diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -10,9 +10,9 @@
   "log_history": [
     {
       "epoch": 1.0,
-      "grad_norm": 1.391684889793396,
+      "grad_norm": 1.549831748008728,
       "learning_rate": 4.9500000000000004e-05,
-      "loss": 1.1481,
+      "loss": 1.1472,
       "step": 106
     },
     {
@@ -37,21 +37,21 @@
       "eval_TIME_number": 34,
       "eval_TIME_precision": 0.0,
       "eval_TIME_recall": 0.0,
-      "eval_loss": 0.7146090865135193,
+      "eval_loss": 0.7123244404792786,
       "eval_overall_accuracy": 0.8365739601279842,
       "eval_overall_f1": 0.0,
       "eval_overall_precision": 0.0,
       "eval_overall_recall": 0.0,
-      "eval_runtime": 0.9362,
-      "eval_samples_per_second": 199.746,
-      "eval_steps_per_second": 3.204,
+      "eval_runtime": 0.3731,
+      "eval_samples_per_second": 501.176,
+      "eval_steps_per_second": 8.04,
       "step": 106
     },
     {
       "epoch": 2.0,
-      "grad_norm": 1.086163878440857,
+      "grad_norm": 1.0792208909988403,
       "learning_rate": 4.9e-05,
-      "loss": 0.6809,
+      "loss": 0.6802,
       "step": 212
     },
     {
@@ -64,9 +64,9 @@
       "eval_ORGANIZATION_number": 65,
       "eval_ORGANIZATION_precision": 0.0,
       "eval_ORGANIZATION_recall": 0.0,
-      "eval_PERSON_f1": 0.012820512820512822,
+      "eval_PERSON_f1": 0.012738853503184716,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.16666666666666666,
+      "eval_PERSON_precision": 0.14285714285714285,
       "eval_PERSON_recall": 0.006666666666666667,
       "eval_QUANTITY_f1": 0.0,
       "eval_QUANTITY_number": 29,
@@ -76,567 +76,567 @@
       "eval_TIME_number": 34,
       "eval_TIME_precision": 0.0,
       "eval_TIME_recall": 0.0,
-      "eval_loss": 0.5919493436813354,
+      "eval_loss": 0.5882730484008789,
       "eval_overall_accuracy": 0.8385429485601772,
-      "eval_overall_f1": 0.005509641873278238,
-      "eval_overall_precision": 0.08333333333333333,
+      "eval_overall_f1": 0.005524861878453039,
+      "eval_overall_precision": 0.09090909090909091,
       "eval_overall_recall": 0.002849002849002849,
-      "eval_runtime": 0.974,
-      "eval_samples_per_second": 191.988,
-      "eval_steps_per_second": 3.08,
+      "eval_runtime": 0.3704,
+      "eval_samples_per_second": 504.825,
+      "eval_steps_per_second": 8.099,
       "step": 212
     },
     {
       "epoch": 3.0,
-      "grad_norm": 1.1405037641525269,
+      "grad_norm": 1.0924339294433594,
       "learning_rate": 4.85e-05,
-      "loss": 0.5553,
+      "loss": 0.56,
       "step": 318
     },
     {
       "epoch": 3.0,
-      "eval_LOCATION_f1": 0.09302325581395347,
+      "eval_LOCATION_f1": 0.1176470588235294,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.3076923076923077,
-      "eval_LOCATION_recall": 0.0547945205479452,
-      "eval_ORGANIZATION_f1": 0.0,
+      "eval_LOCATION_precision": 0.4166666666666667,
+      "eval_LOCATION_recall": 0.0684931506849315,
+      "eval_ORGANIZATION_f1": 0.025974025974025976,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.0,
-      "eval_ORGANIZATION_recall": 0.0,
-      "eval_PERSON_f1": 0.18834080717488788,
+      "eval_ORGANIZATION_precision": 0.08333333333333333,
+      "eval_ORGANIZATION_recall": 0.015384615384615385,
+      "eval_PERSON_f1": 0.17674418604651165,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.2876712328767123,
-      "eval_PERSON_recall": 0.14,
+      "eval_PERSON_precision": 0.2923076923076923,
+      "eval_PERSON_recall": 0.12666666666666668,
       "eval_QUANTITY_f1": 0.0,
       "eval_QUANTITY_number": 29,
       "eval_QUANTITY_precision": 0.0,
       "eval_QUANTITY_recall": 0.0,
-      "eval_TIME_f1": 0.11538461538461539,
-      "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.16666666666666666,
-      "eval_TIME_recall": 0.08823529411764706,
-      "eval_loss": 0.45798516273498535,
-      "eval_overall_accuracy": 0.8599556977602757,
-      "eval_overall_f1": 0.12043010752688173,
-      "eval_overall_precision": 0.24561403508771928,
-      "eval_overall_recall": 0.07977207977207977,
-      "eval_runtime": 1.0183,
-      "eval_samples_per_second": 183.643,
-      "eval_steps_per_second": 2.946,
+      "eval_TIME_f1": 0.25925925925925924,
+      "eval_TIME_number": 34,
+      "eval_TIME_precision": 0.35,
+      "eval_TIME_recall": 0.20588235294117646,
+      "eval_loss": 0.4647313952445984,
+      "eval_overall_accuracy": 0.8616785626384444,
+      "eval_overall_f1": 0.1391304347826087,
+      "eval_overall_precision": 0.29357798165137616,
+      "eval_overall_recall": 0.09116809116809117,
+      "eval_runtime": 0.3714,
+      "eval_samples_per_second": 503.536,
+      "eval_steps_per_second": 8.078,
       "step": 318
     },
     {
       "epoch": 4.0,
-      "grad_norm": 1.2108490467071533,
+      "grad_norm": 1.4220308065414429,
       "learning_rate": 4.8e-05,
-      "loss": 0.4289,
+      "loss": 0.4294,
       "step": 424
     },
     {
       "epoch": 4.0,
-      "eval_LOCATION_f1": 0.40601503759398494,
-      "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.45,
-      "eval_LOCATION_recall": 0.3698630136986301,
-      "eval_ORGANIZATION_f1": 0.14,
-      "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.2,
-      "eval_ORGANIZATION_recall": 0.1076923076923077,
-      "eval_PERSON_f1": 0.6265060240963856,
-      "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.5714285714285714,
-      "eval_PERSON_recall": 0.6933333333333334,
-      "eval_QUANTITY_f1": 0.2173913043478261,
-      "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.29411764705882354,
-      "eval_QUANTITY_recall": 0.1724137931034483,
-      "eval_TIME_f1": 0.5074626865671642,
-      "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.5151515151515151,
-      "eval_TIME_recall": 0.5,
-      "eval_loss": 0.3399774134159088,
-      "eval_overall_accuracy": 0.9059808023627861,
-      "eval_overall_f1": 0.471976401179941,
-      "eval_overall_precision": 0.4892966360856269,
-      "eval_overall_recall": 0.45584045584045585,
-      "eval_runtime": 0.9802,
-      "eval_samples_per_second": 190.773,
-      "eval_steps_per_second": 3.061,
+      "eval_LOCATION_f1": 0.39669421487603307,
+      "eval_LOCATION_number": 73,
+      "eval_LOCATION_precision": 0.5,
+      "eval_LOCATION_recall": 0.3287671232876712,
+      "eval_ORGANIZATION_f1": 0.12903225806451613,
+      "eval_ORGANIZATION_number": 65,
+      "eval_ORGANIZATION_precision": 0.21428571428571427,
+      "eval_ORGANIZATION_recall": 0.09230769230769231,
+      "eval_PERSON_f1": 0.6538461538461539,
+      "eval_PERSON_number": 150,
+      "eval_PERSON_precision": 0.6296296296296297,
+      "eval_PERSON_recall": 0.68,
+      "eval_QUANTITY_f1": 0.13333333333333333,
+      "eval_QUANTITY_number": 29,
+      "eval_QUANTITY_precision": 0.1875,
+      "eval_QUANTITY_recall": 0.10344827586206896,
+      "eval_TIME_f1": 0.5671641791044776,
+      "eval_TIME_number": 34,
+      "eval_TIME_precision": 0.5757575757575758,
+      "eval_TIME_recall": 0.5588235294117647,
+      "eval_loss": 0.3323652446269989,
+      "eval_overall_accuracy": 0.907703667240955,
+      "eval_overall_f1": 0.4827586206896552,
+      "eval_overall_precision": 0.5365853658536586,
+      "eval_overall_recall": 0.43874643874643876,
+      "eval_runtime": 0.374,
+      "eval_samples_per_second": 500.058,
+      "eval_steps_per_second": 8.022,
       "step": 424
     },
     {
       "epoch": 5.0,
-      "grad_norm": 1.8846955299377441,
+      "grad_norm": 1.3654524087905884,
       "learning_rate": 4.75e-05,
-      "loss": 0.3345,
+      "loss": 0.324,
       "step": 530
     },
     {
       "epoch": 5.0,
-      "eval_LOCATION_f1": 0.576923076923077,
-      "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.5421686746987951,
-      "eval_LOCATION_recall": 0.6164383561643836,
-      "eval_ORGANIZATION_f1": 0.4477611940298507,
-      "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.43478260869565216,
-      "eval_ORGANIZATION_recall": 0.46153846153846156,
-      "eval_PERSON_f1": 0.7641791044776118,
-      "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.6918918918918919,
-      "eval_PERSON_recall": 0.8533333333333334,
-      "eval_QUANTITY_f1": 0.4482758620689655,
-      "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.4482758620689655,
-      "eval_QUANTITY_recall": 0.4482758620689655,
-      "eval_TIME_f1": 0.5945945945945946,
-      "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.55,
-      "eval_TIME_recall": 0.6470588235294118,
-      "eval_loss": 0.2660435438156128,
-      "eval_overall_accuracy": 0.9276396751169087,
-      "eval_overall_f1": 0.6287978863936591,
-      "eval_overall_precision": 0.5862068965517241,
-      "eval_overall_recall": 0.6780626780626781,
-      "eval_runtime": 0.9444,
-      "eval_samples_per_second": 198.0,
-      "eval_steps_per_second": 3.176,
+      "eval_LOCATION_f1": 0.564102564102564,
+      "eval_LOCATION_number": 73,
+      "eval_LOCATION_precision": 0.5301204819277109,
+      "eval_LOCATION_recall": 0.6027397260273972,
+      "eval_ORGANIZATION_f1": 0.45255474452554745,
+      "eval_ORGANIZATION_number": 65,
+      "eval_ORGANIZATION_precision": 0.4305555555555556,
+      "eval_ORGANIZATION_recall": 0.47692307692307695,
+      "eval_PERSON_f1": 0.8263473053892216,
+      "eval_PERSON_number": 150,
+      "eval_PERSON_precision": 0.75,
+      "eval_PERSON_recall": 0.92,
+      "eval_QUANTITY_f1": 0.39999999999999997,
+      "eval_QUANTITY_number": 29,
+      "eval_QUANTITY_precision": 0.3870967741935484,
+      "eval_QUANTITY_recall": 0.41379310344827586,
+      "eval_TIME_f1": 0.6849315068493151,
+      "eval_TIME_number": 34,
+      "eval_TIME_precision": 0.6410256410256411,
+      "eval_TIME_recall": 0.7352941176470589,
+      "eval_loss": 0.26078376173973083,
+      "eval_overall_accuracy": 0.9308392813192222,
+      "eval_overall_f1": 0.6578947368421052,
+      "eval_overall_precision": 0.6112469437652812,
+      "eval_overall_recall": 0.7122507122507122,
+      "eval_runtime": 0.3725,
+      "eval_samples_per_second": 502.055,
+      "eval_steps_per_second": 8.054,
       "step": 530
     },
     {
       "epoch": 6.0,
-      "grad_norm": 2.0284557342529297,
+      "grad_norm": 1.6405872106552124,
       "learning_rate": 4.7e-05,
-      "loss": 0.2736,
+      "loss": 0.2665,
       "step": 636
     },
     {
       "epoch": 6.0,
-      "eval_LOCATION_f1": 0.6473988439306358,
+      "eval_LOCATION_f1": 0.6826347305389222,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.56,
-      "eval_LOCATION_recall": 0.7671232876712328,
-      "eval_ORGANIZATION_f1": 0.5694444444444443,
+      "eval_LOCATION_precision": 0.6063829787234043,
+      "eval_LOCATION_recall": 0.7808219178082192,
+      "eval_ORGANIZATION_f1": 0.5394736842105263,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.5189873417721519,
+      "eval_ORGANIZATION_precision": 0.47126436781609193,
       "eval_ORGANIZATION_recall": 0.6307692307692307,
-      "eval_PERSON_f1": 0.8263473053892216,
+      "eval_PERSON_f1": 0.8363636363636363,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.75,
+      "eval_PERSON_precision": 0.7666666666666667,
       "eval_PERSON_recall": 0.92,
-      "eval_QUANTITY_f1": 0.43750000000000006,
+      "eval_QUANTITY_f1": 0.4444444444444445,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.4,
+      "eval_QUANTITY_precision": 0.4117647058823529,
       "eval_QUANTITY_recall": 0.4827586206896552,
-      "eval_TIME_f1": 0.6578947368421053,
+      "eval_TIME_f1": 0.7027027027027027,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.5952380952380952,
-      "eval_TIME_recall": 0.7352941176470589,
-      "eval_loss": 0.23392952978610992,
-      "eval_overall_accuracy": 0.9355156288456805,
-      "eval_overall_f1": 0.6927939317319847,
-      "eval_overall_precision": 0.6227272727272727,
-      "eval_overall_recall": 0.7806267806267806,
-      "eval_runtime": 1.0811,
-      "eval_samples_per_second": 172.97,
-      "eval_steps_per_second": 2.775,
+      "eval_TIME_precision": 0.65,
+      "eval_TIME_recall": 0.7647058823529411,
+      "eval_loss": 0.23299816250801086,
+      "eval_overall_accuracy": 0.934531134629584,
+      "eval_overall_f1": 0.7022900763358779,
+      "eval_overall_precision": 0.6344827586206897,
+      "eval_overall_recall": 0.7863247863247863,
+      "eval_runtime": 0.3778,
+      "eval_samples_per_second": 494.969,
+      "eval_steps_per_second": 7.941,
       "step": 636
     },
     {
       "epoch": 7.0,
-      "grad_norm": 1.2625077962875366,
+      "grad_norm": 1.7495594024658203,
       "learning_rate": 4.6500000000000005e-05,
-      "loss": 0.2339,
+      "loss": 0.2275,
       "step": 742
     },
     {
       "epoch": 7.0,
-      "eval_LOCATION_f1": 0.6896551724137931,
+      "eval_LOCATION_f1": 0.7349397590361444,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.594059405940594,
-      "eval_LOCATION_recall": 0.821917808219178,
-      "eval_ORGANIZATION_f1": 0.5874125874125874,
+      "eval_LOCATION_precision": 0.6559139784946236,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.5660377358490566,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.5384615384615384,
-      "eval_ORGANIZATION_recall": 0.6461538461538462,
-      "eval_PERSON_f1": 0.871165644171779,
+      "eval_ORGANIZATION_precision": 0.4787234042553192,
+      "eval_ORGANIZATION_recall": 0.6923076923076923,
+      "eval_PERSON_f1": 0.8475609756097561,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8068181818181818,
-      "eval_PERSON_recall": 0.9466666666666667,
-      "eval_QUANTITY_f1": 0.4666666666666667,
+      "eval_PERSON_precision": 0.7808988764044944,
+      "eval_PERSON_recall": 0.9266666666666666,
+      "eval_QUANTITY_f1": 0.4444444444444445,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.45161290322580644,
+      "eval_QUANTITY_precision": 0.4117647058823529,
       "eval_QUANTITY_recall": 0.4827586206896552,
-      "eval_TIME_f1": 0.6666666666666667,
+      "eval_TIME_f1": 0.7027027027027027,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.5909090909090909,
+      "eval_TIME_precision": 0.65,
       "eval_TIME_recall": 0.7647058823529411,
-      "eval_loss": 0.19882944226264954,
-      "eval_overall_accuracy": 0.9414225941422594,
-      "eval_overall_f1": 0.7272727272727274,
-      "eval_overall_precision": 0.6604651162790698,
-      "eval_overall_recall": 0.8091168091168092,
-      "eval_runtime": 0.9366,
-      "eval_samples_per_second": 199.653,
-      "eval_steps_per_second": 3.203,
+      "eval_loss": 0.19754259288311005,
+      "eval_overall_accuracy": 0.9392074821560423,
+      "eval_overall_f1": 0.7215189873417722,
+      "eval_overall_precision": 0.6492027334851936,
+      "eval_overall_recall": 0.811965811965812,
+      "eval_runtime": 0.382,
+      "eval_samples_per_second": 489.509,
+      "eval_steps_per_second": 7.853,
       "step": 742
     },
     {
       "epoch": 8.0,
-      "grad_norm": 1.0806405544281006,
+      "grad_norm": 1.2530863285064697,
       "learning_rate": 4.600000000000001e-05,
-      "loss": 0.213,
+      "loss": 0.2049,
       "step": 848
     },
     {
       "epoch": 8.0,
-      "eval_LOCATION_f1": 0.696629213483146,
+      "eval_LOCATION_f1": 0.7283236994219653,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.5904761904761905,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.6133333333333333,
+      "eval_LOCATION_precision": 0.63,
+      "eval_LOCATION_recall": 0.863013698630137,
+      "eval_ORGANIZATION_f1": 0.5256410256410257,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.5411764705882353,
-      "eval_ORGANIZATION_recall": 0.7076923076923077,
-      "eval_PERSON_f1": 0.8695652173913044,
+      "eval_ORGANIZATION_precision": 0.45054945054945056,
+      "eval_ORGANIZATION_recall": 0.6307692307692307,
+      "eval_PERSON_f1": 0.851851851851852,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.813953488372093,
-      "eval_PERSON_recall": 0.9333333333333333,
-      "eval_QUANTITY_f1": 0.4918032786885246,
+      "eval_PERSON_precision": 0.7931034482758621,
+      "eval_PERSON_recall": 0.92,
+      "eval_QUANTITY_f1": 0.5538461538461539,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.46875,
-      "eval_QUANTITY_recall": 0.5172413793103449,
-      "eval_TIME_f1": 0.7088607594936709,
+      "eval_QUANTITY_precision": 0.5,
+      "eval_QUANTITY_recall": 0.6206896551724138,
+      "eval_TIME_f1": 0.72,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.6222222222222222,
-      "eval_TIME_recall": 0.8235294117647058,
-      "eval_loss": 0.1896568089723587,
-      "eval_overall_accuracy": 0.9406842234801871,
-      "eval_overall_f1": 0.7367088607594937,
-      "eval_overall_precision": 0.662870159453303,
-      "eval_overall_recall": 0.8290598290598291,
-      "eval_runtime": 0.9615,
-      "eval_samples_per_second": 194.494,
-      "eval_steps_per_second": 3.12,
+      "eval_TIME_precision": 0.6585365853658537,
+      "eval_TIME_recall": 0.7941176470588235,
+      "eval_loss": 0.1830800473690033,
+      "eval_overall_accuracy": 0.9419148412503077,
+      "eval_overall_f1": 0.7238335435056746,
+      "eval_overall_precision": 0.6493212669683258,
+      "eval_overall_recall": 0.8176638176638177,
+      "eval_runtime": 0.3764,
+      "eval_samples_per_second": 496.877,
+      "eval_steps_per_second": 7.971,
       "step": 848
     },
     {
       "epoch": 9.0,
-      "grad_norm": 2.574251651763916,
+      "grad_norm": 2.133810520172119,
       "learning_rate": 4.55e-05,
-      "loss": 0.1927,
+      "loss": 0.1888,
       "step": 954
     },
     {
       "epoch": 9.0,
-      "eval_LOCATION_f1": 0.7052023121387282,
+      "eval_LOCATION_f1": 0.7261904761904762,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.61,
+      "eval_LOCATION_precision": 0.6421052631578947,
       "eval_LOCATION_recall": 0.8356164383561644,
-      "eval_ORGANIZATION_f1": 0.5960264900662252,
+      "eval_ORGANIZATION_f1": 0.5822784810126582,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.5232558139534884,
-      "eval_ORGANIZATION_recall": 0.6923076923076923,
-      "eval_PERSON_f1": 0.8607594936708859,
+      "eval_ORGANIZATION_precision": 0.4946236559139785,
+      "eval_ORGANIZATION_recall": 0.7076923076923077,
+      "eval_PERSON_f1": 0.8374999999999999,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8192771084337349,
-      "eval_PERSON_recall": 0.9066666666666666,
-      "eval_QUANTITY_f1": 0.5333333333333333,
+      "eval_PERSON_precision": 0.788235294117647,
+      "eval_PERSON_recall": 0.8933333333333333,
+      "eval_QUANTITY_f1": 0.47619047619047616,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.5161290322580645,
-      "eval_QUANTITY_recall": 0.5517241379310345,
-      "eval_TIME_f1": 0.7272727272727273,
-      "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.6511627906976745,
-      "eval_TIME_recall": 0.8235294117647058,
-      "eval_loss": 0.17324766516685486,
-      "eval_overall_accuracy": 0.9419148412503077,
-      "eval_overall_f1": 0.7361647361647361,
-      "eval_overall_precision": 0.6713615023474179,
-      "eval_overall_recall": 0.8148148148148148,
-      "eval_runtime": 0.9772,
-      "eval_samples_per_second": 191.367,
-      "eval_steps_per_second": 3.07,
+      "eval_QUANTITY_precision": 0.4411764705882353,
+      "eval_QUANTITY_recall": 0.5172413793103449,
+      "eval_TIME_f1": 0.7105263157894737,
+      "eval_TIME_number": 34,
+      "eval_TIME_precision": 0.6428571428571429,
+      "eval_TIME_recall": 0.7941176470588235,
+      "eval_loss": 0.17504610121250153,
+      "eval_overall_accuracy": 0.9431454590204282,
+      "eval_overall_f1": 0.7210191082802548,
+      "eval_overall_precision": 0.652073732718894,
+      "eval_overall_recall": 0.8062678062678063,
+      "eval_runtime": 0.3734,
+      "eval_samples_per_second": 500.739,
+      "eval_steps_per_second": 8.033,
       "step": 954
     },
     {
       "epoch": 10.0,
-      "grad_norm": 1.0041288137435913,
+      "grad_norm": 0.8962897658348083,
       "learning_rate": 4.5e-05,
-      "loss": 0.1794,
+      "loss": 0.1759,
       "step": 1060
     },
     {
       "epoch": 10.0,
-      "eval_LOCATION_f1": 0.7710843373493976,
+      "eval_LOCATION_f1": 0.7643312101910829,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.6881720430107527,
-      "eval_LOCATION_recall": 0.8767123287671232,
-      "eval_ORGANIZATION_f1": 0.6206896551724138,
+      "eval_LOCATION_precision": 0.7142857142857143,
+      "eval_LOCATION_recall": 0.821917808219178,
+      "eval_ORGANIZATION_f1": 0.6086956521739131,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.5625,
-      "eval_ORGANIZATION_recall": 0.6923076923076923,
-      "eval_PERSON_f1": 0.8517350157728706,
+      "eval_ORGANIZATION_precision": 0.5104166666666666,
+      "eval_ORGANIZATION_recall": 0.7538461538461538,
+      "eval_PERSON_f1": 0.8348909657320873,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8083832335329342,
-      "eval_PERSON_recall": 0.9,
-      "eval_QUANTITY_f1": 0.523076923076923,
+      "eval_PERSON_precision": 0.783625730994152,
+      "eval_PERSON_recall": 0.8933333333333333,
+      "eval_QUANTITY_f1": 0.6060606060606061,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.4722222222222222,
-      "eval_QUANTITY_recall": 0.5862068965517241,
-      "eval_TIME_f1": 0.8493150684931507,
+      "eval_QUANTITY_precision": 0.5405405405405406,
+      "eval_QUANTITY_recall": 0.6896551724137931,
+      "eval_TIME_f1": 0.7368421052631577,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.7948717948717948,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.16186459362506866,
-      "eval_overall_accuracy": 0.9470834358848141,
-      "eval_overall_f1": 0.7624020887728459,
-      "eval_overall_precision": 0.7036144578313253,
-      "eval_overall_recall": 0.8319088319088319,
-      "eval_runtime": 0.9594,
-      "eval_samples_per_second": 194.91,
-      "eval_steps_per_second": 3.127,
+      "eval_TIME_precision": 0.6666666666666666,
+      "eval_TIME_recall": 0.8235294117647058,
+      "eval_loss": 0.16711482405662537,
+      "eval_overall_accuracy": 0.9458528181146936,
+      "eval_overall_f1": 0.7451984635083226,
+      "eval_overall_precision": 0.6767441860465117,
+      "eval_overall_recall": 0.8290598290598291,
+      "eval_runtime": 0.3754,
+      "eval_samples_per_second": 498.083,
+      "eval_steps_per_second": 7.991,
       "step": 1060
     },
     {
       "epoch": 11.0,
-      "grad_norm": 1.2041752338409424,
+      "grad_norm": 1.1401879787445068,
       "learning_rate": 4.4500000000000004e-05,
-      "loss": 0.1696,
+      "loss": 0.1691,
       "step": 1166
     },
     {
       "epoch": 11.0,
-      "eval_LOCATION_f1": 0.7619047619047619,
+      "eval_LOCATION_f1": 0.7738095238095238,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.6736842105263158,
-      "eval_LOCATION_recall": 0.8767123287671232,
-      "eval_ORGANIZATION_f1": 0.6357615894039734,
+      "eval_LOCATION_precision": 0.6842105263157895,
+      "eval_LOCATION_recall": 0.8904109589041096,
+      "eval_ORGANIZATION_f1": 0.6415094339622641,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.5581395348837209,
-      "eval_ORGANIZATION_recall": 0.7384615384615385,
-      "eval_PERSON_f1": 0.8598726114649682,
+      "eval_ORGANIZATION_precision": 0.5425531914893617,
+      "eval_ORGANIZATION_recall": 0.7846153846153846,
+      "eval_PERSON_f1": 0.8500000000000001,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.823170731707317,
-      "eval_PERSON_recall": 0.9,
-      "eval_QUANTITY_f1": 0.6461538461538462,
+      "eval_PERSON_precision": 0.8,
+      "eval_PERSON_recall": 0.9066666666666666,
+      "eval_QUANTITY_f1": 0.6363636363636365,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.5833333333333334,
+      "eval_QUANTITY_precision": 0.5675675675675675,
       "eval_QUANTITY_recall": 0.7241379310344828,
-      "eval_TIME_f1": 0.8108108108108107,
+      "eval_TIME_f1": 0.7894736842105262,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.75,
+      "eval_TIME_precision": 0.7142857142857143,
       "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.15661031007766724,
-      "eval_overall_accuracy": 0.948806300762983,
-      "eval_overall_f1": 0.7720207253886011,
-      "eval_overall_precision": 0.7078384798099763,
-      "eval_overall_recall": 0.8490028490028491,
-      "eval_runtime": 0.9517,
-      "eval_samples_per_second": 196.498,
-      "eval_steps_per_second": 3.152,
+      "eval_loss": 0.16388720273971558,
+      "eval_overall_accuracy": 0.9463450652227419,
+      "eval_overall_f1": 0.7680608365019013,
+      "eval_overall_precision": 0.6917808219178082,
+      "eval_overall_recall": 0.8632478632478633,
+      "eval_runtime": 0.3746,
+      "eval_samples_per_second": 499.243,
+      "eval_steps_per_second": 8.009,
       "step": 1166
     },
     {
       "epoch": 12.0,
-      "grad_norm": 1.5384461879730225,
+      "grad_norm": 1.234176754951477,
       "learning_rate": 4.4000000000000006e-05,
-      "loss": 0.1593,
+      "loss": 0.1569,
       "step": 1272
     },
     {
       "epoch": 12.0,
-      "eval_LOCATION_f1": 0.7804878048780488,
+      "eval_LOCATION_f1": 0.7901234567901235,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7032967032967034,
+      "eval_LOCATION_precision": 0.7191011235955056,
       "eval_LOCATION_recall": 0.8767123287671232,
-      "eval_ORGANIZATION_f1": 0.6282051282051283,
+      "eval_ORGANIZATION_f1": 0.625,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.5384615384615384,
-      "eval_ORGANIZATION_recall": 0.7538461538461538,
-      "eval_PERSON_f1": 0.8706624605678233,
+      "eval_ORGANIZATION_precision": 0.5263157894736842,
+      "eval_ORGANIZATION_recall": 0.7692307692307693,
+      "eval_PERSON_f1": 0.8385093167701863,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8263473053892215,
-      "eval_PERSON_recall": 0.92,
-      "eval_QUANTITY_f1": 0.7692307692307692,
+      "eval_PERSON_precision": 0.7848837209302325,
+      "eval_PERSON_recall": 0.9,
+      "eval_QUANTITY_f1": 0.6086956521739131,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6944444444444444,
-      "eval_QUANTITY_recall": 0.8620689655172413,
-      "eval_TIME_f1": 0.7894736842105262,
+      "eval_QUANTITY_precision": 0.525,
+      "eval_QUANTITY_recall": 0.7241379310344828,
+      "eval_TIME_f1": 0.7532467532467532,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.7142857142857143,
-      "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.15920695662498474,
-      "eval_overall_accuracy": 0.950775289195176,
-      "eval_overall_f1": 0.7866323907455014,
-      "eval_overall_precision": 0.7166276346604216,
-      "eval_overall_recall": 0.8717948717948718,
-      "eval_runtime": 0.9534,
-      "eval_samples_per_second": 196.149,
-      "eval_steps_per_second": 3.147,
+      "eval_TIME_precision": 0.6744186046511628,
+      "eval_TIME_recall": 0.8529411764705882,
+      "eval_loss": 0.16664613783359528,
+      "eval_overall_accuracy": 0.9458528181146936,
+      "eval_overall_f1": 0.7569620253164556,
+      "eval_overall_precision": 0.6810933940774487,
+      "eval_overall_recall": 0.8518518518518519,
+      "eval_runtime": 0.3735,
+      "eval_samples_per_second": 500.672,
+      "eval_steps_per_second": 8.032,
       "step": 1272
     },
     {
       "epoch": 13.0,
-      "grad_norm": 1.1916617155075073,
+      "grad_norm": 1.1105190515518188,
       "learning_rate": 4.35e-05,
-      "loss": 0.1517,
+      "loss": 0.1512,
       "step": 1378
     },
     {
       "epoch": 13.0,
-      "eval_LOCATION_f1": 0.7924528301886793,
+      "eval_LOCATION_f1": 0.7749999999999999,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7325581395348837,
-      "eval_LOCATION_recall": 0.863013698630137,
-      "eval_ORGANIZATION_f1": 0.6538461538461537,
+      "eval_LOCATION_precision": 0.7126436781609196,
+      "eval_LOCATION_recall": 0.8493150684931506,
+      "eval_ORGANIZATION_f1": 0.6357615894039734,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.5604395604395604,
-      "eval_ORGANIZATION_recall": 0.7846153846153846,
-      "eval_PERSON_f1": 0.8589341692789968,
+      "eval_ORGANIZATION_precision": 0.5581395348837209,
+      "eval_ORGANIZATION_recall": 0.7384615384615385,
+      "eval_PERSON_f1": 0.84375,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8106508875739645,
-      "eval_PERSON_recall": 0.9133333333333333,
-      "eval_QUANTITY_f1": 0.6666666666666666,
+      "eval_PERSON_precision": 0.7941176470588235,
+      "eval_PERSON_recall": 0.9,
+      "eval_QUANTITY_f1": 0.6268656716417911,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.575,
-      "eval_QUANTITY_recall": 0.7931034482758621,
-      "eval_TIME_f1": 0.8378378378378379,
+      "eval_QUANTITY_precision": 0.5526315789473685,
+      "eval_QUANTITY_recall": 0.7241379310344828,
+      "eval_TIME_f1": 0.8108108108108107,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.775,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.15383993089199066,
+      "eval_TIME_precision": 0.75,
+      "eval_TIME_recall": 0.8823529411764706,
+      "eval_loss": 0.1505931168794632,
       "eval_overall_accuracy": 0.9510214127492,
-      "eval_overall_f1": 0.785070785070785,
-      "eval_overall_precision": 0.715962441314554,
-      "eval_overall_recall": 0.8689458689458689,
-      "eval_runtime": 0.957,
-      "eval_samples_per_second": 195.394,
-      "eval_steps_per_second": 3.135,
+      "eval_overall_f1": 0.766839378238342,
+      "eval_overall_precision": 0.7030878859857482,
+      "eval_overall_recall": 0.8433048433048433,
+      "eval_runtime": 0.3701,
+      "eval_samples_per_second": 505.237,
+      "eval_steps_per_second": 8.105,
       "step": 1378
     },
     {
       "epoch": 14.0,
-      "grad_norm": 1.1742634773254395,
+      "grad_norm": 1.2889316082000732,
       "learning_rate": 4.3e-05,
-      "loss": 0.1512,
+      "loss": 0.1492,
       "step": 1484
     },
     {
       "epoch": 14.0,
-      "eval_LOCATION_f1": 0.7804878048780488,
+      "eval_LOCATION_f1": 0.7901234567901235,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7032967032967034,
+      "eval_LOCATION_precision": 0.7191011235955056,
       "eval_LOCATION_recall": 0.8767123287671232,
-      "eval_ORGANIZATION_f1": 0.6219512195121952,
+      "eval_ORGANIZATION_f1": 0.6257668711656441,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.5151515151515151,
+      "eval_ORGANIZATION_precision": 0.5204081632653061,
       "eval_ORGANIZATION_recall": 0.7846153846153846,
-      "eval_PERSON_f1": 0.8535825545171339,
+      "eval_PERSON_f1": 0.8411214953271027,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8011695906432749,
-      "eval_PERSON_recall": 0.9133333333333333,
-      "eval_QUANTITY_f1": 0.7142857142857144,
+      "eval_PERSON_precision": 0.7894736842105263,
+      "eval_PERSON_recall": 0.9,
+      "eval_QUANTITY_f1": 0.6865671641791046,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6097560975609756,
-      "eval_QUANTITY_recall": 0.8620689655172413,
-      "eval_TIME_f1": 0.8051948051948051,
+      "eval_QUANTITY_precision": 0.6052631578947368,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.7894736842105262,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.7209302325581395,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.15704582631587982,
-      "eval_overall_accuracy": 0.9500369185331036,
-      "eval_overall_f1": 0.7738693467336684,
-      "eval_overall_precision": 0.6921348314606741,
-      "eval_overall_recall": 0.8774928774928775,
-      "eval_runtime": 0.9592,
-      "eval_samples_per_second": 194.955,
-      "eval_steps_per_second": 3.128,
+      "eval_TIME_precision": 0.7142857142857143,
+      "eval_TIME_recall": 0.8823529411764706,
+      "eval_loss": 0.15868589282035828,
+      "eval_overall_accuracy": 0.9483140536549348,
+      "eval_overall_f1": 0.7680608365019013,
+      "eval_overall_precision": 0.6917808219178082,
+      "eval_overall_recall": 0.8632478632478633,
+      "eval_runtime": 0.3734,
+      "eval_samples_per_second": 500.783,
+      "eval_steps_per_second": 8.034,
       "step": 1484
     },
     {
       "epoch": 15.0,
-      "grad_norm": 1.3693042993545532,
+      "grad_norm": 1.1166456937789917,
       "learning_rate": 4.25e-05,
-      "loss": 0.1456,
+      "loss": 0.1422,
       "step": 1590
     },
     {
       "epoch": 15.0,
-      "eval_LOCATION_f1": 0.7636363636363637,
+      "eval_LOCATION_f1": 0.7852760736196318,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.6847826086956522,
-      "eval_LOCATION_recall": 0.863013698630137,
-      "eval_ORGANIZATION_f1": 0.6329113924050633,
+      "eval_LOCATION_precision": 0.7111111111111111,
+      "eval_LOCATION_recall": 0.8767123287671232,
+      "eval_ORGANIZATION_f1": 0.6134969325153374,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.5376344086021505,
+      "eval_ORGANIZATION_precision": 0.5102040816326531,
       "eval_ORGANIZATION_recall": 0.7692307692307693,
-      "eval_PERSON_f1": 0.8706624605678233,
+      "eval_PERSON_f1": 0.8500000000000001,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8263473053892215,
-      "eval_PERSON_recall": 0.92,
-      "eval_QUANTITY_f1": 0.6857142857142856,
+      "eval_PERSON_precision": 0.8,
+      "eval_PERSON_recall": 0.9066666666666666,
+      "eval_QUANTITY_f1": 0.7272727272727273,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.5853658536585366,
+      "eval_QUANTITY_precision": 0.6486486486486487,
       "eval_QUANTITY_recall": 0.8275862068965517,
-      "eval_TIME_f1": 0.794871794871795,
+      "eval_TIME_f1": 0.8051948051948051,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.7045454545454546,
+      "eval_TIME_precision": 0.7209302325581395,
       "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.15755978226661682,
-      "eval_overall_accuracy": 0.9502830420871278,
-      "eval_overall_f1": 0.7766497461928934,
-      "eval_overall_precision": 0.700228832951945,
-      "eval_overall_recall": 0.8717948717948718,
-      "eval_runtime": 0.9594,
-      "eval_samples_per_second": 194.913,
-      "eval_steps_per_second": 3.127,
+      "eval_loss": 0.15712317824363708,
+      "eval_overall_accuracy": 0.9483140536549348,
+      "eval_overall_f1": 0.7731305449936628,
+      "eval_overall_precision": 0.6963470319634704,
+      "eval_overall_recall": 0.8689458689458689,
+      "eval_runtime": 0.3753,
+      "eval_samples_per_second": 498.228,
+      "eval_steps_per_second": 7.993,
       "step": 1590
     },
     {
       "epoch": 16.0,
-      "grad_norm": 1.6605894565582275,
+      "grad_norm": 1.7718479633331299,
       "learning_rate": 4.2e-05,
       "loss": 0.1386,
       "step": 1696
     },
     {
       "epoch": 16.0,
-      "eval_LOCATION_f1": 0.7590361445783131,
+      "eval_LOCATION_f1": 0.782608695652174,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.6774193548387096,
+      "eval_LOCATION_precision": 0.7159090909090909,
       "eval_LOCATION_recall": 0.863013698630137,
-      "eval_ORGANIZATION_f1": 0.6451612903225806,
+      "eval_ORGANIZATION_f1": 0.6233766233766234,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.5555555555555556,
-      "eval_ORGANIZATION_recall": 0.7692307692307693,
-      "eval_PERSON_f1": 0.8616352201257862,
+      "eval_ORGANIZATION_precision": 0.5393258426966292,
+      "eval_ORGANIZATION_recall": 0.7384615384615385,
+      "eval_PERSON_f1": 0.8714733542319748,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8154761904761905,
-      "eval_PERSON_recall": 0.9133333333333333,
-      "eval_QUANTITY_f1": 0.676056338028169,
+      "eval_PERSON_precision": 0.8224852071005917,
+      "eval_PERSON_recall": 0.9266666666666666,
+      "eval_QUANTITY_f1": 0.7352941176470588,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.5714285714285714,
-      "eval_QUANTITY_recall": 0.8275862068965517,
-      "eval_TIME_f1": 0.794871794871795,
+      "eval_QUANTITY_precision": 0.6410256410256411,
+      "eval_QUANTITY_recall": 0.8620689655172413,
+      "eval_TIME_f1": 0.7792207792207793,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.7045454545454546,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.1546044647693634,
-      "eval_overall_accuracy": 0.9510214127492,
-      "eval_overall_f1": 0.7741116751269036,
-      "eval_overall_precision": 0.6979405034324943,
+      "eval_TIME_precision": 0.6976744186046512,
+      "eval_TIME_recall": 0.8823529411764706,
+      "eval_loss": 0.1536765843629837,
+      "eval_overall_accuracy": 0.9502830420871278,
+      "eval_overall_f1": 0.7830551989730424,
+      "eval_overall_precision": 0.7126168224299065,
       "eval_overall_recall": 0.8689458689458689,
-      "eval_runtime": 0.9626,
-      "eval_samples_per_second": 194.274,
-      "eval_steps_per_second": 3.117,
+      "eval_runtime": 0.3723,
+      "eval_samples_per_second": 502.251,
+      "eval_steps_per_second": 8.058,
       "step": 1696
     },
     {
       "epoch": 17.0,
-      "grad_norm": 0.814884603023529,
+      "grad_norm": 2.4890553951263428,
       "learning_rate": 4.15e-05,
-      "loss": 0.1366,
+      "loss": 0.1351,
       "step": 1802
     },
     {
@@ -645,232 +645,232 @@
       "eval_LOCATION_number": 73,
       "eval_LOCATION_precision": 0.7411764705882353,
       "eval_LOCATION_recall": 0.863013698630137,
-      "eval_ORGANIZATION_f1": 0.6496815286624205,
+      "eval_ORGANIZATION_f1": 0.6219512195121952,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.5543478260869565,
+      "eval_ORGANIZATION_precision": 0.5151515151515151,
       "eval_ORGANIZATION_recall": 0.7846153846153846,
-      "eval_PERSON_f1": 0.8753993610223643,
+      "eval_PERSON_f1": 0.8698412698412699,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8404907975460123,
+      "eval_PERSON_precision": 0.8303030303030303,
       "eval_PERSON_recall": 0.9133333333333333,
-      "eval_QUANTITY_f1": 0.6666666666666666,
+      "eval_QUANTITY_f1": 0.7076923076923076,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.575,
+      "eval_QUANTITY_precision": 0.6388888888888888,
       "eval_QUANTITY_recall": 0.7931034482758621,
-      "eval_TIME_f1": 0.8051948051948051,
+      "eval_TIME_f1": 0.7894736842105262,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.7209302325581395,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.14499862492084503,
-      "eval_overall_accuracy": 0.9539748953974896,
-      "eval_overall_f1": 0.7881136950904393,
-      "eval_overall_precision": 0.7210401891252955,
-      "eval_overall_recall": 0.8689458689458689,
-      "eval_runtime": 0.9628,
-      "eval_samples_per_second": 194.217,
-      "eval_steps_per_second": 3.116,
+      "eval_TIME_precision": 0.7142857142857143,
+      "eval_TIME_recall": 0.8823529411764706,
+      "eval_loss": 0.15198422968387604,
+      "eval_overall_accuracy": 0.9500369185331036,
+      "eval_overall_f1": 0.7814910025706941,
+      "eval_overall_precision": 0.711943793911007,
+      "eval_overall_recall": 0.8660968660968661,
+      "eval_runtime": 0.3738,
+      "eval_samples_per_second": 500.243,
+      "eval_steps_per_second": 8.025,
       "step": 1802
     },
     {
       "epoch": 18.0,
-      "grad_norm": 1.8611637353897095,
+      "grad_norm": 3.517162561416626,
       "learning_rate": 4.1e-05,
-      "loss": 0.1338,
+      "loss": 0.1397,
       "step": 1908
     },
     {
       "epoch": 18.0,
-      "eval_LOCATION_f1": 0.7924528301886793,
+      "eval_LOCATION_f1": 0.7848101265822783,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7325581395348837,
-      "eval_LOCATION_recall": 0.863013698630137,
-      "eval_ORGANIZATION_f1": 0.6578947368421053,
+      "eval_LOCATION_precision": 0.7294117647058823,
+      "eval_LOCATION_recall": 0.8493150684931506,
+      "eval_ORGANIZATION_f1": 0.6,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.5747126436781609,
-      "eval_ORGANIZATION_recall": 0.7692307692307693,
+      "eval_ORGANIZATION_precision": 0.5052631578947369,
+      "eval_ORGANIZATION_recall": 0.7384615384615385,
       "eval_PERSON_f1": 0.8698412698412699,
       "eval_PERSON_number": 150,
       "eval_PERSON_precision": 0.8303030303030303,
       "eval_PERSON_recall": 0.9133333333333333,
-      "eval_QUANTITY_f1": 0.6567164179104478,
+      "eval_QUANTITY_f1": 0.7575757575757576,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.5789473684210527,
-      "eval_QUANTITY_recall": 0.7586206896551724,
-      "eval_TIME_f1": 0.8421052631578947,
+      "eval_QUANTITY_precision": 0.6756756756756757,
+      "eval_QUANTITY_recall": 0.8620689655172413,
+      "eval_TIME_f1": 0.8157894736842106,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.7619047619047619,
-      "eval_TIME_recall": 0.9411764705882353,
-      "eval_loss": 0.14045144617557526,
-      "eval_overall_accuracy": 0.9552055131676102,
-      "eval_overall_f1": 0.7906371911573472,
-      "eval_overall_precision": 0.7272727272727273,
-      "eval_overall_recall": 0.8660968660968661,
-      "eval_runtime": 1.0246,
-      "eval_samples_per_second": 182.513,
-      "eval_steps_per_second": 2.928,
+      "eval_TIME_precision": 0.7380952380952381,
+      "eval_TIME_recall": 0.9117647058823529,
+      "eval_loss": 0.1488378345966339,
+      "eval_overall_accuracy": 0.9500369185331036,
+      "eval_overall_f1": 0.7819354838709679,
+      "eval_overall_precision": 0.714622641509434,
+      "eval_overall_recall": 0.8632478632478633,
+      "eval_runtime": 0.375,
+      "eval_samples_per_second": 498.682,
+      "eval_steps_per_second": 8.0,
       "step": 1908
     },
     {
       "epoch": 19.0,
-      "grad_norm": 1.1466857194900513,
+      "grad_norm": 0.9577983021736145,
       "learning_rate": 4.05e-05,
-      "loss": 0.1282,
+      "loss": 0.1332,
       "step": 2014
     },
     {
       "epoch": 19.0,
-      "eval_LOCATION_f1": 0.7974683544303799,
+      "eval_LOCATION_f1": 0.802547770700637,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7411764705882353,
+      "eval_LOCATION_precision": 0.75,
       "eval_LOCATION_recall": 0.863013698630137,
-      "eval_ORGANIZATION_f1": 0.6463414634146342,
+      "eval_ORGANIZATION_f1": 0.6296296296296297,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.5353535353535354,
-      "eval_ORGANIZATION_recall": 0.8153846153846154,
-      "eval_PERSON_f1": 0.8553459119496856,
+      "eval_ORGANIZATION_precision": 0.5257731958762887,
+      "eval_ORGANIZATION_recall": 0.7846153846153846,
+      "eval_PERSON_f1": 0.8670886075949367,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8095238095238095,
-      "eval_PERSON_recall": 0.9066666666666666,
-      "eval_QUANTITY_f1": 0.7058823529411765,
+      "eval_PERSON_precision": 0.8253012048192772,
+      "eval_PERSON_recall": 0.9133333333333333,
+      "eval_QUANTITY_f1": 0.6461538461538462,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6153846153846154,
-      "eval_QUANTITY_recall": 0.8275862068965517,
-      "eval_TIME_f1": 0.8051948051948051,
+      "eval_QUANTITY_precision": 0.5833333333333334,
+      "eval_QUANTITY_recall": 0.7241379310344828,
+      "eval_TIME_f1": 0.8266666666666665,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.7209302325581395,
+      "eval_TIME_precision": 0.7560975609756098,
       "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.1484779715538025,
-      "eval_overall_accuracy": 0.9524981540733448,
-      "eval_overall_f1": 0.7821656050955413,
-      "eval_overall_precision": 0.7073732718894009,
-      "eval_overall_recall": 0.8746438746438746,
-      "eval_runtime": 0.9587,
-      "eval_samples_per_second": 195.054,
-      "eval_steps_per_second": 3.129,
+      "eval_loss": 0.14682170748710632,
+      "eval_overall_accuracy": 0.9512675363032242,
+      "eval_overall_f1": 0.7819354838709679,
+      "eval_overall_precision": 0.714622641509434,
+      "eval_overall_recall": 0.8632478632478633,
+      "eval_runtime": 0.3719,
+      "eval_samples_per_second": 502.788,
+      "eval_steps_per_second": 8.066,
       "step": 2014
     },
     {
       "epoch": 20.0,
-      "grad_norm": 1.398795247077942,
+      "grad_norm": 1.0945817232131958,
       "learning_rate": 4e-05,
-      "loss": 0.1253,
+      "loss": 0.1258,
       "step": 2120
     },
     {
       "epoch": 20.0,
-      "eval_LOCATION_f1": 0.802547770700637,
+      "eval_LOCATION_f1": 0.7974683544303799,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.75,
+      "eval_LOCATION_precision": 0.7411764705882353,
       "eval_LOCATION_recall": 0.863013698630137,
-      "eval_ORGANIZATION_f1": 0.6622516556291391,
+      "eval_ORGANIZATION_f1": 0.6216216216216216,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.5813953488372093,
-      "eval_ORGANIZATION_recall": 0.7692307692307693,
-      "eval_PERSON_f1": 0.8726114649681529,
+      "eval_ORGANIZATION_precision": 0.5542168674698795,
+      "eval_ORGANIZATION_recall": 0.7076923076923077,
+      "eval_PERSON_f1": 0.8670886075949367,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8353658536585366,
+      "eval_PERSON_precision": 0.8253012048192772,
       "eval_PERSON_recall": 0.9133333333333333,
-      "eval_QUANTITY_f1": 0.7246376811594203,
+      "eval_QUANTITY_f1": 0.7272727272727273,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.625,
-      "eval_QUANTITY_recall": 0.8620689655172413,
-      "eval_TIME_f1": 0.8157894736842106,
+      "eval_QUANTITY_precision": 0.6486486486486487,
+      "eval_QUANTITY_recall": 0.8275862068965517,
+      "eval_TIME_f1": 0.8266666666666665,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.7380952380952381,
+      "eval_TIME_precision": 0.7560975609756098,
       "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.13943925499916077,
-      "eval_overall_accuracy": 0.9556977602756583,
-      "eval_overall_f1": 0.7979139504563233,
-      "eval_overall_precision": 0.7355769230769231,
-      "eval_overall_recall": 0.8717948717948718,
-      "eval_runtime": 0.9521,
-      "eval_samples_per_second": 196.41,
-      "eval_steps_per_second": 3.151,
+      "eval_loss": 0.14338740706443787,
+      "eval_overall_accuracy": 0.9527442776273689,
+      "eval_overall_f1": 0.7889908256880734,
+      "eval_overall_precision": 0.7305825242718447,
+      "eval_overall_recall": 0.8575498575498576,
+      "eval_runtime": 0.3718,
+      "eval_samples_per_second": 502.95,
+      "eval_steps_per_second": 8.069,
       "step": 2120
     },
     {
       "epoch": 21.0,
-      "grad_norm": 1.8820486068725586,
+      "grad_norm": 1.6282328367233276,
       "learning_rate": 3.9500000000000005e-05,
-      "loss": 0.1245,
+      "loss": 0.1231,
       "step": 2226
     },
     {
       "epoch": 21.0,
-      "eval_LOCATION_f1": 0.8129032258064516,
+      "eval_LOCATION_f1": 0.7848101265822783,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7682926829268293,
-      "eval_LOCATION_recall": 0.863013698630137,
-      "eval_ORGANIZATION_f1": 0.6578947368421053,
+      "eval_LOCATION_precision": 0.7294117647058823,
+      "eval_LOCATION_recall": 0.8493150684931506,
+      "eval_ORGANIZATION_f1": 0.625,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.5747126436781609,
-      "eval_ORGANIZATION_recall": 0.7692307692307693,
-      "eval_PERSON_f1": 0.8726114649681529,
+      "eval_ORGANIZATION_precision": 0.569620253164557,
+      "eval_ORGANIZATION_recall": 0.6923076923076923,
+      "eval_PERSON_f1": 0.8643533123028392,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8353658536585366,
+      "eval_PERSON_precision": 0.8203592814371258,
       "eval_PERSON_recall": 0.9133333333333333,
-      "eval_QUANTITY_f1": 0.7384615384615385,
+      "eval_QUANTITY_f1": 0.7076923076923076,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6666666666666666,
-      "eval_QUANTITY_recall": 0.8275862068965517,
-      "eval_TIME_f1": 0.8533333333333334,
+      "eval_QUANTITY_precision": 0.6388888888888888,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8266666666666665,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.7804878048780488,
-      "eval_TIME_recall": 0.9411764705882353,
-      "eval_loss": 0.13452573120594025,
-      "eval_overall_accuracy": 0.9586512429239479,
-      "eval_overall_f1": 0.8042049934296978,
-      "eval_overall_precision": 0.7463414634146341,
-      "eval_overall_recall": 0.8717948717948718,
-      "eval_runtime": 0.9556,
-      "eval_samples_per_second": 195.683,
-      "eval_steps_per_second": 3.139,
+      "eval_TIME_precision": 0.7560975609756098,
+      "eval_TIME_recall": 0.9117647058823529,
+      "eval_loss": 0.1407139003276825,
+      "eval_overall_accuracy": 0.9544671425055378,
+      "eval_overall_f1": 0.785243741765481,
+      "eval_overall_precision": 0.7303921568627451,
+      "eval_overall_recall": 0.8490028490028491,
+      "eval_runtime": 0.3796,
+      "eval_samples_per_second": 492.652,
+      "eval_steps_per_second": 7.904,
       "step": 2226
     },
     {
       "epoch": 22.0,
-      "grad_norm": 1.1895145177841187,
+      "grad_norm": 1.690437912940979,
       "learning_rate": 3.9000000000000006e-05,
-      "loss": 0.12,
+      "loss": 0.1218,
       "step": 2332
     },
     {
       "epoch": 22.0,
-      "eval_LOCATION_f1": 0.779874213836478,
+      "eval_LOCATION_f1": 0.759493670886076,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7209302325581395,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.6583850931677019,
+      "eval_LOCATION_precision": 0.7058823529411765,
+      "eval_LOCATION_recall": 0.821917808219178,
+      "eval_ORGANIZATION_f1": 0.6144578313253013,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.5520833333333334,
-      "eval_ORGANIZATION_recall": 0.8153846153846154,
-      "eval_PERSON_f1": 0.8714733542319748,
+      "eval_ORGANIZATION_precision": 0.504950495049505,
+      "eval_ORGANIZATION_recall": 0.7846153846153846,
+      "eval_PERSON_f1": 0.85625,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8224852071005917,
-      "eval_PERSON_recall": 0.9266666666666666,
-      "eval_QUANTITY_f1": 0.7246376811594203,
+      "eval_PERSON_precision": 0.8058823529411765,
+      "eval_PERSON_recall": 0.9133333333333333,
+      "eval_QUANTITY_f1": 0.7076923076923076,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.625,
-      "eval_QUANTITY_recall": 0.8620689655172413,
-      "eval_TIME_f1": 0.7894736842105262,
+      "eval_QUANTITY_precision": 0.6388888888888888,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8533333333333334,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.7142857142857143,
-      "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.150462806224823,
-      "eval_overall_accuracy": 0.9522520305193207,
-      "eval_overall_f1": 0.788265306122449,
-      "eval_overall_precision": 0.7136258660508084,
-      "eval_overall_recall": 0.8803418803418803,
-      "eval_runtime": 0.9605,
-      "eval_samples_per_second": 194.682,
-      "eval_steps_per_second": 3.123,
+      "eval_TIME_precision": 0.7804878048780488,
+      "eval_TIME_recall": 0.9411764705882353,
+      "eval_loss": 0.14947740733623505,
+      "eval_overall_accuracy": 0.9500369185331036,
+      "eval_overall_f1": 0.7729591836734694,
+      "eval_overall_precision": 0.6997690531177829,
+      "eval_overall_recall": 0.8632478632478633,
+      "eval_runtime": 0.3754,
+      "eval_samples_per_second": 498.1,
+      "eval_steps_per_second": 7.991,
       "step": 2332
     },
     {
       "epoch": 23.0,
-      "grad_norm": 2.248535633087158,
+      "grad_norm": 1.4524682760238647,
       "learning_rate": 3.85e-05,
-      "loss": 0.1191,
+      "loss": 0.1189,
       "step": 2438
     },
     {
@@ -879,400 +879,400 @@
       "eval_LOCATION_number": 73,
       "eval_LOCATION_precision": 0.7590361445783133,
       "eval_LOCATION_recall": 0.863013698630137,
-      "eval_ORGANIZATION_f1": 0.6538461538461537,
+      "eval_ORGANIZATION_f1": 0.6451612903225806,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.5604395604395604,
-      "eval_ORGANIZATION_recall": 0.7846153846153846,
-      "eval_PERSON_f1": 0.8753993610223643,
+      "eval_ORGANIZATION_precision": 0.5555555555555556,
+      "eval_ORGANIZATION_recall": 0.7692307692307693,
+      "eval_PERSON_f1": 0.8726114649681529,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8404907975460123,
+      "eval_PERSON_precision": 0.8353658536585366,
       "eval_PERSON_recall": 0.9133333333333333,
-      "eval_QUANTITY_f1": 0.7246376811594203,
+      "eval_QUANTITY_f1": 0.696969696969697,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.625,
-      "eval_QUANTITY_recall": 0.8620689655172413,
-      "eval_TIME_f1": 0.8533333333333334,
+      "eval_QUANTITY_precision": 0.6216216216216216,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8800000000000001,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.7804878048780488,
-      "eval_TIME_recall": 0.9411764705882353,
-      "eval_loss": 0.13763977587223053,
+      "eval_TIME_precision": 0.8048780487804879,
+      "eval_TIME_recall": 0.9705882352941176,
+      "eval_loss": 0.13879932463169098,
       "eval_overall_accuracy": 0.9561900073837066,
-      "eval_overall_f1": 0.8010403120936281,
-      "eval_overall_precision": 0.7368421052631579,
-      "eval_overall_recall": 0.8774928774928775,
-      "eval_runtime": 0.9781,
-      "eval_samples_per_second": 191.185,
-      "eval_steps_per_second": 3.067,
+      "eval_overall_f1": 0.7989556135770235,
+      "eval_overall_precision": 0.7373493975903614,
+      "eval_overall_recall": 0.8717948717948718,
+      "eval_runtime": 0.3734,
+      "eval_samples_per_second": 500.853,
+      "eval_steps_per_second": 8.035,
       "step": 2438
     },
     {
       "epoch": 24.0,
-      "grad_norm": 1.4806087017059326,
+      "grad_norm": 1.4798812866210938,
       "learning_rate": 3.8e-05,
-      "loss": 0.1188,
+      "loss": 0.1176,
       "step": 2544
     },
     {
       "epoch": 24.0,
-      "eval_LOCATION_f1": 0.7924528301886793,
+      "eval_LOCATION_f1": 0.7848101265822783,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7325581395348837,
-      "eval_LOCATION_recall": 0.863013698630137,
-      "eval_ORGANIZATION_f1": 0.6753246753246754,
+      "eval_LOCATION_precision": 0.7294117647058823,
+      "eval_LOCATION_recall": 0.8493150684931506,
+      "eval_ORGANIZATION_f1": 0.6459627329192547,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.5842696629213483,
+      "eval_ORGANIZATION_precision": 0.5416666666666666,
       "eval_ORGANIZATION_recall": 0.8,
-      "eval_PERSON_f1": 0.8698412698412699,
+      "eval_PERSON_f1": 0.8643533123028392,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8303030303030303,
+      "eval_PERSON_precision": 0.8203592814371258,
       "eval_PERSON_recall": 0.9133333333333333,
-      "eval_QUANTITY_f1": 0.6376811594202898,
+      "eval_QUANTITY_f1": 0.7058823529411765,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.55,
-      "eval_QUANTITY_recall": 0.7586206896551724,
-      "eval_TIME_f1": 0.8533333333333334,
+      "eval_QUANTITY_precision": 0.6153846153846154,
+      "eval_QUANTITY_recall": 0.8275862068965517,
+      "eval_TIME_f1": 0.7894736842105262,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.7804878048780488,
-      "eval_TIME_recall": 0.9411764705882353,
-      "eval_loss": 0.14500263333320618,
-      "eval_overall_accuracy": 0.9559438838296825,
-      "eval_overall_f1": 0.7927461139896372,
-      "eval_overall_precision": 0.7268408551068883,
-      "eval_overall_recall": 0.8717948717948718,
-      "eval_runtime": 0.9648,
-      "eval_samples_per_second": 193.813,
-      "eval_steps_per_second": 3.109,
+      "eval_TIME_precision": 0.7142857142857143,
+      "eval_TIME_recall": 0.8823529411764706,
+      "eval_loss": 0.14802934229373932,
+      "eval_overall_accuracy": 0.9522520305193207,
+      "eval_overall_f1": 0.782051282051282,
+      "eval_overall_precision": 0.710955710955711,
+      "eval_overall_recall": 0.8689458689458689,
+      "eval_runtime": 0.373,
+      "eval_samples_per_second": 501.394,
+      "eval_steps_per_second": 8.044,
       "step": 2544
     },
     {
       "epoch": 25.0,
-      "grad_norm": 1.1676396131515503,
+      "grad_norm": 1.3379883766174316,
       "learning_rate": 3.7500000000000003e-05,
-      "loss": 0.1158,
+      "loss": 0.115,
       "step": 2650
     },
     {
       "epoch": 25.0,
-      "eval_LOCATION_f1": 0.7848101265822783,
+      "eval_LOCATION_f1": 0.7777777777777778,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7294117647058823,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.6666666666666666,
+      "eval_LOCATION_precision": 0.7078651685393258,
+      "eval_LOCATION_recall": 0.863013698630137,
+      "eval_ORGANIZATION_f1": 0.6335403726708075,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.5714285714285714,
-      "eval_ORGANIZATION_recall": 0.8,
-      "eval_PERSON_f1": 0.8734177215189873,
+      "eval_ORGANIZATION_precision": 0.53125,
+      "eval_ORGANIZATION_recall": 0.7846153846153846,
+      "eval_PERSON_f1": 0.8698412698412699,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8313253012048193,
-      "eval_PERSON_recall": 0.92,
-      "eval_QUANTITY_f1": 0.6857142857142856,
+      "eval_PERSON_precision": 0.8303030303030303,
+      "eval_PERSON_recall": 0.9133333333333333,
+      "eval_QUANTITY_f1": 0.7076923076923076,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.5853658536585366,
-      "eval_QUANTITY_recall": 0.8275862068965517,
-      "eval_TIME_f1": 0.7894736842105262,
+      "eval_QUANTITY_precision": 0.6388888888888888,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8266666666666665,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.7142857142857143,
-      "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.14524681866168976,
-      "eval_overall_accuracy": 0.9544671425055378,
-      "eval_overall_f1": 0.788659793814433,
-      "eval_overall_precision": 0.72,
-      "eval_overall_recall": 0.8717948717948718,
-      "eval_runtime": 0.9488,
-      "eval_samples_per_second": 197.084,
-      "eval_steps_per_second": 3.162,
+      "eval_TIME_precision": 0.7560975609756098,
+      "eval_TIME_recall": 0.9117647058823529,
+      "eval_loss": 0.14569295942783356,
+      "eval_overall_accuracy": 0.9534826482894413,
+      "eval_overall_f1": 0.7840616966580978,
+      "eval_overall_precision": 0.7142857142857143,
+      "eval_overall_recall": 0.8689458689458689,
+      "eval_runtime": 0.3725,
+      "eval_samples_per_second": 502.079,
+      "eval_steps_per_second": 8.055,
       "step": 2650
     },
     {
       "epoch": 26.0,
-      "grad_norm": 2.0401439666748047,
+      "grad_norm": 2.1995344161987305,
       "learning_rate": 3.7e-05,
-      "loss": 0.1162,
+      "loss": 0.1175,
       "step": 2756
     },
     {
       "epoch": 26.0,
-      "eval_LOCATION_f1": 0.7848101265822783,
+      "eval_LOCATION_f1": 0.7973856209150326,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7294117647058823,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.675496688741722,
+      "eval_LOCATION_precision": 0.7625,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.6308724832214766,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.5930232558139535,
-      "eval_ORGANIZATION_recall": 0.7846153846153846,
-      "eval_PERSON_f1": 0.8810289389067525,
+      "eval_ORGANIZATION_precision": 0.5595238095238095,
+      "eval_ORGANIZATION_recall": 0.7230769230769231,
+      "eval_PERSON_f1": 0.879746835443038,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8509316770186336,
-      "eval_PERSON_recall": 0.9133333333333333,
+      "eval_PERSON_precision": 0.8373493975903614,
+      "eval_PERSON_recall": 0.9266666666666666,
       "eval_QUANTITY_f1": 0.7058823529411765,
       "eval_QUANTITY_number": 29,
       "eval_QUANTITY_precision": 0.6153846153846154,
       "eval_QUANTITY_recall": 0.8275862068965517,
-      "eval_TIME_f1": 0.8108108108108107,
+      "eval_TIME_f1": 0.8266666666666665,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.75,
-      "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.13445432484149933,
-      "eval_overall_accuracy": 0.9581589958158996,
-      "eval_overall_f1": 0.7979002624671917,
-      "eval_overall_precision": 0.7396593673965937,
-      "eval_overall_recall": 0.8660968660968661,
-      "eval_runtime": 0.9577,
-      "eval_samples_per_second": 195.269,
-      "eval_steps_per_second": 3.133,
+      "eval_TIME_precision": 0.7560975609756098,
+      "eval_TIME_recall": 0.9117647058823529,
+      "eval_loss": 0.14247295260429382,
+      "eval_overall_accuracy": 0.9554516367216342,
+      "eval_overall_f1": 0.7936925098554534,
+      "eval_overall_precision": 0.7365853658536585,
+      "eval_overall_recall": 0.8603988603988604,
+      "eval_runtime": 0.3735,
+      "eval_samples_per_second": 500.637,
+      "eval_steps_per_second": 8.032,
       "step": 2756
     },
     {
       "epoch": 27.0,
-      "grad_norm": 0.6006813645362854,
+      "grad_norm": 1.194440484046936,
       "learning_rate": 3.65e-05,
-      "loss": 0.1118,
+      "loss": 0.1099,
       "step": 2862
     },
     {
       "epoch": 27.0,
-      "eval_LOCATION_f1": 0.8181818181818182,
+      "eval_LOCATION_f1": 0.7820512820512822,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7777777777777778,
-      "eval_LOCATION_recall": 0.863013698630137,
-      "eval_ORGANIZATION_f1": 0.6666666666666667,
+      "eval_LOCATION_precision": 0.7349397590361446,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.6883116883116883,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.5795454545454546,
-      "eval_ORGANIZATION_recall": 0.7846153846153846,
-      "eval_PERSON_f1": 0.8846153846153846,
+      "eval_ORGANIZATION_precision": 0.5955056179775281,
+      "eval_ORGANIZATION_recall": 0.8153846153846154,
+      "eval_PERSON_f1": 0.8825396825396825,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8518518518518519,
-      "eval_PERSON_recall": 0.92,
-      "eval_QUANTITY_f1": 0.746268656716418,
+      "eval_PERSON_precision": 0.8424242424242424,
+      "eval_PERSON_recall": 0.9266666666666666,
+      "eval_QUANTITY_f1": 0.6666666666666666,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6578947368421053,
-      "eval_QUANTITY_recall": 0.8620689655172413,
-      "eval_TIME_f1": 0.8157894736842106,
+      "eval_QUANTITY_precision": 0.575,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8311688311688312,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.7380952380952381,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.131368950009346,
-      "eval_overall_accuracy": 0.9591434900319961,
-      "eval_overall_f1": 0.8083989501312336,
-      "eval_overall_precision": 0.7493917274939172,
+      "eval_TIME_precision": 0.7441860465116279,
+      "eval_TIME_recall": 0.9411764705882353,
+      "eval_loss": 0.14262671768665314,
+      "eval_overall_accuracy": 0.9556977602756583,
+      "eval_overall_f1": 0.7989623865110247,
+      "eval_overall_precision": 0.7333333333333333,
       "eval_overall_recall": 0.8774928774928775,
-      "eval_runtime": 0.9543,
-      "eval_samples_per_second": 195.947,
-      "eval_steps_per_second": 3.144,
+      "eval_runtime": 0.3711,
+      "eval_samples_per_second": 503.938,
+      "eval_steps_per_second": 8.085,
       "step": 2862
     },
     {
       "epoch": 28.0,
-      "grad_norm": 1.7794803380966187,
+      "grad_norm": 1.36868155002594,
       "learning_rate": 3.6e-05,
-      "loss": 0.1128,
+      "loss": 0.1108,
       "step": 2968
     },
     {
       "epoch": 28.0,
-      "eval_LOCATION_f1": 0.7770700636942677,
+      "eval_LOCATION_f1": 0.7999999999999999,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7261904761904762,
-      "eval_LOCATION_recall": 0.8356164383561644,
-      "eval_ORGANIZATION_f1": 0.6666666666666667,
+      "eval_LOCATION_precision": 0.7560975609756098,
+      "eval_LOCATION_recall": 0.8493150684931506,
+      "eval_ORGANIZATION_f1": 0.6842105263157896,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.5882352941176471,
-      "eval_ORGANIZATION_recall": 0.7692307692307693,
-      "eval_PERSON_f1": 0.8753993610223643,
+      "eval_ORGANIZATION_precision": 0.5977011494252874,
+      "eval_ORGANIZATION_recall": 0.8,
+      "eval_PERSON_f1": 0.8706624605678233,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8404907975460123,
-      "eval_PERSON_recall": 0.9133333333333333,
-      "eval_QUANTITY_f1": 0.716417910447761,
+      "eval_PERSON_precision": 0.8263473053892215,
+      "eval_PERSON_recall": 0.92,
+      "eval_QUANTITY_f1": 0.6666666666666667,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.631578947368421,
-      "eval_QUANTITY_recall": 0.8275862068965517,
-      "eval_TIME_f1": 0.8493150684931507,
+      "eval_QUANTITY_precision": 0.5945945945945946,
+      "eval_QUANTITY_recall": 0.7586206896551724,
+      "eval_TIME_f1": 0.8266666666666665,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.7948717948717948,
+      "eval_TIME_precision": 0.7560975609756098,
       "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.13416561484336853,
-      "eval_overall_accuracy": 0.9576667487078513,
-      "eval_overall_f1": 0.7973684210526317,
-      "eval_overall_precision": 0.7408312958435208,
-      "eval_overall_recall": 0.8632478632478633,
-      "eval_runtime": 0.953,
-      "eval_samples_per_second": 196.219,
-      "eval_steps_per_second": 3.148,
+      "eval_loss": 0.1398727297782898,
+      "eval_overall_accuracy": 0.9554516367216342,
+      "eval_overall_f1": 0.7973856209150327,
+      "eval_overall_precision": 0.7367149758454107,
+      "eval_overall_recall": 0.8689458689458689,
+      "eval_runtime": 0.3749,
+      "eval_samples_per_second": 498.749,
+      "eval_steps_per_second": 8.001,
       "step": 2968
     },
     {
       "epoch": 29.0,
-      "grad_norm": 1.108047366142273,
+      "grad_norm": 0.8069501519203186,
       "learning_rate": 3.55e-05,
-      "loss": 0.1089,
+      "loss": 0.1071,
       "step": 3074
     },
     {
       "epoch": 29.0,
-      "eval_LOCATION_f1": 0.7848101265822783,
+      "eval_LOCATION_f1": 0.782608695652174,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7294117647058823,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.6538461538461537,
+      "eval_LOCATION_precision": 0.7159090909090909,
+      "eval_LOCATION_recall": 0.863013698630137,
+      "eval_ORGANIZATION_f1": 0.6506024096385542,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.5604395604395604,
-      "eval_ORGANIZATION_recall": 0.7846153846153846,
-      "eval_PERSON_f1": 0.8761904761904761,
+      "eval_ORGANIZATION_precision": 0.5346534653465347,
+      "eval_ORGANIZATION_recall": 0.8307692307692308,
+      "eval_PERSON_f1": 0.877742946708464,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8363636363636363,
-      "eval_PERSON_recall": 0.92,
-      "eval_QUANTITY_f1": 0.7384615384615385,
+      "eval_PERSON_precision": 0.8284023668639053,
+      "eval_PERSON_recall": 0.9333333333333333,
+      "eval_QUANTITY_f1": 0.716417910447761,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6666666666666666,
+      "eval_QUANTITY_precision": 0.631578947368421,
       "eval_QUANTITY_recall": 0.8275862068965517,
-      "eval_TIME_f1": 0.8533333333333334,
+      "eval_TIME_f1": 0.8421052631578947,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.7804878048780488,
+      "eval_TIME_precision": 0.7619047619047619,
       "eval_TIME_recall": 0.9411764705882353,
-      "eval_loss": 0.13507266342639923,
-      "eval_overall_accuracy": 0.9584051193699237,
-      "eval_overall_f1": 0.7984395318595578,
-      "eval_overall_precision": 0.7344497607655502,
-      "eval_overall_recall": 0.8746438746438746,
-      "eval_runtime": 0.9682,
-      "eval_samples_per_second": 193.134,
-      "eval_steps_per_second": 3.098,
+      "eval_loss": 0.14704711735248566,
+      "eval_overall_accuracy": 0.9512675363032242,
+      "eval_overall_f1": 0.7934093789607097,
+      "eval_overall_precision": 0.7146118721461188,
+      "eval_overall_recall": 0.8917378917378918,
+      "eval_runtime": 0.3733,
+      "eval_samples_per_second": 500.911,
+      "eval_steps_per_second": 8.036,
       "step": 3074
     },
     {
       "epoch": 30.0,
-      "grad_norm": 1.1064585447311401,
+      "grad_norm": 1.9499945640563965,
       "learning_rate": 3.5e-05,
-      "loss": 0.1054,
+      "loss": 0.1075,
       "step": 3180
     },
     {
       "epoch": 30.0,
-      "eval_LOCATION_f1": 0.8076923076923077,
+      "eval_LOCATION_f1": 0.7999999999999999,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7590361445783133,
-      "eval_LOCATION_recall": 0.863013698630137,
-      "eval_ORGANIZATION_f1": 0.6756756756756758,
+      "eval_LOCATION_precision": 0.7560975609756098,
+      "eval_LOCATION_recall": 0.8493150684931506,
+      "eval_ORGANIZATION_f1": 0.6714285714285715,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6024096385542169,
-      "eval_ORGANIZATION_recall": 0.7692307692307693,
-      "eval_PERSON_f1": 0.8726114649681529,
+      "eval_ORGANIZATION_precision": 0.6266666666666667,
+      "eval_ORGANIZATION_recall": 0.7230769230769231,
+      "eval_PERSON_f1": 0.8616352201257862,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8353658536585366,
+      "eval_PERSON_precision": 0.8154761904761905,
       "eval_PERSON_recall": 0.9133333333333333,
-      "eval_QUANTITY_f1": 0.7272727272727273,
+      "eval_QUANTITY_f1": 0.6461538461538462,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6486486486486487,
-      "eval_QUANTITY_recall": 0.8275862068965517,
-      "eval_TIME_f1": 0.861111111111111,
+      "eval_QUANTITY_precision": 0.5833333333333334,
+      "eval_QUANTITY_recall": 0.7241379310344828,
+      "eval_TIME_f1": 0.9041095890410958,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8157894736842105,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.12900210916996002,
-      "eval_overall_accuracy": 0.9598818606940684,
-      "eval_overall_f1": 0.8068783068783069,
-      "eval_overall_precision": 0.7530864197530864,
-      "eval_overall_recall": 0.8689458689458689,
-      "eval_runtime": 0.9665,
-      "eval_samples_per_second": 193.482,
-      "eval_steps_per_second": 3.104,
+      "eval_TIME_precision": 0.8461538461538461,
+      "eval_TIME_recall": 0.9705882352941176,
+      "eval_loss": 0.13154004514217377,
+      "eval_overall_accuracy": 0.956928378045779,
+      "eval_overall_f1": 0.7989347536617842,
+      "eval_overall_precision": 0.75,
+      "eval_overall_recall": 0.8547008547008547,
+      "eval_runtime": 0.3718,
+      "eval_samples_per_second": 502.965,
+      "eval_steps_per_second": 8.069,
       "step": 3180
     },
     {
       "epoch": 31.0,
-      "grad_norm": 0.547255277633667,
+      "grad_norm": 0.5096124410629272,
       "learning_rate": 3.45e-05,
-      "loss": 0.1029,
+      "loss": 0.1053,
       "step": 3286
     },
     {
       "epoch": 31.0,
-      "eval_LOCATION_f1": 0.8129032258064516,
+      "eval_LOCATION_f1": 0.8051948051948052,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7682926829268293,
-      "eval_LOCATION_recall": 0.863013698630137,
-      "eval_ORGANIZATION_f1": 0.6878980891719746,
+      "eval_LOCATION_precision": 0.7654320987654321,
+      "eval_LOCATION_recall": 0.8493150684931506,
+      "eval_ORGANIZATION_f1": 0.6753246753246754,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.5869565217391305,
-      "eval_ORGANIZATION_recall": 0.8307692307692308,
-      "eval_PERSON_f1": 0.8670886075949367,
+      "eval_ORGANIZATION_precision": 0.5842696629213483,
+      "eval_ORGANIZATION_recall": 0.8,
+      "eval_PERSON_f1": 0.879746835443038,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8253012048192772,
-      "eval_PERSON_recall": 0.9133333333333333,
-      "eval_QUANTITY_f1": 0.7352941176470588,
+      "eval_PERSON_precision": 0.8373493975903614,
+      "eval_PERSON_recall": 0.9266666666666666,
+      "eval_QUANTITY_f1": 0.676470588235294,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6410256410256411,
-      "eval_QUANTITY_recall": 0.8620689655172413,
-      "eval_TIME_f1": 0.7837837837837837,
+      "eval_QUANTITY_precision": 0.5897435897435898,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8266666666666665,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.725,
-      "eval_TIME_recall": 0.8529411764705882,
-      "eval_loss": 0.13610094785690308,
-      "eval_overall_accuracy": 0.9571745015998031,
-      "eval_overall_f1": 0.8,
-      "eval_overall_precision": 0.7350835322195705,
-      "eval_overall_recall": 0.8774928774928775,
-      "eval_runtime": 0.9586,
-      "eval_samples_per_second": 195.083,
-      "eval_steps_per_second": 3.13,
+      "eval_TIME_precision": 0.7560975609756098,
+      "eval_TIME_recall": 0.9117647058823529,
+      "eval_loss": 0.1396467238664627,
+      "eval_overall_accuracy": 0.9554516367216342,
+      "eval_overall_f1": 0.8005215123859191,
+      "eval_overall_precision": 0.7379807692307693,
+      "eval_overall_recall": 0.8746438746438746,
+      "eval_runtime": 0.373,
+      "eval_samples_per_second": 501.294,
+      "eval_steps_per_second": 8.042,
       "step": 3286
     },
     {
       "epoch": 32.0,
-      "grad_norm": 0.32147249579429626,
+      "grad_norm": 0.622959554195404,
       "learning_rate": 3.4000000000000007e-05,
-      "loss": 0.1023,
+      "loss": 0.1027,
       "step": 3392
     },
     {
       "epoch": 32.0,
-      "eval_LOCATION_f1": 0.7848101265822783,
+      "eval_LOCATION_f1": 0.8129032258064516,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7294117647058823,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.6794871794871795,
+      "eval_LOCATION_precision": 0.7682926829268293,
+      "eval_LOCATION_recall": 0.863013698630137,
+      "eval_ORGANIZATION_f1": 0.6748466257668712,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.5824175824175825,
-      "eval_ORGANIZATION_recall": 0.8153846153846154,
-      "eval_PERSON_f1": 0.8734177215189873,
+      "eval_ORGANIZATION_precision": 0.5612244897959183,
+      "eval_ORGANIZATION_recall": 0.8461538461538461,
+      "eval_PERSON_f1": 0.879746835443038,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8313253012048193,
-      "eval_PERSON_recall": 0.92,
-      "eval_QUANTITY_f1": 0.7352941176470588,
+      "eval_PERSON_precision": 0.8373493975903614,
+      "eval_PERSON_recall": 0.9266666666666666,
+      "eval_QUANTITY_f1": 0.696969696969697,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6410256410256411,
-      "eval_QUANTITY_recall": 0.8620689655172413,
-      "eval_TIME_f1": 0.8493150684931507,
+      "eval_QUANTITY_precision": 0.6216216216216216,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8533333333333334,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.7948717948717948,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.13857892155647278,
-      "eval_overall_accuracy": 0.9581589958158996,
-      "eval_overall_f1": 0.8015564202334631,
-      "eval_overall_precision": 0.7357142857142858,
-      "eval_overall_recall": 0.8803418803418803,
-      "eval_runtime": 0.963,
-      "eval_samples_per_second": 194.191,
-      "eval_steps_per_second": 3.115,
+      "eval_TIME_precision": 0.7804878048780488,
+      "eval_TIME_recall": 0.9411764705882353,
+      "eval_loss": 0.14285185933113098,
+      "eval_overall_accuracy": 0.9549593896135861,
+      "eval_overall_f1": 0.8051612903225805,
+      "eval_overall_precision": 0.7358490566037735,
+      "eval_overall_recall": 0.8888888888888888,
+      "eval_runtime": 0.371,
+      "eval_samples_per_second": 504.003,
+      "eval_steps_per_second": 8.086,
       "step": 3392
     },
     {
       "epoch": 33.0,
-      "grad_norm": 1.505850911140442,
+      "grad_norm": 1.2904415130615234,
       "learning_rate": 3.35e-05,
-      "loss": 0.1028,
+      "loss": 0.1036,
       "step": 3498
     },
     {
       "epoch": 33.0,
-      "eval_LOCATION_f1": 0.7820512820512822,
+      "eval_LOCATION_f1": 0.8051948051948052,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7349397590361446,
-      "eval_LOCATION_recall": 0.8356164383561644,
-      "eval_ORGANIZATION_f1": 0.7006369426751592,
+      "eval_LOCATION_precision": 0.7654320987654321,
+      "eval_LOCATION_recall": 0.8493150684931506,
+      "eval_ORGANIZATION_f1": 0.6878980891719746,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.5978260869565217,
-      "eval_ORGANIZATION_recall": 0.8461538461538461,
+      "eval_ORGANIZATION_precision": 0.5869565217391305,
+      "eval_ORGANIZATION_recall": 0.8307692307692308,
       "eval_PERSON_f1": 0.8734177215189873,
       "eval_PERSON_number": 150,
       "eval_PERSON_precision": 0.8313253012048193,
@@ -1281,1351 +1281,1351 @@
       "eval_QUANTITY_number": 29,
       "eval_QUANTITY_precision": 0.631578947368421,
       "eval_QUANTITY_recall": 0.8275862068965517,
-      "eval_TIME_f1": 0.8333333333333333,
+      "eval_TIME_f1": 0.8378378378378379,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.7894736842105263,
-      "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.1360621303319931,
-      "eval_overall_accuracy": 0.9576667487078513,
-      "eval_overall_f1": 0.8020833333333333,
-      "eval_overall_precision": 0.7386091127098321,
-      "eval_overall_recall": 0.8774928774928775,
-      "eval_runtime": 0.9705,
-      "eval_samples_per_second": 192.685,
-      "eval_steps_per_second": 3.091,
+      "eval_TIME_precision": 0.775,
+      "eval_TIME_recall": 0.9117647058823529,
+      "eval_loss": 0.1392923891544342,
+      "eval_overall_accuracy": 0.9559438838296825,
+      "eval_overall_f1": 0.8046875000000001,
+      "eval_overall_precision": 0.7410071942446043,
+      "eval_overall_recall": 0.8803418803418803,
+      "eval_runtime": 0.3722,
+      "eval_samples_per_second": 502.408,
+      "eval_steps_per_second": 8.06,
       "step": 3498
     },
     {
       "epoch": 34.0,
-      "grad_norm": 1.6967480182647705,
+      "grad_norm": 3.785581350326538,
       "learning_rate": 3.3e-05,
-      "loss": 0.0991,
+      "loss": 0.1,
       "step": 3604
     },
     {
       "epoch": 34.0,
-      "eval_LOCATION_f1": 0.7770700636942677,
+      "eval_LOCATION_f1": 0.7999999999999999,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7261904761904762,
-      "eval_LOCATION_recall": 0.8356164383561644,
-      "eval_ORGANIZATION_f1": 0.6710526315789473,
+      "eval_LOCATION_precision": 0.7560975609756098,
+      "eval_LOCATION_recall": 0.8493150684931506,
+      "eval_ORGANIZATION_f1": 0.6838709677419356,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.5862068965517241,
-      "eval_ORGANIZATION_recall": 0.7846153846153846,
-      "eval_PERSON_f1": 0.8761904761904761,
+      "eval_ORGANIZATION_precision": 0.5888888888888889,
+      "eval_ORGANIZATION_recall": 0.8153846153846154,
+      "eval_PERSON_f1": 0.8734177215189873,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8363636363636363,
+      "eval_PERSON_precision": 0.8313253012048193,
       "eval_PERSON_recall": 0.92,
-      "eval_QUANTITY_f1": 0.716417910447761,
+      "eval_QUANTITY_f1": 0.6865671641791046,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.631578947368421,
-      "eval_QUANTITY_recall": 0.8275862068965517,
-      "eval_TIME_f1": 0.8648648648648648,
+      "eval_QUANTITY_precision": 0.6052631578947368,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8533333333333334,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8,
+      "eval_TIME_precision": 0.7804878048780488,
       "eval_TIME_recall": 0.9411764705882353,
-      "eval_loss": 0.13175979256629944,
-      "eval_overall_accuracy": 0.9581589958158996,
-      "eval_overall_f1": 0.7999999999999999,
-      "eval_overall_precision": 0.7391304347826086,
-      "eval_overall_recall": 0.8717948717948718,
-      "eval_runtime": 0.967,
-      "eval_samples_per_second": 193.389,
-      "eval_steps_per_second": 3.103,
+      "eval_loss": 0.13862958550453186,
+      "eval_overall_accuracy": 0.9561900073837066,
+      "eval_overall_f1": 0.8020833333333333,
+      "eval_overall_precision": 0.7386091127098321,
+      "eval_overall_recall": 0.8774928774928775,
+      "eval_runtime": 0.3727,
+      "eval_samples_per_second": 501.785,
+      "eval_steps_per_second": 8.05,
       "step": 3604
     },
     {
       "epoch": 35.0,
-      "grad_norm": 1.3785823583602905,
+      "grad_norm": 0.7286350131034851,
       "learning_rate": 3.2500000000000004e-05,
-      "loss": 0.0983,
+      "loss": 0.1013,
       "step": 3710
     },
     {
       "epoch": 35.0,
-      "eval_LOCATION_f1": 0.7530864197530865,
+      "eval_LOCATION_f1": 0.7948717948717949,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.6853932584269663,
-      "eval_LOCATION_recall": 0.8356164383561644,
-      "eval_ORGANIZATION_f1": 0.6838709677419356,
+      "eval_LOCATION_precision": 0.7469879518072289,
+      "eval_LOCATION_recall": 0.8493150684931506,
+      "eval_ORGANIZATION_f1": 0.6875,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.5888888888888889,
-      "eval_ORGANIZATION_recall": 0.8153846153846154,
-      "eval_PERSON_f1": 0.8761904761904761,
+      "eval_ORGANIZATION_precision": 0.5789473684210527,
+      "eval_ORGANIZATION_recall": 0.8461538461538461,
+      "eval_PERSON_f1": 0.8734177215189873,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8363636363636363,
+      "eval_PERSON_precision": 0.8313253012048193,
       "eval_PERSON_recall": 0.92,
       "eval_QUANTITY_f1": 0.6956521739130435,
       "eval_QUANTITY_number": 29,
       "eval_QUANTITY_precision": 0.6,
       "eval_QUANTITY_recall": 0.8275862068965517,
-      "eval_TIME_f1": 0.8108108108108107,
+      "eval_TIME_f1": 0.8266666666666665,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.75,
-      "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.14071786403656006,
-      "eval_overall_accuracy": 0.9556977602756583,
-      "eval_overall_f1": 0.7896774193548387,
-      "eval_overall_precision": 0.7216981132075472,
-      "eval_overall_recall": 0.8717948717948718,
-      "eval_runtime": 0.9573,
-      "eval_samples_per_second": 195.336,
-      "eval_steps_per_second": 3.134,
+      "eval_TIME_precision": 0.7560975609756098,
+      "eval_TIME_recall": 0.9117647058823529,
+      "eval_loss": 0.1437290757894516,
+      "eval_overall_accuracy": 0.9549593896135861,
+      "eval_overall_f1": 0.7989690721649485,
+      "eval_overall_precision": 0.7294117647058823,
+      "eval_overall_recall": 0.8831908831908832,
+      "eval_runtime": 0.3727,
+      "eval_samples_per_second": 501.777,
+      "eval_steps_per_second": 8.05,
       "step": 3710
     },
     {
       "epoch": 36.0,
-      "grad_norm": 1.251754879951477,
+      "grad_norm": 0.8509505391120911,
       "learning_rate": 3.2000000000000005e-05,
-      "loss": 0.0962,
+      "loss": 0.0988,
       "step": 3816
     },
     {
       "epoch": 36.0,
-      "eval_LOCATION_f1": 0.7948717948717949,
+      "eval_LOCATION_f1": 0.7870967741935484,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7469879518072289,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7169811320754718,
+      "eval_LOCATION_precision": 0.7439024390243902,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.6794871794871795,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6063829787234043,
-      "eval_ORGANIZATION_recall": 0.8769230769230769,
-      "eval_PERSON_f1": 0.8789808917197452,
+      "eval_ORGANIZATION_precision": 0.5824175824175825,
+      "eval_ORGANIZATION_recall": 0.8153846153846154,
+      "eval_PERSON_f1": 0.8761904761904761,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8414634146341463,
+      "eval_PERSON_precision": 0.8363636363636363,
       "eval_PERSON_recall": 0.92,
-      "eval_QUANTITY_f1": 0.7058823529411765,
+      "eval_QUANTITY_f1": 0.676470588235294,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6153846153846154,
-      "eval_QUANTITY_recall": 0.8275862068965517,
-      "eval_TIME_f1": 0.8108108108108107,
+      "eval_QUANTITY_precision": 0.5897435897435898,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8266666666666665,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.75,
-      "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.14177566766738892,
-      "eval_overall_accuracy": 0.956928378045779,
-      "eval_overall_f1": 0.8067444876783397,
-      "eval_overall_precision": 0.7404761904761905,
-      "eval_overall_recall": 0.886039886039886,
-      "eval_runtime": 0.962,
-      "eval_samples_per_second": 194.386,
-      "eval_steps_per_second": 3.118,
+      "eval_TIME_precision": 0.7560975609756098,
+      "eval_TIME_recall": 0.9117647058823529,
+      "eval_loss": 0.14367760717868805,
+      "eval_overall_accuracy": 0.9547132660595619,
+      "eval_overall_f1": 0.7958387516254878,
+      "eval_overall_precision": 0.7320574162679426,
+      "eval_overall_recall": 0.8717948717948718,
+      "eval_runtime": 0.3724,
+      "eval_samples_per_second": 502.107,
+      "eval_steps_per_second": 8.055,
       "step": 3816
     },
     {
       "epoch": 37.0,
-      "grad_norm": 1.2584689855575562,
+      "grad_norm": 1.2342133522033691,
       "learning_rate": 3.15e-05,
-      "loss": 0.0986,
+      "loss": 0.0971,
       "step": 3922
     },
     {
       "epoch": 37.0,
-      "eval_LOCATION_f1": 0.7848101265822783,
+      "eval_LOCATION_f1": 0.7770700636942677,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7294117647058823,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.708860759493671,
+      "eval_LOCATION_precision": 0.7261904761904762,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.6705882352941177,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6021505376344086,
-      "eval_ORGANIZATION_recall": 0.8615384615384616,
-      "eval_PERSON_f1": 0.8734177215189873,
+      "eval_ORGANIZATION_precision": 0.5428571428571428,
+      "eval_ORGANIZATION_recall": 0.8769230769230769,
+      "eval_PERSON_f1": 0.8706624605678233,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8313253012048193,
+      "eval_PERSON_precision": 0.8263473053892215,
       "eval_PERSON_recall": 0.92,
-      "eval_QUANTITY_f1": 0.6956521739130435,
+      "eval_QUANTITY_f1": 0.676470588235294,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6,
-      "eval_QUANTITY_recall": 0.8275862068965517,
-      "eval_TIME_f1": 0.7837837837837837,
+      "eval_QUANTITY_precision": 0.5897435897435898,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.725,
-      "eval_TIME_recall": 0.8529411764705882,
-      "eval_loss": 0.1421373039484024,
-      "eval_overall_accuracy": 0.9556977602756583,
-      "eval_overall_f1": 0.7974193548387097,
-      "eval_overall_precision": 0.7287735849056604,
+      "eval_TIME_precision": 0.7317073170731707,
+      "eval_TIME_recall": 0.8823529411764706,
+      "eval_loss": 0.1551184207201004,
+      "eval_overall_accuracy": 0.9490524243170071,
+      "eval_overall_f1": 0.7852604828462516,
+      "eval_overall_precision": 0.7087155963302753,
       "eval_overall_recall": 0.8803418803418803,
-      "eval_runtime": 0.9499,
-      "eval_samples_per_second": 196.858,
-      "eval_steps_per_second": 3.158,
+      "eval_runtime": 0.3737,
+      "eval_samples_per_second": 500.4,
+      "eval_steps_per_second": 8.028,
       "step": 3922
     },
     {
       "epoch": 38.0,
-      "grad_norm": 0.9512435793876648,
+      "grad_norm": 1.0431395769119263,
       "learning_rate": 3.1e-05,
-      "loss": 0.0971,
+      "loss": 0.0967,
       "step": 4028
     },
     {
       "epoch": 38.0,
-      "eval_LOCATION_f1": 0.7770700636942677,
+      "eval_LOCATION_f1": 0.7973856209150326,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7261904761904762,
+      "eval_LOCATION_precision": 0.7625,
       "eval_LOCATION_recall": 0.8356164383561644,
-      "eval_ORGANIZATION_f1": 0.684931506849315,
+      "eval_ORGANIZATION_f1": 0.6805555555555556,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6172839506172839,
-      "eval_ORGANIZATION_recall": 0.7692307692307693,
-      "eval_PERSON_f1": 0.8734177215189873,
+      "eval_ORGANIZATION_precision": 0.620253164556962,
+      "eval_ORGANIZATION_recall": 0.7538461538461538,
+      "eval_PERSON_f1": 0.8761904761904761,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8313253012048193,
+      "eval_PERSON_precision": 0.8363636363636363,
       "eval_PERSON_recall": 0.92,
-      "eval_QUANTITY_f1": 0.6666666666666666,
+      "eval_QUANTITY_f1": 0.6865671641791046,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.575,
+      "eval_QUANTITY_precision": 0.6052631578947368,
       "eval_QUANTITY_recall": 0.7931034482758621,
-      "eval_TIME_f1": 0.8493150684931507,
+      "eval_TIME_f1": 0.8378378378378379,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.7948717948717948,
+      "eval_TIME_precision": 0.775,
       "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.13257591426372528,
-      "eval_overall_accuracy": 0.9576667487078513,
-      "eval_overall_f1": 0.7963206307490145,
-      "eval_overall_precision": 0.7390243902439024,
-      "eval_overall_recall": 0.8632478632478633,
-      "eval_runtime": 0.9531,
-      "eval_samples_per_second": 196.208,
-      "eval_steps_per_second": 3.148,
+      "eval_loss": 0.13686208426952362,
+      "eval_overall_accuracy": 0.956928378045779,
+      "eval_overall_f1": 0.802124833997344,
+      "eval_overall_precision": 0.7512437810945274,
+      "eval_overall_recall": 0.8603988603988604,
+      "eval_runtime": 0.3727,
+      "eval_samples_per_second": 501.728,
+      "eval_steps_per_second": 8.049,
       "step": 4028
     },
     {
       "epoch": 39.0,
-      "grad_norm": 1.808843731880188,
+      "grad_norm": 1.3837076425552368,
       "learning_rate": 3.05e-05,
-      "loss": 0.0944,
+      "loss": 0.0949,
       "step": 4134
     },
     {
       "epoch": 39.0,
-      "eval_LOCATION_f1": 0.7749999999999999,
+      "eval_LOCATION_f1": 0.7898089171974523,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7126436781609196,
+      "eval_LOCATION_precision": 0.7380952380952381,
       "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7012987012987013,
+      "eval_ORGANIZATION_f1": 0.6832298136645962,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6067415730337079,
-      "eval_ORGANIZATION_recall": 0.8307692307692308,
-      "eval_PERSON_f1": 0.8832807570977917,
+      "eval_ORGANIZATION_precision": 0.5729166666666666,
+      "eval_ORGANIZATION_recall": 0.8461538461538461,
+      "eval_PERSON_f1": 0.8761904761904761,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8383233532934131,
-      "eval_PERSON_recall": 0.9333333333333333,
-      "eval_QUANTITY_f1": 0.676470588235294,
+      "eval_PERSON_precision": 0.8363636363636363,
+      "eval_PERSON_recall": 0.92,
+      "eval_QUANTITY_f1": 0.6666666666666667,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.5897435897435898,
-      "eval_QUANTITY_recall": 0.7931034482758621,
-      "eval_TIME_f1": 0.7837837837837837,
+      "eval_QUANTITY_precision": 0.5945945945945946,
+      "eval_QUANTITY_recall": 0.7586206896551724,
+      "eval_TIME_f1": 0.8266666666666665,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.725,
-      "eval_TIME_recall": 0.8529411764705882,
-      "eval_loss": 0.1387697011232376,
-      "eval_overall_accuracy": 0.9564361309377307,
-      "eval_overall_f1": 0.7968952134540751,
-      "eval_overall_precision": 0.7298578199052133,
+      "eval_TIME_precision": 0.7560975609756098,
+      "eval_TIME_recall": 0.9117647058823529,
+      "eval_loss": 0.13983625173568726,
+      "eval_overall_accuracy": 0.9559438838296825,
+      "eval_overall_f1": 0.7958656330749354,
+      "eval_overall_precision": 0.7281323877068558,
       "eval_overall_recall": 0.8774928774928775,
-      "eval_runtime": 0.9514,
-      "eval_samples_per_second": 196.544,
-      "eval_steps_per_second": 3.153,
+      "eval_runtime": 0.3721,
+      "eval_samples_per_second": 502.567,
+      "eval_steps_per_second": 8.063,
       "step": 4134
     },
     {
       "epoch": 40.0,
-      "grad_norm": 1.092385172843933,
+      "grad_norm": 0.6632676124572754,
       "learning_rate": 3e-05,
-      "loss": 0.093,
+      "loss": 0.0956,
       "step": 4240
     },
     {
       "epoch": 40.0,
-      "eval_LOCATION_f1": 0.7999999999999999,
+      "eval_LOCATION_f1": 0.7770700636942677,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7560975609756098,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7368421052631579,
+      "eval_LOCATION_precision": 0.7261904761904762,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.6973684210526315,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6436781609195402,
-      "eval_ORGANIZATION_recall": 0.8615384615384616,
-      "eval_PERSON_f1": 0.8670886075949367,
+      "eval_ORGANIZATION_precision": 0.6091954022988506,
+      "eval_ORGANIZATION_recall": 0.8153846153846154,
+      "eval_PERSON_f1": 0.8789808917197452,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8253012048192772,
-      "eval_PERSON_recall": 0.9133333333333333,
-      "eval_QUANTITY_f1": 0.6567164179104478,
+      "eval_PERSON_precision": 0.8414634146341463,
+      "eval_PERSON_recall": 0.92,
+      "eval_QUANTITY_f1": 0.676470588235294,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.5789473684210527,
-      "eval_QUANTITY_recall": 0.7586206896551724,
-      "eval_TIME_f1": 0.8333333333333333,
+      "eval_QUANTITY_precision": 0.5897435897435898,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8378378378378379,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.7894736842105263,
-      "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.1326187252998352,
-      "eval_overall_accuracy": 0.958897366477972,
-      "eval_overall_f1": 0.8057742782152231,
-      "eval_overall_precision": 0.7469586374695864,
-      "eval_overall_recall": 0.8746438746438746,
-      "eval_runtime": 0.9497,
-      "eval_samples_per_second": 196.897,
-      "eval_steps_per_second": 3.159,
+      "eval_TIME_precision": 0.775,
+      "eval_TIME_recall": 0.9117647058823529,
+      "eval_loss": 0.14134033024311066,
+      "eval_overall_accuracy": 0.9559438838296825,
+      "eval_overall_f1": 0.7999999999999999,
+      "eval_overall_precision": 0.7391304347826086,
+      "eval_overall_recall": 0.8717948717948718,
+      "eval_runtime": 0.3777,
+      "eval_samples_per_second": 495.154,
+      "eval_steps_per_second": 7.944,
       "step": 4240
     },
     {
       "epoch": 41.0,
-      "grad_norm": 0.6803504824638367,
+      "grad_norm": 0.36149993538856506,
       "learning_rate": 2.95e-05,
-      "loss": 0.0911,
+      "loss": 0.0913,
       "step": 4346
     },
     {
       "epoch": 41.0,
-      "eval_LOCATION_f1": 0.8026315789473685,
+      "eval_LOCATION_f1": 0.7898089171974523,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7721518987341772,
-      "eval_LOCATION_recall": 0.8356164383561644,
-      "eval_ORGANIZATION_f1": 0.7368421052631579,
+      "eval_LOCATION_precision": 0.7380952380952381,
+      "eval_LOCATION_recall": 0.8493150684931506,
+      "eval_ORGANIZATION_f1": 0.7051282051282052,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6436781609195402,
-      "eval_ORGANIZATION_recall": 0.8615384615384616,
-      "eval_PERSON_f1": 0.8761904761904761,
+      "eval_ORGANIZATION_precision": 0.6043956043956044,
+      "eval_ORGANIZATION_recall": 0.8461538461538461,
+      "eval_PERSON_f1": 0.8789808917197452,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8363636363636363,
+      "eval_PERSON_precision": 0.8414634146341463,
       "eval_PERSON_recall": 0.92,
-      "eval_QUANTITY_f1": 0.6666666666666667,
+      "eval_QUANTITY_f1": 0.6865671641791046,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.5945945945945946,
-      "eval_QUANTITY_recall": 0.7586206896551724,
-      "eval_TIME_f1": 0.861111111111111,
+      "eval_QUANTITY_precision": 0.6052631578947368,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8266666666666665,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8157894736842105,
+      "eval_TIME_precision": 0.7560975609756098,
       "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.12640033662319183,
-      "eval_overall_accuracy": 0.9603741078021166,
-      "eval_overall_f1": 0.8137384412153237,
-      "eval_overall_precision": 0.7586206896551724,
-      "eval_overall_recall": 0.8774928774928775,
-      "eval_runtime": 0.955,
-      "eval_samples_per_second": 195.805,
-      "eval_steps_per_second": 3.141,
+      "eval_loss": 0.13970939815044403,
+      "eval_overall_accuracy": 0.9566822544917548,
+      "eval_overall_f1": 0.8036410923276982,
+      "eval_overall_precision": 0.7392344497607656,
+      "eval_overall_recall": 0.8803418803418803,
+      "eval_runtime": 0.3722,
+      "eval_samples_per_second": 502.411,
+      "eval_steps_per_second": 8.06,
       "step": 4346
     },
     {
       "epoch": 42.0,
-      "grad_norm": 0.7038798928260803,
+      "grad_norm": 0.5602841377258301,
       "learning_rate": 2.9e-05,
-      "loss": 0.0915,
+      "loss": 0.0949,
       "step": 4452
     },
     {
       "epoch": 42.0,
-      "eval_LOCATION_f1": 0.8051948051948052,
+      "eval_LOCATION_f1": 0.7870967741935484,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7654320987654321,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7225806451612903,
+      "eval_LOCATION_precision": 0.7439024390243902,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.6923076923076923,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6222222222222222,
-      "eval_ORGANIZATION_recall": 0.8615384615384616,
-      "eval_PERSON_f1": 0.8761904761904761,
+      "eval_ORGANIZATION_precision": 0.5934065934065934,
+      "eval_ORGANIZATION_recall": 0.8307692307692308,
+      "eval_PERSON_f1": 0.8846153846153846,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8363636363636363,
+      "eval_PERSON_precision": 0.8518518518518519,
       "eval_PERSON_recall": 0.92,
-      "eval_QUANTITY_f1": 0.6956521739130435,
+      "eval_QUANTITY_f1": 0.676470588235294,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6,
-      "eval_QUANTITY_recall": 0.8275862068965517,
-      "eval_TIME_f1": 0.8333333333333333,
+      "eval_QUANTITY_precision": 0.5897435897435898,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8378378378378379,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.7894736842105263,
-      "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.13610343635082245,
-      "eval_overall_accuracy": 0.9593896135860202,
-      "eval_overall_f1": 0.8104575163398693,
-      "eval_overall_precision": 0.748792270531401,
-      "eval_overall_recall": 0.8831908831908832,
-      "eval_runtime": 0.9531,
-      "eval_samples_per_second": 196.198,
-      "eval_steps_per_second": 3.148,
+      "eval_TIME_precision": 0.775,
+      "eval_TIME_recall": 0.9117647058823529,
+      "eval_loss": 0.1462649405002594,
+      "eval_overall_accuracy": 0.9547132660595619,
+      "eval_overall_f1": 0.8026143790849672,
+      "eval_overall_precision": 0.7415458937198067,
+      "eval_overall_recall": 0.8746438746438746,
+      "eval_runtime": 0.3742,
+      "eval_samples_per_second": 499.754,
+      "eval_steps_per_second": 8.017,
       "step": 4452
     },
     {
       "epoch": 43.0,
-      "grad_norm": 0.7236311435699463,
+      "grad_norm": 0.9771382808685303,
       "learning_rate": 2.8499999999999998e-05,
-      "loss": 0.0896,
+      "loss": 0.0919,
       "step": 4558
     },
     {
       "epoch": 43.0,
-      "eval_LOCATION_f1": 0.7870967741935484,
+      "eval_LOCATION_f1": 0.7820512820512822,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7439024390243902,
+      "eval_LOCATION_precision": 0.7349397590361446,
       "eval_LOCATION_recall": 0.8356164383561644,
-      "eval_ORGANIZATION_f1": 0.6986301369863014,
+      "eval_ORGANIZATION_f1": 0.684931506849315,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6296296296296297,
-      "eval_ORGANIZATION_recall": 0.7846153846153846,
-      "eval_PERSON_f1": 0.8726114649681529,
+      "eval_ORGANIZATION_precision": 0.6172839506172839,
+      "eval_ORGANIZATION_recall": 0.7692307692307693,
+      "eval_PERSON_f1": 0.8789808917197452,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8353658536585366,
-      "eval_PERSON_recall": 0.9133333333333333,
-      "eval_QUANTITY_f1": 0.7076923076923076,
+      "eval_PERSON_precision": 0.8414634146341463,
+      "eval_PERSON_recall": 0.92,
+      "eval_QUANTITY_f1": 0.6666666666666667,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6388888888888888,
-      "eval_QUANTITY_recall": 0.7931034482758621,
-      "eval_TIME_f1": 0.8333333333333333,
+      "eval_QUANTITY_precision": 0.5945945945945946,
+      "eval_QUANTITY_recall": 0.7586206896551724,
+      "eval_TIME_f1": 0.8266666666666665,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.7894736842105263,
-      "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.1266656517982483,
-      "eval_overall_accuracy": 0.9596357371400444,
-      "eval_overall_f1": 0.8031914893617021,
-      "eval_overall_precision": 0.7531172069825436,
+      "eval_TIME_precision": 0.7560975609756098,
+      "eval_TIME_recall": 0.9117647058823529,
+      "eval_loss": 0.1370939314365387,
+      "eval_overall_accuracy": 0.9566822544917548,
+      "eval_overall_f1": 0.7978863936591809,
+      "eval_overall_precision": 0.7438423645320197,
       "eval_overall_recall": 0.8603988603988604,
-      "eval_runtime": 0.9502,
-      "eval_samples_per_second": 196.803,
-      "eval_steps_per_second": 3.157,
+      "eval_runtime": 0.3709,
+      "eval_samples_per_second": 504.137,
+      "eval_steps_per_second": 8.088,
       "step": 4558
     },
     {
       "epoch": 44.0,
-      "grad_norm": 1.4153372049331665,
+      "grad_norm": 3.2234935760498047,
       "learning_rate": 2.8000000000000003e-05,
-      "loss": 0.0891,
+      "loss": 0.0895,
       "step": 4664
     },
     {
       "epoch": 44.0,
-      "eval_LOCATION_f1": 0.8157894736842105,
+      "eval_LOCATION_f1": 0.7999999999999999,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7848101265822784,
+      "eval_LOCATION_precision": 0.7560975609756098,
       "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7210884353741497,
+      "eval_ORGANIZATION_f1": 0.6758620689655173,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6463414634146342,
-      "eval_ORGANIZATION_recall": 0.8153846153846154,
-      "eval_PERSON_f1": 0.8761904761904761,
+      "eval_ORGANIZATION_precision": 0.6125,
+      "eval_ORGANIZATION_recall": 0.7538461538461538,
+      "eval_PERSON_f1": 0.8825396825396825,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8363636363636363,
-      "eval_PERSON_recall": 0.92,
+      "eval_PERSON_precision": 0.8424242424242424,
+      "eval_PERSON_recall": 0.9266666666666666,
       "eval_QUANTITY_f1": 0.6865671641791046,
       "eval_QUANTITY_number": 29,
       "eval_QUANTITY_precision": 0.6052631578947368,
       "eval_QUANTITY_recall": 0.7931034482758621,
-      "eval_TIME_f1": 0.8219178082191781,
+      "eval_TIME_f1": 0.8648648648648648,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.7692307692307693,
-      "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.12547963857650757,
-      "eval_overall_accuracy": 0.9596357371400444,
-      "eval_overall_f1": 0.8116710875331565,
-      "eval_overall_precision": 0.7593052109181141,
-      "eval_overall_recall": 0.8717948717948718,
-      "eval_runtime": 0.9553,
-      "eval_samples_per_second": 195.759,
-      "eval_steps_per_second": 3.141,
-      "step": 4664
-    },
-    {
-      "epoch": 45.0,
-      "grad_norm": 3.5179989337921143,
-      "learning_rate": 2.7500000000000004e-05,
-      "loss": 0.0882,
+      "eval_TIME_precision": 0.8,
+      "eval_TIME_recall": 0.9411764705882353,
+      "eval_loss": 0.1338210105895996,
+      "eval_overall_accuracy": 0.9576667487078513,
+      "eval_overall_f1": 0.8068783068783069,
+      "eval_overall_precision": 0.7530864197530864,
+      "eval_overall_recall": 0.8689458689458689,
+      "eval_runtime": 0.3714,
+      "eval_samples_per_second": 503.565,
+      "eval_steps_per_second": 8.079,
+      "step": 4664
+    },
+    {
+      "epoch": 45.0,
+      "grad_norm": 3.458214282989502,
+      "learning_rate": 2.7500000000000004e-05,
+      "loss": 0.0899,
       "step": 4770
     },
     {
       "epoch": 45.0,
-      "eval_LOCATION_f1": 0.8157894736842105,
+      "eval_LOCATION_f1": 0.7792207792207793,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7848101265822784,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7320261437908497,
+      "eval_LOCATION_precision": 0.7407407407407407,
+      "eval_LOCATION_recall": 0.821917808219178,
+      "eval_ORGANIZATION_f1": 0.6835443037974683,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6363636363636364,
-      "eval_ORGANIZATION_recall": 0.8615384615384616,
-      "eval_PERSON_f1": 0.8761904761904761,
+      "eval_ORGANIZATION_precision": 0.5806451612903226,
+      "eval_ORGANIZATION_recall": 0.8307692307692308,
+      "eval_PERSON_f1": 0.8789808917197452,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8363636363636363,
+      "eval_PERSON_precision": 0.8414634146341463,
       "eval_PERSON_recall": 0.92,
-      "eval_QUANTITY_f1": 0.7272727272727273,
+      "eval_QUANTITY_f1": 0.6666666666666667,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6486486486486487,
-      "eval_QUANTITY_recall": 0.8275862068965517,
-      "eval_TIME_f1": 0.8493150684931507,
+      "eval_QUANTITY_precision": 0.5945945945945946,
+      "eval_QUANTITY_recall": 0.7586206896551724,
+      "eval_TIME_f1": 0.8533333333333334,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.7948717948717948,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.13158655166625977,
-      "eval_overall_accuracy": 0.9603741078021166,
-      "eval_overall_f1": 0.8194993412384717,
-      "eval_overall_precision": 0.7622549019607843,
-      "eval_overall_recall": 0.886039886039886,
-      "eval_runtime": 0.9562,
-      "eval_samples_per_second": 195.568,
-      "eval_steps_per_second": 3.137,
+      "eval_TIME_precision": 0.7804878048780488,
+      "eval_TIME_recall": 0.9411764705882353,
+      "eval_loss": 0.14024274051189423,
+      "eval_overall_accuracy": 0.956928378045779,
+      "eval_overall_f1": 0.7979139504563233,
+      "eval_overall_precision": 0.7355769230769231,
+      "eval_overall_recall": 0.8717948717948718,
+      "eval_runtime": 0.4103,
+      "eval_samples_per_second": 455.728,
+      "eval_steps_per_second": 7.311,
       "step": 4770
     },
     {
       "epoch": 46.0,
-      "grad_norm": 1.2313103675842285,
+      "grad_norm": 1.2206172943115234,
       "learning_rate": 2.7000000000000002e-05,
-      "loss": 0.0853,
+      "loss": 0.088,
       "step": 4876
     },
     {
       "epoch": 46.0,
-      "eval_LOCATION_f1": 0.8157894736842105,
+      "eval_LOCATION_f1": 0.7922077922077922,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7848101265822784,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7333333333333334,
+      "eval_LOCATION_precision": 0.7530864197530864,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.7019867549668873,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6470588235294118,
-      "eval_ORGANIZATION_recall": 0.8461538461538461,
-      "eval_PERSON_f1": 0.888888888888889,
+      "eval_ORGANIZATION_precision": 0.6162790697674418,
+      "eval_ORGANIZATION_recall": 0.8153846153846154,
+      "eval_PERSON_f1": 0.8846153846153846,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8484848484848485,
-      "eval_PERSON_recall": 0.9333333333333333,
-      "eval_QUANTITY_f1": 0.6984126984126984,
+      "eval_PERSON_precision": 0.8518518518518519,
+      "eval_PERSON_recall": 0.92,
+      "eval_QUANTITY_f1": 0.6461538461538462,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6470588235294118,
-      "eval_QUANTITY_recall": 0.7586206896551724,
-      "eval_TIME_f1": 0.8888888888888888,
+      "eval_QUANTITY_precision": 0.5833333333333334,
+      "eval_QUANTITY_recall": 0.7241379310344828,
+      "eval_TIME_f1": 0.8648648648648648,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8421052631578947,
+      "eval_TIME_precision": 0.8,
       "eval_TIME_recall": 0.9411764705882353,
-      "eval_loss": 0.12368568032979965,
-      "eval_overall_accuracy": 0.9625892197883338,
-      "eval_overall_f1": 0.827127659574468,
-      "eval_overall_precision": 0.7755610972568578,
-      "eval_overall_recall": 0.886039886039886,
-      "eval_runtime": 0.9617,
-      "eval_samples_per_second": 194.451,
-      "eval_steps_per_second": 3.12,
+      "eval_loss": 0.1336318403482437,
+      "eval_overall_accuracy": 0.9584051193699237,
+      "eval_overall_f1": 0.8068783068783069,
+      "eval_overall_precision": 0.7530864197530864,
+      "eval_overall_recall": 0.8689458689458689,
+      "eval_runtime": 0.3724,
+      "eval_samples_per_second": 502.143,
+      "eval_steps_per_second": 8.056,
       "step": 4876
     },
     {
       "epoch": 47.0,
-      "grad_norm": 1.974324107170105,
+      "grad_norm": 1.561812162399292,
       "learning_rate": 2.6500000000000004e-05,
-      "loss": 0.0874,
+      "loss": 0.0896,
       "step": 4982
     },
     {
       "epoch": 47.0,
-      "eval_LOCATION_f1": 0.7924528301886793,
+      "eval_LOCATION_f1": 0.7770700636942677,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7325581395348837,
-      "eval_LOCATION_recall": 0.863013698630137,
-      "eval_ORGANIZATION_f1": 0.7297297297297298,
+      "eval_LOCATION_precision": 0.7261904761904762,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.7,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6506024096385542,
-      "eval_ORGANIZATION_recall": 0.8307692307692308,
-      "eval_PERSON_f1": 0.8769716088328076,
+      "eval_ORGANIZATION_precision": 0.5894736842105263,
+      "eval_ORGANIZATION_recall": 0.8615384615384616,
+      "eval_PERSON_f1": 0.8853503184713376,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8323353293413174,
+      "eval_PERSON_precision": 0.8475609756097561,
       "eval_PERSON_recall": 0.9266666666666666,
-      "eval_QUANTITY_f1": 0.7384615384615385,
+      "eval_QUANTITY_f1": 0.7076923076923076,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6666666666666666,
-      "eval_QUANTITY_recall": 0.8275862068965517,
-      "eval_TIME_f1": 0.8493150684931507,
+      "eval_QUANTITY_precision": 0.6388888888888888,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8421052631578947,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.7948717948717948,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.1277364045381546,
-      "eval_overall_accuracy": 0.9608663549101649,
-      "eval_overall_f1": 0.8162729658792651,
-      "eval_overall_precision": 0.7566909975669099,
+      "eval_TIME_precision": 0.7619047619047619,
+      "eval_TIME_recall": 0.9411764705882353,
+      "eval_loss": 0.14269451797008514,
+      "eval_overall_accuracy": 0.9561900073837066,
+      "eval_overall_f1": 0.805699481865285,
+      "eval_overall_precision": 0.7387173396674585,
       "eval_overall_recall": 0.886039886039886,
-      "eval_runtime": 0.9678,
-      "eval_samples_per_second": 193.219,
-      "eval_steps_per_second": 3.1,
+      "eval_runtime": 0.3722,
+      "eval_samples_per_second": 502.433,
+      "eval_steps_per_second": 8.06,
       "step": 4982
     },
     {
       "epoch": 48.0,
-      "grad_norm": 2.900847911834717,
+      "grad_norm": 3.845472812652588,
       "learning_rate": 2.6000000000000002e-05,
-      "loss": 0.0863,
+      "loss": 0.0897,
       "step": 5088
     },
     {
       "epoch": 48.0,
-      "eval_LOCATION_f1": 0.7974683544303799,
+      "eval_LOCATION_f1": 0.7672955974842768,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7411764705882353,
-      "eval_LOCATION_recall": 0.863013698630137,
-      "eval_ORGANIZATION_f1": 0.728476821192053,
+      "eval_LOCATION_precision": 0.7093023255813954,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.7019867549668873,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6395348837209303,
-      "eval_ORGANIZATION_recall": 0.8461538461538461,
-      "eval_PERSON_f1": 0.879746835443038,
+      "eval_ORGANIZATION_precision": 0.6162790697674418,
+      "eval_ORGANIZATION_recall": 0.8153846153846154,
+      "eval_PERSON_f1": 0.8825396825396825,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8373493975903614,
+      "eval_PERSON_precision": 0.8424242424242424,
       "eval_PERSON_recall": 0.9266666666666666,
-      "eval_QUANTITY_f1": 0.7384615384615385,
+      "eval_QUANTITY_f1": 0.676470588235294,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6666666666666666,
-      "eval_QUANTITY_recall": 0.8275862068965517,
-      "eval_TIME_f1": 0.8219178082191781,
+      "eval_QUANTITY_precision": 0.5897435897435898,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8157894736842106,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.7692307692307693,
-      "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.13069714605808258,
-      "eval_overall_accuracy": 0.9598818606940684,
-      "eval_overall_f1": 0.8152031454783748,
-      "eval_overall_precision": 0.7548543689320388,
-      "eval_overall_recall": 0.886039886039886,
-      "eval_runtime": 0.9521,
-      "eval_samples_per_second": 196.413,
-      "eval_steps_per_second": 3.151,
+      "eval_TIME_precision": 0.7380952380952381,
+      "eval_TIME_recall": 0.9117647058823529,
+      "eval_loss": 0.1460731029510498,
+      "eval_overall_accuracy": 0.9549593896135861,
+      "eval_overall_f1": 0.7984395318595578,
+      "eval_overall_precision": 0.7344497607655502,
+      "eval_overall_recall": 0.8746438746438746,
+      "eval_runtime": 0.3721,
+      "eval_samples_per_second": 502.528,
+      "eval_steps_per_second": 8.062,
       "step": 5088
     },
     {
       "epoch": 49.0,
-      "grad_norm": 0.5791418552398682,
+      "grad_norm": 0.42411187291145325,
       "learning_rate": 2.5500000000000003e-05,
-      "loss": 0.0881,
+      "loss": 0.0836,
       "step": 5194
     },
     {
       "epoch": 49.0,
-      "eval_LOCATION_f1": 0.779874213836478,
+      "eval_LOCATION_f1": 0.7770700636942677,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7209302325581395,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7142857142857143,
+      "eval_LOCATION_precision": 0.7261904761904762,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.6838709677419356,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6179775280898876,
-      "eval_ORGANIZATION_recall": 0.8461538461538461,
-      "eval_PERSON_f1": 0.8706624605678233,
+      "eval_ORGANIZATION_precision": 0.5888888888888889,
+      "eval_ORGANIZATION_recall": 0.8153846153846154,
+      "eval_PERSON_f1": 0.8846153846153846,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8263473053892215,
+      "eval_PERSON_precision": 0.8518518518518519,
       "eval_PERSON_recall": 0.92,
-      "eval_QUANTITY_f1": 0.6956521739130435,
+      "eval_QUANTITY_f1": 0.696969696969697,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6,
-      "eval_QUANTITY_recall": 0.8275862068965517,
-      "eval_TIME_f1": 0.8108108108108107,
+      "eval_QUANTITY_precision": 0.6216216216216216,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.75,
+      "eval_TIME_precision": 0.7317073170731707,
       "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.1359923928976059,
-      "eval_overall_accuracy": 0.9579128722618755,
-      "eval_overall_f1": 0.7994825355756792,
-      "eval_overall_precision": 0.7322274881516587,
-      "eval_overall_recall": 0.8803418803418803,
-      "eval_runtime": 0.9488,
-      "eval_samples_per_second": 197.098,
-      "eval_steps_per_second": 3.162,
+      "eval_loss": 0.14366112649440765,
+      "eval_overall_accuracy": 0.9552055131676102,
+      "eval_overall_f1": 0.7973856209150327,
+      "eval_overall_precision": 0.7367149758454107,
+      "eval_overall_recall": 0.8689458689458689,
+      "eval_runtime": 0.3738,
+      "eval_samples_per_second": 500.237,
+      "eval_steps_per_second": 8.025,
       "step": 5194
     },
     {
       "epoch": 50.0,
-      "grad_norm": 1.3018290996551514,
+      "grad_norm": 1.5665122270584106,
       "learning_rate": 2.5e-05,
-      "loss": 0.0852,
+      "loss": 0.0845,
       "step": 5300
     },
     {
       "epoch": 50.0,
-      "eval_LOCATION_f1": 0.779874213836478,
+      "eval_LOCATION_f1": 0.7770700636942677,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7209302325581395,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7105263157894737,
+      "eval_LOCATION_precision": 0.7261904761904762,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.7051282051282052,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6206896551724138,
-      "eval_ORGANIZATION_recall": 0.8307692307692308,
-      "eval_PERSON_f1": 0.8734177215189873,
+      "eval_ORGANIZATION_precision": 0.6043956043956044,
+      "eval_ORGANIZATION_recall": 0.8461538461538461,
+      "eval_PERSON_f1": 0.8817891373801918,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8313253012048193,
+      "eval_PERSON_precision": 0.8466257668711656,
       "eval_PERSON_recall": 0.92,
-      "eval_QUANTITY_f1": 0.7058823529411765,
+      "eval_QUANTITY_f1": 0.6666666666666667,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6153846153846154,
-      "eval_QUANTITY_recall": 0.8275862068965517,
-      "eval_TIME_f1": 0.8219178082191781,
+      "eval_QUANTITY_precision": 0.5945945945945946,
+      "eval_QUANTITY_recall": 0.7586206896551724,
+      "eval_TIME_f1": 0.8378378378378379,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.7692307692307693,
-      "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.13739532232284546,
-      "eval_overall_accuracy": 0.958897366477972,
-      "eval_overall_f1": 0.8020833333333333,
-      "eval_overall_precision": 0.7386091127098321,
-      "eval_overall_recall": 0.8774928774928775,
-      "eval_runtime": 0.9606,
-      "eval_samples_per_second": 194.667,
-      "eval_steps_per_second": 3.123,
+      "eval_TIME_precision": 0.775,
+      "eval_TIME_recall": 0.9117647058823529,
+      "eval_loss": 0.14279314875602722,
+      "eval_overall_accuracy": 0.9552055131676102,
+      "eval_overall_f1": 0.8015665796344648,
+      "eval_overall_precision": 0.7397590361445783,
+      "eval_overall_recall": 0.8746438746438746,
+      "eval_runtime": 0.3776,
+      "eval_samples_per_second": 495.202,
+      "eval_steps_per_second": 7.944,
       "step": 5300
     },
     {
       "epoch": 51.0,
-      "grad_norm": 1.6092292070388794,
+      "grad_norm": 0.8180804252624512,
       "learning_rate": 2.45e-05,
-      "loss": 0.0816,
+      "loss": 0.0849,
       "step": 5406
     },
     {
       "epoch": 51.0,
-      "eval_LOCATION_f1": 0.7848101265822783,
+      "eval_LOCATION_f1": 0.7870967741935484,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7294117647058823,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7189542483660131,
+      "eval_LOCATION_precision": 0.7439024390243902,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.6973684210526315,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.625,
-      "eval_ORGANIZATION_recall": 0.8461538461538461,
-      "eval_PERSON_f1": 0.8761904761904761,
+      "eval_ORGANIZATION_precision": 0.6091954022988506,
+      "eval_ORGANIZATION_recall": 0.8153846153846154,
+      "eval_PERSON_f1": 0.8817891373801918,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8363636363636363,
+      "eval_PERSON_precision": 0.8466257668711656,
       "eval_PERSON_recall": 0.92,
-      "eval_QUANTITY_f1": 0.716417910447761,
+      "eval_QUANTITY_f1": 0.6666666666666667,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.631578947368421,
-      "eval_QUANTITY_recall": 0.8275862068965517,
-      "eval_TIME_f1": 0.8219178082191781,
+      "eval_QUANTITY_precision": 0.5945945945945946,
+      "eval_QUANTITY_recall": 0.7586206896551724,
+      "eval_TIME_f1": 0.8378378378378379,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.7692307692307693,
-      "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.1346472203731537,
-      "eval_overall_accuracy": 0.9591434900319961,
-      "eval_overall_f1": 0.8067885117493472,
-      "eval_overall_precision": 0.744578313253012,
-      "eval_overall_recall": 0.8803418803418803,
-      "eval_runtime": 0.9667,
-      "eval_samples_per_second": 193.447,
-      "eval_steps_per_second": 3.103,
+      "eval_TIME_precision": 0.775,
+      "eval_TIME_recall": 0.9117647058823529,
+      "eval_loss": 0.1387655884027481,
+      "eval_overall_accuracy": 0.9564361309377307,
+      "eval_overall_f1": 0.8026315789473685,
+      "eval_overall_precision": 0.7457212713936431,
+      "eval_overall_recall": 0.8689458689458689,
+      "eval_runtime": 0.377,
+      "eval_samples_per_second": 496.079,
+      "eval_steps_per_second": 7.958,
       "step": 5406
     },
     {
       "epoch": 52.0,
-      "grad_norm": 1.4530425071716309,
+      "grad_norm": 2.047976016998291,
       "learning_rate": 2.4e-05,
-      "loss": 0.084,
+      "loss": 0.0836,
       "step": 5512
     },
     {
       "epoch": 52.0,
-      "eval_LOCATION_f1": 0.7922077922077922,
+      "eval_LOCATION_f1": 0.7672955974842768,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7530864197530864,
+      "eval_LOCATION_precision": 0.7093023255813954,
       "eval_LOCATION_recall": 0.8356164383561644,
-      "eval_ORGANIZATION_f1": 0.728476821192053,
+      "eval_ORGANIZATION_f1": 0.7019867549668873,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6395348837209303,
-      "eval_ORGANIZATION_recall": 0.8461538461538461,
-      "eval_PERSON_f1": 0.879746835443038,
+      "eval_ORGANIZATION_precision": 0.6162790697674418,
+      "eval_ORGANIZATION_recall": 0.8153846153846154,
+      "eval_PERSON_f1": 0.8846153846153846,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8373493975903614,
-      "eval_PERSON_recall": 0.9266666666666666,
-      "eval_QUANTITY_f1": 0.6956521739130435,
+      "eval_PERSON_precision": 0.8518518518518519,
+      "eval_PERSON_recall": 0.92,
+      "eval_QUANTITY_f1": 0.6865671641791046,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6,
-      "eval_QUANTITY_recall": 0.8275862068965517,
-      "eval_TIME_f1": 0.8333333333333333,
+      "eval_QUANTITY_precision": 0.6052631578947368,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8266666666666665,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.7894736842105263,
-      "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.13276532292366028,
-      "eval_overall_accuracy": 0.9598818606940684,
-      "eval_overall_f1": 0.811023622047244,
-      "eval_overall_precision": 0.7518248175182481,
-      "eval_overall_recall": 0.8803418803418803,
-      "eval_runtime": 0.9609,
-      "eval_samples_per_second": 194.599,
-      "eval_steps_per_second": 3.122,
+      "eval_TIME_precision": 0.7560975609756098,
+      "eval_TIME_recall": 0.9117647058823529,
+      "eval_loss": 0.14351429045200348,
+      "eval_overall_accuracy": 0.9554516367216342,
+      "eval_overall_f1": 0.8010471204188482,
+      "eval_overall_precision": 0.7409200968523002,
+      "eval_overall_recall": 0.8717948717948718,
+      "eval_runtime": 0.3744,
+      "eval_samples_per_second": 499.405,
+      "eval_steps_per_second": 8.012,
       "step": 5512
     },
     {
       "epoch": 53.0,
-      "grad_norm": 2.500049352645874,
+      "grad_norm": 1.9604336023330688,
       "learning_rate": 2.35e-05,
-      "loss": 0.0806,
+      "loss": 0.0838,
       "step": 5618
     },
     {
       "epoch": 53.0,
-      "eval_LOCATION_f1": 0.7721518987341772,
+      "eval_LOCATION_f1": 0.7625000000000001,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7176470588235294,
+      "eval_LOCATION_precision": 0.7011494252873564,
       "eval_LOCATION_recall": 0.8356164383561644,
-      "eval_ORGANIZATION_f1": 0.7189542483660131,
+      "eval_ORGANIZATION_f1": 0.6835443037974683,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.625,
-      "eval_ORGANIZATION_recall": 0.8461538461538461,
-      "eval_PERSON_f1": 0.8679245283018867,
+      "eval_ORGANIZATION_precision": 0.5806451612903226,
+      "eval_ORGANIZATION_recall": 0.8307692307692308,
+      "eval_PERSON_f1": 0.8789808917197452,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8214285714285714,
+      "eval_PERSON_precision": 0.8414634146341463,
       "eval_PERSON_recall": 0.92,
-      "eval_QUANTITY_f1": 0.6956521739130435,
+      "eval_QUANTITY_f1": 0.6666666666666666,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6,
-      "eval_QUANTITY_recall": 0.8275862068965517,
-      "eval_TIME_f1": 0.7945205479452054,
+      "eval_QUANTITY_precision": 0.575,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.7435897435897436,
-      "eval_TIME_recall": 0.8529411764705882,
-      "eval_loss": 0.13982656598091125,
-      "eval_overall_accuracy": 0.9576667487078513,
-      "eval_overall_f1": 0.7963683527885862,
-      "eval_overall_precision": 0.7309523809523809,
-      "eval_overall_recall": 0.8746438746438746,
-      "eval_runtime": 0.9688,
-      "eval_samples_per_second": 193.02,
-      "eval_steps_per_second": 3.097,
+      "eval_TIME_precision": 0.7317073170731707,
+      "eval_TIME_recall": 0.8823529411764706,
+      "eval_loss": 0.14942599833011627,
+      "eval_overall_accuracy": 0.9527442776273689,
+      "eval_overall_f1": 0.788659793814433,
+      "eval_overall_precision": 0.72,
+      "eval_overall_recall": 0.8717948717948718,
+      "eval_runtime": 0.3783,
+      "eval_samples_per_second": 494.359,
+      "eval_steps_per_second": 7.931,
       "step": 5618
     },
     {
       "epoch": 54.0,
-      "grad_norm": 2.2398998737335205,
+      "grad_norm": 1.6018130779266357,
       "learning_rate": 2.3000000000000003e-05,
-      "loss": 0.0827,
+      "loss": 0.0811,
       "step": 5724
     },
     {
       "epoch": 54.0,
-      "eval_LOCATION_f1": 0.7770700636942677,
+      "eval_LOCATION_f1": 0.7654320987654321,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7261904761904762,
-      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_LOCATION_precision": 0.6966292134831461,
+      "eval_LOCATION_recall": 0.8493150684931506,
       "eval_ORGANIZATION_f1": 0.6962025316455697,
       "eval_ORGANIZATION_number": 65,
       "eval_ORGANIZATION_precision": 0.5913978494623656,
       "eval_ORGANIZATION_recall": 0.8461538461538461,
-      "eval_PERSON_f1": 0.8917197452229298,
+      "eval_PERSON_f1": 0.8817891373801918,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8536585365853658,
-      "eval_PERSON_recall": 0.9333333333333333,
+      "eval_PERSON_precision": 0.8466257668711656,
+      "eval_PERSON_recall": 0.92,
       "eval_QUANTITY_f1": 0.7272727272727273,
       "eval_QUANTITY_number": 29,
       "eval_QUANTITY_precision": 0.6486486486486487,
       "eval_QUANTITY_recall": 0.8275862068965517,
-      "eval_TIME_f1": 0.8219178082191781,
+      "eval_TIME_f1": 0.8266666666666665,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.7692307692307693,
-      "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.1386086791753769,
-      "eval_overall_accuracy": 0.9586512429239479,
-      "eval_overall_f1": 0.8072916666666667,
-      "eval_overall_precision": 0.7434052757793765,
+      "eval_TIME_precision": 0.7560975609756098,
+      "eval_TIME_recall": 0.9117647058823529,
+      "eval_loss": 0.14617691934108734,
+      "eval_overall_accuracy": 0.9556977602756583,
+      "eval_overall_f1": 0.8010335917312662,
+      "eval_overall_precision": 0.7328605200945626,
       "eval_overall_recall": 0.8831908831908832,
-      "eval_runtime": 0.9542,
-      "eval_samples_per_second": 195.97,
-      "eval_steps_per_second": 3.144,
+      "eval_runtime": 0.3727,
+      "eval_samples_per_second": 501.802,
+      "eval_steps_per_second": 8.05,
       "step": 5724
     },
     {
       "epoch": 55.0,
-      "grad_norm": 0.9107390642166138,
+      "grad_norm": 1.0216081142425537,
       "learning_rate": 2.25e-05,
-      "loss": 0.0816,
+      "loss": 0.0826,
       "step": 5830
     },
     {
       "epoch": 55.0,
-      "eval_LOCATION_f1": 0.7870967741935484,
+      "eval_LOCATION_f1": 0.7770700636942677,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7439024390243902,
+      "eval_LOCATION_precision": 0.7261904761904762,
       "eval_LOCATION_recall": 0.8356164383561644,
-      "eval_ORGANIZATION_f1": 0.7189542483660131,
+      "eval_ORGANIZATION_f1": 0.6666666666666666,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.625,
-      "eval_ORGANIZATION_recall": 0.8461538461538461,
-      "eval_PERSON_f1": 0.8853503184713376,
+      "eval_ORGANIZATION_precision": 0.5714285714285714,
+      "eval_ORGANIZATION_recall": 0.8,
+      "eval_PERSON_f1": 0.8817891373801918,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8475609756097561,
-      "eval_PERSON_recall": 0.9266666666666666,
-      "eval_QUANTITY_f1": 0.6857142857142856,
+      "eval_PERSON_precision": 0.8466257668711656,
+      "eval_PERSON_recall": 0.92,
+      "eval_QUANTITY_f1": 0.6865671641791046,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.5853658536585366,
-      "eval_QUANTITY_recall": 0.8275862068965517,
-      "eval_TIME_f1": 0.8219178082191781,
+      "eval_QUANTITY_precision": 0.6052631578947368,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8266666666666665,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.7692307692307693,
-      "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.13677756488323212,
-      "eval_overall_accuracy": 0.958897366477972,
-      "eval_overall_f1": 0.807843137254902,
-      "eval_overall_precision": 0.7463768115942029,
-      "eval_overall_recall": 0.8803418803418803,
-      "eval_runtime": 0.9619,
-      "eval_samples_per_second": 194.412,
-      "eval_steps_per_second": 3.119,
+      "eval_TIME_precision": 0.7560975609756098,
+      "eval_TIME_recall": 0.9117647058823529,
+      "eval_loss": 0.14629320800304413,
+      "eval_overall_accuracy": 0.9534826482894413,
+      "eval_overall_f1": 0.7942708333333333,
+      "eval_overall_precision": 0.7314148681055156,
+      "eval_overall_recall": 0.8689458689458689,
+      "eval_runtime": 0.373,
+      "eval_samples_per_second": 501.397,
+      "eval_steps_per_second": 8.044,
       "step": 5830
     },
     {
       "epoch": 56.0,
-      "grad_norm": 2.796041965484619,
+      "grad_norm": 1.5277239084243774,
       "learning_rate": 2.2000000000000003e-05,
-      "loss": 0.0828,
+      "loss": 0.0818,
       "step": 5936
     },
     {
       "epoch": 56.0,
-      "eval_LOCATION_f1": 0.7922077922077922,
+      "eval_LOCATION_f1": 0.7898089171974523,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7530864197530864,
-      "eval_LOCATION_recall": 0.8356164383561644,
-      "eval_ORGANIZATION_f1": 0.7189542483660131,
+      "eval_LOCATION_precision": 0.7380952380952381,
+      "eval_LOCATION_recall": 0.8493150684931506,
+      "eval_ORGANIZATION_f1": 0.6794871794871795,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.625,
-      "eval_ORGANIZATION_recall": 0.8461538461538461,
-      "eval_PERSON_f1": 0.8742138364779874,
-      "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8273809523809523,
-      "eval_PERSON_recall": 0.9266666666666666,
-      "eval_QUANTITY_f1": 0.7692307692307692,
+      "eval_ORGANIZATION_precision": 0.5824175824175825,
+      "eval_ORGANIZATION_recall": 0.8153846153846154,
+      "eval_PERSON_f1": 0.8789808917197452,
+      "eval_PERSON_number": 150,
+      "eval_PERSON_precision": 0.8414634146341463,
+      "eval_PERSON_recall": 0.92,
+      "eval_QUANTITY_f1": 0.696969696969697,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6944444444444444,
-      "eval_QUANTITY_recall": 0.8620689655172413,
-      "eval_TIME_f1": 0.861111111111111,
+      "eval_QUANTITY_precision": 0.6216216216216216,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8378378378378379,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8157894736842105,
+      "eval_TIME_precision": 0.775,
       "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.13076351583003998,
-      "eval_overall_accuracy": 0.9613586020182131,
-      "eval_overall_f1": 0.8162729658792651,
-      "eval_overall_precision": 0.7566909975669099,
-      "eval_overall_recall": 0.886039886039886,
-      "eval_runtime": 0.9668,
-      "eval_samples_per_second": 193.426,
-      "eval_steps_per_second": 3.103,
+      "eval_loss": 0.14746278524398804,
+      "eval_overall_accuracy": 0.9554516367216342,
+      "eval_overall_f1": 0.8005215123859191,
+      "eval_overall_precision": 0.7379807692307693,
+      "eval_overall_recall": 0.8746438746438746,
+      "eval_runtime": 0.3747,
+      "eval_samples_per_second": 499.009,
+      "eval_steps_per_second": 8.005,
       "step": 5936
     },
     {
       "epoch": 57.0,
-      "grad_norm": 1.4414466619491577,
+      "grad_norm": 1.1813913583755493,
       "learning_rate": 2.15e-05,
-      "loss": 0.0813,
+      "loss": 0.0833,
       "step": 6042
     },
     {
       "epoch": 57.0,
-      "eval_LOCATION_f1": 0.8104575163398693,
+      "eval_LOCATION_f1": 0.7692307692307693,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.775,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7114093959731543,
+      "eval_LOCATION_precision": 0.7228915662650602,
+      "eval_LOCATION_recall": 0.821917808219178,
+      "eval_ORGANIZATION_f1": 0.6792452830188681,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6309523809523809,
-      "eval_ORGANIZATION_recall": 0.8153846153846154,
-      "eval_PERSON_f1": 0.8860759493670886,
+      "eval_ORGANIZATION_precision": 0.574468085106383,
+      "eval_ORGANIZATION_recall": 0.8307692307692308,
+      "eval_PERSON_f1": 0.8734177215189873,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8433734939759037,
-      "eval_PERSON_recall": 0.9333333333333333,
-      "eval_QUANTITY_f1": 0.7384615384615385,
+      "eval_PERSON_precision": 0.8313253012048193,
+      "eval_PERSON_recall": 0.92,
+      "eval_QUANTITY_f1": 0.6865671641791046,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6666666666666666,
-      "eval_QUANTITY_recall": 0.8275862068965517,
-      "eval_TIME_f1": 0.8378378378378379,
+      "eval_QUANTITY_precision": 0.6052631578947368,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8266666666666665,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.775,
+      "eval_TIME_precision": 0.7560975609756098,
       "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.12566277384757996,
-      "eval_overall_accuracy": 0.961112478464189,
-      "eval_overall_f1": 0.8190224570673711,
-      "eval_overall_precision": 0.7635467980295566,
-      "eval_overall_recall": 0.8831908831908832,
-      "eval_runtime": 1.0243,
-      "eval_samples_per_second": 182.567,
-      "eval_steps_per_second": 2.929,
+      "eval_loss": 0.1485372930765152,
+      "eval_overall_accuracy": 0.9529904011813931,
+      "eval_overall_f1": 0.7917205692108668,
+      "eval_overall_precision": 0.7251184834123223,
+      "eval_overall_recall": 0.8717948717948718,
+      "eval_runtime": 0.3727,
+      "eval_samples_per_second": 501.711,
+      "eval_steps_per_second": 8.049,
       "step": 6042
     },
     {
       "epoch": 58.0,
-      "grad_norm": 0.5744277834892273,
+      "grad_norm": 0.640755295753479,
       "learning_rate": 2.1e-05,
-      "loss": 0.0797,
+      "loss": 0.0798,
       "step": 6148
     },
     {
       "epoch": 58.0,
-      "eval_LOCATION_f1": 0.8051948051948052,
+      "eval_LOCATION_f1": 0.7692307692307693,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7654320987654321,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7027027027027027,
+      "eval_LOCATION_precision": 0.7228915662650602,
+      "eval_LOCATION_recall": 0.821917808219178,
+      "eval_ORGANIZATION_f1": 0.697986577181208,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6265060240963856,
+      "eval_ORGANIZATION_precision": 0.6190476190476191,
       "eval_ORGANIZATION_recall": 0.8,
-      "eval_PERSON_f1": 0.8832807570977917,
+      "eval_PERSON_f1": 0.887459807073955,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8383233532934131,
-      "eval_PERSON_recall": 0.9333333333333333,
-      "eval_QUANTITY_f1": 0.6956521739130435,
+      "eval_PERSON_precision": 0.8571428571428571,
+      "eval_PERSON_recall": 0.92,
+      "eval_QUANTITY_f1": 0.71875,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6,
-      "eval_QUANTITY_recall": 0.8275862068965517,
-      "eval_TIME_f1": 0.7837837837837837,
+      "eval_QUANTITY_precision": 0.6571428571428571,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8378378378378379,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.725,
-      "eval_TIME_recall": 0.8529411764705882,
-      "eval_loss": 0.12857982516288757,
-      "eval_overall_accuracy": 0.958897366477972,
-      "eval_overall_f1": 0.8057742782152231,
-      "eval_overall_precision": 0.7469586374695864,
-      "eval_overall_recall": 0.8746438746438746,
-      "eval_runtime": 0.9638,
-      "eval_samples_per_second": 194.014,
-      "eval_steps_per_second": 3.113,
+      "eval_TIME_precision": 0.775,
+      "eval_TIME_recall": 0.9117647058823529,
+      "eval_loss": 0.14086660742759705,
+      "eval_overall_accuracy": 0.9564361309377307,
+      "eval_overall_f1": 0.8063660477453581,
+      "eval_overall_precision": 0.7543424317617866,
+      "eval_overall_recall": 0.8660968660968661,
+      "eval_runtime": 0.3766,
+      "eval_samples_per_second": 496.571,
+      "eval_steps_per_second": 7.966,
       "step": 6148
     },
     {
       "epoch": 59.0,
-      "grad_norm": 2.238668203353882,
+      "grad_norm": 2.363363265991211,
       "learning_rate": 2.05e-05,
-      "loss": 0.0782,
+      "loss": 0.0802,
       "step": 6254
     },
     {
       "epoch": 59.0,
-      "eval_LOCATION_f1": 0.7870967741935484,
+      "eval_LOCATION_f1": 0.7770700636942677,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7439024390243902,
+      "eval_LOCATION_precision": 0.7261904761904762,
       "eval_LOCATION_recall": 0.8356164383561644,
-      "eval_ORGANIZATION_f1": 0.7333333333333334,
+      "eval_ORGANIZATION_f1": 0.6883116883116883,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6470588235294118,
-      "eval_ORGANIZATION_recall": 0.8461538461538461,
-      "eval_PERSON_f1": 0.8832807570977917,
+      "eval_ORGANIZATION_precision": 0.5955056179775281,
+      "eval_ORGANIZATION_recall": 0.8153846153846154,
+      "eval_PERSON_f1": 0.887459807073955,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8383233532934131,
-      "eval_PERSON_recall": 0.9333333333333333,
-      "eval_QUANTITY_f1": 0.7272727272727273,
+      "eval_PERSON_precision": 0.8571428571428571,
+      "eval_PERSON_recall": 0.92,
+      "eval_QUANTITY_f1": 0.7076923076923076,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6486486486486487,
-      "eval_QUANTITY_recall": 0.8275862068965517,
-      "eval_TIME_f1": 0.8219178082191781,
+      "eval_QUANTITY_precision": 0.6388888888888888,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8266666666666665,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.7692307692307693,
-      "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.13031935691833496,
-      "eval_overall_accuracy": 0.9608663549101649,
-      "eval_overall_f1": 0.8147174770039421,
-      "eval_overall_precision": 0.7560975609756098,
-      "eval_overall_recall": 0.8831908831908832,
-      "eval_runtime": 0.9586,
-      "eval_samples_per_second": 195.077,
-      "eval_steps_per_second": 3.13,
+      "eval_TIME_precision": 0.7560975609756098,
+      "eval_TIME_recall": 0.9117647058823529,
+      "eval_loss": 0.14503292739391327,
+      "eval_overall_accuracy": 0.9556977602756583,
+      "eval_overall_f1": 0.8031496062992126,
+      "eval_overall_precision": 0.7445255474452555,
+      "eval_overall_recall": 0.8717948717948718,
+      "eval_runtime": 0.374,
+      "eval_samples_per_second": 499.944,
+      "eval_steps_per_second": 8.02,
       "step": 6254
     },
     {
       "epoch": 60.0,
-      "grad_norm": 1.372519612312317,
+      "grad_norm": 1.6086369752883911,
       "learning_rate": 2e-05,
-      "loss": 0.0804,
+      "loss": 0.08,
       "step": 6360
     },
     {
       "epoch": 60.0,
-      "eval_LOCATION_f1": 0.8104575163398693,
+      "eval_LOCATION_f1": 0.7999999999999999,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.775,
+      "eval_LOCATION_precision": 0.7560975609756098,
       "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7189542483660131,
+      "eval_ORGANIZATION_f1": 0.7019867549668873,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.625,
-      "eval_ORGANIZATION_recall": 0.8461538461538461,
-      "eval_PERSON_f1": 0.8769716088328076,
+      "eval_ORGANIZATION_precision": 0.6162790697674418,
+      "eval_ORGANIZATION_recall": 0.8153846153846154,
+      "eval_PERSON_f1": 0.887459807073955,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8323353293413174,
-      "eval_PERSON_recall": 0.9266666666666666,
-      "eval_QUANTITY_f1": 0.7692307692307692,
+      "eval_PERSON_precision": 0.8571428571428571,
+      "eval_PERSON_recall": 0.92,
+      "eval_QUANTITY_f1": 0.6875,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6944444444444444,
-      "eval_QUANTITY_recall": 0.8620689655172413,
-      "eval_TIME_f1": 0.861111111111111,
+      "eval_QUANTITY_precision": 0.6285714285714286,
+      "eval_QUANTITY_recall": 0.7586206896551724,
+      "eval_TIME_f1": 0.8533333333333334,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8157894736842105,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.12692256271839142,
-      "eval_overall_accuracy": 0.961112478464189,
-      "eval_overall_f1": 0.8210526315789473,
-      "eval_overall_precision": 0.7628361858190709,
-      "eval_overall_recall": 0.8888888888888888,
-      "eval_runtime": 0.9613,
-      "eval_samples_per_second": 194.533,
-      "eval_steps_per_second": 3.121,
+      "eval_TIME_precision": 0.7804878048780488,
+      "eval_TIME_recall": 0.9411764705882353,
+      "eval_loss": 0.1365276426076889,
+      "eval_overall_accuracy": 0.9591434900319961,
+      "eval_overall_f1": 0.8121693121693121,
+      "eval_overall_precision": 0.7580246913580246,
+      "eval_overall_recall": 0.8746438746438746,
+      "eval_runtime": 0.3728,
+      "eval_samples_per_second": 501.578,
+      "eval_steps_per_second": 8.047,
       "step": 6360
     },
     {
       "epoch": 61.0,
-      "grad_norm": 2.849003553390503,
+      "grad_norm": 1.1481249332427979,
       "learning_rate": 1.9500000000000003e-05,
-      "loss": 0.0777,
+      "loss": 0.0792,
       "step": 6466
     },
     {
       "epoch": 61.0,
-      "eval_LOCATION_f1": 0.7922077922077922,
+      "eval_LOCATION_f1": 0.7692307692307693,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7530864197530864,
-      "eval_LOCATION_recall": 0.8356164383561644,
-      "eval_ORGANIZATION_f1": 0.7236842105263158,
+      "eval_LOCATION_precision": 0.7228915662650602,
+      "eval_LOCATION_recall": 0.821917808219178,
+      "eval_ORGANIZATION_f1": 0.7152317880794703,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.632183908045977,
-      "eval_ORGANIZATION_recall": 0.8461538461538461,
-      "eval_PERSON_f1": 0.8742138364779874,
+      "eval_ORGANIZATION_precision": 0.627906976744186,
+      "eval_ORGANIZATION_recall": 0.8307692307692308,
+      "eval_PERSON_f1": 0.8846153846153846,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8273809523809523,
-      "eval_PERSON_recall": 0.9266666666666666,
-      "eval_QUANTITY_f1": 0.71875,
+      "eval_PERSON_precision": 0.8518518518518519,
+      "eval_PERSON_recall": 0.92,
+      "eval_QUANTITY_f1": 0.6875,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6571428571428571,
-      "eval_QUANTITY_recall": 0.7931034482758621,
-      "eval_TIME_f1": 0.8732394366197184,
+      "eval_QUANTITY_precision": 0.6285714285714286,
+      "eval_QUANTITY_recall": 0.7586206896551724,
+      "eval_TIME_f1": 0.8533333333333334,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8378378378378378,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.12815138697624207,
-      "eval_overall_accuracy": 0.961112478464189,
-      "eval_overall_f1": 0.8142292490118577,
-      "eval_overall_precision": 0.7573529411764706,
-      "eval_overall_recall": 0.8803418803418803,
-      "eval_runtime": 0.9568,
-      "eval_samples_per_second": 195.448,
-      "eval_steps_per_second": 3.136,
+      "eval_TIME_precision": 0.7804878048780488,
+      "eval_TIME_recall": 0.9411764705882353,
+      "eval_loss": 0.13838407397270203,
+      "eval_overall_accuracy": 0.9576667487078513,
+      "eval_overall_f1": 0.8073878627968336,
+      "eval_overall_precision": 0.7518427518427518,
+      "eval_overall_recall": 0.8717948717948718,
+      "eval_runtime": 0.3773,
+      "eval_samples_per_second": 495.651,
+      "eval_steps_per_second": 7.952,
       "step": 6466
     },
     {
       "epoch": 62.0,
-      "grad_norm": 1.4931484460830688,
+      "grad_norm": 1.3433496952056885,
       "learning_rate": 1.9e-05,
-      "loss": 0.0769,
+      "loss": 0.0776,
       "step": 6572
     },
     {
       "epoch": 62.0,
-      "eval_LOCATION_f1": 0.7922077922077922,
+      "eval_LOCATION_f1": 0.7692307692307693,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7530864197530864,
-      "eval_LOCATION_recall": 0.8356164383561644,
-      "eval_ORGANIZATION_f1": 0.7236842105263158,
+      "eval_LOCATION_precision": 0.7228915662650602,
+      "eval_LOCATION_recall": 0.821917808219178,
+      "eval_ORGANIZATION_f1": 0.7222222222222222,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.632183908045977,
-      "eval_ORGANIZATION_recall": 0.8461538461538461,
-      "eval_PERSON_f1": 0.8825396825396825,
+      "eval_ORGANIZATION_precision": 0.6582278481012658,
+      "eval_ORGANIZATION_recall": 0.8,
+      "eval_PERSON_f1": 0.8846153846153846,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8424242424242424,
-      "eval_PERSON_recall": 0.9266666666666666,
-      "eval_QUANTITY_f1": 0.6956521739130435,
+      "eval_PERSON_precision": 0.8518518518518519,
+      "eval_PERSON_recall": 0.92,
+      "eval_QUANTITY_f1": 0.696969696969697,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6,
-      "eval_QUANTITY_recall": 0.8275862068965517,
-      "eval_TIME_f1": 0.8333333333333333,
+      "eval_QUANTITY_precision": 0.6216216216216216,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8421052631578947,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.7894736842105263,
-      "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.13441848754882812,
-      "eval_overall_accuracy": 0.961112478464189,
-      "eval_overall_f1": 0.811023622047244,
-      "eval_overall_precision": 0.7518248175182481,
-      "eval_overall_recall": 0.8803418803418803,
-      "eval_runtime": 0.9895,
-      "eval_samples_per_second": 188.992,
-      "eval_steps_per_second": 3.032,
+      "eval_TIME_precision": 0.7619047619047619,
+      "eval_TIME_recall": 0.9411764705882353,
+      "eval_loss": 0.14121700823307037,
+      "eval_overall_accuracy": 0.9576667487078513,
+      "eval_overall_f1": 0.8090185676392573,
+      "eval_overall_precision": 0.7568238213399504,
+      "eval_overall_recall": 0.8689458689458689,
+      "eval_runtime": 0.3749,
+      "eval_samples_per_second": 498.784,
+      "eval_steps_per_second": 8.002,
       "step": 6572
     },
     {
       "epoch": 63.0,
-      "grad_norm": 7.7728190422058105,
+      "grad_norm": 2.4079196453094482,
       "learning_rate": 1.85e-05,
-      "loss": 0.0766,
+      "loss": 0.0782,
       "step": 6678
     },
     {
       "epoch": 63.0,
-      "eval_LOCATION_f1": 0.8104575163398693,
+      "eval_LOCATION_f1": 0.7770700636942677,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.775,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7189542483660131,
+      "eval_LOCATION_precision": 0.7261904761904762,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.6878980891719746,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.625,
-      "eval_ORGANIZATION_recall": 0.8461538461538461,
-      "eval_PERSON_f1": 0.8832807570977917,
+      "eval_ORGANIZATION_precision": 0.5869565217391305,
+      "eval_ORGANIZATION_recall": 0.8307692307692308,
+      "eval_PERSON_f1": 0.887459807073955,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8383233532934131,
-      "eval_PERSON_recall": 0.9333333333333333,
-      "eval_QUANTITY_f1": 0.7384615384615385,
+      "eval_PERSON_precision": 0.8571428571428571,
+      "eval_PERSON_recall": 0.92,
+      "eval_QUANTITY_f1": 0.7076923076923076,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6666666666666666,
-      "eval_QUANTITY_recall": 0.8275862068965517,
-      "eval_TIME_f1": 0.861111111111111,
+      "eval_QUANTITY_precision": 0.6388888888888888,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8421052631578947,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8157894736842105,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.13149088621139526,
-      "eval_overall_accuracy": 0.9613586020182131,
-      "eval_overall_f1": 0.8210526315789473,
-      "eval_overall_precision": 0.7628361858190709,
-      "eval_overall_recall": 0.8888888888888888,
-      "eval_runtime": 0.9572,
-      "eval_samples_per_second": 195.358,
-      "eval_steps_per_second": 3.134,
+      "eval_TIME_precision": 0.7619047619047619,
+      "eval_TIME_recall": 0.9411764705882353,
+      "eval_loss": 0.1416482925415039,
+      "eval_overall_accuracy": 0.9554516367216342,
+      "eval_overall_f1": 0.804177545691906,
+      "eval_overall_precision": 0.7421686746987952,
+      "eval_overall_recall": 0.8774928774928775,
+      "eval_runtime": 0.3756,
+      "eval_samples_per_second": 497.871,
+      "eval_steps_per_second": 7.987,
       "step": 6678
     },
     {
       "epoch": 64.0,
-      "grad_norm": 1.2526227235794067,
+      "grad_norm": 0.9945477247238159,
       "learning_rate": 1.8e-05,
-      "loss": 0.0768,
+      "loss": 0.0789,
       "step": 6784
     },
     {
       "epoch": 64.0,
-      "eval_LOCATION_f1": 0.8104575163398693,
+      "eval_LOCATION_f1": 0.7770700636942677,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.775,
-      "eval_LOCATION_recall": 0.8493150684931506,
+      "eval_LOCATION_precision": 0.7261904761904762,
+      "eval_LOCATION_recall": 0.8356164383561644,
       "eval_ORGANIZATION_f1": 0.7123287671232877,
       "eval_ORGANIZATION_number": 65,
       "eval_ORGANIZATION_precision": 0.6419753086419753,
       "eval_ORGANIZATION_recall": 0.8,
-      "eval_PERSON_f1": 0.8825396825396825,
+      "eval_PERSON_f1": 0.887459807073955,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8424242424242424,
-      "eval_PERSON_recall": 0.9266666666666666,
-      "eval_QUANTITY_f1": 0.7619047619047619,
+      "eval_PERSON_precision": 0.8571428571428571,
+      "eval_PERSON_recall": 0.92,
+      "eval_QUANTITY_f1": 0.6875,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.7058823529411765,
-      "eval_QUANTITY_recall": 0.8275862068965517,
-      "eval_TIME_f1": 0.861111111111111,
+      "eval_QUANTITY_precision": 0.6285714285714286,
+      "eval_QUANTITY_recall": 0.7586206896551724,
+      "eval_TIME_f1": 0.8421052631578947,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8157894736842105,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.12420864403247833,
-      "eval_overall_accuracy": 0.9618508491262614,
-      "eval_overall_f1": 0.822429906542056,
-      "eval_overall_precision": 0.7738693467336684,
-      "eval_overall_recall": 0.8774928774928775,
-      "eval_runtime": 0.9596,
-      "eval_samples_per_second": 194.871,
-      "eval_steps_per_second": 3.126,
+      "eval_TIME_precision": 0.7619047619047619,
+      "eval_TIME_recall": 0.9411764705882353,
+      "eval_loss": 0.13378003239631653,
+      "eval_overall_accuracy": 0.9581589958158996,
+      "eval_overall_f1": 0.8090185676392573,
+      "eval_overall_precision": 0.7568238213399504,
+      "eval_overall_recall": 0.8689458689458689,
+      "eval_runtime": 0.3716,
+      "eval_samples_per_second": 503.164,
+      "eval_steps_per_second": 8.072,
       "step": 6784
     },
     {
       "epoch": 65.0,
-      "grad_norm": 1.5504401922225952,
+      "grad_norm": 1.0777437686920166,
       "learning_rate": 1.75e-05,
-      "loss": 0.0772,
+      "loss": 0.0769,
       "step": 6890
     },
     {
       "epoch": 65.0,
-      "eval_LOCATION_f1": 0.7999999999999999,
+      "eval_LOCATION_f1": 0.7721518987341772,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7560975609756098,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.738255033557047,
+      "eval_LOCATION_precision": 0.7176470588235294,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.7123287671232877,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6547619047619048,
-      "eval_ORGANIZATION_recall": 0.8461538461538461,
-      "eval_PERSON_f1": 0.8789808917197452,
+      "eval_ORGANIZATION_precision": 0.6419753086419753,
+      "eval_ORGANIZATION_recall": 0.8,
+      "eval_PERSON_f1": 0.887459807073955,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8414634146341463,
+      "eval_PERSON_precision": 0.8571428571428571,
       "eval_PERSON_recall": 0.92,
-      "eval_QUANTITY_f1": 0.716417910447761,
+      "eval_QUANTITY_f1": 0.7076923076923076,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.631578947368421,
-      "eval_QUANTITY_recall": 0.8275862068965517,
-      "eval_TIME_f1": 0.861111111111111,
+      "eval_QUANTITY_precision": 0.6388888888888888,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8266666666666665,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8157894736842105,
+      "eval_TIME_precision": 0.7560975609756098,
       "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.12913870811462402,
-      "eval_overall_accuracy": 0.9613586020182131,
-      "eval_overall_f1": 0.8190224570673711,
-      "eval_overall_precision": 0.7635467980295566,
-      "eval_overall_recall": 0.8831908831908832,
-      "eval_runtime": 0.9536,
-      "eval_samples_per_second": 196.097,
-      "eval_steps_per_second": 3.146,
+      "eval_loss": 0.13960254192352295,
+      "eval_overall_accuracy": 0.9579128722618755,
+      "eval_overall_f1": 0.8079470198675496,
+      "eval_overall_precision": 0.754950495049505,
+      "eval_overall_recall": 0.8689458689458689,
+      "eval_runtime": 0.3755,
+      "eval_samples_per_second": 497.937,
+      "eval_steps_per_second": 7.988,
       "step": 6890
     },
     {
       "epoch": 66.0,
-      "grad_norm": 1.0265319347381592,
+      "grad_norm": 0.8647677302360535,
       "learning_rate": 1.7000000000000003e-05,
-      "loss": 0.0738,
+      "loss": 0.0757,
       "step": 6996
     },
     {
       "epoch": 66.0,
-      "eval_LOCATION_f1": 0.7820512820512822,
+      "eval_LOCATION_f1": 0.7770700636942677,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7349397590361446,
+      "eval_LOCATION_precision": 0.7261904761904762,
       "eval_LOCATION_recall": 0.8356164383561644,
-      "eval_ORGANIZATION_f1": 0.7236842105263158,
+      "eval_ORGANIZATION_f1": 0.7019867549668873,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.632183908045977,
-      "eval_ORGANIZATION_recall": 0.8461538461538461,
-      "eval_PERSON_f1": 0.8714733542319748,
+      "eval_ORGANIZATION_precision": 0.6162790697674418,
+      "eval_ORGANIZATION_recall": 0.8153846153846154,
+      "eval_PERSON_f1": 0.8846153846153846,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8224852071005917,
-      "eval_PERSON_recall": 0.9266666666666666,
-      "eval_QUANTITY_f1": 0.746268656716418,
+      "eval_PERSON_precision": 0.8518518518518519,
+      "eval_PERSON_recall": 0.92,
+      "eval_QUANTITY_f1": 0.676923076923077,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6578947368421053,
-      "eval_QUANTITY_recall": 0.8620689655172413,
-      "eval_TIME_f1": 0.8493150684931507,
+      "eval_QUANTITY_precision": 0.6111111111111112,
+      "eval_QUANTITY_recall": 0.7586206896551724,
+      "eval_TIME_f1": 0.8421052631578947,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.7948717948717948,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.13372032344341278,
-      "eval_overall_accuracy": 0.9601279842480925,
-      "eval_overall_f1": 0.8109517601043025,
-      "eval_overall_precision": 0.7475961538461539,
-      "eval_overall_recall": 0.886039886039886,
-      "eval_runtime": 0.955,
-      "eval_samples_per_second": 195.812,
-      "eval_steps_per_second": 3.141,
+      "eval_TIME_precision": 0.7619047619047619,
+      "eval_TIME_recall": 0.9411764705882353,
+      "eval_loss": 0.1402267962694168,
+      "eval_overall_accuracy": 0.9576667487078513,
+      "eval_overall_f1": 0.8042049934296978,
+      "eval_overall_precision": 0.7463414634146341,
+      "eval_overall_recall": 0.8717948717948718,
+      "eval_runtime": 0.3717,
+      "eval_samples_per_second": 503.11,
+      "eval_steps_per_second": 8.071,
       "step": 6996
     },
     {
       "epoch": 67.0,
-      "grad_norm": 1.6299247741699219,
+      "grad_norm": 1.0575724840164185,
       "learning_rate": 1.65e-05,
-      "loss": 0.0752,
+      "loss": 0.0753,
       "step": 7102
     },
     {
       "epoch": 67.0,
-      "eval_LOCATION_f1": 0.7999999999999999,
+      "eval_LOCATION_f1": 0.7820512820512822,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7560975609756098,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7297297297297298,
+      "eval_LOCATION_precision": 0.7349397590361446,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.7210884353741497,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6506024096385542,
-      "eval_ORGANIZATION_recall": 0.8307692307692308,
-      "eval_PERSON_f1": 0.8917197452229298,
+      "eval_ORGANIZATION_precision": 0.6463414634146342,
+      "eval_ORGANIZATION_recall": 0.8153846153846154,
+      "eval_PERSON_f1": 0.887459807073955,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8536585365853658,
-      "eval_PERSON_recall": 0.9333333333333333,
-      "eval_QUANTITY_f1": 0.7272727272727273,
+      "eval_PERSON_precision": 0.8571428571428571,
+      "eval_PERSON_recall": 0.92,
+      "eval_QUANTITY_f1": 0.7076923076923076,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6486486486486487,
-      "eval_QUANTITY_recall": 0.8275862068965517,
-      "eval_TIME_f1": 0.8493150684931507,
+      "eval_QUANTITY_precision": 0.6388888888888888,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8533333333333334,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.7948717948717948,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.12958921492099762,
-      "eval_overall_accuracy": 0.9613586020182131,
-      "eval_overall_f1": 0.8227513227513228,
-      "eval_overall_precision": 0.7679012345679013,
-      "eval_overall_recall": 0.886039886039886,
-      "eval_runtime": 0.955,
-      "eval_samples_per_second": 195.809,
-      "eval_steps_per_second": 3.141,
+      "eval_TIME_precision": 0.7804878048780488,
+      "eval_TIME_recall": 0.9411764705882353,
+      "eval_loss": 0.13897380232810974,
+      "eval_overall_accuracy": 0.9593896135860202,
+      "eval_overall_f1": 0.8143236074270558,
+      "eval_overall_precision": 0.7617866004962779,
+      "eval_overall_recall": 0.8746438746438746,
+      "eval_runtime": 0.3722,
+      "eval_samples_per_second": 502.482,
+      "eval_steps_per_second": 8.061,
       "step": 7102
     },
     {
       "epoch": 68.0,
-      "grad_norm": 1.4963232278823853,
+      "grad_norm": 0.9653581976890564,
       "learning_rate": 1.6000000000000003e-05,
-      "loss": 0.0743,
+      "loss": 0.0757,
       "step": 7208
     },
     {
@@ -2634,1288 +2634,1288 @@
       "eval_LOCATION_number": 73,
       "eval_LOCATION_precision": 0.7439024390243902,
       "eval_LOCATION_recall": 0.8356164383561644,
-      "eval_ORGANIZATION_f1": 0.7074829931972789,
+      "eval_ORGANIZATION_f1": 0.7172413793103448,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6341463414634146,
+      "eval_ORGANIZATION_precision": 0.65,
       "eval_ORGANIZATION_recall": 0.8,
-      "eval_PERSON_f1": 0.8945686900958466,
+      "eval_PERSON_f1": 0.887459807073955,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8588957055214724,
-      "eval_PERSON_recall": 0.9333333333333333,
-      "eval_QUANTITY_f1": 0.696969696969697,
+      "eval_PERSON_precision": 0.8571428571428571,
+      "eval_PERSON_recall": 0.92,
+      "eval_QUANTITY_f1": 0.7076923076923076,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6216216216216216,
+      "eval_QUANTITY_precision": 0.6388888888888888,
       "eval_QUANTITY_recall": 0.7931034482758621,
-      "eval_TIME_f1": 0.8219178082191781,
+      "eval_TIME_f1": 0.8533333333333334,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.7692307692307693,
-      "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.13297563791275024,
-      "eval_overall_accuracy": 0.9606202313561408,
-      "eval_overall_f1": 0.8116710875331565,
-      "eval_overall_precision": 0.7593052109181141,
+      "eval_TIME_precision": 0.7804878048780488,
+      "eval_TIME_recall": 0.9411764705882353,
+      "eval_loss": 0.13793830573558807,
+      "eval_overall_accuracy": 0.9591434900319961,
+      "eval_overall_f1": 0.8149134487350199,
+      "eval_overall_precision": 0.765,
       "eval_overall_recall": 0.8717948717948718,
-      "eval_runtime": 0.9533,
-      "eval_samples_per_second": 196.168,
-      "eval_steps_per_second": 3.147,
+      "eval_runtime": 0.3738,
+      "eval_samples_per_second": 500.215,
+      "eval_steps_per_second": 8.025,
       "step": 7208
     },
     {
       "epoch": 69.0,
-      "grad_norm": 1.4221112728118896,
+      "grad_norm": 1.3589030504226685,
       "learning_rate": 1.55e-05,
-      "loss": 0.0758,
+      "loss": 0.0731,
       "step": 7314
     },
     {
       "epoch": 69.0,
-      "eval_LOCATION_f1": 0.8051948051948052,
+      "eval_LOCATION_f1": 0.7770700636942677,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7654320987654321,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.728476821192053,
+      "eval_LOCATION_precision": 0.7261904761904762,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.7320261437908497,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6395348837209303,
-      "eval_ORGANIZATION_recall": 0.8461538461538461,
-      "eval_PERSON_f1": 0.8825396825396825,
+      "eval_ORGANIZATION_precision": 0.6363636363636364,
+      "eval_ORGANIZATION_recall": 0.8615384615384616,
+      "eval_PERSON_f1": 0.887459807073955,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8424242424242424,
-      "eval_PERSON_recall": 0.9266666666666666,
+      "eval_PERSON_precision": 0.8571428571428571,
+      "eval_PERSON_recall": 0.92,
       "eval_QUANTITY_f1": 0.71875,
       "eval_QUANTITY_number": 29,
       "eval_QUANTITY_precision": 0.6571428571428571,
       "eval_QUANTITY_recall": 0.7931034482758621,
-      "eval_TIME_f1": 0.8493150684931507,
+      "eval_TIME_f1": 0.8533333333333334,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.7948717948717948,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.1320372223854065,
-      "eval_overall_accuracy": 0.9613586020182131,
-      "eval_overall_f1": 0.8190224570673711,
-      "eval_overall_precision": 0.7635467980295566,
+      "eval_TIME_precision": 0.7804878048780488,
+      "eval_TIME_recall": 0.9411764705882353,
+      "eval_loss": 0.14166338741779327,
+      "eval_overall_accuracy": 0.9581589958158996,
+      "eval_overall_f1": 0.8157894736842105,
+      "eval_overall_precision": 0.7579462102689487,
       "eval_overall_recall": 0.8831908831908832,
-      "eval_runtime": 0.9574,
-      "eval_samples_per_second": 195.314,
-      "eval_steps_per_second": 3.133,
+      "eval_runtime": 0.3735,
+      "eval_samples_per_second": 500.684,
+      "eval_steps_per_second": 8.032,
       "step": 7314
     },
     {
       "epoch": 70.0,
-      "grad_norm": 0.8428576588630676,
+      "grad_norm": 1.4494452476501465,
       "learning_rate": 1.5e-05,
-      "loss": 0.0729,
+      "loss": 0.0751,
       "step": 7420
     },
     {
       "epoch": 70.0,
-      "eval_LOCATION_f1": 0.8051948051948052,
+      "eval_LOCATION_f1": 0.7870967741935484,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7654320987654321,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7210884353741497,
+      "eval_LOCATION_precision": 0.7439024390243902,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.7172413793103448,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6463414634146342,
-      "eval_ORGANIZATION_recall": 0.8153846153846154,
-      "eval_PERSON_f1": 0.879746835443038,
+      "eval_ORGANIZATION_precision": 0.65,
+      "eval_ORGANIZATION_recall": 0.8,
+      "eval_PERSON_f1": 0.8903225806451613,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8373493975903614,
-      "eval_PERSON_recall": 0.9266666666666666,
-      "eval_QUANTITY_f1": 0.7272727272727273,
+      "eval_PERSON_precision": 0.8625,
+      "eval_PERSON_recall": 0.92,
+      "eval_QUANTITY_f1": 0.6875,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6486486486486487,
-      "eval_QUANTITY_recall": 0.8275862068965517,
-      "eval_TIME_f1": 0.8493150684931507,
+      "eval_QUANTITY_precision": 0.6285714285714286,
+      "eval_QUANTITY_recall": 0.7586206896551724,
+      "eval_TIME_f1": 0.8533333333333334,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.7948717948717948,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.1279895156621933,
-      "eval_overall_accuracy": 0.9613586020182131,
-      "eval_overall_f1": 0.8174603174603174,
-      "eval_overall_precision": 0.762962962962963,
-      "eval_overall_recall": 0.8803418803418803,
-      "eval_runtime": 0.9596,
-      "eval_samples_per_second": 194.864,
-      "eval_steps_per_second": 3.126,
+      "eval_TIME_precision": 0.7804878048780488,
+      "eval_TIME_recall": 0.9411764705882353,
+      "eval_loss": 0.13516362011432648,
+      "eval_overall_accuracy": 0.9591434900319961,
+      "eval_overall_f1": 0.814419225634179,
+      "eval_overall_precision": 0.7663316582914573,
+      "eval_overall_recall": 0.8689458689458689,
+      "eval_runtime": 0.3728,
+      "eval_samples_per_second": 501.549,
+      "eval_steps_per_second": 8.046,
       "step": 7420
     },
     {
       "epoch": 71.0,
-      "grad_norm": 2.0319900512695312,
+      "grad_norm": 2.012171745300293,
       "learning_rate": 1.45e-05,
-      "loss": 0.0728,
+      "loss": 0.0755,
       "step": 7526
     },
     {
       "epoch": 71.0,
-      "eval_LOCATION_f1": 0.7922077922077922,
+      "eval_LOCATION_f1": 0.7820512820512822,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7530864197530864,
+      "eval_LOCATION_precision": 0.7349397590361446,
       "eval_LOCATION_recall": 0.8356164383561644,
-      "eval_ORGANIZATION_f1": 0.7096774193548387,
+      "eval_ORGANIZATION_f1": 0.7189542483660131,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6111111111111112,
+      "eval_ORGANIZATION_precision": 0.625,
       "eval_ORGANIZATION_recall": 0.8461538461538461,
-      "eval_PERSON_f1": 0.8910256410256411,
+      "eval_PERSON_f1": 0.8846153846153846,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8580246913580247,
-      "eval_PERSON_recall": 0.9266666666666666,
+      "eval_PERSON_precision": 0.8518518518518519,
+      "eval_PERSON_recall": 0.92,
       "eval_QUANTITY_f1": 0.696969696969697,
       "eval_QUANTITY_number": 29,
       "eval_QUANTITY_precision": 0.6216216216216216,
       "eval_QUANTITY_recall": 0.7931034482758621,
-      "eval_TIME_f1": 0.8219178082191781,
+      "eval_TIME_f1": 0.8421052631578947,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.7692307692307693,
-      "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.1363026648759842,
-      "eval_overall_accuracy": 0.9598818606940684,
-      "eval_overall_f1": 0.8105263157894738,
-      "eval_overall_precision": 0.7530562347188264,
-      "eval_overall_recall": 0.8774928774928775,
-      "eval_runtime": 0.9593,
-      "eval_samples_per_second": 194.942,
-      "eval_steps_per_second": 3.127,
+      "eval_TIME_precision": 0.7619047619047619,
+      "eval_TIME_recall": 0.9411764705882353,
+      "eval_loss": 0.14387430250644684,
+      "eval_overall_accuracy": 0.9576667487078513,
+      "eval_overall_f1": 0.8099606815203145,
+      "eval_overall_precision": 0.75,
+      "eval_overall_recall": 0.8803418803418803,
+      "eval_runtime": 0.3724,
+      "eval_samples_per_second": 502.104,
+      "eval_steps_per_second": 8.055,
       "step": 7526
     },
     {
       "epoch": 72.0,
-      "grad_norm": 1.2698560953140259,
+      "grad_norm": 1.370118498802185,
       "learning_rate": 1.4000000000000001e-05,
-      "loss": 0.0725,
+      "loss": 0.0745,
       "step": 7632
     },
     {
       "epoch": 72.0,
-      "eval_LOCATION_f1": 0.7721518987341772,
+      "eval_LOCATION_f1": 0.7770700636942677,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7176470588235294,
+      "eval_LOCATION_precision": 0.7261904761904762,
       "eval_LOCATION_recall": 0.8356164383561644,
-      "eval_ORGANIZATION_f1": 0.7333333333333334,
+      "eval_ORGANIZATION_f1": 0.7123287671232877,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6470588235294118,
-      "eval_ORGANIZATION_recall": 0.8461538461538461,
-      "eval_PERSON_f1": 0.8910256410256411,
+      "eval_ORGANIZATION_precision": 0.6419753086419753,
+      "eval_ORGANIZATION_recall": 0.8,
+      "eval_PERSON_f1": 0.8846153846153846,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8580246913580247,
-      "eval_PERSON_recall": 0.9266666666666666,
-      "eval_QUANTITY_f1": 0.7076923076923076,
+      "eval_PERSON_precision": 0.8518518518518519,
+      "eval_PERSON_recall": 0.92,
+      "eval_QUANTITY_f1": 0.676923076923077,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6388888888888888,
-      "eval_QUANTITY_recall": 0.7931034482758621,
-      "eval_TIME_f1": 0.8493150684931507,
+      "eval_QUANTITY_precision": 0.6111111111111112,
+      "eval_QUANTITY_recall": 0.7586206896551724,
+      "eval_TIME_f1": 0.8421052631578947,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.7948717948717948,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.1346251517534256,
-      "eval_overall_accuracy": 0.9603741078021166,
-      "eval_overall_f1": 0.8153034300791556,
-      "eval_overall_precision": 0.7592137592137592,
-      "eval_overall_recall": 0.8803418803418803,
-      "eval_runtime": 0.9608,
-      "eval_samples_per_second": 194.624,
-      "eval_steps_per_second": 3.122,
+      "eval_TIME_precision": 0.7619047619047619,
+      "eval_TIME_recall": 0.9411764705882353,
+      "eval_loss": 0.13974949717521667,
+      "eval_overall_accuracy": 0.9579128722618755,
+      "eval_overall_f1": 0.8068783068783069,
+      "eval_overall_precision": 0.7530864197530864,
+      "eval_overall_recall": 0.8689458689458689,
+      "eval_runtime": 0.3737,
+      "eval_samples_per_second": 500.351,
+      "eval_steps_per_second": 8.027,
       "step": 7632
     },
     {
       "epoch": 73.0,
-      "grad_norm": 0.6609696745872498,
+      "grad_norm": 0.621573805809021,
       "learning_rate": 1.3500000000000001e-05,
-      "loss": 0.0713,
+      "loss": 0.0746,
       "step": 7738
     },
     {
       "epoch": 73.0,
-      "eval_LOCATION_f1": 0.7922077922077922,
+      "eval_LOCATION_f1": 0.7792207792207793,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7530864197530864,
-      "eval_LOCATION_recall": 0.8356164383561644,
-      "eval_ORGANIZATION_f1": 0.7189542483660131,
+      "eval_LOCATION_precision": 0.7407407407407407,
+      "eval_LOCATION_recall": 0.821917808219178,
+      "eval_ORGANIZATION_f1": 0.7172413793103448,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.625,
-      "eval_ORGANIZATION_recall": 0.8461538461538461,
-      "eval_PERSON_f1": 0.8910256410256411,
+      "eval_ORGANIZATION_precision": 0.65,
+      "eval_ORGANIZATION_recall": 0.8,
+      "eval_PERSON_f1": 0.887459807073955,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8580246913580247,
-      "eval_PERSON_recall": 0.9266666666666666,
-      "eval_QUANTITY_f1": 0.7272727272727273,
+      "eval_PERSON_precision": 0.8571428571428571,
+      "eval_PERSON_recall": 0.92,
+      "eval_QUANTITY_f1": 0.7076923076923076,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6486486486486487,
-      "eval_QUANTITY_recall": 0.8275862068965517,
-      "eval_TIME_f1": 0.8493150684931507,
+      "eval_QUANTITY_precision": 0.6388888888888888,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8533333333333334,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.7948717948717948,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.13484306633472443,
-      "eval_overall_accuracy": 0.961112478464189,
-      "eval_overall_f1": 0.8179419525065963,
-      "eval_overall_precision": 0.7616707616707616,
-      "eval_overall_recall": 0.8831908831908832,
-      "eval_runtime": 0.9626,
-      "eval_samples_per_second": 194.273,
-      "eval_steps_per_second": 3.117,
+      "eval_TIME_precision": 0.7804878048780488,
+      "eval_TIME_recall": 0.9411764705882353,
+      "eval_loss": 0.1360926479101181,
+      "eval_overall_accuracy": 0.9591434900319961,
+      "eval_overall_f1": 0.8133333333333332,
+      "eval_overall_precision": 0.7644110275689223,
+      "eval_overall_recall": 0.8689458689458689,
+      "eval_runtime": 0.3745,
+      "eval_samples_per_second": 499.381,
+      "eval_steps_per_second": 8.011,
       "step": 7738
     },
     {
       "epoch": 74.0,
-      "grad_norm": 2.321972131729126,
+      "grad_norm": 1.1147890090942383,
       "learning_rate": 1.3000000000000001e-05,
-      "loss": 0.0717,
+      "loss": 0.0727,
       "step": 7844
     },
     {
       "epoch": 74.0,
-      "eval_LOCATION_f1": 0.8051948051948052,
+      "eval_LOCATION_f1": 0.7870967741935484,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7654320987654321,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7074829931972789,
+      "eval_LOCATION_precision": 0.7439024390243902,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.7172413793103448,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6341463414634146,
+      "eval_ORGANIZATION_precision": 0.65,
       "eval_ORGANIZATION_recall": 0.8,
-      "eval_PERSON_f1": 0.888178913738019,
+      "eval_PERSON_f1": 0.887459807073955,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.852760736196319,
-      "eval_PERSON_recall": 0.9266666666666666,
-      "eval_QUANTITY_f1": 0.75,
+      "eval_PERSON_precision": 0.8571428571428571,
+      "eval_PERSON_recall": 0.92,
+      "eval_QUANTITY_f1": 0.6875,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6857142857142857,
-      "eval_QUANTITY_recall": 0.8275862068965517,
-      "eval_TIME_f1": 0.861111111111111,
+      "eval_QUANTITY_precision": 0.6285714285714286,
+      "eval_QUANTITY_recall": 0.7586206896551724,
+      "eval_TIME_f1": 0.8533333333333334,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8157894736842105,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.13028575479984283,
-      "eval_overall_accuracy": 0.9616047255722373,
-      "eval_overall_f1": 0.8213333333333335,
-      "eval_overall_precision": 0.7719298245614035,
-      "eval_overall_recall": 0.8774928774928775,
-      "eval_runtime": 0.9468,
-      "eval_samples_per_second": 197.515,
-      "eval_steps_per_second": 3.169,
+      "eval_TIME_precision": 0.7804878048780488,
+      "eval_TIME_recall": 0.9411764705882353,
+      "eval_loss": 0.13560639321804047,
+      "eval_overall_accuracy": 0.9596357371400444,
+      "eval_overall_f1": 0.8133333333333332,
+      "eval_overall_precision": 0.7644110275689223,
+      "eval_overall_recall": 0.8689458689458689,
+      "eval_runtime": 0.374,
+      "eval_samples_per_second": 500.038,
+      "eval_steps_per_second": 8.022,
       "step": 7844
     },
     {
       "epoch": 75.0,
-      "grad_norm": 1.0534641742706299,
+      "grad_norm": 0.9206987619400024,
       "learning_rate": 1.25e-05,
-      "loss": 0.074,
+      "loss": 0.0727,
       "step": 7950
     },
     {
       "epoch": 75.0,
-      "eval_LOCATION_f1": 0.7870967741935484,
+      "eval_LOCATION_f1": 0.7770700636942677,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7439024390243902,
+      "eval_LOCATION_precision": 0.7261904761904762,
       "eval_LOCATION_recall": 0.8356164383561644,
-      "eval_ORGANIZATION_f1": 0.7333333333333334,
+      "eval_ORGANIZATION_f1": 0.7162162162162162,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6470588235294118,
-      "eval_ORGANIZATION_recall": 0.8461538461538461,
-      "eval_PERSON_f1": 0.8938906752411575,
+      "eval_ORGANIZATION_precision": 0.6385542168674698,
+      "eval_ORGANIZATION_recall": 0.8153846153846154,
+      "eval_PERSON_f1": 0.887459807073955,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8633540372670807,
-      "eval_PERSON_recall": 0.9266666666666666,
-      "eval_QUANTITY_f1": 0.6666666666666666,
+      "eval_PERSON_precision": 0.8571428571428571,
+      "eval_PERSON_recall": 0.92,
+      "eval_QUANTITY_f1": 0.7076923076923076,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.575,
+      "eval_QUANTITY_precision": 0.6388888888888888,
       "eval_QUANTITY_recall": 0.7931034482758621,
-      "eval_TIME_f1": 0.861111111111111,
+      "eval_TIME_f1": 0.8421052631578947,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8157894736842105,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.13667696714401245,
-      "eval_overall_accuracy": 0.9608663549101649,
-      "eval_overall_f1": 0.8163804491413474,
-      "eval_overall_precision": 0.7610837438423645,
-      "eval_overall_recall": 0.8803418803418803,
-      "eval_runtime": 0.9529,
-      "eval_samples_per_second": 196.237,
-      "eval_steps_per_second": 3.148,
+      "eval_TIME_precision": 0.7619047619047619,
+      "eval_TIME_recall": 0.9411764705882353,
+      "eval_loss": 0.14432945847511292,
+      "eval_overall_accuracy": 0.9574206251538272,
+      "eval_overall_f1": 0.8110964332892999,
+      "eval_overall_precision": 0.7561576354679803,
+      "eval_overall_recall": 0.8746438746438746,
+      "eval_runtime": 0.374,
+      "eval_samples_per_second": 499.951,
+      "eval_steps_per_second": 8.021,
       "step": 7950
     },
     {
       "epoch": 76.0,
-      "grad_norm": 2.212968111038208,
+      "grad_norm": 1.3934582471847534,
       "learning_rate": 1.2e-05,
       "loss": 0.073,
       "step": 8056
     },
     {
       "epoch": 76.0,
-      "eval_LOCATION_f1": 0.7870967741935484,
+      "eval_LOCATION_f1": 0.7770700636942677,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7439024390243902,
+      "eval_LOCATION_precision": 0.7261904761904762,
       "eval_LOCATION_recall": 0.8356164383561644,
-      "eval_ORGANIZATION_f1": 0.728476821192053,
+      "eval_ORGANIZATION_f1": 0.7012987012987013,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6395348837209303,
-      "eval_ORGANIZATION_recall": 0.8461538461538461,
-      "eval_PERSON_f1": 0.8938906752411575,
+      "eval_ORGANIZATION_precision": 0.6067415730337079,
+      "eval_ORGANIZATION_recall": 0.8307692307692308,
+      "eval_PERSON_f1": 0.887459807073955,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8633540372670807,
-      "eval_PERSON_recall": 0.9266666666666666,
-      "eval_QUANTITY_f1": 0.75,
+      "eval_PERSON_precision": 0.8571428571428571,
+      "eval_PERSON_recall": 0.92,
+      "eval_QUANTITY_f1": 0.6875,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6857142857142857,
-      "eval_QUANTITY_recall": 0.8275862068965517,
-      "eval_TIME_f1": 0.861111111111111,
+      "eval_QUANTITY_precision": 0.6285714285714286,
+      "eval_QUANTITY_recall": 0.7586206896551724,
+      "eval_TIME_f1": 0.8,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8157894736842105,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.1331731677055359,
-      "eval_overall_accuracy": 0.9623430962343096,
-      "eval_overall_f1": 0.8233731739707835,
-      "eval_overall_precision": 0.7711442786069652,
-      "eval_overall_recall": 0.8831908831908832,
-      "eval_runtime": 0.9553,
-      "eval_samples_per_second": 195.751,
-      "eval_steps_per_second": 3.14,
+      "eval_TIME_precision": 0.7317073170731707,
+      "eval_TIME_recall": 0.8823529411764706,
+      "eval_loss": 0.14394371211528778,
+      "eval_overall_accuracy": 0.9549593896135861,
+      "eval_overall_f1": 0.8015768725361366,
+      "eval_overall_precision": 0.7439024390243902,
+      "eval_overall_recall": 0.8689458689458689,
+      "eval_runtime": 0.3764,
+      "eval_samples_per_second": 496.875,
+      "eval_steps_per_second": 7.971,
       "step": 8056
     },
     {
       "epoch": 77.0,
-      "grad_norm": 1.6397372484207153,
+      "grad_norm": 1.3345227241516113,
       "learning_rate": 1.1500000000000002e-05,
-      "loss": 0.0714,
+      "loss": 0.0733,
       "step": 8162
     },
     {
       "epoch": 77.0,
-      "eval_LOCATION_f1": 0.7870967741935484,
+      "eval_LOCATION_f1": 0.7792207792207793,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7439024390243902,
-      "eval_LOCATION_recall": 0.8356164383561644,
-      "eval_ORGANIZATION_f1": 0.7333333333333334,
+      "eval_LOCATION_precision": 0.7407407407407407,
+      "eval_LOCATION_recall": 0.821917808219178,
+      "eval_ORGANIZATION_f1": 0.7310344827586207,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6470588235294118,
-      "eval_ORGANIZATION_recall": 0.8461538461538461,
-      "eval_PERSON_f1": 0.888178913738019,
+      "eval_ORGANIZATION_precision": 0.6625,
+      "eval_ORGANIZATION_recall": 0.8153846153846154,
+      "eval_PERSON_f1": 0.887459807073955,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.852760736196319,
-      "eval_PERSON_recall": 0.9266666666666666,
-      "eval_QUANTITY_f1": 0.7384615384615385,
+      "eval_PERSON_precision": 0.8571428571428571,
+      "eval_PERSON_recall": 0.92,
+      "eval_QUANTITY_f1": 0.6875,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6666666666666666,
-      "eval_QUANTITY_recall": 0.8275862068965517,
-      "eval_TIME_f1": 0.861111111111111,
+      "eval_QUANTITY_precision": 0.6285714285714286,
+      "eval_QUANTITY_recall": 0.7586206896551724,
+      "eval_TIME_f1": 0.8266666666666665,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8157894736842105,
+      "eval_TIME_precision": 0.7560975609756098,
       "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.1331976056098938,
-      "eval_overall_accuracy": 0.9616047255722373,
-      "eval_overall_f1": 0.8211920529801325,
-      "eval_overall_precision": 0.7673267326732673,
-      "eval_overall_recall": 0.8831908831908832,
-      "eval_runtime": 0.9558,
-      "eval_samples_per_second": 195.645,
-      "eval_steps_per_second": 3.139,
+      "eval_loss": 0.13728104531764984,
+      "eval_overall_accuracy": 0.9593896135860202,
+      "eval_overall_f1": 0.8117489986648865,
+      "eval_overall_precision": 0.7638190954773869,
+      "eval_overall_recall": 0.8660968660968661,
+      "eval_runtime": 0.3706,
+      "eval_samples_per_second": 504.552,
+      "eval_steps_per_second": 8.094,
       "step": 8162
     },
     {
       "epoch": 78.0,
-      "grad_norm": 2.294029474258423,
+      "grad_norm": 1.4736689329147339,
       "learning_rate": 1.1000000000000001e-05,
-      "loss": 0.0705,
+      "loss": 0.072,
       "step": 8268
     },
     {
       "epoch": 78.0,
-      "eval_LOCATION_f1": 0.8051948051948052,
+      "eval_LOCATION_f1": 0.7870967741935484,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7654320987654321,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.728476821192053,
+      "eval_LOCATION_precision": 0.7439024390243902,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.726027397260274,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6395348837209303,
-      "eval_ORGANIZATION_recall": 0.8461538461538461,
-      "eval_PERSON_f1": 0.8945686900958466,
+      "eval_ORGANIZATION_precision": 0.654320987654321,
+      "eval_ORGANIZATION_recall": 0.8153846153846154,
+      "eval_PERSON_f1": 0.887459807073955,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8588957055214724,
-      "eval_PERSON_recall": 0.9333333333333333,
-      "eval_QUANTITY_f1": 0.746268656716418,
+      "eval_PERSON_precision": 0.8571428571428571,
+      "eval_PERSON_recall": 0.92,
+      "eval_QUANTITY_f1": 0.7076923076923076,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6578947368421053,
-      "eval_QUANTITY_recall": 0.8620689655172413,
-      "eval_TIME_f1": 0.8493150684931507,
+      "eval_QUANTITY_precision": 0.6388888888888888,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8421052631578947,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.7948717948717948,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.135635107755661,
-      "eval_overall_accuracy": 0.9618508491262614,
-      "eval_overall_f1": 0.8258575197889182,
-      "eval_overall_precision": 0.769041769041769,
-      "eval_overall_recall": 0.8917378917378918,
-      "eval_runtime": 0.9599,
-      "eval_samples_per_second": 194.803,
-      "eval_steps_per_second": 3.125,
+      "eval_TIME_precision": 0.7619047619047619,
+      "eval_TIME_recall": 0.9411764705882353,
+      "eval_loss": 0.14146719872951508,
+      "eval_overall_accuracy": 0.9581589958158996,
+      "eval_overall_f1": 0.8154050464807436,
+      "eval_overall_precision": 0.763681592039801,
+      "eval_overall_recall": 0.8746438746438746,
+      "eval_runtime": 0.3724,
+      "eval_samples_per_second": 502.127,
+      "eval_steps_per_second": 8.056,
       "step": 8268
     },
     {
       "epoch": 79.0,
-      "grad_norm": 1.3058913946151733,
+      "grad_norm": 0.9665191769599915,
       "learning_rate": 1.05e-05,
-      "loss": 0.072,
+      "loss": 0.0718,
       "step": 8374
     },
     {
       "epoch": 79.0,
-      "eval_LOCATION_f1": 0.7870967741935484,
+      "eval_LOCATION_f1": 0.7721518987341772,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7439024390243902,
+      "eval_LOCATION_precision": 0.7176470588235294,
       "eval_LOCATION_recall": 0.8356164383561644,
-      "eval_ORGANIZATION_f1": 0.7189542483660131,
+      "eval_ORGANIZATION_f1": 0.717948717948718,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.625,
-      "eval_ORGANIZATION_recall": 0.8461538461538461,
-      "eval_PERSON_f1": 0.8945686900958466,
+      "eval_ORGANIZATION_precision": 0.6153846153846154,
+      "eval_ORGANIZATION_recall": 0.8615384615384616,
+      "eval_PERSON_f1": 0.8817891373801918,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8588957055214724,
-      "eval_PERSON_recall": 0.9333333333333333,
-      "eval_QUANTITY_f1": 0.7384615384615385,
+      "eval_PERSON_precision": 0.8466257668711656,
+      "eval_PERSON_recall": 0.92,
+      "eval_QUANTITY_f1": 0.7076923076923076,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6666666666666666,
-      "eval_QUANTITY_recall": 0.8275862068965517,
-      "eval_TIME_f1": 0.8493150684931507,
+      "eval_QUANTITY_precision": 0.6388888888888888,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.7948717948717948,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.13674825429916382,
-      "eval_overall_accuracy": 0.9616047255722373,
-      "eval_overall_f1": 0.8194993412384717,
-      "eval_overall_precision": 0.7622549019607843,
-      "eval_overall_recall": 0.886039886039886,
-      "eval_runtime": 0.9529,
-      "eval_samples_per_second": 196.243,
-      "eval_steps_per_second": 3.148,
+      "eval_TIME_precision": 0.7317073170731707,
+      "eval_TIME_recall": 0.8823529411764706,
+      "eval_loss": 0.14953278005123138,
+      "eval_overall_accuracy": 0.9552055131676102,
+      "eval_overall_f1": 0.8031290743155151,
+      "eval_overall_precision": 0.7403846153846154,
+      "eval_overall_recall": 0.8774928774928775,
+      "eval_runtime": 0.3713,
+      "eval_samples_per_second": 503.668,
+      "eval_steps_per_second": 8.08,
       "step": 8374
     },
     {
       "epoch": 80.0,
-      "grad_norm": 1.1319706439971924,
+      "grad_norm": 0.4756767153739929,
       "learning_rate": 1e-05,
-      "loss": 0.0691,
+      "loss": 0.0703,
       "step": 8480
     },
     {
       "epoch": 80.0,
-      "eval_LOCATION_f1": 0.7870967741935484,
+      "eval_LOCATION_f1": 0.7820512820512822,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7439024390243902,
+      "eval_LOCATION_precision": 0.7349397590361446,
       "eval_LOCATION_recall": 0.8356164383561644,
-      "eval_ORGANIZATION_f1": 0.7189542483660131,
+      "eval_ORGANIZATION_f1": 0.7432432432432433,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.625,
+      "eval_ORGANIZATION_precision": 0.6626506024096386,
       "eval_ORGANIZATION_recall": 0.8461538461538461,
-      "eval_PERSON_f1": 0.8917197452229298,
+      "eval_PERSON_f1": 0.8817891373801918,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8536585365853658,
-      "eval_PERSON_recall": 0.9333333333333333,
-      "eval_QUANTITY_f1": 0.7384615384615385,
+      "eval_PERSON_precision": 0.8466257668711656,
+      "eval_PERSON_recall": 0.92,
+      "eval_QUANTITY_f1": 0.7076923076923076,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6666666666666666,
-      "eval_QUANTITY_recall": 0.8275862068965517,
-      "eval_TIME_f1": 0.8493150684931507,
+      "eval_QUANTITY_precision": 0.6388888888888888,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.7948717948717948,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.1365245133638382,
-      "eval_overall_accuracy": 0.9616047255722373,
-      "eval_overall_f1": 0.8184210526315789,
-      "eval_overall_precision": 0.7603911980440098,
-      "eval_overall_recall": 0.886039886039886,
-      "eval_runtime": 0.9471,
-      "eval_samples_per_second": 197.451,
-      "eval_steps_per_second": 3.168,
+      "eval_TIME_precision": 0.7317073170731707,
+      "eval_TIME_recall": 0.8823529411764706,
+      "eval_loss": 0.14299984276294708,
+      "eval_overall_accuracy": 0.9571745015998031,
+      "eval_overall_f1": 0.8110964332892999,
+      "eval_overall_precision": 0.7561576354679803,
+      "eval_overall_recall": 0.8746438746438746,
+      "eval_runtime": 0.3781,
+      "eval_samples_per_second": 494.608,
+      "eval_steps_per_second": 7.935,
       "step": 8480
     },
     {
       "epoch": 81.0,
-      "grad_norm": 1.053929090499878,
+      "grad_norm": 0.7018587589263916,
       "learning_rate": 9.5e-06,
-      "loss": 0.0709,
+      "loss": 0.0702,
       "step": 8586
     },
     {
       "epoch": 81.0,
-      "eval_LOCATION_f1": 0.8051948051948052,
+      "eval_LOCATION_f1": 0.7820512820512822,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7654320987654321,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7333333333333334,
+      "eval_LOCATION_precision": 0.7349397590361446,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.7432432432432433,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6470588235294118,
+      "eval_ORGANIZATION_precision": 0.6626506024096386,
       "eval_ORGANIZATION_recall": 0.8461538461538461,
-      "eval_PERSON_f1": 0.8945686900958466,
+      "eval_PERSON_f1": 0.887459807073955,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8588957055214724,
-      "eval_PERSON_recall": 0.9333333333333333,
+      "eval_PERSON_precision": 0.8571428571428571,
+      "eval_PERSON_recall": 0.92,
       "eval_QUANTITY_f1": 0.71875,
       "eval_QUANTITY_number": 29,
       "eval_QUANTITY_precision": 0.6571428571428571,
       "eval_QUANTITY_recall": 0.7931034482758621,
-      "eval_TIME_f1": 0.861111111111111,
+      "eval_TIME_f1": 0.8,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8157894736842105,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.13273945450782776,
-      "eval_overall_accuracy": 0.963081466896382,
-      "eval_overall_f1": 0.8260292164674634,
-      "eval_overall_precision": 0.7736318407960199,
-      "eval_overall_recall": 0.886039886039886,
-      "eval_runtime": 0.9555,
-      "eval_samples_per_second": 195.717,
-      "eval_steps_per_second": 3.14,
+      "eval_TIME_precision": 0.7317073170731707,
+      "eval_TIME_recall": 0.8823529411764706,
+      "eval_loss": 0.14150598645210266,
+      "eval_overall_accuracy": 0.9584051193699237,
+      "eval_overall_f1": 0.8143236074270558,
+      "eval_overall_precision": 0.7617866004962779,
+      "eval_overall_recall": 0.8746438746438746,
+      "eval_runtime": 0.3737,
+      "eval_samples_per_second": 500.383,
+      "eval_steps_per_second": 8.028,
       "step": 8586
     },
     {
       "epoch": 82.0,
-      "grad_norm": 1.5062341690063477,
+      "grad_norm": 1.6355091333389282,
       "learning_rate": 9e-06,
-      "loss": 0.0702,
+      "loss": 0.0721,
       "step": 8692
     },
     {
       "epoch": 82.0,
-      "eval_LOCATION_f1": 0.7922077922077922,
+      "eval_LOCATION_f1": 0.7820512820512822,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7530864197530864,
+      "eval_LOCATION_precision": 0.7349397590361446,
       "eval_LOCATION_recall": 0.8356164383561644,
-      "eval_ORGANIZATION_f1": 0.7333333333333334,
+      "eval_ORGANIZATION_f1": 0.7432432432432433,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6470588235294118,
+      "eval_ORGANIZATION_precision": 0.6626506024096386,
       "eval_ORGANIZATION_recall": 0.8461538461538461,
-      "eval_PERSON_f1": 0.8945686900958466,
+      "eval_PERSON_f1": 0.887459807073955,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8588957055214724,
-      "eval_PERSON_recall": 0.9333333333333333,
-      "eval_QUANTITY_f1": 0.7272727272727273,
+      "eval_PERSON_precision": 0.8571428571428571,
+      "eval_PERSON_recall": 0.92,
+      "eval_QUANTITY_f1": 0.71875,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6486486486486487,
-      "eval_QUANTITY_recall": 0.8275862068965517,
-      "eval_TIME_f1": 0.861111111111111,
+      "eval_QUANTITY_precision": 0.6571428571428571,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8266666666666665,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8157894736842105,
+      "eval_TIME_precision": 0.7560975609756098,
       "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.1342259794473648,
-      "eval_overall_accuracy": 0.9620969726802855,
-      "eval_overall_f1": 0.823841059602649,
-      "eval_overall_precision": 0.7698019801980198,
-      "eval_overall_recall": 0.886039886039886,
-      "eval_runtime": 0.9561,
-      "eval_samples_per_second": 195.594,
-      "eval_steps_per_second": 3.138,
+      "eval_loss": 0.14245130121707916,
+      "eval_overall_accuracy": 0.958897366477972,
+      "eval_overall_f1": 0.8169761273209548,
+      "eval_overall_precision": 0.7642679900744417,
+      "eval_overall_recall": 0.8774928774928775,
+      "eval_runtime": 0.3741,
+      "eval_samples_per_second": 499.836,
+      "eval_steps_per_second": 8.019,
       "step": 8692
     },
     {
       "epoch": 83.0,
-      "grad_norm": 1.0942083597183228,
+      "grad_norm": 0.8933713436126709,
       "learning_rate": 8.500000000000002e-06,
-      "loss": 0.0703,
+      "loss": 0.0712,
       "step": 8798
     },
     {
       "epoch": 83.0,
-      "eval_LOCATION_f1": 0.7922077922077922,
+      "eval_LOCATION_f1": 0.7770700636942677,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7530864197530864,
+      "eval_LOCATION_precision": 0.7261904761904762,
       "eval_LOCATION_recall": 0.8356164383561644,
-      "eval_ORGANIZATION_f1": 0.728476821192053,
+      "eval_ORGANIZATION_f1": 0.7432432432432433,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6395348837209303,
+      "eval_ORGANIZATION_precision": 0.6626506024096386,
       "eval_ORGANIZATION_recall": 0.8461538461538461,
-      "eval_PERSON_f1": 0.8945686900958466,
+      "eval_PERSON_f1": 0.887459807073955,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8588957055214724,
-      "eval_PERSON_recall": 0.9333333333333333,
-      "eval_QUANTITY_f1": 0.7272727272727273,
+      "eval_PERSON_precision": 0.8571428571428571,
+      "eval_PERSON_recall": 0.92,
+      "eval_QUANTITY_f1": 0.71875,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6486486486486487,
-      "eval_QUANTITY_recall": 0.8275862068965517,
-      "eval_TIME_f1": 0.8219178082191781,
+      "eval_QUANTITY_precision": 0.6571428571428571,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8421052631578947,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.7692307692307693,
-      "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.13451319932937622,
-      "eval_overall_accuracy": 0.9620969726802855,
-      "eval_overall_f1": 0.8190224570673711,
-      "eval_overall_precision": 0.7635467980295566,
-      "eval_overall_recall": 0.8831908831908832,
-      "eval_runtime": 0.9561,
-      "eval_samples_per_second": 195.58,
-      "eval_steps_per_second": 3.138,
+      "eval_TIME_precision": 0.7619047619047619,
+      "eval_TIME_recall": 0.9411764705882353,
+      "eval_loss": 0.1438244879245758,
+      "eval_overall_accuracy": 0.9586512429239479,
+      "eval_overall_f1": 0.8174603174603174,
+      "eval_overall_precision": 0.762962962962963,
+      "eval_overall_recall": 0.8803418803418803,
+      "eval_runtime": 0.3755,
+      "eval_samples_per_second": 498.003,
+      "eval_steps_per_second": 7.989,
       "step": 8798
     },
     {
       "epoch": 84.0,
-      "grad_norm": 1.3639521598815918,
+      "grad_norm": 1.0442496538162231,
       "learning_rate": 8.000000000000001e-06,
-      "loss": 0.0687,
+      "loss": 0.0719,
       "step": 8904
     },
     {
       "epoch": 84.0,
-      "eval_LOCATION_f1": 0.8051948051948052,
+      "eval_LOCATION_f1": 0.7770700636942677,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7654320987654321,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7333333333333334,
+      "eval_LOCATION_precision": 0.7261904761904762,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.738255033557047,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6470588235294118,
+      "eval_ORGANIZATION_precision": 0.6547619047619048,
       "eval_ORGANIZATION_recall": 0.8461538461538461,
-      "eval_PERSON_f1": 0.8945686900958466,
+      "eval_PERSON_f1": 0.887459807073955,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8588957055214724,
-      "eval_PERSON_recall": 0.9333333333333333,
-      "eval_QUANTITY_f1": 0.716417910447761,
+      "eval_PERSON_precision": 0.8571428571428571,
+      "eval_PERSON_recall": 0.92,
+      "eval_QUANTITY_f1": 0.7076923076923076,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.631578947368421,
-      "eval_QUANTITY_recall": 0.8275862068965517,
-      "eval_TIME_f1": 0.8333333333333333,
+      "eval_QUANTITY_precision": 0.6388888888888888,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8421052631578947,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.7894736842105263,
-      "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.13550812005996704,
-      "eval_overall_accuracy": 0.9616047255722373,
-      "eval_overall_f1": 0.8227513227513228,
-      "eval_overall_precision": 0.7679012345679013,
-      "eval_overall_recall": 0.886039886039886,
-      "eval_runtime": 0.9603,
-      "eval_samples_per_second": 194.739,
-      "eval_steps_per_second": 3.124,
+      "eval_TIME_precision": 0.7619047619047619,
+      "eval_TIME_recall": 0.9411764705882353,
+      "eval_loss": 0.1436508744955063,
+      "eval_overall_accuracy": 0.9579128722618755,
+      "eval_overall_f1": 0.8153034300791556,
+      "eval_overall_precision": 0.7592137592137592,
+      "eval_overall_recall": 0.8803418803418803,
+      "eval_runtime": 0.3727,
+      "eval_samples_per_second": 501.681,
+      "eval_steps_per_second": 8.048,
       "step": 8904
     },
     {
       "epoch": 85.0,
-      "grad_norm": 1.3763707876205444,
+      "grad_norm": 1.8608384132385254,
       "learning_rate": 7.5e-06,
-      "loss": 0.0707,
+      "loss": 0.0709,
       "step": 9010
     },
     {
       "epoch": 85.0,
-      "eval_LOCATION_f1": 0.7922077922077922,
+      "eval_LOCATION_f1": 0.7770700636942677,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7530864197530864,
+      "eval_LOCATION_precision": 0.7261904761904762,
       "eval_LOCATION_recall": 0.8356164383561644,
-      "eval_ORGANIZATION_f1": 0.7236842105263158,
+      "eval_ORGANIZATION_f1": 0.7297297297297298,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.632183908045977,
-      "eval_ORGANIZATION_recall": 0.8461538461538461,
-      "eval_PERSON_f1": 0.8945686900958466,
+      "eval_ORGANIZATION_precision": 0.6506024096385542,
+      "eval_ORGANIZATION_recall": 0.8307692307692308,
+      "eval_PERSON_f1": 0.887459807073955,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8588957055214724,
-      "eval_PERSON_recall": 0.9333333333333333,
-      "eval_QUANTITY_f1": 0.7575757575757576,
+      "eval_PERSON_precision": 0.8571428571428571,
+      "eval_PERSON_recall": 0.92,
+      "eval_QUANTITY_f1": 0.7076923076923076,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6756756756756757,
-      "eval_QUANTITY_recall": 0.8620689655172413,
-      "eval_TIME_f1": 0.8219178082191781,
+      "eval_QUANTITY_precision": 0.6388888888888888,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.7692307692307693,
+      "eval_TIME_precision": 0.7317073170731707,
       "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.13885249197483063,
-      "eval_overall_accuracy": 0.9618508491262614,
-      "eval_overall_f1": 0.8205804749340371,
-      "eval_overall_precision": 0.7641277641277642,
-      "eval_overall_recall": 0.886039886039886,
-      "eval_runtime": 0.9502,
-      "eval_samples_per_second": 196.794,
-      "eval_steps_per_second": 3.157,
+      "eval_loss": 0.14415280520915985,
+      "eval_overall_accuracy": 0.9574206251538272,
+      "eval_overall_f1": 0.8095238095238095,
+      "eval_overall_precision": 0.7555555555555555,
+      "eval_overall_recall": 0.8717948717948718,
+      "eval_runtime": 0.3737,
+      "eval_samples_per_second": 500.35,
+      "eval_steps_per_second": 8.027,
       "step": 9010
     },
     {
       "epoch": 86.0,
-      "grad_norm": 0.6508387923240662,
+      "grad_norm": 1.1902055740356445,
       "learning_rate": 7.000000000000001e-06,
-      "loss": 0.069,
+      "loss": 0.0699,
       "step": 9116
     },
     {
       "epoch": 86.0,
-      "eval_LOCATION_f1": 0.8051948051948052,
+      "eval_LOCATION_f1": 0.7870967741935484,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7654320987654321,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7333333333333334,
+      "eval_LOCATION_precision": 0.7439024390243902,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.7346938775510204,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6470588235294118,
-      "eval_ORGANIZATION_recall": 0.8461538461538461,
-      "eval_PERSON_f1": 0.8945686900958466,
+      "eval_ORGANIZATION_precision": 0.6585365853658537,
+      "eval_ORGANIZATION_recall": 0.8307692307692308,
+      "eval_PERSON_f1": 0.887459807073955,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8588957055214724,
-      "eval_PERSON_recall": 0.9333333333333333,
-      "eval_QUANTITY_f1": 0.746268656716418,
+      "eval_PERSON_precision": 0.8571428571428571,
+      "eval_PERSON_recall": 0.92,
+      "eval_QUANTITY_f1": 0.7076923076923076,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6578947368421053,
-      "eval_QUANTITY_recall": 0.8620689655172413,
-      "eval_TIME_f1": 0.8333333333333333,
+      "eval_QUANTITY_precision": 0.6388888888888888,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8421052631578947,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.7894736842105263,
-      "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.13471482694149017,
-      "eval_overall_accuracy": 0.9620969726802855,
-      "eval_overall_f1": 0.8253968253968255,
-      "eval_overall_precision": 0.7703703703703704,
-      "eval_overall_recall": 0.8888888888888888,
-      "eval_runtime": 0.9478,
-      "eval_samples_per_second": 197.302,
-      "eval_steps_per_second": 3.165,
+      "eval_TIME_precision": 0.7619047619047619,
+      "eval_TIME_recall": 0.9411764705882353,
+      "eval_loss": 0.14072592556476593,
+      "eval_overall_accuracy": 0.958897366477972,
+      "eval_overall_f1": 0.8169761273209548,
+      "eval_overall_precision": 0.7642679900744417,
+      "eval_overall_recall": 0.8774928774928775,
+      "eval_runtime": 0.3749,
+      "eval_samples_per_second": 498.822,
+      "eval_steps_per_second": 8.002,
       "step": 9116
     },
     {
       "epoch": 87.0,
-      "grad_norm": 1.5169206857681274,
+      "grad_norm": 0.9288851022720337,
       "learning_rate": 6.5000000000000004e-06,
-      "loss": 0.0699,
+      "loss": 0.0681,
       "step": 9222
     },
     {
       "epoch": 87.0,
-      "eval_LOCATION_f1": 0.8051948051948052,
+      "eval_LOCATION_f1": 0.7820512820512822,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7654320987654321,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.728476821192053,
+      "eval_LOCATION_precision": 0.7349397590361446,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.7346938775510204,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6395348837209303,
-      "eval_ORGANIZATION_recall": 0.8461538461538461,
-      "eval_PERSON_f1": 0.8945686900958466,
+      "eval_ORGANIZATION_precision": 0.6585365853658537,
+      "eval_ORGANIZATION_recall": 0.8307692307692308,
+      "eval_PERSON_f1": 0.887459807073955,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8588957055214724,
-      "eval_PERSON_recall": 0.9333333333333333,
-      "eval_QUANTITY_f1": 0.75,
+      "eval_PERSON_precision": 0.8571428571428571,
+      "eval_PERSON_recall": 0.92,
+      "eval_QUANTITY_f1": 0.71875,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6857142857142857,
-      "eval_QUANTITY_recall": 0.8275862068965517,
-      "eval_TIME_f1": 0.861111111111111,
+      "eval_QUANTITY_precision": 0.6571428571428571,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8421052631578947,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8157894736842105,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.13515108823776245,
-      "eval_overall_accuracy": 0.9620969726802855,
-      "eval_overall_f1": 0.8275862068965517,
-      "eval_overall_precision": 0.7741935483870968,
-      "eval_overall_recall": 0.8888888888888888,
-      "eval_runtime": 0.9449,
-      "eval_samples_per_second": 197.905,
-      "eval_steps_per_second": 3.175,
+      "eval_TIME_precision": 0.7619047619047619,
+      "eval_TIME_recall": 0.9411764705882353,
+      "eval_loss": 0.14135389029979706,
+      "eval_overall_accuracy": 0.958897366477972,
+      "eval_overall_f1": 0.8169761273209548,
+      "eval_overall_precision": 0.7642679900744417,
+      "eval_overall_recall": 0.8774928774928775,
+      "eval_runtime": 0.3733,
+      "eval_samples_per_second": 500.974,
+      "eval_steps_per_second": 8.037,
       "step": 9222
     },
     {
       "epoch": 88.0,
-      "grad_norm": 0.8914301991462708,
+      "grad_norm": 1.107254981994629,
       "learning_rate": 6e-06,
-      "loss": 0.0668,
+      "loss": 0.0685,
       "step": 9328
     },
     {
       "epoch": 88.0,
-      "eval_LOCATION_f1": 0.8051948051948052,
+      "eval_LOCATION_f1": 0.7770700636942677,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7654320987654321,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.728476821192053,
+      "eval_LOCATION_precision": 0.7261904761904762,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.7346938775510204,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6395348837209303,
-      "eval_ORGANIZATION_recall": 0.8461538461538461,
-      "eval_PERSON_f1": 0.8945686900958466,
+      "eval_ORGANIZATION_precision": 0.6585365853658537,
+      "eval_ORGANIZATION_recall": 0.8307692307692308,
+      "eval_PERSON_f1": 0.8846153846153846,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8588957055214724,
-      "eval_PERSON_recall": 0.9333333333333333,
-      "eval_QUANTITY_f1": 0.7272727272727273,
+      "eval_PERSON_precision": 0.8518518518518519,
+      "eval_PERSON_recall": 0.92,
+      "eval_QUANTITY_f1": 0.7076923076923076,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6486486486486487,
-      "eval_QUANTITY_recall": 0.8275862068965517,
-      "eval_TIME_f1": 0.861111111111111,
+      "eval_QUANTITY_precision": 0.6388888888888888,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8533333333333334,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8157894736842105,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.1347392499446869,
-      "eval_overall_accuracy": 0.9616047255722373,
-      "eval_overall_f1": 0.8253968253968255,
-      "eval_overall_precision": 0.7703703703703704,
-      "eval_overall_recall": 0.8888888888888888,
-      "eval_runtime": 0.9567,
-      "eval_samples_per_second": 195.462,
-      "eval_steps_per_second": 3.136,
+      "eval_TIME_precision": 0.7804878048780488,
+      "eval_TIME_recall": 0.9411764705882353,
+      "eval_loss": 0.14269116520881653,
+      "eval_overall_accuracy": 0.9579128722618755,
+      "eval_overall_f1": 0.8148148148148149,
+      "eval_overall_precision": 0.7604938271604939,
+      "eval_overall_recall": 0.8774928774928775,
+      "eval_runtime": 0.378,
+      "eval_samples_per_second": 494.729,
+      "eval_steps_per_second": 7.937,
       "step": 9328
     },
     {
       "epoch": 89.0,
-      "grad_norm": 3.5633649826049805,
+      "grad_norm": 2.199923276901245,
       "learning_rate": 5.500000000000001e-06,
-      "loss": 0.0702,
+      "loss": 0.0703,
       "step": 9434
     },
     {
       "epoch": 89.0,
-      "eval_LOCATION_f1": 0.8051948051948052,
+      "eval_LOCATION_f1": 0.7870967741935484,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7654320987654321,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.738255033557047,
+      "eval_LOCATION_precision": 0.7439024390243902,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.7310344827586207,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6547619047619048,
-      "eval_ORGANIZATION_recall": 0.8461538461538461,
-      "eval_PERSON_f1": 0.8917197452229298,
+      "eval_ORGANIZATION_precision": 0.6625,
+      "eval_ORGANIZATION_recall": 0.8153846153846154,
+      "eval_PERSON_f1": 0.887459807073955,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8536585365853658,
-      "eval_PERSON_recall": 0.9333333333333333,
-      "eval_QUANTITY_f1": 0.7384615384615385,
+      "eval_PERSON_precision": 0.8571428571428571,
+      "eval_PERSON_recall": 0.92,
+      "eval_QUANTITY_f1": 0.7076923076923076,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6666666666666666,
-      "eval_QUANTITY_recall": 0.8275862068965517,
-      "eval_TIME_f1": 0.861111111111111,
+      "eval_QUANTITY_precision": 0.6388888888888888,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8648648648648648,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8157894736842105,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.13250109553337097,
-      "eval_overall_accuracy": 0.9620969726802855,
-      "eval_overall_f1": 0.8275862068965517,
-      "eval_overall_precision": 0.7741935483870968,
-      "eval_overall_recall": 0.8888888888888888,
-      "eval_runtime": 0.9522,
-      "eval_samples_per_second": 196.391,
-      "eval_steps_per_second": 3.151,
+      "eval_TIME_precision": 0.8,
+      "eval_TIME_recall": 0.9411764705882353,
+      "eval_loss": 0.14063221216201782,
+      "eval_overall_accuracy": 0.9593896135860202,
+      "eval_overall_f1": 0.8186666666666667,
+      "eval_overall_precision": 0.7694235588972431,
+      "eval_overall_recall": 0.8746438746438746,
+      "eval_runtime": 0.3757,
+      "eval_samples_per_second": 497.751,
+      "eval_steps_per_second": 7.985,
       "step": 9434
     },
     {
       "epoch": 90.0,
-      "grad_norm": 4.408683776855469,
+      "grad_norm": 1.6066433191299438,
       "learning_rate": 5e-06,
-      "loss": 0.0673,
+      "loss": 0.0695,
       "step": 9540
     },
     {
       "epoch": 90.0,
-      "eval_LOCATION_f1": 0.8051948051948052,
+      "eval_LOCATION_f1": 0.7870967741935484,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7654320987654321,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.738255033557047,
+      "eval_LOCATION_precision": 0.7439024390243902,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.726027397260274,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6547619047619048,
-      "eval_ORGANIZATION_recall": 0.8461538461538461,
-      "eval_PERSON_f1": 0.8917197452229298,
+      "eval_ORGANIZATION_precision": 0.654320987654321,
+      "eval_ORGANIZATION_recall": 0.8153846153846154,
+      "eval_PERSON_f1": 0.887459807073955,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8536585365853658,
-      "eval_PERSON_recall": 0.9333333333333333,
-      "eval_QUANTITY_f1": 0.75,
+      "eval_PERSON_precision": 0.8571428571428571,
+      "eval_PERSON_recall": 0.92,
+      "eval_QUANTITY_f1": 0.7076923076923076,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6857142857142857,
-      "eval_QUANTITY_recall": 0.8275862068965517,
-      "eval_TIME_f1": 0.861111111111111,
+      "eval_QUANTITY_precision": 0.6388888888888888,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8767123287671232,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8157894736842105,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.1317829042673111,
-      "eval_overall_accuracy": 0.9623430962343096,
-      "eval_overall_f1": 0.8286852589641434,
-      "eval_overall_precision": 0.7761194029850746,
-      "eval_overall_recall": 0.8888888888888888,
-      "eval_runtime": 0.9498,
-      "eval_samples_per_second": 196.876,
-      "eval_steps_per_second": 3.158,
+      "eval_TIME_precision": 0.8205128205128205,
+      "eval_TIME_recall": 0.9411764705882353,
+      "eval_loss": 0.13843446969985962,
+      "eval_overall_accuracy": 0.9601279842480925,
+      "eval_overall_f1": 0.8186666666666667,
+      "eval_overall_precision": 0.7694235588972431,
+      "eval_overall_recall": 0.8746438746438746,
+      "eval_runtime": 0.3746,
+      "eval_samples_per_second": 499.225,
+      "eval_steps_per_second": 8.009,
       "step": 9540
     },
     {
       "epoch": 91.0,
-      "grad_norm": 1.6093438863754272,
+      "grad_norm": 1.384969711303711,
       "learning_rate": 4.5e-06,
-      "loss": 0.0677,
+      "loss": 0.0684,
       "step": 9646
     },
     {
       "epoch": 91.0,
-      "eval_LOCATION_f1": 0.8051948051948052,
+      "eval_LOCATION_f1": 0.7770700636942677,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7654320987654321,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7333333333333334,
+      "eval_LOCATION_precision": 0.7261904761904762,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.7210884353741497,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6470588235294118,
-      "eval_ORGANIZATION_recall": 0.8461538461538461,
-      "eval_PERSON_f1": 0.8917197452229298,
+      "eval_ORGANIZATION_precision": 0.6463414634146342,
+      "eval_ORGANIZATION_recall": 0.8153846153846154,
+      "eval_PERSON_f1": 0.8817891373801918,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8536585365853658,
-      "eval_PERSON_recall": 0.9333333333333333,
-      "eval_QUANTITY_f1": 0.7384615384615385,
+      "eval_PERSON_precision": 0.8466257668711656,
+      "eval_PERSON_recall": 0.92,
+      "eval_QUANTITY_f1": 0.7076923076923076,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6666666666666666,
-      "eval_QUANTITY_recall": 0.8275862068965517,
-      "eval_TIME_f1": 0.861111111111111,
+      "eval_QUANTITY_precision": 0.6388888888888888,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8421052631578947,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8157894736842105,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.1336798220872879,
-      "eval_overall_accuracy": 0.9623430962343096,
-      "eval_overall_f1": 0.8264900662251655,
-      "eval_overall_precision": 0.7722772277227723,
-      "eval_overall_recall": 0.8888888888888888,
-      "eval_runtime": 1.0231,
-      "eval_samples_per_second": 182.78,
-      "eval_steps_per_second": 2.932,
+      "eval_TIME_precision": 0.7619047619047619,
+      "eval_TIME_recall": 0.9411764705882353,
+      "eval_loss": 0.14361019432544708,
+      "eval_overall_accuracy": 0.956928378045779,
+      "eval_overall_f1": 0.8100263852242744,
+      "eval_overall_precision": 0.7542997542997543,
+      "eval_overall_recall": 0.8746438746438746,
+      "eval_runtime": 0.3739,
+      "eval_samples_per_second": 500.193,
+      "eval_steps_per_second": 8.024,
       "step": 9646
     },
     {
       "epoch": 92.0,
-      "grad_norm": 1.4977636337280273,
+      "grad_norm": 1.5777561664581299,
       "learning_rate": 4.000000000000001e-06,
-      "loss": 0.068,
+      "loss": 0.0681,
       "step": 9752
     },
     {
       "epoch": 92.0,
-      "eval_LOCATION_f1": 0.8051948051948052,
+      "eval_LOCATION_f1": 0.7870967741935484,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7654320987654321,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7333333333333334,
+      "eval_LOCATION_precision": 0.7439024390243902,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.726027397260274,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6470588235294118,
-      "eval_ORGANIZATION_recall": 0.8461538461538461,
-      "eval_PERSON_f1": 0.8917197452229298,
+      "eval_ORGANIZATION_precision": 0.654320987654321,
+      "eval_ORGANIZATION_recall": 0.8153846153846154,
+      "eval_PERSON_f1": 0.887459807073955,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8536585365853658,
-      "eval_PERSON_recall": 0.9333333333333333,
-      "eval_QUANTITY_f1": 0.7384615384615385,
+      "eval_PERSON_precision": 0.8571428571428571,
+      "eval_PERSON_recall": 0.92,
+      "eval_QUANTITY_f1": 0.7076923076923076,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6666666666666666,
-      "eval_QUANTITY_recall": 0.8275862068965517,
-      "eval_TIME_f1": 0.861111111111111,
+      "eval_QUANTITY_precision": 0.6388888888888888,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8533333333333334,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8157894736842105,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.13396261632442474,
-      "eval_overall_accuracy": 0.9623430962343096,
-      "eval_overall_f1": 0.8264900662251655,
-      "eval_overall_precision": 0.7722772277227723,
-      "eval_overall_recall": 0.8888888888888888,
-      "eval_runtime": 0.96,
-      "eval_samples_per_second": 194.788,
-      "eval_steps_per_second": 3.125,
+      "eval_TIME_precision": 0.7804878048780488,
+      "eval_TIME_recall": 0.9411764705882353,
+      "eval_loss": 0.14141333103179932,
+      "eval_overall_accuracy": 0.9591434900319961,
+      "eval_overall_f1": 0.8164893617021275,
+      "eval_overall_precision": 0.7655860349127181,
+      "eval_overall_recall": 0.8746438746438746,
+      "eval_runtime": 0.3754,
+      "eval_samples_per_second": 498.089,
+      "eval_steps_per_second": 7.991,
       "step": 9752
     },
     {
       "epoch": 93.0,
-      "grad_norm": 0.2478037327528,
+      "grad_norm": 0.12837229669094086,
       "learning_rate": 3.5000000000000004e-06,
-      "loss": 0.0673,
+      "loss": 0.0694,
       "step": 9858
     },
     {
       "epoch": 93.0,
-      "eval_LOCATION_f1": 0.8051948051948052,
+      "eval_LOCATION_f1": 0.7820512820512822,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7654320987654321,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7333333333333334,
+      "eval_LOCATION_precision": 0.7349397590361446,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.7346938775510204,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6470588235294118,
-      "eval_ORGANIZATION_recall": 0.8461538461538461,
-      "eval_PERSON_f1": 0.8917197452229298,
+      "eval_ORGANIZATION_precision": 0.6585365853658537,
+      "eval_ORGANIZATION_recall": 0.8307692307692308,
+      "eval_PERSON_f1": 0.887459807073955,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8536585365853658,
-      "eval_PERSON_recall": 0.9333333333333333,
-      "eval_QUANTITY_f1": 0.7272727272727273,
+      "eval_PERSON_precision": 0.8571428571428571,
+      "eval_PERSON_recall": 0.92,
+      "eval_QUANTITY_f1": 0.7076923076923076,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6486486486486487,
-      "eval_QUANTITY_recall": 0.8275862068965517,
-      "eval_TIME_f1": 0.861111111111111,
+      "eval_QUANTITY_precision": 0.6388888888888888,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8157894736842106,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8157894736842105,
+      "eval_TIME_precision": 0.7380952380952381,
       "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.13295292854309082,
-      "eval_overall_accuracy": 0.9618508491262614,
-      "eval_overall_f1": 0.8253968253968255,
-      "eval_overall_precision": 0.7703703703703704,
-      "eval_overall_recall": 0.8888888888888888,
-      "eval_runtime": 0.9596,
-      "eval_samples_per_second": 194.864,
-      "eval_steps_per_second": 3.126,
+      "eval_loss": 0.14285434782505035,
+      "eval_overall_accuracy": 0.9584051193699237,
+      "eval_overall_f1": 0.8132450331125828,
+      "eval_overall_precision": 0.7599009900990099,
+      "eval_overall_recall": 0.8746438746438746,
+      "eval_runtime": 0.3737,
+      "eval_samples_per_second": 500.447,
+      "eval_steps_per_second": 8.029,
       "step": 9858
     },
     {
       "epoch": 94.0,
-      "grad_norm": 1.5132652521133423,
+      "grad_norm": 1.4026833772659302,
       "learning_rate": 3e-06,
-      "loss": 0.0666,
+      "loss": 0.068,
       "step": 9964
     },
     {
       "epoch": 94.0,
-      "eval_LOCATION_f1": 0.8051948051948052,
+      "eval_LOCATION_f1": 0.7820512820512822,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7654320987654321,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7236842105263158,
+      "eval_LOCATION_precision": 0.7349397590361446,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.726027397260274,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.632183908045977,
-      "eval_ORGANIZATION_recall": 0.8461538461538461,
-      "eval_PERSON_f1": 0.8945686900958466,
+      "eval_ORGANIZATION_precision": 0.654320987654321,
+      "eval_ORGANIZATION_recall": 0.8153846153846154,
+      "eval_PERSON_f1": 0.887459807073955,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8588957055214724,
-      "eval_PERSON_recall": 0.9333333333333333,
-      "eval_QUANTITY_f1": 0.7272727272727273,
+      "eval_PERSON_precision": 0.8571428571428571,
+      "eval_PERSON_recall": 0.92,
+      "eval_QUANTITY_f1": 0.7076923076923076,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6486486486486487,
-      "eval_QUANTITY_recall": 0.8275862068965517,
-      "eval_TIME_f1": 0.861111111111111,
+      "eval_QUANTITY_precision": 0.6388888888888888,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8421052631578947,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8157894736842105,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.1347019225358963,
-      "eval_overall_accuracy": 0.9618508491262614,
-      "eval_overall_f1": 0.8243064729194187,
-      "eval_overall_precision": 0.7684729064039408,
-      "eval_overall_recall": 0.8888888888888888,
-      "eval_runtime": 0.958,
-      "eval_samples_per_second": 195.195,
-      "eval_steps_per_second": 3.131,
+      "eval_TIME_precision": 0.7619047619047619,
+      "eval_TIME_recall": 0.9411764705882353,
+      "eval_loss": 0.14125195145606995,
+      "eval_overall_accuracy": 0.9581589958158996,
+      "eval_overall_f1": 0.8143236074270558,
+      "eval_overall_precision": 0.7617866004962779,
+      "eval_overall_recall": 0.8746438746438746,
+      "eval_runtime": 0.3714,
+      "eval_samples_per_second": 503.459,
+      "eval_steps_per_second": 8.077,
       "step": 9964
     },
     {
       "epoch": 95.0,
-      "grad_norm": 0.8759310245513916,
+      "grad_norm": 2.012713670730591,
       "learning_rate": 2.5e-06,
-      "loss": 0.0669,
+      "loss": 0.0682,
       "step": 10070
     },
     {
       "epoch": 95.0,
-      "eval_LOCATION_f1": 0.8051948051948052,
+      "eval_LOCATION_f1": 0.7820512820512822,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7654320987654321,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7333333333333334,
+      "eval_LOCATION_precision": 0.7349397590361446,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.7346938775510204,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6470588235294118,
-      "eval_ORGANIZATION_recall": 0.8461538461538461,
-      "eval_PERSON_f1": 0.8945686900958466,
+      "eval_ORGANIZATION_precision": 0.6585365853658537,
+      "eval_ORGANIZATION_recall": 0.8307692307692308,
+      "eval_PERSON_f1": 0.887459807073955,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8588957055214724,
-      "eval_PERSON_recall": 0.9333333333333333,
-      "eval_QUANTITY_f1": 0.7272727272727273,
+      "eval_PERSON_precision": 0.8571428571428571,
+      "eval_PERSON_recall": 0.92,
+      "eval_QUANTITY_f1": 0.7076923076923076,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6486486486486487,
-      "eval_QUANTITY_recall": 0.8275862068965517,
-      "eval_TIME_f1": 0.861111111111111,
+      "eval_QUANTITY_precision": 0.6388888888888888,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8157894736842106,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8157894736842105,
+      "eval_TIME_precision": 0.7380952380952381,
       "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.13462986052036285,
-      "eval_overall_accuracy": 0.9618508491262614,
-      "eval_overall_f1": 0.8264900662251655,
-      "eval_overall_precision": 0.7722772277227723,
-      "eval_overall_recall": 0.8888888888888888,
-      "eval_runtime": 0.9564,
-      "eval_samples_per_second": 195.523,
-      "eval_steps_per_second": 3.137,
+      "eval_loss": 0.1429029256105423,
+      "eval_overall_accuracy": 0.9581589958158996,
+      "eval_overall_f1": 0.8132450331125828,
+      "eval_overall_precision": 0.7599009900990099,
+      "eval_overall_recall": 0.8746438746438746,
+      "eval_runtime": 0.3742,
+      "eval_samples_per_second": 499.707,
+      "eval_steps_per_second": 8.017,
       "step": 10070
     },
     {
       "epoch": 96.0,
-      "grad_norm": 1.3174265623092651,
+      "grad_norm": 0.9533500671386719,
       "learning_rate": 2.0000000000000003e-06,
-      "loss": 0.0683,
+      "loss": 0.0697,
       "step": 10176
     },
     {
       "epoch": 96.0,
-      "eval_LOCATION_f1": 0.8051948051948052,
+      "eval_LOCATION_f1": 0.7820512820512822,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7654320987654321,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.728476821192053,
+      "eval_LOCATION_precision": 0.7349397590361446,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.726027397260274,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6395348837209303,
-      "eval_ORGANIZATION_recall": 0.8461538461538461,
-      "eval_PERSON_f1": 0.8945686900958466,
+      "eval_ORGANIZATION_precision": 0.654320987654321,
+      "eval_ORGANIZATION_recall": 0.8153846153846154,
+      "eval_PERSON_f1": 0.887459807073955,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8588957055214724,
-      "eval_PERSON_recall": 0.9333333333333333,
-      "eval_QUANTITY_f1": 0.7272727272727273,
+      "eval_PERSON_precision": 0.8571428571428571,
+      "eval_PERSON_recall": 0.92,
+      "eval_QUANTITY_f1": 0.7076923076923076,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6486486486486487,
-      "eval_QUANTITY_recall": 0.8275862068965517,
-      "eval_TIME_f1": 0.861111111111111,
+      "eval_QUANTITY_precision": 0.6388888888888888,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8157894736842106,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8157894736842105,
+      "eval_TIME_precision": 0.7380952380952381,
       "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.13497987389564514,
-      "eval_overall_accuracy": 0.9625892197883338,
-      "eval_overall_f1": 0.8253968253968255,
-      "eval_overall_precision": 0.7703703703703704,
-      "eval_overall_recall": 0.8888888888888888,
-      "eval_runtime": 1.0176,
-      "eval_samples_per_second": 183.773,
-      "eval_steps_per_second": 2.948,
+      "eval_loss": 0.14225105941295624,
+      "eval_overall_accuracy": 0.9579128722618755,
+      "eval_overall_f1": 0.8116710875331565,
+      "eval_overall_precision": 0.7593052109181141,
+      "eval_overall_recall": 0.8717948717948718,
+      "eval_runtime": 0.3776,
+      "eval_samples_per_second": 495.257,
+      "eval_steps_per_second": 7.945,
       "step": 10176
     },
     {
       "epoch": 97.0,
-      "grad_norm": 0.7363091111183167,
+      "grad_norm": 0.8223116397857666,
       "learning_rate": 1.5e-06,
-      "loss": 0.0659,
+      "loss": 0.0672,
       "step": 10282
     },
     {
       "epoch": 97.0,
-      "eval_LOCATION_f1": 0.8051948051948052,
+      "eval_LOCATION_f1": 0.7820512820512822,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7654320987654321,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7333333333333334,
+      "eval_LOCATION_precision": 0.7349397590361446,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.726027397260274,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6470588235294118,
-      "eval_ORGANIZATION_recall": 0.8461538461538461,
-      "eval_PERSON_f1": 0.8945686900958466,
+      "eval_ORGANIZATION_precision": 0.654320987654321,
+      "eval_ORGANIZATION_recall": 0.8153846153846154,
+      "eval_PERSON_f1": 0.887459807073955,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8588957055214724,
-      "eval_PERSON_recall": 0.9333333333333333,
-      "eval_QUANTITY_f1": 0.7272727272727273,
+      "eval_PERSON_precision": 0.8571428571428571,
+      "eval_PERSON_recall": 0.92,
+      "eval_QUANTITY_f1": 0.7076923076923076,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6486486486486487,
-      "eval_QUANTITY_recall": 0.8275862068965517,
-      "eval_TIME_f1": 0.861111111111111,
+      "eval_QUANTITY_precision": 0.6388888888888888,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8157894736842106,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8157894736842105,
+      "eval_TIME_precision": 0.7380952380952381,
       "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.13355118036270142,
-      "eval_overall_accuracy": 0.9618508491262614,
-      "eval_overall_f1": 0.8264900662251655,
-      "eval_overall_precision": 0.7722772277227723,
-      "eval_overall_recall": 0.8888888888888888,
-      "eval_runtime": 0.9627,
-      "eval_samples_per_second": 194.239,
-      "eval_steps_per_second": 3.116,
+      "eval_loss": 0.14218071103096008,
+      "eval_overall_accuracy": 0.9579128722618755,
+      "eval_overall_f1": 0.8116710875331565,
+      "eval_overall_precision": 0.7593052109181141,
+      "eval_overall_recall": 0.8717948717948718,
+      "eval_runtime": 0.3739,
+      "eval_samples_per_second": 500.069,
+      "eval_steps_per_second": 8.022,
       "step": 10282
     },
     {
       "epoch": 98.0,
-      "grad_norm": 0.7795009613037109,
+      "grad_norm": 3.7341294288635254,
       "learning_rate": 1.0000000000000002e-06,
-      "loss": 0.0679,
+      "loss": 0.0674,
       "step": 10388
     },
     {
       "epoch": 98.0,
-      "eval_LOCATION_f1": 0.8051948051948052,
+      "eval_LOCATION_f1": 0.7820512820512822,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7654320987654321,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7333333333333334,
+      "eval_LOCATION_precision": 0.7349397590361446,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.7346938775510204,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6470588235294118,
-      "eval_ORGANIZATION_recall": 0.8461538461538461,
-      "eval_PERSON_f1": 0.8945686900958466,
+      "eval_ORGANIZATION_precision": 0.6585365853658537,
+      "eval_ORGANIZATION_recall": 0.8307692307692308,
+      "eval_PERSON_f1": 0.887459807073955,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8588957055214724,
-      "eval_PERSON_recall": 0.9333333333333333,
-      "eval_QUANTITY_f1": 0.7272727272727273,
+      "eval_PERSON_precision": 0.8571428571428571,
+      "eval_PERSON_recall": 0.92,
+      "eval_QUANTITY_f1": 0.7076923076923076,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6486486486486487,
-      "eval_QUANTITY_recall": 0.8275862068965517,
-      "eval_TIME_f1": 0.861111111111111,
+      "eval_QUANTITY_precision": 0.6388888888888888,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8157894736842106,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8157894736842105,
+      "eval_TIME_precision": 0.7380952380952381,
       "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.1339399665594101,
-      "eval_overall_accuracy": 0.9620969726802855,
-      "eval_overall_f1": 0.8264900662251655,
-      "eval_overall_precision": 0.7722772277227723,
-      "eval_overall_recall": 0.8888888888888888,
-      "eval_runtime": 0.9645,
-      "eval_samples_per_second": 193.887,
-      "eval_steps_per_second": 3.11,
+      "eval_loss": 0.14276809990406036,
+      "eval_overall_accuracy": 0.9584051193699237,
+      "eval_overall_f1": 0.8132450331125828,
+      "eval_overall_precision": 0.7599009900990099,
+      "eval_overall_recall": 0.8746438746438746,
+      "eval_runtime": 0.3733,
+      "eval_samples_per_second": 500.927,
+      "eval_steps_per_second": 8.036,
       "step": 10388
     },
     {
       "epoch": 99.0,
-      "grad_norm": 0.9519725441932678,
+      "grad_norm": 0.9920492768287659,
       "learning_rate": 5.000000000000001e-07,
-      "loss": 0.0674,
+      "loss": 0.0666,
       "step": 10494
     },
     {
       "epoch": 99.0,
-      "eval_LOCATION_f1": 0.8051948051948052,
+      "eval_LOCATION_f1": 0.7820512820512822,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7654320987654321,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7333333333333334,
+      "eval_LOCATION_precision": 0.7349397590361446,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.726027397260274,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6470588235294118,
-      "eval_ORGANIZATION_recall": 0.8461538461538461,
-      "eval_PERSON_f1": 0.8945686900958466,
+      "eval_ORGANIZATION_precision": 0.654320987654321,
+      "eval_ORGANIZATION_recall": 0.8153846153846154,
+      "eval_PERSON_f1": 0.887459807073955,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8588957055214724,
-      "eval_PERSON_recall": 0.9333333333333333,
-      "eval_QUANTITY_f1": 0.7272727272727273,
+      "eval_PERSON_precision": 0.8571428571428571,
+      "eval_PERSON_recall": 0.92,
+      "eval_QUANTITY_f1": 0.7076923076923076,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6486486486486487,
-      "eval_QUANTITY_recall": 0.8275862068965517,
-      "eval_TIME_f1": 0.861111111111111,
+      "eval_QUANTITY_precision": 0.6388888888888888,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8157894736842106,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8157894736842105,
+      "eval_TIME_precision": 0.7380952380952381,
       "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.13475549221038818,
-      "eval_overall_accuracy": 0.9620969726802855,
-      "eval_overall_f1": 0.8264900662251655,
-      "eval_overall_precision": 0.7722772277227723,
-      "eval_overall_recall": 0.8888888888888888,
-      "eval_runtime": 0.9607,
-      "eval_samples_per_second": 194.64,
-      "eval_steps_per_second": 3.123,
+      "eval_loss": 0.14194943010807037,
+      "eval_overall_accuracy": 0.9581589958158996,
+      "eval_overall_f1": 0.8116710875331565,
+      "eval_overall_precision": 0.7593052109181141,
+      "eval_overall_recall": 0.8717948717948718,
+      "eval_runtime": 0.3735,
+      "eval_samples_per_second": 500.685,
+      "eval_steps_per_second": 8.032,
       "step": 10494
     },
     {
       "epoch": 100.0,
-      "grad_norm": 0.5444777607917786,
+      "grad_norm": 0.6790415644645691,
       "learning_rate": 0.0,
-      "loss": 0.0677,
+      "loss": 0.0693,
       "step": 10600
     },
     {
       "epoch": 100.0,
-      "eval_LOCATION_f1": 0.8051948051948052,
+      "eval_LOCATION_f1": 0.7820512820512822,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7654320987654321,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7333333333333334,
+      "eval_LOCATION_precision": 0.7349397590361446,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.726027397260274,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6470588235294118,
-      "eval_ORGANIZATION_recall": 0.8461538461538461,
-      "eval_PERSON_f1": 0.8945686900958466,
+      "eval_ORGANIZATION_precision": 0.654320987654321,
+      "eval_ORGANIZATION_recall": 0.8153846153846154,
+      "eval_PERSON_f1": 0.887459807073955,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8588957055214724,
-      "eval_PERSON_recall": 0.9333333333333333,
-      "eval_QUANTITY_f1": 0.7272727272727273,
+      "eval_PERSON_precision": 0.8571428571428571,
+      "eval_PERSON_recall": 0.92,
+      "eval_QUANTITY_f1": 0.7076923076923076,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6486486486486487,
-      "eval_QUANTITY_recall": 0.8275862068965517,
-      "eval_TIME_f1": 0.861111111111111,
+      "eval_QUANTITY_precision": 0.6388888888888888,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8157894736842106,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8157894736842105,
+      "eval_TIME_precision": 0.7380952380952381,
       "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.1346171796321869,
-      "eval_overall_accuracy": 0.9620969726802855,
-      "eval_overall_f1": 0.8264900662251655,
-      "eval_overall_precision": 0.7722772277227723,
-      "eval_overall_recall": 0.8888888888888888,
-      "eval_runtime": 0.9671,
-      "eval_samples_per_second": 193.362,
-      "eval_steps_per_second": 3.102,
+      "eval_loss": 0.1422806829214096,
+      "eval_overall_accuracy": 0.9581589958158996,
+      "eval_overall_f1": 0.8116710875331565,
+      "eval_overall_precision": 0.7593052109181141,
+      "eval_overall_recall": 0.8717948717948718,
+      "eval_runtime": 0.3745,
+      "eval_samples_per_second": 499.395,
+      "eval_steps_per_second": 8.012,
       "step": 10600
     },
     {
       "epoch": 100.0,
       "step": 10600,
       "total_flos": 4551589111445376.0,
-      "train_loss": 0.12355997449946854,
-      "train_runtime": 1233.063,
-      "train_samples_per_second": 136.895,
-      "train_steps_per_second": 8.596
+      "train_loss": 0.12385802520895904,
+      "train_runtime": 580.0713,
+      "train_samples_per_second": 290.999,
+      "train_steps_per_second": 18.274
     }
   ],
   "logging_steps": 500,