diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,5201 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.0,
+  "eval_steps": 146,
+  "global_step": 1456,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.005494505494505495,
+      "grad_norm": Infinity,
+      "learning_rate": 0.0,
+      "loss": 2.0983,
+      "step": 2
+    },
+    {
+      "epoch": 0.01098901098901099,
+      "grad_norm": 5.516454696655273,
+      "learning_rate": 1.36986301369863e-07,
+      "loss": 2.0688,
+      "step": 4
+    },
+    {
+      "epoch": 0.016483516483516484,
+      "grad_norm": 5.619696617126465,
+      "learning_rate": 2.73972602739726e-07,
+      "loss": 2.0901,
+      "step": 6
+    },
+    {
+      "epoch": 0.02197802197802198,
+      "grad_norm": 5.722362995147705,
+      "learning_rate": 4.1095890410958903e-07,
+      "loss": 2.1,
+      "step": 8
+    },
+    {
+      "epoch": 0.027472527472527472,
+      "grad_norm": 5.279537200927734,
+      "learning_rate": 5.47945205479452e-07,
+      "loss": 2.0519,
+      "step": 10
+    },
+    {
+      "epoch": 0.03296703296703297,
+      "grad_norm": 4.9255170822143555,
+      "learning_rate": 6.849315068493151e-07,
+      "loss": 2.0328,
+      "step": 12
+    },
+    {
+      "epoch": 0.038461538461538464,
+      "grad_norm": 4.8295369148254395,
+      "learning_rate": 8.219178082191781e-07,
+      "loss": 1.9894,
+      "step": 14
+    },
+    {
+      "epoch": 0.04395604395604396,
+      "grad_norm": 6.103366374969482,
+      "learning_rate": 9.589041095890411e-07,
+      "loss": 1.9078,
+      "step": 16
+    },
+    {
+      "epoch": 0.04945054945054945,
+      "grad_norm": 5.239104747772217,
+      "learning_rate": 1.095890410958904e-06,
+      "loss": 1.8896,
+      "step": 18
+    },
+    {
+      "epoch": 0.054945054945054944,
+      "grad_norm": 4.494051933288574,
+      "learning_rate": 1.2328767123287673e-06,
+      "loss": 1.6875,
+      "step": 20
+    },
+    {
+      "epoch": 0.06043956043956044,
+      "grad_norm": 4.799252510070801,
+      "learning_rate": 1.3698630136986302e-06,
+      "loss": 1.6452,
+      "step": 22
+    },
+    {
+      "epoch": 0.06593406593406594,
+      "grad_norm": 4.425687789916992,
+      "learning_rate": 1.5068493150684932e-06,
+      "loss": 1.5281,
+      "step": 24
+    },
+    {
+      "epoch": 0.07142857142857142,
+      "grad_norm": 2.157963514328003,
+      "learning_rate": 1.6438356164383561e-06,
+      "loss": 1.3483,
+      "step": 26
+    },
+    {
+      "epoch": 0.07692307692307693,
+      "grad_norm": 1.2025972604751587,
+      "learning_rate": 1.7808219178082193e-06,
+      "loss": 1.3064,
+      "step": 28
+    },
+    {
+      "epoch": 0.08241758241758242,
+      "grad_norm": 0.9520919322967529,
+      "learning_rate": 1.9178082191780823e-06,
+      "loss": 1.2681,
+      "step": 30
+    },
+    {
+      "epoch": 0.08791208791208792,
+      "grad_norm": 0.6353013515472412,
+      "learning_rate": 2.0547945205479454e-06,
+      "loss": 1.2484,
+      "step": 32
+    },
+    {
+      "epoch": 0.09340659340659341,
+      "grad_norm": 0.46617600321769714,
+      "learning_rate": 2.191780821917808e-06,
+      "loss": 1.24,
+      "step": 34
+    },
+    {
+      "epoch": 0.0989010989010989,
+      "grad_norm": 0.44788283109664917,
+      "learning_rate": 2.3287671232876713e-06,
+      "loss": 1.2342,
+      "step": 36
+    },
+    {
+      "epoch": 0.1043956043956044,
+      "grad_norm": 0.3921670615673065,
+      "learning_rate": 2.4657534246575345e-06,
+      "loss": 1.217,
+      "step": 38
+    },
+    {
+      "epoch": 0.10989010989010989,
+      "grad_norm": 0.3610630929470062,
+      "learning_rate": 2.6027397260273973e-06,
+      "loss": 1.225,
+      "step": 40
+    },
+    {
+      "epoch": 0.11538461538461539,
+      "grad_norm": 0.3526926040649414,
+      "learning_rate": 2.7397260273972604e-06,
+      "loss": 1.2088,
+      "step": 42
+    },
+    {
+      "epoch": 0.12087912087912088,
+      "grad_norm": 0.3451712727546692,
+      "learning_rate": 2.876712328767123e-06,
+      "loss": 1.2198,
+      "step": 44
+    },
+    {
+      "epoch": 0.12637362637362637,
+      "grad_norm": 0.33328160643577576,
+      "learning_rate": 3.0136986301369864e-06,
+      "loss": 1.2173,
+      "step": 46
+    },
+    {
+      "epoch": 0.13186813186813187,
+      "grad_norm": 0.3273652195930481,
+      "learning_rate": 3.1506849315068495e-06,
+      "loss": 1.2088,
+      "step": 48
+    },
+    {
+      "epoch": 0.13736263736263737,
+      "grad_norm": 0.3195970952510834,
+      "learning_rate": 3.2876712328767123e-06,
+      "loss": 1.2054,
+      "step": 50
+    },
+    {
+      "epoch": 0.14285714285714285,
+      "grad_norm": 0.2997814416885376,
+      "learning_rate": 3.4246575342465754e-06,
+      "loss": 1.2046,
+      "step": 52
+    },
+    {
+      "epoch": 0.14835164835164835,
+      "grad_norm": 0.2609250247478485,
+      "learning_rate": 3.5616438356164386e-06,
+      "loss": 1.2058,
+      "step": 54
+    },
+    {
+      "epoch": 0.15384615384615385,
+      "grad_norm": 0.21140357851982117,
+      "learning_rate": 3.6986301369863014e-06,
+      "loss": 1.2001,
+      "step": 56
+    },
+    {
+      "epoch": 0.15934065934065933,
+      "grad_norm": 0.16791734099388123,
+      "learning_rate": 3.8356164383561645e-06,
+      "loss": 1.2017,
+      "step": 58
+    },
+    {
+      "epoch": 0.16483516483516483,
+      "grad_norm": 0.13620080053806305,
+      "learning_rate": 3.972602739726027e-06,
+      "loss": 1.1951,
+      "step": 60
+    },
+    {
+      "epoch": 0.17032967032967034,
+      "grad_norm": 0.12290852516889572,
+      "learning_rate": 4.109589041095891e-06,
+      "loss": 1.2006,
+      "step": 62
+    },
+    {
+      "epoch": 0.17582417582417584,
+      "grad_norm": 0.1265764832496643,
+      "learning_rate": 4.246575342465754e-06,
+      "loss": 1.2009,
+      "step": 64
+    },
+    {
+      "epoch": 0.1813186813186813,
+      "grad_norm": 0.10449671745300293,
+      "learning_rate": 4.383561643835616e-06,
+      "loss": 1.1948,
+      "step": 66
+    },
+    {
+      "epoch": 0.18681318681318682,
+      "grad_norm": 0.09835775941610336,
+      "learning_rate": 4.52054794520548e-06,
+      "loss": 1.186,
+      "step": 68
+    },
+    {
+      "epoch": 0.19230769230769232,
+      "grad_norm": 0.10505539923906326,
+      "learning_rate": 4.657534246575343e-06,
+      "loss": 1.1913,
+      "step": 70
+    },
+    {
+      "epoch": 0.1978021978021978,
+      "grad_norm": 0.13944971561431885,
+      "learning_rate": 4.7945205479452054e-06,
+      "loss": 1.1911,
+      "step": 72
+    },
+    {
+      "epoch": 0.2032967032967033,
+      "grad_norm": 0.1453215479850769,
+      "learning_rate": 4.931506849315069e-06,
+      "loss": 1.1969,
+      "step": 74
+    },
+    {
+      "epoch": 0.2087912087912088,
+      "grad_norm": 0.12393910437822342,
+      "learning_rate": 5.068493150684932e-06,
+      "loss": 1.199,
+      "step": 76
+    },
+    {
+      "epoch": 0.21428571428571427,
+      "grad_norm": 0.10112921893596649,
+      "learning_rate": 5.2054794520547945e-06,
+      "loss": 1.1914,
+      "step": 78
+    },
+    {
+      "epoch": 0.21978021978021978,
+      "grad_norm": 0.1048625037074089,
+      "learning_rate": 5.342465753424658e-06,
+      "loss": 1.1937,
+      "step": 80
+    },
+    {
+      "epoch": 0.22527472527472528,
+      "grad_norm": 0.1040705144405365,
+      "learning_rate": 5.479452054794521e-06,
+      "loss": 1.1889,
+      "step": 82
+    },
+    {
+      "epoch": 0.23076923076923078,
+      "grad_norm": 0.14360405504703522,
+      "learning_rate": 5.6164383561643845e-06,
+      "loss": 1.1946,
+      "step": 84
+    },
+    {
+      "epoch": 0.23626373626373626,
+      "grad_norm": 0.12936367094516754,
+      "learning_rate": 5.753424657534246e-06,
+      "loss": 1.1882,
+      "step": 86
+    },
+    {
+      "epoch": 0.24175824175824176,
+      "grad_norm": 0.08769886940717697,
+      "learning_rate": 5.89041095890411e-06,
+      "loss": 1.1873,
+      "step": 88
+    },
+    {
+      "epoch": 0.24725274725274726,
+      "grad_norm": 0.1197541207075119,
+      "learning_rate": 6.027397260273973e-06,
+      "loss": 1.1863,
+      "step": 90
+    },
+    {
+      "epoch": 0.25274725274725274,
+      "grad_norm": 0.12938760221004486,
+      "learning_rate": 6.164383561643836e-06,
+      "loss": 1.188,
+      "step": 92
+    },
+    {
+      "epoch": 0.25824175824175827,
+      "grad_norm": 0.14707419276237488,
+      "learning_rate": 6.301369863013699e-06,
+      "loss": 1.1886,
+      "step": 94
+    },
+    {
+      "epoch": 0.26373626373626374,
+      "grad_norm": 0.12229153513908386,
+      "learning_rate": 6.438356164383563e-06,
+      "loss": 1.1908,
+      "step": 96
+    },
+    {
+      "epoch": 0.2692307692307692,
+      "grad_norm": 0.1133919209241867,
+      "learning_rate": 6.5753424657534245e-06,
+      "loss": 1.1825,
+      "step": 98
+    },
+    {
+      "epoch": 0.27472527472527475,
+      "grad_norm": 0.12165709584951401,
+      "learning_rate": 6.712328767123288e-06,
+      "loss": 1.1807,
+      "step": 100
+    },
+    {
+      "epoch": 0.2802197802197802,
+      "grad_norm": 0.12554234266281128,
+      "learning_rate": 6.849315068493151e-06,
+      "loss": 1.1831,
+      "step": 102
+    },
+    {
+      "epoch": 0.2857142857142857,
+      "grad_norm": 0.11460216343402863,
+      "learning_rate": 6.9863013698630145e-06,
+      "loss": 1.1865,
+      "step": 104
+    },
+    {
+      "epoch": 0.29120879120879123,
+      "grad_norm": 0.1516668051481247,
+      "learning_rate": 7.123287671232877e-06,
+      "loss": 1.1871,
+      "step": 106
+    },
+    {
+      "epoch": 0.2967032967032967,
+      "grad_norm": 0.15025198459625244,
+      "learning_rate": 7.260273972602741e-06,
+      "loss": 1.1844,
+      "step": 108
+    },
+    {
+      "epoch": 0.3021978021978022,
+      "grad_norm": 0.13930697739124298,
+      "learning_rate": 7.397260273972603e-06,
+      "loss": 1.1803,
+      "step": 110
+    },
+    {
+      "epoch": 0.3076923076923077,
+      "grad_norm": 0.14465640485286713,
+      "learning_rate": 7.534246575342466e-06,
+      "loss": 1.1895,
+      "step": 112
+    },
+    {
+      "epoch": 0.3131868131868132,
+      "grad_norm": 0.12900669872760773,
+      "learning_rate": 7.671232876712329e-06,
+      "loss": 1.1874,
+      "step": 114
+    },
+    {
+      "epoch": 0.31868131868131866,
+      "grad_norm": 0.12059827893972397,
+      "learning_rate": 7.808219178082192e-06,
+      "loss": 1.184,
+      "step": 116
+    },
+    {
+      "epoch": 0.3241758241758242,
+      "grad_norm": 0.11196708679199219,
+      "learning_rate": 7.945205479452055e-06,
+      "loss": 1.1876,
+      "step": 118
+    },
+    {
+      "epoch": 0.32967032967032966,
+      "grad_norm": 0.11420655995607376,
+      "learning_rate": 8.082191780821919e-06,
+      "loss": 1.1799,
+      "step": 120
+    },
+    {
+      "epoch": 0.33516483516483514,
+      "grad_norm": 0.10233011841773987,
+      "learning_rate": 8.219178082191782e-06,
+      "loss": 1.1836,
+      "step": 122
+    },
+    {
+      "epoch": 0.34065934065934067,
+      "grad_norm": 0.13560251891613007,
+      "learning_rate": 8.356164383561644e-06,
+      "loss": 1.1825,
+      "step": 124
+    },
+    {
+      "epoch": 0.34615384615384615,
+      "grad_norm": 0.1417393684387207,
+      "learning_rate": 8.493150684931507e-06,
+      "loss": 1.181,
+      "step": 126
+    },
+    {
+      "epoch": 0.3516483516483517,
+      "grad_norm": 0.12368721514940262,
+      "learning_rate": 8.63013698630137e-06,
+      "loss": 1.1857,
+      "step": 128
+    },
+    {
+      "epoch": 0.35714285714285715,
+      "grad_norm": 0.14341862499713898,
+      "learning_rate": 8.767123287671233e-06,
+      "loss": 1.1839,
+      "step": 130
+    },
+    {
+      "epoch": 0.3626373626373626,
+      "grad_norm": 0.14925454556941986,
+      "learning_rate": 8.904109589041097e-06,
+      "loss": 1.1792,
+      "step": 132
+    },
+    {
+      "epoch": 0.36813186813186816,
+      "grad_norm": 0.1684715300798416,
+      "learning_rate": 9.04109589041096e-06,
+      "loss": 1.1806,
+      "step": 134
+    },
+    {
+      "epoch": 0.37362637362637363,
+      "grad_norm": 0.10095636546611786,
+      "learning_rate": 9.178082191780823e-06,
+      "loss": 1.176,
+      "step": 136
+    },
+    {
+      "epoch": 0.3791208791208791,
+      "grad_norm": 0.11856956034898758,
+      "learning_rate": 9.315068493150685e-06,
+      "loss": 1.1849,
+      "step": 138
+    },
+    {
+      "epoch": 0.38461538461538464,
+      "grad_norm": 0.19565819203853607,
+      "learning_rate": 9.452054794520548e-06,
+      "loss": 1.1803,
+      "step": 140
+    },
+    {
+      "epoch": 0.3901098901098901,
+      "grad_norm": 0.15888364613056183,
+      "learning_rate": 9.589041095890411e-06,
+      "loss": 1.1756,
+      "step": 142
+    },
+    {
+      "epoch": 0.3956043956043956,
+      "grad_norm": 0.16350317001342773,
+      "learning_rate": 9.726027397260275e-06,
+      "loss": 1.1777,
+      "step": 144
+    },
+    {
+      "epoch": 0.4010989010989011,
+      "grad_norm": 0.13230757415294647,
+      "learning_rate": 9.863013698630138e-06,
+      "loss": 1.1843,
+      "step": 146
+    },
+    {
+      "epoch": 0.4010989010989011,
+      "eval_loss": 1.1792316436767578,
+      "eval_runtime": 299.2609,
+      "eval_samples_per_second": 17.296,
+      "eval_steps_per_second": 0.271,
+      "step": 146
+    },
+    {
+      "epoch": 0.4065934065934066,
+      "grad_norm": 0.1627519279718399,
+      "learning_rate": 1e-05,
+      "loss": 1.174,
+      "step": 148
+    },
+    {
+      "epoch": 0.41208791208791207,
+      "grad_norm": 0.13225838541984558,
+      "learning_rate": 9.999942488284598e-06,
+      "loss": 1.1779,
+      "step": 150
+    },
+    {
+      "epoch": 0.4175824175824176,
+      "grad_norm": 0.15863117575645447,
+      "learning_rate": 9.999769954461425e-06,
+      "loss": 1.1767,
+      "step": 152
+    },
+    {
+      "epoch": 0.4230769230769231,
+      "grad_norm": 0.15866652131080627,
+      "learning_rate": 9.999482402499569e-06,
+      "loss": 1.1758,
+      "step": 154
+    },
+    {
+      "epoch": 0.42857142857142855,
+      "grad_norm": 0.1270761787891388,
+      "learning_rate": 9.999079839014074e-06,
+      "loss": 1.1756,
+      "step": 156
+    },
+    {
+      "epoch": 0.4340659340659341,
+      "grad_norm": 0.10430735349655151,
+      "learning_rate": 9.998562273265786e-06,
+      "loss": 1.1728,
+      "step": 158
+    },
+    {
+      "epoch": 0.43956043956043955,
+      "grad_norm": 0.13678255677223206,
+      "learning_rate": 9.997929717161142e-06,
+      "loss": 1.1774,
+      "step": 160
+    },
+    {
+      "epoch": 0.44505494505494503,
+      "grad_norm": 0.13817106187343597,
+      "learning_rate": 9.997182185251896e-06,
+      "loss": 1.1804,
+      "step": 162
+    },
+    {
+      "epoch": 0.45054945054945056,
+      "grad_norm": 0.15956689417362213,
+      "learning_rate": 9.996319694734787e-06,
+      "loss": 1.1802,
+      "step": 164
+    },
+    {
+      "epoch": 0.45604395604395603,
+      "grad_norm": 0.2157265841960907,
+      "learning_rate": 9.995342265451138e-06,
+      "loss": 1.1803,
+      "step": 166
+    },
+    {
+      "epoch": 0.46153846153846156,
+      "grad_norm": 0.17677265405654907,
+      "learning_rate": 9.994249919886402e-06,
+      "loss": 1.1764,
+      "step": 168
+    },
+    {
+      "epoch": 0.46703296703296704,
+      "grad_norm": 0.16796888411045074,
+      "learning_rate": 9.993042683169647e-06,
+      "loss": 1.1702,
+      "step": 170
+    },
+    {
+      "epoch": 0.4725274725274725,
+      "grad_norm": 0.16481636464595795,
+      "learning_rate": 9.991720583072975e-06,
+      "loss": 1.1844,
+      "step": 172
+    },
+    {
+      "epoch": 0.47802197802197804,
+      "grad_norm": 0.19735674560070038,
+      "learning_rate": 9.990283650010883e-06,
+      "loss": 1.1791,
+      "step": 174
+    },
+    {
+      "epoch": 0.4835164835164835,
+      "grad_norm": 0.15785232186317444,
+      "learning_rate": 9.988731917039564e-06,
+      "loss": 1.1783,
+      "step": 176
+    },
+    {
+      "epoch": 0.489010989010989,
+      "grad_norm": 0.17257508635520935,
+      "learning_rate": 9.98706541985615e-06,
+      "loss": 1.1769,
+      "step": 178
+    },
+    {
+      "epoch": 0.4945054945054945,
+      "grad_norm": 0.1660294085741043,
+      "learning_rate": 9.985284196797884e-06,
+      "loss": 1.181,
+      "step": 180
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.1156650260090828,
+      "learning_rate": 9.983388288841246e-06,
+      "loss": 1.1865,
+      "step": 182
+    },
+    {
+      "epoch": 0.5054945054945055,
+      "grad_norm": 0.0954098179936409,
+      "learning_rate": 9.981377739601002e-06,
+      "loss": 1.1787,
+      "step": 184
+    },
+    {
+      "epoch": 0.510989010989011,
+      "grad_norm": 0.0957237258553505,
+      "learning_rate": 9.979252595329204e-06,
+      "loss": 1.1731,
+      "step": 186
+    },
+    {
+      "epoch": 0.5164835164835165,
+      "grad_norm": 0.11179114878177643,
+      "learning_rate": 9.977012904914133e-06,
+      "loss": 1.166,
+      "step": 188
+    },
+    {
+      "epoch": 0.521978021978022,
+      "grad_norm": 0.12088882923126221,
+      "learning_rate": 9.974658719879163e-06,
+      "loss": 1.1706,
+      "step": 190
+    },
+    {
+      "epoch": 0.5274725274725275,
+      "grad_norm": 0.1067051887512207,
+      "learning_rate": 9.972190094381578e-06,
+      "loss": 1.178,
+      "step": 192
+    },
+    {
+      "epoch": 0.532967032967033,
+      "grad_norm": 0.09633362293243408,
+      "learning_rate": 9.96960708521134e-06,
+      "loss": 1.1763,
+      "step": 194
+    },
+    {
+      "epoch": 0.5384615384615384,
+      "grad_norm": 0.11866485327482224,
+      "learning_rate": 9.966909751789758e-06,
+      "loss": 1.1777,
+      "step": 196
+    },
+    {
+      "epoch": 0.5439560439560439,
+      "grad_norm": 0.11669634282588959,
+      "learning_rate": 9.964098156168143e-06,
+      "loss": 1.1753,
+      "step": 198
+    },
+    {
+      "epoch": 0.5494505494505495,
+      "grad_norm": 0.1488696038722992,
+      "learning_rate": 9.96117236302637e-06,
+      "loss": 1.1656,
+      "step": 200
+    },
+    {
+      "epoch": 0.554945054945055,
+      "grad_norm": 0.1773654669523239,
+      "learning_rate": 9.958132439671392e-06,
+      "loss": 1.1809,
+      "step": 202
+    },
+    {
+      "epoch": 0.5604395604395604,
+      "grad_norm": 0.17689326405525208,
+      "learning_rate": 9.954978456035695e-06,
+      "loss": 1.1722,
+      "step": 204
+    },
+    {
+      "epoch": 0.5659340659340659,
+      "grad_norm": 0.13418439030647278,
+      "learning_rate": 9.951710484675677e-06,
+      "loss": 1.167,
+      "step": 206
+    },
+    {
+      "epoch": 0.5714285714285714,
+      "grad_norm": 0.14982537925243378,
+      "learning_rate": 9.948328600769996e-06,
+      "loss": 1.1713,
+      "step": 208
+    },
+    {
+      "epoch": 0.5769230769230769,
+      "grad_norm": 0.12595325708389282,
+      "learning_rate": 9.94483288211783e-06,
+      "loss": 1.1677,
+      "step": 210
+    },
+    {
+      "epoch": 0.5824175824175825,
+      "grad_norm": 0.10920244455337524,
+      "learning_rate": 9.941223409137088e-06,
+      "loss": 1.1727,
+      "step": 212
+    },
+    {
+      "epoch": 0.5879120879120879,
+      "grad_norm": 0.10856951773166656,
+      "learning_rate": 9.937500264862567e-06,
+      "loss": 1.1739,
+      "step": 214
+    },
+    {
+      "epoch": 0.5934065934065934,
+      "grad_norm": 0.13141585886478424,
+      "learning_rate": 9.933663534944029e-06,
+      "loss": 1.176,
+      "step": 216
+    },
+    {
+      "epoch": 0.5989010989010989,
+      "grad_norm": 0.13502047955989838,
+      "learning_rate": 9.929713307644245e-06,
+      "loss": 1.178,
+      "step": 218
+    },
+    {
+      "epoch": 0.6043956043956044,
+      "grad_norm": 0.1468881517648697,
+      "learning_rate": 9.925649673836949e-06,
+      "loss": 1.1819,
+      "step": 220
+    },
+    {
+      "epoch": 0.6098901098901099,
+      "grad_norm": 0.13739052414894104,
+      "learning_rate": 9.921472727004765e-06,
+      "loss": 1.1784,
+      "step": 222
+    },
+    {
+      "epoch": 0.6153846153846154,
+      "grad_norm": 0.13080830872058868,
+      "learning_rate": 9.917182563237045e-06,
+      "loss": 1.1762,
+      "step": 224
+    },
+    {
+      "epoch": 0.6208791208791209,
+      "grad_norm": 0.11544028669595718,
+      "learning_rate": 9.912779281227656e-06,
+      "loss": 1.1733,
+      "step": 226
+    },
+    {
+      "epoch": 0.6263736263736264,
+      "grad_norm": 0.14270079135894775,
+      "learning_rate": 9.908262982272724e-06,
+      "loss": 1.1812,
+      "step": 228
+    },
+    {
+      "epoch": 0.6318681318681318,
+      "grad_norm": 0.11610660701990128,
+      "learning_rate": 9.903633770268286e-06,
+      "loss": 1.1803,
+      "step": 230
+    },
+    {
+      "epoch": 0.6373626373626373,
+      "grad_norm": 0.10949360579252243,
+      "learning_rate": 9.89889175170791e-06,
+      "loss": 1.1776,
+      "step": 232
+    },
+    {
+      "epoch": 0.6428571428571429,
+      "grad_norm": 0.11754804104566574,
+      "learning_rate": 9.894037035680246e-06,
+      "loss": 1.169,
+      "step": 234
+    },
+    {
+      "epoch": 0.6483516483516484,
+      "grad_norm": 0.1328520029783249,
+      "learning_rate": 9.889069733866515e-06,
+      "loss": 1.1663,
+      "step": 236
+    },
+    {
+      "epoch": 0.6538461538461539,
+      "grad_norm": 0.12216369807720184,
+      "learning_rate": 9.883989960537934e-06,
+      "loss": 1.1779,
+      "step": 238
+    },
+    {
+      "epoch": 0.6593406593406593,
+      "grad_norm": 0.11219753324985504,
+      "learning_rate": 9.878797832553093e-06,
+      "loss": 1.1718,
+      "step": 240
+    },
+    {
+      "epoch": 0.6648351648351648,
+      "grad_norm": 0.11396931111812592,
+      "learning_rate": 9.873493469355271e-06,
+      "loss": 1.175,
+      "step": 242
+    },
+    {
+      "epoch": 0.6703296703296703,
+      "grad_norm": 0.104461669921875,
+      "learning_rate": 9.868076992969672e-06,
+      "loss": 1.1761,
+      "step": 244
+    },
+    {
+      "epoch": 0.6758241758241759,
+      "grad_norm": 0.11892971396446228,
+      "learning_rate": 9.862548528000644e-06,
+      "loss": 1.1713,
+      "step": 246
+    },
+    {
+      "epoch": 0.6813186813186813,
+      "grad_norm": 0.10305490344762802,
+      "learning_rate": 9.85690820162878e-06,
+      "loss": 1.1751,
+      "step": 248
+    },
+    {
+      "epoch": 0.6868131868131868,
+      "grad_norm": 0.11825086921453476,
+      "learning_rate": 9.851156143608025e-06,
+      "loss": 1.1749,
+      "step": 250
+    },
+    {
+      "epoch": 0.6923076923076923,
+      "grad_norm": 0.09655775129795074,
+      "learning_rate": 9.845292486262664e-06,
+      "loss": 1.1733,
+      "step": 252
+    },
+    {
+      "epoch": 0.6978021978021978,
+      "grad_norm": 0.09013015031814575,
+      "learning_rate": 9.839317364484295e-06,
+      "loss": 1.1736,
+      "step": 254
+    },
+    {
+      "epoch": 0.7032967032967034,
+      "grad_norm": 0.12731756269931793,
+      "learning_rate": 9.83323091572872e-06,
+      "loss": 1.1735,
+      "step": 256
+    },
+    {
+      "epoch": 0.7087912087912088,
+      "grad_norm": 0.11326012015342712,
+      "learning_rate": 9.827033280012783e-06,
+      "loss": 1.1678,
+      "step": 258
+    },
+    {
+      "epoch": 0.7142857142857143,
+      "grad_norm": 0.09636470675468445,
+      "learning_rate": 9.820724599911147e-06,
+      "loss": 1.1735,
+      "step": 260
+    },
+    {
+      "epoch": 0.7197802197802198,
+      "grad_norm": 0.12517008185386658,
+      "learning_rate": 9.81430502055302e-06,
+      "loss": 1.1839,
+      "step": 262
+    },
+    {
+      "epoch": 0.7252747252747253,
+      "grad_norm": 0.14088405668735504,
+      "learning_rate": 9.807774689618806e-06,
+      "loss": 1.1734,
+      "step": 264
+    },
+    {
+      "epoch": 0.7307692307692307,
+      "grad_norm": 0.09721764922142029,
+      "learning_rate": 9.801133757336726e-06,
+      "loss": 1.1681,
+      "step": 266
+    },
+    {
+      "epoch": 0.7362637362637363,
+      "grad_norm": 0.08106404542922974,
+      "learning_rate": 9.794382376479334e-06,
+      "loss": 1.1684,
+      "step": 268
+    },
+    {
+      "epoch": 0.7417582417582418,
+      "grad_norm": 0.1210760697722435,
+      "learning_rate": 9.787520702360035e-06,
+      "loss": 1.1656,
+      "step": 270
+    },
+    {
+      "epoch": 0.7472527472527473,
+      "grad_norm": 0.12102904915809631,
+      "learning_rate": 9.780548892829486e-06,
+      "loss": 1.1743,
+      "step": 272
+    },
+    {
+      "epoch": 0.7527472527472527,
+      "grad_norm": 0.11343714594841003,
+      "learning_rate": 9.773467108271978e-06,
+      "loss": 1.1756,
+      "step": 274
+    },
+    {
+      "epoch": 0.7582417582417582,
+      "grad_norm": 0.11999528110027313,
+      "learning_rate": 9.766275511601742e-06,
+      "loss": 1.1782,
+      "step": 276
+    },
+    {
+      "epoch": 0.7637362637362637,
+      "grad_norm": 0.148450568318367,
+      "learning_rate": 9.7589742682592e-06,
+      "loss": 1.1796,
+      "step": 278
+    },
+    {
+      "epoch": 0.7692307692307693,
+      "grad_norm": 0.09977131336927414,
+      "learning_rate": 9.751563546207167e-06,
+      "loss": 1.1686,
+      "step": 280
+    },
+    {
+      "epoch": 0.7747252747252747,
+      "grad_norm": 0.12250227481126785,
+      "learning_rate": 9.744043515926975e-06,
+      "loss": 1.1673,
+      "step": 282
+    },
+    {
+      "epoch": 0.7802197802197802,
+      "grad_norm": 0.12362032383680344,
+      "learning_rate": 9.736414350414564e-06,
+      "loss": 1.1707,
+      "step": 284
+    },
+    {
+      "epoch": 0.7857142857142857,
+      "grad_norm": 0.09825081378221512,
+      "learning_rate": 9.72867622517649e-06,
+      "loss": 1.1719,
+      "step": 286
+    },
+    {
+      "epoch": 0.7912087912087912,
+      "grad_norm": 0.13811303675174713,
+      "learning_rate": 9.720829318225897e-06,
+      "loss": 1.1711,
+      "step": 288
+    },
+    {
+      "epoch": 0.7967032967032966,
+      "grad_norm": 0.14676132798194885,
+      "learning_rate": 9.712873810078415e-06,
+      "loss": 1.1728,
+      "step": 290
+    },
+    {
+      "epoch": 0.8021978021978022,
+      "grad_norm": 0.09035024791955948,
+      "learning_rate": 9.704809883748012e-06,
+      "loss": 1.1776,
+      "step": 292
+    },
+    {
+      "epoch": 0.8021978021978022,
+      "eval_loss": 1.1732887029647827,
+      "eval_runtime": 298.7065,
+      "eval_samples_per_second": 17.328,
+      "eval_steps_per_second": 0.271,
+      "step": 292
+    },
+    {
+      "epoch": 0.8076923076923077,
+      "grad_norm": 0.10824787616729736,
+      "learning_rate": 9.696637724742785e-06,
+      "loss": 1.1735,
+      "step": 294
+    },
+    {
+      "epoch": 0.8131868131868132,
+      "grad_norm": 0.10512899607419968,
+      "learning_rate": 9.688357521060685e-06,
+      "loss": 1.1702,
+      "step": 296
+    },
+    {
+      "epoch": 0.8186813186813187,
+      "grad_norm": 0.12007107585668564,
+      "learning_rate": 9.6799694631852e-06,
+      "loss": 1.1687,
+      "step": 298
+    },
+    {
+      "epoch": 0.8241758241758241,
+      "grad_norm": 0.1161542534828186,
+      "learning_rate": 9.67147374408097e-06,
+      "loss": 1.1726,
+      "step": 300
+    },
+    {
+      "epoch": 0.8296703296703297,
+      "grad_norm": 0.16156896948814392,
+      "learning_rate": 9.662870559189344e-06,
+      "loss": 1.1639,
+      "step": 302
+    },
+    {
+      "epoch": 0.8351648351648352,
+      "grad_norm": 0.12350093573331833,
+      "learning_rate": 9.654160106423891e-06,
+      "loss": 1.1711,
+      "step": 304
+    },
+    {
+      "epoch": 0.8406593406593407,
+      "grad_norm": 0.1292959749698639,
+      "learning_rate": 9.645342586165845e-06,
+      "loss": 1.1677,
+      "step": 306
+    },
+    {
+      "epoch": 0.8461538461538461,
+      "grad_norm": 0.09656315296888351,
+      "learning_rate": 9.63641820125949e-06,
+      "loss": 1.1773,
+      "step": 308
+    },
+    {
+      "epoch": 0.8516483516483516,
+      "grad_norm": 0.13036847114562988,
+      "learning_rate": 9.627387157007502e-06,
+      "loss": 1.1819,
+      "step": 310
+    },
+    {
+      "epoch": 0.8571428571428571,
+      "grad_norm": 0.12335841357707977,
+      "learning_rate": 9.618249661166218e-06,
+      "loss": 1.1688,
+      "step": 312
+    },
+    {
+      "epoch": 0.8626373626373627,
+      "grad_norm": 0.13838279247283936,
+      "learning_rate": 9.609005923940865e-06,
+      "loss": 1.1665,
+      "step": 314
+    },
+    {
+      "epoch": 0.8681318681318682,
+      "grad_norm": 0.13689468801021576,
+      "learning_rate": 9.599656157980715e-06,
+      "loss": 1.1696,
+      "step": 316
+    },
+    {
+      "epoch": 0.8736263736263736,
+      "grad_norm": 0.13239113986492157,
+      "learning_rate": 9.590200578374198e-06,
+      "loss": 1.1686,
+      "step": 318
+    },
+    {
+      "epoch": 0.8791208791208791,
+      "grad_norm": 0.09575112909078598,
+      "learning_rate": 9.580639402643957e-06,
+      "loss": 1.1673,
+      "step": 320
+    },
+    {
+      "epoch": 0.8846153846153846,
+      "grad_norm": 0.11522715538740158,
+      "learning_rate": 9.570972850741839e-06,
+      "loss": 1.1685,
+      "step": 322
+    },
+    {
+      "epoch": 0.8901098901098901,
+      "grad_norm": 0.14126405119895935,
+      "learning_rate": 9.561201145043835e-06,
+      "loss": 1.1788,
+      "step": 324
+    },
+    {
+      "epoch": 0.8956043956043956,
+      "grad_norm": 0.17362023890018463,
+      "learning_rate": 9.551324510344972e-06,
+      "loss": 1.1746,
+      "step": 326
+    },
+    {
+      "epoch": 0.9010989010989011,
+      "grad_norm": 0.1431788206100464,
+      "learning_rate": 9.541343173854128e-06,
+      "loss": 1.1719,
+      "step": 328
+    },
+    {
+      "epoch": 0.9065934065934066,
+      "grad_norm": 0.11747945845127106,
+      "learning_rate": 9.531257365188818e-06,
+      "loss": 1.1763,
+      "step": 330
+    },
+    {
+      "epoch": 0.9120879120879121,
+      "grad_norm": 0.10873839259147644,
+      "learning_rate": 9.521067316369903e-06,
+      "loss": 1.173,
+      "step": 332
+    },
+    {
+      "epoch": 0.9175824175824175,
+      "grad_norm": 0.10153867304325104,
+      "learning_rate": 9.510773261816261e-06,
+      "loss": 1.1681,
+      "step": 334
+    },
+    {
+      "epoch": 0.9230769230769231,
+      "grad_norm": 0.07133303582668304,
+      "learning_rate": 9.500375438339384e-06,
+      "loss": 1.1643,
+      "step": 336
+    },
+    {
+      "epoch": 0.9285714285714286,
+      "grad_norm": 0.09183746576309204,
+      "learning_rate": 9.48987408513794e-06,
+      "loss": 1.1788,
+      "step": 338
+    },
+    {
+      "epoch": 0.9340659340659341,
+      "grad_norm": 0.10790558159351349,
+      "learning_rate": 9.47926944379226e-06,
+      "loss": 1.1817,
+      "step": 340
+    },
+    {
+      "epoch": 0.9395604395604396,
+      "grad_norm": 0.09112295508384705,
+      "learning_rate": 9.468561758258795e-06,
+      "loss": 1.1674,
+      "step": 342
+    },
+    {
+      "epoch": 0.945054945054945,
+      "grad_norm": 0.08475514501333237,
+      "learning_rate": 9.457751274864486e-06,
+      "loss": 1.1706,
+      "step": 344
+    },
+    {
+      "epoch": 0.9505494505494505,
+      "grad_norm": 0.10173246264457703,
+      "learning_rate": 9.446838242301113e-06,
+      "loss": 1.1644,
+      "step": 346
+    },
+    {
+      "epoch": 0.9560439560439561,
+      "grad_norm": 0.10165045410394669,
+      "learning_rate": 9.435822911619564e-06,
+      "loss": 1.1694,
+      "step": 348
+    },
+    {
+      "epoch": 0.9615384615384616,
+      "grad_norm": 0.10506884753704071,
+      "learning_rate": 9.424705536224065e-06,
+      "loss": 1.1708,
+      "step": 350
+    },
+    {
+      "epoch": 0.967032967032967,
+      "grad_norm": 0.10648410022258759,
+      "learning_rate": 9.41348637186635e-06,
+      "loss": 1.1712,
+      "step": 352
+    },
+    {
+      "epoch": 0.9725274725274725,
+      "grad_norm": 0.11728779226541519,
+      "learning_rate": 9.40216567663977e-06,
+      "loss": 1.1838,
+      "step": 354
+    },
+    {
+      "epoch": 0.978021978021978,
+      "grad_norm": 0.12955018877983093,
+      "learning_rate": 9.390743710973366e-06,
+      "loss": 1.1773,
+      "step": 356
+    },
+    {
+      "epoch": 0.9835164835164835,
+      "grad_norm": 0.14313377439975739,
+      "learning_rate": 9.379220737625877e-06,
+      "loss": 1.1693,
+      "step": 358
+    },
+    {
+      "epoch": 0.989010989010989,
+      "grad_norm": 0.17154935002326965,
+      "learning_rate": 9.367597021679686e-06,
+      "loss": 1.1674,
+      "step": 360
+    },
+    {
+      "epoch": 0.9945054945054945,
+      "grad_norm": 0.13390299677848816,
+      "learning_rate": 9.35587283053473e-06,
+      "loss": 1.1636,
+      "step": 362
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.13049638271331787,
+      "learning_rate": 9.344048433902351e-06,
+      "loss": 1.1678,
+      "step": 364
+    },
+    {
+      "epoch": 1.0054945054945055,
+      "grad_norm": 0.11013516038656235,
+      "learning_rate": 9.332124103799075e-06,
+      "loss": 1.1627,
+      "step": 366
+    },
+    {
+      "epoch": 1.010989010989011,
+      "grad_norm": 0.11757113039493561,
+      "learning_rate": 9.320100114540382e-06,
+      "loss": 1.1679,
+      "step": 368
+    },
+    {
+      "epoch": 1.0164835164835164,
+      "grad_norm": 0.11947564780712128,
+      "learning_rate": 9.307976742734366e-06,
+      "loss": 1.1767,
+      "step": 370
+    },
+    {
+      "epoch": 1.021978021978022,
+      "grad_norm": 0.1045905202627182,
+      "learning_rate": 9.295754267275393e-06,
+      "loss": 1.1694,
+      "step": 372
+    },
+    {
+      "epoch": 1.0274725274725274,
+      "grad_norm": 0.10813715308904648,
+      "learning_rate": 9.283432969337672e-06,
+      "loss": 1.1752,
+      "step": 374
+    },
+    {
+      "epoch": 1.032967032967033,
+      "grad_norm": 0.11131364107131958,
+      "learning_rate": 9.271013132368799e-06,
+      "loss": 1.1714,
+      "step": 376
+    },
+    {
+      "epoch": 1.0384615384615385,
+      "grad_norm": 0.09973058849573135,
+      "learning_rate": 9.258495042083222e-06,
+      "loss": 1.1696,
+      "step": 378
+    },
+    {
+      "epoch": 1.043956043956044,
+      "grad_norm": 0.10416287928819656,
+      "learning_rate": 9.245878986455684e-06,
+      "loss": 1.1641,
+      "step": 380
+    },
+    {
+      "epoch": 1.0494505494505495,
+      "grad_norm": 0.08854486793279648,
+      "learning_rate": 9.23316525571458e-06,
+      "loss": 1.1649,
+      "step": 382
+    },
+    {
+      "epoch": 1.054945054945055,
+      "grad_norm": 0.1364666223526001,
+      "learning_rate": 9.2203541423353e-06,
+      "loss": 1.1769,
+      "step": 384
+    },
+    {
+      "epoch": 1.0604395604395604,
+      "grad_norm": 0.14936769008636475,
+      "learning_rate": 9.207445941033483e-06,
+      "loss": 1.1686,
+      "step": 386
+    },
+    {
+      "epoch": 1.065934065934066,
+      "grad_norm": 0.18906323611736298,
+      "learning_rate": 9.19444094875825e-06,
+      "loss": 1.1692,
+      "step": 388
+    },
+    {
+      "epoch": 1.0714285714285714,
+      "grad_norm": 0.44419270753860474,
+      "learning_rate": 9.18133946468537e-06,
+      "loss": 1.1746,
+      "step": 390
+    },
+    {
+      "epoch": 1.0769230769230769,
+      "grad_norm": 0.1541990488767624,
+      "learning_rate": 9.16814179021037e-06,
+      "loss": 1.1735,
+      "step": 392
+    },
+    {
+      "epoch": 1.0824175824175823,
+      "grad_norm": 0.10936509072780609,
+      "learning_rate": 9.154848228941607e-06,
+      "loss": 1.1734,
+      "step": 394
+    },
+    {
+      "epoch": 1.0879120879120878,
+      "grad_norm": 0.0885339230298996,
+      "learning_rate": 9.14145908669329e-06,
+      "loss": 1.1702,
+      "step": 396
+    },
+    {
+      "epoch": 1.0934065934065935,
+      "grad_norm": 0.07241977006196976,
+      "learning_rate": 9.127974671478432e-06,
+      "loss": 1.1714,
+      "step": 398
+    },
+    {
+      "epoch": 1.098901098901099,
+      "grad_norm": 0.08955956995487213,
+      "learning_rate": 9.114395293501775e-06,
+      "loss": 1.1712,
+      "step": 400
+    },
+    {
+      "epoch": 1.1043956043956045,
+      "grad_norm": 0.11948683112859726,
+      "learning_rate": 9.100721265152644e-06,
+      "loss": 1.1799,
+      "step": 402
+    },
+    {
+      "epoch": 1.10989010989011,
+      "grad_norm": 0.11843076348304749,
+      "learning_rate": 9.086952900997774e-06,
+      "loss": 1.1716,
+      "step": 404
+    },
+    {
+      "epoch": 1.1153846153846154,
+      "grad_norm": 0.08606066554784775,
+      "learning_rate": 9.073090517774057e-06,
+      "loss": 1.1777,
+      "step": 406
+    },
+    {
+      "epoch": 1.120879120879121,
+      "grad_norm": 0.08759862184524536,
+      "learning_rate": 9.059134434381274e-06,
+      "loss": 1.1698,
+      "step": 408
+    },
+    {
+      "epoch": 1.1263736263736264,
+      "grad_norm": 0.10086913406848907,
+      "learning_rate": 9.045084971874738e-06,
+      "loss": 1.1641,
+      "step": 410
+    },
+    {
+      "epoch": 1.1318681318681318,
+      "grad_norm": 0.11529091000556946,
+      "learning_rate": 9.030942453457928e-06,
+      "loss": 1.1655,
+      "step": 412
+    },
+    {
+      "epoch": 1.1373626373626373,
+      "grad_norm": 0.0781669095158577,
+      "learning_rate": 9.01670720447504e-06,
+      "loss": 1.1678,
+      "step": 414
+    },
+    {
+      "epoch": 1.1428571428571428,
+      "grad_norm": 0.09989538788795471,
+      "learning_rate": 9.00237955240351e-06,
+      "loss": 1.1689,
+      "step": 416
+    },
+    {
+      "epoch": 1.1483516483516483,
+      "grad_norm": 0.12073778361082077,
+      "learning_rate": 8.987959826846479e-06,
+      "loss": 1.1747,
+      "step": 418
+    },
+    {
+      "epoch": 1.1538461538461537,
+      "grad_norm": 0.12605726718902588,
+      "learning_rate": 8.973448359525207e-06,
+      "loss": 1.1725,
+      "step": 420
+    },
+    {
+      "epoch": 1.1593406593406592,
+      "grad_norm": 0.12503278255462646,
+      "learning_rate": 8.958845484271443e-06,
+      "loss": 1.1709,
+      "step": 422
+    },
+    {
+      "epoch": 1.164835164835165,
+      "grad_norm": 0.10487185418605804,
+      "learning_rate": 8.944151537019752e-06,
+      "loss": 1.1566,
+      "step": 424
+    },
+    {
+      "epoch": 1.1703296703296704,
+      "grad_norm": 0.09222570061683655,
+      "learning_rate": 8.929366855799777e-06,
+      "loss": 1.1808,
+      "step": 426
+    },
+    {
+      "epoch": 1.1758241758241759,
+      "grad_norm": 0.10268251597881317,
+      "learning_rate": 8.914491780728471e-06,
+      "loss": 1.1782,
+      "step": 428
+    },
+    {
+      "epoch": 1.1813186813186813,
+      "grad_norm": 0.08701249957084656,
+      "learning_rate": 8.899526654002268e-06,
+      "loss": 1.1716,
+      "step": 430
+    },
+    {
+      "epoch": 1.1868131868131868,
+      "grad_norm": 0.11233299225568771,
+      "learning_rate": 8.88447181988921e-06,
+      "loss": 1.1699,
+      "step": 432
+    },
+    {
+      "epoch": 1.1923076923076923,
+      "grad_norm": 0.09972134232521057,
+      "learning_rate": 8.869327624721033e-06,
+      "loss": 1.1687,
+      "step": 434
+    },
+    {
+      "epoch": 1.1978021978021978,
+      "grad_norm": 0.10448331385850906,
+      "learning_rate": 8.854094416885192e-06,
+      "loss": 1.1656,
+      "step": 436
+    },
+    {
+      "epoch": 1.2032967032967032,
+      "grad_norm": 0.11365893483161926,
+      "learning_rate": 8.838772546816857e-06,
+      "loss": 1.1764,
+      "step": 438
+    },
+    {
+      "epoch": 1.2032967032967032,
+      "eval_loss": 1.1709927320480347,
+      "eval_runtime": 299.4182,
+      "eval_samples_per_second": 17.287,
+      "eval_steps_per_second": 0.271,
+      "step": 438
+    },
+    {
+      "epoch": 1.2087912087912087,
+      "grad_norm": 0.1238979920744896,
+      "learning_rate": 8.823362366990833e-06,
+      "loss": 1.1718,
+      "step": 440
+    },
+    {
+      "epoch": 1.2142857142857142,
+      "grad_norm": 0.12039055675268173,
+      "learning_rate": 8.807864231913475e-06,
+      "loss": 1.1686,
+      "step": 442
+    },
+    {
+      "epoch": 1.2197802197802199,
+      "grad_norm": 0.09191784262657166,
+      "learning_rate": 8.792278498114517e-06,
+      "loss": 1.1702,
+      "step": 444
+    },
+    {
+      "epoch": 1.2252747252747254,
+      "grad_norm": 0.08927666395902634,
+      "learning_rate": 8.77660552413887e-06,
+      "loss": 1.1761,
+      "step": 446
+    },
+    {
+      "epoch": 1.2307692307692308,
+      "grad_norm": 0.08614211529493332,
+      "learning_rate": 8.760845670538387e-06,
+      "loss": 1.1743,
+      "step": 448
+    },
+    {
+      "epoch": 1.2362637362637363,
+      "grad_norm": 0.10282223671674728,
+      "learning_rate": 8.744999299863549e-06,
+      "loss": 1.1705,
+      "step": 450
+    },
+    {
+      "epoch": 1.2417582417582418,
+      "grad_norm": 0.10722998529672623,
+      "learning_rate": 8.729066776655144e-06,
+      "loss": 1.169,
+      "step": 452
+    },
+    {
+      "epoch": 1.2472527472527473,
+      "grad_norm": 0.09755399823188782,
+      "learning_rate": 8.713048467435865e-06,
+      "loss": 1.1732,
+      "step": 454
+    },
+    {
+      "epoch": 1.2527472527472527,
+      "grad_norm": 0.09675360471010208,
+      "learning_rate": 8.696944740701891e-06,
+      "loss": 1.1696,
+      "step": 456
+    },
+    {
+      "epoch": 1.2582417582417582,
+      "grad_norm": 0.08405736833810806,
+      "learning_rate": 8.6807559669144e-06,
+      "loss": 1.1678,
+      "step": 458
+    },
+    {
+      "epoch": 1.2637362637362637,
+      "grad_norm": 0.08441252261400223,
+      "learning_rate": 8.664482518491053e-06,
+      "loss": 1.1684,
+      "step": 460
+    },
+    {
+      "epoch": 1.2692307692307692,
+      "grad_norm": 0.11046714335680008,
+      "learning_rate": 8.648124769797424e-06,
+      "loss": 1.1747,
+      "step": 462
+    },
+    {
+      "epoch": 1.2747252747252746,
+      "grad_norm": 0.10532383620738983,
+      "learning_rate": 8.631683097138386e-06,
+      "loss": 1.1728,
+      "step": 464
+    },
+    {
+      "epoch": 1.2802197802197801,
+      "grad_norm": 0.09521500766277313,
+      "learning_rate": 8.615157878749462e-06,
+      "loss": 1.169,
+      "step": 466
+    },
+    {
+      "epoch": 1.2857142857142856,
+      "grad_norm": 0.09845519065856934,
+      "learning_rate": 8.598549494788111e-06,
+      "loss": 1.1667,
+      "step": 468
+    },
+    {
+      "epoch": 1.2912087912087913,
+      "grad_norm": 0.10502909123897552,
+      "learning_rate": 8.581858327324996e-06,
+      "loss": 1.1709,
+      "step": 470
+    },
+    {
+      "epoch": 1.2967032967032968,
+      "grad_norm": 0.0786222293972969,
+      "learning_rate": 8.565084760335188e-06,
+      "loss": 1.165,
+      "step": 472
+    },
+    {
+      "epoch": 1.3021978021978022,
+      "grad_norm": 0.10599285364151001,
+      "learning_rate": 8.548229179689325e-06,
+      "loss": 1.1718,
+      "step": 474
+    },
+    {
+      "epoch": 1.3076923076923077,
+      "grad_norm": 0.11965076625347137,
+      "learning_rate": 8.531291973144755e-06,
+      "loss": 1.1683,
+      "step": 476
+    },
+    {
+      "epoch": 1.3131868131868132,
+      "grad_norm": 0.10626693069934845,
+      "learning_rate": 8.5142735303366e-06,
+      "loss": 1.1676,
+      "step": 478
+    },
+    {
+      "epoch": 1.3186813186813187,
+      "grad_norm": 0.10863650590181351,
+      "learning_rate": 8.497174242768792e-06,
+      "loss": 1.17,
+      "step": 480
+    },
+    {
+      "epoch": 1.3241758241758241,
+      "grad_norm": 0.11272372305393219,
+      "learning_rate": 8.479994503805079e-06,
+      "loss": 1.1674,
+      "step": 482
+    },
+    {
+      "epoch": 1.3296703296703296,
+      "grad_norm": 0.1116040050983429,
+      "learning_rate": 8.462734708659959e-06,
+      "loss": 1.1711,
+      "step": 484
+    },
+    {
+      "epoch": 1.335164835164835,
+      "grad_norm": 0.1068933829665184,
+      "learning_rate": 8.445395254389605e-06,
+      "loss": 1.179,
+      "step": 486
+    },
+    {
+      "epoch": 1.3406593406593408,
+      "grad_norm": 0.11034112423658371,
+      "learning_rate": 8.427976539882725e-06,
+      "loss": 1.1617,
+      "step": 488
+    },
+    {
+      "epoch": 1.3461538461538463,
+      "grad_norm": 0.11086605489253998,
+      "learning_rate": 8.410478965851371e-06,
+      "loss": 1.171,
+      "step": 490
+    },
+    {
+      "epoch": 1.3516483516483517,
+      "grad_norm": 0.1053035780787468,
+      "learning_rate": 8.39290293482175e-06,
+      "loss": 1.1696,
+      "step": 492
+    },
+    {
+      "epoch": 1.3571428571428572,
+      "grad_norm": 0.11563800275325775,
+      "learning_rate": 8.375248851124937e-06,
+      "loss": 1.1817,
+      "step": 494
+    },
+    {
+      "epoch": 1.3626373626373627,
+      "grad_norm": 0.09770724177360535,
+      "learning_rate": 8.357517120887586e-06,
+      "loss": 1.1726,
+      "step": 496
+    },
+    {
+      "epoch": 1.3681318681318682,
+      "grad_norm": 0.10480888187885284,
+      "learning_rate": 8.339708152022586e-06,
+      "loss": 1.1716,
+      "step": 498
+    },
+    {
+      "epoch": 1.3736263736263736,
+      "grad_norm": 0.0983305424451828,
+      "learning_rate": 8.321822354219677e-06,
+      "loss": 1.1708,
+      "step": 500
+    },
+    {
+      "epoch": 1.379120879120879,
+      "grad_norm": 0.11572114378213882,
+      "learning_rate": 8.303860138936027e-06,
+      "loss": 1.1669,
+      "step": 502
+    },
+    {
+      "epoch": 1.3846153846153846,
+      "grad_norm": 0.15876273810863495,
+      "learning_rate": 8.285821919386758e-06,
+      "loss": 1.172,
+      "step": 504
+    },
+    {
+      "epoch": 1.39010989010989,
+      "grad_norm": 0.09302208572626114,
+      "learning_rate": 8.267708110535449e-06,
+      "loss": 1.1689,
+      "step": 506
+    },
+    {
+      "epoch": 1.3956043956043955,
+      "grad_norm": 0.0982600748538971,
+      "learning_rate": 8.24951912908459e-06,
+      "loss": 1.1683,
+      "step": 508
+    },
+    {
+      "epoch": 1.401098901098901,
+      "grad_norm": 0.1328704059123993,
+      "learning_rate": 8.231255393465993e-06,
+      "loss": 1.1686,
+      "step": 510
+    },
+    {
+      "epoch": 1.4065934065934065,
+      "grad_norm": 0.1403021216392517,
+      "learning_rate": 8.21291732383116e-06,
+      "loss": 1.1654,
+      "step": 512
+    },
+    {
+      "epoch": 1.412087912087912,
+      "grad_norm": 0.10671450942754745,
+      "learning_rate": 8.19450534204163e-06,
+      "loss": 1.1671,
+      "step": 514
+    },
+    {
+      "epoch": 1.4175824175824177,
+      "grad_norm": 0.10965994000434875,
+      "learning_rate": 8.176019871659263e-06,
+      "loss": 1.1791,
+      "step": 516
+    },
+    {
+      "epoch": 1.4230769230769231,
+      "grad_norm": 0.09535852819681168,
+      "learning_rate": 8.157461337936506e-06,
+      "loss": 1.1654,
+      "step": 518
+    },
+    {
+      "epoch": 1.4285714285714286,
+      "grad_norm": 0.09910845011472702,
+      "learning_rate": 8.138830167806601e-06,
+      "loss": 1.1612,
+      "step": 520
+    },
+    {
+      "epoch": 1.434065934065934,
+      "grad_norm": 0.08044712990522385,
+      "learning_rate": 8.120126789873775e-06,
+      "loss": 1.169,
+      "step": 522
+    },
+    {
+      "epoch": 1.4395604395604396,
+      "grad_norm": 0.11208353191614151,
+      "learning_rate": 8.10135163440336e-06,
+      "loss": 1.1676,
+      "step": 524
+    },
+    {
+      "epoch": 1.445054945054945,
+      "grad_norm": 0.09039778262376785,
+      "learning_rate": 8.08250513331192e-06,
+      "loss": 1.1647,
+      "step": 526
+    },
+    {
+      "epoch": 1.4505494505494505,
+      "grad_norm": 0.08799196779727936,
+      "learning_rate": 8.063587720157298e-06,
+      "loss": 1.1594,
+      "step": 528
+    },
+    {
+      "epoch": 1.456043956043956,
+      "grad_norm": 0.09703148156404495,
+      "learning_rate": 8.044599830128643e-06,
+      "loss": 1.1751,
+      "step": 530
+    },
+    {
+      "epoch": 1.4615384615384617,
+      "grad_norm": 0.10245784372091293,
+      "learning_rate": 8.02554190003641e-06,
+      "loss": 1.1746,
+      "step": 532
+    },
+    {
+      "epoch": 1.4670329670329672,
+      "grad_norm": 0.09949609637260437,
+      "learning_rate": 8.006414368302297e-06,
+      "loss": 1.1696,
+      "step": 534
+    },
+    {
+      "epoch": 1.4725274725274726,
+      "grad_norm": 0.11576993763446808,
+      "learning_rate": 7.98721767494917e-06,
+      "loss": 1.1662,
+      "step": 536
+    },
+    {
+      "epoch": 1.478021978021978,
+      "grad_norm": 0.09676701575517654,
+      "learning_rate": 7.967952261590936e-06,
+      "loss": 1.1652,
+      "step": 538
+    },
+    {
+      "epoch": 1.4835164835164836,
+      "grad_norm": 0.10620467364788055,
+      "learning_rate": 7.94861857142238e-06,
+      "loss": 1.1659,
+      "step": 540
+    },
+    {
+      "epoch": 1.489010989010989,
+      "grad_norm": 0.09917207062244415,
+      "learning_rate": 7.929217049208977e-06,
+      "loss": 1.175,
+      "step": 542
+    },
+    {
+      "epoch": 1.4945054945054945,
+      "grad_norm": 0.131545290350914,
+      "learning_rate": 7.90974814127666e-06,
+      "loss": 1.1619,
+      "step": 544
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.16173675656318665,
+      "learning_rate": 7.890212295501542e-06,
+      "loss": 1.1718,
+      "step": 546
+    },
+    {
+      "epoch": 1.5054945054945055,
+      "grad_norm": 0.11190006136894226,
+      "learning_rate": 7.870609961299627e-06,
+      "loss": 1.1707,
+      "step": 548
+    },
+    {
+      "epoch": 1.510989010989011,
+      "grad_norm": 0.0947834774851799,
+      "learning_rate": 7.850941589616458e-06,
+      "loss": 1.1703,
+      "step": 550
+    },
+    {
+      "epoch": 1.5164835164835164,
+      "grad_norm": 0.12736476957798004,
+      "learning_rate": 7.831207632916757e-06,
+      "loss": 1.1743,
+      "step": 552
+    },
+    {
+      "epoch": 1.521978021978022,
+      "grad_norm": 0.17213790118694305,
+      "learning_rate": 7.811408545174001e-06,
+      "loss": 1.1763,
+      "step": 554
+    },
+    {
+      "epoch": 1.5274725274725274,
+      "grad_norm": 0.26195263862609863,
+      "learning_rate": 7.791544781859993e-06,
+      "loss": 1.1741,
+      "step": 556
+    },
+    {
+      "epoch": 1.5329670329670328,
+      "grad_norm": 0.1335473209619522,
+      "learning_rate": 7.771616799934372e-06,
+      "loss": 1.169,
+      "step": 558
+    },
+    {
+      "epoch": 1.5384615384615383,
+      "grad_norm": 0.11248943954706192,
+      "learning_rate": 7.751625057834107e-06,
+      "loss": 1.1611,
+      "step": 560
+    },
+    {
+      "epoch": 1.5439560439560438,
+      "grad_norm": 0.12497013807296753,
+      "learning_rate": 7.731570015462953e-06,
+      "loss": 1.1657,
+      "step": 562
+    },
+    {
+      "epoch": 1.5494505494505495,
+      "grad_norm": 0.10039670765399933,
+      "learning_rate": 7.711452134180865e-06,
+      "loss": 1.1689,
+      "step": 564
+    },
+    {
+      "epoch": 1.554945054945055,
+      "grad_norm": 0.09104856103658676,
+      "learning_rate": 7.691271876793387e-06,
+      "loss": 1.1691,
+      "step": 566
+    },
+    {
+      "epoch": 1.5604395604395604,
+      "grad_norm": 0.10671708732843399,
+      "learning_rate": 7.67102970754101e-06,
+      "loss": 1.1622,
+      "step": 568
+    },
+    {
+      "epoch": 1.565934065934066,
+      "grad_norm": 0.10622277110815048,
+      "learning_rate": 7.65072609208848e-06,
+      "loss": 1.1716,
+      "step": 570
+    },
+    {
+      "epoch": 1.5714285714285714,
+      "grad_norm": 0.09694940596818924,
+      "learning_rate": 7.630361497514104e-06,
+      "loss": 1.1681,
+      "step": 572
+    },
+    {
+      "epoch": 1.5769230769230769,
+      "grad_norm": 0.11600632965564728,
+      "learning_rate": 7.6099363922989845e-06,
+      "loss": 1.169,
+      "step": 574
+    },
+    {
+      "epoch": 1.5824175824175826,
+      "grad_norm": 0.10654427111148834,
+      "learning_rate": 7.5894512463162595e-06,
+      "loss": 1.1711,
+      "step": 576
+    },
+    {
+      "epoch": 1.587912087912088,
+      "grad_norm": 0.08246386051177979,
+      "learning_rate": 7.568906530820281e-06,
+      "loss": 1.1661,
+      "step": 578
+    },
+    {
+      "epoch": 1.5934065934065935,
+      "grad_norm": 0.09424301981925964,
+      "learning_rate": 7.5483027184357825e-06,
+      "loss": 1.166,
+      "step": 580
+    },
+    {
+      "epoch": 1.598901098901099,
+      "grad_norm": 0.08908534795045853,
+      "learning_rate": 7.527640283147003e-06,
+      "loss": 1.1698,
+      "step": 582
+    },
+    {
+      "epoch": 1.6043956043956045,
+      "grad_norm": 0.10692695528268814,
+      "learning_rate": 7.50691970028678e-06,
+      "loss": 1.1701,
+      "step": 584
+    },
+    {
+      "epoch": 1.6043956043956045,
+      "eval_loss": 1.1695584058761597,
+      "eval_runtime": 299.2734,
+      "eval_samples_per_second": 17.295,
+      "eval_steps_per_second": 0.271,
+      "step": 584
+    },
+    {
+      "epoch": 1.60989010989011,
+      "grad_norm": 0.09625241905450821,
+      "learning_rate": 7.486141446525619e-06,
+      "loss": 1.1734,
+      "step": 586
+    },
+    {
+      "epoch": 1.6153846153846154,
+      "grad_norm": 0.09392037242650986,
+      "learning_rate": 7.465305999860728e-06,
+      "loss": 1.1698,
+      "step": 588
+    },
+    {
+      "epoch": 1.620879120879121,
+      "grad_norm": 0.08401469886302948,
+      "learning_rate": 7.444413839605017e-06,
+      "loss": 1.1659,
+      "step": 590
+    },
+    {
+      "epoch": 1.6263736263736264,
+      "grad_norm": 0.11073325574398041,
+      "learning_rate": 7.423465446376079e-06,
+      "loss": 1.166,
+      "step": 592
+    },
+    {
+      "epoch": 1.6318681318681318,
+      "grad_norm": 0.09533084183931351,
+      "learning_rate": 7.402461302085121e-06,
+      "loss": 1.1734,
+      "step": 594
+    },
+    {
+      "epoch": 1.6373626373626373,
+      "grad_norm": 0.08371090888977051,
+      "learning_rate": 7.381401889925894e-06,
+      "loss": 1.167,
+      "step": 596
+    },
+    {
+      "epoch": 1.6428571428571428,
+      "grad_norm": 0.0968126654624939,
+      "learning_rate": 7.360287694363566e-06,
+      "loss": 1.1679,
+      "step": 598
+    },
+    {
+      "epoch": 1.6483516483516483,
+      "grad_norm": 0.09040292352437973,
+      "learning_rate": 7.3391192011235764e-06,
+      "loss": 1.1754,
+      "step": 600
+    },
+    {
+      "epoch": 1.6538461538461537,
+      "grad_norm": 0.09428130835294724,
+      "learning_rate": 7.317896897180472e-06,
+      "loss": 1.1713,
+      "step": 602
+    },
+    {
+      "epoch": 1.6593406593406592,
+      "grad_norm": 0.08703873306512833,
+      "learning_rate": 7.296621270746691e-06,
+      "loss": 1.1688,
+      "step": 604
+    },
+    {
+      "epoch": 1.6648351648351647,
+      "grad_norm": 0.09342585504055023,
+      "learning_rate": 7.275292811261346e-06,
+      "loss": 1.1651,
+      "step": 606
+    },
+    {
+      "epoch": 1.6703296703296702,
+      "grad_norm": 0.11370475590229034,
+      "learning_rate": 7.253912009378953e-06,
+      "loss": 1.1721,
+      "step": 608
+    },
+    {
+      "epoch": 1.6758241758241759,
+      "grad_norm": 0.0810871347784996,
+      "learning_rate": 7.2324793569581474e-06,
+      "loss": 1.174,
+      "step": 610
+    },
+    {
+      "epoch": 1.6813186813186813,
+      "grad_norm": 0.09331346303224564,
+      "learning_rate": 7.210995347050372e-06,
+      "loss": 1.1691,
+      "step": 612
+    },
+    {
+      "epoch": 1.6868131868131868,
+      "grad_norm": 0.1314326524734497,
+      "learning_rate": 7.189460473888535e-06,
+      "loss": 1.1672,
+      "step": 614
+    },
+    {
+      "epoch": 1.6923076923076923,
+      "grad_norm": 0.12525762617588043,
+      "learning_rate": 7.167875232875632e-06,
+      "loss": 1.1645,
+      "step": 616
+    },
+    {
+      "epoch": 1.6978021978021978,
+      "grad_norm": 0.13776274025440216,
+      "learning_rate": 7.146240120573358e-06,
+      "loss": 1.1693,
+      "step": 618
+    },
+    {
+      "epoch": 1.7032967032967035,
+      "grad_norm": 0.1336820125579834,
+      "learning_rate": 7.124555634690684e-06,
+      "loss": 1.1649,
+      "step": 620
+    },
+    {
+      "epoch": 1.708791208791209,
+      "grad_norm": 0.12508928775787354,
+      "learning_rate": 7.1028222740724e-06,
+      "loss": 1.169,
+      "step": 622
+    },
+    {
+      "epoch": 1.7142857142857144,
+      "grad_norm": 0.10577461123466492,
+      "learning_rate": 7.081040538687649e-06,
+      "loss": 1.1575,
+      "step": 624
+    },
+    {
+      "epoch": 1.7197802197802199,
+      "grad_norm": 0.11875636875629425,
+      "learning_rate": 7.059210929618416e-06,
+      "loss": 1.1666,
+      "step": 626
+    },
+    {
+      "epoch": 1.7252747252747254,
+      "grad_norm": 0.10840770602226257,
+      "learning_rate": 7.037333949048005e-06,
+      "loss": 1.1664,
+      "step": 628
+    },
+    {
+      "epoch": 1.7307692307692308,
+      "grad_norm": 0.11352021992206573,
+      "learning_rate": 7.0154101002494914e-06,
+      "loss": 1.1643,
+      "step": 630
+    },
+    {
+      "epoch": 1.7362637362637363,
+      "grad_norm": 0.08095169812440872,
+      "learning_rate": 6.993439887574133e-06,
+      "loss": 1.1691,
+      "step": 632
+    },
+    {
+      "epoch": 1.7417582417582418,
+      "grad_norm": 0.09175344556570053,
+      "learning_rate": 6.971423816439782e-06,
+      "loss": 1.1664,
+      "step": 634
+    },
+    {
+      "epoch": 1.7472527472527473,
+      "grad_norm": 0.07321044057607651,
+      "learning_rate": 6.949362393319239e-06,
+      "loss": 1.1649,
+      "step": 636
+    },
+    {
+      "epoch": 1.7527472527472527,
+      "grad_norm": 0.10566896945238113,
+      "learning_rate": 6.927256125728624e-06,
+      "loss": 1.167,
+      "step": 638
+    },
+    {
+      "epoch": 1.7582417582417582,
+      "grad_norm": 0.10230179876089096,
+      "learning_rate": 6.905105522215684e-06,
+      "loss": 1.1569,
+      "step": 640
+    },
+    {
+      "epoch": 1.7637362637362637,
+      "grad_norm": 0.09293391555547714,
+      "learning_rate": 6.8829110923481e-06,
+      "loss": 1.1677,
+      "step": 642
+    },
+    {
+      "epoch": 1.7692307692307692,
+      "grad_norm": 0.08087108284235,
+      "learning_rate": 6.8606733467017675e-06,
+      "loss": 1.1691,
+      "step": 644
+    },
+    {
+      "epoch": 1.7747252747252746,
+      "grad_norm": 0.08191601186990738,
+      "learning_rate": 6.838392796849042e-06,
+      "loss": 1.1683,
+      "step": 646
+    },
+    {
+      "epoch": 1.7802197802197801,
+      "grad_norm": 0.08746813237667084,
+      "learning_rate": 6.816069955346986e-06,
+      "loss": 1.1643,
+      "step": 648
+    },
+    {
+      "epoch": 1.7857142857142856,
+      "grad_norm": 0.10125511139631271,
+      "learning_rate": 6.7937053357255585e-06,
+      "loss": 1.1716,
+      "step": 650
+    },
+    {
+      "epoch": 1.791208791208791,
+      "grad_norm": 0.08800845593214035,
+      "learning_rate": 6.771299452475818e-06,
+      "loss": 1.1721,
+      "step": 652
+    },
+    {
+      "epoch": 1.7967032967032965,
+      "grad_norm": 0.09477917104959488,
+      "learning_rate": 6.748852821038075e-06,
+      "loss": 1.1649,
+      "step": 654
+    },
+    {
+      "epoch": 1.8021978021978022,
+      "grad_norm": 0.10864371061325073,
+      "learning_rate": 6.7263659577900375e-06,
+      "loss": 1.1621,
+      "step": 656
+    },
+    {
+      "epoch": 1.8076923076923077,
+      "grad_norm": 0.07818944752216339,
+      "learning_rate": 6.703839380034945e-06,
+      "loss": 1.1667,
+      "step": 658
+    },
+    {
+      "epoch": 1.8131868131868132,
+      "grad_norm": 0.0869673416018486,
+      "learning_rate": 6.681273605989643e-06,
+      "loss": 1.1692,
+      "step": 660
+    },
+    {
+      "epoch": 1.8186813186813187,
+      "grad_norm": 0.10159242898225784,
+      "learning_rate": 6.6586691547726855e-06,
+      "loss": 1.1708,
+      "step": 662
+    },
+    {
+      "epoch": 1.8241758241758241,
+      "grad_norm": 0.20149534940719604,
+      "learning_rate": 6.636026546392374e-06,
+      "loss": 1.175,
+      "step": 664
+    },
+    {
+      "epoch": 1.8296703296703298,
+      "grad_norm": 0.35707324743270874,
+      "learning_rate": 6.613346301734813e-06,
+      "loss": 1.1556,
+      "step": 666
+    },
+    {
+      "epoch": 1.8351648351648353,
+      "grad_norm": 0.10949152708053589,
+      "learning_rate": 6.590628942551909e-06,
+      "loss": 1.1696,
+      "step": 668
+    },
+    {
+      "epoch": 1.8406593406593408,
+      "grad_norm": 0.07581349462270737,
+      "learning_rate": 6.567874991449383e-06,
+      "loss": 1.1701,
+      "step": 670
+    },
+    {
+      "epoch": 1.8461538461538463,
+      "grad_norm": 0.11389974504709244,
+      "learning_rate": 6.545084971874738e-06,
+      "loss": 1.1721,
+      "step": 672
+    },
+    {
+      "epoch": 1.8516483516483517,
+      "grad_norm": 0.10378382354974747,
+      "learning_rate": 6.522259408105223e-06,
+      "loss": 1.1674,
+      "step": 674
+    },
+    {
+      "epoch": 1.8571428571428572,
+      "grad_norm": 0.09723298251628876,
+      "learning_rate": 6.499398825235767e-06,
+      "loss": 1.1696,
+      "step": 676
+    },
+    {
+      "epoch": 1.8626373626373627,
+      "grad_norm": 0.10554076731204987,
+      "learning_rate": 6.476503749166903e-06,
+      "loss": 1.1674,
+      "step": 678
+    },
+    {
+      "epoch": 1.8681318681318682,
+      "grad_norm": 0.09231211990118027,
+      "learning_rate": 6.453574706592676e-06,
+      "loss": 1.1652,
+      "step": 680
+    },
+    {
+      "epoch": 1.8736263736263736,
+      "grad_norm": 0.09046723693609238,
+      "learning_rate": 6.4306122249885105e-06,
+      "loss": 1.1716,
+      "step": 682
+    },
+    {
+      "epoch": 1.879120879120879,
+      "grad_norm": 0.09619590640068054,
+      "learning_rate": 6.407616832599091e-06,
+      "loss": 1.1699,
+      "step": 684
+    },
+    {
+      "epoch": 1.8846153846153846,
+      "grad_norm": 0.08878965675830841,
+      "learning_rate": 6.384589058426201e-06,
+      "loss": 1.1701,
+      "step": 686
+    },
+    {
+      "epoch": 1.89010989010989,
+      "grad_norm": 0.0882963240146637,
+      "learning_rate": 6.36152943221656e-06,
+      "loss": 1.1757,
+      "step": 688
+    },
+    {
+      "epoch": 1.8956043956043955,
+      "grad_norm": 0.10285267233848572,
+      "learning_rate": 6.338438484449632e-06,
+      "loss": 1.1612,
+      "step": 690
+    },
+    {
+      "epoch": 1.901098901098901,
+      "grad_norm": 0.08342345803976059,
+      "learning_rate": 6.31531674632542e-06,
+      "loss": 1.1696,
+      "step": 692
+    },
+    {
+      "epoch": 1.9065934065934065,
+      "grad_norm": 0.09322002530097961,
+      "learning_rate": 6.292164749752256e-06,
+      "loss": 1.1733,
+      "step": 694
+    },
+    {
+      "epoch": 1.912087912087912,
+      "grad_norm": 0.10594601929187775,
+      "learning_rate": 6.268983027334557e-06,
+      "loss": 1.1675,
+      "step": 696
+    },
+    {
+      "epoch": 1.9175824175824174,
+      "grad_norm": 0.10415042191743851,
+      "learning_rate": 6.245772112360568e-06,
+      "loss": 1.1706,
+      "step": 698
+    },
+    {
+      "epoch": 1.9230769230769231,
+      "grad_norm": 0.08474233746528625,
+      "learning_rate": 6.222532538790107e-06,
+      "loss": 1.1717,
+      "step": 700
+    },
+    {
+      "epoch": 1.9285714285714286,
+      "grad_norm": 0.09690573066473007,
+      "learning_rate": 6.199264841242267e-06,
+      "loss": 1.1579,
+      "step": 702
+    },
+    {
+      "epoch": 1.934065934065934,
+      "grad_norm": 0.09174709767103195,
+      "learning_rate": 6.17596955498313e-06,
+      "loss": 1.1707,
+      "step": 704
+    },
+    {
+      "epoch": 1.9395604395604396,
+      "grad_norm": 0.09256250411272049,
+      "learning_rate": 6.1526472159134454e-06,
+      "loss": 1.1675,
+      "step": 706
+    },
+    {
+      "epoch": 1.945054945054945,
+      "grad_norm": 0.115385040640831,
+      "learning_rate": 6.129298360556304e-06,
+      "loss": 1.1666,
+      "step": 708
+    },
+    {
+      "epoch": 1.9505494505494505,
+      "grad_norm": 0.09870624542236328,
+      "learning_rate": 6.105923526044794e-06,
+      "loss": 1.1712,
+      "step": 710
+    },
+    {
+      "epoch": 1.9560439560439562,
+      "grad_norm": 0.09220367670059204,
+      "learning_rate": 6.08252325010965e-06,
+      "loss": 1.1702,
+      "step": 712
+    },
+    {
+      "epoch": 1.9615384615384617,
+      "grad_norm": 0.08229216188192368,
+      "learning_rate": 6.059098071066874e-06,
+      "loss": 1.1641,
+      "step": 714
+    },
+    {
+      "epoch": 1.9670329670329672,
+      "grad_norm": 0.08601760119199753,
+      "learning_rate": 6.035648527805359e-06,
+      "loss": 1.1658,
+      "step": 716
+    },
+    {
+      "epoch": 1.9725274725274726,
+      "grad_norm": 0.09920581430196762,
+      "learning_rate": 6.012175159774488e-06,
+      "loss": 1.1627,
+      "step": 718
+    },
+    {
+      "epoch": 1.978021978021978,
+      "grad_norm": 0.09022627770900726,
+      "learning_rate": 5.988678506971726e-06,
+      "loss": 1.1723,
+      "step": 720
+    },
+    {
+      "epoch": 1.9835164835164836,
+      "grad_norm": 0.08053141087293625,
+      "learning_rate": 5.965159109930196e-06,
+      "loss": 1.1785,
+      "step": 722
+    },
+    {
+      "epoch": 1.989010989010989,
+      "grad_norm": 0.07670550793409348,
+      "learning_rate": 5.941617509706247e-06,
+      "loss": 1.1606,
+      "step": 724
+    },
+    {
+      "epoch": 1.9945054945054945,
+      "grad_norm": 0.0779750868678093,
+      "learning_rate": 5.9180542478670025e-06,
+      "loss": 1.1624,
+      "step": 726
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.0955984815955162,
+      "learning_rate": 5.894469866477905e-06,
+      "loss": 1.1647,
+      "step": 728
+    },
+    {
+      "epoch": 2.0054945054945055,
+      "grad_norm": 0.08561510592699051,
+      "learning_rate": 5.87086490809025e-06,
+      "loss": 1.1675,
+      "step": 730
+    },
+    {
+      "epoch": 2.0054945054945055,
+      "eval_loss": 1.168563961982727,
+      "eval_runtime": 298.1262,
+      "eval_samples_per_second": 17.362,
+      "eval_steps_per_second": 0.272,
+      "step": 730
+    },
+    {
+      "epoch": 2.010989010989011,
+      "grad_norm": 0.08887989073991776,
+      "learning_rate": 5.847239915728695e-06,
+      "loss": 1.1711,
+      "step": 732
+    },
+    {
+      "epoch": 2.0164835164835164,
+      "grad_norm": 0.1793489307165146,
+      "learning_rate": 5.823595432878775e-06,
+      "loss": 1.1689,
+      "step": 734
+    },
+    {
+      "epoch": 2.021978021978022,
+      "grad_norm": 0.14795880019664764,
+      "learning_rate": 5.799932003474398e-06,
+      "loss": 1.1692,
+      "step": 736
+    },
+    {
+      "epoch": 2.0274725274725274,
+      "grad_norm": 0.08862569183111191,
+      "learning_rate": 5.776250171885329e-06,
+      "loss": 1.1653,
+      "step": 738
+    },
+    {
+      "epoch": 2.032967032967033,
+      "grad_norm": 0.10424422472715378,
+      "learning_rate": 5.752550482904674e-06,
+      "loss": 1.1651,
+      "step": 740
+    },
+    {
+      "epoch": 2.0384615384615383,
+      "grad_norm": 0.0930488184094429,
+      "learning_rate": 5.728833481736339e-06,
+      "loss": 1.1666,
+      "step": 742
+    },
+    {
+      "epoch": 2.043956043956044,
+      "grad_norm": 0.1007128432393074,
+      "learning_rate": 5.705099713982491e-06,
+      "loss": 1.1737,
+      "step": 744
+    },
+    {
+      "epoch": 2.0494505494505493,
+      "grad_norm": 0.0997442975640297,
+      "learning_rate": 5.6813497256310124e-06,
+      "loss": 1.1701,
+      "step": 746
+    },
+    {
+      "epoch": 2.0549450549450547,
+      "grad_norm": 0.09021608531475067,
+      "learning_rate": 5.6575840630429295e-06,
+      "loss": 1.1677,
+      "step": 748
+    },
+    {
+      "epoch": 2.0604395604395602,
+      "grad_norm": 0.08044509589672089,
+      "learning_rate": 5.633803272939851e-06,
+      "loss": 1.1696,
+      "step": 750
+    },
+    {
+      "epoch": 2.065934065934066,
+      "grad_norm": 0.08909393101930618,
+      "learning_rate": 5.610007902391387e-06,
+      "loss": 1.1616,
+      "step": 752
+    },
+    {
+      "epoch": 2.0714285714285716,
+      "grad_norm": 0.08932027965784073,
+      "learning_rate": 5.586198498802577e-06,
+      "loss": 1.1623,
+      "step": 754
+    },
+    {
+      "epoch": 2.076923076923077,
+      "grad_norm": 0.1681545227766037,
+      "learning_rate": 5.562375609901273e-06,
+      "loss": 1.1654,
+      "step": 756
+    },
+    {
+      "epoch": 2.0824175824175826,
+      "grad_norm": 0.1894630342721939,
+      "learning_rate": 5.538539783725556e-06,
+      "loss": 1.1681,
+      "step": 758
+    },
+    {
+      "epoch": 2.087912087912088,
+      "grad_norm": 0.11349444836378098,
+      "learning_rate": 5.51469156861113e-06,
+      "loss": 1.1688,
+      "step": 760
+    },
+    {
+      "epoch": 2.0934065934065935,
+      "grad_norm": 0.08764316141605377,
+      "learning_rate": 5.490831513178698e-06,
+      "loss": 1.1718,
+      "step": 762
+    },
+    {
+      "epoch": 2.098901098901099,
+      "grad_norm": 0.08182983100414276,
+      "learning_rate": 5.466960166321348e-06,
+      "loss": 1.1698,
+      "step": 764
+    },
+    {
+      "epoch": 2.1043956043956045,
+      "grad_norm": 0.08446374535560608,
+      "learning_rate": 5.44307807719192e-06,
+      "loss": 1.1605,
+      "step": 766
+    },
+    {
+      "epoch": 2.10989010989011,
+      "grad_norm": 0.06986381858587265,
+      "learning_rate": 5.4191857951903825e-06,
+      "loss": 1.1661,
+      "step": 768
+    },
+    {
+      "epoch": 2.1153846153846154,
+      "grad_norm": 0.09340260177850723,
+      "learning_rate": 5.395283869951184e-06,
+      "loss": 1.1685,
+      "step": 770
+    },
+    {
+      "epoch": 2.120879120879121,
+      "grad_norm": 0.10494975745677948,
+      "learning_rate": 5.371372851330612e-06,
+      "loss": 1.1748,
+      "step": 772
+    },
+    {
+      "epoch": 2.1263736263736264,
+      "grad_norm": 0.09624402970075607,
+      "learning_rate": 5.347453289394146e-06,
+      "loss": 1.1683,
+      "step": 774
+    },
+    {
+      "epoch": 2.131868131868132,
+      "grad_norm": 0.09158363193273544,
+      "learning_rate": 5.3235257344037996e-06,
+      "loss": 1.1693,
+      "step": 776
+    },
+    {
+      "epoch": 2.1373626373626373,
+      "grad_norm": 0.07997617870569229,
+      "learning_rate": 5.29959073680547e-06,
+      "loss": 1.1671,
+      "step": 778
+    },
+    {
+      "epoch": 2.142857142857143,
+      "grad_norm": 0.08307652175426483,
+      "learning_rate": 5.275648847216263e-06,
+      "loss": 1.1648,
+      "step": 780
+    },
+    {
+      "epoch": 2.1483516483516483,
+      "grad_norm": 0.09365742653608322,
+      "learning_rate": 5.251700616411836e-06,
+      "loss": 1.1738,
+      "step": 782
+    },
+    {
+      "epoch": 2.1538461538461537,
+      "grad_norm": 0.09766220301389694,
+      "learning_rate": 5.22774659531372e-06,
+      "loss": 1.1649,
+      "step": 784
+    },
+    {
+      "epoch": 2.159340659340659,
+      "grad_norm": 0.10807793587446213,
+      "learning_rate": 5.203787334976655e-06,
+      "loss": 1.1728,
+      "step": 786
+    },
+    {
+      "epoch": 2.1648351648351647,
+      "grad_norm": 0.07477546483278275,
+      "learning_rate": 5.179823386575908e-06,
+      "loss": 1.1701,
+      "step": 788
+    },
+    {
+      "epoch": 2.17032967032967,
+      "grad_norm": 0.08150489628314972,
+      "learning_rate": 5.155855301394585e-06,
+      "loss": 1.1672,
+      "step": 790
+    },
+    {
+      "epoch": 2.1758241758241756,
+      "grad_norm": 0.11139123886823654,
+      "learning_rate": 5.131883630810966e-06,
+      "loss": 1.172,
+      "step": 792
+    },
+    {
+      "epoch": 2.181318681318681,
+      "grad_norm": 0.09428671002388,
+      "learning_rate": 5.107908926285813e-06,
+      "loss": 1.1662,
+      "step": 794
+    },
+    {
+      "epoch": 2.186813186813187,
+      "grad_norm": 0.11421363055706024,
+      "learning_rate": 5.083931739349675e-06,
+      "loss": 1.1666,
+      "step": 796
+    },
+    {
+      "epoch": 2.1923076923076925,
+      "grad_norm": 0.08080089837312698,
+      "learning_rate": 5.059952621590216e-06,
+      "loss": 1.1681,
+      "step": 798
+    },
+    {
+      "epoch": 2.197802197802198,
+      "grad_norm": 0.09634223580360413,
+      "learning_rate": 5.035972124639511e-06,
+      "loss": 1.1654,
+      "step": 800
+    },
+    {
+      "epoch": 2.2032967032967035,
+      "grad_norm": 0.08151613175868988,
+      "learning_rate": 5.011990800161369e-06,
+      "loss": 1.1668,
+      "step": 802
+    },
+    {
+      "epoch": 2.208791208791209,
+      "grad_norm": 0.09357167035341263,
+      "learning_rate": 4.988009199838632e-06,
+      "loss": 1.1611,
+      "step": 804
+    },
+    {
+      "epoch": 2.2142857142857144,
+      "grad_norm": 0.08290430158376694,
+      "learning_rate": 4.96402787536049e-06,
+      "loss": 1.1562,
+      "step": 806
+    },
+    {
+      "epoch": 2.21978021978022,
+      "grad_norm": 0.09985602647066116,
+      "learning_rate": 4.940047378409786e-06,
+      "loss": 1.1716,
+      "step": 808
+    },
+    {
+      "epoch": 2.2252747252747254,
+      "grad_norm": 0.08841745555400848,
+      "learning_rate": 4.9160682606503255e-06,
+      "loss": 1.1723,
+      "step": 810
+    },
+    {
+      "epoch": 2.230769230769231,
+      "grad_norm": 0.08120245486497879,
+      "learning_rate": 4.892091073714189e-06,
+      "loss": 1.1642,
+      "step": 812
+    },
+    {
+      "epoch": 2.2362637362637363,
+      "grad_norm": 0.08893705904483795,
+      "learning_rate": 4.868116369189033e-06,
+      "loss": 1.1614,
+      "step": 814
+    },
+    {
+      "epoch": 2.241758241758242,
+      "grad_norm": 0.16162922978401184,
+      "learning_rate": 4.844144698605418e-06,
+      "loss": 1.1682,
+      "step": 816
+    },
+    {
+      "epoch": 2.2472527472527473,
+      "grad_norm": 0.08067131042480469,
+      "learning_rate": 4.820176613424095e-06,
+      "loss": 1.1676,
+      "step": 818
+    },
+    {
+      "epoch": 2.2527472527472527,
+      "grad_norm": 0.11158166825771332,
+      "learning_rate": 4.796212665023345e-06,
+      "loss": 1.1731,
+      "step": 820
+    },
+    {
+      "epoch": 2.258241758241758,
+      "grad_norm": 0.08565322309732437,
+      "learning_rate": 4.7722534046862805e-06,
+      "loss": 1.1624,
+      "step": 822
+    },
+    {
+      "epoch": 2.2637362637362637,
+      "grad_norm": 0.09021101146936417,
+      "learning_rate": 4.748299383588167e-06,
+      "loss": 1.1649,
+      "step": 824
+    },
+    {
+      "epoch": 2.269230769230769,
+      "grad_norm": 0.09416112303733826,
+      "learning_rate": 4.7243511527837374e-06,
+      "loss": 1.1643,
+      "step": 826
+    },
+    {
+      "epoch": 2.2747252747252746,
+      "grad_norm": 0.10305804759263992,
+      "learning_rate": 4.7004092631945315e-06,
+      "loss": 1.1709,
+      "step": 828
+    },
+    {
+      "epoch": 2.28021978021978,
+      "grad_norm": 0.11213865131139755,
+      "learning_rate": 4.6764742655962e-06,
+      "loss": 1.1684,
+      "step": 830
+    },
+    {
+      "epoch": 2.2857142857142856,
+      "grad_norm": 0.10108992457389832,
+      "learning_rate": 4.652546710605857e-06,
+      "loss": 1.1619,
+      "step": 832
+    },
+    {
+      "epoch": 2.291208791208791,
+      "grad_norm": 0.08697344362735748,
+      "learning_rate": 4.628627148669391e-06,
+      "loss": 1.1686,
+      "step": 834
+    },
+    {
+      "epoch": 2.2967032967032965,
+      "grad_norm": 0.10428237169981003,
+      "learning_rate": 4.604716130048818e-06,
+      "loss": 1.1668,
+      "step": 836
+    },
+    {
+      "epoch": 2.302197802197802,
+      "grad_norm": 0.08793749660253525,
+      "learning_rate": 4.580814204809618e-06,
+      "loss": 1.1713,
+      "step": 838
+    },
+    {
+      "epoch": 2.3076923076923075,
+      "grad_norm": 0.11877471953630447,
+      "learning_rate": 4.5569219228080805e-06,
+      "loss": 1.162,
+      "step": 840
+    },
+    {
+      "epoch": 2.313186813186813,
+      "grad_norm": 0.08249559253454208,
+      "learning_rate": 4.5330398336786526e-06,
+      "loss": 1.1678,
+      "step": 842
+    },
+    {
+      "epoch": 2.3186813186813184,
+      "grad_norm": 0.11602500081062317,
+      "learning_rate": 4.509168486821304e-06,
+      "loss": 1.164,
+      "step": 844
+    },
+    {
+      "epoch": 2.3241758241758244,
+      "grad_norm": 0.11740805953741074,
+      "learning_rate": 4.48530843138887e-06,
+      "loss": 1.1643,
+      "step": 846
+    },
+    {
+      "epoch": 2.32967032967033,
+      "grad_norm": 0.10009398311376572,
+      "learning_rate": 4.4614602162744455e-06,
+      "loss": 1.1629,
+      "step": 848
+    },
+    {
+      "epoch": 2.3351648351648353,
+      "grad_norm": 0.07898547500371933,
+      "learning_rate": 4.4376243900987296e-06,
+      "loss": 1.1568,
+      "step": 850
+    },
+    {
+      "epoch": 2.340659340659341,
+      "grad_norm": 0.08736059069633484,
+      "learning_rate": 4.413801501197424e-06,
+      "loss": 1.1703,
+      "step": 852
+    },
+    {
+      "epoch": 2.3461538461538463,
+      "grad_norm": 0.08960308879613876,
+      "learning_rate": 4.389992097608613e-06,
+      "loss": 1.1632,
+      "step": 854
+    },
+    {
+      "epoch": 2.3516483516483517,
+      "grad_norm": 0.10276441276073456,
+      "learning_rate": 4.366196727060152e-06,
+      "loss": 1.1634,
+      "step": 856
+    },
+    {
+      "epoch": 2.357142857142857,
+      "grad_norm": 0.07904151827096939,
+      "learning_rate": 4.342415936957073e-06,
+      "loss": 1.1709,
+      "step": 858
+    },
+    {
+      "epoch": 2.3626373626373627,
+      "grad_norm": 0.07296542823314667,
+      "learning_rate": 4.318650274368989e-06,
+      "loss": 1.1672,
+      "step": 860
+    },
+    {
+      "epoch": 2.368131868131868,
+      "grad_norm": 0.08600784838199615,
+      "learning_rate": 4.294900286017509e-06,
+      "loss": 1.1612,
+      "step": 862
+    },
+    {
+      "epoch": 2.3736263736263736,
+      "grad_norm": 0.08018805086612701,
+      "learning_rate": 4.271166518263662e-06,
+      "loss": 1.1722,
+      "step": 864
+    },
+    {
+      "epoch": 2.379120879120879,
+      "grad_norm": 0.07395070046186447,
+      "learning_rate": 4.247449517095329e-06,
+      "loss": 1.1711,
+      "step": 866
+    },
+    {
+      "epoch": 2.3846153846153846,
+      "grad_norm": 0.07601243257522583,
+      "learning_rate": 4.223749828114672e-06,
+      "loss": 1.1771,
+      "step": 868
+    },
+    {
+      "epoch": 2.39010989010989,
+      "grad_norm": 0.09423944354057312,
+      "learning_rate": 4.2000679965256045e-06,
+      "loss": 1.1603,
+      "step": 870
+    },
+    {
+      "epoch": 2.3956043956043955,
+      "grad_norm": 0.08968156576156616,
+      "learning_rate": 4.176404567121225e-06,
+      "loss": 1.1608,
+      "step": 872
+    },
+    {
+      "epoch": 2.401098901098901,
+      "grad_norm": 0.10625026375055313,
+      "learning_rate": 4.152760084271305e-06,
+      "loss": 1.1574,
+      "step": 874
+    },
+    {
+      "epoch": 2.4065934065934065,
+      "grad_norm": 0.18610498309135437,
+      "learning_rate": 4.129135091909752e-06,
+      "loss": 1.1747,
+      "step": 876
+    },
+    {
+      "epoch": 2.4065934065934065,
+      "eval_loss": 1.1678768396377563,
+      "eval_runtime": 299.619,
+      "eval_samples_per_second": 17.275,
+      "eval_steps_per_second": 0.27,
+      "step": 876
+    },
+    {
+      "epoch": 2.412087912087912,
+      "grad_norm": 0.15964192152023315,
+      "learning_rate": 4.105530133522096e-06,
+      "loss": 1.1678,
+      "step": 878
+    },
+    {
+      "epoch": 2.4175824175824174,
+      "grad_norm": 0.11698123812675476,
+      "learning_rate": 4.081945752133e-06,
+      "loss": 1.1678,
+      "step": 880
+    },
+    {
+      "epoch": 2.423076923076923,
+      "grad_norm": 0.16632910072803497,
+      "learning_rate": 4.058382490293755e-06,
+      "loss": 1.1722,
+      "step": 882
+    },
+    {
+      "epoch": 2.4285714285714284,
+      "grad_norm": 0.17577840387821198,
+      "learning_rate": 4.034840890069805e-06,
+      "loss": 1.1643,
+      "step": 884
+    },
+    {
+      "epoch": 2.4340659340659343,
+      "grad_norm": 0.17302070558071136,
+      "learning_rate": 4.0113214930282765e-06,
+      "loss": 1.1641,
+      "step": 886
+    },
+    {
+      "epoch": 2.4395604395604398,
+      "grad_norm": 0.2563433349132538,
+      "learning_rate": 3.987824840225512e-06,
+      "loss": 1.1678,
+      "step": 888
+    },
+    {
+      "epoch": 2.4450549450549453,
+      "grad_norm": 0.19088977575302124,
+      "learning_rate": 3.964351472194642e-06,
+      "loss": 1.174,
+      "step": 890
+    },
+    {
+      "epoch": 2.4505494505494507,
+      "grad_norm": 0.22029054164886475,
+      "learning_rate": 3.940901928933127e-06,
+      "loss": 1.1641,
+      "step": 892
+    },
+    {
+      "epoch": 2.456043956043956,
+      "grad_norm": 0.26073768734931946,
+      "learning_rate": 3.917476749890351e-06,
+      "loss": 1.1594,
+      "step": 894
+    },
+    {
+      "epoch": 2.4615384615384617,
+      "grad_norm": 0.14887195825576782,
+      "learning_rate": 3.894076473955207e-06,
+      "loss": 1.1696,
+      "step": 896
+    },
+    {
+      "epoch": 2.467032967032967,
+      "grad_norm": 0.09114201366901398,
+      "learning_rate": 3.8707016394436985e-06,
+      "loss": 1.1686,
+      "step": 898
+    },
+    {
+      "epoch": 2.4725274725274726,
+      "grad_norm": 0.12883096933364868,
+      "learning_rate": 3.847352784086556e-06,
+      "loss": 1.1614,
+      "step": 900
+    },
+    {
+      "epoch": 2.478021978021978,
+      "grad_norm": 0.10688427090644836,
+      "learning_rate": 3.8240304450168716e-06,
+      "loss": 1.1583,
+      "step": 902
+    },
+    {
+      "epoch": 2.4835164835164836,
+      "grad_norm": 0.08104757964611053,
+      "learning_rate": 3.8007351587577342e-06,
+      "loss": 1.1713,
+      "step": 904
+    },
+    {
+      "epoch": 2.489010989010989,
+      "grad_norm": 0.11496758460998535,
+      "learning_rate": 3.777467461209895e-06,
+      "loss": 1.1662,
+      "step": 906
+    },
+    {
+      "epoch": 2.4945054945054945,
+      "grad_norm": 0.08005277067422867,
+      "learning_rate": 3.754227887639434e-06,
+      "loss": 1.1683,
+      "step": 908
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 0.08510690182447433,
+      "learning_rate": 3.7310169726654444e-06,
+      "loss": 1.1603,
+      "step": 910
+    },
+    {
+      "epoch": 2.5054945054945055,
+      "grad_norm": 0.08159387856721878,
+      "learning_rate": 3.707835250247745e-06,
+      "loss": 1.1703,
+      "step": 912
+    },
+    {
+      "epoch": 2.510989010989011,
+      "grad_norm": 0.08646809309720993,
+      "learning_rate": 3.684683253674583e-06,
+      "loss": 1.1744,
+      "step": 914
+    },
+    {
+      "epoch": 2.5164835164835164,
+      "grad_norm": 0.10196442157030106,
+      "learning_rate": 3.6615615155503703e-06,
+      "loss": 1.1607,
+      "step": 916
+    },
+    {
+      "epoch": 2.521978021978022,
+      "grad_norm": 0.07023598998785019,
+      "learning_rate": 3.638470567783442e-06,
+      "loss": 1.17,
+      "step": 918
+    },
+    {
+      "epoch": 2.5274725274725274,
+      "grad_norm": 0.09794013947248459,
+      "learning_rate": 3.615410941573799e-06,
+      "loss": 1.1713,
+      "step": 920
+    },
+    {
+      "epoch": 2.532967032967033,
+      "grad_norm": 0.07678301632404327,
+      "learning_rate": 3.59238316740091e-06,
+      "loss": 1.1673,
+      "step": 922
+    },
+    {
+      "epoch": 2.5384615384615383,
+      "grad_norm": 0.09355033934116364,
+      "learning_rate": 3.5693877750114903e-06,
+      "loss": 1.1639,
+      "step": 924
+    },
+    {
+      "epoch": 2.543956043956044,
+      "grad_norm": 0.08433817327022552,
+      "learning_rate": 3.546425293407324e-06,
+      "loss": 1.1614,
+      "step": 926
+    },
+    {
+      "epoch": 2.5494505494505493,
+      "grad_norm": 0.08144976943731308,
+      "learning_rate": 3.523496250833098e-06,
+      "loss": 1.1599,
+      "step": 928
+    },
+    {
+      "epoch": 2.5549450549450547,
+      "grad_norm": 0.0796588808298111,
+      "learning_rate": 3.5006011747642366e-06,
+      "loss": 1.1667,
+      "step": 930
+    },
+    {
+      "epoch": 2.5604395604395602,
+      "grad_norm": 0.08032579720020294,
+      "learning_rate": 3.4777405918947795e-06,
+      "loss": 1.1612,
+      "step": 932
+    },
+    {
+      "epoch": 2.5659340659340657,
+      "grad_norm": 0.08575133234262466,
+      "learning_rate": 3.4549150281252635e-06,
+      "loss": 1.1695,
+      "step": 934
+    },
+    {
+      "epoch": 2.571428571428571,
+      "grad_norm": 0.08064404129981995,
+      "learning_rate": 3.4321250085506174e-06,
+      "loss": 1.1698,
+      "step": 936
+    },
+    {
+      "epoch": 2.5769230769230766,
+      "grad_norm": 0.07455819100141525,
+      "learning_rate": 3.4093710574480926e-06,
+      "loss": 1.1643,
+      "step": 938
+    },
+    {
+      "epoch": 2.5824175824175826,
+      "grad_norm": 0.08538255095481873,
+      "learning_rate": 3.386653698265189e-06,
+      "loss": 1.1593,
+      "step": 940
+    },
+    {
+      "epoch": 2.587912087912088,
+      "grad_norm": 0.08684508502483368,
+      "learning_rate": 3.3639734536076263e-06,
+      "loss": 1.1651,
+      "step": 942
+    },
+    {
+      "epoch": 2.5934065934065935,
+      "grad_norm": 0.08089859038591385,
+      "learning_rate": 3.341330845227316e-06,
+      "loss": 1.165,
+      "step": 944
+    },
+    {
+      "epoch": 2.598901098901099,
+      "grad_norm": 0.07942517846822739,
+      "learning_rate": 3.3187263940103587e-06,
+      "loss": 1.1685,
+      "step": 946
+    },
+    {
+      "epoch": 2.6043956043956045,
+      "grad_norm": 0.08427475392818451,
+      "learning_rate": 3.296160619965056e-06,
+      "loss": 1.1547,
+      "step": 948
+    },
+    {
+      "epoch": 2.60989010989011,
+      "grad_norm": 0.08193778246641159,
+      "learning_rate": 3.2736340422099633e-06,
+      "loss": 1.1651,
+      "step": 950
+    },
+    {
+      "epoch": 2.6153846153846154,
+      "grad_norm": 0.0785454586148262,
+      "learning_rate": 3.2511471789619274e-06,
+      "loss": 1.169,
+      "step": 952
+    },
+    {
+      "epoch": 2.620879120879121,
+      "grad_norm": 0.08144666999578476,
+      "learning_rate": 3.228700547524184e-06,
+      "loss": 1.1656,
+      "step": 954
+    },
+    {
+      "epoch": 2.6263736263736264,
+      "grad_norm": 0.08215656876564026,
+      "learning_rate": 3.206294664274443e-06,
+      "loss": 1.163,
+      "step": 956
+    },
+    {
+      "epoch": 2.631868131868132,
+      "grad_norm": 0.07643819600343704,
+      "learning_rate": 3.183930044653014e-06,
+      "loss": 1.1609,
+      "step": 958
+    },
+    {
+      "epoch": 2.6373626373626373,
+      "grad_norm": 0.06851556152105331,
+      "learning_rate": 3.1616072031509594e-06,
+      "loss": 1.168,
+      "step": 960
+    },
+    {
+      "epoch": 2.642857142857143,
+      "grad_norm": 0.07472985982894897,
+      "learning_rate": 3.139326653298236e-06,
+      "loss": 1.1672,
+      "step": 962
+    },
+    {
+      "epoch": 2.6483516483516483,
+      "grad_norm": 0.08239593356847763,
+      "learning_rate": 3.117088907651902e-06,
+      "loss": 1.1668,
+      "step": 964
+    },
+    {
+      "epoch": 2.6538461538461537,
+      "grad_norm": 0.07118227332830429,
+      "learning_rate": 3.094894477784318e-06,
+      "loss": 1.163,
+      "step": 966
+    },
+    {
+      "epoch": 2.659340659340659,
+      "grad_norm": 0.08010434359312057,
+      "learning_rate": 3.0727438742713766e-06,
+      "loss": 1.1674,
+      "step": 968
+    },
+    {
+      "epoch": 2.6648351648351647,
+      "grad_norm": 0.06395678967237473,
+      "learning_rate": 3.0506376066807632e-06,
+      "loss": 1.1709,
+      "step": 970
+    },
+    {
+      "epoch": 2.67032967032967,
+      "grad_norm": 0.07488574087619781,
+      "learning_rate": 3.028576183560221e-06,
+      "loss": 1.1616,
+      "step": 972
+    },
+    {
+      "epoch": 2.675824175824176,
+      "grad_norm": 0.0717063769698143,
+      "learning_rate": 3.006560112425867e-06,
+      "loss": 1.159,
+      "step": 974
+    },
+    {
+      "epoch": 2.6813186813186816,
+      "grad_norm": 0.0665455088019371,
+      "learning_rate": 2.9845898997505102e-06,
+      "loss": 1.1717,
+      "step": 976
+    },
+    {
+      "epoch": 2.686813186813187,
+      "grad_norm": 0.07839926332235336,
+      "learning_rate": 2.962666050951997e-06,
+      "loss": 1.1635,
+      "step": 978
+    },
+    {
+      "epoch": 2.6923076923076925,
+      "grad_norm": 0.08876995742321014,
+      "learning_rate": 2.940789070381587e-06,
+      "loss": 1.1634,
+      "step": 980
+    },
+    {
+      "epoch": 2.697802197802198,
+      "grad_norm": 0.09154446423053741,
+      "learning_rate": 2.918959461312353e-06,
+      "loss": 1.1651,
+      "step": 982
+    },
+    {
+      "epoch": 2.7032967032967035,
+      "grad_norm": 0.08424372225999832,
+      "learning_rate": 2.897177725927599e-06,
+      "loss": 1.153,
+      "step": 984
+    },
+    {
+      "epoch": 2.708791208791209,
+      "grad_norm": 0.11759259551763535,
+      "learning_rate": 2.8754443653093186e-06,
+      "loss": 1.1662,
+      "step": 986
+    },
+    {
+      "epoch": 2.7142857142857144,
+      "grad_norm": 0.08856125921010971,
+      "learning_rate": 2.853759879426644e-06,
+      "loss": 1.1644,
+      "step": 988
+    },
+    {
+      "epoch": 2.71978021978022,
+      "grad_norm": 0.08224651217460632,
+      "learning_rate": 2.8321247671243695e-06,
+      "loss": 1.1628,
+      "step": 990
+    },
+    {
+      "epoch": 2.7252747252747254,
+      "grad_norm": 0.11355423927307129,
+      "learning_rate": 2.8105395261114666e-06,
+      "loss": 1.1663,
+      "step": 992
+    },
+    {
+      "epoch": 2.730769230769231,
+      "grad_norm": 0.06467260420322418,
+      "learning_rate": 2.7890046529496284e-06,
+      "loss": 1.1736,
+      "step": 994
+    },
+    {
+      "epoch": 2.7362637362637363,
+      "grad_norm": 0.08972840011119843,
+      "learning_rate": 2.7675206430418542e-06,
+      "loss": 1.1692,
+      "step": 996
+    },
+    {
+      "epoch": 2.741758241758242,
+      "grad_norm": 0.09491213411092758,
+      "learning_rate": 2.7460879906210485e-06,
+      "loss": 1.1707,
+      "step": 998
+    },
+    {
+      "epoch": 2.7472527472527473,
+      "grad_norm": 0.07063695043325424,
+      "learning_rate": 2.7247071887386544e-06,
+      "loss": 1.1605,
+      "step": 1000
+    },
+    {
+      "epoch": 2.7527472527472527,
+      "grad_norm": 0.07039818912744522,
+      "learning_rate": 2.70337872925331e-06,
+      "loss": 1.1648,
+      "step": 1002
+    },
+    {
+      "epoch": 2.758241758241758,
+      "grad_norm": 0.07515694946050644,
+      "learning_rate": 2.68210310281953e-06,
+      "loss": 1.1563,
+      "step": 1004
+    },
+    {
+      "epoch": 2.7637362637362637,
+      "grad_norm": 0.06818992644548416,
+      "learning_rate": 2.6608807988764252e-06,
+      "loss": 1.1735,
+      "step": 1006
+    },
+    {
+      "epoch": 2.769230769230769,
+      "grad_norm": 0.0827600434422493,
+      "learning_rate": 2.6397123056364364e-06,
+      "loss": 1.1698,
+      "step": 1008
+    },
+    {
+      "epoch": 2.7747252747252746,
+      "grad_norm": 0.10222798585891724,
+      "learning_rate": 2.618598110074105e-06,
+      "loss": 1.1689,
+      "step": 1010
+    },
+    {
+      "epoch": 2.78021978021978,
+      "grad_norm": 0.0823441818356514,
+      "learning_rate": 2.5975386979148792e-06,
+      "loss": 1.166,
+      "step": 1012
+    },
+    {
+      "epoch": 2.7857142857142856,
+      "grad_norm": 0.0735846534371376,
+      "learning_rate": 2.576534553623925e-06,
+      "loss": 1.1592,
+      "step": 1014
+    },
+    {
+      "epoch": 2.791208791208791,
+      "grad_norm": 0.07179060578346252,
+      "learning_rate": 2.5555861603949832e-06,
+      "loss": 1.1755,
+      "step": 1016
+    },
+    {
+      "epoch": 2.7967032967032965,
+      "grad_norm": 0.08374089747667313,
+      "learning_rate": 2.534694000139273e-06,
+      "loss": 1.1757,
+      "step": 1018
+    },
+    {
+      "epoch": 2.802197802197802,
+      "grad_norm": 0.07065007835626602,
+      "learning_rate": 2.513858553474382e-06,
+      "loss": 1.1698,
+      "step": 1020
+    },
+    {
+      "epoch": 2.8076923076923075,
+      "grad_norm": 0.08045931160449982,
+      "learning_rate": 2.4930802997132213e-06,
+      "loss": 1.1582,
+      "step": 1022
+    },
+    {
+      "epoch": 2.8076923076923075,
+      "eval_loss": 1.1672074794769287,
+      "eval_runtime": 299.934,
+      "eval_samples_per_second": 17.257,
+      "eval_steps_per_second": 0.27,
+      "step": 1022
+    },
+    {
+      "epoch": 2.813186813186813,
+      "grad_norm": 0.07823072373867035,
+      "learning_rate": 2.4723597168529984e-06,
+      "loss": 1.1596,
+      "step": 1024
+    },
+    {
+      "epoch": 2.8186813186813184,
+      "grad_norm": 0.08508284389972687,
+      "learning_rate": 2.4516972815642166e-06,
+      "loss": 1.1723,
+      "step": 1026
+    },
+    {
+      "epoch": 2.824175824175824,
+      "grad_norm": 0.07922125607728958,
+      "learning_rate": 2.4310934691797207e-06,
+      "loss": 1.1617,
+      "step": 1028
+    },
+    {
+      "epoch": 2.82967032967033,
+      "grad_norm": 0.0768875777721405,
+      "learning_rate": 2.410548753683743e-06,
+      "loss": 1.1664,
+      "step": 1030
+    },
+    {
+      "epoch": 2.8351648351648353,
+      "grad_norm": 0.07595735043287277,
+      "learning_rate": 2.390063607701016e-06,
+      "loss": 1.1731,
+      "step": 1032
+    },
+    {
+      "epoch": 2.840659340659341,
+      "grad_norm": 0.08126144856214523,
+      "learning_rate": 2.369638502485897e-06,
+      "loss": 1.163,
+      "step": 1034
+    },
+    {
+      "epoch": 2.8461538461538463,
+      "grad_norm": 0.06850147992372513,
+      "learning_rate": 2.3492739079115214e-06,
+      "loss": 1.1685,
+      "step": 1036
+    },
+    {
+      "epoch": 2.8516483516483517,
+      "grad_norm": 0.08522983640432358,
+      "learning_rate": 2.3289702924589914e-06,
+      "loss": 1.1686,
+      "step": 1038
+    },
+    {
+      "epoch": 2.857142857142857,
+      "grad_norm": 0.08155310899019241,
+      "learning_rate": 2.3087281232066134e-06,
+      "loss": 1.1653,
+      "step": 1040
+    },
+    {
+      "epoch": 2.8626373626373627,
+      "grad_norm": 0.07040958106517792,
+      "learning_rate": 2.2885478658191364e-06,
+      "loss": 1.1557,
+      "step": 1042
+    },
+    {
+      "epoch": 2.868131868131868,
+      "grad_norm": 0.0749848261475563,
+      "learning_rate": 2.268429984537048e-06,
+      "loss": 1.1724,
+      "step": 1044
+    },
+    {
+      "epoch": 2.8736263736263736,
+      "grad_norm": 0.06682237237691879,
+      "learning_rate": 2.248374942165894e-06,
+      "loss": 1.158,
+      "step": 1046
+    },
+    {
+      "epoch": 2.879120879120879,
+      "grad_norm": 0.0770927369594574,
+      "learning_rate": 2.2283832000656304e-06,
+      "loss": 1.1676,
+      "step": 1048
+    },
+    {
+      "epoch": 2.8846153846153846,
+      "grad_norm": 0.08294253051280975,
+      "learning_rate": 2.2084552181400087e-06,
+      "loss": 1.1654,
+      "step": 1050
+    },
+    {
+      "epoch": 2.89010989010989,
+      "grad_norm": 0.08596468716859818,
+      "learning_rate": 2.188591454826e-06,
+      "loss": 1.1689,
+      "step": 1052
+    },
+    {
+      "epoch": 2.8956043956043955,
+      "grad_norm": 0.08016978204250336,
+      "learning_rate": 2.168792367083243e-06,
+      "loss": 1.1614,
+      "step": 1054
+    },
+    {
+      "epoch": 2.901098901098901,
+      "grad_norm": 0.0838996022939682,
+      "learning_rate": 2.1490584103835433e-06,
+      "loss": 1.1658,
+      "step": 1056
+    },
+    {
+      "epoch": 2.9065934065934065,
+      "grad_norm": 0.07020293921232224,
+      "learning_rate": 2.1293900387003742e-06,
+      "loss": 1.1594,
+      "step": 1058
+    },
+    {
+      "epoch": 2.912087912087912,
+      "grad_norm": 0.07320253551006317,
+      "learning_rate": 2.109787704498459e-06,
+      "loss": 1.1635,
+      "step": 1060
+    },
+    {
+      "epoch": 2.9175824175824174,
+      "grad_norm": 0.0821409672498703,
+      "learning_rate": 2.0902518587233418e-06,
+      "loss": 1.1628,
+      "step": 1062
+    },
+    {
+      "epoch": 2.9230769230769234,
+      "grad_norm": 0.07211437821388245,
+      "learning_rate": 2.0707829507910237e-06,
+      "loss": 1.1689,
+      "step": 1064
+    },
+    {
+      "epoch": 2.928571428571429,
+      "grad_norm": 0.08253902941942215,
+      "learning_rate": 2.051381428577622e-06,
+      "loss": 1.1654,
+      "step": 1066
+    },
+    {
+      "epoch": 2.9340659340659343,
+      "grad_norm": 0.080472432076931,
+      "learning_rate": 2.0320477384090665e-06,
+      "loss": 1.1707,
+      "step": 1068
+    },
+    {
+      "epoch": 2.9395604395604398,
+      "grad_norm": 0.08331170678138733,
+      "learning_rate": 2.012782325050831e-06,
+      "loss": 1.1729,
+      "step": 1070
+    },
+    {
+      "epoch": 2.9450549450549453,
+      "grad_norm": 0.07666671276092529,
+      "learning_rate": 1.9935856316977044e-06,
+      "loss": 1.1715,
+      "step": 1072
+    },
+    {
+      "epoch": 2.9505494505494507,
+      "grad_norm": 0.09519699960947037,
+      "learning_rate": 1.9744580999635902e-06,
+      "loss": 1.1629,
+      "step": 1074
+    },
+    {
+      "epoch": 2.956043956043956,
+      "grad_norm": 0.0872369259595871,
+      "learning_rate": 1.9554001698713572e-06,
+      "loss": 1.1683,
+      "step": 1076
+    },
+    {
+      "epoch": 2.9615384615384617,
+      "grad_norm": 0.07113870233297348,
+      "learning_rate": 1.936412279842705e-06,
+      "loss": 1.1637,
+      "step": 1078
+    },
+    {
+      "epoch": 2.967032967032967,
+      "grad_norm": 0.08545450866222382,
+      "learning_rate": 1.9174948666880805e-06,
+      "loss": 1.1627,
+      "step": 1080
+    },
+    {
+      "epoch": 2.9725274725274726,
+      "grad_norm": 0.08711759746074677,
+      "learning_rate": 1.8986483655966408e-06,
+      "loss": 1.1582,
+      "step": 1082
+    },
+    {
+      "epoch": 2.978021978021978,
+      "grad_norm": 0.08870179951190948,
+      "learning_rate": 1.879873210126229e-06,
+      "loss": 1.167,
+      "step": 1084
+    },
+    {
+      "epoch": 2.9835164835164836,
+      "grad_norm": 0.1006346121430397,
+      "learning_rate": 1.8611698321933991e-06,
+      "loss": 1.1724,
+      "step": 1086
+    },
+    {
+      "epoch": 2.989010989010989,
+      "grad_norm": 0.07408854365348816,
+      "learning_rate": 1.8425386620634961e-06,
+      "loss": 1.1705,
+      "step": 1088
+    },
+    {
+      "epoch": 2.9945054945054945,
+      "grad_norm": 0.09146919846534729,
+      "learning_rate": 1.8239801283407393e-06,
+      "loss": 1.16,
+      "step": 1090
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.0709662064909935,
+      "learning_rate": 1.8054946579583732e-06,
+      "loss": 1.1636,
+      "step": 1092
+    },
+    {
+      "epoch": 3.0054945054945055,
+      "grad_norm": 0.10273056477308273,
+      "learning_rate": 1.787082676168842e-06,
+      "loss": 1.1647,
+      "step": 1094
+    },
+    {
+      "epoch": 3.010989010989011,
+      "grad_norm": 0.07023092359304428,
+      "learning_rate": 1.7687446065340074e-06,
+      "loss": 1.162,
+      "step": 1096
+    },
+    {
+      "epoch": 3.0164835164835164,
+      "grad_norm": 0.08073507994413376,
+      "learning_rate": 1.7504808709154104e-06,
+      "loss": 1.1697,
+      "step": 1098
+    },
+    {
+      "epoch": 3.021978021978022,
+      "grad_norm": 0.07398983091115952,
+      "learning_rate": 1.7322918894645525e-06,
+      "loss": 1.1637,
+      "step": 1100
+    },
+    {
+      "epoch": 3.0274725274725274,
+      "grad_norm": 0.08476796001195908,
+      "learning_rate": 1.7141780806132429e-06,
+      "loss": 1.161,
+      "step": 1102
+    },
+    {
+      "epoch": 3.032967032967033,
+      "grad_norm": 0.07763465493917465,
+      "learning_rate": 1.696139861063974e-06,
+      "loss": 1.1685,
+      "step": 1104
+    },
+    {
+      "epoch": 3.0384615384615383,
+      "grad_norm": 0.08244740962982178,
+      "learning_rate": 1.6781776457803227e-06,
+      "loss": 1.1683,
+      "step": 1106
+    },
+    {
+      "epoch": 3.043956043956044,
+      "grad_norm": 0.09578699618577957,
+      "learning_rate": 1.660291847977415e-06,
+      "loss": 1.1677,
+      "step": 1108
+    },
+    {
+      "epoch": 3.0494505494505493,
+      "grad_norm": 0.07818014919757843,
+      "learning_rate": 1.6424828791124159e-06,
+      "loss": 1.166,
+      "step": 1110
+    },
+    {
+      "epoch": 3.0549450549450547,
+      "grad_norm": 0.07996879518032074,
+      "learning_rate": 1.624751148875065e-06,
+      "loss": 1.1732,
+      "step": 1112
+    },
+    {
+      "epoch": 3.0604395604395602,
+      "grad_norm": 0.06753943115472794,
+      "learning_rate": 1.6070970651782514e-06,
+      "loss": 1.1628,
+      "step": 1114
+    },
+    {
+      "epoch": 3.065934065934066,
+      "grad_norm": 0.06478522717952728,
+      "learning_rate": 1.5895210341486279e-06,
+      "loss": 1.1607,
+      "step": 1116
+    },
+    {
+      "epoch": 3.0714285714285716,
+      "grad_norm": 0.07904413342475891,
+      "learning_rate": 1.5720234601172767e-06,
+      "loss": 1.1655,
+      "step": 1118
+    },
+    {
+      "epoch": 3.076923076923077,
+      "grad_norm": 0.07776004076004028,
+      "learning_rate": 1.5546047456103964e-06,
+      "loss": 1.1677,
+      "step": 1120
+    },
+    {
+      "epoch": 3.0824175824175826,
+      "grad_norm": 0.07198958098888397,
+      "learning_rate": 1.537265291340042e-06,
+      "loss": 1.1636,
+      "step": 1122
+    },
+    {
+      "epoch": 3.087912087912088,
+      "grad_norm": 0.07310649752616882,
+      "learning_rate": 1.5200054961949233e-06,
+      "loss": 1.1601,
+      "step": 1124
+    },
+    {
+      "epoch": 3.0934065934065935,
+      "grad_norm": 0.07591935992240906,
+      "learning_rate": 1.5028257572312105e-06,
+      "loss": 1.1649,
+      "step": 1126
+    },
+    {
+      "epoch": 3.098901098901099,
+      "grad_norm": 0.07674799114465714,
+      "learning_rate": 1.485726469663401e-06,
+      "loss": 1.1704,
+      "step": 1128
+    },
+    {
+      "epoch": 3.1043956043956045,
+      "grad_norm": 0.09393850713968277,
+      "learning_rate": 1.468708026855245e-06,
+      "loss": 1.1692,
+      "step": 1130
+    },
+    {
+      "epoch": 3.10989010989011,
+      "grad_norm": 0.07599301636219025,
+      "learning_rate": 1.4517708203106763e-06,
+      "loss": 1.1645,
+      "step": 1132
+    },
+    {
+      "epoch": 3.1153846153846154,
+      "grad_norm": 0.08221649378538132,
+      "learning_rate": 1.4349152396648153e-06,
+      "loss": 1.1631,
+      "step": 1134
+    },
+    {
+      "epoch": 3.120879120879121,
+      "grad_norm": 0.07295921444892883,
+      "learning_rate": 1.4181416726750052e-06,
+      "loss": 1.1612,
+      "step": 1136
+    },
+    {
+      "epoch": 3.1263736263736264,
+      "grad_norm": 0.079580157995224,
+      "learning_rate": 1.4014505052118893e-06,
+      "loss": 1.1599,
+      "step": 1138
+    },
+    {
+      "epoch": 3.131868131868132,
+      "grad_norm": 0.09730138629674911,
+      "learning_rate": 1.3848421212505404e-06,
+      "loss": 1.1632,
+      "step": 1140
+    },
+    {
+      "epoch": 3.1373626373626373,
+      "grad_norm": 0.09287888556718826,
+      "learning_rate": 1.3683169028616155e-06,
+      "loss": 1.1602,
+      "step": 1142
+    },
+    {
+      "epoch": 3.142857142857143,
+      "grad_norm": 0.07959942519664764,
+      "learning_rate": 1.3518752302025773e-06,
+      "loss": 1.1629,
+      "step": 1144
+    },
+    {
+      "epoch": 3.1483516483516483,
+      "grad_norm": 0.07988713681697845,
+      "learning_rate": 1.3355174815089477e-06,
+      "loss": 1.1641,
+      "step": 1146
+    },
+    {
+      "epoch": 3.1538461538461537,
+      "grad_norm": 0.06947878748178482,
+      "learning_rate": 1.3192440330856005e-06,
+      "loss": 1.1614,
+      "step": 1148
+    },
+    {
+      "epoch": 3.159340659340659,
+      "grad_norm": 0.06804593652486801,
+      "learning_rate": 1.30305525929811e-06,
+      "loss": 1.1666,
+      "step": 1150
+    },
+    {
+      "epoch": 3.1648351648351647,
+      "grad_norm": 0.07112333178520203,
+      "learning_rate": 1.2869515325641357e-06,
+      "loss": 1.1592,
+      "step": 1152
+    },
+    {
+      "epoch": 3.17032967032967,
+      "grad_norm": 0.07313236594200134,
+      "learning_rate": 1.2709332233448573e-06,
+      "loss": 1.1686,
+      "step": 1154
+    },
+    {
+      "epoch": 3.1758241758241756,
+      "grad_norm": 0.07543834298849106,
+      "learning_rate": 1.2550007001364518e-06,
+      "loss": 1.1626,
+      "step": 1156
+    },
+    {
+      "epoch": 3.181318681318681,
+      "grad_norm": 0.07730558514595032,
+      "learning_rate": 1.239154329461615e-06,
+      "loss": 1.164,
+      "step": 1158
+    },
+    {
+      "epoch": 3.186813186813187,
+      "grad_norm": 0.07528182864189148,
+      "learning_rate": 1.223394475861131e-06,
+      "loss": 1.1621,
+      "step": 1160
+    },
+    {
+      "epoch": 3.1923076923076925,
+      "grad_norm": 0.07378531992435455,
+      "learning_rate": 1.207721501885486e-06,
+      "loss": 1.162,
+      "step": 1162
+    },
+    {
+      "epoch": 3.197802197802198,
+      "grad_norm": 0.07334302365779877,
+      "learning_rate": 1.1921357680865258e-06,
+      "loss": 1.1642,
+      "step": 1164
+    },
+    {
+      "epoch": 3.2032967032967035,
+      "grad_norm": 0.07697130739688873,
+      "learning_rate": 1.1766376330091684e-06,
+      "loss": 1.1647,
+      "step": 1166
+    },
+    {
+      "epoch": 3.208791208791209,
+      "grad_norm": 0.0726918876171112,
+      "learning_rate": 1.1612274531831463e-06,
+      "loss": 1.1719,
+      "step": 1168
+    },
+    {
+      "epoch": 3.208791208791209,
+      "eval_loss": 1.1669209003448486,
+      "eval_runtime": 299.9103,
+      "eval_samples_per_second": 17.258,
+      "eval_steps_per_second": 0.27,
+      "step": 1168
+    },
+    {
+      "epoch": 3.2142857142857144,
+      "grad_norm": 0.07847239077091217,
+      "learning_rate": 1.1459055831148074e-06,
+      "loss": 1.1651,
+      "step": 1170
+    },
+    {
+      "epoch": 3.21978021978022,
+      "grad_norm": 0.06886722892522812,
+      "learning_rate": 1.1306723752789672e-06,
+      "loss": 1.1648,
+      "step": 1172
+    },
+    {
+      "epoch": 3.2252747252747254,
+      "grad_norm": 0.0860794335603714,
+      "learning_rate": 1.1155281801107897e-06,
+      "loss": 1.168,
+      "step": 1174
+    },
+    {
+      "epoch": 3.230769230769231,
+      "grad_norm": 0.07369523495435715,
+      "learning_rate": 1.1004733459977325e-06,
+      "loss": 1.1669,
+      "step": 1176
+    },
+    {
+      "epoch": 3.2362637362637363,
+      "grad_norm": 0.07247929275035858,
+      "learning_rate": 1.0855082192715294e-06,
+      "loss": 1.168,
+      "step": 1178
+    },
+    {
+      "epoch": 3.241758241758242,
+      "grad_norm": 0.07475174218416214,
+      "learning_rate": 1.0706331442002226e-06,
+      "loss": 1.1622,
+      "step": 1180
+    },
+    {
+      "epoch": 3.2472527472527473,
+      "grad_norm": 0.07838430255651474,
+      "learning_rate": 1.0558484629802502e-06,
+      "loss": 1.171,
+      "step": 1182
+    },
+    {
+      "epoch": 3.2527472527472527,
+      "grad_norm": 0.0696970596909523,
+      "learning_rate": 1.041154515728559e-06,
+      "loss": 1.1621,
+      "step": 1184
+    },
+    {
+      "epoch": 3.258241758241758,
+      "grad_norm": 0.07396616786718369,
+      "learning_rate": 1.0265516404747943e-06,
+      "loss": 1.1641,
+      "step": 1186
+    },
+    {
+      "epoch": 3.2637362637362637,
+      "grad_norm": 0.08104917407035828,
+      "learning_rate": 1.0120401731535213e-06,
+      "loss": 1.1663,
+      "step": 1188
+    },
+    {
+      "epoch": 3.269230769230769,
+      "grad_norm": 0.0944487676024437,
+      "learning_rate": 9.976204475964907e-07,
+      "loss": 1.1618,
+      "step": 1190
+    },
+    {
+      "epoch": 3.2747252747252746,
+      "grad_norm": 0.07517724484205246,
+      "learning_rate": 9.832927955249605e-07,
+      "loss": 1.1726,
+      "step": 1192
+    },
+    {
+      "epoch": 3.28021978021978,
+      "grad_norm": 0.08082670718431473,
+      "learning_rate": 9.690575465420733e-07,
+      "loss": 1.1689,
+      "step": 1194
+    },
+    {
+      "epoch": 3.2857142857142856,
+      "grad_norm": 0.07103955000638962,
+      "learning_rate": 9.549150281252633e-07,
+      "loss": 1.1724,
+      "step": 1196
+    },
+    {
+      "epoch": 3.291208791208791,
+      "grad_norm": 0.08387453854084015,
+      "learning_rate": 9.408655656187282e-07,
+      "loss": 1.1598,
+      "step": 1198
+    },
+    {
+      "epoch": 3.2967032967032965,
+      "grad_norm": 0.07405807077884674,
+      "learning_rate": 9.269094822259439e-07,
+      "loss": 1.1648,
+      "step": 1200
+    },
+    {
+      "epoch": 3.302197802197802,
+      "grad_norm": 0.07046142220497131,
+      "learning_rate": 9.130470990022283e-07,
+      "loss": 1.1707,
+      "step": 1202
+    },
+    {
+      "epoch": 3.3076923076923075,
+      "grad_norm": 0.07296803593635559,
+      "learning_rate": 8.992787348473575e-07,
+      "loss": 1.1642,
+      "step": 1204
+    },
+    {
+      "epoch": 3.313186813186813,
+      "grad_norm": 0.0734080895781517,
+      "learning_rate": 8.856047064982276e-07,
+      "loss": 1.1558,
+      "step": 1206
+    },
+    {
+      "epoch": 3.3186813186813184,
+      "grad_norm": 0.07395216077566147,
+      "learning_rate": 8.720253285215685e-07,
+      "loss": 1.1721,
+      "step": 1208
+    },
+    {
+      "epoch": 3.3241758241758244,
+      "grad_norm": 0.07748089730739594,
+      "learning_rate": 8.585409133067119e-07,
+      "loss": 1.1653,
+      "step": 1210
+    },
+    {
+      "epoch": 3.32967032967033,
+      "grad_norm": 0.07397377490997314,
+      "learning_rate": 8.451517710583934e-07,
+      "loss": 1.1623,
+      "step": 1212
+    },
+    {
+      "epoch": 3.3351648351648353,
+      "grad_norm": 0.07272295653820038,
+      "learning_rate": 8.318582097896316e-07,
+      "loss": 1.1643,
+      "step": 1214
+    },
+    {
+      "epoch": 3.340659340659341,
+      "grad_norm": 0.07338716834783554,
+      "learning_rate": 8.18660535314631e-07,
+      "loss": 1.1612,
+      "step": 1216
+    },
+    {
+      "epoch": 3.3461538461538463,
+      "grad_norm": 0.06233609840273857,
+      "learning_rate": 8.055590512417499e-07,
+      "loss": 1.1642,
+      "step": 1218
+    },
+    {
+      "epoch": 3.3516483516483517,
+      "grad_norm": 0.07398121803998947,
+      "learning_rate": 7.925540589665187e-07,
+      "loss": 1.1719,
+      "step": 1220
+    },
+    {
+      "epoch": 3.357142857142857,
+      "grad_norm": 0.0798347070813179,
+      "learning_rate": 7.796458576647015e-07,
+      "loss": 1.1669,
+      "step": 1222
+    },
+    {
+      "epoch": 3.3626373626373627,
+      "grad_norm": 0.06861409544944763,
+      "learning_rate": 7.668347442854218e-07,
+      "loss": 1.1698,
+      "step": 1224
+    },
+    {
+      "epoch": 3.368131868131868,
+      "grad_norm": 0.06391950696706772,
+      "learning_rate": 7.541210135443188e-07,
+      "loss": 1.166,
+      "step": 1226
+    },
+    {
+      "epoch": 3.3736263736263736,
+      "grad_norm": 0.07030785083770752,
+      "learning_rate": 7.415049579167783e-07,
+      "loss": 1.166,
+      "step": 1228
+    },
+    {
+      "epoch": 3.379120879120879,
+      "grad_norm": 0.07540637254714966,
+      "learning_rate": 7.289868676312023e-07,
+      "loss": 1.1643,
+      "step": 1230
+    },
+    {
+      "epoch": 3.3846153846153846,
+      "grad_norm": 0.06893002241849899,
+      "learning_rate": 7.165670306623296e-07,
+      "loss": 1.163,
+      "step": 1232
+    },
+    {
+      "epoch": 3.39010989010989,
+      "grad_norm": 0.07210598886013031,
+      "learning_rate": 7.042457327246088e-07,
+      "loss": 1.1648,
+      "step": 1234
+    },
+    {
+      "epoch": 3.3956043956043955,
+      "grad_norm": 0.076958067715168,
+      "learning_rate": 6.920232572656349e-07,
+      "loss": 1.1635,
+      "step": 1236
+    },
+    {
+      "epoch": 3.401098901098901,
+      "grad_norm": 0.07737283408641815,
+      "learning_rate": 6.79899885459619e-07,
+      "loss": 1.161,
+      "step": 1238
+    },
+    {
+      "epoch": 3.4065934065934065,
+      "grad_norm": 0.07089677453041077,
+      "learning_rate": 6.678758962009241e-07,
+      "loss": 1.1595,
+      "step": 1240
+    },
+    {
+      "epoch": 3.412087912087912,
+      "grad_norm": 0.06980301439762115,
+      "learning_rate": 6.559515660976506e-07,
+      "loss": 1.1646,
+      "step": 1242
+    },
+    {
+      "epoch": 3.4175824175824174,
+      "grad_norm": 0.0859365463256836,
+      "learning_rate": 6.441271694652701e-07,
+      "loss": 1.1653,
+      "step": 1244
+    },
+    {
+      "epoch": 3.423076923076923,
+      "grad_norm": 0.0688340812921524,
+      "learning_rate": 6.32402978320315e-07,
+      "loss": 1.171,
+      "step": 1246
+    },
+    {
+      "epoch": 3.4285714285714284,
+      "grad_norm": 0.07327734678983688,
+      "learning_rate": 6.207792623741249e-07,
+      "loss": 1.1664,
+      "step": 1248
+    },
+    {
+      "epoch": 3.4340659340659343,
+      "grad_norm": 0.06622574478387833,
+      "learning_rate": 6.092562890266341e-07,
+      "loss": 1.162,
+      "step": 1250
+    },
+    {
+      "epoch": 3.4395604395604398,
+      "grad_norm": 0.06825845688581467,
+      "learning_rate": 5.97834323360233e-07,
+      "loss": 1.1601,
+      "step": 1252
+    },
+    {
+      "epoch": 3.4450549450549453,
+      "grad_norm": 0.0715174600481987,
+      "learning_rate": 5.86513628133652e-07,
+      "loss": 1.1653,
+      "step": 1254
+    },
+    {
+      "epoch": 3.4505494505494507,
+      "grad_norm": 0.06850449740886688,
+      "learning_rate": 5.75294463775935e-07,
+      "loss": 1.1625,
+      "step": 1256
+    },
+    {
+      "epoch": 3.456043956043956,
+      "grad_norm": 0.061297595500946045,
+      "learning_rate": 5.641770883804365e-07,
+      "loss": 1.168,
+      "step": 1258
+    },
+    {
+      "epoch": 3.4615384615384617,
+      "grad_norm": 0.07319982349872589,
+      "learning_rate": 5.531617576988879e-07,
+      "loss": 1.1693,
+      "step": 1260
+    },
+    {
+      "epoch": 3.467032967032967,
+      "grad_norm": 0.07959607243537903,
+      "learning_rate": 5.422487251355146e-07,
+      "loss": 1.1625,
+      "step": 1262
+    },
+    {
+      "epoch": 3.4725274725274726,
+      "grad_norm": 0.06803479790687561,
+      "learning_rate": 5.314382417412062e-07,
+      "loss": 1.162,
+      "step": 1264
+    },
+    {
+      "epoch": 3.478021978021978,
+      "grad_norm": 0.07737120240926743,
+      "learning_rate": 5.207305562077403e-07,
+      "loss": 1.1705,
+      "step": 1266
+    },
+    {
+      "epoch": 3.4835164835164836,
+      "grad_norm": 0.06910370290279388,
+      "learning_rate": 5.101259148620618e-07,
+      "loss": 1.1619,
+      "step": 1268
+    },
+    {
+      "epoch": 3.489010989010989,
+      "grad_norm": 0.06398583203554153,
+      "learning_rate": 4.99624561660616e-07,
+      "loss": 1.1659,
+      "step": 1270
+    },
+    {
+      "epoch": 3.4945054945054945,
+      "grad_norm": 0.06311172246932983,
+      "learning_rate": 4.892267381837396e-07,
+      "loss": 1.1595,
+      "step": 1272
+    },
+    {
+      "epoch": 3.5,
+      "grad_norm": 0.06687135994434357,
+      "learning_rate": 4.789326836300983e-07,
+      "loss": 1.1639,
+      "step": 1274
+    },
+    {
+      "epoch": 3.5054945054945055,
+      "grad_norm": 0.06371094286441803,
+      "learning_rate": 4.687426348111834e-07,
+      "loss": 1.1644,
+      "step": 1276
+    },
+    {
+      "epoch": 3.510989010989011,
+      "grad_norm": 0.0776761993765831,
+      "learning_rate": 4.586568261458729e-07,
+      "loss": 1.1745,
+      "step": 1278
+    },
+    {
+      "epoch": 3.5164835164835164,
+      "grad_norm": 0.06816533952951431,
+      "learning_rate": 4.486754896550288e-07,
+      "loss": 1.1636,
+      "step": 1280
+    },
+    {
+      "epoch": 3.521978021978022,
+      "grad_norm": 0.07598984241485596,
+      "learning_rate": 4.3879885495616505e-07,
+      "loss": 1.1657,
+      "step": 1282
+    },
+    {
+      "epoch": 3.5274725274725274,
+      "grad_norm": 0.06942006200551987,
+      "learning_rate": 4.290271492581627e-07,
+      "loss": 1.1638,
+      "step": 1284
+    },
+    {
+      "epoch": 3.532967032967033,
+      "grad_norm": 0.06620027124881744,
+      "learning_rate": 4.1936059735604497e-07,
+      "loss": 1.1698,
+      "step": 1286
+    },
+    {
+      "epoch": 3.5384615384615383,
+      "grad_norm": 0.07451992481946945,
+      "learning_rate": 4.0979942162580387e-07,
+      "loss": 1.1663,
+      "step": 1288
+    },
+    {
+      "epoch": 3.543956043956044,
+      "grad_norm": 0.07096972316503525,
+      "learning_rate": 4.003438420192873e-07,
+      "loss": 1.1627,
+      "step": 1290
+    },
+    {
+      "epoch": 3.5494505494505493,
+      "grad_norm": 0.0655079111456871,
+      "learning_rate": 3.9099407605913576e-07,
+      "loss": 1.1645,
+      "step": 1292
+    },
+    {
+      "epoch": 3.5549450549450547,
+      "grad_norm": 0.06677290052175522,
+      "learning_rate": 3.8175033883378233e-07,
+      "loss": 1.1642,
+      "step": 1294
+    },
+    {
+      "epoch": 3.5604395604395602,
+      "grad_norm": 0.06589250266551971,
+      "learning_rate": 3.7261284299249967e-07,
+      "loss": 1.1661,
+      "step": 1296
+    },
+    {
+      "epoch": 3.5659340659340657,
+      "grad_norm": 0.06369265913963318,
+      "learning_rate": 3.63581798740511e-07,
+      "loss": 1.1591,
+      "step": 1298
+    },
+    {
+      "epoch": 3.571428571428571,
+      "grad_norm": 0.06584078073501587,
+      "learning_rate": 3.5465741383415684e-07,
+      "loss": 1.1669,
+      "step": 1300
+    },
+    {
+      "epoch": 3.5769230769230766,
+      "grad_norm": 0.07220979779958725,
+      "learning_rate": 3.4583989357611037e-07,
+      "loss": 1.1637,
+      "step": 1302
+    },
+    {
+      "epoch": 3.5824175824175826,
+      "grad_norm": 0.07124118506908417,
+      "learning_rate": 3.371294408106585e-07,
+      "loss": 1.1685,
+      "step": 1304
+    },
+    {
+      "epoch": 3.587912087912088,
+      "grad_norm": 0.0632823258638382,
+      "learning_rate": 3.285262559190322e-07,
+      "loss": 1.1646,
+      "step": 1306
+    },
+    {
+      "epoch": 3.5934065934065935,
+      "grad_norm": 0.07331771403551102,
+      "learning_rate": 3.20030536814801e-07,
+      "loss": 1.1639,
+      "step": 1308
+    },
+    {
+      "epoch": 3.598901098901099,
+      "grad_norm": 0.07538831979036331,
+      "learning_rate": 3.1164247893931575e-07,
+      "loss": 1.1642,
+      "step": 1310
+    },
+    {
+      "epoch": 3.6043956043956045,
+      "grad_norm": 0.06919172406196594,
+      "learning_rate": 3.033622752572157e-07,
+      "loss": 1.165,
+      "step": 1312
+    },
+    {
+      "epoch": 3.60989010989011,
+      "grad_norm": 0.0683453157544136,
+      "learning_rate": 2.951901162519877e-07,
+      "loss": 1.1657,
+      "step": 1314
+    },
+    {
+      "epoch": 3.60989010989011,
+      "eval_loss": 1.1666840314865112,
+      "eval_runtime": 300.2612,
+      "eval_samples_per_second": 17.238,
+      "eval_steps_per_second": 0.27,
+      "step": 1314
+    },
+    {
+      "epoch": 3.6153846153846154,
+      "grad_norm": 0.07349937409162521,
+      "learning_rate": 2.8712618992158656e-07,
+      "loss": 1.167,
+      "step": 1316
+    },
+    {
+      "epoch": 3.620879120879121,
+      "grad_norm": 0.0776103138923645,
+      "learning_rate": 2.791706817741041e-07,
+      "loss": 1.1628,
+      "step": 1318
+    },
+    {
+      "epoch": 3.6263736263736264,
+      "grad_norm": 0.06213715299963951,
+      "learning_rate": 2.7132377482351037e-07,
+      "loss": 1.1623,
+      "step": 1320
+    },
+    {
+      "epoch": 3.631868131868132,
+      "grad_norm": 0.06590726226568222,
+      "learning_rate": 2.635856495854372e-07,
+      "loss": 1.1675,
+      "step": 1322
+    },
+    {
+      "epoch": 3.6373626373626373,
+      "grad_norm": 0.06559966504573822,
+      "learning_rate": 2.5595648407302496e-07,
+      "loss": 1.1692,
+      "step": 1324
+    },
+    {
+      "epoch": 3.642857142857143,
+      "grad_norm": 0.0632384866476059,
+      "learning_rate": 2.484364537928341e-07,
+      "loss": 1.1632,
+      "step": 1326
+    },
+    {
+      "epoch": 3.6483516483516483,
+      "grad_norm": 0.07011737674474716,
+      "learning_rate": 2.41025731740801e-07,
+      "loss": 1.1626,
+      "step": 1328
+    },
+    {
+      "epoch": 3.6538461538461537,
+      "grad_norm": 0.06871969997882843,
+      "learning_rate": 2.3372448839825978e-07,
+      "loss": 1.1597,
+      "step": 1330
+    },
+    {
+      "epoch": 3.659340659340659,
+      "grad_norm": 0.06817732751369476,
+      "learning_rate": 2.2653289172802295e-07,
+      "loss": 1.1714,
+      "step": 1332
+    },
+    {
+      "epoch": 3.6648351648351647,
+      "grad_norm": 0.06385420262813568,
+      "learning_rate": 2.194511071705141e-07,
+      "loss": 1.1747,
+      "step": 1334
+    },
+    {
+      "epoch": 3.67032967032967,
+      "grad_norm": 0.06753863394260406,
+      "learning_rate": 2.1247929763996534e-07,
+      "loss": 1.1602,
+      "step": 1336
+    },
+    {
+      "epoch": 3.675824175824176,
+      "grad_norm": 0.0678897351026535,
+      "learning_rate": 2.0561762352066638e-07,
+      "loss": 1.1685,
+      "step": 1338
+    },
+    {
+      "epoch": 3.6813186813186816,
+      "grad_norm": 0.06596899032592773,
+      "learning_rate": 1.988662426632765e-07,
+      "loss": 1.1657,
+      "step": 1340
+    },
+    {
+      "epoch": 3.686813186813187,
+      "grad_norm": 0.0690767765045166,
+      "learning_rate": 1.922253103811944e-07,
+      "loss": 1.167,
+      "step": 1342
+    },
+    {
+      "epoch": 3.6923076923076925,
+      "grad_norm": 0.07675404101610184,
+      "learning_rate": 1.85694979446982e-07,
+      "loss": 1.1661,
+      "step": 1344
+    },
+    {
+      "epoch": 3.697802197802198,
+      "grad_norm": 0.06820071488618851,
+      "learning_rate": 1.7927540008885414e-07,
+      "loss": 1.1635,
+      "step": 1346
+    },
+    {
+      "epoch": 3.7032967032967035,
+      "grad_norm": 0.07236689329147339,
+      "learning_rate": 1.729667199872187e-07,
+      "loss": 1.1581,
+      "step": 1348
+    },
+    {
+      "epoch": 3.708791208791209,
+      "grad_norm": 0.06689772009849548,
+      "learning_rate": 1.6676908427128103e-07,
+      "loss": 1.1617,
+      "step": 1350
+    },
+    {
+      "epoch": 3.7142857142857144,
+      "grad_norm": 0.06681732088327408,
+      "learning_rate": 1.6068263551570596e-07,
+      "loss": 1.1732,
+      "step": 1352
+    },
+    {
+      "epoch": 3.71978021978022,
+      "grad_norm": 0.07003802061080933,
+      "learning_rate": 1.5470751373733773e-07,
+      "loss": 1.1631,
+      "step": 1354
+    },
+    {
+      "epoch": 3.7252747252747254,
+      "grad_norm": 0.06234800070524216,
+      "learning_rate": 1.488438563919764e-07,
+      "loss": 1.1696,
+      "step": 1356
+    },
+    {
+      "epoch": 3.730769230769231,
+      "grad_norm": 0.0727643072605133,
+      "learning_rate": 1.4309179837122045e-07,
+      "loss": 1.1654,
+      "step": 1358
+    },
+    {
+      "epoch": 3.7362637362637363,
+      "grad_norm": 0.0751447007060051,
+      "learning_rate": 1.374514719993575e-07,
+      "loss": 1.1647,
+      "step": 1360
+    },
+    {
+      "epoch": 3.741758241758242,
+      "grad_norm": 0.061782509088516235,
+      "learning_rate": 1.3192300703032733e-07,
+      "loss": 1.1672,
+      "step": 1362
+    },
+    {
+      "epoch": 3.7472527472527473,
+      "grad_norm": 0.06823485344648361,
+      "learning_rate": 1.2650653064473106e-07,
+      "loss": 1.1682,
+      "step": 1364
+    },
+    {
+      "epoch": 3.7527472527472527,
+      "grad_norm": 0.061277277767658234,
+      "learning_rate": 1.2120216744690716e-07,
+      "loss": 1.1671,
+      "step": 1366
+    },
+    {
+      "epoch": 3.758241758241758,
+      "grad_norm": 0.06526022404432297,
+      "learning_rate": 1.1601003946206723e-07,
+      "loss": 1.1643,
+      "step": 1368
+    },
+    {
+      "epoch": 3.7637362637362637,
+      "grad_norm": 0.06104410067200661,
+      "learning_rate": 1.1093026613348601e-07,
+      "loss": 1.1574,
+      "step": 1370
+    },
+    {
+      "epoch": 3.769230769230769,
+      "grad_norm": 0.07134535163640976,
+      "learning_rate": 1.0596296431975406e-07,
+      "loss": 1.1614,
+      "step": 1372
+    },
+    {
+      "epoch": 3.7747252747252746,
+      "grad_norm": 0.06213730573654175,
+      "learning_rate": 1.0110824829209164e-07,
+      "loss": 1.1657,
+      "step": 1374
+    },
+    {
+      "epoch": 3.78021978021978,
+      "grad_norm": 0.06847725808620453,
+      "learning_rate": 9.636622973171583e-08,
+      "loss": 1.1663,
+      "step": 1376
+    },
+    {
+      "epoch": 3.7857142857142856,
+      "grad_norm": 0.07414616644382477,
+      "learning_rate": 9.17370177272775e-08,
+      "loss": 1.1668,
+      "step": 1378
+    },
+    {
+      "epoch": 3.791208791208791,
+      "grad_norm": 0.06581508368253708,
+      "learning_rate": 8.72207187723445e-08,
+      "loss": 1.1693,
+      "step": 1380
+    },
+    {
+      "epoch": 3.7967032967032965,
+      "grad_norm": 0.06397808343172073,
+      "learning_rate": 8.281743676295639e-08,
+      "loss": 1.159,
+      "step": 1382
+    },
+    {
+      "epoch": 3.802197802197802,
+      "grad_norm": 0.06523909419775009,
+      "learning_rate": 7.852727299523577e-08,
+      "loss": 1.1562,
+      "step": 1384
+    },
+    {
+      "epoch": 3.8076923076923075,
+      "grad_norm": 0.06779211014509201,
+      "learning_rate": 7.435032616305238e-08,
+      "loss": 1.1727,
+      "step": 1386
+    },
+    {
+      "epoch": 3.813186813186813,
+      "grad_norm": 0.06538354605436325,
+      "learning_rate": 7.028669235575714e-08,
+      "loss": 1.16,
+      "step": 1388
+    },
+    {
+      "epoch": 3.8186813186813184,
+      "grad_norm": 0.06357073783874512,
+      "learning_rate": 6.633646505597113e-08,
+      "loss": 1.1637,
+      "step": 1390
+    },
+    {
+      "epoch": 3.824175824175824,
+      "grad_norm": 0.0635320246219635,
+      "learning_rate": 6.249973513743345e-08,
+      "loss": 1.1639,
+      "step": 1392
+    },
+    {
+      "epoch": 3.82967032967033,
+      "grad_norm": 0.06830117851495743,
+      "learning_rate": 5.8776590862911764e-08,
+      "loss": 1.1585,
+      "step": 1394
+    },
+    {
+      "epoch": 3.8351648351648353,
+      "grad_norm": 0.06444601714611053,
+      "learning_rate": 5.5167117882171104e-08,
+      "loss": 1.1655,
+      "step": 1396
+    },
+    {
+      "epoch": 3.840659340659341,
+      "grad_norm": 0.06375352293252945,
+      "learning_rate": 5.167139923000553e-08,
+      "loss": 1.167,
+      "step": 1398
+    },
+    {
+      "epoch": 3.8461538461538463,
+      "grad_norm": 0.06707257032394409,
+      "learning_rate": 4.828951532432457e-08,
+      "loss": 1.171,
+      "step": 1400
+    },
+    {
+      "epoch": 3.8516483516483517,
+      "grad_norm": 0.06371629983186722,
+      "learning_rate": 4.5021543964306466e-08,
+      "loss": 1.1648,
+      "step": 1402
+    },
+    {
+      "epoch": 3.857142857142857,
+      "grad_norm": 0.0662907212972641,
+      "learning_rate": 4.186756032860728e-08,
+      "loss": 1.1609,
+      "step": 1404
+    },
+    {
+      "epoch": 3.8626373626373627,
+      "grad_norm": 0.06402067840099335,
+      "learning_rate": 3.8827636973630126e-08,
+      "loss": 1.1653,
+      "step": 1406
+    },
+    {
+      "epoch": 3.868131868131868,
+      "grad_norm": 0.07026324421167374,
+      "learning_rate": 3.590184383185758e-08,
+      "loss": 1.1583,
+      "step": 1408
+    },
+    {
+      "epoch": 3.8736263736263736,
+      "grad_norm": 0.06799634546041489,
+      "learning_rate": 3.309024821024354e-08,
+      "loss": 1.1688,
+      "step": 1410
+    },
+    {
+      "epoch": 3.879120879120879,
+      "grad_norm": 0.06444913148880005,
+      "learning_rate": 3.039291478866169e-08,
+      "loss": 1.1616,
+      "step": 1412
+    },
+    {
+      "epoch": 3.8846153846153846,
+      "grad_norm": 0.06899133324623108,
+      "learning_rate": 2.7809905618422227e-08,
+      "loss": 1.1598,
+      "step": 1414
+    },
+    {
+      "epoch": 3.89010989010989,
+      "grad_norm": 0.06785506010055542,
+      "learning_rate": 2.534128012083914e-08,
+      "loss": 1.1532,
+      "step": 1416
+    },
+    {
+      "epoch": 3.8956043956043955,
+      "grad_norm": 0.06701900064945221,
+      "learning_rate": 2.298709508586794e-08,
+      "loss": 1.1654,
+      "step": 1418
+    },
+    {
+      "epoch": 3.901098901098901,
+      "grad_norm": 0.06285455822944641,
+      "learning_rate": 2.074740467079672e-08,
+      "loss": 1.157,
+      "step": 1420
+    },
+    {
+      "epoch": 3.9065934065934065,
+      "grad_norm": 0.06540726125240326,
+      "learning_rate": 1.862226039899995e-08,
+      "loss": 1.1608,
+      "step": 1422
+    },
+    {
+      "epoch": 3.912087912087912,
+      "grad_norm": 0.07258068770170212,
+      "learning_rate": 1.661171115875493e-08,
+      "loss": 1.1696,
+      "step": 1424
+    },
+    {
+      "epoch": 3.9175824175824174,
+      "grad_norm": 0.06025758385658264,
+      "learning_rate": 1.4715803202116075e-08,
+      "loss": 1.1644,
+      "step": 1426
+    },
+    {
+      "epoch": 3.9230769230769234,
+      "grad_norm": 0.07302309572696686,
+      "learning_rate": 1.2934580143851294e-08,
+      "loss": 1.1621,
+      "step": 1428
+    },
+    {
+      "epoch": 3.928571428571429,
+      "grad_norm": 0.05992519110441208,
+      "learning_rate": 1.1268082960436688e-08,
+      "loss": 1.1689,
+      "step": 1430
+    },
+    {
+      "epoch": 3.9340659340659343,
+      "grad_norm": 0.06481453776359558,
+      "learning_rate": 9.716349989118412e-09,
+      "loss": 1.166,
+      "step": 1432
+    },
+    {
+      "epoch": 3.9395604395604398,
+      "grad_norm": 0.07549932599067688,
+      "learning_rate": 8.279416927026163e-09,
+      "loss": 1.159,
+      "step": 1434
+    },
+    {
+      "epoch": 3.9450549450549453,
+      "grad_norm": 0.06642387062311172,
+      "learning_rate": 6.9573168303532775e-09,
+      "loss": 1.1602,
+      "step": 1436
+    },
+    {
+      "epoch": 3.9505494505494507,
+      "grad_norm": 0.0685223713517189,
+      "learning_rate": 5.750080113598455e-09,
+      "loss": 1.1659,
+      "step": 1438
+    },
+    {
+      "epoch": 3.956043956043956,
+      "grad_norm": 0.06598740071058273,
+      "learning_rate": 4.65773454886298e-09,
+      "loss": 1.1607,
+      "step": 1440
+    },
+    {
+      "epoch": 3.9615384615384617,
+      "grad_norm": 0.06653392314910889,
+      "learning_rate": 3.6803052652134572e-09,
+      "loss": 1.1545,
+      "step": 1442
+    },
+    {
+      "epoch": 3.967032967032967,
+      "grad_norm": 0.06884676963090897,
+      "learning_rate": 2.817814748104497e-09,
+      "loss": 1.1557,
+      "step": 1444
+    },
+    {
+      "epoch": 3.9725274725274726,
+      "grad_norm": 0.06552927196025848,
+      "learning_rate": 2.070282838859683e-09,
+      "loss": 1.1591,
+      "step": 1446
+    },
+    {
+      "epoch": 3.978021978021978,
+      "grad_norm": 0.06410259753465652,
+      "learning_rate": 1.4377267342158274e-09,
+      "loss": 1.169,
+      "step": 1448
+    },
+    {
+      "epoch": 3.9835164835164836,
+      "grad_norm": 0.06548061966896057,
+      "learning_rate": 9.201609859271765e-10,
+      "loss": 1.1665,
+      "step": 1450
+    },
+    {
+      "epoch": 3.989010989010989,
+      "grad_norm": 0.06399548798799515,
+      "learning_rate": 5.17597500432343e-10,
+      "loss": 1.1717,
+      "step": 1452
+    },
+    {
+      "epoch": 3.9945054945054945,
+      "grad_norm": 0.06421104073524475,
+      "learning_rate": 2.3004553857675082e-10,
+      "loss": 1.1725,
+      "step": 1454
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.06589093804359436,
+      "learning_rate": 5.751171540391287e-11,
+      "loss": 1.1709,
+      "step": 1456
+    }
+  ],
+  "logging_steps": 2,
+  "max_steps": 1456,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 4,
+  "save_steps": 364,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 3.7105900771564585e+18,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}