "epoch": 0.0026775510204081632,
+      "grad_norm": 2.775689125061035,
+      "learning_rate": 9.998954931417376e-06,
+      "loss": 0.9145,
+      "step": 82
+    },
+    {
+      "epoch": 0.0027102040816326532,
+      "grad_norm": 2.695878744125366,
+      "learning_rate": 9.998922273024167e-06,
+      "loss": 0.9075,
+      "step": 83
+    },
+    {
+      "epoch": 0.002742857142857143,
+      "grad_norm": 2.748655080795288,
+      "learning_rate": 9.99888961463096e-06,
+      "loss": 0.966,
+      "step": 84
+    },
+    {
+      "epoch": 0.002775510204081633,
+      "grad_norm": 2.5372986793518066,
+      "learning_rate": 9.998856956237754e-06,
+      "loss": 0.9541,
+      "step": 85
+    },
+    {
+      "epoch": 0.0028081632653061224,
+      "grad_norm": 2.6666107177734375,
+      "learning_rate": 9.998824297844547e-06,
+      "loss": 0.9678,
+      "step": 86
+    },
+    {
+      "epoch": 0.0028408163265306124,
+      "grad_norm": 2.450654983520508,
+      "learning_rate": 9.99879163945134e-06,
+      "loss": 0.9404,
+      "step": 87
+    },
+    {
+      "epoch": 0.002873469387755102,
+      "grad_norm": 2.6407573223114014,
+      "learning_rate": 9.998758981058134e-06,
+      "loss": 0.9661,
+      "step": 88
+    },
+    {
+      "epoch": 0.002906122448979592,
+      "grad_norm": 2.589488983154297,
+      "learning_rate": 9.998726322664925e-06,
+      "loss": 0.9528,
+      "step": 89
+    },
+    {
+      "epoch": 0.0029387755102040815,
+      "grad_norm": 3.260467767715454,
+      "learning_rate": 9.998693664271718e-06,
+      "loss": 0.9625,
+      "step": 90
+    },
+    {
+      "epoch": 0.0029714285714285715,
+      "grad_norm": 2.8341681957244873,
+      "learning_rate": 9.998661005878512e-06,
+      "loss": 0.9088,
+      "step": 91
+    },
+    {
+      "epoch": 0.003004081632653061,
+      "grad_norm": 3.00934100151062,
+      "learning_rate": 9.998628347485305e-06,
+      "loss": 0.9148,
+      "step": 92
+    },
+    {
+      "epoch": 0.003036734693877551,
+      "grad_norm": 2.8187415599823,
+      "learning_rate": 9.998595689092098e-06,
+      "loss": 0.9028,
+      "step": 93
+    },
+    {
+      "epoch": 0.0030693877551020406,
+      "grad_norm": 2.617946147918701,
+      "learning_rate": 9.99856303069889e-06,
+      "loss": 0.8886,
+      "step": 94
+    },
+    {
+      "epoch": 0.0031020408163265306,
+      "grad_norm": 2.7720468044281006,
+      "learning_rate": 9.998530372305683e-06,
+      "loss": 0.9417,
+      "step": 95
+    },
+    {
+      "epoch": 0.0031346938775510206,
+      "grad_norm": 2.8178319931030273,
+      "learning_rate": 9.998497713912476e-06,
+      "loss": 0.9012,
+      "step": 96
+    },
+    {
+      "epoch": 0.00316734693877551,
+      "grad_norm": 2.9068961143493652,
+      "learning_rate": 9.99846505551927e-06,
+      "loss": 0.9318,
+      "step": 97
+    },
+    {
+      "epoch": 0.0032,
+      "grad_norm": 3.227482557296753,
+      "learning_rate": 9.998432397126063e-06,
+      "loss": 0.8956,
+      "step": 98
+    },
+    {
+      "epoch": 0.0032326530612244897,
+      "grad_norm": 2.8666388988494873,
+      "learning_rate": 9.998399738732854e-06,
+      "loss": 0.9288,
+      "step": 99
+    },
+    {
+      "epoch": 0.0032653061224489797,
+      "grad_norm": 2.861967086791992,
+      "learning_rate": 9.998367080339647e-06,
+      "loss": 0.8923,
+      "step": 100
+    },
+    {
+      "epoch": 0.0032653061224489797,
+      "eval_loss": 0.9108777046203613,
+      "eval_runtime": 73.7729,
+      "eval_samples_per_second": 1.356,
+      "eval_steps_per_second": 1.356,
+      "step": 100
+    },
+    {
+      "epoch": 0.0032979591836734693,
+      "grad_norm": 3.026766777038574,
+      "learning_rate": 9.99833442194644e-06,
+      "loss": 0.9085,
+      "step": 101
+    },
+    {
+      "epoch": 0.0033306122448979593,
+      "grad_norm": 3.0358951091766357,
+      "learning_rate": 9.998301763553234e-06,
+      "loss": 0.9019,
+      "step": 102
+    },
+    {
+      "epoch": 0.003363265306122449,
+      "grad_norm": 2.7646968364715576,
+      "learning_rate": 9.998269105160027e-06,
+      "loss": 0.8954,
+      "step": 103
+    },
+    {
+      "epoch": 0.003395918367346939,
+      "grad_norm": 2.880887746810913,
+      "learning_rate": 9.99823644676682e-06,
+      "loss": 0.9257,
+      "step": 104
+    },
+    {
+      "epoch": 0.0034285714285714284,
+      "grad_norm": 3.1517140865325928,
+      "learning_rate": 9.998203788373614e-06,
+      "loss": 0.8951,
+      "step": 105
+    },
+    {
+      "epoch": 0.0034612244897959184,
+      "grad_norm": 2.7021565437316895,
+      "learning_rate": 9.998171129980407e-06,
+      "loss": 0.9521,
+      "step": 106
+    },
+    {
+      "epoch": 0.003493877551020408,
+      "grad_norm": 2.860952854156494,
+      "learning_rate": 9.998138471587198e-06,
+      "loss": 0.902,
+      "step": 107
+    },
+    {
+      "epoch": 0.003526530612244898,
+      "grad_norm": 3.0253970623016357,
+      "learning_rate": 9.998105813193992e-06,
+      "loss": 0.8935,
+      "step": 108
+    },
+    {
+      "epoch": 0.003559183673469388,
+      "grad_norm": 2.776489734649658,
+      "learning_rate": 9.998073154800785e-06,
+      "loss": 0.9323,
+      "step": 109
+    },
+    {
+      "epoch": 0.0035918367346938775,
+      "grad_norm": 2.3988196849823,
+      "learning_rate": 9.998040496407578e-06,
+      "loss": 0.8951,
+      "step": 110
+    },
+    {
+      "epoch": 0.0036244897959183675,
+      "grad_norm": 2.6600584983825684,
+      "learning_rate": 9.998007838014371e-06,
+      "loss": 0.8913,
+      "step": 111
+    },
+    {
+      "epoch": 0.003657142857142857,
+      "grad_norm": 2.4915781021118164,
+      "learning_rate": 9.997975179621163e-06,
+      "loss": 0.8814,
+      "step": 112
+    },
+    {
+      "epoch": 0.003689795918367347,
+      "grad_norm": 2.7426276206970215,
+      "learning_rate": 9.997942521227956e-06,
+      "loss": 0.9027,
+      "step": 113
+    },
+    {
+      "epoch": 0.0037224489795918366,
+      "grad_norm": 2.80908465385437,
+      "learning_rate": 9.99790986283475e-06,
+      "loss": 0.9033,
+      "step": 114
+    },
+    {
+      "epoch": 0.0037551020408163266,
+      "grad_norm": 2.5127768516540527,
+      "learning_rate": 9.997877204441543e-06,
+      "loss": 0.8886,
+      "step": 115
+    },
+    {
+      "epoch": 0.003787755102040816,
+      "grad_norm": 2.7205052375793457,
+      "learning_rate": 9.997844546048336e-06,
+      "loss": 0.9199,
+      "step": 116
+    },
+    {
+      "epoch": 0.003820408163265306,
+      "grad_norm": 2.6644845008850098,
+      "learning_rate": 9.997811887655127e-06,
+      "loss": 0.8722,
+      "step": 117
+    },
+    {
+      "epoch": 0.0038530612244897957,
+      "grad_norm": 3.05825138092041,
+      "learning_rate": 9.99777922926192e-06,
+      "loss": 0.8926,
+      "step": 118
+    },
+    {
+      "epoch": 0.0038857142857142857,
+      "grad_norm": 2.8610692024230957,
+      "learning_rate": 9.997746570868714e-06,
+      "loss": 0.886,
+      "step": 119
+    },
+    {
+      "epoch": 0.003918367346938775,
+      "grad_norm": 2.8283677101135254,
+      "learning_rate": 9.997713912475507e-06,
+      "loss": 0.8838,
+      "step": 120
+    },
+    {
+      "epoch": 0.003951020408163266,
+      "grad_norm": 3.36906361579895,
+      "learning_rate": 9.9976812540823e-06,
+      "loss": 0.8614,
+      "step": 121
+    },
+    {
+      "epoch": 0.003983673469387755,
+      "grad_norm": 2.949343681335449,
+      "learning_rate": 9.997648595689093e-06,
+      "loss": 0.8915,
+      "step": 122
+    },
+    {
+      "epoch": 0.004016326530612245,
+      "grad_norm": 2.986492156982422,
+      "learning_rate": 9.997615937295885e-06,
+      "loss": 0.9053,
+      "step": 123
+    },
+    {
+      "epoch": 0.004048979591836734,
+      "grad_norm": 2.7720727920532227,
+      "learning_rate": 9.997583278902678e-06,
+      "loss": 0.8915,
+      "step": 124
+    },
+    {
+      "epoch": 0.004081632653061225,
+      "grad_norm": 2.5769472122192383,
+      "learning_rate": 9.997550620509471e-06,
+      "loss": 0.8845,
+      "step": 125
+    },
+    {
+      "epoch": 0.004114285714285714,
+      "grad_norm": 2.9634106159210205,
+      "learning_rate": 9.997517962116265e-06,
+      "loss": 0.9057,
+      "step": 126
+    },
+    {
+      "epoch": 0.004146938775510204,
+      "grad_norm": 2.6103193759918213,
+      "learning_rate": 9.997485303723058e-06,
+      "loss": 0.8843,
+      "step": 127
+    },
+    {
+      "epoch": 0.0041795918367346935,
+      "grad_norm": 2.892089366912842,
+      "learning_rate": 9.997452645329851e-06,
+      "loss": 0.8875,
+      "step": 128
+    },
+    {
+      "epoch": 0.004212244897959184,
+      "grad_norm": 3.0502076148986816,
+      "learning_rate": 9.997419986936644e-06,
+      "loss": 0.8654,
+      "step": 129
+    },
+    {
+      "epoch": 0.0042448979591836735,
+      "grad_norm": 2.968538522720337,
+      "learning_rate": 9.997387328543436e-06,
+      "loss": 0.9211,
+      "step": 130
+    },
+    {
+      "epoch": 0.004277551020408163,
+      "grad_norm": 2.7077767848968506,
+      "learning_rate": 9.99735467015023e-06,
+      "loss": 0.849,
+      "step": 131
+    },
+    {
+      "epoch": 0.004310204081632653,
+      "grad_norm": 2.8962769508361816,
+      "learning_rate": 9.997322011757022e-06,
+      "loss": 0.8844,
+      "step": 132
+    },
+    {
+      "epoch": 0.004342857142857143,
+      "grad_norm": 2.5692780017852783,
+      "learning_rate": 9.997289353363816e-06,
+      "loss": 0.8494,
+      "step": 133
+    },
+    {
+      "epoch": 0.004375510204081633,
+      "grad_norm": 2.603320837020874,
+      "learning_rate": 9.997256694970609e-06,
+      "loss": 0.8994,
+      "step": 134
+    },
+    {
+      "epoch": 0.004408163265306122,
+      "grad_norm": 2.762920618057251,
+      "learning_rate": 9.9972240365774e-06,
+      "loss": 0.8894,
+      "step": 135
+    },
+    {
+      "epoch": 0.004440816326530613,
+      "grad_norm": 2.7908272743225098,
+      "learning_rate": 9.997191378184194e-06,
+      "loss": 0.888,
+      "step": 136
+    },
+    {
+      "epoch": 0.004473469387755102,
+      "grad_norm": 2.405191421508789,
+      "learning_rate": 9.997158719790987e-06,
+      "loss": 0.8953,
+      "step": 137
+    },
+    {
+      "epoch": 0.004506122448979592,
+      "grad_norm": 3.086392879486084,
+      "learning_rate": 9.99712606139778e-06,
+      "loss": 0.9024,
+      "step": 138
+    },
+    {
+      "epoch": 0.004538775510204081,
+      "grad_norm": 3.10508394241333,
+      "learning_rate": 9.997093403004573e-06,
+      "loss": 0.8636,
+      "step": 139
+    },
+    {
+      "epoch": 0.004571428571428572,
+      "grad_norm": 2.7642314434051514,
+      "learning_rate": 9.997060744611365e-06,
+      "loss": 0.8589,
+      "step": 140
+    },
+    {
+      "epoch": 0.004604081632653061,
+      "grad_norm": 2.9966800212860107,
+      "learning_rate": 9.997028086218158e-06,
+      "loss": 0.8564,
+      "step": 141
+    },
+    {
+      "epoch": 0.004636734693877551,
+      "grad_norm": 2.998603582382202,
+      "learning_rate": 9.996995427824951e-06,
+      "loss": 0.8596,
+      "step": 142
+    },
+    {
+      "epoch": 0.0046693877551020405,
+      "grad_norm": 2.815833330154419,
+      "learning_rate": 9.996962769431745e-06,
+      "loss": 0.9006,
+      "step": 143
+    },
+    {
+      "epoch": 0.004702040816326531,
+      "grad_norm": 2.654766798019409,
+      "learning_rate": 9.996930111038538e-06,
+      "loss": 0.8683,
+      "step": 144
+    },
+    {
+      "epoch": 0.0047346938775510205,
+      "grad_norm": 2.9405479431152344,
+      "learning_rate": 9.996897452645331e-06,
+      "loss": 0.8756,
+      "step": 145
+    },
+    {
+      "epoch": 0.00476734693877551,
+      "grad_norm": 3.789085626602173,
+      "learning_rate": 9.996864794252124e-06,
+      "loss": 0.8991,
+      "step": 146
+    },
+    {
+      "epoch": 0.0048,
+      "grad_norm": 3.062678813934326,
+      "learning_rate": 9.996832135858918e-06,
+      "loss": 0.8461,
+      "step": 147
+    },
+    {
+      "epoch": 0.00483265306122449,
+      "grad_norm": 2.656879425048828,
+      "learning_rate": 9.996799477465709e-06,
+      "loss": 0.8755,
+      "step": 148
+    },
+    {
+      "epoch": 0.00486530612244898,
+      "grad_norm": 2.66681170463562,
+      "learning_rate": 9.996766819072502e-06,
+      "loss": 0.8882,
+      "step": 149
+    },
+    {
+      "epoch": 0.004897959183673469,
+      "grad_norm": 2.69744873046875,
+      "learning_rate": 9.996734160679296e-06,
+      "loss": 0.8904,
+      "step": 150
+    },
+    {
+      "epoch": 0.004897959183673469,
+      "eval_loss": 0.8796091675758362,
+      "eval_runtime": 74.0605,
+      "eval_samples_per_second": 1.35,
+      "eval_steps_per_second": 1.35,
+      "step": 150
+    },
+    {
+      "epoch": 0.0049306122448979596,
+      "grad_norm": 2.9235658645629883,
+      "learning_rate": 9.996701502286089e-06,
+      "loss": 0.819,
+      "step": 151
+    },
+    {
+      "epoch": 0.004963265306122449,
+      "grad_norm": 3.2201597690582275,
+      "learning_rate": 9.996668843892882e-06,
+      "loss": 0.8964,
+      "step": 152
+    },
+    {
+      "epoch": 0.004995918367346939,
+      "grad_norm": 2.7854557037353516,
+      "learning_rate": 9.996636185499674e-06,
+      "loss": 0.8886,
+      "step": 153
+    },
+    {
+      "epoch": 0.005028571428571428,
+      "grad_norm": 2.4900546073913574,
+      "learning_rate": 9.996603527106467e-06,
+      "loss": 0.8496,
+      "step": 154
+    },
+    {
+      "epoch": 0.005061224489795919,
+      "grad_norm": 2.7506489753723145,
+      "learning_rate": 9.99657086871326e-06,
+      "loss": 0.9044,
+      "step": 155
+    },
+    {
+      "epoch": 0.005093877551020408,
+      "grad_norm": 2.8616607189178467,
+      "learning_rate": 9.996538210320053e-06,
+      "loss": 0.8685,
+      "step": 156
+    },
+    {
+      "epoch": 0.005126530612244898,
+      "grad_norm": 2.814704656600952,
+      "learning_rate": 9.996505551926847e-06,
+      "loss": 0.8771,
+      "step": 157
+    },
+    {
+      "epoch": 0.005159183673469387,
+      "grad_norm": 3.006065845489502,
+      "learning_rate": 9.996472893533638e-06,
+      "loss": 0.8571,
+      "step": 158
+    },
+    {
+      "epoch": 0.005191836734693878,
+      "grad_norm": 2.821923017501831,
+      "learning_rate": 9.996440235140431e-06,
+      "loss": 0.8629,
+      "step": 159
+    },
+    {
+      "epoch": 0.005224489795918367,
+      "grad_norm": 3.278881072998047,
+      "learning_rate": 9.996407576747225e-06,
+      "loss": 0.893,
+      "step": 160
+    },
+    {
+      "epoch": 0.005257142857142857,
+      "grad_norm": 2.744616985321045,
+      "learning_rate": 9.996374918354018e-06,
+      "loss": 0.8682,
+      "step": 161
+    },
+    {
+      "epoch": 0.005289795918367347,
+      "grad_norm": 2.697544574737549,
+      "learning_rate": 9.996342259960811e-06,
+      "loss": 0.8974,
+      "step": 162
+    },
+    {
+      "epoch": 0.005322448979591837,
+      "grad_norm": 2.9768247604370117,
+      "learning_rate": 9.996309601567604e-06,
+      "loss": 0.8931,
+      "step": 163
+    },
+    {
+      "epoch": 0.0053551020408163265,
+      "grad_norm": 3.027183771133423,
+      "learning_rate": 9.996276943174396e-06,
+      "loss": 0.8897,
+      "step": 164
+    },
+    {
+      "epoch": 0.005387755102040816,
+      "grad_norm": 2.7222235202789307,
+      "learning_rate": 9.996244284781189e-06,
+      "loss": 0.8787,
+      "step": 165
+    },
+    {
+      "epoch": 0.0054204081632653065,
+      "grad_norm": 2.7184181213378906,
+      "learning_rate": 9.996211626387982e-06,
+      "loss": 0.8832,
+      "step": 166
+    },
+    {
+      "epoch": 0.005453061224489796,
+      "grad_norm": 2.8517744541168213,
+      "learning_rate": 9.996178967994776e-06,
+      "loss": 0.8463,
+      "step": 167
+    },
+    {
+      "epoch": 0.005485714285714286,
+      "grad_norm": 3.104189395904541,
+      "learning_rate": 9.996146309601569e-06,
+      "loss": 0.8724,
+      "step": 168
+    },
+    {
+      "epoch": 0.005518367346938775,
+      "grad_norm": 2.9883522987365723,
+      "learning_rate": 9.996113651208362e-06,
+      "loss": 0.8654,
+      "step": 169
+    },
+    {
+      "epoch": 0.005551020408163266,
+      "grad_norm": 3.018421173095703,
+      "learning_rate": 9.996080992815155e-06,
+      "loss": 0.8836,
+      "step": 170
+    },
+    {
+      "epoch": 0.005583673469387755,
+      "grad_norm": 2.795041561126709,
+      "learning_rate": 9.996048334421947e-06,
+      "loss": 0.8843,
+      "step": 171
+    },
+    {
+      "epoch": 0.005616326530612245,
+      "grad_norm": 2.594553232192993,
+      "learning_rate": 9.99601567602874e-06,
+      "loss": 0.8874,
+      "step": 172
+    },
+    {
+      "epoch": 0.005648979591836734,
+      "grad_norm": 2.946117877960205,
+      "learning_rate": 9.995983017635533e-06,
+      "loss": 0.8265,
+      "step": 173
+    },
+    {
+      "epoch": 0.005681632653061225,
+      "grad_norm": 3.060215950012207,
+      "learning_rate": 9.995950359242327e-06,
+      "loss": 0.8872,
+      "step": 174
+    },
+    {
+      "epoch": 0.005714285714285714,
+      "grad_norm": 2.5649561882019043,
+      "learning_rate": 9.99591770084912e-06,
+      "loss": 0.898,
+      "step": 175
+    },
+    {
+      "epoch": 0.005746938775510204,
+      "grad_norm": 2.7028987407684326,
+      "learning_rate": 9.995885042455911e-06,
+      "loss": 0.8168,
+      "step": 176
+    },
+    {
+      "epoch": 0.005779591836734694,
+      "grad_norm": 2.918105125427246,
+      "learning_rate": 9.995852384062705e-06,
+      "loss": 0.9006,
+      "step": 177
+    },
+    {
+      "epoch": 0.005812244897959184,
+      "grad_norm": 3.0559136867523193,
+      "learning_rate": 9.995819725669498e-06,
+      "loss": 0.9106,
+      "step": 178
+    },
+    {
+      "epoch": 0.005844897959183673,
+      "grad_norm": 2.7586793899536133,
+      "learning_rate": 9.995787067276291e-06,
+      "loss": 0.8643,
+      "step": 179
+    },
+    {
+      "epoch": 0.005877551020408163,
+      "grad_norm": 2.6476991176605225,
+      "learning_rate": 9.995754408883084e-06,
+      "loss": 0.8952,
+      "step": 180
+    },
+    {
+      "epoch": 0.005910204081632653,
+      "grad_norm": 2.624241352081299,
+      "learning_rate": 9.995721750489876e-06,
+      "loss": 0.8836,
+      "step": 181
+    },
+    {
+      "epoch": 0.005942857142857143,
+      "grad_norm": 3.0197315216064453,
+      "learning_rate": 9.995689092096669e-06,
+      "loss": 0.87,
+      "step": 182
+    },
+    {
+      "epoch": 0.0059755102040816325,
+      "grad_norm": 2.9282002449035645,
+      "learning_rate": 9.995656433703462e-06,
+      "loss": 0.8904,
+      "step": 183
+    },
+    {
+      "epoch": 0.006008163265306122,
+      "grad_norm": 2.6132211685180664,
+      "learning_rate": 9.995623775310255e-06,
+      "loss": 0.8565,
+      "step": 184
+    },
+    {
+      "epoch": 0.0060408163265306125,
+      "grad_norm": 2.6105284690856934,
+      "learning_rate": 9.995591116917049e-06,
+      "loss": 0.8842,
+      "step": 185
+    },
+    {
+      "epoch": 0.006073469387755102,
+      "grad_norm": 2.868211030960083,
+      "learning_rate": 9.995558458523842e-06,
+      "loss": 0.8575,
+      "step": 186
+    },
+    {
+      "epoch": 0.006106122448979592,
+      "grad_norm": 2.7017822265625,
+      "learning_rate": 9.995525800130633e-06,
+      "loss": 0.8429,
+      "step": 187
+    },
+    {
+      "epoch": 0.006138775510204081,
+      "grad_norm": 2.7130422592163086,
+      "learning_rate": 9.995493141737427e-06,
+      "loss": 0.8451,
+      "step": 188
+    },
+    {
+      "epoch": 0.006171428571428572,
+      "grad_norm": 2.681940793991089,
+      "learning_rate": 9.99546048334422e-06,
+      "loss": 0.866,
+      "step": 189
+    },
+    {
+      "epoch": 0.006204081632653061,
+      "grad_norm": 2.6012067794799805,
+      "learning_rate": 9.995427824951013e-06,
+      "loss": 0.8575,
+      "step": 190
+    },
+    {
+      "epoch": 0.006236734693877551,
+      "grad_norm": 2.7982876300811768,
+      "learning_rate": 9.995395166557806e-06,
+      "loss": 0.8818,
+      "step": 191
+    },
+    {
+      "epoch": 0.006269387755102041,
+      "grad_norm": 2.7027719020843506,
+      "learning_rate": 9.9953625081646e-06,
+      "loss": 0.8638,
+      "step": 192
+    },
+    {
+      "epoch": 0.006302040816326531,
+      "grad_norm": 2.6796092987060547,
+      "learning_rate": 9.995329849771393e-06,
+      "loss": 0.8473,
+      "step": 193
+    },
+    {
+      "epoch": 0.00633469387755102,
+      "grad_norm": 2.882390022277832,
+      "learning_rate": 9.995297191378184e-06,
+      "loss": 0.8854,
+      "step": 194
+    },
+    {
+      "epoch": 0.00636734693877551,
+      "grad_norm": 2.970914125442505,
+      "learning_rate": 9.995264532984978e-06,
+      "loss": 0.8499,
+      "step": 195
+    },
+    {
+      "epoch": 0.0064,
+      "grad_norm": 2.7777442932128906,
+      "learning_rate": 9.995231874591771e-06,
+      "loss": 0.8761,
+      "step": 196
+    },
+    {
+      "epoch": 0.00643265306122449,
+      "grad_norm": 2.6736974716186523,
+      "learning_rate": 9.995199216198564e-06,
+      "loss": 0.8645,
+      "step": 197
+    },
+    {
+      "epoch": 0.006465306122448979,
+      "grad_norm": 2.9803805351257324,
+      "learning_rate": 9.995166557805357e-06,
+      "loss": 0.8257,
+      "step": 198
+    },
+    {
+      "epoch": 0.006497959183673469,
+      "grad_norm": 2.7522194385528564,
+      "learning_rate": 9.995133899412149e-06,
+      "loss": 0.8467,
+      "step": 199
+    },
+    {
+      "epoch": 0.006530612244897959,
+      "grad_norm": 2.6975879669189453,
+      "learning_rate": 9.995101241018942e-06,
+      "loss": 0.872,
+      "step": 200
+    },
+    {
+      "epoch": 0.006530612244897959,
+      "eval_loss": 0.8632190823554993,
+      "eval_runtime": 74.7709,
+      "eval_samples_per_second": 1.337,
+      "eval_steps_per_second": 1.337,
+      "step": 200
+    },
+    {
+      "epoch": 0.006563265306122449,
+      "grad_norm": 2.8385777473449707,
+      "learning_rate": 9.995068582625735e-06,
+      "loss": 0.8592,
+      "step": 201
+    },
+    {
+      "epoch": 0.0065959183673469386,
+      "grad_norm": 2.8501343727111816,
+      "learning_rate": 9.995035924232529e-06,
+      "loss": 0.8516,
+      "step": 202
+    },
+    {
+      "epoch": 0.006628571428571429,
+      "grad_norm": 2.841095447540283,
+      "learning_rate": 9.995003265839322e-06,
+      "loss": 0.8589,
+      "step": 203
+    },
+    {
+      "epoch": 0.0066612244897959185,
+      "grad_norm": 2.7659661769866943,
+      "learning_rate": 9.994970607446113e-06,
+      "loss": 0.8434,
+      "step": 204
+    },
+    {
+      "epoch": 0.006693877551020408,
+      "grad_norm": 2.892019748687744,
+      "learning_rate": 9.994937949052907e-06,
+      "loss": 0.8896,
+      "step": 205
+    },
+    {
+      "epoch": 0.006726530612244898,
+      "grad_norm": 2.7645156383514404,
+      "learning_rate": 9.9949052906597e-06,
+      "loss": 0.8654,
+      "step": 206
+    },
+    {
+      "epoch": 0.006759183673469388,
+      "grad_norm": 2.7840843200683594,
+      "learning_rate": 9.994872632266493e-06,
+      "loss": 0.8558,
+      "step": 207
+    },
+    {
+      "epoch": 0.006791836734693878,
+      "grad_norm": 2.843766212463379,
+      "learning_rate": 9.994839973873286e-06,
+      "loss": 0.8482,
+      "step": 208
+    },
+    {
+      "epoch": 0.006824489795918367,
+      "grad_norm": 2.7724156379699707,
+      "learning_rate": 9.99480731548008e-06,
+      "loss": 0.8664,
+      "step": 209
+    },
+    {
+      "epoch": 0.006857142857142857,
+      "grad_norm": 2.8699095249176025,
+      "learning_rate": 9.994774657086873e-06,
+      "loss": 0.8506,
+      "step": 210
+    },
+    {
+      "epoch": 0.006889795918367347,
+      "grad_norm": 2.988041877746582,
+      "learning_rate": 9.994741998693666e-06,
+      "loss": 0.8781,
+      "step": 211
+    },
+    {
+      "epoch": 0.006922448979591837,
+      "grad_norm": 2.597806930541992,
+      "learning_rate": 9.994709340300458e-06,
+      "loss": 0.8586,
+      "step": 212
+    },
+    {
+      "epoch": 0.006955102040816326,
+      "grad_norm": 2.7888684272766113,
+      "learning_rate": 9.994676681907251e-06,
+      "loss": 0.8431,
+      "step": 213
+    },
+    {
+      "epoch": 0.006987755102040816,
+      "grad_norm": 2.80710768699646,
+      "learning_rate": 9.994644023514044e-06,
+      "loss": 0.8911,
+      "step": 214
+    },
+    {
+      "epoch": 0.007020408163265306,
+      "grad_norm": 2.6957030296325684,
+      "learning_rate": 9.994611365120837e-06,
+      "loss": 0.8626,
+      "step": 215
+    },
+    {
+      "epoch": 0.007053061224489796,
+      "grad_norm": 2.8306329250335693,
+      "learning_rate": 9.99457870672763e-06,
+      "loss": 0.8619,
+      "step": 216
+    },
+    {
+      "epoch": 0.0070857142857142855,
+      "grad_norm": 2.735477924346924,
+      "learning_rate": 9.994546048334422e-06,
+      "loss": 0.8661,
+      "step": 217
+    },
+    {
+      "epoch": 0.007118367346938776,
+      "grad_norm": 2.737856388092041,
+      "learning_rate": 9.994513389941215e-06,
+      "loss": 0.8718,
+      "step": 218
+    },
+    {
+      "epoch": 0.0071510204081632655,
+      "grad_norm": 2.743589401245117,
+      "learning_rate": 9.994480731548009e-06,
+      "loss": 0.8781,
+      "step": 219
+    },
+    {
+      "epoch": 0.007183673469387755,
+      "grad_norm": 2.8569047451019287,
+      "learning_rate": 9.994448073154802e-06,
+      "loss": 0.8512,
+      "step": 220
+    },
+    {
+      "epoch": 0.007216326530612245,
+      "grad_norm": 3.129603385925293,
+      "learning_rate": 9.994415414761595e-06,
+      "loss": 0.8839,
+      "step": 221
+    },
+    {
+      "epoch": 0.007248979591836735,
+      "grad_norm": 2.8460073471069336,
+      "learning_rate": 9.994382756368387e-06,
+      "loss": 0.8568,
+      "step": 222
+    },
+    {
+      "epoch": 0.007281632653061225,
+      "grad_norm": 3.024534225463867,
+      "learning_rate": 9.99435009797518e-06,
+      "loss": 0.8672,
+      "step": 223
+    },
+    {
+      "epoch": 0.007314285714285714,
+      "grad_norm": 3.0231759548187256,
+      "learning_rate": 9.994317439581973e-06,
+      "loss": 0.8899,
+      "step": 224
+    },
+    {
+      "epoch": 0.007346938775510204,
+      "grad_norm": 2.8645284175872803,
+      "learning_rate": 9.994284781188766e-06,
+      "loss": 0.8319,
+      "step": 225
+    },
+    {
+      "epoch": 0.007379591836734694,
+      "grad_norm": 2.741297483444214,
+      "learning_rate": 9.99425212279556e-06,
+      "loss": 0.8726,
+      "step": 226
+    },
+    {
+      "epoch": 0.007412244897959184,
+      "grad_norm": 2.8565313816070557,
+      "learning_rate": 9.994219464402353e-06,
+      "loss": 0.8807,
+      "step": 227
+    },
+    {
+      "epoch": 0.007444897959183673,
+      "grad_norm": 2.9202141761779785,
+      "learning_rate": 9.994186806009144e-06,
+      "loss": 0.8098,
+      "step": 228
+    },
+    {
+      "epoch": 0.007477551020408164,
+      "grad_norm": 2.955820083618164,
+      "learning_rate": 9.994154147615938e-06,
+      "loss": 0.8638,
+      "step": 229
+    },
+    {
+      "epoch": 0.007510204081632653,
+      "grad_norm": 2.7195394039154053,
+      "learning_rate": 9.99412148922273e-06,
+      "loss": 0.8591,
+      "step": 230
+    },
+    {
+      "epoch": 0.007542857142857143,
+      "grad_norm": 2.91768479347229,
+      "learning_rate": 9.994088830829524e-06,
+      "loss": 0.8774,
+      "step": 231
+    },
+    {
+      "epoch": 0.007575510204081632,
+      "grad_norm": 2.7152891159057617,
+      "learning_rate": 9.994056172436317e-06,
+      "loss": 0.8488,
+      "step": 232
+    },
+    {
+      "epoch": 0.007608163265306123,
+      "grad_norm": 2.5756800174713135,
+      "learning_rate": 9.99402351404311e-06,
+      "loss": 0.8598,
+      "step": 233
+    },
+    {
+      "epoch": 0.007640816326530612,
+      "grad_norm": 2.603003978729248,
+      "learning_rate": 9.993990855649904e-06,
+      "loss": 0.8473,
+      "step": 234
+    },
+    {
+      "epoch": 0.007673469387755102,
+      "grad_norm": 2.758294105529785,
+      "learning_rate": 9.993958197256695e-06,
+      "loss": 0.8624,
+      "step": 235
+    },
+    {
+      "epoch": 0.0077061224489795915,
+      "grad_norm": 2.8660902976989746,
+      "learning_rate": 9.993925538863489e-06,
+      "loss": 0.8564,
+      "step": 236
+    },
+    {
+      "epoch": 0.007738775510204082,
+      "grad_norm": 2.8767125606536865,
+      "learning_rate": 9.993892880470282e-06,
+      "loss": 0.8578,
+      "step": 237
+    },
+    {
+      "epoch": 0.0077714285714285715,
+      "grad_norm": 2.906562089920044,
+      "learning_rate": 9.993860222077075e-06,
+      "loss": 0.8383,
+      "step": 238
+    },
+    {
+      "epoch": 0.007804081632653061,
+      "grad_norm": 2.74647855758667,
+      "learning_rate": 9.993827563683868e-06,
+      "loss": 0.8656,
+      "step": 239
+    },
+    {
+      "epoch": 0.00783673469387755,
+      "grad_norm": 3.078082323074341,
+      "learning_rate": 9.99379490529066e-06,
+      "loss": 0.8244,
+      "step": 240
+    },
+    {
+      "epoch": 0.00786938775510204,
+      "grad_norm": 2.7712790966033936,
+      "learning_rate": 9.993762246897453e-06,
+      "loss": 0.8536,
+      "step": 241
+    },
+    {
+      "epoch": 0.007902040816326531,
+      "grad_norm": 2.852389335632324,
+      "learning_rate": 9.993729588504246e-06,
+      "loss": 0.8496,
+      "step": 242
+    },
+    {
+      "epoch": 0.007934693877551021,
+      "grad_norm": 2.7907660007476807,
+      "learning_rate": 9.99369693011104e-06,
+      "loss": 0.8993,
+      "step": 243
+    },
+    {
+      "epoch": 0.00796734693877551,
+      "grad_norm": 2.8219473361968994,
+      "learning_rate": 9.993664271717833e-06,
+      "loss": 0.8652,
+      "step": 244
+    },
+    {
+      "epoch": 0.008,
+      "grad_norm": 2.6678013801574707,
+      "learning_rate": 9.993631613324624e-06,
+      "loss": 0.8636,
+      "step": 245
+    },
+    {
+      "epoch": 0.00803265306122449,
+      "grad_norm": 2.724881887435913,
+      "learning_rate": 9.993598954931417e-06,
+      "loss": 0.8559,
+      "step": 246
+    },
+    {
+      "epoch": 0.00806530612244898,
+      "grad_norm": 2.8275516033172607,
+      "learning_rate": 9.99356629653821e-06,
+      "loss": 0.8646,
+      "step": 247
+    },
+    {
+      "epoch": 0.008097959183673469,
+      "grad_norm": 2.8110673427581787,
+      "learning_rate": 9.993533638145004e-06,
+      "loss": 0.8713,
+      "step": 248
+    },
+    {
+      "epoch": 0.008130612244897958,
+      "grad_norm": 2.652221202850342,
+      "learning_rate": 9.993500979751797e-06,
+      "loss": 0.8373,
+      "step": 249
+    },
+    {
+      "epoch": 0.00816326530612245,
+      "grad_norm": 2.6357979774475098,
+      "learning_rate": 9.99346832135859e-06,
+      "loss": 0.8537,
+      "step": 250
+    },
+    {
+      "epoch": 0.00816326530612245,
+      "eval_loss": 0.8545005917549133,
+      "eval_runtime": 74.0963,
+      "eval_samples_per_second": 1.35,
+      "eval_steps_per_second": 1.35,
+      "step": 250
+    },
+    {
+      "epoch": 0.00819591836734694,
+      "grad_norm": 2.6132781505584717,
+      "learning_rate": 9.993435662965384e-06,
+      "loss": 0.8592,
+      "step": 251
+    },
+    {
+      "epoch": 0.008228571428571429,
+      "grad_norm": 2.942228078842163,
+      "learning_rate": 9.993403004572177e-06,
+      "loss": 0.7732,
+      "step": 252
+    },
+    {
+      "epoch": 0.008261224489795918,
+      "grad_norm": 2.5452983379364014,
+      "learning_rate": 9.993370346178968e-06,
+      "loss": 0.8666,
+      "step": 253
+    },
+    {
+      "epoch": 0.008293877551020408,
+      "grad_norm": 2.863976240158081,
+      "learning_rate": 9.993337687785762e-06,
+      "loss": 0.8232,
+      "step": 254
+    },
+    {
+      "epoch": 0.008326530612244898,
+      "grad_norm": 2.640972137451172,
+      "learning_rate": 9.993305029392555e-06,
+      "loss": 0.8703,
+      "step": 255
+    },
+    {
+      "epoch": 0.008359183673469387,
+      "grad_norm": 2.98362135887146,
+      "learning_rate": 9.993272370999348e-06,
+      "loss": 0.8165,
+      "step": 256
+    },
+    {
+      "epoch": 0.008391836734693878,
+      "grad_norm": 2.848294258117676,
+      "learning_rate": 9.993239712606141e-06,
+      "loss": 0.9104,
+      "step": 257
+    },
+    {
+      "epoch": 0.008424489795918368,
+      "grad_norm": 2.7992308139801025,
+      "learning_rate": 9.993207054212933e-06,
+      "loss": 0.8146,
+      "step": 258
+    },
+    {
+      "epoch": 0.008457142857142858,
+      "grad_norm": 2.905052900314331,
+      "learning_rate": 9.993174395819726e-06,
+      "loss": 0.8606,
+      "step": 259
+    },
+    {
+      "epoch": 0.008489795918367347,
+      "grad_norm": 2.97420334815979,
+      "learning_rate": 9.99314173742652e-06,
+      "loss": 0.7909,
+      "step": 260
+    },
+    {
+      "epoch": 0.008522448979591837,
+      "grad_norm": 2.7139410972595215,
+      "learning_rate": 9.993109079033313e-06,
+      "loss": 0.8593,
+      "step": 261
+    },
+    {
+      "epoch": 0.008555102040816326,
+      "grad_norm": 2.7178266048431396,
+      "learning_rate": 9.993076420640106e-06,
+      "loss": 0.8785,
+      "step": 262
+    },
+    {
+      "epoch": 0.008587755102040816,
+      "grad_norm": 2.6858327388763428,
+      "learning_rate": 9.993043762246897e-06,
+      "loss": 0.846,
+      "step": 263
+    },
+    {
+      "epoch": 0.008620408163265305,
+      "grad_norm": 2.813204288482666,
+      "learning_rate": 9.99301110385369e-06,
+      "loss": 0.8386,
+      "step": 264
+    },
+    {
+      "epoch": 0.008653061224489797,
+      "grad_norm": 2.795591115951538,
+      "learning_rate": 9.992978445460484e-06,
+      "loss": 0.8417,
+      "step": 265
+    },
+    {
+      "epoch": 0.008685714285714286,
+      "grad_norm": 2.8094005584716797,
+      "learning_rate": 9.992945787067277e-06,
+      "loss": 0.8396,
+      "step": 266
+    },
+    {
+      "epoch": 0.008718367346938776,
+      "grad_norm": 2.723392963409424,
+      "learning_rate": 9.99291312867407e-06,
+      "loss": 0.8357,
+      "step": 267
+    },
+    {
+      "epoch": 0.008751020408163265,
+      "grad_norm": 2.9623281955718994,
+      "learning_rate": 9.992880470280864e-06,
+      "loss": 0.8874,
+      "step": 268
+    },
+    {
+      "epoch": 0.008783673469387755,
+      "grad_norm": 2.8696374893188477,
+      "learning_rate": 9.992847811887655e-06,
+      "loss": 0.8595,
+      "step": 269
+    },
+    {
+      "epoch": 0.008816326530612244,
+      "grad_norm": 2.906726837158203,
+      "learning_rate": 9.992815153494448e-06,
+      "loss": 0.8736,
+      "step": 270
+    },
+    {
+      "epoch": 0.008848979591836734,
+      "grad_norm": 2.814854383468628,
+      "learning_rate": 9.992782495101242e-06,
+      "loss": 0.8603,
+      "step": 271
+    },
+    {
+      "epoch": 0.008881632653061225,
+      "grad_norm": 2.799457550048828,
+      "learning_rate": 9.992749836708035e-06,
+      "loss": 0.8469,
+      "step": 272
+    },
+    {
+      "epoch": 0.008914285714285715,
+      "grad_norm": 2.755631923675537,
+      "learning_rate": 9.992717178314828e-06,
+      "loss": 0.8702,
+      "step": 273
+    },
+    {
+      "epoch": 0.008946938775510204,
+      "grad_norm": 2.9200055599212646,
+      "learning_rate": 9.992684519921621e-06,
+      "loss": 0.8407,
+      "step": 274
+    },
+    {
+      "epoch": 0.008979591836734694,
+      "grad_norm": 3.2251176834106445,
+      "learning_rate": 9.992651861528415e-06,
+      "loss": 0.8572,
+      "step": 275
+    },
+    {
+      "epoch": 0.009012244897959184,
+      "grad_norm": 3.8850901126861572,
+      "learning_rate": 9.992619203135206e-06,
+      "loss": 0.8458,
+      "step": 276
+    },
+    {
+      "epoch": 0.009044897959183673,
+      "grad_norm": 2.8082637786865234,
+      "learning_rate": 9.992586544742e-06,
+      "loss": 0.8415,
+      "step": 277
+    },
+    {
+      "epoch": 0.009077551020408163,
+      "grad_norm": 2.8923559188842773,
+      "learning_rate": 9.992553886348793e-06,
+      "loss": 0.8116,
+      "step": 278
+    },
+    {
+      "epoch": 0.009110204081632652,
+      "grad_norm": 2.849003791809082,
+      "learning_rate": 9.992521227955586e-06,
+      "loss": 0.834,
+      "step": 279
+    },
+    {
+      "epoch": 0.009142857142857144,
+      "grad_norm": 2.7201125621795654,
+      "learning_rate": 9.992488569562379e-06,
+      "loss": 0.8518,
+      "step": 280
+    },
+    {
+      "epoch": 0.009175510204081633,
+      "grad_norm": 2.9798922538757324,
+      "learning_rate": 9.99245591116917e-06,
+      "loss": 0.8691,
+      "step": 281
+    },
+    {
+      "epoch": 0.009208163265306123,
+      "grad_norm": 2.805250406265259,
+      "learning_rate": 9.992423252775964e-06,
+      "loss": 0.8537,
+      "step": 282
+    },
+    {
+      "epoch": 0.009240816326530612,
+      "grad_norm": 2.6364924907684326,
+      "learning_rate": 9.992390594382757e-06,
+      "loss": 0.8368,
+      "step": 283
+    },
+    {
+      "epoch": 0.009273469387755102,
+      "grad_norm": 2.692296028137207,
+      "learning_rate": 9.99235793598955e-06,
+      "loss": 0.8373,
+      "step": 284
+    },
+    {
+      "epoch": 0.009306122448979591,
+      "grad_norm": 2.876549005508423,
+      "learning_rate": 9.992325277596344e-06,
+      "loss": 0.8348,
+      "step": 285
+    },
+    {
+      "epoch": 0.009338775510204081,
+      "grad_norm": 2.77553129196167,
+      "learning_rate": 9.992292619203135e-06,
+      "loss": 0.8377,
+      "step": 286
+    },
+    {
+      "epoch": 0.009371428571428572,
+      "grad_norm": 2.900949478149414,
+      "learning_rate": 9.992259960809928e-06,
+      "loss": 0.8469,
+      "step": 287
+    },
+    {
+      "epoch": 0.009404081632653062,
+      "grad_norm": 2.99700927734375,
+      "learning_rate": 9.992227302416722e-06,
+      "loss": 0.8926,
+      "step": 288
+    },
+    {
+      "epoch": 0.009436734693877551,
+      "grad_norm": 2.738802194595337,
+      "learning_rate": 9.992194644023515e-06,
+      "loss": 0.8511,
+      "step": 289
+    },
+    {
+      "epoch": 0.009469387755102041,
+      "grad_norm": 2.7471158504486084,
+      "learning_rate": 9.992161985630308e-06,
+      "loss": 0.8449,
+      "step": 290
+    },
+    {
+      "epoch": 0.00950204081632653,
+      "grad_norm": 2.7274839878082275,
+      "learning_rate": 9.992129327237101e-06,
+      "loss": 0.8308,
+      "step": 291
+    },
+    {
+      "epoch": 0.00953469387755102,
+      "grad_norm": 2.554326057434082,
+      "learning_rate": 9.992096668843893e-06,
+      "loss": 0.868,
+      "step": 292
+    },
+    {
+      "epoch": 0.00956734693877551,
+      "grad_norm": 2.578237295150757,
+      "learning_rate": 9.992064010450688e-06,
+      "loss": 0.8557,
+      "step": 293
+    },
+    {
+      "epoch": 0.0096,
+      "grad_norm": 3.0283281803131104,
+      "learning_rate": 9.99203135205748e-06,
+      "loss": 0.8289,
+      "step": 294
+    },
+    {
+      "epoch": 0.00963265306122449,
+      "grad_norm": 2.785170316696167,
+      "learning_rate": 9.991998693664272e-06,
+      "loss": 0.878,
+      "step": 295
+    },
+    {
+      "epoch": 0.00966530612244898,
+      "grad_norm": 2.8442277908325195,
+      "learning_rate": 9.991966035271066e-06,
+      "loss": 0.8282,
+      "step": 296
+    },
+    {
+      "epoch": 0.00969795918367347,
+      "grad_norm": 2.847158908843994,
+      "learning_rate": 9.991933376877859e-06,
+      "loss": 0.844,
+      "step": 297
+    },
+    {
+      "epoch": 0.00973061224489796,
+      "grad_norm": 2.6960904598236084,
+      "learning_rate": 9.991900718484652e-06,
+      "loss": 0.863,
+      "step": 298
+    },
+    {
+      "epoch": 0.009763265306122449,
+      "grad_norm": 2.97739839553833,
+      "learning_rate": 9.991868060091444e-06,
+      "loss": 0.8448,
+      "step": 299
+    },
+    {
+      "epoch": 0.009795918367346938,
+      "grad_norm": 2.9316720962524414,
+      "learning_rate": 9.991835401698237e-06,
+      "loss": 0.8512,
+      "step": 300
+    },
+    {
+      "epoch": 0.009795918367346938,
+      "eval_loss": 0.8451775312423706,
+      "eval_runtime": 73.1659,
+      "eval_samples_per_second": 1.367,
+      "eval_steps_per_second": 1.367,
+      "step": 300
+    },
+    {
+      "epoch": 0.009828571428571428,
+      "grad_norm": 3.2117464542388916,
+      "learning_rate": 9.99180274330503e-06,
+      "loss": 0.852,
+      "step": 301
+    },
+    {
+      "epoch": 0.009861224489795919,
+      "grad_norm": 2.7478814125061035,
+      "learning_rate": 9.991770084911823e-06,
+      "loss": 0.8454,
+      "step": 302
+    },
+    {
+      "epoch": 0.009893877551020409,
+      "grad_norm": 2.7774980068206787,
+      "learning_rate": 9.991737426518617e-06,
+      "loss": 0.8581,
+      "step": 303
+    },
+    {
+      "epoch": 0.009926530612244898,
+      "grad_norm": 2.691710948944092,
+      "learning_rate": 9.991704768125408e-06,
+      "loss": 0.8483,
+      "step": 304
+    },
+    {
+      "epoch": 0.009959183673469388,
+      "grad_norm": 2.803921937942505,
+      "learning_rate": 9.991672109732201e-06,
+      "loss": 0.8394,
+      "step": 305
+    },
+    {
+      "epoch": 0.009991836734693877,
+      "grad_norm": 2.9755606651306152,
+      "learning_rate": 9.991639451338995e-06,
+      "loss": 0.8245,
+      "step": 306
+    },
+    {
+      "epoch": 0.010024489795918367,
+      "grad_norm": 3.09531569480896,
+      "learning_rate": 9.991606792945788e-06,
+      "loss": 0.8701,
+      "step": 307
+    },
+    {
+      "epoch": 0.010057142857142857,
+      "grad_norm": 2.8512625694274902,
+      "learning_rate": 9.991574134552581e-06,
+      "loss": 0.8597,
+      "step": 308
+    },
+    {
+      "epoch": 0.010089795918367346,
+      "grad_norm": 2.7423555850982666,
+      "learning_rate": 9.991541476159374e-06,
+      "loss": 0.8116,
+      "step": 309
+    },
+    {
+      "epoch": 0.010122448979591837,
+      "grad_norm": 2.8539059162139893,
+      "learning_rate": 9.991508817766166e-06,
+      "loss": 0.841,
+      "step": 310
+    },
+    {
+      "epoch": 0.010155102040816327,
+      "grad_norm": 2.843325138092041,
+      "learning_rate": 9.99147615937296e-06,
+      "loss": 0.8491,
+      "step": 311
+    },
+    {
+      "epoch": 0.010187755102040816,
+      "grad_norm": 2.6367223262786865,
+      "learning_rate": 9.991443500979752e-06,
+      "loss": 0.8452,
+      "step": 312
+    },
+    {
+      "epoch": 0.010220408163265306,
+      "grad_norm": 2.6435811519622803,
+      "learning_rate": 9.991410842586546e-06,
+      "loss": 0.8415,
+      "step": 313
+    },
+    {
+      "epoch": 0.010253061224489796,
+      "grad_norm": 2.727060317993164,
+      "learning_rate": 9.991378184193339e-06,
+      "loss": 0.7985,
+      "step": 314
+    },
+    {
+      "epoch": 0.010285714285714285,
+      "grad_norm": 2.9816014766693115,
+      "learning_rate": 9.991345525800132e-06,
+      "loss": 0.8129,
+      "step": 315
+    },
+    {
+      "epoch": 0.010318367346938775,
+      "grad_norm": 3.3140499591827393,
+      "learning_rate": 9.991312867406925e-06,
+      "loss": 0.8822,
+      "step": 316
+    },
+    {
+      "epoch": 0.010351020408163266,
+      "grad_norm": 2.8612375259399414,
+      "learning_rate": 9.991280209013717e-06,
+      "loss": 0.7966,
+      "step": 317
+    },
+    {
+      "epoch": 0.010383673469387756,
+      "grad_norm": 2.692972183227539,
+      "learning_rate": 9.99124755062051e-06,
+      "loss": 0.8288,
+      "step": 318
+    },
+    {
+      "epoch": 0.010416326530612245,
+      "grad_norm": 2.980607271194458,
+      "learning_rate": 9.991214892227303e-06,
+      "loss": 0.8238,
+      "step": 319
+    },
+    {
+      "epoch": 0.010448979591836735,
+      "grad_norm": 3.0936498641967773,
+      "learning_rate": 9.991182233834097e-06,
+      "loss": 0.8338,
+      "step": 320
+    },
+    {
+      "epoch": 0.010481632653061224,
+      "grad_norm": 2.7482964992523193,
+      "learning_rate": 9.99114957544089e-06,
+      "loss": 0.8511,
+      "step": 321
+    },
+    {
+      "epoch": 0.010514285714285714,
+      "grad_norm": 2.6563820838928223,
+      "learning_rate": 9.991116917047681e-06,
+      "loss": 0.8568,
+      "step": 322
+    },
+    {
+      "epoch": 0.010546938775510203,
+      "grad_norm": 2.7284178733825684,
+      "learning_rate": 9.991084258654475e-06,
+      "loss": 0.8615,
+      "step": 323
+    },
+    {
+      "epoch": 0.010579591836734695,
+      "grad_norm": 2.7720155715942383,
+      "learning_rate": 9.991051600261268e-06,
+      "loss": 0.8459,
+      "step": 324
+    },
+    {
+      "epoch": 0.010612244897959184,
+      "grad_norm": 2.6759071350097656,
+      "learning_rate": 9.991018941868061e-06,
+      "loss": 0.8489,
+      "step": 325
+    },
+    {
+      "epoch": 0.010644897959183674,
+      "grad_norm": 2.8001279830932617,
+      "learning_rate": 9.990986283474854e-06,
+      "loss": 0.83,
+      "step": 326
+    },
+    {
+      "epoch": 0.010677551020408163,
+      "grad_norm": 2.7164299488067627,
+      "learning_rate": 9.990953625081646e-06,
+      "loss": 0.8059,
+      "step": 327
+    },
+    {
+      "epoch": 0.010710204081632653,
+      "grad_norm": 2.833035707473755,
+      "learning_rate": 9.990920966688439e-06,
+      "loss": 0.8394,
+      "step": 328
+    },
+    {
+      "epoch": 0.010742857142857143,
+      "grad_norm": 2.6928770542144775,
+      "learning_rate": 9.990888308295232e-06,
+      "loss": 0.8297,
+      "step": 329
+    },
+    {
+      "epoch": 0.010775510204081632,
+      "grad_norm": 2.7575488090515137,
+      "learning_rate": 9.990855649902026e-06,
+      "loss": 0.8149,
+      "step": 330
+    },
+    {
+      "epoch": 0.010808163265306122,
+      "grad_norm": 2.7353413105010986,
+      "learning_rate": 9.990822991508819e-06,
+      "loss": 0.839,
+      "step": 331
+    },
+    {
+      "epoch": 0.010840816326530613,
+      "grad_norm": 2.7132623195648193,
+      "learning_rate": 9.990790333115612e-06,
+      "loss": 0.8452,
+      "step": 332
+    },
+    {
+      "epoch": 0.010873469387755103,
+      "grad_norm": 2.8830623626708984,
+      "learning_rate": 9.990757674722404e-06,
+      "loss": 0.8156,
+      "step": 333
+    },
+    {
+      "epoch": 0.010906122448979592,
+      "grad_norm": 2.7021992206573486,
+      "learning_rate": 9.990725016329197e-06,
+      "loss": 0.8368,
+      "step": 334
+    },
+    {
+      "epoch": 0.010938775510204082,
+      "grad_norm": 2.6805949211120605,
+      "learning_rate": 9.99069235793599e-06,
+      "loss": 0.8402,
+      "step": 335
+    },
+    {
+      "epoch": 0.010971428571428571,
+      "grad_norm": 3.0374648571014404,
+      "learning_rate": 9.990659699542783e-06,
+      "loss": 0.8332,
+      "step": 336
+    },
+    {
+      "epoch": 0.01100408163265306,
+      "grad_norm": 2.9531807899475098,
+      "learning_rate": 9.990627041149577e-06,
+      "loss": 0.8451,
+      "step": 337
+    },
+    {
+      "epoch": 0.01103673469387755,
+      "grad_norm": 2.8179969787597656,
+      "learning_rate": 9.99059438275637e-06,
+      "loss": 0.8336,
+      "step": 338
+    },
+    {
+      "epoch": 0.011069387755102042,
+      "grad_norm": 3.139448642730713,
+      "learning_rate": 9.990561724363163e-06,
+      "loss": 0.865,
+      "step": 339
+    },
+    {
+      "epoch": 0.011102040816326531,
+      "grad_norm": 3.040363073348999,
+      "learning_rate": 9.990529065969955e-06,
+      "loss": 0.8387,
+      "step": 340
+    },
+    {
+      "epoch": 0.01113469387755102,
+      "grad_norm": 2.700171709060669,
+      "learning_rate": 9.990496407576748e-06,
+      "loss": 0.8219,
+      "step": 341
+    },
+    {
+      "epoch": 0.01116734693877551,
+      "grad_norm": 2.7326748371124268,
+      "learning_rate": 9.990463749183541e-06,
+      "loss": 0.7956,
+      "step": 342
+    },
+    {
+      "epoch": 0.0112,
+      "grad_norm": 2.7744908332824707,
+      "learning_rate": 9.990431090790334e-06,
+      "loss": 0.8133,
+      "step": 343
+    },
+    {
+      "epoch": 0.01123265306122449,
+      "grad_norm": 2.8565139770507812,
+      "learning_rate": 9.990398432397128e-06,
+      "loss": 0.825,
+      "step": 344
+    },
+    {
+      "epoch": 0.011265306122448979,
+      "grad_norm": 2.9909799098968506,
+      "learning_rate": 9.990365774003919e-06,
+      "loss": 0.8683,
+      "step": 345
+    },
+    {
+      "epoch": 0.011297959183673469,
+      "grad_norm": 2.642660140991211,
+      "learning_rate": 9.990333115610712e-06,
+      "loss": 0.8598,
+      "step": 346
+    },
+    {
+      "epoch": 0.01133061224489796,
+      "grad_norm": 2.7017788887023926,
+      "learning_rate": 9.990300457217506e-06,
+      "loss": 0.8164,
+      "step": 347
+    },
+    {
+      "epoch": 0.01136326530612245,
+      "grad_norm": 2.7391109466552734,
+      "learning_rate": 9.990267798824299e-06,
+      "loss": 0.8498,
+      "step": 348
+    },
+    {
+      "epoch": 0.011395918367346939,
+      "grad_norm": 2.7465527057647705,
+      "learning_rate": 9.990235140431092e-06,
+      "loss": 0.8225,
+      "step": 349
+    },
+    {
+      "epoch": 0.011428571428571429,
+      "grad_norm": 2.6583974361419678,
+      "learning_rate": 9.990202482037884e-06,
+      "loss": 0.8505,
+      "step": 350
+    },
+    {
+      "epoch": 0.011428571428571429,
+      "eval_loss": 0.8395382165908813,
+      "eval_runtime": 73.8611,
+      "eval_samples_per_second": 1.354,
+      "eval_steps_per_second": 1.354,
+      "step": 350
+    },
+    {
+      "epoch": 0.011461224489795918,
+      "grad_norm": 2.7981696128845215,
+      "learning_rate": 9.990169823644677e-06,
+      "loss": 0.8312,
+      "step": 351
+    },
+    {
+      "epoch": 0.011493877551020408,
+      "grad_norm": 2.924536943435669,
+      "learning_rate": 9.99013716525147e-06,
+      "loss": 0.8498,
+      "step": 352
+    },
+    {
+      "epoch": 0.011526530612244897,
+      "grad_norm": 2.819368362426758,
+      "learning_rate": 9.990104506858263e-06,
+      "loss": 0.8022,
+      "step": 353
+    },
+    {
+      "epoch": 0.011559183673469389,
+      "grad_norm": 2.9508090019226074,
+      "learning_rate": 9.990071848465056e-06,
+      "loss": 0.8363,
+      "step": 354
+    },
+    {
+      "epoch": 0.011591836734693878,
+      "grad_norm": 2.643820285797119,
+      "learning_rate": 9.99003919007185e-06,
+      "loss": 0.8172,
+      "step": 355
+    },
+    {
+      "epoch": 0.011624489795918368,
+      "grad_norm": 2.5882935523986816,
+      "learning_rate": 9.990006531678641e-06,
+      "loss": 0.8095,
+      "step": 356
+    },
+    {
+      "epoch": 0.011657142857142857,
+      "grad_norm": 2.968686103820801,
+      "learning_rate": 9.989973873285436e-06,
+      "loss": 0.8673,
+      "step": 357
+    },
+    {
+      "epoch": 0.011689795918367347,
+      "grad_norm": 2.8162810802459717,
+      "learning_rate": 9.989941214892228e-06,
+      "loss": 0.8693,
+      "step": 358
+    },
+    {
+      "epoch": 0.011722448979591836,
+      "grad_norm": 2.8215367794036865,
+      "learning_rate": 9.989908556499021e-06,
+      "loss": 0.8053,
+      "step": 359
+    },
+    {
+      "epoch": 0.011755102040816326,
+      "grad_norm": 2.7967352867126465,
+      "learning_rate": 9.989875898105814e-06,
+      "loss": 0.8094,
+      "step": 360
+    },
+    {
+      "epoch": 0.011787755102040815,
+      "grad_norm": 2.873476505279541,
+      "learning_rate": 9.989843239712607e-06,
+      "loss": 0.8174,
+      "step": 361
+    },
+    {
+      "epoch": 0.011820408163265307,
+      "grad_norm": 2.8655476570129395,
+      "learning_rate": 9.9898105813194e-06,
+      "loss": 0.7988,
+      "step": 362
+    },
+    {
+      "epoch": 0.011853061224489796,
+      "grad_norm": 2.794279098510742,
+      "learning_rate": 9.989777922926192e-06,
+      "loss": 0.8404,
+      "step": 363
+    },
+    {
+      "epoch": 0.011885714285714286,
+      "grad_norm": 2.8534135818481445,
+      "learning_rate": 9.989745264532985e-06,
+      "loss": 0.8474,
+      "step": 364
+    },
+    {
+      "epoch": 0.011918367346938775,
+      "grad_norm": 2.716606855392456,
+      "learning_rate": 9.989712606139779e-06,
+      "loss": 0.8256,
+      "step": 365
+    },
+    {
+      "epoch": 0.011951020408163265,
+      "grad_norm": 2.8670265674591064,
+      "learning_rate": 9.989679947746572e-06,
+      "loss": 0.834,
+      "step": 366
+    },
+    {
+      "epoch": 0.011983673469387755,
+      "grad_norm": 2.9551408290863037,
+      "learning_rate": 9.989647289353365e-06,
+      "loss": 0.8295,
+      "step": 367
+    },
+    {
+      "epoch": 0.012016326530612244,
+      "grad_norm": 3.209449529647827,
+      "learning_rate": 9.989614630960157e-06,
+      "loss": 0.8167,
+      "step": 368
+    },
+    {
+      "epoch": 0.012048979591836735,
+      "grad_norm": 2.8796558380126953,
+      "learning_rate": 9.98958197256695e-06,
+      "loss": 0.8245,
+      "step": 369
+    },
+    {
+      "epoch": 0.012081632653061225,
+      "grad_norm": 2.9785306453704834,
+      "learning_rate": 9.989549314173743e-06,
+      "loss": 0.7946,
+      "step": 370
+    },
+    {
+      "epoch": 0.012114285714285715,
+      "grad_norm": 2.8382294178009033,
+      "learning_rate": 9.989516655780536e-06,
+      "loss": 0.8392,
+      "step": 371
+    },
+    {
+      "epoch": 0.012146938775510204,
+      "grad_norm": 3.2182369232177734,
+      "learning_rate": 9.98948399738733e-06,
+      "loss": 0.8343,
+      "step": 372
+    },
+    {
+      "epoch": 0.012179591836734694,
+      "grad_norm": 2.9380509853363037,
+      "learning_rate": 9.989451338994123e-06,
+      "loss": 0.8417,
+      "step": 373
+    },
+    {
+      "epoch": 0.012212244897959183,
+      "grad_norm": 2.6060733795166016,
+      "learning_rate": 9.989418680600914e-06,
+      "loss": 0.8471,
+      "step": 374
+    },
+    {
+      "epoch": 0.012244897959183673,
+      "grad_norm": 3.0596020221710205,
+      "learning_rate": 9.989386022207708e-06,
+      "loss": 0.8408,
+      "step": 375
+    },
+    {
+      "epoch": 0.012277551020408162,
+      "grad_norm": 2.91416335105896,
+      "learning_rate": 9.989353363814501e-06,
+      "loss": 0.8648,
+      "step": 376
+    },
+    {
+      "epoch": 0.012310204081632654,
+      "grad_norm": 2.8798558712005615,
+      "learning_rate": 9.989320705421294e-06,
+      "loss": 0.7871,
+      "step": 377
+    },
+    {
+      "epoch": 0.012342857142857143,
+      "grad_norm": 2.9666454792022705,
+      "learning_rate": 9.989288047028087e-06,
+      "loss": 0.8461,
+      "step": 378
+    },
+    {
+      "epoch": 0.012375510204081633,
+      "grad_norm": 2.9145092964172363,
+      "learning_rate": 9.98925538863488e-06,
+      "loss": 0.8288,
+      "step": 379
+    },
+    {
+      "epoch": 0.012408163265306122,
+      "grad_norm": 2.844569444656372,
+      "learning_rate": 9.989222730241674e-06,
+      "loss": 0.8685,
+      "step": 380
+    },
+    {
+      "epoch": 0.012440816326530612,
+      "grad_norm": 2.790707588195801,
+      "learning_rate": 9.989190071848465e-06,
+      "loss": 0.8467,
+      "step": 381
+    },
+    {
+      "epoch": 0.012473469387755102,
+      "grad_norm": 2.7228894233703613,
+      "learning_rate": 9.989157413455259e-06,
+      "loss": 0.8149,
+      "step": 382
+    },
+    {
+      "epoch": 0.012506122448979591,
+      "grad_norm": 2.8930270671844482,
+      "learning_rate": 9.989124755062052e-06,
+      "loss": 0.8521,
+      "step": 383
+    },
+    {
+      "epoch": 0.012538775510204082,
+      "grad_norm": 2.7201409339904785,
+      "learning_rate": 9.989092096668845e-06,
+      "loss": 0.8303,
+      "step": 384
+    },
+    {
+      "epoch": 0.012571428571428572,
+      "grad_norm": 2.558375597000122,
+      "learning_rate": 9.989059438275638e-06,
+      "loss": 0.8165,
+      "step": 385
+    },
+    {
+      "epoch": 0.012604081632653062,
+      "grad_norm": 2.893265724182129,
+      "learning_rate": 9.98902677988243e-06,
+      "loss": 0.833,
+      "step": 386
+    },
+    {
+      "epoch": 0.012636734693877551,
+      "grad_norm": 2.6960701942443848,
+      "learning_rate": 9.988994121489223e-06,
+      "loss": 0.7972,
+      "step": 387
+    },
+    {
+      "epoch": 0.01266938775510204,
+      "grad_norm": 2.5734570026397705,
+      "learning_rate": 9.988961463096016e-06,
+      "loss": 0.8748,
+      "step": 388
+    },
+    {
+      "epoch": 0.01270204081632653,
+      "grad_norm": 2.993227243423462,
+      "learning_rate": 9.98892880470281e-06,
+      "loss": 0.8405,
+      "step": 389
+    },
+    {
+      "epoch": 0.01273469387755102,
+      "grad_norm": 3.2019479274749756,
+      "learning_rate": 9.988896146309603e-06,
+      "loss": 0.8423,
+      "step": 390
+    },
+    {
+      "epoch": 0.012767346938775511,
+      "grad_norm": 3.2083091735839844,
+      "learning_rate": 9.988863487916394e-06,
+      "loss": 0.8088,
+      "step": 391
+    },
+    {
+      "epoch": 0.0128,
+      "grad_norm": 3.1379964351654053,
+      "learning_rate": 9.988830829523188e-06,
+      "loss": 0.8681,
+      "step": 392
+    },
+    {
+      "epoch": 0.01283265306122449,
+      "grad_norm": 2.7037453651428223,
+      "learning_rate": 9.98879817112998e-06,
+      "loss": 0.8379,
+      "step": 393
+    },
+    {
+      "epoch": 0.01286530612244898,
+      "grad_norm": 2.828984022140503,
+      "learning_rate": 9.988765512736774e-06,
+      "loss": 0.8148,
+      "step": 394
+    },
+    {
+      "epoch": 0.01289795918367347,
+      "grad_norm": 2.9447624683380127,
+      "learning_rate": 9.988732854343567e-06,
+      "loss": 0.8263,
+      "step": 395
+    },
+    {
+      "epoch": 0.012930612244897959,
+      "grad_norm": 2.7229952812194824,
+      "learning_rate": 9.98870019595036e-06,
+      "loss": 0.8613,
+      "step": 396
+    },
+    {
+      "epoch": 0.012963265306122448,
+      "grad_norm": 2.820530652999878,
+      "learning_rate": 9.988667537557152e-06,
+      "loss": 0.8174,
+      "step": 397
+    },
+    {
+      "epoch": 0.012995918367346938,
+      "grad_norm": 2.897592067718506,
+      "learning_rate": 9.988634879163947e-06,
+      "loss": 0.8416,
+      "step": 398
+    },
+    {
+      "epoch": 0.01302857142857143,
+      "grad_norm": 3.032390832901001,
+      "learning_rate": 9.988602220770739e-06,
+      "loss": 0.842,
+      "step": 399
+    },
+    {
+      "epoch": 0.013061224489795919,
+      "grad_norm": 2.949514389038086,
+      "learning_rate": 9.988569562377532e-06,
+      "loss": 0.8283,
+      "step": 400
+    },
+    {
+      "epoch": 0.013061224489795919,
+      "eval_loss": 0.8356085419654846,
+      "eval_runtime": 76.0573,
+      "eval_samples_per_second": 1.315,
+      "eval_steps_per_second": 1.315,
+      "step": 400
+    },
+    {
+      "epoch": 0.013093877551020408,
+      "grad_norm": 2.811276912689209,
+      "learning_rate": 9.988536903984325e-06,
+      "loss": 0.7906,
+      "step": 401
+    },
+    {
+      "epoch": 0.013126530612244898,
+      "grad_norm": 2.9961533546447754,
+      "learning_rate": 9.988504245591118e-06,
+      "loss": 0.839,
+      "step": 402
+    },
+    {
+      "epoch": 0.013159183673469388,
+      "grad_norm": 3.0735530853271484,
+      "learning_rate": 9.988471587197911e-06,
+      "loss": 0.8053,
+      "step": 403
+    },
+    {
+      "epoch": 0.013191836734693877,
+      "grad_norm": 2.793745279312134,
+      "learning_rate": 9.988438928804703e-06,
+      "loss": 0.8231,
+      "step": 404
+    },
+    {
+      "epoch": 0.013224489795918367,
+      "grad_norm": 3.1552298069000244,
+      "learning_rate": 9.988406270411496e-06,
+      "loss": 0.8328,
+      "step": 405
+    },
+    {
+      "epoch": 0.013257142857142858,
+      "grad_norm": 2.984865427017212,
+      "learning_rate": 9.98837361201829e-06,
+      "loss": 0.811,
+      "step": 406
+    },
+    {
+      "epoch": 0.013289795918367348,
+      "grad_norm": 2.8718035221099854,
+      "learning_rate": 9.988340953625083e-06,
+      "loss": 0.853,
+      "step": 407
+    },
+    {
+      "epoch": 0.013322448979591837,
+      "grad_norm": 3.0865938663482666,
+      "learning_rate": 9.988308295231876e-06,
+      "loss": 0.8003,
+      "step": 408
+    },
+    {
+      "epoch": 0.013355102040816327,
+      "grad_norm": 2.7381820678710938,
+      "learning_rate": 9.988275636838668e-06,
+      "loss": 0.8184,
+      "step": 409
+    },
+    {
+      "epoch": 0.013387755102040816,
+      "grad_norm": 2.8170037269592285,
+      "learning_rate": 9.98824297844546e-06,
+      "loss": 0.8311,
+      "step": 410
+    },
+    {
+      "epoch": 0.013420408163265306,
+      "grad_norm": 2.9140095710754395,
+      "learning_rate": 9.988210320052254e-06,
+      "loss": 0.834,
+      "step": 411
+    },
+    {
+      "epoch": 0.013453061224489795,
+      "grad_norm": 2.933525800704956,
+      "learning_rate": 9.988177661659047e-06,
+      "loss": 0.8294,
+      "step": 412
+    },
+    {
+      "epoch": 0.013485714285714285,
+      "grad_norm": 2.8974623680114746,
+      "learning_rate": 9.98814500326584e-06,
+      "loss": 0.8704,
+      "step": 413
+    },
+    {
+      "epoch": 0.013518367346938776,
+      "grad_norm": 2.794966697692871,
+      "learning_rate": 9.988112344872634e-06,
+      "loss": 0.8579,
+      "step": 414
+    },
+    {
+      "epoch": 0.013551020408163266,
+      "grad_norm": 2.856208324432373,
+      "learning_rate": 9.988079686479425e-06,
+      "loss": 0.8239,
+      "step": 415
+    },
+    {
+      "epoch": 0.013583673469387755,
+      "grad_norm": 2.999901533126831,
+      "learning_rate": 9.988047028086218e-06,
+      "loss": 0.8166,
+      "step": 416
+    },
+    {
+      "epoch": 0.013616326530612245,
+      "grad_norm": 2.8142268657684326,
+      "learning_rate": 9.988014369693012e-06,
+      "loss": 0.8343,
+      "step": 417
+    },
+    {
+      "epoch": 0.013648979591836734,
+      "grad_norm": 2.808567523956299,
+      "learning_rate": 9.987981711299805e-06,
+      "loss": 0.8414,
+      "step": 418
+    },
+    {
+      "epoch": 0.013681632653061224,
+      "grad_norm": 2.6575918197631836,
+      "learning_rate": 9.987949052906598e-06,
+      "loss": 0.8256,
+      "step": 419
+    },
+    {
+      "epoch": 0.013714285714285714,
+      "grad_norm": 2.9415652751922607,
+      "learning_rate": 9.987916394513391e-06,
+      "loss": 0.811,
+      "step": 420
+    },
+    {
+      "epoch": 0.013746938775510205,
+      "grad_norm": 2.7710697650909424,
+      "learning_rate": 9.987883736120185e-06,
+      "loss": 0.8132,
+      "step": 421
+    },
+    {
+      "epoch": 0.013779591836734694,
+      "grad_norm": 2.6907401084899902,
+      "learning_rate": 9.987851077726976e-06,
+      "loss": 0.8022,
+      "step": 422
+    },
+    {
+      "epoch": 0.013812244897959184,
+      "grad_norm": 3.010432481765747,
+      "learning_rate": 9.98781841933377e-06,
+      "loss": 0.8196,
+      "step": 423
+    },
+    {
+      "epoch": 0.013844897959183674,
+      "grad_norm": 2.868598461151123,
+      "learning_rate": 9.987785760940563e-06,
+      "loss": 0.8234,
+      "step": 424
+    },
+    {
+      "epoch": 0.013877551020408163,
+      "grad_norm": 2.9146206378936768,
+      "learning_rate": 9.987753102547356e-06,
+      "loss": 0.8357,
+      "step": 425
+    },
+    {
+      "epoch": 0.013910204081632653,
+      "grad_norm": 3.148287534713745,
+      "learning_rate": 9.987720444154149e-06,
+      "loss": 0.8168,
+      "step": 426
+    },
+    {
+      "epoch": 0.013942857142857142,
+      "grad_norm": 3.0107545852661133,
+      "learning_rate": 9.98768778576094e-06,
+      "loss": 0.8313,
+      "step": 427
+    },
+    {
+      "epoch": 0.013975510204081632,
+      "grad_norm": 2.7397043704986572,
+      "learning_rate": 9.987655127367734e-06,
+      "loss": 0.825,
+      "step": 428
+    },
+    {
+      "epoch": 0.014008163265306123,
+      "grad_norm": 2.8118340969085693,
+      "learning_rate": 9.987622468974527e-06,
+      "loss": 0.8101,
+      "step": 429
+    },
+    {
+      "epoch": 0.014040816326530613,
+      "grad_norm": 2.862879514694214,
+      "learning_rate": 9.98758981058132e-06,
+      "loss": 0.8504,
+      "step": 430
+    },
+    {
+      "epoch": 0.014073469387755102,
+      "grad_norm": 3.0519521236419678,
+      "learning_rate": 9.987557152188114e-06,
+      "loss": 0.8337,
+      "step": 431
+    },
+    {
+      "epoch": 0.014106122448979592,
+      "grad_norm": 2.8947372436523438,
+      "learning_rate": 9.987524493794905e-06,
+      "loss": 0.8041,
+      "step": 432
+    },
+    {
+      "epoch": 0.014138775510204081,
+      "grad_norm": 2.7874197959899902,
+      "learning_rate": 9.987491835401698e-06,
+      "loss": 0.8143,
+      "step": 433
+    },
+    {
+      "epoch": 0.014171428571428571,
+      "grad_norm": 3.011812686920166,
+      "learning_rate": 9.987459177008492e-06,
+      "loss": 0.7936,
+      "step": 434
+    },
+    {
+      "epoch": 0.01420408163265306,
+      "grad_norm": 2.8212246894836426,
+      "learning_rate": 9.987426518615285e-06,
+      "loss": 0.8028,
+      "step": 435
+    },
+    {
+      "epoch": 0.014236734693877552,
+      "grad_norm": 2.7385244369506836,
+      "learning_rate": 9.987393860222078e-06,
+      "loss": 0.8292,
+      "step": 436
+    },
+    {
+      "epoch": 0.014269387755102041,
+      "grad_norm": 2.7048816680908203,
+      "learning_rate": 9.987361201828871e-06,
+      "loss": 0.7567,
+      "step": 437
+    },
+    {
+      "epoch": 0.014302040816326531,
+      "grad_norm": 3.001534938812256,
+      "learning_rate": 9.987328543435663e-06,
+      "loss": 0.7897,
+      "step": 438
+    },
+    {
+      "epoch": 0.01433469387755102,
+      "grad_norm": 3.1388514041900635,
+      "learning_rate": 9.987295885042456e-06,
+      "loss": 0.8373,
+      "step": 439
+    },
+    {
+      "epoch": 0.01436734693877551,
+      "grad_norm": 2.902858018875122,
+      "learning_rate": 9.98726322664925e-06,
+      "loss": 0.8342,
+      "step": 440
+    },
+    {
+      "epoch": 0.0144,
+      "grad_norm": 2.871511459350586,
+      "learning_rate": 9.987230568256043e-06,
+      "loss": 0.8107,
+      "step": 441
+    },
+    {
+      "epoch": 0.01443265306122449,
+      "grad_norm": 3.0739245414733887,
+      "learning_rate": 9.987197909862836e-06,
+      "loss": 0.8171,
+      "step": 442
+    },
+    {
+      "epoch": 0.014465306122448979,
+      "grad_norm": 2.9680142402648926,
+      "learning_rate": 9.987165251469629e-06,
+      "loss": 0.8492,
+      "step": 443
+    },
+    {
+      "epoch": 0.01449795918367347,
+      "grad_norm": 2.763535737991333,
+      "learning_rate": 9.987132593076422e-06,
+      "loss": 0.8201,
+      "step": 444
+    },
+    {
+      "epoch": 0.01453061224489796,
+      "grad_norm": 2.8160109519958496,
+      "learning_rate": 9.987099934683214e-06,
+      "loss": 0.8175,
+      "step": 445
+    },
+    {
+      "epoch": 0.01456326530612245,
+      "grad_norm": 2.746919870376587,
+      "learning_rate": 9.987067276290007e-06,
+      "loss": 0.8121,
+      "step": 446
+    },
+    {
+      "epoch": 0.014595918367346939,
+      "grad_norm": 2.9417343139648438,
+      "learning_rate": 9.9870346178968e-06,
+      "loss": 0.8211,
+      "step": 447
+    },
+    {
+      "epoch": 0.014628571428571428,
+      "grad_norm": 2.886810302734375,
+      "learning_rate": 9.987001959503594e-06,
+      "loss": 0.7889,
+      "step": 448
+    },
+    {
+      "epoch": 0.014661224489795918,
+      "grad_norm": 2.89780592918396,
+      "learning_rate": 9.986969301110387e-06,
+      "loss": 0.8386,
+      "step": 449
+    },
+    {
+      "epoch": 0.014693877551020407,
+      "grad_norm": 2.8015475273132324,
+      "learning_rate": 9.986936642717178e-06,
+      "loss": 0.8023,
+      "step": 450
+    },
+    {
+      "epoch": 0.014693877551020407,
+      "eval_loss": 0.83209228515625,
+      "eval_runtime": 76.371,
+      "eval_samples_per_second": 1.309,
+      "eval_steps_per_second": 1.309,
+      "step": 450
+    },
+    {
+      "epoch": 0.014726530612244899,
+      "grad_norm": 3.0286951065063477,
+      "learning_rate": 9.986903984323972e-06,
+      "loss": 0.795,
+      "step": 451
+    },
+    {
+      "epoch": 0.014759183673469388,
+      "grad_norm": 2.798675060272217,
+      "learning_rate": 9.986871325930765e-06,
+      "loss": 0.778,
+      "step": 452
+    },
+    {
+      "epoch": 0.014791836734693878,
+      "grad_norm": 2.9744062423706055,
+      "learning_rate": 9.986838667537558e-06,
+      "loss": 0.8034,
+      "step": 453
+    },
+    {
+      "epoch": 0.014824489795918367,
+      "grad_norm": 3.131807804107666,
+      "learning_rate": 9.986806009144351e-06,
+      "loss": 0.8147,
+      "step": 454
+    },
+    {
+      "epoch": 0.014857142857142857,
+      "grad_norm": 2.759378671646118,
+      "learning_rate": 9.986773350751145e-06,
+      "loss": 0.7798,
+      "step": 455
+    },
+    {
+      "epoch": 0.014889795918367347,
+      "grad_norm": 2.79740047454834,
+      "learning_rate": 9.986740692357936e-06,
+      "loss": 0.793,
+      "step": 456
+    },
+    {
+      "epoch": 0.014922448979591836,
+      "grad_norm": 3.367506742477417,
+      "learning_rate": 9.98670803396473e-06,
+      "loss": 0.8295,
+      "step": 457
+    },
+    {
+      "epoch": 0.014955102040816327,
+      "grad_norm": 2.965843677520752,
+      "learning_rate": 9.986675375571523e-06,
+      "loss": 0.7908,
+      "step": 458
+    },
+    {
+      "epoch": 0.014987755102040817,
+      "grad_norm": 2.815945625305176,
+      "learning_rate": 9.986642717178316e-06,
+      "loss": 0.821,
+      "step": 459
+    },
+    {
+      "epoch": 0.015020408163265307,
+      "grad_norm": 2.8080756664276123,
+      "learning_rate": 9.986610058785109e-06,
+      "loss": 0.7952,
+      "step": 460
+    },
+    {
+      "epoch": 0.015053061224489796,
+      "grad_norm": 3.2081828117370605,
+      "learning_rate": 9.9865774003919e-06,
+      "loss": 0.8306,
+      "step": 461
+    },
+    {
+      "epoch": 0.015085714285714286,
+      "grad_norm": 2.866701602935791,
+      "learning_rate": 9.986544741998695e-06,
+      "loss": 0.8438,
+      "step": 462
+    },
+    {
+      "epoch": 0.015118367346938775,
+      "grad_norm": 2.806741952896118,
+      "learning_rate": 9.986512083605487e-06,
+      "loss": 0.8082,
+      "step": 463
+    },
+    {
+      "epoch": 0.015151020408163265,
+      "grad_norm": 2.8352231979370117,
+      "learning_rate": 9.98647942521228e-06,
+      "loss": 0.7805,
+      "step": 464
+    },
+    {
+      "epoch": 0.015183673469387754,
+      "grad_norm": 2.838469982147217,
+      "learning_rate": 9.986446766819073e-06,
+      "loss": 0.844,
+      "step": 465
+    },
+    {
+      "epoch": 0.015216326530612246,
+      "grad_norm": 2.7868754863739014,
+      "learning_rate": 9.986414108425867e-06,
+      "loss": 0.8078,
+      "step": 466
+    },
+    {
+      "epoch": 0.015248979591836735,
+      "grad_norm": 2.809109687805176,
+      "learning_rate": 9.98638145003266e-06,
+      "loss": 0.8425,
+      "step": 467
+    },
+    {
+      "epoch": 0.015281632653061225,
+      "grad_norm": 2.7377090454101562,
+      "learning_rate": 9.986348791639451e-06,
+      "loss": 0.8332,
+      "step": 468
+    },
+    {
+      "epoch": 0.015314285714285714,
+      "grad_norm": 2.863290786743164,
+      "learning_rate": 9.986316133246245e-06,
+      "loss": 0.8352,
+      "step": 469
+    },
+    {
+      "epoch": 0.015346938775510204,
+      "grad_norm": 2.720506429672241,
+      "learning_rate": 9.986283474853038e-06,
+      "loss": 0.8114,
+      "step": 470
+    },
+    {
+      "epoch": 0.015379591836734693,
+      "grad_norm": 2.7052462100982666,
+      "learning_rate": 9.986250816459831e-06,
+      "loss": 0.8093,
+      "step": 471
+    },
+    {
+      "epoch": 0.015412244897959183,
+      "grad_norm": 3.0364644527435303,
+      "learning_rate": 9.986218158066624e-06,
+      "loss": 0.8156,
+      "step": 472
+    },
+    {
+      "epoch": 0.015444897959183674,
+      "grad_norm": 2.9919137954711914,
+      "learning_rate": 9.986185499673416e-06,
+      "loss": 0.7906,
+      "step": 473
+    },
+    {
+      "epoch": 0.015477551020408164,
+      "grad_norm": 3.102023124694824,
+      "learning_rate": 9.98615284128021e-06,
+      "loss": 0.8247,
+      "step": 474
+    },
+    {
+      "epoch": 0.015510204081632653,
+      "grad_norm": 2.8093349933624268,
+      "learning_rate": 9.986120182887002e-06,
+      "loss": 0.8339,
+      "step": 475
+    },
+    {
+      "epoch": 0.015542857142857143,
+      "grad_norm": 2.893361806869507,
+      "learning_rate": 9.986087524493796e-06,
+      "loss": 0.7698,
+      "step": 476
+    },
+    {
+      "epoch": 0.015575510204081633,
+      "grad_norm": 2.813911199569702,
+      "learning_rate": 9.986054866100589e-06,
+      "loss": 0.8204,
+      "step": 477
+    },
+    {
+      "epoch": 0.015608163265306122,
+      "grad_norm": 2.7959954738616943,
+      "learning_rate": 9.986022207707382e-06,
+      "loss": 0.8099,
+      "step": 478
+    },
+    {
+      "epoch": 0.015640816326530613,
+      "grad_norm": 2.637242317199707,
+      "learning_rate": 9.985989549314174e-06,
+      "loss": 0.8218,
+      "step": 479
+    },
+    {
+      "epoch": 0.0156734693877551,
+      "grad_norm": 2.694908380508423,
+      "learning_rate": 9.985956890920967e-06,
+      "loss": 0.843,
+      "step": 480
+    },
+    {
+      "epoch": 0.015706122448979593,
+      "grad_norm": 2.65104603767395,
+      "learning_rate": 9.98592423252776e-06,
+      "loss": 0.8142,
+      "step": 481
+    },
+    {
+      "epoch": 0.01573877551020408,
+      "grad_norm": 2.7440900802612305,
+      "learning_rate": 9.985891574134553e-06,
+      "loss": 0.8079,
+      "step": 482
+    },
+    {
+      "epoch": 0.01577142857142857,
+      "grad_norm": 2.7440459728240967,
+      "learning_rate": 9.985858915741347e-06,
+      "loss": 0.8106,
+      "step": 483
+    },
+    {
+      "epoch": 0.015804081632653063,
+      "grad_norm": 2.8990628719329834,
+      "learning_rate": 9.98582625734814e-06,
+      "loss": 0.7954,
+      "step": 484
+    },
+    {
+      "epoch": 0.01583673469387755,
+      "grad_norm": 2.8682186603546143,
+      "learning_rate": 9.985793598954933e-06,
+      "loss": 0.7892,
+      "step": 485
+    },
+    {
+      "epoch": 0.015869387755102042,
+      "grad_norm": 3.1332759857177734,
+      "learning_rate": 9.985760940561725e-06,
+      "loss": 0.8182,
+      "step": 486
+    },
+    {
+      "epoch": 0.01590204081632653,
+      "grad_norm": 2.9516282081604004,
+      "learning_rate": 9.985728282168518e-06,
+      "loss": 0.7985,
+      "step": 487
+    },
+    {
+      "epoch": 0.01593469387755102,
+      "grad_norm": 3.3294899463653564,
+      "learning_rate": 9.985695623775311e-06,
+      "loss": 0.8368,
+      "step": 488
+    },
+    {
+      "epoch": 0.01596734693877551,
+      "grad_norm": 2.95438814163208,
+      "learning_rate": 9.985662965382104e-06,
+      "loss": 0.8151,
+      "step": 489
+    },
+    {
+      "epoch": 0.016,
+      "grad_norm": 2.7308006286621094,
+      "learning_rate": 9.985630306988898e-06,
+      "loss": 0.827,
+      "step": 490
+    },
+    {
+      "epoch": 0.016032653061224488,
+      "grad_norm": 2.695218801498413,
+      "learning_rate": 9.985597648595689e-06,
+      "loss": 0.7846,
+      "step": 491
+    },
+    {
+      "epoch": 0.01606530612244898,
+      "grad_norm": 2.842834949493408,
+      "learning_rate": 9.985564990202482e-06,
+      "loss": 0.783,
+      "step": 492
+    },
+    {
+      "epoch": 0.01609795918367347,
+      "grad_norm": 2.944671869277954,
+      "learning_rate": 9.985532331809276e-06,
+      "loss": 0.8273,
+      "step": 493
+    },
+    {
+      "epoch": 0.01613061224489796,
+      "grad_norm": 2.8893423080444336,
+      "learning_rate": 9.985499673416069e-06,
+      "loss": 0.8131,
+      "step": 494
+    },
+    {
+      "epoch": 0.01616326530612245,
+      "grad_norm": 2.917097568511963,
+      "learning_rate": 9.985467015022862e-06,
+      "loss": 0.8491,
+      "step": 495
+    },
+    {
+      "epoch": 0.016195918367346938,
+      "grad_norm": 2.678619384765625,
+      "learning_rate": 9.985434356629654e-06,
+      "loss": 0.7789,
+      "step": 496
+    },
+    {
+      "epoch": 0.01622857142857143,
+      "grad_norm": 2.86249041557312,
+      "learning_rate": 9.985401698236447e-06,
+      "loss": 0.8461,
+      "step": 497
+    },
+    {
+      "epoch": 0.016261224489795917,
+      "grad_norm": 2.9272475242614746,
+      "learning_rate": 9.98536903984324e-06,
+      "loss": 0.8047,
+      "step": 498
+    },
+    {
+      "epoch": 0.016293877551020408,
+      "grad_norm": 2.894817590713501,
+      "learning_rate": 9.985336381450033e-06,
+      "loss": 0.8157,
+      "step": 499
+    },
+    {
+      "epoch": 0.0163265306122449,
+      "grad_norm": 2.9469430446624756,
+      "learning_rate": 9.985303723056827e-06,
+      "loss": 0.8409,
+      "step": 500
+    },
+    {
+      "epoch": 0.0163265306122449,
+      "eval_loss": 0.8293061852455139,
+      "eval_runtime": 76.9824,
+      "eval_samples_per_second": 1.299,
+      "eval_steps_per_second": 1.299,
+      "step": 500
+    },
+    {
+      "epoch": 0.016359183673469387,
+      "grad_norm": 2.7225022315979004,
+      "learning_rate": 9.98527106466362e-06,
+      "loss": 0.7978,
+      "step": 501
+    },
+    {
+      "epoch": 0.01639183673469388,
+      "grad_norm": 2.6931450366973877,
+      "learning_rate": 9.985238406270411e-06,
+      "loss": 0.8089,
+      "step": 502
+    },
+    {
+      "epoch": 0.016424489795918366,
+      "grad_norm": 2.697438955307007,
+      "learning_rate": 9.985205747877206e-06,
+      "loss": 0.8231,
+      "step": 503
+    },
+    {
+      "epoch": 0.016457142857142858,
+      "grad_norm": 2.997002124786377,
+      "learning_rate": 9.985173089483998e-06,
+      "loss": 0.8211,
+      "step": 504
+    },
+    {
+      "epoch": 0.016489795918367346,
+      "grad_norm": 2.952876567840576,
+      "learning_rate": 9.985140431090791e-06,
+      "loss": 0.786,
+      "step": 505
+    },
+    {
+      "epoch": 0.016522448979591837,
+      "grad_norm": 3.085167407989502,
+      "learning_rate": 9.985107772697584e-06,
+      "loss": 0.8261,
+      "step": 506
+    },
+    {
+      "epoch": 0.016555102040816328,
+      "grad_norm": 2.8937699794769287,
+      "learning_rate": 9.985075114304378e-06,
+      "loss": 0.8009,
+      "step": 507
+    },
+    {
+      "epoch": 0.016587755102040816,
+      "grad_norm": 2.934492588043213,
+      "learning_rate": 9.98504245591117e-06,
+      "loss": 0.8059,
+      "step": 508
+    },
+    {
+      "epoch": 0.016620408163265307,
+      "grad_norm": 2.9281036853790283,
+      "learning_rate": 9.985009797517962e-06,
+      "loss": 0.8488,
+      "step": 509
+    },
+    {
+      "epoch": 0.016653061224489795,
+      "grad_norm": 2.8904941082000732,
+      "learning_rate": 9.984977139124756e-06,
+      "loss": 0.8431,
+      "step": 510
+    },
+    {
+      "epoch": 0.016685714285714286,
+      "grad_norm": 2.7994799613952637,
+      "learning_rate": 9.984944480731549e-06,
+      "loss": 0.8339,
+      "step": 511
+    },
+    {
+      "epoch": 0.016718367346938774,
+      "grad_norm": 3.0456082820892334,
+      "learning_rate": 9.984911822338342e-06,
+      "loss": 0.8077,
+      "step": 512
+    },
+    {
+      "epoch": 0.016751020408163265,
+      "grad_norm": 2.969820261001587,
+      "learning_rate": 9.984879163945135e-06,
+      "loss": 0.8254,
+      "step": 513
+    },
+    {
+      "epoch": 0.016783673469387757,
+      "grad_norm": 2.691143035888672,
+      "learning_rate": 9.984846505551927e-06,
+      "loss": 0.8208,
+      "step": 514
+    },
+    {
+      "epoch": 0.016816326530612245,
+      "grad_norm": 3.0714621543884277,
+      "learning_rate": 9.98481384715872e-06,
+      "loss": 0.8094,
+      "step": 515
+    },
+    {
+      "epoch": 0.016848979591836736,
+      "grad_norm": 3.146872043609619,
+      "learning_rate": 9.984781188765513e-06,
+      "loss": 0.8295,
+      "step": 516
+    },
+    {
+      "epoch": 0.016881632653061224,
+      "grad_norm": 3.0401713848114014,
+      "learning_rate": 9.984748530372307e-06,
+      "loss": 0.8438,
+      "step": 517
+    },
+    {
+      "epoch": 0.016914285714285715,
+      "grad_norm": 3.1565845012664795,
+      "learning_rate": 9.9847158719791e-06,
+      "loss": 0.8238,
+      "step": 518
+    },
+    {
+      "epoch": 0.016946938775510203,
+      "grad_norm": 3.035167932510376,
+      "learning_rate": 9.984683213585893e-06,
+      "loss": 0.7907,
+      "step": 519
+    },
+    {
+      "epoch": 0.016979591836734694,
+      "grad_norm": 2.9173972606658936,
+      "learning_rate": 9.984650555192685e-06,
+      "loss": 0.8298,
+      "step": 520
+    },
+    {
+      "epoch": 0.017012244897959182,
+      "grad_norm": 3.0475914478302,
+      "learning_rate": 9.984617896799478e-06,
+      "loss": 0.7896,
+      "step": 521
+    },
+    {
+      "epoch": 0.017044897959183673,
+      "grad_norm": 2.8846919536590576,
+      "learning_rate": 9.984585238406271e-06,
+      "loss": 0.8268,
+      "step": 522
+    },
+    {
+      "epoch": 0.017077551020408165,
+      "grad_norm": 2.9851925373077393,
+      "learning_rate": 9.984552580013064e-06,
+      "loss": 0.8166,
+      "step": 523
+    },
+    {
+      "epoch": 0.017110204081632652,
+      "grad_norm": 2.960608720779419,
+      "learning_rate": 9.984519921619857e-06,
+      "loss": 0.838,
+      "step": 524
+    },
+    {
+      "epoch": 0.017142857142857144,
+      "grad_norm": 2.8663570880889893,
+      "learning_rate": 9.984487263226649e-06,
+      "loss": 0.8134,
+      "step": 525
+    },
+    {
+      "epoch": 0.01717551020408163,
+      "grad_norm": 3.2871882915496826,
+      "learning_rate": 9.984454604833444e-06,
+      "loss": 0.842,
+      "step": 526
+    },
+    {
+      "epoch": 0.017208163265306123,
+      "grad_norm": 2.9747695922851562,
+      "learning_rate": 9.984421946440235e-06,
+      "loss": 0.8371,
+      "step": 527
+    },
+    {
+      "epoch": 0.01724081632653061,
+      "grad_norm": 3.065366744995117,
+      "learning_rate": 9.984389288047029e-06,
+      "loss": 0.799,
+      "step": 528
+    },
+    {
+      "epoch": 0.017273469387755102,
+      "grad_norm": 3.3719398975372314,
+      "learning_rate": 9.984356629653822e-06,
+      "loss": 0.7875,
+      "step": 529
+    },
+    {
+      "epoch": 0.017306122448979593,
+      "grad_norm": 3.0407257080078125,
+      "learning_rate": 9.984323971260615e-06,
+      "loss": 0.833,
+      "step": 530
+    },
+    {
+      "epoch": 0.01733877551020408,
+      "grad_norm": 2.992464780807495,
+      "learning_rate": 9.984291312867408e-06,
+      "loss": 0.83,
+      "step": 531
+    },
+    {
+      "epoch": 0.017371428571428572,
+      "grad_norm": 2.655125379562378,
+      "learning_rate": 9.9842586544742e-06,
+      "loss": 0.8243,
+      "step": 532
+    },
+    {
+      "epoch": 0.01740408163265306,
+      "grad_norm": 2.7782280445098877,
+      "learning_rate": 9.984225996080993e-06,
+      "loss": 0.8279,
+      "step": 533
+    },
+    {
+      "epoch": 0.01743673469387755,
+      "grad_norm": 2.6449782848358154,
+      "learning_rate": 9.984193337687786e-06,
+      "loss": 0.8517,
+      "step": 534
+    },
+    {
+      "epoch": 0.01746938775510204,
+      "grad_norm": 3.011317491531372,
+      "learning_rate": 9.98416067929458e-06,
+      "loss": 0.7763,
+      "step": 535
+    },
+    {
+      "epoch": 0.01750204081632653,
+      "grad_norm": 2.915745496749878,
+      "learning_rate": 9.984128020901373e-06,
+      "loss": 0.8283,
+      "step": 536
+    },
+    {
+      "epoch": 0.017534693877551022,
+      "grad_norm": 2.7176997661590576,
+      "learning_rate": 9.984095362508164e-06,
+      "loss": 0.7846,
+      "step": 537
+    },
+    {
+      "epoch": 0.01756734693877551,
+      "grad_norm": 2.9343252182006836,
+      "learning_rate": 9.984062704114958e-06,
+      "loss": 0.7733,
+      "step": 538
+    },
+    {
+      "epoch": 0.0176,
+      "grad_norm": 3.041198492050171,
+      "learning_rate": 9.984030045721751e-06,
+      "loss": 0.8024,
+      "step": 539
+    },
+    {
+      "epoch": 0.01763265306122449,
+      "grad_norm": 2.8848764896392822,
+      "learning_rate": 9.983997387328544e-06,
+      "loss": 0.8064,
+      "step": 540
+    },
+    {
+      "epoch": 0.01766530612244898,
+      "grad_norm": 2.847885847091675,
+      "learning_rate": 9.983964728935337e-06,
+      "loss": 0.7638,
+      "step": 541
+    },
+    {
+      "epoch": 0.017697959183673468,
+      "grad_norm": 2.7816402912139893,
+      "learning_rate": 9.98393207054213e-06,
+      "loss": 0.7839,
+      "step": 542
+    },
+    {
+      "epoch": 0.01773061224489796,
+      "grad_norm": 2.8443193435668945,
+      "learning_rate": 9.983899412148922e-06,
+      "loss": 0.8005,
+      "step": 543
+    },
+    {
+      "epoch": 0.01776326530612245,
+      "grad_norm": 2.923306941986084,
+      "learning_rate": 9.983866753755717e-06,
+      "loss": 0.8048,
+      "step": 544
+    },
+    {
+      "epoch": 0.01779591836734694,
+      "grad_norm": 3.0685269832611084,
+      "learning_rate": 9.983834095362509e-06,
+      "loss": 0.7676,
+      "step": 545
+    },
+    {
+      "epoch": 0.01782857142857143,
+      "grad_norm": 2.841459274291992,
+      "learning_rate": 9.983801436969302e-06,
+      "loss": 0.7828,
+      "step": 546
+    },
+    {
+      "epoch": 0.017861224489795918,
+      "grad_norm": 3.1108107566833496,
+      "learning_rate": 9.983768778576095e-06,
+      "loss": 0.8287,
+      "step": 547
+    },
+    {
+      "epoch": 0.01789387755102041,
+      "grad_norm": 2.8307478427886963,
+      "learning_rate": 9.983736120182888e-06,
+      "loss": 0.7956,
+      "step": 548
+    },
+    {
+      "epoch": 0.017926530612244897,
+      "grad_norm": 2.83941650390625,
+      "learning_rate": 9.983703461789682e-06,
+      "loss": 0.7963,
+      "step": 549
+    },
+    {
+      "epoch": 0.017959183673469388,
+      "grad_norm": 2.710853338241577,
+      "learning_rate": 9.983670803396473e-06,
+      "loss": 0.8023,
+      "step": 550
+    },
+    {
+      "epoch": 0.017959183673469388,
+      "eval_loss": 0.8259984850883484,
+      "eval_runtime": 98.251,
+      "eval_samples_per_second": 1.018,
+      "eval_steps_per_second": 1.018,
+      "step": 550
+    },
+    {
+      "epoch": 0.01799183673469388,
+      "grad_norm": 2.7674920558929443,
+      "learning_rate": 9.983638145003266e-06,
+      "loss": 0.8121,
+      "step": 551
+    },
+    {
+      "epoch": 0.018024489795918367,
+      "grad_norm": 2.756206512451172,
+      "learning_rate": 9.98360548661006e-06,
+      "loss": 0.7886,
+      "step": 552
+    },
+    {
+      "epoch": 0.01805714285714286,
+      "grad_norm": 3.000251054763794,
+      "learning_rate": 9.983572828216853e-06,
+      "loss": 0.8181,
+      "step": 553
+    },
+    {
+      "epoch": 0.018089795918367346,
+      "grad_norm": 2.7608935832977295,
+      "learning_rate": 9.983540169823646e-06,
+      "loss": 0.8047,
+      "step": 554
+    },
+    {
+      "epoch": 0.018122448979591838,
+      "grad_norm": 2.7886579036712646,
+      "learning_rate": 9.983507511430438e-06,
+      "loss": 0.8023,
+      "step": 555
+    },
+    {
+      "epoch": 0.018155102040816325,
+      "grad_norm": 2.7562673091888428,
+      "learning_rate": 9.983474853037231e-06,
+      "loss": 0.7935,
+      "step": 556
+    },
+    {
+      "epoch": 0.018187755102040817,
+      "grad_norm": 2.8150694370269775,
+      "learning_rate": 9.983442194644024e-06,
+      "loss": 0.8252,
+      "step": 557
+    },
+    {
+      "epoch": 0.018220408163265304,
+      "grad_norm": 3.069070339202881,
+      "learning_rate": 9.983409536250817e-06,
+      "loss": 0.8352,
+      "step": 558
+    },
+    {
+      "epoch": 0.018253061224489796,
+      "grad_norm": 3.065143585205078,
+      "learning_rate": 9.98337687785761e-06,
+      "loss": 0.8322,
+      "step": 559
+    },
+    {
+      "epoch": 0.018285714285714287,
+      "grad_norm": 2.808361530303955,
+      "learning_rate": 9.983344219464404e-06,
+      "loss": 0.8316,
+      "step": 560
+    },
+    {
+      "epoch": 0.018318367346938775,
+      "grad_norm": 3.086597204208374,
+      "learning_rate": 9.983311561071195e-06,
+      "loss": 0.8063,
+      "step": 561
+    },
+    {
+      "epoch": 0.018351020408163266,
+      "grad_norm": 2.8510525226593018,
+      "learning_rate": 9.983278902677989e-06,
+      "loss": 0.7735,
+      "step": 562
+    },
+    {
+      "epoch": 0.018383673469387754,
+      "grad_norm": 2.8289899826049805,
+      "learning_rate": 9.983246244284782e-06,
+      "loss": 0.831,
+      "step": 563
+    },
+    {
+      "epoch": 0.018416326530612245,
+      "grad_norm": 2.6663060188293457,
+      "learning_rate": 9.983213585891575e-06,
+      "loss": 0.8109,
+      "step": 564
+    },
+    {
+      "epoch": 0.018448979591836733,
+      "grad_norm": 2.8458051681518555,
+      "learning_rate": 9.983180927498368e-06,
+      "loss": 0.8069,
+      "step": 565
+    },
+    {
+      "epoch": 0.018481632653061224,
+      "grad_norm": 3.2364461421966553,
+      "learning_rate": 9.98314826910516e-06,
+      "loss": 0.8214,
+      "step": 566
+    },
+    {
+      "epoch": 0.018514285714285716,
+      "grad_norm": 3.0581226348876953,
+      "learning_rate": 9.983115610711955e-06,
+      "loss": 0.8115,
+      "step": 567
+    },
+    {
+      "epoch": 0.018546938775510204,
+      "grad_norm": 2.7239277362823486,
+      "learning_rate": 9.983082952318746e-06,
+      "loss": 0.8248,
+      "step": 568
+    },
+    {
+      "epoch": 0.018579591836734695,
+      "grad_norm": 2.827193260192871,
+      "learning_rate": 9.98305029392554e-06,
+      "loss": 0.8282,
+      "step": 569
+    },
+    {
+      "epoch": 0.018612244897959183,
+      "grad_norm": 3.018669605255127,
+      "learning_rate": 9.983017635532333e-06,
+      "loss": 0.8196,
+      "step": 570
+    },
+    {
+      "epoch": 0.018644897959183674,
+      "grad_norm": 2.9022018909454346,
+      "learning_rate": 9.982984977139126e-06,
+      "loss": 0.7748,
+      "step": 571
+    },
+    {
+      "epoch": 0.018677551020408162,
+      "grad_norm": 2.7094552516937256,
+      "learning_rate": 9.98295231874592e-06,
+      "loss": 0.8381,
+      "step": 572
+    },
+    {
+      "epoch": 0.018710204081632653,
+      "grad_norm": 3.020749568939209,
+      "learning_rate": 9.98291966035271e-06,
+      "loss": 0.7954,
+      "step": 573
+    },
+    {
+      "epoch": 0.018742857142857144,
+      "grad_norm": 2.9567782878875732,
+      "learning_rate": 9.982887001959504e-06,
+      "loss": 0.8197,
+      "step": 574
+    },
+    {
+      "epoch": 0.018775510204081632,
+      "grad_norm": 2.939793348312378,
+      "learning_rate": 9.982854343566297e-06,
+      "loss": 0.7891,
+      "step": 575
+    },
+    {
+      "epoch": 0.018808163265306124,
+      "grad_norm": 2.851130485534668,
+      "learning_rate": 9.98282168517309e-06,
+      "loss": 0.8669,
+      "step": 576
+    },
+    {
+      "epoch": 0.01884081632653061,
+      "grad_norm": 2.745495080947876,
+      "learning_rate": 9.982789026779884e-06,
+      "loss": 0.8028,
+      "step": 577
+    },
+    {
+      "epoch": 0.018873469387755103,
+      "grad_norm": 2.83685302734375,
+      "learning_rate": 9.982756368386675e-06,
+      "loss": 0.7981,
+      "step": 578
+    },
+    {
+      "epoch": 0.01890612244897959,
+      "grad_norm": 2.845992088317871,
+      "learning_rate": 9.982723709993469e-06,
+      "loss": 0.7752,
+      "step": 579
+    },
+    {
+      "epoch": 0.018938775510204082,
+      "grad_norm": 3.1280617713928223,
+      "learning_rate": 9.982691051600262e-06,
+      "loss": 0.824,
+      "step": 580
+    },
+    {
+      "epoch": 0.018971428571428573,
+      "grad_norm": 2.8821194171905518,
+      "learning_rate": 9.982658393207055e-06,
+      "loss": 0.7925,
+      "step": 581
+    },
+    {
+      "epoch": 0.01900408163265306,
+      "grad_norm": 2.827406167984009,
+      "learning_rate": 9.982625734813848e-06,
+      "loss": 0.8311,
+      "step": 582
+    },
+    {
+      "epoch": 0.019036734693877552,
+      "grad_norm": 2.7628026008605957,
+      "learning_rate": 9.982593076420641e-06,
+      "loss": 0.8228,
+      "step": 583
+    },
+    {
+      "epoch": 0.01906938775510204,
+      "grad_norm": 2.8712172508239746,
+      "learning_rate": 9.982560418027433e-06,
+      "loss": 0.8014,
+      "step": 584
+    },
+    {
+      "epoch": 0.01910204081632653,
+      "grad_norm": 2.85799241065979,
+      "learning_rate": 9.982527759634226e-06,
+      "loss": 0.8281,
+      "step": 585
+    },
+    {
+      "epoch": 0.01913469387755102,
+      "grad_norm": 2.917358160018921,
+      "learning_rate": 9.98249510124102e-06,
+      "loss": 0.8049,
+      "step": 586
+    },
+    {
+      "epoch": 0.01916734693877551,
+      "grad_norm": 2.8612101078033447,
+      "learning_rate": 9.982462442847813e-06,
+      "loss": 0.7645,
+      "step": 587
+    },
+    {
+      "epoch": 0.0192,
+      "grad_norm": 2.8899152278900146,
+      "learning_rate": 9.982429784454606e-06,
+      "loss": 0.7655,
+      "step": 588
+    },
+    {
+      "epoch": 0.01923265306122449,
+      "grad_norm": 2.761504888534546,
+      "learning_rate": 9.9823971260614e-06,
+      "loss": 0.784,
+      "step": 589
+    },
+    {
+      "epoch": 0.01926530612244898,
+      "grad_norm": 3.1624319553375244,
+      "learning_rate": 9.982364467668192e-06,
+      "loss": 0.8538,
+      "step": 590
+    },
+    {
+      "epoch": 0.01929795918367347,
+      "grad_norm": 2.840982675552368,
+      "learning_rate": 9.982331809274984e-06,
+      "loss": 0.7879,
+      "step": 591
+    },
+    {
+      "epoch": 0.01933061224489796,
+      "grad_norm": 2.8063557147979736,
+      "learning_rate": 9.982299150881777e-06,
+      "loss": 0.7858,
+      "step": 592
+    },
+    {
+      "epoch": 0.019363265306122448,
+      "grad_norm": 2.761817455291748,
+      "learning_rate": 9.98226649248857e-06,
+      "loss": 0.7932,
+      "step": 593
+    },
+    {
+      "epoch": 0.01939591836734694,
+      "grad_norm": 2.845803737640381,
+      "learning_rate": 9.982233834095364e-06,
+      "loss": 0.8315,
+      "step": 594
+    },
+    {
+      "epoch": 0.019428571428571427,
+      "grad_norm": 2.820237398147583,
+      "learning_rate": 9.982201175702157e-06,
+      "loss": 0.8115,
+      "step": 595
+    },
+    {
+      "epoch": 0.01946122448979592,
+      "grad_norm": 2.8963451385498047,
+      "learning_rate": 9.982168517308948e-06,
+      "loss": 0.7965,
+      "step": 596
+    },
+    {
+      "epoch": 0.01949387755102041,
+      "grad_norm": 2.8076157569885254,
+      "learning_rate": 9.982135858915742e-06,
+      "loss": 0.8205,
+      "step": 597
+    },
+    {
+      "epoch": 0.019526530612244897,
+      "grad_norm": 2.9265244007110596,
+      "learning_rate": 9.982103200522535e-06,
+      "loss": 0.7975,
+      "step": 598
+    },
+    {
+      "epoch": 0.01955918367346939,
+      "grad_norm": 3.044168472290039,
+      "learning_rate": 9.982070542129328e-06,
+      "loss": 0.7738,
+      "step": 599
+    },
+    {
+      "epoch": 0.019591836734693877,
+      "grad_norm": 2.991462469100952,
+      "learning_rate": 9.982037883736121e-06,
+      "loss": 0.808,
+      "step": 600
+    },
+    {
+      "epoch": 0.019591836734693877,
+      "eval_loss": 0.8216409087181091,
+      "eval_runtime": 96.6305,
+      "eval_samples_per_second": 1.035,
+      "eval_steps_per_second": 1.035,
+      "step": 600
+    },
+    {
+      "epoch": 0.019624489795918368,
+      "grad_norm": 2.856177568435669,
+      "learning_rate": 9.982005225342915e-06,
+      "loss": 0.8002,
+      "step": 601
+    },
+    {
+      "epoch": 0.019657142857142856,
+      "grad_norm": 2.881652355194092,
+      "learning_rate": 9.981972566949706e-06,
+      "loss": 0.8012,
+      "step": 602
+    },
+    {
+      "epoch": 0.019689795918367347,
+      "grad_norm": 2.9178857803344727,
+      "learning_rate": 9.9819399085565e-06,
+      "loss": 0.7724,
+      "step": 603
+    },
+    {
+      "epoch": 0.019722448979591838,
+      "grad_norm": 2.7211968898773193,
+      "learning_rate": 9.981907250163293e-06,
+      "loss": 0.798,
+      "step": 604
+    },
+    {
+      "epoch": 0.019755102040816326,
+      "grad_norm": 3.050656795501709,
+      "learning_rate": 9.981874591770086e-06,
+      "loss": 0.8564,
+      "step": 605
+    },
+    {
+      "epoch": 0.019787755102040817,
+      "grad_norm": 2.9224750995635986,
+      "learning_rate": 9.981841933376879e-06,
+      "loss": 0.8058,
+      "step": 606
+    },
+    {
+      "epoch": 0.019820408163265305,
+      "grad_norm": 2.9750068187713623,
+      "learning_rate": 9.98180927498367e-06,
+      "loss": 0.7774,
+      "step": 607
+    },
+    {
+      "epoch": 0.019853061224489797,
+      "grad_norm": 3.0843162536621094,
+      "learning_rate": 9.981776616590466e-06,
+      "loss": 0.8017,
+      "step": 608
+    },
+    {
+      "epoch": 0.019885714285714284,
+      "grad_norm": 3.043372631072998,
+      "learning_rate": 9.981743958197257e-06,
+      "loss": 0.7763,
+      "step": 609
+    },
+    {
+      "epoch": 0.019918367346938776,
+      "grad_norm": 2.8992819786071777,
+      "learning_rate": 9.98171129980405e-06,
+      "loss": 0.8483,
+      "step": 610
+    },
+    {
+      "epoch": 0.019951020408163267,
+      "grad_norm": 2.5121912956237793,
+      "learning_rate": 9.981678641410844e-06,
+      "loss": 0.8014,
+      "step": 611
+    },
+    {
+      "epoch": 0.019983673469387755,
+      "grad_norm": 2.8126091957092285,
+      "learning_rate": 9.981645983017637e-06,
+      "loss": 0.8,
+      "step": 612
+    },
+    {
+      "epoch": 0.020016326530612246,
+      "grad_norm": 2.760281801223755,
+      "learning_rate": 9.98161332462443e-06,
+      "loss": 0.8089,
+      "step": 613
+    },
+    {
+      "epoch": 0.020048979591836734,
+      "grad_norm": 2.75252103805542,
+      "learning_rate": 9.981580666231222e-06,
+      "loss": 0.7991,
+      "step": 614
+    },
+    {
+      "epoch": 0.020081632653061225,
+      "grad_norm": 2.7450368404388428,
+      "learning_rate": 9.981548007838015e-06,
+      "loss": 0.7995,
+      "step": 615
+    },
+    {
+      "epoch": 0.020114285714285713,
+      "grad_norm": 2.752061367034912,
+      "learning_rate": 9.981515349444808e-06,
+      "loss": 0.8167,
+      "step": 616
+    },
+    {
+      "epoch": 0.020146938775510204,
+      "grad_norm": 2.882789134979248,
+      "learning_rate": 9.981482691051601e-06,
+      "loss": 0.8438,
+      "step": 617
+    },
+    {
+      "epoch": 0.020179591836734692,
+      "grad_norm": 2.8994760513305664,
+      "learning_rate": 9.981450032658395e-06,
+      "loss": 0.7794,
+      "step": 618
+    },
+    {
+      "epoch": 0.020212244897959183,
+      "grad_norm": 2.925473928451538,
+      "learning_rate": 9.981417374265186e-06,
+      "loss": 0.798,
+      "step": 619
+    },
+    {
+      "epoch": 0.020244897959183675,
+      "grad_norm": 2.909290075302124,
+      "learning_rate": 9.98138471587198e-06,
+      "loss": 0.8487,
+      "step": 620
+    },
+    {
+      "epoch": 0.020277551020408163,
+      "grad_norm": 3.004021644592285,
+      "learning_rate": 9.981352057478773e-06,
+      "loss": 0.8214,
+      "step": 621
+    },
+    {
+      "epoch": 0.020310204081632654,
+      "grad_norm": 2.850106954574585,
+      "learning_rate": 9.981319399085566e-06,
+      "loss": 0.8158,
+      "step": 622
+    },
+    {
+      "epoch": 0.02034285714285714,
+      "grad_norm": 2.903092622756958,
+      "learning_rate": 9.981286740692359e-06,
+      "loss": 0.8078,
+      "step": 623
+    },
+    {
+      "epoch": 0.020375510204081633,
+      "grad_norm": 2.9145193099975586,
+      "learning_rate": 9.981254082299152e-06,
+      "loss": 0.8224,
+      "step": 624
+    },
+    {
+      "epoch": 0.02040816326530612,
+      "grad_norm": 3.0473721027374268,
+      "learning_rate": 9.981221423905944e-06,
+      "loss": 0.7681,
+      "step": 625
+    },
+    {
+      "epoch": 0.020440816326530612,
+      "grad_norm": 2.8615925312042236,
+      "learning_rate": 9.981188765512737e-06,
+      "loss": 0.815,
+      "step": 626
+    },
+    {
+      "epoch": 0.020473469387755103,
+      "grad_norm": 2.8572824001312256,
+      "learning_rate": 9.98115610711953e-06,
+      "loss": 0.7672,
+      "step": 627
+    },
+    {
+      "epoch": 0.02050612244897959,
+      "grad_norm": 2.9113521575927734,
+      "learning_rate": 9.981123448726324e-06,
+      "loss": 0.7952,
+      "step": 628
+    },
+    {
+      "epoch": 0.020538775510204083,
+      "grad_norm": 3.0004873275756836,
+      "learning_rate": 9.981090790333117e-06,
+      "loss": 0.8181,
+      "step": 629
+    },
+    {
+      "epoch": 0.02057142857142857,
+      "grad_norm": 2.9250986576080322,
+      "learning_rate": 9.981058131939908e-06,
+      "loss": 0.7681,
+      "step": 630
+    },
+    {
+      "epoch": 0.02060408163265306,
+      "grad_norm": 3.1994357109069824,
+      "learning_rate": 9.981025473546703e-06,
+      "loss": 0.8292,
+      "step": 631
+    },
+    {
+      "epoch": 0.02063673469387755,
+      "grad_norm": 3.061445713043213,
+      "learning_rate": 9.980992815153495e-06,
+      "loss": 0.8241,
+      "step": 632
+    },
+    {
+      "epoch": 0.02066938775510204,
+      "grad_norm": 3.1536760330200195,
+      "learning_rate": 9.980960156760288e-06,
+      "loss": 0.7889,
+      "step": 633
+    },
+    {
+      "epoch": 0.020702040816326532,
+      "grad_norm": 2.7358641624450684,
+      "learning_rate": 9.980927498367081e-06,
+      "loss": 0.8167,
+      "step": 634
+    },
+    {
+      "epoch": 0.02073469387755102,
+      "grad_norm": 2.952697992324829,
+      "learning_rate": 9.980894839973874e-06,
+      "loss": 0.7741,
+      "step": 635
+    },
+    {
+      "epoch": 0.02076734693877551,
+      "grad_norm": 3.003535747528076,
+      "learning_rate": 9.980862181580668e-06,
+      "loss": 0.8101,
+      "step": 636
+    },
+    {
+      "epoch": 0.0208,
+      "grad_norm": 2.895209789276123,
+      "learning_rate": 9.98082952318746e-06,
+      "loss": 0.8025,
+      "step": 637
+    },
+    {
+      "epoch": 0.02083265306122449,
+      "grad_norm": 2.974769115447998,
+      "learning_rate": 9.980796864794252e-06,
+      "loss": 0.8236,
+      "step": 638
+    },
+    {
+      "epoch": 0.020865306122448978,
+      "grad_norm": 2.8859550952911377,
+      "learning_rate": 9.980764206401046e-06,
+      "loss": 0.8214,
+      "step": 639
+    },
+    {
+      "epoch": 0.02089795918367347,
+      "grad_norm": 3.1916568279266357,
+      "learning_rate": 9.980731548007839e-06,
+      "loss": 0.7618,
+      "step": 640
+    },
+    {
+      "epoch": 0.02093061224489796,
+      "grad_norm": 3.1790311336517334,
+      "learning_rate": 9.980698889614632e-06,
+      "loss": 0.779,
+      "step": 641
+    },
+    {
+      "epoch": 0.02096326530612245,
+      "grad_norm": 9.082115173339844,
+      "learning_rate": 9.980666231221424e-06,
+      "loss": 0.7972,
+      "step": 642
+    },
+    {
+      "epoch": 0.02099591836734694,
+      "grad_norm": 3.0865111351013184,
+      "learning_rate": 9.980633572828217e-06,
+      "loss": 0.7972,
+      "step": 643
+    },
+    {
+      "epoch": 0.021028571428571428,
+      "grad_norm": 3.3228487968444824,
+      "learning_rate": 9.98060091443501e-06,
+      "loss": 0.8245,
+      "step": 644
+    },
+    {
+      "epoch": 0.02106122448979592,
+      "grad_norm": 2.7116079330444336,
+      "learning_rate": 9.980568256041803e-06,
+      "loss": 0.8129,
+      "step": 645
+    },
+    {
+      "epoch": 0.021093877551020407,
+      "grad_norm": 2.9123730659484863,
+      "learning_rate": 9.980535597648597e-06,
+      "loss": 0.832,
+      "step": 646
+    },
+    {
+      "epoch": 0.021126530612244898,
+      "grad_norm": 3.1061103343963623,
+      "learning_rate": 9.98050293925539e-06,
+      "loss": 0.8067,
+      "step": 647
+    },
+    {
+      "epoch": 0.02115918367346939,
+      "grad_norm": 3.2617485523223877,
+      "learning_rate": 9.980470280862181e-06,
+      "loss": 0.7897,
+      "step": 648
+    },
+    {
+      "epoch": 0.021191836734693877,
+      "grad_norm": 3.0000083446502686,
+      "learning_rate": 9.980437622468976e-06,
+      "loss": 0.7663,
+      "step": 649
+    },
+    {
+      "epoch": 0.02122448979591837,
+      "grad_norm": 3.029799461364746,
+      "learning_rate": 9.980404964075768e-06,
+      "loss": 0.7688,
+      "step": 650
+    },
+    {
+      "epoch": 0.02122448979591837,
+      "eval_loss": 0.8194996118545532,
+      "eval_runtime": 74.2652,
+      "eval_samples_per_second": 1.347,
+      "eval_steps_per_second": 1.347,
+      "step": 650
+    },
+    {
+      "epoch": 0.021257142857142856,
+      "grad_norm": 2.762091875076294,
+      "learning_rate": 9.980372305682561e-06,
+      "loss": 0.8051,
+      "step": 651
+    },
+    {
+      "epoch": 0.021289795918367348,
+      "grad_norm": 3.030320167541504,
+      "learning_rate": 9.980339647289354e-06,
+      "loss": 0.8201,
+      "step": 652
+    },
+    {
+      "epoch": 0.021322448979591836,
+      "grad_norm": 2.9875736236572266,
+      "learning_rate": 9.980306988896148e-06,
+      "loss": 0.7841,
+      "step": 653
+    },
+    {
+      "epoch": 0.021355102040816327,
+      "grad_norm": 2.847587823867798,
+      "learning_rate": 9.980274330502941e-06,
+      "loss": 0.7807,
+      "step": 654
+    },
+    {
+      "epoch": 0.021387755102040815,
+      "grad_norm": 3.0852084159851074,
+      "learning_rate": 9.980241672109732e-06,
+      "loss": 0.7606,
+      "step": 655
+    },
+    {
+      "epoch": 0.021420408163265306,
+      "grad_norm": 3.159208297729492,
+      "learning_rate": 9.980209013716526e-06,
+      "loss": 0.8011,
+      "step": 656
+    },
+    {
+      "epoch": 0.021453061224489797,
+      "grad_norm": 3.063185214996338,
+      "learning_rate": 9.980176355323319e-06,
+      "loss": 0.7839,
+      "step": 657
+    },
+    {
+      "epoch": 0.021485714285714285,
+      "grad_norm": 2.9843311309814453,
+      "learning_rate": 9.980143696930112e-06,
+      "loss": 0.7961,
+      "step": 658
+    },
+    {
+      "epoch": 0.021518367346938776,
+      "grad_norm": 3.0662293434143066,
+      "learning_rate": 9.980111038536905e-06,
+      "loss": 0.8065,
+      "step": 659
+    },
+    {
+      "epoch": 0.021551020408163264,
+      "grad_norm": 2.9915835857391357,
+      "learning_rate": 9.980078380143697e-06,
+      "loss": 0.8022,
+      "step": 660
+    },
+    {
+      "epoch": 0.021583673469387755,
+      "grad_norm": 2.7856531143188477,
+      "learning_rate": 9.98004572175049e-06,
+      "loss": 0.7853,
+      "step": 661
+    },
+    {
+      "epoch": 0.021616326530612243,
+      "grad_norm": 2.8300893306732178,
+      "learning_rate": 9.980013063357283e-06,
+      "loss": 0.8097,
+      "step": 662
+    },
+    {
+      "epoch": 0.021648979591836735,
+      "grad_norm": 2.7422966957092285,
+      "learning_rate": 9.979980404964077e-06,
+      "loss": 0.8307,
+      "step": 663
+    },
+    {
+      "epoch": 0.021681632653061226,
+      "grad_norm": 2.80092453956604,
+      "learning_rate": 9.97994774657087e-06,
+      "loss": 0.7932,
+      "step": 664
+    },
+    {
+      "epoch": 0.021714285714285714,
+      "grad_norm": 2.8894505500793457,
+      "learning_rate": 9.979915088177663e-06,
+      "loss": 0.8042,
+      "step": 665
+    },
+    {
+      "epoch": 0.021746938775510205,
+      "grad_norm": 2.7492778301239014,
+      "learning_rate": 9.979882429784455e-06,
+      "loss": 0.7817,
+      "step": 666
+    },
+    {
+      "epoch": 0.021779591836734693,
+      "grad_norm": 2.734226703643799,
+      "learning_rate": 9.979849771391248e-06,
+      "loss": 0.7984,
+      "step": 667
+    },
+    {
+      "epoch": 0.021812244897959184,
+      "grad_norm": 2.7127978801727295,
+      "learning_rate": 9.979817112998041e-06,
+      "loss": 0.7955,
+      "step": 668
+    },
+    {
+      "epoch": 0.021844897959183672,
+      "grad_norm": 2.881192207336426,
+      "learning_rate": 9.979784454604834e-06,
+      "loss": 0.8257,
+      "step": 669
+    },
+    {
+      "epoch": 0.021877551020408163,
+      "grad_norm": 3.02278995513916,
+      "learning_rate": 9.979751796211628e-06,
+      "loss": 0.7892,
+      "step": 670
+    },
+    {
+      "epoch": 0.021910204081632655,
+      "grad_norm": 3.157317876815796,
+      "learning_rate": 9.979719137818419e-06,
+      "loss": 0.8227,
+      "step": 671
+    },
+    {
+      "epoch": 0.021942857142857142,
+      "grad_norm": 2.876741409301758,
+      "learning_rate": 9.979686479425214e-06,
+      "loss": 0.8116,
+      "step": 672
+    },
+    {
+      "epoch": 0.021975510204081634,
+      "grad_norm": 3.1728057861328125,
+      "learning_rate": 9.979653821032006e-06,
+      "loss": 0.8207,
+      "step": 673
+    },
+    {
+      "epoch": 0.02200816326530612,
+      "grad_norm": 3.2653119564056396,
+      "learning_rate": 9.979621162638799e-06,
+      "loss": 0.8087,
+      "step": 674
+    },
+    {
+      "epoch": 0.022040816326530613,
+      "grad_norm": 3.0209734439849854,
+      "learning_rate": 9.979588504245592e-06,
+      "loss": 0.7958,
+      "step": 675
+    },
+    {
+      "epoch": 0.0220734693877551,
+      "grad_norm": 2.9444637298583984,
+      "learning_rate": 9.979555845852385e-06,
+      "loss": 0.8031,
+      "step": 676
+    },
+    {
+      "epoch": 0.022106122448979592,
+      "grad_norm": 2.8222014904022217,
+      "learning_rate": 9.979523187459179e-06,
+      "loss": 0.7842,
+      "step": 677
+    },
+    {
+      "epoch": 0.022138775510204083,
+      "grad_norm": 3.034346342086792,
+      "learning_rate": 9.97949052906597e-06,
+      "loss": 0.7979,
+      "step": 678
+    },
+    {
+      "epoch": 0.02217142857142857,
+      "grad_norm": 2.9454312324523926,
+      "learning_rate": 9.979457870672763e-06,
+      "loss": 0.8209,
+      "step": 679
+    },
+    {
+      "epoch": 0.022204081632653062,
+      "grad_norm": 2.9699203968048096,
+      "learning_rate": 9.979425212279557e-06,
+      "loss": 0.7783,
+      "step": 680
+    },
+    {
+      "epoch": 0.02223673469387755,
+      "grad_norm": 3.0586094856262207,
+      "learning_rate": 9.97939255388635e-06,
+      "loss": 0.8257,
+      "step": 681
+    },
+    {
+      "epoch": 0.02226938775510204,
+      "grad_norm": 3.1194658279418945,
+      "learning_rate": 9.979359895493143e-06,
+      "loss": 0.7824,
+      "step": 682
+    },
+    {
+      "epoch": 0.02230204081632653,
+      "grad_norm": 2.950078248977661,
+      "learning_rate": 9.979327237099935e-06,
+      "loss": 0.7866,
+      "step": 683
+    },
+    {
+      "epoch": 0.02233469387755102,
+      "grad_norm": 3.135443925857544,
+      "learning_rate": 9.979294578706728e-06,
+      "loss": 0.786,
+      "step": 684
+    },
+    {
+      "epoch": 0.02236734693877551,
+      "grad_norm": 3.1549785137176514,
+      "learning_rate": 9.979261920313521e-06,
+      "loss": 0.8063,
+      "step": 685
+    },
+    {
+      "epoch": 0.0224,
+      "grad_norm": 3.064702033996582,
+      "learning_rate": 9.979229261920314e-06,
+      "loss": 0.7789,
+      "step": 686
+    },
+    {
+      "epoch": 0.02243265306122449,
+      "grad_norm": 2.9797706604003906,
+      "learning_rate": 9.979196603527108e-06,
+      "loss": 0.8022,
+      "step": 687
+    },
+    {
+      "epoch": 0.02246530612244898,
+      "grad_norm": 3.0248265266418457,
+      "learning_rate": 9.9791639451339e-06,
+      "loss": 0.7935,
+      "step": 688
+    },
+    {
+      "epoch": 0.02249795918367347,
+      "grad_norm": 2.865000009536743,
+      "learning_rate": 9.979131286740692e-06,
+      "loss": 0.7539,
+      "step": 689
+    },
+    {
+      "epoch": 0.022530612244897958,
+      "grad_norm": 2.7899985313415527,
+      "learning_rate": 9.979098628347487e-06,
+      "loss": 0.766,
+      "step": 690
+    },
+    {
+      "epoch": 0.02256326530612245,
+      "grad_norm": 3.016523838043213,
+      "learning_rate": 9.979065969954279e-06,
+      "loss": 0.7904,
+      "step": 691
+    },
+    {
+      "epoch": 0.022595918367346937,
+      "grad_norm": 2.954990863800049,
+      "learning_rate": 9.979033311561072e-06,
+      "loss": 0.8276,
+      "step": 692
+    },
+    {
+      "epoch": 0.02262857142857143,
+      "grad_norm": 2.932527780532837,
+      "learning_rate": 9.979000653167865e-06,
+      "loss": 0.7614,
+      "step": 693
+    },
+    {
+      "epoch": 0.02266122448979592,
+      "grad_norm": 2.892082691192627,
+      "learning_rate": 9.978967994774657e-06,
+      "loss": 0.8026,
+      "step": 694
+    },
+    {
+      "epoch": 0.022693877551020408,
+      "grad_norm": 2.8288159370422363,
+      "learning_rate": 9.978935336381452e-06,
+      "loss": 0.7956,
+      "step": 695
+    },
+    {
+      "epoch": 0.0227265306122449,
+      "grad_norm": 2.8160433769226074,
+      "learning_rate": 9.978902677988243e-06,
+      "loss": 0.7765,
+      "step": 696
+    },
+    {
+      "epoch": 0.022759183673469387,
+      "grad_norm": 2.8383493423461914,
+      "learning_rate": 9.978870019595036e-06,
+      "loss": 0.7926,
+      "step": 697
+    },
+    {
+      "epoch": 0.022791836734693878,
+      "grad_norm": 3.058093786239624,
+      "learning_rate": 9.97883736120183e-06,
+      "loss": 0.7878,
+      "step": 698
+    },
+    {
+      "epoch": 0.022824489795918366,
+      "grad_norm": 3.0352556705474854,
+      "learning_rate": 9.978804702808623e-06,
+      "loss": 0.7885,
+      "step": 699
+    },
+    {
+      "epoch": 0.022857142857142857,
+      "grad_norm": 2.9252421855926514,
+      "learning_rate": 9.978772044415416e-06,
+      "loss": 0.7759,
+      "step": 700
+    },
+    {
+      "epoch": 0.022857142857142857,
+      "eval_loss": 0.8187767267227173,
+      "eval_runtime": 74.1789,
+      "eval_samples_per_second": 1.348,
+      "eval_steps_per_second": 1.348,
+      "step": 700
+    },
+    {
+      "epoch": 0.02288979591836735,
+      "grad_norm": 2.860393762588501,
+      "learning_rate": 9.978739386022208e-06,
+      "loss": 0.8177,
+      "step": 701
+    },
+    {
+      "epoch": 0.022922448979591836,
+      "grad_norm": 2.960848331451416,
+      "learning_rate": 9.978706727629001e-06,
+      "loss": 0.8298,
+      "step": 702
+    },
+    {
+      "epoch": 0.022955102040816328,
+      "grad_norm": 2.8897740840911865,
+      "learning_rate": 9.978674069235794e-06,
+      "loss": 0.7625,
+      "step": 703
+    },
+    {
+      "epoch": 0.022987755102040815,
+      "grad_norm": 3.0318431854248047,
+      "learning_rate": 9.978641410842587e-06,
+      "loss": 0.7759,
+      "step": 704
+    },
+    {
+      "epoch": 0.023020408163265307,
+      "grad_norm": 2.959118366241455,
+      "learning_rate": 9.97860875244938e-06,
+      "loss": 0.7921,
+      "step": 705
+    },
+    {
+      "epoch": 0.023053061224489795,
+      "grad_norm": 2.871290922164917,
+      "learning_rate": 9.978576094056174e-06,
+      "loss": 0.7943,
+      "step": 706
+    },
+    {
+      "epoch": 0.023085714285714286,
+      "grad_norm": 2.699939489364624,
+      "learning_rate": 9.978543435662965e-06,
+      "loss": 0.7708,
+      "step": 707
+    },
+    {
+      "epoch": 0.023118367346938777,
+      "grad_norm": 2.8150787353515625,
+      "learning_rate": 9.978510777269759e-06,
+      "loss": 0.7888,
+      "step": 708
+    },
+    {
+      "epoch": 0.023151020408163265,
+      "grad_norm": 2.9636101722717285,
+      "learning_rate": 9.978478118876552e-06,
+      "loss": 0.7861,
+      "step": 709
+    },
+    {
+      "epoch": 0.023183673469387756,
+      "grad_norm": 3.122624397277832,
+      "learning_rate": 9.978445460483345e-06,
+      "loss": 0.7967,
+      "step": 710
+    },
+    {
+      "epoch": 0.023216326530612244,
+      "grad_norm": 3.070082902908325,
+      "learning_rate": 9.978412802090138e-06,
+      "loss": 0.793,
+      "step": 711
+    },
+    {
+      "epoch": 0.023248979591836735,
+      "grad_norm": 3.063530206680298,
+      "learning_rate": 9.97838014369693e-06,
+      "loss": 0.8115,
+      "step": 712
+    },
+    {
+      "epoch": 0.023281632653061223,
+      "grad_norm": 2.8320116996765137,
+      "learning_rate": 9.978347485303725e-06,
+      "loss": 0.831,
+      "step": 713
+    },
+    {
+      "epoch": 0.023314285714285714,
+      "grad_norm": 2.9792158603668213,
+      "learning_rate": 9.978314826910516e-06,
+      "loss": 0.745,
+      "step": 714
+    },
+    {
+      "epoch": 0.023346938775510206,
+      "grad_norm": 2.9203131198883057,
+      "learning_rate": 9.97828216851731e-06,
+      "loss": 0.7912,
+      "step": 715
+    },
+    {
+      "epoch": 0.023379591836734694,
+      "grad_norm": 3.151167631149292,
+      "learning_rate": 9.978249510124103e-06,
+      "loss": 0.7687,
+      "step": 716
+    },
+    {
+      "epoch": 0.023412244897959185,
+      "grad_norm": 2.8722281455993652,
+      "learning_rate": 9.978216851730896e-06,
+      "loss": 0.7975,
+      "step": 717
+    },
+    {
+      "epoch": 0.023444897959183673,
+      "grad_norm": 2.989642858505249,
+      "learning_rate": 9.97818419333769e-06,
+      "loss": 0.8136,
+      "step": 718
+    },
+    {
+      "epoch": 0.023477551020408164,
+      "grad_norm": 3.2293035984039307,
+      "learning_rate": 9.978151534944481e-06,
+      "loss": 0.8469,
+      "step": 719
+    },
+    {
+      "epoch": 0.023510204081632652,
+      "grad_norm": 3.016899585723877,
+      "learning_rate": 9.978118876551274e-06,
+      "loss": 0.7901,
+      "step": 720
+    },
+    {
+      "epoch": 0.023542857142857143,
+      "grad_norm": 3.194211006164551,
+      "learning_rate": 9.978086218158067e-06,
+      "loss": 0.8112,
+      "step": 721
+    },
+    {
+      "epoch": 0.02357551020408163,
+      "grad_norm": 2.810654640197754,
+      "learning_rate": 9.97805355976486e-06,
+      "loss": 0.7878,
+      "step": 722
+    },
+    {
+      "epoch": 0.023608163265306122,
+      "grad_norm": 2.788594961166382,
+      "learning_rate": 9.978020901371654e-06,
+      "loss": 0.8245,
+      "step": 723
+    },
+    {
+      "epoch": 0.023640816326530614,
+      "grad_norm": 2.976698875427246,
+      "learning_rate": 9.977988242978445e-06,
+      "loss": 0.7869,
+      "step": 724
+    },
+    {
+      "epoch": 0.0236734693877551,
+      "grad_norm": 2.7467331886291504,
+      "learning_rate": 9.977955584585239e-06,
+      "loss": 0.7837,
+      "step": 725
+    },
+    {
+      "epoch": 0.023706122448979593,
+      "grad_norm": 2.8149194717407227,
+      "learning_rate": 9.977922926192032e-06,
+      "loss": 0.7713,
+      "step": 726
+    },
+    {
+      "epoch": 0.02373877551020408,
+      "grad_norm": 3.07743501663208,
+      "learning_rate": 9.977890267798825e-06,
+      "loss": 0.7519,
+      "step": 727
+    },
+    {
+      "epoch": 0.023771428571428572,
+      "grad_norm": 2.779167413711548,
+      "learning_rate": 9.977857609405618e-06,
+      "loss": 0.8015,
+      "step": 728
+    },
+    {
+      "epoch": 0.02380408163265306,
+      "grad_norm": 2.861452579498291,
+      "learning_rate": 9.977824951012412e-06,
+      "loss": 0.8097,
+      "step": 729
+    },
+    {
+      "epoch": 0.02383673469387755,
+      "grad_norm": 2.9489521980285645,
+      "learning_rate": 9.977792292619203e-06,
+      "loss": 0.7988,
+      "step": 730
+    },
+    {
+      "epoch": 0.023869387755102042,
+      "grad_norm": 2.7786002159118652,
+      "learning_rate": 9.977759634225996e-06,
+      "loss": 0.7828,
+      "step": 731
+    },
+    {
+      "epoch": 0.02390204081632653,
+      "grad_norm": 2.9412615299224854,
+      "learning_rate": 9.97772697583279e-06,
+      "loss": 0.7684,
+      "step": 732
+    },
+    {
+      "epoch": 0.02393469387755102,
+      "grad_norm": 2.9088659286499023,
+      "learning_rate": 9.977694317439583e-06,
+      "loss": 0.7899,
+      "step": 733
+    },
+    {
+      "epoch": 0.02396734693877551,
+      "grad_norm": 2.811553955078125,
+      "learning_rate": 9.977661659046376e-06,
+      "loss": 0.7705,
+      "step": 734
+    },
+    {
+      "epoch": 0.024,
+      "grad_norm": 2.7078611850738525,
+      "learning_rate": 9.977629000653168e-06,
+      "loss": 0.7803,
+      "step": 735
+    },
+    {
+      "epoch": 0.02403265306122449,
+      "grad_norm": 2.9198622703552246,
+      "learning_rate": 9.977596342259963e-06,
+      "loss": 0.794,
+      "step": 736
+    },
+    {
+      "epoch": 0.02406530612244898,
+      "grad_norm": 3.1856372356414795,
+      "learning_rate": 9.977563683866754e-06,
+      "loss": 0.8146,
+      "step": 737
+    },
+    {
+      "epoch": 0.02409795918367347,
+      "grad_norm": 2.914483070373535,
+      "learning_rate": 9.977531025473547e-06,
+      "loss": 0.813,
+      "step": 738
+    },
+    {
+      "epoch": 0.02413061224489796,
+      "grad_norm": 2.837502956390381,
+      "learning_rate": 9.97749836708034e-06,
+      "loss": 0.7807,
+      "step": 739
+    },
+    {
+      "epoch": 0.02416326530612245,
+      "grad_norm": 2.7452003955841064,
+      "learning_rate": 9.977465708687134e-06,
+      "loss": 0.8272,
+      "step": 740
+    },
+    {
+      "epoch": 0.024195918367346938,
+      "grad_norm": 2.922658681869507,
+      "learning_rate": 9.977433050293927e-06,
+      "loss": 0.8096,
+      "step": 741
+    },
+    {
+      "epoch": 0.02422857142857143,
+      "grad_norm": 2.9923369884490967,
+      "learning_rate": 9.977400391900719e-06,
+      "loss": 0.8232,
+      "step": 742
+    },
+    {
+      "epoch": 0.024261224489795917,
+      "grad_norm": 2.879516363143921,
+      "learning_rate": 9.977367733507512e-06,
+      "loss": 0.8128,
+      "step": 743
+    },
+    {
+      "epoch": 0.02429387755102041,
+      "grad_norm": 2.7809741497039795,
+      "learning_rate": 9.977335075114305e-06,
+      "loss": 0.7824,
+      "step": 744
+    },
+    {
+      "epoch": 0.0243265306122449,
+      "grad_norm": 2.8293521404266357,
+      "learning_rate": 9.977302416721098e-06,
+      "loss": 0.7637,
+      "step": 745
+    },
+    {
+      "epoch": 0.024359183673469387,
+      "grad_norm": 2.900975227355957,
+      "learning_rate": 9.977269758327891e-06,
+      "loss": 0.8065,
+      "step": 746
+    },
+    {
+      "epoch": 0.02439183673469388,
+      "grad_norm": 2.8670005798339844,
+      "learning_rate": 9.977237099934683e-06,
+      "loss": 0.7747,
+      "step": 747
+    },
+    {
+      "epoch": 0.024424489795918367,
+      "grad_norm": 2.987907886505127,
+      "learning_rate": 9.977204441541476e-06,
+      "loss": 0.7907,
+      "step": 748
+    },
+    {
+      "epoch": 0.024457142857142858,
+      "grad_norm": 2.8523452281951904,
+      "learning_rate": 9.97717178314827e-06,
+      "loss": 0.781,
+      "step": 749
+    },
+    {
+      "epoch": 0.024489795918367346,
+      "grad_norm": 2.9037506580352783,
+      "learning_rate": 9.977139124755063e-06,
+      "loss": 0.7505,
+      "step": 750
+    },
+    {
+      "epoch": 0.024489795918367346,
+      "eval_loss": 0.8159348964691162,
+      "eval_runtime": 74.1956,
+      "eval_samples_per_second": 1.348,
+      "eval_steps_per_second": 1.348,
+      "step": 750
+    },
+    {
+      "epoch": 0.024522448979591837,
+      "grad_norm": 2.8492870330810547,
+      "learning_rate": 9.977106466361856e-06,
+      "loss": 0.8186,
+      "step": 751
+    },
+    {
+      "epoch": 0.024555102040816325,
+      "grad_norm": 3.0403406620025635,
+      "learning_rate": 9.97707380796865e-06,
+      "loss": 0.8152,
+      "step": 752
+    },
+    {
+      "epoch": 0.024587755102040816,
+      "grad_norm": 2.8419723510742188,
+      "learning_rate": 9.97704114957544e-06,
+      "loss": 0.7908,
+      "step": 753
+    },
+    {
+      "epoch": 0.024620408163265307,
+      "grad_norm": 3.0176303386688232,
+      "learning_rate": 9.977008491182236e-06,
+      "loss": 0.7921,
+      "step": 754
+    },
+    {
+      "epoch": 0.024653061224489795,
+      "grad_norm": 2.928659439086914,
+      "learning_rate": 9.976975832789027e-06,
+      "loss": 0.7784,
+      "step": 755
+    },
+    {
+      "epoch": 0.024685714285714287,
+      "grad_norm": 2.800809621810913,
+      "learning_rate": 9.97694317439582e-06,
+      "loss": 0.7362,
+      "step": 756
+    },
+    {
+      "epoch": 0.024718367346938774,
+      "grad_norm": 2.698071002960205,
+      "learning_rate": 9.976910516002614e-06,
+      "loss": 0.8046,
+      "step": 757
+    },
+    {
+      "epoch": 0.024751020408163266,
+      "grad_norm": 2.8206839561462402,
+      "learning_rate": 9.976877857609407e-06,
+      "loss": 0.8169,
+      "step": 758
+    },
+    {
+      "epoch": 0.024783673469387753,
+      "grad_norm": 2.927217960357666,
+      "learning_rate": 9.9768451992162e-06,
+      "loss": 0.7816,
+      "step": 759
+    },
+    {
+      "epoch": 0.024816326530612245,
+      "grad_norm": 2.9845449924468994,
+      "learning_rate": 9.976812540822992e-06,
+      "loss": 0.801,
+      "step": 760
+    },
+    {
+      "epoch": 0.024848979591836736,
+      "grad_norm": 2.881765604019165,
+      "learning_rate": 9.976779882429785e-06,
+      "loss": 0.8052,
+      "step": 761
+    },
+    {
+      "epoch": 0.024881632653061224,
+      "grad_norm": 2.9570679664611816,
+      "learning_rate": 9.976747224036578e-06,
+      "loss": 0.8044,
+      "step": 762
+    },
+    {
+      "epoch": 0.024914285714285715,
+      "grad_norm": 3.069812774658203,
+      "learning_rate": 9.976714565643371e-06,
+      "loss": 0.8166,
+      "step": 763
+    },
+    {
+      "epoch": 0.024946938775510203,
+      "grad_norm": 3.0275254249572754,
+      "learning_rate": 9.976681907250165e-06,
+      "loss": 0.7835,
+      "step": 764
+    },
+    {
+      "epoch": 0.024979591836734694,
+      "grad_norm": 3.007643461227417,
+      "learning_rate": 9.976649248856956e-06,
+      "loss": 0.8088,
+      "step": 765
+    },
+    {
+      "epoch": 0.025012244897959182,
+      "grad_norm": 2.788733720779419,
+      "learning_rate": 9.97661659046375e-06,
+      "loss": 0.7852,
+      "step": 766
+    },
+    {
+      "epoch": 0.025044897959183673,
+      "grad_norm": 3.187893867492676,
+      "learning_rate": 9.976583932070543e-06,
+      "loss": 0.7887,
+      "step": 767
+    },
+    {
+      "epoch": 0.025077551020408165,
+      "grad_norm": 3.0475502014160156,
+      "learning_rate": 9.976551273677336e-06,
+      "loss": 0.7837,
+      "step": 768
+    },
+    {
+      "epoch": 0.025110204081632653,
+      "grad_norm": 2.9664242267608643,
+      "learning_rate": 9.976518615284129e-06,
+      "loss": 0.8021,
+      "step": 769
+    },
+    {
+      "epoch": 0.025142857142857144,
+      "grad_norm": 3.1118979454040527,
+      "learning_rate": 9.976485956890922e-06,
+      "loss": 0.7992,
+      "step": 770
+    },
+    {
+      "epoch": 0.02517551020408163,
+      "grad_norm": 2.909301280975342,
+      "learning_rate": 9.976453298497714e-06,
+      "loss": 0.743,
+      "step": 771
+    },
+    {
+      "epoch": 0.025208163265306123,
+      "grad_norm": 2.9557557106018066,
+      "learning_rate": 9.976420640104507e-06,
+      "loss": 0.8007,
+      "step": 772
+    },
+    {
+      "epoch": 0.02524081632653061,
+      "grad_norm": 3.108922243118286,
+      "learning_rate": 9.9763879817113e-06,
+      "loss": 0.8127,
+      "step": 773
+    },
+    {
+      "epoch": 0.025273469387755102,
+      "grad_norm": 2.927797555923462,
+      "learning_rate": 9.976355323318094e-06,
+      "loss": 0.7577,
+      "step": 774
+    },
+    {
+      "epoch": 0.025306122448979593,
+      "grad_norm": 2.929468870162964,
+      "learning_rate": 9.976322664924887e-06,
+      "loss": 0.7662,
+      "step": 775
+    },
+    {
+      "epoch": 0.02533877551020408,
+      "grad_norm": 3.0343034267425537,
+      "learning_rate": 9.976290006531678e-06,
+      "loss": 0.8283,
+      "step": 776
+    },
+    {
+      "epoch": 0.025371428571428573,
+      "grad_norm": 3.136800765991211,
+      "learning_rate": 9.976257348138473e-06,
+      "loss": 0.7935,
+      "step": 777
+    },
+    {
+      "epoch": 0.02540408163265306,
+      "grad_norm": 2.916213035583496,
+      "learning_rate": 9.976224689745265e-06,
+      "loss": 0.7651,
+      "step": 778
+    },
+    {
+      "epoch": 0.02543673469387755,
+      "grad_norm": 2.929622173309326,
+      "learning_rate": 9.976192031352058e-06,
+      "loss": 0.7677,
+      "step": 779
+    },
+    {
+      "epoch": 0.02546938775510204,
+      "grad_norm": 3.2419509887695312,
+      "learning_rate": 9.976159372958851e-06,
+      "loss": 0.7789,
+      "step": 780
+    },
+    {
+      "epoch": 0.02550204081632653,
+      "grad_norm": 2.8888614177703857,
+      "learning_rate": 9.976126714565645e-06,
+      "loss": 0.7735,
+      "step": 781
+    },
+    {
+      "epoch": 0.025534693877551022,
+      "grad_norm": 3.1505699157714844,
+      "learning_rate": 9.976094056172438e-06,
+      "loss": 0.8128,
+      "step": 782
+    },
+    {
+      "epoch": 0.02556734693877551,
+      "grad_norm": 3.030273675918579,
+      "learning_rate": 9.97606139777923e-06,
+      "loss": 0.7678,
+      "step": 783
+    },
+    {
+      "epoch": 0.0256,
+      "grad_norm": 3.0003929138183594,
+      "learning_rate": 9.976028739386023e-06,
+      "loss": 0.8076,
+      "step": 784
+    },
+    {
+      "epoch": 0.02563265306122449,
+      "grad_norm": 2.957667350769043,
+      "learning_rate": 9.975996080992816e-06,
+      "loss": 0.7886,
+      "step": 785
+    },
+    {
+      "epoch": 0.02566530612244898,
+      "grad_norm": 3.0816612243652344,
+      "learning_rate": 9.975963422599609e-06,
+      "loss": 0.7871,
+      "step": 786
+    },
+    {
+      "epoch": 0.025697959183673468,
+      "grad_norm": 2.947930097579956,
+      "learning_rate": 9.975930764206402e-06,
+      "loss": 0.8288,
+      "step": 787
+    },
+    {
+      "epoch": 0.02573061224489796,
+      "grad_norm": 3.121185302734375,
+      "learning_rate": 9.975898105813194e-06,
+      "loss": 0.793,
+      "step": 788
+    },
+    {
+      "epoch": 0.025763265306122447,
+      "grad_norm": 3.0452706813812256,
+      "learning_rate": 9.975865447419987e-06,
+      "loss": 0.7819,
+      "step": 789
+    },
+    {
+      "epoch": 0.02579591836734694,
+      "grad_norm": 2.903444528579712,
+      "learning_rate": 9.97583278902678e-06,
+      "loss": 0.798,
+      "step": 790
+    },
+    {
+      "epoch": 0.02582857142857143,
+      "grad_norm": 3.088996410369873,
+      "learning_rate": 9.975800130633574e-06,
+      "loss": 0.8171,
+      "step": 791
+    },
+    {
+      "epoch": 0.025861224489795918,
+      "grad_norm": 2.957293748855591,
+      "learning_rate": 9.975767472240367e-06,
+      "loss": 0.7862,
+      "step": 792
+    },
+    {
+      "epoch": 0.02589387755102041,
+      "grad_norm": 2.7569499015808105,
+      "learning_rate": 9.97573481384716e-06,
+      "loss": 0.7899,
+      "step": 793
+    },
+    {
+      "epoch": 0.025926530612244897,
+      "grad_norm": 2.931257724761963,
+      "learning_rate": 9.975702155453952e-06,
+      "loss": 0.8051,
+      "step": 794
+    },
+    {
+      "epoch": 0.025959183673469388,
+      "grad_norm": 2.9962334632873535,
+      "learning_rate": 9.975669497060746e-06,
+      "loss": 0.7914,
+      "step": 795
+    },
+    {
+      "epoch": 0.025991836734693876,
+      "grad_norm": 3.016993761062622,
+      "learning_rate": 9.975636838667538e-06,
+      "loss": 0.8011,
+      "step": 796
+    },
+    {
+      "epoch": 0.026024489795918367,
+      "grad_norm": 3.066710948944092,
+      "learning_rate": 9.975604180274331e-06,
+      "loss": 0.7885,
+      "step": 797
+    },
+    {
+      "epoch": 0.02605714285714286,
+      "grad_norm": 2.9981632232666016,
+      "learning_rate": 9.975571521881125e-06,
+      "loss": 0.7968,
+      "step": 798
+    },
+    {
+      "epoch": 0.026089795918367346,
+      "grad_norm": 2.971733331680298,
+      "learning_rate": 9.975538863487916e-06,
+      "loss": 0.7997,
+      "step": 799
+    },
+    {
+      "epoch": 0.026122448979591838,
+      "grad_norm": 3.0647032260894775,
+      "learning_rate": 9.975506205094711e-06,
+      "loss": 0.8122,
+      "step": 800
+    },
+    {
+      "epoch": 0.026122448979591838,
+      "eval_loss": 0.8141899108886719,
+      "eval_runtime": 73.6209,
+      "eval_samples_per_second": 1.358,
+      "eval_steps_per_second": 1.358,
+      "step": 800
+    },
+    {
+      "epoch": 0.026155102040816326,
+      "grad_norm": 3.015854597091675,
+      "learning_rate": 9.975473546701503e-06,
+      "loss": 0.8062,
+      "step": 801
+    },
+    {
+      "epoch": 0.026187755102040817,
+      "grad_norm": 3.065284013748169,
+      "learning_rate": 9.975440888308296e-06,
+      "loss": 0.7973,
+      "step": 802
+    },
+    {
+      "epoch": 0.026220408163265305,
+      "grad_norm": 2.8390414714813232,
+      "learning_rate": 9.975408229915089e-06,
+      "loss": 0.7943,
+      "step": 803
+    },
+    {
+      "epoch": 0.026253061224489796,
+      "grad_norm": 3.254967451095581,
+      "learning_rate": 9.975375571521882e-06,
+      "loss": 0.8059,
+      "step": 804
+    },
+    {
+      "epoch": 0.026285714285714287,
+      "grad_norm": 3.0705697536468506,
+      "learning_rate": 9.975342913128675e-06,
+      "loss": 0.7871,
+      "step": 805
+    },
+    {
+      "epoch": 0.026318367346938775,
+      "grad_norm": 2.900188446044922,
+      "learning_rate": 9.975310254735467e-06,
+      "loss": 0.7672,
+      "step": 806
+    },
+    {
+      "epoch": 0.026351020408163266,
+      "grad_norm": 2.8207056522369385,
+      "learning_rate": 9.97527759634226e-06,
+      "loss": 0.7921,
+      "step": 807
+    },
+    {
+      "epoch": 0.026383673469387754,
+      "grad_norm": 2.891328811645508,
+      "learning_rate": 9.975244937949053e-06,
+      "loss": 0.8177,
+      "step": 808
+    },
+    {
+      "epoch": 0.026416326530612246,
+      "grad_norm": 2.974545955657959,
+      "learning_rate": 9.975212279555847e-06,
+      "loss": 0.7773,
+      "step": 809
+    },
+    {
+      "epoch": 0.026448979591836733,
+      "grad_norm": 2.8445703983306885,
+      "learning_rate": 9.97517962116264e-06,
+      "loss": 0.8131,
+      "step": 810
+    },
+    {
+      "epoch": 0.026481632653061225,
+      "grad_norm": 2.93383526802063,
+      "learning_rate": 9.975146962769433e-06,
+      "loss": 0.7693,
+      "step": 811
+    },
+    {
+      "epoch": 0.026514285714285716,
+      "grad_norm": 3.0080511569976807,
+      "learning_rate": 9.975114304376225e-06,
+      "loss": 0.7593,
+      "step": 812
+    },
+    {
+      "epoch": 0.026546938775510204,
+      "grad_norm": 2.967305898666382,
+      "learning_rate": 9.975081645983018e-06,
+      "loss": 0.7773,
+      "step": 813
+    },
+    {
+      "epoch": 0.026579591836734695,
+      "grad_norm": 2.9201626777648926,
+      "learning_rate": 9.975048987589811e-06,
+      "loss": 0.7908,
+      "step": 814
+    },
+    {
+      "epoch": 0.026612244897959183,
+      "grad_norm": 3.0869758129119873,
+      "learning_rate": 9.975016329196604e-06,
+      "loss": 0.8015,
+      "step": 815
+    },
+    {
+      "epoch": 0.026644897959183674,
+      "grad_norm": 3.120110511779785,
+      "learning_rate": 9.974983670803398e-06,
+      "loss": 0.7993,
+      "step": 816
+    },
+    {
+      "epoch": 0.026677551020408162,
+      "grad_norm": 3.1260945796966553,
+      "learning_rate": 9.97495101241019e-06,
+      "loss": 0.8407,
+      "step": 817
+    },
+    {
+      "epoch": 0.026710204081632653,
+      "grad_norm": 2.858689785003662,
+      "learning_rate": 9.974918354016984e-06,
+      "loss": 0.8001,
+      "step": 818
+    },
+    {
+      "epoch": 0.02674285714285714,
+      "grad_norm": 2.854126214981079,
+      "learning_rate": 9.974885695623776e-06,
+      "loss": 0.754,
+      "step": 819
+    },
+    {
+      "epoch": 0.026775510204081632,
+      "grad_norm": 2.932650566101074,
+      "learning_rate": 9.974853037230569e-06,
+      "loss": 0.8051,
+      "step": 820
+    },
+    {
+      "epoch": 0.026808163265306124,
+      "grad_norm": 3.063242197036743,
+      "learning_rate": 9.974820378837362e-06,
+      "loss": 0.8114,
+      "step": 821
+    },
+    {
+      "epoch": 0.02684081632653061,
+      "grad_norm": 2.8537795543670654,
+      "learning_rate": 9.974787720444155e-06,
+      "loss": 0.7842,
+      "step": 822
+    },
+    {
+      "epoch": 0.026873469387755103,
+      "grad_norm": 2.986083745956421,
+      "learning_rate": 9.974755062050949e-06,
+      "loss": 0.7949,
+      "step": 823
+    },
+    {
+      "epoch": 0.02690612244897959,
+      "grad_norm": 3.0178754329681396,
+      "learning_rate": 9.97472240365774e-06,
+      "loss": 0.8005,
+      "step": 824
+    },
+    {
+      "epoch": 0.026938775510204082,
+      "grad_norm": 3.1746835708618164,
+      "learning_rate": 9.974689745264533e-06,
+      "loss": 0.7994,
+      "step": 825
+    },
+    {
+      "epoch": 0.02697142857142857,
+      "grad_norm": 2.8827731609344482,
+      "learning_rate": 9.974657086871327e-06,
+      "loss": 0.7764,
+      "step": 826
+    },
+    {
+      "epoch": 0.02700408163265306,
+      "grad_norm": 2.7529618740081787,
+      "learning_rate": 9.97462442847812e-06,
+      "loss": 0.8092,
+      "step": 827
+    },
+    {
+      "epoch": 0.027036734693877552,
+      "grad_norm": 2.8854501247406006,
+      "learning_rate": 9.974591770084913e-06,
+      "loss": 0.7713,
+      "step": 828
+    },
+    {
+      "epoch": 0.02706938775510204,
+      "grad_norm": 2.8492770195007324,
+      "learning_rate": 9.974559111691705e-06,
+      "loss": 0.75,
+      "step": 829
+    },
+    {
+      "epoch": 0.02710204081632653,
+      "grad_norm": 2.9942445755004883,
+      "learning_rate": 9.974526453298498e-06,
+      "loss": 0.7838,
+      "step": 830
+    },
+    {
+      "epoch": 0.02713469387755102,
+      "grad_norm": 2.9380862712860107,
+      "learning_rate": 9.974493794905291e-06,
+      "loss": 0.8111,
+      "step": 831
+    },
+    {
+      "epoch": 0.02716734693877551,
+      "grad_norm": 2.8200275897979736,
+      "learning_rate": 9.974461136512084e-06,
+      "loss": 0.7818,
+      "step": 832
+    },
+    {
+      "epoch": 0.0272,
+      "grad_norm": 3.0098073482513428,
+      "learning_rate": 9.974428478118878e-06,
+      "loss": 0.7786,
+      "step": 833
+    },
+    {
+      "epoch": 0.02723265306122449,
+      "grad_norm": 2.96885347366333,
+      "learning_rate": 9.97439581972567e-06,
+      "loss": 0.762,
+      "step": 834
+    },
+    {
+      "epoch": 0.02726530612244898,
+      "grad_norm": 2.9910387992858887,
+      "learning_rate": 9.974363161332462e-06,
+      "loss": 0.8142,
+      "step": 835
+    },
+    {
+      "epoch": 0.02729795918367347,
+      "grad_norm": 3.1752495765686035,
+      "learning_rate": 9.974330502939257e-06,
+      "loss": 0.8031,
+      "step": 836
+    },
+    {
+      "epoch": 0.02733061224489796,
+      "grad_norm": 2.9318926334381104,
+      "learning_rate": 9.974297844546049e-06,
+      "loss": 0.7873,
+      "step": 837
+    },
+    {
+      "epoch": 0.027363265306122448,
+      "grad_norm": 2.8536527156829834,
+      "learning_rate": 9.974265186152842e-06,
+      "loss": 0.7878,
+      "step": 838
+    },
+    {
+      "epoch": 0.02739591836734694,
+      "grad_norm": 2.7921385765075684,
+      "learning_rate": 9.974232527759635e-06,
+      "loss": 0.8042,
+      "step": 839
+    },
+    {
+      "epoch": 0.027428571428571427,
+      "grad_norm": 2.97587251663208,
+      "learning_rate": 9.974199869366427e-06,
+      "loss": 0.8044,
+      "step": 840
+    },
+    {
+      "epoch": 0.02746122448979592,
+      "grad_norm": 2.879415273666382,
+      "learning_rate": 9.974167210973222e-06,
+      "loss": 0.7902,
+      "step": 841
+    },
+    {
+      "epoch": 0.02749387755102041,
+      "grad_norm": 2.930534839630127,
+      "learning_rate": 9.974134552580013e-06,
+      "loss": 0.7809,
+      "step": 842
+    },
+    {
+      "epoch": 0.027526530612244898,
+      "grad_norm": 2.8264150619506836,
+      "learning_rate": 9.974101894186807e-06,
+      "loss": 0.7565,
+      "step": 843
+    },
+    {
+      "epoch": 0.02755918367346939,
+      "grad_norm": 2.879453182220459,
+      "learning_rate": 9.9740692357936e-06,
+      "loss": 0.8139,
+      "step": 844
+    },
+    {
+      "epoch": 0.027591836734693877,
+      "grad_norm": 2.805103063583374,
+      "learning_rate": 9.974036577400393e-06,
+      "loss": 0.798,
+      "step": 845
+    },
+    {
+      "epoch": 0.027624489795918368,
+      "grad_norm": 2.8710720539093018,
+      "learning_rate": 9.974003919007186e-06,
+      "loss": 0.7412,
+      "step": 846
+    },
+    {
+      "epoch": 0.027657142857142856,
+      "grad_norm": 2.9521493911743164,
+      "learning_rate": 9.973971260613978e-06,
+      "loss": 0.7943,
+      "step": 847
+    },
+    {
+      "epoch": 0.027689795918367347,
+      "grad_norm": 3.062432050704956,
+      "learning_rate": 9.973938602220771e-06,
+      "loss": 0.7884,
+      "step": 848
+    },
+    {
+      "epoch": 0.02772244897959184,
+      "grad_norm": 2.98431658744812,
+      "learning_rate": 9.973905943827564e-06,
+      "loss": 0.7643,
+      "step": 849
+    },
+    {
+      "epoch": 0.027755102040816326,
+      "grad_norm": 2.852004289627075,
+      "learning_rate": 9.973873285434358e-06,
+      "loss": 0.8126,
+      "step": 850
+    },
+    {
+      "epoch": 0.027755102040816326,
+      "eval_loss": 0.8143028020858765,
+      "eval_runtime": 85.3794,
+      "eval_samples_per_second": 1.171,
+      "eval_steps_per_second": 1.171,
+      "step": 850
+    },
+    {
+      "epoch": 0.027787755102040818,
+      "grad_norm": 3.2032933235168457,
+      "learning_rate": 9.97384062704115e-06,
+      "loss": 0.7847,
+      "step": 851
+    },
+    {
+      "epoch": 0.027820408163265305,
+      "grad_norm": 3.0217716693878174,
+      "learning_rate": 9.973807968647944e-06,
+      "loss": 0.7812,
+      "step": 852
+    },
+    {
+      "epoch": 0.027853061224489797,
+      "grad_norm": 2.961899995803833,
+      "learning_rate": 9.973775310254736e-06,
+      "loss": 0.7917,
+      "step": 853
+    },
+    {
+      "epoch": 0.027885714285714285,
+      "grad_norm": 2.9481585025787354,
+      "learning_rate": 9.973742651861529e-06,
+      "loss": 0.7866,
+      "step": 854
+    },
+    {
+      "epoch": 0.027918367346938776,
+      "grad_norm": 2.8054749965667725,
+      "learning_rate": 9.973709993468322e-06,
+      "loss": 0.7808,
+      "step": 855
+    },
+    {
+      "epoch": 0.027951020408163264,
+      "grad_norm": 3.106367349624634,
+      "learning_rate": 9.973677335075115e-06,
+      "loss": 0.8281,
+      "step": 856
+    },
+    {
+      "epoch": 0.027983673469387755,
+      "grad_norm": 3.7496819496154785,
+      "learning_rate": 9.973644676681908e-06,
+      "loss": 0.7851,
+      "step": 857
+    },
+    {
+      "epoch": 0.028016326530612246,
+      "grad_norm": 2.7985293865203857,
+      "learning_rate": 9.9736120182887e-06,
+      "loss": 0.7843,
+      "step": 858
+    },
+    {
+      "epoch": 0.028048979591836734,
+      "grad_norm": 2.8734278678894043,
+      "learning_rate": 9.973579359895495e-06,
+      "loss": 0.7482,
+      "step": 859
+    },
+    {
+      "epoch": 0.028081632653061225,
+      "grad_norm": 3.0287320613861084,
+      "learning_rate": 9.973546701502287e-06,
+      "loss": 0.7901,
+      "step": 860
+    },
+    {
+      "epoch": 0.028114285714285713,
+      "grad_norm": 3.2069811820983887,
+      "learning_rate": 9.97351404310908e-06,
+      "loss": 0.7762,
+      "step": 861
+    },
+    {
+      "epoch": 0.028146938775510204,
+      "grad_norm": 3.193441152572632,
+      "learning_rate": 9.973481384715873e-06,
+      "loss": 0.7867,
+      "step": 862
+    },
+    {
+      "epoch": 0.028179591836734692,
+      "grad_norm": 2.932454824447632,
+      "learning_rate": 9.973448726322665e-06,
+      "loss": 0.801,
+      "step": 863
+    },
+    {
+      "epoch": 0.028212244897959184,
+      "grad_norm": 3.009593963623047,
+      "learning_rate": 9.97341606792946e-06,
+      "loss": 0.7612,
+      "step": 864
+    },
+    {
+      "epoch": 0.028244897959183675,
+      "grad_norm": 3.2580459117889404,
+      "learning_rate": 9.973383409536251e-06,
+      "loss": 0.7295,
+      "step": 865
+    },
+    {
+      "epoch": 0.028277551020408163,
+      "grad_norm": 2.934058904647827,
+      "learning_rate": 9.973350751143044e-06,
+      "loss": 0.7766,
+      "step": 866
+    },
+    {
+      "epoch": 0.028310204081632654,
+      "grad_norm": 3.0753285884857178,
+      "learning_rate": 9.973318092749837e-06,
+      "loss": 0.7647,
+      "step": 867
+    },
+    {
+      "epoch": 0.028342857142857142,
+      "grad_norm": 3.1178665161132812,
+      "learning_rate": 9.97328543435663e-06,
+      "loss": 0.8071,
+      "step": 868
+    },
+    {
+      "epoch": 0.028375510204081633,
+      "grad_norm": 3.235008955001831,
+      "learning_rate": 9.973252775963424e-06,
+      "loss": 0.7899,
+      "step": 869
+    },
+    {
+      "epoch": 0.02840816326530612,
+      "grad_norm": 3.198587417602539,
+      "learning_rate": 9.973220117570215e-06,
+      "loss": 0.7918,
+      "step": 870
+    },
+    {
+      "epoch": 0.028440816326530612,
+      "grad_norm": 3.1364388465881348,
+      "learning_rate": 9.973187459177009e-06,
+      "loss": 0.8178,
+      "step": 871
+    },
+    {
+      "epoch": 0.028473469387755104,
+      "grad_norm": 3.1181514263153076,
+      "learning_rate": 9.973154800783802e-06,
+      "loss": 0.7922,
+      "step": 872
+    },
+    {
+      "epoch": 0.02850612244897959,
+      "grad_norm": 3.0627710819244385,
+      "learning_rate": 9.973122142390595e-06,
+      "loss": 0.8063,
+      "step": 873
+    },
+    {
+      "epoch": 0.028538775510204083,
+      "grad_norm": 3.042802572250366,
+      "learning_rate": 9.973089483997388e-06,
+      "loss": 0.8304,
+      "step": 874
+    },
+    {
+      "epoch": 0.02857142857142857,
+      "grad_norm": 2.935323715209961,
+      "learning_rate": 9.973056825604182e-06,
+      "loss": 0.7983,
+      "step": 875
+    },
+    {
+      "epoch": 0.028604081632653062,
+      "grad_norm": 2.9549400806427,
+      "learning_rate": 9.973024167210973e-06,
+      "loss": 0.8063,
+      "step": 876
+    },
+    {
+      "epoch": 0.02863673469387755,
+      "grad_norm": 2.7875542640686035,
+      "learning_rate": 9.972991508817766e-06,
+      "loss": 0.7981,
+      "step": 877
+    },
+    {
+      "epoch": 0.02866938775510204,
+      "grad_norm": 2.9394519329071045,
+      "learning_rate": 9.97295885042456e-06,
+      "loss": 0.792,
+      "step": 878
+    },
+    {
+      "epoch": 0.028702040816326532,
+      "grad_norm": 3.0494203567504883,
+      "learning_rate": 9.972926192031353e-06,
+      "loss": 0.8128,
+      "step": 879
+    },
+    {
+      "epoch": 0.02873469387755102,
+      "grad_norm": 2.9686169624328613,
+      "learning_rate": 9.972893533638146e-06,
+      "loss": 0.7797,
+      "step": 880
+    },
+    {
+      "epoch": 0.02876734693877551,
+      "grad_norm": 2.964944839477539,
+      "learning_rate": 9.972860875244938e-06,
+      "loss": 0.7494,
+      "step": 881
+    },
+    {
+      "epoch": 0.0288,
+      "grad_norm": 2.9027297496795654,
+      "learning_rate": 9.972828216851733e-06,
+      "loss": 0.7526,
+      "step": 882
+    },
+    {
+      "epoch": 0.02883265306122449,
+      "grad_norm": 3.0997474193573,
+      "learning_rate": 9.972795558458524e-06,
+      "loss": 0.7878,
+      "step": 883
+    },
+    {
+      "epoch": 0.02886530612244898,
+      "grad_norm": 2.999011516571045,
+      "learning_rate": 9.972762900065317e-06,
+      "loss": 0.785,
+      "step": 884
+    },
+    {
+      "epoch": 0.02889795918367347,
+      "grad_norm": 3.1621158123016357,
+      "learning_rate": 9.97273024167211e-06,
+      "loss": 0.8338,
+      "step": 885
+    },
+    {
+      "epoch": 0.028930612244897957,
+      "grad_norm": 2.943216323852539,
+      "learning_rate": 9.972697583278904e-06,
+      "loss": 0.7844,
+      "step": 886
+    },
+    {
+      "epoch": 0.02896326530612245,
+      "grad_norm": 3.0958337783813477,
+      "learning_rate": 9.972664924885697e-06,
+      "loss": 0.7895,
+      "step": 887
+    },
+    {
+      "epoch": 0.02899591836734694,
+      "grad_norm": 2.982701539993286,
+      "learning_rate": 9.972632266492489e-06,
+      "loss": 0.7783,
+      "step": 888
+    },
+    {
+      "epoch": 0.029028571428571428,
+      "grad_norm": 2.9064059257507324,
+      "learning_rate": 9.972599608099282e-06,
+      "loss": 0.7915,
+      "step": 889
+    },
+    {
+      "epoch": 0.02906122448979592,
+      "grad_norm": 2.8960936069488525,
+      "learning_rate": 9.972566949706075e-06,
+      "loss": 0.7914,
+      "step": 890
+    },
+    {
+      "epoch": 0.029093877551020407,
+      "grad_norm": 2.881122350692749,
+      "learning_rate": 9.972534291312868e-06,
+      "loss": 0.7673,
+      "step": 891
+    },
+    {
+      "epoch": 0.0291265306122449,
+      "grad_norm": 2.9828569889068604,
+      "learning_rate": 9.972501632919662e-06,
+      "loss": 0.8089,
+      "step": 892
+    },
+    {
+      "epoch": 0.029159183673469386,
+      "grad_norm": 3.0593910217285156,
+      "learning_rate": 9.972468974526453e-06,
+      "loss": 0.8079,
+      "step": 893
+    },
+    {
+      "epoch": 0.029191836734693877,
+      "grad_norm": 2.993431568145752,
+      "learning_rate": 9.972436316133246e-06,
+      "loss": 0.7855,
+      "step": 894
+    },
+    {
+      "epoch": 0.02922448979591837,
+      "grad_norm": 3.132528066635132,
+      "learning_rate": 9.97240365774004e-06,
+      "loss": 0.7948,
+      "step": 895
+    },
+    {
+      "epoch": 0.029257142857142857,
+      "grad_norm": 3.0954887866973877,
+      "learning_rate": 9.972370999346833e-06,
+      "loss": 0.765,
+      "step": 896
+    },
+    {
+      "epoch": 0.029289795918367348,
+      "grad_norm": 3.0066490173339844,
+      "learning_rate": 9.972338340953626e-06,
+      "loss": 0.7616,
+      "step": 897
+    },
+    {
+      "epoch": 0.029322448979591836,
+      "grad_norm": 3.05830717086792,
+      "learning_rate": 9.97230568256042e-06,
+      "loss": 0.7585,
+      "step": 898
+    },
+    {
+      "epoch": 0.029355102040816327,
+      "grad_norm": 3.2171273231506348,
+      "learning_rate": 9.972273024167211e-06,
+      "loss": 0.8138,
+      "step": 899
+    },
+    {
+      "epoch": 0.029387755102040815,
+      "grad_norm": 2.9297590255737305,
+      "learning_rate": 9.972240365774006e-06,
+      "loss": 0.762,
+      "step": 900
+    },
+    {
+      "epoch": 0.029387755102040815,
+      "eval_loss": 0.8132917284965515,
+      "eval_runtime": 76.0316,
+      "eval_samples_per_second": 1.315,
+      "eval_steps_per_second": 1.315,
+      "step": 900
+    },
+    {
+      "epoch": 0.029420408163265306,
+      "grad_norm": 2.8318891525268555,
+      "learning_rate": 9.972207707380797e-06,
+      "loss": 0.7571,
+      "step": 901
+    },
+    {
+      "epoch": 0.029453061224489797,
+      "grad_norm": 3.222458600997925,
+      "learning_rate": 9.97217504898759e-06,
+      "loss": 0.7869,
+      "step": 902
+    },
+    {
+      "epoch": 0.029485714285714285,
+      "grad_norm": 3.2787888050079346,
+      "learning_rate": 9.972142390594384e-06,
+      "loss": 0.7922,
+      "step": 903
+    },
+    {
+      "epoch": 0.029518367346938777,
+      "grad_norm": 2.9158124923706055,
+      "learning_rate": 9.972109732201175e-06,
+      "loss": 0.8008,
+      "step": 904
+    },
+    {
+      "epoch": 0.029551020408163264,
+      "grad_norm": 2.931711435317993,
+      "learning_rate": 9.97207707380797e-06,
+      "loss": 0.7958,
+      "step": 905
+    },
+    {
+      "epoch": 0.029583673469387756,
+      "grad_norm": 3.1747889518737793,
+      "learning_rate": 9.972044415414762e-06,
+      "loss": 0.7953,
+      "step": 906
+    },
+    {
+      "epoch": 0.029616326530612243,
+      "grad_norm": 3.23126482963562,
+      "learning_rate": 9.972011757021555e-06,
+      "loss": 0.8133,
+      "step": 907
+    },
+    {
+      "epoch": 0.029648979591836735,
+      "grad_norm": 2.8572208881378174,
+      "learning_rate": 9.971979098628348e-06,
+      "loss": 0.7832,
+      "step": 908
+    },
+    {
+      "epoch": 0.029681632653061226,
+      "grad_norm": 3.004584312438965,
+      "learning_rate": 9.971946440235142e-06,
+      "loss": 0.8025,
+      "step": 909
+    },
+    {
+      "epoch": 0.029714285714285714,
+      "grad_norm": 2.7122766971588135,
+      "learning_rate": 9.971913781841935e-06,
+      "loss": 0.7819,
+      "step": 910
+    },
+    {
+      "epoch": 0.029746938775510205,
+      "grad_norm": 2.9016056060791016,
+      "learning_rate": 9.971881123448726e-06,
+      "loss": 0.7852,
+      "step": 911
+    },
+    {
+      "epoch": 0.029779591836734693,
+      "grad_norm": 2.985293388366699,
+      "learning_rate": 9.97184846505552e-06,
+      "loss": 0.8294,
+      "step": 912
+    },
+    {
+      "epoch": 0.029812244897959184,
+      "grad_norm": 2.9722225666046143,
+      "learning_rate": 9.971815806662313e-06,
+      "loss": 0.7924,
+      "step": 913
+    },
+    {
+      "epoch": 0.029844897959183672,
+      "grad_norm": 2.8796117305755615,
+      "learning_rate": 9.971783148269106e-06,
+      "loss": 0.7549,
+      "step": 914
+    },
+    {
+      "epoch": 0.029877551020408163,
+      "grad_norm": 3.0465636253356934,
+      "learning_rate": 9.9717504898759e-06,
+      "loss": 0.7875,
+      "step": 915
+    },
+    {
+      "epoch": 0.029910204081632655,
+      "grad_norm": 3.179034948348999,
+      "learning_rate": 9.971717831482692e-06,
+      "loss": 0.774,
+      "step": 916
+    },
+    {
+      "epoch": 0.029942857142857143,
+      "grad_norm": 3.2824554443359375,
+      "learning_rate": 9.971685173089484e-06,
+      "loss": 0.788,
+      "step": 917
+    },
+    {
+      "epoch": 0.029975510204081634,
+      "grad_norm": 3.2578866481781006,
+      "learning_rate": 9.971652514696277e-06,
+      "loss": 0.7672,
+      "step": 918
+    },
+    {
+      "epoch": 0.03000816326530612,
+      "grad_norm": 3.1106038093566895,
+      "learning_rate": 9.97161985630307e-06,
+      "loss": 0.8133,
+      "step": 919
+    },
+    {
+      "epoch": 0.030040816326530613,
+      "grad_norm": 2.948033094406128,
+      "learning_rate": 9.971587197909864e-06,
+      "loss": 0.7819,
+      "step": 920
+    },
+    {
+      "epoch": 0.0300734693877551,
+      "grad_norm": 3.094900131225586,
+      "learning_rate": 9.971554539516657e-06,
+      "loss": 0.7833,
+      "step": 921
+    },
+    {
+      "epoch": 0.030106122448979592,
+      "grad_norm": 2.9197640419006348,
+      "learning_rate": 9.971521881123449e-06,
+      "loss": 0.7804,
+      "step": 922
+    },
+    {
+      "epoch": 0.03013877551020408,
+      "grad_norm": 3.011596441268921,
+      "learning_rate": 9.971489222730243e-06,
+      "loss": 0.7706,
+      "step": 923
+    },
+    {
+      "epoch": 0.03017142857142857,
+      "grad_norm": 2.9059317111968994,
+      "learning_rate": 9.971456564337035e-06,
+      "loss": 0.7594,
+      "step": 924
+    },
+    {
+      "epoch": 0.030204081632653063,
+      "grad_norm": 3.198932409286499,
+      "learning_rate": 9.971423905943828e-06,
+      "loss": 0.7876,
+      "step": 925
+    },
+    {
+      "epoch": 0.03023673469387755,
+      "grad_norm": 3.1033425331115723,
+      "learning_rate": 9.971391247550621e-06,
+      "loss": 0.7628,
+      "step": 926
+    },
+    {
+      "epoch": 0.03026938775510204,
+      "grad_norm": 3.210116147994995,
+      "learning_rate": 9.971358589157413e-06,
+      "loss": 0.748,
+      "step": 927
+    },
+    {
+      "epoch": 0.03030204081632653,
+      "grad_norm": 2.9255874156951904,
+      "learning_rate": 9.971325930764208e-06,
+      "loss": 0.7654,
+      "step": 928
+    },
+    {
+      "epoch": 0.03033469387755102,
+      "grad_norm": 2.949495553970337,
+      "learning_rate": 9.971293272371e-06,
+      "loss": 0.7875,
+      "step": 929
+    },
+    {
+      "epoch": 0.03036734693877551,
+      "grad_norm": 2.8590776920318604,
+      "learning_rate": 9.971260613977793e-06,
+      "loss": 0.7365,
+      "step": 930
+    },
+    {
+      "epoch": 0.0304,
+      "grad_norm": 2.9772932529449463,
+      "learning_rate": 9.971227955584586e-06,
+      "loss": 0.7683,
+      "step": 931
+    },
+    {
+      "epoch": 0.03043265306122449,
+      "grad_norm": 3.1256070137023926,
+      "learning_rate": 9.97119529719138e-06,
+      "loss": 0.8071,
+      "step": 932
+    },
+    {
+      "epoch": 0.03046530612244898,
+      "grad_norm": 3.2088918685913086,
+      "learning_rate": 9.971162638798172e-06,
+      "loss": 0.7676,
+      "step": 933
+    },
+    {
+      "epoch": 0.03049795918367347,
+      "grad_norm": 2.9030396938323975,
+      "learning_rate": 9.971129980404964e-06,
+      "loss": 0.8101,
+      "step": 934
+    },
+    {
+      "epoch": 0.030530612244897958,
+      "grad_norm": 2.968740940093994,
+      "learning_rate": 9.971097322011757e-06,
+      "loss": 0.7716,
+      "step": 935
+    },
+    {
+      "epoch": 0.03056326530612245,
+      "grad_norm": 2.748076915740967,
+      "learning_rate": 9.97106466361855e-06,
+      "loss": 0.7889,
+      "step": 936
+    },
+    {
+      "epoch": 0.030595918367346937,
+      "grad_norm": 2.941471815109253,
+      "learning_rate": 9.971032005225344e-06,
+      "loss": 0.7453,
+      "step": 937
+    },
+    {
+      "epoch": 0.03062857142857143,
+      "grad_norm": 2.9281535148620605,
+      "learning_rate": 9.970999346832137e-06,
+      "loss": 0.7891,
+      "step": 938
+    },
+    {
+      "epoch": 0.03066122448979592,
+      "grad_norm": 2.85978627204895,
+      "learning_rate": 9.97096668843893e-06,
+      "loss": 0.7772,
+      "step": 939
+    },
+    {
+      "epoch": 0.030693877551020408,
+      "grad_norm": 3.195918560028076,
+      "learning_rate": 9.970934030045722e-06,
+      "loss": 0.8068,
+      "step": 940
+    },
+    {
+      "epoch": 0.0307265306122449,
+      "grad_norm": 3.1167869567871094,
+      "learning_rate": 9.970901371652517e-06,
+      "loss": 0.7549,
+      "step": 941
+    },
+    {
+      "epoch": 0.030759183673469387,
+      "grad_norm": 2.9762868881225586,
+      "learning_rate": 9.970868713259308e-06,
+      "loss": 0.7923,
+      "step": 942
+    },
+    {
+      "epoch": 0.030791836734693878,
+      "grad_norm": 3.1077170372009277,
+      "learning_rate": 9.970836054866101e-06,
+      "loss": 0.7636,
+      "step": 943
+    },
+    {
+      "epoch": 0.030824489795918366,
+      "grad_norm": 3.062537431716919,
+      "learning_rate": 9.970803396472895e-06,
+      "loss": 0.7796,
+      "step": 944
+    },
+    {
+      "epoch": 0.030857142857142857,
+      "grad_norm": 3.2072839736938477,
+      "learning_rate": 9.970770738079686e-06,
+      "loss": 0.793,
+      "step": 945
+    },
+    {
+      "epoch": 0.03088979591836735,
+      "grad_norm": 3.26112699508667,
+      "learning_rate": 9.970738079686481e-06,
+      "loss": 0.8107,
+      "step": 946
+    },
+    {
+      "epoch": 0.030922448979591836,
+      "grad_norm": 2.9786770343780518,
+      "learning_rate": 9.970705421293273e-06,
+      "loss": 0.8025,
+      "step": 947
+    },
+    {
+      "epoch": 0.030955102040816328,
+      "grad_norm": 3.0765345096588135,
+      "learning_rate": 9.970672762900066e-06,
+      "loss": 0.8047,
+      "step": 948
+    },
+    {
+      "epoch": 0.030987755102040816,
+      "grad_norm": 2.958984375,
+      "learning_rate": 9.970640104506859e-06,
+      "loss": 0.762,
+      "step": 949
+    },
+    {
+      "epoch": 0.031020408163265307,
+      "grad_norm": 3.08040452003479,
+      "learning_rate": 9.970607446113652e-06,
+      "loss": 0.7928,
+      "step": 950
+    },
+    {
+      "epoch": 0.031020408163265307,
+      "eval_loss": 0.8111075758934021,
+      "eval_runtime": 74.5279,
+      "eval_samples_per_second": 1.342,
+      "eval_steps_per_second": 1.342,
+      "step": 950
+    },
+    {
+      "epoch": 0.031053061224489795,
+      "grad_norm": 2.965144395828247,
+      "learning_rate": 9.970574787720446e-06,
+      "loss": 0.8011,
+      "step": 951
+    },
+    {
+      "epoch": 0.031085714285714286,
+      "grad_norm": 2.9179527759552,
+      "learning_rate": 9.970542129327237e-06,
+      "loss": 0.7703,
+      "step": 952
+    },
+    {
+      "epoch": 0.031118367346938774,
+      "grad_norm": 2.9587607383728027,
+      "learning_rate": 9.97050947093403e-06,
+      "loss": 0.7536,
+      "step": 953
+    },
+    {
+      "epoch": 0.031151020408163265,
+      "grad_norm": 2.9393138885498047,
+      "learning_rate": 9.970476812540824e-06,
+      "loss": 0.7922,
+      "step": 954
+    },
+    {
+      "epoch": 0.031183673469387756,
+      "grad_norm": 3.013162612915039,
+      "learning_rate": 9.970444154147617e-06,
+      "loss": 0.7668,
+      "step": 955
+    },
+    {
+      "epoch": 0.031216326530612244,
+      "grad_norm": 2.8993349075317383,
+      "learning_rate": 9.97041149575441e-06,
+      "loss": 0.7559,
+      "step": 956
+    },
+    {
+      "epoch": 0.031248979591836736,
+      "grad_norm": 3.0861432552337646,
+      "learning_rate": 9.970378837361203e-06,
+      "loss": 0.7632,
+      "step": 957
+    },
+    {
+      "epoch": 0.03128163265306123,
+      "grad_norm": 2.9803247451782227,
+      "learning_rate": 9.970346178967995e-06,
+      "loss": 0.7933,
+      "step": 958
+    },
+    {
+      "epoch": 0.03131428571428571,
+      "grad_norm": 3.0155863761901855,
+      "learning_rate": 9.970313520574788e-06,
+      "loss": 0.8215,
+      "step": 959
+    },
+    {
+      "epoch": 0.0313469387755102,
+      "grad_norm": 3.0787782669067383,
+      "learning_rate": 9.970280862181581e-06,
+      "loss": 0.7936,
+      "step": 960
+    },
+    {
+      "epoch": 0.031379591836734694,
+      "grad_norm": 3.0619027614593506,
+      "learning_rate": 9.970248203788375e-06,
+      "loss": 0.7958,
+      "step": 961
+    },
+    {
+      "epoch": 0.031412244897959185,
+      "grad_norm": 2.7557358741760254,
+      "learning_rate": 9.970215545395168e-06,
+      "loss": 0.761,
+      "step": 962
+    },
+    {
+      "epoch": 0.031444897959183676,
+      "grad_norm": 2.8766896724700928,
+      "learning_rate": 9.97018288700196e-06,
+      "loss": 0.7563,
+      "step": 963
+    },
+    {
+      "epoch": 0.03147755102040816,
+      "grad_norm": 2.976086139678955,
+      "learning_rate": 9.970150228608754e-06,
+      "loss": 0.7839,
+      "step": 964
+    },
+    {
+      "epoch": 0.03151020408163265,
+      "grad_norm": 2.788193941116333,
+      "learning_rate": 9.970117570215546e-06,
+      "loss": 0.789,
+      "step": 965
+    },
+    {
+      "epoch": 0.03154285714285714,
+      "grad_norm": 3.0085878372192383,
+      "learning_rate": 9.970084911822339e-06,
+      "loss": 0.771,
+      "step": 966
+    },
+    {
+      "epoch": 0.031575510204081635,
+      "grad_norm": 3.1194570064544678,
+      "learning_rate": 9.970052253429132e-06,
+      "loss": 0.7889,
+      "step": 967
+    },
+    {
+      "epoch": 0.031608163265306126,
+      "grad_norm": 2.8725526332855225,
+      "learning_rate": 9.970019595035924e-06,
+      "loss": 0.7882,
+      "step": 968
+    },
+    {
+      "epoch": 0.03164081632653061,
+      "grad_norm": 2.982837677001953,
+      "learning_rate": 9.969986936642719e-06,
+      "loss": 0.7674,
+      "step": 969
+    },
+    {
+      "epoch": 0.0316734693877551,
+      "grad_norm": 2.9641151428222656,
+      "learning_rate": 9.96995427824951e-06,
+      "loss": 0.7871,
+      "step": 970
+    },
+    {
+      "epoch": 0.03170612244897959,
+      "grad_norm": 2.973365068435669,
+      "learning_rate": 9.969921619856304e-06,
+      "loss": 0.7939,
+      "step": 971
+    },
+    {
+      "epoch": 0.031738775510204084,
+      "grad_norm": 3.0161428451538086,
+      "learning_rate": 9.969888961463097e-06,
+      "loss": 0.7865,
+      "step": 972
+    },
+    {
+      "epoch": 0.03177142857142857,
+      "grad_norm": 3.369993209838867,
+      "learning_rate": 9.96985630306989e-06,
+      "loss": 0.8231,
+      "step": 973
+    },
+    {
+      "epoch": 0.03180408163265306,
+      "grad_norm": 3.0183799266815186,
+      "learning_rate": 9.969823644676683e-06,
+      "loss": 0.7853,
+      "step": 974
+    },
+    {
+      "epoch": 0.03183673469387755,
+      "grad_norm": 3.244966983795166,
+      "learning_rate": 9.969790986283475e-06,
+      "loss": 0.7607,
+      "step": 975
+    },
+    {
+      "epoch": 0.03186938775510204,
+      "grad_norm": 3.0273585319519043,
+      "learning_rate": 9.969758327890268e-06,
+      "loss": 0.77,
+      "step": 976
+    },
+    {
+      "epoch": 0.031902040816326534,
+      "grad_norm": 3.168429136276245,
+      "learning_rate": 9.969725669497061e-06,
+      "loss": 0.7537,
+      "step": 977
+    },
+    {
+      "epoch": 0.03193469387755102,
+      "grad_norm": 3.006373882293701,
+      "learning_rate": 9.969693011103854e-06,
+      "loss": 0.7927,
+      "step": 978
+    },
+    {
+      "epoch": 0.03196734693877551,
+      "grad_norm": 3.1100571155548096,
+      "learning_rate": 9.969660352710648e-06,
+      "loss": 0.7846,
+      "step": 979
+    },
+    {
+      "epoch": 0.032,
+      "grad_norm": 2.941803216934204,
+      "learning_rate": 9.969627694317441e-06,
+      "loss": 0.7821,
+      "step": 980
+    },
+    {
+      "epoch": 0.03203265306122449,
+      "grad_norm": 3.077153444290161,
+      "learning_rate": 9.969595035924232e-06,
+      "loss": 0.8004,
+      "step": 981
+    },
+    {
+      "epoch": 0.032065306122448976,
+      "grad_norm": 2.9901161193847656,
+      "learning_rate": 9.969562377531026e-06,
+      "loss": 0.7848,
+      "step": 982
+    },
+    {
+      "epoch": 0.03209795918367347,
+      "grad_norm": 3.0473811626434326,
+      "learning_rate": 9.969529719137819e-06,
+      "loss": 0.7823,
+      "step": 983
+    },
+    {
+      "epoch": 0.03213061224489796,
+      "grad_norm": 2.894256353378296,
+      "learning_rate": 9.969497060744612e-06,
+      "loss": 0.7757,
+      "step": 984
+    },
+    {
+      "epoch": 0.03216326530612245,
+      "grad_norm": 3.0119597911834717,
+      "learning_rate": 9.969464402351405e-06,
+      "loss": 0.7799,
+      "step": 985
+    },
+    {
+      "epoch": 0.03219591836734694,
+      "grad_norm": 3.020019292831421,
+      "learning_rate": 9.969431743958197e-06,
+      "loss": 0.7817,
+      "step": 986
+    },
+    {
+      "epoch": 0.032228571428571426,
+      "grad_norm": 3.0470404624938965,
+      "learning_rate": 9.969399085564992e-06,
+      "loss": 0.7873,
+      "step": 987
+    },
+    {
+      "epoch": 0.03226122448979592,
+      "grad_norm": 2.9759464263916016,
+      "learning_rate": 9.969366427171783e-06,
+      "loss": 0.7857,
+      "step": 988
+    },
+    {
+      "epoch": 0.03229387755102041,
+      "grad_norm": 3.1986935138702393,
+      "learning_rate": 9.969333768778577e-06,
+      "loss": 0.786,
+      "step": 989
+    },
+    {
+      "epoch": 0.0323265306122449,
+      "grad_norm": 3.0360336303710938,
+      "learning_rate": 9.96930111038537e-06,
+      "loss": 0.8174,
+      "step": 990
+    },
+    {
+      "epoch": 0.03235918367346939,
+      "grad_norm": 3.093979597091675,
+      "learning_rate": 9.969268451992163e-06,
+      "loss": 0.7619,
+      "step": 991
+    },
+    {
+      "epoch": 0.032391836734693875,
+      "grad_norm": 2.935920238494873,
+      "learning_rate": 9.969235793598956e-06,
+      "loss": 0.7661,
+      "step": 992
+    },
+    {
+      "epoch": 0.03242448979591837,
+      "grad_norm": 3.1429708003997803,
+      "learning_rate": 9.969203135205748e-06,
+      "loss": 0.7981,
+      "step": 993
+    },
+    {
+      "epoch": 0.03245714285714286,
+      "grad_norm": 2.8831772804260254,
+      "learning_rate": 9.969170476812541e-06,
+      "loss": 0.7969,
+      "step": 994
+    },
+    {
+      "epoch": 0.03248979591836735,
+      "grad_norm": 2.8683125972747803,
+      "learning_rate": 9.969137818419334e-06,
+      "loss": 0.7688,
+      "step": 995
+    },
+    {
+      "epoch": 0.032522448979591834,
+      "grad_norm": 3.026094436645508,
+      "learning_rate": 9.969105160026128e-06,
+      "loss": 0.782,
+      "step": 996
+    },
+    {
+      "epoch": 0.032555102040816325,
+      "grad_norm": 3.065110445022583,
+      "learning_rate": 9.969072501632921e-06,
+      "loss": 0.782,
+      "step": 997
+    },
+    {
+      "epoch": 0.032587755102040816,
+      "grad_norm": 2.9125099182128906,
+      "learning_rate": 9.969039843239714e-06,
+      "loss": 0.8056,
+      "step": 998
+    },
+    {
+      "epoch": 0.03262040816326531,
+      "grad_norm": 2.978609800338745,
+      "learning_rate": 9.969007184846506e-06,
+      "loss": 0.7839,
+      "step": 999
+    },
+    {
+      "epoch": 0.0326530612244898,
+      "grad_norm": 3.037384510040283,
+      "learning_rate": 9.968974526453299e-06,
+      "loss": 0.7877,
+      "step": 1000
+    },
+    {
+      "epoch": 0.0326530612244898,
+      "eval_loss": 0.8111925721168518,
+      "eval_runtime": 72.9629,
+      "eval_samples_per_second": 1.371,
+      "eval_steps_per_second": 1.371,
+      "step": 1000
+    },
+    {
+      "epoch": 0.03268571428571428,
+      "grad_norm": 2.9627959728240967,
+      "learning_rate": 9.968941868060092e-06,
+      "loss": 0.7855,
+      "step": 1001
+    },
+    {
+      "epoch": 0.032718367346938775,
+      "grad_norm": 3.2459042072296143,
+      "learning_rate": 9.968909209666885e-06,
+      "loss": 0.7897,
+      "step": 1002
+    },
+    {
+      "epoch": 0.032751020408163266,
+      "grad_norm": 3.0746281147003174,
+      "learning_rate": 9.968876551273679e-06,
+      "loss": 0.7415,
+      "step": 1003
+    },
+    {
+      "epoch": 0.03278367346938776,
+      "grad_norm": 3.005556583404541,
+      "learning_rate": 9.96884389288047e-06,
+      "loss": 0.8075,
+      "step": 1004
+    },
+    {
+      "epoch": 0.03281632653061224,
+      "grad_norm": 2.9875948429107666,
+      "learning_rate": 9.968811234487265e-06,
+      "loss": 0.7982,
+      "step": 1005
+    },
+    {
+      "epoch": 0.03284897959183673,
+      "grad_norm": 2.8231279850006104,
+      "learning_rate": 9.968778576094057e-06,
+      "loss": 0.758,
+      "step": 1006
+    },
+    {
+      "epoch": 0.032881632653061224,
+      "grad_norm": 2.8995821475982666,
+      "learning_rate": 9.96874591770085e-06,
+      "loss": 0.7653,
+      "step": 1007
+    },
+    {
+      "epoch": 0.032914285714285715,
+      "grad_norm": 3.0024149417877197,
+      "learning_rate": 9.968713259307643e-06,
+      "loss": 0.7777,
+      "step": 1008
+    },
+    {
+      "epoch": 0.03294693877551021,
+      "grad_norm": 3.019351005554199,
+      "learning_rate": 9.968680600914435e-06,
+      "loss": 0.7921,
+      "step": 1009
+    },
+    {
+      "epoch": 0.03297959183673469,
+      "grad_norm": 2.916240930557251,
+      "learning_rate": 9.96864794252123e-06,
+      "loss": 0.75,
+      "step": 1010
+    },
+    {
+      "epoch": 0.03301224489795918,
+      "grad_norm": 3.005613088607788,
+      "learning_rate": 9.968615284128021e-06,
+      "loss": 0.7943,
+      "step": 1011
+    },
+    {
+      "epoch": 0.033044897959183674,
+      "grad_norm": 2.9690959453582764,
+      "learning_rate": 9.968582625734814e-06,
+      "loss": 0.7855,
+      "step": 1012
+    },
+    {
+      "epoch": 0.033077551020408165,
+      "grad_norm": 3.0370657444000244,
+      "learning_rate": 9.968549967341608e-06,
+      "loss": 0.7567,
+      "step": 1013
+    },
+    {
+      "epoch": 0.033110204081632656,
+      "grad_norm": 3.010715961456299,
+      "learning_rate": 9.9685173089484e-06,
+      "loss": 0.7545,
+      "step": 1014
+    },
+    {
+      "epoch": 0.03314285714285714,
+      "grad_norm": 3.1019158363342285,
+      "learning_rate": 9.968484650555194e-06,
+      "loss": 0.7881,
+      "step": 1015
+    },
+    {
+      "epoch": 0.03317551020408163,
+      "grad_norm": 2.9990556240081787,
+      "learning_rate": 9.968451992161986e-06,
+      "loss": 0.7892,
+      "step": 1016
+    },
+    {
+      "epoch": 0.03320816326530612,
+      "grad_norm": 2.9631762504577637,
+      "learning_rate": 9.968419333768779e-06,
+      "loss": 0.7459,
+      "step": 1017
+    },
+    {
+      "epoch": 0.033240816326530614,
+      "grad_norm": 3.1492321491241455,
+      "learning_rate": 9.968386675375572e-06,
+      "loss": 0.796,
+      "step": 1018
+    },
+    {
+      "epoch": 0.0332734693877551,
+      "grad_norm": 3.0348122119903564,
+      "learning_rate": 9.968354016982365e-06,
+      "loss": 0.795,
+      "step": 1019
+    },
+    {
+      "epoch": 0.03330612244897959,
+      "grad_norm": 3.2612578868865967,
+      "learning_rate": 9.968321358589159e-06,
+      "loss": 0.7781,
+      "step": 1020
+    },
+    {
+      "epoch": 0.03333877551020408,
+      "grad_norm": 3.0605742931365967,
+      "learning_rate": 9.968288700195952e-06,
+      "loss": 0.7329,
+      "step": 1021
+    },
+    {
+      "epoch": 0.03337142857142857,
+      "grad_norm": 3.0548489093780518,
+      "learning_rate": 9.968256041802743e-06,
+      "loss": 0.7601,
+      "step": 1022
+    },
+    {
+      "epoch": 0.033404081632653064,
+      "grad_norm": 2.946451187133789,
+      "learning_rate": 9.968223383409537e-06,
+      "loss": 0.7756,
+      "step": 1023
+    },
+    {
+      "epoch": 0.03343673469387755,
+      "grad_norm": 3.0444273948669434,
+      "learning_rate": 9.96819072501633e-06,
+      "loss": 0.762,
+      "step": 1024
+    },
+    {
+      "epoch": 0.03346938775510204,
+      "grad_norm": 3.207759380340576,
+      "learning_rate": 9.968158066623123e-06,
+      "loss": 0.7755,
+      "step": 1025
+    },
+    {
+      "epoch": 0.03350204081632653,
+      "grad_norm": 3.0307605266571045,
+      "learning_rate": 9.968125408229916e-06,
+      "loss": 0.7604,
+      "step": 1026
+    },
+    {
+      "epoch": 0.03353469387755102,
+      "grad_norm": 2.975473165512085,
+      "learning_rate": 9.968092749836708e-06,
+      "loss": 0.7929,
+      "step": 1027
+    },
+    {
+      "epoch": 0.033567346938775514,
+      "grad_norm": 3.00290846824646,
+      "learning_rate": 9.968060091443503e-06,
+      "loss": 0.8294,
+      "step": 1028
+    },
+    {
+      "epoch": 0.0336,
+      "grad_norm": 3.235747814178467,
+      "learning_rate": 9.968027433050294e-06,
+      "loss": 0.7807,
+      "step": 1029
+    },
+    {
+      "epoch": 0.03363265306122449,
+      "grad_norm": 3.031163215637207,
+      "learning_rate": 9.967994774657088e-06,
+      "loss": 0.7549,
+      "step": 1030
+    },
+    {
+      "epoch": 0.03366530612244898,
+      "grad_norm": 2.9528584480285645,
+      "learning_rate": 9.96796211626388e-06,
+      "loss": 0.7542,
+      "step": 1031
+    },
+    {
+      "epoch": 0.03369795918367347,
+      "grad_norm": 3.0693178176879883,
+      "learning_rate": 9.967929457870672e-06,
+      "loss": 0.7773,
+      "step": 1032
+    },
+    {
+      "epoch": 0.033730612244897956,
+      "grad_norm": 3.159449338912964,
+      "learning_rate": 9.967896799477467e-06,
+      "loss": 0.7998,
+      "step": 1033
+    },
+    {
+      "epoch": 0.03376326530612245,
+      "grad_norm": 3.0627989768981934,
+      "learning_rate": 9.967864141084259e-06,
+      "loss": 0.7624,
+      "step": 1034
+    },
+    {
+      "epoch": 0.03379591836734694,
+      "grad_norm": 3.088207721710205,
+      "learning_rate": 9.967831482691052e-06,
+      "loss": 0.7757,
+      "step": 1035
+    },
+    {
+      "epoch": 0.03382857142857143,
+      "grad_norm": 3.1921637058258057,
+      "learning_rate": 9.967798824297845e-06,
+      "loss": 0.7923,
+      "step": 1036
+    },
+    {
+      "epoch": 0.03386122448979592,
+      "grad_norm": 3.039994955062866,
+      "learning_rate": 9.967766165904638e-06,
+      "loss": 0.7628,
+      "step": 1037
+    },
+    {
+      "epoch": 0.033893877551020406,
+      "grad_norm": 3.221714973449707,
+      "learning_rate": 9.967733507511432e-06,
+      "loss": 0.7684,
+      "step": 1038
+    },
+    {
+      "epoch": 0.0339265306122449,
+      "grad_norm": 2.9829065799713135,
+      "learning_rate": 9.967700849118223e-06,
+      "loss": 0.7956,
+      "step": 1039
+    },
+    {
+      "epoch": 0.03395918367346939,
+      "grad_norm": 2.9304864406585693,
+      "learning_rate": 9.967668190725016e-06,
+      "loss": 0.7865,
+      "step": 1040
+    },
+    {
+      "epoch": 0.03399183673469388,
+      "grad_norm": 3.0041542053222656,
+      "learning_rate": 9.96763553233181e-06,
+      "loss": 0.8113,
+      "step": 1041
+    },
+    {
+      "epoch": 0.034024489795918364,
+      "grad_norm": 3.1308107376098633,
+      "learning_rate": 9.967602873938603e-06,
+      "loss": 0.7622,
+      "step": 1042
+    },
+    {
+      "epoch": 0.034057142857142855,
+      "grad_norm": 3.0761616230010986,
+      "learning_rate": 9.967570215545396e-06,
+      "loss": 0.7822,
+      "step": 1043
+    },
+    {
+      "epoch": 0.03408979591836735,
+      "grad_norm": 3.073542594909668,
+      "learning_rate": 9.96753755715219e-06,
+      "loss": 0.782,
+      "step": 1044
+    },
+    {
+      "epoch": 0.03412244897959184,
+      "grad_norm": 2.954620361328125,
+      "learning_rate": 9.967504898758981e-06,
+      "loss": 0.7686,
+      "step": 1045
+    },
+    {
+      "epoch": 0.03415510204081633,
+      "grad_norm": 3.1652603149414062,
+      "learning_rate": 9.967472240365776e-06,
+      "loss": 0.7743,
+      "step": 1046
+    },
+    {
+      "epoch": 0.034187755102040814,
+      "grad_norm": 2.8833162784576416,
+      "learning_rate": 9.967439581972567e-06,
+      "loss": 0.7741,
+      "step": 1047
+    },
+    {
+      "epoch": 0.034220408163265305,
+      "grad_norm": 2.910767078399658,
+      "learning_rate": 9.96740692357936e-06,
+      "loss": 0.7768,
+      "step": 1048
+    },
+    {
+      "epoch": 0.034253061224489796,
+      "grad_norm": 3.0124878883361816,
+      "learning_rate": 9.967374265186154e-06,
+      "loss": 0.7656,
+      "step": 1049
+    },
+    {
+      "epoch": 0.03428571428571429,
+      "grad_norm": 3.200578451156616,
+      "learning_rate": 9.967341606792945e-06,
+      "loss": 0.7979,
+      "step": 1050
+    },
+    {
+      "epoch": 0.03428571428571429,
+      "eval_loss": 0.809950053691864,
+      "eval_runtime": 78.3268,
+      "eval_samples_per_second": 1.277,
+      "eval_steps_per_second": 1.277,
+      "step": 1050
+    },
+    {
+      "epoch": 0.03431836734693878,
+      "grad_norm": 3.1058106422424316,
+      "learning_rate": 9.96730894839974e-06,
+      "loss": 0.7929,
+      "step": 1051
+    },
+    {
+      "epoch": 0.03435102040816326,
+      "grad_norm": 2.981287956237793,
+      "learning_rate": 9.967276290006532e-06,
+      "loss": 0.7922,
+      "step": 1052
+    },
+    {
+      "epoch": 0.034383673469387754,
+      "grad_norm": 3.0732994079589844,
+      "learning_rate": 9.967243631613325e-06,
+      "loss": 0.7598,
+      "step": 1053
+    },
+    {
+      "epoch": 0.034416326530612246,
+      "grad_norm": 2.965428352355957,
+      "learning_rate": 9.967210973220118e-06,
+      "loss": 0.7796,
+      "step": 1054
+    },
+    {
+      "epoch": 0.03444897959183674,
+      "grad_norm": 3.047433853149414,
+      "learning_rate": 9.967178314826912e-06,
+      "loss": 0.7806,
+      "step": 1055
+    },
+    {
+      "epoch": 0.03448163265306122,
+      "grad_norm": 2.9787747859954834,
+      "learning_rate": 9.967145656433705e-06,
+      "loss": 0.7548,
+      "step": 1056
+    },
+    {
+      "epoch": 0.03451428571428571,
+      "grad_norm": 3.3675661087036133,
+      "learning_rate": 9.967112998040496e-06,
+      "loss": 0.7764,
+      "step": 1057
+    },
+    {
+      "epoch": 0.034546938775510204,
+      "grad_norm": 2.90116024017334,
+      "learning_rate": 9.96708033964729e-06,
+      "loss": 0.7819,
+      "step": 1058
+    },
+    {
+      "epoch": 0.034579591836734695,
+      "grad_norm": 2.965744972229004,
+      "learning_rate": 9.967047681254083e-06,
+      "loss": 0.7603,
+      "step": 1059
+    },
+    {
+      "epoch": 0.034612244897959187,
+      "grad_norm": 3.116170644760132,
+      "learning_rate": 9.967015022860876e-06,
+      "loss": 0.7885,
+      "step": 1060
+    },
+    {
+      "epoch": 0.03464489795918367,
+      "grad_norm": 3.1526293754577637,
+      "learning_rate": 9.96698236446767e-06,
+      "loss": 0.7752,
+      "step": 1061
+    },
+    {
+      "epoch": 0.03467755102040816,
+      "grad_norm": 3.0937249660491943,
+      "learning_rate": 9.966949706074463e-06,
+      "loss": 0.7845,
+      "step": 1062
+    },
+    {
+      "epoch": 0.034710204081632653,
+      "grad_norm": 3.069850444793701,
+      "learning_rate": 9.966917047681254e-06,
+      "loss": 0.7645,
+      "step": 1063
+    },
+    {
+      "epoch": 0.034742857142857145,
+      "grad_norm": 2.979881763458252,
+      "learning_rate": 9.966884389288047e-06,
+      "loss": 0.7687,
+      "step": 1064
+    },
+    {
+      "epoch": 0.034775510204081636,
+      "grad_norm": 3.1711268424987793,
+      "learning_rate": 9.96685173089484e-06,
+      "loss": 0.8136,
+      "step": 1065
+    },
+    {
+      "epoch": 0.03480816326530612,
+      "grad_norm": 3.197852373123169,
+      "learning_rate": 9.966819072501634e-06,
+      "loss": 0.8152,
+      "step": 1066
+    },
+    {
+      "epoch": 0.03484081632653061,
+      "grad_norm": 3.098540782928467,
+      "learning_rate": 9.966786414108427e-06,
+      "loss": 0.7645,
+      "step": 1067
+    },
+    {
+      "epoch": 0.0348734693877551,
+      "grad_norm": 3.1754651069641113,
+      "learning_rate": 9.966753755715219e-06,
+      "loss": 0.7663,
+      "step": 1068
+    },
+    {
+      "epoch": 0.034906122448979594,
+      "grad_norm": 3.0712950229644775,
+      "learning_rate": 9.966721097322014e-06,
+      "loss": 0.7851,
+      "step": 1069
+    },
+    {
+      "epoch": 0.03493877551020408,
+      "grad_norm": 3.285234212875366,
+      "learning_rate": 9.966688438928805e-06,
+      "loss": 0.774,
+      "step": 1070
+    },
+    {
+      "epoch": 0.03497142857142857,
+      "grad_norm": 3.2550666332244873,
+      "learning_rate": 9.966655780535598e-06,
+      "loss": 0.751,
+      "step": 1071
+    },
+    {
+      "epoch": 0.03500408163265306,
+      "grad_norm": 3.238384246826172,
+      "learning_rate": 9.966623122142392e-06,
+      "loss": 0.7666,
+      "step": 1072
+    },
+    {
+      "epoch": 0.03503673469387755,
+      "grad_norm": 3.0862374305725098,
+      "learning_rate": 9.966590463749183e-06,
+      "loss": 0.7981,
+      "step": 1073
+    },
+    {
+      "epoch": 0.035069387755102044,
+      "grad_norm": 3.172941207885742,
+      "learning_rate": 9.966557805355978e-06,
+      "loss": 0.7549,
+      "step": 1074
+    },
+    {
+      "epoch": 0.03510204081632653,
+      "grad_norm": 3.060302495956421,
+      "learning_rate": 9.96652514696277e-06,
+      "loss": 0.7835,
+      "step": 1075
+    },
+    {
+      "epoch": 0.03513469387755102,
+      "grad_norm": 3.122617244720459,
+      "learning_rate": 9.966492488569563e-06,
+      "loss": 0.7952,
+      "step": 1076
+    },
+    {
+      "epoch": 0.03516734693877551,
+      "grad_norm": 2.904637336730957,
+      "learning_rate": 9.966459830176356e-06,
+      "loss": 0.7595,
+      "step": 1077
+    },
+    {
+      "epoch": 0.0352,
+      "grad_norm": 3.1672565937042236,
+      "learning_rate": 9.96642717178315e-06,
+      "loss": 0.7926,
+      "step": 1078
+    },
+    {
+      "epoch": 0.035232653061224486,
+      "grad_norm": 3.0276424884796143,
+      "learning_rate": 9.966394513389943e-06,
+      "loss": 0.7801,
+      "step": 1079
+    },
+    {
+      "epoch": 0.03526530612244898,
+      "grad_norm": 3.0199339389801025,
+      "learning_rate": 9.966361854996734e-06,
+      "loss": 0.7977,
+      "step": 1080
+    },
+    {
+      "epoch": 0.03529795918367347,
+      "grad_norm": 3.1580443382263184,
+      "learning_rate": 9.966329196603527e-06,
+      "loss": 0.7638,
+      "step": 1081
+    },
+    {
+      "epoch": 0.03533061224489796,
+      "grad_norm": 2.9399259090423584,
+      "learning_rate": 9.96629653821032e-06,
+      "loss": 0.7594,
+      "step": 1082
+    },
+    {
+      "epoch": 0.03536326530612245,
+      "grad_norm": 3.1555962562561035,
+      "learning_rate": 9.966263879817114e-06,
+      "loss": 0.7688,
+      "step": 1083
+    },
+    {
+      "epoch": 0.035395918367346936,
+      "grad_norm": 3.03648042678833,
+      "learning_rate": 9.966231221423907e-06,
+      "loss": 0.8,
+      "step": 1084
+    },
+    {
+      "epoch": 0.03542857142857143,
+      "grad_norm": 3.142136573791504,
+      "learning_rate": 9.9661985630307e-06,
+      "loss": 0.7502,
+      "step": 1085
+    },
+    {
+      "epoch": 0.03546122448979592,
+      "grad_norm": 3.103031635284424,
+      "learning_rate": 9.966165904637492e-06,
+      "loss": 0.7979,
+      "step": 1086
+    },
+    {
+      "epoch": 0.03549387755102041,
+      "grad_norm": 3.0145888328552246,
+      "learning_rate": 9.966133246244287e-06,
+      "loss": 0.7669,
+      "step": 1087
+    },
+    {
+      "epoch": 0.0355265306122449,
+      "grad_norm": 3.1782283782958984,
+      "learning_rate": 9.966100587851078e-06,
+      "loss": 0.7677,
+      "step": 1088
+    },
+    {
+      "epoch": 0.035559183673469386,
+      "grad_norm": 3.102893829345703,
+      "learning_rate": 9.966067929457871e-06,
+      "loss": 0.7651,
+      "step": 1089
+    },
+    {
+      "epoch": 0.03559183673469388,
+      "grad_norm": 3.0069074630737305,
+      "learning_rate": 9.966035271064665e-06,
+      "loss": 0.7729,
+      "step": 1090
+    },
+    {
+      "epoch": 0.03562448979591837,
+      "grad_norm": 3.022388458251953,
+      "learning_rate": 9.966002612671456e-06,
+      "loss": 0.7537,
+      "step": 1091
+    },
+    {
+      "epoch": 0.03565714285714286,
+      "grad_norm": 3.0949289798736572,
+      "learning_rate": 9.965969954278251e-06,
+      "loss": 0.77,
+      "step": 1092
+    },
+    {
+      "epoch": 0.035689795918367344,
+      "grad_norm": 3.48496413230896,
+      "learning_rate": 9.965937295885043e-06,
+      "loss": 0.7671,
+      "step": 1093
+    },
+    {
+      "epoch": 0.035722448979591835,
+      "grad_norm": 3.1143581867218018,
+      "learning_rate": 9.965904637491836e-06,
+      "loss": 0.7451,
+      "step": 1094
+    },
+    {
+      "epoch": 0.035755102040816326,
+      "grad_norm": 2.93046498298645,
+      "learning_rate": 9.96587197909863e-06,
+      "loss": 0.7878,
+      "step": 1095
+    },
+    {
+      "epoch": 0.03578775510204082,
+      "grad_norm": 3.0415232181549072,
+      "learning_rate": 9.96583932070542e-06,
+      "loss": 0.7465,
+      "step": 1096
+    },
+    {
+      "epoch": 0.03582040816326531,
+      "grad_norm": 3.1707770824432373,
+      "learning_rate": 9.965806662312216e-06,
+      "loss": 0.7548,
+      "step": 1097
+    },
+    {
+      "epoch": 0.03585306122448979,
+      "grad_norm": 3.0290682315826416,
+      "learning_rate": 9.965774003919007e-06,
+      "loss": 0.7703,
+      "step": 1098
+    },
+    {
+      "epoch": 0.035885714285714285,
+      "grad_norm": 2.925924777984619,
+      "learning_rate": 9.9657413455258e-06,
+      "loss": 0.7653,
+      "step": 1099
+    },
+    {
+      "epoch": 0.035918367346938776,
+      "grad_norm": 3.2360856533050537,
+      "learning_rate": 9.965708687132594e-06,
+      "loss": 0.7705,
+      "step": 1100
+    },
+    {
+      "epoch": 0.035918367346938776,
+      "eval_loss": 0.8095739483833313,
+      "eval_runtime": 95.0564,
+      "eval_samples_per_second": 1.052,
+      "eval_steps_per_second": 1.052,
+      "step": 1100
+    },
+    {
+      "epoch": 0.03595102040816327,
+      "grad_norm": 3.276266574859619,
+      "learning_rate": 9.965676028739387e-06,
+      "loss": 0.7926,
+      "step": 1101
+    },
+    {
+      "epoch": 0.03598367346938776,
+      "grad_norm": 3.069849967956543,
+      "learning_rate": 9.96564337034618e-06,
+      "loss": 0.7966,
+      "step": 1102
+    },
+    {
+      "epoch": 0.03601632653061224,
+      "grad_norm": 3.484065532684326,
+      "learning_rate": 9.965610711952973e-06,
+      "loss": 0.7744,
+      "step": 1103
+    },
+    {
+      "epoch": 0.036048979591836734,
+      "grad_norm": 3.468071699142456,
+      "learning_rate": 9.965578053559765e-06,
+      "loss": 0.7575,
+      "step": 1104
+    },
+    {
+      "epoch": 0.036081632653061226,
+      "grad_norm": 3.8356101512908936,
+      "learning_rate": 9.965545395166558e-06,
+      "loss": 0.7817,
+      "step": 1105
+    },
+    {
+      "epoch": 0.03611428571428572,
+      "grad_norm": 3.2512879371643066,
+      "learning_rate": 9.965512736773351e-06,
+      "loss": 0.7356,
+      "step": 1106
+    },
+    {
+      "epoch": 0.0361469387755102,
+      "grad_norm": 3.004352569580078,
+      "learning_rate": 9.965480078380145e-06,
+      "loss": 0.7286,
+      "step": 1107
+    },
+    {
+      "epoch": 0.03617959183673469,
+      "grad_norm": 3.194286823272705,
+      "learning_rate": 9.965447419986938e-06,
+      "loss": 0.7474,
+      "step": 1108
+    },
+    {
+      "epoch": 0.036212244897959184,
+      "grad_norm": 3.509319543838501,
+      "learning_rate": 9.96541476159373e-06,
+      "loss": 0.7888,
+      "step": 1109
+    },
+    {
+      "epoch": 0.036244897959183675,
+      "grad_norm": 3.528698205947876,
+      "learning_rate": 9.965382103200524e-06,
+      "loss": 0.794,
+      "step": 1110
+    },
+    {
+      "epoch": 0.036277551020408166,
+      "grad_norm": 3.2880067825317383,
+      "learning_rate": 9.965349444807316e-06,
+      "loss": 0.7408,
+      "step": 1111
+    },
+    {
+      "epoch": 0.03631020408163265,
+      "grad_norm": 3.031257152557373,
+      "learning_rate": 9.965316786414109e-06,
+      "loss": 0.7847,
+      "step": 1112
+    },
+    {
+      "epoch": 0.03634285714285714,
+      "grad_norm": 3.2594382762908936,
+      "learning_rate": 9.965284128020902e-06,
+      "loss": 0.7911,
+      "step": 1113
+    },
+    {
+      "epoch": 0.03637551020408163,
+      "grad_norm": 3.2341842651367188,
+      "learning_rate": 9.965251469627694e-06,
+      "loss": 0.7412,
+      "step": 1114
+    },
+    {
+      "epoch": 0.036408163265306125,
+      "grad_norm": 3.251246213912964,
+      "learning_rate": 9.965218811234489e-06,
+      "loss": 0.7549,
+      "step": 1115
+    },
+    {
+      "epoch": 0.03644081632653061,
+      "grad_norm": 3.0365469455718994,
+      "learning_rate": 9.96518615284128e-06,
+      "loss": 0.78,
+      "step": 1116
+    },
+    {
+      "epoch": 0.0364734693877551,
+      "grad_norm": 3.059936285018921,
+      "learning_rate": 9.965153494448074e-06,
+      "loss": 0.805,
+      "step": 1117
+    },
+    {
+      "epoch": 0.03650612244897959,
+      "grad_norm": 3.0952272415161133,
+      "learning_rate": 9.965120836054867e-06,
+      "loss": 0.7563,
+      "step": 1118
+    },
+    {
+      "epoch": 0.03653877551020408,
+      "grad_norm": 3.136528491973877,
+      "learning_rate": 9.96508817766166e-06,
+      "loss": 0.7767,
+      "step": 1119
+    },
+    {
+      "epoch": 0.036571428571428574,
+      "grad_norm": 2.979304313659668,
+      "learning_rate": 9.965055519268453e-06,
+      "loss": 0.8002,
+      "step": 1120
+    },
+    {
+      "epoch": 0.03660408163265306,
+      "grad_norm": 2.8801610469818115,
+      "learning_rate": 9.965022860875245e-06,
+      "loss": 0.7793,
+      "step": 1121
+    },
+    {
+      "epoch": 0.03663673469387755,
+      "grad_norm": 2.9416167736053467,
+      "learning_rate": 9.964990202482038e-06,
+      "loss": 0.7703,
+      "step": 1122
+    },
+    {
+      "epoch": 0.03666938775510204,
+      "grad_norm": 3.1201093196868896,
+      "learning_rate": 9.964957544088831e-06,
+      "loss": 0.7699,
+      "step": 1123
+    },
+    {
+      "epoch": 0.03670204081632653,
+      "grad_norm": 3.1061949729919434,
+      "learning_rate": 9.964924885695625e-06,
+      "loss": 0.7974,
+      "step": 1124
+    },
+    {
+      "epoch": 0.036734693877551024,
+      "grad_norm": 3.0530447959899902,
+      "learning_rate": 9.964892227302418e-06,
+      "loss": 0.7576,
+      "step": 1125
+    },
+    {
+      "epoch": 0.03676734693877551,
+      "grad_norm": 3.002209186553955,
+      "learning_rate": 9.964859568909211e-06,
+      "loss": 0.7699,
+      "step": 1126
+    },
+    {
+      "epoch": 0.0368,
+      "grad_norm": 3.1953022480010986,
+      "learning_rate": 9.964826910516003e-06,
+      "loss": 0.7153,
+      "step": 1127
+    },
+    {
+      "epoch": 0.03683265306122449,
+      "grad_norm": 3.091996192932129,
+      "learning_rate": 9.964794252122796e-06,
+      "loss": 0.7368,
+      "step": 1128
+    },
+    {
+      "epoch": 0.03686530612244898,
+      "grad_norm": 3.173081398010254,
+      "learning_rate": 9.964761593729589e-06,
+      "loss": 0.7839,
+      "step": 1129
+    },
+    {
+      "epoch": 0.036897959183673466,
+      "grad_norm": 3.0133304595947266,
+      "learning_rate": 9.964728935336382e-06,
+      "loss": 0.7827,
+      "step": 1130
+    },
+    {
+      "epoch": 0.03693061224489796,
+      "grad_norm": 3.1596643924713135,
+      "learning_rate": 9.964696276943176e-06,
+      "loss": 0.7649,
+      "step": 1131
+    },
+    {
+      "epoch": 0.03696326530612245,
+      "grad_norm": 3.2854256629943848,
+      "learning_rate": 9.964663618549967e-06,
+      "loss": 0.751,
+      "step": 1132
+    },
+    {
+      "epoch": 0.03699591836734694,
+      "grad_norm": 2.8992016315460205,
+      "learning_rate": 9.964630960156762e-06,
+      "loss": 0.7535,
+      "step": 1133
+    },
+    {
+      "epoch": 0.03702857142857143,
+      "grad_norm": 2.9541709423065186,
+      "learning_rate": 9.964598301763554e-06,
+      "loss": 0.7433,
+      "step": 1134
+    },
+    {
+      "epoch": 0.037061224489795916,
+      "grad_norm": 2.9385082721710205,
+      "learning_rate": 9.964565643370347e-06,
+      "loss": 0.8005,
+      "step": 1135
+    },
+    {
+      "epoch": 0.03709387755102041,
+      "grad_norm": 3.042072057723999,
+      "learning_rate": 9.96453298497714e-06,
+      "loss": 0.7819,
+      "step": 1136
+    },
+    {
+      "epoch": 0.0371265306122449,
+      "grad_norm": 2.9819841384887695,
+      "learning_rate": 9.964500326583932e-06,
+      "loss": 0.7911,
+      "step": 1137
+    },
+    {
+      "epoch": 0.03715918367346939,
+      "grad_norm": 3.242607831954956,
+      "learning_rate": 9.964467668190726e-06,
+      "loss": 0.7926,
+      "step": 1138
+    },
+    {
+      "epoch": 0.037191836734693874,
+      "grad_norm": 2.9996449947357178,
+      "learning_rate": 9.964435009797518e-06,
+      "loss": 0.7904,
+      "step": 1139
+    },
+    {
+      "epoch": 0.037224489795918365,
+      "grad_norm": 3.1378376483917236,
+      "learning_rate": 9.964402351404311e-06,
+      "loss": 0.8,
+      "step": 1140
+    },
+    {
+      "epoch": 0.03725714285714286,
+      "grad_norm": 2.9573757648468018,
+      "learning_rate": 9.964369693011105e-06,
+      "loss": 0.7589,
+      "step": 1141
+    },
+    {
+      "epoch": 0.03728979591836735,
+      "grad_norm": 3.122229814529419,
+      "learning_rate": 9.964337034617898e-06,
+      "loss": 0.7852,
+      "step": 1142
+    },
+    {
+      "epoch": 0.03732244897959184,
+      "grad_norm": 3.3379294872283936,
+      "learning_rate": 9.964304376224691e-06,
+      "loss": 0.7868,
+      "step": 1143
+    },
+    {
+      "epoch": 0.037355102040816324,
+      "grad_norm": 3.2145421504974365,
+      "learning_rate": 9.964271717831484e-06,
+      "loss": 0.7516,
+      "step": 1144
+    },
+    {
+      "epoch": 0.037387755102040815,
+      "grad_norm": 3.068854808807373,
+      "learning_rate": 9.964239059438276e-06,
+      "loss": 0.7539,
+      "step": 1145
+    },
+    {
+      "epoch": 0.037420408163265306,
+      "grad_norm": 3.1995043754577637,
+      "learning_rate": 9.964206401045069e-06,
+      "loss": 0.7927,
+      "step": 1146
+    },
+    {
+      "epoch": 0.0374530612244898,
+      "grad_norm": 3.0854032039642334,
+      "learning_rate": 9.964173742651862e-06,
+      "loss": 0.7608,
+      "step": 1147
+    },
+    {
+      "epoch": 0.03748571428571429,
+      "grad_norm": 3.3977510929107666,
+      "learning_rate": 9.964141084258655e-06,
+      "loss": 0.779,
+      "step": 1148
+    },
+    {
+      "epoch": 0.03751836734693877,
+      "grad_norm": 3.0493321418762207,
+      "learning_rate": 9.964108425865449e-06,
+      "loss": 0.7329,
+      "step": 1149
+    },
+    {
+      "epoch": 0.037551020408163265,
+      "grad_norm": 2.8519468307495117,
+      "learning_rate": 9.96407576747224e-06,
+      "loss": 0.7954,
+      "step": 1150
+    },
+    {
+      "epoch": 0.037551020408163265,
+      "eval_loss": 0.8091681003570557,
+      "eval_runtime": 85.0018,
+      "eval_samples_per_second": 1.176,
+      "eval_steps_per_second": 1.176,
+      "step": 1150
+    },
+    {
+      "epoch": 0.037583673469387756,
+      "grad_norm": 2.9106438159942627,
+      "learning_rate": 9.964043109079035e-06,
+      "loss": 0.7796,
+      "step": 1151
+    },
+    {
+      "epoch": 0.03761632653061225,
+      "grad_norm": 3.1851396560668945,
+      "learning_rate": 9.964010450685827e-06,
+      "loss": 0.7718,
+      "step": 1152
+    },
+    {
+      "epoch": 0.03764897959183673,
+      "grad_norm": 3.101126194000244,
+      "learning_rate": 9.96397779229262e-06,
+      "loss": 0.8017,
+      "step": 1153
+    },
+    {
+      "epoch": 0.03768163265306122,
+      "grad_norm": 3.1867284774780273,
+      "learning_rate": 9.963945133899413e-06,
+      "loss": 0.7973,
+      "step": 1154
+    },
+    {
+      "epoch": 0.037714285714285714,
+      "grad_norm": 2.98366379737854,
+      "learning_rate": 9.963912475506205e-06,
+      "loss": 0.8105,
+      "step": 1155
+    },
+    {
+      "epoch": 0.037746938775510205,
+      "grad_norm": 2.998530387878418,
+      "learning_rate": 9.963879817113e-06,
+      "loss": 0.7738,
+      "step": 1156
+    },
+    {
+      "epoch": 0.0377795918367347,
+      "grad_norm": 3.0798680782318115,
+      "learning_rate": 9.963847158719791e-06,
+      "loss": 0.7693,
+      "step": 1157
+    },
+    {
+      "epoch": 0.03781224489795918,
+      "grad_norm": 3.2727184295654297,
+      "learning_rate": 9.963814500326584e-06,
+      "loss": 0.7271,
+      "step": 1158
+    },
+    {
+      "epoch": 0.03784489795918367,
+      "grad_norm": 3.05658221244812,
+      "learning_rate": 9.963781841933378e-06,
+      "loss": 0.7759,
+      "step": 1159
+    },
+    {
+      "epoch": 0.037877551020408164,
+      "grad_norm": 3.0974197387695312,
+      "learning_rate": 9.963749183540171e-06,
+      "loss": 0.7844,
+      "step": 1160
+    },
+    {
+      "epoch": 0.037910204081632655,
+      "grad_norm": 3.1191930770874023,
+      "learning_rate": 9.963716525146964e-06,
+      "loss": 0.8066,
+      "step": 1161
+    },
+    {
+      "epoch": 0.037942857142857146,
+      "grad_norm": 3.0258705615997314,
+      "learning_rate": 9.963683866753756e-06,
+      "loss": 0.7373,
+      "step": 1162
+    },
+    {
+      "epoch": 0.03797551020408163,
+      "grad_norm": 3.1126925945281982,
+      "learning_rate": 9.963651208360549e-06,
+      "loss": 0.7577,
+      "step": 1163
+    },
+    {
+      "epoch": 0.03800816326530612,
+      "grad_norm": 3.1630795001983643,
+      "learning_rate": 9.963618549967342e-06,
+      "loss": 0.7333,
+      "step": 1164
+    },
+    {
+      "epoch": 0.03804081632653061,
+      "grad_norm": 3.100018262863159,
+      "learning_rate": 9.963585891574135e-06,
+      "loss": 0.7697,
+      "step": 1165
+    },
+    {
+      "epoch": 0.038073469387755104,
+      "grad_norm": 3.369335889816284,
+      "learning_rate": 9.963553233180929e-06,
+      "loss": 0.7788,
+      "step": 1166
+    },
+    {
+      "epoch": 0.03810612244897959,
+      "grad_norm": 3.357944965362549,
+      "learning_rate": 9.963520574787722e-06,
+      "loss": 0.7744,
+      "step": 1167
+    },
+    {
+      "epoch": 0.03813877551020408,
+      "grad_norm": 3.253232717514038,
+      "learning_rate": 9.963487916394513e-06,
+      "loss": 0.752,
+      "step": 1168
+    },
+    {
+      "epoch": 0.03817142857142857,
+      "grad_norm": 2.9491517543792725,
+      "learning_rate": 9.963455258001307e-06,
+      "loss": 0.7657,
+      "step": 1169
+    },
+    {
+      "epoch": 0.03820408163265306,
+      "grad_norm": 3.1153948307037354,
+      "learning_rate": 9.9634225996081e-06,
+      "loss": 0.7545,
+      "step": 1170
+    },
+    {
+      "epoch": 0.038236734693877554,
+      "grad_norm": 3.0334460735321045,
+      "learning_rate": 9.963389941214893e-06,
+      "loss": 0.7999,
+      "step": 1171
+    },
+    {
+      "epoch": 0.03826938775510204,
+      "grad_norm": 3.0638813972473145,
+      "learning_rate": 9.963357282821686e-06,
+      "loss": 0.7922,
+      "step": 1172
+    },
+    {
+      "epoch": 0.03830204081632653,
+      "grad_norm": 3.0759074687957764,
+      "learning_rate": 9.963324624428478e-06,
+      "loss": 0.7521,
+      "step": 1173
+    },
+    {
+      "epoch": 0.03833469387755102,
+      "grad_norm": 3.0248312950134277,
+      "learning_rate": 9.963291966035273e-06,
+      "loss": 0.7695,
+      "step": 1174
+    },
+    {
+      "epoch": 0.03836734693877551,
+      "grad_norm": 3.101550817489624,
+      "learning_rate": 9.963259307642064e-06,
+      "loss": 0.7446,
+      "step": 1175
+    },
+    {
+      "epoch": 0.0384,
+      "grad_norm": 3.0817489624023438,
+      "learning_rate": 9.963226649248858e-06,
+      "loss": 0.7653,
+      "step": 1176
+    },
+    {
+      "epoch": 0.03843265306122449,
+      "grad_norm": 3.051537036895752,
+      "learning_rate": 9.96319399085565e-06,
+      "loss": 0.7723,
+      "step": 1177
+    },
+    {
+      "epoch": 0.03846530612244898,
+      "grad_norm": 3.030776023864746,
+      "learning_rate": 9.963161332462442e-06,
+      "loss": 0.7762,
+      "step": 1178
+    },
+    {
+      "epoch": 0.03849795918367347,
+      "grad_norm": 3.1081271171569824,
+      "learning_rate": 9.963128674069237e-06,
+      "loss": 0.8198,
+      "step": 1179
+    },
+    {
+      "epoch": 0.03853061224489796,
+      "grad_norm": 3.1396594047546387,
+      "learning_rate": 9.963096015676029e-06,
+      "loss": 0.7389,
+      "step": 1180
+    },
+    {
+      "epoch": 0.038563265306122446,
+      "grad_norm": 2.977698564529419,
+      "learning_rate": 9.963063357282822e-06,
+      "loss": 0.7961,
+      "step": 1181
+    },
+    {
+      "epoch": 0.03859591836734694,
+      "grad_norm": 3.048220157623291,
+      "learning_rate": 9.963030698889615e-06,
+      "loss": 0.8025,
+      "step": 1182
+    },
+    {
+      "epoch": 0.03862857142857143,
+      "grad_norm": 3.087770700454712,
+      "learning_rate": 9.962998040496409e-06,
+      "loss": 0.6993,
+      "step": 1183
+    },
+    {
+      "epoch": 0.03866122448979592,
+      "grad_norm": 3.0148611068725586,
+      "learning_rate": 9.962965382103202e-06,
+      "loss": 0.7555,
+      "step": 1184
+    },
+    {
+      "epoch": 0.03869387755102041,
+      "grad_norm": 3.4241836071014404,
+      "learning_rate": 9.962932723709993e-06,
+      "loss": 0.7876,
+      "step": 1185
+    },
+    {
+      "epoch": 0.038726530612244896,
+      "grad_norm": 3.200383424758911,
+      "learning_rate": 9.962900065316787e-06,
+      "loss": 0.7739,
+      "step": 1186
+    },
+    {
+      "epoch": 0.03875918367346939,
+      "grad_norm": 3.0834004878997803,
+      "learning_rate": 9.96286740692358e-06,
+      "loss": 0.7524,
+      "step": 1187
+    },
+    {
+      "epoch": 0.03879183673469388,
+      "grad_norm": 2.9796390533447266,
+      "learning_rate": 9.962834748530373e-06,
+      "loss": 0.7495,
+      "step": 1188
+    },
+    {
+      "epoch": 0.03882448979591837,
+      "grad_norm": 3.1456053256988525,
+      "learning_rate": 9.962802090137166e-06,
+      "loss": 0.7939,
+      "step": 1189
+    },
+    {
+      "epoch": 0.038857142857142854,
+      "grad_norm": 3.227130651473999,
+      "learning_rate": 9.96276943174396e-06,
+      "loss": 0.7283,
+      "step": 1190
+    },
+    {
+      "epoch": 0.038889795918367345,
+      "grad_norm": 3.2170636653900146,
+      "learning_rate": 9.962736773350751e-06,
+      "loss": 0.7481,
+      "step": 1191
+    },
+    {
+      "epoch": 0.03892244897959184,
+      "grad_norm": 6.70438814163208,
+      "learning_rate": 9.962704114957546e-06,
+      "loss": 0.7565,
+      "step": 1192
+    },
+    {
+      "epoch": 0.03895510204081633,
+      "grad_norm": 3.2916693687438965,
+      "learning_rate": 9.962671456564338e-06,
+      "loss": 0.7732,
+      "step": 1193
+    },
+    {
+      "epoch": 0.03898775510204082,
+      "grad_norm": 3.4526243209838867,
+      "learning_rate": 9.96263879817113e-06,
+      "loss": 0.7563,
+      "step": 1194
+    },
+    {
+      "epoch": 0.039020408163265304,
+      "grad_norm": 3.188997507095337,
+      "learning_rate": 9.962606139777924e-06,
+      "loss": 0.771,
+      "step": 1195
+    },
+    {
+      "epoch": 0.039053061224489795,
+      "grad_norm": 3.0594732761383057,
+      "learning_rate": 9.962573481384716e-06,
+      "loss": 0.7433,
+      "step": 1196
+    },
+    {
+      "epoch": 0.039085714285714286,
+      "grad_norm": 3.239617109298706,
+      "learning_rate": 9.96254082299151e-06,
+      "loss": 0.7823,
+      "step": 1197
+    },
+    {
+      "epoch": 0.03911836734693878,
+      "grad_norm": 2.983693838119507,
+      "learning_rate": 9.962508164598302e-06,
+      "loss": 0.7557,
+      "step": 1198
+    },
+    {
+      "epoch": 0.03915102040816327,
+      "grad_norm": 2.9180068969726562,
+      "learning_rate": 9.962475506205095e-06,
+      "loss": 0.7915,
+      "step": 1199
+    },
+    {
+      "epoch": 0.03918367346938775,
+      "grad_norm": 3.102025270462036,
+      "learning_rate": 9.962442847811888e-06,
+      "loss": 0.764,
+      "step": 1200
+    },
+    {
+      "epoch": 0.03918367346938775,
+      "eval_loss": 0.807881772518158,
+      "eval_runtime": 85.0203,
+      "eval_samples_per_second": 1.176,
+      "eval_steps_per_second": 1.176,
+      "step": 1200
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 306250,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 50,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.305701335141549e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}