diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -1,4524 +1,1584 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 15.0,
+  "epoch": 10.0,
   "eval_steps": 500,
-  "global_step": 3120,
+  "global_step": 1040,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.004807692307692308,
-      "grad_norm": 2.21875,
-      "learning_rate": 6.41025641025641e-07,
-      "loss": 3.0298,
+      "epoch": 0.009615384615384616,
+      "grad_norm": 4.21875,
+      "learning_rate": 1.9230769230769234e-06,
+      "loss": 3.0033,
       "step": 1
     },
     {
-      "epoch": 0.02403846153846154,
-      "grad_norm": 2.1875,
-      "learning_rate": 3.205128205128205e-06,
-      "loss": 2.9989,
+      "epoch": 0.04807692307692308,
+      "grad_norm": 4.0,
+      "learning_rate": 9.615384615384616e-06,
+      "loss": 3.0098,
       "step": 5
     },
     {
-      "epoch": 0.04807692307692308,
-      "grad_norm": 2.0625,
-      "learning_rate": 6.41025641025641e-06,
-      "loss": 3.0133,
+      "epoch": 0.09615384615384616,
+      "grad_norm": 3.15625,
+      "learning_rate": 1.923076923076923e-05,
+      "loss": 2.9448,
       "step": 10
     },
     {
-      "epoch": 0.07211538461538461,
-      "grad_norm": 2.609375,
-      "learning_rate": 9.615384615384616e-06,
-      "loss": 3.0227,
+      "epoch": 0.14423076923076922,
+      "grad_norm": 4.15625,
+      "learning_rate": 2.8846153846153845e-05,
+      "loss": 2.7074,
       "step": 15
     },
     {
-      "epoch": 0.09615384615384616,
-      "grad_norm": 2.015625,
-      "learning_rate": 1.282051282051282e-05,
-      "loss": 2.9874,
+      "epoch": 0.19230769230769232,
+      "grad_norm": 1.8984375,
+      "learning_rate": 3.846153846153846e-05,
+      "loss": 2.4264,
       "step": 20
     },
     {
-      "epoch": 0.1201923076923077,
-      "grad_norm": 2.234375,
-      "learning_rate": 1.602564102564103e-05,
-      "loss": 2.9698,
+      "epoch": 0.2403846153846154,
+      "grad_norm": 3.75,
+      "learning_rate": 4.8076923076923084e-05,
+      "loss": 2.1972,
       "step": 25
     },
     {
-      "epoch": 0.14423076923076922,
-      "grad_norm": 1.8828125,
-      "learning_rate": 1.923076923076923e-05,
-      "loss": 2.872,
+      "epoch": 0.28846153846153844,
+      "grad_norm": 1.3671875,
+      "learning_rate": 5.769230769230769e-05,
+      "loss": 1.9962,
       "step": 30
     },
     {
-      "epoch": 0.16826923076923078,
-      "grad_norm": 1.8359375,
-      "learning_rate": 2.2435897435897437e-05,
-      "loss": 2.7651,
+      "epoch": 0.33653846153846156,
+      "grad_norm": 5.53125,
+      "learning_rate": 6.730769230769232e-05,
+      "loss": 1.8086,
       "step": 35
     },
     {
-      "epoch": 0.19230769230769232,
-      "grad_norm": 1.578125,
-      "learning_rate": 2.564102564102564e-05,
-      "loss": 2.6545,
+      "epoch": 0.38461538461538464,
+      "grad_norm": 0.65625,
+      "learning_rate": 7.692307692307693e-05,
+      "loss": 1.6425,
       "step": 40
     },
     {
-      "epoch": 0.21634615384615385,
-      "grad_norm": 1.3125,
-      "learning_rate": 2.8846153846153845e-05,
-      "loss": 2.4986,
+      "epoch": 0.4326923076923077,
+      "grad_norm": 0.60546875,
+      "learning_rate": 8.653846153846155e-05,
+      "loss": 1.5058,
       "step": 45
     },
     {
-      "epoch": 0.2403846153846154,
-      "grad_norm": 1.1171875,
-      "learning_rate": 3.205128205128206e-05,
-      "loss": 2.4122,
+      "epoch": 0.4807692307692308,
+      "grad_norm": 0.5625,
+      "learning_rate": 9.615384615384617e-05,
+      "loss": 1.401,
       "step": 50
     },
     {
-      "epoch": 0.2644230769230769,
-      "grad_norm": 1.078125,
-      "learning_rate": 3.525641025641026e-05,
-      "loss": 2.301,
+      "epoch": 0.5288461538461539,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.00010576923076923077,
+      "loss": 1.3223,
       "step": 55
     },
     {
-      "epoch": 0.28846153846153844,
-      "grad_norm": 1.8828125,
-      "learning_rate": 3.846153846153846e-05,
-      "loss": 2.1926,
+      "epoch": 0.5769230769230769,
+      "grad_norm": 0.2265625,
+      "learning_rate": 0.00011538461538461538,
+      "loss": 1.2631,
       "step": 60
     },
     {
-      "epoch": 0.3125,
-      "grad_norm": 1.0078125,
-      "learning_rate": 4.166666666666667e-05,
-      "loss": 2.0844,
+      "epoch": 0.625,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.000125,
+      "loss": 1.2263,
       "step": 65
     },
     {
-      "epoch": 0.33653846153846156,
-      "grad_norm": 0.78125,
-      "learning_rate": 4.4871794871794874e-05,
-      "loss": 1.9755,
+      "epoch": 0.6730769230769231,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.00013461538461538464,
+      "loss": 1.1954,
       "step": 70
     },
     {
-      "epoch": 0.3605769230769231,
-      "grad_norm": 1.0703125,
-      "learning_rate": 4.8076923076923084e-05,
-      "loss": 1.8602,
+      "epoch": 0.7211538461538461,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.00014423076923076924,
+      "loss": 1.1671,
       "step": 75
     },
     {
-      "epoch": 0.38461538461538464,
-      "grad_norm": 1.4296875,
-      "learning_rate": 5.128205128205128e-05,
-      "loss": 1.7791,
+      "epoch": 0.7692307692307693,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 0.00015384615384615385,
+      "loss": 1.1505,
       "step": 80
     },
     {
-      "epoch": 0.40865384615384615,
-      "grad_norm": 4.3125,
-      "learning_rate": 5.448717948717948e-05,
-      "loss": 1.698,
+      "epoch": 0.8173076923076923,
+      "grad_norm": 0.88671875,
+      "learning_rate": 0.00016346153846153846,
+      "loss": 1.1211,
       "step": 85
     },
     {
-      "epoch": 0.4326923076923077,
-      "grad_norm": 1.375,
-      "learning_rate": 5.769230769230769e-05,
-      "loss": 1.6469,
+      "epoch": 0.8653846153846154,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.0001730769230769231,
+      "loss": 1.1226,
       "step": 90
     },
     {
-      "epoch": 0.4567307692307692,
-      "grad_norm": 0.4375,
-      "learning_rate": 6.089743589743589e-05,
-      "loss": 1.5753,
+      "epoch": 0.9134615384615384,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.0001826923076923077,
+      "loss": 1.114,
       "step": 95
     },
     {
-      "epoch": 0.4807692307692308,
-      "grad_norm": 1.234375,
-      "learning_rate": 6.410256410256412e-05,
-      "loss": 1.5187,
+      "epoch": 0.9615384615384616,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.00019230769230769233,
+      "loss": 1.0978,
       "step": 100
     },
     {
-      "epoch": 0.5048076923076923,
-      "grad_norm": 0.490234375,
-      "learning_rate": 6.730769230769232e-05,
-      "loss": 1.4681,
+      "epoch": 1.0,
+      "eval_loss": 2.483147621154785,
+      "eval_runtime": 0.4801,
+      "eval_samples_per_second": 20.827,
+      "eval_steps_per_second": 2.083,
+      "step": 104
+    },
+    {
+      "epoch": 1.0096153846153846,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.0001999994367286727,
+      "loss": 1.0851,
       "step": 105
     },
     {
-      "epoch": 0.5288461538461539,
-      "grad_norm": 0.53125,
-      "learning_rate": 7.051282051282052e-05,
-      "loss": 1.4296,
+      "epoch": 1.0576923076923077,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.00019997972289848503,
+      "loss": 1.0778,
       "step": 110
     },
     {
-      "epoch": 0.5528846153846154,
-      "grad_norm": 0.396484375,
-      "learning_rate": 7.371794871794872e-05,
-      "loss": 1.3838,
+      "epoch": 1.1057692307692308,
+      "grad_norm": 0.91015625,
+      "learning_rate": 0.00019993185184710165,
+      "loss": 1.0708,
       "step": 115
     },
     {
-      "epoch": 0.5769230769230769,
-      "grad_norm": 0.37109375,
-      "learning_rate": 7.692307692307693e-05,
-      "loss": 1.3593,
+      "epoch": 1.1538461538461537,
+      "grad_norm": 1.484375,
+      "learning_rate": 0.00019985583705641418,
+      "loss": 1.0615,
       "step": 120
     },
     {
-      "epoch": 0.6009615384615384,
-      "grad_norm": 0.34375,
-      "learning_rate": 8.012820512820514e-05,
-      "loss": 1.3277,
+      "epoch": 1.2019230769230769,
+      "grad_norm": 0.625,
+      "learning_rate": 0.00019975169993441627,
+      "loss": 1.0549,
       "step": 125
     },
     {
-      "epoch": 0.625,
-      "grad_norm": 0.400390625,
-      "learning_rate": 8.333333333333334e-05,
-      "loss": 1.3011,
+      "epoch": 1.25,
+      "grad_norm": 0.416015625,
+      "learning_rate": 0.00019961946980917456,
+      "loss": 1.039,
       "step": 130
     },
     {
-      "epoch": 0.6490384615384616,
-      "grad_norm": 0.765625,
-      "learning_rate": 8.653846153846155e-05,
-      "loss": 1.2784,
+      "epoch": 1.2980769230769231,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.0001994591839205691,
+      "loss": 1.0429,
       "step": 135
     },
     {
-      "epoch": 0.6730769230769231,
-      "grad_norm": 0.279296875,
-      "learning_rate": 8.974358974358975e-05,
-      "loss": 1.2583,
+      "epoch": 1.3461538461538463,
+      "grad_norm": 0.478515625,
+      "learning_rate": 0.0001992708874098054,
+      "loss": 1.0433,
       "step": 140
     },
     {
-      "epoch": 0.6971153846153846,
-      "grad_norm": 0.353515625,
-      "learning_rate": 9.294871794871795e-05,
-      "loss": 1.2399,
+      "epoch": 1.3942307692307692,
+      "grad_norm": 0.40234375,
+      "learning_rate": 0.00019905463330670143,
+      "loss": 1.0285,
       "step": 145
     },
     {
-      "epoch": 0.7211538461538461,
-      "grad_norm": 0.357421875,
-      "learning_rate": 9.615384615384617e-05,
-      "loss": 1.2212,
+      "epoch": 1.4423076923076923,
+      "grad_norm": 0.37109375,
+      "learning_rate": 0.0001988104825147528,
+      "loss": 1.0263,
       "step": 150
     },
     {
-      "epoch": 0.7451923076923077,
-      "grad_norm": 0.421875,
-      "learning_rate": 9.935897435897437e-05,
-      "loss": 1.2205,
+      "epoch": 1.4903846153846154,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.0001985385037939806,
+      "loss": 1.0219,
       "step": 155
     },
     {
-      "epoch": 0.7692307692307693,
-      "grad_norm": 0.73046875,
-      "learning_rate": 0.00010256410256410256,
-      "loss": 1.2004,
+      "epoch": 1.5384615384615383,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.00019823877374156647,
+      "loss": 1.0197,
       "step": 160
     },
     {
-      "epoch": 0.7932692307692307,
-      "grad_norm": 0.8125,
-      "learning_rate": 0.00010576923076923077,
-      "loss": 1.1779,
+      "epoch": 1.5865384615384617,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.00019791137677028082,
+      "loss": 1.0135,
       "step": 165
     },
     {
-      "epoch": 0.8173076923076923,
-      "grad_norm": 0.76953125,
-      "learning_rate": 0.00010897435897435896,
-      "loss": 1.1786,
+      "epoch": 1.6346153846153846,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.00019755640508470942,
+      "loss": 1.0121,
       "step": 170
     },
     {
-      "epoch": 0.8413461538461539,
-      "grad_norm": 0.392578125,
-      "learning_rate": 0.00011217948717948718,
-      "loss": 1.1734,
+      "epoch": 1.6826923076923077,
+      "grad_norm": 0.419921875,
+      "learning_rate": 0.00019717395865528602,
+      "loss": 1.0105,
       "step": 175
     },
     {
-      "epoch": 0.8653846153846154,
-      "grad_norm": 0.451171875,
-      "learning_rate": 0.00011538461538461538,
-      "loss": 1.1658,
+      "epoch": 1.7307692307692308,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.00019676414519013781,
+      "loss": 1.0087,
       "step": 180
     },
     {
-      "epoch": 0.8894230769230769,
-      "grad_norm": 0.73046875,
-      "learning_rate": 0.0001185897435897436,
-      "loss": 1.1622,
+      "epoch": 1.7788461538461537,
+      "grad_norm": 0.79296875,
+      "learning_rate": 0.00019632708010475165,
+      "loss": 1.0026,
       "step": 185
     },
     {
-      "epoch": 0.9134615384615384,
-      "grad_norm": 0.55078125,
-      "learning_rate": 0.00012179487179487179,
-      "loss": 1.1533,
+      "epoch": 1.8269230769230769,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.00019586288648946947,
+      "loss": 1.0078,
       "step": 190
     },
     {
-      "epoch": 0.9375,
-      "grad_norm": 0.39453125,
-      "learning_rate": 0.000125,
-      "loss": 1.1445,
+      "epoch": 1.875,
+      "grad_norm": 0.21484375,
+      "learning_rate": 0.0001953716950748227,
+      "loss": 1.0,
       "step": 195
     },
     {
-      "epoch": 0.9615384615384616,
-      "grad_norm": 0.4375,
-      "learning_rate": 0.00012820512820512823,
-      "loss": 1.1324,
+      "epoch": 1.9230769230769231,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.00019485364419471454,
+      "loss": 0.9957,
       "step": 200
     },
     {
-      "epoch": 0.9855769230769231,
-      "grad_norm": 0.69921875,
-      "learning_rate": 0.0001314102564102564,
-      "loss": 1.1289,
+      "epoch": 1.9711538461538463,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.0001943088797474612,
+      "loss": 0.9985,
       "step": 205
     },
     {
-      "epoch": 1.0,
-      "eval_loss": 2.516178607940674,
-      "eval_runtime": 0.5377,
-      "eval_samples_per_second": 18.598,
-      "eval_steps_per_second": 1.86,
+      "epoch": 2.0,
+      "eval_loss": 2.4666388034820557,
+      "eval_runtime": 0.4867,
+      "eval_samples_per_second": 20.548,
+      "eval_steps_per_second": 2.055,
       "step": 208
     },
     {
-      "epoch": 1.0096153846153846,
-      "grad_norm": 0.279296875,
-      "learning_rate": 0.00013461538461538464,
-      "loss": 1.1241,
+      "epoch": 2.019230769230769,
+      "grad_norm": 0.38671875,
+      "learning_rate": 0.00019373755515470254,
+      "loss": 0.9897,
       "step": 210
     },
     {
-      "epoch": 1.0336538461538463,
-      "grad_norm": 0.310546875,
-      "learning_rate": 0.00013782051282051284,
-      "loss": 1.1188,
+      "epoch": 2.0673076923076925,
+      "grad_norm": 0.453125,
+      "learning_rate": 0.00019313983131819407,
+      "loss": 0.9717,
       "step": 215
     },
     {
-      "epoch": 1.0576923076923077,
-      "grad_norm": 0.267578125,
-      "learning_rate": 0.00014102564102564104,
-      "loss": 1.1222,
+      "epoch": 2.1153846153846154,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.00019251587657449236,
+      "loss": 0.9697,
       "step": 220
     },
     {
-      "epoch": 1.0817307692307692,
-      "grad_norm": 0.392578125,
-      "learning_rate": 0.00014423076923076924,
-      "loss": 1.1088,
+      "epoch": 2.1634615384615383,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.0001918658666475465,
+      "loss": 0.9697,
       "step": 225
     },
     {
-      "epoch": 1.1057692307692308,
-      "grad_norm": 0.28125,
-      "learning_rate": 0.00014743589743589745,
-      "loss": 1.113,
+      "epoch": 2.2115384615384617,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.00019118998459920902,
+      "loss": 0.9735,
       "step": 230
     },
     {
-      "epoch": 1.1298076923076923,
-      "grad_norm": 0.326171875,
-      "learning_rate": 0.00015064102564102565,
-      "loss": 1.0952,
+      "epoch": 2.2596153846153846,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.0001904884207776804,
+      "loss": 0.9675,
       "step": 235
     },
     {
-      "epoch": 1.1538461538461537,
-      "grad_norm": 0.7109375,
-      "learning_rate": 0.00015384615384615385,
-      "loss": 1.1041,
+      "epoch": 2.3076923076923075,
+      "grad_norm": 0.41015625,
+      "learning_rate": 0.0001897613727639014,
+      "loss": 0.9654,
       "step": 240
     },
     {
-      "epoch": 1.1778846153846154,
-      "grad_norm": 1.109375,
-      "learning_rate": 0.00015705128205128205,
-      "loss": 1.0966,
+      "epoch": 2.355769230769231,
+      "grad_norm": 0.25,
+      "learning_rate": 0.00018900904531590846,
+      "loss": 0.9655,
       "step": 245
     },
     {
-      "epoch": 1.2019230769230769,
-      "grad_norm": 0.6875,
-      "learning_rate": 0.00016025641025641028,
-      "loss": 1.0921,
+      "epoch": 2.4038461538461537,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.0001882316503111678,
+      "loss": 0.9653,
       "step": 250
     },
     {
-      "epoch": 1.2259615384615385,
-      "grad_norm": 0.9296875,
-      "learning_rate": 0.00016346153846153846,
-      "loss": 1.0783,
+      "epoch": 2.451923076923077,
+      "grad_norm": 0.81640625,
+      "learning_rate": 0.00018742940668690464,
+      "loss": 0.9729,
       "step": 255
     },
     {
-      "epoch": 1.25,
-      "grad_norm": 1.0390625,
-      "learning_rate": 0.0001666666666666667,
-      "loss": 1.0807,
+      "epoch": 2.5,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00018660254037844388,
+      "loss": 0.9619,
       "step": 260
     },
     {
-      "epoch": 1.2740384615384617,
-      "grad_norm": 0.42578125,
-      "learning_rate": 0.00016987179487179486,
-      "loss": 1.0819,
+      "epoch": 2.5480769230769234,
+      "grad_norm": 0.30859375,
+      "learning_rate": 0.00018575128425558023,
+      "loss": 0.9702,
       "step": 265
     },
     {
-      "epoch": 1.2980769230769231,
-      "grad_norm": 0.61328125,
-      "learning_rate": 0.0001730769230769231,
-      "loss": 1.0849,
+      "epoch": 2.5961538461538463,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.00018487587805699526,
+      "loss": 0.964,
       "step": 270
     },
     {
-      "epoch": 1.3221153846153846,
-      "grad_norm": 0.65234375,
-      "learning_rate": 0.0001762820512820513,
-      "loss": 1.0831,
+      "epoch": 2.644230769230769,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.0001839765683227398,
+      "loss": 0.9603,
       "step": 275
     },
     {
-      "epoch": 1.3461538461538463,
-      "grad_norm": 0.482421875,
-      "learning_rate": 0.0001794871794871795,
-      "loss": 1.0799,
+      "epoch": 2.6923076923076925,
+      "grad_norm": 0.455078125,
+      "learning_rate": 0.00018305360832480117,
+      "loss": 0.9593,
       "step": 280
     },
     {
-      "epoch": 1.3701923076923077,
-      "grad_norm": 0.58203125,
-      "learning_rate": 0.0001826923076923077,
-      "loss": 1.0656,
+      "epoch": 2.7403846153846154,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.00018210725799577439,
+      "loss": 0.963,
       "step": 285
     },
     {
-      "epoch": 1.3942307692307692,
-      "grad_norm": 1.0078125,
-      "learning_rate": 0.0001858974358974359,
-      "loss": 1.0645,
+      "epoch": 2.7884615384615383,
+      "grad_norm": 0.76953125,
+      "learning_rate": 0.00018113778385565733,
+      "loss": 0.9626,
       "step": 290
     },
     {
-      "epoch": 1.4182692307692308,
-      "grad_norm": 0.94140625,
-      "learning_rate": 0.0001891025641025641,
-      "loss": 1.0651,
+      "epoch": 2.8365384615384617,
+      "grad_norm": 1.515625,
+      "learning_rate": 0.00018014545893679115,
+      "loss": 0.9655,
       "step": 295
     },
     {
-      "epoch": 1.4423076923076923,
-      "grad_norm": 0.333984375,
-      "learning_rate": 0.00019230769230769233,
-      "loss": 1.0636,
+      "epoch": 2.8846153846153846,
+      "grad_norm": 0.44921875,
+      "learning_rate": 0.0001791305627069662,
+      "loss": 0.9599,
       "step": 300
     },
     {
-      "epoch": 1.4663461538461537,
-      "grad_norm": 0.451171875,
-      "learning_rate": 0.0001955128205128205,
-      "loss": 1.0583,
+      "epoch": 2.9326923076923075,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.00017809338099071577,
+      "loss": 0.9557,
       "step": 305
     },
     {
-      "epoch": 1.4903846153846154,
-      "grad_norm": 0.56640625,
-      "learning_rate": 0.00019871794871794874,
-      "loss": 1.0605,
+      "epoch": 2.980769230769231,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.00017703420588881946,
+      "loss": 0.9543,
       "step": 310
     },
     {
-      "epoch": 1.5144230769230769,
-      "grad_norm": 0.54296875,
-      "learning_rate": 0.0001999994367286727,
-      "loss": 1.0652,
+      "epoch": 3.0,
+      "eval_loss": 2.4561338424682617,
+      "eval_runtime": 0.4747,
+      "eval_samples_per_second": 21.065,
+      "eval_steps_per_second": 2.107,
+      "step": 312
+    },
+    {
+      "epoch": 3.0288461538461537,
+      "grad_norm": 0.2451171875,
+      "learning_rate": 0.0001759533356960391,
+      "loss": 0.946,
       "step": 315
     },
     {
-      "epoch": 1.5384615384615383,
-      "grad_norm": 0.462890625,
-      "learning_rate": 0.00019999599453798524,
-      "loss": 1.047,
+      "epoch": 3.076923076923077,
+      "grad_norm": 0.2255859375,
+      "learning_rate": 0.00017485107481711012,
+      "loss": 0.9384,
       "step": 320
     },
     {
-      "epoch": 1.5625,
-      "grad_norm": 0.345703125,
-      "learning_rate": 0.00019998942319271077,
-      "loss": 1.0475,
+      "epoch": 3.125,
+      "grad_norm": 0.2470703125,
+      "learning_rate": 0.0001737277336810124,
+      "loss": 0.9299,
       "step": 325
     },
     {
-      "epoch": 1.5865384615384617,
-      "grad_norm": 0.65234375,
-      "learning_rate": 0.00019997972289848503,
-      "loss": 1.0528,
+      "epoch": 3.173076923076923,
+      "grad_norm": 0.2255859375,
+      "learning_rate": 0.00017258362865354426,
+      "loss": 0.9322,
       "step": 330
     },
     {
-      "epoch": 1.6105769230769231,
-      "grad_norm": 0.310546875,
-      "learning_rate": 0.0001999668939588574,
-      "loss": 1.0473,
+      "epoch": 3.2211538461538463,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.00017141908194822446,
+      "loss": 0.9328,
       "step": 335
     },
     {
-      "epoch": 1.6346153846153846,
-      "grad_norm": 0.765625,
-      "learning_rate": 0.0001999509367752813,
-      "loss": 1.0484,
+      "epoch": 3.269230769230769,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.00017023442153554777,
+      "loss": 0.9307,
       "step": 340
     },
     {
-      "epoch": 1.6586538461538463,
-      "grad_norm": 0.490234375,
-      "learning_rate": 0.00019993185184710165,
-      "loss": 1.049,
+      "epoch": 3.3173076923076925,
+      "grad_norm": 0.2177734375,
+      "learning_rate": 0.00016902998105061844,
+      "loss": 0.9273,
       "step": 345
     },
     {
-      "epoch": 1.6826923076923077,
-      "grad_norm": 0.66015625,
-      "learning_rate": 0.00019990963977153936,
-      "loss": 1.0437,
+      "epoch": 3.3653846153846154,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.0001678060996991891,
+      "loss": 0.9324,
       "step": 350
     },
     {
-      "epoch": 1.7067307692307692,
-      "grad_norm": 0.349609375,
-      "learning_rate": 0.00019988430124367237,
-      "loss": 1.0392,
-      "step": 355
+      "epoch": 3.4134615384615383,
+      "grad_norm": 0.234375,
+      "learning_rate": 0.00016656312216213034,
+      "loss": 0.9233,
+      "step": 355
     },
     {
-      "epoch": 1.7307692307692308,
-      "grad_norm": 0.439453125,
-      "learning_rate": 0.00019985583705641418,
-      "loss": 1.0471,
+      "epoch": 3.4615384615384617,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.0001653013984983585,
+      "loss": 0.925,
       "step": 360
     },
     {
-      "epoch": 1.7548076923076923,
-      "grad_norm": 0.380859375,
-      "learning_rate": 0.0001998242481004889,
-      "loss": 1.0384,
+      "epoch": 3.5096153846153846,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00016402128404624882,
+      "loss": 0.9351,
       "step": 365
     },
     {
-      "epoch": 1.7788461538461537,
-      "grad_norm": 0.7265625,
-      "learning_rate": 0.00019978953536440336,
-      "loss": 1.0339,
+      "epoch": 3.5576923076923075,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00016272313932356162,
+      "loss": 0.9266,
       "step": 370
     },
     {
-      "epoch": 1.8028846153846154,
-      "grad_norm": 0.67578125,
-      "learning_rate": 0.00019975169993441627,
-      "loss": 1.0365,
+      "epoch": 3.605769230769231,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.0001614073299259101,
+      "loss": 0.9259,
       "step": 375
     },
     {
-      "epoch": 1.8269230769230769,
-      "grad_norm": 0.341796875,
-      "learning_rate": 0.0001997107429945041,
-      "loss": 1.0452,
+      "epoch": 3.6538461538461537,
+      "grad_norm": 0.2109375,
+      "learning_rate": 0.0001600742264237979,
+      "loss": 0.9322,
       "step": 380
     },
     {
-      "epoch": 1.8509615384615383,
-      "grad_norm": 0.51953125,
-      "learning_rate": 0.00019966666582632423,
-      "loss": 1.0371,
+      "epoch": 3.7019230769230766,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.0001587242042582554,
+      "loss": 0.9327,
       "step": 385
     },
     {
-      "epoch": 1.875,
-      "grad_norm": 0.4921875,
-      "learning_rate": 0.00019961946980917456,
-      "loss": 1.0315,
+      "epoch": 3.75,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.0001573576436351046,
+      "loss": 0.9308,
       "step": 390
     },
     {
-      "epoch": 1.8990384615384617,
-      "grad_norm": 0.84765625,
-      "learning_rate": 0.00019956915641995061,
-      "loss": 1.0327,
+      "epoch": 3.7980769230769234,
+      "grad_norm": 0.3984375,
+      "learning_rate": 0.00015597492941788222,
+      "loss": 0.9265,
       "step": 395
     },
     {
-      "epoch": 1.9230769230769231,
-      "grad_norm": 0.478515625,
-      "learning_rate": 0.0001995157272330992,
-      "loss": 1.0252,
+      "epoch": 3.8461538461538463,
+      "grad_norm": 0.41015625,
+      "learning_rate": 0.00015457645101945046,
+      "loss": 0.9273,
       "step": 400
     },
     {
-      "epoch": 1.9471153846153846,
-      "grad_norm": 0.318359375,
-      "learning_rate": 0.0001994591839205691,
-      "loss": 1.0302,
+      "epoch": 3.894230769230769,
+      "grad_norm": 0.25,
+      "learning_rate": 0.00015316260229232727,
+      "loss": 0.9285,
       "step": 405
     },
     {
-      "epoch": 1.9711538461538463,
-      "grad_norm": 0.353515625,
-      "learning_rate": 0.00019939952825175888,
-      "loss": 1.0339,
+      "epoch": 3.9423076923076925,
+      "grad_norm": 0.2138671875,
+      "learning_rate": 0.00015173378141776568,
+      "loss": 0.9243,
       "step": 410
     },
     {
-      "epoch": 1.9951923076923077,
-      "grad_norm": 0.62109375,
-      "learning_rate": 0.00019933676209346138,
-      "loss": 1.0298,
+      "epoch": 3.9903846153846154,
+      "grad_norm": 0.2236328125,
+      "learning_rate": 0.00015029039079361448,
+      "loss": 0.92,
       "step": 415
     },
     {
-      "epoch": 2.0,
-      "eval_loss": 2.4574124813079834,
-      "eval_runtime": 0.5373,
-      "eval_samples_per_second": 18.61,
-      "eval_steps_per_second": 1.861,
+      "epoch": 4.0,
+      "eval_loss": 2.4799063205718994,
+      "eval_runtime": 0.4817,
+      "eval_samples_per_second": 20.76,
+      "eval_steps_per_second": 2.076,
       "step": 416
     },
     {
-      "epoch": 2.019230769230769,
-      "grad_norm": 0.396484375,
-      "learning_rate": 0.0001992708874098054,
-      "loss": 1.0204,
+      "epoch": 4.038461538461538,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.00014883283692099112,
+      "loss": 0.9122,
       "step": 420
     },
     {
-      "epoch": 2.043269230769231,
-      "grad_norm": 0.40625,
-      "learning_rate": 0.00019920190626219423,
-      "loss": 1.0129,
+      "epoch": 4.086538461538462,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.00014736153028979893,
+      "loss": 0.9007,
       "step": 425
     },
     {
-      "epoch": 2.0673076923076925,
-      "grad_norm": 0.6171875,
-      "learning_rate": 0.00019912982080924103,
-      "loss": 1.0075,
+      "epoch": 4.134615384615385,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.00014587688526312143,
+      "loss": 0.9098,
       "step": 430
     },
     {
-      "epoch": 2.0913461538461537,
-      "grad_norm": 0.4375,
-      "learning_rate": 0.00019905463330670143,
-      "loss": 1.019,
+      "epoch": 4.1826923076923075,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.00014437931996052518,
+      "loss": 0.8987,
       "step": 435
     },
     {
-      "epoch": 2.1153846153846154,
-      "grad_norm": 0.44140625,
-      "learning_rate": 0.00019897634610740287,
-      "loss": 0.9987,
+      "epoch": 4.230769230769231,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.00014286925614030542,
+      "loss": 0.9044,
       "step": 440
     },
     {
-      "epoch": 2.139423076923077,
-      "grad_norm": 0.380859375,
-      "learning_rate": 0.00019889496166117094,
-      "loss": 1.016,
+      "epoch": 4.278846153846154,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.00014134711908070631,
+      "loss": 0.8988,
       "step": 445
     },
     {
-      "epoch": 2.1634615384615383,
-      "grad_norm": 0.53515625,
-      "learning_rate": 0.0001988104825147528,
-      "loss": 1.0047,
+      "epoch": 4.326923076923077,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.0001398133374601501,
+      "loss": 0.9037,
       "step": 450
     },
     {
-      "epoch": 2.1875,
-      "grad_norm": 0.86328125,
-      "learning_rate": 0.00019872291131173742,
-      "loss": 1.0157,
+      "epoch": 4.375,
+      "grad_norm": 0.248046875,
+      "learning_rate": 0.000138268343236509,
+      "loss": 0.9034,
       "step": 455
     },
     {
-      "epoch": 2.2115384615384617,
-      "grad_norm": 0.328125,
-      "learning_rate": 0.00019863225079247285,
-      "loss": 1.0115,
+      "epoch": 4.423076923076923,
+      "grad_norm": 0.25,
+      "learning_rate": 0.00013671257152545277,
+      "loss": 0.8994,
       "step": 460
     },
     {
-      "epoch": 2.235576923076923,
-      "grad_norm": 0.82421875,
-      "learning_rate": 0.0001985385037939806,
-      "loss": 1.0117,
+      "epoch": 4.471153846153846,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.00013514646047790775,
+      "loss": 0.9013,
       "step": 465
     },
     {
-      "epoch": 2.2596153846153846,
-      "grad_norm": 0.3515625,
-      "learning_rate": 0.00019844167324986657,
-      "loss": 1.0038,
+      "epoch": 4.519230769230769,
+      "grad_norm": 0.23828125,
+      "learning_rate": 0.0001335704511566605,
+      "loss": 0.9069,
       "step": 470
     },
     {
-      "epoch": 2.2836538461538463,
-      "grad_norm": 0.5390625,
-      "learning_rate": 0.00019834176219022965,
-      "loss": 1.0031,
+      "epoch": 4.5673076923076925,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.00013198498741214166,
+      "loss": 0.9028,
       "step": 475
     },
     {
-      "epoch": 2.3076923076923075,
-      "grad_norm": 0.455078125,
-      "learning_rate": 0.00019823877374156647,
-      "loss": 1.0061,
+      "epoch": 4.615384615384615,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.0001303905157574247,
+      "loss": 0.9059,
       "step": 480
     },
     {
-      "epoch": 2.331730769230769,
-      "grad_norm": 0.34375,
-      "learning_rate": 0.00019813271112667396,
-      "loss": 1.0035,
+      "epoch": 4.663461538461538,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.00012878748524247462,
+      "loss": 0.9004,
       "step": 485
     },
     {
-      "epoch": 2.355769230769231,
-      "grad_norm": 0.43359375,
-      "learning_rate": 0.00019802357766454827,
-      "loss": 1.0054,
+      "epoch": 4.711538461538462,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00012717634732768243,
+      "loss": 0.905,
       "step": 490
     },
     {
-      "epoch": 2.3798076923076925,
-      "grad_norm": 0.3671875,
-      "learning_rate": 0.00019791137677028082,
-      "loss": 1.0012,
+      "epoch": 4.759615384615385,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.0001255575557567207,
+      "loss": 0.9035,
       "step": 495
     },
     {
-      "epoch": 2.4038461538461537,
-      "grad_norm": 0.31640625,
-      "learning_rate": 0.00019779611195495177,
-      "loss": 1.0052,
+      "epoch": 4.8076923076923075,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.0001239315664287558,
+      "loss": 0.8974,
       "step": 500
     },
     {
-      "epoch": 2.4278846153846154,
-      "grad_norm": 0.33203125,
-      "learning_rate": 0.00019767778682551975,
-      "loss": 1.0097,
+      "epoch": 4.855769230769231,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.00012229883727005365,
+      "loss": 0.9034,
       "step": 505
     },
     {
-      "epoch": 2.451923076923077,
-      "grad_norm": 0.353515625,
-      "learning_rate": 0.00019755640508470942,
-      "loss": 1.0108,
+      "epoch": 4.903846153846154,
+      "grad_norm": 0.2431640625,
+      "learning_rate": 0.00012065982810501404,
+      "loss": 0.9056,
       "step": 510
     },
     {
-      "epoch": 2.4759615384615383,
-      "grad_norm": 0.337890625,
-      "learning_rate": 0.0001974319705308951,
-      "loss": 0.9998,
+      "epoch": 4.951923076923077,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.00011901500052667068,
+      "loss": 0.902,
       "step": 515
     },
     {
-      "epoch": 2.5,
-      "grad_norm": 0.67578125,
-      "learning_rate": 0.00019730448705798239,
-      "loss": 0.9987,
+      "epoch": 5.0,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.00011736481776669306,
+      "loss": 0.9016,
       "step": 520
     },
     {
-      "epoch": 2.5240384615384617,
-      "grad_norm": 0.90625,
-      "learning_rate": 0.00019717395865528602,
-      "loss": 1.0114,
+      "epoch": 5.0,
+      "eval_loss": 2.4989540576934814,
+      "eval_runtime": 0.4791,
+      "eval_samples_per_second": 20.875,
+      "eval_steps_per_second": 2.087,
+      "step": 520
+    },
+    {
+      "epoch": 5.048076923076923,
+      "grad_norm": 0.25,
+      "learning_rate": 0.00011570974456492678,
+      "loss": 0.8816,
       "step": 525
     },
     {
-      "epoch": 2.5480769230769234,
-      "grad_norm": 0.416015625,
-      "learning_rate": 0.00019704038940740505,
-      "loss": 1.0076,
+      "epoch": 5.096153846153846,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.00011405024703850929,
+      "loss": 0.8777,
       "step": 530
     },
     {
-      "epoch": 2.5721153846153846,
-      "grad_norm": 0.349609375,
-      "learning_rate": 0.00019690378349409519,
-      "loss": 1.0035,
+      "epoch": 5.144230769230769,
+      "grad_norm": 0.2265625,
+      "learning_rate": 0.00011238679255059752,
+      "loss": 0.8809,
       "step": 535
     },
     {
-      "epoch": 2.5961538461538463,
-      "grad_norm": 0.453125,
-      "learning_rate": 0.00019676414519013781,
-      "loss": 1.0012,
+      "epoch": 5.1923076923076925,
+      "grad_norm": 0.2421875,
+      "learning_rate": 0.00011071984957874479,
+      "loss": 0.882,
       "step": 540
     },
     {
-      "epoch": 2.6201923076923075,
-      "grad_norm": 0.40625,
-      "learning_rate": 0.0001966214788652064,
-      "loss": 0.9969,
+      "epoch": 5.240384615384615,
+      "grad_norm": 0.236328125,
+      "learning_rate": 0.0001090498875829638,
+      "loss": 0.8787,
       "step": 545
     },
     {
-      "epoch": 2.644230769230769,
-      "grad_norm": 0.85546875,
-      "learning_rate": 0.0001964757889837296,
-      "loss": 0.9995,
+      "epoch": 5.288461538461538,
+      "grad_norm": 0.2333984375,
+      "learning_rate": 0.00010737737687351284,
+      "loss": 0.8806,
       "step": 550
     },
     {
-      "epoch": 2.668269230769231,
-      "grad_norm": 0.70703125,
-      "learning_rate": 0.00019632708010475165,
-      "loss": 0.9986,
+      "epoch": 5.336538461538462,
+      "grad_norm": 0.2275390625,
+      "learning_rate": 0.00010570278847844275,
+      "loss": 0.8859,
       "step": 555
     },
     {
-      "epoch": 2.6923076923076925,
-      "grad_norm": 0.63671875,
-      "learning_rate": 0.0001961753568817896,
-      "loss": 0.9974,
+      "epoch": 5.384615384615385,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.00010402659401094152,
+      "loss": 0.8832,
       "step": 560
     },
     {
-      "epoch": 2.7163461538461537,
-      "grad_norm": 0.34765625,
-      "learning_rate": 0.00019602062406268788,
-      "loss": 0.999,
+      "epoch": 5.4326923076923075,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.00010234926553651422,
+      "loss": 0.8803,
       "step": 565
     },
     {
-      "epoch": 2.7403846153846154,
-      "grad_norm": 0.5390625,
-      "learning_rate": 0.00019586288648946947,
-      "loss": 1.0028,
+      "epoch": 5.480769230769231,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.00010067127544003563,
+      "loss": 0.882,
       "step": 570
     },
     {
-      "epoch": 2.7644230769230766,
-      "grad_norm": 0.8515625,
-      "learning_rate": 0.00019570214909818466,
-      "loss": 0.9927,
+      "epoch": 5.528846153846154,
+      "grad_norm": 0.2275390625,
+      "learning_rate": 9.899309629271246e-05,
+      "loss": 0.8797,
       "step": 575
     },
     {
-      "epoch": 2.7884615384615383,
-      "grad_norm": 0.451171875,
-      "learning_rate": 0.0001955384169187563,
-      "loss": 0.9972,
+      "epoch": 5.576923076923077,
+      "grad_norm": 0.2734375,
+      "learning_rate": 9.73152007189939e-05,
+      "loss": 0.8818,
       "step": 580
     },
     {
-      "epoch": 2.8125,
-      "grad_norm": 0.291015625,
-      "learning_rate": 0.0001953716950748227,
-      "loss": 1.0003,
+      "epoch": 5.625,
+      "grad_norm": 0.3828125,
+      "learning_rate": 9.563806126346642e-05,
+      "loss": 0.8864,
       "step": 585
     },
     {
-      "epoch": 2.8365384615384617,
-      "grad_norm": 0.63671875,
-      "learning_rate": 0.00019520198878357703,
-      "loss": 0.9956,
+      "epoch": 5.673076923076923,
+      "grad_norm": 0.2451171875,
+      "learning_rate": 9.396215025777139e-05,
+      "loss": 0.8876,
       "step": 590
     },
     {
-      "epoch": 2.8605769230769234,
-      "grad_norm": 0.609375,
-      "learning_rate": 0.00019502930335560414,
-      "loss": 0.9846,
+      "epoch": 5.721153846153846,
+      "grad_norm": 0.279296875,
+      "learning_rate": 9.22879396875828e-05,
+      "loss": 0.8843,
       "step": 595
     },
     {
-      "epoch": 2.8846153846153846,
-      "grad_norm": 0.435546875,
-      "learning_rate": 0.00019485364419471454,
-      "loss": 1.0018,
+      "epoch": 5.769230769230769,
+      "grad_norm": 0.294921875,
+      "learning_rate": 9.061590105968208e-05,
+      "loss": 0.8832,
       "step": 600
     },
     {
-      "epoch": 2.9086538461538463,
-      "grad_norm": 0.345703125,
-      "learning_rate": 0.00019467501679777511,
-      "loss": 0.9883,
+      "epoch": 5.8173076923076925,
+      "grad_norm": 0.359375,
+      "learning_rate": 8.894650526916803e-05,
+      "loss": 0.884,
       "step": 605
     },
     {
-      "epoch": 2.9326923076923075,
-      "grad_norm": 0.5703125,
-      "learning_rate": 0.00019449342675453707,
-      "loss": 0.9916,
+      "epoch": 5.865384615384615,
+      "grad_norm": 0.388671875,
+      "learning_rate": 8.728022246683894e-05,
+      "loss": 0.8814,
       "step": 610
     },
     {
-      "epoch": 2.956730769230769,
-      "grad_norm": 0.384765625,
-      "learning_rate": 0.0001943088797474612,
-      "loss": 0.9885,
+      "epoch": 5.913461538461538,
+      "grad_norm": 0.33203125,
+      "learning_rate": 8.561752192678443e-05,
+      "loss": 0.8826,
       "step": 615
     },
     {
-      "epoch": 2.980769230769231,
-      "grad_norm": 0.427734375,
-      "learning_rate": 0.00019412138155154002,
-      "loss": 0.9905,
+      "epoch": 5.961538461538462,
+      "grad_norm": 0.328125,
+      "learning_rate": 8.395887191422397e-05,
+      "loss": 0.8871,
       "step": 620
     },
     {
-      "epoch": 3.0,
-      "eval_loss": 2.445496082305908,
-      "eval_runtime": 0.5354,
-      "eval_samples_per_second": 18.679,
-      "eval_steps_per_second": 1.868,
+      "epoch": 6.0,
+      "eval_loss": 2.5249791145324707,
+      "eval_runtime": 0.4816,
+      "eval_samples_per_second": 20.766,
+      "eval_steps_per_second": 2.077,
       "step": 624
     },
     {
-      "epoch": 3.0048076923076925,
-      "grad_norm": 0.6640625,
-      "learning_rate": 0.00019393093803411686,
-      "loss": 0.9988,
+      "epoch": 6.009615384615385,
+      "grad_norm": 0.234375,
+      "learning_rate": 8.23047395536298e-05,
+      "loss": 0.8778,
       "step": 625
     },
     {
-      "epoch": 3.0288461538461537,
-      "grad_norm": 0.578125,
-      "learning_rate": 0.00019373755515470254,
-      "loss": 0.971,
+      "epoch": 6.0576923076923075,
+      "grad_norm": 0.2373046875,
+      "learning_rate": 8.065559069717088e-05,
+      "loss": 0.8638,
       "step": 630
     },
     {
-      "epoch": 3.0528846153846154,
-      "grad_norm": 0.6015625,
-      "learning_rate": 0.0001935412389647887,
-      "loss": 0.9808,
+      "epoch": 6.105769230769231,
+      "grad_norm": 0.248046875,
+      "learning_rate": 7.901188979351526e-05,
+      "loss": 0.8612,
       "step": 635
     },
     {
-      "epoch": 3.076923076923077,
-      "grad_norm": 0.66015625,
-      "learning_rate": 0.0001933419956076584,
-      "loss": 0.9832,
+      "epoch": 6.153846153846154,
+      "grad_norm": 0.2431640625,
+      "learning_rate": 7.73740997570278e-05,
+      "loss": 0.8638,
       "step": 640
     },
     {
-      "epoch": 3.1009615384615383,
-      "grad_norm": 0.52734375,
-      "learning_rate": 0.00019313983131819407,
-      "loss": 0.9735,
+      "epoch": 6.201923076923077,
+      "grad_norm": 0.224609375,
+      "learning_rate": 7.574268183739989e-05,
+      "loss": 0.866,
       "step": 645
     },
     {
-      "epoch": 3.125,
-      "grad_norm": 0.392578125,
-      "learning_rate": 0.00019293475242268223,
-      "loss": 0.9739,
+      "epoch": 6.25,
+      "grad_norm": 0.236328125,
+      "learning_rate": 7.411809548974792e-05,
+      "loss": 0.8673,
       "step": 650
     },
     {
-      "epoch": 3.1490384615384617,
-      "grad_norm": 0.55078125,
-      "learning_rate": 0.00019272676533861564,
-      "loss": 0.9864,
+      "epoch": 6.298076923076923,
+      "grad_norm": 0.263671875,
+      "learning_rate": 7.250079824521743e-05,
+      "loss": 0.87,
       "step": 655
     },
     {
-      "epoch": 3.173076923076923,
-      "grad_norm": 0.8125,
-      "learning_rate": 0.00019251587657449236,
-      "loss": 0.9697,
+      "epoch": 6.346153846153846,
+      "grad_norm": 0.30859375,
+      "learning_rate": 7.089124558212871e-05,
+      "loss": 0.8634,
       "step": 660
     },
     {
-      "epoch": 3.1971153846153846,
-      "grad_norm": 0.4765625,
-      "learning_rate": 0.00019230209272961215,
-      "loss": 0.9854,
+      "epoch": 6.394230769230769,
+      "grad_norm": 0.23828125,
+      "learning_rate": 6.928989079770094e-05,
+      "loss": 0.8688,
       "step": 665
     },
     {
-      "epoch": 3.2211538461538463,
-      "grad_norm": 0.73828125,
-      "learning_rate": 0.0001920854204938699,
-      "loss": 0.9705,
+      "epoch": 6.4423076923076925,
+      "grad_norm": 0.234375,
+      "learning_rate": 6.769718488039023e-05,
+      "loss": 0.864,
       "step": 670
     },
     {
-      "epoch": 3.2451923076923075,
-      "grad_norm": 0.443359375,
-      "learning_rate": 0.0001918658666475465,
-      "loss": 0.9778,
+      "epoch": 6.490384615384615,
+      "grad_norm": 0.236328125,
+      "learning_rate": 6.611357638287823e-05,
+      "loss": 0.8698,
       "step": 675
     },
     {
-      "epoch": 3.269230769230769,
-      "grad_norm": 0.57421875,
-      "learning_rate": 0.00019164343806109632,
-      "loss": 0.9731,
+      "epoch": 6.538461538461538,
+      "grad_norm": 0.2431640625,
+      "learning_rate": 6.453951129574644e-05,
+      "loss": 0.865,
       "step": 680
     },
     {
-      "epoch": 3.293269230769231,
-      "grad_norm": 0.466796875,
-      "learning_rate": 0.0001914181416949325,
-      "loss": 0.969,
+      "epoch": 6.586538461538462,
+      "grad_norm": 0.23046875,
+      "learning_rate": 6.297543292187215e-05,
+      "loss": 0.874,
       "step": 685
     },
     {
-      "epoch": 3.3173076923076925,
-      "grad_norm": 0.4765625,
-      "learning_rate": 0.00019118998459920902,
-      "loss": 0.9752,
+      "epoch": 6.634615384615385,
+      "grad_norm": 0.2294921875,
+      "learning_rate": 6.142178175158149e-05,
+      "loss": 0.8694,
       "step": 690
     },
     {
-      "epoch": 3.3413461538461537,
-      "grad_norm": 0.455078125,
-      "learning_rate": 0.00019095897391360005,
-      "loss": 0.98,
+      "epoch": 6.6826923076923075,
+      "grad_norm": 0.2421875,
+      "learning_rate": 5.9878995338594224e-05,
+      "loss": 0.8688,
       "step": 695
     },
     {
-      "epoch": 3.3653846153846154,
-      "grad_norm": 0.69921875,
-      "learning_rate": 0.00019072511686707663,
-      "loss": 0.9743,
+      "epoch": 6.730769230769231,
+      "grad_norm": 0.23046875,
+      "learning_rate": 5.834750817679606e-05,
+      "loss": 0.8671,
       "step": 700
     },
     {
-      "epoch": 3.389423076923077,
-      "grad_norm": 0.328125,
-      "learning_rate": 0.0001904884207776804,
-      "loss": 0.9641,
+      "epoch": 6.778846153846154,
+      "grad_norm": 0.240234375,
+      "learning_rate": 5.682775157787213e-05,
+      "loss": 0.8666,
       "step": 705
     },
     {
-      "epoch": 3.4134615384615383,
-      "grad_norm": 0.3359375,
-      "learning_rate": 0.00019024889305229456,
-      "loss": 0.9713,
+      "epoch": 6.826923076923077,
+      "grad_norm": 0.2158203125,
+      "learning_rate": 5.5320153549837415e-05,
+      "loss": 0.8623,
       "step": 710
     },
     {
-      "epoch": 3.4375,
-      "grad_norm": 0.33984375,
-      "learning_rate": 0.00019000654118641211,
-      "loss": 0.976,
+      "epoch": 6.875,
+      "grad_norm": 0.21484375,
+      "learning_rate": 5.382513867649663e-05,
+      "loss": 0.8677,
       "step": 715
     },
     {
-      "epoch": 3.4615384615384617,
-      "grad_norm": 0.349609375,
-      "learning_rate": 0.0001897613727639014,
-      "loss": 0.9612,
+      "epoch": 6.923076923076923,
+      "grad_norm": 0.2138671875,
+      "learning_rate": 5.234312799786921e-05,
+      "loss": 0.8654,
       "step": 720
     },
     {
-      "epoch": 3.485576923076923,
-      "grad_norm": 0.357421875,
-      "learning_rate": 0.00018951339545676866,
-      "loss": 0.9762,
+      "epoch": 6.971153846153846,
+      "grad_norm": 0.208984375,
+      "learning_rate": 5.087453889161229e-05,
+      "loss": 0.8635,
       "step": 725
     },
     {
-      "epoch": 3.5096153846153846,
-      "grad_norm": 0.33984375,
-      "learning_rate": 0.00018926261702491797,
-      "loss": 0.9816,
+      "epoch": 7.0,
+      "eval_loss": 2.5363287925720215,
+      "eval_runtime": 0.4813,
+      "eval_samples_per_second": 20.776,
+      "eval_steps_per_second": 2.078,
+      "step": 728
+    },
+    {
+      "epoch": 7.019230769230769,
+      "grad_norm": 0.2373046875,
+      "learning_rate": 4.9419784955474524e-05,
+      "loss": 0.8611,
       "step": 730
     },
     {
-      "epoch": 3.5336538461538463,
-      "grad_norm": 0.3515625,
-      "learning_rate": 0.00018900904531590846,
-      "loss": 0.9667,
+      "epoch": 7.0673076923076925,
+      "grad_norm": 0.2333984375,
+      "learning_rate": 4.797927589081509e-05,
+      "loss": 0.8528,
       "step": 735
     },
     {
-      "epoch": 3.5576923076923075,
-      "grad_norm": 0.419921875,
-      "learning_rate": 0.00018875268826470872,
-      "loss": 0.9744,
+      "epoch": 7.115384615384615,
+      "grad_norm": 0.24609375,
+      "learning_rate": 4.6553417387219886e-05,
+      "loss": 0.858,
       "step": 740
     },
     {
-      "epoch": 3.581730769230769,
-      "grad_norm": 0.57421875,
-      "learning_rate": 0.00018849355389344855,
-      "loss": 0.9664,
+      "epoch": 7.163461538461538,
+      "grad_norm": 0.240234375,
+      "learning_rate": 4.514261100824709e-05,
+      "loss": 0.8525,
       "step": 745
     },
     {
-      "epoch": 3.605769230769231,
-      "grad_norm": 0.333984375,
-      "learning_rate": 0.0001882316503111678,
-      "loss": 0.9722,
+      "epoch": 7.211538461538462,
+      "grad_norm": 0.2373046875,
+      "learning_rate": 4.374725407833532e-05,
+      "loss": 0.8593,
       "step": 750
     },
     {
-      "epoch": 3.6298076923076925,
-      "grad_norm": 0.32421875,
-      "learning_rate": 0.0001879669857135627,
-      "loss": 0.9845,
+      "epoch": 7.259615384615385,
+      "grad_norm": 0.244140625,
+      "learning_rate": 4.236773957090548e-05,
+      "loss": 0.8563,
       "step": 755
     },
     {
-      "epoch": 3.6538461538461537,
-      "grad_norm": 0.349609375,
-      "learning_rate": 0.00018769956838272936,
-      "loss": 0.9671,
+      "epoch": 7.3076923076923075,
+      "grad_norm": 0.259765625,
+      "learning_rate": 4.100445599768774e-05,
+      "loss": 0.8562,
       "step": 760
     },
     {
-      "epoch": 3.6778846153846154,
-      "grad_norm": 0.482421875,
-      "learning_rate": 0.00018742940668690464,
-      "loss": 0.9803,
+      "epoch": 7.355769230769231,
+      "grad_norm": 0.2177734375,
+      "learning_rate": 3.96577872993053e-05,
+      "loss": 0.8574,
       "step": 765
     },
     {
-      "epoch": 3.7019230769230766,
-      "grad_norm": 0.40234375,
-      "learning_rate": 0.00018715650908020427,
-      "loss": 0.9729,
+      "epoch": 7.403846153846154,
+      "grad_norm": 0.224609375,
+      "learning_rate": 3.832811273714569e-05,
+      "loss": 0.85,
       "step": 770
     },
     {
-      "epoch": 3.7259615384615383,
-      "grad_norm": 0.39453125,
-      "learning_rate": 0.00018688088410235833,
-      "loss": 0.9688,
+      "epoch": 7.451923076923077,
+      "grad_norm": 0.2158203125,
+      "learning_rate": 3.701580678654925e-05,
+      "loss": 0.8537,
       "step": 775
     },
     {
-      "epoch": 3.75,
-      "grad_norm": 0.375,
-      "learning_rate": 0.00018660254037844388,
-      "loss": 0.9785,
+      "epoch": 7.5,
+      "grad_norm": 0.25390625,
+      "learning_rate": 3.5721239031346066e-05,
+      "loss": 0.8579,
       "step": 780
     },
     {
-      "epoch": 3.7740384615384617,
-      "grad_norm": 0.462890625,
-      "learning_rate": 0.0001863214866186152,
-      "loss": 0.9762,
+      "epoch": 7.548076923076923,
+      "grad_norm": 0.2099609375,
+      "learning_rate": 3.4444774059770536e-05,
+      "loss": 0.8567,
       "step": 785
     },
     {
-      "epoch": 3.7980769230769234,
-      "grad_norm": 0.3515625,
-      "learning_rate": 0.00018603773161783124,
-      "loss": 0.9631,
+      "epoch": 7.596153846153846,
+      "grad_norm": 0.2177734375,
+      "learning_rate": 3.318677136178228e-05,
+      "loss": 0.8515,
       "step": 790
     },
     {
-      "epoch": 3.8221153846153846,
-      "grad_norm": 0.396484375,
-      "learning_rate": 0.00018575128425558023,
-      "loss": 0.9763,
+      "epoch": 7.644230769230769,
+      "grad_norm": 0.224609375,
+      "learning_rate": 3.1947585227823394e-05,
+      "loss": 0.8519,
       "step": 795
     },
     {
-      "epoch": 3.8461538461538463,
-      "grad_norm": 0.41796875,
-      "learning_rate": 0.00018546215349560203,
-      "loss": 0.9652,
+      "epoch": 7.6923076923076925,
+      "grad_norm": 0.2216796875,
+      "learning_rate": 3.072756464904006e-05,
+      "loss": 0.8611,
       "step": 800
     },
     {
-      "epoch": 3.8701923076923075,
-      "grad_norm": 0.6875,
-      "learning_rate": 0.00018517034838560746,
-      "loss": 0.972,
+      "epoch": 7.740384615384615,
+      "grad_norm": 0.21484375,
+      "learning_rate": 2.9527053218996037e-05,
+      "loss": 0.855,
       "step": 805
     },
     {
-      "epoch": 3.894230769230769,
-      "grad_norm": 0.64453125,
-      "learning_rate": 0.00018487587805699526,
-      "loss": 0.9714,
+      "epoch": 7.788461538461538,
+      "grad_norm": 0.220703125,
+      "learning_rate": 2.8346389036906828e-05,
+      "loss": 0.8529,
       "step": 810
     },
     {
-      "epoch": 3.918269230769231,
-      "grad_norm": 0.4296875,
-      "learning_rate": 0.00018457875172456632,
-      "loss": 0.9705,
+      "epoch": 7.836538461538462,
+      "grad_norm": 0.21484375,
+      "learning_rate": 2.7185904612421176e-05,
+      "loss": 0.8506,
       "step": 815
     },
     {
-      "epoch": 3.9423076923076925,
-      "grad_norm": 0.51953125,
-      "learning_rate": 0.00018427897868623534,
-      "loss": 0.9664,
+      "epoch": 7.884615384615385,
+      "grad_norm": 0.2236328125,
+      "learning_rate": 2.6045926771976303e-05,
+      "loss": 0.8563,
       "step": 820
     },
     {
-      "epoch": 3.9663461538461537,
-      "grad_norm": 0.39453125,
-      "learning_rate": 0.0001839765683227398,
-      "loss": 0.9602,
+      "epoch": 7.9326923076923075,
+      "grad_norm": 0.2109375,
+      "learning_rate": 2.492677656675414e-05,
+      "loss": 0.861,
       "step": 825
     },
     {
-      "epoch": 3.9903846153846154,
-      "grad_norm": 0.73828125,
-      "learning_rate": 0.00018367153009734655,
-      "loss": 0.9668,
+      "epoch": 7.980769230769231,
+      "grad_norm": 0.2177734375,
+      "learning_rate": 2.382876918226409e-05,
+      "loss": 0.8535,
       "step": 830
     },
     {
-      "epoch": 4.0,
-      "eval_loss": 2.4518158435821533,
-      "eval_runtime": 0.5401,
-      "eval_samples_per_second": 18.515,
-      "eval_steps_per_second": 1.852,
+      "epoch": 8.0,
+      "eval_loss": 2.5545527935028076,
+      "eval_runtime": 0.4771,
+      "eval_samples_per_second": 20.959,
+      "eval_steps_per_second": 2.096,
       "step": 832
     },
     {
-      "epoch": 4.014423076923077,
-      "grad_norm": 0.3671875,
-      "learning_rate": 0.00018336387355555553,
-      "loss": 0.9635,
+      "epoch": 8.028846153846153,
+      "grad_norm": 0.2158203125,
+      "learning_rate": 2.2752213849577188e-05,
+      "loss": 0.8505,
       "step": 835
     },
     {
-      "epoch": 4.038461538461538,
-      "grad_norm": 0.482421875,
-      "learning_rate": 0.00018305360832480117,
-      "loss": 0.9579,
+      "epoch": 8.076923076923077,
+      "grad_norm": 0.2177734375,
+      "learning_rate": 2.1697413758237784e-05,
+      "loss": 0.8473,
       "step": 840
     },
     {
-      "epoch": 4.0625,
-      "grad_norm": 0.90234375,
-      "learning_rate": 0.00018274074411415105,
-      "loss": 0.9545,
+      "epoch": 8.125,
+      "grad_norm": 0.220703125,
+      "learning_rate": 2.0664665970876496e-05,
+      "loss": 0.8438,
       "step": 845
     },
     {
-      "epoch": 4.086538461538462,
-      "grad_norm": 0.59765625,
-      "learning_rate": 0.00018242529071400214,
-      "loss": 0.9485,
+      "epoch": 8.173076923076923,
+      "grad_norm": 0.212890625,
+      "learning_rate": 1.965426133954854e-05,
+      "loss": 0.8551,
       "step": 850
     },
     {
-      "epoch": 4.110576923076923,
-      "grad_norm": 0.59765625,
-      "learning_rate": 0.00018210725799577439,
-      "loss": 0.9624,
+      "epoch": 8.221153846153847,
+      "grad_norm": 0.2119140625,
+      "learning_rate": 1.8666484423821373e-05,
+      "loss": 0.8528,
       "step": 855
     },
     {
-      "epoch": 4.134615384615385,
-      "grad_norm": 0.328125,
-      "learning_rate": 0.00018178665591160172,
-      "loss": 0.9596,
+      "epoch": 8.26923076923077,
+      "grad_norm": 0.2236328125,
+      "learning_rate": 1.7701613410634365e-05,
+      "loss": 0.848,
       "step": 860
     },
     {
-      "epoch": 4.158653846153846,
-      "grad_norm": 0.318359375,
-      "learning_rate": 0.0001814634944940209,
-      "loss": 0.9468,
+      "epoch": 8.317307692307692,
+      "grad_norm": 0.216796875,
+      "learning_rate": 1.6759920035953093e-05,
+      "loss": 0.8475,
       "step": 865
     },
     {
-      "epoch": 4.1826923076923075,
-      "grad_norm": 0.33984375,
-      "learning_rate": 0.00018113778385565733,
-      "loss": 0.9489,
+      "epoch": 8.365384615384615,
+      "grad_norm": 0.208984375,
+      "learning_rate": 1.584166950824061e-05,
+      "loss": 0.8535,
       "step": 870
     },
     {
-      "epoch": 4.206730769230769,
-      "grad_norm": 0.353515625,
-      "learning_rate": 0.00018080953418890853,
-      "loss": 0.9569,
+      "epoch": 8.413461538461538,
+      "grad_norm": 0.2119140625,
+      "learning_rate": 1.4947120433767047e-05,
+      "loss": 0.8508,
       "step": 875
     },
     {
-      "epoch": 4.230769230769231,
-      "grad_norm": 0.3515625,
-      "learning_rate": 0.00018047875576562557,
-      "loss": 0.9524,
+      "epoch": 8.461538461538462,
+      "grad_norm": 0.212890625,
+      "learning_rate": 1.4076524743778319e-05,
+      "loss": 0.8459,
       "step": 880
     },
     {
-      "epoch": 4.2548076923076925,
-      "grad_norm": 0.4140625,
-      "learning_rate": 0.00018014545893679115,
-      "loss": 0.9491,
+      "epoch": 8.509615384615385,
+      "grad_norm": 0.21484375,
+      "learning_rate": 1.3230127623545064e-05,
+      "loss": 0.8481,
       "step": 885
     },
     {
-      "epoch": 4.278846153846154,
-      "grad_norm": 0.484375,
-      "learning_rate": 0.0001798096541321961,
-      "loss": 0.9489,
+      "epoch": 8.557692307692308,
+      "grad_norm": 0.2109375,
+      "learning_rate": 1.2408167443311214e-05,
+      "loss": 0.8508,
       "step": 890
     },
     {
-      "epoch": 4.302884615384615,
-      "grad_norm": 0.51953125,
-      "learning_rate": 0.00017947135186011275,
-      "loss": 0.9489,
+      "epoch": 8.60576923076923,
+      "grad_norm": 0.212890625,
+      "learning_rate": 1.1610875691161915e-05,
+      "loss": 0.8489,
       "step": 895
     },
     {
-      "epoch": 4.326923076923077,
-      "grad_norm": 0.439453125,
-      "learning_rate": 0.0001791305627069662,
-      "loss": 0.9573,
+      "epoch": 8.653846153846153,
+      "grad_norm": 0.2060546875,
+      "learning_rate": 1.083847690782972e-05,
+      "loss": 0.8423,
       "step": 900
     },
     {
-      "epoch": 4.350961538461538,
-      "grad_norm": 0.408203125,
-      "learning_rate": 0.00017878729733700302,
-      "loss": 0.9541,
+      "epoch": 8.701923076923077,
+      "grad_norm": 0.22265625,
+      "learning_rate": 1.0091188623457415e-05,
+      "loss": 0.8562,
       "step": 905
     },
     {
-      "epoch": 4.375,
-      "grad_norm": 0.349609375,
-      "learning_rate": 0.00017844156649195759,
-      "loss": 0.9512,
+      "epoch": 8.75,
+      "grad_norm": 0.2138671875,
+      "learning_rate": 9.369221296335006e-06,
+      "loss": 0.8515,
       "step": 910
     },
     {
-      "epoch": 4.399038461538462,
-      "grad_norm": 0.33984375,
-      "learning_rate": 0.00017809338099071577,
-      "loss": 0.9462,
+      "epoch": 8.798076923076923,
+      "grad_norm": 0.2138671875,
+      "learning_rate": 8.672778253628621e-06,
+      "loss": 0.8497,
       "step": 915
     },
     {
-      "epoch": 4.423076923076923,
-      "grad_norm": 0.62890625,
-      "learning_rate": 0.0001777427517289766,
-      "loss": 0.9524,
+      "epoch": 8.846153846153847,
+      "grad_norm": 0.208984375,
+      "learning_rate": 8.002055634117578e-06,
+      "loss": 0.8469,
       "step": 920
     },
     {
-      "epoch": 4.447115384615385,
-      "grad_norm": 0.40625,
-      "learning_rate": 0.0001773896896789112,
-      "loss": 0.954,
+      "epoch": 8.89423076923077,
+      "grad_norm": 0.208984375,
+      "learning_rate": 7.357242332955916e-06,
+      "loss": 0.8509,
       "step": 925
     },
     {
-      "epoch": 4.471153846153846,
-      "grad_norm": 0.32421875,
-      "learning_rate": 0.00017703420588881946,
-      "loss": 0.9477,
+      "epoch": 8.942307692307692,
+      "grad_norm": 0.208984375,
+      "learning_rate": 6.738519948473976e-06,
+      "loss": 0.8545,
       "step": 930
     },
     {
-      "epoch": 4.4951923076923075,
-      "grad_norm": 0.365234375,
-      "learning_rate": 0.00017667631148278421,
-      "loss": 0.9556,
+      "epoch": 8.990384615384615,
+      "grad_norm": 0.205078125,
+      "learning_rate": 6.146062731035129e-06,
+      "loss": 0.845,
       "step": 935
     },
     {
-      "epoch": 4.519230769230769,
-      "grad_norm": 0.3515625,
-      "learning_rate": 0.00017631601766032336,
-      "loss": 0.9576,
+      "epoch": 9.0,
+      "eval_loss": 2.5566022396087646,
+      "eval_runtime": 0.4819,
+      "eval_samples_per_second": 20.752,
+      "eval_steps_per_second": 2.075,
+      "step": 936
+    },
+    {
+      "epoch": 9.038461538461538,
+      "grad_norm": 0.2138671875,
+      "learning_rate": 5.580037533961546e-06,
+      "loss": 0.8439,
       "step": 940
     },
     {
-      "epoch": 4.543269230769231,
-      "grad_norm": 0.353515625,
-      "learning_rate": 0.0001759533356960391,
-      "loss": 0.9548,
+      "epoch": 9.086538461538462,
+      "grad_norm": 0.2060546875,
+      "learning_rate": 5.040603766543594e-06,
+      "loss": 0.8496,
       "step": 945
     },
     {
-      "epoch": 4.5673076923076925,
-      "grad_norm": 0.494140625,
-      "learning_rate": 0.00017558827693926534,
-      "loss": 0.9495,
+      "epoch": 9.134615384615385,
+      "grad_norm": 0.208984375,
+      "learning_rate": 4.527913349145441e-06,
+      "loss": 0.8483,
       "step": 950
     },
     {
-      "epoch": 4.591346153846154,
-      "grad_norm": 0.3359375,
-      "learning_rate": 0.00017522085281371257,
-      "loss": 0.9556,
+      "epoch": 9.182692307692308,
+      "grad_norm": 0.2060546875,
+      "learning_rate": 4.042110670419763e-06,
+      "loss": 0.8508,
       "step": 955
     },
     {
-      "epoch": 4.615384615384615,
-      "grad_norm": 0.35546875,
-      "learning_rate": 0.00017485107481711012,
-      "loss": 0.9541,
+      "epoch": 9.23076923076923,
+      "grad_norm": 0.205078125,
+      "learning_rate": 3.5833325466437694e-06,
+      "loss": 0.8508,
       "step": 960
     },
     {
-      "epoch": 4.639423076923077,
-      "grad_norm": 0.376953125,
-      "learning_rate": 0.0001744789545208467,
-      "loss": 0.9444,
+      "epoch": 9.278846153846153,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 3.1517081831876737e-06,
+      "loss": 0.8437,
       "step": 965
     },
     {
-      "epoch": 4.663461538461538,
-      "grad_norm": 0.306640625,
-      "learning_rate": 0.00017410450356960795,
-      "loss": 0.9557,
+      "epoch": 9.326923076923077,
+      "grad_norm": 0.212890625,
+      "learning_rate": 2.7473591381266708e-06,
+      "loss": 0.845,
       "step": 970
     },
     {
-      "epoch": 4.6875,
-      "grad_norm": 0.443359375,
-      "learning_rate": 0.0001737277336810124,
-      "loss": 0.9543,
+      "epoch": 9.375,
+      "grad_norm": 0.2060546875,
+      "learning_rate": 2.3703992880066638e-06,
+      "loss": 0.8428,
       "step": 975
     },
     {
-      "epoch": 4.711538461538462,
-      "grad_norm": 0.44921875,
-      "learning_rate": 0.0001733486566452446,
-      "loss": 0.9512,
+      "epoch": 9.423076923076923,
+      "grad_norm": 0.20703125,
+      "learning_rate": 2.0209347957732328e-06,
+      "loss": 0.8495,
       "step": 980
     },
     {
-      "epoch": 4.735576923076923,
-      "grad_norm": 0.34765625,
-      "learning_rate": 0.00017296728432468603,
-      "loss": 0.9538,
+      "epoch": 9.471153846153847,
+      "grad_norm": 0.2138671875,
+      "learning_rate": 1.6990640808730696e-06,
+      "loss": 0.846,
       "step": 985
     },
     {
-      "epoch": 4.759615384615385,
-      "grad_norm": 0.3671875,
-      "learning_rate": 0.00017258362865354426,
-      "loss": 0.9517,
+      "epoch": 9.51923076923077,
+      "grad_norm": 0.2109375,
+      "learning_rate": 1.404877791536141e-06,
+      "loss": 0.8475,
       "step": 990
     },
     {
-      "epoch": 4.783653846153846,
-      "grad_norm": 0.376953125,
-      "learning_rate": 0.00017219770163747917,
-      "loss": 0.9446,
+      "epoch": 9.567307692307692,
+      "grad_norm": 0.203125,
+      "learning_rate": 1.1384587792465872e-06,
+      "loss": 0.8478,
       "step": 995
     },
     {
-      "epoch": 4.8076923076923075,
-      "grad_norm": 0.431640625,
-      "learning_rate": 0.0001718095153532274,
-      "loss": 0.9484,
+      "epoch": 9.615384615384615,
+      "grad_norm": 0.21875,
+      "learning_rate": 8.998820754091531e-07,
+      "loss": 0.844,
       "step": 1000
     },
     {
-      "epoch": 4.831730769230769,
-      "grad_norm": 0.353515625,
-      "learning_rate": 0.00017141908194822446,
-      "loss": 0.9498,
+      "epoch": 9.663461538461538,
+      "grad_norm": 0.2060546875,
+      "learning_rate": 6.892148702183133e-07,
+      "loss": 0.8478,
       "step": 1005
     },
     {
-      "epoch": 4.855769230769231,
-      "grad_norm": 0.53125,
-      "learning_rate": 0.00017102641364022457,
-      "loss": 0.9554,
+      "epoch": 9.711538461538462,
+      "grad_norm": 0.2060546875,
+      "learning_rate": 5.065164937354428e-07,
+      "loss": 0.8459,
       "step": 1010
     },
     {
-      "epoch": 4.8798076923076925,
-      "grad_norm": 0.337890625,
-      "learning_rate": 0.00017063152271691817,
-      "loss": 0.9543,
+      "epoch": 9.759615384615385,
+      "grad_norm": 0.2041015625,
+      "learning_rate": 3.5183839917972697e-07,
+      "loss": 0.8475,
       "step": 1015
     },
     {
-      "epoch": 4.903846153846154,
-      "grad_norm": 0.34375,
-      "learning_rate": 0.00017023442153554777,
-      "loss": 0.9542,
+      "epoch": 9.807692307692308,
+      "grad_norm": 0.2216796875,
+      "learning_rate": 2.2522414843748618e-07,
+      "loss": 0.8533,
       "step": 1020
     },
     {
-      "epoch": 4.927884615384615,
-      "grad_norm": 0.3828125,
-      "learning_rate": 0.00016983512252252085,
-      "loss": 0.9569,
+      "epoch": 9.85576923076923,
+      "grad_norm": 0.205078125,
+      "learning_rate": 1.2670939979384512e-07,
+      "loss": 0.8547,
       "step": 1025
     },
     {
-      "epoch": 4.951923076923077,
-      "grad_norm": 0.435546875,
-      "learning_rate": 0.00016943363817302135,
-      "loss": 0.9428,
+      "epoch": 9.903846153846153,
+      "grad_norm": 0.2080078125,
+      "learning_rate": 5.632189789027687e-08,
+      "loss": 0.8463,
       "step": 1030
     },
     {
-      "epoch": 4.975961538461538,
-      "grad_norm": 0.365234375,
-      "learning_rate": 0.00016902998105061844,
-      "loss": 0.9489,
+      "epoch": 9.951923076923077,
+      "grad_norm": 0.2060546875,
+      "learning_rate": 1.4081465910975588e-08,
+      "loss": 0.8418,
       "step": 1035
     },
     {
-      "epoch": 5.0,
-      "grad_norm": 0.52734375,
-      "learning_rate": 0.0001686241637868734,
-      "loss": 0.9507,
+      "epoch": 10.0,
+      "grad_norm": 0.2080078125,
+      "learning_rate": 0.0,
+      "loss": 0.853,
       "step": 1040
     },
     {
-      "epoch": 5.0,
-      "eval_loss": 2.457782745361328,
-      "eval_runtime": 0.5339,
-      "eval_samples_per_second": 18.731,
-      "eval_steps_per_second": 1.873,
+      "epoch": 10.0,
+      "eval_loss": 2.5572547912597656,
+      "eval_runtime": 0.4824,
+      "eval_samples_per_second": 20.728,
+      "eval_steps_per_second": 2.073,
       "step": 1040
     },
     {
-      "epoch": 5.024038461538462,
-      "grad_norm": 0.361328125,
-      "learning_rate": 0.00016821619908094436,
-      "loss": 0.9336,
-      "step": 1045
-    },
-    {
-      "epoch": 5.048076923076923,
-      "grad_norm": 0.41796875,
-      "learning_rate": 0.0001678060996991891,
-      "loss": 0.9358,
-      "step": 1050
-    },
-    {
-      "epoch": 5.072115384615385,
-      "grad_norm": 0.34765625,
-      "learning_rate": 0.0001673938784747652,
-      "loss": 0.9309,
-      "step": 1055
-    },
-    {
-      "epoch": 5.096153846153846,
-      "grad_norm": 0.404296875,
-      "learning_rate": 0.00016697954830722868,
-      "loss": 0.9316,
-      "step": 1060
-    },
-    {
-      "epoch": 5.1201923076923075,
-      "grad_norm": 0.5,
-      "learning_rate": 0.00016656312216213034,
-      "loss": 0.9284,
-      "step": 1065
-    },
-    {
-      "epoch": 5.144230769230769,
-      "grad_norm": 0.439453125,
-      "learning_rate": 0.00016614461307061,
-      "loss": 0.9415,
-      "step": 1070
-    },
-    {
-      "epoch": 5.168269230769231,
-      "grad_norm": 0.357421875,
-      "learning_rate": 0.00016572403412898855,
-      "loss": 0.9395,
-      "step": 1075
-    },
-    {
-      "epoch": 5.1923076923076925,
-      "grad_norm": 0.57421875,
-      "learning_rate": 0.0001653013984983585,
-      "loss": 0.933,
-      "step": 1080
-    },
-    {
-      "epoch": 5.216346153846154,
-      "grad_norm": 0.314453125,
-      "learning_rate": 0.0001648767194041718,
-      "loss": 0.9352,
-      "step": 1085
-    },
-    {
-      "epoch": 5.240384615384615,
-      "grad_norm": 0.369140625,
-      "learning_rate": 0.00016445001013582608,
-      "loss": 0.9298,
-      "step": 1090
-    },
-    {
-      "epoch": 5.264423076923077,
-      "grad_norm": 0.3359375,
-      "learning_rate": 0.00016402128404624882,
-      "loss": 0.9302,
-      "step": 1095
-    },
-    {
-      "epoch": 5.288461538461538,
-      "grad_norm": 0.34765625,
-      "learning_rate": 0.0001635905545514795,
-      "loss": 0.9386,
-      "step": 1100
-    },
-    {
-      "epoch": 5.3125,
-      "grad_norm": 0.439453125,
-      "learning_rate": 0.00016315783513024977,
-      "loss": 0.944,
-      "step": 1105
-    },
-    {
-      "epoch": 5.336538461538462,
-      "grad_norm": 0.408203125,
-      "learning_rate": 0.00016272313932356162,
-      "loss": 0.9366,
-      "step": 1110
-    },
-    {
-      "epoch": 5.360576923076923,
-      "grad_norm": 0.333984375,
-      "learning_rate": 0.00016228648073426366,
-      "loss": 0.9417,
-      "step": 1115
-    },
-    {
-      "epoch": 5.384615384615385,
-      "grad_norm": 0.5859375,
-      "learning_rate": 0.0001618478730266255,
-      "loss": 0.9355,
-      "step": 1120
-    },
-    {
-      "epoch": 5.408653846153846,
-      "grad_norm": 0.33984375,
-      "learning_rate": 0.0001614073299259101,
-      "loss": 0.9407,
-      "step": 1125
-    },
-    {
-      "epoch": 5.4326923076923075,
-      "grad_norm": 0.34375,
-      "learning_rate": 0.00016096486521794434,
-      "loss": 0.9268,
-      "step": 1130
-    },
-    {
-      "epoch": 5.456730769230769,
-      "grad_norm": 0.451171875,
-      "learning_rate": 0.0001605204927486874,
-      "loss": 0.9381,
-      "step": 1135
-    },
-    {
-      "epoch": 5.480769230769231,
-      "grad_norm": 0.412109375,
-      "learning_rate": 0.0001600742264237979,
-      "loss": 0.9338,
-      "step": 1140
-    },
-    {
-      "epoch": 5.5048076923076925,
-      "grad_norm": 0.4375,
-      "learning_rate": 0.0001596260802081983,
-      "loss": 0.9306,
-      "step": 1145
-    },
-    {
-      "epoch": 5.528846153846154,
-      "grad_norm": 0.35546875,
-      "learning_rate": 0.0001591760681256382,
-      "loss": 0.9347,
-      "step": 1150
-    },
-    {
-      "epoch": 5.552884615384615,
-      "grad_norm": 0.44140625,
-      "learning_rate": 0.0001587242042582554,
-      "loss": 0.9364,
-      "step": 1155
-    },
-    {
-      "epoch": 5.576923076923077,
-      "grad_norm": 0.400390625,
-      "learning_rate": 0.00015827050274613513,
-      "loss": 0.935,
-      "step": 1160
-    },
-    {
-      "epoch": 5.600961538461538,
-      "grad_norm": 0.458984375,
-      "learning_rate": 0.00015781497778686785,
-      "loss": 0.9417,
-      "step": 1165
-    },
-    {
-      "epoch": 5.625,
-      "grad_norm": 0.55078125,
-      "learning_rate": 0.0001573576436351046,
-      "loss": 0.9384,
-      "step": 1170
-    },
-    {
-      "epoch": 5.649038461538462,
-      "grad_norm": 0.408203125,
-      "learning_rate": 0.00015689851460211126,
-      "loss": 0.9324,
-      "step": 1175
-    },
-    {
-      "epoch": 5.673076923076923,
-      "grad_norm": 0.337890625,
-      "learning_rate": 0.0001564376050553205,
-      "loss": 0.9502,
-      "step": 1180
-    },
-    {
-      "epoch": 5.697115384615385,
-      "grad_norm": 0.54296875,
-      "learning_rate": 0.00015597492941788222,
-      "loss": 0.9392,
-      "step": 1185
-    },
-    {
-      "epoch": 5.721153846153846,
-      "grad_norm": 0.443359375,
-      "learning_rate": 0.0001555105021682123,
-      "loss": 0.937,
-      "step": 1190
-    },
-    {
-      "epoch": 5.7451923076923075,
-      "grad_norm": 0.4609375,
-      "learning_rate": 0.00015504433783953945,
-      "loss": 0.9291,
-      "step": 1195
-    },
-    {
-      "epoch": 5.769230769230769,
-      "grad_norm": 0.42578125,
-      "learning_rate": 0.00015457645101945046,
-      "loss": 0.9435,
-      "step": 1200
-    },
-    {
-      "epoch": 5.793269230769231,
-      "grad_norm": 0.462890625,
-      "learning_rate": 0.00015410685634943366,
-      "loss": 0.9378,
-      "step": 1205
-    },
-    {
-      "epoch": 5.8173076923076925,
-      "grad_norm": 0.486328125,
-      "learning_rate": 0.00015363556852442085,
-      "loss": 0.9344,
-      "step": 1210
-    },
-    {
-      "epoch": 5.841346153846154,
-      "grad_norm": 0.4375,
-      "learning_rate": 0.00015316260229232727,
-      "loss": 0.9357,
-      "step": 1215
-    },
-    {
-      "epoch": 5.865384615384615,
-      "grad_norm": 0.34375,
-      "learning_rate": 0.00015268797245359035,
-      "loss": 0.9321,
-      "step": 1220
-    },
-    {
-      "epoch": 5.889423076923077,
-      "grad_norm": 0.5390625,
-      "learning_rate": 0.00015221169386070636,
-      "loss": 0.9354,
-      "step": 1225
-    },
-    {
-      "epoch": 5.913461538461538,
-      "grad_norm": 0.439453125,
-      "learning_rate": 0.00015173378141776568,
-      "loss": 0.9362,
-      "step": 1230
-    },
-    {
-      "epoch": 5.9375,
-      "grad_norm": 0.55859375,
-      "learning_rate": 0.00015125425007998653,
-      "loss": 0.939,
-      "step": 1235
-    },
-    {
-      "epoch": 5.961538461538462,
-      "grad_norm": 0.50390625,
-      "learning_rate": 0.0001507731148532468,
-      "loss": 0.9405,
-      "step": 1240
-    },
-    {
-      "epoch": 5.985576923076923,
-      "grad_norm": 0.33203125,
-      "learning_rate": 0.00015029039079361448,
-      "loss": 0.9348,
-      "step": 1245
-    },
-    {
-      "epoch": 6.0,
-      "eval_loss": 2.4684882164001465,
-      "eval_runtime": 0.536,
-      "eval_samples_per_second": 18.656,
-      "eval_steps_per_second": 1.866,
-      "step": 1248
-    },
-    {
-      "epoch": 6.009615384615385,
-      "grad_norm": 0.376953125,
-      "learning_rate": 0.00014980609300687683,
-      "loss": 0.9265,
-      "step": 1250
-    },
-    {
-      "epoch": 6.033653846153846,
-      "grad_norm": 0.41015625,
-      "learning_rate": 0.0001493202366480672,
-      "loss": 0.9186,
-      "step": 1255
-    },
-    {
-      "epoch": 6.0576923076923075,
-      "grad_norm": 0.326171875,
-      "learning_rate": 0.00014883283692099112,
-      "loss": 0.9204,
-      "step": 1260
-    },
-    {
-      "epoch": 6.081730769230769,
-      "grad_norm": 0.5078125,
-      "learning_rate": 0.00014834390907775044,
-      "loss": 0.9106,
-      "step": 1265
-    },
-    {
-      "epoch": 6.105769230769231,
-      "grad_norm": 0.453125,
-      "learning_rate": 0.000147853468418266,
-      "loss": 0.9207,
-      "step": 1270
-    },
-    {
-      "epoch": 6.1298076923076925,
-      "grad_norm": 0.5,
-      "learning_rate": 0.00014736153028979893,
-      "loss": 0.9117,
-      "step": 1275
-    },
-    {
-      "epoch": 6.153846153846154,
-      "grad_norm": 0.392578125,
-      "learning_rate": 0.00014686811008647038,
-      "loss": 0.9282,
-      "step": 1280
-    },
-    {
-      "epoch": 6.177884615384615,
-      "grad_norm": 0.421875,
-      "learning_rate": 0.0001463732232487797,
-      "loss": 0.9232,
-      "step": 1285
-    },
-    {
-      "epoch": 6.201923076923077,
-      "grad_norm": 0.515625,
-      "learning_rate": 0.00014587688526312143,
-      "loss": 0.919,
-      "step": 1290
-    },
-    {
-      "epoch": 6.225961538461538,
-      "grad_norm": 0.357421875,
-      "learning_rate": 0.00014537911166130044,
-      "loss": 0.9202,
-      "step": 1295
-    },
-    {
-      "epoch": 6.25,
-      "grad_norm": 0.365234375,
-      "learning_rate": 0.00014487991802004623,
-      "loss": 0.9282,
-      "step": 1300
-    },
-    {
-      "epoch": 6.274038461538462,
-      "grad_norm": 0.345703125,
-      "learning_rate": 0.00014437931996052518,
-      "loss": 0.9208,
-      "step": 1305
-    },
-    {
-      "epoch": 6.298076923076923,
-      "grad_norm": 0.36328125,
-      "learning_rate": 0.00014387733314785193,
-      "loss": 0.9317,
-      "step": 1310
-    },
-    {
-      "epoch": 6.322115384615385,
-      "grad_norm": 0.498046875,
-      "learning_rate": 0.00014337397329059902,
-      "loss": 0.9224,
-      "step": 1315
-    },
-    {
-      "epoch": 6.346153846153846,
-      "grad_norm": 0.46875,
-      "learning_rate": 0.00014286925614030542,
-      "loss": 0.9182,
-      "step": 1320
-    },
-    {
-      "epoch": 6.3701923076923075,
-      "grad_norm": 0.48046875,
-      "learning_rate": 0.00014236319749098367,
-      "loss": 0.9213,
-      "step": 1325
-    },
-    {
-      "epoch": 6.394230769230769,
-      "grad_norm": 0.349609375,
-      "learning_rate": 0.00014185581317862546,
-      "loss": 0.9285,
-      "step": 1330
-    },
-    {
-      "epoch": 6.418269230769231,
-      "grad_norm": 0.419921875,
-      "learning_rate": 0.00014134711908070631,
-      "loss": 0.9149,
-      "step": 1335
-    },
-    {
-      "epoch": 6.4423076923076925,
-      "grad_norm": 0.458984375,
-      "learning_rate": 0.00014083713111568842,
-      "loss": 0.9262,
-      "step": 1340
-    },
-    {
-      "epoch": 6.466346153846154,
-      "grad_norm": 0.3515625,
-      "learning_rate": 0.00014032586524252293,
-      "loss": 0.9226,
-      "step": 1345
-    },
-    {
-      "epoch": 6.490384615384615,
-      "grad_norm": 0.365234375,
-      "learning_rate": 0.0001398133374601501,
-      "loss": 0.9301,
-      "step": 1350
-    },
-    {
-      "epoch": 6.514423076923077,
-      "grad_norm": 0.330078125,
-      "learning_rate": 0.00013929956380699904,
-      "loss": 0.917,
-      "step": 1355
-    },
-    {
-      "epoch": 6.538461538461538,
-      "grad_norm": 0.328125,
-      "learning_rate": 0.0001387845603604855,
-      "loss": 0.9263,
-      "step": 1360
-    },
-    {
-      "epoch": 6.5625,
-      "grad_norm": 0.51953125,
-      "learning_rate": 0.000138268343236509,
-      "loss": 0.9333,
-      "step": 1365
-    },
-    {
-      "epoch": 6.586538461538462,
-      "grad_norm": 0.40234375,
-      "learning_rate": 0.00013775092858894837,
-      "loss": 0.9286,
-      "step": 1370
-    },
-    {
-      "epoch": 6.610576923076923,
-      "grad_norm": 0.333984375,
-      "learning_rate": 0.0001372323326091563,
-      "loss": 0.9285,
-      "step": 1375
-    },
-    {
-      "epoch": 6.634615384615385,
-      "grad_norm": 0.40625,
-      "learning_rate": 0.00013671257152545277,
-      "loss": 0.9234,
-      "step": 1380
-    },
-    {
-      "epoch": 6.658653846153846,
-      "grad_norm": 0.369140625,
-      "learning_rate": 0.000136191661602617,
-      "loss": 0.9334,
-      "step": 1385
-    },
-    {
-      "epoch": 6.6826923076923075,
-      "grad_norm": 0.4375,
-      "learning_rate": 0.00013566961914137867,
-      "loss": 0.9172,
-      "step": 1390
-    },
-    {
-      "epoch": 6.706730769230769,
-      "grad_norm": 0.34375,
-      "learning_rate": 0.00013514646047790775,
-      "loss": 0.9258,
-      "step": 1395
-    },
-    {
-      "epoch": 6.730769230769231,
-      "grad_norm": 0.396484375,
-      "learning_rate": 0.00013462220198330328,
-      "loss": 0.9203,
-      "step": 1400
-    },
-    {
-      "epoch": 6.7548076923076925,
-      "grad_norm": 0.322265625,
-      "learning_rate": 0.0001340968600630811,
-      "loss": 0.9202,
-      "step": 1405
-    },
-    {
-      "epoch": 6.778846153846154,
-      "grad_norm": 0.34765625,
-      "learning_rate": 0.0001335704511566605,
-      "loss": 0.9247,
-      "step": 1410
-    },
-    {
-      "epoch": 6.802884615384615,
-      "grad_norm": 0.380859375,
-      "learning_rate": 0.0001330429917368497,
-      "loss": 0.9181,
-      "step": 1415
-    },
-    {
-      "epoch": 6.826923076923077,
-      "grad_norm": 0.349609375,
-      "learning_rate": 0.0001325144983093305,
-      "loss": 0.9177,
-      "step": 1420
-    },
-    {
-      "epoch": 6.850961538461538,
-      "grad_norm": 0.484375,
-      "learning_rate": 0.00013198498741214166,
-      "loss": 0.9294,
-      "step": 1425
-    },
-    {
-      "epoch": 6.875,
-      "grad_norm": 0.380859375,
-      "learning_rate": 0.00013145447561516138,
-      "loss": 0.9198,
-      "step": 1430
-    },
-    {
-      "epoch": 6.899038461538462,
-      "grad_norm": 0.3359375,
-      "learning_rate": 0.00013092297951958884,
-      "loss": 0.9282,
-      "step": 1435
-    },
-    {
-      "epoch": 6.923076923076923,
-      "grad_norm": 0.337890625,
-      "learning_rate": 0.0001303905157574247,
-      "loss": 0.9139,
-      "step": 1440
-    },
-    {
-      "epoch": 6.947115384615385,
-      "grad_norm": 0.349609375,
-      "learning_rate": 0.00012985710099095062,
-      "loss": 0.917,
-      "step": 1445
-    },
-    {
-      "epoch": 6.971153846153846,
-      "grad_norm": 0.3515625,
-      "learning_rate": 0.00012932275191220776,
-      "loss": 0.9184,
-      "step": 1450
-    },
-    {
-      "epoch": 6.9951923076923075,
-      "grad_norm": 0.36328125,
-      "learning_rate": 0.00012878748524247462,
-      "loss": 0.9236,
-      "step": 1455
-    },
-    {
-      "epoch": 7.0,
-      "eval_loss": 2.4788994789123535,
-      "eval_runtime": 0.5329,
-      "eval_samples_per_second": 18.765,
-      "eval_steps_per_second": 1.876,
-      "step": 1456
-    },
-    {
-      "epoch": 7.019230769230769,
-      "grad_norm": 0.361328125,
-      "learning_rate": 0.0001282513177317437,
-      "loss": 0.9072,
-      "step": 1460
-    },
-    {
-      "epoch": 7.043269230769231,
-      "grad_norm": 0.421875,
-      "learning_rate": 0.00012771426615819725,
-      "loss": 0.9064,
-      "step": 1465
-    },
-    {
-      "epoch": 7.0673076923076925,
-      "grad_norm": 0.4609375,
-      "learning_rate": 0.00012717634732768243,
-      "loss": 0.9101,
-      "step": 1470
-    },
-    {
-      "epoch": 7.091346153846154,
-      "grad_norm": 0.4765625,
-      "learning_rate": 0.00012663757807318521,
-      "loss": 0.9122,
-      "step": 1475
-    },
-    {
-      "epoch": 7.115384615384615,
-      "grad_norm": 0.322265625,
-      "learning_rate": 0.00012609797525430373,
-      "loss": 0.9166,
-      "step": 1480
-    },
-    {
-      "epoch": 7.139423076923077,
-      "grad_norm": 0.39453125,
-      "learning_rate": 0.0001255575557567207,
-      "loss": 0.9082,
-      "step": 1485
-    },
-    {
-      "epoch": 7.163461538461538,
-      "grad_norm": 0.439453125,
-      "learning_rate": 0.00012501633649167495,
-      "loss": 0.9077,
-      "step": 1490
-    },
-    {
-      "epoch": 7.1875,
-      "grad_norm": 0.361328125,
-      "learning_rate": 0.0001244743343954324,
-      "loss": 0.9136,
-      "step": 1495
-    },
-    {
-      "epoch": 7.211538461538462,
-      "grad_norm": 0.40234375,
-      "learning_rate": 0.0001239315664287558,
-      "loss": 0.915,
-      "step": 1500
-    },
-    {
-      "epoch": 7.235576923076923,
-      "grad_norm": 0.388671875,
-      "learning_rate": 0.0001233880495763742,
-      "loss": 0.9126,
-      "step": 1505
-    },
-    {
-      "epoch": 7.259615384615385,
-      "grad_norm": 0.4375,
-      "learning_rate": 0.00012284380084645139,
-      "loss": 0.9109,
-      "step": 1510
-    },
-    {
-      "epoch": 7.283653846153846,
-      "grad_norm": 0.4296875,
-      "learning_rate": 0.00012229883727005365,
-      "loss": 0.9113,
-      "step": 1515
-    },
-    {
-      "epoch": 7.3076923076923075,
-      "grad_norm": 0.416015625,
-      "learning_rate": 0.00012175317590061674,
-      "loss": 0.912,
-      "step": 1520
-    },
-    {
-      "epoch": 7.331730769230769,
-      "grad_norm": 0.390625,
-      "learning_rate": 0.00012120683381341247,
-      "loss": 0.9107,
-      "step": 1525
-    },
-    {
-      "epoch": 7.355769230769231,
-      "grad_norm": 0.333984375,
-      "learning_rate": 0.00012065982810501404,
-      "loss": 0.9192,
-      "step": 1530
-    },
-    {
-      "epoch": 7.3798076923076925,
-      "grad_norm": 0.337890625,
-      "learning_rate": 0.00012011217589276131,
-      "loss": 0.9079,
-      "step": 1535
-    },
-    {
-      "epoch": 7.403846153846154,
-      "grad_norm": 0.337890625,
-      "learning_rate": 0.00011956389431422507,
-      "loss": 0.9035,
-      "step": 1540
-    },
-    {
-      "epoch": 7.427884615384615,
-      "grad_norm": 0.357421875,
-      "learning_rate": 0.00011901500052667068,
-      "loss": 0.9129,
-      "step": 1545
-    },
-    {
-      "epoch": 7.451923076923077,
-      "grad_norm": 0.458984375,
-      "learning_rate": 0.00011846551170652127,
-      "loss": 0.909,
-      "step": 1550
-    },
-    {
-      "epoch": 7.475961538461538,
-      "grad_norm": 0.48828125,
-      "learning_rate": 0.00011791544504882021,
-      "loss": 0.9175,
-      "step": 1555
-    },
-    {
-      "epoch": 7.5,
-      "grad_norm": 0.396484375,
-      "learning_rate": 0.00011736481776669306,
-      "loss": 0.9111,
-      "step": 1560
-    },
-    {
-      "epoch": 7.524038461538462,
-      "grad_norm": 0.419921875,
-      "learning_rate": 0.00011681364709080875,
-      "loss": 0.9123,
-      "step": 1565
-    },
-    {
-      "epoch": 7.548076923076923,
-      "grad_norm": 0.390625,
-      "learning_rate": 0.0001162619502688407,
-      "loss": 0.9156,
-      "step": 1570
-    },
-    {
-      "epoch": 7.572115384615385,
-      "grad_norm": 0.36328125,
-      "learning_rate": 0.00011570974456492678,
-      "loss": 0.9074,
-      "step": 1575
-    },
-    {
-      "epoch": 7.596153846153846,
-      "grad_norm": 0.34765625,
-      "learning_rate": 0.00011515704725912926,
-      "loss": 0.9095,
-      "step": 1580
-    },
-    {
-      "epoch": 7.6201923076923075,
-      "grad_norm": 0.494140625,
-      "learning_rate": 0.00011460387564689404,
-      "loss": 0.9114,
-      "step": 1585
-    },
-    {
-      "epoch": 7.644230769230769,
-      "grad_norm": 0.45703125,
-      "learning_rate": 0.00011405024703850929,
-      "loss": 0.9068,
-      "step": 1590
-    },
-    {
-      "epoch": 7.668269230769231,
-      "grad_norm": 0.455078125,
-      "learning_rate": 0.00011349617875856395,
-      "loss": 0.9257,
-      "step": 1595
-    },
-    {
-      "epoch": 7.6923076923076925,
-      "grad_norm": 0.416015625,
-      "learning_rate": 0.00011294168814540553,
-      "loss": 0.9116,
-      "step": 1600
-    },
-    {
-      "epoch": 7.716346153846154,
-      "grad_norm": 0.486328125,
-      "learning_rate": 0.00011238679255059752,
-      "loss": 0.9114,
-      "step": 1605
-    },
-    {
-      "epoch": 7.740384615384615,
-      "grad_norm": 0.5,
-      "learning_rate": 0.00011183150933837632,
-      "loss": 0.9133,
-      "step": 1610
-    },
-    {
-      "epoch": 7.764423076923077,
-      "grad_norm": 0.39453125,
-      "learning_rate": 0.00011127585588510809,
-      "loss": 0.9108,
-      "step": 1615
-    },
-    {
-      "epoch": 7.788461538461538,
-      "grad_norm": 0.341796875,
-      "learning_rate": 0.00011071984957874479,
-      "loss": 0.9093,
-      "step": 1620
-    },
-    {
-      "epoch": 7.8125,
-      "grad_norm": 0.390625,
-      "learning_rate": 0.00011016350781828019,
-      "loss": 0.9083,
-      "step": 1625
-    },
-    {
-      "epoch": 7.836538461538462,
-      "grad_norm": 0.359375,
-      "learning_rate": 0.00010960684801320536,
-      "loss": 0.9058,
-      "step": 1630
-    },
-    {
-      "epoch": 7.860576923076923,
-      "grad_norm": 0.361328125,
-      "learning_rate": 0.0001090498875829638,
-      "loss": 0.9101,
-      "step": 1635
-    },
-    {
-      "epoch": 7.884615384615385,
-      "grad_norm": 0.357421875,
-      "learning_rate": 0.00010849264395640649,
-      "loss": 0.9151,
-      "step": 1640
-    },
-    {
-      "epoch": 7.908653846153846,
-      "grad_norm": 0.361328125,
-      "learning_rate": 0.0001079351345712464,
-      "loss": 0.9154,
-      "step": 1645
-    },
-    {
-      "epoch": 7.9326923076923075,
-      "grad_norm": 0.3203125,
-      "learning_rate": 0.00010737737687351284,
-      "loss": 0.9197,
-      "step": 1650
-    },
-    {
-      "epoch": 7.956730769230769,
-      "grad_norm": 0.412109375,
-      "learning_rate": 0.00010681938831700546,
-      "loss": 0.9058,
-      "step": 1655
-    },
-    {
-      "epoch": 7.980769230769231,
-      "grad_norm": 0.4296875,
-      "learning_rate": 0.0001062611863627482,
-      "loss": 0.9156,
-      "step": 1660
-    },
-    {
-      "epoch": 8.0,
-      "eval_loss": 2.483135938644409,
-      "eval_runtime": 0.5406,
-      "eval_samples_per_second": 18.499,
-      "eval_steps_per_second": 1.85,
-      "step": 1664
-    },
-    {
-      "epoch": 8.004807692307692,
-      "grad_norm": 0.408203125,
-      "learning_rate": 0.00010570278847844275,
-      "loss": 0.9095,
-      "step": 1665
-    },
-    {
-      "epoch": 8.028846153846153,
-      "grad_norm": 0.35546875,
-      "learning_rate": 0.00010514421213792205,
-      "loss": 0.8983,
-      "step": 1670
-    },
-    {
-      "epoch": 8.052884615384615,
-      "grad_norm": 0.4296875,
-      "learning_rate": 0.00010458547482060341,
-      "loss": 0.9019,
-      "step": 1675
-    },
-    {
-      "epoch": 8.076923076923077,
-      "grad_norm": 0.51953125,
-      "learning_rate": 0.00010402659401094152,
-      "loss": 0.8957,
-      "step": 1680
-    },
-    {
-      "epoch": 8.100961538461538,
-      "grad_norm": 0.32421875,
-      "learning_rate": 0.00010346758719788143,
-      "loss": 0.8918,
-      "step": 1685
-    },
-    {
-      "epoch": 8.125,
-      "grad_norm": 0.365234375,
-      "learning_rate": 0.00010290847187431113,
-      "loss": 0.8979,
-      "step": 1690
-    },
-    {
-      "epoch": 8.149038461538462,
-      "grad_norm": 0.3671875,
-      "learning_rate": 0.00010234926553651422,
-      "loss": 0.9099,
-      "step": 1695
-    },
-    {
-      "epoch": 8.173076923076923,
-      "grad_norm": 0.3125,
-      "learning_rate": 0.00010178998568362243,
-      "loss": 0.9042,
-      "step": 1700
-    },
-    {
-      "epoch": 8.197115384615385,
-      "grad_norm": 0.322265625,
-      "learning_rate": 0.00010123064981706789,
-      "loss": 0.9094,
-      "step": 1705
-    },
-    {
-      "epoch": 8.221153846153847,
-      "grad_norm": 0.376953125,
-      "learning_rate": 0.00010067127544003563,
-      "loss": 0.9003,
-      "step": 1710
-    },
-    {
-      "epoch": 8.245192307692308,
-      "grad_norm": 0.34375,
-      "learning_rate": 0.00010011188005691572,
-      "loss": 0.902,
-      "step": 1715
-    },
-    {
-      "epoch": 8.26923076923077,
-      "grad_norm": 0.388671875,
-      "learning_rate": 9.955248117275566e-05,
-      "loss": 0.8979,
-      "step": 1720
-    },
-    {
-      "epoch": 8.29326923076923,
-      "grad_norm": 0.357421875,
-      "learning_rate": 9.899309629271246e-05,
-      "loss": 0.9012,
-      "step": 1725
-    },
-    {
-      "epoch": 8.317307692307692,
-      "grad_norm": 0.390625,
-      "learning_rate": 9.843374292150488e-05,
-      "loss": 0.8998,
-      "step": 1730
-    },
-    {
-      "epoch": 8.341346153846153,
-      "grad_norm": 0.38671875,
-      "learning_rate": 9.787443856286566e-05,
-      "loss": 0.9049,
-      "step": 1735
-    },
-    {
-      "epoch": 8.365384615384615,
-      "grad_norm": 0.328125,
-      "learning_rate": 9.73152007189939e-05,
-      "loss": 0.9092,
-      "step": 1740
-    },
-    {
-      "epoch": 8.389423076923077,
-      "grad_norm": 0.400390625,
-      "learning_rate": 9.675604689000718e-05,
-      "loss": 0.9001,
-      "step": 1745
-    },
-    {
-      "epoch": 8.413461538461538,
-      "grad_norm": 0.435546875,
-      "learning_rate": 9.619699457339405e-05,
-      "loss": 0.9068,
-      "step": 1750
-    },
-    {
-      "epoch": 8.4375,
-      "grad_norm": 0.369140625,
-      "learning_rate": 9.563806126346642e-05,
-      "loss": 0.8946,
-      "step": 1755
-    },
-    {
-      "epoch": 8.461538461538462,
-      "grad_norm": 0.443359375,
-      "learning_rate": 9.507926445081219e-05,
-      "loss": 0.9051,
-      "step": 1760
-    },
-    {
-      "epoch": 8.485576923076923,
-      "grad_norm": 0.33203125,
-      "learning_rate": 9.45206216217479e-05,
-      "loss": 0.901,
-      "step": 1765
-    },
-    {
-      "epoch": 8.509615384615385,
-      "grad_norm": 0.3359375,
-      "learning_rate": 9.396215025777139e-05,
-      "loss": 0.9021,
-      "step": 1770
-    },
-    {
-      "epoch": 8.533653846153847,
-      "grad_norm": 0.38671875,
-      "learning_rate": 9.340386783501507e-05,
-      "loss": 0.9089,
-      "step": 1775
-    },
-    {
-      "epoch": 8.557692307692308,
-      "grad_norm": 0.3515625,
-      "learning_rate": 9.284579182369867e-05,
-      "loss": 0.9018,
-      "step": 1780
-    },
-    {
-      "epoch": 8.58173076923077,
-      "grad_norm": 0.375,
-      "learning_rate": 9.22879396875828e-05,
-      "loss": 0.9037,
-      "step": 1785
-    },
-    {
-      "epoch": 8.60576923076923,
-      "grad_norm": 0.37109375,
-      "learning_rate": 9.173032888342244e-05,
-      "loss": 0.9022,
-      "step": 1790
-    },
-    {
-      "epoch": 8.629807692307692,
-      "grad_norm": 0.400390625,
-      "learning_rate": 9.117297686042053e-05,
-      "loss": 0.8989,
-      "step": 1795
-    },
-    {
-      "epoch": 8.653846153846153,
-      "grad_norm": 0.345703125,
-      "learning_rate": 9.061590105968208e-05,
-      "loss": 0.8929,
-      "step": 1800
-    },
-    {
-      "epoch": 8.677884615384615,
-      "grad_norm": 0.376953125,
-      "learning_rate": 9.005911891366831e-05,
-      "loss": 0.9076,
-      "step": 1805
-    },
-    {
-      "epoch": 8.701923076923077,
-      "grad_norm": 0.4140625,
-      "learning_rate": 8.950264784565112e-05,
-      "loss": 0.9146,
-      "step": 1810
-    },
-    {
-      "epoch": 8.725961538461538,
-      "grad_norm": 0.4921875,
-      "learning_rate": 8.894650526916803e-05,
-      "loss": 0.9038,
-      "step": 1815
-    },
-    {
-      "epoch": 8.75,
-      "grad_norm": 0.396484375,
-      "learning_rate": 8.839070858747697e-05,
-      "loss": 0.9074,
-      "step": 1820
-    },
-    {
-      "epoch": 8.774038461538462,
-      "grad_norm": 0.5703125,
-      "learning_rate": 8.783527519301204e-05,
-      "loss": 0.9063,
-      "step": 1825
-    },
-    {
-      "epoch": 8.798076923076923,
-      "grad_norm": 0.462890625,
-      "learning_rate": 8.728022246683894e-05,
-      "loss": 0.9026,
-      "step": 1830
-    },
-    {
-      "epoch": 8.822115384615385,
-      "grad_norm": 0.46875,
-      "learning_rate": 8.672556777811121e-05,
-      "loss": 0.901,
-      "step": 1835
-    },
-    {
-      "epoch": 8.846153846153847,
-      "grad_norm": 0.337890625,
-      "learning_rate": 8.617132848352671e-05,
-      "loss": 0.9013,
-      "step": 1840
-    },
-    {
-      "epoch": 8.870192307692308,
-      "grad_norm": 0.353515625,
-      "learning_rate": 8.561752192678443e-05,
-      "loss": 0.907,
-      "step": 1845
-    },
-    {
-      "epoch": 8.89423076923077,
-      "grad_norm": 0.341796875,
-      "learning_rate": 8.506416543804182e-05,
-      "loss": 0.9035,
-      "step": 1850
-    },
-    {
-      "epoch": 8.91826923076923,
-      "grad_norm": 0.359375,
-      "learning_rate": 8.451127633337242e-05,
-      "loss": 0.9075,
-      "step": 1855
-    },
-    {
-      "epoch": 8.942307692307692,
-      "grad_norm": 0.365234375,
-      "learning_rate": 8.395887191422397e-05,
-      "loss": 0.9115,
-      "step": 1860
-    },
-    {
-      "epoch": 8.966346153846153,
-      "grad_norm": 0.376953125,
-      "learning_rate": 8.340696946687716e-05,
-      "loss": 0.901,
-      "step": 1865
-    },
-    {
-      "epoch": 8.990384615384615,
-      "grad_norm": 0.357421875,
-      "learning_rate": 8.285558626190447e-05,
-      "loss": 0.8987,
-      "step": 1870
-    },
-    {
-      "epoch": 9.0,
-      "eval_loss": 2.4963114261627197,
-      "eval_runtime": 0.5347,
-      "eval_samples_per_second": 18.702,
-      "eval_steps_per_second": 1.87,
-      "step": 1872
-    },
-    {
-      "epoch": 9.014423076923077,
-      "grad_norm": 0.3125,
-      "learning_rate": 8.23047395536298e-05,
-      "loss": 0.8889,
-      "step": 1875
-    },
-    {
-      "epoch": 9.038461538461538,
-      "grad_norm": 0.33203125,
-      "learning_rate": 8.175444657958876e-05,
-      "loss": 0.8942,
-      "step": 1880
-    },
-    {
-      "epoch": 9.0625,
-      "grad_norm": 0.380859375,
-      "learning_rate": 8.120472455998882e-05,
-      "loss": 0.8981,
-      "step": 1885
-    },
-    {
-      "epoch": 9.086538461538462,
-      "grad_norm": 0.37890625,
-      "learning_rate": 8.065559069717088e-05,
-      "loss": 0.8926,
-      "step": 1890
-    },
-    {
-      "epoch": 9.110576923076923,
-      "grad_norm": 0.34765625,
-      "learning_rate": 8.010706217507069e-05,
-      "loss": 0.8869,
-      "step": 1895
-    },
-    {
-      "epoch": 9.134615384615385,
-      "grad_norm": 0.3828125,
-      "learning_rate": 7.955915615868111e-05,
-      "loss": 0.901,
-      "step": 1900
-    },
-    {
-      "epoch": 9.158653846153847,
-      "grad_norm": 0.44921875,
-      "learning_rate": 7.901188979351526e-05,
-      "loss": 0.8941,
-      "step": 1905
-    },
-    {
-      "epoch": 9.182692307692308,
-      "grad_norm": 0.39453125,
-      "learning_rate": 7.846528020506957e-05,
-      "loss": 0.9016,
-      "step": 1910
-    },
-    {
-      "epoch": 9.20673076923077,
-      "grad_norm": 0.33203125,
-      "learning_rate": 7.791934449828828e-05,
-      "loss": 0.8986,
-      "step": 1915
-    },
-    {
-      "epoch": 9.23076923076923,
-      "grad_norm": 0.4140625,
-      "learning_rate": 7.73740997570278e-05,
-      "loss": 0.8971,
-      "step": 1920
-    },
-    {
-      "epoch": 9.254807692307692,
-      "grad_norm": 0.34765625,
-      "learning_rate": 7.682956304352243e-05,
-      "loss": 0.8915,
-      "step": 1925
-    },
-    {
-      "epoch": 9.278846153846153,
-      "grad_norm": 0.494140625,
-      "learning_rate": 7.628575139785024e-05,
-      "loss": 0.8903,
-      "step": 1930
-    },
-    {
-      "epoch": 9.302884615384615,
-      "grad_norm": 0.34765625,
-      "learning_rate": 7.574268183739989e-05,
-      "loss": 0.8958,
-      "step": 1935
-    },
-    {
-      "epoch": 9.326923076923077,
-      "grad_norm": 0.43359375,
-      "learning_rate": 7.520037135633816e-05,
-      "loss": 0.8892,
-      "step": 1940
-    },
-    {
-      "epoch": 9.350961538461538,
-      "grad_norm": 0.359375,
-      "learning_rate": 7.46588369250781e-05,
-      "loss": 0.8915,
-      "step": 1945
-    },
-    {
-      "epoch": 9.375,
-      "grad_norm": 0.337890625,
-      "learning_rate": 7.411809548974792e-05,
-      "loss": 0.8903,
-      "step": 1950
-    },
-    {
-      "epoch": 9.399038461538462,
-      "grad_norm": 0.330078125,
-      "learning_rate": 7.357816397166091e-05,
-      "loss": 0.9009,
-      "step": 1955
-    },
-    {
-      "epoch": 9.423076923076923,
-      "grad_norm": 0.365234375,
-      "learning_rate": 7.303905926678564e-05,
-      "loss": 0.893,
-      "step": 1960
-    },
-    {
-      "epoch": 9.447115384615385,
-      "grad_norm": 0.341796875,
-      "learning_rate": 7.250079824521743e-05,
-      "loss": 0.8884,
-      "step": 1965
-    },
-    {
-      "epoch": 9.471153846153847,
-      "grad_norm": 0.365234375,
-      "learning_rate": 7.196339775065042e-05,
-      "loss": 0.8997,
-      "step": 1970
-    },
-    {
-      "epoch": 9.495192307692308,
-      "grad_norm": 0.326171875,
-      "learning_rate": 7.142687459985042e-05,
-      "loss": 0.898,
-      "step": 1975
-    },
-    {
-      "epoch": 9.51923076923077,
-      "grad_norm": 0.345703125,
-      "learning_rate": 7.089124558212871e-05,
-      "loss": 0.894,
-      "step": 1980
-    },
-    {
-      "epoch": 9.54326923076923,
-      "grad_norm": 0.35546875,
-      "learning_rate": 7.035652745881664e-05,
-      "loss": 0.8972,
-      "step": 1985
-    },
-    {
-      "epoch": 9.567307692307692,
-      "grad_norm": 0.359375,
-      "learning_rate": 6.982273696274106e-05,
-      "loss": 0.8951,
-      "step": 1990
-    },
-    {
-      "epoch": 9.591346153846153,
-      "grad_norm": 0.3515625,
-      "learning_rate": 6.928989079770094e-05,
-      "loss": 0.8947,
-      "step": 1995
-    },
-    {
-      "epoch": 9.615384615384615,
-      "grad_norm": 0.337890625,
-      "learning_rate": 6.875800563794425e-05,
-      "loss": 0.8904,
-      "step": 2000
-    },
-    {
-      "epoch": 9.639423076923077,
-      "grad_norm": 0.353515625,
-      "learning_rate": 6.822709812764664e-05,
-      "loss": 0.8921,
-      "step": 2005
-    },
-    {
-      "epoch": 9.663461538461538,
-      "grad_norm": 0.32421875,
-      "learning_rate": 6.769718488039023e-05,
-      "loss": 0.8993,
-      "step": 2010
-    },
-    {
-      "epoch": 9.6875,
-      "grad_norm": 0.322265625,
-      "learning_rate": 6.71682824786439e-05,
-      "loss": 0.8927,
-      "step": 2015
-    },
-    {
-      "epoch": 9.711538461538462,
-      "grad_norm": 0.326171875,
-      "learning_rate": 6.664040747324437e-05,
-      "loss": 0.8955,
-      "step": 2020
-    },
-    {
-      "epoch": 9.735576923076923,
-      "grad_norm": 0.33984375,
-      "learning_rate": 6.611357638287823e-05,
-      "loss": 0.8981,
-      "step": 2025
-    },
-    {
-      "epoch": 9.759615384615385,
-      "grad_norm": 0.392578125,
-      "learning_rate": 6.558780569356507e-05,
-      "loss": 0.8953,
-      "step": 2030
-    },
-    {
-      "epoch": 9.783653846153847,
-      "grad_norm": 0.34375,
-      "learning_rate": 6.506311185814152e-05,
-      "loss": 0.9053,
-      "step": 2035
-    },
-    {
-      "epoch": 9.807692307692308,
-      "grad_norm": 0.33984375,
-      "learning_rate": 6.453951129574644e-05,
-      "loss": 0.8988,
-      "step": 2040
-    },
-    {
-      "epoch": 9.83173076923077,
-      "grad_norm": 0.36328125,
-      "learning_rate": 6.40170203913072e-05,
-      "loss": 0.8978,
-      "step": 2045
-    },
-    {
-      "epoch": 9.85576923076923,
-      "grad_norm": 0.32421875,
-      "learning_rate": 6.349565549502676e-05,
-      "loss": 0.9107,
-      "step": 2050
-    },
-    {
-      "epoch": 9.879807692307692,
-      "grad_norm": 0.33203125,
-      "learning_rate": 6.297543292187215e-05,
-      "loss": 0.8935,
-      "step": 2055
-    },
-    {
-      "epoch": 9.903846153846153,
-      "grad_norm": 0.3359375,
-      "learning_rate": 6.245636895106402e-05,
-      "loss": 0.8963,
-      "step": 2060
-    },
-    {
-      "epoch": 9.927884615384615,
-      "grad_norm": 0.376953125,
-      "learning_rate": 6.193847982556698e-05,
-      "loss": 0.896,
-      "step": 2065
-    },
-    {
-      "epoch": 9.951923076923077,
-      "grad_norm": 0.328125,
-      "learning_rate": 6.142178175158149e-05,
-      "loss": 0.8851,
-      "step": 2070
-    },
-    {
-      "epoch": 9.975961538461538,
-      "grad_norm": 0.33984375,
-      "learning_rate": 6.090629089803668e-05,
-      "loss": 0.9031,
-      "step": 2075
-    },
-    {
-      "epoch": 10.0,
-      "grad_norm": 0.34375,
-      "learning_rate": 6.039202339608432e-05,
-      "loss": 0.9008,
-      "step": 2080
-    },
-    {
-      "epoch": 10.0,
-      "eval_loss": 2.5021300315856934,
-      "eval_runtime": 0.5361,
-      "eval_samples_per_second": 18.653,
-      "eval_steps_per_second": 1.865,
-      "step": 2080
-    },
-    {
-      "epoch": 10.024038461538462,
-      "grad_norm": 0.330078125,
-      "learning_rate": 5.9878995338594224e-05,
-      "loss": 0.8909,
-      "step": 2085
-    },
-    {
-      "epoch": 10.048076923076923,
-      "grad_norm": 0.3359375,
-      "learning_rate": 5.9367222779650334e-05,
-      "loss": 0.8932,
-      "step": 2090
-    },
-    {
-      "epoch": 10.072115384615385,
-      "grad_norm": 0.328125,
-      "learning_rate": 5.8856721734048706e-05,
-      "loss": 0.892,
-      "step": 2095
-    },
-    {
-      "epoch": 10.096153846153847,
-      "grad_norm": 0.384765625,
-      "learning_rate": 5.834750817679606e-05,
-      "loss": 0.8887,
-      "step": 2100
-    },
-    {
-      "epoch": 10.120192307692308,
-      "grad_norm": 0.328125,
-      "learning_rate": 5.783959804260997e-05,
-      "loss": 0.8847,
-      "step": 2105
-    },
-    {
-      "epoch": 10.14423076923077,
-      "grad_norm": 0.3203125,
-      "learning_rate": 5.733300722542045e-05,
-      "loss": 0.8805,
-      "step": 2110
-    },
-    {
-      "epoch": 10.16826923076923,
-      "grad_norm": 0.33203125,
-      "learning_rate": 5.682775157787213e-05,
-      "loss": 0.8819,
-      "step": 2115
-    },
-    {
-      "epoch": 10.192307692307692,
-      "grad_norm": 0.3671875,
-      "learning_rate": 5.6323846910828735e-05,
-      "loss": 0.8786,
-      "step": 2120
-    },
-    {
-      "epoch": 10.216346153846153,
-      "grad_norm": 0.345703125,
-      "learning_rate": 5.582130899287775e-05,
-      "loss": 0.8907,
-      "step": 2125
-    },
-    {
-      "epoch": 10.240384615384615,
-      "grad_norm": 0.427734375,
-      "learning_rate": 5.5320153549837415e-05,
-      "loss": 0.892,
-      "step": 2130
-    },
-    {
-      "epoch": 10.264423076923077,
-      "grad_norm": 0.359375,
-      "learning_rate": 5.482039626426447e-05,
-      "loss": 0.8916,
-      "step": 2135
-    },
-    {
-      "epoch": 10.288461538461538,
-      "grad_norm": 0.3828125,
-      "learning_rate": 5.432205277496327e-05,
-      "loss": 0.8822,
-      "step": 2140
-    },
-    {
-      "epoch": 10.3125,
-      "grad_norm": 0.3359375,
-      "learning_rate": 5.382513867649663e-05,
-      "loss": 0.8959,
-      "step": 2145
-    },
-    {
-      "epoch": 10.336538461538462,
-      "grad_norm": 0.3359375,
-      "learning_rate": 5.33296695186977e-05,
-      "loss": 0.8913,
-      "step": 2150
-    },
-    {
-      "epoch": 10.360576923076923,
-      "grad_norm": 0.333984375,
-      "learning_rate": 5.2835660806183405e-05,
-      "loss": 0.891,
-      "step": 2155
-    },
-    {
-      "epoch": 10.384615384615385,
-      "grad_norm": 0.337890625,
-      "learning_rate": 5.234312799786921e-05,
-      "loss": 0.8862,
-      "step": 2160
-    },
-    {
-      "epoch": 10.408653846153847,
-      "grad_norm": 0.357421875,
-      "learning_rate": 5.18520865064855e-05,
-      "loss": 0.8965,
-      "step": 2165
-    },
-    {
-      "epoch": 10.432692307692308,
-      "grad_norm": 0.34765625,
-      "learning_rate": 5.1362551698094964e-05,
-      "loss": 0.9014,
-      "step": 2170
-    },
-    {
-      "epoch": 10.45673076923077,
-      "grad_norm": 0.3671875,
-      "learning_rate": 5.087453889161229e-05,
-      "loss": 0.8867,
-      "step": 2175
-    },
-    {
-      "epoch": 10.48076923076923,
-      "grad_norm": 0.359375,
-      "learning_rate": 5.0388063358324134e-05,
-      "loss": 0.8859,
-      "step": 2180
-    },
-    {
-      "epoch": 10.504807692307692,
-      "grad_norm": 0.326171875,
-      "learning_rate": 4.990314032141178e-05,
-      "loss": 0.8908,
-      "step": 2185
-    },
-    {
-      "epoch": 10.528846153846153,
-      "grad_norm": 0.34375,
-      "learning_rate": 4.9419784955474524e-05,
-      "loss": 0.9005,
-      "step": 2190
-    },
-    {
-      "epoch": 10.552884615384615,
-      "grad_norm": 0.3359375,
-      "learning_rate": 4.8938012386054725e-05,
-      "loss": 0.8939,
-      "step": 2195
-    },
-    {
-      "epoch": 10.576923076923077,
-      "grad_norm": 0.333984375,
-      "learning_rate": 4.845783768916482e-05,
-      "loss": 0.8977,
-      "step": 2200
-    },
-    {
-      "epoch": 10.600961538461538,
-      "grad_norm": 0.31640625,
-      "learning_rate": 4.797927589081509e-05,
-      "loss": 0.8986,
-      "step": 2205
-    },
-    {
-      "epoch": 10.625,
-      "grad_norm": 0.36328125,
-      "learning_rate": 4.7502341966544e-05,
-      "loss": 0.8952,
-      "step": 2210
-    },
-    {
-      "epoch": 10.649038461538462,
-      "grad_norm": 0.341796875,
-      "learning_rate": 4.702705084094898e-05,
-      "loss": 0.8968,
-      "step": 2215
-    },
-    {
-      "epoch": 10.673076923076923,
-      "grad_norm": 0.330078125,
-      "learning_rate": 4.6553417387219886e-05,
-      "loss": 0.8996,
-      "step": 2220
-    },
-    {
-      "epoch": 10.697115384615385,
-      "grad_norm": 0.408203125,
-      "learning_rate": 4.608145642667336e-05,
-      "loss": 0.8928,
-      "step": 2225
-    },
-    {
-      "epoch": 10.721153846153847,
-      "grad_norm": 0.330078125,
-      "learning_rate": 4.5611182728288895e-05,
-      "loss": 0.8858,
-      "step": 2230
-    },
-    {
-      "epoch": 10.745192307692308,
-      "grad_norm": 0.33203125,
-      "learning_rate": 4.514261100824709e-05,
-      "loss": 0.8941,
-      "step": 2235
-    },
-    {
-      "epoch": 10.76923076923077,
-      "grad_norm": 0.341796875,
-      "learning_rate": 4.467575592946864e-05,
-      "loss": 0.8852,
-      "step": 2240
-    },
-    {
-      "epoch": 10.79326923076923,
-      "grad_norm": 0.3984375,
-      "learning_rate": 4.4210632101155847e-05,
-      "loss": 0.8874,
-      "step": 2245
-    },
-    {
-      "epoch": 10.817307692307692,
-      "grad_norm": 0.37109375,
-      "learning_rate": 4.374725407833532e-05,
-      "loss": 0.8955,
-      "step": 2250
-    },
-    {
-      "epoch": 10.841346153846153,
-      "grad_norm": 0.33984375,
-      "learning_rate": 4.328563636140256e-05,
-      "loss": 0.8823,
-      "step": 2255
-    },
-    {
-      "epoch": 10.865384615384615,
-      "grad_norm": 0.33203125,
-      "learning_rate": 4.282579339566802e-05,
-      "loss": 0.8892,
-      "step": 2260
-    },
-    {
-      "epoch": 10.889423076923077,
-      "grad_norm": 0.359375,
-      "learning_rate": 4.236773957090548e-05,
-      "loss": 0.8883,
-      "step": 2265
-    },
-    {
-      "epoch": 10.913461538461538,
-      "grad_norm": 0.330078125,
-      "learning_rate": 4.1911489220901236e-05,
-      "loss": 0.8946,
-      "step": 2270
-    },
-    {
-      "epoch": 10.9375,
-      "grad_norm": 0.33203125,
-      "learning_rate": 4.145705662300595e-05,
-      "loss": 0.8769,
-      "step": 2275
-    },
-    {
-      "epoch": 10.961538461538462,
-      "grad_norm": 0.32421875,
-      "learning_rate": 4.100445599768774e-05,
-      "loss": 0.8848,
-      "step": 2280
-    },
-    {
-      "epoch": 10.985576923076923,
-      "grad_norm": 0.37109375,
-      "learning_rate": 4.055370150808703e-05,
-      "loss": 0.8976,
-      "step": 2285
-    },
-    {
-      "epoch": 11.0,
-      "eval_loss": 2.5049757957458496,
-      "eval_runtime": 0.5344,
-      "eval_samples_per_second": 18.714,
-      "eval_steps_per_second": 1.871,
-      "step": 2288
-    },
-    {
-      "epoch": 11.009615384615385,
-      "grad_norm": 0.34765625,
-      "learning_rate": 4.0104807259573716e-05,
-      "loss": 0.8811,
-      "step": 2290
-    },
-    {
-      "epoch": 11.033653846153847,
-      "grad_norm": 0.330078125,
-      "learning_rate": 3.96577872993053e-05,
-      "loss": 0.8846,
-      "step": 2295
-    },
-    {
-      "epoch": 11.057692307692308,
-      "grad_norm": 0.34375,
-      "learning_rate": 3.9212655615787804e-05,
-      "loss": 0.894,
-      "step": 2300
-    },
-    {
-      "epoch": 11.08173076923077,
-      "grad_norm": 0.337890625,
-      "learning_rate": 3.876942613843757e-05,
-      "loss": 0.8879,
-      "step": 2305
-    },
-    {
-      "epoch": 11.10576923076923,
-      "grad_norm": 0.33203125,
-      "learning_rate": 3.832811273714569e-05,
-      "loss": 0.8876,
-      "step": 2310
-    },
-    {
-      "epoch": 11.129807692307692,
-      "grad_norm": 0.33984375,
-      "learning_rate": 3.78887292218439e-05,
-      "loss": 0.8936,
-      "step": 2315
-    },
-    {
-      "epoch": 11.153846153846153,
-      "grad_norm": 0.32421875,
-      "learning_rate": 3.745128934207225e-05,
-      "loss": 0.8858,
-      "step": 2320
-    },
-    {
-      "epoch": 11.177884615384615,
-      "grad_norm": 0.357421875,
-      "learning_rate": 3.701580678654925e-05,
-      "loss": 0.8879,
-      "step": 2325
-    },
-    {
-      "epoch": 11.201923076923077,
-      "grad_norm": 0.328125,
-      "learning_rate": 3.6582295182742964e-05,
-      "loss": 0.8887,
-      "step": 2330
-    },
-    {
-      "epoch": 11.225961538461538,
-      "grad_norm": 0.33984375,
-      "learning_rate": 3.6150768096445074e-05,
-      "loss": 0.8793,
-      "step": 2335
-    },
-    {
-      "epoch": 11.25,
-      "grad_norm": 0.33203125,
-      "learning_rate": 3.5721239031346066e-05,
-      "loss": 0.8744,
-      "step": 2340
-    },
-    {
-      "epoch": 11.274038461538462,
-      "grad_norm": 0.4140625,
-      "learning_rate": 3.529372142861277e-05,
-      "loss": 0.8903,
-      "step": 2345
-    },
-    {
-      "epoch": 11.298076923076923,
-      "grad_norm": 0.33984375,
-      "learning_rate": 3.4868228666467704e-05,
-      "loss": 0.8892,
-      "step": 2350
-    },
-    {
-      "epoch": 11.322115384615385,
-      "grad_norm": 0.35546875,
-      "learning_rate": 3.4444774059770536e-05,
-      "loss": 0.896,
-      "step": 2355
-    },
-    {
-      "epoch": 11.346153846153847,
-      "grad_norm": 0.357421875,
-      "learning_rate": 3.402337085960119e-05,
-      "loss": 0.8944,
-      "step": 2360
-    },
-    {
-      "epoch": 11.370192307692308,
-      "grad_norm": 0.3203125,
-      "learning_rate": 3.3604032252845474e-05,
-      "loss": 0.884,
-      "step": 2365
-    },
-    {
-      "epoch": 11.39423076923077,
-      "grad_norm": 0.32421875,
-      "learning_rate": 3.318677136178228e-05,
-      "loss": 0.8879,
-      "step": 2370
-    },
-    {
-      "epoch": 11.41826923076923,
-      "grad_norm": 0.330078125,
-      "learning_rate": 3.2771601243672825e-05,
-      "loss": 0.8893,
-      "step": 2375
-    },
-    {
-      "epoch": 11.442307692307692,
-      "grad_norm": 0.318359375,
-      "learning_rate": 3.235853489035241e-05,
-      "loss": 0.8884,
-      "step": 2380
-    },
-    {
-      "epoch": 11.466346153846153,
-      "grad_norm": 0.34375,
-      "learning_rate": 3.1947585227823394e-05,
-      "loss": 0.8873,
-      "step": 2385
-    },
-    {
-      "epoch": 11.490384615384615,
-      "grad_norm": 0.341796875,
-      "learning_rate": 3.153876511585122e-05,
-      "loss": 0.8853,
-      "step": 2390
-    },
-    {
-      "epoch": 11.514423076923077,
-      "grad_norm": 0.345703125,
-      "learning_rate": 3.113208734756151e-05,
-      "loss": 0.8937,
-      "step": 2395
-    },
-    {
-      "epoch": 11.538461538461538,
-      "grad_norm": 0.32421875,
-      "learning_rate": 3.072756464904006e-05,
-      "loss": 0.8798,
-      "step": 2400
-    },
-    {
-      "epoch": 11.5625,
-      "grad_norm": 0.3515625,
-      "learning_rate": 3.032520967893453e-05,
-      "loss": 0.877,
-      "step": 2405
-    },
-    {
-      "epoch": 11.586538461538462,
-      "grad_norm": 0.322265625,
-      "learning_rate": 2.9925035028058134e-05,
-      "loss": 0.8895,
-      "step": 2410
-    },
-    {
-      "epoch": 11.610576923076923,
-      "grad_norm": 0.33203125,
-      "learning_rate": 2.9527053218996037e-05,
-      "loss": 0.8875,
-      "step": 2415
-    },
-    {
-      "epoch": 11.634615384615385,
-      "grad_norm": 0.30859375,
-      "learning_rate": 2.9131276705713006e-05,
-      "loss": 0.8855,
-      "step": 2420
-    },
-    {
-      "epoch": 11.658653846153847,
-      "grad_norm": 0.3984375,
-      "learning_rate": 2.8737717873164094e-05,
-      "loss": 0.8879,
-      "step": 2425
-    },
-    {
-      "epoch": 11.682692307692308,
-      "grad_norm": 0.3515625,
-      "learning_rate": 2.8346389036906828e-05,
-      "loss": 0.8854,
-      "step": 2430
-    },
-    {
-      "epoch": 11.70673076923077,
-      "grad_norm": 0.328125,
-      "learning_rate": 2.795730244271596e-05,
-      "loss": 0.8852,
-      "step": 2435
-    },
-    {
-      "epoch": 11.73076923076923,
-      "grad_norm": 0.333984375,
-      "learning_rate": 2.7570470266200176e-05,
-      "loss": 0.891,
-      "step": 2440
-    },
-    {
-      "epoch": 11.754807692307692,
-      "grad_norm": 0.353515625,
-      "learning_rate": 2.7185904612421176e-05,
-      "loss": 0.8836,
-      "step": 2445
-    },
-    {
-      "epoch": 11.778846153846153,
-      "grad_norm": 0.32421875,
-      "learning_rate": 2.68036175155147e-05,
-      "loss": 0.8893,
-      "step": 2450
-    },
-    {
-      "epoch": 11.802884615384615,
-      "grad_norm": 0.376953125,
-      "learning_rate": 2.64236209383142e-05,
-      "loss": 0.8855,
-      "step": 2455
-    },
-    {
-      "epoch": 11.826923076923077,
-      "grad_norm": 0.3984375,
-      "learning_rate": 2.6045926771976303e-05,
-      "loss": 0.8822,
-      "step": 2460
-    },
-    {
-      "epoch": 11.850961538461538,
-      "grad_norm": 0.326171875,
-      "learning_rate": 2.5670546835608665e-05,
-      "loss": 0.8855,
-      "step": 2465
-    },
-    {
-      "epoch": 11.875,
-      "grad_norm": 0.337890625,
-      "learning_rate": 2.529749287590042e-05,
-      "loss": 0.8925,
-      "step": 2470
-    },
-    {
-      "epoch": 11.899038461538462,
-      "grad_norm": 0.330078125,
-      "learning_rate": 2.492677656675414e-05,
-      "loss": 0.8835,
-      "step": 2475
-    },
-    {
-      "epoch": 11.923076923076923,
-      "grad_norm": 0.328125,
-      "learning_rate": 2.4558409508920986e-05,
-      "loss": 0.8874,
-      "step": 2480
-    },
-    {
-      "epoch": 11.947115384615385,
-      "grad_norm": 0.328125,
-      "learning_rate": 2.4192403229637283e-05,
-      "loss": 0.8845,
-      "step": 2485
-    },
-    {
-      "epoch": 11.971153846153847,
-      "grad_norm": 0.34765625,
-      "learning_rate": 2.382876918226409e-05,
-      "loss": 0.884,
-      "step": 2490
-    },
-    {
-      "epoch": 11.995192307692308,
-      "grad_norm": 0.33984375,
-      "learning_rate": 2.3467518745928707e-05,
-      "loss": 0.8941,
-      "step": 2495
-    },
-    {
-      "epoch": 12.0,
-      "eval_loss": 2.51073956489563,
-      "eval_runtime": 0.5344,
-      "eval_samples_per_second": 18.713,
-      "eval_steps_per_second": 1.871,
-      "step": 2496
-    },
-    {
-      "epoch": 12.01923076923077,
-      "grad_norm": 0.3359375,
-      "learning_rate": 2.3108663225168435e-05,
-      "loss": 0.8827,
-      "step": 2500
-    },
-    {
-      "epoch": 12.04326923076923,
-      "grad_norm": 0.34375,
-      "learning_rate": 2.2752213849577188e-05,
-      "loss": 0.8774,
-      "step": 2505
-    },
-    {
-      "epoch": 12.067307692307692,
-      "grad_norm": 0.322265625,
-      "learning_rate": 2.239818177345364e-05,
-      "loss": 0.8794,
-      "step": 2510
-    },
-    {
-      "epoch": 12.091346153846153,
-      "grad_norm": 0.38671875,
-      "learning_rate": 2.204657807545252e-05,
-      "loss": 0.8883,
-      "step": 2515
-    },
-    {
-      "epoch": 12.115384615384615,
-      "grad_norm": 0.3203125,
-      "learning_rate": 2.1697413758237784e-05,
-      "loss": 0.8913,
-      "step": 2520
-    },
-    {
-      "epoch": 12.139423076923077,
-      "grad_norm": 0.34765625,
-      "learning_rate": 2.1350699748138326e-05,
-      "loss": 0.8774,
-      "step": 2525
-    },
-    {
-      "epoch": 12.163461538461538,
-      "grad_norm": 0.337890625,
-      "learning_rate": 2.1006446894806065e-05,
-      "loss": 0.8835,
-      "step": 2530
-    },
-    {
-      "epoch": 12.1875,
-      "grad_norm": 0.337890625,
-      "learning_rate": 2.0664665970876496e-05,
-      "loss": 0.898,
-      "step": 2535
-    },
-    {
-      "epoch": 12.211538461538462,
-      "grad_norm": 0.3359375,
-      "learning_rate": 2.032536767163141e-05,
-      "loss": 0.8826,
-      "step": 2540
-    },
-    {
-      "epoch": 12.235576923076923,
-      "grad_norm": 0.3359375,
-      "learning_rate": 1.9988562614664406e-05,
-      "loss": 0.8829,
-      "step": 2545
-    },
-    {
-      "epoch": 12.259615384615385,
-      "grad_norm": 0.353515625,
-      "learning_rate": 1.965426133954854e-05,
-      "loss": 0.8831,
-      "step": 2550
-    },
-    {
-      "epoch": 12.283653846153847,
-      "grad_norm": 0.322265625,
-      "learning_rate": 1.932247430750651e-05,
-      "loss": 0.8863,
-      "step": 2555
-    },
-    {
-      "epoch": 12.307692307692308,
-      "grad_norm": 0.341796875,
-      "learning_rate": 1.8993211901083353e-05,
-      "loss": 0.8851,
-      "step": 2560
-    },
-    {
-      "epoch": 12.33173076923077,
-      "grad_norm": 0.328125,
-      "learning_rate": 1.8666484423821373e-05,
-      "loss": 0.8853,
-      "step": 2565
-    },
-    {
-      "epoch": 12.35576923076923,
-      "grad_norm": 0.33203125,
-      "learning_rate": 1.8342302099938057e-05,
-      "loss": 0.8879,
-      "step": 2570
-    },
-    {
-      "epoch": 12.379807692307692,
-      "grad_norm": 0.328125,
-      "learning_rate": 1.8020675074005723e-05,
-      "loss": 0.8847,
-      "step": 2575
-    },
-    {
-      "epoch": 12.403846153846153,
-      "grad_norm": 0.3359375,
-      "learning_rate": 1.7701613410634365e-05,
-      "loss": 0.8883,
-      "step": 2580
-    },
-    {
-      "epoch": 12.427884615384615,
-      "grad_norm": 0.333984375,
-      "learning_rate": 1.7385127094156594e-05,
-      "loss": 0.8906,
-      "step": 2585
-    },
-    {
-      "epoch": 12.451923076923077,
-      "grad_norm": 0.33203125,
-      "learning_rate": 1.7071226028315113e-05,
-      "loss": 0.8795,
-      "step": 2590
-    },
-    {
-      "epoch": 12.475961538461538,
-      "grad_norm": 0.359375,
-      "learning_rate": 1.6759920035953093e-05,
-      "loss": 0.8835,
-      "step": 2595
-    },
-    {
-      "epoch": 12.5,
-      "grad_norm": 0.32421875,
-      "learning_rate": 1.6451218858706374e-05,
-      "loss": 0.8825,
-      "step": 2600
-    },
-    {
-      "epoch": 12.524038461538462,
-      "grad_norm": 0.34375,
-      "learning_rate": 1.614513215669896e-05,
-      "loss": 0.8869,
-      "step": 2605
-    },
-    {
-      "epoch": 12.548076923076923,
-      "grad_norm": 0.32421875,
-      "learning_rate": 1.584166950824061e-05,
-      "loss": 0.8794,
-      "step": 2610
-    },
-    {
-      "epoch": 12.572115384615385,
-      "grad_norm": 0.337890625,
-      "learning_rate": 1.5540840409527045e-05,
-      "loss": 0.8836,
-      "step": 2615
-    },
-    {
-      "epoch": 12.596153846153847,
-      "grad_norm": 0.318359375,
-      "learning_rate": 1.5242654274342894e-05,
-      "loss": 0.891,
-      "step": 2620
-    },
-    {
-      "epoch": 12.620192307692308,
-      "grad_norm": 0.3203125,
-      "learning_rate": 1.4947120433767047e-05,
-      "loss": 0.8896,
-      "step": 2625
-    },
-    {
-      "epoch": 12.64423076923077,
-      "grad_norm": 0.330078125,
-      "learning_rate": 1.4654248135880621e-05,
-      "loss": 0.8919,
-      "step": 2630
-    },
-    {
-      "epoch": 12.66826923076923,
-      "grad_norm": 0.318359375,
-      "learning_rate": 1.4364046545477672e-05,
-      "loss": 0.8864,
-      "step": 2635
-    },
-    {
-      "epoch": 12.692307692307692,
-      "grad_norm": 0.3203125,
-      "learning_rate": 1.4076524743778319e-05,
-      "loss": 0.8865,
-      "step": 2640
-    },
-    {
-      "epoch": 12.716346153846153,
-      "grad_norm": 0.31640625,
-      "learning_rate": 1.3791691728144606e-05,
-      "loss": 0.8829,
-      "step": 2645
-    },
-    {
-      "epoch": 12.740384615384615,
-      "grad_norm": 0.328125,
-      "learning_rate": 1.350955641179893e-05,
-      "loss": 0.8856,
-      "step": 2650
-    },
-    {
-      "epoch": 12.764423076923077,
-      "grad_norm": 0.3203125,
-      "learning_rate": 1.3230127623545064e-05,
-      "loss": 0.8815,
-      "step": 2655
-    },
-    {
-      "epoch": 12.788461538461538,
-      "grad_norm": 0.353515625,
-      "learning_rate": 1.295341410749208e-05,
-      "loss": 0.8922,
-      "step": 2660
-    },
-    {
-      "epoch": 12.8125,
-      "grad_norm": 0.337890625,
-      "learning_rate": 1.2679424522780426e-05,
-      "loss": 0.8803,
-      "step": 2665
-    },
-    {
-      "epoch": 12.836538461538462,
-      "grad_norm": 0.330078125,
-      "learning_rate": 1.2408167443311214e-05,
-      "loss": 0.8835,
-      "step": 2670
-    },
-    {
-      "epoch": 12.860576923076923,
-      "grad_norm": 0.314453125,
-      "learning_rate": 1.2139651357477788e-05,
-      "loss": 0.8831,
-      "step": 2675
-    },
-    {
-      "epoch": 12.884615384615385,
-      "grad_norm": 0.328125,
-      "learning_rate": 1.1873884667900125e-05,
-      "loss": 0.8905,
-      "step": 2680
-    },
-    {
-      "epoch": 12.908653846153847,
-      "grad_norm": 0.318359375,
-      "learning_rate": 1.1610875691161915e-05,
-      "loss": 0.8874,
-      "step": 2685
-    },
-    {
-      "epoch": 12.932692307692308,
-      "grad_norm": 0.3125,
-      "learning_rate": 1.1350632657550253e-05,
-      "loss": 0.8754,
-      "step": 2690
-    },
-    {
-      "epoch": 12.95673076923077,
-      "grad_norm": 0.32421875,
-      "learning_rate": 1.109316371079816e-05,
-      "loss": 0.887,
-      "step": 2695
-    },
-    {
-      "epoch": 12.98076923076923,
-      "grad_norm": 0.326171875,
-      "learning_rate": 1.083847690782972e-05,
-      "loss": 0.8878,
-      "step": 2700
-    },
-    {
-      "epoch": 13.0,
-      "eval_loss": 2.5122883319854736,
-      "eval_runtime": 0.5322,
-      "eval_samples_per_second": 18.791,
-      "eval_steps_per_second": 1.879,
-      "step": 2704
-    },
-    {
-      "epoch": 13.004807692307692,
-      "grad_norm": 0.32421875,
-      "learning_rate": 1.0586580218507946e-05,
-      "loss": 0.8835,
-      "step": 2705
-    },
-    {
-      "epoch": 13.028846153846153,
-      "grad_norm": 0.326171875,
-      "learning_rate": 1.0337481525385362e-05,
-      "loss": 0.8878,
-      "step": 2710
-    },
-    {
-      "epoch": 13.052884615384615,
-      "grad_norm": 0.34375,
-      "learning_rate": 1.0091188623457415e-05,
-      "loss": 0.8848,
-      "step": 2715
-    },
-    {
-      "epoch": 13.076923076923077,
-      "grad_norm": 0.3203125,
-      "learning_rate": 9.8477092199184e-06,
-      "loss": 0.8916,
-      "step": 2720
-    },
-    {
-      "epoch": 13.100961538461538,
-      "grad_norm": 0.32421875,
-      "learning_rate": 9.607050933920459e-06,
-      "loss": 0.8831,
-      "step": 2725
-    },
-    {
-      "epoch": 13.125,
-      "grad_norm": 0.32421875,
-      "learning_rate": 9.369221296335006e-06,
-      "loss": 0.8942,
-      "step": 2730
-    },
-    {
-      "epoch": 13.149038461538462,
-      "grad_norm": 0.3359375,
-      "learning_rate": 9.134227749517177e-06,
-      "loss": 0.8944,
-      "step": 2735
-    },
-    {
-      "epoch": 13.173076923076923,
-      "grad_norm": 0.318359375,
-      "learning_rate": 8.902077647072881e-06,
-      "loss": 0.8764,
-      "step": 2740
-    },
-    {
-      "epoch": 13.197115384615385,
-      "grad_norm": 0.31640625,
-      "learning_rate": 8.672778253628621e-06,
-      "loss": 0.8817,
-      "step": 2745
-    },
-    {
-      "epoch": 13.221153846153847,
-      "grad_norm": 0.333984375,
-      "learning_rate": 8.446336744604378e-06,
-      "loss": 0.8873,
-      "step": 2750
-    },
-    {
-      "epoch": 13.245192307692308,
-      "grad_norm": 0.318359375,
-      "learning_rate": 8.222760205988766e-06,
-      "loss": 0.88,
-      "step": 2755
-    },
-    {
-      "epoch": 13.26923076923077,
-      "grad_norm": 0.318359375,
-      "learning_rate": 8.002055634117578e-06,
-      "loss": 0.8794,
-      "step": 2760
-    },
-    {
-      "epoch": 13.29326923076923,
-      "grad_norm": 0.3203125,
-      "learning_rate": 7.78422993545468e-06,
-      "loss": 0.8854,
-      "step": 2765
-    },
-    {
-      "epoch": 13.317307692307692,
-      "grad_norm": 0.328125,
-      "learning_rate": 7.569289926375933e-06,
-      "loss": 0.8839,
-      "step": 2770
-    },
-    {
-      "epoch": 13.341346153846153,
-      "grad_norm": 0.341796875,
-      "learning_rate": 7.357242332955916e-06,
-      "loss": 0.8878,
-      "step": 2775
-    },
-    {
-      "epoch": 13.365384615384615,
-      "grad_norm": 0.328125,
-      "learning_rate": 7.148093790757371e-06,
-      "loss": 0.8866,
-      "step": 2780
-    },
-    {
-      "epoch": 13.389423076923077,
-      "grad_norm": 0.341796875,
-      "learning_rate": 6.941850844623688e-06,
-      "loss": 0.8907,
-      "step": 2785
-    },
-    {
-      "epoch": 13.413461538461538,
-      "grad_norm": 0.326171875,
-      "learning_rate": 6.738519948473976e-06,
-      "loss": 0.8923,
-      "step": 2790
-    },
-    {
-      "epoch": 13.4375,
-      "grad_norm": 0.337890625,
-      "learning_rate": 6.538107465101162e-06,
-      "loss": 0.8786,
-      "step": 2795
-    },
-    {
-      "epoch": 13.461538461538462,
-      "grad_norm": 0.322265625,
-      "learning_rate": 6.3406196659728465e-06,
-      "loss": 0.8785,
-      "step": 2800
-    },
-    {
-      "epoch": 13.485576923076923,
-      "grad_norm": 0.3203125,
-      "learning_rate": 6.146062731035129e-06,
-      "loss": 0.8755,
-      "step": 2805
-    },
-    {
-      "epoch": 13.509615384615385,
-      "grad_norm": 0.322265625,
-      "learning_rate": 5.954442748519073e-06,
-      "loss": 0.8818,
-      "step": 2810
-    },
-    {
-      "epoch": 13.533653846153847,
-      "grad_norm": 0.3359375,
-      "learning_rate": 5.765765714750371e-06,
-      "loss": 0.8773,
-      "step": 2815
-    },
-    {
-      "epoch": 13.557692307692308,
-      "grad_norm": 0.341796875,
-      "learning_rate": 5.580037533961546e-06,
-      "loss": 0.8927,
-      "step": 2820
-    },
-    {
-      "epoch": 13.58173076923077,
-      "grad_norm": 0.3359375,
-      "learning_rate": 5.397264018107295e-06,
-      "loss": 0.8942,
-      "step": 2825
-    },
-    {
-      "epoch": 13.60576923076923,
-      "grad_norm": 0.3359375,
-      "learning_rate": 5.217450886682584e-06,
-      "loss": 0.8837,
-      "step": 2830
-    },
-    {
-      "epoch": 13.629807692307692,
-      "grad_norm": 0.330078125,
-      "learning_rate": 5.040603766543594e-06,
-      "loss": 0.8774,
-      "step": 2835
-    },
-    {
-      "epoch": 13.653846153846153,
-      "grad_norm": 0.3359375,
-      "learning_rate": 4.866728191731829e-06,
-      "loss": 0.8844,
-      "step": 2840
-    },
-    {
-      "epoch": 13.677884615384615,
-      "grad_norm": 0.330078125,
-      "learning_rate": 4.695829603300694e-06,
-      "loss": 0.8878,
-      "step": 2845
-    },
-    {
-      "epoch": 13.701923076923077,
-      "grad_norm": 0.322265625,
-      "learning_rate": 4.527913349145441e-06,
-      "loss": 0.882,
-      "step": 2850
-    },
-    {
-      "epoch": 13.725961538461538,
-      "grad_norm": 0.326171875,
-      "learning_rate": 4.362984683835702e-06,
-      "loss": 0.8775,
-      "step": 2855
-    },
-    {
-      "epoch": 13.75,
-      "grad_norm": 0.333984375,
-      "learning_rate": 4.20104876845111e-06,
-      "loss": 0.8757,
-      "step": 2860
-    },
-    {
-      "epoch": 13.774038461538462,
-      "grad_norm": 0.314453125,
-      "learning_rate": 4.042110670419763e-06,
-      "loss": 0.8863,
-      "step": 2865
-    },
-    {
-      "epoch": 13.798076923076923,
-      "grad_norm": 0.3359375,
-      "learning_rate": 3.886175363359646e-06,
-      "loss": 0.8863,
-      "step": 2870
-    },
-    {
-      "epoch": 13.822115384615385,
-      "grad_norm": 0.3203125,
-      "learning_rate": 3.733247726923039e-06,
-      "loss": 0.8933,
-      "step": 2875
-    },
-    {
-      "epoch": 13.846153846153847,
-      "grad_norm": 0.30859375,
-      "learning_rate": 3.5833325466437694e-06,
-      "loss": 0.8857,
-      "step": 2880
-    },
-    {
-      "epoch": 13.870192307692308,
-      "grad_norm": 0.31640625,
-      "learning_rate": 3.436434513787479e-06,
-      "loss": 0.8709,
-      "step": 2885
-    },
-    {
-      "epoch": 13.89423076923077,
-      "grad_norm": 0.330078125,
-      "learning_rate": 3.2925582252048338e-06,
-      "loss": 0.8779,
-      "step": 2890
-    },
-    {
-      "epoch": 13.91826923076923,
-      "grad_norm": 0.326171875,
-      "learning_rate": 3.1517081831876737e-06,
-      "loss": 0.8878,
-      "step": 2895
-    },
-    {
-      "epoch": 13.942307692307692,
-      "grad_norm": 0.31640625,
-      "learning_rate": 3.013888795328057e-06,
-      "loss": 0.8875,
-      "step": 2900
-    },
-    {
-      "epoch": 13.966346153846153,
-      "grad_norm": 0.33203125,
-      "learning_rate": 2.879104374380448e-06,
-      "loss": 0.8884,
-      "step": 2905
-    },
-    {
-      "epoch": 13.990384615384615,
-      "grad_norm": 0.341796875,
-      "learning_rate": 2.7473591381266708e-06,
-      "loss": 0.8896,
-      "step": 2910
-    },
-    {
-      "epoch": 14.0,
-      "eval_loss": 2.5119779109954834,
-      "eval_runtime": 0.5387,
-      "eval_samples_per_second": 18.562,
-      "eval_steps_per_second": 1.856,
-      "step": 2912
-    },
-    {
-      "epoch": 14.014423076923077,
-      "grad_norm": 0.34375,
-      "learning_rate": 2.618657209243969e-06,
-      "loss": 0.8718,
-      "step": 2915
-    },
-    {
-      "epoch": 14.038461538461538,
-      "grad_norm": 0.314453125,
-      "learning_rate": 2.4930026151759766e-06,
-      "loss": 0.8855,
-      "step": 2920
-    },
-    {
-      "epoch": 14.0625,
-      "grad_norm": 0.345703125,
-      "learning_rate": 2.3703992880066638e-06,
-      "loss": 0.888,
-      "step": 2925
-    },
-    {
-      "epoch": 14.086538461538462,
-      "grad_norm": 0.35546875,
-      "learning_rate": 2.250851064337367e-06,
-      "loss": 0.8745,
-      "step": 2930
-    },
-    {
-      "epoch": 14.110576923076923,
-      "grad_norm": 0.369140625,
-      "learning_rate": 2.1343616851666305e-06,
-      "loss": 0.8862,
-      "step": 2935
-    },
-    {
-      "epoch": 14.134615384615385,
-      "grad_norm": 0.31640625,
-      "learning_rate": 2.0209347957732328e-06,
-      "loss": 0.8744,
-      "step": 2940
-    },
-    {
-      "epoch": 14.158653846153847,
-      "grad_norm": 0.318359375,
-      "learning_rate": 1.9105739456020454e-06,
-      "loss": 0.887,
-      "step": 2945
-    },
-    {
-      "epoch": 14.182692307692308,
-      "grad_norm": 0.31640625,
-      "learning_rate": 1.8032825881530213e-06,
-      "loss": 0.885,
-      "step": 2950
-    },
-    {
-      "epoch": 14.20673076923077,
-      "grad_norm": 0.326171875,
-      "learning_rate": 1.6990640808730696e-06,
-      "loss": 0.8811,
-      "step": 2955
-    },
-    {
-      "epoch": 14.23076923076923,
-      "grad_norm": 0.326171875,
-      "learning_rate": 1.5979216850509848e-06,
-      "loss": 0.8877,
-      "step": 2960
-    },
-    {
-      "epoch": 14.254807692307692,
-      "grad_norm": 0.318359375,
-      "learning_rate": 1.4998585657155173e-06,
-      "loss": 0.887,
-      "step": 2965
-    },
-    {
-      "epoch": 14.278846153846153,
-      "grad_norm": 0.330078125,
-      "learning_rate": 1.404877791536141e-06,
-      "loss": 0.8912,
-      "step": 2970
-    },
-    {
-      "epoch": 14.302884615384615,
-      "grad_norm": 0.322265625,
-      "learning_rate": 1.3129823347271753e-06,
-      "loss": 0.8926,
-      "step": 2975
-    },
-    {
-      "epoch": 14.326923076923077,
-      "grad_norm": 0.330078125,
-      "learning_rate": 1.2241750709546917e-06,
-      "loss": 0.8829,
-      "step": 2980
-    },
-    {
-      "epoch": 14.350961538461538,
-      "grad_norm": 0.3125,
-      "learning_rate": 1.1384587792465872e-06,
-      "loss": 0.8864,
-      "step": 2985
-    },
-    {
-      "epoch": 14.375,
-      "grad_norm": 0.318359375,
-      "learning_rate": 1.055836141905553e-06,
-      "loss": 0.8787,
-      "step": 2990
-    },
-    {
-      "epoch": 14.399038461538462,
-      "grad_norm": 0.318359375,
-      "learning_rate": 9.763097444252078e-07,
-      "loss": 0.8863,
-      "step": 2995
-    },
-    {
-      "epoch": 14.423076923076923,
-      "grad_norm": 0.31640625,
-      "learning_rate": 8.998820754091531e-07,
-      "loss": 0.8844,
-      "step": 3000
-    },
-    {
-      "epoch": 14.447115384615385,
-      "grad_norm": 0.33203125,
-      "learning_rate": 8.265555264930891e-07,
-      "loss": 0.8895,
-      "step": 3005
-    },
-    {
-      "epoch": 14.471153846153847,
-      "grad_norm": 0.31640625,
-      "learning_rate": 7.563323922699983e-07,
-      "loss": 0.893,
-      "step": 3010
-    },
-    {
-      "epoch": 14.495192307692308,
-      "grad_norm": 0.33203125,
-      "learning_rate": 6.892148702183133e-07,
-      "loss": 0.8808,
-      "step": 3015
-    },
-    {
-      "epoch": 14.51923076923077,
-      "grad_norm": 0.322265625,
-      "learning_rate": 6.25205060633205e-07,
-      "loss": 0.8786,
-      "step": 3020
-    },
-    {
-      "epoch": 14.54326923076923,
-      "grad_norm": 0.3125,
-      "learning_rate": 5.643049665607691e-07,
-      "loss": 0.8847,
-      "step": 3025
-    },
-    {
-      "epoch": 14.567307692307692,
-      "grad_norm": 0.3125,
-      "learning_rate": 5.065164937354428e-07,
-      "loss": 0.8757,
-      "step": 3030
-    },
-    {
-      "epoch": 14.591346153846153,
-      "grad_norm": 0.328125,
-      "learning_rate": 4.5184145052029615e-07,
-      "loss": 0.8823,
-      "step": 3035
-    },
-    {
-      "epoch": 14.615384615384615,
-      "grad_norm": 0.32421875,
-      "learning_rate": 4.0028154785050063e-07,
-      "loss": 0.884,
-      "step": 3040
-    },
-    {
-      "epoch": 14.639423076923077,
-      "grad_norm": 0.32421875,
-      "learning_rate": 3.5183839917972697e-07,
-      "loss": 0.8832,
-      "step": 3045
-    },
-    {
-      "epoch": 14.663461538461538,
-      "grad_norm": 0.333984375,
-      "learning_rate": 3.065135204296965e-07,
-      "loss": 0.8867,
-      "step": 3050
-    },
-    {
-      "epoch": 14.6875,
-      "grad_norm": 0.375,
-      "learning_rate": 2.643083299427751e-07,
-      "loss": 0.8779,
-      "step": 3055
-    },
-    {
-      "epoch": 14.711538461538462,
-      "grad_norm": 0.3203125,
-      "learning_rate": 2.2522414843748618e-07,
-      "loss": 0.8855,
-      "step": 3060
-    },
-    {
-      "epoch": 14.735576923076923,
-      "grad_norm": 0.333984375,
-      "learning_rate": 1.8926219896728825e-07,
-      "loss": 0.882,
-      "step": 3065
-    },
-    {
-      "epoch": 14.759615384615385,
-      "grad_norm": 0.322265625,
-      "learning_rate": 1.5642360688225e-07,
-      "loss": 0.8838,
-      "step": 3070
-    },
-    {
-      "epoch": 14.783653846153847,
-      "grad_norm": 0.314453125,
-      "learning_rate": 1.2670939979384512e-07,
-      "loss": 0.889,
-      "step": 3075
-    },
-    {
-      "epoch": 14.807692307692308,
-      "grad_norm": 0.337890625,
-      "learning_rate": 1.0012050754277802e-07,
-      "loss": 0.8875,
-      "step": 3080
-    },
-    {
-      "epoch": 14.83173076923077,
-      "grad_norm": 0.353515625,
-      "learning_rate": 7.665776216992938e-08,
-      "loss": 0.8861,
-      "step": 3085
-    },
-    {
-      "epoch": 14.85576923076923,
-      "grad_norm": 0.326171875,
-      "learning_rate": 5.632189789027687e-08,
-      "loss": 0.8823,
-      "step": 3090
-    },
-    {
-      "epoch": 14.879807692307692,
-      "grad_norm": 0.32421875,
-      "learning_rate": 3.911355106994696e-08,
-      "loss": 0.887,
-      "step": 3095
-    },
-    {
-      "epoch": 14.903846153846153,
-      "grad_norm": 0.314453125,
-      "learning_rate": 2.5033260206275277e-08,
-      "loss": 0.8832,
-      "step": 3100
-    },
-    {
-      "epoch": 14.927884615384615,
-      "grad_norm": 0.3125,
-      "learning_rate": 1.4081465910975588e-08,
-      "loss": 0.8871,
-      "step": 3105
-    },
-    {
-      "epoch": 14.951923076923077,
-      "grad_norm": 0.345703125,
-      "learning_rate": 6.25851089636198e-09,
-      "loss": 0.8929,
-      "step": 3110
-    },
-    {
-      "epoch": 14.975961538461538,
-      "grad_norm": 0.31640625,
-      "learning_rate": 1.5646399645796728e-09,
-      "loss": 0.884,
-      "step": 3115
-    },
-    {
-      "epoch": 15.0,
-      "grad_norm": 0.34765625,
-      "learning_rate": 0.0,
-      "loss": 0.8797,
-      "step": 3120
-    },
-    {
-      "epoch": 15.0,
-      "eval_loss": 2.511939525604248,
-      "eval_runtime": 0.5362,
-      "eval_samples_per_second": 18.65,
-      "eval_steps_per_second": 1.865,
-      "step": 3120
-    },
-    {
-      "epoch": 15.0,
-      "step": 3120,
-      "total_flos": 2.443473144564941e+18,
-      "train_loss": 0.9840393823690904,
-      "train_runtime": 10705.8296,
-      "train_samples_per_second": 18.636,
-      "train_steps_per_second": 0.291
+      "epoch": 10.0,
+      "step": 1040,
+      "total_flos": 1.6530423313701274e+18,
+      "train_loss": 0.977329820394516,
+      "train_runtime": 7126.3412,
+      "train_samples_per_second": 18.665,
+      "train_steps_per_second": 0.146
     }
   ],
   "logging_steps": 5,
-  "max_steps": 3120,
+  "max_steps": 1040,
   "num_input_tokens_seen": 0,
-  "num_train_epochs": 15,
+  "num_train_epochs": 10,
   "save_steps": 100,
-  "total_flos": 2.443473144564941e+18,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.6530423313701274e+18,
   "train_batch_size": 8,
   "trial_name": null,
   "trial_params": null