diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,21523 @@
+{
+  "best_global_step": 5900,
+  "best_metric": 1.0105234384536743,
+  "best_model_checkpoint": "./outputs/checkpoint-5900",
+  "epoch": 2.176326530612245,
+  "eval_steps": 100,
+  "global_step": 6000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0007256235827664399,
+      "grad_norm": 0.88671875,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 1.5116,
+      "step": 2
+    },
+    {
+      "epoch": 0.0014512471655328798,
+      "grad_norm": 1.8046875,
+      "learning_rate": 6e-06,
+      "loss": 1.7699,
+      "step": 4
+    },
+    {
+      "epoch": 0.0021768707482993197,
+      "grad_norm": 0.7890625,
+      "learning_rate": 1e-05,
+      "loss": 1.4788,
+      "step": 6
+    },
+    {
+      "epoch": 0.0029024943310657597,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.4000000000000001e-05,
+      "loss": 1.6688,
+      "step": 8
+    },
+    {
+      "epoch": 0.0036281179138321997,
+      "grad_norm": 1.7421875,
+      "learning_rate": 1.8e-05,
+      "loss": 1.798,
+      "step": 10
+    },
+    {
+      "epoch": 0.004353741496598639,
+      "grad_norm": 1.171875,
+      "learning_rate": 2.2000000000000003e-05,
+      "loss": 1.6686,
+      "step": 12
+    },
+    {
+      "epoch": 0.005079365079365079,
+      "grad_norm": 1.1484375,
+      "learning_rate": 2.6000000000000002e-05,
+      "loss": 1.8021,
+      "step": 14
+    },
+    {
+      "epoch": 0.005804988662131519,
+      "grad_norm": 0.7890625,
+      "learning_rate": 3e-05,
+      "loss": 1.6768,
+      "step": 16
+    },
+    {
+      "epoch": 0.006530612244897959,
+      "grad_norm": 0.83203125,
+      "learning_rate": 3.4000000000000007e-05,
+      "loss": 1.4667,
+      "step": 18
+    },
+    {
+      "epoch": 0.0072562358276643995,
+      "grad_norm": 0.98828125,
+      "learning_rate": 3.8e-05,
+      "loss": 1.6074,
+      "step": 20
+    },
+    {
+      "epoch": 0.00798185941043084,
+      "grad_norm": 0.90234375,
+      "learning_rate": 4.2e-05,
+      "loss": 1.6353,
+      "step": 22
+    },
+    {
+      "epoch": 0.008707482993197279,
+      "grad_norm": 0.92578125,
+      "learning_rate": 4.600000000000001e-05,
+      "loss": 1.5452,
+      "step": 24
+    },
+    {
+      "epoch": 0.00943310657596372,
+      "grad_norm": 0.87109375,
+      "learning_rate": 5e-05,
+      "loss": 1.6205,
+      "step": 26
+    },
+    {
+      "epoch": 0.010158730158730159,
+      "grad_norm": 0.61328125,
+      "learning_rate": 5.4000000000000005e-05,
+      "loss": 1.4568,
+      "step": 28
+    },
+    {
+      "epoch": 0.010884353741496598,
+      "grad_norm": 0.69921875,
+      "learning_rate": 5.8e-05,
+      "loss": 1.4901,
+      "step": 30
+    },
+    {
+      "epoch": 0.011609977324263039,
+      "grad_norm": 0.98046875,
+      "learning_rate": 6.2e-05,
+      "loss": 1.422,
+      "step": 32
+    },
+    {
+      "epoch": 0.012335600907029478,
+      "grad_norm": 0.984375,
+      "learning_rate": 6.6e-05,
+      "loss": 1.406,
+      "step": 34
+    },
+    {
+      "epoch": 0.013061224489795919,
+      "grad_norm": 0.69921875,
+      "learning_rate": 7e-05,
+      "loss": 1.2822,
+      "step": 36
+    },
+    {
+      "epoch": 0.013786848072562358,
+      "grad_norm": 0.86328125,
+      "learning_rate": 7.4e-05,
+      "loss": 1.3677,
+      "step": 38
+    },
+    {
+      "epoch": 0.014512471655328799,
+      "grad_norm": 0.78125,
+      "learning_rate": 7.800000000000001e-05,
+      "loss": 1.3597,
+      "step": 40
+    },
+    {
+      "epoch": 0.015238095238095238,
+      "grad_norm": 0.828125,
+      "learning_rate": 8.2e-05,
+      "loss": 1.263,
+      "step": 42
+    },
+    {
+      "epoch": 0.01596371882086168,
+      "grad_norm": 0.75390625,
+      "learning_rate": 8.6e-05,
+      "loss": 1.314,
+      "step": 44
+    },
+    {
+      "epoch": 0.016689342403628116,
+      "grad_norm": 0.46484375,
+      "learning_rate": 9e-05,
+      "loss": 1.2784,
+      "step": 46
+    },
+    {
+      "epoch": 0.017414965986394557,
+      "grad_norm": 0.65625,
+      "learning_rate": 9.4e-05,
+      "loss": 1.1945,
+      "step": 48
+    },
+    {
+      "epoch": 0.018140589569160998,
+      "grad_norm": 0.5703125,
+      "learning_rate": 9.8e-05,
+      "loss": 1.2637,
+      "step": 50
+    },
+    {
+      "epoch": 0.01886621315192744,
+      "grad_norm": 0.625,
+      "learning_rate": 0.00010200000000000001,
+      "loss": 1.1792,
+      "step": 52
+    },
+    {
+      "epoch": 0.019591836734693877,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00010600000000000002,
+      "loss": 1.1901,
+      "step": 54
+    },
+    {
+      "epoch": 0.020317460317460317,
+      "grad_norm": 0.392578125,
+      "learning_rate": 0.00011000000000000002,
+      "loss": 1.1808,
+      "step": 56
+    },
+    {
+      "epoch": 0.02104308390022676,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00011399999999999999,
+      "loss": 1.134,
+      "step": 58
+    },
+    {
+      "epoch": 0.021768707482993196,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.000118,
+      "loss": 1.1071,
+      "step": 60
+    },
+    {
+      "epoch": 0.022494331065759637,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.000122,
+      "loss": 1.1066,
+      "step": 62
+    },
+    {
+      "epoch": 0.023219954648526078,
+      "grad_norm": 0.30859375,
+      "learning_rate": 0.000126,
+      "loss": 1.1922,
+      "step": 64
+    },
+    {
+      "epoch": 0.02394557823129252,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.00013000000000000002,
+      "loss": 1.1028,
+      "step": 66
+    },
+    {
+      "epoch": 0.024671201814058956,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.000134,
+      "loss": 1.0819,
+      "step": 68
+    },
+    {
+      "epoch": 0.025396825396825397,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.000138,
+      "loss": 1.1744,
+      "step": 70
+    },
+    {
+      "epoch": 0.026122448979591838,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.000142,
+      "loss": 1.0784,
+      "step": 72
+    },
+    {
+      "epoch": 0.026848072562358275,
+      "grad_norm": 0.21875,
+      "learning_rate": 0.000146,
+      "loss": 1.0396,
+      "step": 74
+    },
+    {
+      "epoch": 0.027573696145124716,
+      "grad_norm": 0.2333984375,
+      "learning_rate": 0.00015000000000000001,
+      "loss": 1.2014,
+      "step": 76
+    },
+    {
+      "epoch": 0.028299319727891157,
+      "grad_norm": 0.23828125,
+      "learning_rate": 0.000154,
+      "loss": 1.1024,
+      "step": 78
+    },
+    {
+      "epoch": 0.029024943310657598,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.00015800000000000002,
+      "loss": 0.9752,
+      "step": 80
+    },
+    {
+      "epoch": 0.029750566893424035,
+      "grad_norm": 0.2373046875,
+      "learning_rate": 0.000162,
+      "loss": 1.0902,
+      "step": 82
+    },
+    {
+      "epoch": 0.030476190476190476,
+      "grad_norm": 0.19921875,
+      "learning_rate": 0.000166,
+      "loss": 1.0058,
+      "step": 84
+    },
+    {
+      "epoch": 0.031201814058956917,
+      "grad_norm": 0.2197265625,
+      "learning_rate": 0.00017,
+      "loss": 1.0682,
+      "step": 86
+    },
+    {
+      "epoch": 0.03192743764172336,
+      "grad_norm": 0.2119140625,
+      "learning_rate": 0.000174,
+      "loss": 1.0874,
+      "step": 88
+    },
+    {
+      "epoch": 0.0326530612244898,
+      "grad_norm": 0.458984375,
+      "learning_rate": 0.00017800000000000002,
+      "loss": 1.0739,
+      "step": 90
+    },
+    {
+      "epoch": 0.03337868480725623,
+      "grad_norm": 0.212890625,
+      "learning_rate": 0.000182,
+      "loss": 1.0259,
+      "step": 92
+    },
+    {
+      "epoch": 0.034104308390022674,
+      "grad_norm": 0.2470703125,
+      "learning_rate": 0.00018600000000000002,
+      "loss": 1.0279,
+      "step": 94
+    },
+    {
+      "epoch": 0.034829931972789115,
+      "grad_norm": 0.2001953125,
+      "learning_rate": 0.00019,
+      "loss": 1.0263,
+      "step": 96
+    },
+    {
+      "epoch": 0.035555555555555556,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.000194,
+      "loss": 1.0206,
+      "step": 98
+    },
+    {
+      "epoch": 0.036281179138321996,
+      "grad_norm": 0.2353515625,
+      "learning_rate": 0.00019800000000000002,
+      "loss": 1.1275,
+      "step": 100
+    },
+    {
+      "epoch": 0.036281179138321996,
+      "eval_loss": 1.0672887563705444,
+      "eval_runtime": 99.6091,
+      "eval_samples_per_second": 18.071,
+      "eval_steps_per_second": 0.572,
+      "step": 100
+    },
+    {
+      "epoch": 0.03700680272108844,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.0001999755231917758,
+      "loss": 1.1141,
+      "step": 102
+    },
+    {
+      "epoch": 0.03773242630385488,
+      "grad_norm": 0.2080078125,
+      "learning_rate": 0.00019992656957532737,
+      "loss": 1.0543,
+      "step": 104
+    },
+    {
+      "epoch": 0.03845804988662131,
+      "grad_norm": 0.20703125,
+      "learning_rate": 0.00019987761595887898,
+      "loss": 1.0389,
+      "step": 106
+    },
+    {
+      "epoch": 0.03918367346938775,
+      "grad_norm": 0.1982421875,
+      "learning_rate": 0.00019982866234243056,
+      "loss": 1.0699,
+      "step": 108
+    },
+    {
+      "epoch": 0.039909297052154194,
+      "grad_norm": 0.20703125,
+      "learning_rate": 0.00019977970872598214,
+      "loss": 1.0756,
+      "step": 110
+    },
+    {
+      "epoch": 0.040634920634920635,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 0.00019973075510953373,
+      "loss": 1.0309,
+      "step": 112
+    },
+    {
+      "epoch": 0.041360544217687076,
+      "grad_norm": 0.1884765625,
+      "learning_rate": 0.0001996818014930853,
+      "loss": 0.9821,
+      "step": 114
+    },
+    {
+      "epoch": 0.04208616780045352,
+      "grad_norm": 0.2353515625,
+      "learning_rate": 0.0001996328478766369,
+      "loss": 1.0897,
+      "step": 116
+    },
+    {
+      "epoch": 0.04281179138321996,
+      "grad_norm": 0.2080078125,
+      "learning_rate": 0.0001995838942601885,
+      "loss": 1.0576,
+      "step": 118
+    },
+    {
+      "epoch": 0.04353741496598639,
+      "grad_norm": 0.2001953125,
+      "learning_rate": 0.00019953494064374008,
+      "loss": 1.0482,
+      "step": 120
+    },
+    {
+      "epoch": 0.04426303854875283,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.00019948598702729164,
+      "loss": 1.0018,
+      "step": 122
+    },
+    {
+      "epoch": 0.04498866213151927,
+      "grad_norm": 0.1865234375,
+      "learning_rate": 0.00019943703341084322,
+      "loss": 1.0637,
+      "step": 124
+    },
+    {
+      "epoch": 0.045714285714285714,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.00019938807979439483,
+      "loss": 1.0471,
+      "step": 126
+    },
+    {
+      "epoch": 0.046439909297052155,
+      "grad_norm": 0.1875,
+      "learning_rate": 0.0001993391261779464,
+      "loss": 0.9958,
+      "step": 128
+    },
+    {
+      "epoch": 0.047165532879818596,
+      "grad_norm": 0.201171875,
+      "learning_rate": 0.000199290172561498,
+      "loss": 1.0396,
+      "step": 130
+    },
+    {
+      "epoch": 0.04789115646258504,
+      "grad_norm": 0.2333984375,
+      "learning_rate": 0.00019924121894504958,
+      "loss": 1.0407,
+      "step": 132
+    },
+    {
+      "epoch": 0.04861678004535147,
+      "grad_norm": 0.201171875,
+      "learning_rate": 0.00019919226532860116,
+      "loss": 1.043,
+      "step": 134
+    },
+    {
+      "epoch": 0.04934240362811791,
+      "grad_norm": 0.193359375,
+      "learning_rate": 0.00019914331171215274,
+      "loss": 1.0859,
+      "step": 136
+    },
+    {
+      "epoch": 0.05006802721088435,
+      "grad_norm": 0.181640625,
+      "learning_rate": 0.00019909435809570435,
+      "loss": 1.1251,
+      "step": 138
+    },
+    {
+      "epoch": 0.050793650793650794,
+      "grad_norm": 0.2431640625,
+      "learning_rate": 0.0001990454044792559,
+      "loss": 1.0666,
+      "step": 140
+    },
+    {
+      "epoch": 0.051519274376417235,
+      "grad_norm": 0.203125,
+      "learning_rate": 0.0001989964508628075,
+      "loss": 0.9962,
+      "step": 142
+    },
+    {
+      "epoch": 0.052244897959183675,
+      "grad_norm": 0.1923828125,
+      "learning_rate": 0.00019894749724635907,
+      "loss": 1.1262,
+      "step": 144
+    },
+    {
+      "epoch": 0.052970521541950116,
+      "grad_norm": 0.1875,
+      "learning_rate": 0.00019889854362991068,
+      "loss": 1.0174,
+      "step": 146
+    },
+    {
+      "epoch": 0.05369614512471655,
+      "grad_norm": 0.181640625,
+      "learning_rate": 0.00019884959001346226,
+      "loss": 1.0509,
+      "step": 148
+    },
+    {
+      "epoch": 0.05442176870748299,
+      "grad_norm": 0.21875,
+      "learning_rate": 0.00019880063639701382,
+      "loss": 1.05,
+      "step": 150
+    },
+    {
+      "epoch": 0.05514739229024943,
+      "grad_norm": 0.2109375,
+      "learning_rate": 0.00019875168278056543,
+      "loss": 1.134,
+      "step": 152
+    },
+    {
+      "epoch": 0.05587301587301587,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.000198702729164117,
+      "loss": 0.9511,
+      "step": 154
+    },
+    {
+      "epoch": 0.056598639455782314,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.0001986537755476686,
+      "loss": 0.9989,
+      "step": 156
+    },
+    {
+      "epoch": 0.057324263038548755,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.0001986048219312202,
+      "loss": 0.9466,
+      "step": 158
+    },
+    {
+      "epoch": 0.058049886621315196,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.00019855586831477175,
+      "loss": 0.9605,
+      "step": 160
+    },
+    {
+      "epoch": 0.05877551020408163,
+      "grad_norm": 0.2001953125,
+      "learning_rate": 0.00019850691469832334,
+      "loss": 1.0511,
+      "step": 162
+    },
+    {
+      "epoch": 0.05950113378684807,
+      "grad_norm": 0.205078125,
+      "learning_rate": 0.00019845796108187492,
+      "loss": 1.0575,
+      "step": 164
+    },
+    {
+      "epoch": 0.06022675736961451,
+      "grad_norm": 0.189453125,
+      "learning_rate": 0.00019840900746542653,
+      "loss": 1.0211,
+      "step": 166
+    },
+    {
+      "epoch": 0.06095238095238095,
+      "grad_norm": 0.1923828125,
+      "learning_rate": 0.0001983600538489781,
+      "loss": 1.0435,
+      "step": 168
+    },
+    {
+      "epoch": 0.06167800453514739,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 0.00019831110023252967,
+      "loss": 0.9651,
+      "step": 170
+    },
+    {
+      "epoch": 0.062403628117913834,
+      "grad_norm": 0.17578125,
+      "learning_rate": 0.00019826214661608128,
+      "loss": 0.9683,
+      "step": 172
+    },
+    {
+      "epoch": 0.06312925170068028,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.00019821319299963286,
+      "loss": 1.0064,
+      "step": 174
+    },
+    {
+      "epoch": 0.06385487528344672,
+      "grad_norm": 0.1875,
+      "learning_rate": 0.00019816423938318444,
+      "loss": 1.0669,
+      "step": 176
+    },
+    {
+      "epoch": 0.06458049886621316,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.00019811528576673605,
+      "loss": 1.0931,
+      "step": 178
+    },
+    {
+      "epoch": 0.0653061224489796,
+      "grad_norm": 0.19921875,
+      "learning_rate": 0.0001980663321502876,
+      "loss": 1.04,
+      "step": 180
+    },
+    {
+      "epoch": 0.06603174603174604,
+      "grad_norm": 0.1884765625,
+      "learning_rate": 0.0001980173785338392,
+      "loss": 0.9917,
+      "step": 182
+    },
+    {
+      "epoch": 0.06675736961451247,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.00019796842491739077,
+      "loss": 0.9985,
+      "step": 184
+    },
+    {
+      "epoch": 0.0674829931972789,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.00019791947130094238,
+      "loss": 0.9511,
+      "step": 186
+    },
+    {
+      "epoch": 0.06820861678004535,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.00019787051768449396,
+      "loss": 0.9866,
+      "step": 188
+    },
+    {
+      "epoch": 0.06893424036281179,
+      "grad_norm": 0.1982421875,
+      "learning_rate": 0.00019782156406804552,
+      "loss": 1.0879,
+      "step": 190
+    },
+    {
+      "epoch": 0.06965986394557823,
+      "grad_norm": 0.181640625,
+      "learning_rate": 0.00019777261045159713,
+      "loss": 1.006,
+      "step": 192
+    },
+    {
+      "epoch": 0.07038548752834467,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.0001977236568351487,
+      "loss": 1.0737,
+      "step": 194
+    },
+    {
+      "epoch": 0.07111111111111111,
+      "grad_norm": 0.203125,
+      "learning_rate": 0.0001976747032187003,
+      "loss": 0.9719,
+      "step": 196
+    },
+    {
+      "epoch": 0.07183673469387755,
+      "grad_norm": 0.21484375,
+      "learning_rate": 0.0001976257496022519,
+      "loss": 1.0539,
+      "step": 198
+    },
+    {
+      "epoch": 0.07256235827664399,
+      "grad_norm": 0.19921875,
+      "learning_rate": 0.00019757679598580345,
+      "loss": 1.0863,
+      "step": 200
+    },
+    {
+      "epoch": 0.07256235827664399,
+      "eval_loss": 1.0431190729141235,
+      "eval_runtime": 99.2841,
+      "eval_samples_per_second": 18.13,
+      "eval_steps_per_second": 0.574,
+      "step": 200
+    },
+    {
+      "epoch": 0.07328798185941043,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 0.00019752784236935504,
+      "loss": 1.1022,
+      "step": 202
+    },
+    {
+      "epoch": 0.07401360544217687,
+      "grad_norm": 0.177734375,
+      "learning_rate": 0.00019747888875290662,
+      "loss": 0.9781,
+      "step": 204
+    },
+    {
+      "epoch": 0.07473922902494332,
+      "grad_norm": 0.21875,
+      "learning_rate": 0.00019742993513645823,
+      "loss": 1.0735,
+      "step": 206
+    },
+    {
+      "epoch": 0.07546485260770976,
+      "grad_norm": 0.1865234375,
+      "learning_rate": 0.0001973809815200098,
+      "loss": 0.957,
+      "step": 208
+    },
+    {
+      "epoch": 0.0761904761904762,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.00019733202790356137,
+      "loss": 1.0229,
+      "step": 210
+    },
+    {
+      "epoch": 0.07691609977324262,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 0.00019728307428711298,
+      "loss": 0.9876,
+      "step": 212
+    },
+    {
+      "epoch": 0.07764172335600907,
+      "grad_norm": 0.1884765625,
+      "learning_rate": 0.00019723412067066456,
+      "loss": 1.0047,
+      "step": 214
+    },
+    {
+      "epoch": 0.0783673469387755,
+      "grad_norm": 0.1875,
+      "learning_rate": 0.00019718516705421614,
+      "loss": 0.9995,
+      "step": 216
+    },
+    {
+      "epoch": 0.07909297052154195,
+      "grad_norm": 0.1923828125,
+      "learning_rate": 0.00019713621343776772,
+      "loss": 1.0447,
+      "step": 218
+    },
+    {
+      "epoch": 0.07981859410430839,
+      "grad_norm": 0.193359375,
+      "learning_rate": 0.0001970872598213193,
+      "loss": 1.0969,
+      "step": 220
+    },
+    {
+      "epoch": 0.08054421768707483,
+      "grad_norm": 0.203125,
+      "learning_rate": 0.0001970383062048709,
+      "loss": 0.9948,
+      "step": 222
+    },
+    {
+      "epoch": 0.08126984126984127,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.00019698935258842247,
+      "loss": 0.9642,
+      "step": 224
+    },
+    {
+      "epoch": 0.08199546485260771,
+      "grad_norm": 0.19140625,
+      "learning_rate": 0.00019694039897197408,
+      "loss": 1.0414,
+      "step": 226
+    },
+    {
+      "epoch": 0.08272108843537415,
+      "grad_norm": 0.1943359375,
+      "learning_rate": 0.00019689144535552563,
+      "loss": 0.9293,
+      "step": 228
+    },
+    {
+      "epoch": 0.08344671201814059,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 0.00019684249173907722,
+      "loss": 0.9743,
+      "step": 230
+    },
+    {
+      "epoch": 0.08417233560090703,
+      "grad_norm": 0.189453125,
+      "learning_rate": 0.00019679353812262883,
+      "loss": 1.0206,
+      "step": 232
+    },
+    {
+      "epoch": 0.08489795918367347,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.0001967445845061804,
+      "loss": 0.9956,
+      "step": 234
+    },
+    {
+      "epoch": 0.08562358276643992,
+      "grad_norm": 0.185546875,
+      "learning_rate": 0.000196695630889732,
+      "loss": 1.0204,
+      "step": 236
+    },
+    {
+      "epoch": 0.08634920634920636,
+      "grad_norm": 0.177734375,
+      "learning_rate": 0.00019664667727328357,
+      "loss": 0.9644,
+      "step": 238
+    },
+    {
+      "epoch": 0.08707482993197278,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.00019659772365683515,
+      "loss": 0.9574,
+      "step": 240
+    },
+    {
+      "epoch": 0.08780045351473922,
+      "grad_norm": 0.181640625,
+      "learning_rate": 0.00019654877004038674,
+      "loss": 1.0699,
+      "step": 242
+    },
+    {
+      "epoch": 0.08852607709750566,
+      "grad_norm": 0.193359375,
+      "learning_rate": 0.00019649981642393832,
+      "loss": 1.0396,
+      "step": 244
+    },
+    {
+      "epoch": 0.0892517006802721,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.00019645086280748993,
+      "loss": 0.9685,
+      "step": 246
+    },
+    {
+      "epoch": 0.08997732426303855,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.00019640190919104148,
+      "loss": 0.9963,
+      "step": 248
+    },
+    {
+      "epoch": 0.09070294784580499,
+      "grad_norm": 0.181640625,
+      "learning_rate": 0.00019635295557459307,
+      "loss": 1.0085,
+      "step": 250
+    },
+    {
+      "epoch": 0.09142857142857143,
+      "grad_norm": 0.216796875,
+      "learning_rate": 0.00019630400195814468,
+      "loss": 1.0444,
+      "step": 252
+    },
+    {
+      "epoch": 0.09215419501133787,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.00019625504834169626,
+      "loss": 0.9959,
+      "step": 254
+    },
+    {
+      "epoch": 0.09287981859410431,
+      "grad_norm": 0.189453125,
+      "learning_rate": 0.00019620609472524784,
+      "loss": 1.0705,
+      "step": 256
+    },
+    {
+      "epoch": 0.09360544217687075,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.00019615714110879942,
+      "loss": 1.0762,
+      "step": 258
+    },
+    {
+      "epoch": 0.09433106575963719,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 0.000196108187492351,
+      "loss": 1.0227,
+      "step": 260
+    },
+    {
+      "epoch": 0.09505668934240363,
+      "grad_norm": 0.2080078125,
+      "learning_rate": 0.00019605923387590259,
+      "loss": 1.0503,
+      "step": 262
+    },
+    {
+      "epoch": 0.09578231292517007,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.00019601028025945417,
+      "loss": 1.0013,
+      "step": 264
+    },
+    {
+      "epoch": 0.09650793650793651,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.00019596132664300578,
+      "loss": 1.0844,
+      "step": 266
+    },
+    {
+      "epoch": 0.09723356009070294,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.00019591237302655733,
+      "loss": 0.9824,
+      "step": 268
+    },
+    {
+      "epoch": 0.09795918367346938,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.00019586341941010892,
+      "loss": 0.9688,
+      "step": 270
+    },
+    {
+      "epoch": 0.09868480725623582,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.00019581446579366052,
+      "loss": 0.923,
+      "step": 272
+    },
+    {
+      "epoch": 0.09941043083900226,
+      "grad_norm": 0.2197265625,
+      "learning_rate": 0.0001957655121772121,
+      "loss": 0.9817,
+      "step": 274
+    },
+    {
+      "epoch": 0.1001360544217687,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.0001957165585607637,
+      "loss": 0.9691,
+      "step": 276
+    },
+    {
+      "epoch": 0.10086167800453515,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.00019566760494431527,
+      "loss": 1.019,
+      "step": 278
+    },
+    {
+      "epoch": 0.10158730158730159,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.00019561865132786685,
+      "loss": 1.008,
+      "step": 280
+    },
+    {
+      "epoch": 0.10231292517006803,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.00019556969771141844,
+      "loss": 0.9477,
+      "step": 282
+    },
+    {
+      "epoch": 0.10303854875283447,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.00019552074409497002,
+      "loss": 0.962,
+      "step": 284
+    },
+    {
+      "epoch": 0.10376417233560091,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.0001954717904785216,
+      "loss": 0.9685,
+      "step": 286
+    },
+    {
+      "epoch": 0.10448979591836735,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.00019542283686207318,
+      "loss": 1.015,
+      "step": 288
+    },
+    {
+      "epoch": 0.10521541950113379,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.00019537388324562477,
+      "loss": 0.9902,
+      "step": 290
+    },
+    {
+      "epoch": 0.10594104308390023,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.00019532492962917637,
+      "loss": 1.0309,
+      "step": 292
+    },
+    {
+      "epoch": 0.10666666666666667,
+      "grad_norm": 0.1953125,
+      "learning_rate": 0.00019527597601272796,
+      "loss": 1.0179,
+      "step": 294
+    },
+    {
+      "epoch": 0.1073922902494331,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.0001952270223962795,
+      "loss": 1.0208,
+      "step": 296
+    },
+    {
+      "epoch": 0.10811791383219954,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.00019517806877983112,
+      "loss": 1.0252,
+      "step": 298
+    },
+    {
+      "epoch": 0.10884353741496598,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.0001951291151633827,
+      "loss": 1.0185,
+      "step": 300
+    },
+    {
+      "epoch": 0.10884353741496598,
+      "eval_loss": 1.036349892616272,
+      "eval_runtime": 91.1673,
+      "eval_samples_per_second": 19.744,
+      "eval_steps_per_second": 0.625,
+      "step": 300
+    },
+    {
+      "epoch": 0.10956916099773242,
+      "grad_norm": 0.177734375,
+      "learning_rate": 0.00019508016154693429,
+      "loss": 0.9886,
+      "step": 302
+    },
+    {
+      "epoch": 0.11029478458049886,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.00019503120793048587,
+      "loss": 1.0495,
+      "step": 304
+    },
+    {
+      "epoch": 0.1110204081632653,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.00019498225431403745,
+      "loss": 1.0057,
+      "step": 306
+    },
+    {
+      "epoch": 0.11174603174603175,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.00019493330069758903,
+      "loss": 1.042,
+      "step": 308
+    },
+    {
+      "epoch": 0.11247165532879819,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.00019488434708114062,
+      "loss": 1.0619,
+      "step": 310
+    },
+    {
+      "epoch": 0.11319727891156463,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.00019483539346469222,
+      "loss": 1.0055,
+      "step": 312
+    },
+    {
+      "epoch": 0.11392290249433107,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 0.0001947864398482438,
+      "loss": 0.9609,
+      "step": 314
+    },
+    {
+      "epoch": 0.11464852607709751,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.00019473748623179536,
+      "loss": 1.0478,
+      "step": 316
+    },
+    {
+      "epoch": 0.11537414965986395,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 0.00019468853261534697,
+      "loss": 1.0008,
+      "step": 318
+    },
+    {
+      "epoch": 0.11609977324263039,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.00019463957899889855,
+      "loss": 0.9464,
+      "step": 320
+    },
+    {
+      "epoch": 0.11682539682539683,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.00019459062538245014,
+      "loss": 1.0368,
+      "step": 322
+    },
+    {
+      "epoch": 0.11755102040816326,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.00019454167176600172,
+      "loss": 1.0134,
+      "step": 324
+    },
+    {
+      "epoch": 0.1182766439909297,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.0001944927181495533,
+      "loss": 0.9533,
+      "step": 326
+    },
+    {
+      "epoch": 0.11900226757369614,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.00019444376453310488,
+      "loss": 0.9526,
+      "step": 328
+    },
+    {
+      "epoch": 0.11972789115646258,
+      "grad_norm": 0.18359375,
+      "learning_rate": 0.00019439481091665647,
+      "loss": 0.9875,
+      "step": 330
+    },
+    {
+      "epoch": 0.12045351473922902,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.00019434585730020807,
+      "loss": 1.0052,
+      "step": 332
+    },
+    {
+      "epoch": 0.12117913832199546,
+      "grad_norm": 0.1943359375,
+      "learning_rate": 0.00019429690368375966,
+      "loss": 1.0672,
+      "step": 334
+    },
+    {
+      "epoch": 0.1219047619047619,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.0001942479500673112,
+      "loss": 0.9938,
+      "step": 336
+    },
+    {
+      "epoch": 0.12263038548752835,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.00019419899645086282,
+      "loss": 0.9938,
+      "step": 338
+    },
+    {
+      "epoch": 0.12335600907029479,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.0001941500428344144,
+      "loss": 0.9733,
+      "step": 340
+    },
+    {
+      "epoch": 0.12408163265306123,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.00019410108921796599,
+      "loss": 0.9917,
+      "step": 342
+    },
+    {
+      "epoch": 0.12480725623582767,
+      "grad_norm": 0.2041015625,
+      "learning_rate": 0.00019405213560151757,
+      "loss": 0.9594,
+      "step": 344
+    },
+    {
+      "epoch": 0.1255328798185941,
+      "grad_norm": 0.177734375,
+      "learning_rate": 0.00019400318198506915,
+      "loss": 1.0255,
+      "step": 346
+    },
+    {
+      "epoch": 0.12625850340136055,
+      "grad_norm": 0.2294921875,
+      "learning_rate": 0.00019395422836862073,
+      "loss": 1.0643,
+      "step": 348
+    },
+    {
+      "epoch": 0.12698412698412698,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.00019390527475217232,
+      "loss": 0.9907,
+      "step": 350
+    },
+    {
+      "epoch": 0.12770975056689343,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.00019385632113572392,
+      "loss": 1.0455,
+      "step": 352
+    },
+    {
+      "epoch": 0.12843537414965986,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.0001938073675192755,
+      "loss": 0.9612,
+      "step": 354
+    },
+    {
+      "epoch": 0.1291609977324263,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.00019375841390282706,
+      "loss": 0.9235,
+      "step": 356
+    },
+    {
+      "epoch": 0.12988662131519274,
+      "grad_norm": 0.193359375,
+      "learning_rate": 0.00019370946028637867,
+      "loss": 1.009,
+      "step": 358
+    },
+    {
+      "epoch": 0.1306122448979592,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.00019366050666993025,
+      "loss": 1.0014,
+      "step": 360
+    },
+    {
+      "epoch": 0.13133786848072562,
+      "grad_norm": 0.173828125,
+      "learning_rate": 0.00019361155305348184,
+      "loss": 0.9666,
+      "step": 362
+    },
+    {
+      "epoch": 0.13206349206349208,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.00019356259943703342,
+      "loss": 1.0091,
+      "step": 364
+    },
+    {
+      "epoch": 0.1327891156462585,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.000193513645820585,
+      "loss": 1.0238,
+      "step": 366
+    },
+    {
+      "epoch": 0.13351473922902493,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.00019346469220413658,
+      "loss": 1.0511,
+      "step": 368
+    },
+    {
+      "epoch": 0.13424036281179139,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.00019341573858768817,
+      "loss": 0.986,
+      "step": 370
+    },
+    {
+      "epoch": 0.1349659863945578,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.00019336678497123977,
+      "loss": 1.0094,
+      "step": 372
+    },
+    {
+      "epoch": 0.13569160997732427,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.00019331783135479133,
+      "loss": 1.0596,
+      "step": 374
+    },
+    {
+      "epoch": 0.1364172335600907,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.0001932688777383429,
+      "loss": 0.9845,
+      "step": 376
+    },
+    {
+      "epoch": 0.13714285714285715,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.00019321992412189452,
+      "loss": 0.9855,
+      "step": 378
+    },
+    {
+      "epoch": 0.13786848072562358,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.0001931709705054461,
+      "loss": 1.039,
+      "step": 380
+    },
+    {
+      "epoch": 0.13859410430839003,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.00019312201688899769,
+      "loss": 1.0108,
+      "step": 382
+    },
+    {
+      "epoch": 0.13931972789115646,
+      "grad_norm": 0.18359375,
+      "learning_rate": 0.00019307306327254927,
+      "loss": 0.977,
+      "step": 384
+    },
+    {
+      "epoch": 0.1400453514739229,
+      "grad_norm": 0.17578125,
+      "learning_rate": 0.00019302410965610085,
+      "loss": 0.9754,
+      "step": 386
+    },
+    {
+      "epoch": 0.14077097505668934,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.00019297515603965243,
+      "loss": 1.1274,
+      "step": 388
+    },
+    {
+      "epoch": 0.1414965986394558,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.00019292620242320401,
+      "loss": 1.0126,
+      "step": 390
+    },
+    {
+      "epoch": 0.14222222222222222,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.00019287724880675562,
+      "loss": 1.0009,
+      "step": 392
+    },
+    {
+      "epoch": 0.14294784580498865,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.00019282829519030718,
+      "loss": 0.9729,
+      "step": 394
+    },
+    {
+      "epoch": 0.1436734693877551,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.00019277934157385876,
+      "loss": 1.0157,
+      "step": 396
+    },
+    {
+      "epoch": 0.14439909297052153,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.00019273038795741037,
+      "loss": 0.9222,
+      "step": 398
+    },
+    {
+      "epoch": 0.14512471655328799,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.00019268143434096195,
+      "loss": 0.9869,
+      "step": 400
+    },
+    {
+      "epoch": 0.14512471655328799,
+      "eval_loss": 1.033414363861084,
+      "eval_runtime": 94.5288,
+      "eval_samples_per_second": 19.042,
+      "eval_steps_per_second": 0.603,
+      "step": 400
+    },
+    {
+      "epoch": 0.1458503401360544,
+      "grad_norm": 0.173828125,
+      "learning_rate": 0.00019263248072451354,
+      "loss": 1.0265,
+      "step": 402
+    },
+    {
+      "epoch": 0.14657596371882087,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.00019258352710806512,
+      "loss": 1.0718,
+      "step": 404
+    },
+    {
+      "epoch": 0.1473015873015873,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.0001925345734916167,
+      "loss": 1.0709,
+      "step": 406
+    },
+    {
+      "epoch": 0.14802721088435375,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.00019248561987516828,
+      "loss": 1.0135,
+      "step": 408
+    },
+    {
+      "epoch": 0.14875283446712018,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.00019243666625871986,
+      "loss": 0.9483,
+      "step": 410
+    },
+    {
+      "epoch": 0.14947845804988663,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.00019238771264227147,
+      "loss": 1.011,
+      "step": 412
+    },
+    {
+      "epoch": 0.15020408163265306,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.00019233875902582303,
+      "loss": 1.0434,
+      "step": 414
+    },
+    {
+      "epoch": 0.1509297052154195,
+      "grad_norm": 0.185546875,
+      "learning_rate": 0.0001922898054093746,
+      "loss": 1.0594,
+      "step": 416
+    },
+    {
+      "epoch": 0.15165532879818594,
+      "grad_norm": 0.18359375,
+      "learning_rate": 0.00019224085179292622,
+      "loss": 0.998,
+      "step": 418
+    },
+    {
+      "epoch": 0.1523809523809524,
+      "grad_norm": 0.173828125,
+      "learning_rate": 0.0001921918981764778,
+      "loss": 1.0772,
+      "step": 420
+    },
+    {
+      "epoch": 0.15310657596371882,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.00019214294456002939,
+      "loss": 0.9089,
+      "step": 422
+    },
+    {
+      "epoch": 0.15383219954648525,
+      "grad_norm": 0.203125,
+      "learning_rate": 0.00019209399094358097,
+      "loss": 1.0361,
+      "step": 424
+    },
+    {
+      "epoch": 0.1545578231292517,
+      "grad_norm": 0.1865234375,
+      "learning_rate": 0.00019204503732713255,
+      "loss": 1.0325,
+      "step": 426
+    },
+    {
+      "epoch": 0.15528344671201813,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.00019199608371068413,
+      "loss": 1.0237,
+      "step": 428
+    },
+    {
+      "epoch": 0.15600907029478459,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.00019194713009423571,
+      "loss": 0.9513,
+      "step": 430
+    },
+    {
+      "epoch": 0.156734693877551,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.0001918981764777873,
+      "loss": 0.9614,
+      "step": 432
+    },
+    {
+      "epoch": 0.15746031746031747,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.00019184922286133888,
+      "loss": 0.9507,
+      "step": 434
+    },
+    {
+      "epoch": 0.1581859410430839,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.00019180026924489046,
+      "loss": 0.9842,
+      "step": 436
+    },
+    {
+      "epoch": 0.15891156462585035,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.00019175131562844207,
+      "loss": 1.0938,
+      "step": 438
+    },
+    {
+      "epoch": 0.15963718820861678,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.00019170236201199365,
+      "loss": 0.9905,
+      "step": 440
+    },
+    {
+      "epoch": 0.16036281179138323,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.0001916534083955452,
+      "loss": 1.049,
+      "step": 442
+    },
+    {
+      "epoch": 0.16108843537414966,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.00019160445477909682,
+      "loss": 0.9727,
+      "step": 444
+    },
+    {
+      "epoch": 0.1618140589569161,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.0001915555011626484,
+      "loss": 1.0255,
+      "step": 446
+    },
+    {
+      "epoch": 0.16253968253968254,
+      "grad_norm": 0.2197265625,
+      "learning_rate": 0.00019150654754619998,
+      "loss": 0.9454,
+      "step": 448
+    },
+    {
+      "epoch": 0.16326530612244897,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.00019145759392975156,
+      "loss": 1.0037,
+      "step": 450
+    },
+    {
+      "epoch": 0.16399092970521542,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.00019140864031330315,
+      "loss": 0.9791,
+      "step": 452
+    },
+    {
+      "epoch": 0.16471655328798185,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.00019135968669685473,
+      "loss": 1.0033,
+      "step": 454
+    },
+    {
+      "epoch": 0.1654421768707483,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.0001913107330804063,
+      "loss": 1.0124,
+      "step": 456
+    },
+    {
+      "epoch": 0.16616780045351473,
+      "grad_norm": 0.173828125,
+      "learning_rate": 0.00019126177946395792,
+      "loss": 1.0341,
+      "step": 458
+    },
+    {
+      "epoch": 0.16689342403628118,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.0001912128258475095,
+      "loss": 0.999,
+      "step": 460
+    },
+    {
+      "epoch": 0.1676190476190476,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.00019116387223106106,
+      "loss": 1.0343,
+      "step": 462
+    },
+    {
+      "epoch": 0.16834467120181407,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.00019111491861461267,
+      "loss": 1.1115,
+      "step": 464
+    },
+    {
+      "epoch": 0.1690702947845805,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.00019106596499816425,
+      "loss": 1.0324,
+      "step": 466
+    },
+    {
+      "epoch": 0.16979591836734695,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.00019101701138171583,
+      "loss": 1.0122,
+      "step": 468
+    },
+    {
+      "epoch": 0.17052154195011338,
+      "grad_norm": 0.18359375,
+      "learning_rate": 0.00019096805776526741,
+      "loss": 1.075,
+      "step": 470
+    },
+    {
+      "epoch": 0.17124716553287983,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.000190919104148819,
+      "loss": 1.117,
+      "step": 472
+    },
+    {
+      "epoch": 0.17197278911564626,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 0.00019087015053237058,
+      "loss": 0.9855,
+      "step": 474
+    },
+    {
+      "epoch": 0.1726984126984127,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.00019082119691592216,
+      "loss": 0.9953,
+      "step": 476
+    },
+    {
+      "epoch": 0.17342403628117914,
+      "grad_norm": 0.1953125,
+      "learning_rate": 0.00019077224329947377,
+      "loss": 1.0023,
+      "step": 478
+    },
+    {
+      "epoch": 0.17414965986394557,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.00019072328968302535,
+      "loss": 1.0291,
+      "step": 480
+    },
+    {
+      "epoch": 0.17487528344671202,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.0001906743360665769,
+      "loss": 0.9518,
+      "step": 482
+    },
+    {
+      "epoch": 0.17560090702947845,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.00019062538245012852,
+      "loss": 1.0413,
+      "step": 484
+    },
+    {
+      "epoch": 0.1763265306122449,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.0001905764288336801,
+      "loss": 1.0636,
+      "step": 486
+    },
+    {
+      "epoch": 0.17705215419501133,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.00019052747521723168,
+      "loss": 1.041,
+      "step": 488
+    },
+    {
+      "epoch": 0.17777777777777778,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.00019047852160078326,
+      "loss": 1.0237,
+      "step": 490
+    },
+    {
+      "epoch": 0.1785034013605442,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.00019042956798433485,
+      "loss": 0.9516,
+      "step": 492
+    },
+    {
+      "epoch": 0.17922902494331067,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.00019038061436788643,
+      "loss": 1.0947,
+      "step": 494
+    },
+    {
+      "epoch": 0.1799546485260771,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.000190331660751438,
+      "loss": 1.1305,
+      "step": 496
+    },
+    {
+      "epoch": 0.18068027210884355,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.00019028270713498962,
+      "loss": 1.0146,
+      "step": 498
+    },
+    {
+      "epoch": 0.18140589569160998,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.0001902337535185412,
+      "loss": 0.9402,
+      "step": 500
+    },
+    {
+      "epoch": 0.18140589569160998,
+      "eval_loss": 1.030946969985962,
+      "eval_runtime": 94.9838,
+      "eval_samples_per_second": 18.951,
+      "eval_steps_per_second": 0.6,
+      "step": 500
+    },
+    {
+      "epoch": 0.18213151927437643,
+      "grad_norm": 0.177734375,
+      "learning_rate": 0.00019018479990209276,
+      "loss": 0.9476,
+      "step": 502
+    },
+    {
+      "epoch": 0.18285714285714286,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.00019013584628564437,
+      "loss": 1.1158,
+      "step": 504
+    },
+    {
+      "epoch": 0.18358276643990928,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.00019008689266919595,
+      "loss": 1.0539,
+      "step": 506
+    },
+    {
+      "epoch": 0.18430839002267574,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.00019003793905274753,
+      "loss": 1.0797,
+      "step": 508
+    },
+    {
+      "epoch": 0.18503401360544217,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.00018998898543629911,
+      "loss": 0.9487,
+      "step": 510
+    },
+    {
+      "epoch": 0.18575963718820862,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.0001899400318198507,
+      "loss": 1.0215,
+      "step": 512
+    },
+    {
+      "epoch": 0.18648526077097505,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.00018989107820340228,
+      "loss": 1.0063,
+      "step": 514
+    },
+    {
+      "epoch": 0.1872108843537415,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.00018984212458695386,
+      "loss": 1.108,
+      "step": 516
+    },
+    {
+      "epoch": 0.18793650793650793,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.00018979317097050547,
+      "loss": 1.0557,
+      "step": 518
+    },
+    {
+      "epoch": 0.18866213151927438,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.00018974421735405703,
+      "loss": 0.977,
+      "step": 520
+    },
+    {
+      "epoch": 0.1893877551020408,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.0001896952637376086,
+      "loss": 1.0308,
+      "step": 522
+    },
+    {
+      "epoch": 0.19011337868480727,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.00018964631012116022,
+      "loss": 1.0537,
+      "step": 524
+    },
+    {
+      "epoch": 0.1908390022675737,
+      "grad_norm": 0.189453125,
+      "learning_rate": 0.0001895973565047118,
+      "loss": 1.0358,
+      "step": 526
+    },
+    {
+      "epoch": 0.19156462585034015,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.00018954840288826338,
+      "loss": 1.0018,
+      "step": 528
+    },
+    {
+      "epoch": 0.19229024943310657,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.00018949944927181496,
+      "loss": 1.0255,
+      "step": 530
+    },
+    {
+      "epoch": 0.19301587301587303,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.00018945049565536655,
+      "loss": 0.9378,
+      "step": 532
+    },
+    {
+      "epoch": 0.19374149659863946,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.00018940154203891813,
+      "loss": 1.0373,
+      "step": 534
+    },
+    {
+      "epoch": 0.19446712018140588,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.0001893525884224697,
+      "loss": 1.0026,
+      "step": 536
+    },
+    {
+      "epoch": 0.19519274376417234,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 0.00018930363480602132,
+      "loss": 1.0897,
+      "step": 538
+    },
+    {
+      "epoch": 0.19591836734693877,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.00018925468118957288,
+      "loss": 0.9296,
+      "step": 540
+    },
+    {
+      "epoch": 0.19664399092970522,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.00018920572757312446,
+      "loss": 1.0485,
+      "step": 542
+    },
+    {
+      "epoch": 0.19736961451247165,
+      "grad_norm": 0.1875,
+      "learning_rate": 0.00018915677395667607,
+      "loss": 0.9125,
+      "step": 544
+    },
+    {
+      "epoch": 0.1980952380952381,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.00018910782034022765,
+      "loss": 1.0227,
+      "step": 546
+    },
+    {
+      "epoch": 0.19882086167800453,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.00018905886672377923,
+      "loss": 0.9335,
+      "step": 548
+    },
+    {
+      "epoch": 0.19954648526077098,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.00018900991310733081,
+      "loss": 1.0549,
+      "step": 550
+    },
+    {
+      "epoch": 0.2002721088435374,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 0.0001889609594908824,
+      "loss": 0.9836,
+      "step": 552
+    },
+    {
+      "epoch": 0.20099773242630387,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.00018891200587443398,
+      "loss": 1.0058,
+      "step": 554
+    },
+    {
+      "epoch": 0.2017233560090703,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.00018886305225798556,
+      "loss": 1.0077,
+      "step": 556
+    },
+    {
+      "epoch": 0.20244897959183675,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.00018881409864153717,
+      "loss": 0.9855,
+      "step": 558
+    },
+    {
+      "epoch": 0.20317460317460317,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.00018876514502508873,
+      "loss": 0.9711,
+      "step": 560
+    },
+    {
+      "epoch": 0.2039002267573696,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.0001887161914086403,
+      "loss": 0.9793,
+      "step": 562
+    },
+    {
+      "epoch": 0.20462585034013606,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.00018866723779219192,
+      "loss": 1.0408,
+      "step": 564
+    },
+    {
+      "epoch": 0.20535147392290248,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.0001886182841757435,
+      "loss": 1.0226,
+      "step": 566
+    },
+    {
+      "epoch": 0.20607709750566894,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.00018856933055929508,
+      "loss": 0.9565,
+      "step": 568
+    },
+    {
+      "epoch": 0.20680272108843537,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.00018852037694284666,
+      "loss": 1.0553,
+      "step": 570
+    },
+    {
+      "epoch": 0.20752834467120182,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.00018847142332639825,
+      "loss": 1.0416,
+      "step": 572
+    },
+    {
+      "epoch": 0.20825396825396825,
+      "grad_norm": 0.181640625,
+      "learning_rate": 0.00018842246970994983,
+      "loss": 1.0078,
+      "step": 574
+    },
+    {
+      "epoch": 0.2089795918367347,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.0001883735160935014,
+      "loss": 0.9935,
+      "step": 576
+    },
+    {
+      "epoch": 0.20970521541950113,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.000188324562477053,
+      "loss": 1.0732,
+      "step": 578
+    },
+    {
+      "epoch": 0.21043083900226758,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.00018827560886060458,
+      "loss": 1.013,
+      "step": 580
+    },
+    {
+      "epoch": 0.211156462585034,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.00018822665524415616,
+      "loss": 1.0011,
+      "step": 582
+    },
+    {
+      "epoch": 0.21188208616780047,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.00018817770162770777,
+      "loss": 1.0494,
+      "step": 584
+    },
+    {
+      "epoch": 0.2126077097505669,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.00018812874801125935,
+      "loss": 0.9439,
+      "step": 586
+    },
+    {
+      "epoch": 0.21333333333333335,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.0001880797943948109,
+      "loss": 1.0595,
+      "step": 588
+    },
+    {
+      "epoch": 0.21405895691609977,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.00018803084077836251,
+      "loss": 0.9564,
+      "step": 590
+    },
+    {
+      "epoch": 0.2147845804988662,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.0001879818871619141,
+      "loss": 1.1101,
+      "step": 592
+    },
+    {
+      "epoch": 0.21551020408163266,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.00018793293354546568,
+      "loss": 0.9792,
+      "step": 594
+    },
+    {
+      "epoch": 0.21623582766439908,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.00018788397992901726,
+      "loss": 0.9356,
+      "step": 596
+    },
+    {
+      "epoch": 0.21696145124716554,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.00018783502631256884,
+      "loss": 0.9553,
+      "step": 598
+    },
+    {
+      "epoch": 0.21768707482993196,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.00018778607269612043,
+      "loss": 0.9454,
+      "step": 600
+    },
+    {
+      "epoch": 0.21768707482993196,
+      "eval_loss": 1.0292859077453613,
+      "eval_runtime": 95.3967,
+      "eval_samples_per_second": 18.869,
+      "eval_steps_per_second": 0.598,
+      "step": 600
+    },
+    {
+      "epoch": 0.21841269841269842,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.000187737119079672,
+      "loss": 1.0517,
+      "step": 602
+    },
+    {
+      "epoch": 0.21913832199546485,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.00018768816546322362,
+      "loss": 1.0417,
+      "step": 604
+    },
+    {
+      "epoch": 0.2198639455782313,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.0001876392118467752,
+      "loss": 0.9824,
+      "step": 606
+    },
+    {
+      "epoch": 0.22058956916099773,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.00018759025823032675,
+      "loss": 1.0043,
+      "step": 608
+    },
+    {
+      "epoch": 0.22131519274376418,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.00018754130461387836,
+      "loss": 0.94,
+      "step": 610
+    },
+    {
+      "epoch": 0.2220408163265306,
+      "grad_norm": 0.185546875,
+      "learning_rate": 0.00018749235099742995,
+      "loss": 0.9731,
+      "step": 612
+    },
+    {
+      "epoch": 0.22276643990929706,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.00018744339738098153,
+      "loss": 1.0113,
+      "step": 614
+    },
+    {
+      "epoch": 0.2234920634920635,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.0001873944437645331,
+      "loss": 1.0831,
+      "step": 616
+    },
+    {
+      "epoch": 0.22421768707482992,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.0001873454901480847,
+      "loss": 1.0206,
+      "step": 618
+    },
+    {
+      "epoch": 0.22494331065759637,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.00018729653653163628,
+      "loss": 1.0225,
+      "step": 620
+    },
+    {
+      "epoch": 0.2256689342403628,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.00018724758291518786,
+      "loss": 0.9866,
+      "step": 622
+    },
+    {
+      "epoch": 0.22639455782312926,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.00018719862929873947,
+      "loss": 1.0209,
+      "step": 624
+    },
+    {
+      "epoch": 0.22712018140589568,
+      "grad_norm": 0.173828125,
+      "learning_rate": 0.00018714967568229105,
+      "loss": 0.9636,
+      "step": 626
+    },
+    {
+      "epoch": 0.22784580498866214,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.0001871007220658426,
+      "loss": 0.9683,
+      "step": 628
+    },
+    {
+      "epoch": 0.22857142857142856,
+      "grad_norm": 0.2099609375,
+      "learning_rate": 0.00018705176844939421,
+      "loss": 0.9994,
+      "step": 630
+    },
+    {
+      "epoch": 0.22929705215419502,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.0001870028148329458,
+      "loss": 1.0061,
+      "step": 632
+    },
+    {
+      "epoch": 0.23002267573696145,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.00018695386121649738,
+      "loss": 0.9647,
+      "step": 634
+    },
+    {
+      "epoch": 0.2307482993197279,
+      "grad_norm": 0.181640625,
+      "learning_rate": 0.00018690490760004896,
+      "loss": 1.0194,
+      "step": 636
+    },
+    {
+      "epoch": 0.23147392290249433,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.00018685595398360054,
+      "loss": 1.0266,
+      "step": 638
+    },
+    {
+      "epoch": 0.23219954648526078,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.00018680700036715213,
+      "loss": 1.011,
+      "step": 640
+    },
+    {
+      "epoch": 0.2329251700680272,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.0001867580467507037,
+      "loss": 0.9558,
+      "step": 642
+    },
+    {
+      "epoch": 0.23365079365079366,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.00018670909313425532,
+      "loss": 1.0053,
+      "step": 644
+    },
+    {
+      "epoch": 0.2343764172335601,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.0001866601395178069,
+      "loss": 1.054,
+      "step": 646
+    },
+    {
+      "epoch": 0.23510204081632652,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.00018661118590135845,
+      "loss": 0.9628,
+      "step": 648
+    },
+    {
+      "epoch": 0.23582766439909297,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.00018656223228491006,
+      "loss": 1.0158,
+      "step": 650
+    },
+    {
+      "epoch": 0.2365532879818594,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.00018651327866846165,
+      "loss": 0.9922,
+      "step": 652
+    },
+    {
+      "epoch": 0.23727891156462586,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.00018646432505201323,
+      "loss": 0.9373,
+      "step": 654
+    },
+    {
+      "epoch": 0.23800453514739228,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.0001864153714355648,
+      "loss": 0.993,
+      "step": 656
+    },
+    {
+      "epoch": 0.23873015873015874,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.0001863664178191164,
+      "loss": 1.0862,
+      "step": 658
+    },
+    {
+      "epoch": 0.23945578231292516,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.00018631746420266798,
+      "loss": 0.9666,
+      "step": 660
+    },
+    {
+      "epoch": 0.24018140589569162,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.00018626851058621956,
+      "loss": 0.9805,
+      "step": 662
+    },
+    {
+      "epoch": 0.24090702947845805,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.00018621955696977117,
+      "loss": 0.9948,
+      "step": 664
+    },
+    {
+      "epoch": 0.2416326530612245,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.00018617060335332272,
+      "loss": 1.0119,
+      "step": 666
+    },
+    {
+      "epoch": 0.24235827664399093,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.0001861216497368743,
+      "loss": 1.041,
+      "step": 668
+    },
+    {
+      "epoch": 0.24308390022675738,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.00018607269612042591,
+      "loss": 0.9779,
+      "step": 670
+    },
+    {
+      "epoch": 0.2438095238095238,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.0001860237425039775,
+      "loss": 1.0314,
+      "step": 672
+    },
+    {
+      "epoch": 0.24453514739229024,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.00018597478888752908,
+      "loss": 1.0456,
+      "step": 674
+    },
+    {
+      "epoch": 0.2452607709750567,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.00018592583527108066,
+      "loss": 1.0908,
+      "step": 676
+    },
+    {
+      "epoch": 0.24598639455782312,
+      "grad_norm": 0.2216796875,
+      "learning_rate": 0.00018587688165463224,
+      "loss": 0.9864,
+      "step": 678
+    },
+    {
+      "epoch": 0.24671201814058957,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.00018582792803818383,
+      "loss": 1.0069,
+      "step": 680
+    },
+    {
+      "epoch": 0.247437641723356,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.0001857789744217354,
+      "loss": 1.0085,
+      "step": 682
+    },
+    {
+      "epoch": 0.24816326530612245,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.00018573002080528702,
+      "loss": 0.9846,
+      "step": 684
+    },
+    {
+      "epoch": 0.24888888888888888,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 0.00018568106718883857,
+      "loss": 1.0301,
+      "step": 686
+    },
+    {
+      "epoch": 0.24961451247165534,
+      "grad_norm": 0.1865234375,
+      "learning_rate": 0.00018563211357239015,
+      "loss": 1.031,
+      "step": 688
+    },
+    {
+      "epoch": 0.2503401360544218,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.00018558315995594176,
+      "loss": 1.043,
+      "step": 690
+    },
+    {
+      "epoch": 0.2510657596371882,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.00018553420633949335,
+      "loss": 0.9125,
+      "step": 692
+    },
+    {
+      "epoch": 0.25179138321995465,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.00018548525272304493,
+      "loss": 1.0135,
+      "step": 694
+    },
+    {
+      "epoch": 0.2525170068027211,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.0001854362991065965,
+      "loss": 0.9936,
+      "step": 696
+    },
+    {
+      "epoch": 0.25324263038548755,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.0001853873454901481,
+      "loss": 0.9794,
+      "step": 698
+    },
+    {
+      "epoch": 0.25396825396825395,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.00018533839187369968,
+      "loss": 1.0237,
+      "step": 700
+    },
+    {
+      "epoch": 0.25396825396825395,
+      "eval_loss": 1.0280758142471313,
+      "eval_runtime": 92.7311,
+      "eval_samples_per_second": 19.411,
+      "eval_steps_per_second": 0.615,
+      "step": 700
+    },
+    {
+      "epoch": 0.2546938775510204,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.00018528943825725126,
+      "loss": 1.069,
+      "step": 702
+    },
+    {
+      "epoch": 0.25541950113378686,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.00018524048464080287,
+      "loss": 0.902,
+      "step": 704
+    },
+    {
+      "epoch": 0.25614512471655326,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.00018519153102435442,
+      "loss": 1.0793,
+      "step": 706
+    },
+    {
+      "epoch": 0.2568707482993197,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.000185142577407906,
+      "loss": 1.0871,
+      "step": 708
+    },
+    {
+      "epoch": 0.2575963718820862,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.0001850936237914576,
+      "loss": 0.9444,
+      "step": 710
+    },
+    {
+      "epoch": 0.2583219954648526,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.0001850446701750092,
+      "loss": 1.045,
+      "step": 712
+    },
+    {
+      "epoch": 0.259047619047619,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.00018499571655856078,
+      "loss": 1.0274,
+      "step": 714
+    },
+    {
+      "epoch": 0.2597732426303855,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.00018494676294211236,
+      "loss": 1.0513,
+      "step": 716
+    },
+    {
+      "epoch": 0.26049886621315194,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.00018489780932566394,
+      "loss": 0.9794,
+      "step": 718
+    },
+    {
+      "epoch": 0.2612244897959184,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.00018484885570921552,
+      "loss": 0.9927,
+      "step": 720
+    },
+    {
+      "epoch": 0.2619501133786848,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.0001847999020927671,
+      "loss": 0.9424,
+      "step": 722
+    },
+    {
+      "epoch": 0.26267573696145124,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.0001847509484763187,
+      "loss": 0.9492,
+      "step": 724
+    },
+    {
+      "epoch": 0.2634013605442177,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.00018470199485987027,
+      "loss": 1.0054,
+      "step": 726
+    },
+    {
+      "epoch": 0.26412698412698415,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.00018465304124342185,
+      "loss": 1.0401,
+      "step": 728
+    },
+    {
+      "epoch": 0.26485260770975055,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.00018460408762697346,
+      "loss": 0.9785,
+      "step": 730
+    },
+    {
+      "epoch": 0.265578231292517,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.00018455513401052505,
+      "loss": 1.0867,
+      "step": 732
+    },
+    {
+      "epoch": 0.26630385487528346,
+      "grad_norm": 0.2001953125,
+      "learning_rate": 0.0001845061803940766,
+      "loss": 0.9375,
+      "step": 734
+    },
+    {
+      "epoch": 0.26702947845804986,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.0001844572267776282,
+      "loss": 0.9576,
+      "step": 736
+    },
+    {
+      "epoch": 0.2677551020408163,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.0001844082731611798,
+      "loss": 1.0314,
+      "step": 738
+    },
+    {
+      "epoch": 0.26848072562358277,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.00018435931954473137,
+      "loss": 0.9826,
+      "step": 740
+    },
+    {
+      "epoch": 0.2692063492063492,
+      "grad_norm": 0.201171875,
+      "learning_rate": 0.00018431036592828296,
+      "loss": 1.0339,
+      "step": 742
+    },
+    {
+      "epoch": 0.2699319727891156,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.00018426141231183454,
+      "loss": 1.0327,
+      "step": 744
+    },
+    {
+      "epoch": 0.2706575963718821,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.00018421245869538612,
+      "loss": 1.029,
+      "step": 746
+    },
+    {
+      "epoch": 0.27138321995464854,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.0001841635050789377,
+      "loss": 1.0364,
+      "step": 748
+    },
+    {
+      "epoch": 0.272108843537415,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.0001841145514624893,
+      "loss": 0.925,
+      "step": 750
+    },
+    {
+      "epoch": 0.2728344671201814,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.0001840655978460409,
+      "loss": 0.9998,
+      "step": 752
+    },
+    {
+      "epoch": 0.27356009070294784,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 0.00018401664422959245,
+      "loss": 1.0414,
+      "step": 754
+    },
+    {
+      "epoch": 0.2742857142857143,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.00018396769061314406,
+      "loss": 1.0585,
+      "step": 756
+    },
+    {
+      "epoch": 0.2750113378684807,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.00018391873699669564,
+      "loss": 0.945,
+      "step": 758
+    },
+    {
+      "epoch": 0.27573696145124715,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 0.00018386978338024722,
+      "loss": 1.0171,
+      "step": 760
+    },
+    {
+      "epoch": 0.2764625850340136,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.0001838208297637988,
+      "loss": 1.0176,
+      "step": 762
+    },
+    {
+      "epoch": 0.27718820861678006,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.0001837718761473504,
+      "loss": 0.9225,
+      "step": 764
+    },
+    {
+      "epoch": 0.27791383219954646,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.00018372292253090197,
+      "loss": 1.0185,
+      "step": 766
+    },
+    {
+      "epoch": 0.2786394557823129,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.00018367396891445355,
+      "loss": 1.0516,
+      "step": 768
+    },
+    {
+      "epoch": 0.27936507936507937,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.00018362501529800516,
+      "loss": 0.972,
+      "step": 770
+    },
+    {
+      "epoch": 0.2800907029478458,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.00018357606168155675,
+      "loss": 1.0903,
+      "step": 772
+    },
+    {
+      "epoch": 0.2808163265306122,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.0001835271080651083,
+      "loss": 1.0122,
+      "step": 774
+    },
+    {
+      "epoch": 0.2815419501133787,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.0001834781544486599,
+      "loss": 1.0724,
+      "step": 776
+    },
+    {
+      "epoch": 0.28226757369614514,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.0001834292008322115,
+      "loss": 1.02,
+      "step": 778
+    },
+    {
+      "epoch": 0.2829931972789116,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.00018338024721576307,
+      "loss": 0.9654,
+      "step": 780
+    },
+    {
+      "epoch": 0.283718820861678,
+      "grad_norm": 0.173828125,
+      "learning_rate": 0.00018333129359931466,
+      "loss": 1.0622,
+      "step": 782
+    },
+    {
+      "epoch": 0.28444444444444444,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.00018328233998286624,
+      "loss": 0.9866,
+      "step": 784
+    },
+    {
+      "epoch": 0.2851700680272109,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.00018323338636641782,
+      "loss": 1.048,
+      "step": 786
+    },
+    {
+      "epoch": 0.2858956916099773,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.0001831844327499694,
+      "loss": 0.9802,
+      "step": 788
+    },
+    {
+      "epoch": 0.28662131519274375,
+      "grad_norm": 0.17578125,
+      "learning_rate": 0.000183135479133521,
+      "loss": 0.9756,
+      "step": 790
+    },
+    {
+      "epoch": 0.2873469387755102,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.0001830865255170726,
+      "loss": 1.0143,
+      "step": 792
+    },
+    {
+      "epoch": 0.28807256235827666,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.00018303757190062415,
+      "loss": 1.0164,
+      "step": 794
+    },
+    {
+      "epoch": 0.28879818594104306,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.00018298861828417576,
+      "loss": 0.9813,
+      "step": 796
+    },
+    {
+      "epoch": 0.2895238095238095,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.00018293966466772734,
+      "loss": 1.0568,
+      "step": 798
+    },
+    {
+      "epoch": 0.29024943310657597,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.00018289071105127892,
+      "loss": 0.9557,
+      "step": 800
+    },
+    {
+      "epoch": 0.29024943310657597,
+      "eval_loss": 1.0269455909729004,
+      "eval_runtime": 94.2392,
+      "eval_samples_per_second": 19.1,
+      "eval_steps_per_second": 0.605,
+      "step": 800
+    },
+    {
+      "epoch": 0.2909750566893424,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.0001828417574348305,
+      "loss": 1.0491,
+      "step": 802
+    },
+    {
+      "epoch": 0.2917006802721088,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.0001827928038183821,
+      "loss": 0.9705,
+      "step": 804
+    },
+    {
+      "epoch": 0.2924263038548753,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.00018274385020193367,
+      "loss": 1.0566,
+      "step": 806
+    },
+    {
+      "epoch": 0.29315192743764173,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.00018269489658548525,
+      "loss": 0.9782,
+      "step": 808
+    },
+    {
+      "epoch": 0.2938775510204082,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.00018264594296903686,
+      "loss": 0.9967,
+      "step": 810
+    },
+    {
+      "epoch": 0.2946031746031746,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.00018259698935258842,
+      "loss": 1.0371,
+      "step": 812
+    },
+    {
+      "epoch": 0.29532879818594104,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.00018254803573614,
+      "loss": 0.9803,
+      "step": 814
+    },
+    {
+      "epoch": 0.2960544217687075,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.0001824990821196916,
+      "loss": 0.9969,
+      "step": 816
+    },
+    {
+      "epoch": 0.2967800453514739,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.0001824501285032432,
+      "loss": 1.0483,
+      "step": 818
+    },
+    {
+      "epoch": 0.29750566893424035,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.00018240117488679477,
+      "loss": 1.0164,
+      "step": 820
+    },
+    {
+      "epoch": 0.2982312925170068,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.00018235222127034636,
+      "loss": 0.9407,
+      "step": 822
+    },
+    {
+      "epoch": 0.29895691609977326,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.00018230326765389794,
+      "loss": 1.0397,
+      "step": 824
+    },
+    {
+      "epoch": 0.29968253968253966,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.00018225431403744952,
+      "loss": 0.9518,
+      "step": 826
+    },
+    {
+      "epoch": 0.3004081632653061,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.0001822053604210011,
+      "loss": 1.0551,
+      "step": 828
+    },
+    {
+      "epoch": 0.30113378684807257,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.0001821564068045527,
+      "loss": 0.9826,
+      "step": 830
+    },
+    {
+      "epoch": 0.301859410430839,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.00018210745318810427,
+      "loss": 0.945,
+      "step": 832
+    },
+    {
+      "epoch": 0.3025850340136054,
+      "grad_norm": 0.1875,
+      "learning_rate": 0.00018205849957165585,
+      "loss": 0.9464,
+      "step": 834
+    },
+    {
+      "epoch": 0.3033106575963719,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.00018200954595520746,
+      "loss": 0.982,
+      "step": 836
+    },
+    {
+      "epoch": 0.30403628117913833,
+      "grad_norm": 0.173828125,
+      "learning_rate": 0.00018196059233875904,
+      "loss": 1.0529,
+      "step": 838
+    },
+    {
+      "epoch": 0.3047619047619048,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.00018191163872231062,
+      "loss": 0.9982,
+      "step": 840
+    },
+    {
+      "epoch": 0.3054875283446712,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.0001818626851058622,
+      "loss": 0.946,
+      "step": 842
+    },
+    {
+      "epoch": 0.30621315192743764,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.0001818137314894138,
+      "loss": 1.0053,
+      "step": 844
+    },
+    {
+      "epoch": 0.3069387755102041,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.00018176477787296537,
+      "loss": 0.9307,
+      "step": 846
+    },
+    {
+      "epoch": 0.3076643990929705,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.00018171582425651695,
+      "loss": 0.9697,
+      "step": 848
+    },
+    {
+      "epoch": 0.30839002267573695,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.00018166687064006856,
+      "loss": 0.9617,
+      "step": 850
+    },
+    {
+      "epoch": 0.3091156462585034,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.00018161791702362012,
+      "loss": 0.9534,
+      "step": 852
+    },
+    {
+      "epoch": 0.30984126984126986,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.0001815689634071717,
+      "loss": 1.0296,
+      "step": 854
+    },
+    {
+      "epoch": 0.31056689342403626,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.0001815200097907233,
+      "loss": 1.0481,
+      "step": 856
+    },
+    {
+      "epoch": 0.3112925170068027,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.0001814710561742749,
+      "loss": 1.0118,
+      "step": 858
+    },
+    {
+      "epoch": 0.31201814058956917,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.00018142210255782647,
+      "loss": 1.0305,
+      "step": 860
+    },
+    {
+      "epoch": 0.3127437641723356,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.00018137314894137806,
+      "loss": 1.0597,
+      "step": 862
+    },
+    {
+      "epoch": 0.313469387755102,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.00018132419532492964,
+      "loss": 1.0554,
+      "step": 864
+    },
+    {
+      "epoch": 0.3141950113378685,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.00018127524170848122,
+      "loss": 0.9902,
+      "step": 866
+    },
+    {
+      "epoch": 0.31492063492063493,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.0001812262880920328,
+      "loss": 0.9705,
+      "step": 868
+    },
+    {
+      "epoch": 0.31564625850340133,
+      "grad_norm": 0.173828125,
+      "learning_rate": 0.00018117733447558439,
+      "loss": 1.0712,
+      "step": 870
+    },
+    {
+      "epoch": 0.3163718820861678,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.00018112838085913597,
+      "loss": 1.0368,
+      "step": 872
+    },
+    {
+      "epoch": 0.31709750566893424,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.00018107942724268755,
+      "loss": 0.9667,
+      "step": 874
+    },
+    {
+      "epoch": 0.3178231292517007,
+      "grad_norm": 0.17578125,
+      "learning_rate": 0.00018103047362623916,
+      "loss": 0.9888,
+      "step": 876
+    },
+    {
+      "epoch": 0.3185487528344671,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.00018098152000979074,
+      "loss": 0.9979,
+      "step": 878
+    },
+    {
+      "epoch": 0.31927437641723355,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.0001809325663933423,
+      "loss": 0.9947,
+      "step": 880
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.0001808836127768939,
+      "loss": 1.0054,
+      "step": 882
+    },
+    {
+      "epoch": 0.32072562358276646,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.0001808346591604455,
+      "loss": 1.0014,
+      "step": 884
+    },
+    {
+      "epoch": 0.32145124716553286,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 0.00018078570554399707,
+      "loss": 0.9482,
+      "step": 886
+    },
+    {
+      "epoch": 0.3221768707482993,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.00018073675192754865,
+      "loss": 1.0337,
+      "step": 888
+    },
+    {
+      "epoch": 0.32290249433106577,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.00018068779831110024,
+      "loss": 0.9075,
+      "step": 890
+    },
+    {
+      "epoch": 0.3236281179138322,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.00018063884469465182,
+      "loss": 0.9731,
+      "step": 892
+    },
+    {
+      "epoch": 0.3243537414965986,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.0001805898910782034,
+      "loss": 1.0288,
+      "step": 894
+    },
+    {
+      "epoch": 0.3250793650793651,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.000180540937461755,
+      "loss": 0.9984,
+      "step": 896
+    },
+    {
+      "epoch": 0.32580498866213153,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.0001804919838453066,
+      "loss": 0.9527,
+      "step": 898
+    },
+    {
+      "epoch": 0.32653061224489793,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.00018044303022885815,
+      "loss": 0.9399,
+      "step": 900
+    },
+    {
+      "epoch": 0.32653061224489793,
+      "eval_loss": 1.0258883237838745,
+      "eval_runtime": 95.9536,
+      "eval_samples_per_second": 18.759,
+      "eval_steps_per_second": 0.594,
+      "step": 900
+    },
+    {
+      "epoch": 0.3272562358276644,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.00018039407661240976,
+      "loss": 0.9891,
+      "step": 902
+    },
+    {
+      "epoch": 0.32798185941043084,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.00018034512299596134,
+      "loss": 0.9403,
+      "step": 904
+    },
+    {
+      "epoch": 0.3287074829931973,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.00018029616937951292,
+      "loss": 1.0682,
+      "step": 906
+    },
+    {
+      "epoch": 0.3294331065759637,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.0001802472157630645,
+      "loss": 1.0294,
+      "step": 908
+    },
+    {
+      "epoch": 0.33015873015873015,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.00018019826214661609,
+      "loss": 0.9923,
+      "step": 910
+    },
+    {
+      "epoch": 0.3308843537414966,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.00018014930853016767,
+      "loss": 0.9585,
+      "step": 912
+    },
+    {
+      "epoch": 0.33160997732426306,
+      "grad_norm": 0.2041015625,
+      "learning_rate": 0.00018010035491371925,
+      "loss": 1.0097,
+      "step": 914
+    },
+    {
+      "epoch": 0.33233560090702946,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.00018005140129727086,
+      "loss": 0.9935,
+      "step": 916
+    },
+    {
+      "epoch": 0.3330612244897959,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.00018000244768082244,
+      "loss": 0.9699,
+      "step": 918
+    },
+    {
+      "epoch": 0.33378684807256237,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.000179953494064374,
+      "loss": 1.0012,
+      "step": 920
+    },
+    {
+      "epoch": 0.3345124716553288,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.0001799045404479256,
+      "loss": 0.9984,
+      "step": 922
+    },
+    {
+      "epoch": 0.3352380952380952,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.0001798555868314772,
+      "loss": 0.978,
+      "step": 924
+    },
+    {
+      "epoch": 0.3359637188208617,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.00017980663321502877,
+      "loss": 1.0188,
+      "step": 926
+    },
+    {
+      "epoch": 0.33668934240362813,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.00017975767959858035,
+      "loss": 0.9777,
+      "step": 928
+    },
+    {
+      "epoch": 0.33741496598639453,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.00017970872598213194,
+      "loss": 1.042,
+      "step": 930
+    },
+    {
+      "epoch": 0.338140589569161,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.00017965977236568352,
+      "loss": 1.0074,
+      "step": 932
+    },
+    {
+      "epoch": 0.33886621315192744,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.0001796108187492351,
+      "loss": 1.0228,
+      "step": 934
+    },
+    {
+      "epoch": 0.3395918367346939,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.0001795618651327867,
+      "loss": 1.0906,
+      "step": 936
+    },
+    {
+      "epoch": 0.3403174603174603,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.0001795129115163383,
+      "loss": 1.0117,
+      "step": 938
+    },
+    {
+      "epoch": 0.34104308390022675,
+      "grad_norm": 0.185546875,
+      "learning_rate": 0.00017946395789988985,
+      "loss": 1.0274,
+      "step": 940
+    },
+    {
+      "epoch": 0.3417687074829932,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.00017941500428344146,
+      "loss": 1.0271,
+      "step": 942
+    },
+    {
+      "epoch": 0.34249433106575966,
+      "grad_norm": 0.181640625,
+      "learning_rate": 0.00017936605066699304,
+      "loss": 1.0064,
+      "step": 944
+    },
+    {
+      "epoch": 0.34321995464852606,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.00017931709705054462,
+      "loss": 0.9696,
+      "step": 946
+    },
+    {
+      "epoch": 0.3439455782312925,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.0001792681434340962,
+      "loss": 1.0467,
+      "step": 948
+    },
+    {
+      "epoch": 0.34467120181405897,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.00017921918981764779,
+      "loss": 1.0357,
+      "step": 950
+    },
+    {
+      "epoch": 0.3453968253968254,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.00017917023620119937,
+      "loss": 0.9136,
+      "step": 952
+    },
+    {
+      "epoch": 0.3461224489795918,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.00017912128258475095,
+      "loss": 1.036,
+      "step": 954
+    },
+    {
+      "epoch": 0.3468480725623583,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.00017907232896830256,
+      "loss": 0.9946,
+      "step": 956
+    },
+    {
+      "epoch": 0.34757369614512473,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.00017902337535185411,
+      "loss": 0.9912,
+      "step": 958
+    },
+    {
+      "epoch": 0.34829931972789113,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.0001789744217354057,
+      "loss": 0.9606,
+      "step": 960
+    },
+    {
+      "epoch": 0.3490249433106576,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.0001789254681189573,
+      "loss": 0.9763,
+      "step": 962
+    },
+    {
+      "epoch": 0.34975056689342404,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.0001788765145025089,
+      "loss": 0.9389,
+      "step": 964
+    },
+    {
+      "epoch": 0.3504761904761905,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.00017882756088606047,
+      "loss": 0.9526,
+      "step": 966
+    },
+    {
+      "epoch": 0.3512018140589569,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.00017877860726961205,
+      "loss": 0.9523,
+      "step": 968
+    },
+    {
+      "epoch": 0.35192743764172335,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.00017872965365316364,
+      "loss": 1.0858,
+      "step": 970
+    },
+    {
+      "epoch": 0.3526530612244898,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.00017868070003671522,
+      "loss": 0.9991,
+      "step": 972
+    },
+    {
+      "epoch": 0.35337868480725626,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.0001786317464202668,
+      "loss": 0.9704,
+      "step": 974
+    },
+    {
+      "epoch": 0.35410430839002266,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.0001785827928038184,
+      "loss": 0.9472,
+      "step": 976
+    },
+    {
+      "epoch": 0.3548299319727891,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.00017853383918736996,
+      "loss": 1.0397,
+      "step": 978
+    },
+    {
+      "epoch": 0.35555555555555557,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.00017848488557092155,
+      "loss": 1.0412,
+      "step": 980
+    },
+    {
+      "epoch": 0.35628117913832197,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.00017843593195447316,
+      "loss": 1.07,
+      "step": 982
+    },
+    {
+      "epoch": 0.3570068027210884,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.00017838697833802474,
+      "loss": 0.947,
+      "step": 984
+    },
+    {
+      "epoch": 0.3577324263038549,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.00017833802472157632,
+      "loss": 0.9768,
+      "step": 986
+    },
+    {
+      "epoch": 0.35845804988662133,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.0001782890711051279,
+      "loss": 1.0229,
+      "step": 988
+    },
+    {
+      "epoch": 0.35918367346938773,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.00017824011748867949,
+      "loss": 1.0429,
+      "step": 990
+    },
+    {
+      "epoch": 0.3599092970521542,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.00017819116387223107,
+      "loss": 0.9842,
+      "step": 992
+    },
+    {
+      "epoch": 0.36063492063492064,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.00017814221025578265,
+      "loss": 0.981,
+      "step": 994
+    },
+    {
+      "epoch": 0.3613605442176871,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.00017809325663933426,
+      "loss": 1.0028,
+      "step": 996
+    },
+    {
+      "epoch": 0.3620861678004535,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.00017804430302288581,
+      "loss": 1.061,
+      "step": 998
+    },
+    {
+      "epoch": 0.36281179138321995,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.0001779953494064374,
+      "loss": 1.0082,
+      "step": 1000
+    },
+    {
+      "epoch": 0.36281179138321995,
+      "eval_loss": 1.0250780582427979,
+      "eval_runtime": 92.3729,
+      "eval_samples_per_second": 19.486,
+      "eval_steps_per_second": 0.617,
+      "step": 1000
+    },
+    {
+      "epoch": 0.3635374149659864,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.000177946395789989,
+      "loss": 0.9744,
+      "step": 1002
+    },
+    {
+      "epoch": 0.36426303854875286,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.0001778974421735406,
+      "loss": 1.0106,
+      "step": 1004
+    },
+    {
+      "epoch": 0.36498866213151926,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.00017784848855709217,
+      "loss": 0.8815,
+      "step": 1006
+    },
+    {
+      "epoch": 0.3657142857142857,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.00017779953494064375,
+      "loss": 0.9473,
+      "step": 1008
+    },
+    {
+      "epoch": 0.36643990929705217,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.00017775058132419534,
+      "loss": 0.9883,
+      "step": 1010
+    },
+    {
+      "epoch": 0.36716553287981857,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.00017770162770774692,
+      "loss": 1.0715,
+      "step": 1012
+    },
+    {
+      "epoch": 0.367891156462585,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.0001776526740912985,
+      "loss": 0.9768,
+      "step": 1014
+    },
+    {
+      "epoch": 0.3686167800453515,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.00017760372047485008,
+      "loss": 0.9155,
+      "step": 1016
+    },
+    {
+      "epoch": 0.36934240362811793,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.00017755476685840166,
+      "loss": 1.051,
+      "step": 1018
+    },
+    {
+      "epoch": 0.37006802721088433,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.00017750581324195325,
+      "loss": 1.0361,
+      "step": 1020
+    },
+    {
+      "epoch": 0.3707936507936508,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.00017745685962550486,
+      "loss": 1.024,
+      "step": 1022
+    },
+    {
+      "epoch": 0.37151927437641724,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.00017740790600905644,
+      "loss": 1.0117,
+      "step": 1024
+    },
+    {
+      "epoch": 0.3722448979591837,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.000177358952392608,
+      "loss": 0.992,
+      "step": 1026
+    },
+    {
+      "epoch": 0.3729705215419501,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.0001773099987761596,
+      "loss": 1.0872,
+      "step": 1028
+    },
+    {
+      "epoch": 0.37369614512471655,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.00017726104515971118,
+      "loss": 0.9335,
+      "step": 1030
+    },
+    {
+      "epoch": 0.374421768707483,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.00017721209154326277,
+      "loss": 1.017,
+      "step": 1032
+    },
+    {
+      "epoch": 0.37514739229024946,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.00017716313792681435,
+      "loss": 0.9909,
+      "step": 1034
+    },
+    {
+      "epoch": 0.37587301587301586,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.00017711418431036593,
+      "loss": 1.0097,
+      "step": 1036
+    },
+    {
+      "epoch": 0.3765986394557823,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.00017706523069391751,
+      "loss": 0.9652,
+      "step": 1038
+    },
+    {
+      "epoch": 0.37732426303854877,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.0001770162770774691,
+      "loss": 0.9893,
+      "step": 1040
+    },
+    {
+      "epoch": 0.37804988662131517,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.0001769673234610207,
+      "loss": 1.0294,
+      "step": 1042
+    },
+    {
+      "epoch": 0.3787755102040816,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.0001769183698445723,
+      "loss": 1.0322,
+      "step": 1044
+    },
+    {
+      "epoch": 0.3795011337868481,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.00017686941622812384,
+      "loss": 0.908,
+      "step": 1046
+    },
+    {
+      "epoch": 0.38022675736961453,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.00017682046261167545,
+      "loss": 1.0802,
+      "step": 1048
+    },
+    {
+      "epoch": 0.38095238095238093,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.00017677150899522703,
+      "loss": 1.1756,
+      "step": 1050
+    },
+    {
+      "epoch": 0.3816780045351474,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.00017672255537877862,
+      "loss": 0.9534,
+      "step": 1052
+    },
+    {
+      "epoch": 0.38240362811791384,
+      "grad_norm": 0.177734375,
+      "learning_rate": 0.0001766736017623302,
+      "loss": 0.9512,
+      "step": 1054
+    },
+    {
+      "epoch": 0.3831292517006803,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.00017662464814588178,
+      "loss": 1.0466,
+      "step": 1056
+    },
+    {
+      "epoch": 0.3838548752834467,
+      "grad_norm": 0.173828125,
+      "learning_rate": 0.00017657569452943336,
+      "loss": 1.0233,
+      "step": 1058
+    },
+    {
+      "epoch": 0.38458049886621315,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.00017652674091298495,
+      "loss": 1.0718,
+      "step": 1060
+    },
+    {
+      "epoch": 0.3853061224489796,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.00017647778729653656,
+      "loss": 1.0447,
+      "step": 1062
+    },
+    {
+      "epoch": 0.38603174603174606,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.00017642883368008814,
+      "loss": 1.0104,
+      "step": 1064
+    },
+    {
+      "epoch": 0.38675736961451246,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.0001763798800636397,
+      "loss": 0.9774,
+      "step": 1066
+    },
+    {
+      "epoch": 0.3874829931972789,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.0001763309264471913,
+      "loss": 0.9544,
+      "step": 1068
+    },
+    {
+      "epoch": 0.38820861678004537,
+      "grad_norm": 0.1884765625,
+      "learning_rate": 0.00017628197283074288,
+      "loss": 1.0006,
+      "step": 1070
+    },
+    {
+      "epoch": 0.38893424036281177,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.00017623301921429447,
+      "loss": 0.9301,
+      "step": 1072
+    },
+    {
+      "epoch": 0.3896598639455782,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.00017618406559784605,
+      "loss": 1.0312,
+      "step": 1074
+    },
+    {
+      "epoch": 0.3903854875283447,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.00017613511198139763,
+      "loss": 0.9978,
+      "step": 1076
+    },
+    {
+      "epoch": 0.39111111111111113,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.00017608615836494921,
+      "loss": 0.9708,
+      "step": 1078
+    },
+    {
+      "epoch": 0.39183673469387753,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.0001760372047485008,
+      "loss": 0.9779,
+      "step": 1080
+    },
+    {
+      "epoch": 0.392562358276644,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.0001759882511320524,
+      "loss": 0.9583,
+      "step": 1082
+    },
+    {
+      "epoch": 0.39328798185941044,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.000175939297515604,
+      "loss": 1.0361,
+      "step": 1084
+    },
+    {
+      "epoch": 0.3940136054421769,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.00017589034389915554,
+      "loss": 1.061,
+      "step": 1086
+    },
+    {
+      "epoch": 0.3947392290249433,
+      "grad_norm": 0.177734375,
+      "learning_rate": 0.00017584139028270715,
+      "loss": 1.0351,
+      "step": 1088
+    },
+    {
+      "epoch": 0.39546485260770975,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.00017579243666625873,
+      "loss": 0.9836,
+      "step": 1090
+    },
+    {
+      "epoch": 0.3961904761904762,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.00017574348304981032,
+      "loss": 0.9366,
+      "step": 1092
+    },
+    {
+      "epoch": 0.3969160997732426,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.0001756945294333619,
+      "loss": 0.9386,
+      "step": 1094
+    },
+    {
+      "epoch": 0.39764172335600906,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.00017564557581691348,
+      "loss": 0.9872,
+      "step": 1096
+    },
+    {
+      "epoch": 0.3983673469387755,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.00017559662220046506,
+      "loss": 0.962,
+      "step": 1098
+    },
+    {
+      "epoch": 0.39909297052154197,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.00017554766858401665,
+      "loss": 1.0224,
+      "step": 1100
+    },
+    {
+      "epoch": 0.39909297052154197,
+      "eval_loss": 1.024340033531189,
+      "eval_runtime": 96.4085,
+      "eval_samples_per_second": 18.671,
+      "eval_steps_per_second": 0.591,
+      "step": 1100
+    },
+    {
+      "epoch": 0.39981859410430837,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.00017549871496756826,
+      "loss": 0.9655,
+      "step": 1102
+    },
+    {
+      "epoch": 0.4005442176870748,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.0001754497613511198,
+      "loss": 0.9732,
+      "step": 1104
+    },
+    {
+      "epoch": 0.4012698412698413,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.0001754008077346714,
+      "loss": 1.0525,
+      "step": 1106
+    },
+    {
+      "epoch": 0.40199546485260773,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.000175351854118223,
+      "loss": 0.9244,
+      "step": 1108
+    },
+    {
+      "epoch": 0.40272108843537413,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.00017530290050177458,
+      "loss": 0.9976,
+      "step": 1110
+    },
+    {
+      "epoch": 0.4034467120181406,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.00017525394688532617,
+      "loss": 0.9802,
+      "step": 1112
+    },
+    {
+      "epoch": 0.40417233560090704,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.00017520499326887775,
+      "loss": 0.9932,
+      "step": 1114
+    },
+    {
+      "epoch": 0.4048979591836735,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.00017515603965242933,
+      "loss": 1.0189,
+      "step": 1116
+    },
+    {
+      "epoch": 0.4056235827664399,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.00017510708603598091,
+      "loss": 0.908,
+      "step": 1118
+    },
+    {
+      "epoch": 0.40634920634920635,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.0001750581324195325,
+      "loss": 0.9909,
+      "step": 1120
+    },
+    {
+      "epoch": 0.4070748299319728,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.0001750091788030841,
+      "loss": 1.0264,
+      "step": 1122
+    },
+    {
+      "epoch": 0.4078004535147392,
+      "grad_norm": 0.1865234375,
+      "learning_rate": 0.00017496022518663566,
+      "loss": 1.0606,
+      "step": 1124
+    },
+    {
+      "epoch": 0.40852607709750566,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.00017491127157018724,
+      "loss": 0.9924,
+      "step": 1126
+    },
+    {
+      "epoch": 0.4092517006802721,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.00017486231795373885,
+      "loss": 0.9948,
+      "step": 1128
+    },
+    {
+      "epoch": 0.40997732426303857,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.00017481336433729043,
+      "loss": 0.9702,
+      "step": 1130
+    },
+    {
+      "epoch": 0.41070294784580497,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.00017476441072084202,
+      "loss": 1.0459,
+      "step": 1132
+    },
+    {
+      "epoch": 0.4114285714285714,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.0001747154571043936,
+      "loss": 0.9733,
+      "step": 1134
+    },
+    {
+      "epoch": 0.4121541950113379,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.00017466650348794518,
+      "loss": 0.9991,
+      "step": 1136
+    },
+    {
+      "epoch": 0.41287981859410433,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.00017461754987149676,
+      "loss": 0.9936,
+      "step": 1138
+    },
+    {
+      "epoch": 0.41360544217687073,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.00017456859625504835,
+      "loss": 0.9779,
+      "step": 1140
+    },
+    {
+      "epoch": 0.4143310657596372,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.00017451964263859996,
+      "loss": 0.996,
+      "step": 1142
+    },
+    {
+      "epoch": 0.41505668934240364,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.0001744706890221515,
+      "loss": 1.0079,
+      "step": 1144
+    },
+    {
+      "epoch": 0.4157823129251701,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.0001744217354057031,
+      "loss": 0.9805,
+      "step": 1146
+    },
+    {
+      "epoch": 0.4165079365079365,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.0001743727817892547,
+      "loss": 0.9657,
+      "step": 1148
+    },
+    {
+      "epoch": 0.41723356009070295,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.00017432382817280628,
+      "loss": 1.0011,
+      "step": 1150
+    },
+    {
+      "epoch": 0.4179591836734694,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.00017427487455635787,
+      "loss": 1.0128,
+      "step": 1152
+    },
+    {
+      "epoch": 0.4186848072562358,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.00017422592093990945,
+      "loss": 0.97,
+      "step": 1154
+    },
+    {
+      "epoch": 0.41941043083900226,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.00017417696732346103,
+      "loss": 1.0179,
+      "step": 1156
+    },
+    {
+      "epoch": 0.4201360544217687,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.0001741280137070126,
+      "loss": 1.0459,
+      "step": 1158
+    },
+    {
+      "epoch": 0.42086167800453517,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 0.0001740790600905642,
+      "loss": 1.0039,
+      "step": 1160
+    },
+    {
+      "epoch": 0.42158730158730157,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.00017403010647411578,
+      "loss": 1.0498,
+      "step": 1162
+    },
+    {
+      "epoch": 0.422312925170068,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.00017398115285766736,
+      "loss": 1.0111,
+      "step": 1164
+    },
+    {
+      "epoch": 0.4230385487528345,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.00017393219924121894,
+      "loss": 0.9945,
+      "step": 1166
+    },
+    {
+      "epoch": 0.42376417233560093,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.00017388324562477055,
+      "loss": 0.9439,
+      "step": 1168
+    },
+    {
+      "epoch": 0.42448979591836733,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.00017383429200832213,
+      "loss": 0.9019,
+      "step": 1170
+    },
+    {
+      "epoch": 0.4252154195011338,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.0001737853383918737,
+      "loss": 0.9996,
+      "step": 1172
+    },
+    {
+      "epoch": 0.42594104308390024,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.0001737363847754253,
+      "loss": 1.0023,
+      "step": 1174
+    },
+    {
+      "epoch": 0.4266666666666667,
+      "grad_norm": 0.181640625,
+      "learning_rate": 0.00017368743115897688,
+      "loss": 0.9721,
+      "step": 1176
+    },
+    {
+      "epoch": 0.4273922902494331,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.00017363847754252846,
+      "loss": 1.0203,
+      "step": 1178
+    },
+    {
+      "epoch": 0.42811791383219955,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.00017358952392608005,
+      "loss": 0.9972,
+      "step": 1180
+    },
+    {
+      "epoch": 0.428843537414966,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.00017354057030963163,
+      "loss": 0.9853,
+      "step": 1182
+    },
+    {
+      "epoch": 0.4295691609977324,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.0001734916166931832,
+      "loss": 0.9784,
+      "step": 1184
+    },
+    {
+      "epoch": 0.43029478458049886,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.0001734426630767348,
+      "loss": 1.0201,
+      "step": 1186
+    },
+    {
+      "epoch": 0.4310204081632653,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.0001733937094602864,
+      "loss": 1.0091,
+      "step": 1188
+    },
+    {
+      "epoch": 0.43174603174603177,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.00017334475584383798,
+      "loss": 1.0313,
+      "step": 1190
+    },
+    {
+      "epoch": 0.43247165532879817,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.00017329580222738954,
+      "loss": 1.0487,
+      "step": 1192
+    },
+    {
+      "epoch": 0.4331972789115646,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.00017324684861094115,
+      "loss": 1.0059,
+      "step": 1194
+    },
+    {
+      "epoch": 0.4339229024943311,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.00017319789499449273,
+      "loss": 0.9841,
+      "step": 1196
+    },
+    {
+      "epoch": 0.43464852607709753,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.0001731489413780443,
+      "loss": 0.9891,
+      "step": 1198
+    },
+    {
+      "epoch": 0.43537414965986393,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.0001730999877615959,
+      "loss": 1.0482,
+      "step": 1200
+    },
+    {
+      "epoch": 0.43537414965986393,
+      "eval_loss": 1.023897647857666,
+      "eval_runtime": 91.99,
+      "eval_samples_per_second": 19.567,
+      "eval_steps_per_second": 0.62,
+      "step": 1200
+    },
+    {
+      "epoch": 0.4360997732426304,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.00017305103414514748,
+      "loss": 0.9969,
+      "step": 1202
+    },
+    {
+      "epoch": 0.43682539682539684,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.00017300208052869906,
+      "loss": 0.9979,
+      "step": 1204
+    },
+    {
+      "epoch": 0.43755102040816324,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.00017295312691225064,
+      "loss": 0.9974,
+      "step": 1206
+    },
+    {
+      "epoch": 0.4382766439909297,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.00017290417329580225,
+      "loss": 1.0932,
+      "step": 1208
+    },
+    {
+      "epoch": 0.43900226757369615,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.00017285521967935383,
+      "loss": 0.9845,
+      "step": 1210
+    },
+    {
+      "epoch": 0.4397278911564626,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.0001728062660629054,
+      "loss": 1.0405,
+      "step": 1212
+    },
+    {
+      "epoch": 0.440453514739229,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.000172757312446457,
+      "loss": 0.9646,
+      "step": 1214
+    },
+    {
+      "epoch": 0.44117913832199546,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.00017270835883000858,
+      "loss": 0.9535,
+      "step": 1216
+    },
+    {
+      "epoch": 0.4419047619047619,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.00017265940521356016,
+      "loss": 1.0106,
+      "step": 1218
+    },
+    {
+      "epoch": 0.44263038548752837,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.00017261045159711175,
+      "loss": 0.9388,
+      "step": 1220
+    },
+    {
+      "epoch": 0.44335600907029477,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.00017256149798066333,
+      "loss": 0.8825,
+      "step": 1222
+    },
+    {
+      "epoch": 0.4440816326530612,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.0001725125443642149,
+      "loss": 0.98,
+      "step": 1224
+    },
+    {
+      "epoch": 0.4448072562358277,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.0001724635907477665,
+      "loss": 0.996,
+      "step": 1226
+    },
+    {
+      "epoch": 0.44553287981859413,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.0001724146371313181,
+      "loss": 0.9907,
+      "step": 1228
+    },
+    {
+      "epoch": 0.44625850340136053,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.00017236568351486968,
+      "loss": 1.0013,
+      "step": 1230
+    },
+    {
+      "epoch": 0.446984126984127,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.00017231672989842124,
+      "loss": 0.9736,
+      "step": 1232
+    },
+    {
+      "epoch": 0.44770975056689344,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.00017226777628197285,
+      "loss": 1.0385,
+      "step": 1234
+    },
+    {
+      "epoch": 0.44843537414965984,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.00017221882266552443,
+      "loss": 0.9977,
+      "step": 1236
+    },
+    {
+      "epoch": 0.4491609977324263,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.000172169869049076,
+      "loss": 1.0035,
+      "step": 1238
+    },
+    {
+      "epoch": 0.44988662131519275,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.0001721209154326276,
+      "loss": 1.0349,
+      "step": 1240
+    },
+    {
+      "epoch": 0.4506122448979592,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.00017207196181617918,
+      "loss": 1.0325,
+      "step": 1242
+    },
+    {
+      "epoch": 0.4513378684807256,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.00017202300819973076,
+      "loss": 1.0148,
+      "step": 1244
+    },
+    {
+      "epoch": 0.45206349206349206,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.00017197405458328234,
+      "loss": 0.997,
+      "step": 1246
+    },
+    {
+      "epoch": 0.4527891156462585,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 0.00017192510096683395,
+      "loss": 0.913,
+      "step": 1248
+    },
+    {
+      "epoch": 0.45351473922902497,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.0001718761473503855,
+      "loss": 0.9991,
+      "step": 1250
+    },
+    {
+      "epoch": 0.45424036281179136,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.0001718271937339371,
+      "loss": 1.0035,
+      "step": 1252
+    },
+    {
+      "epoch": 0.4549659863945578,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.0001717782401174887,
+      "loss": 0.9321,
+      "step": 1254
+    },
+    {
+      "epoch": 0.4556916099773243,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.00017172928650104028,
+      "loss": 0.9722,
+      "step": 1256
+    },
+    {
+      "epoch": 0.45641723356009073,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.00017168033288459186,
+      "loss": 1.0485,
+      "step": 1258
+    },
+    {
+      "epoch": 0.45714285714285713,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.00017163137926814345,
+      "loss": 0.953,
+      "step": 1260
+    },
+    {
+      "epoch": 0.4578684807256236,
+      "grad_norm": 0.1884765625,
+      "learning_rate": 0.00017158242565169503,
+      "loss": 1.0478,
+      "step": 1262
+    },
+    {
+      "epoch": 0.45859410430839004,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.0001715334720352466,
+      "loss": 0.8863,
+      "step": 1264
+    },
+    {
+      "epoch": 0.45931972789115644,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.0001714845184187982,
+      "loss": 0.9771,
+      "step": 1266
+    },
+    {
+      "epoch": 0.4600453514739229,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.0001714355648023498,
+      "loss": 0.9525,
+      "step": 1268
+    },
+    {
+      "epoch": 0.46077097505668935,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.00017138661118590136,
+      "loss": 1.0151,
+      "step": 1270
+    },
+    {
+      "epoch": 0.4614965986394558,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.00017133765756945294,
+      "loss": 1.0558,
+      "step": 1272
+    },
+    {
+      "epoch": 0.4622222222222222,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.00017128870395300455,
+      "loss": 1.0136,
+      "step": 1274
+    },
+    {
+      "epoch": 0.46294784580498866,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.00017123975033655613,
+      "loss": 0.9975,
+      "step": 1276
+    },
+    {
+      "epoch": 0.4636734693877551,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.0001711907967201077,
+      "loss": 1.028,
+      "step": 1278
+    },
+    {
+      "epoch": 0.46439909297052157,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.0001711418431036593,
+      "loss": 0.9967,
+      "step": 1280
+    },
+    {
+      "epoch": 0.46512471655328796,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.00017109288948721088,
+      "loss": 0.939,
+      "step": 1282
+    },
+    {
+      "epoch": 0.4658503401360544,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.00017104393587076246,
+      "loss": 1.011,
+      "step": 1284
+    },
+    {
+      "epoch": 0.4665759637188209,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.00017099498225431404,
+      "loss": 1.0439,
+      "step": 1286
+    },
+    {
+      "epoch": 0.46730158730158733,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.00017094602863786565,
+      "loss": 1.0081,
+      "step": 1288
+    },
+    {
+      "epoch": 0.46802721088435373,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.0001708970750214172,
+      "loss": 0.9553,
+      "step": 1290
+    },
+    {
+      "epoch": 0.4687528344671202,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.0001708481214049688,
+      "loss": 0.9824,
+      "step": 1292
+    },
+    {
+      "epoch": 0.46947845804988664,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.0001707991677885204,
+      "loss": 1.045,
+      "step": 1294
+    },
+    {
+      "epoch": 0.47020408163265304,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.00017075021417207198,
+      "loss": 1.0157,
+      "step": 1296
+    },
+    {
+      "epoch": 0.4709297052154195,
+      "grad_norm": 0.185546875,
+      "learning_rate": 0.00017070126055562356,
+      "loss": 0.9727,
+      "step": 1298
+    },
+    {
+      "epoch": 0.47165532879818595,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.00017065230693917515,
+      "loss": 0.9486,
+      "step": 1300
+    },
+    {
+      "epoch": 0.47165532879818595,
+      "eval_loss": 1.0229827165603638,
+      "eval_runtime": 91.9183,
+      "eval_samples_per_second": 19.583,
+      "eval_steps_per_second": 0.62,
+      "step": 1300
+    },
+    {
+      "epoch": 0.4723809523809524,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.00017060335332272673,
+      "loss": 1.0029,
+      "step": 1302
+    },
+    {
+      "epoch": 0.4731065759637188,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.0001705543997062783,
+      "loss": 1.0371,
+      "step": 1304
+    },
+    {
+      "epoch": 0.47383219954648526,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.0001705054460898299,
+      "loss": 0.9764,
+      "step": 1306
+    },
+    {
+      "epoch": 0.4745578231292517,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.00017045649247338147,
+      "loss": 0.9555,
+      "step": 1308
+    },
+    {
+      "epoch": 0.47528344671201816,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.00017040753885693306,
+      "loss": 1.0557,
+      "step": 1310
+    },
+    {
+      "epoch": 0.47600907029478456,
+      "grad_norm": 0.18359375,
+      "learning_rate": 0.00017035858524048464,
+      "loss": 0.9758,
+      "step": 1312
+    },
+    {
+      "epoch": 0.476734693877551,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.00017030963162403625,
+      "loss": 1.0147,
+      "step": 1314
+    },
+    {
+      "epoch": 0.4774603174603175,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.00017026067800758783,
+      "loss": 1.0386,
+      "step": 1316
+    },
+    {
+      "epoch": 0.4781859410430839,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.00017021172439113939,
+      "loss": 0.9744,
+      "step": 1318
+    },
+    {
+      "epoch": 0.47891156462585033,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.000170162770774691,
+      "loss": 0.9863,
+      "step": 1320
+    },
+    {
+      "epoch": 0.4796371882086168,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.00017011381715824258,
+      "loss": 1.0313,
+      "step": 1322
+    },
+    {
+      "epoch": 0.48036281179138324,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.00017006486354179416,
+      "loss": 0.9102,
+      "step": 1324
+    },
+    {
+      "epoch": 0.48108843537414964,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.00017001590992534574,
+      "loss": 1.0352,
+      "step": 1326
+    },
+    {
+      "epoch": 0.4818140589569161,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.00016996695630889732,
+      "loss": 1.011,
+      "step": 1328
+    },
+    {
+      "epoch": 0.48253968253968255,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.0001699180026924489,
+      "loss": 0.9854,
+      "step": 1330
+    },
+    {
+      "epoch": 0.483265306122449,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.0001698690490760005,
+      "loss": 0.9816,
+      "step": 1332
+    },
+    {
+      "epoch": 0.4839909297052154,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.0001698200954595521,
+      "loss": 0.9738,
+      "step": 1334
+    },
+    {
+      "epoch": 0.48471655328798186,
+      "grad_norm": 0.19140625,
+      "learning_rate": 0.00016977114184310368,
+      "loss": 1.0348,
+      "step": 1336
+    },
+    {
+      "epoch": 0.4854421768707483,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.00016972218822665524,
+      "loss": 1.0062,
+      "step": 1338
+    },
+    {
+      "epoch": 0.48616780045351476,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.00016967323461020684,
+      "loss": 0.9665,
+      "step": 1340
+    },
+    {
+      "epoch": 0.48689342403628116,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.00016962428099375843,
+      "loss": 0.978,
+      "step": 1342
+    },
+    {
+      "epoch": 0.4876190476190476,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.00016957532737731,
+      "loss": 1.0053,
+      "step": 1344
+    },
+    {
+      "epoch": 0.4883446712018141,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.0001695263737608616,
+      "loss": 1.0164,
+      "step": 1346
+    },
+    {
+      "epoch": 0.4890702947845805,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.00016947742014441317,
+      "loss": 1.0096,
+      "step": 1348
+    },
+    {
+      "epoch": 0.4897959183673469,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.00016942846652796476,
+      "loss": 0.9065,
+      "step": 1350
+    },
+    {
+      "epoch": 0.4905215419501134,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.00016937951291151634,
+      "loss": 0.9667,
+      "step": 1352
+    },
+    {
+      "epoch": 0.49124716553287984,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.00016933055929506795,
+      "loss": 1.0211,
+      "step": 1354
+    },
+    {
+      "epoch": 0.49197278911564624,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.00016928160567861953,
+      "loss": 1.1053,
+      "step": 1356
+    },
+    {
+      "epoch": 0.4926984126984127,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.00016923265206217109,
+      "loss": 1.0523,
+      "step": 1358
+    },
+    {
+      "epoch": 0.49342403628117915,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.0001691836984457227,
+      "loss": 0.9851,
+      "step": 1360
+    },
+    {
+      "epoch": 0.4941496598639456,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.00016913474482927428,
+      "loss": 1.0227,
+      "step": 1362
+    },
+    {
+      "epoch": 0.494875283446712,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.00016908579121282586,
+      "loss": 0.9792,
+      "step": 1364
+    },
+    {
+      "epoch": 0.49560090702947845,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.00016903683759637744,
+      "loss": 0.9671,
+      "step": 1366
+    },
+    {
+      "epoch": 0.4963265306122449,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.00016898788397992902,
+      "loss": 0.9982,
+      "step": 1368
+    },
+    {
+      "epoch": 0.49705215419501136,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.0001689389303634806,
+      "loss": 0.9745,
+      "step": 1370
+    },
+    {
+      "epoch": 0.49777777777777776,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.0001688899767470322,
+      "loss": 1.0174,
+      "step": 1372
+    },
+    {
+      "epoch": 0.4985034013605442,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.0001688410231305838,
+      "loss": 1.0175,
+      "step": 1374
+    },
+    {
+      "epoch": 0.4992290249433107,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.00016879206951413538,
+      "loss": 0.9615,
+      "step": 1376
+    },
+    {
+      "epoch": 0.4999546485260771,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.00016874311589768694,
+      "loss": 0.9905,
+      "step": 1378
+    },
+    {
+      "epoch": 0.5006802721088436,
+      "grad_norm": 0.2158203125,
+      "learning_rate": 0.00016869416228123854,
+      "loss": 0.9977,
+      "step": 1380
+    },
+    {
+      "epoch": 0.50140589569161,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.00016864520866479013,
+      "loss": 0.9374,
+      "step": 1382
+    },
+    {
+      "epoch": 0.5021315192743764,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.0001685962550483417,
+      "loss": 1.0049,
+      "step": 1384
+    },
+    {
+      "epoch": 0.5028571428571429,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.0001685473014318933,
+      "loss": 0.9839,
+      "step": 1386
+    },
+    {
+      "epoch": 0.5035827664399093,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.00016849834781544487,
+      "loss": 0.9599,
+      "step": 1388
+    },
+    {
+      "epoch": 0.5043083900226757,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.00016844939419899646,
+      "loss": 0.9604,
+      "step": 1390
+    },
+    {
+      "epoch": 0.5050340136054422,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.00016840044058254804,
+      "loss": 0.9192,
+      "step": 1392
+    },
+    {
+      "epoch": 0.5057596371882086,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.00016835148696609965,
+      "loss": 0.9461,
+      "step": 1394
+    },
+    {
+      "epoch": 0.5064852607709751,
+      "grad_norm": 0.17578125,
+      "learning_rate": 0.0001683025333496512,
+      "loss": 0.9682,
+      "step": 1396
+    },
+    {
+      "epoch": 0.5072108843537415,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.00016825357973320279,
+      "loss": 1.0531,
+      "step": 1398
+    },
+    {
+      "epoch": 0.5079365079365079,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 0.0001682046261167544,
+      "loss": 0.998,
+      "step": 1400
+    },
+    {
+      "epoch": 0.5079365079365079,
+      "eval_loss": 1.0224273204803467,
+      "eval_runtime": 90.8291,
+      "eval_samples_per_second": 19.817,
+      "eval_steps_per_second": 0.628,
+      "step": 1400
+    },
+    {
+      "epoch": 0.5086621315192744,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.00016815567250030598,
+      "loss": 1.0047,
+      "step": 1402
+    },
+    {
+      "epoch": 0.5093877551020408,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.00016810671888385756,
+      "loss": 0.9845,
+      "step": 1404
+    },
+    {
+      "epoch": 0.5101133786848072,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.00016805776526740914,
+      "loss": 0.9928,
+      "step": 1406
+    },
+    {
+      "epoch": 0.5108390022675737,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.00016800881165096072,
+      "loss": 0.971,
+      "step": 1408
+    },
+    {
+      "epoch": 0.5115646258503401,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.0001679598580345123,
+      "loss": 0.9549,
+      "step": 1410
+    },
+    {
+      "epoch": 0.5122902494331065,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.0001679109044180639,
+      "loss": 1.056,
+      "step": 1412
+    },
+    {
+      "epoch": 0.513015873015873,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.0001678619508016155,
+      "loss": 1.037,
+      "step": 1414
+    },
+    {
+      "epoch": 0.5137414965986394,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.00016781299718516705,
+      "loss": 1.0098,
+      "step": 1416
+    },
+    {
+      "epoch": 0.514467120181406,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.00016776404356871864,
+      "loss": 0.9179,
+      "step": 1418
+    },
+    {
+      "epoch": 0.5151927437641723,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.00016771508995227024,
+      "loss": 1.0747,
+      "step": 1420
+    },
+    {
+      "epoch": 0.5159183673469387,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.00016766613633582183,
+      "loss": 0.9267,
+      "step": 1422
+    },
+    {
+      "epoch": 0.5166439909297053,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.0001676171827193734,
+      "loss": 0.9201,
+      "step": 1424
+    },
+    {
+      "epoch": 0.5173696145124717,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.000167568229102925,
+      "loss": 1.0046,
+      "step": 1426
+    },
+    {
+      "epoch": 0.518095238095238,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.00016751927548647657,
+      "loss": 1.0547,
+      "step": 1428
+    },
+    {
+      "epoch": 0.5188208616780046,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.00016747032187002816,
+      "loss": 0.9742,
+      "step": 1430
+    },
+    {
+      "epoch": 0.519546485260771,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.00016742136825357974,
+      "loss": 1.0289,
+      "step": 1432
+    },
+    {
+      "epoch": 0.5202721088435374,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.00016737241463713135,
+      "loss": 0.9975,
+      "step": 1434
+    },
+    {
+      "epoch": 0.5209977324263039,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.0001673234610206829,
+      "loss": 1.0441,
+      "step": 1436
+    },
+    {
+      "epoch": 0.5217233560090703,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.00016727450740423449,
+      "loss": 1.0447,
+      "step": 1438
+    },
+    {
+      "epoch": 0.5224489795918368,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 0.0001672255537877861,
+      "loss": 1.083,
+      "step": 1440
+    },
+    {
+      "epoch": 0.5231746031746032,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.00016717660017133768,
+      "loss": 0.9575,
+      "step": 1442
+    },
+    {
+      "epoch": 0.5239002267573696,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.00016712764655488926,
+      "loss": 1.0298,
+      "step": 1444
+    },
+    {
+      "epoch": 0.5246258503401361,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.00016707869293844084,
+      "loss": 1.0205,
+      "step": 1446
+    },
+    {
+      "epoch": 0.5253514739229025,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.00016702973932199242,
+      "loss": 0.9916,
+      "step": 1448
+    },
+    {
+      "epoch": 0.5260770975056689,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.000166980785705544,
+      "loss": 1.1116,
+      "step": 1450
+    },
+    {
+      "epoch": 0.5268027210884354,
+      "grad_norm": 0.44140625,
+      "learning_rate": 0.0001669318320890956,
+      "loss": 1.072,
+      "step": 1452
+    },
+    {
+      "epoch": 0.5275283446712018,
+      "grad_norm": 0.1943359375,
+      "learning_rate": 0.00016688287847264717,
+      "loss": 0.9555,
+      "step": 1454
+    },
+    {
+      "epoch": 0.5282539682539683,
+      "grad_norm": 0.2177734375,
+      "learning_rate": 0.00016683392485619875,
+      "loss": 0.9393,
+      "step": 1456
+    },
+    {
+      "epoch": 0.5289795918367347,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.00016678497123975033,
+      "loss": 0.9854,
+      "step": 1458
+    },
+    {
+      "epoch": 0.5297052154195011,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.00016673601762330194,
+      "loss": 1.0133,
+      "step": 1460
+    },
+    {
+      "epoch": 0.5304308390022676,
+      "grad_norm": 0.177734375,
+      "learning_rate": 0.00016668706400685353,
+      "loss": 1.0171,
+      "step": 1462
+    },
+    {
+      "epoch": 0.531156462585034,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.00016663811039040508,
+      "loss": 0.9813,
+      "step": 1464
+    },
+    {
+      "epoch": 0.5318820861678004,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.0001665891567739567,
+      "loss": 1.0612,
+      "step": 1466
+    },
+    {
+      "epoch": 0.5326077097505669,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.00016654020315750827,
+      "loss": 0.9901,
+      "step": 1468
+    },
+    {
+      "epoch": 0.5333333333333333,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.00016649124954105986,
+      "loss": 1.0232,
+      "step": 1470
+    },
+    {
+      "epoch": 0.5340589569160997,
+      "grad_norm": 0.17578125,
+      "learning_rate": 0.00016644229592461144,
+      "loss": 1.0505,
+      "step": 1472
+    },
+    {
+      "epoch": 0.5347845804988662,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.00016639334230816302,
+      "loss": 0.9528,
+      "step": 1474
+    },
+    {
+      "epoch": 0.5355102040816326,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.0001663443886917146,
+      "loss": 1.0136,
+      "step": 1476
+    },
+    {
+      "epoch": 0.5362358276643991,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.00016629543507526618,
+      "loss": 0.9827,
+      "step": 1478
+    },
+    {
+      "epoch": 0.5369614512471655,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.0001662464814588178,
+      "loss": 0.9911,
+      "step": 1480
+    },
+    {
+      "epoch": 0.5376870748299319,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.00016619752784236938,
+      "loss": 0.9756,
+      "step": 1482
+    },
+    {
+      "epoch": 0.5384126984126985,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.00016614857422592093,
+      "loss": 0.9821,
+      "step": 1484
+    },
+    {
+      "epoch": 0.5391383219954649,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.00016609962060947254,
+      "loss": 0.9637,
+      "step": 1486
+    },
+    {
+      "epoch": 0.5398639455782313,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.00016605066699302412,
+      "loss": 1.0195,
+      "step": 1488
+    },
+    {
+      "epoch": 0.5405895691609978,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.0001660017133765757,
+      "loss": 1.0409,
+      "step": 1490
+    },
+    {
+      "epoch": 0.5413151927437642,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.0001659527597601273,
+      "loss": 1.0361,
+      "step": 1492
+    },
+    {
+      "epoch": 0.5420408163265306,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.00016590380614367887,
+      "loss": 1.0197,
+      "step": 1494
+    },
+    {
+      "epoch": 0.5427664399092971,
+      "grad_norm": 0.177734375,
+      "learning_rate": 0.00016585485252723045,
+      "loss": 0.9975,
+      "step": 1496
+    },
+    {
+      "epoch": 0.5434920634920635,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.00016580589891078203,
+      "loss": 0.9776,
+      "step": 1498
+    },
+    {
+      "epoch": 0.54421768707483,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.00016575694529433364,
+      "loss": 0.9366,
+      "step": 1500
+    },
+    {
+      "epoch": 0.54421768707483,
+      "eval_loss": 1.0217424631118774,
+      "eval_runtime": 92.5726,
+      "eval_samples_per_second": 19.444,
+      "eval_steps_per_second": 0.616,
+      "step": 1500
+    },
+    {
+      "epoch": 0.5449433106575964,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.00016570799167788523,
+      "loss": 1.0161,
+      "step": 1502
+    },
+    {
+      "epoch": 0.5456689342403628,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.00016565903806143678,
+      "loss": 1.07,
+      "step": 1504
+    },
+    {
+      "epoch": 0.5463945578231293,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.0001656100844449884,
+      "loss": 0.8992,
+      "step": 1506
+    },
+    {
+      "epoch": 0.5471201814058957,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.00016556113082853997,
+      "loss": 1.123,
+      "step": 1508
+    },
+    {
+      "epoch": 0.5478458049886621,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.00016551217721209156,
+      "loss": 1.1036,
+      "step": 1510
+    },
+    {
+      "epoch": 0.5485714285714286,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.00016546322359564314,
+      "loss": 0.8912,
+      "step": 1512
+    },
+    {
+      "epoch": 0.549297052154195,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.00016541426997919472,
+      "loss": 0.9678,
+      "step": 1514
+    },
+    {
+      "epoch": 0.5500226757369614,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.0001653653163627463,
+      "loss": 0.9695,
+      "step": 1516
+    },
+    {
+      "epoch": 0.5507482993197279,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.00016531636274629788,
+      "loss": 0.9806,
+      "step": 1518
+    },
+    {
+      "epoch": 0.5514739229024943,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.0001652674091298495,
+      "loss": 1.0622,
+      "step": 1520
+    },
+    {
+      "epoch": 0.5521995464852608,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.00016521845551340108,
+      "loss": 0.9587,
+      "step": 1522
+    },
+    {
+      "epoch": 0.5529251700680272,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.00016516950189695263,
+      "loss": 0.9624,
+      "step": 1524
+    },
+    {
+      "epoch": 0.5536507936507936,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.00016512054828050424,
+      "loss": 0.9404,
+      "step": 1526
+    },
+    {
+      "epoch": 0.5543764172335601,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.00016507159466405582,
+      "loss": 1.0409,
+      "step": 1528
+    },
+    {
+      "epoch": 0.5551020408163265,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.0001650226410476074,
+      "loss": 1.0547,
+      "step": 1530
+    },
+    {
+      "epoch": 0.5558276643990929,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.00016497368743115896,
+      "loss": 0.9696,
+      "step": 1532
+    },
+    {
+      "epoch": 0.5565532879818594,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.00016492473381471057,
+      "loss": 1.032,
+      "step": 1534
+    },
+    {
+      "epoch": 0.5572789115646258,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.00016487578019826215,
+      "loss": 1.0726,
+      "step": 1536
+    },
+    {
+      "epoch": 0.5580045351473923,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.00016482682658181373,
+      "loss": 1.0213,
+      "step": 1538
+    },
+    {
+      "epoch": 0.5587301587301587,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.00016477787296536534,
+      "loss": 0.9804,
+      "step": 1540
+    },
+    {
+      "epoch": 0.5594557823129251,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.0001647289193489169,
+      "loss": 1.0306,
+      "step": 1542
+    },
+    {
+      "epoch": 0.5601814058956917,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.00016467996573246848,
+      "loss": 1.0537,
+      "step": 1544
+    },
+    {
+      "epoch": 0.560907029478458,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.0001646310121160201,
+      "loss": 0.928,
+      "step": 1546
+    },
+    {
+      "epoch": 0.5616326530612245,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.00016458205849957167,
+      "loss": 1.0409,
+      "step": 1548
+    },
+    {
+      "epoch": 0.562358276643991,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.00016453310488312326,
+      "loss": 0.966,
+      "step": 1550
+    },
+    {
+      "epoch": 0.5630839002267574,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.0001644841512666748,
+      "loss": 1.0074,
+      "step": 1552
+    },
+    {
+      "epoch": 0.5638095238095238,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.00016443519765022642,
+      "loss": 0.9354,
+      "step": 1554
+    },
+    {
+      "epoch": 0.5645351473922903,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.000164386244033778,
+      "loss": 1.0507,
+      "step": 1556
+    },
+    {
+      "epoch": 0.5652607709750567,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.00016433729041732958,
+      "loss": 1.0014,
+      "step": 1558
+    },
+    {
+      "epoch": 0.5659863945578232,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.0001642883368008812,
+      "loss": 0.9745,
+      "step": 1560
+    },
+    {
+      "epoch": 0.5667120181405896,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.00016423938318443275,
+      "loss": 0.9293,
+      "step": 1562
+    },
+    {
+      "epoch": 0.567437641723356,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.00016419042956798433,
+      "loss": 1.1054,
+      "step": 1564
+    },
+    {
+      "epoch": 0.5681632653061225,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.00016414147595153594,
+      "loss": 0.9644,
+      "step": 1566
+    },
+    {
+      "epoch": 0.5688888888888889,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.00016409252233508752,
+      "loss": 1.0018,
+      "step": 1568
+    },
+    {
+      "epoch": 0.5696145124716553,
+      "grad_norm": 0.19140625,
+      "learning_rate": 0.0001640435687186391,
+      "loss": 1.0597,
+      "step": 1570
+    },
+    {
+      "epoch": 0.5703401360544218,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.00016399461510219066,
+      "loss": 1.0581,
+      "step": 1572
+    },
+    {
+      "epoch": 0.5710657596371882,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.00016394566148574227,
+      "loss": 1.0389,
+      "step": 1574
+    },
+    {
+      "epoch": 0.5717913832199546,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.00016389670786929385,
+      "loss": 0.9944,
+      "step": 1576
+    },
+    {
+      "epoch": 0.5725170068027211,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.00016384775425284543,
+      "loss": 1.0125,
+      "step": 1578
+    },
+    {
+      "epoch": 0.5732426303854875,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.00016379880063639704,
+      "loss": 1.0272,
+      "step": 1580
+    },
+    {
+      "epoch": 0.573968253968254,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.0001637498470199486,
+      "loss": 1.0639,
+      "step": 1582
+    },
+    {
+      "epoch": 0.5746938775510204,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.00016370089340350018,
+      "loss": 0.9577,
+      "step": 1584
+    },
+    {
+      "epoch": 0.5754195011337868,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.0001636519397870518,
+      "loss": 1.0295,
+      "step": 1586
+    },
+    {
+      "epoch": 0.5761451247165533,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.00016360298617060337,
+      "loss": 1.0028,
+      "step": 1588
+    },
+    {
+      "epoch": 0.5768707482993197,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.00016355403255415496,
+      "loss": 1.0054,
+      "step": 1590
+    },
+    {
+      "epoch": 0.5775963718820861,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.00016350507893770654,
+      "loss": 0.9994,
+      "step": 1592
+    },
+    {
+      "epoch": 0.5783219954648526,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.00016345612532125812,
+      "loss": 0.932,
+      "step": 1594
+    },
+    {
+      "epoch": 0.579047619047619,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.0001634071717048097,
+      "loss": 0.9427,
+      "step": 1596
+    },
+    {
+      "epoch": 0.5797732426303854,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.00016335821808836128,
+      "loss": 0.9373,
+      "step": 1598
+    },
+    {
+      "epoch": 0.5804988662131519,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.00016330926447191287,
+      "loss": 1.0353,
+      "step": 1600
+    },
+    {
+      "epoch": 0.5804988662131519,
+      "eval_loss": 1.0210474729537964,
+      "eval_runtime": 93.6809,
+      "eval_samples_per_second": 19.214,
+      "eval_steps_per_second": 0.608,
+      "step": 1600
+    },
+    {
+      "epoch": 0.5812244897959183,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.00016326031085546445,
+      "loss": 0.9768,
+      "step": 1602
+    },
+    {
+      "epoch": 0.5819501133786849,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.00016321135723901603,
+      "loss": 0.9428,
+      "step": 1604
+    },
+    {
+      "epoch": 0.5826757369614513,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.00016316240362256764,
+      "loss": 1.0214,
+      "step": 1606
+    },
+    {
+      "epoch": 0.5834013605442177,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.00016311345000611922,
+      "loss": 0.92,
+      "step": 1608
+    },
+    {
+      "epoch": 0.5841269841269842,
+      "grad_norm": 0.18359375,
+      "learning_rate": 0.00016306449638967078,
+      "loss": 1.0035,
+      "step": 1610
+    },
+    {
+      "epoch": 0.5848526077097506,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.0001630155427732224,
+      "loss": 1.0347,
+      "step": 1612
+    },
+    {
+      "epoch": 0.585578231292517,
+      "grad_norm": 0.185546875,
+      "learning_rate": 0.00016296658915677397,
+      "loss": 1.0052,
+      "step": 1614
+    },
+    {
+      "epoch": 0.5863038548752835,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.00016291763554032555,
+      "loss": 0.9912,
+      "step": 1616
+    },
+    {
+      "epoch": 0.5870294784580499,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.00016286868192387713,
+      "loss": 0.9186,
+      "step": 1618
+    },
+    {
+      "epoch": 0.5877551020408164,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.00016281972830742872,
+      "loss": 1.0071,
+      "step": 1620
+    },
+    {
+      "epoch": 0.5884807256235828,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.0001627707746909803,
+      "loss": 0.9302,
+      "step": 1622
+    },
+    {
+      "epoch": 0.5892063492063492,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.00016272182107453188,
+      "loss": 1.0591,
+      "step": 1624
+    },
+    {
+      "epoch": 0.5899319727891157,
+      "grad_norm": 0.173828125,
+      "learning_rate": 0.0001626728674580835,
+      "loss": 0.9493,
+      "step": 1626
+    },
+    {
+      "epoch": 0.5906575963718821,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 0.00016262391384163507,
+      "loss": 1.0083,
+      "step": 1628
+    },
+    {
+      "epoch": 0.5913832199546485,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.00016257496022518663,
+      "loss": 1.0483,
+      "step": 1630
+    },
+    {
+      "epoch": 0.592108843537415,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.00016252600660873824,
+      "loss": 1.0442,
+      "step": 1632
+    },
+    {
+      "epoch": 0.5928344671201814,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.00016247705299228982,
+      "loss": 1.0041,
+      "step": 1634
+    },
+    {
+      "epoch": 0.5935600907029478,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.0001624280993758414,
+      "loss": 0.8814,
+      "step": 1636
+    },
+    {
+      "epoch": 0.5942857142857143,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.00016237914575939298,
+      "loss": 0.9601,
+      "step": 1638
+    },
+    {
+      "epoch": 0.5950113378684807,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.00016233019214294457,
+      "loss": 1.0641,
+      "step": 1640
+    },
+    {
+      "epoch": 0.5957369614512472,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.00016228123852649615,
+      "loss": 0.928,
+      "step": 1642
+    },
+    {
+      "epoch": 0.5964625850340136,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.00016223228491004773,
+      "loss": 0.9936,
+      "step": 1644
+    },
+    {
+      "epoch": 0.59718820861678,
+      "grad_norm": 0.177734375,
+      "learning_rate": 0.00016218333129359934,
+      "loss": 1.0311,
+      "step": 1646
+    },
+    {
+      "epoch": 0.5979138321995465,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.00016213437767715092,
+      "loss": 1.0138,
+      "step": 1648
+    },
+    {
+      "epoch": 0.5986394557823129,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.00016208542406070248,
+      "loss": 0.9511,
+      "step": 1650
+    },
+    {
+      "epoch": 0.5993650793650793,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.0001620364704442541,
+      "loss": 0.9711,
+      "step": 1652
+    },
+    {
+      "epoch": 0.6000907029478458,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 0.00016198751682780567,
+      "loss": 0.9674,
+      "step": 1654
+    },
+    {
+      "epoch": 0.6008163265306122,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.00016193856321135725,
+      "loss": 1.0381,
+      "step": 1656
+    },
+    {
+      "epoch": 0.6015419501133786,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.00016188960959490883,
+      "loss": 0.9523,
+      "step": 1658
+    },
+    {
+      "epoch": 0.6022675736961451,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.00016184065597846042,
+      "loss": 0.9676,
+      "step": 1660
+    },
+    {
+      "epoch": 0.6029931972789115,
+      "grad_norm": 0.17578125,
+      "learning_rate": 0.000161791702362012,
+      "loss": 0.9896,
+      "step": 1662
+    },
+    {
+      "epoch": 0.603718820861678,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.00016174274874556358,
+      "loss": 0.9777,
+      "step": 1664
+    },
+    {
+      "epoch": 0.6044444444444445,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.0001616937951291152,
+      "loss": 0.9839,
+      "step": 1666
+    },
+    {
+      "epoch": 0.6051700680272109,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.00016164484151266677,
+      "loss": 0.9354,
+      "step": 1668
+    },
+    {
+      "epoch": 0.6058956916099774,
+      "grad_norm": 0.173828125,
+      "learning_rate": 0.00016159588789621833,
+      "loss": 0.9798,
+      "step": 1670
+    },
+    {
+      "epoch": 0.6066213151927438,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.00016154693427976994,
+      "loss": 0.99,
+      "step": 1672
+    },
+    {
+      "epoch": 0.6073469387755102,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.00016149798066332152,
+      "loss": 1.0575,
+      "step": 1674
+    },
+    {
+      "epoch": 0.6080725623582767,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 0.0001614490270468731,
+      "loss": 1.0201,
+      "step": 1676
+    },
+    {
+      "epoch": 0.6087981859410431,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.00016140007343042466,
+      "loss": 0.9888,
+      "step": 1678
+    },
+    {
+      "epoch": 0.6095238095238096,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.00016135111981397627,
+      "loss": 0.9927,
+      "step": 1680
+    },
+    {
+      "epoch": 0.610249433106576,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.00016130216619752785,
+      "loss": 1.0186,
+      "step": 1682
+    },
+    {
+      "epoch": 0.6109750566893424,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.00016125321258107943,
+      "loss": 1.0128,
+      "step": 1684
+    },
+    {
+      "epoch": 0.6117006802721089,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.00016120425896463104,
+      "loss": 1.0212,
+      "step": 1686
+    },
+    {
+      "epoch": 0.6124263038548753,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.0001611553053481826,
+      "loss": 0.9624,
+      "step": 1688
+    },
+    {
+      "epoch": 0.6131519274376417,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.00016110635173173418,
+      "loss": 1.0341,
+      "step": 1690
+    },
+    {
+      "epoch": 0.6138775510204082,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.0001610573981152858,
+      "loss": 0.9915,
+      "step": 1692
+    },
+    {
+      "epoch": 0.6146031746031746,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.00016100844449883737,
+      "loss": 1.0157,
+      "step": 1694
+    },
+    {
+      "epoch": 0.615328798185941,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.00016095949088238895,
+      "loss": 1.0553,
+      "step": 1696
+    },
+    {
+      "epoch": 0.6160544217687075,
+      "grad_norm": 0.185546875,
+      "learning_rate": 0.0001609105372659405,
+      "loss": 0.9294,
+      "step": 1698
+    },
+    {
+      "epoch": 0.6167800453514739,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.00016086158364949212,
+      "loss": 0.9272,
+      "step": 1700
+    },
+    {
+      "epoch": 0.6167800453514739,
+      "eval_loss": 1.0209547281265259,
+      "eval_runtime": 91.8934,
+      "eval_samples_per_second": 19.588,
+      "eval_steps_per_second": 0.62,
+      "step": 1700
+    },
+    {
+      "epoch": 0.6175056689342404,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.0001608126300330437,
+      "loss": 0.976,
+      "step": 1702
+    },
+    {
+      "epoch": 0.6182312925170068,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.00016076367641659528,
+      "loss": 0.9733,
+      "step": 1704
+    },
+    {
+      "epoch": 0.6189569160997732,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.0001607147228001469,
+      "loss": 1.0233,
+      "step": 1706
+    },
+    {
+      "epoch": 0.6196825396825397,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.00016066576918369845,
+      "loss": 0.9805,
+      "step": 1708
+    },
+    {
+      "epoch": 0.6204081632653061,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.00016061681556725003,
+      "loss": 0.9321,
+      "step": 1710
+    },
+    {
+      "epoch": 0.6211337868480725,
+      "grad_norm": 0.1875,
+      "learning_rate": 0.00016056786195080164,
+      "loss": 1.0678,
+      "step": 1712
+    },
+    {
+      "epoch": 0.621859410430839,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.00016051890833435322,
+      "loss": 1.0146,
+      "step": 1714
+    },
+    {
+      "epoch": 0.6225850340136054,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.0001604699547179048,
+      "loss": 0.8791,
+      "step": 1716
+    },
+    {
+      "epoch": 0.6233106575963718,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.00016042100110145636,
+      "loss": 0.9568,
+      "step": 1718
+    },
+    {
+      "epoch": 0.6240362811791383,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.00016037204748500797,
+      "loss": 0.9577,
+      "step": 1720
+    },
+    {
+      "epoch": 0.6247619047619047,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.00016032309386855955,
+      "loss": 0.9506,
+      "step": 1722
+    },
+    {
+      "epoch": 0.6254875283446713,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.00016027414025211113,
+      "loss": 0.9523,
+      "step": 1724
+    },
+    {
+      "epoch": 0.6262131519274377,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.00016022518663566274,
+      "loss": 0.997,
+      "step": 1726
+    },
+    {
+      "epoch": 0.626938775510204,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.0001601762330192143,
+      "loss": 0.9419,
+      "step": 1728
+    },
+    {
+      "epoch": 0.6276643990929706,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.00016012727940276588,
+      "loss": 0.9835,
+      "step": 1730
+    },
+    {
+      "epoch": 0.628390022675737,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.0001600783257863175,
+      "loss": 0.9472,
+      "step": 1732
+    },
+    {
+      "epoch": 0.6291156462585034,
+      "grad_norm": 0.177734375,
+      "learning_rate": 0.00016002937216986907,
+      "loss": 1.0306,
+      "step": 1734
+    },
+    {
+      "epoch": 0.6298412698412699,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.00015998041855342065,
+      "loss": 1.0021,
+      "step": 1736
+    },
+    {
+      "epoch": 0.6305668934240363,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.0001599314649369722,
+      "loss": 0.9455,
+      "step": 1738
+    },
+    {
+      "epoch": 0.6312925170068027,
+      "grad_norm": 0.173828125,
+      "learning_rate": 0.00015988251132052382,
+      "loss": 1.0614,
+      "step": 1740
+    },
+    {
+      "epoch": 0.6320181405895692,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.0001598335577040754,
+      "loss": 0.9934,
+      "step": 1742
+    },
+    {
+      "epoch": 0.6327437641723356,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.00015978460408762698,
+      "loss": 1.0372,
+      "step": 1744
+    },
+    {
+      "epoch": 0.6334693877551021,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.00015973565047117856,
+      "loss": 1.0382,
+      "step": 1746
+    },
+    {
+      "epoch": 0.6341950113378685,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.00015968669685473015,
+      "loss": 1.001,
+      "step": 1748
+    },
+    {
+      "epoch": 0.6349206349206349,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.00015963774323828173,
+      "loss": 0.9544,
+      "step": 1750
+    },
+    {
+      "epoch": 0.6356462585034014,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.00015958878962183334,
+      "loss": 0.9517,
+      "step": 1752
+    },
+    {
+      "epoch": 0.6363718820861678,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.00015953983600538492,
+      "loss": 0.9706,
+      "step": 1754
+    },
+    {
+      "epoch": 0.6370975056689342,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.00015949088238893647,
+      "loss": 1.0153,
+      "step": 1756
+    },
+    {
+      "epoch": 0.6378231292517007,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.00015944192877248806,
+      "loss": 0.9559,
+      "step": 1758
+    },
+    {
+      "epoch": 0.6385487528344671,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.00015939297515603967,
+      "loss": 0.9877,
+      "step": 1760
+    },
+    {
+      "epoch": 0.6392743764172336,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.00015934402153959125,
+      "loss": 0.9975,
+      "step": 1762
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.00015929506792314283,
+      "loss": 0.9666,
+      "step": 1764
+    },
+    {
+      "epoch": 0.6407256235827664,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.0001592461143066944,
+      "loss": 1.028,
+      "step": 1766
+    },
+    {
+      "epoch": 0.6414512471655329,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.000159197160690246,
+      "loss": 1.0233,
+      "step": 1768
+    },
+    {
+      "epoch": 0.6421768707482993,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.00015914820707379758,
+      "loss": 1.017,
+      "step": 1770
+    },
+    {
+      "epoch": 0.6429024943310657,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.0001590992534573492,
+      "loss": 1.0094,
+      "step": 1772
+    },
+    {
+      "epoch": 0.6436281179138322,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.00015905029984090077,
+      "loss": 0.9779,
+      "step": 1774
+    },
+    {
+      "epoch": 0.6443537414965986,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.00015900134622445232,
+      "loss": 0.9917,
+      "step": 1776
+    },
+    {
+      "epoch": 0.645079365079365,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.0001589523926080039,
+      "loss": 0.9526,
+      "step": 1778
+    },
+    {
+      "epoch": 0.6458049886621315,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.00015890343899155552,
+      "loss": 1.0131,
+      "step": 1780
+    },
+    {
+      "epoch": 0.6465306122448979,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.0001588544853751071,
+      "loss": 1.053,
+      "step": 1782
+    },
+    {
+      "epoch": 0.6472562358276645,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.00015880553175865868,
+      "loss": 0.9761,
+      "step": 1784
+    },
+    {
+      "epoch": 0.6479818594104308,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.00015875657814221026,
+      "loss": 0.9547,
+      "step": 1786
+    },
+    {
+      "epoch": 0.6487074829931972,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.00015870762452576184,
+      "loss": 1.0081,
+      "step": 1788
+    },
+    {
+      "epoch": 0.6494331065759638,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 0.00015865867090931343,
+      "loss": 1.007,
+      "step": 1790
+    },
+    {
+      "epoch": 0.6501587301587302,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.00015860971729286504,
+      "loss": 1.0005,
+      "step": 1792
+    },
+    {
+      "epoch": 0.6508843537414966,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.00015856076367641662,
+      "loss": 0.9823,
+      "step": 1794
+    },
+    {
+      "epoch": 0.6516099773242631,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.00015851181005996817,
+      "loss": 0.9324,
+      "step": 1796
+    },
+    {
+      "epoch": 0.6523356009070295,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.00015846285644351976,
+      "loss": 1.0323,
+      "step": 1798
+    },
+    {
+      "epoch": 0.6530612244897959,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.00015841390282707137,
+      "loss": 0.942,
+      "step": 1800
+    },
+    {
+      "epoch": 0.6530612244897959,
+      "eval_loss": 1.0201084613800049,
+      "eval_runtime": 94.9174,
+      "eval_samples_per_second": 18.964,
+      "eval_steps_per_second": 0.601,
+      "step": 1800
+    },
+    {
+      "epoch": 0.6537868480725624,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.00015836494921062295,
+      "loss": 0.9913,
+      "step": 1802
+    },
+    {
+      "epoch": 0.6545124716553288,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.00015831599559417453,
+      "loss": 1.098,
+      "step": 1804
+    },
+    {
+      "epoch": 0.6552380952380953,
+      "grad_norm": 0.173828125,
+      "learning_rate": 0.0001582670419777261,
+      "loss": 1.0758,
+      "step": 1806
+    },
+    {
+      "epoch": 0.6559637188208617,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.0001582180883612777,
+      "loss": 0.9549,
+      "step": 1808
+    },
+    {
+      "epoch": 0.6566893424036281,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.00015816913474482928,
+      "loss": 1.0207,
+      "step": 1810
+    },
+    {
+      "epoch": 0.6574149659863946,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.0001581201811283809,
+      "loss": 0.9206,
+      "step": 1812
+    },
+    {
+      "epoch": 0.658140589569161,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.00015807122751193247,
+      "loss": 0.9818,
+      "step": 1814
+    },
+    {
+      "epoch": 0.6588662131519274,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.00015802227389548402,
+      "loss": 1.0962,
+      "step": 1816
+    },
+    {
+      "epoch": 0.6595918367346939,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.0001579733202790356,
+      "loss": 0.9518,
+      "step": 1818
+    },
+    {
+      "epoch": 0.6603174603174603,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.00015792436666258722,
+      "loss": 0.9924,
+      "step": 1820
+    },
+    {
+      "epoch": 0.6610430839002267,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.0001578754130461388,
+      "loss": 0.9963,
+      "step": 1822
+    },
+    {
+      "epoch": 0.6617687074829932,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.00015782645942969035,
+      "loss": 0.983,
+      "step": 1824
+    },
+    {
+      "epoch": 0.6624943310657596,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.00015777750581324196,
+      "loss": 1.0437,
+      "step": 1826
+    },
+    {
+      "epoch": 0.6632199546485261,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.00015772855219679354,
+      "loss": 1.0547,
+      "step": 1828
+    },
+    {
+      "epoch": 0.6639455782312925,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.00015767959858034513,
+      "loss": 0.9898,
+      "step": 1830
+    },
+    {
+      "epoch": 0.6646712018140589,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.00015763064496389674,
+      "loss": 1.0343,
+      "step": 1832
+    },
+    {
+      "epoch": 0.6653968253968254,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.0001575816913474483,
+      "loss": 1.0069,
+      "step": 1834
+    },
+    {
+      "epoch": 0.6661224489795918,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.00015753273773099987,
+      "loss": 1.032,
+      "step": 1836
+    },
+    {
+      "epoch": 0.6668480725623582,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.00015748378411455146,
+      "loss": 0.9829,
+      "step": 1838
+    },
+    {
+      "epoch": 0.6675736961451247,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.00015743483049810307,
+      "loss": 1.0174,
+      "step": 1840
+    },
+    {
+      "epoch": 0.6682993197278911,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.00015738587688165465,
+      "loss": 1.0235,
+      "step": 1842
+    },
+    {
+      "epoch": 0.6690249433106576,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.0001573369232652062,
+      "loss": 0.9988,
+      "step": 1844
+    },
+    {
+      "epoch": 0.669750566893424,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.0001572879696487578,
+      "loss": 0.9146,
+      "step": 1846
+    },
+    {
+      "epoch": 0.6704761904761904,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.0001572390160323094,
+      "loss": 1.0098,
+      "step": 1848
+    },
+    {
+      "epoch": 0.671201814058957,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.00015719006241586098,
+      "loss": 0.9822,
+      "step": 1850
+    },
+    {
+      "epoch": 0.6719274376417234,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.00015714110879941259,
+      "loss": 1.0375,
+      "step": 1852
+    },
+    {
+      "epoch": 0.6726530612244898,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.00015709215518296414,
+      "loss": 1.0528,
+      "step": 1854
+    },
+    {
+      "epoch": 0.6733786848072563,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.00015704320156651572,
+      "loss": 0.9579,
+      "step": 1856
+    },
+    {
+      "epoch": 0.6741043083900227,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.0001569942479500673,
+      "loss": 1.0379,
+      "step": 1858
+    },
+    {
+      "epoch": 0.6748299319727891,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.00015694529433361892,
+      "loss": 1.0233,
+      "step": 1860
+    },
+    {
+      "epoch": 0.6755555555555556,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.0001568963407171705,
+      "loss": 0.9767,
+      "step": 1862
+    },
+    {
+      "epoch": 0.676281179138322,
+      "grad_norm": 0.173828125,
+      "learning_rate": 0.00015684738710072205,
+      "loss": 0.9701,
+      "step": 1864
+    },
+    {
+      "epoch": 0.6770068027210885,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.00015679843348427366,
+      "loss": 0.9249,
+      "step": 1866
+    },
+    {
+      "epoch": 0.6777324263038549,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.00015674947986782524,
+      "loss": 0.9929,
+      "step": 1868
+    },
+    {
+      "epoch": 0.6784580498866213,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.00015670052625137683,
+      "loss": 0.937,
+      "step": 1870
+    },
+    {
+      "epoch": 0.6791836734693878,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.00015665157263492844,
+      "loss": 0.9915,
+      "step": 1872
+    },
+    {
+      "epoch": 0.6799092970521542,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.00015660261901848,
+      "loss": 1.0443,
+      "step": 1874
+    },
+    {
+      "epoch": 0.6806349206349206,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.00015655366540203157,
+      "loss": 1.051,
+      "step": 1876
+    },
+    {
+      "epoch": 0.6813605442176871,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.00015650471178558316,
+      "loss": 0.9451,
+      "step": 1878
+    },
+    {
+      "epoch": 0.6820861678004535,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.00015645575816913477,
+      "loss": 0.9998,
+      "step": 1880
+    },
+    {
+      "epoch": 0.6828117913832199,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.00015640680455268635,
+      "loss": 0.9396,
+      "step": 1882
+    },
+    {
+      "epoch": 0.6835374149659864,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.0001563578509362379,
+      "loss": 1.0181,
+      "step": 1884
+    },
+    {
+      "epoch": 0.6842630385487528,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.0001563088973197895,
+      "loss": 1.0061,
+      "step": 1886
+    },
+    {
+      "epoch": 0.6849886621315193,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.0001562599437033411,
+      "loss": 0.9953,
+      "step": 1888
+    },
+    {
+      "epoch": 0.6857142857142857,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.00015621099008689268,
+      "loss": 1.0486,
+      "step": 1890
+    },
+    {
+      "epoch": 0.6864399092970521,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.00015616203647044426,
+      "loss": 0.9057,
+      "step": 1892
+    },
+    {
+      "epoch": 0.6871655328798186,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.00015611308285399584,
+      "loss": 0.9876,
+      "step": 1894
+    },
+    {
+      "epoch": 0.687891156462585,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.00015606412923754742,
+      "loss": 0.9982,
+      "step": 1896
+    },
+    {
+      "epoch": 0.6886167800453514,
+      "grad_norm": 0.173828125,
+      "learning_rate": 0.000156015175621099,
+      "loss": 0.9909,
+      "step": 1898
+    },
+    {
+      "epoch": 0.6893424036281179,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.00015596622200465062,
+      "loss": 0.9472,
+      "step": 1900
+    },
+    {
+      "epoch": 0.6893424036281179,
+      "eval_loss": 1.019601821899414,
+      "eval_runtime": 96.3922,
+      "eval_samples_per_second": 18.674,
+      "eval_steps_per_second": 0.591,
+      "step": 1900
+    },
+    {
+      "epoch": 0.6900680272108843,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.00015591726838820217,
+      "loss": 0.9808,
+      "step": 1902
+    },
+    {
+      "epoch": 0.6907936507936508,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.00015586831477175375,
+      "loss": 1.0067,
+      "step": 1904
+    },
+    {
+      "epoch": 0.6915192743764172,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.00015581936115530536,
+      "loss": 0.941,
+      "step": 1906
+    },
+    {
+      "epoch": 0.6922448979591836,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.00015577040753885694,
+      "loss": 0.944,
+      "step": 1908
+    },
+    {
+      "epoch": 0.6929705215419502,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.00015572145392240853,
+      "loss": 0.9849,
+      "step": 1910
+    },
+    {
+      "epoch": 0.6936961451247166,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.0001556725003059601,
+      "loss": 0.9713,
+      "step": 1912
+    },
+    {
+      "epoch": 0.694421768707483,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.0001556235466895117,
+      "loss": 1.0218,
+      "step": 1914
+    },
+    {
+      "epoch": 0.6951473922902495,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.00015557459307306327,
+      "loss": 0.9998,
+      "step": 1916
+    },
+    {
+      "epoch": 0.6958730158730159,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.00015552563945661486,
+      "loss": 1.0357,
+      "step": 1918
+    },
+    {
+      "epoch": 0.6965986394557823,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.00015547668584016647,
+      "loss": 0.9338,
+      "step": 1920
+    },
+    {
+      "epoch": 0.6973242630385488,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.00015542773222371802,
+      "loss": 1.0691,
+      "step": 1922
+    },
+    {
+      "epoch": 0.6980498866213152,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.0001553787786072696,
+      "loss": 0.994,
+      "step": 1924
+    },
+    {
+      "epoch": 0.6987755102040817,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.0001553298249908212,
+      "loss": 0.9554,
+      "step": 1926
+    },
+    {
+      "epoch": 0.6995011337868481,
+      "grad_norm": 0.19921875,
+      "learning_rate": 0.0001552808713743728,
+      "loss": 1.0239,
+      "step": 1928
+    },
+    {
+      "epoch": 0.7002267573696145,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.00015523191775792438,
+      "loss": 1.0417,
+      "step": 1930
+    },
+    {
+      "epoch": 0.700952380952381,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.00015518296414147596,
+      "loss": 0.9733,
+      "step": 1932
+    },
+    {
+      "epoch": 0.7016780045351474,
+      "grad_norm": 0.17578125,
+      "learning_rate": 0.00015513401052502754,
+      "loss": 1.0096,
+      "step": 1934
+    },
+    {
+      "epoch": 0.7024036281179138,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.00015508505690857912,
+      "loss": 0.9957,
+      "step": 1936
+    },
+    {
+      "epoch": 0.7031292517006803,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.0001550361032921307,
+      "loss": 0.9878,
+      "step": 1938
+    },
+    {
+      "epoch": 0.7038548752834467,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.00015498714967568232,
+      "loss": 0.9317,
+      "step": 1940
+    },
+    {
+      "epoch": 0.7045804988662131,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.00015493819605923387,
+      "loss": 1.0407,
+      "step": 1942
+    },
+    {
+      "epoch": 0.7053061224489796,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.00015488924244278545,
+      "loss": 1.0233,
+      "step": 1944
+    },
+    {
+      "epoch": 0.706031746031746,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.00015484028882633706,
+      "loss": 0.9254,
+      "step": 1946
+    },
+    {
+      "epoch": 0.7067573696145125,
+      "grad_norm": 0.220703125,
+      "learning_rate": 0.00015479133520988864,
+      "loss": 0.9546,
+      "step": 1948
+    },
+    {
+      "epoch": 0.7074829931972789,
+      "grad_norm": 0.208984375,
+      "learning_rate": 0.00015474238159344023,
+      "loss": 0.9609,
+      "step": 1950
+    },
+    {
+      "epoch": 0.7082086167800453,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.0001546934279769918,
+      "loss": 1.0335,
+      "step": 1952
+    },
+    {
+      "epoch": 0.7089342403628118,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.0001546444743605434,
+      "loss": 0.952,
+      "step": 1954
+    },
+    {
+      "epoch": 0.7096598639455782,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.00015459552074409497,
+      "loss": 0.9557,
+      "step": 1956
+    },
+    {
+      "epoch": 0.7103854875283446,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.00015454656712764656,
+      "loss": 0.967,
+      "step": 1958
+    },
+    {
+      "epoch": 0.7111111111111111,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.00015449761351119816,
+      "loss": 1.0578,
+      "step": 1960
+    },
+    {
+      "epoch": 0.7118367346938775,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.00015444865989474972,
+      "loss": 1.0424,
+      "step": 1962
+    },
+    {
+      "epoch": 0.7125623582766439,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.0001543997062783013,
+      "loss": 1.0075,
+      "step": 1964
+    },
+    {
+      "epoch": 0.7132879818594104,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.0001543507526618529,
+      "loss": 0.9869,
+      "step": 1966
+    },
+    {
+      "epoch": 0.7140136054421768,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.0001543017990454045,
+      "loss": 0.9439,
+      "step": 1968
+    },
+    {
+      "epoch": 0.7147392290249434,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.00015425284542895605,
+      "loss": 1.004,
+      "step": 1970
+    },
+    {
+      "epoch": 0.7154648526077098,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.00015420389181250766,
+      "loss": 0.9863,
+      "step": 1972
+    },
+    {
+      "epoch": 0.7161904761904762,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.00015415493819605924,
+      "loss": 0.9668,
+      "step": 1974
+    },
+    {
+      "epoch": 0.7169160997732427,
+      "grad_norm": 0.2373046875,
+      "learning_rate": 0.00015410598457961082,
+      "loss": 0.9949,
+      "step": 1976
+    },
+    {
+      "epoch": 0.7176417233560091,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.0001540570309631624,
+      "loss": 0.959,
+      "step": 1978
+    },
+    {
+      "epoch": 0.7183673469387755,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.000154008077346714,
+      "loss": 1.0601,
+      "step": 1980
+    },
+    {
+      "epoch": 0.719092970521542,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.00015395912373026557,
+      "loss": 1.0484,
+      "step": 1982
+    },
+    {
+      "epoch": 0.7198185941043084,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.00015391017011381715,
+      "loss": 0.9602,
+      "step": 1984
+    },
+    {
+      "epoch": 0.7205442176870749,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.00015386121649736876,
+      "loss": 0.9339,
+      "step": 1986
+    },
+    {
+      "epoch": 0.7212698412698413,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.00015381226288092034,
+      "loss": 1.0522,
+      "step": 1988
+    },
+    {
+      "epoch": 0.7219954648526077,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.0001537633092644719,
+      "loss": 0.9165,
+      "step": 1990
+    },
+    {
+      "epoch": 0.7227210884353742,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.0001537143556480235,
+      "loss": 0.9953,
+      "step": 1992
+    },
+    {
+      "epoch": 0.7234467120181406,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.0001536654020315751,
+      "loss": 0.9712,
+      "step": 1994
+    },
+    {
+      "epoch": 0.724172335600907,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.00015361644841512667,
+      "loss": 1.0144,
+      "step": 1996
+    },
+    {
+      "epoch": 0.7248979591836735,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.00015356749479867826,
+      "loss": 1.0188,
+      "step": 1998
+    },
+    {
+      "epoch": 0.7256235827664399,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.00015351854118222984,
+      "loss": 1.0385,
+      "step": 2000
+    },
+    {
+      "epoch": 0.7256235827664399,
+      "eval_loss": 1.0189679861068726,
+      "eval_runtime": 100.5257,
+      "eval_samples_per_second": 17.906,
+      "eval_steps_per_second": 0.567,
+      "step": 2000
+    },
+    {
+      "epoch": 0.7263492063492063,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.00015346958756578142,
+      "loss": 0.9979,
+      "step": 2002
+    },
+    {
+      "epoch": 0.7270748299319728,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.000153420633949333,
+      "loss": 0.9475,
+      "step": 2004
+    },
+    {
+      "epoch": 0.7278004535147392,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.0001533716803328846,
+      "loss": 1.0247,
+      "step": 2006
+    },
+    {
+      "epoch": 0.7285260770975057,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.0001533227267164362,
+      "loss": 1.0989,
+      "step": 2008
+    },
+    {
+      "epoch": 0.7292517006802721,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.00015327377309998775,
+      "loss": 0.9775,
+      "step": 2010
+    },
+    {
+      "epoch": 0.7299773242630385,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.00015322481948353936,
+      "loss": 1.0218,
+      "step": 2012
+    },
+    {
+      "epoch": 0.730702947845805,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.00015317586586709094,
+      "loss": 0.943,
+      "step": 2014
+    },
+    {
+      "epoch": 0.7314285714285714,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.00015312691225064252,
+      "loss": 0.9455,
+      "step": 2016
+    },
+    {
+      "epoch": 0.7321541950113378,
+      "grad_norm": 0.177734375,
+      "learning_rate": 0.0001530779586341941,
+      "loss": 0.9995,
+      "step": 2018
+    },
+    {
+      "epoch": 0.7328798185941043,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.0001530290050177457,
+      "loss": 1.0052,
+      "step": 2020
+    },
+    {
+      "epoch": 0.7336054421768707,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.00015298005140129727,
+      "loss": 1.0352,
+      "step": 2022
+    },
+    {
+      "epoch": 0.7343310657596371,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.00015293109778484885,
+      "loss": 1.0545,
+      "step": 2024
+    },
+    {
+      "epoch": 0.7350566893424036,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.00015288214416840046,
+      "loss": 1.0652,
+      "step": 2026
+    },
+    {
+      "epoch": 0.73578231292517,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.00015283319055195204,
+      "loss": 1.0658,
+      "step": 2028
+    },
+    {
+      "epoch": 0.7365079365079366,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.0001527842369355036,
+      "loss": 1.0105,
+      "step": 2030
+    },
+    {
+      "epoch": 0.737233560090703,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.0001527352833190552,
+      "loss": 1.0369,
+      "step": 2032
+    },
+    {
+      "epoch": 0.7379591836734694,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.0001526863297026068,
+      "loss": 1.0044,
+      "step": 2034
+    },
+    {
+      "epoch": 0.7386848072562359,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.00015263737608615837,
+      "loss": 0.9738,
+      "step": 2036
+    },
+    {
+      "epoch": 0.7394104308390023,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.00015258842246970996,
+      "loss": 1.0011,
+      "step": 2038
+    },
+    {
+      "epoch": 0.7401360544217687,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.00015253946885326154,
+      "loss": 0.8985,
+      "step": 2040
+    },
+    {
+      "epoch": 0.7408616780045352,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.00015249051523681312,
+      "loss": 1.0394,
+      "step": 2042
+    },
+    {
+      "epoch": 0.7415873015873016,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.0001524415616203647,
+      "loss": 1.02,
+      "step": 2044
+    },
+    {
+      "epoch": 0.742312925170068,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.0001523926080039163,
+      "loss": 0.9851,
+      "step": 2046
+    },
+    {
+      "epoch": 0.7430385487528345,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.00015234365438746787,
+      "loss": 1.0116,
+      "step": 2048
+    },
+    {
+      "epoch": 0.7437641723356009,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.00015229470077101945,
+      "loss": 1.041,
+      "step": 2050
+    },
+    {
+      "epoch": 0.7444897959183674,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.00015224574715457106,
+      "loss": 0.9993,
+      "step": 2052
+    },
+    {
+      "epoch": 0.7452154195011338,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.00015219679353812264,
+      "loss": 0.9596,
+      "step": 2054
+    },
+    {
+      "epoch": 0.7459410430839002,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.00015214783992167422,
+      "loss": 1.0042,
+      "step": 2056
+    },
+    {
+      "epoch": 0.7466666666666667,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.0001520988863052258,
+      "loss": 0.9341,
+      "step": 2058
+    },
+    {
+      "epoch": 0.7473922902494331,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.0001520499326887774,
+      "loss": 0.9923,
+      "step": 2060
+    },
+    {
+      "epoch": 0.7481179138321995,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.00015200097907232897,
+      "loss": 1.0059,
+      "step": 2062
+    },
+    {
+      "epoch": 0.748843537414966,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.00015195202545588055,
+      "loss": 0.9428,
+      "step": 2064
+    },
+    {
+      "epoch": 0.7495691609977324,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.00015190307183943216,
+      "loss": 0.9996,
+      "step": 2066
+    },
+    {
+      "epoch": 0.7502947845804989,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.00015185411822298372,
+      "loss": 0.9503,
+      "step": 2068
+    },
+    {
+      "epoch": 0.7510204081632653,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.0001518051646065353,
+      "loss": 1.0125,
+      "step": 2070
+    },
+    {
+      "epoch": 0.7517460317460317,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.0001517562109900869,
+      "loss": 0.9977,
+      "step": 2072
+    },
+    {
+      "epoch": 0.7524716553287982,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.0001517072573736385,
+      "loss": 1.0281,
+      "step": 2074
+    },
+    {
+      "epoch": 0.7531972789115646,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.00015165830375719007,
+      "loss": 0.9786,
+      "step": 2076
+    },
+    {
+      "epoch": 0.753922902494331,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.00015160935014074166,
+      "loss": 0.9559,
+      "step": 2078
+    },
+    {
+      "epoch": 0.7546485260770975,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.00015156039652429324,
+      "loss": 0.9641,
+      "step": 2080
+    },
+    {
+      "epoch": 0.7553741496598639,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.00015151144290784482,
+      "loss": 0.9441,
+      "step": 2082
+    },
+    {
+      "epoch": 0.7560997732426303,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.0001514624892913964,
+      "loss": 0.999,
+      "step": 2084
+    },
+    {
+      "epoch": 0.7568253968253968,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 0.000151413535674948,
+      "loss": 1.0128,
+      "step": 2086
+    },
+    {
+      "epoch": 0.7575510204081632,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.00015136458205849957,
+      "loss": 1.0748,
+      "step": 2088
+    },
+    {
+      "epoch": 0.7582766439909298,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.00015131562844205115,
+      "loss": 0.8835,
+      "step": 2090
+    },
+    {
+      "epoch": 0.7590022675736962,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.00015126667482560276,
+      "loss": 0.9798,
+      "step": 2092
+    },
+    {
+      "epoch": 0.7597278911564626,
+      "grad_norm": 0.173828125,
+      "learning_rate": 0.00015121772120915434,
+      "loss": 0.961,
+      "step": 2094
+    },
+    {
+      "epoch": 0.7604535147392291,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.00015116876759270592,
+      "loss": 0.9972,
+      "step": 2096
+    },
+    {
+      "epoch": 0.7611791383219955,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.0001511198139762575,
+      "loss": 0.9814,
+      "step": 2098
+    },
+    {
+      "epoch": 0.7619047619047619,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.0001510708603598091,
+      "loss": 0.9646,
+      "step": 2100
+    },
+    {
+      "epoch": 0.7619047619047619,
+      "eval_loss": 1.0186907052993774,
+      "eval_runtime": 95.5861,
+      "eval_samples_per_second": 18.831,
+      "eval_steps_per_second": 0.596,
+      "step": 2100
+    },
+    {
+      "epoch": 0.7626303854875284,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.00015102190674336067,
+      "loss": 1.0291,
+      "step": 2102
+    },
+    {
+      "epoch": 0.7633560090702948,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.00015097295312691225,
+      "loss": 0.944,
+      "step": 2104
+    },
+    {
+      "epoch": 0.7640816326530612,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.00015092399951046386,
+      "loss": 0.9971,
+      "step": 2106
+    },
+    {
+      "epoch": 0.7648072562358277,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.00015087504589401542,
+      "loss": 0.9816,
+      "step": 2108
+    },
+    {
+      "epoch": 0.7655328798185941,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.000150826092277567,
+      "loss": 1.0105,
+      "step": 2110
+    },
+    {
+      "epoch": 0.7662585034013606,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.0001507771386611186,
+      "loss": 1.0059,
+      "step": 2112
+    },
+    {
+      "epoch": 0.766984126984127,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.0001507281850446702,
+      "loss": 0.9797,
+      "step": 2114
+    },
+    {
+      "epoch": 0.7677097505668934,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.00015067923142822175,
+      "loss": 0.9623,
+      "step": 2116
+    },
+    {
+      "epoch": 0.7684353741496599,
+      "grad_norm": 0.17578125,
+      "learning_rate": 0.00015063027781177335,
+      "loss": 1.0352,
+      "step": 2118
+    },
+    {
+      "epoch": 0.7691609977324263,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.00015058132419532494,
+      "loss": 1.0687,
+      "step": 2120
+    },
+    {
+      "epoch": 0.7698866213151927,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.00015053237057887652,
+      "loss": 0.9323,
+      "step": 2122
+    },
+    {
+      "epoch": 0.7706122448979592,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.0001504834169624281,
+      "loss": 0.9436,
+      "step": 2124
+    },
+    {
+      "epoch": 0.7713378684807256,
+      "grad_norm": 0.18359375,
+      "learning_rate": 0.00015043446334597968,
+      "loss": 1.0228,
+      "step": 2126
+    },
+    {
+      "epoch": 0.7720634920634921,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.00015038550972953127,
+      "loss": 0.9885,
+      "step": 2128
+    },
+    {
+      "epoch": 0.7727891156462585,
+      "grad_norm": 0.173828125,
+      "learning_rate": 0.00015033655611308285,
+      "loss": 0.9654,
+      "step": 2130
+    },
+    {
+      "epoch": 0.7735147392290249,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.00015028760249663446,
+      "loss": 1.1141,
+      "step": 2132
+    },
+    {
+      "epoch": 0.7742403628117914,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.00015023864888018604,
+      "loss": 0.9624,
+      "step": 2134
+    },
+    {
+      "epoch": 0.7749659863945578,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.0001501896952637376,
+      "loss": 0.9589,
+      "step": 2136
+    },
+    {
+      "epoch": 0.7756916099773242,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 0.0001501407416472892,
+      "loss": 1.0223,
+      "step": 2138
+    },
+    {
+      "epoch": 0.7764172335600907,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.0001500917880308408,
+      "loss": 0.9818,
+      "step": 2140
+    },
+    {
+      "epoch": 0.7771428571428571,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.00015004283441439237,
+      "loss": 0.9146,
+      "step": 2142
+    },
+    {
+      "epoch": 0.7778684807256235,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.00014999388079794395,
+      "loss": 0.9904,
+      "step": 2144
+    },
+    {
+      "epoch": 0.77859410430839,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.00014994492718149553,
+      "loss": 0.9883,
+      "step": 2146
+    },
+    {
+      "epoch": 0.7793197278911564,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.00014989597356504712,
+      "loss": 1.0015,
+      "step": 2148
+    },
+    {
+      "epoch": 0.780045351473923,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.0001498470199485987,
+      "loss": 1.0113,
+      "step": 2150
+    },
+    {
+      "epoch": 0.7807709750566894,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.0001497980663321503,
+      "loss": 0.9982,
+      "step": 2152
+    },
+    {
+      "epoch": 0.7814965986394558,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.0001497491127157019,
+      "loss": 1.0084,
+      "step": 2154
+    },
+    {
+      "epoch": 0.7822222222222223,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.00014970015909925345,
+      "loss": 0.9613,
+      "step": 2156
+    },
+    {
+      "epoch": 0.7829478458049887,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.00014965120548280505,
+      "loss": 1.0032,
+      "step": 2158
+    },
+    {
+      "epoch": 0.7836734693877551,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.00014960225186635664,
+      "loss": 0.979,
+      "step": 2160
+    },
+    {
+      "epoch": 0.7843990929705216,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.00014955329824990822,
+      "loss": 0.9816,
+      "step": 2162
+    },
+    {
+      "epoch": 0.785124716553288,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.0001495043446334598,
+      "loss": 0.9398,
+      "step": 2164
+    },
+    {
+      "epoch": 0.7858503401360544,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.00014945539101701138,
+      "loss": 1.0336,
+      "step": 2166
+    },
+    {
+      "epoch": 0.7865759637188209,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.00014940643740056297,
+      "loss": 0.9288,
+      "step": 2168
+    },
+    {
+      "epoch": 0.7873015873015873,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.00014935748378411455,
+      "loss": 1.0003,
+      "step": 2170
+    },
+    {
+      "epoch": 0.7880272108843538,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.00014930853016766616,
+      "loss": 0.9418,
+      "step": 2172
+    },
+    {
+      "epoch": 0.7887528344671202,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.00014925957655121774,
+      "loss": 1.0262,
+      "step": 2174
+    },
+    {
+      "epoch": 0.7894784580498866,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.0001492106229347693,
+      "loss": 1.0496,
+      "step": 2176
+    },
+    {
+      "epoch": 0.7902040816326531,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.0001491616693183209,
+      "loss": 0.937,
+      "step": 2178
+    },
+    {
+      "epoch": 0.7909297052154195,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.0001491127157018725,
+      "loss": 0.9422,
+      "step": 2180
+    },
+    {
+      "epoch": 0.7916553287981859,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.00014906376208542407,
+      "loss": 0.9921,
+      "step": 2182
+    },
+    {
+      "epoch": 0.7923809523809524,
+      "grad_norm": 0.173828125,
+      "learning_rate": 0.00014901480846897565,
+      "loss": 0.9662,
+      "step": 2184
+    },
+    {
+      "epoch": 0.7931065759637188,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 0.00014896585485252723,
+      "loss": 1.0241,
+      "step": 2186
+    },
+    {
+      "epoch": 0.7938321995464852,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.00014891690123607882,
+      "loss": 1.0144,
+      "step": 2188
+    },
+    {
+      "epoch": 0.7945578231292517,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.0001488679476196304,
+      "loss": 0.9861,
+      "step": 2190
+    },
+    {
+      "epoch": 0.7952834467120181,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.000148818994003182,
+      "loss": 0.9249,
+      "step": 2192
+    },
+    {
+      "epoch": 0.7960090702947846,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.00014877004038673356,
+      "loss": 0.9389,
+      "step": 2194
+    },
+    {
+      "epoch": 0.796734693877551,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.00014872108677028515,
+      "loss": 0.9589,
+      "step": 2196
+    },
+    {
+      "epoch": 0.7974603174603174,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.00014867213315383675,
+      "loss": 1.0548,
+      "step": 2198
+    },
+    {
+      "epoch": 0.7981859410430839,
+      "grad_norm": 0.189453125,
+      "learning_rate": 0.00014862317953738834,
+      "loss": 1.0526,
+      "step": 2200
+    },
+    {
+      "epoch": 0.7981859410430839,
+      "eval_loss": 1.0182089805603027,
+      "eval_runtime": 95.9846,
+      "eval_samples_per_second": 18.753,
+      "eval_steps_per_second": 0.594,
+      "step": 2200
+    },
+    {
+      "epoch": 0.7989115646258503,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.00014857422592093992,
+      "loss": 1.0364,
+      "step": 2202
+    },
+    {
+      "epoch": 0.7996371882086167,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.0001485252723044915,
+      "loss": 0.9324,
+      "step": 2204
+    },
+    {
+      "epoch": 0.8003628117913832,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.00014847631868804308,
+      "loss": 0.936,
+      "step": 2206
+    },
+    {
+      "epoch": 0.8010884353741496,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.00014842736507159467,
+      "loss": 1.0559,
+      "step": 2208
+    },
+    {
+      "epoch": 0.8018140589569162,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.00014837841145514625,
+      "loss": 1.0164,
+      "step": 2210
+    },
+    {
+      "epoch": 0.8025396825396826,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.00014832945783869786,
+      "loss": 1.0198,
+      "step": 2212
+    },
+    {
+      "epoch": 0.803265306122449,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.0001482805042222494,
+      "loss": 1.0224,
+      "step": 2214
+    },
+    {
+      "epoch": 0.8039909297052155,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.000148231550605801,
+      "loss": 0.9483,
+      "step": 2216
+    },
+    {
+      "epoch": 0.8047165532879819,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.0001481825969893526,
+      "loss": 1.0488,
+      "step": 2218
+    },
+    {
+      "epoch": 0.8054421768707483,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.0001481336433729042,
+      "loss": 0.9822,
+      "step": 2220
+    },
+    {
+      "epoch": 0.8061678004535148,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.00014808468975645577,
+      "loss": 0.969,
+      "step": 2222
+    },
+    {
+      "epoch": 0.8068934240362812,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.00014803573614000735,
+      "loss": 0.9803,
+      "step": 2224
+    },
+    {
+      "epoch": 0.8076190476190476,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.00014798678252355893,
+      "loss": 0.9859,
+      "step": 2226
+    },
+    {
+      "epoch": 0.8083446712018141,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.00014793782890711052,
+      "loss": 1.0081,
+      "step": 2228
+    },
+    {
+      "epoch": 0.8090702947845805,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.0001478888752906621,
+      "loss": 0.9469,
+      "step": 2230
+    },
+    {
+      "epoch": 0.809795918367347,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.0001478399216742137,
+      "loss": 1.001,
+      "step": 2232
+    },
+    {
+      "epoch": 0.8105215419501134,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.00014779096805776526,
+      "loss": 1.0429,
+      "step": 2234
+    },
+    {
+      "epoch": 0.8112471655328798,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.00014774201444131684,
+      "loss": 0.9483,
+      "step": 2236
+    },
+    {
+      "epoch": 0.8119727891156463,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.00014769306082486845,
+      "loss": 1.0031,
+      "step": 2238
+    },
+    {
+      "epoch": 0.8126984126984127,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.00014764410720842004,
+      "loss": 0.9813,
+      "step": 2240
+    },
+    {
+      "epoch": 0.8134240362811791,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.00014759515359197162,
+      "loss": 0.986,
+      "step": 2242
+    },
+    {
+      "epoch": 0.8141496598639456,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.0001475461999755232,
+      "loss": 1.0328,
+      "step": 2244
+    },
+    {
+      "epoch": 0.814875283446712,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.00014749724635907478,
+      "loss": 1.032,
+      "step": 2246
+    },
+    {
+      "epoch": 0.8156009070294784,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.00014744829274262637,
+      "loss": 0.9657,
+      "step": 2248
+    },
+    {
+      "epoch": 0.8163265306122449,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.00014739933912617795,
+      "loss": 0.9797,
+      "step": 2250
+    },
+    {
+      "epoch": 0.8170521541950113,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 0.00014735038550972956,
+      "loss": 0.9919,
+      "step": 2252
+    },
+    {
+      "epoch": 0.8177777777777778,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.0001473014318932811,
+      "loss": 1.0706,
+      "step": 2254
+    },
+    {
+      "epoch": 0.8185034013605442,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.0001472524782768327,
+      "loss": 0.9914,
+      "step": 2256
+    },
+    {
+      "epoch": 0.8192290249433106,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.0001472035246603843,
+      "loss": 1.0764,
+      "step": 2258
+    },
+    {
+      "epoch": 0.8199546485260771,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.0001471545710439359,
+      "loss": 0.973,
+      "step": 2260
+    },
+    {
+      "epoch": 0.8206802721088435,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.00014710561742748744,
+      "loss": 0.9797,
+      "step": 2262
+    },
+    {
+      "epoch": 0.8214058956916099,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.00014705666381103905,
+      "loss": 0.9685,
+      "step": 2264
+    },
+    {
+      "epoch": 0.8221315192743764,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.00014700771019459063,
+      "loss": 1.0145,
+      "step": 2266
+    },
+    {
+      "epoch": 0.8228571428571428,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.00014695875657814222,
+      "loss": 0.9714,
+      "step": 2268
+    },
+    {
+      "epoch": 0.8235827664399092,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.0001469098029616938,
+      "loss": 0.9954,
+      "step": 2270
+    },
+    {
+      "epoch": 0.8243083900226758,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.00014686084934524538,
+      "loss": 1.1002,
+      "step": 2272
+    },
+    {
+      "epoch": 0.8250340136054422,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.00014681189572879696,
+      "loss": 0.9418,
+      "step": 2274
+    },
+    {
+      "epoch": 0.8257596371882087,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.00014676294211234854,
+      "loss": 1.044,
+      "step": 2276
+    },
+    {
+      "epoch": 0.8264852607709751,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.00014671398849590015,
+      "loss": 1.0346,
+      "step": 2278
+    },
+    {
+      "epoch": 0.8272108843537415,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 0.00014666503487945174,
+      "loss": 1.027,
+      "step": 2280
+    },
+    {
+      "epoch": 0.827936507936508,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.0001466160812630033,
+      "loss": 0.9886,
+      "step": 2282
+    },
+    {
+      "epoch": 0.8286621315192744,
+      "grad_norm": 0.1884765625,
+      "learning_rate": 0.0001465671276465549,
+      "loss": 1.0345,
+      "step": 2284
+    },
+    {
+      "epoch": 0.8293877551020408,
+      "grad_norm": 0.177734375,
+      "learning_rate": 0.00014651817403010648,
+      "loss": 1.0252,
+      "step": 2286
+    },
+    {
+      "epoch": 0.8301133786848073,
+      "grad_norm": 0.1865234375,
+      "learning_rate": 0.00014646922041365807,
+      "loss": 1.0475,
+      "step": 2288
+    },
+    {
+      "epoch": 0.8308390022675737,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.00014642026679720965,
+      "loss": 1.0286,
+      "step": 2290
+    },
+    {
+      "epoch": 0.8315646258503402,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.00014637131318076123,
+      "loss": 0.9599,
+      "step": 2292
+    },
+    {
+      "epoch": 0.8322902494331066,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.0001463223595643128,
+      "loss": 1.0277,
+      "step": 2294
+    },
+    {
+      "epoch": 0.833015873015873,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.0001462734059478644,
+      "loss": 0.9699,
+      "step": 2296
+    },
+    {
+      "epoch": 0.8337414965986395,
+      "grad_norm": 0.193359375,
+      "learning_rate": 0.000146224452331416,
+      "loss": 0.9157,
+      "step": 2298
+    },
+    {
+      "epoch": 0.8344671201814059,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.00014617549871496759,
+      "loss": 1.0078,
+      "step": 2300
+    },
+    {
+      "epoch": 0.8344671201814059,
+      "eval_loss": 1.018009901046753,
+      "eval_runtime": 97.4465,
+      "eval_samples_per_second": 18.472,
+      "eval_steps_per_second": 0.585,
+      "step": 2300
+    },
+    {
+      "epoch": 0.8351927437641723,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.00014612654509851914,
+      "loss": 0.9975,
+      "step": 2302
+    },
+    {
+      "epoch": 0.8359183673469388,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.00014607759148207075,
+      "loss": 1.0193,
+      "step": 2304
+    },
+    {
+      "epoch": 0.8366439909297052,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.00014602863786562233,
+      "loss": 0.9847,
+      "step": 2306
+    },
+    {
+      "epoch": 0.8373696145124716,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.00014597968424917392,
+      "loss": 1.0365,
+      "step": 2308
+    },
+    {
+      "epoch": 0.8380952380952381,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.0001459307306327255,
+      "loss": 0.9204,
+      "step": 2310
+    },
+    {
+      "epoch": 0.8388208616780045,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.00014588177701627708,
+      "loss": 0.973,
+      "step": 2312
+    },
+    {
+      "epoch": 0.839546485260771,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.00014583282339982866,
+      "loss": 1.0135,
+      "step": 2314
+    },
+    {
+      "epoch": 0.8402721088435374,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.00014578386978338024,
+      "loss": 0.9772,
+      "step": 2316
+    },
+    {
+      "epoch": 0.8409977324263038,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.00014573491616693185,
+      "loss": 0.9281,
+      "step": 2318
+    },
+    {
+      "epoch": 0.8417233560090703,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.00014568596255048344,
+      "loss": 1.0221,
+      "step": 2320
+    },
+    {
+      "epoch": 0.8424489795918367,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 0.000145637008934035,
+      "loss": 0.972,
+      "step": 2322
+    },
+    {
+      "epoch": 0.8431746031746031,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.0001455880553175866,
+      "loss": 1.0392,
+      "step": 2324
+    },
+    {
+      "epoch": 0.8439002267573696,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 0.00014553910170113818,
+      "loss": 1.0201,
+      "step": 2326
+    },
+    {
+      "epoch": 0.844625850340136,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.00014549014808468977,
+      "loss": 0.9544,
+      "step": 2328
+    },
+    {
+      "epoch": 0.8453514739229024,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.00014544119446824135,
+      "loss": 0.9848,
+      "step": 2330
+    },
+    {
+      "epoch": 0.846077097505669,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.00014539224085179293,
+      "loss": 0.9847,
+      "step": 2332
+    },
+    {
+      "epoch": 0.8468027210884354,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.0001453432872353445,
+      "loss": 1.0878,
+      "step": 2334
+    },
+    {
+      "epoch": 0.8475283446712019,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.0001452943336188961,
+      "loss": 0.9742,
+      "step": 2336
+    },
+    {
+      "epoch": 0.8482539682539683,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.0001452453800024477,
+      "loss": 1.06,
+      "step": 2338
+    },
+    {
+      "epoch": 0.8489795918367347,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.00014519642638599926,
+      "loss": 0.9542,
+      "step": 2340
+    },
+    {
+      "epoch": 0.8497052154195012,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.00014514747276955084,
+      "loss": 0.9095,
+      "step": 2342
+    },
+    {
+      "epoch": 0.8504308390022676,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.00014509851915310245,
+      "loss": 1.0327,
+      "step": 2344
+    },
+    {
+      "epoch": 0.851156462585034,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.00014504956553665403,
+      "loss": 0.932,
+      "step": 2346
+    },
+    {
+      "epoch": 0.8518820861678005,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.00014500061192020562,
+      "loss": 0.9619,
+      "step": 2348
+    },
+    {
+      "epoch": 0.8526077097505669,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.0001449516583037572,
+      "loss": 0.9473,
+      "step": 2350
+    },
+    {
+      "epoch": 0.8533333333333334,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.00014490270468730878,
+      "loss": 1.0192,
+      "step": 2352
+    },
+    {
+      "epoch": 0.8540589569160998,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.00014485375107086036,
+      "loss": 0.9686,
+      "step": 2354
+    },
+    {
+      "epoch": 0.8547845804988662,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.00014480479745441194,
+      "loss": 1.0355,
+      "step": 2356
+    },
+    {
+      "epoch": 0.8555102040816327,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.00014475584383796355,
+      "loss": 0.9756,
+      "step": 2358
+    },
+    {
+      "epoch": 0.8562358276643991,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.0001447068902215151,
+      "loss": 1.0697,
+      "step": 2360
+    },
+    {
+      "epoch": 0.8569614512471655,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.0001446579366050667,
+      "loss": 0.9959,
+      "step": 2362
+    },
+    {
+      "epoch": 0.857687074829932,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.0001446089829886183,
+      "loss": 0.984,
+      "step": 2364
+    },
+    {
+      "epoch": 0.8584126984126984,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.00014456002937216988,
+      "loss": 0.9598,
+      "step": 2366
+    },
+    {
+      "epoch": 0.8591383219954648,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.00014451107575572147,
+      "loss": 1.0026,
+      "step": 2368
+    },
+    {
+      "epoch": 0.8598639455782313,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.00014446212213927305,
+      "loss": 0.9883,
+      "step": 2370
+    },
+    {
+      "epoch": 0.8605895691609977,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.00014441316852282463,
+      "loss": 1.0565,
+      "step": 2372
+    },
+    {
+      "epoch": 0.8613151927437642,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.0001443642149063762,
+      "loss": 0.9496,
+      "step": 2374
+    },
+    {
+      "epoch": 0.8620408163265306,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.0001443152612899278,
+      "loss": 0.961,
+      "step": 2376
+    },
+    {
+      "epoch": 0.862766439909297,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.0001442663076734794,
+      "loss": 0.9518,
+      "step": 2378
+    },
+    {
+      "epoch": 0.8634920634920635,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.00014421735405703096,
+      "loss": 0.9264,
+      "step": 2380
+    },
+    {
+      "epoch": 0.8642176870748299,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.00014416840044058254,
+      "loss": 1.0389,
+      "step": 2382
+    },
+    {
+      "epoch": 0.8649433106575963,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.00014411944682413415,
+      "loss": 1.0283,
+      "step": 2384
+    },
+    {
+      "epoch": 0.8656689342403628,
+      "grad_norm": 0.193359375,
+      "learning_rate": 0.00014407049320768573,
+      "loss": 1.0456,
+      "step": 2386
+    },
+    {
+      "epoch": 0.8663945578231292,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.00014402153959123732,
+      "loss": 1.007,
+      "step": 2388
+    },
+    {
+      "epoch": 0.8671201814058956,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.0001439725859747889,
+      "loss": 1.0643,
+      "step": 2390
+    },
+    {
+      "epoch": 0.8678458049886622,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.00014392363235834048,
+      "loss": 1.0052,
+      "step": 2392
+    },
+    {
+      "epoch": 0.8685714285714285,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.00014387467874189206,
+      "loss": 0.9164,
+      "step": 2394
+    },
+    {
+      "epoch": 0.8692970521541951,
+      "grad_norm": 0.181640625,
+      "learning_rate": 0.00014382572512544364,
+      "loss": 0.9513,
+      "step": 2396
+    },
+    {
+      "epoch": 0.8700226757369615,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.00014377677150899525,
+      "loss": 0.9772,
+      "step": 2398
+    },
+    {
+      "epoch": 0.8707482993197279,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.0001437278178925468,
+      "loss": 1.0452,
+      "step": 2400
+    },
+    {
+      "epoch": 0.8707482993197279,
+      "eval_loss": 1.0175837278366089,
+      "eval_runtime": 93.47,
+      "eval_samples_per_second": 19.258,
+      "eval_steps_per_second": 0.61,
+      "step": 2400
+    },
+    {
+      "epoch": 0.8714739229024944,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.0001436788642760984,
+      "loss": 0.9611,
+      "step": 2402
+    },
+    {
+      "epoch": 0.8721995464852608,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.00014362991065965,
+      "loss": 0.9569,
+      "step": 2404
+    },
+    {
+      "epoch": 0.8729251700680272,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.00014358095704320158,
+      "loss": 1.0252,
+      "step": 2406
+    },
+    {
+      "epoch": 0.8736507936507937,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.00014353200342675314,
+      "loss": 1.0249,
+      "step": 2408
+    },
+    {
+      "epoch": 0.8743764172335601,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.00014348304981030475,
+      "loss": 1.034,
+      "step": 2410
+    },
+    {
+      "epoch": 0.8751020408163265,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.00014343409619385633,
+      "loss": 0.9909,
+      "step": 2412
+    },
+    {
+      "epoch": 0.875827664399093,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.0001433851425774079,
+      "loss": 0.9933,
+      "step": 2414
+    },
+    {
+      "epoch": 0.8765532879818594,
+      "grad_norm": 0.21875,
+      "learning_rate": 0.0001433361889609595,
+      "loss": 0.8944,
+      "step": 2416
+    },
+    {
+      "epoch": 0.8772789115646259,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.00014328723534451108,
+      "loss": 1.0102,
+      "step": 2418
+    },
+    {
+      "epoch": 0.8780045351473923,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.00014323828172806266,
+      "loss": 0.959,
+      "step": 2420
+    },
+    {
+      "epoch": 0.8787301587301587,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.00014318932811161424,
+      "loss": 0.9532,
+      "step": 2422
+    },
+    {
+      "epoch": 0.8794557823129252,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.00014314037449516585,
+      "loss": 0.9811,
+      "step": 2424
+    },
+    {
+      "epoch": 0.8801814058956916,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.00014309142087871743,
+      "loss": 0.9697,
+      "step": 2426
+    },
+    {
+      "epoch": 0.880907029478458,
+      "grad_norm": 0.18359375,
+      "learning_rate": 0.000143042467262269,
+      "loss": 1.0135,
+      "step": 2428
+    },
+    {
+      "epoch": 0.8816326530612245,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.0001429935136458206,
+      "loss": 1.0377,
+      "step": 2430
+    },
+    {
+      "epoch": 0.8823582766439909,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.00014294456002937218,
+      "loss": 0.9046,
+      "step": 2432
+    },
+    {
+      "epoch": 0.8830839002267574,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.00014289560641292376,
+      "loss": 0.9787,
+      "step": 2434
+    },
+    {
+      "epoch": 0.8838095238095238,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.00014284665279647534,
+      "loss": 1.0511,
+      "step": 2436
+    },
+    {
+      "epoch": 0.8845351473922902,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.00014279769918002693,
+      "loss": 0.9777,
+      "step": 2438
+    },
+    {
+      "epoch": 0.8852607709750567,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.0001427487455635785,
+      "loss": 0.9929,
+      "step": 2440
+    },
+    {
+      "epoch": 0.8859863945578231,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.0001426997919471301,
+      "loss": 1.077,
+      "step": 2442
+    },
+    {
+      "epoch": 0.8867120181405895,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.0001426508383306817,
+      "loss": 1.0122,
+      "step": 2444
+    },
+    {
+      "epoch": 0.887437641723356,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.00014260188471423328,
+      "loss": 0.9913,
+      "step": 2446
+    },
+    {
+      "epoch": 0.8881632653061224,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.00014255293109778484,
+      "loss": 0.9736,
+      "step": 2448
+    },
+    {
+      "epoch": 0.8888888888888888,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.00014250397748133645,
+      "loss": 0.9983,
+      "step": 2450
+    },
+    {
+      "epoch": 0.8896145124716553,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 0.00014245502386488803,
+      "loss": 1.033,
+      "step": 2452
+    },
+    {
+      "epoch": 0.8903401360544217,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.0001424060702484396,
+      "loss": 1.0804,
+      "step": 2454
+    },
+    {
+      "epoch": 0.8910657596371883,
+      "grad_norm": 0.17578125,
+      "learning_rate": 0.0001423571166319912,
+      "loss": 0.9584,
+      "step": 2456
+    },
+    {
+      "epoch": 0.8917913832199547,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.00014230816301554278,
+      "loss": 1.0151,
+      "step": 2458
+    },
+    {
+      "epoch": 0.8925170068027211,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.00014225920939909436,
+      "loss": 1.0216,
+      "step": 2460
+    },
+    {
+      "epoch": 0.8932426303854876,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.00014221025578264594,
+      "loss": 1.0113,
+      "step": 2462
+    },
+    {
+      "epoch": 0.893968253968254,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.00014216130216619755,
+      "loss": 1.0047,
+      "step": 2464
+    },
+    {
+      "epoch": 0.8946938775510204,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.00014211234854974913,
+      "loss": 0.9717,
+      "step": 2466
+    },
+    {
+      "epoch": 0.8954195011337869,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.0001420633949333007,
+      "loss": 1.0086,
+      "step": 2468
+    },
+    {
+      "epoch": 0.8961451247165533,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.0001420144413168523,
+      "loss": 0.9738,
+      "step": 2470
+    },
+    {
+      "epoch": 0.8968707482993197,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.00014196548770040388,
+      "loss": 1.0254,
+      "step": 2472
+    },
+    {
+      "epoch": 0.8975963718820862,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.00014191653408395546,
+      "loss": 0.9363,
+      "step": 2474
+    },
+    {
+      "epoch": 0.8983219954648526,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.00014186758046750704,
+      "loss": 1.0181,
+      "step": 2476
+    },
+    {
+      "epoch": 0.8990476190476191,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.00014181862685105863,
+      "loss": 0.9986,
+      "step": 2478
+    },
+    {
+      "epoch": 0.8997732426303855,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.0001417696732346102,
+      "loss": 1.0503,
+      "step": 2480
+    },
+    {
+      "epoch": 0.9004988662131519,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.0001417207196181618,
+      "loss": 1.0875,
+      "step": 2482
+    },
+    {
+      "epoch": 0.9012244897959184,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.0001416717660017134,
+      "loss": 1.0045,
+      "step": 2484
+    },
+    {
+      "epoch": 0.9019501133786848,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.00014162281238526496,
+      "loss": 1.0325,
+      "step": 2486
+    },
+    {
+      "epoch": 0.9026757369614512,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.00014157385876881654,
+      "loss": 0.9717,
+      "step": 2488
+    },
+    {
+      "epoch": 0.9034013605442177,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.00014152490515236815,
+      "loss": 1.0202,
+      "step": 2490
+    },
+    {
+      "epoch": 0.9041269841269841,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.00014147595153591973,
+      "loss": 0.9743,
+      "step": 2492
+    },
+    {
+      "epoch": 0.9048526077097505,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.0001414269979194713,
+      "loss": 0.9956,
+      "step": 2494
+    },
+    {
+      "epoch": 0.905578231292517,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.0001413780443030229,
+      "loss": 0.9782,
+      "step": 2496
+    },
+    {
+      "epoch": 0.9063038548752834,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.00014132909068657448,
+      "loss": 1.0001,
+      "step": 2498
+    },
+    {
+      "epoch": 0.9070294784580499,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.00014128013707012606,
+      "loss": 0.9439,
+      "step": 2500
+    },
+    {
+      "epoch": 0.9070294784580499,
+      "eval_loss": 1.0171304941177368,
+      "eval_runtime": 93.0998,
+      "eval_samples_per_second": 19.334,
+      "eval_steps_per_second": 0.612,
+      "step": 2500
+    },
+    {
+      "epoch": 0.9077551020408163,
+      "grad_norm": 0.181640625,
+      "learning_rate": 0.00014123118345367764,
+      "loss": 1.0005,
+      "step": 2502
+    },
+    {
+      "epoch": 0.9084807256235827,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.00014118222983722925,
+      "loss": 0.9762,
+      "step": 2504
+    },
+    {
+      "epoch": 0.9092063492063492,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.0001411332762207808,
+      "loss": 1.0189,
+      "step": 2506
+    },
+    {
+      "epoch": 0.9099319727891156,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.0001410843226043324,
+      "loss": 1.0226,
+      "step": 2508
+    },
+    {
+      "epoch": 0.910657596371882,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.000141035368987884,
+      "loss": 1.008,
+      "step": 2510
+    },
+    {
+      "epoch": 0.9113832199546485,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.00014098641537143558,
+      "loss": 0.9309,
+      "step": 2512
+    },
+    {
+      "epoch": 0.912108843537415,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.00014093746175498716,
+      "loss": 0.9321,
+      "step": 2514
+    },
+    {
+      "epoch": 0.9128344671201815,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.00014088850813853874,
+      "loss": 1.0389,
+      "step": 2516
+    },
+    {
+      "epoch": 0.9135600907029479,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.00014083955452209033,
+      "loss": 0.9912,
+      "step": 2518
+    },
+    {
+      "epoch": 0.9142857142857143,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.0001407906009056419,
+      "loss": 1.0086,
+      "step": 2520
+    },
+    {
+      "epoch": 0.9150113378684808,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.0001407416472891935,
+      "loss": 1.0779,
+      "step": 2522
+    },
+    {
+      "epoch": 0.9157369614512472,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.0001406926936727451,
+      "loss": 1.0576,
+      "step": 2524
+    },
+    {
+      "epoch": 0.9164625850340136,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.00014064374005629666,
+      "loss": 1.0101,
+      "step": 2526
+    },
+    {
+      "epoch": 0.9171882086167801,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.00014059478643984824,
+      "loss": 1.0382,
+      "step": 2528
+    },
+    {
+      "epoch": 0.9179138321995465,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.00014054583282339985,
+      "loss": 1.0193,
+      "step": 2530
+    },
+    {
+      "epoch": 0.9186394557823129,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.00014049687920695143,
+      "loss": 1.0049,
+      "step": 2532
+    },
+    {
+      "epoch": 0.9193650793650794,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.000140447925590503,
+      "loss": 0.9785,
+      "step": 2534
+    },
+    {
+      "epoch": 0.9200907029478458,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.0001403989719740546,
+      "loss": 0.9122,
+      "step": 2536
+    },
+    {
+      "epoch": 0.9208163265306123,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.00014035001835760618,
+      "loss": 0.9331,
+      "step": 2538
+    },
+    {
+      "epoch": 0.9215419501133787,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.00014030106474115776,
+      "loss": 1.0167,
+      "step": 2540
+    },
+    {
+      "epoch": 0.9222675736961451,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.00014025211112470934,
+      "loss": 0.9788,
+      "step": 2542
+    },
+    {
+      "epoch": 0.9229931972789116,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.00014020315750826095,
+      "loss": 1.0077,
+      "step": 2544
+    },
+    {
+      "epoch": 0.923718820861678,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.0001401542038918125,
+      "loss": 1.0014,
+      "step": 2546
+    },
+    {
+      "epoch": 0.9244444444444444,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.0001401052502753641,
+      "loss": 0.9479,
+      "step": 2548
+    },
+    {
+      "epoch": 0.9251700680272109,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.0001400562966589157,
+      "loss": 1.0187,
+      "step": 2550
+    },
+    {
+      "epoch": 0.9258956916099773,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.00014000734304246728,
+      "loss": 0.9687,
+      "step": 2552
+    },
+    {
+      "epoch": 0.9266213151927437,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.00013995838942601883,
+      "loss": 0.9767,
+      "step": 2554
+    },
+    {
+      "epoch": 0.9273469387755102,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.00013990943580957044,
+      "loss": 1.0263,
+      "step": 2556
+    },
+    {
+      "epoch": 0.9280725623582766,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.00013986048219312203,
+      "loss": 0.979,
+      "step": 2558
+    },
+    {
+      "epoch": 0.9287981859410431,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.0001398115285766736,
+      "loss": 1.0121,
+      "step": 2560
+    },
+    {
+      "epoch": 0.9295238095238095,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.0001397625749602252,
+      "loss": 0.9887,
+      "step": 2562
+    },
+    {
+      "epoch": 0.9302494331065759,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.00013971362134377677,
+      "loss": 0.949,
+      "step": 2564
+    },
+    {
+      "epoch": 0.9309750566893424,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.00013966466772732835,
+      "loss": 0.9792,
+      "step": 2566
+    },
+    {
+      "epoch": 0.9317006802721088,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.00013961571411087994,
+      "loss": 0.9836,
+      "step": 2568
+    },
+    {
+      "epoch": 0.9324263038548752,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.00013956676049443155,
+      "loss": 0.9845,
+      "step": 2570
+    },
+    {
+      "epoch": 0.9331519274376417,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.00013951780687798313,
+      "loss": 0.9387,
+      "step": 2572
+    },
+    {
+      "epoch": 0.9338775510204081,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.00013946885326153468,
+      "loss": 1.0167,
+      "step": 2574
+    },
+    {
+      "epoch": 0.9346031746031747,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.0001394198996450863,
+      "loss": 1.0307,
+      "step": 2576
+    },
+    {
+      "epoch": 0.9353287981859411,
+      "grad_norm": 0.173828125,
+      "learning_rate": 0.00013937094602863788,
+      "loss": 1.0411,
+      "step": 2578
+    },
+    {
+      "epoch": 0.9360544217687075,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.00013932199241218946,
+      "loss": 0.9697,
+      "step": 2580
+    },
+    {
+      "epoch": 0.936780045351474,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.00013927303879574104,
+      "loss": 0.9579,
+      "step": 2582
+    },
+    {
+      "epoch": 0.9375056689342404,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.00013922408517929262,
+      "loss": 0.9208,
+      "step": 2584
+    },
+    {
+      "epoch": 0.9382312925170068,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.0001391751315628442,
+      "loss": 1.0053,
+      "step": 2586
+    },
+    {
+      "epoch": 0.9389569160997733,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.0001391261779463958,
+      "loss": 0.9873,
+      "step": 2588
+    },
+    {
+      "epoch": 0.9396825396825397,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.0001390772243299474,
+      "loss": 0.9857,
+      "step": 2590
+    },
+    {
+      "epoch": 0.9404081632653061,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.00013902827071349898,
+      "loss": 1.0317,
+      "step": 2592
+    },
+    {
+      "epoch": 0.9411337868480726,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.00013897931709705053,
+      "loss": 0.939,
+      "step": 2594
+    },
+    {
+      "epoch": 0.941859410430839,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.00013893036348060214,
+      "loss": 1.006,
+      "step": 2596
+    },
+    {
+      "epoch": 0.9425850340136055,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.00013888140986415373,
+      "loss": 1.083,
+      "step": 2598
+    },
+    {
+      "epoch": 0.9433106575963719,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.0001388324562477053,
+      "loss": 1.0546,
+      "step": 2600
+    },
+    {
+      "epoch": 0.9433106575963719,
+      "eval_loss": 1.016438603401184,
+      "eval_runtime": 97.3894,
+      "eval_samples_per_second": 18.482,
+      "eval_steps_per_second": 0.585,
+      "step": 2600
+    },
+    {
+      "epoch": 0.9440362811791383,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.0001387835026312569,
+      "loss": 0.9987,
+      "step": 2602
+    },
+    {
+      "epoch": 0.9447619047619048,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.00013873454901480847,
+      "loss": 0.9945,
+      "step": 2604
+    },
+    {
+      "epoch": 0.9454875283446712,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.00013868559539836005,
+      "loss": 1.0031,
+      "step": 2606
+    },
+    {
+      "epoch": 0.9462131519274376,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.00013863664178191164,
+      "loss": 1.0133,
+      "step": 2608
+    },
+    {
+      "epoch": 0.9469387755102041,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.00013858768816546325,
+      "loss": 1.0281,
+      "step": 2610
+    },
+    {
+      "epoch": 0.9476643990929705,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.00013853873454901483,
+      "loss": 1.0062,
+      "step": 2612
+    },
+    {
+      "epoch": 0.9483900226757369,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.00013848978093256638,
+      "loss": 0.9394,
+      "step": 2614
+    },
+    {
+      "epoch": 0.9491156462585034,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.000138440827316118,
+      "loss": 0.978,
+      "step": 2616
+    },
+    {
+      "epoch": 0.9498412698412698,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.00013839187369966958,
+      "loss": 0.9216,
+      "step": 2618
+    },
+    {
+      "epoch": 0.9505668934240363,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.00013834292008322116,
+      "loss": 0.9806,
+      "step": 2620
+    },
+    {
+      "epoch": 0.9512925170068027,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.00013829396646677274,
+      "loss": 1.0237,
+      "step": 2622
+    },
+    {
+      "epoch": 0.9520181405895691,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.00013824501285032432,
+      "loss": 0.9156,
+      "step": 2624
+    },
+    {
+      "epoch": 0.9527437641723356,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.0001381960592338759,
+      "loss": 0.9625,
+      "step": 2626
+    },
+    {
+      "epoch": 0.953469387755102,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.0001381471056174275,
+      "loss": 1.0259,
+      "step": 2628
+    },
+    {
+      "epoch": 0.9541950113378684,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.0001380981520009791,
+      "loss": 0.921,
+      "step": 2630
+    },
+    {
+      "epoch": 0.954920634920635,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.00013804919838453065,
+      "loss": 0.9542,
+      "step": 2632
+    },
+    {
+      "epoch": 0.9556462585034013,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.00013800024476808223,
+      "loss": 1.0174,
+      "step": 2634
+    },
+    {
+      "epoch": 0.9563718820861677,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.00013795129115163384,
+      "loss": 1.0429,
+      "step": 2636
+    },
+    {
+      "epoch": 0.9570975056689343,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.00013790233753518543,
+      "loss": 0.9952,
+      "step": 2638
+    },
+    {
+      "epoch": 0.9578231292517007,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.000137853383918737,
+      "loss": 0.957,
+      "step": 2640
+    },
+    {
+      "epoch": 0.9585487528344672,
+      "grad_norm": 0.1884765625,
+      "learning_rate": 0.0001378044303022886,
+      "loss": 1.0364,
+      "step": 2642
+    },
+    {
+      "epoch": 0.9592743764172336,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.00013775547668584017,
+      "loss": 0.9722,
+      "step": 2644
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.00013770652306939175,
+      "loss": 0.8664,
+      "step": 2646
+    },
+    {
+      "epoch": 0.9607256235827665,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.00013765756945294334,
+      "loss": 1.0287,
+      "step": 2648
+    },
+    {
+      "epoch": 0.9614512471655329,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.00013760861583649495,
+      "loss": 0.9959,
+      "step": 2650
+    },
+    {
+      "epoch": 0.9621768707482993,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.0001375596622200465,
+      "loss": 0.9511,
+      "step": 2652
+    },
+    {
+      "epoch": 0.9629024943310658,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.00013751070860359808,
+      "loss": 0.9145,
+      "step": 2654
+    },
+    {
+      "epoch": 0.9636281179138322,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.0001374617549871497,
+      "loss": 1.0201,
+      "step": 2656
+    },
+    {
+      "epoch": 0.9643537414965987,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.00013741280137070128,
+      "loss": 0.9204,
+      "step": 2658
+    },
+    {
+      "epoch": 0.9650793650793651,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.00013736384775425286,
+      "loss": 0.9881,
+      "step": 2660
+    },
+    {
+      "epoch": 0.9658049886621315,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.00013731489413780444,
+      "loss": 1.0235,
+      "step": 2662
+    },
+    {
+      "epoch": 0.966530612244898,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.00013726594052135602,
+      "loss": 0.9197,
+      "step": 2664
+    },
+    {
+      "epoch": 0.9672562358276644,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.0001372169869049076,
+      "loss": 0.957,
+      "step": 2666
+    },
+    {
+      "epoch": 0.9679818594104308,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.0001371680332884592,
+      "loss": 1.0025,
+      "step": 2668
+    },
+    {
+      "epoch": 0.9687074829931973,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.0001371190796720108,
+      "loss": 0.9623,
+      "step": 2670
+    },
+    {
+      "epoch": 0.9694331065759637,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.00013707012605556235,
+      "loss": 0.9854,
+      "step": 2672
+    },
+    {
+      "epoch": 0.9701587301587301,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.00013702117243911393,
+      "loss": 0.9908,
+      "step": 2674
+    },
+    {
+      "epoch": 0.9708843537414966,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.00013697221882266554,
+      "loss": 0.9643,
+      "step": 2676
+    },
+    {
+      "epoch": 0.971609977324263,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.00013692326520621713,
+      "loss": 1.0681,
+      "step": 2678
+    },
+    {
+      "epoch": 0.9723356009070295,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.0001368743115897687,
+      "loss": 0.9858,
+      "step": 2680
+    },
+    {
+      "epoch": 0.9730612244897959,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.0001368253579733203,
+      "loss": 0.9974,
+      "step": 2682
+    },
+    {
+      "epoch": 0.9737868480725623,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.00013677640435687187,
+      "loss": 0.9672,
+      "step": 2684
+    },
+    {
+      "epoch": 0.9745124716553288,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.00013672745074042345,
+      "loss": 0.9644,
+      "step": 2686
+    },
+    {
+      "epoch": 0.9752380952380952,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.00013667849712397504,
+      "loss": 0.9687,
+      "step": 2688
+    },
+    {
+      "epoch": 0.9759637188208616,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.00013662954350752665,
+      "loss": 0.9365,
+      "step": 2690
+    },
+    {
+      "epoch": 0.9766893424036281,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.0001365805898910782,
+      "loss": 1.0098,
+      "step": 2692
+    },
+    {
+      "epoch": 0.9774149659863945,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.00013653163627462978,
+      "loss": 1.0377,
+      "step": 2694
+    },
+    {
+      "epoch": 0.978140589569161,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.0001364826826581814,
+      "loss": 0.9091,
+      "step": 2696
+    },
+    {
+      "epoch": 0.9788662131519275,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.00013643372904173298,
+      "loss": 1.0475,
+      "step": 2698
+    },
+    {
+      "epoch": 0.9795918367346939,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.00013638477542528453,
+      "loss": 0.9671,
+      "step": 2700
+    },
+    {
+      "epoch": 0.9795918367346939,
+      "eval_loss": 1.0159815549850464,
+      "eval_runtime": 97.2766,
+      "eval_samples_per_second": 18.504,
+      "eval_steps_per_second": 0.586,
+      "step": 2700
+    },
+    {
+      "epoch": 0.9803174603174604,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.00013633582180883614,
+      "loss": 0.9728,
+      "step": 2702
+    },
+    {
+      "epoch": 0.9810430839002268,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.00013628686819238772,
+      "loss": 1.0255,
+      "step": 2704
+    },
+    {
+      "epoch": 0.9817687074829932,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.0001362379145759393,
+      "loss": 1.0246,
+      "step": 2706
+    },
+    {
+      "epoch": 0.9824943310657597,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.00013618896095949089,
+      "loss": 0.9583,
+      "step": 2708
+    },
+    {
+      "epoch": 0.9832199546485261,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.00013614000734304247,
+      "loss": 0.9655,
+      "step": 2710
+    },
+    {
+      "epoch": 0.9839455782312925,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.00013609105372659405,
+      "loss": 0.9796,
+      "step": 2712
+    },
+    {
+      "epoch": 0.984671201814059,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.00013604210011014563,
+      "loss": 0.9351,
+      "step": 2714
+    },
+    {
+      "epoch": 0.9853968253968254,
+      "grad_norm": 0.201171875,
+      "learning_rate": 0.00013599314649369724,
+      "loss": 1.0364,
+      "step": 2716
+    },
+    {
+      "epoch": 0.9861224489795918,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.00013594419287724882,
+      "loss": 1.031,
+      "step": 2718
+    },
+    {
+      "epoch": 0.9868480725623583,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.00013589523926080038,
+      "loss": 0.9863,
+      "step": 2720
+    },
+    {
+      "epoch": 0.9875736961451247,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.000135846285644352,
+      "loss": 0.9365,
+      "step": 2722
+    },
+    {
+      "epoch": 0.9882993197278912,
+      "grad_norm": 0.177734375,
+      "learning_rate": 0.00013579733202790357,
+      "loss": 1.0187,
+      "step": 2724
+    },
+    {
+      "epoch": 0.9890249433106576,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.00013574837841145515,
+      "loss": 1.0717,
+      "step": 2726
+    },
+    {
+      "epoch": 0.989750566893424,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.00013569942479500674,
+      "loss": 1.0162,
+      "step": 2728
+    },
+    {
+      "epoch": 0.9904761904761905,
+      "grad_norm": 0.181640625,
+      "learning_rate": 0.00013565047117855832,
+      "loss": 1.0003,
+      "step": 2730
+    },
+    {
+      "epoch": 0.9912018140589569,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.0001356015175621099,
+      "loss": 0.9634,
+      "step": 2732
+    },
+    {
+      "epoch": 0.9919274376417233,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.00013555256394566148,
+      "loss": 0.9042,
+      "step": 2734
+    },
+    {
+      "epoch": 0.9926530612244898,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.0001355036103292131,
+      "loss": 1.0227,
+      "step": 2736
+    },
+    {
+      "epoch": 0.9933786848072562,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.00013545465671276467,
+      "loss": 0.9887,
+      "step": 2738
+    },
+    {
+      "epoch": 0.9941043083900227,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.00013540570309631623,
+      "loss": 0.9897,
+      "step": 2740
+    },
+    {
+      "epoch": 0.9948299319727891,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.00013535674947986784,
+      "loss": 0.9965,
+      "step": 2742
+    },
+    {
+      "epoch": 0.9955555555555555,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.00013530779586341942,
+      "loss": 0.9902,
+      "step": 2744
+    },
+    {
+      "epoch": 0.996281179138322,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.000135258842246971,
+      "loss": 1.0522,
+      "step": 2746
+    },
+    {
+      "epoch": 0.9970068027210884,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.00013520988863052259,
+      "loss": 0.9595,
+      "step": 2748
+    },
+    {
+      "epoch": 0.9977324263038548,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.00013516093501407417,
+      "loss": 0.9819,
+      "step": 2750
+    },
+    {
+      "epoch": 0.9984580498866213,
+      "grad_norm": 0.185546875,
+      "learning_rate": 0.00013511198139762575,
+      "loss": 1.1001,
+      "step": 2752
+    },
+    {
+      "epoch": 0.9991836734693877,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.00013506302778117733,
+      "loss": 0.9502,
+      "step": 2754
+    },
+    {
+      "epoch": 0.9999092970521541,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.00013501407416472894,
+      "loss": 0.9582,
+      "step": 2756
+    },
+    {
+      "epoch": 1.0003628117913832,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.00013496512054828052,
+      "loss": 0.9314,
+      "step": 2758
+    },
+    {
+      "epoch": 1.0010884353741496,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.00013491616693183208,
+      "loss": 0.9523,
+      "step": 2760
+    },
+    {
+      "epoch": 1.001814058956916,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.0001348672133153837,
+      "loss": 1.0069,
+      "step": 2762
+    },
+    {
+      "epoch": 1.0025396825396826,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.00013481825969893527,
+      "loss": 0.9828,
+      "step": 2764
+    },
+    {
+      "epoch": 1.003265306122449,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.00013476930608248685,
+      "loss": 1.0188,
+      "step": 2766
+    },
+    {
+      "epoch": 1.0039909297052154,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.00013472035246603844,
+      "loss": 0.9474,
+      "step": 2768
+    },
+    {
+      "epoch": 1.0047165532879818,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.00013467139884959002,
+      "loss": 0.9772,
+      "step": 2770
+    },
+    {
+      "epoch": 1.0054421768707482,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.0001346224452331416,
+      "loss": 1.0774,
+      "step": 2772
+    },
+    {
+      "epoch": 1.0061678004535148,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.00013457349161669318,
+      "loss": 0.9454,
+      "step": 2774
+    },
+    {
+      "epoch": 1.0068934240362812,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.0001345245380002448,
+      "loss": 0.9687,
+      "step": 2776
+    },
+    {
+      "epoch": 1.0076190476190476,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.00013447558438379635,
+      "loss": 1.0119,
+      "step": 2778
+    },
+    {
+      "epoch": 1.008344671201814,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.00013442663076734793,
+      "loss": 1.0194,
+      "step": 2780
+    },
+    {
+      "epoch": 1.0090702947845804,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.00013437767715089954,
+      "loss": 0.9619,
+      "step": 2782
+    },
+    {
+      "epoch": 1.0097959183673468,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.00013432872353445112,
+      "loss": 0.9702,
+      "step": 2784
+    },
+    {
+      "epoch": 1.0105215419501135,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.0001342797699180027,
+      "loss": 0.9513,
+      "step": 2786
+    },
+    {
+      "epoch": 1.0112471655328799,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.00013423081630155429,
+      "loss": 1.0061,
+      "step": 2788
+    },
+    {
+      "epoch": 1.0119727891156463,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.00013418186268510587,
+      "loss": 0.9964,
+      "step": 2790
+    },
+    {
+      "epoch": 1.0126984126984127,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.00013413290906865745,
+      "loss": 0.9677,
+      "step": 2792
+    },
+    {
+      "epoch": 1.013424036281179,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.00013408395545220903,
+      "loss": 0.9397,
+      "step": 2794
+    },
+    {
+      "epoch": 1.0141496598639457,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.00013403500183576064,
+      "loss": 0.9438,
+      "step": 2796
+    },
+    {
+      "epoch": 1.014875283446712,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.0001339860482193122,
+      "loss": 0.9445,
+      "step": 2798
+    },
+    {
+      "epoch": 1.0156009070294785,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.00013393709460286378,
+      "loss": 0.9987,
+      "step": 2800
+    },
+    {
+      "epoch": 1.0156009070294785,
+      "eval_loss": 1.0157527923583984,
+      "eval_runtime": 93.9867,
+      "eval_samples_per_second": 19.152,
+      "eval_steps_per_second": 0.606,
+      "step": 2800
+    },
+    {
+      "epoch": 1.0163265306122449,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.0001338881409864154,
+      "loss": 1.0336,
+      "step": 2802
+    },
+    {
+      "epoch": 1.0170521541950113,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.00013383918736996697,
+      "loss": 1.0242,
+      "step": 2804
+    },
+    {
+      "epoch": 1.0177777777777777,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.00013379023375351855,
+      "loss": 0.9604,
+      "step": 2806
+    },
+    {
+      "epoch": 1.0185034013605443,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.00013374128013707014,
+      "loss": 1.0008,
+      "step": 2808
+    },
+    {
+      "epoch": 1.0192290249433107,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.00013369232652062172,
+      "loss": 0.9853,
+      "step": 2810
+    },
+    {
+      "epoch": 1.019954648526077,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.0001336433729041733,
+      "loss": 1.1074,
+      "step": 2812
+    },
+    {
+      "epoch": 1.0206802721088435,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.00013359441928772488,
+      "loss": 1.0543,
+      "step": 2814
+    },
+    {
+      "epoch": 1.0214058956916099,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.0001335454656712765,
+      "loss": 0.9545,
+      "step": 2816
+    },
+    {
+      "epoch": 1.0221315192743765,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.00013349651205482805,
+      "loss": 0.9777,
+      "step": 2818
+    },
+    {
+      "epoch": 1.022857142857143,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.00013344755843837963,
+      "loss": 1.002,
+      "step": 2820
+    },
+    {
+      "epoch": 1.0235827664399093,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.00013339860482193124,
+      "loss": 0.9998,
+      "step": 2822
+    },
+    {
+      "epoch": 1.0243083900226757,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.00013334965120548282,
+      "loss": 0.9978,
+      "step": 2824
+    },
+    {
+      "epoch": 1.025034013605442,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.0001333006975890344,
+      "loss": 1.0295,
+      "step": 2826
+    },
+    {
+      "epoch": 1.0257596371882087,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.00013325174397258599,
+      "loss": 1.0145,
+      "step": 2828
+    },
+    {
+      "epoch": 1.0264852607709751,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.00013320279035613757,
+      "loss": 0.9537,
+      "step": 2830
+    },
+    {
+      "epoch": 1.0272108843537415,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.00013315383673968915,
+      "loss": 1.091,
+      "step": 2832
+    },
+    {
+      "epoch": 1.027936507936508,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.00013310488312324073,
+      "loss": 1.0578,
+      "step": 2834
+    },
+    {
+      "epoch": 1.0286621315192743,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.00013305592950679234,
+      "loss": 0.9262,
+      "step": 2836
+    },
+    {
+      "epoch": 1.0293877551020407,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.0001330069758903439,
+      "loss": 0.9779,
+      "step": 2838
+    },
+    {
+      "epoch": 1.0301133786848073,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.00013295802227389548,
+      "loss": 0.9869,
+      "step": 2840
+    },
+    {
+      "epoch": 1.0308390022675737,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.0001329090686574471,
+      "loss": 0.9595,
+      "step": 2842
+    },
+    {
+      "epoch": 1.0315646258503401,
+      "grad_norm": 0.197265625,
+      "learning_rate": 0.00013286011504099867,
+      "loss": 1.0107,
+      "step": 2844
+    },
+    {
+      "epoch": 1.0322902494331065,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.00013281116142455023,
+      "loss": 1.0039,
+      "step": 2846
+    },
+    {
+      "epoch": 1.033015873015873,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.00013276220780810184,
+      "loss": 0.9353,
+      "step": 2848
+    },
+    {
+      "epoch": 1.0337414965986396,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.00013271325419165342,
+      "loss": 0.9496,
+      "step": 2850
+    },
+    {
+      "epoch": 1.034467120181406,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.000132664300575205,
+      "loss": 1.0133,
+      "step": 2852
+    },
+    {
+      "epoch": 1.0351927437641724,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.00013261534695875658,
+      "loss": 0.9736,
+      "step": 2854
+    },
+    {
+      "epoch": 1.0359183673469388,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.00013256639334230816,
+      "loss": 1.0096,
+      "step": 2856
+    },
+    {
+      "epoch": 1.0366439909297052,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.00013251743972585975,
+      "loss": 1.0358,
+      "step": 2858
+    },
+    {
+      "epoch": 1.0373696145124716,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.00013246848610941133,
+      "loss": 1.0608,
+      "step": 2860
+    },
+    {
+      "epoch": 1.0380952380952382,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.00013241953249296294,
+      "loss": 0.9904,
+      "step": 2862
+    },
+    {
+      "epoch": 1.0388208616780046,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.00013237057887651452,
+      "loss": 0.9738,
+      "step": 2864
+    },
+    {
+      "epoch": 1.039546485260771,
+      "grad_norm": 0.185546875,
+      "learning_rate": 0.00013232162526006608,
+      "loss": 1.0697,
+      "step": 2866
+    },
+    {
+      "epoch": 1.0402721088435374,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.00013227267164361769,
+      "loss": 0.9685,
+      "step": 2868
+    },
+    {
+      "epoch": 1.0409977324263038,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.00013222371802716927,
+      "loss": 1.0041,
+      "step": 2870
+    },
+    {
+      "epoch": 1.0417233560090704,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.00013217476441072085,
+      "loss": 0.9809,
+      "step": 2872
+    },
+    {
+      "epoch": 1.0424489795918368,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.00013212581079427243,
+      "loss": 0.9908,
+      "step": 2874
+    },
+    {
+      "epoch": 1.0431746031746032,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.00013207685717782401,
+      "loss": 1.032,
+      "step": 2876
+    },
+    {
+      "epoch": 1.0439002267573696,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.0001320279035613756,
+      "loss": 0.9646,
+      "step": 2878
+    },
+    {
+      "epoch": 1.044625850340136,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.00013197894994492718,
+      "loss": 0.9965,
+      "step": 2880
+    },
+    {
+      "epoch": 1.0453514739229024,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.0001319299963284788,
+      "loss": 0.9963,
+      "step": 2882
+    },
+    {
+      "epoch": 1.046077097505669,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.00013188104271203037,
+      "loss": 1.0057,
+      "step": 2884
+    },
+    {
+      "epoch": 1.0468027210884354,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.00013183208909558193,
+      "loss": 0.9545,
+      "step": 2886
+    },
+    {
+      "epoch": 1.0475283446712018,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.00013178313547913354,
+      "loss": 0.9863,
+      "step": 2888
+    },
+    {
+      "epoch": 1.0482539682539682,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.00013173418186268512,
+      "loss": 1.0419,
+      "step": 2890
+    },
+    {
+      "epoch": 1.0489795918367346,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.0001316852282462367,
+      "loss": 1.013,
+      "step": 2892
+    },
+    {
+      "epoch": 1.0497052154195012,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.00013163627462978828,
+      "loss": 1.096,
+      "step": 2894
+    },
+    {
+      "epoch": 1.0504308390022676,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.00013158732101333986,
+      "loss": 1.0023,
+      "step": 2896
+    },
+    {
+      "epoch": 1.051156462585034,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.00013153836739689145,
+      "loss": 0.9743,
+      "step": 2898
+    },
+    {
+      "epoch": 1.0518820861678004,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.00013148941378044303,
+      "loss": 1.0481,
+      "step": 2900
+    },
+    {
+      "epoch": 1.0518820861678004,
+      "eval_loss": 1.0157631635665894,
+      "eval_runtime": 98.609,
+      "eval_samples_per_second": 18.254,
+      "eval_steps_per_second": 0.578,
+      "step": 2900
+    },
+    {
+      "epoch": 1.0526077097505668,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.00013144046016399464,
+      "loss": 1.0343,
+      "step": 2902
+    },
+    {
+      "epoch": 1.0533333333333332,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.00013139150654754622,
+      "loss": 0.9971,
+      "step": 2904
+    },
+    {
+      "epoch": 1.0540589569160999,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.00013134255293109778,
+      "loss": 0.9762,
+      "step": 2906
+    },
+    {
+      "epoch": 1.0547845804988663,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.00013129359931464939,
+      "loss": 1.1036,
+      "step": 2908
+    },
+    {
+      "epoch": 1.0555102040816327,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.00013124464569820097,
+      "loss": 0.9814,
+      "step": 2910
+    },
+    {
+      "epoch": 1.056235827664399,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.00013119569208175255,
+      "loss": 0.9185,
+      "step": 2912
+    },
+    {
+      "epoch": 1.0569614512471655,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.00013114673846530413,
+      "loss": 0.9613,
+      "step": 2914
+    },
+    {
+      "epoch": 1.057687074829932,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.00013109778484885571,
+      "loss": 0.9276,
+      "step": 2916
+    },
+    {
+      "epoch": 1.0584126984126985,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.0001310488312324073,
+      "loss": 0.9268,
+      "step": 2918
+    },
+    {
+      "epoch": 1.0591383219954649,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.00013099987761595888,
+      "loss": 1.0498,
+      "step": 2920
+    },
+    {
+      "epoch": 1.0598639455782313,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.0001309509239995105,
+      "loss": 1.0223,
+      "step": 2922
+    },
+    {
+      "epoch": 1.0605895691609977,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.00013090197038306204,
+      "loss": 0.9899,
+      "step": 2924
+    },
+    {
+      "epoch": 1.061315192743764,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.00013085301676661363,
+      "loss": 0.9639,
+      "step": 2926
+    },
+    {
+      "epoch": 1.0620408163265307,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.00013080406315016524,
+      "loss": 0.9369,
+      "step": 2928
+    },
+    {
+      "epoch": 1.062766439909297,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.00013075510953371682,
+      "loss": 0.93,
+      "step": 2930
+    },
+    {
+      "epoch": 1.0634920634920635,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.0001307061559172684,
+      "loss": 0.993,
+      "step": 2932
+    },
+    {
+      "epoch": 1.0642176870748299,
+      "grad_norm": 0.181640625,
+      "learning_rate": 0.00013065720230081998,
+      "loss": 1.0847,
+      "step": 2934
+    },
+    {
+      "epoch": 1.0649433106575963,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.00013060824868437156,
+      "loss": 1.003,
+      "step": 2936
+    },
+    {
+      "epoch": 1.065668934240363,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.00013055929506792315,
+      "loss": 1.0098,
+      "step": 2938
+    },
+    {
+      "epoch": 1.0663945578231293,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.00013051034145147473,
+      "loss": 0.9617,
+      "step": 2940
+    },
+    {
+      "epoch": 1.0671201814058957,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.00013046138783502634,
+      "loss": 0.9282,
+      "step": 2942
+    },
+    {
+      "epoch": 1.067845804988662,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.0001304124342185779,
+      "loss": 0.9462,
+      "step": 2944
+    },
+    {
+      "epoch": 1.0685714285714285,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.00013036348060212948,
+      "loss": 1.0117,
+      "step": 2946
+    },
+    {
+      "epoch": 1.069297052154195,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.00013031452698568109,
+      "loss": 0.9634,
+      "step": 2948
+    },
+    {
+      "epoch": 1.0700226757369615,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.00013026557336923267,
+      "loss": 1.0243,
+      "step": 2950
+    },
+    {
+      "epoch": 1.070748299319728,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.00013021661975278425,
+      "loss": 0.9924,
+      "step": 2952
+    },
+    {
+      "epoch": 1.0714739229024943,
+      "grad_norm": 0.177734375,
+      "learning_rate": 0.00013016766613633583,
+      "loss": 0.9808,
+      "step": 2954
+    },
+    {
+      "epoch": 1.0721995464852607,
+      "grad_norm": 0.181640625,
+      "learning_rate": 0.00013011871251988741,
+      "loss": 1.0503,
+      "step": 2956
+    },
+    {
+      "epoch": 1.0729251700680271,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.000130069758903439,
+      "loss": 1.042,
+      "step": 2958
+    },
+    {
+      "epoch": 1.0736507936507937,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.00013002080528699058,
+      "loss": 1.0484,
+      "step": 2960
+    },
+    {
+      "epoch": 1.0743764172335601,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.0001299718516705422,
+      "loss": 0.925,
+      "step": 2962
+    },
+    {
+      "epoch": 1.0751020408163265,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.00012992289805409374,
+      "loss": 1.0459,
+      "step": 2964
+    },
+    {
+      "epoch": 1.075827664399093,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 0.00012987394443764533,
+      "loss": 0.9794,
+      "step": 2966
+    },
+    {
+      "epoch": 1.0765532879818593,
+      "grad_norm": 0.1884765625,
+      "learning_rate": 0.00012982499082119694,
+      "loss": 0.887,
+      "step": 2968
+    },
+    {
+      "epoch": 1.0772789115646257,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.00012977603720474852,
+      "loss": 1.0116,
+      "step": 2970
+    },
+    {
+      "epoch": 1.0780045351473924,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.0001297270835883001,
+      "loss": 0.9688,
+      "step": 2972
+    },
+    {
+      "epoch": 1.0787301587301588,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.00012967812997185168,
+      "loss": 1.024,
+      "step": 2974
+    },
+    {
+      "epoch": 1.0794557823129252,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.00012962917635540326,
+      "loss": 1.0757,
+      "step": 2976
+    },
+    {
+      "epoch": 1.0801814058956916,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.00012958022273895485,
+      "loss": 0.9665,
+      "step": 2978
+    },
+    {
+      "epoch": 1.080907029478458,
+      "grad_norm": 0.18359375,
+      "learning_rate": 0.00012953126912250643,
+      "loss": 0.9317,
+      "step": 2980
+    },
+    {
+      "epoch": 1.0816326530612246,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.00012948231550605804,
+      "loss": 0.9634,
+      "step": 2982
+    },
+    {
+      "epoch": 1.082358276643991,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.0001294333618896096,
+      "loss": 0.9887,
+      "step": 2984
+    },
+    {
+      "epoch": 1.0830839002267574,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.00012938440827316118,
+      "loss": 0.8876,
+      "step": 2986
+    },
+    {
+      "epoch": 1.0838095238095238,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.00012933545465671279,
+      "loss": 0.9634,
+      "step": 2988
+    },
+    {
+      "epoch": 1.0845351473922902,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.00012928650104026437,
+      "loss": 0.9834,
+      "step": 2990
+    },
+    {
+      "epoch": 1.0852607709750566,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.00012923754742381592,
+      "loss": 0.99,
+      "step": 2992
+    },
+    {
+      "epoch": 1.0859863945578232,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.00012918859380736753,
+      "loss": 0.9375,
+      "step": 2994
+    },
+    {
+      "epoch": 1.0867120181405896,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.00012913964019091911,
+      "loss": 0.9757,
+      "step": 2996
+    },
+    {
+      "epoch": 1.087437641723356,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.0001290906865744707,
+      "loss": 0.9934,
+      "step": 2998
+    },
+    {
+      "epoch": 1.0881632653061224,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.00012904173295802228,
+      "loss": 0.9745,
+      "step": 3000
+    },
+    {
+      "epoch": 1.0881632653061224,
+      "eval_loss": 1.0154409408569336,
+      "eval_runtime": 92.0958,
+      "eval_samples_per_second": 19.545,
+      "eval_steps_per_second": 0.619,
+      "step": 3000
+    },
+    {
+      "epoch": 1.0888888888888888,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.00012899277934157386,
+      "loss": 0.9854,
+      "step": 3002
+    },
+    {
+      "epoch": 1.0896145124716554,
+      "grad_norm": 0.19140625,
+      "learning_rate": 0.00012894382572512544,
+      "loss": 0.9796,
+      "step": 3004
+    },
+    {
+      "epoch": 1.0903401360544218,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.00012889487210867703,
+      "loss": 0.9706,
+      "step": 3006
+    },
+    {
+      "epoch": 1.0910657596371882,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.00012884591849222864,
+      "loss": 0.9691,
+      "step": 3008
+    },
+    {
+      "epoch": 1.0917913832199546,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.00012879696487578022,
+      "loss": 1.0203,
+      "step": 3010
+    },
+    {
+      "epoch": 1.092517006802721,
+      "grad_norm": 0.1865234375,
+      "learning_rate": 0.00012874801125933177,
+      "loss": 0.9979,
+      "step": 3012
+    },
+    {
+      "epoch": 1.0932426303854874,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.00012869905764288338,
+      "loss": 1.0656,
+      "step": 3014
+    },
+    {
+      "epoch": 1.093968253968254,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.00012865010402643496,
+      "loss": 0.9869,
+      "step": 3016
+    },
+    {
+      "epoch": 1.0946938775510204,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.00012860115040998655,
+      "loss": 1.0094,
+      "step": 3018
+    },
+    {
+      "epoch": 1.0954195011337868,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.00012855219679353813,
+      "loss": 1.0243,
+      "step": 3020
+    },
+    {
+      "epoch": 1.0961451247165532,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.0001285032431770897,
+      "loss": 0.9527,
+      "step": 3022
+    },
+    {
+      "epoch": 1.0968707482993196,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.0001284542895606413,
+      "loss": 1.0238,
+      "step": 3024
+    },
+    {
+      "epoch": 1.0975963718820863,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.00012840533594419288,
+      "loss": 0.9843,
+      "step": 3026
+    },
+    {
+      "epoch": 1.0983219954648527,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.00012835638232774448,
+      "loss": 0.9041,
+      "step": 3028
+    },
+    {
+      "epoch": 1.099047619047619,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.00012830742871129607,
+      "loss": 0.9152,
+      "step": 3030
+    },
+    {
+      "epoch": 1.0997732426303855,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.00012825847509484762,
+      "loss": 0.9771,
+      "step": 3032
+    },
+    {
+      "epoch": 1.1004988662131518,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.00012820952147839923,
+      "loss": 0.952,
+      "step": 3034
+    },
+    {
+      "epoch": 1.1012244897959185,
+      "grad_norm": 0.17578125,
+      "learning_rate": 0.00012816056786195081,
+      "loss": 0.9888,
+      "step": 3036
+    },
+    {
+      "epoch": 1.1019501133786849,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.0001281116142455024,
+      "loss": 0.9741,
+      "step": 3038
+    },
+    {
+      "epoch": 1.1026757369614513,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.00012806266062905398,
+      "loss": 0.9094,
+      "step": 3040
+    },
+    {
+      "epoch": 1.1034013605442177,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.00012801370701260556,
+      "loss": 0.9386,
+      "step": 3042
+    },
+    {
+      "epoch": 1.104126984126984,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.00012796475339615714,
+      "loss": 0.9503,
+      "step": 3044
+    },
+    {
+      "epoch": 1.1048526077097505,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.00012791579977970873,
+      "loss": 0.9595,
+      "step": 3046
+    },
+    {
+      "epoch": 1.105578231292517,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.00012786684616326033,
+      "loss": 0.9771,
+      "step": 3048
+    },
+    {
+      "epoch": 1.1063038548752835,
+      "grad_norm": 0.17578125,
+      "learning_rate": 0.00012781789254681192,
+      "loss": 1.054,
+      "step": 3050
+    },
+    {
+      "epoch": 1.1070294784580499,
+      "grad_norm": 0.173828125,
+      "learning_rate": 0.00012776893893036347,
+      "loss": 0.9665,
+      "step": 3052
+    },
+    {
+      "epoch": 1.1077551020408163,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.00012771998531391508,
+      "loss": 1.0094,
+      "step": 3054
+    },
+    {
+      "epoch": 1.1084807256235827,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.00012767103169746666,
+      "loss": 0.9462,
+      "step": 3056
+    },
+    {
+      "epoch": 1.1092063492063493,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.00012762207808101825,
+      "loss": 1.0243,
+      "step": 3058
+    },
+    {
+      "epoch": 1.1099319727891157,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.00012757312446456983,
+      "loss": 0.9708,
+      "step": 3060
+    },
+    {
+      "epoch": 1.110657596371882,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.0001275241708481214,
+      "loss": 0.9986,
+      "step": 3062
+    },
+    {
+      "epoch": 1.1113832199546485,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.000127475217231673,
+      "loss": 0.9911,
+      "step": 3064
+    },
+    {
+      "epoch": 1.112108843537415,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.00012742626361522458,
+      "loss": 1.0371,
+      "step": 3066
+    },
+    {
+      "epoch": 1.1128344671201813,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.00012737730999877618,
+      "loss": 1.0138,
+      "step": 3068
+    },
+    {
+      "epoch": 1.113560090702948,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.00012732835638232774,
+      "loss": 0.9498,
+      "step": 3070
+    },
+    {
+      "epoch": 1.1142857142857143,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.00012727940276587932,
+      "loss": 1.0003,
+      "step": 3072
+    },
+    {
+      "epoch": 1.1150113378684807,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.00012723044914943093,
+      "loss": 1.0018,
+      "step": 3074
+    },
+    {
+      "epoch": 1.1157369614512471,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.00012718149553298251,
+      "loss": 0.9182,
+      "step": 3076
+    },
+    {
+      "epoch": 1.1164625850340135,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.0001271325419165341,
+      "loss": 0.9443,
+      "step": 3078
+    },
+    {
+      "epoch": 1.1171882086167801,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.00012708358830008568,
+      "loss": 1.0236,
+      "step": 3080
+    },
+    {
+      "epoch": 1.1179138321995465,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.00012703463468363726,
+      "loss": 0.977,
+      "step": 3082
+    },
+    {
+      "epoch": 1.118639455782313,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.00012698568106718884,
+      "loss": 0.9371,
+      "step": 3084
+    },
+    {
+      "epoch": 1.1193650793650793,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.00012693672745074043,
+      "loss": 0.9309,
+      "step": 3086
+    },
+    {
+      "epoch": 1.1200907029478457,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.00012688777383429203,
+      "loss": 0.9547,
+      "step": 3088
+    },
+    {
+      "epoch": 1.1208163265306124,
+      "grad_norm": 0.23828125,
+      "learning_rate": 0.0001268388202178436,
+      "loss": 1.0593,
+      "step": 3090
+    },
+    {
+      "epoch": 1.1215419501133788,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.00012678986660139517,
+      "loss": 0.9916,
+      "step": 3092
+    },
+    {
+      "epoch": 1.1222675736961452,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.00012674091298494678,
+      "loss": 0.9708,
+      "step": 3094
+    },
+    {
+      "epoch": 1.1229931972789116,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 0.00012669195936849836,
+      "loss": 0.9498,
+      "step": 3096
+    },
+    {
+      "epoch": 1.123718820861678,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.00012664300575204995,
+      "loss": 0.9388,
+      "step": 3098
+    },
+    {
+      "epoch": 1.1244444444444444,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.00012659405213560153,
+      "loss": 0.9372,
+      "step": 3100
+    },
+    {
+      "epoch": 1.1244444444444444,
+      "eval_loss": 1.0154296159744263,
+      "eval_runtime": 89.182,
+      "eval_samples_per_second": 20.183,
+      "eval_steps_per_second": 0.639,
+      "step": 3100
+    },
+    {
+      "epoch": 1.125170068027211,
+      "grad_norm": 0.19140625,
+      "learning_rate": 0.0001265450985191531,
+      "loss": 0.9911,
+      "step": 3102
+    },
+    {
+      "epoch": 1.1258956916099774,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.0001264961449027047,
+      "loss": 1.0097,
+      "step": 3104
+    },
+    {
+      "epoch": 1.1266213151927438,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.00012644719128625628,
+      "loss": 0.9942,
+      "step": 3106
+    },
+    {
+      "epoch": 1.1273469387755102,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.00012639823766980788,
+      "loss": 0.9891,
+      "step": 3108
+    },
+    {
+      "epoch": 1.1280725623582766,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.00012634928405335944,
+      "loss": 1.0132,
+      "step": 3110
+    },
+    {
+      "epoch": 1.1287981859410432,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.00012630033043691102,
+      "loss": 0.9039,
+      "step": 3112
+    },
+    {
+      "epoch": 1.1295238095238096,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.00012625137682046263,
+      "loss": 0.9398,
+      "step": 3114
+    },
+    {
+      "epoch": 1.130249433106576,
+      "grad_norm": 0.1923828125,
+      "learning_rate": 0.00012620242320401421,
+      "loss": 0.9643,
+      "step": 3116
+    },
+    {
+      "epoch": 1.1309750566893424,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.0001261534695875658,
+      "loss": 0.9252,
+      "step": 3118
+    },
+    {
+      "epoch": 1.1317006802721088,
+      "grad_norm": 0.17578125,
+      "learning_rate": 0.00012610451597111738,
+      "loss": 0.9849,
+      "step": 3120
+    },
+    {
+      "epoch": 1.1324263038548752,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.00012605556235466896,
+      "loss": 1.004,
+      "step": 3122
+    },
+    {
+      "epoch": 1.1331519274376418,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 0.00012600660873822054,
+      "loss": 1.0351,
+      "step": 3124
+    },
+    {
+      "epoch": 1.1338775510204082,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.00012595765512177213,
+      "loss": 0.9665,
+      "step": 3126
+    },
+    {
+      "epoch": 1.1346031746031746,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.00012590870150532373,
+      "loss": 1.0028,
+      "step": 3128
+    },
+    {
+      "epoch": 1.135328798185941,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.0001258597478888753,
+      "loss": 0.9811,
+      "step": 3130
+    },
+    {
+      "epoch": 1.1360544217687074,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.00012581079427242687,
+      "loss": 1.01,
+      "step": 3132
+    },
+    {
+      "epoch": 1.136780045351474,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.00012576184065597848,
+      "loss": 0.9534,
+      "step": 3134
+    },
+    {
+      "epoch": 1.1375056689342404,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.00012571288703953006,
+      "loss": 0.9672,
+      "step": 3136
+    },
+    {
+      "epoch": 1.1382312925170068,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.00012566393342308162,
+      "loss": 1.0314,
+      "step": 3138
+    },
+    {
+      "epoch": 1.1389569160997732,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 0.00012561497980663323,
+      "loss": 0.96,
+      "step": 3140
+    },
+    {
+      "epoch": 1.1396825396825396,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.0001255660261901848,
+      "loss": 0.9722,
+      "step": 3142
+    },
+    {
+      "epoch": 1.140408163265306,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.0001255170725737364,
+      "loss": 1.063,
+      "step": 3144
+    },
+    {
+      "epoch": 1.1411337868480727,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.00012546811895728798,
+      "loss": 0.987,
+      "step": 3146
+    },
+    {
+      "epoch": 1.141859410430839,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.00012541916534083956,
+      "loss": 0.9693,
+      "step": 3148
+    },
+    {
+      "epoch": 1.1425850340136054,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.00012537021172439114,
+      "loss": 0.9687,
+      "step": 3150
+    },
+    {
+      "epoch": 1.1433106575963718,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.00012532125810794272,
+      "loss": 0.9822,
+      "step": 3152
+    },
+    {
+      "epoch": 1.1440362811791382,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.00012527230449149433,
+      "loss": 0.9652,
+      "step": 3154
+    },
+    {
+      "epoch": 1.1447619047619049,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.0001252233508750459,
+      "loss": 0.8838,
+      "step": 3156
+    },
+    {
+      "epoch": 1.1454875283446713,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.00012517439725859747,
+      "loss": 0.9707,
+      "step": 3158
+    },
+    {
+      "epoch": 1.1462131519274377,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.00012512544364214908,
+      "loss": 1.0355,
+      "step": 3160
+    },
+    {
+      "epoch": 1.146938775510204,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.00012507649002570066,
+      "loss": 1.0814,
+      "step": 3162
+    },
+    {
+      "epoch": 1.1476643990929705,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.00012502753640925224,
+      "loss": 1.005,
+      "step": 3164
+    },
+    {
+      "epoch": 1.1483900226757369,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.00012497858279280382,
+      "loss": 0.9581,
+      "step": 3166
+    },
+    {
+      "epoch": 1.1491156462585035,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.0001249296291763554,
+      "loss": 1.0175,
+      "step": 3168
+    },
+    {
+      "epoch": 1.1498412698412699,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.000124880675559907,
+      "loss": 0.9524,
+      "step": 3170
+    },
+    {
+      "epoch": 1.1505668934240363,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.00012483172194345857,
+      "loss": 1.0104,
+      "step": 3172
+    },
+    {
+      "epoch": 1.1512925170068027,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.00012478276832701018,
+      "loss": 0.9864,
+      "step": 3174
+    },
+    {
+      "epoch": 1.152018140589569,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.00012473381471056176,
+      "loss": 1.0231,
+      "step": 3176
+    },
+    {
+      "epoch": 1.1527437641723357,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.00012468486109411332,
+      "loss": 0.9632,
+      "step": 3178
+    },
+    {
+      "epoch": 1.153469387755102,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.00012463590747766493,
+      "loss": 0.9549,
+      "step": 3180
+    },
+    {
+      "epoch": 1.1541950113378685,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.0001245869538612165,
+      "loss": 1.0362,
+      "step": 3182
+    },
+    {
+      "epoch": 1.154920634920635,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.0001245380002447681,
+      "loss": 0.9991,
+      "step": 3184
+    },
+    {
+      "epoch": 1.1556462585034013,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.00012448904662831967,
+      "loss": 0.9353,
+      "step": 3186
+    },
+    {
+      "epoch": 1.1563718820861677,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.00012444009301187126,
+      "loss": 0.9651,
+      "step": 3188
+    },
+    {
+      "epoch": 1.1570975056689343,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.00012439113939542284,
+      "loss": 0.9336,
+      "step": 3190
+    },
+    {
+      "epoch": 1.1578231292517007,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.00012434218577897442,
+      "loss": 0.9326,
+      "step": 3192
+    },
+    {
+      "epoch": 1.1585487528344671,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.00012429323216252603,
+      "loss": 1.0304,
+      "step": 3194
+    },
+    {
+      "epoch": 1.1592743764172335,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.0001242442785460776,
+      "loss": 0.9833,
+      "step": 3196
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.00012419532492962917,
+      "loss": 1.0228,
+      "step": 3198
+    },
+    {
+      "epoch": 1.1607256235827665,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.00012414637131318078,
+      "loss": 0.988,
+      "step": 3200
+    },
+    {
+      "epoch": 1.1607256235827665,
+      "eval_loss": 1.014972448348999,
+      "eval_runtime": 94.878,
+      "eval_samples_per_second": 18.972,
+      "eval_steps_per_second": 0.601,
+      "step": 3200
+    },
+    {
+      "epoch": 1.161451247165533,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.00012409741769673236,
+      "loss": 0.9443,
+      "step": 3202
+    },
+    {
+      "epoch": 1.1621768707482993,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.00012404846408028394,
+      "loss": 1.0144,
+      "step": 3204
+    },
+    {
+      "epoch": 1.1629024943310657,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.0001239995104638355,
+      "loss": 0.8949,
+      "step": 3206
+    },
+    {
+      "epoch": 1.1636281179138321,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.0001239505568473871,
+      "loss": 0.9979,
+      "step": 3208
+    },
+    {
+      "epoch": 1.1643537414965985,
+      "grad_norm": 0.173828125,
+      "learning_rate": 0.0001239016032309387,
+      "loss": 0.985,
+      "step": 3210
+    },
+    {
+      "epoch": 1.1650793650793652,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.00012385264961449027,
+      "loss": 1.018,
+      "step": 3212
+    },
+    {
+      "epoch": 1.1658049886621316,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 0.00012380369599804188,
+      "loss": 0.9736,
+      "step": 3214
+    },
+    {
+      "epoch": 1.166530612244898,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.00012375474238159344,
+      "loss": 0.9888,
+      "step": 3216
+    },
+    {
+      "epoch": 1.1672562358276644,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.00012370578876514502,
+      "loss": 0.9882,
+      "step": 3218
+    },
+    {
+      "epoch": 1.1679818594104308,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.00012365683514869663,
+      "loss": 1.0883,
+      "step": 3220
+    },
+    {
+      "epoch": 1.1687074829931974,
+      "grad_norm": 0.185546875,
+      "learning_rate": 0.0001236078815322482,
+      "loss": 1.0099,
+      "step": 3222
+    },
+    {
+      "epoch": 1.1694331065759638,
+      "grad_norm": 0.203125,
+      "learning_rate": 0.0001235589279157998,
+      "loss": 0.991,
+      "step": 3224
+    },
+    {
+      "epoch": 1.1701587301587302,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.00012350997429935135,
+      "loss": 0.9971,
+      "step": 3226
+    },
+    {
+      "epoch": 1.1708843537414966,
+      "grad_norm": 0.181640625,
+      "learning_rate": 0.00012346102068290296,
+      "loss": 1.0944,
+      "step": 3228
+    },
+    {
+      "epoch": 1.171609977324263,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.00012341206706645454,
+      "loss": 1.0144,
+      "step": 3230
+    },
+    {
+      "epoch": 1.1723356009070294,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.00012336311345000612,
+      "loss": 0.9596,
+      "step": 3232
+    },
+    {
+      "epoch": 1.173061224489796,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.00012331415983355773,
+      "loss": 1.0307,
+      "step": 3234
+    },
+    {
+      "epoch": 1.1737868480725624,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.00012326520621710929,
+      "loss": 0.9639,
+      "step": 3236
+    },
+    {
+      "epoch": 1.1745124716553288,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.00012321625260066087,
+      "loss": 0.9618,
+      "step": 3238
+    },
+    {
+      "epoch": 1.1752380952380952,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.00012316729898421248,
+      "loss": 0.9739,
+      "step": 3240
+    },
+    {
+      "epoch": 1.1759637188208616,
+      "grad_norm": 0.173828125,
+      "learning_rate": 0.00012311834536776406,
+      "loss": 1.0316,
+      "step": 3242
+    },
+    {
+      "epoch": 1.1766893424036282,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.00012306939175131564,
+      "loss": 0.9901,
+      "step": 3244
+    },
+    {
+      "epoch": 1.1774149659863946,
+      "grad_norm": 0.173828125,
+      "learning_rate": 0.0001230204381348672,
+      "loss": 1.0439,
+      "step": 3246
+    },
+    {
+      "epoch": 1.178140589569161,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.0001229714845184188,
+      "loss": 0.9763,
+      "step": 3248
+    },
+    {
+      "epoch": 1.1788662131519274,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.0001229225309019704,
+      "loss": 0.9723,
+      "step": 3250
+    },
+    {
+      "epoch": 1.1795918367346938,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.00012287357728552197,
+      "loss": 1.1122,
+      "step": 3252
+    },
+    {
+      "epoch": 1.1803174603174602,
+      "grad_norm": 0.18359375,
+      "learning_rate": 0.00012282462366907358,
+      "loss": 1.0547,
+      "step": 3254
+    },
+    {
+      "epoch": 1.1810430839002268,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.00012277567005262514,
+      "loss": 0.9226,
+      "step": 3256
+    },
+    {
+      "epoch": 1.1817687074829932,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.00012272671643617672,
+      "loss": 0.9439,
+      "step": 3258
+    },
+    {
+      "epoch": 1.1824943310657596,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.00012267776281972833,
+      "loss": 1.0307,
+      "step": 3260
+    },
+    {
+      "epoch": 1.183219954648526,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.0001226288092032799,
+      "loss": 1.0502,
+      "step": 3262
+    },
+    {
+      "epoch": 1.1839455782312924,
+      "grad_norm": 0.1923828125,
+      "learning_rate": 0.0001225798555868315,
+      "loss": 1.0492,
+      "step": 3264
+    },
+    {
+      "epoch": 1.184671201814059,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.00012253090197038305,
+      "loss": 0.962,
+      "step": 3266
+    },
+    {
+      "epoch": 1.1853968253968254,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.00012248194835393466,
+      "loss": 0.9904,
+      "step": 3268
+    },
+    {
+      "epoch": 1.1861224489795918,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.00012243299473748624,
+      "loss": 0.9677,
+      "step": 3270
+    },
+    {
+      "epoch": 1.1868480725623582,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.00012238404112103782,
+      "loss": 1.0684,
+      "step": 3272
+    },
+    {
+      "epoch": 1.1875736961451246,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.00012233508750458943,
+      "loss": 1.0299,
+      "step": 3274
+    },
+    {
+      "epoch": 1.188299319727891,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.00012228613388814099,
+      "loss": 0.98,
+      "step": 3276
+    },
+    {
+      "epoch": 1.1890249433106577,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.00012223718027169257,
+      "loss": 1.0087,
+      "step": 3278
+    },
+    {
+      "epoch": 1.189750566893424,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.00012218822665524418,
+      "loss": 1.0092,
+      "step": 3280
+    },
+    {
+      "epoch": 1.1904761904761905,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.00012213927303879576,
+      "loss": 1.0301,
+      "step": 3282
+    },
+    {
+      "epoch": 1.1912018140589569,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.00012209031942234731,
+      "loss": 1.0265,
+      "step": 3284
+    },
+    {
+      "epoch": 1.1919274376417233,
+      "grad_norm": 0.25,
+      "learning_rate": 0.00012204136580589891,
+      "loss": 0.9792,
+      "step": 3286
+    },
+    {
+      "epoch": 1.1926530612244899,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.0001219924121894505,
+      "loss": 0.9287,
+      "step": 3288
+    },
+    {
+      "epoch": 1.1933786848072563,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.00012194345857300209,
+      "loss": 0.9637,
+      "step": 3290
+    },
+    {
+      "epoch": 1.1941043083900227,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.00012189450495655368,
+      "loss": 0.984,
+      "step": 3292
+    },
+    {
+      "epoch": 1.194829931972789,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.00012184555134010524,
+      "loss": 1.0349,
+      "step": 3294
+    },
+    {
+      "epoch": 1.1955555555555555,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.00012179659772365684,
+      "loss": 0.9786,
+      "step": 3296
+    },
+    {
+      "epoch": 1.1962811791383219,
+      "grad_norm": 0.19921875,
+      "learning_rate": 0.00012174764410720843,
+      "loss": 0.9701,
+      "step": 3298
+    },
+    {
+      "epoch": 1.1970068027210885,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.00012169869049076001,
+      "loss": 1.0021,
+      "step": 3300
+    },
+    {
+      "epoch": 1.1970068027210885,
+      "eval_loss": 1.014991044998169,
+      "eval_runtime": 91.949,
+      "eval_samples_per_second": 19.576,
+      "eval_steps_per_second": 0.62,
+      "step": 3300
+    },
+    {
+      "epoch": 1.197732426303855,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.00012164973687431161,
+      "loss": 0.8863,
+      "step": 3302
+    },
+    {
+      "epoch": 1.1984580498866213,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.00012160078325786316,
+      "loss": 1.0103,
+      "step": 3304
+    },
+    {
+      "epoch": 1.1991836734693877,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.00012155182964141476,
+      "loss": 0.9499,
+      "step": 3306
+    },
+    {
+      "epoch": 1.199909297052154,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.00012150287602496636,
+      "loss": 1.0125,
+      "step": 3308
+    },
+    {
+      "epoch": 1.2006349206349207,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.00012145392240851794,
+      "loss": 0.9703,
+      "step": 3310
+    },
+    {
+      "epoch": 1.2013605442176871,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.00012140496879206953,
+      "loss": 0.9431,
+      "step": 3312
+    },
+    {
+      "epoch": 1.2020861678004535,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.00012135601517562109,
+      "loss": 0.9832,
+      "step": 3314
+    },
+    {
+      "epoch": 1.20281179138322,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.00012130706155917269,
+      "loss": 0.9843,
+      "step": 3316
+    },
+    {
+      "epoch": 1.2035374149659863,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.00012125810794272428,
+      "loss": 0.9591,
+      "step": 3318
+    },
+    {
+      "epoch": 1.2042630385487527,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.00012120915432627586,
+      "loss": 0.9599,
+      "step": 3320
+    },
+    {
+      "epoch": 1.2049886621315193,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.00012116020070982746,
+      "loss": 1.0021,
+      "step": 3322
+    },
+    {
+      "epoch": 1.2057142857142857,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.00012111124709337901,
+      "loss": 1.0136,
+      "step": 3324
+    },
+    {
+      "epoch": 1.2064399092970521,
+      "grad_norm": 0.205078125,
+      "learning_rate": 0.00012106229347693061,
+      "loss": 0.9422,
+      "step": 3326
+    },
+    {
+      "epoch": 1.2071655328798185,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.0001210133398604822,
+      "loss": 1.0347,
+      "step": 3328
+    },
+    {
+      "epoch": 1.2078911564625852,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.00012096438624403379,
+      "loss": 1.0006,
+      "step": 3330
+    },
+    {
+      "epoch": 1.2086167800453516,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.00012091543262758538,
+      "loss": 0.9784,
+      "step": 3332
+    },
+    {
+      "epoch": 1.209342403628118,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.00012086647901113694,
+      "loss": 1.0127,
+      "step": 3334
+    },
+    {
+      "epoch": 1.2100680272108844,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.00012081752539468854,
+      "loss": 1.0538,
+      "step": 3336
+    },
+    {
+      "epoch": 1.2107936507936508,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.00012076857177824013,
+      "loss": 0.9858,
+      "step": 3338
+    },
+    {
+      "epoch": 1.2115192743764172,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.00012071961816179171,
+      "loss": 0.978,
+      "step": 3340
+    },
+    {
+      "epoch": 1.2122448979591836,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.00012067066454534331,
+      "loss": 0.9754,
+      "step": 3342
+    },
+    {
+      "epoch": 1.2129705215419502,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.00012062171092889486,
+      "loss": 0.9934,
+      "step": 3344
+    },
+    {
+      "epoch": 1.2136961451247166,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.00012057275731244646,
+      "loss": 0.9599,
+      "step": 3346
+    },
+    {
+      "epoch": 1.214421768707483,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.00012052380369599806,
+      "loss": 1.0248,
+      "step": 3348
+    },
+    {
+      "epoch": 1.2151473922902494,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.00012047485007954964,
+      "loss": 1.0095,
+      "step": 3350
+    },
+    {
+      "epoch": 1.215873015873016,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.00012042589646310121,
+      "loss": 0.9968,
+      "step": 3352
+    },
+    {
+      "epoch": 1.2165986394557824,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.00012037694284665279,
+      "loss": 0.9093,
+      "step": 3354
+    },
+    {
+      "epoch": 1.2173242630385488,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.00012032798923020439,
+      "loss": 0.9017,
+      "step": 3356
+    },
+    {
+      "epoch": 1.2180498866213152,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.00012027903561375598,
+      "loss": 0.9775,
+      "step": 3358
+    },
+    {
+      "epoch": 1.2187755102040816,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.00012023008199730756,
+      "loss": 1.0245,
+      "step": 3360
+    },
+    {
+      "epoch": 1.219501133786848,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.00012018112838085913,
+      "loss": 1.0026,
+      "step": 3362
+    },
+    {
+      "epoch": 1.2202267573696144,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.00012013217476441071,
+      "loss": 0.991,
+      "step": 3364
+    },
+    {
+      "epoch": 1.220952380952381,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.00012008322114796231,
+      "loss": 0.9144,
+      "step": 3366
+    },
+    {
+      "epoch": 1.2216780045351474,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.0001200342675315139,
+      "loss": 0.9246,
+      "step": 3368
+    },
+    {
+      "epoch": 1.2224036281179138,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.00011998531391506549,
+      "loss": 0.991,
+      "step": 3370
+    },
+    {
+      "epoch": 1.2231292517006802,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.00011993636029861706,
+      "loss": 1.0354,
+      "step": 3372
+    },
+    {
+      "epoch": 1.2238548752834468,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.00011988740668216864,
+      "loss": 1.0143,
+      "step": 3374
+    },
+    {
+      "epoch": 1.2245804988662132,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.00011983845306572024,
+      "loss": 1.0311,
+      "step": 3376
+    },
+    {
+      "epoch": 1.2253061224489796,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.00011978949944927183,
+      "loss": 0.9852,
+      "step": 3378
+    },
+    {
+      "epoch": 1.226031746031746,
+      "grad_norm": 0.18359375,
+      "learning_rate": 0.00011974054583282341,
+      "loss": 0.9433,
+      "step": 3380
+    },
+    {
+      "epoch": 1.2267573696145124,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.00011969159221637498,
+      "loss": 0.9716,
+      "step": 3382
+    },
+    {
+      "epoch": 1.2274829931972788,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.00011964263859992656,
+      "loss": 0.9286,
+      "step": 3384
+    },
+    {
+      "epoch": 1.2282086167800454,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.00011959368498347816,
+      "loss": 0.9936,
+      "step": 3386
+    },
+    {
+      "epoch": 1.2289342403628118,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.00011954473136702976,
+      "loss": 0.9766,
+      "step": 3388
+    },
+    {
+      "epoch": 1.2296598639455782,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.00011949577775058134,
+      "loss": 0.9575,
+      "step": 3390
+    },
+    {
+      "epoch": 1.2303854875283446,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.00011944682413413291,
+      "loss": 0.9791,
+      "step": 3392
+    },
+    {
+      "epoch": 1.231111111111111,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.00011939787051768449,
+      "loss": 1.0045,
+      "step": 3394
+    },
+    {
+      "epoch": 1.2318367346938777,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.00011934891690123609,
+      "loss": 1.0042,
+      "step": 3396
+    },
+    {
+      "epoch": 1.232562358276644,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.00011929996328478768,
+      "loss": 0.9742,
+      "step": 3398
+    },
+    {
+      "epoch": 1.2332879818594105,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 0.00011925100966833926,
+      "loss": 0.9748,
+      "step": 3400
+    },
+    {
+      "epoch": 1.2332879818594105,
+      "eval_loss": 1.0148086547851562,
+      "eval_runtime": 94.1179,
+      "eval_samples_per_second": 19.125,
+      "eval_steps_per_second": 0.606,
+      "step": 3400
+    },
+    {
+      "epoch": 1.2340136054421769,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.00011920205605189083,
+      "loss": 0.9734,
+      "step": 3402
+    },
+    {
+      "epoch": 1.2347392290249433,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.00011915310243544241,
+      "loss": 0.9926,
+      "step": 3404
+    },
+    {
+      "epoch": 1.2354648526077097,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.00011910414881899401,
+      "loss": 0.9998,
+      "step": 3406
+    },
+    {
+      "epoch": 1.2361904761904763,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.0001190551952025456,
+      "loss": 0.9564,
+      "step": 3408
+    },
+    {
+      "epoch": 1.2369160997732427,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.00011900624158609719,
+      "loss": 0.9322,
+      "step": 3410
+    },
+    {
+      "epoch": 1.237641723356009,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.00011895728796964876,
+      "loss": 0.9525,
+      "step": 3412
+    },
+    {
+      "epoch": 1.2383673469387755,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.00011890833435320034,
+      "loss": 1.0165,
+      "step": 3414
+    },
+    {
+      "epoch": 1.2390929705215419,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.00011885938073675194,
+      "loss": 0.9926,
+      "step": 3416
+    },
+    {
+      "epoch": 1.2398185941043085,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.00011881042712030353,
+      "loss": 0.9599,
+      "step": 3418
+    },
+    {
+      "epoch": 1.240544217687075,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.00011876147350385511,
+      "loss": 0.9723,
+      "step": 3420
+    },
+    {
+      "epoch": 1.2412698412698413,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.00011871251988740668,
+      "loss": 1.0077,
+      "step": 3422
+    },
+    {
+      "epoch": 1.2419954648526077,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.00011866356627095826,
+      "loss": 0.9708,
+      "step": 3424
+    },
+    {
+      "epoch": 1.242721088435374,
+      "grad_norm": 0.17578125,
+      "learning_rate": 0.00011861461265450986,
+      "loss": 1.0205,
+      "step": 3426
+    },
+    {
+      "epoch": 1.2434467120181405,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.00011856565903806146,
+      "loss": 0.9795,
+      "step": 3428
+    },
+    {
+      "epoch": 1.2441723356009071,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.00011851670542161301,
+      "loss": 1.0165,
+      "step": 3430
+    },
+    {
+      "epoch": 1.2448979591836735,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.00011846775180516461,
+      "loss": 1.0622,
+      "step": 3432
+    },
+    {
+      "epoch": 1.24562358276644,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.00011841879818871619,
+      "loss": 1.0099,
+      "step": 3434
+    },
+    {
+      "epoch": 1.2463492063492063,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.00011836984457226779,
+      "loss": 0.9006,
+      "step": 3436
+    },
+    {
+      "epoch": 1.2470748299319727,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.00011832089095581938,
+      "loss": 1.0503,
+      "step": 3438
+    },
+    {
+      "epoch": 1.2478004535147393,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.00011827193733937094,
+      "loss": 0.9561,
+      "step": 3440
+    },
+    {
+      "epoch": 1.2485260770975057,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.00011822298372292253,
+      "loss": 0.9966,
+      "step": 3442
+    },
+    {
+      "epoch": 1.2492517006802721,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.00011817403010647411,
+      "loss": 1.0036,
+      "step": 3444
+    },
+    {
+      "epoch": 1.2499773242630385,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.00011812507649002571,
+      "loss": 1.0254,
+      "step": 3446
+    },
+    {
+      "epoch": 1.250702947845805,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.0001180761228735773,
+      "loss": 0.9544,
+      "step": 3448
+    },
+    {
+      "epoch": 1.2514285714285713,
+      "grad_norm": 0.17578125,
+      "learning_rate": 0.00011802716925712886,
+      "loss": 0.9625,
+      "step": 3450
+    },
+    {
+      "epoch": 1.2521541950113377,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.00011797821564068046,
+      "loss": 0.9375,
+      "step": 3452
+    },
+    {
+      "epoch": 1.2528798185941044,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.00011792926202423204,
+      "loss": 1.0041,
+      "step": 3454
+    },
+    {
+      "epoch": 1.2536054421768708,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.00011788030840778364,
+      "loss": 0.9661,
+      "step": 3456
+    },
+    {
+      "epoch": 1.2543310657596372,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.00011783135479133523,
+      "loss": 1.0311,
+      "step": 3458
+    },
+    {
+      "epoch": 1.2550566893424036,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.00011778240117488679,
+      "loss": 0.9824,
+      "step": 3460
+    },
+    {
+      "epoch": 1.2557823129251702,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.00011773344755843838,
+      "loss": 0.9351,
+      "step": 3462
+    },
+    {
+      "epoch": 1.2565079365079366,
+      "grad_norm": 0.181640625,
+      "learning_rate": 0.00011768449394198996,
+      "loss": 1.0692,
+      "step": 3464
+    },
+    {
+      "epoch": 1.257233560090703,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.00011763554032554156,
+      "loss": 1.0367,
+      "step": 3466
+    },
+    {
+      "epoch": 1.2579591836734694,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.00011758658670909316,
+      "loss": 0.9531,
+      "step": 3468
+    },
+    {
+      "epoch": 1.2586848072562358,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.00011753763309264471,
+      "loss": 1.0415,
+      "step": 3470
+    },
+    {
+      "epoch": 1.2594104308390022,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.00011748867947619631,
+      "loss": 0.9577,
+      "step": 3472
+    },
+    {
+      "epoch": 1.2601360544217686,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.00011743972585974789,
+      "loss": 1.0151,
+      "step": 3474
+    },
+    {
+      "epoch": 1.2608616780045352,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.00011739077224329948,
+      "loss": 0.9575,
+      "step": 3476
+    },
+    {
+      "epoch": 1.2615873015873016,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.00011734181862685108,
+      "loss": 0.9676,
+      "step": 3478
+    },
+    {
+      "epoch": 1.262312925170068,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.00011729286501040264,
+      "loss": 0.907,
+      "step": 3480
+    },
+    {
+      "epoch": 1.2630385487528344,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.00011724391139395423,
+      "loss": 0.9871,
+      "step": 3482
+    },
+    {
+      "epoch": 1.263764172335601,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.00011719495777750581,
+      "loss": 0.9981,
+      "step": 3484
+    },
+    {
+      "epoch": 1.2644897959183674,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.00011714600416105741,
+      "loss": 0.993,
+      "step": 3486
+    },
+    {
+      "epoch": 1.2652154195011338,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.000117097050544609,
+      "loss": 1.0024,
+      "step": 3488
+    },
+    {
+      "epoch": 1.2659410430839002,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.00011704809692816056,
+      "loss": 0.9617,
+      "step": 3490
+    },
+    {
+      "epoch": 1.2666666666666666,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.00011699914331171216,
+      "loss": 0.9186,
+      "step": 3492
+    },
+    {
+      "epoch": 1.267392290249433,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.00011695018969526374,
+      "loss": 0.9795,
+      "step": 3494
+    },
+    {
+      "epoch": 1.2681179138321996,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.00011690123607881533,
+      "loss": 1.0009,
+      "step": 3496
+    },
+    {
+      "epoch": 1.268843537414966,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.0001168522824623669,
+      "loss": 1.001,
+      "step": 3498
+    },
+    {
+      "epoch": 1.2695691609977324,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.00011680332884591849,
+      "loss": 1.0089,
+      "step": 3500
+    },
+    {
+      "epoch": 1.2695691609977324,
+      "eval_loss": 1.0145820379257202,
+      "eval_runtime": 87.8413,
+      "eval_samples_per_second": 20.491,
+      "eval_steps_per_second": 0.649,
+      "step": 3500
+    },
+    {
+      "epoch": 1.2702947845804988,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.00011675437522947008,
+      "loss": 1.0022,
+      "step": 3502
+    },
+    {
+      "epoch": 1.2710204081632652,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.00011670542161302166,
+      "loss": 1.0038,
+      "step": 3504
+    },
+    {
+      "epoch": 1.2717460317460318,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.00011665646799657326,
+      "loss": 0.9343,
+      "step": 3506
+    },
+    {
+      "epoch": 1.2724716553287982,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.00011660751438012483,
+      "loss": 1.0078,
+      "step": 3508
+    },
+    {
+      "epoch": 1.2731972789115646,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.00011655856076367641,
+      "loss": 0.9831,
+      "step": 3510
+    },
+    {
+      "epoch": 1.273922902494331,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.000116509607147228,
+      "loss": 1.0197,
+      "step": 3512
+    },
+    {
+      "epoch": 1.2746485260770974,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.00011646065353077959,
+      "loss": 1.0065,
+      "step": 3514
+    },
+    {
+      "epoch": 1.2753741496598638,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.00011641169991433118,
+      "loss": 0.9438,
+      "step": 3516
+    },
+    {
+      "epoch": 1.2760997732426305,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.00011636274629788275,
+      "loss": 1.0004,
+      "step": 3518
+    },
+    {
+      "epoch": 1.2768253968253969,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.00011631379268143434,
+      "loss": 0.9518,
+      "step": 3520
+    },
+    {
+      "epoch": 1.2775510204081633,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.00011626483906498593,
+      "loss": 0.9643,
+      "step": 3522
+    },
+    {
+      "epoch": 1.2782766439909297,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.00011621588544853751,
+      "loss": 0.977,
+      "step": 3524
+    },
+    {
+      "epoch": 1.2790022675736963,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.00011616693183208911,
+      "loss": 1.0163,
+      "step": 3526
+    },
+    {
+      "epoch": 1.2797278911564627,
+      "grad_norm": 0.2109375,
+      "learning_rate": 0.00011611797821564068,
+      "loss": 0.9853,
+      "step": 3528
+    },
+    {
+      "epoch": 1.280453514739229,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.00011606902459919226,
+      "loss": 1.0327,
+      "step": 3530
+    },
+    {
+      "epoch": 1.2811791383219955,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.00011602007098274386,
+      "loss": 1.039,
+      "step": 3532
+    },
+    {
+      "epoch": 1.2819047619047619,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.00011597111736629544,
+      "loss": 1.0176,
+      "step": 3534
+    },
+    {
+      "epoch": 1.2826303854875283,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.00011592216374984703,
+      "loss": 0.9863,
+      "step": 3536
+    },
+    {
+      "epoch": 1.2833560090702947,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.0001158732101333986,
+      "loss": 0.9882,
+      "step": 3538
+    },
+    {
+      "epoch": 1.2840816326530613,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.00011582425651695019,
+      "loss": 1.0326,
+      "step": 3540
+    },
+    {
+      "epoch": 1.2848072562358277,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.00011577530290050178,
+      "loss": 0.9556,
+      "step": 3542
+    },
+    {
+      "epoch": 1.285532879818594,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.00011572634928405336,
+      "loss": 0.9931,
+      "step": 3544
+    },
+    {
+      "epoch": 1.2862585034013605,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.00011567739566760496,
+      "loss": 0.9452,
+      "step": 3546
+    },
+    {
+      "epoch": 1.2869841269841271,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.00011562844205115653,
+      "loss": 0.9842,
+      "step": 3548
+    },
+    {
+      "epoch": 1.2877097505668935,
+      "grad_norm": 0.197265625,
+      "learning_rate": 0.00011557948843470811,
+      "loss": 1.0206,
+      "step": 3550
+    },
+    {
+      "epoch": 1.28843537414966,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.0001155305348182597,
+      "loss": 0.909,
+      "step": 3552
+    },
+    {
+      "epoch": 1.2891609977324263,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.00011548158120181129,
+      "loss": 1.0186,
+      "step": 3554
+    },
+    {
+      "epoch": 1.2898866213151927,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.00011543262758536288,
+      "loss": 1.0197,
+      "step": 3556
+    },
+    {
+      "epoch": 1.2906122448979591,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.00011538367396891445,
+      "loss": 0.9857,
+      "step": 3558
+    },
+    {
+      "epoch": 1.2913378684807255,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.00011533472035246604,
+      "loss": 0.9722,
+      "step": 3560
+    },
+    {
+      "epoch": 1.2920634920634921,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.00011528576673601763,
+      "loss": 1.0129,
+      "step": 3562
+    },
+    {
+      "epoch": 1.2927891156462585,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.00011523681311956921,
+      "loss": 0.9774,
+      "step": 3564
+    },
+    {
+      "epoch": 1.293514739229025,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.00011518785950312081,
+      "loss": 1.0214,
+      "step": 3566
+    },
+    {
+      "epoch": 1.2942403628117913,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.00011513890588667238,
+      "loss": 0.9524,
+      "step": 3568
+    },
+    {
+      "epoch": 1.294965986394558,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.00011508995227022396,
+      "loss": 0.9879,
+      "step": 3570
+    },
+    {
+      "epoch": 1.2956916099773244,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.00011504099865377556,
+      "loss": 0.946,
+      "step": 3572
+    },
+    {
+      "epoch": 1.2964172335600908,
+      "grad_norm": 0.17578125,
+      "learning_rate": 0.00011499204503732714,
+      "loss": 1.0078,
+      "step": 3574
+    },
+    {
+      "epoch": 1.2971428571428572,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.00011494309142087871,
+      "loss": 1.0155,
+      "step": 3576
+    },
+    {
+      "epoch": 1.2978684807256236,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.0001148941378044303,
+      "loss": 0.9912,
+      "step": 3578
+    },
+    {
+      "epoch": 1.29859410430839,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.00011484518418798189,
+      "loss": 0.9606,
+      "step": 3580
+    },
+    {
+      "epoch": 1.2993197278911564,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.00011479623057153348,
+      "loss": 0.9986,
+      "step": 3582
+    },
+    {
+      "epoch": 1.300045351473923,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.00011474727695508506,
+      "loss": 0.9684,
+      "step": 3584
+    },
+    {
+      "epoch": 1.3007709750566894,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.00011469832333863663,
+      "loss": 0.9675,
+      "step": 3586
+    },
+    {
+      "epoch": 1.3014965986394558,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.00011464936972218823,
+      "loss": 0.971,
+      "step": 3588
+    },
+    {
+      "epoch": 1.3022222222222222,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.00011460041610573981,
+      "loss": 0.9238,
+      "step": 3590
+    },
+    {
+      "epoch": 1.3029478458049888,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.0001145514624892914,
+      "loss": 0.9196,
+      "step": 3592
+    },
+    {
+      "epoch": 1.3036734693877552,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.00011450250887284299,
+      "loss": 0.9861,
+      "step": 3594
+    },
+    {
+      "epoch": 1.3043990929705216,
+      "grad_norm": 0.185546875,
+      "learning_rate": 0.00011445355525639456,
+      "loss": 1.0305,
+      "step": 3596
+    },
+    {
+      "epoch": 1.305124716553288,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.00011440460163994615,
+      "loss": 0.9368,
+      "step": 3598
+    },
+    {
+      "epoch": 1.3058503401360544,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.00011435564802349774,
+      "loss": 0.9788,
+      "step": 3600
+    },
+    {
+      "epoch": 1.3058503401360544,
+      "eval_loss": 1.014317274093628,
+      "eval_runtime": 87.2865,
+      "eval_samples_per_second": 20.622,
+      "eval_steps_per_second": 0.653,
+      "step": 3600
+    },
+    {
+      "epoch": 1.3065759637188208,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.00011430669440704933,
+      "loss": 0.9431,
+      "step": 3602
+    },
+    {
+      "epoch": 1.3073015873015872,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.00011425774079060091,
+      "loss": 0.9195,
+      "step": 3604
+    },
+    {
+      "epoch": 1.3080272108843538,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.00011420878717415248,
+      "loss": 0.9597,
+      "step": 3606
+    },
+    {
+      "epoch": 1.3087528344671202,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.00011415983355770408,
+      "loss": 0.9325,
+      "step": 3608
+    },
+    {
+      "epoch": 1.3094784580498866,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.00011411087994125566,
+      "loss": 0.9972,
+      "step": 3610
+    },
+    {
+      "epoch": 1.310204081632653,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.00011406192632480726,
+      "loss": 1.0023,
+      "step": 3612
+    },
+    {
+      "epoch": 1.3109297052154196,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.00011401297270835884,
+      "loss": 0.9872,
+      "step": 3614
+    },
+    {
+      "epoch": 1.311655328798186,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.00011396401909191041,
+      "loss": 1.0391,
+      "step": 3616
+    },
+    {
+      "epoch": 1.3123809523809524,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.000113915065475462,
+      "loss": 0.9912,
+      "step": 3618
+    },
+    {
+      "epoch": 1.3131065759637188,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.00011386611185901359,
+      "loss": 1.0563,
+      "step": 3620
+    },
+    {
+      "epoch": 1.3138321995464852,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.00011381715824256518,
+      "loss": 0.9985,
+      "step": 3622
+    },
+    {
+      "epoch": 1.3145578231292516,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.00011376820462611676,
+      "loss": 0.9683,
+      "step": 3624
+    },
+    {
+      "epoch": 1.315283446712018,
+      "grad_norm": 0.17578125,
+      "learning_rate": 0.00011371925100966833,
+      "loss": 0.9682,
+      "step": 3626
+    },
+    {
+      "epoch": 1.3160090702947846,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.00011367029739321993,
+      "loss": 1.0611,
+      "step": 3628
+    },
+    {
+      "epoch": 1.316734693877551,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.00011362134377677151,
+      "loss": 0.9994,
+      "step": 3630
+    },
+    {
+      "epoch": 1.3174603174603174,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.0001135723901603231,
+      "loss": 0.9085,
+      "step": 3632
+    },
+    {
+      "epoch": 1.3181859410430838,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.00011352343654387469,
+      "loss": 0.9891,
+      "step": 3634
+    },
+    {
+      "epoch": 1.3189115646258505,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.00011347448292742626,
+      "loss": 1.0091,
+      "step": 3636
+    },
+    {
+      "epoch": 1.3196371882086169,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.00011342552931097785,
+      "loss": 0.9623,
+      "step": 3638
+    },
+    {
+      "epoch": 1.3203628117913833,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.00011337657569452944,
+      "loss": 1.0065,
+      "step": 3640
+    },
+    {
+      "epoch": 1.3210884353741497,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.00011332762207808103,
+      "loss": 0.9271,
+      "step": 3642
+    },
+    {
+      "epoch": 1.321814058956916,
+      "grad_norm": 0.17578125,
+      "learning_rate": 0.0001132786684616326,
+      "loss": 0.9958,
+      "step": 3644
+    },
+    {
+      "epoch": 1.3225396825396825,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.00011322971484518418,
+      "loss": 0.9128,
+      "step": 3646
+    },
+    {
+      "epoch": 1.3232653061224489,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.00011318076122873578,
+      "loss": 0.9164,
+      "step": 3648
+    },
+    {
+      "epoch": 1.3239909297052155,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.00011313180761228736,
+      "loss": 1.0168,
+      "step": 3650
+    },
+    {
+      "epoch": 1.3247165532879819,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.00011308285399583896,
+      "loss": 0.9683,
+      "step": 3652
+    },
+    {
+      "epoch": 1.3254421768707483,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.00011303390037939052,
+      "loss": 0.971,
+      "step": 3654
+    },
+    {
+      "epoch": 1.3261678004535147,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.00011298494676294211,
+      "loss": 0.9244,
+      "step": 3656
+    },
+    {
+      "epoch": 1.3268934240362813,
+      "grad_norm": 0.2060546875,
+      "learning_rate": 0.0001129359931464937,
+      "loss": 0.9584,
+      "step": 3658
+    },
+    {
+      "epoch": 1.3276190476190477,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.00011288703953004529,
+      "loss": 0.882,
+      "step": 3660
+    },
+    {
+      "epoch": 1.328344671201814,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.00011283808591359688,
+      "loss": 0.9848,
+      "step": 3662
+    },
+    {
+      "epoch": 1.3290702947845805,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.00011278913229714845,
+      "loss": 1.1064,
+      "step": 3664
+    },
+    {
+      "epoch": 1.329795918367347,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.00011274017868070003,
+      "loss": 0.9595,
+      "step": 3666
+    },
+    {
+      "epoch": 1.3305215419501133,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.00011269122506425163,
+      "loss": 0.9298,
+      "step": 3668
+    },
+    {
+      "epoch": 1.3312471655328797,
+      "grad_norm": 0.185546875,
+      "learning_rate": 0.00011264227144780321,
+      "loss": 1.0159,
+      "step": 3670
+    },
+    {
+      "epoch": 1.3319727891156463,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.0001125933178313548,
+      "loss": 0.9367,
+      "step": 3672
+    },
+    {
+      "epoch": 1.3326984126984127,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.00011254436421490637,
+      "loss": 0.9847,
+      "step": 3674
+    },
+    {
+      "epoch": 1.3334240362811791,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.00011249541059845796,
+      "loss": 0.9506,
+      "step": 3676
+    },
+    {
+      "epoch": 1.3341496598639455,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.00011244645698200955,
+      "loss": 1.0063,
+      "step": 3678
+    },
+    {
+      "epoch": 1.3348752834467121,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.00011239750336556114,
+      "loss": 0.9438,
+      "step": 3680
+    },
+    {
+      "epoch": 1.3356009070294785,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.00011234854974911273,
+      "loss": 0.968,
+      "step": 3682
+    },
+    {
+      "epoch": 1.336326530612245,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.0001122995961326643,
+      "loss": 0.9975,
+      "step": 3684
+    },
+    {
+      "epoch": 1.3370521541950113,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.00011225064251621588,
+      "loss": 1.0121,
+      "step": 3686
+    },
+    {
+      "epoch": 1.3377777777777777,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.00011220168889976748,
+      "loss": 0.9627,
+      "step": 3688
+    },
+    {
+      "epoch": 1.3385034013605441,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.00011215273528331906,
+      "loss": 1.0203,
+      "step": 3690
+    },
+    {
+      "epoch": 1.3392290249433105,
+      "grad_norm": 0.185546875,
+      "learning_rate": 0.00011210378166687066,
+      "loss": 1.0594,
+      "step": 3692
+    },
+    {
+      "epoch": 1.3399546485260772,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.00011205482805042222,
+      "loss": 0.9768,
+      "step": 3694
+    },
+    {
+      "epoch": 1.3406802721088436,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.00011200587443397381,
+      "loss": 1.027,
+      "step": 3696
+    },
+    {
+      "epoch": 1.34140589569161,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.0001119569208175254,
+      "loss": 1.0098,
+      "step": 3698
+    },
+    {
+      "epoch": 1.3421315192743764,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.00011190796720107698,
+      "loss": 0.975,
+      "step": 3700
+    },
+    {
+      "epoch": 1.3421315192743764,
+      "eval_loss": 1.0139893293380737,
+      "eval_runtime": 87.2842,
+      "eval_samples_per_second": 20.622,
+      "eval_steps_per_second": 0.653,
+      "step": 3700
+    },
+    {
+      "epoch": 1.342857142857143,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.00011185901358462858,
+      "loss": 0.9606,
+      "step": 3702
+    },
+    {
+      "epoch": 1.3435827664399094,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 0.00011181005996818015,
+      "loss": 1.0558,
+      "step": 3704
+    },
+    {
+      "epoch": 1.3443083900226758,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.00011176110635173173,
+      "loss": 0.9632,
+      "step": 3706
+    },
+    {
+      "epoch": 1.3450340136054422,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.00011171215273528333,
+      "loss": 1.0016,
+      "step": 3708
+    },
+    {
+      "epoch": 1.3457596371882086,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.00011166319911883491,
+      "loss": 0.9078,
+      "step": 3710
+    },
+    {
+      "epoch": 1.346485260770975,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.0001116142455023865,
+      "loss": 1.0287,
+      "step": 3712
+    },
+    {
+      "epoch": 1.3472108843537414,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.00011156529188593807,
+      "loss": 0.9694,
+      "step": 3714
+    },
+    {
+      "epoch": 1.347936507936508,
+      "grad_norm": 0.2001953125,
+      "learning_rate": 0.00011151633826948966,
+      "loss": 0.9519,
+      "step": 3716
+    },
+    {
+      "epoch": 1.3486621315192744,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.00011146738465304125,
+      "loss": 0.9624,
+      "step": 3718
+    },
+    {
+      "epoch": 1.3493877551020408,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.00011141843103659283,
+      "loss": 0.9302,
+      "step": 3720
+    },
+    {
+      "epoch": 1.3501133786848072,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.0001113694774201444,
+      "loss": 0.9119,
+      "step": 3722
+    },
+    {
+      "epoch": 1.3508390022675738,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.000111320523803696,
+      "loss": 0.9269,
+      "step": 3724
+    },
+    {
+      "epoch": 1.3515646258503402,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.00011127157018724758,
+      "loss": 1.0261,
+      "step": 3726
+    },
+    {
+      "epoch": 1.3522902494331066,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.00011122261657079918,
+      "loss": 1.0123,
+      "step": 3728
+    },
+    {
+      "epoch": 1.353015873015873,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.00011117366295435076,
+      "loss": 1.0042,
+      "step": 3730
+    },
+    {
+      "epoch": 1.3537414965986394,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.00011112470933790233,
+      "loss": 0.9707,
+      "step": 3732
+    },
+    {
+      "epoch": 1.3544671201814058,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.00011107575572145392,
+      "loss": 0.9504,
+      "step": 3734
+    },
+    {
+      "epoch": 1.3551927437641722,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.0001110268021050055,
+      "loss": 0.9999,
+      "step": 3736
+    },
+    {
+      "epoch": 1.3559183673469388,
+      "grad_norm": 0.189453125,
+      "learning_rate": 0.0001109778484885571,
+      "loss": 1.0107,
+      "step": 3738
+    },
+    {
+      "epoch": 1.3566439909297052,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.00011092889487210868,
+      "loss": 0.9823,
+      "step": 3740
+    },
+    {
+      "epoch": 1.3573696145124716,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.00011087994125566025,
+      "loss": 0.9654,
+      "step": 3742
+    },
+    {
+      "epoch": 1.358095238095238,
+      "grad_norm": 0.2099609375,
+      "learning_rate": 0.00011083098763921185,
+      "loss": 0.983,
+      "step": 3744
+    },
+    {
+      "epoch": 1.3588208616780046,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.00011078203402276343,
+      "loss": 1.0468,
+      "step": 3746
+    },
+    {
+      "epoch": 1.359546485260771,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.00011073308040631503,
+      "loss": 0.9342,
+      "step": 3748
+    },
+    {
+      "epoch": 1.3602721088435374,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.00011068412678986661,
+      "loss": 0.9798,
+      "step": 3750
+    },
+    {
+      "epoch": 1.3609977324263038,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.00011063517317341818,
+      "loss": 0.9705,
+      "step": 3752
+    },
+    {
+      "epoch": 1.3617233560090702,
+      "grad_norm": 0.1884765625,
+      "learning_rate": 0.00011058621955696977,
+      "loss": 1.0327,
+      "step": 3754
+    },
+    {
+      "epoch": 1.3624489795918366,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.00011053726594052136,
+      "loss": 1.0191,
+      "step": 3756
+    },
+    {
+      "epoch": 1.363174603174603,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.00011048831232407295,
+      "loss": 0.9705,
+      "step": 3758
+    },
+    {
+      "epoch": 1.3639002267573697,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.00011043935870762453,
+      "loss": 0.9572,
+      "step": 3760
+    },
+    {
+      "epoch": 1.364625850340136,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.0001103904050911761,
+      "loss": 0.9237,
+      "step": 3762
+    },
+    {
+      "epoch": 1.3653514739229025,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.0001103414514747277,
+      "loss": 0.8971,
+      "step": 3764
+    },
+    {
+      "epoch": 1.3660770975056689,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.00011029249785827928,
+      "loss": 0.9106,
+      "step": 3766
+    },
+    {
+      "epoch": 1.3668027210884355,
+      "grad_norm": 0.216796875,
+      "learning_rate": 0.00011024354424183088,
+      "loss": 1.0726,
+      "step": 3768
+    },
+    {
+      "epoch": 1.3675283446712019,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.00011019459062538246,
+      "loss": 0.946,
+      "step": 3770
+    },
+    {
+      "epoch": 1.3682539682539683,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.00011014563700893403,
+      "loss": 0.936,
+      "step": 3772
+    },
+    {
+      "epoch": 1.3689795918367347,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.00011009668339248562,
+      "loss": 0.9972,
+      "step": 3774
+    },
+    {
+      "epoch": 1.369705215419501,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.0001100477297760372,
+      "loss": 1.0203,
+      "step": 3776
+    },
+    {
+      "epoch": 1.3704308390022675,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 0.0001099987761595888,
+      "loss": 1.016,
+      "step": 3778
+    },
+    {
+      "epoch": 1.3711564625850339,
+      "grad_norm": 0.18359375,
+      "learning_rate": 0.00010994982254314038,
+      "loss": 0.9996,
+      "step": 3780
+    },
+    {
+      "epoch": 1.3718820861678005,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.00010990086892669195,
+      "loss": 0.9701,
+      "step": 3782
+    },
+    {
+      "epoch": 1.372607709750567,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.00010985191531024355,
+      "loss": 0.9935,
+      "step": 3784
+    },
+    {
+      "epoch": 1.3733333333333333,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.00010980296169379513,
+      "loss": 0.9978,
+      "step": 3786
+    },
+    {
+      "epoch": 1.3740589569160997,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.00010975400807734673,
+      "loss": 0.9602,
+      "step": 3788
+    },
+    {
+      "epoch": 1.3747845804988663,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.0001097050544608983,
+      "loss": 0.9894,
+      "step": 3790
+    },
+    {
+      "epoch": 1.3755102040816327,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.00010965610084444988,
+      "loss": 0.9851,
+      "step": 3792
+    },
+    {
+      "epoch": 1.3762358276643991,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.00010960714722800147,
+      "loss": 1.0069,
+      "step": 3794
+    },
+    {
+      "epoch": 1.3769614512471655,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.00010955819361155306,
+      "loss": 0.9355,
+      "step": 3796
+    },
+    {
+      "epoch": 1.377687074829932,
+      "grad_norm": 0.1884765625,
+      "learning_rate": 0.00010950923999510465,
+      "loss": 1.0524,
+      "step": 3798
+    },
+    {
+      "epoch": 1.3784126984126983,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.00010946028637865622,
+      "loss": 0.9241,
+      "step": 3800
+    },
+    {
+      "epoch": 1.3784126984126983,
+      "eval_loss": 1.0139381885528564,
+      "eval_runtime": 88.7285,
+      "eval_samples_per_second": 20.287,
+      "eval_steps_per_second": 0.642,
+      "step": 3800
+    },
+    {
+      "epoch": 1.379138321995465,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.0001094113327622078,
+      "loss": 0.9646,
+      "step": 3802
+    },
+    {
+      "epoch": 1.3798639455782313,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.0001093623791457594,
+      "loss": 0.9912,
+      "step": 3804
+    },
+    {
+      "epoch": 1.3805895691609977,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.00010931342552931098,
+      "loss": 1.1027,
+      "step": 3806
+    },
+    {
+      "epoch": 1.3813151927437641,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.00010926447191286258,
+      "loss": 1.0653,
+      "step": 3808
+    },
+    {
+      "epoch": 1.3820408163265305,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.00010921551829641415,
+      "loss": 0.8741,
+      "step": 3810
+    },
+    {
+      "epoch": 1.3827664399092972,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.00010916656467996573,
+      "loss": 1.0081,
+      "step": 3812
+    },
+    {
+      "epoch": 1.3834920634920636,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.00010911761106351732,
+      "loss": 1.0412,
+      "step": 3814
+    },
+    {
+      "epoch": 1.38421768707483,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.0001090686574470689,
+      "loss": 1.0026,
+      "step": 3816
+    },
+    {
+      "epoch": 1.3849433106575963,
+      "grad_norm": 0.177734375,
+      "learning_rate": 0.0001090197038306205,
+      "loss": 1.0484,
+      "step": 3818
+    },
+    {
+      "epoch": 1.3856689342403627,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.00010897075021417207,
+      "loss": 1.0069,
+      "step": 3820
+    },
+    {
+      "epoch": 1.3863945578231291,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.00010892179659772365,
+      "loss": 0.9699,
+      "step": 3822
+    },
+    {
+      "epoch": 1.3871201814058958,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.00010887284298127525,
+      "loss": 0.9712,
+      "step": 3824
+    },
+    {
+      "epoch": 1.3878458049886622,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.00010882388936482683,
+      "loss": 0.9634,
+      "step": 3826
+    },
+    {
+      "epoch": 1.3885714285714286,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.00010877493574837843,
+      "loss": 0.9306,
+      "step": 3828
+    },
+    {
+      "epoch": 1.389297052154195,
+      "grad_norm": 0.18359375,
+      "learning_rate": 0.00010872598213193,
+      "loss": 0.9489,
+      "step": 3830
+    },
+    {
+      "epoch": 1.3900226757369616,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.00010867702851548158,
+      "loss": 1.0097,
+      "step": 3832
+    },
+    {
+      "epoch": 1.390748299319728,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.00010862807489903317,
+      "loss": 0.9603,
+      "step": 3834
+    },
+    {
+      "epoch": 1.3914739229024944,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.00010857912128258476,
+      "loss": 0.944,
+      "step": 3836
+    },
+    {
+      "epoch": 1.3921995464852608,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.00010853016766613635,
+      "loss": 0.9558,
+      "step": 3838
+    },
+    {
+      "epoch": 1.3929251700680272,
+      "grad_norm": 0.181640625,
+      "learning_rate": 0.00010848121404968792,
+      "loss": 0.98,
+      "step": 3840
+    },
+    {
+      "epoch": 1.3936507936507936,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.0001084322604332395,
+      "loss": 1.0133,
+      "step": 3842
+    },
+    {
+      "epoch": 1.39437641723356,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.0001083833068167911,
+      "loss": 1.0192,
+      "step": 3844
+    },
+    {
+      "epoch": 1.3951020408163266,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.00010833435320034268,
+      "loss": 1.0125,
+      "step": 3846
+    },
+    {
+      "epoch": 1.395827664399093,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.00010828539958389428,
+      "loss": 0.9468,
+      "step": 3848
+    },
+    {
+      "epoch": 1.3965532879818594,
+      "grad_norm": 0.1884765625,
+      "learning_rate": 0.00010823644596744585,
+      "loss": 0.9093,
+      "step": 3850
+    },
+    {
+      "epoch": 1.3972789115646258,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.00010818749235099743,
+      "loss": 0.9601,
+      "step": 3852
+    },
+    {
+      "epoch": 1.3980045351473924,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.00010813853873454902,
+      "loss": 0.9524,
+      "step": 3854
+    },
+    {
+      "epoch": 1.3987301587301588,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.0001080895851181006,
+      "loss": 0.9915,
+      "step": 3856
+    },
+    {
+      "epoch": 1.3994557823129252,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.0001080406315016522,
+      "loss": 0.9875,
+      "step": 3858
+    },
+    {
+      "epoch": 1.4001814058956916,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.00010799167788520377,
+      "loss": 0.9353,
+      "step": 3860
+    },
+    {
+      "epoch": 1.400907029478458,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.00010794272426875535,
+      "loss": 0.974,
+      "step": 3862
+    },
+    {
+      "epoch": 1.4016326530612244,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.00010789377065230695,
+      "loss": 0.9781,
+      "step": 3864
+    },
+    {
+      "epoch": 1.4023582766439908,
+      "grad_norm": 0.181640625,
+      "learning_rate": 0.00010784481703585853,
+      "loss": 0.9633,
+      "step": 3866
+    },
+    {
+      "epoch": 1.4030839002267574,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.0001077958634194101,
+      "loss": 0.909,
+      "step": 3868
+    },
+    {
+      "epoch": 1.4038095238095238,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.0001077469098029617,
+      "loss": 1.0195,
+      "step": 3870
+    },
+    {
+      "epoch": 1.4045351473922902,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.00010769795618651328,
+      "loss": 1.0009,
+      "step": 3872
+    },
+    {
+      "epoch": 1.4052607709750566,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.00010764900257006487,
+      "loss": 0.941,
+      "step": 3874
+    },
+    {
+      "epoch": 1.4059863945578233,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.00010760004895361646,
+      "loss": 0.8787,
+      "step": 3876
+    },
+    {
+      "epoch": 1.4067120181405897,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.00010755109533716802,
+      "loss": 1.0049,
+      "step": 3878
+    },
+    {
+      "epoch": 1.407437641723356,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.00010750214172071962,
+      "loss": 1.0413,
+      "step": 3880
+    },
+    {
+      "epoch": 1.4081632653061225,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.0001074531881042712,
+      "loss": 1.0002,
+      "step": 3882
+    },
+    {
+      "epoch": 1.4088888888888889,
+      "grad_norm": 0.19140625,
+      "learning_rate": 0.0001074042344878228,
+      "loss": 1.0016,
+      "step": 3884
+    },
+    {
+      "epoch": 1.4096145124716553,
+      "grad_norm": 0.1884765625,
+      "learning_rate": 0.00010735528087137438,
+      "loss": 0.9948,
+      "step": 3886
+    },
+    {
+      "epoch": 1.4103401360544217,
+      "grad_norm": 0.173828125,
+      "learning_rate": 0.00010730632725492595,
+      "loss": 0.9512,
+      "step": 3888
+    },
+    {
+      "epoch": 1.4110657596371883,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.00010725737363847755,
+      "loss": 0.9985,
+      "step": 3890
+    },
+    {
+      "epoch": 1.4117913832199547,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.00010720842002202913,
+      "loss": 0.9329,
+      "step": 3892
+    },
+    {
+      "epoch": 1.412517006802721,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.00010715946640558072,
+      "loss": 0.9697,
+      "step": 3894
+    },
+    {
+      "epoch": 1.4132426303854875,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.0001071105127891323,
+      "loss": 0.9691,
+      "step": 3896
+    },
+    {
+      "epoch": 1.413968253968254,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.00010706155917268387,
+      "loss": 0.9706,
+      "step": 3898
+    },
+    {
+      "epoch": 1.4146938775510205,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.00010701260555623547,
+      "loss": 1.0023,
+      "step": 3900
+    },
+    {
+      "epoch": 1.4146938775510205,
+      "eval_loss": 1.013636589050293,
+      "eval_runtime": 92.7582,
+      "eval_samples_per_second": 19.405,
+      "eval_steps_per_second": 0.615,
+      "step": 3900
+    },
+    {
+      "epoch": 1.415419501133787,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.00010696365193978705,
+      "loss": 0.9634,
+      "step": 3902
+    },
+    {
+      "epoch": 1.4161451247165533,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.00010691469832333865,
+      "loss": 0.9808,
+      "step": 3904
+    },
+    {
+      "epoch": 1.4168707482993197,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.00010686574470689023,
+      "loss": 0.9637,
+      "step": 3906
+    },
+    {
+      "epoch": 1.417596371882086,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.0001068167910904418,
+      "loss": 0.9661,
+      "step": 3908
+    },
+    {
+      "epoch": 1.4183219954648525,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.0001067678374739934,
+      "loss": 0.9844,
+      "step": 3910
+    },
+    {
+      "epoch": 1.4190476190476191,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.00010671888385754498,
+      "loss": 0.9415,
+      "step": 3912
+    },
+    {
+      "epoch": 1.4197732426303855,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.00010666993024109657,
+      "loss": 1.0491,
+      "step": 3914
+    },
+    {
+      "epoch": 1.420498866213152,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 0.00010662097662464816,
+      "loss": 1.007,
+      "step": 3916
+    },
+    {
+      "epoch": 1.4212244897959183,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.00010657202300819972,
+      "loss": 0.9802,
+      "step": 3918
+    },
+    {
+      "epoch": 1.421950113378685,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.00010652306939175132,
+      "loss": 1.0043,
+      "step": 3920
+    },
+    {
+      "epoch": 1.4226757369614513,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.0001064741157753029,
+      "loss": 1.0034,
+      "step": 3922
+    },
+    {
+      "epoch": 1.4234013605442177,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.0001064251621588545,
+      "loss": 0.9387,
+      "step": 3924
+    },
+    {
+      "epoch": 1.4241269841269841,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.00010637620854240608,
+      "loss": 0.9047,
+      "step": 3926
+    },
+    {
+      "epoch": 1.4248526077097505,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.00010632725492595765,
+      "loss": 0.9024,
+      "step": 3928
+    },
+    {
+      "epoch": 1.425578231292517,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.00010627830130950925,
+      "loss": 1.024,
+      "step": 3930
+    },
+    {
+      "epoch": 1.4263038548752833,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.00010622934769306083,
+      "loss": 0.9421,
+      "step": 3932
+    },
+    {
+      "epoch": 1.42702947845805,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.00010618039407661242,
+      "loss": 1.0215,
+      "step": 3934
+    },
+    {
+      "epoch": 1.4277551020408163,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.00010613144046016399,
+      "loss": 0.9815,
+      "step": 3936
+    },
+    {
+      "epoch": 1.4284807256235827,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.00010608248684371557,
+      "loss": 0.9606,
+      "step": 3938
+    },
+    {
+      "epoch": 1.4292063492063491,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.00010603353322726717,
+      "loss": 0.9918,
+      "step": 3940
+    },
+    {
+      "epoch": 1.4299319727891158,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.00010598457961081875,
+      "loss": 0.9695,
+      "step": 3942
+    },
+    {
+      "epoch": 1.4306575963718822,
+      "grad_norm": 0.177734375,
+      "learning_rate": 0.00010593562599437035,
+      "loss": 0.9732,
+      "step": 3944
+    },
+    {
+      "epoch": 1.4313832199546486,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.00010588667237792192,
+      "loss": 1.0474,
+      "step": 3946
+    },
+    {
+      "epoch": 1.432108843537415,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.0001058377187614735,
+      "loss": 1.0052,
+      "step": 3948
+    },
+    {
+      "epoch": 1.4328344671201814,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.0001057887651450251,
+      "loss": 0.9909,
+      "step": 3950
+    },
+    {
+      "epoch": 1.4335600907029478,
+      "grad_norm": 0.2109375,
+      "learning_rate": 0.00010573981152857668,
+      "loss": 0.9774,
+      "step": 3952
+    },
+    {
+      "epoch": 1.4342857142857142,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.00010569085791212827,
+      "loss": 0.9736,
+      "step": 3954
+    },
+    {
+      "epoch": 1.4350113378684808,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.00010564190429567984,
+      "loss": 1.0008,
+      "step": 3956
+    },
+    {
+      "epoch": 1.4357369614512472,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.00010559295067923142,
+      "loss": 0.9877,
+      "step": 3958
+    },
+    {
+      "epoch": 1.4364625850340136,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.00010554399706278302,
+      "loss": 0.9768,
+      "step": 3960
+    },
+    {
+      "epoch": 1.43718820861678,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.0001054950434463346,
+      "loss": 0.9943,
+      "step": 3962
+    },
+    {
+      "epoch": 1.4379138321995466,
+      "grad_norm": 0.1865234375,
+      "learning_rate": 0.0001054460898298862,
+      "loss": 1.0235,
+      "step": 3964
+    },
+    {
+      "epoch": 1.438639455782313,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.00010539713621343777,
+      "loss": 1.0468,
+      "step": 3966
+    },
+    {
+      "epoch": 1.4393650793650794,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.00010534818259698935,
+      "loss": 0.9501,
+      "step": 3968
+    },
+    {
+      "epoch": 1.4400907029478458,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.00010529922898054095,
+      "loss": 1.0001,
+      "step": 3970
+    },
+    {
+      "epoch": 1.4408163265306122,
+      "grad_norm": 0.1865234375,
+      "learning_rate": 0.00010525027536409253,
+      "loss": 0.9847,
+      "step": 3972
+    },
+    {
+      "epoch": 1.4415419501133786,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.00010520132174764412,
+      "loss": 0.9209,
+      "step": 3974
+    },
+    {
+      "epoch": 1.442267573696145,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.00010515236813119569,
+      "loss": 0.969,
+      "step": 3976
+    },
+    {
+      "epoch": 1.4429931972789116,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.00010510341451474727,
+      "loss": 0.8542,
+      "step": 3978
+    },
+    {
+      "epoch": 1.443718820861678,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.00010505446089829887,
+      "loss": 0.9088,
+      "step": 3980
+    },
+    {
+      "epoch": 1.4444444444444444,
+      "grad_norm": 0.1875,
+      "learning_rate": 0.00010500550728185045,
+      "loss": 0.9864,
+      "step": 3982
+    },
+    {
+      "epoch": 1.4451700680272108,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.00010495655366540205,
+      "loss": 1.0233,
+      "step": 3984
+    },
+    {
+      "epoch": 1.4458956916099774,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.00010490760004895362,
+      "loss": 0.9392,
+      "step": 3986
+    },
+    {
+      "epoch": 1.4466213151927438,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.0001048586464325052,
+      "loss": 0.9399,
+      "step": 3988
+    },
+    {
+      "epoch": 1.4473469387755102,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.0001048096928160568,
+      "loss": 1.0332,
+      "step": 3990
+    },
+    {
+      "epoch": 1.4480725623582766,
+      "grad_norm": 0.2060546875,
+      "learning_rate": 0.00010476073919960838,
+      "loss": 1.0093,
+      "step": 3992
+    },
+    {
+      "epoch": 1.448798185941043,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.00010471178558315997,
+      "loss": 0.9698,
+      "step": 3994
+    },
+    {
+      "epoch": 1.4495238095238094,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.00010466283196671154,
+      "loss": 0.9719,
+      "step": 3996
+    },
+    {
+      "epoch": 1.4502494331065758,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.00010461387835026312,
+      "loss": 1.0843,
+      "step": 3998
+    },
+    {
+      "epoch": 1.4509750566893425,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.00010456492473381472,
+      "loss": 0.9405,
+      "step": 4000
+    },
+    {
+      "epoch": 1.4509750566893425,
+      "eval_loss": 1.0138306617736816,
+      "eval_runtime": 90.2991,
+      "eval_samples_per_second": 19.934,
+      "eval_steps_per_second": 0.631,
+      "step": 4000
+    },
+    {
+      "epoch": 1.4517006802721089,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.0001045159711173663,
+      "loss": 0.9971,
+      "step": 4002
+    },
+    {
+      "epoch": 1.4524263038548753,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.0001044670175009179,
+      "loss": 0.9311,
+      "step": 4004
+    },
+    {
+      "epoch": 1.4531519274376417,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.00010441806388446947,
+      "loss": 0.9054,
+      "step": 4006
+    },
+    {
+      "epoch": 1.4538775510204083,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.00010436911026802105,
+      "loss": 1.0426,
+      "step": 4008
+    },
+    {
+      "epoch": 1.4546031746031747,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.00010432015665157264,
+      "loss": 0.9125,
+      "step": 4010
+    },
+    {
+      "epoch": 1.455328798185941,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.00010427120303512423,
+      "loss": 0.9421,
+      "step": 4012
+    },
+    {
+      "epoch": 1.4560544217687075,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.0001042222494186758,
+      "loss": 0.9698,
+      "step": 4014
+    },
+    {
+      "epoch": 1.4567800453514739,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.00010417329580222739,
+      "loss": 1.0262,
+      "step": 4016
+    },
+    {
+      "epoch": 1.4575056689342403,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.00010412434218577897,
+      "loss": 0.9472,
+      "step": 4018
+    },
+    {
+      "epoch": 1.4582312925170067,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.00010407538856933057,
+      "loss": 0.981,
+      "step": 4020
+    },
+    {
+      "epoch": 1.4589569160997733,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.00010402643495288215,
+      "loss": 0.9012,
+      "step": 4022
+    },
+    {
+      "epoch": 1.4596825396825397,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.00010397748133643372,
+      "loss": 0.9633,
+      "step": 4024
+    },
+    {
+      "epoch": 1.460408163265306,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.00010392852771998532,
+      "loss": 0.9586,
+      "step": 4026
+    },
+    {
+      "epoch": 1.4611337868480725,
+      "grad_norm": 0.193359375,
+      "learning_rate": 0.0001038795741035369,
+      "loss": 1.0336,
+      "step": 4028
+    },
+    {
+      "epoch": 1.4618594104308391,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.0001038306204870885,
+      "loss": 0.9896,
+      "step": 4030
+    },
+    {
+      "epoch": 1.4625850340136055,
+      "grad_norm": 0.185546875,
+      "learning_rate": 0.00010378166687064008,
+      "loss": 1.0079,
+      "step": 4032
+    },
+    {
+      "epoch": 1.463310657596372,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.00010373271325419165,
+      "loss": 0.9626,
+      "step": 4034
+    },
+    {
+      "epoch": 1.4640362811791383,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.00010368375963774324,
+      "loss": 0.9859,
+      "step": 4036
+    },
+    {
+      "epoch": 1.4647619047619047,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.00010363480602129482,
+      "loss": 0.9595,
+      "step": 4038
+    },
+    {
+      "epoch": 1.465487528344671,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.00010358585240484642,
+      "loss": 0.9691,
+      "step": 4040
+    },
+    {
+      "epoch": 1.4662131519274375,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.000103536898788398,
+      "loss": 0.983,
+      "step": 4042
+    },
+    {
+      "epoch": 1.4669387755102041,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.00010348794517194957,
+      "loss": 1.0172,
+      "step": 4044
+    },
+    {
+      "epoch": 1.4676643990929705,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.00010343899155550117,
+      "loss": 0.9945,
+      "step": 4046
+    },
+    {
+      "epoch": 1.468390022675737,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.00010339003793905275,
+      "loss": 0.9569,
+      "step": 4048
+    },
+    {
+      "epoch": 1.4691156462585033,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.00010334108432260434,
+      "loss": 0.9445,
+      "step": 4050
+    },
+    {
+      "epoch": 1.46984126984127,
+      "grad_norm": 0.18359375,
+      "learning_rate": 0.00010329213070615593,
+      "loss": 1.0193,
+      "step": 4052
+    },
+    {
+      "epoch": 1.4705668934240363,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.0001032431770897075,
+      "loss": 1.0227,
+      "step": 4054
+    },
+    {
+      "epoch": 1.4712925170068027,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.00010319422347325909,
+      "loss": 0.9067,
+      "step": 4056
+    },
+    {
+      "epoch": 1.4720181405895691,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.00010314526985681067,
+      "loss": 0.9907,
+      "step": 4058
+    },
+    {
+      "epoch": 1.4727437641723355,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.00010309631624036227,
+      "loss": 0.963,
+      "step": 4060
+    },
+    {
+      "epoch": 1.473469387755102,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.00010304736262391385,
+      "loss": 0.9769,
+      "step": 4062
+    },
+    {
+      "epoch": 1.4741950113378683,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.00010299840900746542,
+      "loss": 0.9689,
+      "step": 4064
+    },
+    {
+      "epoch": 1.474920634920635,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.00010294945539101702,
+      "loss": 0.9922,
+      "step": 4066
+    },
+    {
+      "epoch": 1.4756462585034014,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.0001029005017745686,
+      "loss": 0.9429,
+      "step": 4068
+    },
+    {
+      "epoch": 1.4763718820861678,
+      "grad_norm": 0.1884765625,
+      "learning_rate": 0.0001028515481581202,
+      "loss": 1.0386,
+      "step": 4070
+    },
+    {
+      "epoch": 1.4770975056689342,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.00010280259454167178,
+      "loss": 0.9666,
+      "step": 4072
+    },
+    {
+      "epoch": 1.4778231292517008,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.00010275364092522335,
+      "loss": 1.0141,
+      "step": 4074
+    },
+    {
+      "epoch": 1.4785487528344672,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.00010270468730877494,
+      "loss": 0.9448,
+      "step": 4076
+    },
+    {
+      "epoch": 1.4792743764172336,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.00010265573369232652,
+      "loss": 1.0185,
+      "step": 4078
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.00010260678007587812,
+      "loss": 0.9499,
+      "step": 4080
+    },
+    {
+      "epoch": 1.4807256235827664,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.00010255782645942969,
+      "loss": 0.9646,
+      "step": 4082
+    },
+    {
+      "epoch": 1.4814512471655328,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.00010250887284298127,
+      "loss": 1.0299,
+      "step": 4084
+    },
+    {
+      "epoch": 1.4821768707482994,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.00010245991922653287,
+      "loss": 0.9877,
+      "step": 4086
+    },
+    {
+      "epoch": 1.4829024943310658,
+      "grad_norm": 0.173828125,
+      "learning_rate": 0.00010241096561008445,
+      "loss": 0.9352,
+      "step": 4088
+    },
+    {
+      "epoch": 1.4836281179138322,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.00010236201199363604,
+      "loss": 0.9672,
+      "step": 4090
+    },
+    {
+      "epoch": 1.4843537414965986,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.00010231305837718761,
+      "loss": 0.9205,
+      "step": 4092
+    },
+    {
+      "epoch": 1.485079365079365,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.0001022641047607392,
+      "loss": 1.0447,
+      "step": 4094
+    },
+    {
+      "epoch": 1.4858049886621316,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.00010221515114429079,
+      "loss": 0.9848,
+      "step": 4096
+    },
+    {
+      "epoch": 1.486530612244898,
+      "grad_norm": 0.1875,
+      "learning_rate": 0.00010216619752784237,
+      "loss": 0.9774,
+      "step": 4098
+    },
+    {
+      "epoch": 1.4872562358276644,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.00010211724391139397,
+      "loss": 0.9378,
+      "step": 4100
+    },
+    {
+      "epoch": 1.4872562358276644,
+      "eval_loss": 1.0134267807006836,
+      "eval_runtime": 90.0252,
+      "eval_samples_per_second": 19.994,
+      "eval_steps_per_second": 0.633,
+      "step": 4100
+    },
+    {
+      "epoch": 1.4879818594104308,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.00010206829029494554,
+      "loss": 0.9782,
+      "step": 4102
+    },
+    {
+      "epoch": 1.4887074829931972,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.00010201933667849712,
+      "loss": 1.0532,
+      "step": 4104
+    },
+    {
+      "epoch": 1.4894331065759636,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.00010197038306204872,
+      "loss": 0.9129,
+      "step": 4106
+    },
+    {
+      "epoch": 1.4901587301587302,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.0001019214294456003,
+      "loss": 0.9277,
+      "step": 4108
+    },
+    {
+      "epoch": 1.4908843537414966,
+      "grad_norm": 0.17578125,
+      "learning_rate": 0.0001018724758291519,
+      "loss": 0.9731,
+      "step": 4110
+    },
+    {
+      "epoch": 1.491609977324263,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.00010182352221270346,
+      "loss": 0.9863,
+      "step": 4112
+    },
+    {
+      "epoch": 1.4923356009070294,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.00010177456859625505,
+      "loss": 1.1322,
+      "step": 4114
+    },
+    {
+      "epoch": 1.4930612244897958,
+      "grad_norm": 0.181640625,
+      "learning_rate": 0.00010172561497980664,
+      "loss": 0.9763,
+      "step": 4116
+    },
+    {
+      "epoch": 1.4937868480725625,
+      "grad_norm": 0.173828125,
+      "learning_rate": 0.00010167666136335822,
+      "loss": 0.978,
+      "step": 4118
+    },
+    {
+      "epoch": 1.4945124716553289,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.00010162770774690982,
+      "loss": 1.0198,
+      "step": 4120
+    },
+    {
+      "epoch": 1.4952380952380953,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.00010157875413046139,
+      "loss": 0.8959,
+      "step": 4122
+    },
+    {
+      "epoch": 1.4959637188208617,
+      "grad_norm": 0.173828125,
+      "learning_rate": 0.00010152980051401297,
+      "loss": 0.9886,
+      "step": 4124
+    },
+    {
+      "epoch": 1.496689342403628,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.00010148084689756457,
+      "loss": 0.9965,
+      "step": 4126
+    },
+    {
+      "epoch": 1.4974149659863945,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.00010143189328111615,
+      "loss": 0.961,
+      "step": 4128
+    },
+    {
+      "epoch": 1.498140589569161,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.00010138293966466774,
+      "loss": 0.9631,
+      "step": 4130
+    },
+    {
+      "epoch": 1.4988662131519275,
+      "grad_norm": 0.208984375,
+      "learning_rate": 0.00010133398604821931,
+      "loss": 1.0086,
+      "step": 4132
+    },
+    {
+      "epoch": 1.4995918367346939,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.0001012850324317709,
+      "loss": 0.9727,
+      "step": 4134
+    },
+    {
+      "epoch": 1.5003174603174603,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.00010123607881532249,
+      "loss": 0.9614,
+      "step": 4136
+    },
+    {
+      "epoch": 1.501043083900227,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.00010118712519887407,
+      "loss": 0.9524,
+      "step": 4138
+    },
+    {
+      "epoch": 1.5017687074829933,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.00010113817158242567,
+      "loss": 0.9623,
+      "step": 4140
+    },
+    {
+      "epoch": 1.5024943310657597,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.00010108921796597724,
+      "loss": 0.9599,
+      "step": 4142
+    },
+    {
+      "epoch": 1.503219954648526,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.00010104026434952882,
+      "loss": 0.9384,
+      "step": 4144
+    },
+    {
+      "epoch": 1.5039455782312925,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.00010099131073308042,
+      "loss": 0.9673,
+      "step": 4146
+    },
+    {
+      "epoch": 1.504671201814059,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.000100942357116632,
+      "loss": 0.9375,
+      "step": 4148
+    },
+    {
+      "epoch": 1.5053968253968253,
+      "grad_norm": 0.181640625,
+      "learning_rate": 0.0001008934035001836,
+      "loss": 0.884,
+      "step": 4150
+    },
+    {
+      "epoch": 1.5061224489795917,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.00010084444988373516,
+      "loss": 0.9254,
+      "step": 4152
+    },
+    {
+      "epoch": 1.5068480725623583,
+      "grad_norm": 0.203125,
+      "learning_rate": 0.00010079549626728675,
+      "loss": 1.0317,
+      "step": 4154
+    },
+    {
+      "epoch": 1.5075736961451247,
+      "grad_norm": 0.17578125,
+      "learning_rate": 0.00010074654265083834,
+      "loss": 0.9853,
+      "step": 4156
+    },
+    {
+      "epoch": 1.508299319727891,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.00010069758903438992,
+      "loss": 0.9551,
+      "step": 4158
+    },
+    {
+      "epoch": 1.5090249433106577,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.00010064863541794149,
+      "loss": 1.0439,
+      "step": 4160
+    },
+    {
+      "epoch": 1.5097505668934241,
+      "grad_norm": 0.1875,
+      "learning_rate": 0.00010059968180149309,
+      "loss": 0.9087,
+      "step": 4162
+    },
+    {
+      "epoch": 1.5104761904761905,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.00010055072818504467,
+      "loss": 0.9663,
+      "step": 4164
+    },
+    {
+      "epoch": 1.511201814058957,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.00010050177456859627,
+      "loss": 0.9691,
+      "step": 4166
+    },
+    {
+      "epoch": 1.5119274376417233,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.00010045282095214785,
+      "loss": 1.0104,
+      "step": 4168
+    },
+    {
+      "epoch": 1.5126530612244897,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.00010040386733569942,
+      "loss": 1.022,
+      "step": 4170
+    },
+    {
+      "epoch": 1.5133786848072561,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.00010035491371925101,
+      "loss": 1.0086,
+      "step": 4172
+    },
+    {
+      "epoch": 1.5141043083900225,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.0001003059601028026,
+      "loss": 0.9908,
+      "step": 4174
+    },
+    {
+      "epoch": 1.5148299319727891,
+      "grad_norm": 0.22265625,
+      "learning_rate": 0.00010025700648635419,
+      "loss": 0.9026,
+      "step": 4176
+    },
+    {
+      "epoch": 1.5155555555555555,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.00010020805286990577,
+      "loss": 1.0208,
+      "step": 4178
+    },
+    {
+      "epoch": 1.516281179138322,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.00010015909925345734,
+      "loss": 0.9172,
+      "step": 4180
+    },
+    {
+      "epoch": 1.5170068027210886,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.00010011014563700894,
+      "loss": 0.9348,
+      "step": 4182
+    },
+    {
+      "epoch": 1.517732426303855,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.00010006119202056052,
+      "loss": 0.9899,
+      "step": 4184
+    },
+    {
+      "epoch": 1.5184580498866214,
+      "grad_norm": 0.1953125,
+      "learning_rate": 0.00010001223840411212,
+      "loss": 1.0367,
+      "step": 4186
+    },
+    {
+      "epoch": 1.5191836734693878,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 9.996328478766368e-05,
+      "loss": 0.9138,
+      "step": 4188
+    },
+    {
+      "epoch": 1.5199092970521542,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 9.991433117121528e-05,
+      "loss": 1.0288,
+      "step": 4190
+    },
+    {
+      "epoch": 1.5206349206349206,
+      "grad_norm": 0.158203125,
+      "learning_rate": 9.986537755476686e-05,
+      "loss": 0.9677,
+      "step": 4192
+    },
+    {
+      "epoch": 1.521360544217687,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 9.981642393831845e-05,
+      "loss": 1.0642,
+      "step": 4194
+    },
+    {
+      "epoch": 1.5220861678004534,
+      "grad_norm": 0.16796875,
+      "learning_rate": 9.976747032187004e-05,
+      "loss": 1.0481,
+      "step": 4196
+    },
+    {
+      "epoch": 1.52281179138322,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 9.971851670542161e-05,
+      "loss": 1.0118,
+      "step": 4198
+    },
+    {
+      "epoch": 1.5235374149659864,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 9.96695630889732e-05,
+      "loss": 0.9331,
+      "step": 4200
+    },
+    {
+      "epoch": 1.5235374149659864,
+      "eval_loss": 1.013152003288269,
+      "eval_runtime": 89.7624,
+      "eval_samples_per_second": 20.053,
+      "eval_steps_per_second": 0.635,
+      "step": 4200
+    },
+    {
+      "epoch": 1.5242630385487528,
+      "grad_norm": 0.16796875,
+      "learning_rate": 9.962060947252479e-05,
+      "loss": 1.0175,
+      "step": 4202
+    },
+    {
+      "epoch": 1.5249886621315194,
+      "grad_norm": 0.16796875,
+      "learning_rate": 9.957165585607637e-05,
+      "loss": 1.0108,
+      "step": 4204
+    },
+    {
+      "epoch": 1.5257142857142858,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 9.952270223962795e-05,
+      "loss": 1.0153,
+      "step": 4206
+    },
+    {
+      "epoch": 1.5264399092970522,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 9.947374862317953e-05,
+      "loss": 1.0924,
+      "step": 4208
+    },
+    {
+      "epoch": 1.5271655328798186,
+      "grad_norm": 0.162109375,
+      "learning_rate": 9.942479500673113e-05,
+      "loss": 0.9743,
+      "step": 4210
+    },
+    {
+      "epoch": 1.527891156462585,
+      "grad_norm": 0.177734375,
+      "learning_rate": 9.937584139028271e-05,
+      "loss": 0.9311,
+      "step": 4212
+    },
+    {
+      "epoch": 1.5286167800453514,
+      "grad_norm": 0.158203125,
+      "learning_rate": 9.93268877738343e-05,
+      "loss": 0.9674,
+      "step": 4214
+    },
+    {
+      "epoch": 1.5293424036281178,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 9.927793415738588e-05,
+      "loss": 0.9447,
+      "step": 4216
+    },
+    {
+      "epoch": 1.5300680272108842,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 9.922898054093746e-05,
+      "loss": 1.0299,
+      "step": 4218
+    },
+    {
+      "epoch": 1.5307936507936508,
+      "grad_norm": 0.166015625,
+      "learning_rate": 9.918002692448906e-05,
+      "loss": 0.954,
+      "step": 4220
+    },
+    {
+      "epoch": 1.5315192743764172,
+      "grad_norm": 0.171875,
+      "learning_rate": 9.913107330804064e-05,
+      "loss": 1.012,
+      "step": 4222
+    },
+    {
+      "epoch": 1.5322448979591838,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 9.908211969159222e-05,
+      "loss": 1.0334,
+      "step": 4224
+    },
+    {
+      "epoch": 1.5329705215419502,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 9.90331660751438e-05,
+      "loss": 0.9732,
+      "step": 4226
+    },
+    {
+      "epoch": 1.5336961451247166,
+      "grad_norm": 0.15234375,
+      "learning_rate": 9.898421245869538e-05,
+      "loss": 0.9949,
+      "step": 4228
+    },
+    {
+      "epoch": 1.534421768707483,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 9.893525884224698e-05,
+      "loss": 1.0049,
+      "step": 4230
+    },
+    {
+      "epoch": 1.5351473922902494,
+      "grad_norm": 0.158203125,
+      "learning_rate": 9.888630522579856e-05,
+      "loss": 0.9727,
+      "step": 4232
+    },
+    {
+      "epoch": 1.5358730158730158,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 9.883735160935014e-05,
+      "loss": 0.9889,
+      "step": 4234
+    },
+    {
+      "epoch": 1.5365986394557822,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 9.878839799290173e-05,
+      "loss": 0.9699,
+      "step": 4236
+    },
+    {
+      "epoch": 1.5373242630385486,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 9.873944437645331e-05,
+      "loss": 0.9659,
+      "step": 4238
+    },
+    {
+      "epoch": 1.538049886621315,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 9.86904907600049e-05,
+      "loss": 0.9404,
+      "step": 4240
+    },
+    {
+      "epoch": 1.5387755102040817,
+      "grad_norm": 0.17578125,
+      "learning_rate": 9.864153714355649e-05,
+      "loss": 0.9822,
+      "step": 4242
+    },
+    {
+      "epoch": 1.539501133786848,
+      "grad_norm": 0.154296875,
+      "learning_rate": 9.859258352710807e-05,
+      "loss": 0.9127,
+      "step": 4244
+    },
+    {
+      "epoch": 1.5402267573696147,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 9.854362991065965e-05,
+      "loss": 1.0608,
+      "step": 4246
+    },
+    {
+      "epoch": 1.540952380952381,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 9.849467629421123e-05,
+      "loss": 1.0184,
+      "step": 4248
+    },
+    {
+      "epoch": 1.5416780045351475,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 9.844572267776282e-05,
+      "loss": 1.0275,
+      "step": 4250
+    },
+    {
+      "epoch": 1.5424036281179139,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 9.839676906131441e-05,
+      "loss": 0.9631,
+      "step": 4252
+    },
+    {
+      "epoch": 1.5431292517006803,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 9.8347815444866e-05,
+      "loss": 0.9986,
+      "step": 4254
+    },
+    {
+      "epoch": 1.5438548752834467,
+      "grad_norm": 0.16015625,
+      "learning_rate": 9.829886182841758e-05,
+      "loss": 0.9216,
+      "step": 4256
+    },
+    {
+      "epoch": 1.544580498866213,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 9.824990821196916e-05,
+      "loss": 0.939,
+      "step": 4258
+    },
+    {
+      "epoch": 1.5453061224489795,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 9.820095459552074e-05,
+      "loss": 1.049,
+      "step": 4260
+    },
+    {
+      "epoch": 1.5460317460317459,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 9.815200097907234e-05,
+      "loss": 0.9569,
+      "step": 4262
+    },
+    {
+      "epoch": 1.5467573696145125,
+      "grad_norm": 0.17578125,
+      "learning_rate": 9.810304736262392e-05,
+      "loss": 1.0119,
+      "step": 4264
+    },
+    {
+      "epoch": 1.5474829931972789,
+      "grad_norm": 0.1875,
+      "learning_rate": 9.80540937461755e-05,
+      "loss": 1.1073,
+      "step": 4266
+    },
+    {
+      "epoch": 1.5482086167800455,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 9.800514012972708e-05,
+      "loss": 0.9696,
+      "step": 4268
+    },
+    {
+      "epoch": 1.548934240362812,
+      "grad_norm": 0.154296875,
+      "learning_rate": 9.795618651327867e-05,
+      "loss": 0.8951,
+      "step": 4270
+    },
+    {
+      "epoch": 1.5496598639455783,
+      "grad_norm": 0.193359375,
+      "learning_rate": 9.790723289683026e-05,
+      "loss": 0.94,
+      "step": 4272
+    },
+    {
+      "epoch": 1.5503854875283447,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 9.785827928038184e-05,
+      "loss": 1.0004,
+      "step": 4274
+    },
+    {
+      "epoch": 1.551111111111111,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 9.780932566393343e-05,
+      "loss": 0.9606,
+      "step": 4276
+    },
+    {
+      "epoch": 1.5518367346938775,
+      "grad_norm": 0.162109375,
+      "learning_rate": 9.776037204748501e-05,
+      "loss": 1.0272,
+      "step": 4278
+    },
+    {
+      "epoch": 1.552562358276644,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 9.771141843103659e-05,
+      "loss": 0.9806,
+      "step": 4280
+    },
+    {
+      "epoch": 1.5532879818594103,
+      "grad_norm": 0.1484375,
+      "learning_rate": 9.766246481458819e-05,
+      "loss": 0.8835,
+      "step": 4282
+    },
+    {
+      "epoch": 1.5540136054421767,
+      "grad_norm": 0.16015625,
+      "learning_rate": 9.761351119813976e-05,
+      "loss": 0.9617,
+      "step": 4284
+    },
+    {
+      "epoch": 1.5547392290249433,
+      "grad_norm": 0.154296875,
+      "learning_rate": 9.756455758169135e-05,
+      "loss": 1.047,
+      "step": 4286
+    },
+    {
+      "epoch": 1.5554648526077097,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 9.751560396524293e-05,
+      "loss": 0.9978,
+      "step": 4288
+    },
+    {
+      "epoch": 1.5561904761904763,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 9.746665034879452e-05,
+      "loss": 0.9766,
+      "step": 4290
+    },
+    {
+      "epoch": 1.5569160997732427,
+      "grad_norm": 0.173828125,
+      "learning_rate": 9.741769673234611e-05,
+      "loss": 1.0446,
+      "step": 4292
+    },
+    {
+      "epoch": 1.5576417233560091,
+      "grad_norm": 0.177734375,
+      "learning_rate": 9.736874311589768e-05,
+      "loss": 1.0301,
+      "step": 4294
+    },
+    {
+      "epoch": 1.5583673469387755,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 9.731978949944928e-05,
+      "loss": 1.0035,
+      "step": 4296
+    },
+    {
+      "epoch": 1.559092970521542,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 9.727083588300086e-05,
+      "loss": 0.9809,
+      "step": 4298
+    },
+    {
+      "epoch": 1.5598185941043083,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 9.722188226655244e-05,
+      "loss": 0.9959,
+      "step": 4300
+    },
+    {
+      "epoch": 1.5598185941043083,
+      "eval_loss": 1.0130585432052612,
+      "eval_runtime": 87.5971,
+      "eval_samples_per_second": 20.549,
+      "eval_steps_per_second": 0.651,
+      "step": 4300
+    },
+    {
+      "epoch": 1.5605442176870747,
+      "grad_norm": 0.15234375,
+      "learning_rate": 9.717292865010404e-05,
+      "loss": 1.0011,
+      "step": 4302
+    },
+    {
+      "epoch": 1.5612698412698411,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 9.71239750336556e-05,
+      "loss": 0.934,
+      "step": 4304
+    },
+    {
+      "epoch": 1.5619954648526078,
+      "grad_norm": 0.14453125,
+      "learning_rate": 9.70750214172072e-05,
+      "loss": 0.9783,
+      "step": 4306
+    },
+    {
+      "epoch": 1.5627210884353742,
+      "grad_norm": 0.1796875,
+      "learning_rate": 9.702606780075878e-05,
+      "loss": 0.995,
+      "step": 4308
+    },
+    {
+      "epoch": 1.5634467120181406,
+      "grad_norm": 0.162109375,
+      "learning_rate": 9.697711418431037e-05,
+      "loss": 0.932,
+      "step": 4310
+    },
+    {
+      "epoch": 1.5641723356009072,
+      "grad_norm": 0.154296875,
+      "learning_rate": 9.692816056786196e-05,
+      "loss": 1.0029,
+      "step": 4312
+    },
+    {
+      "epoch": 1.5648979591836736,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 9.687920695141353e-05,
+      "loss": 0.9971,
+      "step": 4314
+    },
+    {
+      "epoch": 1.56562358276644,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 9.683025333496513e-05,
+      "loss": 0.9928,
+      "step": 4316
+    },
+    {
+      "epoch": 1.5663492063492064,
+      "grad_norm": 0.16015625,
+      "learning_rate": 9.678129971851671e-05,
+      "loss": 0.9279,
+      "step": 4318
+    },
+    {
+      "epoch": 1.5670748299319728,
+      "grad_norm": 0.17578125,
+      "learning_rate": 9.673234610206829e-05,
+      "loss": 0.937,
+      "step": 4320
+    },
+    {
+      "epoch": 1.5678004535147392,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 9.668339248561989e-05,
+      "loss": 1.1021,
+      "step": 4322
+    },
+    {
+      "epoch": 1.5685260770975056,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 9.663443886917146e-05,
+      "loss": 0.938,
+      "step": 4324
+    },
+    {
+      "epoch": 1.569251700680272,
+      "grad_norm": 0.177734375,
+      "learning_rate": 9.658548525272305e-05,
+      "loss": 0.992,
+      "step": 4326
+    },
+    {
+      "epoch": 1.5699773242630386,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 9.653653163627463e-05,
+      "loss": 1.0522,
+      "step": 4328
+    },
+    {
+      "epoch": 1.570702947845805,
+      "grad_norm": 0.16015625,
+      "learning_rate": 9.648757801982622e-05,
+      "loss": 1.0242,
+      "step": 4330
+    },
+    {
+      "epoch": 1.5714285714285714,
+      "grad_norm": 0.154296875,
+      "learning_rate": 9.643862440337781e-05,
+      "loss": 0.9797,
+      "step": 4332
+    },
+    {
+      "epoch": 1.572154195011338,
+      "grad_norm": 0.185546875,
+      "learning_rate": 9.638967078692938e-05,
+      "loss": 1.0288,
+      "step": 4334
+    },
+    {
+      "epoch": 1.5728798185941044,
+      "grad_norm": 0.162109375,
+      "learning_rate": 9.634071717048098e-05,
+      "loss": 0.9567,
+      "step": 4336
+    },
+    {
+      "epoch": 1.5736054421768708,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 9.629176355403256e-05,
+      "loss": 1.0486,
+      "step": 4338
+    },
+    {
+      "epoch": 1.5743310657596372,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 9.624280993758414e-05,
+      "loss": 0.9999,
+      "step": 4340
+    },
+    {
+      "epoch": 1.5750566893424036,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 9.619385632113574e-05,
+      "loss": 0.9742,
+      "step": 4342
+    },
+    {
+      "epoch": 1.57578231292517,
+      "grad_norm": 0.154296875,
+      "learning_rate": 9.61449027046873e-05,
+      "loss": 0.9984,
+      "step": 4344
+    },
+    {
+      "epoch": 1.5765079365079364,
+      "grad_norm": 0.15625,
+      "learning_rate": 9.60959490882389e-05,
+      "loss": 0.987,
+      "step": 4346
+    },
+    {
+      "epoch": 1.5772335600907028,
+      "grad_norm": 0.158203125,
+      "learning_rate": 9.604699547179048e-05,
+      "loss": 1.0147,
+      "step": 4348
+    },
+    {
+      "epoch": 1.5779591836734694,
+      "grad_norm": 0.169921875,
+      "learning_rate": 9.599804185534207e-05,
+      "loss": 0.9514,
+      "step": 4350
+    },
+    {
+      "epoch": 1.5786848072562358,
+      "grad_norm": 0.150390625,
+      "learning_rate": 9.594908823889365e-05,
+      "loss": 0.8686,
+      "step": 4352
+    },
+    {
+      "epoch": 1.5794104308390022,
+      "grad_norm": 0.173828125,
+      "learning_rate": 9.590013462244523e-05,
+      "loss": 0.9818,
+      "step": 4354
+    },
+    {
+      "epoch": 1.5801360544217689,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 9.585118100599683e-05,
+      "loss": 0.9206,
+      "step": 4356
+    },
+    {
+      "epoch": 1.5808616780045353,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 9.580222738954841e-05,
+      "loss": 1.0003,
+      "step": 4358
+    },
+    {
+      "epoch": 1.5815873015873017,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 9.575327377309999e-05,
+      "loss": 0.9899,
+      "step": 4360
+    },
+    {
+      "epoch": 1.582312925170068,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 9.570432015665157e-05,
+      "loss": 0.9102,
+      "step": 4362
+    },
+    {
+      "epoch": 1.5830385487528345,
+      "grad_norm": 0.1484375,
+      "learning_rate": 9.565536654020316e-05,
+      "loss": 0.9594,
+      "step": 4364
+    },
+    {
+      "epoch": 1.5837641723356009,
+      "grad_norm": 0.21875,
+      "learning_rate": 9.560641292375475e-05,
+      "loss": 0.9382,
+      "step": 4366
+    },
+    {
+      "epoch": 1.5844897959183672,
+      "grad_norm": 0.158203125,
+      "learning_rate": 9.555745930730633e-05,
+      "loss": 1.0231,
+      "step": 4368
+    },
+    {
+      "epoch": 1.5852154195011336,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 9.550850569085792e-05,
+      "loss": 0.9838,
+      "step": 4370
+    },
+    {
+      "epoch": 1.5859410430839003,
+      "grad_norm": 0.17578125,
+      "learning_rate": 9.54595520744095e-05,
+      "loss": 1.0085,
+      "step": 4372
+    },
+    {
+      "epoch": 1.5866666666666667,
+      "grad_norm": 0.1640625,
+      "learning_rate": 9.541059845796108e-05,
+      "loss": 0.9167,
+      "step": 4374
+    },
+    {
+      "epoch": 1.587392290249433,
+      "grad_norm": 0.181640625,
+      "learning_rate": 9.536164484151268e-05,
+      "loss": 0.9559,
+      "step": 4376
+    },
+    {
+      "epoch": 1.5881179138321997,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 9.531269122506426e-05,
+      "loss": 0.9368,
+      "step": 4378
+    },
+    {
+      "epoch": 1.588843537414966,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 9.526373760861584e-05,
+      "loss": 0.9922,
+      "step": 4380
+    },
+    {
+      "epoch": 1.5895691609977325,
+      "grad_norm": 0.1640625,
+      "learning_rate": 9.521478399216742e-05,
+      "loss": 0.9523,
+      "step": 4382
+    },
+    {
+      "epoch": 1.5902947845804989,
+      "grad_norm": 0.17578125,
+      "learning_rate": 9.5165830375719e-05,
+      "loss": 0.9917,
+      "step": 4384
+    },
+    {
+      "epoch": 1.5910204081632653,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 9.51168767592706e-05,
+      "loss": 0.9945,
+      "step": 4386
+    },
+    {
+      "epoch": 1.5917460317460317,
+      "grad_norm": 0.205078125,
+      "learning_rate": 9.506792314282218e-05,
+      "loss": 1.0831,
+      "step": 4388
+    },
+    {
+      "epoch": 1.592471655328798,
+      "grad_norm": 0.1982421875,
+      "learning_rate": 9.501896952637377e-05,
+      "loss": 1.0329,
+      "step": 4390
+    },
+    {
+      "epoch": 1.5931972789115645,
+      "grad_norm": 0.171875,
+      "learning_rate": 9.497001590992535e-05,
+      "loss": 0.8841,
+      "step": 4392
+    },
+    {
+      "epoch": 1.593922902494331,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 9.492106229347693e-05,
+      "loss": 0.8609,
+      "step": 4394
+    },
+    {
+      "epoch": 1.5946485260770975,
+      "grad_norm": 0.16796875,
+      "learning_rate": 9.487210867702851e-05,
+      "loss": 0.9959,
+      "step": 4396
+    },
+    {
+      "epoch": 1.595374149659864,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 9.482315506058011e-05,
+      "loss": 0.9974,
+      "step": 4398
+    },
+    {
+      "epoch": 1.5960997732426305,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 9.477420144413169e-05,
+      "loss": 0.9548,
+      "step": 4400
+    },
+    {
+      "epoch": 1.5960997732426305,
+      "eval_loss": 1.0129448175430298,
+      "eval_runtime": 87.6312,
+      "eval_samples_per_second": 20.541,
+      "eval_steps_per_second": 0.65,
+      "step": 4400
+    },
+    {
+      "epoch": 1.596825396825397,
+      "grad_norm": 0.166015625,
+      "learning_rate": 9.472524782768327e-05,
+      "loss": 1.0032,
+      "step": 4402
+    },
+    {
+      "epoch": 1.5975510204081633,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 9.467629421123486e-05,
+      "loss": 0.9867,
+      "step": 4404
+    },
+    {
+      "epoch": 1.5982766439909297,
+      "grad_norm": 0.162109375,
+      "learning_rate": 9.462734059478644e-05,
+      "loss": 1.0104,
+      "step": 4406
+    },
+    {
+      "epoch": 1.5990022675736961,
+      "grad_norm": 0.177734375,
+      "learning_rate": 9.457838697833803e-05,
+      "loss": 0.881,
+      "step": 4408
+    },
+    {
+      "epoch": 1.5997278911564625,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 9.452943336188962e-05,
+      "loss": 0.9992,
+      "step": 4410
+    },
+    {
+      "epoch": 1.600453514739229,
+      "grad_norm": 0.2236328125,
+      "learning_rate": 9.44804797454412e-05,
+      "loss": 0.9666,
+      "step": 4412
+    },
+    {
+      "epoch": 1.6011791383219953,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 9.443152612899278e-05,
+      "loss": 1.0162,
+      "step": 4414
+    },
+    {
+      "epoch": 1.601904761904762,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 9.438257251254436e-05,
+      "loss": 0.8902,
+      "step": 4416
+    },
+    {
+      "epoch": 1.6026303854875283,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 9.433361889609596e-05,
+      "loss": 0.9424,
+      "step": 4418
+    },
+    {
+      "epoch": 1.6033560090702947,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 9.428466527964754e-05,
+      "loss": 0.9927,
+      "step": 4420
+    },
+    {
+      "epoch": 1.6040816326530614,
+      "grad_norm": 0.177734375,
+      "learning_rate": 9.423571166319912e-05,
+      "loss": 0.9972,
+      "step": 4422
+    },
+    {
+      "epoch": 1.6048072562358278,
+      "grad_norm": 0.19921875,
+      "learning_rate": 9.41867580467507e-05,
+      "loss": 0.938,
+      "step": 4424
+    },
+    {
+      "epoch": 1.6055328798185942,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 9.413780443030229e-05,
+      "loss": 0.8708,
+      "step": 4426
+    },
+    {
+      "epoch": 1.6062585034013606,
+      "grad_norm": 0.1796875,
+      "learning_rate": 9.408885081385388e-05,
+      "loss": 1.0519,
+      "step": 4428
+    },
+    {
+      "epoch": 1.606984126984127,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 9.403989719740545e-05,
+      "loss": 0.9837,
+      "step": 4430
+    },
+    {
+      "epoch": 1.6077097505668934,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 9.399094358095705e-05,
+      "loss": 1.0007,
+      "step": 4432
+    },
+    {
+      "epoch": 1.6084353741496598,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 9.394198996450863e-05,
+      "loss": 0.9906,
+      "step": 4434
+    },
+    {
+      "epoch": 1.6091609977324262,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 9.389303634806021e-05,
+      "loss": 0.9736,
+      "step": 4436
+    },
+    {
+      "epoch": 1.6098866213151928,
+      "grad_norm": 0.193359375,
+      "learning_rate": 9.384408273161181e-05,
+      "loss": 1.0087,
+      "step": 4438
+    },
+    {
+      "epoch": 1.6106122448979592,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 9.379512911516338e-05,
+      "loss": 0.9447,
+      "step": 4440
+    },
+    {
+      "epoch": 1.6113378684807256,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 9.374617549871497e-05,
+      "loss": 1.0428,
+      "step": 4442
+    },
+    {
+      "epoch": 1.6120634920634922,
+      "grad_norm": 0.154296875,
+      "learning_rate": 9.369722188226656e-05,
+      "loss": 0.9664,
+      "step": 4444
+    },
+    {
+      "epoch": 1.6127891156462586,
+      "grad_norm": 0.173828125,
+      "learning_rate": 9.364826826581814e-05,
+      "loss": 0.9886,
+      "step": 4446
+    },
+    {
+      "epoch": 1.613514739229025,
+      "grad_norm": 0.19921875,
+      "learning_rate": 9.359931464936973e-05,
+      "loss": 1.0257,
+      "step": 4448
+    },
+    {
+      "epoch": 1.6142403628117914,
+      "grad_norm": 0.15234375,
+      "learning_rate": 9.35503610329213e-05,
+      "loss": 0.9166,
+      "step": 4450
+    },
+    {
+      "epoch": 1.6149659863945578,
+      "grad_norm": 0.201171875,
+      "learning_rate": 9.35014074164729e-05,
+      "loss": 1.0876,
+      "step": 4452
+    },
+    {
+      "epoch": 1.6156916099773242,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 9.345245380002448e-05,
+      "loss": 0.9517,
+      "step": 4454
+    },
+    {
+      "epoch": 1.6164172335600906,
+      "grad_norm": 0.154296875,
+      "learning_rate": 9.340350018357606e-05,
+      "loss": 0.9019,
+      "step": 4456
+    },
+    {
+      "epoch": 1.617142857142857,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 9.335454656712766e-05,
+      "loss": 0.9497,
+      "step": 4458
+    },
+    {
+      "epoch": 1.6178684807256236,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 9.330559295067923e-05,
+      "loss": 0.9503,
+      "step": 4460
+    },
+    {
+      "epoch": 1.61859410430839,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 9.325663933423082e-05,
+      "loss": 0.9898,
+      "step": 4462
+    },
+    {
+      "epoch": 1.6193197278911564,
+      "grad_norm": 0.15234375,
+      "learning_rate": 9.32076857177824e-05,
+      "loss": 0.9641,
+      "step": 4464
+    },
+    {
+      "epoch": 1.620045351473923,
+      "grad_norm": 0.1865234375,
+      "learning_rate": 9.315873210133399e-05,
+      "loss": 0.963,
+      "step": 4466
+    },
+    {
+      "epoch": 1.6207709750566894,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 9.310977848488558e-05,
+      "loss": 0.9416,
+      "step": 4468
+    },
+    {
+      "epoch": 1.6214965986394558,
+      "grad_norm": 0.162109375,
+      "learning_rate": 9.306082486843715e-05,
+      "loss": 1.0424,
+      "step": 4470
+    },
+    {
+      "epoch": 1.6222222222222222,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 9.301187125198875e-05,
+      "loss": 0.9659,
+      "step": 4472
+    },
+    {
+      "epoch": 1.6229478458049886,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 9.296291763554033e-05,
+      "loss": 0.8367,
+      "step": 4474
+    },
+    {
+      "epoch": 1.623673469387755,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 9.291396401909191e-05,
+      "loss": 0.9668,
+      "step": 4476
+    },
+    {
+      "epoch": 1.6243990929705214,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 9.286501040264351e-05,
+      "loss": 0.9655,
+      "step": 4478
+    },
+    {
+      "epoch": 1.6251247165532878,
+      "grad_norm": 0.189453125,
+      "learning_rate": 9.281605678619508e-05,
+      "loss": 0.9554,
+      "step": 4480
+    },
+    {
+      "epoch": 1.6258503401360545,
+      "grad_norm": 0.158203125,
+      "learning_rate": 9.276710316974667e-05,
+      "loss": 0.9243,
+      "step": 4482
+    },
+    {
+      "epoch": 1.6265759637188208,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 9.271814955329826e-05,
+      "loss": 0.9596,
+      "step": 4484
+    },
+    {
+      "epoch": 1.6273015873015872,
+      "grad_norm": 0.15625,
+      "learning_rate": 9.266919593684984e-05,
+      "loss": 0.9765,
+      "step": 4486
+    },
+    {
+      "epoch": 1.6280272108843539,
+      "grad_norm": 0.16015625,
+      "learning_rate": 9.262024232040143e-05,
+      "loss": 0.9482,
+      "step": 4488
+    },
+    {
+      "epoch": 1.6287528344671203,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 9.2571288703953e-05,
+      "loss": 0.9621,
+      "step": 4490
+    },
+    {
+      "epoch": 1.6294784580498867,
+      "grad_norm": 0.18359375,
+      "learning_rate": 9.25223350875046e-05,
+      "loss": 1.0181,
+      "step": 4492
+    },
+    {
+      "epoch": 1.630204081632653,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 9.247338147105618e-05,
+      "loss": 0.9446,
+      "step": 4494
+    },
+    {
+      "epoch": 1.6309297052154195,
+      "grad_norm": 0.162109375,
+      "learning_rate": 9.242442785460776e-05,
+      "loss": 0.9495,
+      "step": 4496
+    },
+    {
+      "epoch": 1.6316553287981859,
+      "grad_norm": 0.169921875,
+      "learning_rate": 9.237547423815934e-05,
+      "loss": 1.0283,
+      "step": 4498
+    },
+    {
+      "epoch": 1.6323809523809523,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 9.232652062171093e-05,
+      "loss": 1.0349,
+      "step": 4500
+    },
+    {
+      "epoch": 1.6323809523809523,
+      "eval_loss": 1.0125901699066162,
+      "eval_runtime": 86.0097,
+      "eval_samples_per_second": 20.928,
+      "eval_steps_per_second": 0.663,
+      "step": 4500
+    },
+    {
+      "epoch": 1.6331065759637187,
+      "grad_norm": 0.169921875,
+      "learning_rate": 9.227756700526252e-05,
+      "loss": 1.0119,
+      "step": 4502
+    },
+    {
+      "epoch": 1.6338321995464853,
+      "grad_norm": 0.162109375,
+      "learning_rate": 9.22286133888141e-05,
+      "loss": 0.9775,
+      "step": 4504
+    },
+    {
+      "epoch": 1.6345578231292517,
+      "grad_norm": 0.177734375,
+      "learning_rate": 9.217965977236569e-05,
+      "loss": 0.9544,
+      "step": 4506
+    },
+    {
+      "epoch": 1.6352834467120183,
+      "grad_norm": 0.16796875,
+      "learning_rate": 9.213070615591727e-05,
+      "loss": 0.941,
+      "step": 4508
+    },
+    {
+      "epoch": 1.6360090702947847,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 9.208175253946885e-05,
+      "loss": 0.9603,
+      "step": 4510
+    },
+    {
+      "epoch": 1.636734693877551,
+      "grad_norm": 0.154296875,
+      "learning_rate": 9.203279892302045e-05,
+      "loss": 0.9505,
+      "step": 4512
+    },
+    {
+      "epoch": 1.6374603174603175,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 9.198384530657203e-05,
+      "loss": 1.0291,
+      "step": 4514
+    },
+    {
+      "epoch": 1.638185941043084,
+      "grad_norm": 0.1923828125,
+      "learning_rate": 9.193489169012361e-05,
+      "loss": 0.8945,
+      "step": 4516
+    },
+    {
+      "epoch": 1.6389115646258503,
+      "grad_norm": 0.166015625,
+      "learning_rate": 9.18859380736752e-05,
+      "loss": 1.0053,
+      "step": 4518
+    },
+    {
+      "epoch": 1.6396371882086167,
+      "grad_norm": 0.17578125,
+      "learning_rate": 9.183698445722678e-05,
+      "loss": 0.9705,
+      "step": 4520
+    },
+    {
+      "epoch": 1.640362811791383,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 9.178803084077837e-05,
+      "loss": 0.9802,
+      "step": 4522
+    },
+    {
+      "epoch": 1.6410884353741495,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 9.173907722432996e-05,
+      "loss": 0.9674,
+      "step": 4524
+    },
+    {
+      "epoch": 1.6418140589569161,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 9.169012360788154e-05,
+      "loss": 1.0555,
+      "step": 4526
+    },
+    {
+      "epoch": 1.6425396825396825,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 9.164116999143312e-05,
+      "loss": 0.9737,
+      "step": 4528
+    },
+    {
+      "epoch": 1.6432653061224491,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 9.15922163749847e-05,
+      "loss": 0.9983,
+      "step": 4530
+    },
+    {
+      "epoch": 1.6439909297052155,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 9.15432627585363e-05,
+      "loss": 0.9307,
+      "step": 4532
+    },
+    {
+      "epoch": 1.644716553287982,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 9.149430914208788e-05,
+      "loss": 0.9447,
+      "step": 4534
+    },
+    {
+      "epoch": 1.6454421768707483,
+      "grad_norm": 0.15234375,
+      "learning_rate": 9.144535552563946e-05,
+      "loss": 0.9611,
+      "step": 4536
+    },
+    {
+      "epoch": 1.6461678004535147,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 9.139640190919104e-05,
+      "loss": 1.0478,
+      "step": 4538
+    },
+    {
+      "epoch": 1.6468934240362811,
+      "grad_norm": 0.17578125,
+      "learning_rate": 9.134744829274263e-05,
+      "loss": 1.002,
+      "step": 4540
+    },
+    {
+      "epoch": 1.6476190476190475,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 9.129849467629421e-05,
+      "loss": 0.9249,
+      "step": 4542
+    },
+    {
+      "epoch": 1.648344671201814,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 9.12495410598458e-05,
+      "loss": 0.9697,
+      "step": 4544
+    },
+    {
+      "epoch": 1.6490702947845803,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 9.120058744339739e-05,
+      "loss": 0.9693,
+      "step": 4546
+    },
+    {
+      "epoch": 1.649795918367347,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 9.115163382694897e-05,
+      "loss": 0.9989,
+      "step": 4548
+    },
+    {
+      "epoch": 1.6505215419501134,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 9.110268021050055e-05,
+      "loss": 1.0239,
+      "step": 4550
+    },
+    {
+      "epoch": 1.65124716553288,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 9.105372659405213e-05,
+      "loss": 0.9383,
+      "step": 4552
+    },
+    {
+      "epoch": 1.6519727891156464,
+      "grad_norm": 0.1796875,
+      "learning_rate": 9.100477297760373e-05,
+      "loss": 0.9475,
+      "step": 4554
+    },
+    {
+      "epoch": 1.6526984126984128,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 9.095581936115531e-05,
+      "loss": 0.9904,
+      "step": 4556
+    },
+    {
+      "epoch": 1.6534240362811792,
+      "grad_norm": 0.16796875,
+      "learning_rate": 9.09068657447069e-05,
+      "loss": 0.9174,
+      "step": 4558
+    },
+    {
+      "epoch": 1.6541496598639456,
+      "grad_norm": 0.166015625,
+      "learning_rate": 9.085791212825848e-05,
+      "loss": 1.0161,
+      "step": 4560
+    },
+    {
+      "epoch": 1.654875283446712,
+      "grad_norm": 0.16015625,
+      "learning_rate": 9.080895851181006e-05,
+      "loss": 1.1105,
+      "step": 4562
+    },
+    {
+      "epoch": 1.6556009070294784,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 9.076000489536165e-05,
+      "loss": 1.0257,
+      "step": 4564
+    },
+    {
+      "epoch": 1.6563265306122448,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 9.071105127891324e-05,
+      "loss": 0.9142,
+      "step": 4566
+    },
+    {
+      "epoch": 1.6570521541950112,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 9.066209766246482e-05,
+      "loss": 0.9849,
+      "step": 4568
+    },
+    {
+      "epoch": 1.6577777777777778,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 9.06131440460164e-05,
+      "loss": 0.9778,
+      "step": 4570
+    },
+    {
+      "epoch": 1.6585034013605442,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 9.056419042956798e-05,
+      "loss": 0.9673,
+      "step": 4572
+    },
+    {
+      "epoch": 1.6592290249433108,
+      "grad_norm": 0.16015625,
+      "learning_rate": 9.051523681311958e-05,
+      "loss": 1.0083,
+      "step": 4574
+    },
+    {
+      "epoch": 1.6599546485260772,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 9.046628319667115e-05,
+      "loss": 0.9852,
+      "step": 4576
+    },
+    {
+      "epoch": 1.6606802721088436,
+      "grad_norm": 0.19921875,
+      "learning_rate": 9.041732958022274e-05,
+      "loss": 0.9384,
+      "step": 4578
+    },
+    {
+      "epoch": 1.66140589569161,
+      "grad_norm": 0.166015625,
+      "learning_rate": 9.036837596377433e-05,
+      "loss": 1.0178,
+      "step": 4580
+    },
+    {
+      "epoch": 1.6621315192743764,
+      "grad_norm": 0.166015625,
+      "learning_rate": 9.031942234732591e-05,
+      "loss": 0.9881,
+      "step": 4582
+    },
+    {
+      "epoch": 1.6628571428571428,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 9.02704687308775e-05,
+      "loss": 1.0091,
+      "step": 4584
+    },
+    {
+      "epoch": 1.6635827664399092,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 9.022151511442907e-05,
+      "loss": 1.0278,
+      "step": 4586
+    },
+    {
+      "epoch": 1.6643083900226756,
+      "grad_norm": 0.1640625,
+      "learning_rate": 9.017256149798067e-05,
+      "loss": 0.9655,
+      "step": 4588
+    },
+    {
+      "epoch": 1.6650340136054422,
+      "grad_norm": 0.15625,
+      "learning_rate": 9.012360788153225e-05,
+      "loss": 0.9821,
+      "step": 4590
+    },
+    {
+      "epoch": 1.6657596371882086,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 9.007465426508383e-05,
+      "loss": 1.0376,
+      "step": 4592
+    },
+    {
+      "epoch": 1.666485260770975,
+      "grad_norm": 0.150390625,
+      "learning_rate": 9.002570064863543e-05,
+      "loss": 0.9686,
+      "step": 4594
+    },
+    {
+      "epoch": 1.6672108843537417,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 8.9976747032187e-05,
+      "loss": 1.0581,
+      "step": 4596
+    },
+    {
+      "epoch": 1.667936507936508,
+      "grad_norm": 0.201171875,
+      "learning_rate": 8.99277934157386e-05,
+      "loss": 0.9976,
+      "step": 4598
+    },
+    {
+      "epoch": 1.6686621315192744,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 8.987883979929018e-05,
+      "loss": 0.9585,
+      "step": 4600
+    },
+    {
+      "epoch": 1.6686621315192744,
+      "eval_loss": 1.0123869180679321,
+      "eval_runtime": 90.9583,
+      "eval_samples_per_second": 19.789,
+      "eval_steps_per_second": 0.627,
+      "step": 4600
+    },
+    {
+      "epoch": 1.6693877551020408,
+      "grad_norm": 0.15234375,
+      "learning_rate": 8.982988618284176e-05,
+      "loss": 0.9106,
+      "step": 4602
+    },
+    {
+      "epoch": 1.6701133786848072,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 8.978093256639335e-05,
+      "loss": 0.9744,
+      "step": 4604
+    },
+    {
+      "epoch": 1.6708390022675736,
+      "grad_norm": 0.1875,
+      "learning_rate": 8.973197894994492e-05,
+      "loss": 0.9738,
+      "step": 4606
+    },
+    {
+      "epoch": 1.67156462585034,
+      "grad_norm": 0.158203125,
+      "learning_rate": 8.968302533349652e-05,
+      "loss": 0.9827,
+      "step": 4608
+    },
+    {
+      "epoch": 1.6722902494331064,
+      "grad_norm": 0.220703125,
+      "learning_rate": 8.96340717170481e-05,
+      "loss": 1.0603,
+      "step": 4610
+    },
+    {
+      "epoch": 1.673015873015873,
+      "grad_norm": 0.162109375,
+      "learning_rate": 8.958511810059968e-05,
+      "loss": 0.9742,
+      "step": 4612
+    },
+    {
+      "epoch": 1.6737414965986395,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 8.953616448415128e-05,
+      "loss": 0.9745,
+      "step": 4614
+    },
+    {
+      "epoch": 1.6744671201814059,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 8.948721086770285e-05,
+      "loss": 1.039,
+      "step": 4616
+    },
+    {
+      "epoch": 1.6751927437641725,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 8.943825725125444e-05,
+      "loss": 0.917,
+      "step": 4618
+    },
+    {
+      "epoch": 1.6759183673469389,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 8.938930363480603e-05,
+      "loss": 1.0145,
+      "step": 4620
+    },
+    {
+      "epoch": 1.6766439909297053,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 8.934035001835761e-05,
+      "loss": 0.8931,
+      "step": 4622
+    },
+    {
+      "epoch": 1.6773696145124717,
+      "grad_norm": 0.169921875,
+      "learning_rate": 8.92913964019092e-05,
+      "loss": 0.9811,
+      "step": 4624
+    },
+    {
+      "epoch": 1.678095238095238,
+      "grad_norm": 0.185546875,
+      "learning_rate": 8.924244278546077e-05,
+      "loss": 0.9463,
+      "step": 4626
+    },
+    {
+      "epoch": 1.6788208616780045,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 8.919348916901237e-05,
+      "loss": 0.9064,
+      "step": 4628
+    },
+    {
+      "epoch": 1.6795464852607709,
+      "grad_norm": 0.177734375,
+      "learning_rate": 8.914453555256395e-05,
+      "loss": 1.0477,
+      "step": 4630
+    },
+    {
+      "epoch": 1.6802721088435373,
+      "grad_norm": 0.201171875,
+      "learning_rate": 8.909558193611553e-05,
+      "loss": 1.0279,
+      "step": 4632
+    },
+    {
+      "epoch": 1.680997732426304,
+      "grad_norm": 0.1640625,
+      "learning_rate": 8.904662831966713e-05,
+      "loss": 0.9726,
+      "step": 4634
+    },
+    {
+      "epoch": 1.6817233560090703,
+      "grad_norm": 0.1484375,
+      "learning_rate": 8.89976747032187e-05,
+      "loss": 0.9378,
+      "step": 4636
+    },
+    {
+      "epoch": 1.6824489795918367,
+      "grad_norm": 0.177734375,
+      "learning_rate": 8.89487210867703e-05,
+      "loss": 0.9877,
+      "step": 4638
+    },
+    {
+      "epoch": 1.6831746031746033,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 8.889976747032188e-05,
+      "loss": 0.9537,
+      "step": 4640
+    },
+    {
+      "epoch": 1.6839002267573697,
+      "grad_norm": 0.181640625,
+      "learning_rate": 8.885081385387346e-05,
+      "loss": 0.9996,
+      "step": 4642
+    },
+    {
+      "epoch": 1.6846258503401361,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 8.880186023742504e-05,
+      "loss": 0.9682,
+      "step": 4644
+    },
+    {
+      "epoch": 1.6853514739229025,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 8.875290662097662e-05,
+      "loss": 0.995,
+      "step": 4646
+    },
+    {
+      "epoch": 1.686077097505669,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 8.870395300452822e-05,
+      "loss": 0.9594,
+      "step": 4648
+    },
+    {
+      "epoch": 1.6868027210884353,
+      "grad_norm": 0.16796875,
+      "learning_rate": 8.86549993880798e-05,
+      "loss": 0.9323,
+      "step": 4650
+    },
+    {
+      "epoch": 1.6875283446712017,
+      "grad_norm": 0.171875,
+      "learning_rate": 8.860604577163138e-05,
+      "loss": 1.0264,
+      "step": 4652
+    },
+    {
+      "epoch": 1.6882539682539681,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 8.855709215518297e-05,
+      "loss": 0.9587,
+      "step": 4654
+    },
+    {
+      "epoch": 1.6889795918367347,
+      "grad_norm": 0.16796875,
+      "learning_rate": 8.850813853873455e-05,
+      "loss": 0.9402,
+      "step": 4656
+    },
+    {
+      "epoch": 1.6897052154195011,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 8.845918492228614e-05,
+      "loss": 0.9673,
+      "step": 4658
+    },
+    {
+      "epoch": 1.6904308390022675,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 8.841023130583773e-05,
+      "loss": 0.9555,
+      "step": 4660
+    },
+    {
+      "epoch": 1.6911564625850342,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 8.836127768938931e-05,
+      "loss": 0.9361,
+      "step": 4662
+    },
+    {
+      "epoch": 1.6918820861678006,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 8.831232407294089e-05,
+      "loss": 0.9679,
+      "step": 4664
+    },
+    {
+      "epoch": 1.692607709750567,
+      "grad_norm": 0.166015625,
+      "learning_rate": 8.826337045649247e-05,
+      "loss": 0.9455,
+      "step": 4666
+    },
+    {
+      "epoch": 1.6933333333333334,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 8.821441684004407e-05,
+      "loss": 0.9773,
+      "step": 4668
+    },
+    {
+      "epoch": 1.6940589569160998,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 8.816546322359565e-05,
+      "loss": 0.95,
+      "step": 4670
+    },
+    {
+      "epoch": 1.6947845804988662,
+      "grad_norm": 0.150390625,
+      "learning_rate": 8.811650960714723e-05,
+      "loss": 0.9737,
+      "step": 4672
+    },
+    {
+      "epoch": 1.6955102040816326,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 8.806755599069882e-05,
+      "loss": 1.0247,
+      "step": 4674
+    },
+    {
+      "epoch": 1.696235827664399,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 8.80186023742504e-05,
+      "loss": 0.9881,
+      "step": 4676
+    },
+    {
+      "epoch": 1.6969614512471656,
+      "grad_norm": 0.173828125,
+      "learning_rate": 8.7969648757802e-05,
+      "loss": 1.0309,
+      "step": 4678
+    },
+    {
+      "epoch": 1.697687074829932,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 8.792069514135358e-05,
+      "loss": 0.9641,
+      "step": 4680
+    },
+    {
+      "epoch": 1.6984126984126984,
+      "grad_norm": 0.1640625,
+      "learning_rate": 8.787174152490516e-05,
+      "loss": 0.9768,
+      "step": 4682
+    },
+    {
+      "epoch": 1.699138321995465,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 8.782278790845674e-05,
+      "loss": 0.9526,
+      "step": 4684
+    },
+    {
+      "epoch": 1.6998639455782314,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 8.777383429200832e-05,
+      "loss": 1.0107,
+      "step": 4686
+    },
+    {
+      "epoch": 1.7005895691609978,
+      "grad_norm": 0.17578125,
+      "learning_rate": 8.77248806755599e-05,
+      "loss": 1.0372,
+      "step": 4688
+    },
+    {
+      "epoch": 1.7013151927437642,
+      "grad_norm": 0.16015625,
+      "learning_rate": 8.76759270591115e-05,
+      "loss": 0.949,
+      "step": 4690
+    },
+    {
+      "epoch": 1.7020408163265306,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 8.762697344266308e-05,
+      "loss": 0.9846,
+      "step": 4692
+    },
+    {
+      "epoch": 1.702766439909297,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 8.757801982621467e-05,
+      "loss": 0.9488,
+      "step": 4694
+    },
+    {
+      "epoch": 1.7034920634920634,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 8.752906620976625e-05,
+      "loss": 0.9439,
+      "step": 4696
+    },
+    {
+      "epoch": 1.7042176870748298,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 8.748011259331783e-05,
+      "loss": 0.9623,
+      "step": 4698
+    },
+    {
+      "epoch": 1.7049433106575964,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 8.743115897686943e-05,
+      "loss": 1.0633,
+      "step": 4700
+    },
+    {
+      "epoch": 1.7049433106575964,
+      "eval_loss": 1.0122075080871582,
+      "eval_runtime": 84.9355,
+      "eval_samples_per_second": 21.193,
+      "eval_steps_per_second": 0.671,
+      "step": 4700
+    },
+    {
+      "epoch": 1.7056689342403628,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 8.738220536042101e-05,
+      "loss": 0.9102,
+      "step": 4702
+    },
+    {
+      "epoch": 1.7063945578231292,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 8.733325174397259e-05,
+      "loss": 0.9428,
+      "step": 4704
+    },
+    {
+      "epoch": 1.7071201814058958,
+      "grad_norm": 0.15625,
+      "learning_rate": 8.728429812752417e-05,
+      "loss": 0.9601,
+      "step": 4706
+    },
+    {
+      "epoch": 1.7078458049886622,
+      "grad_norm": 0.16015625,
+      "learning_rate": 8.723534451107576e-05,
+      "loss": 0.9997,
+      "step": 4708
+    },
+    {
+      "epoch": 1.7085714285714286,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 8.718639089462735e-05,
+      "loss": 0.9719,
+      "step": 4710
+    },
+    {
+      "epoch": 1.709297052154195,
+      "grad_norm": 0.162109375,
+      "learning_rate": 8.713743727817893e-05,
+      "loss": 0.938,
+      "step": 4712
+    },
+    {
+      "epoch": 1.7100226757369614,
+      "grad_norm": 0.16796875,
+      "learning_rate": 8.708848366173052e-05,
+      "loss": 0.9769,
+      "step": 4714
+    },
+    {
+      "epoch": 1.7107482993197278,
+      "grad_norm": 0.18359375,
+      "learning_rate": 8.70395300452821e-05,
+      "loss": 0.9675,
+      "step": 4716
+    },
+    {
+      "epoch": 1.7114739229024942,
+      "grad_norm": 0.1865234375,
+      "learning_rate": 8.699057642883368e-05,
+      "loss": 0.9881,
+      "step": 4718
+    },
+    {
+      "epoch": 1.7121995464852606,
+      "grad_norm": 0.1640625,
+      "learning_rate": 8.694162281238528e-05,
+      "loss": 1.0473,
+      "step": 4720
+    },
+    {
+      "epoch": 1.7129251700680272,
+      "grad_norm": 0.15234375,
+      "learning_rate": 8.689266919593684e-05,
+      "loss": 0.9488,
+      "step": 4722
+    },
+    {
+      "epoch": 1.7136507936507936,
+      "grad_norm": 0.1796875,
+      "learning_rate": 8.684371557948844e-05,
+      "loss": 1.0057,
+      "step": 4724
+    },
+    {
+      "epoch": 1.71437641723356,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 8.679476196304002e-05,
+      "loss": 0.9249,
+      "step": 4726
+    },
+    {
+      "epoch": 1.7151020408163267,
+      "grad_norm": 0.1943359375,
+      "learning_rate": 8.67458083465916e-05,
+      "loss": 0.9699,
+      "step": 4728
+    },
+    {
+      "epoch": 1.715827664399093,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 8.66968547301432e-05,
+      "loss": 1.001,
+      "step": 4730
+    },
+    {
+      "epoch": 1.7165532879818595,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 8.664790111369477e-05,
+      "loss": 0.901,
+      "step": 4732
+    },
+    {
+      "epoch": 1.7172789115646259,
+      "grad_norm": 0.173828125,
+      "learning_rate": 8.659894749724637e-05,
+      "loss": 1.026,
+      "step": 4734
+    },
+    {
+      "epoch": 1.7180045351473923,
+      "grad_norm": 0.173828125,
+      "learning_rate": 8.654999388079795e-05,
+      "loss": 0.9686,
+      "step": 4736
+    },
+    {
+      "epoch": 1.7187301587301587,
+      "grad_norm": 0.203125,
+      "learning_rate": 8.650104026434953e-05,
+      "loss": 1.0217,
+      "step": 4738
+    },
+    {
+      "epoch": 1.719455782312925,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 8.645208664790113e-05,
+      "loss": 1.0212,
+      "step": 4740
+    },
+    {
+      "epoch": 1.7201814058956915,
+      "grad_norm": 0.162109375,
+      "learning_rate": 8.64031330314527e-05,
+      "loss": 0.9317,
+      "step": 4742
+    },
+    {
+      "epoch": 1.720907029478458,
+      "grad_norm": 0.158203125,
+      "learning_rate": 8.635417941500429e-05,
+      "loss": 0.9294,
+      "step": 4744
+    },
+    {
+      "epoch": 1.7216326530612245,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 8.630522579855587e-05,
+      "loss": 0.9838,
+      "step": 4746
+    },
+    {
+      "epoch": 1.7223582766439909,
+      "grad_norm": 0.1796875,
+      "learning_rate": 8.625627218210746e-05,
+      "loss": 0.9376,
+      "step": 4748
+    },
+    {
+      "epoch": 1.7230839002267575,
+      "grad_norm": 0.177734375,
+      "learning_rate": 8.620731856565905e-05,
+      "loss": 1.0101,
+      "step": 4750
+    },
+    {
+      "epoch": 1.723809523809524,
+      "grad_norm": 0.16796875,
+      "learning_rate": 8.615836494921062e-05,
+      "loss": 0.9394,
+      "step": 4752
+    },
+    {
+      "epoch": 1.7245351473922903,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 8.610941133276222e-05,
+      "loss": 1.0197,
+      "step": 4754
+    },
+    {
+      "epoch": 1.7252607709750567,
+      "grad_norm": 0.166015625,
+      "learning_rate": 8.60604577163138e-05,
+      "loss": 1.0194,
+      "step": 4756
+    },
+    {
+      "epoch": 1.725986394557823,
+      "grad_norm": 0.1865234375,
+      "learning_rate": 8.601150409986538e-05,
+      "loss": 1.0242,
+      "step": 4758
+    },
+    {
+      "epoch": 1.7267120181405895,
+      "grad_norm": 0.185546875,
+      "learning_rate": 8.596255048341698e-05,
+      "loss": 0.9163,
+      "step": 4760
+    },
+    {
+      "epoch": 1.727437641723356,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 8.591359686696854e-05,
+      "loss": 0.9777,
+      "step": 4762
+    },
+    {
+      "epoch": 1.7281632653061223,
+      "grad_norm": 0.162109375,
+      "learning_rate": 8.586464325052014e-05,
+      "loss": 0.9872,
+      "step": 4764
+    },
+    {
+      "epoch": 1.728888888888889,
+      "grad_norm": 0.162109375,
+      "learning_rate": 8.581568963407172e-05,
+      "loss": 1.059,
+      "step": 4766
+    },
+    {
+      "epoch": 1.7296145124716553,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 8.57667360176233e-05,
+      "loss": 1.0383,
+      "step": 4768
+    },
+    {
+      "epoch": 1.7303401360544217,
+      "grad_norm": 0.158203125,
+      "learning_rate": 8.57177824011749e-05,
+      "loss": 0.9363,
+      "step": 4770
+    },
+    {
+      "epoch": 1.7310657596371883,
+      "grad_norm": 0.154296875,
+      "learning_rate": 8.566882878472647e-05,
+      "loss": 0.8923,
+      "step": 4772
+    },
+    {
+      "epoch": 1.7317913832199547,
+      "grad_norm": 0.1640625,
+      "learning_rate": 8.561987516827807e-05,
+      "loss": 0.9878,
+      "step": 4774
+    },
+    {
+      "epoch": 1.7325170068027211,
+      "grad_norm": 0.1875,
+      "learning_rate": 8.557092155182965e-05,
+      "loss": 1.0398,
+      "step": 4776
+    },
+    {
+      "epoch": 1.7332426303854875,
+      "grad_norm": 0.19140625,
+      "learning_rate": 8.552196793538123e-05,
+      "loss": 0.9416,
+      "step": 4778
+    },
+    {
+      "epoch": 1.733968253968254,
+      "grad_norm": 0.1953125,
+      "learning_rate": 8.547301431893283e-05,
+      "loss": 1.051,
+      "step": 4780
+    },
+    {
+      "epoch": 1.7346938775510203,
+      "grad_norm": 0.212890625,
+      "learning_rate": 8.54240607024844e-05,
+      "loss": 1.0494,
+      "step": 4782
+    },
+    {
+      "epoch": 1.7354195011337867,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 8.537510708603599e-05,
+      "loss": 1.0535,
+      "step": 4784
+    },
+    {
+      "epoch": 1.7361451247165531,
+      "grad_norm": 0.173828125,
+      "learning_rate": 8.532615346958757e-05,
+      "loss": 1.0028,
+      "step": 4786
+    },
+    {
+      "epoch": 1.7368707482993198,
+      "grad_norm": 0.189453125,
+      "learning_rate": 8.527719985313915e-05,
+      "loss": 1.0047,
+      "step": 4788
+    },
+    {
+      "epoch": 1.7375963718820862,
+      "grad_norm": 0.166015625,
+      "learning_rate": 8.522824623669074e-05,
+      "loss": 0.9975,
+      "step": 4790
+    },
+    {
+      "epoch": 1.7383219954648526,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 8.517929262024232e-05,
+      "loss": 1.0143,
+      "step": 4792
+    },
+    {
+      "epoch": 1.7390476190476192,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 8.513033900379392e-05,
+      "loss": 0.948,
+      "step": 4794
+    },
+    {
+      "epoch": 1.7397732426303856,
+      "grad_norm": 0.1923828125,
+      "learning_rate": 8.50813853873455e-05,
+      "loss": 0.9424,
+      "step": 4796
+    },
+    {
+      "epoch": 1.740498866213152,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 8.503243177089708e-05,
+      "loss": 0.9775,
+      "step": 4798
+    },
+    {
+      "epoch": 1.7412244897959184,
+      "grad_norm": 0.19140625,
+      "learning_rate": 8.498347815444866e-05,
+      "loss": 0.9664,
+      "step": 4800
+    },
+    {
+      "epoch": 1.7412244897959184,
+      "eval_loss": 1.012211561203003,
+      "eval_runtime": 102.4092,
+      "eval_samples_per_second": 17.577,
+      "eval_steps_per_second": 0.557,
+      "step": 4800
+    },
+    {
+      "epoch": 1.7419501133786848,
+      "grad_norm": 0.18359375,
+      "learning_rate": 8.493452453800024e-05,
+      "loss": 0.9895,
+      "step": 4802
+    },
+    {
+      "epoch": 1.7426757369614512,
+      "grad_norm": 0.1484375,
+      "learning_rate": 8.488557092155184e-05,
+      "loss": 0.995,
+      "step": 4804
+    },
+    {
+      "epoch": 1.7434013605442176,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 8.483661730510342e-05,
+      "loss": 0.9932,
+      "step": 4806
+    },
+    {
+      "epoch": 1.744126984126984,
+      "grad_norm": 0.1484375,
+      "learning_rate": 8.4787663688655e-05,
+      "loss": 1.005,
+      "step": 4808
+    },
+    {
+      "epoch": 1.7448526077097506,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 8.473871007220659e-05,
+      "loss": 0.9672,
+      "step": 4810
+    },
+    {
+      "epoch": 1.745578231292517,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 8.468975645575817e-05,
+      "loss": 0.9399,
+      "step": 4812
+    },
+    {
+      "epoch": 1.7463038548752836,
+      "grad_norm": 0.15625,
+      "learning_rate": 8.464080283930977e-05,
+      "loss": 0.9642,
+      "step": 4814
+    },
+    {
+      "epoch": 1.74702947845805,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 8.459184922286135e-05,
+      "loss": 0.9698,
+      "step": 4816
+    },
+    {
+      "epoch": 1.7477551020408164,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 8.454289560641293e-05,
+      "loss": 0.981,
+      "step": 4818
+    },
+    {
+      "epoch": 1.7484807256235828,
+      "grad_norm": 0.169921875,
+      "learning_rate": 8.449394198996451e-05,
+      "loss": 0.9702,
+      "step": 4820
+    },
+    {
+      "epoch": 1.7492063492063492,
+      "grad_norm": 0.150390625,
+      "learning_rate": 8.44449883735161e-05,
+      "loss": 0.9315,
+      "step": 4822
+    },
+    {
+      "epoch": 1.7499319727891156,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 8.439603475706769e-05,
+      "loss": 0.9465,
+      "step": 4824
+    },
+    {
+      "epoch": 1.750657596371882,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 8.434708114061927e-05,
+      "loss": 1.013,
+      "step": 4826
+    },
+    {
+      "epoch": 1.7513832199546484,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 8.429812752417085e-05,
+      "loss": 0.9432,
+      "step": 4828
+    },
+    {
+      "epoch": 1.7521088435374148,
+      "grad_norm": 0.181640625,
+      "learning_rate": 8.424917390772244e-05,
+      "loss": 1.0176,
+      "step": 4830
+    },
+    {
+      "epoch": 1.7528344671201814,
+      "grad_norm": 0.171875,
+      "learning_rate": 8.420022029127402e-05,
+      "loss": 1.0079,
+      "step": 4832
+    },
+    {
+      "epoch": 1.7535600907029478,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 8.41512666748256e-05,
+      "loss": 0.9487,
+      "step": 4834
+    },
+    {
+      "epoch": 1.7542857142857144,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 8.41023130583772e-05,
+      "loss": 0.945,
+      "step": 4836
+    },
+    {
+      "epoch": 1.7550113378684808,
+      "grad_norm": 0.166015625,
+      "learning_rate": 8.405335944192878e-05,
+      "loss": 0.9486,
+      "step": 4838
+    },
+    {
+      "epoch": 1.7557369614512472,
+      "grad_norm": 0.1796875,
+      "learning_rate": 8.400440582548036e-05,
+      "loss": 0.9777,
+      "step": 4840
+    },
+    {
+      "epoch": 1.7564625850340136,
+      "grad_norm": 0.16015625,
+      "learning_rate": 8.395545220903194e-05,
+      "loss": 0.9026,
+      "step": 4842
+    },
+    {
+      "epoch": 1.75718820861678,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 8.390649859258353e-05,
+      "loss": 1.082,
+      "step": 4844
+    },
+    {
+      "epoch": 1.7579138321995464,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 8.385754497613512e-05,
+      "loss": 0.9822,
+      "step": 4846
+    },
+    {
+      "epoch": 1.7586394557823128,
+      "grad_norm": 0.162109375,
+      "learning_rate": 8.38085913596867e-05,
+      "loss": 0.9299,
+      "step": 4848
+    },
+    {
+      "epoch": 1.7593650793650792,
+      "grad_norm": 0.1640625,
+      "learning_rate": 8.375963774323829e-05,
+      "loss": 0.9228,
+      "step": 4850
+    },
+    {
+      "epoch": 1.7600907029478456,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 8.371068412678987e-05,
+      "loss": 0.985,
+      "step": 4852
+    },
+    {
+      "epoch": 1.7608163265306123,
+      "grad_norm": 0.181640625,
+      "learning_rate": 8.366173051034145e-05,
+      "loss": 0.9734,
+      "step": 4854
+    },
+    {
+      "epoch": 1.7615419501133787,
+      "grad_norm": 0.1640625,
+      "learning_rate": 8.361277689389305e-05,
+      "loss": 0.9293,
+      "step": 4856
+    },
+    {
+      "epoch": 1.7622675736961453,
+      "grad_norm": 0.158203125,
+      "learning_rate": 8.356382327744463e-05,
+      "loss": 0.99,
+      "step": 4858
+    },
+    {
+      "epoch": 1.7629931972789117,
+      "grad_norm": 0.1640625,
+      "learning_rate": 8.351486966099621e-05,
+      "loss": 0.9639,
+      "step": 4860
+    },
+    {
+      "epoch": 1.763718820861678,
+      "grad_norm": 0.177734375,
+      "learning_rate": 8.34659160445478e-05,
+      "loss": 1.0291,
+      "step": 4862
+    },
+    {
+      "epoch": 1.7644444444444445,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 8.341696242809938e-05,
+      "loss": 0.9175,
+      "step": 4864
+    },
+    {
+      "epoch": 1.7651700680272109,
+      "grad_norm": 0.17578125,
+      "learning_rate": 8.336800881165097e-05,
+      "loss": 1.0067,
+      "step": 4866
+    },
+    {
+      "epoch": 1.7658956916099773,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 8.331905519520254e-05,
+      "loss": 0.9996,
+      "step": 4868
+    },
+    {
+      "epoch": 1.7666213151927437,
+      "grad_norm": 0.16796875,
+      "learning_rate": 8.327010157875414e-05,
+      "loss": 0.9372,
+      "step": 4870
+    },
+    {
+      "epoch": 1.76734693877551,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 8.322114796230572e-05,
+      "loss": 0.9316,
+      "step": 4872
+    },
+    {
+      "epoch": 1.7680725623582765,
+      "grad_norm": 0.15234375,
+      "learning_rate": 8.31721943458573e-05,
+      "loss": 1.0023,
+      "step": 4874
+    },
+    {
+      "epoch": 1.768798185941043,
+      "grad_norm": 0.177734375,
+      "learning_rate": 8.31232407294089e-05,
+      "loss": 1.0662,
+      "step": 4876
+    },
+    {
+      "epoch": 1.7695238095238095,
+      "grad_norm": 0.1796875,
+      "learning_rate": 8.307428711296047e-05,
+      "loss": 0.9817,
+      "step": 4878
+    },
+    {
+      "epoch": 1.7702494331065761,
+      "grad_norm": 0.162109375,
+      "learning_rate": 8.302533349651206e-05,
+      "loss": 0.9366,
+      "step": 4880
+    },
+    {
+      "epoch": 1.7709750566893425,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 8.297637988006364e-05,
+      "loss": 0.956,
+      "step": 4882
+    },
+    {
+      "epoch": 1.771700680272109,
+      "grad_norm": 0.140625,
+      "learning_rate": 8.292742626361523e-05,
+      "loss": 0.9326,
+      "step": 4884
+    },
+    {
+      "epoch": 1.7724263038548753,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 8.287847264716682e-05,
+      "loss": 0.9756,
+      "step": 4886
+    },
+    {
+      "epoch": 1.7731519274376417,
+      "grad_norm": 0.173828125,
+      "learning_rate": 8.282951903071839e-05,
+      "loss": 1.0435,
+      "step": 4888
+    },
+    {
+      "epoch": 1.7738775510204081,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 8.278056541426999e-05,
+      "loss": 1.0294,
+      "step": 4890
+    },
+    {
+      "epoch": 1.7746031746031745,
+      "grad_norm": 0.154296875,
+      "learning_rate": 8.273161179782157e-05,
+      "loss": 0.9823,
+      "step": 4892
+    },
+    {
+      "epoch": 1.775328798185941,
+      "grad_norm": 0.162109375,
+      "learning_rate": 8.268265818137315e-05,
+      "loss": 0.9292,
+      "step": 4894
+    },
+    {
+      "epoch": 1.7760544217687075,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 8.263370456492475e-05,
+      "loss": 0.9869,
+      "step": 4896
+    },
+    {
+      "epoch": 1.776780045351474,
+      "grad_norm": 0.16015625,
+      "learning_rate": 8.258475094847632e-05,
+      "loss": 0.9646,
+      "step": 4898
+    },
+    {
+      "epoch": 1.7775056689342403,
+      "grad_norm": 0.1875,
+      "learning_rate": 8.253579733202791e-05,
+      "loss": 0.9385,
+      "step": 4900
+    },
+    {
+      "epoch": 1.7775056689342403,
+      "eval_loss": 1.0119349956512451,
+      "eval_runtime": 86.664,
+      "eval_samples_per_second": 20.77,
+      "eval_steps_per_second": 0.658,
+      "step": 4900
+    },
+    {
+      "epoch": 1.778231292517007,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 8.248684371557948e-05,
+      "loss": 0.9462,
+      "step": 4902
+    },
+    {
+      "epoch": 1.7789569160997734,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 8.243789009913108e-05,
+      "loss": 1.0036,
+      "step": 4904
+    },
+    {
+      "epoch": 1.7796825396825398,
+      "grad_norm": 0.166015625,
+      "learning_rate": 8.238893648268267e-05,
+      "loss": 0.9715,
+      "step": 4906
+    },
+    {
+      "epoch": 1.7804081632653062,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 8.233998286623424e-05,
+      "loss": 1.0437,
+      "step": 4908
+    },
+    {
+      "epoch": 1.7811337868480726,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 8.229102924978584e-05,
+      "loss": 0.9423,
+      "step": 4910
+    },
+    {
+      "epoch": 1.781859410430839,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 8.22420756333374e-05,
+      "loss": 0.9602,
+      "step": 4912
+    },
+    {
+      "epoch": 1.7825850340136054,
+      "grad_norm": 0.16015625,
+      "learning_rate": 8.2193122016889e-05,
+      "loss": 0.9675,
+      "step": 4914
+    },
+    {
+      "epoch": 1.7833106575963718,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 8.21441684004406e-05,
+      "loss": 0.9956,
+      "step": 4916
+    },
+    {
+      "epoch": 1.7840362811791384,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 8.209521478399217e-05,
+      "loss": 0.9582,
+      "step": 4918
+    },
+    {
+      "epoch": 1.7847619047619048,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 8.204626116754376e-05,
+      "loss": 0.9657,
+      "step": 4920
+    },
+    {
+      "epoch": 1.7854875283446712,
+      "grad_norm": 0.169921875,
+      "learning_rate": 8.199730755109533e-05,
+      "loss": 0.9727,
+      "step": 4922
+    },
+    {
+      "epoch": 1.7862131519274378,
+      "grad_norm": 0.185546875,
+      "learning_rate": 8.194835393464693e-05,
+      "loss": 0.9675,
+      "step": 4924
+    },
+    {
+      "epoch": 1.7869387755102042,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 8.189940031819852e-05,
+      "loss": 0.9393,
+      "step": 4926
+    },
+    {
+      "epoch": 1.7876643990929706,
+      "grad_norm": 0.16015625,
+      "learning_rate": 8.185044670175009e-05,
+      "loss": 0.9847,
+      "step": 4928
+    },
+    {
+      "epoch": 1.788390022675737,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 8.180149308530169e-05,
+      "loss": 0.9224,
+      "step": 4930
+    },
+    {
+      "epoch": 1.7891156462585034,
+      "grad_norm": 0.162109375,
+      "learning_rate": 8.175253946885327e-05,
+      "loss": 1.0141,
+      "step": 4932
+    },
+    {
+      "epoch": 1.7898412698412698,
+      "grad_norm": 0.173828125,
+      "learning_rate": 8.170358585240485e-05,
+      "loss": 1.0221,
+      "step": 4934
+    },
+    {
+      "epoch": 1.7905668934240362,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 8.165463223595643e-05,
+      "loss": 0.9043,
+      "step": 4936
+    },
+    {
+      "epoch": 1.7912925170068026,
+      "grad_norm": 0.173828125,
+      "learning_rate": 8.160567861950802e-05,
+      "loss": 0.9836,
+      "step": 4938
+    },
+    {
+      "epoch": 1.7920181405895692,
+      "grad_norm": 0.1640625,
+      "learning_rate": 8.155672500305961e-05,
+      "loss": 0.9173,
+      "step": 4940
+    },
+    {
+      "epoch": 1.7927437641723356,
+      "grad_norm": 0.16796875,
+      "learning_rate": 8.15077713866112e-05,
+      "loss": 0.9854,
+      "step": 4942
+    },
+    {
+      "epoch": 1.793469387755102,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 8.145881777016278e-05,
+      "loss": 0.9927,
+      "step": 4944
+    },
+    {
+      "epoch": 1.7941950113378686,
+      "grad_norm": 0.16796875,
+      "learning_rate": 8.140986415371436e-05,
+      "loss": 0.9674,
+      "step": 4946
+    },
+    {
+      "epoch": 1.794920634920635,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 8.136091053726594e-05,
+      "loss": 0.9844,
+      "step": 4948
+    },
+    {
+      "epoch": 1.7956462585034014,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 8.131195692081754e-05,
+      "loss": 0.8923,
+      "step": 4950
+    },
+    {
+      "epoch": 1.7963718820861678,
+      "grad_norm": 0.154296875,
+      "learning_rate": 8.126300330436912e-05,
+      "loss": 0.9434,
+      "step": 4952
+    },
+    {
+      "epoch": 1.7970975056689342,
+      "grad_norm": 0.185546875,
+      "learning_rate": 8.12140496879207e-05,
+      "loss": 1.0141,
+      "step": 4954
+    },
+    {
+      "epoch": 1.7978231292517006,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 8.116509607147228e-05,
+      "loss": 0.9718,
+      "step": 4956
+    },
+    {
+      "epoch": 1.798548752834467,
+      "grad_norm": 0.1884765625,
+      "learning_rate": 8.111614245502387e-05,
+      "loss": 1.1004,
+      "step": 4958
+    },
+    {
+      "epoch": 1.7992743764172334,
+      "grad_norm": 0.169921875,
+      "learning_rate": 8.106718883857546e-05,
+      "loss": 0.9479,
+      "step": 4960
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.154296875,
+      "learning_rate": 8.101823522212704e-05,
+      "loss": 0.9463,
+      "step": 4962
+    },
+    {
+      "epoch": 1.8007256235827664,
+      "grad_norm": 0.171875,
+      "learning_rate": 8.096928160567863e-05,
+      "loss": 0.9659,
+      "step": 4964
+    },
+    {
+      "epoch": 1.8014512471655328,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 8.092032798923021e-05,
+      "loss": 1.0237,
+      "step": 4966
+    },
+    {
+      "epoch": 1.8021768707482995,
+      "grad_norm": 0.17578125,
+      "learning_rate": 8.087137437278179e-05,
+      "loss": 1.0192,
+      "step": 4968
+    },
+    {
+      "epoch": 1.8029024943310659,
+      "grad_norm": 0.1640625,
+      "learning_rate": 8.082242075633339e-05,
+      "loss": 0.9587,
+      "step": 4970
+    },
+    {
+      "epoch": 1.8036281179138323,
+      "grad_norm": 0.16796875,
+      "learning_rate": 8.077346713988497e-05,
+      "loss": 1.0048,
+      "step": 4972
+    },
+    {
+      "epoch": 1.8043537414965987,
+      "grad_norm": 0.18359375,
+      "learning_rate": 8.072451352343655e-05,
+      "loss": 0.9607,
+      "step": 4974
+    },
+    {
+      "epoch": 1.805079365079365,
+      "grad_norm": 0.1640625,
+      "learning_rate": 8.067555990698813e-05,
+      "loss": 1.0168,
+      "step": 4976
+    },
+    {
+      "epoch": 1.8058049886621315,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 8.062660629053972e-05,
+      "loss": 0.9538,
+      "step": 4978
+    },
+    {
+      "epoch": 1.8065306122448979,
+      "grad_norm": 0.1796875,
+      "learning_rate": 8.05776526740913e-05,
+      "loss": 0.9468,
+      "step": 4980
+    },
+    {
+      "epoch": 1.8072562358276643,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 8.05286990576429e-05,
+      "loss": 0.995,
+      "step": 4982
+    },
+    {
+      "epoch": 1.8079818594104309,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 8.047974544119448e-05,
+      "loss": 0.954,
+      "step": 4984
+    },
+    {
+      "epoch": 1.8087074829931973,
+      "grad_norm": 0.1484375,
+      "learning_rate": 8.043079182474606e-05,
+      "loss": 0.957,
+      "step": 4986
+    },
+    {
+      "epoch": 1.8094331065759637,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 8.038183820829764e-05,
+      "loss": 0.9804,
+      "step": 4988
+    },
+    {
+      "epoch": 1.8101587301587303,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 8.033288459184922e-05,
+      "loss": 0.9968,
+      "step": 4990
+    },
+    {
+      "epoch": 1.8108843537414967,
+      "grad_norm": 0.1640625,
+      "learning_rate": 8.028393097540082e-05,
+      "loss": 0.9741,
+      "step": 4992
+    },
+    {
+      "epoch": 1.811609977324263,
+      "grad_norm": 0.1982421875,
+      "learning_rate": 8.02349773589524e-05,
+      "loss": 0.9552,
+      "step": 4994
+    },
+    {
+      "epoch": 1.8123356009070295,
+      "grad_norm": 0.16015625,
+      "learning_rate": 8.018602374250398e-05,
+      "loss": 1.0086,
+      "step": 4996
+    },
+    {
+      "epoch": 1.813061224489796,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 8.013707012605557e-05,
+      "loss": 0.9693,
+      "step": 4998
+    },
+    {
+      "epoch": 1.8137868480725623,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 8.008811650960715e-05,
+      "loss": 1.0028,
+      "step": 5000
+    },
+    {
+      "epoch": 1.8137868480725623,
+      "eval_loss": 1.0117406845092773,
+      "eval_runtime": 82.6702,
+      "eval_samples_per_second": 21.773,
+      "eval_steps_per_second": 0.689,
+      "step": 5000
+    },
+    {
+      "epoch": 1.8145124716553287,
+      "grad_norm": 0.15234375,
+      "learning_rate": 8.003916289315874e-05,
+      "loss": 0.968,
+      "step": 5002
+    },
+    {
+      "epoch": 1.815238095238095,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 7.999020927671033e-05,
+      "loss": 1.0029,
+      "step": 5004
+    },
+    {
+      "epoch": 1.8159637188208617,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 7.994125566026191e-05,
+      "loss": 0.9718,
+      "step": 5006
+    },
+    {
+      "epoch": 1.8166893424036281,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 7.989230204381349e-05,
+      "loss": 0.9443,
+      "step": 5008
+    },
+    {
+      "epoch": 1.8174149659863945,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 7.984334842736507e-05,
+      "loss": 1.054,
+      "step": 5010
+    },
+    {
+      "epoch": 1.8181405895691611,
+      "grad_norm": 0.171875,
+      "learning_rate": 7.979439481091667e-05,
+      "loss": 1.0275,
+      "step": 5012
+    },
+    {
+      "epoch": 1.8188662131519275,
+      "grad_norm": 0.181640625,
+      "learning_rate": 7.974544119446824e-05,
+      "loss": 1.0257,
+      "step": 5014
+    },
+    {
+      "epoch": 1.819591836734694,
+      "grad_norm": 0.16015625,
+      "learning_rate": 7.969648757801983e-05,
+      "loss": 0.9781,
+      "step": 5016
+    },
+    {
+      "epoch": 1.8203174603174603,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 7.964753396157142e-05,
+      "loss": 0.9828,
+      "step": 5018
+    },
+    {
+      "epoch": 1.8210430839002267,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 7.9598580345123e-05,
+      "loss": 0.9198,
+      "step": 5020
+    },
+    {
+      "epoch": 1.8217687074829931,
+      "grad_norm": 0.173828125,
+      "learning_rate": 7.95496267286746e-05,
+      "loss": 0.9939,
+      "step": 5022
+    },
+    {
+      "epoch": 1.8224943310657595,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 7.950067311222616e-05,
+      "loss": 0.9844,
+      "step": 5024
+    },
+    {
+      "epoch": 1.823219954648526,
+      "grad_norm": 0.1875,
+      "learning_rate": 7.945171949577776e-05,
+      "loss": 0.9858,
+      "step": 5026
+    },
+    {
+      "epoch": 1.8239455782312926,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 7.940276587932934e-05,
+      "loss": 0.9681,
+      "step": 5028
+    },
+    {
+      "epoch": 1.824671201814059,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 7.935381226288092e-05,
+      "loss": 1.0324,
+      "step": 5030
+    },
+    {
+      "epoch": 1.8253968253968254,
+      "grad_norm": 0.216796875,
+      "learning_rate": 7.930485864643252e-05,
+      "loss": 1.0411,
+      "step": 5032
+    },
+    {
+      "epoch": 1.826122448979592,
+      "grad_norm": 0.2080078125,
+      "learning_rate": 7.925590502998409e-05,
+      "loss": 0.9887,
+      "step": 5034
+    },
+    {
+      "epoch": 1.8268480725623584,
+      "grad_norm": 0.15625,
+      "learning_rate": 7.920695141353568e-05,
+      "loss": 0.985,
+      "step": 5036
+    },
+    {
+      "epoch": 1.8275736961451248,
+      "grad_norm": 0.1923828125,
+      "learning_rate": 7.915799779708727e-05,
+      "loss": 0.9819,
+      "step": 5038
+    },
+    {
+      "epoch": 1.8282993197278912,
+      "grad_norm": 0.1640625,
+      "learning_rate": 7.910904418063885e-05,
+      "loss": 0.9861,
+      "step": 5040
+    },
+    {
+      "epoch": 1.8290249433106576,
+      "grad_norm": 0.255859375,
+      "learning_rate": 7.906009056419044e-05,
+      "loss": 1.0396,
+      "step": 5042
+    },
+    {
+      "epoch": 1.829750566893424,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 7.901113694774201e-05,
+      "loss": 1.0312,
+      "step": 5044
+    },
+    {
+      "epoch": 1.8304761904761904,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 7.896218333129361e-05,
+      "loss": 1.0328,
+      "step": 5046
+    },
+    {
+      "epoch": 1.8312018140589568,
+      "grad_norm": 0.169921875,
+      "learning_rate": 7.891322971484518e-05,
+      "loss": 0.9466,
+      "step": 5048
+    },
+    {
+      "epoch": 1.8319274376417234,
+      "grad_norm": 0.1796875,
+      "learning_rate": 7.886427609839677e-05,
+      "loss": 0.9832,
+      "step": 5050
+    },
+    {
+      "epoch": 1.8326530612244898,
+      "grad_norm": 0.2060546875,
+      "learning_rate": 7.881532248194837e-05,
+      "loss": 0.9828,
+      "step": 5052
+    },
+    {
+      "epoch": 1.8333786848072562,
+      "grad_norm": 0.15234375,
+      "learning_rate": 7.876636886549994e-05,
+      "loss": 0.9619,
+      "step": 5054
+    },
+    {
+      "epoch": 1.8341043083900228,
+      "grad_norm": 0.15625,
+      "learning_rate": 7.871741524905153e-05,
+      "loss": 0.8759,
+      "step": 5056
+    },
+    {
+      "epoch": 1.8348299319727892,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 7.86684616326031e-05,
+      "loss": 1.0589,
+      "step": 5058
+    },
+    {
+      "epoch": 1.8355555555555556,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 7.86195080161547e-05,
+      "loss": 1.0069,
+      "step": 5060
+    },
+    {
+      "epoch": 1.836281179138322,
+      "grad_norm": 0.17578125,
+      "learning_rate": 7.857055439970629e-05,
+      "loss": 0.9726,
+      "step": 5062
+    },
+    {
+      "epoch": 1.8370068027210884,
+      "grad_norm": 0.173828125,
+      "learning_rate": 7.852160078325786e-05,
+      "loss": 0.9907,
+      "step": 5064
+    },
+    {
+      "epoch": 1.8377324263038548,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 7.847264716680946e-05,
+      "loss": 0.9354,
+      "step": 5066
+    },
+    {
+      "epoch": 1.8384580498866212,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 7.842369355036103e-05,
+      "loss": 0.9583,
+      "step": 5068
+    },
+    {
+      "epoch": 1.8391836734693876,
+      "grad_norm": 0.15234375,
+      "learning_rate": 7.837473993391262e-05,
+      "loss": 0.9899,
+      "step": 5070
+    },
+    {
+      "epoch": 1.8399092970521542,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 7.832578631746422e-05,
+      "loss": 1.0061,
+      "step": 5072
+    },
+    {
+      "epoch": 1.8406349206349206,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 7.827683270101579e-05,
+      "loss": 0.8678,
+      "step": 5074
+    },
+    {
+      "epoch": 1.841360544217687,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 7.822787908456738e-05,
+      "loss": 0.9876,
+      "step": 5076
+    },
+    {
+      "epoch": 1.8420861678004536,
+      "grad_norm": 0.158203125,
+      "learning_rate": 7.817892546811895e-05,
+      "loss": 0.9555,
+      "step": 5078
+    },
+    {
+      "epoch": 1.84281179138322,
+      "grad_norm": 0.169921875,
+      "learning_rate": 7.812997185167055e-05,
+      "loss": 1.0482,
+      "step": 5080
+    },
+    {
+      "epoch": 1.8435374149659864,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 7.808101823522213e-05,
+      "loss": 0.9854,
+      "step": 5082
+    },
+    {
+      "epoch": 1.8442630385487528,
+      "grad_norm": 0.173828125,
+      "learning_rate": 7.803206461877371e-05,
+      "loss": 0.9896,
+      "step": 5084
+    },
+    {
+      "epoch": 1.8449886621315192,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 7.798311100232531e-05,
+      "loss": 0.9083,
+      "step": 5086
+    },
+    {
+      "epoch": 1.8457142857142856,
+      "grad_norm": 0.1875,
+      "learning_rate": 7.793415738587688e-05,
+      "loss": 0.9821,
+      "step": 5088
+    },
+    {
+      "epoch": 1.846439909297052,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 7.788520376942847e-05,
+      "loss": 1.0252,
+      "step": 5090
+    },
+    {
+      "epoch": 1.8471655328798184,
+      "grad_norm": 0.1640625,
+      "learning_rate": 7.783625015298005e-05,
+      "loss": 1.0168,
+      "step": 5092
+    },
+    {
+      "epoch": 1.847891156462585,
+      "grad_norm": 0.18359375,
+      "learning_rate": 7.778729653653164e-05,
+      "loss": 1.0525,
+      "step": 5094
+    },
+    {
+      "epoch": 1.8486167800453515,
+      "grad_norm": 0.16796875,
+      "learning_rate": 7.773834292008323e-05,
+      "loss": 0.946,
+      "step": 5096
+    },
+    {
+      "epoch": 1.8493424036281179,
+      "grad_norm": 0.2001953125,
+      "learning_rate": 7.76893893036348e-05,
+      "loss": 0.9013,
+      "step": 5098
+    },
+    {
+      "epoch": 1.8500680272108845,
+      "grad_norm": 0.1875,
+      "learning_rate": 7.76404356871864e-05,
+      "loss": 0.9918,
+      "step": 5100
+    },
+    {
+      "epoch": 1.8500680272108845,
+      "eval_loss": 1.0116716623306274,
+      "eval_runtime": 85.8661,
+      "eval_samples_per_second": 20.963,
+      "eval_steps_per_second": 0.664,
+      "step": 5100
+    },
+    {
+      "epoch": 1.8507936507936509,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 7.759148207073798e-05,
+      "loss": 0.988,
+      "step": 5102
+    },
+    {
+      "epoch": 1.8515192743764173,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 7.754252845428956e-05,
+      "loss": 0.9003,
+      "step": 5104
+    },
+    {
+      "epoch": 1.8522448979591837,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 7.749357483784116e-05,
+      "loss": 0.9411,
+      "step": 5106
+    },
+    {
+      "epoch": 1.85297052154195,
+      "grad_norm": 0.16015625,
+      "learning_rate": 7.744462122139273e-05,
+      "loss": 0.9319,
+      "step": 5108
+    },
+    {
+      "epoch": 1.8536961451247165,
+      "grad_norm": 0.189453125,
+      "learning_rate": 7.739566760494432e-05,
+      "loss": 1.046,
+      "step": 5110
+    },
+    {
+      "epoch": 1.8544217687074829,
+      "grad_norm": 0.14453125,
+      "learning_rate": 7.73467139884959e-05,
+      "loss": 0.8999,
+      "step": 5112
+    },
+    {
+      "epoch": 1.8551473922902493,
+      "grad_norm": 0.169921875,
+      "learning_rate": 7.729776037204749e-05,
+      "loss": 1.0543,
+      "step": 5114
+    },
+    {
+      "epoch": 1.855873015873016,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 7.724880675559908e-05,
+      "loss": 1.0118,
+      "step": 5116
+    },
+    {
+      "epoch": 1.8565986394557823,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 7.719985313915065e-05,
+      "loss": 1.0171,
+      "step": 5118
+    },
+    {
+      "epoch": 1.857324263038549,
+      "grad_norm": 0.158203125,
+      "learning_rate": 7.715089952270225e-05,
+      "loss": 0.9521,
+      "step": 5120
+    },
+    {
+      "epoch": 1.8580498866213153,
+      "grad_norm": 0.154296875,
+      "learning_rate": 7.710194590625383e-05,
+      "loss": 0.9663,
+      "step": 5122
+    },
+    {
+      "epoch": 1.8587755102040817,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 7.705299228980541e-05,
+      "loss": 0.9721,
+      "step": 5124
+    },
+    {
+      "epoch": 1.8595011337868481,
+      "grad_norm": 0.17578125,
+      "learning_rate": 7.7004038673357e-05,
+      "loss": 1.0055,
+      "step": 5126
+    },
+    {
+      "epoch": 1.8602267573696145,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 7.695508505690858e-05,
+      "loss": 0.9677,
+      "step": 5128
+    },
+    {
+      "epoch": 1.860952380952381,
+      "grad_norm": 0.16796875,
+      "learning_rate": 7.690613144046017e-05,
+      "loss": 0.9718,
+      "step": 5130
+    },
+    {
+      "epoch": 1.8616780045351473,
+      "grad_norm": 0.1865234375,
+      "learning_rate": 7.685717782401175e-05,
+      "loss": 0.9995,
+      "step": 5132
+    },
+    {
+      "epoch": 1.8624036281179137,
+      "grad_norm": 0.154296875,
+      "learning_rate": 7.680822420756334e-05,
+      "loss": 0.9169,
+      "step": 5134
+    },
+    {
+      "epoch": 1.8631292517006801,
+      "grad_norm": 0.166015625,
+      "learning_rate": 7.675927059111492e-05,
+      "loss": 0.9085,
+      "step": 5136
+    },
+    {
+      "epoch": 1.8638548752834467,
+      "grad_norm": 0.1796875,
+      "learning_rate": 7.67103169746665e-05,
+      "loss": 0.9904,
+      "step": 5138
+    },
+    {
+      "epoch": 1.8645804988662131,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 7.66613633582181e-05,
+      "loss": 1.0268,
+      "step": 5140
+    },
+    {
+      "epoch": 1.8653061224489798,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 7.661240974176968e-05,
+      "loss": 1.0152,
+      "step": 5142
+    },
+    {
+      "epoch": 1.8660317460317462,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 7.656345612532126e-05,
+      "loss": 1.0148,
+      "step": 5144
+    },
+    {
+      "epoch": 1.8667573696145126,
+      "grad_norm": 0.16015625,
+      "learning_rate": 7.651450250887284e-05,
+      "loss": 0.9883,
+      "step": 5146
+    },
+    {
+      "epoch": 1.867482993197279,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 7.646554889242443e-05,
+      "loss": 1.0816,
+      "step": 5148
+    },
+    {
+      "epoch": 1.8682086167800454,
+      "grad_norm": 0.17578125,
+      "learning_rate": 7.641659527597602e-05,
+      "loss": 0.9106,
+      "step": 5150
+    },
+    {
+      "epoch": 1.8689342403628117,
+      "grad_norm": 0.177734375,
+      "learning_rate": 7.63676416595276e-05,
+      "loss": 0.9509,
+      "step": 5152
+    },
+    {
+      "epoch": 1.8696598639455781,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 7.631868804307919e-05,
+      "loss": 0.8867,
+      "step": 5154
+    },
+    {
+      "epoch": 1.8703854875283445,
+      "grad_norm": 0.1953125,
+      "learning_rate": 7.626973442663077e-05,
+      "loss": 1.0658,
+      "step": 5156
+    },
+    {
+      "epoch": 1.871111111111111,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 7.622078081018235e-05,
+      "loss": 0.9254,
+      "step": 5158
+    },
+    {
+      "epoch": 1.8718367346938776,
+      "grad_norm": 0.15234375,
+      "learning_rate": 7.617182719373393e-05,
+      "loss": 0.9928,
+      "step": 5160
+    },
+    {
+      "epoch": 1.872562358276644,
+      "grad_norm": 0.16796875,
+      "learning_rate": 7.612287357728553e-05,
+      "loss": 0.9523,
+      "step": 5162
+    },
+    {
+      "epoch": 1.8732879818594106,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 7.607391996083711e-05,
+      "loss": 1.0135,
+      "step": 5164
+    },
+    {
+      "epoch": 1.874013605442177,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 7.60249663443887e-05,
+      "loss": 1.0446,
+      "step": 5166
+    },
+    {
+      "epoch": 1.8747392290249434,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 7.597601272794028e-05,
+      "loss": 0.9577,
+      "step": 5168
+    },
+    {
+      "epoch": 1.8754648526077098,
+      "grad_norm": 0.16796875,
+      "learning_rate": 7.592705911149186e-05,
+      "loss": 1.0376,
+      "step": 5170
+    },
+    {
+      "epoch": 1.8761904761904762,
+      "grad_norm": 0.15625,
+      "learning_rate": 7.587810549504345e-05,
+      "loss": 0.9016,
+      "step": 5172
+    },
+    {
+      "epoch": 1.8769160997732426,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 7.582915187859504e-05,
+      "loss": 0.9314,
+      "step": 5174
+    },
+    {
+      "epoch": 1.877641723356009,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 7.578019826214662e-05,
+      "loss": 0.9535,
+      "step": 5176
+    },
+    {
+      "epoch": 1.8783673469387754,
+      "grad_norm": 0.166015625,
+      "learning_rate": 7.57312446456982e-05,
+      "loss": 1.0041,
+      "step": 5178
+    },
+    {
+      "epoch": 1.879092970521542,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 7.568229102924978e-05,
+      "loss": 0.9118,
+      "step": 5180
+    },
+    {
+      "epoch": 1.8798185941043084,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 7.563333741280138e-05,
+      "loss": 0.9734,
+      "step": 5182
+    },
+    {
+      "epoch": 1.8805442176870748,
+      "grad_norm": 0.162109375,
+      "learning_rate": 7.558438379635296e-05,
+      "loss": 0.941,
+      "step": 5184
+    },
+    {
+      "epoch": 1.8812698412698414,
+      "grad_norm": 0.173828125,
+      "learning_rate": 7.553543017990454e-05,
+      "loss": 1.0394,
+      "step": 5186
+    },
+    {
+      "epoch": 1.8819954648526078,
+      "grad_norm": 0.154296875,
+      "learning_rate": 7.548647656345613e-05,
+      "loss": 0.9472,
+      "step": 5188
+    },
+    {
+      "epoch": 1.8827210884353742,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 7.543752294700771e-05,
+      "loss": 0.9045,
+      "step": 5190
+    },
+    {
+      "epoch": 1.8834467120181406,
+      "grad_norm": 0.18359375,
+      "learning_rate": 7.53885693305593e-05,
+      "loss": 1.0148,
+      "step": 5192
+    },
+    {
+      "epoch": 1.884172335600907,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 7.533961571411087e-05,
+      "loss": 0.9824,
+      "step": 5194
+    },
+    {
+      "epoch": 1.8848979591836734,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 7.529066209766247e-05,
+      "loss": 1.0019,
+      "step": 5196
+    },
+    {
+      "epoch": 1.8856235827664398,
+      "grad_norm": 0.216796875,
+      "learning_rate": 7.524170848121405e-05,
+      "loss": 1.0268,
+      "step": 5198
+    },
+    {
+      "epoch": 1.8863492063492062,
+      "grad_norm": 0.166015625,
+      "learning_rate": 7.519275486476563e-05,
+      "loss": 1.0249,
+      "step": 5200
+    },
+    {
+      "epoch": 1.8863492063492062,
+      "eval_loss": 1.011323094367981,
+      "eval_runtime": 85.9575,
+      "eval_samples_per_second": 20.941,
+      "eval_steps_per_second": 0.663,
+      "step": 5200
+    },
+    {
+      "epoch": 1.8870748299319728,
+      "grad_norm": 0.181640625,
+      "learning_rate": 7.514380124831723e-05,
+      "loss": 0.9658,
+      "step": 5202
+    },
+    {
+      "epoch": 1.8878004535147392,
+      "grad_norm": 0.16796875,
+      "learning_rate": 7.50948476318688e-05,
+      "loss": 1.0172,
+      "step": 5204
+    },
+    {
+      "epoch": 1.8885260770975056,
+      "grad_norm": 0.1875,
+      "learning_rate": 7.50458940154204e-05,
+      "loss": 0.9279,
+      "step": 5206
+    },
+    {
+      "epoch": 1.8892517006802723,
+      "grad_norm": 0.162109375,
+      "learning_rate": 7.499694039897198e-05,
+      "loss": 0.9781,
+      "step": 5208
+    },
+    {
+      "epoch": 1.8899773242630387,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 7.494798678252356e-05,
+      "loss": 1.0788,
+      "step": 5210
+    },
+    {
+      "epoch": 1.890702947845805,
+      "grad_norm": 0.13671875,
+      "learning_rate": 7.489903316607515e-05,
+      "loss": 0.9595,
+      "step": 5212
+    },
+    {
+      "epoch": 1.8914285714285715,
+      "grad_norm": 0.158203125,
+      "learning_rate": 7.485007954962672e-05,
+      "loss": 0.9927,
+      "step": 5214
+    },
+    {
+      "epoch": 1.8921541950113379,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 7.480112593317832e-05,
+      "loss": 1.0254,
+      "step": 5216
+    },
+    {
+      "epoch": 1.8928798185941043,
+      "grad_norm": 0.1923828125,
+      "learning_rate": 7.47521723167299e-05,
+      "loss": 1.0419,
+      "step": 5218
+    },
+    {
+      "epoch": 1.8936054421768707,
+      "grad_norm": 0.205078125,
+      "learning_rate": 7.470321870028148e-05,
+      "loss": 0.9921,
+      "step": 5220
+    },
+    {
+      "epoch": 1.894331065759637,
+      "grad_norm": 0.17578125,
+      "learning_rate": 7.465426508383308e-05,
+      "loss": 0.9195,
+      "step": 5222
+    },
+    {
+      "epoch": 1.8950566893424037,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 7.460531146738465e-05,
+      "loss": 1.0073,
+      "step": 5224
+    },
+    {
+      "epoch": 1.89578231292517,
+      "grad_norm": 0.19921875,
+      "learning_rate": 7.455635785093624e-05,
+      "loss": 1.0088,
+      "step": 5226
+    },
+    {
+      "epoch": 1.8965079365079365,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 7.450740423448783e-05,
+      "loss": 0.938,
+      "step": 5228
+    },
+    {
+      "epoch": 1.897233560090703,
+      "grad_norm": 0.146484375,
+      "learning_rate": 7.445845061803941e-05,
+      "loss": 0.9745,
+      "step": 5230
+    },
+    {
+      "epoch": 1.8979591836734695,
+      "grad_norm": 0.16796875,
+      "learning_rate": 7.4409497001591e-05,
+      "loss": 0.9514,
+      "step": 5232
+    },
+    {
+      "epoch": 1.898684807256236,
+      "grad_norm": 0.173828125,
+      "learning_rate": 7.436054338514257e-05,
+      "loss": 0.9912,
+      "step": 5234
+    },
+    {
+      "epoch": 1.8994104308390023,
+      "grad_norm": 0.1640625,
+      "learning_rate": 7.431158976869417e-05,
+      "loss": 0.9794,
+      "step": 5236
+    },
+    {
+      "epoch": 1.9001360544217687,
+      "grad_norm": 0.2099609375,
+      "learning_rate": 7.426263615224575e-05,
+      "loss": 1.1444,
+      "step": 5238
+    },
+    {
+      "epoch": 1.900861678004535,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 7.421368253579733e-05,
+      "loss": 0.9846,
+      "step": 5240
+    },
+    {
+      "epoch": 1.9015873015873015,
+      "grad_norm": 0.18359375,
+      "learning_rate": 7.416472891934893e-05,
+      "loss": 0.9898,
+      "step": 5242
+    },
+    {
+      "epoch": 1.902312925170068,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 7.41157753029005e-05,
+      "loss": 0.9763,
+      "step": 5244
+    },
+    {
+      "epoch": 1.9030385487528345,
+      "grad_norm": 0.171875,
+      "learning_rate": 7.40668216864521e-05,
+      "loss": 0.9776,
+      "step": 5246
+    },
+    {
+      "epoch": 1.903764172335601,
+      "grad_norm": 0.17578125,
+      "learning_rate": 7.401786807000368e-05,
+      "loss": 0.9866,
+      "step": 5248
+    },
+    {
+      "epoch": 1.9044897959183673,
+      "grad_norm": 0.166015625,
+      "learning_rate": 7.396891445355526e-05,
+      "loss": 0.9892,
+      "step": 5250
+    },
+    {
+      "epoch": 1.905215419501134,
+      "grad_norm": 0.15625,
+      "learning_rate": 7.391996083710685e-05,
+      "loss": 0.9734,
+      "step": 5252
+    },
+    {
+      "epoch": 1.9059410430839003,
+      "grad_norm": 0.1953125,
+      "learning_rate": 7.387100722065842e-05,
+      "loss": 1.0002,
+      "step": 5254
+    },
+    {
+      "epoch": 1.9066666666666667,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 7.382205360421002e-05,
+      "loss": 0.9108,
+      "step": 5256
+    },
+    {
+      "epoch": 1.9073922902494331,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 7.37730999877616e-05,
+      "loss": 0.9652,
+      "step": 5258
+    },
+    {
+      "epoch": 1.9081179138321995,
+      "grad_norm": 0.2041015625,
+      "learning_rate": 7.372414637131318e-05,
+      "loss": 0.9549,
+      "step": 5260
+    },
+    {
+      "epoch": 1.908843537414966,
+      "grad_norm": 0.16796875,
+      "learning_rate": 7.367519275486478e-05,
+      "loss": 1.0341,
+      "step": 5262
+    },
+    {
+      "epoch": 1.9095691609977323,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 7.362623913841635e-05,
+      "loss": 0.9873,
+      "step": 5264
+    },
+    {
+      "epoch": 1.9102947845804987,
+      "grad_norm": 0.173828125,
+      "learning_rate": 7.357728552196794e-05,
+      "loss": 0.9916,
+      "step": 5266
+    },
+    {
+      "epoch": 1.9110204081632653,
+      "grad_norm": 0.162109375,
+      "learning_rate": 7.352833190551953e-05,
+      "loss": 0.96,
+      "step": 5268
+    },
+    {
+      "epoch": 1.9117460317460317,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 7.347937828907111e-05,
+      "loss": 0.9318,
+      "step": 5270
+    },
+    {
+      "epoch": 1.9124716553287981,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 7.343042467262269e-05,
+      "loss": 0.9651,
+      "step": 5272
+    },
+    {
+      "epoch": 1.9131972789115648,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 7.338147105617427e-05,
+      "loss": 0.9995,
+      "step": 5274
+    },
+    {
+      "epoch": 1.9139229024943312,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 7.333251743972587e-05,
+      "loss": 1.0009,
+      "step": 5276
+    },
+    {
+      "epoch": 1.9146485260770976,
+      "grad_norm": 0.171875,
+      "learning_rate": 7.328356382327745e-05,
+      "loss": 1.0127,
+      "step": 5278
+    },
+    {
+      "epoch": 1.915374149659864,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 7.323461020682903e-05,
+      "loss": 0.9979,
+      "step": 5280
+    },
+    {
+      "epoch": 1.9160997732426304,
+      "grad_norm": 0.2001953125,
+      "learning_rate": 7.318565659038062e-05,
+      "loss": 1.057,
+      "step": 5282
+    },
+    {
+      "epoch": 1.9168253968253968,
+      "grad_norm": 0.16015625,
+      "learning_rate": 7.31367029739322e-05,
+      "loss": 1.0084,
+      "step": 5284
+    },
+    {
+      "epoch": 1.9175510204081632,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 7.308774935748379e-05,
+      "loss": 1.0373,
+      "step": 5286
+    },
+    {
+      "epoch": 1.9182766439909296,
+      "grad_norm": 0.1923828125,
+      "learning_rate": 7.303879574103538e-05,
+      "loss": 1.0019,
+      "step": 5288
+    },
+    {
+      "epoch": 1.9190022675736962,
+      "grad_norm": 0.18359375,
+      "learning_rate": 7.298984212458696e-05,
+      "loss": 0.9535,
+      "step": 5290
+    },
+    {
+      "epoch": 1.9197278911564626,
+      "grad_norm": 0.169921875,
+      "learning_rate": 7.294088850813854e-05,
+      "loss": 0.9318,
+      "step": 5292
+    },
+    {
+      "epoch": 1.920453514739229,
+      "grad_norm": 0.1484375,
+      "learning_rate": 7.289193489169012e-05,
+      "loss": 0.9384,
+      "step": 5294
+    },
+    {
+      "epoch": 1.9211791383219956,
+      "grad_norm": 0.177734375,
+      "learning_rate": 7.284298127524172e-05,
+      "loss": 0.9661,
+      "step": 5296
+    },
+    {
+      "epoch": 1.921904761904762,
+      "grad_norm": 0.166015625,
+      "learning_rate": 7.27940276587933e-05,
+      "loss": 0.9253,
+      "step": 5298
+    },
+    {
+      "epoch": 1.9226303854875284,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 7.274507404234488e-05,
+      "loss": 0.9973,
+      "step": 5300
+    },
+    {
+      "epoch": 1.9226303854875284,
+      "eval_loss": 1.0112332105636597,
+      "eval_runtime": 86.071,
+      "eval_samples_per_second": 20.913,
+      "eval_steps_per_second": 0.662,
+      "step": 5300
+    },
+    {
+      "epoch": 1.9233560090702948,
+      "grad_norm": 0.1796875,
+      "learning_rate": 7.269612042589646e-05,
+      "loss": 1.0532,
+      "step": 5302
+    },
+    {
+      "epoch": 1.9240816326530612,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 7.264716680944805e-05,
+      "loss": 0.9237,
+      "step": 5304
+    },
+    {
+      "epoch": 1.9248072562358276,
+      "grad_norm": 0.15625,
+      "learning_rate": 7.259821319299963e-05,
+      "loss": 0.9588,
+      "step": 5306
+    },
+    {
+      "epoch": 1.925532879818594,
+      "grad_norm": 0.16796875,
+      "learning_rate": 7.254925957655123e-05,
+      "loss": 1.0311,
+      "step": 5308
+    },
+    {
+      "epoch": 1.9262585034013604,
+      "grad_norm": 0.15625,
+      "learning_rate": 7.250030596010281e-05,
+      "loss": 0.901,
+      "step": 5310
+    },
+    {
+      "epoch": 1.926984126984127,
+      "grad_norm": 0.18359375,
+      "learning_rate": 7.245135234365439e-05,
+      "loss": 1.0205,
+      "step": 5312
+    },
+    {
+      "epoch": 1.9277097505668934,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 7.240239872720597e-05,
+      "loss": 0.958,
+      "step": 5314
+    },
+    {
+      "epoch": 1.9284353741496598,
+      "grad_norm": 0.17578125,
+      "learning_rate": 7.235344511075755e-05,
+      "loss": 0.996,
+      "step": 5316
+    },
+    {
+      "epoch": 1.9291609977324264,
+      "grad_norm": 0.162109375,
+      "learning_rate": 7.230449149430915e-05,
+      "loss": 0.9822,
+      "step": 5318
+    },
+    {
+      "epoch": 1.9298866213151928,
+      "grad_norm": 0.1484375,
+      "learning_rate": 7.225553787786073e-05,
+      "loss": 0.9584,
+      "step": 5320
+    },
+    {
+      "epoch": 1.9306122448979592,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 7.220658426141231e-05,
+      "loss": 0.9161,
+      "step": 5322
+    },
+    {
+      "epoch": 1.9313378684807256,
+      "grad_norm": 0.158203125,
+      "learning_rate": 7.21576306449639e-05,
+      "loss": 0.9766,
+      "step": 5324
+    },
+    {
+      "epoch": 1.932063492063492,
+      "grad_norm": 0.166015625,
+      "learning_rate": 7.210867702851548e-05,
+      "loss": 1.0074,
+      "step": 5326
+    },
+    {
+      "epoch": 1.9327891156462584,
+      "grad_norm": 0.166015625,
+      "learning_rate": 7.205972341206708e-05,
+      "loss": 0.9183,
+      "step": 5328
+    },
+    {
+      "epoch": 1.9335147392290248,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 7.201076979561866e-05,
+      "loss": 1.01,
+      "step": 5330
+    },
+    {
+      "epoch": 1.9342403628117912,
+      "grad_norm": 0.18359375,
+      "learning_rate": 7.196181617917024e-05,
+      "loss": 0.969,
+      "step": 5332
+    },
+    {
+      "epoch": 1.9349659863945579,
+      "grad_norm": 0.177734375,
+      "learning_rate": 7.191286256272182e-05,
+      "loss": 1.0358,
+      "step": 5334
+    },
+    {
+      "epoch": 1.9356916099773243,
+      "grad_norm": 0.1865234375,
+      "learning_rate": 7.18639089462734e-05,
+      "loss": 1.0049,
+      "step": 5336
+    },
+    {
+      "epoch": 1.9364172335600907,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 7.1814955329825e-05,
+      "loss": 0.9693,
+      "step": 5338
+    },
+    {
+      "epoch": 1.9371428571428573,
+      "grad_norm": 0.14453125,
+      "learning_rate": 7.176600171337657e-05,
+      "loss": 0.9089,
+      "step": 5340
+    },
+    {
+      "epoch": 1.9378684807256237,
+      "grad_norm": 0.18359375,
+      "learning_rate": 7.171704809692816e-05,
+      "loss": 0.9832,
+      "step": 5342
+    },
+    {
+      "epoch": 1.93859410430839,
+      "grad_norm": 0.162109375,
+      "learning_rate": 7.166809448047975e-05,
+      "loss": 0.9502,
+      "step": 5344
+    },
+    {
+      "epoch": 1.9393197278911565,
+      "grad_norm": 0.15625,
+      "learning_rate": 7.161914086403133e-05,
+      "loss": 0.9835,
+      "step": 5346
+    },
+    {
+      "epoch": 1.9400453514739229,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 7.157018724758293e-05,
+      "loss": 0.9826,
+      "step": 5348
+    },
+    {
+      "epoch": 1.9407709750566893,
+      "grad_norm": 0.166015625,
+      "learning_rate": 7.15212336311345e-05,
+      "loss": 0.9625,
+      "step": 5350
+    },
+    {
+      "epoch": 1.9414965986394557,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 7.147228001468609e-05,
+      "loss": 0.9491,
+      "step": 5352
+    },
+    {
+      "epoch": 1.942222222222222,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 7.142332639823767e-05,
+      "loss": 1.0431,
+      "step": 5354
+    },
+    {
+      "epoch": 1.9429478458049887,
+      "grad_norm": 0.15625,
+      "learning_rate": 7.137437278178925e-05,
+      "loss": 1.0504,
+      "step": 5356
+    },
+    {
+      "epoch": 1.943673469387755,
+      "grad_norm": 0.1865234375,
+      "learning_rate": 7.132541916534085e-05,
+      "loss": 1.0378,
+      "step": 5358
+    },
+    {
+      "epoch": 1.9443990929705215,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 7.127646554889242e-05,
+      "loss": 0.949,
+      "step": 5360
+    },
+    {
+      "epoch": 1.9451247165532881,
+      "grad_norm": 0.1640625,
+      "learning_rate": 7.122751193244401e-05,
+      "loss": 0.9973,
+      "step": 5362
+    },
+    {
+      "epoch": 1.9458503401360545,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 7.11785583159956e-05,
+      "loss": 0.996,
+      "step": 5364
+    },
+    {
+      "epoch": 1.946575963718821,
+      "grad_norm": 0.171875,
+      "learning_rate": 7.112960469954718e-05,
+      "loss": 1.0218,
+      "step": 5366
+    },
+    {
+      "epoch": 1.9473015873015873,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 7.108065108309878e-05,
+      "loss": 1.0081,
+      "step": 5368
+    },
+    {
+      "epoch": 1.9480272108843537,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 7.103169746665034e-05,
+      "loss": 0.9438,
+      "step": 5370
+    },
+    {
+      "epoch": 1.94875283446712,
+      "grad_norm": 0.2109375,
+      "learning_rate": 7.098274385020194e-05,
+      "loss": 0.9731,
+      "step": 5372
+    },
+    {
+      "epoch": 1.9494784580498865,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 7.093379023375352e-05,
+      "loss": 0.9133,
+      "step": 5374
+    },
+    {
+      "epoch": 1.950204081632653,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 7.08848366173051e-05,
+      "loss": 0.946,
+      "step": 5376
+    },
+    {
+      "epoch": 1.9509297052154195,
+      "grad_norm": 0.19140625,
+      "learning_rate": 7.08358830008567e-05,
+      "loss": 0.9632,
+      "step": 5378
+    },
+    {
+      "epoch": 1.951655328798186,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 7.078692938440827e-05,
+      "loss": 0.9542,
+      "step": 5380
+    },
+    {
+      "epoch": 1.9523809523809523,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 7.073797576795986e-05,
+      "loss": 0.9328,
+      "step": 5382
+    },
+    {
+      "epoch": 1.953106575963719,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 7.068902215151145e-05,
+      "loss": 0.9566,
+      "step": 5384
+    },
+    {
+      "epoch": 1.9538321995464853,
+      "grad_norm": 0.1865234375,
+      "learning_rate": 7.064006853506303e-05,
+      "loss": 0.9863,
+      "step": 5386
+    },
+    {
+      "epoch": 1.9545578231292517,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 7.059111491861462e-05,
+      "loss": 0.9328,
+      "step": 5388
+    },
+    {
+      "epoch": 1.9552834467120181,
+      "grad_norm": 0.169921875,
+      "learning_rate": 7.05421613021662e-05,
+      "loss": 0.9978,
+      "step": 5390
+    },
+    {
+      "epoch": 1.9560090702947845,
+      "grad_norm": 0.177734375,
+      "learning_rate": 7.049320768571779e-05,
+      "loss": 1.0063,
+      "step": 5392
+    },
+    {
+      "epoch": 1.956734693877551,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 7.044425406926937e-05,
+      "loss": 0.9898,
+      "step": 5394
+    },
+    {
+      "epoch": 1.9574603174603173,
+      "grad_norm": 0.166015625,
+      "learning_rate": 7.039530045282095e-05,
+      "loss": 0.9487,
+      "step": 5396
+    },
+    {
+      "epoch": 1.9581859410430837,
+      "grad_norm": 0.162109375,
+      "learning_rate": 7.034634683637255e-05,
+      "loss": 0.9949,
+      "step": 5398
+    },
+    {
+      "epoch": 1.9589115646258504,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 7.029739321992412e-05,
+      "loss": 0.9803,
+      "step": 5400
+    },
+    {
+      "epoch": 1.9589115646258504,
+      "eval_loss": 1.011096715927124,
+      "eval_runtime": 82.3524,
+      "eval_samples_per_second": 21.857,
+      "eval_steps_per_second": 0.692,
+      "step": 5400
+    },
+    {
+      "epoch": 1.9596371882086168,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 7.024843960347571e-05,
+      "loss": 0.9387,
+      "step": 5402
+    },
+    {
+      "epoch": 1.9603628117913834,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 7.01994859870273e-05,
+      "loss": 0.9136,
+      "step": 5404
+    },
+    {
+      "epoch": 1.9610884353741498,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 7.015053237057888e-05,
+      "loss": 0.9976,
+      "step": 5406
+    },
+    {
+      "epoch": 1.9618140589569162,
+      "grad_norm": 0.150390625,
+      "learning_rate": 7.010157875413047e-05,
+      "loss": 0.9514,
+      "step": 5408
+    },
+    {
+      "epoch": 1.9625396825396826,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 7.005262513768204e-05,
+      "loss": 0.9502,
+      "step": 5410
+    },
+    {
+      "epoch": 1.963265306122449,
+      "grad_norm": 0.16015625,
+      "learning_rate": 7.000367152123364e-05,
+      "loss": 0.9255,
+      "step": 5412
+    },
+    {
+      "epoch": 1.9639909297052154,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 6.995471790478522e-05,
+      "loss": 0.9789,
+      "step": 5414
+    },
+    {
+      "epoch": 1.9647165532879818,
+      "grad_norm": 0.15234375,
+      "learning_rate": 6.99057642883368e-05,
+      "loss": 0.9001,
+      "step": 5416
+    },
+    {
+      "epoch": 1.9654421768707482,
+      "grad_norm": 0.150390625,
+      "learning_rate": 6.985681067188839e-05,
+      "loss": 1.0297,
+      "step": 5418
+    },
+    {
+      "epoch": 1.9661678004535146,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 6.980785705543997e-05,
+      "loss": 0.9494,
+      "step": 5420
+    },
+    {
+      "epoch": 1.9668934240362812,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 6.975890343899156e-05,
+      "loss": 0.9232,
+      "step": 5422
+    },
+    {
+      "epoch": 1.9676190476190476,
+      "grad_norm": 0.158203125,
+      "learning_rate": 6.970994982254315e-05,
+      "loss": 0.9913,
+      "step": 5424
+    },
+    {
+      "epoch": 1.9683446712018142,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 6.966099620609473e-05,
+      "loss": 0.9597,
+      "step": 5426
+    },
+    {
+      "epoch": 1.9690702947845806,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 6.961204258964631e-05,
+      "loss": 0.9634,
+      "step": 5428
+    },
+    {
+      "epoch": 1.969795918367347,
+      "grad_norm": 0.1943359375,
+      "learning_rate": 6.95630889731979e-05,
+      "loss": 1.027,
+      "step": 5430
+    },
+    {
+      "epoch": 1.9705215419501134,
+      "grad_norm": 0.17578125,
+      "learning_rate": 6.951413535674949e-05,
+      "loss": 0.886,
+      "step": 5432
+    },
+    {
+      "epoch": 1.9712471655328798,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 6.946518174030107e-05,
+      "loss": 1.0277,
+      "step": 5434
+    },
+    {
+      "epoch": 1.9719727891156462,
+      "grad_norm": 0.166015625,
+      "learning_rate": 6.941622812385265e-05,
+      "loss": 1.018,
+      "step": 5436
+    },
+    {
+      "epoch": 1.9726984126984126,
+      "grad_norm": 0.17578125,
+      "learning_rate": 6.936727450740424e-05,
+      "loss": 0.9642,
+      "step": 5438
+    },
+    {
+      "epoch": 1.973424036281179,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 6.931832089095582e-05,
+      "loss": 0.9943,
+      "step": 5440
+    },
+    {
+      "epoch": 1.9741496598639454,
+      "grad_norm": 0.166015625,
+      "learning_rate": 6.926936727450741e-05,
+      "loss": 0.9534,
+      "step": 5442
+    },
+    {
+      "epoch": 1.974875283446712,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 6.9220413658059e-05,
+      "loss": 0.9578,
+      "step": 5444
+    },
+    {
+      "epoch": 1.9756009070294784,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 6.917146004161058e-05,
+      "loss": 0.9278,
+      "step": 5446
+    },
+    {
+      "epoch": 1.976326530612245,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 6.912250642516216e-05,
+      "loss": 1.0084,
+      "step": 5448
+    },
+    {
+      "epoch": 1.9770521541950115,
+      "grad_norm": 0.171875,
+      "learning_rate": 6.907355280871374e-05,
+      "loss": 0.9462,
+      "step": 5450
+    },
+    {
+      "epoch": 1.9777777777777779,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 6.902459919226533e-05,
+      "loss": 0.9877,
+      "step": 5452
+    },
+    {
+      "epoch": 1.9785034013605443,
+      "grad_norm": 0.169921875,
+      "learning_rate": 6.897564557581692e-05,
+      "loss": 0.9353,
+      "step": 5454
+    },
+    {
+      "epoch": 1.9792290249433107,
+      "grad_norm": 0.15625,
+      "learning_rate": 6.89266919593685e-05,
+      "loss": 1.0031,
+      "step": 5456
+    },
+    {
+      "epoch": 1.979954648526077,
+      "grad_norm": 0.171875,
+      "learning_rate": 6.887773834292009e-05,
+      "loss": 0.9874,
+      "step": 5458
+    },
+    {
+      "epoch": 1.9806802721088435,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 6.882878472647167e-05,
+      "loss": 0.9689,
+      "step": 5460
+    },
+    {
+      "epoch": 1.9814058956916099,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 6.877983111002325e-05,
+      "loss": 0.9945,
+      "step": 5462
+    },
+    {
+      "epoch": 1.9821315192743763,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 6.873087749357485e-05,
+      "loss": 0.9695,
+      "step": 5464
+    },
+    {
+      "epoch": 1.9828571428571429,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 6.868192387712643e-05,
+      "loss": 0.9294,
+      "step": 5466
+    },
+    {
+      "epoch": 1.9835827664399093,
+      "grad_norm": 0.16015625,
+      "learning_rate": 6.863297026067801e-05,
+      "loss": 1.0133,
+      "step": 5468
+    },
+    {
+      "epoch": 1.984308390022676,
+      "grad_norm": 0.171875,
+      "learning_rate": 6.85840166442296e-05,
+      "loss": 0.9074,
+      "step": 5470
+    },
+    {
+      "epoch": 1.9850340136054423,
+      "grad_norm": 0.1982421875,
+      "learning_rate": 6.853506302778118e-05,
+      "loss": 0.9792,
+      "step": 5472
+    },
+    {
+      "epoch": 1.9857596371882087,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 6.848610941133277e-05,
+      "loss": 1.0607,
+      "step": 5474
+    },
+    {
+      "epoch": 1.986485260770975,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 6.843715579488435e-05,
+      "loss": 0.9649,
+      "step": 5476
+    },
+    {
+      "epoch": 1.9872108843537415,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 6.838820217843594e-05,
+      "loss": 0.9928,
+      "step": 5478
+    },
+    {
+      "epoch": 1.987936507936508,
+      "grad_norm": 0.2119140625,
+      "learning_rate": 6.833924856198752e-05,
+      "loss": 0.9035,
+      "step": 5480
+    },
+    {
+      "epoch": 1.9886621315192743,
+      "grad_norm": 0.1875,
+      "learning_rate": 6.82902949455391e-05,
+      "loss": 1.0789,
+      "step": 5482
+    },
+    {
+      "epoch": 1.9893877551020407,
+      "grad_norm": 0.181640625,
+      "learning_rate": 6.82413413290907e-05,
+      "loss": 0.9973,
+      "step": 5484
+    },
+    {
+      "epoch": 1.9901133786848073,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 6.819238771264227e-05,
+      "loss": 1.0186,
+      "step": 5486
+    },
+    {
+      "epoch": 1.9908390022675737,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 6.814343409619386e-05,
+      "loss": 0.9575,
+      "step": 5488
+    },
+    {
+      "epoch": 1.99156462585034,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 6.809448047974544e-05,
+      "loss": 0.9181,
+      "step": 5490
+    },
+    {
+      "epoch": 1.9922902494331067,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 6.804552686329703e-05,
+      "loss": 0.9937,
+      "step": 5492
+    },
+    {
+      "epoch": 1.9930158730158731,
+      "grad_norm": 0.1923828125,
+      "learning_rate": 6.799657324684862e-05,
+      "loss": 0.9643,
+      "step": 5494
+    },
+    {
+      "epoch": 1.9937414965986395,
+      "grad_norm": 0.154296875,
+      "learning_rate": 6.794761963040019e-05,
+      "loss": 0.9727,
+      "step": 5496
+    },
+    {
+      "epoch": 1.994467120181406,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 6.789866601395179e-05,
+      "loss": 0.9577,
+      "step": 5498
+    },
+    {
+      "epoch": 1.9951927437641723,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 6.784971239750337e-05,
+      "loss": 0.9759,
+      "step": 5500
+    },
+    {
+      "epoch": 1.9951927437641723,
+      "eval_loss": 1.0108295679092407,
+      "eval_runtime": 88.2979,
+      "eval_samples_per_second": 20.386,
+      "eval_steps_per_second": 0.646,
+      "step": 5500
+    },
+    {
+      "epoch": 1.9959183673469387,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 6.780075878105495e-05,
+      "loss": 1.0464,
+      "step": 5502
+    },
+    {
+      "epoch": 1.9966439909297051,
+      "grad_norm": 0.177734375,
+      "learning_rate": 6.775180516460655e-05,
+      "loss": 1.0008,
+      "step": 5504
+    },
+    {
+      "epoch": 1.9973696145124715,
+      "grad_norm": 0.19140625,
+      "learning_rate": 6.770285154815812e-05,
+      "loss": 0.9546,
+      "step": 5506
+    },
+    {
+      "epoch": 1.9980952380952381,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 6.765389793170971e-05,
+      "loss": 0.9799,
+      "step": 5508
+    },
+    {
+      "epoch": 1.9988208616780045,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 6.760494431526129e-05,
+      "loss": 1.0274,
+      "step": 5510
+    },
+    {
+      "epoch": 1.999546485260771,
+      "grad_norm": 0.16015625,
+      "learning_rate": 6.755599069881288e-05,
+      "loss": 0.9454,
+      "step": 5512
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.3046875,
+      "learning_rate": 6.750703708236447e-05,
+      "loss": 0.9132,
+      "step": 5514
+    },
+    {
+      "epoch": 2.0007256235827664,
+      "grad_norm": 0.19140625,
+      "learning_rate": 6.745808346591604e-05,
+      "loss": 0.9337,
+      "step": 5516
+    },
+    {
+      "epoch": 2.001451247165533,
+      "grad_norm": 0.150390625,
+      "learning_rate": 6.740912984946764e-05,
+      "loss": 0.9514,
+      "step": 5518
+    },
+    {
+      "epoch": 2.002176870748299,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 6.736017623301922e-05,
+      "loss": 1.0014,
+      "step": 5520
+    },
+    {
+      "epoch": 2.0029024943310656,
+      "grad_norm": 0.1796875,
+      "learning_rate": 6.73112226165708e-05,
+      "loss": 0.9509,
+      "step": 5522
+    },
+    {
+      "epoch": 2.003628117913832,
+      "grad_norm": 0.154296875,
+      "learning_rate": 6.72622690001224e-05,
+      "loss": 1.0404,
+      "step": 5524
+    },
+    {
+      "epoch": 2.004353741496599,
+      "grad_norm": 0.16796875,
+      "learning_rate": 6.721331538367396e-05,
+      "loss": 0.9281,
+      "step": 5526
+    },
+    {
+      "epoch": 2.0050793650793652,
+      "grad_norm": 0.16796875,
+      "learning_rate": 6.716436176722556e-05,
+      "loss": 0.9993,
+      "step": 5528
+    },
+    {
+      "epoch": 2.0058049886621316,
+      "grad_norm": 0.162109375,
+      "learning_rate": 6.711540815077714e-05,
+      "loss": 1.0137,
+      "step": 5530
+    },
+    {
+      "epoch": 2.006530612244898,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 6.706645453432873e-05,
+      "loss": 0.9299,
+      "step": 5532
+    },
+    {
+      "epoch": 2.0072562358276644,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 6.701750091788032e-05,
+      "loss": 0.9612,
+      "step": 5534
+    },
+    {
+      "epoch": 2.007981859410431,
+      "grad_norm": 0.171875,
+      "learning_rate": 6.696854730143189e-05,
+      "loss": 1.0255,
+      "step": 5536
+    },
+    {
+      "epoch": 2.0087074829931972,
+      "grad_norm": 0.158203125,
+      "learning_rate": 6.691959368498349e-05,
+      "loss": 0.9482,
+      "step": 5538
+    },
+    {
+      "epoch": 2.0094331065759636,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 6.687064006853507e-05,
+      "loss": 0.9571,
+      "step": 5540
+    },
+    {
+      "epoch": 2.01015873015873,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 6.682168645208665e-05,
+      "loss": 0.9913,
+      "step": 5542
+    },
+    {
+      "epoch": 2.0108843537414964,
+      "grad_norm": 0.169921875,
+      "learning_rate": 6.677273283563825e-05,
+      "loss": 0.9695,
+      "step": 5544
+    },
+    {
+      "epoch": 2.011609977324263,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 6.672377921918981e-05,
+      "loss": 0.9805,
+      "step": 5546
+    },
+    {
+      "epoch": 2.0123356009070297,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 6.667482560274141e-05,
+      "loss": 0.9468,
+      "step": 5548
+    },
+    {
+      "epoch": 2.013061224489796,
+      "grad_norm": 0.1484375,
+      "learning_rate": 6.662587198629299e-05,
+      "loss": 0.9305,
+      "step": 5550
+    },
+    {
+      "epoch": 2.0137868480725625,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 6.657691836984458e-05,
+      "loss": 0.956,
+      "step": 5552
+    },
+    {
+      "epoch": 2.014512471655329,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 6.652796475339617e-05,
+      "loss": 0.9536,
+      "step": 5554
+    },
+    {
+      "epoch": 2.0152380952380953,
+      "grad_norm": 0.173828125,
+      "learning_rate": 6.647901113694774e-05,
+      "loss": 0.9555,
+      "step": 5556
+    },
+    {
+      "epoch": 2.0159637188208617,
+      "grad_norm": 0.150390625,
+      "learning_rate": 6.643005752049934e-05,
+      "loss": 0.9548,
+      "step": 5558
+    },
+    {
+      "epoch": 2.016689342403628,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 6.638110390405092e-05,
+      "loss": 1.0376,
+      "step": 5560
+    },
+    {
+      "epoch": 2.0174149659863945,
+      "grad_norm": 0.173828125,
+      "learning_rate": 6.63321502876025e-05,
+      "loss": 0.9717,
+      "step": 5562
+    },
+    {
+      "epoch": 2.018140589569161,
+      "grad_norm": 0.1865234375,
+      "learning_rate": 6.628319667115408e-05,
+      "loss": 1.0095,
+      "step": 5564
+    },
+    {
+      "epoch": 2.0188662131519273,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 6.623424305470566e-05,
+      "loss": 0.9583,
+      "step": 5566
+    },
+    {
+      "epoch": 2.0195918367346937,
+      "grad_norm": 0.1640625,
+      "learning_rate": 6.618528943825726e-05,
+      "loss": 1.0386,
+      "step": 5568
+    },
+    {
+      "epoch": 2.0203174603174605,
+      "grad_norm": 0.18359375,
+      "learning_rate": 6.613633582180884e-05,
+      "loss": 1.0481,
+      "step": 5570
+    },
+    {
+      "epoch": 2.021043083900227,
+      "grad_norm": 0.15625,
+      "learning_rate": 6.608738220536043e-05,
+      "loss": 0.9812,
+      "step": 5572
+    },
+    {
+      "epoch": 2.0217687074829933,
+      "grad_norm": 0.16796875,
+      "learning_rate": 6.603842858891201e-05,
+      "loss": 0.9529,
+      "step": 5574
+    },
+    {
+      "epoch": 2.0224943310657597,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 6.598947497246359e-05,
+      "loss": 0.9702,
+      "step": 5576
+    },
+    {
+      "epoch": 2.023219954648526,
+      "grad_norm": 0.1796875,
+      "learning_rate": 6.594052135601519e-05,
+      "loss": 1.0463,
+      "step": 5578
+    },
+    {
+      "epoch": 2.0239455782312925,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 6.589156773956677e-05,
+      "loss": 0.9735,
+      "step": 5580
+    },
+    {
+      "epoch": 2.024671201814059,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 6.584261412311835e-05,
+      "loss": 0.9631,
+      "step": 5582
+    },
+    {
+      "epoch": 2.0253968253968253,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 6.579366050666993e-05,
+      "loss": 1.0417,
+      "step": 5584
+    },
+    {
+      "epoch": 2.0261224489795917,
+      "grad_norm": 0.171875,
+      "learning_rate": 6.574470689022151e-05,
+      "loss": 0.9679,
+      "step": 5586
+    },
+    {
+      "epoch": 2.026848072562358,
+      "grad_norm": 0.15625,
+      "learning_rate": 6.569575327377311e-05,
+      "loss": 0.9433,
+      "step": 5588
+    },
+    {
+      "epoch": 2.0275736961451245,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 6.564679965732469e-05,
+      "loss": 1.1191,
+      "step": 5590
+    },
+    {
+      "epoch": 2.0282993197278913,
+      "grad_norm": 0.166015625,
+      "learning_rate": 6.559784604087628e-05,
+      "loss": 1.0056,
+      "step": 5592
+    },
+    {
+      "epoch": 2.0290249433106577,
+      "grad_norm": 0.140625,
+      "learning_rate": 6.554889242442786e-05,
+      "loss": 0.8881,
+      "step": 5594
+    },
+    {
+      "epoch": 2.029750566893424,
+      "grad_norm": 0.15234375,
+      "learning_rate": 6.549993880797944e-05,
+      "loss": 1.005,
+      "step": 5596
+    },
+    {
+      "epoch": 2.0304761904761905,
+      "grad_norm": 0.16015625,
+      "learning_rate": 6.545098519153102e-05,
+      "loss": 0.9251,
+      "step": 5598
+    },
+    {
+      "epoch": 2.031201814058957,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 6.540203157508262e-05,
+      "loss": 0.9907,
+      "step": 5600
+    },
+    {
+      "epoch": 2.031201814058957,
+      "eval_loss": 1.0105645656585693,
+      "eval_runtime": 88.0197,
+      "eval_samples_per_second": 20.45,
+      "eval_steps_per_second": 0.648,
+      "step": 5600
+    },
+    {
+      "epoch": 2.0319274376417233,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 6.53530779586342e-05,
+      "loss": 1.0097,
+      "step": 5602
+    },
+    {
+      "epoch": 2.0326530612244897,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 6.530412434218578e-05,
+      "loss": 0.9781,
+      "step": 5604
+    },
+    {
+      "epoch": 2.033378684807256,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 6.525517072573736e-05,
+      "loss": 0.9432,
+      "step": 5606
+    },
+    {
+      "epoch": 2.0341043083900225,
+      "grad_norm": 0.1875,
+      "learning_rate": 6.520621710928895e-05,
+      "loss": 0.9518,
+      "step": 5608
+    },
+    {
+      "epoch": 2.034829931972789,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 6.515726349284054e-05,
+      "loss": 0.9617,
+      "step": 5610
+    },
+    {
+      "epoch": 2.0355555555555553,
+      "grad_norm": 0.181640625,
+      "learning_rate": 6.510830987639212e-05,
+      "loss": 0.954,
+      "step": 5612
+    },
+    {
+      "epoch": 2.036281179138322,
+      "grad_norm": 0.173828125,
+      "learning_rate": 6.505935625994371e-05,
+      "loss": 1.0424,
+      "step": 5614
+    },
+    {
+      "epoch": 2.0370068027210886,
+      "grad_norm": 0.251953125,
+      "learning_rate": 6.501040264349529e-05,
+      "loss": 1.0404,
+      "step": 5616
+    },
+    {
+      "epoch": 2.037732426303855,
+      "grad_norm": 0.181640625,
+      "learning_rate": 6.496144902704687e-05,
+      "loss": 0.989,
+      "step": 5618
+    },
+    {
+      "epoch": 2.0384580498866214,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 6.491249541059847e-05,
+      "loss": 0.9792,
+      "step": 5620
+    },
+    {
+      "epoch": 2.039183673469388,
+      "grad_norm": 0.169921875,
+      "learning_rate": 6.486354179415005e-05,
+      "loss": 1.0176,
+      "step": 5622
+    },
+    {
+      "epoch": 2.039909297052154,
+      "grad_norm": 0.18359375,
+      "learning_rate": 6.481458817770163e-05,
+      "loss": 0.9985,
+      "step": 5624
+    },
+    {
+      "epoch": 2.0406349206349206,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 6.476563456125321e-05,
+      "loss": 0.9782,
+      "step": 5626
+    },
+    {
+      "epoch": 2.041360544217687,
+      "grad_norm": 0.158203125,
+      "learning_rate": 6.47166809448048e-05,
+      "loss": 0.9238,
+      "step": 5628
+    },
+    {
+      "epoch": 2.0420861678004534,
+      "grad_norm": 0.162109375,
+      "learning_rate": 6.466772732835639e-05,
+      "loss": 1.0303,
+      "step": 5630
+    },
+    {
+      "epoch": 2.0428117913832198,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 6.461877371190796e-05,
+      "loss": 1.0011,
+      "step": 5632
+    },
+    {
+      "epoch": 2.043537414965986,
+      "grad_norm": 0.162109375,
+      "learning_rate": 6.456982009545956e-05,
+      "loss": 0.981,
+      "step": 5634
+    },
+    {
+      "epoch": 2.044263038548753,
+      "grad_norm": 0.19140625,
+      "learning_rate": 6.452086647901114e-05,
+      "loss": 0.9464,
+      "step": 5636
+    },
+    {
+      "epoch": 2.0449886621315194,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 6.447191286256272e-05,
+      "loss": 1.0066,
+      "step": 5638
+    },
+    {
+      "epoch": 2.045714285714286,
+      "grad_norm": 0.16015625,
+      "learning_rate": 6.442295924611432e-05,
+      "loss": 0.9959,
+      "step": 5640
+    },
+    {
+      "epoch": 2.046439909297052,
+      "grad_norm": 0.16796875,
+      "learning_rate": 6.437400562966589e-05,
+      "loss": 0.942,
+      "step": 5642
+    },
+    {
+      "epoch": 2.0471655328798186,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 6.432505201321748e-05,
+      "loss": 0.9856,
+      "step": 5644
+    },
+    {
+      "epoch": 2.047891156462585,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 6.427609839676906e-05,
+      "loss": 0.9828,
+      "step": 5646
+    },
+    {
+      "epoch": 2.0486167800453514,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 6.422714478032065e-05,
+      "loss": 0.9923,
+      "step": 5648
+    },
+    {
+      "epoch": 2.049342403628118,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 6.417819116387224e-05,
+      "loss": 1.0303,
+      "step": 5650
+    },
+    {
+      "epoch": 2.050068027210884,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 6.412923754742381e-05,
+      "loss": 1.0702,
+      "step": 5652
+    },
+    {
+      "epoch": 2.0507936507936506,
+      "grad_norm": 0.21875,
+      "learning_rate": 6.408028393097541e-05,
+      "loss": 1.0109,
+      "step": 5654
+    },
+    {
+      "epoch": 2.0515192743764175,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 6.403133031452699e-05,
+      "loss": 0.9457,
+      "step": 5656
+    },
+    {
+      "epoch": 2.052244897959184,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 6.398237669807857e-05,
+      "loss": 1.0746,
+      "step": 5658
+    },
+    {
+      "epoch": 2.0529705215419503,
+      "grad_norm": 0.16796875,
+      "learning_rate": 6.393342308163017e-05,
+      "loss": 0.9641,
+      "step": 5660
+    },
+    {
+      "epoch": 2.0536961451247167,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 6.388446946518174e-05,
+      "loss": 1.0007,
+      "step": 5662
+    },
+    {
+      "epoch": 2.054421768707483,
+      "grad_norm": 0.203125,
+      "learning_rate": 6.383551584873333e-05,
+      "loss": 0.9972,
+      "step": 5664
+    },
+    {
+      "epoch": 2.0551473922902495,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 6.378656223228491e-05,
+      "loss": 1.0813,
+      "step": 5666
+    },
+    {
+      "epoch": 2.055873015873016,
+      "grad_norm": 0.154296875,
+      "learning_rate": 6.37376086158365e-05,
+      "loss": 0.9023,
+      "step": 5668
+    },
+    {
+      "epoch": 2.0565986394557823,
+      "grad_norm": 0.154296875,
+      "learning_rate": 6.368865499938809e-05,
+      "loss": 0.9559,
+      "step": 5670
+    },
+    {
+      "epoch": 2.0573242630385487,
+      "grad_norm": 0.150390625,
+      "learning_rate": 6.363970138293966e-05,
+      "loss": 0.9024,
+      "step": 5672
+    },
+    {
+      "epoch": 2.058049886621315,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 6.359074776649126e-05,
+      "loss": 0.9186,
+      "step": 5674
+    },
+    {
+      "epoch": 2.0587755102040814,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 6.354179415004284e-05,
+      "loss": 1.0022,
+      "step": 5676
+    },
+    {
+      "epoch": 2.0595011337868483,
+      "grad_norm": 0.1865234375,
+      "learning_rate": 6.349284053359442e-05,
+      "loss": 1.0076,
+      "step": 5678
+    },
+    {
+      "epoch": 2.0602267573696147,
+      "grad_norm": 0.166015625,
+      "learning_rate": 6.344388691714602e-05,
+      "loss": 0.9731,
+      "step": 5680
+    },
+    {
+      "epoch": 2.060952380952381,
+      "grad_norm": 0.166015625,
+      "learning_rate": 6.339493330069759e-05,
+      "loss": 0.9956,
+      "step": 5682
+    },
+    {
+      "epoch": 2.0616780045351475,
+      "grad_norm": 0.18359375,
+      "learning_rate": 6.334597968424918e-05,
+      "loss": 0.918,
+      "step": 5684
+    },
+    {
+      "epoch": 2.062403628117914,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 6.329702606780076e-05,
+      "loss": 0.9229,
+      "step": 5686
+    },
+    {
+      "epoch": 2.0631292517006803,
+      "grad_norm": 0.15234375,
+      "learning_rate": 6.324807245135235e-05,
+      "loss": 0.9616,
+      "step": 5688
+    },
+    {
+      "epoch": 2.0638548752834467,
+      "grad_norm": 0.177734375,
+      "learning_rate": 6.319911883490394e-05,
+      "loss": 1.0174,
+      "step": 5690
+    },
+    {
+      "epoch": 2.064580498866213,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 6.315016521845551e-05,
+      "loss": 1.0449,
+      "step": 5692
+    },
+    {
+      "epoch": 2.0653061224489795,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 6.310121160200711e-05,
+      "loss": 0.9916,
+      "step": 5694
+    },
+    {
+      "epoch": 2.066031746031746,
+      "grad_norm": 0.166015625,
+      "learning_rate": 6.305225798555869e-05,
+      "loss": 0.9486,
+      "step": 5696
+    },
+    {
+      "epoch": 2.0667573696145123,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 6.300330436911027e-05,
+      "loss": 0.9569,
+      "step": 5698
+    },
+    {
+      "epoch": 2.067482993197279,
+      "grad_norm": 0.1640625,
+      "learning_rate": 6.295435075266187e-05,
+      "loss": 0.9016,
+      "step": 5700
+    },
+    {
+      "epoch": 2.067482993197279,
+      "eval_loss": 1.0106786489486694,
+      "eval_runtime": 89.0366,
+      "eval_samples_per_second": 20.216,
+      "eval_steps_per_second": 0.64,
+      "step": 5700
+    },
+    {
+      "epoch": 2.0682086167800455,
+      "grad_norm": 0.1640625,
+      "learning_rate": 6.290539713621344e-05,
+      "loss": 0.9435,
+      "step": 5702
+    },
+    {
+      "epoch": 2.068934240362812,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 6.285644351976503e-05,
+      "loss": 1.0447,
+      "step": 5704
+    },
+    {
+      "epoch": 2.0696598639455783,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 6.280748990331661e-05,
+      "loss": 0.9554,
+      "step": 5706
+    },
+    {
+      "epoch": 2.0703854875283447,
+      "grad_norm": 0.162109375,
+      "learning_rate": 6.27585362868682e-05,
+      "loss": 1.03,
+      "step": 5708
+    },
+    {
+      "epoch": 2.071111111111111,
+      "grad_norm": 0.18359375,
+      "learning_rate": 6.270958267041978e-05,
+      "loss": 0.9301,
+      "step": 5710
+    },
+    {
+      "epoch": 2.0718367346938775,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 6.266062905397136e-05,
+      "loss": 1.0046,
+      "step": 5712
+    },
+    {
+      "epoch": 2.072562358276644,
+      "grad_norm": 0.185546875,
+      "learning_rate": 6.261167543752296e-05,
+      "loss": 1.0349,
+      "step": 5714
+    },
+    {
+      "epoch": 2.0732879818594103,
+      "grad_norm": 0.171875,
+      "learning_rate": 6.256272182107454e-05,
+      "loss": 1.0612,
+      "step": 5716
+    },
+    {
+      "epoch": 2.0740136054421767,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 6.251376820462612e-05,
+      "loss": 0.9338,
+      "step": 5718
+    },
+    {
+      "epoch": 2.074739229024943,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 6.24648145881777e-05,
+      "loss": 1.0275,
+      "step": 5720
+    },
+    {
+      "epoch": 2.07546485260771,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 6.241586097172929e-05,
+      "loss": 0.9196,
+      "step": 5722
+    },
+    {
+      "epoch": 2.0761904761904764,
+      "grad_norm": 0.150390625,
+      "learning_rate": 6.236690735528088e-05,
+      "loss": 0.9807,
+      "step": 5724
+    },
+    {
+      "epoch": 2.0769160997732428,
+      "grad_norm": 0.171875,
+      "learning_rate": 6.231795373883246e-05,
+      "loss": 0.9343,
+      "step": 5726
+    },
+    {
+      "epoch": 2.077641723356009,
+      "grad_norm": 0.181640625,
+      "learning_rate": 6.226900012238405e-05,
+      "loss": 0.9618,
+      "step": 5728
+    },
+    {
+      "epoch": 2.0783673469387756,
+      "grad_norm": 0.18359375,
+      "learning_rate": 6.222004650593563e-05,
+      "loss": 0.9586,
+      "step": 5730
+    },
+    {
+      "epoch": 2.079092970521542,
+      "grad_norm": 0.1640625,
+      "learning_rate": 6.217109288948721e-05,
+      "loss": 1.0001,
+      "step": 5732
+    },
+    {
+      "epoch": 2.0798185941043084,
+      "grad_norm": 0.18359375,
+      "learning_rate": 6.21221392730388e-05,
+      "loss": 1.0526,
+      "step": 5734
+    },
+    {
+      "epoch": 2.0805442176870748,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 6.207318565659039e-05,
+      "loss": 0.9504,
+      "step": 5736
+    },
+    {
+      "epoch": 2.081269841269841,
+      "grad_norm": 0.150390625,
+      "learning_rate": 6.202423204014197e-05,
+      "loss": 0.923,
+      "step": 5738
+    },
+    {
+      "epoch": 2.0819954648526076,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 6.197527842369355e-05,
+      "loss": 0.9957,
+      "step": 5740
+    },
+    {
+      "epoch": 2.082721088435374,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 6.192632480724514e-05,
+      "loss": 0.8884,
+      "step": 5742
+    },
+    {
+      "epoch": 2.083446712018141,
+      "grad_norm": 0.162109375,
+      "learning_rate": 6.187737119079672e-05,
+      "loss": 0.9332,
+      "step": 5744
+    },
+    {
+      "epoch": 2.084172335600907,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 6.182841757434831e-05,
+      "loss": 0.9743,
+      "step": 5746
+    },
+    {
+      "epoch": 2.0848979591836736,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 6.17794639578999e-05,
+      "loss": 0.9578,
+      "step": 5748
+    },
+    {
+      "epoch": 2.08562358276644,
+      "grad_norm": 0.1865234375,
+      "learning_rate": 6.173051034145148e-05,
+      "loss": 0.9793,
+      "step": 5750
+    },
+    {
+      "epoch": 2.0863492063492064,
+      "grad_norm": 0.162109375,
+      "learning_rate": 6.168155672500306e-05,
+      "loss": 0.9264,
+      "step": 5752
+    },
+    {
+      "epoch": 2.087074829931973,
+      "grad_norm": 0.171875,
+      "learning_rate": 6.163260310855464e-05,
+      "loss": 0.92,
+      "step": 5754
+    },
+    {
+      "epoch": 2.087800453514739,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 6.158364949210624e-05,
+      "loss": 1.0329,
+      "step": 5756
+    },
+    {
+      "epoch": 2.0885260770975056,
+      "grad_norm": 0.1796875,
+      "learning_rate": 6.153469587565782e-05,
+      "loss": 0.9942,
+      "step": 5758
+    },
+    {
+      "epoch": 2.089251700680272,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 6.14857422592094e-05,
+      "loss": 0.9291,
+      "step": 5760
+    },
+    {
+      "epoch": 2.0899773242630384,
+      "grad_norm": 0.150390625,
+      "learning_rate": 6.143678864276099e-05,
+      "loss": 0.9504,
+      "step": 5762
+    },
+    {
+      "epoch": 2.090702947845805,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 6.138783502631257e-05,
+      "loss": 0.9671,
+      "step": 5764
+    },
+    {
+      "epoch": 2.0914285714285716,
+      "grad_norm": 0.20703125,
+      "learning_rate": 6.133888140986416e-05,
+      "loss": 1.0002,
+      "step": 5766
+    },
+    {
+      "epoch": 2.092154195011338,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 6.128992779341575e-05,
+      "loss": 0.9597,
+      "step": 5768
+    },
+    {
+      "epoch": 2.0928798185941044,
+      "grad_norm": 0.1884765625,
+      "learning_rate": 6.124097417696733e-05,
+      "loss": 1.0265,
+      "step": 5770
+    },
+    {
+      "epoch": 2.093605442176871,
+      "grad_norm": 0.1796875,
+      "learning_rate": 6.119202056051891e-05,
+      "loss": 1.035,
+      "step": 5772
+    },
+    {
+      "epoch": 2.0943310657596372,
+      "grad_norm": 0.166015625,
+      "learning_rate": 6.114306694407049e-05,
+      "loss": 0.9861,
+      "step": 5774
+    },
+    {
+      "epoch": 2.0950566893424036,
+      "grad_norm": 0.173828125,
+      "learning_rate": 6.109411332762209e-05,
+      "loss": 1.011,
+      "step": 5776
+    },
+    {
+      "epoch": 2.09578231292517,
+      "grad_norm": 0.162109375,
+      "learning_rate": 6.104515971117366e-05,
+      "loss": 0.9615,
+      "step": 5778
+    },
+    {
+      "epoch": 2.0965079365079364,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 6.099620609472525e-05,
+      "loss": 1.0444,
+      "step": 5780
+    },
+    {
+      "epoch": 2.097233560090703,
+      "grad_norm": 0.17578125,
+      "learning_rate": 6.094725247827684e-05,
+      "loss": 0.9458,
+      "step": 5782
+    },
+    {
+      "epoch": 2.0979591836734692,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 6.089829886182842e-05,
+      "loss": 0.9296,
+      "step": 5784
+    },
+    {
+      "epoch": 2.0986848072562356,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 6.084934524538001e-05,
+      "loss": 0.8853,
+      "step": 5786
+    },
+    {
+      "epoch": 2.0994104308390025,
+      "grad_norm": 0.2177734375,
+      "learning_rate": 6.080039162893158e-05,
+      "loss": 0.9348,
+      "step": 5788
+    },
+    {
+      "epoch": 2.100136054421769,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 6.075143801248318e-05,
+      "loss": 0.934,
+      "step": 5790
+    },
+    {
+      "epoch": 2.1008616780045353,
+      "grad_norm": 0.1875,
+      "learning_rate": 6.070248439603477e-05,
+      "loss": 0.9812,
+      "step": 5792
+    },
+    {
+      "epoch": 2.1015873015873017,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 6.065353077958634e-05,
+      "loss": 0.9687,
+      "step": 5794
+    },
+    {
+      "epoch": 2.102312925170068,
+      "grad_norm": 0.150390625,
+      "learning_rate": 6.060457716313793e-05,
+      "loss": 0.9097,
+      "step": 5796
+    },
+    {
+      "epoch": 2.1030385487528345,
+      "grad_norm": 0.166015625,
+      "learning_rate": 6.055562354668951e-05,
+      "loss": 0.9202,
+      "step": 5798
+    },
+    {
+      "epoch": 2.103764172335601,
+      "grad_norm": 0.17578125,
+      "learning_rate": 6.05066699302411e-05,
+      "loss": 0.9264,
+      "step": 5800
+    },
+    {
+      "epoch": 2.103764172335601,
+      "eval_loss": 1.010640025138855,
+      "eval_runtime": 86.5952,
+      "eval_samples_per_second": 20.786,
+      "eval_steps_per_second": 0.658,
+      "step": 5800
+    },
+    {
+      "epoch": 2.1044897959183673,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 6.045771631379269e-05,
+      "loss": 0.9798,
+      "step": 5802
+    },
+    {
+      "epoch": 2.1052154195011337,
+      "grad_norm": 0.1640625,
+      "learning_rate": 6.040876269734427e-05,
+      "loss": 0.953,
+      "step": 5804
+    },
+    {
+      "epoch": 2.1059410430839,
+      "grad_norm": 0.169921875,
+      "learning_rate": 6.035980908089586e-05,
+      "loss": 0.9906,
+      "step": 5806
+    },
+    {
+      "epoch": 2.1066666666666665,
+      "grad_norm": 0.197265625,
+      "learning_rate": 6.031085546444743e-05,
+      "loss": 0.9762,
+      "step": 5808
+    },
+    {
+      "epoch": 2.1073922902494333,
+      "grad_norm": 0.166015625,
+      "learning_rate": 6.026190184799903e-05,
+      "loss": 0.9851,
+      "step": 5810
+    },
+    {
+      "epoch": 2.1081179138321997,
+      "grad_norm": 0.171875,
+      "learning_rate": 6.0212948231550604e-05,
+      "loss": 0.9887,
+      "step": 5812
+    },
+    {
+      "epoch": 2.108843537414966,
+      "grad_norm": 0.169921875,
+      "learning_rate": 6.016399461510219e-05,
+      "loss": 0.9786,
+      "step": 5814
+    },
+    {
+      "epoch": 2.1095691609977325,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 6.011504099865378e-05,
+      "loss": 0.952,
+      "step": 5816
+    },
+    {
+      "epoch": 2.110294784580499,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 6.006608738220536e-05,
+      "loss": 1.0122,
+      "step": 5818
+    },
+    {
+      "epoch": 2.1110204081632653,
+      "grad_norm": 0.15625,
+      "learning_rate": 6.001713376575695e-05,
+      "loss": 0.9715,
+      "step": 5820
+    },
+    {
+      "epoch": 2.1117460317460317,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 5.996818014930853e-05,
+      "loss": 1.005,
+      "step": 5822
+    },
+    {
+      "epoch": 2.112471655328798,
+      "grad_norm": 0.1875,
+      "learning_rate": 5.991922653286012e-05,
+      "loss": 1.021,
+      "step": 5824
+    },
+    {
+      "epoch": 2.1131972789115645,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 5.987027291641171e-05,
+      "loss": 0.9739,
+      "step": 5826
+    },
+    {
+      "epoch": 2.113922902494331,
+      "grad_norm": 0.173828125,
+      "learning_rate": 5.982131929996328e-05,
+      "loss": 0.9251,
+      "step": 5828
+    },
+    {
+      "epoch": 2.1146485260770973,
+      "grad_norm": 0.1640625,
+      "learning_rate": 5.977236568351488e-05,
+      "loss": 1.0101,
+      "step": 5830
+    },
+    {
+      "epoch": 2.115374149659864,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 5.9723412067066454e-05,
+      "loss": 0.9652,
+      "step": 5832
+    },
+    {
+      "epoch": 2.1160997732426305,
+      "grad_norm": 0.16015625,
+      "learning_rate": 5.967445845061804e-05,
+      "loss": 0.9139,
+      "step": 5834
+    },
+    {
+      "epoch": 2.116825396825397,
+      "grad_norm": 0.158203125,
+      "learning_rate": 5.962550483416963e-05,
+      "loss": 1.0025,
+      "step": 5836
+    },
+    {
+      "epoch": 2.1175510204081633,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 5.957655121772121e-05,
+      "loss": 0.973,
+      "step": 5838
+    },
+    {
+      "epoch": 2.1182766439909297,
+      "grad_norm": 0.1484375,
+      "learning_rate": 5.95275976012728e-05,
+      "loss": 0.918,
+      "step": 5840
+    },
+    {
+      "epoch": 2.119002267573696,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 5.947864398482438e-05,
+      "loss": 0.9193,
+      "step": 5842
+    },
+    {
+      "epoch": 2.1197278911564625,
+      "grad_norm": 0.177734375,
+      "learning_rate": 5.942969036837597e-05,
+      "loss": 0.949,
+      "step": 5844
+    },
+    {
+      "epoch": 2.120453514739229,
+      "grad_norm": 0.1796875,
+      "learning_rate": 5.938073675192756e-05,
+      "loss": 0.9686,
+      "step": 5846
+    },
+    {
+      "epoch": 2.1211791383219953,
+      "grad_norm": 0.1796875,
+      "learning_rate": 5.933178313547913e-05,
+      "loss": 1.0197,
+      "step": 5848
+    },
+    {
+      "epoch": 2.1219047619047617,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 5.928282951903073e-05,
+      "loss": 0.9537,
+      "step": 5850
+    },
+    {
+      "epoch": 2.122630385487528,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 5.9233875902582304e-05,
+      "loss": 0.9571,
+      "step": 5852
+    },
+    {
+      "epoch": 2.123356009070295,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 5.918492228613389e-05,
+      "loss": 0.9315,
+      "step": 5854
+    },
+    {
+      "epoch": 2.1240816326530614,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 5.913596866968547e-05,
+      "loss": 0.9558,
+      "step": 5856
+    },
+    {
+      "epoch": 2.124807256235828,
+      "grad_norm": 0.16015625,
+      "learning_rate": 5.908701505323706e-05,
+      "loss": 0.9256,
+      "step": 5858
+    },
+    {
+      "epoch": 2.125532879818594,
+      "grad_norm": 0.21875,
+      "learning_rate": 5.903806143678865e-05,
+      "loss": 0.9814,
+      "step": 5860
+    },
+    {
+      "epoch": 2.1262585034013606,
+      "grad_norm": 0.1982421875,
+      "learning_rate": 5.898910782034023e-05,
+      "loss": 1.0241,
+      "step": 5862
+    },
+    {
+      "epoch": 2.126984126984127,
+      "grad_norm": 0.16796875,
+      "learning_rate": 5.894015420389182e-05,
+      "loss": 0.9556,
+      "step": 5864
+    },
+    {
+      "epoch": 2.1277097505668934,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 5.889120058744339e-05,
+      "loss": 1.0065,
+      "step": 5866
+    },
+    {
+      "epoch": 2.1284353741496598,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 5.884224697099498e-05,
+      "loss": 0.9229,
+      "step": 5868
+    },
+    {
+      "epoch": 2.129160997732426,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 5.879329335454658e-05,
+      "loss": 0.8913,
+      "step": 5870
+    },
+    {
+      "epoch": 2.1298866213151926,
+      "grad_norm": 0.193359375,
+      "learning_rate": 5.8744339738098153e-05,
+      "loss": 0.9706,
+      "step": 5872
+    },
+    {
+      "epoch": 2.130612244897959,
+      "grad_norm": 0.1796875,
+      "learning_rate": 5.869538612164974e-05,
+      "loss": 0.9607,
+      "step": 5874
+    },
+    {
+      "epoch": 2.131337868480726,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 5.864643250520132e-05,
+      "loss": 0.9314,
+      "step": 5876
+    },
+    {
+      "epoch": 2.132063492063492,
+      "grad_norm": 0.173828125,
+      "learning_rate": 5.859747888875291e-05,
+      "loss": 0.97,
+      "step": 5878
+    },
+    {
+      "epoch": 2.1327891156462586,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 5.85485252723045e-05,
+      "loss": 0.9887,
+      "step": 5880
+    },
+    {
+      "epoch": 2.133514739229025,
+      "grad_norm": 0.162109375,
+      "learning_rate": 5.849957165585608e-05,
+      "loss": 1.013,
+      "step": 5882
+    },
+    {
+      "epoch": 2.1342403628117914,
+      "grad_norm": 0.15625,
+      "learning_rate": 5.845061803940767e-05,
+      "loss": 0.9536,
+      "step": 5884
+    },
+    {
+      "epoch": 2.134965986394558,
+      "grad_norm": 0.166015625,
+      "learning_rate": 5.840166442295924e-05,
+      "loss": 0.9715,
+      "step": 5886
+    },
+    {
+      "epoch": 2.135691609977324,
+      "grad_norm": 0.15234375,
+      "learning_rate": 5.835271080651083e-05,
+      "loss": 1.0258,
+      "step": 5888
+    },
+    {
+      "epoch": 2.1364172335600906,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 5.8303757190062414e-05,
+      "loss": 0.9495,
+      "step": 5890
+    },
+    {
+      "epoch": 2.137142857142857,
+      "grad_norm": 0.166015625,
+      "learning_rate": 5.8254803573614e-05,
+      "loss": 0.9514,
+      "step": 5892
+    },
+    {
+      "epoch": 2.1378684807256234,
+      "grad_norm": 0.181640625,
+      "learning_rate": 5.820584995716559e-05,
+      "loss": 1.001,
+      "step": 5894
+    },
+    {
+      "epoch": 2.13859410430839,
+      "grad_norm": 0.177734375,
+      "learning_rate": 5.815689634071717e-05,
+      "loss": 0.9775,
+      "step": 5896
+    },
+    {
+      "epoch": 2.1393197278911567,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 5.810794272426876e-05,
+      "loss": 0.9368,
+      "step": 5898
+    },
+    {
+      "epoch": 2.140045351473923,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 5.805898910782034e-05,
+      "loss": 0.9397,
+      "step": 5900
+    },
+    {
+      "epoch": 2.140045351473923,
+      "eval_loss": 1.0105234384536743,
+      "eval_runtime": 86.09,
+      "eval_samples_per_second": 20.908,
+      "eval_steps_per_second": 0.662,
+      "step": 5900
+    },
+    {
+      "epoch": 2.1407709750566895,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 5.801003549137193e-05,
+      "loss": 1.0924,
+      "step": 5902
+    },
+    {
+      "epoch": 2.141496598639456,
+      "grad_norm": 0.1796875,
+      "learning_rate": 5.796108187492352e-05,
+      "loss": 0.9779,
+      "step": 5904
+    },
+    {
+      "epoch": 2.1422222222222222,
+      "grad_norm": 0.169921875,
+      "learning_rate": 5.791212825847509e-05,
+      "loss": 0.9686,
+      "step": 5906
+    },
+    {
+      "epoch": 2.1429478458049886,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 5.786317464202668e-05,
+      "loss": 0.9359,
+      "step": 5908
+    },
+    {
+      "epoch": 2.143673469387755,
+      "grad_norm": 0.185546875,
+      "learning_rate": 5.7814221025578264e-05,
+      "loss": 0.982,
+      "step": 5910
+    },
+    {
+      "epoch": 2.1443990929705214,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 5.776526740912985e-05,
+      "loss": 0.887,
+      "step": 5912
+    },
+    {
+      "epoch": 2.145124716553288,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 5.771631379268144e-05,
+      "loss": 0.9527,
+      "step": 5914
+    },
+    {
+      "epoch": 2.1458503401360542,
+      "grad_norm": 0.1875,
+      "learning_rate": 5.766736017623302e-05,
+      "loss": 0.9901,
+      "step": 5916
+    },
+    {
+      "epoch": 2.1465759637188206,
+      "grad_norm": 0.169921875,
+      "learning_rate": 5.761840655978461e-05,
+      "loss": 1.0351,
+      "step": 5918
+    },
+    {
+      "epoch": 2.1473015873015875,
+      "grad_norm": 0.16796875,
+      "learning_rate": 5.756945294333619e-05,
+      "loss": 1.0367,
+      "step": 5920
+    },
+    {
+      "epoch": 2.148027210884354,
+      "grad_norm": 0.1796875,
+      "learning_rate": 5.752049932688778e-05,
+      "loss": 0.9757,
+      "step": 5922
+    },
+    {
+      "epoch": 2.1487528344671203,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 5.7471545710439354e-05,
+      "loss": 0.9131,
+      "step": 5924
+    },
+    {
+      "epoch": 2.1494784580498867,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 5.742259209399094e-05,
+      "loss": 0.9754,
+      "step": 5926
+    },
+    {
+      "epoch": 2.150204081632653,
+      "grad_norm": 0.1640625,
+      "learning_rate": 5.737363847754253e-05,
+      "loss": 1.0091,
+      "step": 5928
+    },
+    {
+      "epoch": 2.1509297052154195,
+      "grad_norm": 0.197265625,
+      "learning_rate": 5.7324684861094114e-05,
+      "loss": 1.0215,
+      "step": 5930
+    },
+    {
+      "epoch": 2.151655328798186,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 5.72757312446457e-05,
+      "loss": 0.9583,
+      "step": 5932
+    },
+    {
+      "epoch": 2.1523809523809523,
+      "grad_norm": 0.1796875,
+      "learning_rate": 5.722677762819728e-05,
+      "loss": 1.0407,
+      "step": 5934
+    },
+    {
+      "epoch": 2.1531065759637187,
+      "grad_norm": 0.1484375,
+      "learning_rate": 5.717782401174887e-05,
+      "loss": 0.8784,
+      "step": 5936
+    },
+    {
+      "epoch": 2.153832199546485,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 5.712887039530046e-05,
+      "loss": 0.9976,
+      "step": 5938
+    },
+    {
+      "epoch": 2.1545578231292515,
+      "grad_norm": 0.1923828125,
+      "learning_rate": 5.707991677885204e-05,
+      "loss": 0.9972,
+      "step": 5940
+    },
+    {
+      "epoch": 2.1552834467120183,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 5.703096316240363e-05,
+      "loss": 0.9872,
+      "step": 5942
+    },
+    {
+      "epoch": 2.1560090702947847,
+      "grad_norm": 0.169921875,
+      "learning_rate": 5.6982009545955204e-05,
+      "loss": 0.9176,
+      "step": 5944
+    },
+    {
+      "epoch": 2.156734693877551,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 5.693305592950679e-05,
+      "loss": 0.9286,
+      "step": 5946
+    },
+    {
+      "epoch": 2.1574603174603175,
+      "grad_norm": 0.16015625,
+      "learning_rate": 5.688410231305838e-05,
+      "loss": 0.9204,
+      "step": 5948
+    },
+    {
+      "epoch": 2.158185941043084,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 5.6835148696609964e-05,
+      "loss": 0.9486,
+      "step": 5950
+    },
+    {
+      "epoch": 2.1589115646258503,
+      "grad_norm": 0.158203125,
+      "learning_rate": 5.678619508016155e-05,
+      "loss": 1.0603,
+      "step": 5952
+    },
+    {
+      "epoch": 2.1596371882086167,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 5.673724146371313e-05,
+      "loss": 0.9545,
+      "step": 5954
+    },
+    {
+      "epoch": 2.160362811791383,
+      "grad_norm": 0.1884765625,
+      "learning_rate": 5.668828784726472e-05,
+      "loss": 1.0123,
+      "step": 5956
+    },
+    {
+      "epoch": 2.1610884353741495,
+      "grad_norm": 0.150390625,
+      "learning_rate": 5.66393342308163e-05,
+      "loss": 0.9411,
+      "step": 5958
+    },
+    {
+      "epoch": 2.161814058956916,
+      "grad_norm": 0.15625,
+      "learning_rate": 5.659038061436789e-05,
+      "loss": 0.9927,
+      "step": 5960
+    },
+    {
+      "epoch": 2.1625396825396823,
+      "grad_norm": 0.166015625,
+      "learning_rate": 5.654142699791948e-05,
+      "loss": 0.9126,
+      "step": 5962
+    },
+    {
+      "epoch": 2.163265306122449,
+      "grad_norm": 0.193359375,
+      "learning_rate": 5.6492473381471054e-05,
+      "loss": 0.9676,
+      "step": 5964
+    },
+    {
+      "epoch": 2.1639909297052156,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 5.644351976502264e-05,
+      "loss": 0.9483,
+      "step": 5966
+    },
+    {
+      "epoch": 2.164716553287982,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 5.6394566148574225e-05,
+      "loss": 0.9706,
+      "step": 5968
+    },
+    {
+      "epoch": 2.1654421768707484,
+      "grad_norm": 0.162109375,
+      "learning_rate": 5.6345612532125814e-05,
+      "loss": 0.978,
+      "step": 5970
+    },
+    {
+      "epoch": 2.1661678004535148,
+      "grad_norm": 0.19140625,
+      "learning_rate": 5.62966589156774e-05,
+      "loss": 0.9953,
+      "step": 5972
+    },
+    {
+      "epoch": 2.166893424036281,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 5.624770529922898e-05,
+      "loss": 0.9638,
+      "step": 5974
+    },
+    {
+      "epoch": 2.1676190476190476,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 5.619875168278057e-05,
+      "loss": 1.0009,
+      "step": 5976
+    },
+    {
+      "epoch": 2.168344671201814,
+      "grad_norm": 0.1875,
+      "learning_rate": 5.614979806633215e-05,
+      "loss": 1.0743,
+      "step": 5978
+    },
+    {
+      "epoch": 2.1690702947845804,
+      "grad_norm": 0.1640625,
+      "learning_rate": 5.610084444988374e-05,
+      "loss": 0.9951,
+      "step": 5980
+    },
+    {
+      "epoch": 2.1697959183673468,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 5.605189083343533e-05,
+      "loss": 0.9734,
+      "step": 5982
+    },
+    {
+      "epoch": 2.170521541950113,
+      "grad_norm": 0.18359375,
+      "learning_rate": 5.6002937216986903e-05,
+      "loss": 1.0389,
+      "step": 5984
+    },
+    {
+      "epoch": 2.17124716553288,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 5.595398360053849e-05,
+      "loss": 1.0822,
+      "step": 5986
+    },
+    {
+      "epoch": 2.1719727891156464,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 5.5905029984090075e-05,
+      "loss": 0.9547,
+      "step": 5988
+    },
+    {
+      "epoch": 2.172698412698413,
+      "grad_norm": 0.15625,
+      "learning_rate": 5.5856076367641664e-05,
+      "loss": 0.9614,
+      "step": 5990
+    },
+    {
+      "epoch": 2.173424036281179,
+      "grad_norm": 0.177734375,
+      "learning_rate": 5.580712275119325e-05,
+      "loss": 0.9657,
+      "step": 5992
+    },
+    {
+      "epoch": 2.1741496598639456,
+      "grad_norm": 0.177734375,
+      "learning_rate": 5.575816913474483e-05,
+      "loss": 0.9888,
+      "step": 5994
+    },
+    {
+      "epoch": 2.174875283446712,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 5.570921551829642e-05,
+      "loss": 0.9204,
+      "step": 5996
+    },
+    {
+      "epoch": 2.1756009070294784,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 5.5660261901848e-05,
+      "loss": 1.0059,
+      "step": 5998
+    },
+    {
+      "epoch": 2.176326530612245,
+      "grad_norm": 0.1640625,
+      "learning_rate": 5.561130828539959e-05,
+      "loss": 1.0289,
+      "step": 6000
+    },
+    {
+      "epoch": 2.176326530612245,
+      "eval_loss": 1.0105350017547607,
+      "eval_runtime": 89.1474,
+      "eval_samples_per_second": 20.191,
+      "eval_steps_per_second": 0.639,
+      "step": 6000
+    }
+  ],
+  "logging_steps": 2,
+  "max_steps": 8271,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 100,
+  "stateful_callbacks": {
+    "EarlyStoppingCallback": {
+      "args": {
+        "early_stopping_patience": 3,
+        "early_stopping_threshold": 0.0
+      },
+      "attributes": {
+        "early_stopping_patience_counter": 1
+      }
+    },
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.0187451388598944e+19,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}