diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -10,1497 +10,1497 @@
   "log_history": [
     {
       "epoch": 1.0,
-      "grad_norm": 1.1324371099472046,
+      "grad_norm": 1.0955090522766113,
       "learning_rate": 4.9500000000000004e-05,
-      "loss": 0.8428,
+      "loss": 0.8608,
       "step": 96
     },
     {
       "epoch": 1.0,
-      "eval_LOCATION_f1": 0.07272727272727272,
-      "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.23529411764705882,
-      "eval_LOCATION_recall": 0.043010752688172046,
-      "eval_ORGANIZATION_f1": 0.2706766917293233,
-      "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.2317596566523605,
-      "eval_ORGANIZATION_recall": 0.3253012048192771,
-      "eval_PERSON_f1": 0.31309904153354634,
-      "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.28654970760233917,
-      "eval_PERSON_recall": 0.34507042253521125,
-      "eval_loss": 0.3884907364845276,
-      "eval_overall_accuracy": 0.869684499314129,
-      "eval_overall_f1": 0.2603406326034063,
-      "eval_overall_precision": 0.25415676959619954,
-      "eval_overall_recall": 0.26683291770573564,
-      "eval_runtime": 0.6812,
-      "eval_samples_per_second": 249.575,
-      "eval_steps_per_second": 4.404,
+      "eval_LOCATION_f1": 0.03960396039603961,
+      "eval_LOCATION_number": 93,
+      "eval_LOCATION_precision": 0.25,
+      "eval_LOCATION_recall": 0.021505376344086023,
+      "eval_ORGANIZATION_f1": 0.22727272727272724,
+      "eval_ORGANIZATION_number": 166,
+      "eval_ORGANIZATION_precision": 0.21505376344086022,
+      "eval_ORGANIZATION_recall": 0.24096385542168675,
+      "eval_PERSON_f1": 0.2815884476534296,
+      "eval_PERSON_number": 142,
+      "eval_PERSON_precision": 0.28888888888888886,
+      "eval_PERSON_recall": 0.2746478873239437,
+      "eval_loss": 0.3943306505680084,
+      "eval_overall_accuracy": 0.8650205761316873,
+      "eval_overall_f1": 0.2219178082191781,
+      "eval_overall_precision": 0.24620060790273557,
+      "eval_overall_recall": 0.20199501246882792,
+      "eval_runtime": 0.2929,
+      "eval_samples_per_second": 580.415,
+      "eval_steps_per_second": 10.243,
       "step": 96
     },
     {
       "epoch": 2.0,
-      "grad_norm": 0.9713420867919922,
+      "grad_norm": 0.8960049152374268,
       "learning_rate": 4.9e-05,
-      "loss": 0.3617,
+      "loss": 0.3625,
       "step": 192
     },
     {
       "epoch": 2.0,
-      "eval_LOCATION_f1": 0.3596491228070176,
-      "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.3037037037037037,
-      "eval_LOCATION_recall": 0.44086021505376344,
-      "eval_ORGANIZATION_f1": 0.5814696485623003,
-      "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.6190476190476191,
-      "eval_ORGANIZATION_recall": 0.5481927710843374,
-      "eval_PERSON_f1": 0.5706051873198846,
-      "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.48292682926829267,
-      "eval_PERSON_recall": 0.6971830985915493,
-      "eval_loss": 0.21714521944522858,
-      "eval_overall_accuracy": 0.9316872427983539,
-      "eval_overall_f1": 0.5202702702702703,
-      "eval_overall_precision": 0.47433264887063653,
-      "eval_overall_recall": 0.5760598503740648,
-      "eval_runtime": 0.636,
-      "eval_samples_per_second": 267.295,
-      "eval_steps_per_second": 4.717,
+      "eval_LOCATION_f1": 0.4050632911392405,
+      "eval_LOCATION_number": 93,
+      "eval_LOCATION_precision": 0.3333333333333333,
+      "eval_LOCATION_recall": 0.5161290322580645,
+      "eval_ORGANIZATION_f1": 0.6056782334384858,
+      "eval_ORGANIZATION_number": 166,
+      "eval_ORGANIZATION_precision": 0.6357615894039735,
+      "eval_ORGANIZATION_recall": 0.5783132530120482,
+      "eval_PERSON_f1": 0.6726726726726726,
+      "eval_PERSON_number": 142,
+      "eval_PERSON_precision": 0.5863874345549738,
+      "eval_PERSON_recall": 0.7887323943661971,
+      "eval_loss": 0.21519558131694794,
+      "eval_overall_accuracy": 0.9404663923182441,
+      "eval_overall_f1": 0.5772266065388952,
+      "eval_overall_precision": 0.5267489711934157,
+      "eval_overall_recall": 0.6384039900249376,
+      "eval_runtime": 0.2904,
+      "eval_samples_per_second": 585.432,
+      "eval_steps_per_second": 10.331,
       "step": 192
     },
     {
       "epoch": 3.0,
-      "grad_norm": 0.8544023036956787,
+      "grad_norm": 0.8648101687431335,
       "learning_rate": 4.85e-05,
-      "loss": 0.1955,
+      "loss": 0.1958,
       "step": 288
     },
     {
       "epoch": 3.0,
-      "eval_LOCATION_f1": 0.7727272727272727,
-      "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.8192771084337349,
-      "eval_LOCATION_recall": 0.7311827956989247,
-      "eval_ORGANIZATION_f1": 0.7717391304347826,
-      "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.7029702970297029,
-      "eval_ORGANIZATION_recall": 0.8554216867469879,
-      "eval_PERSON_f1": 0.968421052631579,
-      "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.965034965034965,
-      "eval_PERSON_recall": 0.971830985915493,
-      "eval_loss": 0.1000150665640831,
-      "eval_overall_accuracy": 0.9681755829903979,
-      "eval_overall_f1": 0.8395657418576598,
-      "eval_overall_precision": 0.8130841121495327,
-      "eval_overall_recall": 0.8678304239401496,
-      "eval_runtime": 0.6673,
-      "eval_samples_per_second": 254.774,
-      "eval_steps_per_second": 4.496,
+      "eval_LOCATION_f1": 0.7909604519774012,
+      "eval_LOCATION_number": 93,
+      "eval_LOCATION_precision": 0.8333333333333334,
+      "eval_LOCATION_recall": 0.7526881720430108,
+      "eval_ORGANIZATION_f1": 0.7342465753424658,
+      "eval_ORGANIZATION_number": 166,
+      "eval_ORGANIZATION_precision": 0.6733668341708543,
+      "eval_ORGANIZATION_recall": 0.8072289156626506,
+      "eval_PERSON_f1": 0.9655172413793103,
+      "eval_PERSON_number": 142,
+      "eval_PERSON_precision": 0.9459459459459459,
+      "eval_PERSON_recall": 0.9859154929577465,
+      "eval_loss": 0.09694702178239822,
+      "eval_overall_accuracy": 0.9676268861454047,
+      "eval_overall_f1": 0.826923076923077,
+      "eval_overall_precision": 0.7981438515081206,
+      "eval_overall_recall": 0.85785536159601,
+      "eval_runtime": 0.2959,
+      "eval_samples_per_second": 574.554,
+      "eval_steps_per_second": 10.139,
       "step": 288
     },
     {
       "epoch": 4.0,
-      "grad_norm": 1.1749948263168335,
+      "grad_norm": 1.2493984699249268,
       "learning_rate": 4.8e-05,
-      "loss": 0.1335,
+      "loss": 0.1348,
       "step": 384
     },
     {
       "epoch": 4.0,
-      "eval_LOCATION_f1": 0.7959183673469389,
+      "eval_LOCATION_f1": 0.7536231884057972,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.7572815533980582,
+      "eval_LOCATION_precision": 0.6842105263157895,
       "eval_LOCATION_recall": 0.8387096774193549,
-      "eval_ORGANIZATION_f1": 0.8109589041095889,
+      "eval_ORGANIZATION_f1": 0.8033240997229917,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.7437185929648241,
-      "eval_ORGANIZATION_recall": 0.891566265060241,
-      "eval_PERSON_f1": 0.9652777777777778,
-      "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.952054794520548,
-      "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.08597932755947113,
-      "eval_overall_accuracy": 0.9714677640603566,
-      "eval_overall_f1": 0.8598351001177856,
-      "eval_overall_precision": 0.8147321428571429,
-      "eval_overall_recall": 0.9102244389027432,
-      "eval_runtime": 0.6604,
-      "eval_samples_per_second": 257.436,
-      "eval_steps_per_second": 4.543,
+      "eval_ORGANIZATION_precision": 0.7435897435897436,
+      "eval_ORGANIZATION_recall": 0.8734939759036144,
+      "eval_PERSON_f1": 0.979020979020979,
+      "eval_PERSON_number": 142,
+      "eval_PERSON_precision": 0.9722222222222222,
+      "eval_PERSON_recall": 0.9859154929577465,
+      "eval_loss": 0.08365878462791443,
+      "eval_overall_accuracy": 0.9711934156378601,
+      "eval_overall_f1": 0.8501170960187353,
+      "eval_overall_precision": 0.8013245033112583,
+      "eval_overall_recall": 0.9052369077306733,
+      "eval_runtime": 0.2924,
+      "eval_samples_per_second": 581.361,
+      "eval_steps_per_second": 10.259,
       "step": 384
     },
     {
       "epoch": 5.0,
-      "grad_norm": 0.6283699870109558,
+      "grad_norm": 0.48399850726127625,
       "learning_rate": 4.75e-05,
-      "loss": 0.1071,
+      "loss": 0.1081,
       "step": 480
     },
     {
       "epoch": 5.0,
-      "eval_LOCATION_f1": 0.8686868686868686,
+      "eval_LOCATION_f1": 0.8633879781420764,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.819047619047619,
-      "eval_LOCATION_recall": 0.9247311827956989,
-      "eval_ORGANIZATION_f1": 0.880952380952381,
+      "eval_LOCATION_precision": 0.8777777777777778,
+      "eval_LOCATION_recall": 0.8494623655913979,
+      "eval_ORGANIZATION_f1": 0.8563049853372433,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.8705882352941177,
-      "eval_ORGANIZATION_recall": 0.891566265060241,
-      "eval_PERSON_f1": 0.9754385964912281,
+      "eval_ORGANIZATION_precision": 0.8342857142857143,
+      "eval_ORGANIZATION_recall": 0.8795180722891566,
+      "eval_PERSON_f1": 0.971830985915493,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.972027972027972,
-      "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.06288647651672363,
-      "eval_overall_accuracy": 0.9813443072702333,
-      "eval_overall_f1": 0.910866910866911,
-      "eval_overall_precision": 0.8923444976076556,
-      "eval_overall_recall": 0.9301745635910225,
-      "eval_runtime": 0.6328,
-      "eval_samples_per_second": 268.66,
-      "eval_steps_per_second": 4.741,
+      "eval_PERSON_precision": 0.971830985915493,
+      "eval_PERSON_recall": 0.971830985915493,
+      "eval_loss": 0.06440697610378265,
+      "eval_overall_accuracy": 0.9807956104252401,
+      "eval_overall_f1": 0.8985148514851485,
+      "eval_overall_precision": 0.8918918918918919,
+      "eval_overall_recall": 0.9052369077306733,
+      "eval_runtime": 0.2962,
+      "eval_samples_per_second": 573.926,
+      "eval_steps_per_second": 10.128,
       "step": 480
     },
     {
       "epoch": 6.0,
-      "grad_norm": 0.7255609035491943,
+      "grad_norm": 0.6719767451286316,
       "learning_rate": 4.7e-05,
-      "loss": 0.0956,
+      "loss": 0.0952,
       "step": 576
     },
     {
       "epoch": 6.0,
-      "eval_LOCATION_f1": 0.8682926829268293,
+      "eval_LOCATION_f1": 0.8195121951219512,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.7946428571428571,
-      "eval_LOCATION_recall": 0.956989247311828,
-      "eval_ORGANIZATION_f1": 0.8875739644970415,
+      "eval_LOCATION_precision": 0.75,
+      "eval_LOCATION_recall": 0.9032258064516129,
+      "eval_ORGANIZATION_f1": 0.8580060422960725,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.872093023255814,
-      "eval_ORGANIZATION_recall": 0.9036144578313253,
+      "eval_ORGANIZATION_precision": 0.8606060606060606,
+      "eval_ORGANIZATION_recall": 0.8554216867469879,
       "eval_PERSON_f1": 0.9823321554770318,
       "eval_PERSON_number": 142,
       "eval_PERSON_precision": 0.9858156028368794,
       "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.053115665912628174,
-      "eval_overall_accuracy": 0.9829903978052126,
-      "eval_overall_f1": 0.9152542372881356,
-      "eval_overall_precision": 0.8894117647058823,
-      "eval_overall_recall": 0.942643391521197,
-      "eval_runtime": 0.627,
-      "eval_samples_per_second": 271.122,
-      "eval_steps_per_second": 4.785,
+      "eval_loss": 0.05366777628660202,
+      "eval_overall_accuracy": 0.9824417009602194,
+      "eval_overall_f1": 0.8913308913308914,
+      "eval_overall_precision": 0.8732057416267942,
+      "eval_overall_recall": 0.9102244389027432,
+      "eval_runtime": 0.2937,
+      "eval_samples_per_second": 578.83,
+      "eval_steps_per_second": 10.215,
       "step": 576
     },
     {
       "epoch": 7.0,
-      "grad_norm": 1.3008713722229004,
+      "grad_norm": 1.178981900215149,
       "learning_rate": 4.6500000000000005e-05,
-      "loss": 0.083,
+      "loss": 0.0875,
       "step": 672
     },
     {
       "epoch": 7.0,
-      "eval_LOCATION_f1": 0.8663101604278075,
+      "eval_LOCATION_f1": 0.849740932642487,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.8617021276595744,
-      "eval_LOCATION_recall": 0.8709677419354839,
-      "eval_ORGANIZATION_f1": 0.8725212464589235,
+      "eval_LOCATION_precision": 0.82,
+      "eval_LOCATION_recall": 0.8817204301075269,
+      "eval_ORGANIZATION_f1": 0.888888888888889,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.8235294117647058,
-      "eval_ORGANIZATION_recall": 0.927710843373494,
-      "eval_PERSON_f1": 0.9823321554770318,
+      "eval_ORGANIZATION_precision": 0.8432432432432433,
+      "eval_ORGANIZATION_recall": 0.9397590361445783,
+      "eval_PERSON_f1": 0.9788732394366197,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9858156028368794,
+      "eval_PERSON_precision": 0.9788732394366197,
       "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.05037233233451843,
-      "eval_overall_accuracy": 0.9832647462277092,
-      "eval_overall_f1": 0.9088699878493317,
-      "eval_overall_precision": 0.8862559241706162,
-      "eval_overall_recall": 0.9326683291770573,
-      "eval_runtime": 0.6164,
-      "eval_samples_per_second": 275.782,
-      "eval_steps_per_second": 4.867,
+      "eval_loss": 0.05959905683994293,
+      "eval_overall_accuracy": 0.9816186556927298,
+      "eval_overall_f1": 0.9106280193236714,
+      "eval_overall_precision": 0.882903981264637,
+      "eval_overall_recall": 0.940149625935162,
+      "eval_runtime": 0.2926,
+      "eval_samples_per_second": 581.049,
+      "eval_steps_per_second": 10.254,
       "step": 672
     },
     {
       "epoch": 8.0,
-      "grad_norm": 0.7086864113807678,
+      "grad_norm": 0.6990514993667603,
       "learning_rate": 4.600000000000001e-05,
-      "loss": 0.0784,
+      "loss": 0.073,
       "step": 768
     },
     {
       "epoch": 8.0,
-      "eval_LOCATION_f1": 0.8502415458937198,
+      "eval_LOCATION_f1": 0.883248730964467,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.7719298245614035,
-      "eval_LOCATION_recall": 0.946236559139785,
-      "eval_ORGANIZATION_f1": 0.8753799392097266,
+      "eval_LOCATION_precision": 0.8365384615384616,
+      "eval_LOCATION_recall": 0.9354838709677419,
+      "eval_ORGANIZATION_f1": 0.9032258064516129,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.8834355828220859,
-      "eval_ORGANIZATION_recall": 0.8674698795180723,
-      "eval_PERSON_f1": 0.9823321554770318,
+      "eval_ORGANIZATION_precision": 0.88,
+      "eval_ORGANIZATION_recall": 0.927710843373494,
+      "eval_PERSON_f1": 0.9788732394366197,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9858156028368794,
+      "eval_PERSON_precision": 0.9788732394366197,
       "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.05147531256079674,
-      "eval_overall_accuracy": 0.9818930041152263,
-      "eval_overall_f1": 0.9059829059829061,
-      "eval_overall_precision": 0.8875598086124402,
-      "eval_overall_recall": 0.9251870324189526,
-      "eval_runtime": 0.626,
-      "eval_samples_per_second": 271.558,
-      "eval_steps_per_second": 4.792,
+      "eval_loss": 0.05377694219350815,
+      "eval_overall_accuracy": 0.9835390946502057,
+      "eval_overall_f1": 0.9245742092457422,
+      "eval_overall_precision": 0.9026128266033254,
+      "eval_overall_recall": 0.9476309226932669,
+      "eval_runtime": 0.2917,
+      "eval_samples_per_second": 582.8,
+      "eval_steps_per_second": 10.285,
       "step": 768
     },
     {
       "epoch": 9.0,
-      "grad_norm": 0.5022886395454407,
+      "grad_norm": 0.3934107720851898,
       "learning_rate": 4.55e-05,
-      "loss": 0.072,
+      "loss": 0.0696,
       "step": 864
     },
     {
       "epoch": 9.0,
-      "eval_LOCATION_f1": 0.8958333333333334,
+      "eval_LOCATION_f1": 0.8900523560209423,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.8686868686868687,
-      "eval_LOCATION_recall": 0.9247311827956989,
-      "eval_ORGANIZATION_f1": 0.9009009009009009,
+      "eval_LOCATION_precision": 0.8673469387755102,
+      "eval_LOCATION_recall": 0.9139784946236559,
+      "eval_ORGANIZATION_f1": 0.88955223880597,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.8982035928143712,
-      "eval_ORGANIZATION_recall": 0.9036144578313253,
-      "eval_PERSON_f1": 0.9823321554770318,
+      "eval_ORGANIZATION_precision": 0.8816568047337278,
+      "eval_ORGANIZATION_recall": 0.8975903614457831,
+      "eval_PERSON_f1": 0.971830985915493,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9858156028368794,
-      "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.04265189915895462,
-      "eval_overall_accuracy": 0.9860082304526749,
-      "eval_overall_f1": 0.9282178217821783,
-      "eval_overall_precision": 0.9213759213759214,
-      "eval_overall_recall": 0.9351620947630923,
-      "eval_runtime": 0.6359,
-      "eval_samples_per_second": 267.337,
-      "eval_steps_per_second": 4.718,
+      "eval_PERSON_precision": 0.971830985915493,
+      "eval_PERSON_recall": 0.971830985915493,
+      "eval_loss": 0.04754691570997238,
+      "eval_overall_accuracy": 0.9857338820301783,
+      "eval_overall_f1": 0.9185185185185184,
+      "eval_overall_precision": 0.9095354523227384,
+      "eval_overall_recall": 0.9276807980049875,
+      "eval_runtime": 0.2908,
+      "eval_samples_per_second": 584.625,
+      "eval_steps_per_second": 10.317,
       "step": 864
     },
     {
       "epoch": 10.0,
-      "grad_norm": 0.7925252318382263,
+      "grad_norm": 0.8873964548110962,
       "learning_rate": 4.5e-05,
-      "loss": 0.0688,
+      "loss": 0.064,
       "step": 960
     },
     {
       "epoch": 10.0,
-      "eval_LOCATION_f1": 0.89,
+      "eval_LOCATION_f1": 0.922279792746114,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.8317757009345794,
+      "eval_LOCATION_precision": 0.89,
       "eval_LOCATION_recall": 0.956989247311828,
-      "eval_ORGANIZATION_f1": 0.9107142857142857,
+      "eval_ORGANIZATION_f1": 0.9179331306990882,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9,
-      "eval_ORGANIZATION_recall": 0.9216867469879518,
-      "eval_PERSON_f1": 0.968421052631579,
+      "eval_ORGANIZATION_precision": 0.9263803680981595,
+      "eval_ORGANIZATION_recall": 0.9096385542168675,
+      "eval_PERSON_f1": 0.9754385964912281,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.965034965034965,
-      "eval_PERSON_recall": 0.971830985915493,
-      "eval_loss": 0.046313099563121796,
-      "eval_overall_accuracy": 0.9849108367626886,
-      "eval_overall_f1": 0.9257003654080391,
-      "eval_overall_precision": 0.9047619047619048,
-      "eval_overall_recall": 0.9476309226932669,
-      "eval_runtime": 0.635,
-      "eval_samples_per_second": 267.722,
-      "eval_steps_per_second": 4.725,
+      "eval_PERSON_precision": 0.972027972027972,
+      "eval_PERSON_recall": 0.9788732394366197,
+      "eval_loss": 0.052041731774806976,
+      "eval_overall_accuracy": 0.9868312757201646,
+      "eval_overall_f1": 0.9392812887236679,
+      "eval_overall_precision": 0.9334975369458128,
+      "eval_overall_recall": 0.9451371571072319,
+      "eval_runtime": 0.2893,
+      "eval_samples_per_second": 587.659,
+      "eval_steps_per_second": 10.37,
       "step": 960
     },
     {
       "epoch": 11.0,
-      "grad_norm": 0.376903772354126,
+      "grad_norm": 0.5631869435310364,
       "learning_rate": 4.4500000000000004e-05,
-      "loss": 0.0621,
+      "loss": 0.0626,
       "step": 1056
     },
     {
       "epoch": 11.0,
-      "eval_LOCATION_f1": 0.914572864321608,
+      "eval_LOCATION_f1": 0.9175257731958764,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.8584905660377359,
-      "eval_LOCATION_recall": 0.978494623655914,
-      "eval_ORGANIZATION_f1": 0.9174311926605505,
+      "eval_LOCATION_precision": 0.8811881188118812,
+      "eval_LOCATION_recall": 0.956989247311828,
+      "eval_ORGANIZATION_f1": 0.9169230769230768,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9316770186335404,
-      "eval_ORGANIZATION_recall": 0.9036144578313253,
-      "eval_PERSON_f1": 0.9823321554770318,
+      "eval_ORGANIZATION_precision": 0.9371069182389937,
+      "eval_ORGANIZATION_recall": 0.8975903614457831,
+      "eval_PERSON_f1": 0.968421052631579,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9858156028368794,
-      "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.04419618099927902,
-      "eval_overall_accuracy": 0.9860082304526749,
-      "eval_overall_f1": 0.9394313967861558,
-      "eval_overall_precision": 0.9313725490196079,
-      "eval_overall_recall": 0.9476309226932669,
-      "eval_runtime": 0.6308,
-      "eval_samples_per_second": 269.488,
-      "eval_steps_per_second": 4.756,
+      "eval_PERSON_precision": 0.965034965034965,
+      "eval_PERSON_recall": 0.971830985915493,
+      "eval_loss": 0.043987125158309937,
+      "eval_overall_accuracy": 0.9879286694101509,
+      "eval_overall_f1": 0.9353233830845771,
+      "eval_overall_precision": 0.9330024813895782,
+      "eval_overall_recall": 0.9376558603491272,
+      "eval_runtime": 0.2913,
+      "eval_samples_per_second": 583.542,
+      "eval_steps_per_second": 10.298,
       "step": 1056
     },
     {
       "epoch": 12.0,
-      "grad_norm": 0.7413063645362854,
+      "grad_norm": 0.559025228023529,
       "learning_rate": 4.4000000000000006e-05,
-      "loss": 0.0595,
+      "loss": 0.0554,
       "step": 1152
     },
     {
       "epoch": 12.0,
-      "eval_LOCATION_f1": 0.9374999999999999,
+      "eval_LOCATION_f1": 0.9139784946236559,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.9090909090909091,
-      "eval_LOCATION_recall": 0.967741935483871,
-      "eval_ORGANIZATION_f1": 0.9235474006116209,
+      "eval_LOCATION_precision": 0.9139784946236559,
+      "eval_LOCATION_recall": 0.9139784946236559,
+      "eval_ORGANIZATION_f1": 0.9058823529411766,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.937888198757764,
-      "eval_ORGANIZATION_recall": 0.9096385542168675,
-      "eval_PERSON_f1": 0.9752650176678446,
+      "eval_ORGANIZATION_precision": 0.8850574712643678,
+      "eval_ORGANIZATION_recall": 0.927710843373494,
+      "eval_PERSON_f1": 0.9754385964912281,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9787234042553191,
-      "eval_PERSON_recall": 0.971830985915493,
-      "eval_loss": 0.045279767364263535,
-      "eval_overall_accuracy": 0.9873799725651577,
-      "eval_overall_f1": 0.9451371571072319,
-      "eval_overall_precision": 0.9451371571072319,
-      "eval_overall_recall": 0.9451371571072319,
-      "eval_runtime": 0.628,
-      "eval_samples_per_second": 270.712,
-      "eval_steps_per_second": 4.777,
+      "eval_PERSON_precision": 0.972027972027972,
+      "eval_PERSON_recall": 0.9788732394366197,
+      "eval_loss": 0.04547671601176262,
+      "eval_overall_accuracy": 0.9871056241426612,
+      "eval_overall_f1": 0.9321824907521579,
+      "eval_overall_precision": 0.9219512195121952,
+      "eval_overall_recall": 0.942643391521197,
+      "eval_runtime": 0.2923,
+      "eval_samples_per_second": 581.688,
+      "eval_steps_per_second": 10.265,
       "step": 1152
     },
     {
       "epoch": 13.0,
-      "grad_norm": 0.575614869594574,
+      "grad_norm": 1.0373061895370483,
       "learning_rate": 4.35e-05,
-      "loss": 0.0576,
+      "loss": 0.0548,
       "step": 1248
     },
     {
       "epoch": 13.0,
-      "eval_LOCATION_f1": 0.91,
+      "eval_LOCATION_f1": 0.9197860962566845,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.8504672897196262,
-      "eval_LOCATION_recall": 0.978494623655914,
-      "eval_ORGANIZATION_f1": 0.9268292682926829,
+      "eval_LOCATION_precision": 0.9148936170212766,
+      "eval_LOCATION_recall": 0.9247311827956989,
+      "eval_ORGANIZATION_f1": 0.9317507418397626,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9382716049382716,
-      "eval_ORGANIZATION_recall": 0.9156626506024096,
-      "eval_PERSON_f1": 0.9788732394366197,
+      "eval_ORGANIZATION_precision": 0.9181286549707602,
+      "eval_ORGANIZATION_recall": 0.9457831325301205,
+      "eval_PERSON_f1": 0.9754385964912281,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9788732394366197,
+      "eval_PERSON_precision": 0.972027972027972,
       "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.04553345590829849,
-      "eval_overall_accuracy": 0.9868312757201646,
-      "eval_overall_f1": 0.9408866995073891,
-      "eval_overall_precision": 0.9294403892944039,
+      "eval_loss": 0.04833203926682472,
+      "eval_overall_accuracy": 0.988477366255144,
+      "eval_overall_f1": 0.9443757725587144,
+      "eval_overall_precision": 0.9362745098039216,
       "eval_overall_recall": 0.9526184538653366,
-      "eval_runtime": 0.6217,
-      "eval_samples_per_second": 273.463,
-      "eval_steps_per_second": 4.826,
+      "eval_runtime": 0.2913,
+      "eval_samples_per_second": 583.67,
+      "eval_steps_per_second": 10.3,
       "step": 1248
     },
     {
       "epoch": 14.0,
-      "grad_norm": 0.5367061495780945,
+      "grad_norm": 0.2547779381275177,
       "learning_rate": 4.3e-05,
-      "loss": 0.054,
+      "loss": 0.0504,
       "step": 1344
     },
     {
       "epoch": 14.0,
-      "eval_LOCATION_f1": 0.9361702127659575,
+      "eval_LOCATION_f1": 0.9247311827956989,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.9263157894736842,
-      "eval_LOCATION_recall": 0.946236559139785,
-      "eval_ORGANIZATION_f1": 0.9212121212121213,
+      "eval_LOCATION_precision": 0.9247311827956989,
+      "eval_LOCATION_recall": 0.9247311827956989,
+      "eval_ORGANIZATION_f1": 0.9461077844311376,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.926829268292683,
-      "eval_ORGANIZATION_recall": 0.9156626506024096,
-      "eval_PERSON_f1": 0.971830985915493,
+      "eval_ORGANIZATION_precision": 0.9404761904761905,
+      "eval_ORGANIZATION_recall": 0.9518072289156626,
+      "eval_PERSON_f1": 0.968421052631579,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.971830985915493,
+      "eval_PERSON_precision": 0.965034965034965,
       "eval_PERSON_recall": 0.971830985915493,
-      "eval_loss": 0.041482534259557724,
-      "eval_overall_accuracy": 0.9879286694101509,
-      "eval_overall_f1": 0.942643391521197,
-      "eval_overall_precision": 0.942643391521197,
-      "eval_overall_recall": 0.942643391521197,
-      "eval_runtime": 0.6192,
-      "eval_samples_per_second": 274.558,
-      "eval_steps_per_second": 4.845,
+      "eval_loss": 0.04443012550473213,
+      "eval_overall_accuracy": 0.9895747599451303,
+      "eval_overall_f1": 0.9490683229813663,
+      "eval_overall_precision": 0.9455445544554455,
+      "eval_overall_recall": 0.9526184538653366,
+      "eval_runtime": 0.2937,
+      "eval_samples_per_second": 578.827,
+      "eval_steps_per_second": 10.215,
       "step": 1344
     },
     {
       "epoch": 15.0,
-      "grad_norm": 0.7122671604156494,
+      "grad_norm": 0.5698651075363159,
       "learning_rate": 4.25e-05,
-      "loss": 0.0491,
+      "loss": 0.0494,
       "step": 1440
     },
     {
       "epoch": 15.0,
-      "eval_LOCATION_f1": 0.925531914893617,
+      "eval_LOCATION_f1": 0.9354838709677419,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.9157894736842105,
+      "eval_LOCATION_precision": 0.9354838709677419,
       "eval_LOCATION_recall": 0.9354838709677419,
-      "eval_ORGANIZATION_f1": 0.9221556886227544,
+      "eval_ORGANIZATION_f1": 0.9317507418397626,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9166666666666666,
-      "eval_ORGANIZATION_recall": 0.927710843373494,
-      "eval_PERSON_f1": 0.9823321554770318,
+      "eval_ORGANIZATION_precision": 0.9181286549707602,
+      "eval_ORGANIZATION_recall": 0.9457831325301205,
+      "eval_PERSON_f1": 0.9754385964912281,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9858156028368794,
+      "eval_PERSON_precision": 0.972027972027972,
       "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.036706916987895966,
-      "eval_overall_accuracy": 0.988477366255144,
-      "eval_overall_f1": 0.9440993788819876,
-      "eval_overall_precision": 0.9405940594059405,
-      "eval_overall_recall": 0.9476309226932669,
-      "eval_runtime": 0.6348,
-      "eval_samples_per_second": 267.784,
-      "eval_steps_per_second": 4.726,
+      "eval_loss": 0.042747754603624344,
+      "eval_overall_accuracy": 0.9887517146776406,
+      "eval_overall_f1": 0.9480198019801981,
+      "eval_overall_precision": 0.941031941031941,
+      "eval_overall_recall": 0.9551122194513716,
+      "eval_runtime": 0.2936,
+      "eval_samples_per_second": 578.975,
+      "eval_steps_per_second": 10.217,
       "step": 1440
     },
     {
       "epoch": 16.0,
-      "grad_norm": 1.1978126764297485,
+      "grad_norm": 0.5443034172058105,
       "learning_rate": 4.2e-05,
-      "loss": 0.0468,
+      "loss": 0.0445,
       "step": 1536
     },
     {
       "epoch": 16.0,
-      "eval_LOCATION_f1": 0.9128205128205128,
+      "eval_LOCATION_f1": 0.8989898989898989,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.8725490196078431,
+      "eval_LOCATION_precision": 0.8476190476190476,
       "eval_LOCATION_recall": 0.956989247311828,
-      "eval_ORGANIZATION_f1": 0.9216867469879518,
+      "eval_ORGANIZATION_f1": 0.9123867069486404,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9216867469879518,
-      "eval_ORGANIZATION_recall": 0.9216867469879518,
-      "eval_PERSON_f1": 0.9823321554770318,
+      "eval_ORGANIZATION_precision": 0.9151515151515152,
+      "eval_ORGANIZATION_recall": 0.9096385542168675,
+      "eval_PERSON_f1": 0.968421052631579,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9858156028368794,
-      "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.04370429739356041,
-      "eval_overall_accuracy": 0.9865569272976681,
-      "eval_overall_f1": 0.9407407407407408,
-      "eval_overall_precision": 0.9315403422982885,
-      "eval_overall_recall": 0.9501246882793017,
-      "eval_runtime": 0.6383,
-      "eval_samples_per_second": 266.338,
-      "eval_steps_per_second": 4.7,
+      "eval_PERSON_precision": 0.965034965034965,
+      "eval_PERSON_recall": 0.971830985915493,
+      "eval_loss": 0.049201611429452896,
+      "eval_overall_accuracy": 0.9873799725651577,
+      "eval_overall_f1": 0.9287469287469288,
+      "eval_overall_precision": 0.9152542372881356,
+      "eval_overall_recall": 0.942643391521197,
+      "eval_runtime": 0.2919,
+      "eval_samples_per_second": 582.42,
+      "eval_steps_per_second": 10.278,
       "step": 1536
     },
     {
       "epoch": 17.0,
-      "grad_norm": 0.49218180775642395,
+      "grad_norm": 0.4349294602870941,
       "learning_rate": 4.15e-05,
-      "loss": 0.0466,
+      "loss": 0.0457,
       "step": 1632
     },
     {
       "epoch": 17.0,
-      "eval_LOCATION_f1": 0.9157894736842105,
+      "eval_LOCATION_f1": 0.9270833333333334,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.8969072164948454,
-      "eval_LOCATION_recall": 0.9354838709677419,
-      "eval_ORGANIZATION_f1": 0.913946587537092,
+      "eval_LOCATION_precision": 0.898989898989899,
+      "eval_LOCATION_recall": 0.956989247311828,
+      "eval_ORGANIZATION_f1": 0.9397590361445783,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9005847953216374,
-      "eval_ORGANIZATION_recall": 0.927710843373494,
-      "eval_PERSON_f1": 0.9752650176678446,
+      "eval_ORGANIZATION_precision": 0.9397590361445783,
+      "eval_ORGANIZATION_recall": 0.9397590361445783,
+      "eval_PERSON_f1": 0.9788732394366197,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9787234042553191,
-      "eval_PERSON_recall": 0.971830985915493,
-      "eval_loss": 0.04270460084080696,
-      "eval_overall_accuracy": 0.9876543209876543,
-      "eval_overall_f1": 0.9358024691358026,
-      "eval_overall_precision": 0.9266503667481663,
-      "eval_overall_recall": 0.9451371571072319,
-      "eval_runtime": 0.6391,
-      "eval_samples_per_second": 265.99,
-      "eval_steps_per_second": 4.694,
+      "eval_PERSON_precision": 0.9788732394366197,
+      "eval_PERSON_recall": 0.9788732394366197,
+      "eval_loss": 0.04454689472913742,
+      "eval_overall_accuracy": 0.9890260631001372,
+      "eval_overall_f1": 0.9504950495049505,
+      "eval_overall_precision": 0.9434889434889435,
+      "eval_overall_recall": 0.9576059850374065,
+      "eval_runtime": 0.2923,
+      "eval_samples_per_second": 581.575,
+      "eval_steps_per_second": 10.263,
       "step": 1632
     },
     {
       "epoch": 18.0,
-      "grad_norm": 0.3642722964286804,
+      "grad_norm": 0.2558753192424774,
       "learning_rate": 4.1e-05,
-      "loss": 0.0442,
+      "loss": 0.0433,
       "step": 1728
     },
     {
       "epoch": 18.0,
-      "eval_LOCATION_f1": 0.9263157894736843,
+      "eval_LOCATION_f1": 0.9270833333333334,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.9072164948453608,
-      "eval_LOCATION_recall": 0.946236559139785,
-      "eval_ORGANIZATION_f1": 0.9151515151515153,
+      "eval_LOCATION_precision": 0.898989898989899,
+      "eval_LOCATION_recall": 0.956989247311828,
+      "eval_ORGANIZATION_f1": 0.9305135951661632,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9207317073170732,
-      "eval_ORGANIZATION_recall": 0.9096385542168675,
-      "eval_PERSON_f1": 0.9752650176678446,
+      "eval_ORGANIZATION_precision": 0.9333333333333333,
+      "eval_ORGANIZATION_recall": 0.927710843373494,
+      "eval_PERSON_f1": 0.9788732394366197,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9787234042553191,
-      "eval_PERSON_recall": 0.971830985915493,
-      "eval_loss": 0.03979617729783058,
-      "eval_overall_accuracy": 0.9876543209876543,
-      "eval_overall_f1": 0.9389788293897883,
-      "eval_overall_precision": 0.9378109452736318,
-      "eval_overall_recall": 0.940149625935162,
-      "eval_runtime": 0.6398,
-      "eval_samples_per_second": 265.693,
-      "eval_steps_per_second": 4.689,
+      "eval_PERSON_precision": 0.9788732394366197,
+      "eval_PERSON_recall": 0.9788732394366197,
+      "eval_loss": 0.05001579597592354,
+      "eval_overall_accuracy": 0.9895747599451303,
+      "eval_overall_f1": 0.946716232961586,
+      "eval_overall_precision": 0.9408866995073891,
+      "eval_overall_recall": 0.9526184538653366,
+      "eval_runtime": 0.291,
+      "eval_samples_per_second": 584.152,
+      "eval_steps_per_second": 10.309,
       "step": 1728
     },
     {
       "epoch": 19.0,
-      "grad_norm": 1.037690281867981,
+      "grad_norm": 0.414883553981781,
       "learning_rate": 4.05e-05,
-      "loss": 0.0407,
+      "loss": 0.0395,
       "step": 1824
     },
     {
       "epoch": 19.0,
-      "eval_LOCATION_f1": 0.9430051813471503,
+      "eval_LOCATION_f1": 0.9304812834224598,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.91,
-      "eval_LOCATION_recall": 0.978494623655914,
-      "eval_ORGANIZATION_f1": 0.9263803680981595,
+      "eval_LOCATION_precision": 0.925531914893617,
+      "eval_LOCATION_recall": 0.9354838709677419,
+      "eval_ORGANIZATION_f1": 0.9285714285714286,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.94375,
-      "eval_ORGANIZATION_recall": 0.9096385542168675,
-      "eval_PERSON_f1": 0.9752650176678446,
+      "eval_ORGANIZATION_precision": 0.9176470588235294,
+      "eval_ORGANIZATION_recall": 0.9397590361445783,
+      "eval_PERSON_f1": 0.9788732394366197,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9787234042553191,
-      "eval_PERSON_recall": 0.971830985915493,
-      "eval_loss": 0.04469582438468933,
-      "eval_overall_accuracy": 0.9887517146776406,
-      "eval_overall_f1": 0.9476309226932669,
-      "eval_overall_precision": 0.9476309226932669,
-      "eval_overall_recall": 0.9476309226932669,
-      "eval_runtime": 0.6332,
-      "eval_samples_per_second": 268.466,
-      "eval_steps_per_second": 4.738,
+      "eval_PERSON_precision": 0.9788732394366197,
+      "eval_PERSON_recall": 0.9788732394366197,
+      "eval_loss": 0.04901066794991493,
+      "eval_overall_accuracy": 0.9890260631001372,
+      "eval_overall_f1": 0.946716232961586,
+      "eval_overall_precision": 0.9408866995073891,
+      "eval_overall_recall": 0.9526184538653366,
+      "eval_runtime": 0.2903,
+      "eval_samples_per_second": 585.592,
+      "eval_steps_per_second": 10.334,
       "step": 1824
     },
     {
       "epoch": 20.0,
-      "grad_norm": 0.3398281931877136,
+      "grad_norm": 0.6924071311950684,
       "learning_rate": 4e-05,
-      "loss": 0.0391,
+      "loss": 0.0375,
       "step": 1920
     },
     {
       "epoch": 20.0,
-      "eval_LOCATION_f1": 0.9045226130653267,
+      "eval_LOCATION_f1": 0.9263157894736843,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.8490566037735849,
-      "eval_LOCATION_recall": 0.967741935483871,
-      "eval_ORGANIZATION_f1": 0.9057750759878419,
+      "eval_LOCATION_precision": 0.9072164948453608,
+      "eval_LOCATION_recall": 0.946236559139785,
+      "eval_ORGANIZATION_f1": 0.9289940828402367,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9141104294478528,
-      "eval_ORGANIZATION_recall": 0.8975903614457831,
-      "eval_PERSON_f1": 0.9752650176678446,
+      "eval_ORGANIZATION_precision": 0.9127906976744186,
+      "eval_ORGANIZATION_recall": 0.9457831325301205,
+      "eval_PERSON_f1": 0.9788732394366197,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9787234042553191,
-      "eval_PERSON_recall": 0.971830985915493,
-      "eval_loss": 0.04723087698221207,
-      "eval_overall_accuracy": 0.9857338820301783,
-      "eval_overall_f1": 0.9297163995067818,
-      "eval_overall_precision": 0.9195121951219513,
-      "eval_overall_recall": 0.940149625935162,
-      "eval_runtime": 0.6325,
-      "eval_samples_per_second": 268.763,
-      "eval_steps_per_second": 4.743,
+      "eval_PERSON_precision": 0.9788732394366197,
+      "eval_PERSON_recall": 0.9788732394366197,
+      "eval_loss": 0.04936488717794418,
+      "eval_overall_accuracy": 0.9879286694101509,
+      "eval_overall_f1": 0.9458128078817734,
+      "eval_overall_precision": 0.9343065693430657,
+      "eval_overall_recall": 0.9576059850374065,
+      "eval_runtime": 0.2926,
+      "eval_samples_per_second": 581.005,
+      "eval_steps_per_second": 10.253,
       "step": 1920
     },
     {
       "epoch": 21.0,
-      "grad_norm": 0.9918632507324219,
+      "grad_norm": 0.32161277532577515,
       "learning_rate": 3.9500000000000005e-05,
-      "loss": 0.0387,
+      "loss": 0.0394,
       "step": 2016
     },
     {
       "epoch": 21.0,
-      "eval_LOCATION_f1": 0.9191919191919192,
+      "eval_LOCATION_f1": 0.9319371727748691,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.8666666666666667,
-      "eval_LOCATION_recall": 0.978494623655914,
-      "eval_ORGANIZATION_f1": 0.8888888888888888,
+      "eval_LOCATION_precision": 0.9081632653061225,
+      "eval_LOCATION_recall": 0.956989247311828,
+      "eval_ORGANIZATION_f1": 0.9281437125748503,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9113924050632911,
-      "eval_ORGANIZATION_recall": 0.8674698795180723,
-      "eval_PERSON_f1": 0.9823321554770318,
+      "eval_ORGANIZATION_precision": 0.9226190476190477,
+      "eval_ORGANIZATION_recall": 0.9337349397590361,
+      "eval_PERSON_f1": 0.971830985915493,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9858156028368794,
-      "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.046005215495824814,
-      "eval_overall_accuracy": 0.9873799725651577,
-      "eval_overall_f1": 0.929192546583851,
-      "eval_overall_precision": 0.9257425742574258,
-      "eval_overall_recall": 0.9326683291770573,
-      "eval_runtime": 0.6246,
-      "eval_samples_per_second": 272.187,
-      "eval_steps_per_second": 4.803,
-      "step": 2016
-    },
-    {
+      "eval_PERSON_precision": 0.971830985915493,
+      "eval_PERSON_recall": 0.971830985915493,
+      "eval_loss": 0.05034565553069115,
+      "eval_overall_accuracy": 0.9887517146776406,
+      "eval_overall_f1": 0.9443757725587144,
+      "eval_overall_precision": 0.9362745098039216,
+      "eval_overall_recall": 0.9526184538653366,
+      "eval_runtime": 0.2909,
+      "eval_samples_per_second": 584.394,
+      "eval_steps_per_second": 10.313,
+      "step": 2016
+    },
+    {
       "epoch": 22.0,
-      "grad_norm": 0.2701110541820526,
+      "grad_norm": 0.15593692660331726,
       "learning_rate": 3.9000000000000006e-05,
-      "loss": 0.0343,
+      "loss": 0.0364,
       "step": 2112
     },
     {
       "epoch": 22.0,
-      "eval_LOCATION_f1": 0.9473684210526316,
+      "eval_LOCATION_f1": 0.9214659685863874,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.9278350515463918,
-      "eval_LOCATION_recall": 0.967741935483871,
-      "eval_ORGANIZATION_f1": 0.9244712990936556,
+      "eval_LOCATION_precision": 0.8979591836734694,
+      "eval_LOCATION_recall": 0.946236559139785,
+      "eval_ORGANIZATION_f1": 0.920353982300885,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9272727272727272,
-      "eval_ORGANIZATION_recall": 0.9216867469879518,
-      "eval_PERSON_f1": 0.9823321554770318,
+      "eval_ORGANIZATION_precision": 0.9017341040462428,
+      "eval_ORGANIZATION_recall": 0.9397590361445783,
+      "eval_PERSON_f1": 0.9788732394366197,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9858156028368794,
+      "eval_PERSON_precision": 0.9788732394366197,
       "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.04117260500788689,
-      "eval_overall_accuracy": 0.9890260631001372,
-      "eval_overall_f1": 0.9502487562189055,
-      "eval_overall_precision": 0.9478908188585607,
-      "eval_overall_recall": 0.9526184538653366,
-      "eval_runtime": 0.6316,
-      "eval_samples_per_second": 269.142,
-      "eval_steps_per_second": 4.75,
+      "eval_loss": 0.04979310929775238,
+      "eval_overall_accuracy": 0.9879286694101509,
+      "eval_overall_f1": 0.941031941031941,
+      "eval_overall_precision": 0.927360774818402,
+      "eval_overall_recall": 0.9551122194513716,
+      "eval_runtime": 0.2951,
+      "eval_samples_per_second": 576.049,
+      "eval_steps_per_second": 10.166,
       "step": 2112
     },
     {
       "epoch": 23.0,
-      "grad_norm": 0.3225027322769165,
+      "grad_norm": 1.1374483108520508,
       "learning_rate": 3.85e-05,
-      "loss": 0.0331,
+      "loss": 0.035,
       "step": 2208
     },
     {
       "epoch": 23.0,
-      "eval_LOCATION_f1": 0.9157894736842105,
+      "eval_LOCATION_f1": 0.9109947643979057,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.8969072164948454,
+      "eval_LOCATION_precision": 0.8877551020408163,
       "eval_LOCATION_recall": 0.9354838709677419,
-      "eval_ORGANIZATION_f1": 0.9174311926605505,
+      "eval_ORGANIZATION_f1": 0.924924924924925,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9316770186335404,
-      "eval_ORGANIZATION_recall": 0.9036144578313253,
-      "eval_PERSON_f1": 0.9752650176678446,
+      "eval_ORGANIZATION_precision": 0.9221556886227545,
+      "eval_ORGANIZATION_recall": 0.927710843373494,
+      "eval_PERSON_f1": 0.971830985915493,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9787234042553191,
+      "eval_PERSON_precision": 0.971830985915493,
       "eval_PERSON_recall": 0.971830985915493,
-      "eval_loss": 0.04424041137099266,
-      "eval_overall_accuracy": 0.9879286694101509,
-      "eval_overall_f1": 0.9375,
-      "eval_overall_precision": 0.9398496240601504,
-      "eval_overall_recall": 0.9351620947630923,
-      "eval_runtime": 0.6279,
-      "eval_samples_per_second": 270.749,
-      "eval_steps_per_second": 4.778,
+      "eval_loss": 0.04968388378620148,
+      "eval_overall_accuracy": 0.9876543209876543,
+      "eval_overall_f1": 0.9381188118811881,
+      "eval_overall_precision": 0.9312039312039312,
+      "eval_overall_recall": 0.9451371571072319,
+      "eval_runtime": 0.2946,
+      "eval_samples_per_second": 577.009,
+      "eval_steps_per_second": 10.183,
       "step": 2208
     },
     {
       "epoch": 24.0,
-      "grad_norm": 0.8494559526443481,
+      "grad_norm": 0.7729804515838623,
       "learning_rate": 3.8e-05,
-      "loss": 0.0329,
+      "loss": 0.0355,
       "step": 2304
     },
     {
       "epoch": 24.0,
-      "eval_LOCATION_f1": 0.9473684210526316,
+      "eval_LOCATION_f1": 0.9361702127659575,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.9278350515463918,
-      "eval_LOCATION_recall": 0.967741935483871,
-      "eval_ORGANIZATION_f1": 0.9309309309309309,
+      "eval_LOCATION_precision": 0.9263157894736842,
+      "eval_LOCATION_recall": 0.946236559139785,
+      "eval_ORGANIZATION_f1": 0.923076923076923,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9281437125748503,
-      "eval_ORGANIZATION_recall": 0.9337349397590361,
-      "eval_PERSON_f1": 0.9752650176678446,
+      "eval_ORGANIZATION_precision": 0.9069767441860465,
+      "eval_ORGANIZATION_recall": 0.9397590361445783,
+      "eval_PERSON_f1": 0.9823321554770318,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9787234042553191,
-      "eval_PERSON_recall": 0.971830985915493,
-      "eval_loss": 0.04214971885085106,
-      "eval_overall_accuracy": 0.9882030178326474,
-      "eval_overall_f1": 0.9503722084367245,
-      "eval_overall_precision": 0.945679012345679,
+      "eval_PERSON_precision": 0.9858156028368794,
+      "eval_PERSON_recall": 0.9788732394366197,
+      "eval_loss": 0.04491155222058296,
+      "eval_overall_accuracy": 0.9879286694101509,
+      "eval_overall_f1": 0.9468479604449938,
+      "eval_overall_precision": 0.9387254901960784,
       "eval_overall_recall": 0.9551122194513716,
-      "eval_runtime": 0.6347,
-      "eval_samples_per_second": 267.825,
-      "eval_steps_per_second": 4.726,
+      "eval_runtime": 0.2916,
+      "eval_samples_per_second": 582.943,
+      "eval_steps_per_second": 10.287,
       "step": 2304
     },
     {
       "epoch": 25.0,
-      "grad_norm": 0.9196493625640869,
+      "grad_norm": 0.5651174783706665,
       "learning_rate": 3.7500000000000003e-05,
-      "loss": 0.0336,
+      "loss": 0.0316,
       "step": 2400
     },
     {
       "epoch": 25.0,
-      "eval_LOCATION_f1": 0.9374999999999999,
+      "eval_LOCATION_f1": 0.9128205128205128,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.9090909090909091,
-      "eval_LOCATION_recall": 0.967741935483871,
-      "eval_ORGANIZATION_f1": 0.9325153374233127,
+      "eval_LOCATION_precision": 0.8725490196078431,
+      "eval_LOCATION_recall": 0.956989247311828,
+      "eval_ORGANIZATION_f1": 0.9085365853658537,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.95,
-      "eval_ORGANIZATION_recall": 0.9156626506024096,
-      "eval_PERSON_f1": 0.9788732394366197,
+      "eval_ORGANIZATION_precision": 0.9197530864197531,
+      "eval_ORGANIZATION_recall": 0.8975903614457831,
+      "eval_PERSON_f1": 0.971830985915493,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9788732394366197,
-      "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.047626204788684845,
-      "eval_overall_accuracy": 0.9890260631001372,
-      "eval_overall_f1": 0.9501246882793017,
-      "eval_overall_precision": 0.9501246882793017,
-      "eval_overall_recall": 0.9501246882793017,
-      "eval_runtime": 0.6357,
-      "eval_samples_per_second": 267.402,
-      "eval_steps_per_second": 4.719,
+      "eval_PERSON_precision": 0.971830985915493,
+      "eval_PERSON_recall": 0.971830985915493,
+      "eval_loss": 0.046948954463005066,
+      "eval_overall_accuracy": 0.9879286694101509,
+      "eval_overall_f1": 0.9318463444857497,
+      "eval_overall_precision": 0.9261083743842364,
+      "eval_overall_recall": 0.9376558603491272,
+      "eval_runtime": 0.2914,
+      "eval_samples_per_second": 583.339,
+      "eval_steps_per_second": 10.294,
       "step": 2400
     },
     {
       "epoch": 26.0,
-      "grad_norm": 0.2330346554517746,
+      "grad_norm": 0.0639706701040268,
       "learning_rate": 3.7e-05,
-      "loss": 0.0313,
+      "loss": 0.0309,
       "step": 2496
     },
     {
       "epoch": 26.0,
-      "eval_LOCATION_f1": 0.946808510638298,
+      "eval_LOCATION_f1": 0.9354838709677419,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.9368421052631579,
-      "eval_LOCATION_recall": 0.956989247311828,
-      "eval_ORGANIZATION_f1": 0.9235294117647058,
+      "eval_LOCATION_precision": 0.9354838709677419,
+      "eval_LOCATION_recall": 0.9354838709677419,
+      "eval_ORGANIZATION_f1": 0.9337349397590361,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9022988505747126,
-      "eval_ORGANIZATION_recall": 0.9457831325301205,
-      "eval_PERSON_f1": 0.9823321554770318,
+      "eval_ORGANIZATION_precision": 0.9337349397590361,
+      "eval_ORGANIZATION_recall": 0.9337349397590361,
+      "eval_PERSON_f1": 0.968421052631579,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9858156028368794,
-      "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.04078715667128563,
-      "eval_overall_accuracy": 0.9893004115226337,
-      "eval_overall_f1": 0.9494451294697903,
-      "eval_overall_precision": 0.9390243902439024,
-      "eval_overall_recall": 0.9600997506234414,
-      "eval_runtime": 0.6352,
-      "eval_samples_per_second": 267.628,
-      "eval_steps_per_second": 4.723,
+      "eval_PERSON_precision": 0.965034965034965,
+      "eval_PERSON_recall": 0.971830985915493,
+      "eval_loss": 0.04571140930056572,
+      "eval_overall_accuracy": 0.9895747599451303,
+      "eval_overall_f1": 0.9464508094645082,
+      "eval_overall_precision": 0.945273631840796,
+      "eval_overall_recall": 0.9476309226932669,
+      "eval_runtime": 0.2914,
+      "eval_samples_per_second": 583.325,
+      "eval_steps_per_second": 10.294,
       "step": 2496
     },
     {
       "epoch": 27.0,
-      "grad_norm": 0.33721673488616943,
+      "grad_norm": 1.063063383102417,
       "learning_rate": 3.65e-05,
-      "loss": 0.0287,
+      "loss": 0.0293,
       "step": 2592
     },
     {
       "epoch": 27.0,
-      "eval_LOCATION_f1": 0.9473684210526316,
+      "eval_LOCATION_f1": 0.9166666666666667,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.9278350515463918,
-      "eval_LOCATION_recall": 0.967741935483871,
-      "eval_ORGANIZATION_f1": 0.9272727272727274,
+      "eval_LOCATION_precision": 0.8888888888888888,
+      "eval_LOCATION_recall": 0.946236559139785,
+      "eval_ORGANIZATION_f1": 0.9285714285714286,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9329268292682927,
-      "eval_ORGANIZATION_recall": 0.9216867469879518,
-      "eval_PERSON_f1": 0.9752650176678446,
+      "eval_ORGANIZATION_precision": 0.9176470588235294,
+      "eval_ORGANIZATION_recall": 0.9397590361445783,
+      "eval_PERSON_f1": 0.9754385964912281,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9787234042553191,
-      "eval_PERSON_recall": 0.971830985915493,
-      "eval_loss": 0.04490111395716667,
-      "eval_overall_accuracy": 0.9898491083676269,
-      "eval_overall_f1": 0.9489414694894147,
-      "eval_overall_precision": 0.9477611940298507,
-      "eval_overall_recall": 0.9501246882793017,
-      "eval_runtime": 0.6359,
-      "eval_samples_per_second": 267.352,
-      "eval_steps_per_second": 4.718,
+      "eval_PERSON_precision": 0.972027972027972,
+      "eval_PERSON_recall": 0.9788732394366197,
+      "eval_loss": 0.04741787165403366,
+      "eval_overall_accuracy": 0.9876543209876543,
+      "eval_overall_f1": 0.942189421894219,
+      "eval_overall_precision": 0.9296116504854369,
+      "eval_overall_recall": 0.9551122194513716,
+      "eval_runtime": 0.2945,
+      "eval_samples_per_second": 577.282,
+      "eval_steps_per_second": 10.187,
       "step": 2592
     },
     {
       "epoch": 28.0,
-      "grad_norm": 0.46604397892951965,
+      "grad_norm": 0.6515504121780396,
       "learning_rate": 3.6e-05,
-      "loss": 0.03,
+      "loss": 0.0299,
       "step": 2688
     },
     {
       "epoch": 28.0,
-      "eval_LOCATION_f1": 0.9263157894736843,
+      "eval_LOCATION_f1": 0.9473684210526316,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.9072164948453608,
-      "eval_LOCATION_recall": 0.946236559139785,
-      "eval_ORGANIZATION_f1": 0.9262536873156342,
+      "eval_LOCATION_precision": 0.9278350515463918,
+      "eval_LOCATION_recall": 0.967741935483871,
+      "eval_ORGANIZATION_f1": 0.920353982300885,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9075144508670521,
-      "eval_ORGANIZATION_recall": 0.9457831325301205,
-      "eval_PERSON_f1": 0.9752650176678446,
+      "eval_ORGANIZATION_precision": 0.9017341040462428,
+      "eval_ORGANIZATION_recall": 0.9397590361445783,
+      "eval_PERSON_f1": 0.971830985915493,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9787234042553191,
+      "eval_PERSON_precision": 0.971830985915493,
       "eval_PERSON_recall": 0.971830985915493,
-      "eval_loss": 0.042864300310611725,
-      "eval_overall_accuracy": 0.9887517146776406,
-      "eval_overall_f1": 0.9433497536945813,
-      "eval_overall_precision": 0.9318734793187348,
-      "eval_overall_recall": 0.9551122194513716,
-      "eval_runtime": 0.6323,
-      "eval_samples_per_second": 268.849,
-      "eval_steps_per_second": 4.744,
+      "eval_loss": 0.050067901611328125,
+      "eval_overall_accuracy": 0.9882030178326474,
+      "eval_overall_f1": 0.944649446494465,
+      "eval_overall_precision": 0.9320388349514563,
+      "eval_overall_recall": 0.9576059850374065,
+      "eval_runtime": 0.2919,
+      "eval_samples_per_second": 582.353,
+      "eval_steps_per_second": 10.277,
       "step": 2688
     },
     {
       "epoch": 29.0,
-      "grad_norm": 0.5060299038887024,
+      "grad_norm": 0.7020676136016846,
       "learning_rate": 3.55e-05,
-      "loss": 0.031,
+      "loss": 0.0264,
       "step": 2784
     },
     {
       "epoch": 29.0,
-      "eval_LOCATION_f1": 0.946236559139785,
+      "eval_LOCATION_f1": 0.9175257731958764,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.946236559139785,
-      "eval_LOCATION_recall": 0.946236559139785,
-      "eval_ORGANIZATION_f1": 0.9321533923303834,
+      "eval_LOCATION_precision": 0.8811881188118812,
+      "eval_LOCATION_recall": 0.956989247311828,
+      "eval_ORGANIZATION_f1": 0.9281437125748503,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9132947976878613,
-      "eval_ORGANIZATION_recall": 0.9518072289156626,
-      "eval_PERSON_f1": 0.9752650176678446,
+      "eval_ORGANIZATION_precision": 0.9226190476190477,
+      "eval_ORGANIZATION_recall": 0.9337349397590361,
+      "eval_PERSON_f1": 0.9788732394366197,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9787234042553191,
-      "eval_PERSON_recall": 0.971830985915493,
-      "eval_loss": 0.04162987321615219,
-      "eval_overall_accuracy": 0.9898491083676269,
-      "eval_overall_f1": 0.9504950495049505,
-      "eval_overall_precision": 0.9434889434889435,
-      "eval_overall_recall": 0.9576059850374065,
-      "eval_runtime": 0.7055,
-      "eval_samples_per_second": 240.973,
-      "eval_steps_per_second": 4.252,
+      "eval_PERSON_precision": 0.9788732394366197,
+      "eval_PERSON_recall": 0.9788732394366197,
+      "eval_loss": 0.0471595823764801,
+      "eval_overall_accuracy": 0.9882030178326474,
+      "eval_overall_f1": 0.9433497536945813,
+      "eval_overall_precision": 0.9318734793187348,
+      "eval_overall_recall": 0.9551122194513716,
+      "eval_runtime": 0.2947,
+      "eval_samples_per_second": 576.863,
+      "eval_steps_per_second": 10.18,
       "step": 2784
     },
     {
       "epoch": 30.0,
-      "grad_norm": 1.1173897981643677,
+      "grad_norm": 0.7510899901390076,
       "learning_rate": 3.5e-05,
-      "loss": 0.027,
+      "loss": 0.0257,
       "step": 2880
     },
     {
       "epoch": 30.0,
-      "eval_LOCATION_f1": 0.9513513513513514,
+      "eval_LOCATION_f1": 0.925531914893617,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.9565217391304348,
-      "eval_LOCATION_recall": 0.946236559139785,
-      "eval_ORGANIZATION_f1": 0.9321533923303834,
+      "eval_LOCATION_precision": 0.9157894736842105,
+      "eval_LOCATION_recall": 0.9354838709677419,
+      "eval_ORGANIZATION_f1": 0.9212827988338192,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9132947976878613,
+      "eval_ORGANIZATION_precision": 0.8926553672316384,
       "eval_ORGANIZATION_recall": 0.9518072289156626,
-      "eval_PERSON_f1": 0.9823321554770318,
+      "eval_PERSON_f1": 0.9788732394366197,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9858156028368794,
+      "eval_PERSON_precision": 0.9788732394366197,
       "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.040551040321588516,
-      "eval_overall_accuracy": 0.9909465020576131,
-      "eval_overall_f1": 0.9541511771995044,
-      "eval_overall_precision": 0.9482758620689655,
-      "eval_overall_recall": 0.9600997506234414,
-      "eval_runtime": 0.6794,
-      "eval_samples_per_second": 250.21,
-      "eval_steps_per_second": 4.415,
+      "eval_loss": 0.05023224279284477,
+      "eval_overall_accuracy": 0.9876543209876543,
+      "eval_overall_f1": 0.9423312883435583,
+      "eval_overall_precision": 0.927536231884058,
+      "eval_overall_recall": 0.9576059850374065,
+      "eval_runtime": 0.2921,
+      "eval_samples_per_second": 581.974,
+      "eval_steps_per_second": 10.27,
       "step": 2880
     },
     {
       "epoch": 31.0,
-      "grad_norm": 0.17513635754585266,
+      "grad_norm": 0.40647125244140625,
       "learning_rate": 3.45e-05,
-      "loss": 0.0257,
+      "loss": 0.0262,
       "step": 2976
     },
     {
       "epoch": 31.0,
-      "eval_LOCATION_f1": 0.9473684210526316,
+      "eval_LOCATION_f1": 0.911917098445596,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.9278350515463918,
-      "eval_LOCATION_recall": 0.967741935483871,
-      "eval_ORGANIZATION_f1": 0.911854103343465,
+      "eval_LOCATION_precision": 0.88,
+      "eval_LOCATION_recall": 0.946236559139785,
+      "eval_ORGANIZATION_f1": 0.9189189189189191,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9202453987730062,
-      "eval_ORGANIZATION_recall": 0.9036144578313253,
-      "eval_PERSON_f1": 0.9823321554770318,
+      "eval_ORGANIZATION_precision": 0.9161676646706587,
+      "eval_ORGANIZATION_recall": 0.9216867469879518,
+      "eval_PERSON_f1": 0.9754385964912281,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9858156028368794,
+      "eval_PERSON_precision": 0.972027972027972,
       "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.043482180684804916,
-      "eval_overall_accuracy": 0.9901234567901235,
-      "eval_overall_f1": 0.9451371571072319,
-      "eval_overall_precision": 0.9451371571072319,
-      "eval_overall_recall": 0.9451371571072319,
-      "eval_runtime": 0.6311,
-      "eval_samples_per_second": 269.372,
-      "eval_steps_per_second": 4.754,
+      "eval_loss": 0.050489392131567,
+      "eval_overall_accuracy": 0.9873799725651577,
+      "eval_overall_f1": 0.9371146732429101,
+      "eval_overall_precision": 0.926829268292683,
+      "eval_overall_recall": 0.9476309226932669,
+      "eval_runtime": 0.2918,
+      "eval_samples_per_second": 582.684,
+      "eval_steps_per_second": 10.283,
       "step": 2976
     },
     {
       "epoch": 32.0,
-      "grad_norm": 0.40757057070732117,
+      "grad_norm": 0.44217225909233093,
       "learning_rate": 3.4000000000000007e-05,
-      "loss": 0.0277,
+      "loss": 0.0263,
       "step": 3072
     },
     {
       "epoch": 32.0,
-      "eval_LOCATION_f1": 0.9424083769633509,
+      "eval_LOCATION_f1": 0.9263157894736843,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.9183673469387755,
-      "eval_LOCATION_recall": 0.967741935483871,
-      "eval_ORGANIZATION_f1": 0.8996960486322187,
+      "eval_LOCATION_precision": 0.9072164948453608,
+      "eval_LOCATION_recall": 0.946236559139785,
+      "eval_ORGANIZATION_f1": 0.9235474006116209,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9079754601226994,
-      "eval_ORGANIZATION_recall": 0.891566265060241,
-      "eval_PERSON_f1": 0.9752650176678446,
+      "eval_ORGANIZATION_precision": 0.937888198757764,
+      "eval_ORGANIZATION_recall": 0.9096385542168675,
+      "eval_PERSON_f1": 0.9788732394366197,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9787234042553191,
-      "eval_PERSON_recall": 0.971830985915493,
-      "eval_loss": 0.04837553948163986,
-      "eval_overall_accuracy": 0.9887517146776406,
-      "eval_overall_f1": 0.9364881693648817,
-      "eval_overall_precision": 0.9353233830845771,
-      "eval_overall_recall": 0.9376558603491272,
-      "eval_runtime": 0.6374,
-      "eval_samples_per_second": 266.721,
-      "eval_steps_per_second": 4.707,
+      "eval_PERSON_precision": 0.9788732394366197,
+      "eval_PERSON_recall": 0.9788732394366197,
+      "eval_loss": 0.05121176689863205,
+      "eval_overall_accuracy": 0.988477366255144,
+      "eval_overall_f1": 0.9438202247191011,
+      "eval_overall_precision": 0.945,
+      "eval_overall_recall": 0.942643391521197,
+      "eval_runtime": 0.2907,
+      "eval_samples_per_second": 584.728,
+      "eval_steps_per_second": 10.319,
       "step": 3072
     },
     {
       "epoch": 33.0,
-      "grad_norm": 0.5286445617675781,
+      "grad_norm": 0.46512940526008606,
       "learning_rate": 3.35e-05,
-      "loss": 0.024,
+      "loss": 0.0246,
       "step": 3168
     },
     {
       "epoch": 33.0,
-      "eval_LOCATION_f1": 0.9270833333333334,
+      "eval_LOCATION_f1": 0.9361702127659575,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.898989898989899,
-      "eval_LOCATION_recall": 0.956989247311828,
-      "eval_ORGANIZATION_f1": 0.896969696969697,
+      "eval_LOCATION_precision": 0.9263157894736842,
+      "eval_LOCATION_recall": 0.946236559139785,
+      "eval_ORGANIZATION_f1": 0.923076923076923,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9024390243902439,
-      "eval_ORGANIZATION_recall": 0.891566265060241,
-      "eval_PERSON_f1": 0.9823321554770318,
+      "eval_ORGANIZATION_precision": 0.9069767441860465,
+      "eval_ORGANIZATION_recall": 0.9397590361445783,
+      "eval_PERSON_f1": 0.9752650176678446,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9858156028368794,
-      "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.04499199241399765,
-      "eval_overall_accuracy": 0.9887517146776406,
-      "eval_overall_f1": 0.9341614906832298,
-      "eval_overall_precision": 0.9306930693069307,
-      "eval_overall_recall": 0.9376558603491272,
-      "eval_runtime": 0.6314,
-      "eval_samples_per_second": 269.238,
-      "eval_steps_per_second": 4.751,
+      "eval_PERSON_precision": 0.9787234042553191,
+      "eval_PERSON_recall": 0.971830985915493,
+      "eval_loss": 0.04934880882501602,
+      "eval_overall_accuracy": 0.9893004115226337,
+      "eval_overall_f1": 0.9443757725587144,
+      "eval_overall_precision": 0.9362745098039216,
+      "eval_overall_recall": 0.9526184538653366,
+      "eval_runtime": 0.2912,
+      "eval_samples_per_second": 583.778,
+      "eval_steps_per_second": 10.302,
       "step": 3168
     },
     {
       "epoch": 34.0,
-      "grad_norm": 0.7948114275932312,
+      "grad_norm": 0.3086019456386566,
       "learning_rate": 3.3e-05,
-      "loss": 0.0227,
+      "loss": 0.0242,
       "step": 3264
     },
     {
       "epoch": 34.0,
-      "eval_LOCATION_f1": 0.9473684210526316,
+      "eval_LOCATION_f1": 0.922279792746114,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.9278350515463918,
-      "eval_LOCATION_recall": 0.967741935483871,
-      "eval_ORGANIZATION_f1": 0.9156626506024096,
+      "eval_LOCATION_precision": 0.89,
+      "eval_LOCATION_recall": 0.956989247311828,
+      "eval_ORGANIZATION_f1": 0.9161676646706587,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9156626506024096,
-      "eval_ORGANIZATION_recall": 0.9156626506024096,
+      "eval_ORGANIZATION_precision": 0.9107142857142857,
+      "eval_ORGANIZATION_recall": 0.9216867469879518,
       "eval_PERSON_f1": 0.9823321554770318,
       "eval_PERSON_number": 142,
       "eval_PERSON_precision": 0.9858156028368794,
       "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.05086701363325119,
-      "eval_overall_accuracy": 0.988477366255144,
-      "eval_overall_f1": 0.946583850931677,
-      "eval_overall_precision": 0.943069306930693,
+      "eval_loss": 0.04890041425824165,
+      "eval_overall_accuracy": 0.9876543209876543,
+      "eval_overall_f1": 0.9407407407407408,
+      "eval_overall_precision": 0.9315403422982885,
       "eval_overall_recall": 0.9501246882793017,
-      "eval_runtime": 0.6371,
-      "eval_samples_per_second": 266.844,
-      "eval_steps_per_second": 4.709,
+      "eval_runtime": 0.2886,
+      "eval_samples_per_second": 589.007,
+      "eval_steps_per_second": 10.394,
       "step": 3264
     },
     {
       "epoch": 35.0,
-      "grad_norm": 0.3472101390361786,
+      "grad_norm": 0.2681832015514374,
       "learning_rate": 3.2500000000000004e-05,
-      "loss": 0.0226,
+      "loss": 0.0217,
       "step": 3360
     },
     {
       "epoch": 35.0,
-      "eval_LOCATION_f1": 0.9374999999999999,
+      "eval_LOCATION_f1": 0.9304812834224598,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.9090909090909091,
-      "eval_LOCATION_recall": 0.967741935483871,
-      "eval_ORGANIZATION_f1": 0.9085545722713864,
+      "eval_LOCATION_precision": 0.925531914893617,
+      "eval_LOCATION_recall": 0.9354838709677419,
+      "eval_ORGANIZATION_f1": 0.911764705882353,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.8901734104046243,
-      "eval_ORGANIZATION_recall": 0.927710843373494,
-      "eval_PERSON_f1": 0.9823321554770318,
+      "eval_ORGANIZATION_precision": 0.8908045977011494,
+      "eval_ORGANIZATION_recall": 0.9337349397590361,
+      "eval_PERSON_f1": 0.9647887323943662,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9858156028368794,
-      "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.052043646574020386,
+      "eval_PERSON_precision": 0.9647887323943662,
+      "eval_PERSON_recall": 0.9647887323943662,
+      "eval_loss": 0.05158211290836334,
       "eval_overall_accuracy": 0.9871056241426612,
-      "eval_overall_f1": 0.941031941031941,
-      "eval_overall_precision": 0.927360774818402,
-      "eval_overall_recall": 0.9551122194513716,
-      "eval_runtime": 0.6348,
-      "eval_samples_per_second": 267.78,
-      "eval_steps_per_second": 4.726,
+      "eval_overall_f1": 0.9346485819975339,
+      "eval_overall_precision": 0.9243902439024391,
+      "eval_overall_recall": 0.9451371571072319,
+      "eval_runtime": 0.2907,
+      "eval_samples_per_second": 584.806,
+      "eval_steps_per_second": 10.32,
       "step": 3360
     },
     {
       "epoch": 36.0,
-      "grad_norm": 0.432446151971817,
+      "grad_norm": 0.3382647633552551,
       "learning_rate": 3.2000000000000005e-05,
-      "loss": 0.0239,
+      "loss": 0.0221,
       "step": 3456
     },
     {
       "epoch": 36.0,
-      "eval_LOCATION_f1": 0.9108910891089109,
+      "eval_LOCATION_f1": 0.922279792746114,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.8440366972477065,
-      "eval_LOCATION_recall": 0.989247311827957,
-      "eval_ORGANIZATION_f1": 0.8909657320872274,
+      "eval_LOCATION_precision": 0.89,
+      "eval_LOCATION_recall": 0.956989247311828,
+      "eval_ORGANIZATION_f1": 0.9240121580547112,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9225806451612903,
-      "eval_ORGANIZATION_recall": 0.8614457831325302,
+      "eval_ORGANIZATION_precision": 0.9325153374233128,
+      "eval_ORGANIZATION_recall": 0.9156626506024096,
       "eval_PERSON_f1": 0.971830985915493,
       "eval_PERSON_number": 142,
       "eval_PERSON_precision": 0.971830985915493,
       "eval_PERSON_recall": 0.971830985915493,
-      "eval_loss": 0.06132661551237106,
-      "eval_overall_accuracy": 0.9862825788751715,
-      "eval_overall_f1": 0.9244114002478315,
-      "eval_overall_precision": 0.9187192118226601,
-      "eval_overall_recall": 0.9301745635910225,
-      "eval_runtime": 0.627,
-      "eval_samples_per_second": 271.131,
-      "eval_steps_per_second": 4.785,
+      "eval_loss": 0.056546807289123535,
+      "eval_overall_accuracy": 0.9876543209876543,
+      "eval_overall_f1": 0.9404466501240694,
+      "eval_overall_precision": 0.9358024691358025,
+      "eval_overall_recall": 0.9451371571072319,
+      "eval_runtime": 0.2901,
+      "eval_samples_per_second": 586.08,
+      "eval_steps_per_second": 10.343,
       "step": 3456
     },
     {
       "epoch": 37.0,
-      "grad_norm": 1.2789398431777954,
+      "grad_norm": 1.3838005065917969,
       "learning_rate": 3.15e-05,
-      "loss": 0.0222,
+      "loss": 0.0219,
       "step": 3552
     },
     {
       "epoch": 37.0,
-      "eval_LOCATION_f1": 0.911917098445596,
+      "eval_LOCATION_f1": 0.9319371727748691,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.88,
-      "eval_LOCATION_recall": 0.946236559139785,
-      "eval_ORGANIZATION_f1": 0.8982035928143712,
+      "eval_LOCATION_precision": 0.9081632653061225,
+      "eval_LOCATION_recall": 0.956989247311828,
+      "eval_ORGANIZATION_f1": 0.9156626506024096,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.8928571428571429,
-      "eval_ORGANIZATION_recall": 0.9036144578313253,
-      "eval_PERSON_f1": 0.9823321554770318,
+      "eval_ORGANIZATION_precision": 0.9156626506024096,
+      "eval_ORGANIZATION_recall": 0.9156626506024096,
+      "eval_PERSON_f1": 0.9752650176678446,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9858156028368794,
-      "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.048861872404813766,
-      "eval_overall_accuracy": 0.9876543209876543,
-      "eval_overall_f1": 0.9308641975308642,
-      "eval_overall_precision": 0.921760391198044,
-      "eval_overall_recall": 0.940149625935162,
-      "eval_runtime": 0.6344,
-      "eval_samples_per_second": 267.981,
-      "eval_steps_per_second": 4.729,
+      "eval_PERSON_precision": 0.9787234042553191,
+      "eval_PERSON_recall": 0.971830985915493,
+      "eval_loss": 0.04845535010099411,
+      "eval_overall_accuracy": 0.9893004115226337,
+      "eval_overall_f1": 0.9404466501240694,
+      "eval_overall_precision": 0.9358024691358025,
+      "eval_overall_recall": 0.9451371571072319,
+      "eval_runtime": 0.2893,
+      "eval_samples_per_second": 587.579,
+      "eval_steps_per_second": 10.369,
       "step": 3552
     },
     {
       "epoch": 38.0,
-      "grad_norm": 0.11196000874042511,
+      "grad_norm": 0.24594035744667053,
       "learning_rate": 3.1e-05,
-      "loss": 0.0221,
+      "loss": 0.0217,
       "step": 3648
     },
     {
       "epoch": 38.0,
-      "eval_LOCATION_f1": 0.9292929292929293,
+      "eval_LOCATION_f1": 0.9424083769633509,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.8761904761904762,
-      "eval_LOCATION_recall": 0.989247311827957,
-      "eval_ORGANIZATION_f1": 0.9057750759878419,
+      "eval_LOCATION_precision": 0.9183673469387755,
+      "eval_LOCATION_recall": 0.967741935483871,
+      "eval_ORGANIZATION_f1": 0.918429003021148,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9141104294478528,
-      "eval_ORGANIZATION_recall": 0.8975903614457831,
-      "eval_PERSON_f1": 0.9823321554770318,
+      "eval_ORGANIZATION_precision": 0.9212121212121213,
+      "eval_ORGANIZATION_recall": 0.9156626506024096,
+      "eval_PERSON_f1": 0.9681978798586572,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9858156028368794,
-      "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.052511900663375854,
-      "eval_overall_accuracy": 0.9868312757201646,
-      "eval_overall_f1": 0.9382716049382714,
-      "eval_overall_precision": 0.9290953545232273,
-      "eval_overall_recall": 0.9476309226932669,
-      "eval_runtime": 0.6225,
-      "eval_samples_per_second": 273.092,
-      "eval_steps_per_second": 4.819,
+      "eval_PERSON_precision": 0.9716312056737588,
+      "eval_PERSON_recall": 0.9647887323943662,
+      "eval_loss": 0.052869848906993866,
+      "eval_overall_accuracy": 0.9882030178326474,
+      "eval_overall_f1": 0.9416149068322982,
+      "eval_overall_precision": 0.9381188118811881,
+      "eval_overall_recall": 0.9451371571072319,
+      "eval_runtime": 0.2898,
+      "eval_samples_per_second": 586.556,
+      "eval_steps_per_second": 10.351,
       "step": 3648
     },
     {
       "epoch": 39.0,
-      "grad_norm": 0.2599373757839203,
+      "grad_norm": 0.40099620819091797,
       "learning_rate": 3.05e-05,
-      "loss": 0.0225,
+      "loss": 0.0236,
       "step": 3744
     },
     {
       "epoch": 39.0,
-      "eval_LOCATION_f1": 0.923076923076923,
+      "eval_LOCATION_f1": 0.9304812834224598,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.8823529411764706,
-      "eval_LOCATION_recall": 0.967741935483871,
-      "eval_ORGANIZATION_f1": 0.8975903614457831,
+      "eval_LOCATION_precision": 0.925531914893617,
+      "eval_LOCATION_recall": 0.9354838709677419,
+      "eval_ORGANIZATION_f1": 0.9107142857142857,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.8975903614457831,
-      "eval_ORGANIZATION_recall": 0.8975903614457831,
-      "eval_PERSON_f1": 0.9823321554770318,
+      "eval_ORGANIZATION_precision": 0.9,
+      "eval_ORGANIZATION_recall": 0.9216867469879518,
+      "eval_PERSON_f1": 0.9681978798586572,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9858156028368794,
-      "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.050379928201436996,
-      "eval_overall_accuracy": 0.9860082304526749,
-      "eval_overall_f1": 0.9333333333333333,
-      "eval_overall_precision": 0.9242053789731052,
-      "eval_overall_recall": 0.942643391521197,
-      "eval_runtime": 0.6319,
-      "eval_samples_per_second": 269.039,
-      "eval_steps_per_second": 4.748,
+      "eval_PERSON_precision": 0.9716312056737588,
+      "eval_PERSON_recall": 0.9647887323943662,
+      "eval_loss": 0.05043813958764076,
+      "eval_overall_accuracy": 0.9873799725651577,
+      "eval_overall_f1": 0.9354838709677419,
+      "eval_overall_precision": 0.9308641975308642,
+      "eval_overall_recall": 0.940149625935162,
+      "eval_runtime": 0.288,
+      "eval_samples_per_second": 590.367,
+      "eval_steps_per_second": 10.418,
       "step": 3744
     },
     {
       "epoch": 40.0,
-      "grad_norm": 0.2717348039150238,
+      "grad_norm": 0.19064545631408691,
       "learning_rate": 3e-05,
-      "loss": 0.0194,
+      "loss": 0.0205,
       "step": 3840
     },
     {
       "epoch": 40.0,
-      "eval_LOCATION_f1": 0.9270833333333334,
+      "eval_LOCATION_f1": 0.90625,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.898989898989899,
-      "eval_LOCATION_recall": 0.956989247311828,
-      "eval_ORGANIZATION_f1": 0.918429003021148,
+      "eval_LOCATION_precision": 0.8787878787878788,
+      "eval_LOCATION_recall": 0.9354838709677419,
+      "eval_ORGANIZATION_f1": 0.9134328358208955,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9212121212121213,
-      "eval_ORGANIZATION_recall": 0.9156626506024096,
-      "eval_PERSON_f1": 0.9752650176678446,
+      "eval_ORGANIZATION_precision": 0.9053254437869822,
+      "eval_ORGANIZATION_recall": 0.9216867469879518,
+      "eval_PERSON_f1": 0.9614035087719298,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9787234042553191,
-      "eval_PERSON_recall": 0.971830985915493,
-      "eval_loss": 0.04696699604392052,
-      "eval_overall_accuracy": 0.9890260631001372,
-      "eval_overall_f1": 0.9404466501240694,
-      "eval_overall_precision": 0.9358024691358025,
-      "eval_overall_recall": 0.9451371571072319,
-      "eval_runtime": 0.6215,
-      "eval_samples_per_second": 273.521,
-      "eval_steps_per_second": 4.827,
+      "eval_PERSON_precision": 0.958041958041958,
+      "eval_PERSON_recall": 0.9647887323943662,
+      "eval_loss": 0.053412336856126785,
+      "eval_overall_accuracy": 0.9876543209876543,
+      "eval_overall_f1": 0.9285714285714285,
+      "eval_overall_precision": 0.9172749391727494,
+      "eval_overall_recall": 0.940149625935162,
+      "eval_runtime": 0.2891,
+      "eval_samples_per_second": 588.081,
+      "eval_steps_per_second": 10.378,
       "step": 3840
     },
     {
       "epoch": 41.0,
-      "grad_norm": 0.43004825711250305,
+      "grad_norm": 0.02482328750193119,
       "learning_rate": 2.95e-05,
-      "loss": 0.0206,
+      "loss": 0.0193,
       "step": 3936
     },
     {
       "epoch": 41.0,
-      "eval_LOCATION_f1": 0.9157894736842105,
+      "eval_LOCATION_f1": 0.9072164948453608,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.8969072164948454,
-      "eval_LOCATION_recall": 0.9354838709677419,
-      "eval_ORGANIZATION_f1": 0.9,
+      "eval_LOCATION_precision": 0.8712871287128713,
+      "eval_LOCATION_recall": 0.946236559139785,
+      "eval_ORGANIZATION_f1": 0.890207715133531,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.8793103448275862,
-      "eval_ORGANIZATION_recall": 0.9216867469879518,
-      "eval_PERSON_f1": 0.9823321554770318,
+      "eval_ORGANIZATION_precision": 0.8771929824561403,
+      "eval_ORGANIZATION_recall": 0.9036144578313253,
+      "eval_PERSON_f1": 0.9787234042553192,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9858156028368794,
-      "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.048884760588407516,
-      "eval_overall_accuracy": 0.9865569272976681,
-      "eval_overall_f1": 0.932349323493235,
-      "eval_overall_precision": 0.9199029126213593,
-      "eval_overall_recall": 0.9451371571072319,
-      "eval_runtime": 0.6384,
-      "eval_samples_per_second": 266.27,
-      "eval_steps_per_second": 4.699,
+      "eval_PERSON_precision": 0.9857142857142858,
+      "eval_PERSON_recall": 0.971830985915493,
+      "eval_loss": 0.0644853487610817,
+      "eval_overall_accuracy": 0.9860082304526749,
+      "eval_overall_f1": 0.9249692496924969,
+      "eval_overall_precision": 0.912621359223301,
+      "eval_overall_recall": 0.9376558603491272,
+      "eval_runtime": 0.2901,
+      "eval_samples_per_second": 586.09,
+      "eval_steps_per_second": 10.343,
       "step": 3936
     },
     {
       "epoch": 42.0,
-      "grad_norm": 0.7156493663787842,
+      "grad_norm": 0.5300124287605286,
       "learning_rate": 2.9e-05,
-      "loss": 0.0208,
+      "loss": 0.0198,
       "step": 4032
     },
     {
       "epoch": 42.0,
-      "eval_LOCATION_f1": 0.9326424870466321,
+      "eval_LOCATION_f1": 0.922279792746114,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.9,
-      "eval_LOCATION_recall": 0.967741935483871,
-      "eval_ORGANIZATION_f1": 0.9101796407185628,
+      "eval_LOCATION_precision": 0.89,
+      "eval_LOCATION_recall": 0.956989247311828,
+      "eval_ORGANIZATION_f1": 0.8948948948948949,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9047619047619048,
-      "eval_ORGANIZATION_recall": 0.9156626506024096,
-      "eval_PERSON_f1": 0.9752650176678446,
+      "eval_ORGANIZATION_precision": 0.8922155688622755,
+      "eval_ORGANIZATION_recall": 0.8975903614457831,
+      "eval_PERSON_f1": 0.9823321554770318,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9787234042553191,
-      "eval_PERSON_recall": 0.971830985915493,
-      "eval_loss": 0.05095763877034187,
-      "eval_overall_accuracy": 0.9879286694101509,
-      "eval_overall_f1": 0.9382716049382714,
-      "eval_overall_precision": 0.9290953545232273,
-      "eval_overall_recall": 0.9476309226932669,
-      "eval_runtime": 0.6338,
-      "eval_samples_per_second": 268.241,
-      "eval_steps_per_second": 4.734,
+      "eval_PERSON_precision": 0.9858156028368794,
+      "eval_PERSON_recall": 0.9788732394366197,
+      "eval_loss": 0.0561443567276001,
+      "eval_overall_accuracy": 0.9860082304526749,
+      "eval_overall_f1": 0.9320148331273177,
+      "eval_overall_precision": 0.9240196078431373,
+      "eval_overall_recall": 0.940149625935162,
+      "eval_runtime": 0.2902,
+      "eval_samples_per_second": 585.717,
+      "eval_steps_per_second": 10.336,
       "step": 4032
     },
     {
       "epoch": 43.0,
-      "grad_norm": 0.23692376911640167,
+      "grad_norm": 0.8966781497001648,
       "learning_rate": 2.8499999999999998e-05,
-      "loss": 0.0178,
+      "loss": 0.018,
       "step": 4128
     },
     {
       "epoch": 43.0,
-      "eval_LOCATION_f1": 0.9278350515463919,
+      "eval_LOCATION_f1": 0.9368421052631579,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.8910891089108911,
-      "eval_LOCATION_recall": 0.967741935483871,
-      "eval_ORGANIZATION_f1": 0.9382716049382716,
+      "eval_LOCATION_precision": 0.9175257731958762,
+      "eval_LOCATION_recall": 0.956989247311828,
+      "eval_ORGANIZATION_f1": 0.9123867069486404,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9620253164556962,
-      "eval_ORGANIZATION_recall": 0.9156626506024096,
+      "eval_ORGANIZATION_precision": 0.9151515151515152,
+      "eval_ORGANIZATION_recall": 0.9096385542168675,
       "eval_PERSON_f1": 0.9752650176678446,
       "eval_PERSON_number": 142,
       "eval_PERSON_precision": 0.9787234042553191,
       "eval_PERSON_recall": 0.971830985915493,
-      "eval_loss": 0.04964771866798401,
-      "eval_overall_accuracy": 0.9895747599451303,
-      "eval_overall_f1": 0.9488139825218477,
-      "eval_overall_precision": 0.95,
-      "eval_overall_recall": 0.9476309226932669,
-      "eval_runtime": 0.6373,
-      "eval_samples_per_second": 266.748,
-      "eval_steps_per_second": 4.707,
+      "eval_loss": 0.05398855730891228,
+      "eval_overall_accuracy": 0.9882030178326474,
+      "eval_overall_f1": 0.9402985074626866,
+      "eval_overall_precision": 0.9379652605459057,
+      "eval_overall_recall": 0.942643391521197,
+      "eval_runtime": 0.2865,
+      "eval_samples_per_second": 593.421,
+      "eval_steps_per_second": 10.472,
       "step": 4128
     },
     {
       "epoch": 44.0,
-      "grad_norm": 0.20943285524845123,
+      "grad_norm": 0.3985178768634796,
       "learning_rate": 2.8000000000000003e-05,
-      "loss": 0.0184,
+      "loss": 0.0188,
       "step": 4224
     },
     {
       "epoch": 44.0,
-      "eval_LOCATION_f1": 0.9381443298969072,
+      "eval_LOCATION_f1": 0.9109947643979057,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.900990099009901,
-      "eval_LOCATION_recall": 0.978494623655914,
-      "eval_ORGANIZATION_f1": 0.9101796407185628,
+      "eval_LOCATION_precision": 0.8877551020408163,
+      "eval_LOCATION_recall": 0.9354838709677419,
+      "eval_ORGANIZATION_f1": 0.8975903614457831,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9047619047619048,
-      "eval_ORGANIZATION_recall": 0.9156626506024096,
-      "eval_PERSON_f1": 0.9752650176678446,
+      "eval_ORGANIZATION_precision": 0.8975903614457831,
+      "eval_ORGANIZATION_recall": 0.8975903614457831,
+      "eval_PERSON_f1": 0.9788732394366197,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9787234042553191,
-      "eval_PERSON_recall": 0.971830985915493,
-      "eval_loss": 0.05329016596078873,
-      "eval_overall_accuracy": 0.9882030178326474,
-      "eval_overall_f1": 0.9395807644882862,
-      "eval_overall_precision": 0.9292682926829269,
-      "eval_overall_recall": 0.9501246882793017,
-      "eval_runtime": 0.6294,
-      "eval_samples_per_second": 270.096,
-      "eval_steps_per_second": 4.766,
+      "eval_PERSON_precision": 0.9788732394366197,
+      "eval_PERSON_recall": 0.9788732394366197,
+      "eval_loss": 0.05714266002178192,
+      "eval_overall_accuracy": 0.9868312757201646,
+      "eval_overall_f1": 0.9293680297397771,
+      "eval_overall_precision": 0.9236453201970444,
+      "eval_overall_recall": 0.9351620947630923,
+      "eval_runtime": 0.2944,
+      "eval_samples_per_second": 577.415,
+      "eval_steps_per_second": 10.19,
       "step": 4224
     },
     {
       "epoch": 45.0,
-      "grad_norm": 0.8311372399330139,
+      "grad_norm": 0.04688119515776634,
       "learning_rate": 2.7500000000000004e-05,
-      "loss": 0.0172,
+      "loss": 0.0177,
       "step": 4320
     },
     {
       "epoch": 45.0,
-      "eval_LOCATION_f1": 0.9479166666666667,
+      "eval_LOCATION_f1": 0.9263157894736843,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.9191919191919192,
-      "eval_LOCATION_recall": 0.978494623655914,
-      "eval_ORGANIZATION_f1": 0.9329268292682927,
+      "eval_LOCATION_precision": 0.9072164948453608,
+      "eval_LOCATION_recall": 0.946236559139785,
+      "eval_ORGANIZATION_f1": 0.9030303030303031,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9444444444444444,
-      "eval_ORGANIZATION_recall": 0.9216867469879518,
-      "eval_PERSON_f1": 0.9752650176678446,
+      "eval_ORGANIZATION_precision": 0.9085365853658537,
+      "eval_ORGANIZATION_recall": 0.8975903614457831,
+      "eval_PERSON_f1": 0.9788732394366197,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9787234042553191,
-      "eval_PERSON_recall": 0.971830985915493,
-      "eval_loss": 0.05253069847822189,
-      "eval_overall_accuracy": 0.9882030178326474,
-      "eval_overall_f1": 0.9514321295143213,
-      "eval_overall_precision": 0.9502487562189055,
-      "eval_overall_recall": 0.9526184538653366,
-      "eval_runtime": 0.6282,
-      "eval_samples_per_second": 270.603,
-      "eval_steps_per_second": 4.775,
+      "eval_PERSON_precision": 0.9788732394366197,
+      "eval_PERSON_recall": 0.9788732394366197,
+      "eval_loss": 0.05286408215761185,
+      "eval_overall_accuracy": 0.9862825788751715,
+      "eval_overall_f1": 0.9353233830845771,
+      "eval_overall_precision": 0.9330024813895782,
+      "eval_overall_recall": 0.9376558603491272,
+      "eval_runtime": 0.2902,
+      "eval_samples_per_second": 585.832,
+      "eval_steps_per_second": 10.338,
       "step": 4320
     },
     {
       "epoch": 46.0,
-      "grad_norm": 0.6656791567802429,
+      "grad_norm": 0.19999141991138458,
       "learning_rate": 2.7000000000000002e-05,
-      "loss": 0.0159,
+      "loss": 0.0178,
       "step": 4416
     },
     {
       "epoch": 46.0,
-      "eval_LOCATION_f1": 0.9533678756476685,
+      "eval_LOCATION_f1": 0.914572864321608,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.92,
-      "eval_LOCATION_recall": 0.989247311827957,
-      "eval_ORGANIZATION_f1": 0.9151515151515153,
+      "eval_LOCATION_precision": 0.8584905660377359,
+      "eval_LOCATION_recall": 0.978494623655914,
+      "eval_ORGANIZATION_f1": 0.8895705521472392,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9207317073170732,
-      "eval_ORGANIZATION_recall": 0.9096385542168675,
+      "eval_ORGANIZATION_precision": 0.90625,
+      "eval_ORGANIZATION_recall": 0.8734939759036144,
       "eval_PERSON_f1": 0.9823321554770318,
       "eval_PERSON_number": 142,
       "eval_PERSON_precision": 0.9858156028368794,
       "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.052427615970373154,
-      "eval_overall_accuracy": 0.988477366255144,
-      "eval_overall_f1": 0.9478908188585606,
-      "eval_overall_precision": 0.9432098765432099,
-      "eval_overall_recall": 0.9526184538653366,
-      "eval_runtime": 0.6393,
-      "eval_samples_per_second": 265.902,
-      "eval_steps_per_second": 4.692,
+      "eval_loss": 0.053873803466558456,
+      "eval_overall_accuracy": 0.9873799725651577,
+      "eval_overall_f1": 0.9282178217821783,
+      "eval_overall_precision": 0.9213759213759214,
+      "eval_overall_recall": 0.9351620947630923,
+      "eval_runtime": 0.2867,
+      "eval_samples_per_second": 593.025,
+      "eval_steps_per_second": 10.465,
       "step": 4416
     },
     {
       "epoch": 47.0,
-      "grad_norm": 0.03922390192747116,
+      "grad_norm": 0.04654427990317345,
       "learning_rate": 2.6500000000000004e-05,
-      "loss": 0.0174,
+      "loss": 0.0168,
       "step": 4512
     },
     {
       "epoch": 47.0,
-      "eval_LOCATION_f1": 0.9417989417989417,
+      "eval_LOCATION_f1": 0.9319371727748691,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.9270833333333334,
+      "eval_LOCATION_precision": 0.9081632653061225,
       "eval_LOCATION_recall": 0.956989247311828,
-      "eval_ORGANIZATION_f1": 0.9014925373134328,
+      "eval_ORGANIZATION_f1": 0.9166666666666667,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.893491124260355,
-      "eval_ORGANIZATION_recall": 0.9096385542168675,
-      "eval_PERSON_f1": 0.9752650176678446,
+      "eval_ORGANIZATION_precision": 0.9058823529411765,
+      "eval_ORGANIZATION_recall": 0.927710843373494,
+      "eval_PERSON_f1": 0.9823321554770318,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9787234042553191,
-      "eval_PERSON_recall": 0.971830985915493,
-      "eval_loss": 0.051573678851127625,
-      "eval_overall_accuracy": 0.9879286694101509,
-      "eval_overall_f1": 0.9368029739776952,
-      "eval_overall_precision": 0.9310344827586207,
-      "eval_overall_recall": 0.942643391521197,
-      "eval_runtime": 0.6335,
-      "eval_samples_per_second": 268.33,
-      "eval_steps_per_second": 4.735,
+      "eval_PERSON_precision": 0.9858156028368794,
+      "eval_PERSON_recall": 0.9788732394366197,
+      "eval_loss": 0.05449910834431648,
+      "eval_overall_accuracy": 0.9873799725651577,
+      "eval_overall_f1": 0.9432098765432098,
+      "eval_overall_precision": 0.9339853300733496,
+      "eval_overall_recall": 0.9526184538653366,
+      "eval_runtime": 0.288,
+      "eval_samples_per_second": 590.188,
+      "eval_steps_per_second": 10.415,
       "step": 4512
     },
     {
       "epoch": 48.0,
-      "grad_norm": 0.08619959652423859,
+      "grad_norm": 0.35884276032447815,
       "learning_rate": 2.6000000000000002e-05,
-      "loss": 0.0176,
+      "loss": 0.016,
       "step": 4608
     },
     {
       "epoch": 48.0,
-      "eval_LOCATION_f1": 0.9417989417989417,
+      "eval_LOCATION_f1": 0.9270833333333334,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.9270833333333334,
+      "eval_LOCATION_precision": 0.898989898989899,
       "eval_LOCATION_recall": 0.956989247311828,
-      "eval_ORGANIZATION_f1": 0.924924924924925,
+      "eval_ORGANIZATION_f1": 0.9096385542168675,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9221556886227545,
-      "eval_ORGANIZATION_recall": 0.927710843373494,
-      "eval_PERSON_f1": 0.9752650176678446,
+      "eval_ORGANIZATION_precision": 0.9096385542168675,
+      "eval_ORGANIZATION_recall": 0.9096385542168675,
+      "eval_PERSON_f1": 0.9823321554770318,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9787234042553191,
-      "eval_PERSON_recall": 0.971830985915493,
-      "eval_loss": 0.04628601670265198,
-      "eval_overall_accuracy": 0.99039780521262,
-      "eval_overall_f1": 0.946583850931677,
-      "eval_overall_precision": 0.943069306930693,
-      "eval_overall_recall": 0.9501246882793017,
-      "eval_runtime": 0.6378,
-      "eval_samples_per_second": 266.532,
-      "eval_steps_per_second": 4.704,
+      "eval_PERSON_precision": 0.9858156028368794,
+      "eval_PERSON_recall": 0.9788732394366197,
+      "eval_loss": 0.05373828485608101,
+      "eval_overall_accuracy": 0.988477366255144,
+      "eval_overall_f1": 0.9392812887236679,
+      "eval_overall_precision": 0.9334975369458128,
+      "eval_overall_recall": 0.9451371571072319,
+      "eval_runtime": 0.2902,
+      "eval_samples_per_second": 585.775,
+      "eval_steps_per_second": 10.337,
       "step": 4608
     },
     {
       "epoch": 49.0,
-      "grad_norm": 0.017897285521030426,
+      "grad_norm": 0.010812529362738132,
       "learning_rate": 2.5500000000000003e-05,
-      "loss": 0.0156,
+      "loss": 0.0167,
       "step": 4704
     },
     {
@@ -1509,587 +1509,587 @@
       "eval_LOCATION_number": 93,
       "eval_LOCATION_precision": 0.9081632653061225,
       "eval_LOCATION_recall": 0.956989247311828,
-      "eval_ORGANIZATION_f1": 0.9207317073170733,
+      "eval_ORGANIZATION_f1": 0.927710843373494,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9320987654320988,
-      "eval_ORGANIZATION_recall": 0.9096385542168675,
-      "eval_PERSON_f1": 0.9752650176678446,
+      "eval_ORGANIZATION_precision": 0.927710843373494,
+      "eval_ORGANIZATION_recall": 0.927710843373494,
+      "eval_PERSON_f1": 0.9823321554770318,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9787234042553191,
-      "eval_PERSON_recall": 0.971830985915493,
-      "eval_loss": 0.05306881293654442,
+      "eval_PERSON_precision": 0.9858156028368794,
+      "eval_PERSON_recall": 0.9788732394366197,
+      "eval_loss": 0.05074891448020935,
       "eval_overall_accuracy": 0.9893004115226337,
-      "eval_overall_f1": 0.942643391521197,
-      "eval_overall_precision": 0.942643391521197,
-      "eval_overall_recall": 0.942643391521197,
-      "eval_runtime": 0.6253,
-      "eval_samples_per_second": 271.873,
-      "eval_steps_per_second": 4.798,
+      "eval_overall_f1": 0.9478908188585606,
+      "eval_overall_precision": 0.9432098765432099,
+      "eval_overall_recall": 0.9526184538653366,
+      "eval_runtime": 0.2898,
+      "eval_samples_per_second": 586.57,
+      "eval_steps_per_second": 10.351,
       "step": 4704
     },
     {
       "epoch": 50.0,
-      "grad_norm": 0.15169987082481384,
+      "grad_norm": 0.36007794737815857,
       "learning_rate": 2.5e-05,
-      "loss": 0.017,
+      "loss": 0.0157,
       "step": 4800
     },
     {
       "epoch": 50.0,
-      "eval_LOCATION_f1": 0.9326424870466321,
+      "eval_LOCATION_f1": 0.9109947643979057,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.9,
-      "eval_LOCATION_recall": 0.967741935483871,
-      "eval_ORGANIZATION_f1": 0.918429003021148,
+      "eval_LOCATION_precision": 0.8877551020408163,
+      "eval_LOCATION_recall": 0.9354838709677419,
+      "eval_ORGANIZATION_f1": 0.9080118694362017,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9212121212121213,
-      "eval_ORGANIZATION_recall": 0.9156626506024096,
+      "eval_ORGANIZATION_precision": 0.8947368421052632,
+      "eval_ORGANIZATION_recall": 0.9216867469879518,
       "eval_PERSON_f1": 0.9823321554770318,
       "eval_PERSON_number": 142,
       "eval_PERSON_precision": 0.9858156028368794,
       "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.05259489268064499,
-      "eval_overall_accuracy": 0.9879286694101509,
-      "eval_overall_f1": 0.9442379182156134,
-      "eval_overall_precision": 0.9384236453201971,
-      "eval_overall_recall": 0.9501246882793017,
-      "eval_runtime": 0.6281,
-      "eval_samples_per_second": 270.654,
-      "eval_steps_per_second": 4.776,
+      "eval_loss": 0.05863998457789421,
+      "eval_overall_accuracy": 0.9868312757201646,
+      "eval_overall_f1": 0.9346485819975339,
+      "eval_overall_precision": 0.9243902439024391,
+      "eval_overall_recall": 0.9451371571072319,
+      "eval_runtime": 0.2884,
+      "eval_samples_per_second": 589.476,
+      "eval_steps_per_second": 10.403,
       "step": 4800
     },
     {
       "epoch": 51.0,
-      "grad_norm": 0.4998176097869873,
+      "grad_norm": 0.11987640708684921,
       "learning_rate": 2.45e-05,
-      "loss": 0.0156,
+      "loss": 0.015,
       "step": 4896
     },
     {
       "epoch": 51.0,
-      "eval_LOCATION_f1": 0.923076923076923,
+      "eval_LOCATION_f1": 0.90625,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.8823529411764706,
-      "eval_LOCATION_recall": 0.967741935483871,
-      "eval_ORGANIZATION_f1": 0.920245398773006,
+      "eval_LOCATION_precision": 0.8787878787878788,
+      "eval_LOCATION_recall": 0.9354838709677419,
+      "eval_ORGANIZATION_f1": 0.9189189189189191,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9375,
-      "eval_ORGANIZATION_recall": 0.9036144578313253,
-      "eval_PERSON_f1": 0.9823321554770318,
+      "eval_ORGANIZATION_precision": 0.9161676646706587,
+      "eval_ORGANIZATION_recall": 0.9216867469879518,
+      "eval_PERSON_f1": 0.971830985915493,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9858156028368794,
-      "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.05702535808086395,
-      "eval_overall_accuracy": 0.9871056241426612,
-      "eval_overall_f1": 0.9427860696517413,
-      "eval_overall_precision": 0.9404466501240695,
-      "eval_overall_recall": 0.9451371571072319,
-      "eval_runtime": 0.6253,
-      "eval_samples_per_second": 271.881,
-      "eval_steps_per_second": 4.798,
+      "eval_PERSON_precision": 0.971830985915493,
+      "eval_PERSON_recall": 0.971830985915493,
+      "eval_loss": 0.052376989275217056,
+      "eval_overall_accuracy": 0.9873799725651577,
+      "eval_overall_f1": 0.9344870210135972,
+      "eval_overall_precision": 0.9264705882352942,
+      "eval_overall_recall": 0.942643391521197,
+      "eval_runtime": 0.2909,
+      "eval_samples_per_second": 584.325,
+      "eval_steps_per_second": 10.312,
       "step": 4896
     },
     {
       "epoch": 52.0,
-      "grad_norm": 0.4189845025539398,
+      "grad_norm": 0.8939554691314697,
       "learning_rate": 2.4e-05,
-      "loss": 0.0152,
+      "loss": 0.0155,
       "step": 4992
     },
     {
       "epoch": 52.0,
-      "eval_LOCATION_f1": 0.9263157894736843,
+      "eval_LOCATION_f1": 0.9025641025641027,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.9072164948453608,
+      "eval_LOCATION_precision": 0.8627450980392157,
       "eval_LOCATION_recall": 0.946236559139785,
-      "eval_ORGANIZATION_f1": 0.9272727272727274,
+      "eval_ORGANIZATION_f1": 0.9041916167664671,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9329268292682927,
-      "eval_ORGANIZATION_recall": 0.9216867469879518,
-      "eval_PERSON_f1": 0.9752650176678446,
+      "eval_ORGANIZATION_precision": 0.8988095238095238,
+      "eval_ORGANIZATION_recall": 0.9096385542168675,
+      "eval_PERSON_f1": 0.9787234042553192,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9787234042553191,
+      "eval_PERSON_precision": 0.9857142857142858,
       "eval_PERSON_recall": 0.971830985915493,
-      "eval_loss": 0.05183970928192139,
-      "eval_overall_accuracy": 0.9898491083676269,
-      "eval_overall_f1": 0.9439601494396015,
-      "eval_overall_precision": 0.9427860696517413,
-      "eval_overall_recall": 0.9451371571072319,
-      "eval_runtime": 0.6336,
-      "eval_samples_per_second": 268.303,
-      "eval_steps_per_second": 4.735,
+      "eval_loss": 0.06287980824708939,
+      "eval_overall_accuracy": 0.9862825788751715,
+      "eval_overall_f1": 0.9297163995067818,
+      "eval_overall_precision": 0.9195121951219513,
+      "eval_overall_recall": 0.940149625935162,
+      "eval_runtime": 0.2901,
+      "eval_samples_per_second": 585.943,
+      "eval_steps_per_second": 10.34,
       "step": 4992
     },
     {
       "epoch": 53.0,
-      "grad_norm": 1.3798094987869263,
+      "grad_norm": 0.20487843453884125,
       "learning_rate": 2.35e-05,
-      "loss": 0.017,
+      "loss": 0.0156,
       "step": 5088
     },
     {
       "epoch": 53.0,
-      "eval_LOCATION_f1": 0.9263157894736843,
+      "eval_LOCATION_f1": 0.9015544041450777,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.9072164948453608,
-      "eval_LOCATION_recall": 0.946236559139785,
-      "eval_ORGANIZATION_f1": 0.9309309309309309,
+      "eval_LOCATION_precision": 0.87,
+      "eval_LOCATION_recall": 0.9354838709677419,
+      "eval_ORGANIZATION_f1": 0.9216867469879518,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9281437125748503,
-      "eval_ORGANIZATION_recall": 0.9337349397590361,
-      "eval_PERSON_f1": 0.9823321554770318,
+      "eval_ORGANIZATION_precision": 0.9216867469879518,
+      "eval_ORGANIZATION_recall": 0.9216867469879518,
+      "eval_PERSON_f1": 0.9752650176678446,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9858156028368794,
-      "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.048745788633823395,
-      "eval_overall_accuracy": 0.9895747599451303,
-      "eval_overall_f1": 0.9478908188585606,
-      "eval_overall_precision": 0.9432098765432099,
-      "eval_overall_recall": 0.9526184538653366,
-      "eval_runtime": 0.632,
-      "eval_samples_per_second": 268.983,
-      "eval_steps_per_second": 4.747,
+      "eval_PERSON_precision": 0.9787234042553191,
+      "eval_PERSON_recall": 0.971830985915493,
+      "eval_loss": 0.06014350429177284,
+      "eval_overall_accuracy": 0.9868312757201646,
+      "eval_overall_f1": 0.9356435643564357,
+      "eval_overall_precision": 0.9287469287469288,
+      "eval_overall_recall": 0.942643391521197,
+      "eval_runtime": 0.2903,
+      "eval_samples_per_second": 585.535,
+      "eval_steps_per_second": 10.333,
       "step": 5088
     },
     {
       "epoch": 54.0,
-      "grad_norm": 0.7631996870040894,
+      "grad_norm": 0.055074214935302734,
       "learning_rate": 2.3000000000000003e-05,
-      "loss": 0.0143,
+      "loss": 0.0141,
       "step": 5184
     },
     {
       "epoch": 54.0,
-      "eval_LOCATION_f1": 0.911917098445596,
+      "eval_LOCATION_f1": 0.9081632653061223,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.88,
-      "eval_LOCATION_recall": 0.946236559139785,
-      "eval_ORGANIZATION_f1": 0.9212121212121213,
+      "eval_LOCATION_precision": 0.8640776699029126,
+      "eval_LOCATION_recall": 0.956989247311828,
+      "eval_ORGANIZATION_f1": 0.9085545722713864,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.926829268292683,
-      "eval_ORGANIZATION_recall": 0.9156626506024096,
-      "eval_PERSON_f1": 0.9752650176678446,
+      "eval_ORGANIZATION_precision": 0.8901734104046243,
+      "eval_ORGANIZATION_recall": 0.927710843373494,
+      "eval_PERSON_f1": 0.9787234042553192,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9787234042553191,
+      "eval_PERSON_precision": 0.9857142857142858,
       "eval_PERSON_recall": 0.971830985915493,
-      "eval_loss": 0.054195113480091095,
-      "eval_overall_accuracy": 0.9876543209876543,
-      "eval_overall_f1": 0.9379652605459058,
-      "eval_overall_precision": 0.9333333333333333,
-      "eval_overall_recall": 0.942643391521197,
-      "eval_runtime": 0.6274,
-      "eval_samples_per_second": 270.938,
-      "eval_steps_per_second": 4.781,
+      "eval_loss": 0.07066118717193604,
+      "eval_overall_accuracy": 0.9849108367626886,
+      "eval_overall_f1": 0.9326805385556914,
+      "eval_overall_precision": 0.9158653846153846,
+      "eval_overall_recall": 0.9501246882793017,
+      "eval_runtime": 0.2903,
+      "eval_samples_per_second": 585.543,
+      "eval_steps_per_second": 10.333,
       "step": 5184
     },
     {
       "epoch": 55.0,
-      "grad_norm": 0.7094045281410217,
+      "grad_norm": 0.6687442064285278,
       "learning_rate": 2.25e-05,
-      "loss": 0.0152,
+      "loss": 0.0133,
       "step": 5280
     },
     {
       "epoch": 55.0,
-      "eval_LOCATION_f1": 0.9270833333333334,
+      "eval_LOCATION_f1": 0.922279792746114,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.898989898989899,
+      "eval_LOCATION_precision": 0.89,
       "eval_LOCATION_recall": 0.956989247311828,
-      "eval_ORGANIZATION_f1": 0.924924924924925,
+      "eval_ORGANIZATION_f1": 0.9146341463414634,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9221556886227545,
-      "eval_ORGANIZATION_recall": 0.927710843373494,
-      "eval_PERSON_f1": 0.9716312056737589,
+      "eval_ORGANIZATION_precision": 0.9259259259259259,
+      "eval_ORGANIZATION_recall": 0.9036144578313253,
+      "eval_PERSON_f1": 0.9752650176678446,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9785714285714285,
-      "eval_PERSON_recall": 0.9647887323943662,
-      "eval_loss": 0.052083536982536316,
-      "eval_overall_accuracy": 0.9887517146776406,
-      "eval_overall_f1": 0.9417596034696407,
-      "eval_overall_precision": 0.9359605911330049,
-      "eval_overall_recall": 0.9476309226932669,
-      "eval_runtime": 0.6299,
-      "eval_samples_per_second": 269.891,
-      "eval_steps_per_second": 4.763,
+      "eval_PERSON_precision": 0.9787234042553191,
+      "eval_PERSON_recall": 0.971830985915493,
+      "eval_loss": 0.06329189985990524,
+      "eval_overall_accuracy": 0.9868312757201646,
+      "eval_overall_f1": 0.9378109452736318,
+      "eval_overall_precision": 0.9354838709677419,
+      "eval_overall_recall": 0.940149625935162,
+      "eval_runtime": 0.2896,
+      "eval_samples_per_second": 587.102,
+      "eval_steps_per_second": 10.361,
       "step": 5280
     },
     {
       "epoch": 56.0,
-      "grad_norm": 0.5432300567626953,
+      "grad_norm": 0.35132700204849243,
       "learning_rate": 2.2000000000000003e-05,
-      "loss": 0.0123,
+      "loss": 0.013,
       "step": 5376
     },
     {
       "epoch": 56.0,
-      "eval_LOCATION_f1": 0.9263157894736843,
+      "eval_LOCATION_f1": 0.9214659685863874,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.9072164948453608,
+      "eval_LOCATION_precision": 0.8979591836734694,
       "eval_LOCATION_recall": 0.946236559139785,
-      "eval_ORGANIZATION_f1": 0.9272727272727274,
+      "eval_ORGANIZATION_f1": 0.9216867469879518,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9329268292682927,
+      "eval_ORGANIZATION_precision": 0.9216867469879518,
       "eval_ORGANIZATION_recall": 0.9216867469879518,
-      "eval_PERSON_f1": 0.9752650176678446,
+      "eval_PERSON_f1": 0.9681978798586572,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9787234042553191,
-      "eval_PERSON_recall": 0.971830985915493,
-      "eval_loss": 0.05357722193002701,
-      "eval_overall_accuracy": 0.9887517146776406,
-      "eval_overall_f1": 0.9439601494396015,
-      "eval_overall_precision": 0.9427860696517413,
-      "eval_overall_recall": 0.9451371571072319,
-      "eval_runtime": 0.6253,
-      "eval_samples_per_second": 271.851,
-      "eval_steps_per_second": 4.797,
+      "eval_PERSON_precision": 0.9716312056737588,
+      "eval_PERSON_recall": 0.9647887323943662,
+      "eval_loss": 0.059241946786642075,
+      "eval_overall_accuracy": 0.9879286694101509,
+      "eval_overall_f1": 0.9379652605459058,
+      "eval_overall_precision": 0.9333333333333333,
+      "eval_overall_recall": 0.942643391521197,
+      "eval_runtime": 0.2893,
+      "eval_samples_per_second": 587.723,
+      "eval_steps_per_second": 10.372,
       "step": 5376
     },
     {
       "epoch": 57.0,
-      "grad_norm": 0.9852465391159058,
+      "grad_norm": 0.6489832401275635,
       "learning_rate": 2.15e-05,
-      "loss": 0.0145,
+      "loss": 0.0143,
       "step": 5472
     },
     {
       "epoch": 57.0,
-      "eval_LOCATION_f1": 0.9312169312169313,
+      "eval_LOCATION_f1": 0.9128205128205128,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.9166666666666666,
-      "eval_LOCATION_recall": 0.946236559139785,
-      "eval_ORGANIZATION_f1": 0.918429003021148,
+      "eval_LOCATION_precision": 0.8725490196078431,
+      "eval_LOCATION_recall": 0.956989247311828,
+      "eval_ORGANIZATION_f1": 0.9207317073170733,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9212121212121213,
-      "eval_ORGANIZATION_recall": 0.9156626506024096,
-      "eval_PERSON_f1": 0.9752650176678446,
+      "eval_ORGANIZATION_precision": 0.9320987654320988,
+      "eval_ORGANIZATION_recall": 0.9096385542168675,
+      "eval_PERSON_f1": 0.9788732394366197,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9787234042553191,
-      "eval_PERSON_recall": 0.971830985915493,
-      "eval_loss": 0.05067111924290657,
-      "eval_overall_accuracy": 0.9898491083676269,
-      "eval_overall_f1": 0.941469489414695,
-      "eval_overall_precision": 0.9402985074626866,
-      "eval_overall_recall": 0.942643391521197,
-      "eval_runtime": 0.7174,
-      "eval_samples_per_second": 236.961,
-      "eval_steps_per_second": 4.182,
+      "eval_PERSON_precision": 0.9788732394366197,
+      "eval_PERSON_recall": 0.9788732394366197,
+      "eval_loss": 0.0567299947142601,
+      "eval_overall_accuracy": 0.9887517146776406,
+      "eval_overall_f1": 0.9392812887236679,
+      "eval_overall_precision": 0.9334975369458128,
+      "eval_overall_recall": 0.9451371571072319,
+      "eval_runtime": 0.2876,
+      "eval_samples_per_second": 591.033,
+      "eval_steps_per_second": 10.43,
       "step": 5472
     },
     {
       "epoch": 58.0,
-      "grad_norm": 0.5912477374076843,
+      "grad_norm": 0.1813511699438095,
       "learning_rate": 2.1e-05,
-      "loss": 0.0138,
+      "loss": 0.0135,
       "step": 5568
     },
     {
       "epoch": 58.0,
-      "eval_LOCATION_f1": 0.9214659685863874,
+      "eval_LOCATION_f1": 0.9175257731958764,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.8979591836734694,
-      "eval_LOCATION_recall": 0.946236559139785,
-      "eval_ORGANIZATION_f1": 0.9179331306990882,
+      "eval_LOCATION_precision": 0.8811881188118812,
+      "eval_LOCATION_recall": 0.956989247311828,
+      "eval_ORGANIZATION_f1": 0.9240121580547112,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9263803680981595,
-      "eval_ORGANIZATION_recall": 0.9096385542168675,
-      "eval_PERSON_f1": 0.9681978798586572,
+      "eval_ORGANIZATION_precision": 0.9325153374233128,
+      "eval_ORGANIZATION_recall": 0.9156626506024096,
+      "eval_PERSON_f1": 0.9823321554770318,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9716312056737588,
-      "eval_PERSON_recall": 0.9647887323943662,
-      "eval_loss": 0.054992906749248505,
-      "eval_overall_accuracy": 0.9893004115226337,
-      "eval_overall_f1": 0.9364881693648817,
-      "eval_overall_precision": 0.9353233830845771,
-      "eval_overall_recall": 0.9376558603491272,
-      "eval_runtime": 0.636,
-      "eval_samples_per_second": 267.284,
-      "eval_steps_per_second": 4.717,
+      "eval_PERSON_precision": 0.9858156028368794,
+      "eval_PERSON_recall": 0.9788732394366197,
+      "eval_loss": 0.06323638558387756,
+      "eval_overall_accuracy": 0.9876543209876543,
+      "eval_overall_f1": 0.9429280397022332,
+      "eval_overall_precision": 0.9382716049382716,
+      "eval_overall_recall": 0.9476309226932669,
+      "eval_runtime": 0.2942,
+      "eval_samples_per_second": 577.895,
+      "eval_steps_per_second": 10.198,
       "step": 5568
     },
     {
       "epoch": 59.0,
-      "grad_norm": 0.44573214650154114,
+      "grad_norm": 0.027425482869148254,
       "learning_rate": 2.05e-05,
-      "loss": 0.0122,
+      "loss": 0.0111,
       "step": 5664
     },
     {
       "epoch": 59.0,
-      "eval_LOCATION_f1": 0.9381443298969072,
+      "eval_LOCATION_f1": 0.9263157894736843,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.900990099009901,
-      "eval_LOCATION_recall": 0.978494623655914,
-      "eval_ORGANIZATION_f1": 0.9235474006116209,
+      "eval_LOCATION_precision": 0.9072164948453608,
+      "eval_LOCATION_recall": 0.946236559139785,
+      "eval_ORGANIZATION_f1": 0.9174311926605505,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.937888198757764,
-      "eval_ORGANIZATION_recall": 0.9096385542168675,
-      "eval_PERSON_f1": 0.9752650176678446,
+      "eval_ORGANIZATION_precision": 0.9316770186335404,
+      "eval_ORGANIZATION_recall": 0.9036144578313253,
+      "eval_PERSON_f1": 0.9681978798586572,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9787234042553191,
-      "eval_PERSON_recall": 0.971830985915493,
-      "eval_loss": 0.05316930636763573,
-      "eval_overall_accuracy": 0.9882030178326474,
-      "eval_overall_f1": 0.945273631840796,
-      "eval_overall_precision": 0.9429280397022333,
-      "eval_overall_recall": 0.9476309226932669,
-      "eval_runtime": 0.6375,
-      "eval_samples_per_second": 266.686,
-      "eval_steps_per_second": 4.706,
+      "eval_PERSON_precision": 0.9716312056737588,
+      "eval_PERSON_recall": 0.9647887323943662,
+      "eval_loss": 0.06284485757350922,
+      "eval_overall_accuracy": 0.9871056241426612,
+      "eval_overall_f1": 0.9375,
+      "eval_overall_precision": 0.9398496240601504,
+      "eval_overall_recall": 0.9351620947630923,
+      "eval_runtime": 0.2885,
+      "eval_samples_per_second": 589.346,
+      "eval_steps_per_second": 10.4,
       "step": 5664
     },
     {
       "epoch": 60.0,
-      "grad_norm": 0.14488360285758972,
+      "grad_norm": 0.3114117383956909,
       "learning_rate": 2e-05,
-      "loss": 0.0144,
+      "loss": 0.0108,
       "step": 5760
     },
     {
       "epoch": 60.0,
-      "eval_LOCATION_f1": 0.9285714285714286,
+      "eval_LOCATION_f1": 0.9157894736842105,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.883495145631068,
-      "eval_LOCATION_recall": 0.978494623655914,
-      "eval_ORGANIZATION_f1": 0.9135802469135801,
+      "eval_LOCATION_precision": 0.8969072164948454,
+      "eval_LOCATION_recall": 0.9354838709677419,
+      "eval_ORGANIZATION_f1": 0.9129129129129129,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9367088607594937,
-      "eval_ORGANIZATION_recall": 0.891566265060241,
-      "eval_PERSON_f1": 0.971830985915493,
+      "eval_ORGANIZATION_precision": 0.9101796407185628,
+      "eval_ORGANIZATION_recall": 0.9156626506024096,
+      "eval_PERSON_f1": 0.9681978798586572,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.971830985915493,
-      "eval_PERSON_recall": 0.971830985915493,
-      "eval_loss": 0.05720693990588188,
-      "eval_overall_accuracy": 0.9882030178326474,
-      "eval_overall_f1": 0.9378109452736318,
-      "eval_overall_precision": 0.9354838709677419,
-      "eval_overall_recall": 0.940149625935162,
-      "eval_runtime": 0.6336,
-      "eval_samples_per_second": 268.304,
-      "eval_steps_per_second": 4.735,
+      "eval_PERSON_precision": 0.9716312056737588,
+      "eval_PERSON_recall": 0.9647887323943662,
+      "eval_loss": 0.05974132940173149,
+      "eval_overall_accuracy": 0.9868312757201646,
+      "eval_overall_f1": 0.9330024813895781,
+      "eval_overall_precision": 0.928395061728395,
+      "eval_overall_recall": 0.9376558603491272,
+      "eval_runtime": 0.2915,
+      "eval_samples_per_second": 583.278,
+      "eval_steps_per_second": 10.293,
       "step": 5760
     },
     {
       "epoch": 61.0,
-      "grad_norm": 0.5334123373031616,
+      "grad_norm": 0.14523079991340637,
       "learning_rate": 1.9500000000000003e-05,
-      "loss": 0.0126,
+      "loss": 0.0144,
       "step": 5856
     },
     {
       "epoch": 61.0,
-      "eval_LOCATION_f1": 0.9574468085106383,
+      "eval_LOCATION_f1": 0.9166666666666667,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.9473684210526315,
-      "eval_LOCATION_recall": 0.967741935483871,
-      "eval_ORGANIZATION_f1": 0.9333333333333333,
+      "eval_LOCATION_precision": 0.8888888888888888,
+      "eval_LOCATION_recall": 0.946236559139785,
+      "eval_ORGANIZATION_f1": 0.9169230769230768,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9390243902439024,
-      "eval_ORGANIZATION_recall": 0.927710843373494,
-      "eval_PERSON_f1": 0.971830985915493,
+      "eval_ORGANIZATION_precision": 0.9371069182389937,
+      "eval_ORGANIZATION_recall": 0.8975903614457831,
+      "eval_PERSON_f1": 0.9681978798586572,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.971830985915493,
-      "eval_PERSON_recall": 0.971830985915493,
-      "eval_loss": 0.048754770308732986,
-      "eval_overall_accuracy": 0.9906721536351166,
-      "eval_overall_f1": 0.9526184538653366,
-      "eval_overall_precision": 0.9526184538653366,
-      "eval_overall_recall": 0.9526184538653366,
-      "eval_runtime": 0.63,
-      "eval_samples_per_second": 269.825,
-      "eval_steps_per_second": 4.762,
+      "eval_PERSON_precision": 0.9716312056737588,
+      "eval_PERSON_recall": 0.9647887323943662,
+      "eval_loss": 0.06437654048204422,
+      "eval_overall_accuracy": 0.9860082304526749,
+      "eval_overall_f1": 0.935,
+      "eval_overall_precision": 0.9373433583959899,
+      "eval_overall_recall": 0.9326683291770573,
+      "eval_runtime": 0.2896,
+      "eval_samples_per_second": 587.057,
+      "eval_steps_per_second": 10.36,
       "step": 5856
     },
     {
       "epoch": 62.0,
-      "grad_norm": 1.1560488939285278,
+      "grad_norm": 0.9911012649536133,
       "learning_rate": 1.9e-05,
-      "loss": 0.0113,
+      "loss": 0.0124,
       "step": 5952
     },
     {
       "epoch": 62.0,
-      "eval_LOCATION_f1": 0.9278350515463919,
+      "eval_LOCATION_f1": 0.9109947643979057,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.8910891089108911,
-      "eval_LOCATION_recall": 0.967741935483871,
-      "eval_ORGANIZATION_f1": 0.9079754601226994,
+      "eval_LOCATION_precision": 0.8877551020408163,
+      "eval_LOCATION_recall": 0.9354838709677419,
+      "eval_ORGANIZATION_f1": 0.9166666666666667,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.925,
-      "eval_ORGANIZATION_recall": 0.891566265060241,
-      "eval_PERSON_f1": 0.971830985915493,
+      "eval_ORGANIZATION_precision": 0.9058823529411765,
+      "eval_ORGANIZATION_recall": 0.927710843373494,
+      "eval_PERSON_f1": 0.9787234042553192,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.971830985915493,
+      "eval_PERSON_precision": 0.9857142857142858,
       "eval_PERSON_recall": 0.971830985915493,
-      "eval_loss": 0.056291624903678894,
-      "eval_overall_accuracy": 0.9887517146776406,
-      "eval_overall_f1": 0.9353233830845771,
-      "eval_overall_precision": 0.9330024813895782,
-      "eval_overall_recall": 0.9376558603491272,
-      "eval_runtime": 0.626,
-      "eval_samples_per_second": 271.571,
-      "eval_steps_per_second": 4.792,
+      "eval_loss": 0.05878928676247597,
+      "eval_overall_accuracy": 0.9871056241426612,
+      "eval_overall_f1": 0.9369592088998765,
+      "eval_overall_precision": 0.928921568627451,
+      "eval_overall_recall": 0.9451371571072319,
+      "eval_runtime": 0.2911,
+      "eval_samples_per_second": 583.943,
+      "eval_steps_per_second": 10.305,
       "step": 5952
     },
     {
       "epoch": 63.0,
-      "grad_norm": 0.4997645616531372,
+      "grad_norm": 0.3812105059623718,
       "learning_rate": 1.85e-05,
-      "loss": 0.0109,
+      "loss": 0.0117,
       "step": 6048
     },
     {
       "epoch": 63.0,
-      "eval_LOCATION_f1": 0.9128205128205128,
+      "eval_LOCATION_f1": 0.9214659685863874,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.8725490196078431,
-      "eval_LOCATION_recall": 0.956989247311828,
-      "eval_ORGANIZATION_f1": 0.8895705521472392,
+      "eval_LOCATION_precision": 0.8979591836734694,
+      "eval_LOCATION_recall": 0.946236559139785,
+      "eval_ORGANIZATION_f1": 0.9281437125748503,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.90625,
-      "eval_ORGANIZATION_recall": 0.8734939759036144,
-      "eval_PERSON_f1": 0.9823321554770318,
+      "eval_ORGANIZATION_precision": 0.9226190476190477,
+      "eval_ORGANIZATION_recall": 0.9337349397590361,
+      "eval_PERSON_f1": 0.9787234042553192,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9858156028368794,
-      "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.0592079721391201,
-      "eval_overall_accuracy": 0.9865569272976681,
-      "eval_overall_f1": 0.927860696517413,
-      "eval_overall_precision": 0.9255583126550868,
-      "eval_overall_recall": 0.9301745635910225,
-      "eval_runtime": 0.635,
-      "eval_samples_per_second": 267.73,
-      "eval_steps_per_second": 4.725,
+      "eval_PERSON_precision": 0.9857142857142858,
+      "eval_PERSON_recall": 0.971830985915493,
+      "eval_loss": 0.0626605674624443,
+      "eval_overall_accuracy": 0.9879286694101509,
+      "eval_overall_f1": 0.9442379182156134,
+      "eval_overall_precision": 0.9384236453201971,
+      "eval_overall_recall": 0.9501246882793017,
+      "eval_runtime": 0.2883,
+      "eval_samples_per_second": 589.687,
+      "eval_steps_per_second": 10.406,
       "step": 6048
     },
     {
       "epoch": 64.0,
-      "grad_norm": 0.4493540823459625,
+      "grad_norm": 0.28910690546035767,
       "learning_rate": 1.8e-05,
-      "loss": 0.0124,
+      "loss": 0.0101,
       "step": 6144
     },
     {
       "epoch": 64.0,
-      "eval_LOCATION_f1": 0.9304812834224598,
+      "eval_LOCATION_f1": 0.9263157894736843,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.925531914893617,
-      "eval_LOCATION_recall": 0.9354838709677419,
-      "eval_ORGANIZATION_f1": 0.9030303030303031,
+      "eval_LOCATION_precision": 0.9072164948453608,
+      "eval_LOCATION_recall": 0.946236559139785,
+      "eval_ORGANIZATION_f1": 0.924924924924925,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9085365853658537,
-      "eval_ORGANIZATION_recall": 0.8975903614457831,
-      "eval_PERSON_f1": 0.971830985915493,
+      "eval_ORGANIZATION_precision": 0.9221556886227545,
+      "eval_ORGANIZATION_recall": 0.927710843373494,
+      "eval_PERSON_f1": 0.9787234042553192,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.971830985915493,
+      "eval_PERSON_precision": 0.9857142857142858,
       "eval_PERSON_recall": 0.971830985915493,
-      "eval_loss": 0.05676347389817238,
-      "eval_overall_accuracy": 0.9876543209876543,
-      "eval_overall_f1": 0.933832709113608,
-      "eval_overall_precision": 0.935,
-      "eval_overall_recall": 0.9326683291770573,
-      "eval_runtime": 0.6373,
-      "eval_samples_per_second": 266.761,
-      "eval_steps_per_second": 4.708,
+      "eval_loss": 0.0623946376144886,
+      "eval_overall_accuracy": 0.9871056241426612,
+      "eval_overall_f1": 0.9440993788819876,
+      "eval_overall_precision": 0.9405940594059405,
+      "eval_overall_recall": 0.9476309226932669,
+      "eval_runtime": 0.2886,
+      "eval_samples_per_second": 588.967,
+      "eval_steps_per_second": 10.394,
       "step": 6144
     },
     {
       "epoch": 65.0,
-      "grad_norm": 0.3998490273952484,
+      "grad_norm": 0.6301209926605225,
       "learning_rate": 1.75e-05,
-      "loss": 0.0126,
+      "loss": 0.0117,
       "step": 6240
     },
     {
       "epoch": 65.0,
-      "eval_LOCATION_f1": 0.9270833333333334,
+      "eval_LOCATION_f1": 0.9166666666666667,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.898989898989899,
-      "eval_LOCATION_recall": 0.956989247311828,
-      "eval_ORGANIZATION_f1": 0.8929663608562691,
+      "eval_LOCATION_precision": 0.8888888888888888,
+      "eval_LOCATION_recall": 0.946236559139785,
+      "eval_ORGANIZATION_f1": 0.9101796407185628,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.906832298136646,
-      "eval_ORGANIZATION_recall": 0.8795180722891566,
-      "eval_PERSON_f1": 0.9823321554770318,
+      "eval_ORGANIZATION_precision": 0.9047619047619048,
+      "eval_ORGANIZATION_recall": 0.9156626506024096,
+      "eval_PERSON_f1": 0.9787234042553192,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9858156028368794,
-      "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.05590180307626724,
-      "eval_overall_accuracy": 0.9876543209876543,
-      "eval_overall_f1": 0.9326683291770573,
-      "eval_overall_precision": 0.9326683291770573,
-      "eval_overall_recall": 0.9326683291770573,
-      "eval_runtime": 0.6342,
-      "eval_samples_per_second": 268.035,
-      "eval_steps_per_second": 4.73,
+      "eval_PERSON_precision": 0.9857142857142858,
+      "eval_PERSON_recall": 0.971830985915493,
+      "eval_loss": 0.06263893097639084,
+      "eval_overall_accuracy": 0.9871056241426612,
+      "eval_overall_f1": 0.9356435643564357,
+      "eval_overall_precision": 0.9287469287469288,
+      "eval_overall_recall": 0.942643391521197,
+      "eval_runtime": 0.2889,
+      "eval_samples_per_second": 588.45,
+      "eval_steps_per_second": 10.384,
       "step": 6240
     },
     {
       "epoch": 66.0,
-      "grad_norm": 0.5854397416114807,
+      "grad_norm": 0.4729289412498474,
       "learning_rate": 1.7000000000000003e-05,
-      "loss": 0.0112,
+      "loss": 0.0108,
       "step": 6336
     },
     {
       "epoch": 66.0,
-      "eval_LOCATION_f1": 0.9361702127659575,
+      "eval_LOCATION_f1": 0.9052631578947367,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.9263157894736842,
-      "eval_LOCATION_recall": 0.946236559139785,
-      "eval_ORGANIZATION_f1": 0.9123867069486404,
+      "eval_LOCATION_precision": 0.8865979381443299,
+      "eval_LOCATION_recall": 0.9247311827956989,
+      "eval_ORGANIZATION_f1": 0.9179331306990882,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9151515151515152,
+      "eval_ORGANIZATION_precision": 0.9263803680981595,
       "eval_ORGANIZATION_recall": 0.9096385542168675,
-      "eval_PERSON_f1": 0.9752650176678446,
+      "eval_PERSON_f1": 0.9647887323943662,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9787234042553191,
-      "eval_PERSON_recall": 0.971830985915493,
-      "eval_loss": 0.057300012558698654,
-      "eval_overall_accuracy": 0.988477366255144,
-      "eval_overall_f1": 0.940149625935162,
-      "eval_overall_precision": 0.940149625935162,
-      "eval_overall_recall": 0.940149625935162,
-      "eval_runtime": 0.6275,
-      "eval_samples_per_second": 270.902,
-      "eval_steps_per_second": 4.781,
+      "eval_PERSON_precision": 0.9647887323943662,
+      "eval_PERSON_recall": 0.9647887323943662,
+      "eval_loss": 0.05901220068335533,
+      "eval_overall_accuracy": 0.9876543209876543,
+      "eval_overall_f1": 0.9315068493150686,
+      "eval_overall_precision": 0.9303482587064676,
+      "eval_overall_recall": 0.9326683291770573,
+      "eval_runtime": 0.2942,
+      "eval_samples_per_second": 577.799,
+      "eval_steps_per_second": 10.196,
       "step": 6336
     },
     {
       "epoch": 67.0,
-      "grad_norm": 0.009959683753550053,
+      "grad_norm": 0.20351357758045197,
       "learning_rate": 1.65e-05,
-      "loss": 0.0112,
+      "loss": 0.0104,
       "step": 6432
     },
     {
       "epoch": 67.0,
-      "eval_LOCATION_f1": 0.9035532994923857,
+      "eval_LOCATION_f1": 0.9270833333333334,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.8557692307692307,
+      "eval_LOCATION_precision": 0.898989898989899,
       "eval_LOCATION_recall": 0.956989247311828,
-      "eval_ORGANIZATION_f1": 0.8848484848484849,
+      "eval_ORGANIZATION_f1": 0.9161676646706587,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.8902439024390244,
-      "eval_ORGANIZATION_recall": 0.8795180722891566,
-      "eval_PERSON_f1": 0.9823321554770318,
+      "eval_ORGANIZATION_precision": 0.9107142857142857,
+      "eval_ORGANIZATION_recall": 0.9216867469879518,
+      "eval_PERSON_f1": 0.9787234042553192,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9858156028368794,
-      "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.06168980151414871,
-      "eval_overall_accuracy": 0.9862825788751715,
-      "eval_overall_f1": 0.9234567901234567,
-      "eval_overall_precision": 0.9144254278728606,
-      "eval_overall_recall": 0.9326683291770573,
-      "eval_runtime": 0.6306,
-      "eval_samples_per_second": 269.564,
-      "eval_steps_per_second": 4.757,
+      "eval_PERSON_precision": 0.9857142857142858,
+      "eval_PERSON_recall": 0.971830985915493,
+      "eval_loss": 0.06248219311237335,
+      "eval_overall_accuracy": 0.9876543209876543,
+      "eval_overall_f1": 0.9405940594059405,
+      "eval_overall_precision": 0.9336609336609336,
+      "eval_overall_recall": 0.9476309226932669,
+      "eval_runtime": 0.2905,
+      "eval_samples_per_second": 585.283,
+      "eval_steps_per_second": 10.329,
       "step": 6432
     },
     {
       "epoch": 68.0,
-      "grad_norm": 0.43869954347610474,
+      "grad_norm": 0.7123875617980957,
       "learning_rate": 1.6000000000000003e-05,
-      "loss": 0.0117,
+      "loss": 0.0119,
       "step": 6528
     },
     {
@@ -2098,1024 +2098,1024 @@
       "eval_LOCATION_number": 93,
       "eval_LOCATION_precision": 0.9081632653061225,
       "eval_LOCATION_recall": 0.956989247311828,
-      "eval_ORGANIZATION_f1": 0.9146341463414634,
+      "eval_ORGANIZATION_f1": 0.9166666666666667,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9259259259259259,
-      "eval_ORGANIZATION_recall": 0.9036144578313253,
-      "eval_PERSON_f1": 0.9823321554770318,
+      "eval_ORGANIZATION_precision": 0.9058823529411765,
+      "eval_ORGANIZATION_recall": 0.927710843373494,
+      "eval_PERSON_f1": 0.9787234042553192,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9858156028368794,
-      "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.05341716110706329,
-      "eval_overall_accuracy": 0.988477366255144,
-      "eval_overall_f1": 0.942643391521197,
-      "eval_overall_precision": 0.942643391521197,
-      "eval_overall_recall": 0.942643391521197,
-      "eval_runtime": 0.64,
-      "eval_samples_per_second": 265.619,
-      "eval_steps_per_second": 4.687,
+      "eval_PERSON_precision": 0.9857142857142858,
+      "eval_PERSON_recall": 0.971830985915493,
+      "eval_loss": 0.06472578644752502,
+      "eval_overall_accuracy": 0.9873799725651577,
+      "eval_overall_f1": 0.9419035846724352,
+      "eval_overall_precision": 0.9338235294117647,
+      "eval_overall_recall": 0.9501246882793017,
+      "eval_runtime": 0.2879,
+      "eval_samples_per_second": 590.566,
+      "eval_steps_per_second": 10.422,
       "step": 6528
     },
     {
       "epoch": 69.0,
-      "grad_norm": 0.6488298177719116,
+      "grad_norm": 0.5229490399360657,
       "learning_rate": 1.55e-05,
-      "loss": 0.0101,
+      "loss": 0.011,
       "step": 6624
     },
     {
       "epoch": 69.0,
-      "eval_LOCATION_f1": 0.9417989417989417,
+      "eval_LOCATION_f1": 0.9214659685863874,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.9270833333333334,
-      "eval_LOCATION_recall": 0.956989247311828,
+      "eval_LOCATION_precision": 0.8979591836734694,
+      "eval_LOCATION_recall": 0.946236559139785,
       "eval_ORGANIZATION_f1": 0.9156626506024096,
       "eval_ORGANIZATION_number": 166,
       "eval_ORGANIZATION_precision": 0.9156626506024096,
       "eval_ORGANIZATION_recall": 0.9156626506024096,
-      "eval_PERSON_f1": 0.9823321554770318,
+      "eval_PERSON_f1": 0.9752650176678446,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9858156028368794,
-      "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.05710820481181145,
-      "eval_overall_accuracy": 0.9890260631001372,
-      "eval_overall_f1": 0.945273631840796,
-      "eval_overall_precision": 0.9429280397022333,
-      "eval_overall_recall": 0.9476309226932669,
-      "eval_runtime": 0.6329,
-      "eval_samples_per_second": 268.611,
-      "eval_steps_per_second": 4.74,
+      "eval_PERSON_precision": 0.9787234042553191,
+      "eval_PERSON_recall": 0.971830985915493,
+      "eval_loss": 0.05594140291213989,
+      "eval_overall_accuracy": 0.9887517146776406,
+      "eval_overall_f1": 0.9379652605459058,
+      "eval_overall_precision": 0.9333333333333333,
+      "eval_overall_recall": 0.942643391521197,
+      "eval_runtime": 0.2896,
+      "eval_samples_per_second": 586.973,
+      "eval_steps_per_second": 10.358,
       "step": 6624
     },
     {
       "epoch": 70.0,
-      "grad_norm": 0.07458376884460449,
+      "grad_norm": 0.36743980646133423,
       "learning_rate": 1.5e-05,
-      "loss": 0.0092,
+      "loss": 0.0096,
       "step": 6720
     },
     {
       "epoch": 70.0,
-      "eval_LOCATION_f1": 0.9374999999999999,
+      "eval_LOCATION_f1": 0.9270833333333334,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.9090909090909091,
-      "eval_LOCATION_recall": 0.967741935483871,
-      "eval_ORGANIZATION_f1": 0.9240121580547112,
+      "eval_LOCATION_precision": 0.898989898989899,
+      "eval_LOCATION_recall": 0.956989247311828,
+      "eval_ORGANIZATION_f1": 0.9129129129129129,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9325153374233128,
+      "eval_ORGANIZATION_precision": 0.9101796407185628,
       "eval_ORGANIZATION_recall": 0.9156626506024096,
-      "eval_PERSON_f1": 0.9823321554770318,
+      "eval_PERSON_f1": 0.9787234042553192,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9858156028368794,
-      "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.05593955144286156,
-      "eval_overall_accuracy": 0.9893004115226337,
-      "eval_overall_f1": 0.9477611940298507,
-      "eval_overall_precision": 0.9454094292803971,
-      "eval_overall_recall": 0.9501246882793017,
-      "eval_runtime": 0.631,
-      "eval_samples_per_second": 269.43,
-      "eval_steps_per_second": 4.755,
+      "eval_PERSON_precision": 0.9857142857142858,
+      "eval_PERSON_recall": 0.971830985915493,
+      "eval_loss": 0.05785645917057991,
+      "eval_overall_accuracy": 0.9882030178326474,
+      "eval_overall_f1": 0.9392812887236679,
+      "eval_overall_precision": 0.9334975369458128,
+      "eval_overall_recall": 0.9451371571072319,
+      "eval_runtime": 0.2897,
+      "eval_samples_per_second": 586.718,
+      "eval_steps_per_second": 10.354,
       "step": 6720
     },
     {
       "epoch": 71.0,
-      "grad_norm": 1.7527589797973633,
+      "grad_norm": 1.1272770166397095,
       "learning_rate": 1.45e-05,
-      "loss": 0.011,
+      "loss": 0.0103,
       "step": 6816
     },
     {
       "epoch": 71.0,
-      "eval_LOCATION_f1": 0.946808510638298,
+      "eval_LOCATION_f1": 0.9214659685863874,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.9368421052631579,
-      "eval_LOCATION_recall": 0.956989247311828,
-      "eval_ORGANIZATION_f1": 0.9272727272727274,
+      "eval_LOCATION_precision": 0.8979591836734694,
+      "eval_LOCATION_recall": 0.946236559139785,
+      "eval_ORGANIZATION_f1": 0.9047619047619048,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9329268292682927,
-      "eval_ORGANIZATION_recall": 0.9216867469879518,
-      "eval_PERSON_f1": 0.9823321554770318,
+      "eval_ORGANIZATION_precision": 0.8941176470588236,
+      "eval_ORGANIZATION_recall": 0.9156626506024096,
+      "eval_PERSON_f1": 0.9787234042553192,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9858156028368794,
-      "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.061140261590480804,
-      "eval_overall_accuracy": 0.9879286694101509,
-      "eval_overall_f1": 0.951310861423221,
-      "eval_overall_precision": 0.9525,
-      "eval_overall_recall": 0.9501246882793017,
-      "eval_runtime": 0.6251,
-      "eval_samples_per_second": 271.935,
-      "eval_steps_per_second": 4.799,
+      "eval_PERSON_precision": 0.9857142857142858,
+      "eval_PERSON_recall": 0.971830985915493,
+      "eval_loss": 0.06020239740610123,
+      "eval_overall_accuracy": 0.9876543209876543,
+      "eval_overall_f1": 0.9344870210135972,
+      "eval_overall_precision": 0.9264705882352942,
+      "eval_overall_recall": 0.942643391521197,
+      "eval_runtime": 0.2903,
+      "eval_samples_per_second": 585.692,
+      "eval_steps_per_second": 10.336,
       "step": 6816
     },
     {
       "epoch": 72.0,
-      "grad_norm": 0.09629790484905243,
+      "grad_norm": 0.5486194491386414,
       "learning_rate": 1.4000000000000001e-05,
-      "loss": 0.0097,
+      "loss": 0.0091,
       "step": 6912
     },
     {
       "epoch": 72.0,
-      "eval_LOCATION_f1": 0.9312169312169313,
+      "eval_LOCATION_f1": 0.9270833333333334,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.9166666666666666,
-      "eval_LOCATION_recall": 0.946236559139785,
-      "eval_ORGANIZATION_f1": 0.9268292682926829,
+      "eval_LOCATION_precision": 0.898989898989899,
+      "eval_LOCATION_recall": 0.956989247311828,
+      "eval_ORGANIZATION_f1": 0.9212121212121213,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9382716049382716,
+      "eval_ORGANIZATION_precision": 0.926829268292683,
       "eval_ORGANIZATION_recall": 0.9156626506024096,
-      "eval_PERSON_f1": 0.9823321554770318,
+      "eval_PERSON_f1": 0.9752650176678446,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9858156028368794,
-      "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.055803388357162476,
-      "eval_overall_accuracy": 0.9893004115226337,
-      "eval_overall_f1": 0.9475000000000001,
-      "eval_overall_precision": 0.949874686716792,
+      "eval_PERSON_precision": 0.9787234042553191,
+      "eval_PERSON_recall": 0.971830985915493,
+      "eval_loss": 0.0652991458773613,
+      "eval_overall_accuracy": 0.9876543209876543,
+      "eval_overall_f1": 0.9416149068322982,
+      "eval_overall_precision": 0.9381188118811881,
       "eval_overall_recall": 0.9451371571072319,
-      "eval_runtime": 0.6294,
-      "eval_samples_per_second": 270.101,
-      "eval_steps_per_second": 4.766,
+      "eval_runtime": 0.2905,
+      "eval_samples_per_second": 585.264,
+      "eval_steps_per_second": 10.328,
       "step": 6912
     },
     {
       "epoch": 73.0,
-      "grad_norm": 0.45607447624206543,
+      "grad_norm": 1.128568172454834,
       "learning_rate": 1.3500000000000001e-05,
-      "loss": 0.0102,
+      "loss": 0.0093,
       "step": 7008
     },
     {
       "epoch": 73.0,
-      "eval_LOCATION_f1": 0.9214659685863874,
+      "eval_LOCATION_f1": 0.9166666666666667,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.8979591836734694,
+      "eval_LOCATION_precision": 0.8888888888888888,
       "eval_LOCATION_recall": 0.946236559139785,
-      "eval_ORGANIZATION_f1": 0.9207317073170733,
+      "eval_ORGANIZATION_f1": 0.9096385542168675,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9320987654320988,
+      "eval_ORGANIZATION_precision": 0.9096385542168675,
       "eval_ORGANIZATION_recall": 0.9096385542168675,
-      "eval_PERSON_f1": 0.9823321554770318,
+      "eval_PERSON_f1": 0.9681978798586572,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9858156028368794,
-      "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.05601666495203972,
-      "eval_overall_accuracy": 0.9879286694101509,
-      "eval_overall_f1": 0.942643391521197,
-      "eval_overall_precision": 0.942643391521197,
-      "eval_overall_recall": 0.942643391521197,
-      "eval_runtime": 0.632,
-      "eval_samples_per_second": 268.986,
-      "eval_steps_per_second": 4.747,
+      "eval_PERSON_precision": 0.9716312056737588,
+      "eval_PERSON_recall": 0.9647887323943662,
+      "eval_loss": 0.06982634961605072,
+      "eval_overall_accuracy": 0.9865569272976681,
+      "eval_overall_f1": 0.9318463444857497,
+      "eval_overall_precision": 0.9261083743842364,
+      "eval_overall_recall": 0.9376558603491272,
+      "eval_runtime": 0.2922,
+      "eval_samples_per_second": 581.854,
+      "eval_steps_per_second": 10.268,
       "step": 7008
     },
     {
       "epoch": 74.0,
-      "grad_norm": 0.02336580492556095,
+      "grad_norm": 0.033605970442295074,
       "learning_rate": 1.3000000000000001e-05,
-      "loss": 0.0098,
+      "loss": 0.0091,
       "step": 7104
     },
     {
       "epoch": 74.0,
-      "eval_LOCATION_f1": 0.9424083769633509,
+      "eval_LOCATION_f1": 0.9263157894736843,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.9183673469387755,
-      "eval_LOCATION_recall": 0.967741935483871,
-      "eval_ORGANIZATION_f1": 0.918429003021148,
+      "eval_LOCATION_precision": 0.9072164948453608,
+      "eval_LOCATION_recall": 0.946236559139785,
+      "eval_ORGANIZATION_f1": 0.9074626865671641,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9212121212121213,
+      "eval_ORGANIZATION_precision": 0.8994082840236687,
       "eval_ORGANIZATION_recall": 0.9156626506024096,
-      "eval_PERSON_f1": 0.9823321554770318,
+      "eval_PERSON_f1": 0.9752650176678446,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9858156028368794,
-      "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.05871947109699249,
-      "eval_overall_accuracy": 0.9882030178326474,
-      "eval_overall_f1": 0.946583850931677,
-      "eval_overall_precision": 0.943069306930693,
-      "eval_overall_recall": 0.9501246882793017,
-      "eval_runtime": 0.6932,
-      "eval_samples_per_second": 245.254,
-      "eval_steps_per_second": 4.328,
+      "eval_PERSON_precision": 0.9787234042553191,
+      "eval_PERSON_recall": 0.971830985915493,
+      "eval_loss": 0.06609321385622025,
+      "eval_overall_accuracy": 0.9871056241426612,
+      "eval_overall_f1": 0.9356435643564357,
+      "eval_overall_precision": 0.9287469287469288,
+      "eval_overall_recall": 0.942643391521197,
+      "eval_runtime": 0.2892,
+      "eval_samples_per_second": 587.781,
+      "eval_steps_per_second": 10.373,
       "step": 7104
     },
     {
       "epoch": 75.0,
-      "grad_norm": 0.5431827902793884,
+      "grad_norm": 0.5324849486351013,
       "learning_rate": 1.25e-05,
-      "loss": 0.0105,
+      "loss": 0.0093,
       "step": 7200
     },
     {
       "epoch": 75.0,
-      "eval_LOCATION_f1": 0.9175257731958764,
+      "eval_LOCATION_f1": 0.90625,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.8811881188118812,
-      "eval_LOCATION_recall": 0.956989247311828,
-      "eval_ORGANIZATION_f1": 0.9057750759878419,
+      "eval_LOCATION_precision": 0.8787878787878788,
+      "eval_LOCATION_recall": 0.9354838709677419,
+      "eval_ORGANIZATION_f1": 0.9009009009009009,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9141104294478528,
-      "eval_ORGANIZATION_recall": 0.8975903614457831,
-      "eval_PERSON_f1": 0.9823321554770318,
+      "eval_ORGANIZATION_precision": 0.8982035928143712,
+      "eval_ORGANIZATION_recall": 0.9036144578313253,
+      "eval_PERSON_f1": 0.9681978798586572,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9858156028368794,
-      "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.05846357345581055,
-      "eval_overall_accuracy": 0.9873799725651577,
-      "eval_overall_f1": 0.9354838709677419,
-      "eval_overall_precision": 0.9308641975308642,
-      "eval_overall_recall": 0.940149625935162,
-      "eval_runtime": 0.6414,
-      "eval_samples_per_second": 265.034,
-      "eval_steps_per_second": 4.677,
+      "eval_PERSON_precision": 0.9716312056737588,
+      "eval_PERSON_recall": 0.9647887323943662,
+      "eval_loss": 0.07267607003450394,
+      "eval_overall_accuracy": 0.9854595336076818,
+      "eval_overall_f1": 0.9257425742574258,
+      "eval_overall_precision": 0.918918918918919,
+      "eval_overall_recall": 0.9326683291770573,
+      "eval_runtime": 0.2896,
+      "eval_samples_per_second": 586.938,
+      "eval_steps_per_second": 10.358,
       "step": 7200
     },
     {
       "epoch": 76.0,
-      "grad_norm": 0.803601086139679,
+      "grad_norm": 0.02331961691379547,
       "learning_rate": 1.2e-05,
-      "loss": 0.01,
+      "loss": 0.0087,
       "step": 7296
     },
     {
       "epoch": 76.0,
-      "eval_LOCATION_f1": 0.946808510638298,
+      "eval_LOCATION_f1": 0.922279792746114,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.9368421052631579,
+      "eval_LOCATION_precision": 0.89,
       "eval_LOCATION_recall": 0.956989247311828,
-      "eval_ORGANIZATION_f1": 0.9244712990936556,
+      "eval_ORGANIZATION_f1": 0.9129129129129129,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9272727272727272,
-      "eval_ORGANIZATION_recall": 0.9216867469879518,
-      "eval_PERSON_f1": 0.9823321554770318,
+      "eval_ORGANIZATION_precision": 0.9101796407185628,
+      "eval_ORGANIZATION_recall": 0.9156626506024096,
+      "eval_PERSON_f1": 0.9752650176678446,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9858156028368794,
-      "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.059869080781936646,
-      "eval_overall_accuracy": 0.9893004115226337,
-      "eval_overall_f1": 0.9501246882793017,
-      "eval_overall_precision": 0.9501246882793017,
-      "eval_overall_recall": 0.9501246882793017,
-      "eval_runtime": 0.6266,
-      "eval_samples_per_second": 271.309,
-      "eval_steps_per_second": 4.788,
+      "eval_PERSON_precision": 0.9787234042553191,
+      "eval_PERSON_recall": 0.971830985915493,
+      "eval_loss": 0.06968007981777191,
+      "eval_overall_accuracy": 0.9868312757201646,
+      "eval_overall_f1": 0.9369592088998765,
+      "eval_overall_precision": 0.928921568627451,
+      "eval_overall_recall": 0.9451371571072319,
+      "eval_runtime": 0.2896,
+      "eval_samples_per_second": 587.037,
+      "eval_steps_per_second": 10.359,
       "step": 7296
     },
     {
       "epoch": 77.0,
-      "grad_norm": 0.8899322152137756,
+      "grad_norm": 1.1283886432647705,
       "learning_rate": 1.1500000000000002e-05,
-      "loss": 0.0084,
+      "loss": 0.0104,
       "step": 7392
     },
     {
       "epoch": 77.0,
-      "eval_LOCATION_f1": 0.9374999999999999,
+      "eval_LOCATION_f1": 0.911917098445596,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.9090909090909091,
-      "eval_LOCATION_recall": 0.967741935483871,
-      "eval_ORGANIZATION_f1": 0.918429003021148,
+      "eval_LOCATION_precision": 0.88,
+      "eval_LOCATION_recall": 0.946236559139785,
+      "eval_ORGANIZATION_f1": 0.9212121212121213,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9212121212121213,
+      "eval_ORGANIZATION_precision": 0.926829268292683,
       "eval_ORGANIZATION_recall": 0.9156626506024096,
-      "eval_PERSON_f1": 0.9823321554770318,
+      "eval_PERSON_f1": 0.9787234042553192,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9858156028368794,
-      "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.060152728110551834,
-      "eval_overall_accuracy": 0.9893004115226337,
-      "eval_overall_f1": 0.9454094292803972,
-      "eval_overall_precision": 0.9407407407407408,
-      "eval_overall_recall": 0.9501246882793017,
-      "eval_runtime": 0.629,
-      "eval_samples_per_second": 270.252,
-      "eval_steps_per_second": 4.769,
+      "eval_PERSON_precision": 0.9857142857142858,
+      "eval_PERSON_recall": 0.971830985915493,
+      "eval_loss": 0.06767893582582474,
+      "eval_overall_accuracy": 0.9879286694101509,
+      "eval_overall_f1": 0.9391304347826087,
+      "eval_overall_precision": 0.9356435643564357,
+      "eval_overall_recall": 0.942643391521197,
+      "eval_runtime": 0.2891,
+      "eval_samples_per_second": 588.107,
+      "eval_steps_per_second": 10.378,
       "step": 7392
     },
     {
       "epoch": 78.0,
-      "grad_norm": 0.5667243599891663,
+      "grad_norm": 0.6260883212089539,
       "learning_rate": 1.1000000000000001e-05,
       "loss": 0.0089,
       "step": 7488
     },
     {
       "epoch": 78.0,
-      "eval_LOCATION_f1": 0.922279792746114,
+      "eval_LOCATION_f1": 0.9109947643979057,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.89,
-      "eval_LOCATION_recall": 0.956989247311828,
-      "eval_ORGANIZATION_f1": 0.9151515151515153,
+      "eval_LOCATION_precision": 0.8877551020408163,
+      "eval_LOCATION_recall": 0.9354838709677419,
+      "eval_ORGANIZATION_f1": 0.906906906906907,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9207317073170732,
+      "eval_ORGANIZATION_precision": 0.9041916167664671,
       "eval_ORGANIZATION_recall": 0.9096385542168675,
-      "eval_PERSON_f1": 0.9823321554770318,
+      "eval_PERSON_f1": 0.9681978798586572,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9858156028368794,
-      "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.06323432922363281,
-      "eval_overall_accuracy": 0.9879286694101509,
-      "eval_overall_f1": 0.9404466501240694,
-      "eval_overall_precision": 0.9358024691358025,
-      "eval_overall_recall": 0.9451371571072319,
-      "eval_runtime": 0.63,
-      "eval_samples_per_second": 269.835,
-      "eval_steps_per_second": 4.762,
+      "eval_PERSON_precision": 0.9716312056737588,
+      "eval_PERSON_recall": 0.9647887323943662,
+      "eval_loss": 0.06810528039932251,
+      "eval_overall_accuracy": 0.9862825788751715,
+      "eval_overall_f1": 0.9293680297397771,
+      "eval_overall_precision": 0.9236453201970444,
+      "eval_overall_recall": 0.9351620947630923,
+      "eval_runtime": 0.2872,
+      "eval_samples_per_second": 591.943,
+      "eval_steps_per_second": 10.446,
       "step": 7488
     },
     {
       "epoch": 79.0,
-      "grad_norm": 0.020913848653435707,
+      "grad_norm": 0.1493467390537262,
       "learning_rate": 1.05e-05,
-      "loss": 0.0086,
+      "loss": 0.0092,
       "step": 7584
     },
     {
       "epoch": 79.0,
-      "eval_LOCATION_f1": 0.9263157894736843,
+      "eval_LOCATION_f1": 0.9166666666666667,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.9072164948453608,
+      "eval_LOCATION_precision": 0.8888888888888888,
       "eval_LOCATION_recall": 0.946236559139785,
-      "eval_ORGANIZATION_f1": 0.9235474006116209,
+      "eval_ORGANIZATION_f1": 0.9014925373134328,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.937888198757764,
+      "eval_ORGANIZATION_precision": 0.893491124260355,
       "eval_ORGANIZATION_recall": 0.9096385542168675,
-      "eval_PERSON_f1": 0.9823321554770318,
+      "eval_PERSON_f1": 0.9787234042553192,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9858156028368794,
-      "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.05881747603416443,
-      "eval_overall_accuracy": 0.988477366255144,
-      "eval_overall_f1": 0.9450000000000001,
-      "eval_overall_precision": 0.9473684210526315,
-      "eval_overall_recall": 0.942643391521197,
-      "eval_runtime": 0.6316,
-      "eval_samples_per_second": 269.163,
-      "eval_steps_per_second": 4.75,
+      "eval_PERSON_precision": 0.9857142857142858,
+      "eval_PERSON_recall": 0.971830985915493,
+      "eval_loss": 0.06890777498483658,
+      "eval_overall_accuracy": 0.9862825788751715,
+      "eval_overall_f1": 0.9320148331273177,
+      "eval_overall_precision": 0.9240196078431373,
+      "eval_overall_recall": 0.940149625935162,
+      "eval_runtime": 0.291,
+      "eval_samples_per_second": 584.258,
+      "eval_steps_per_second": 10.31,
       "step": 7584
     },
     {
       "epoch": 80.0,
-      "grad_norm": 0.7093363404273987,
+      "grad_norm": 0.6644344329833984,
       "learning_rate": 1e-05,
-      "loss": 0.009,
+      "loss": 0.0069,
       "step": 7680
     },
     {
       "epoch": 80.0,
-      "eval_LOCATION_f1": 0.9368421052631579,
+      "eval_LOCATION_f1": 0.9166666666666667,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.9175257731958762,
-      "eval_LOCATION_recall": 0.956989247311828,
-      "eval_ORGANIZATION_f1": 0.9244712990936556,
+      "eval_LOCATION_precision": 0.8888888888888888,
+      "eval_LOCATION_recall": 0.946236559139785,
+      "eval_ORGANIZATION_f1": 0.9101796407185628,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9272727272727272,
-      "eval_ORGANIZATION_recall": 0.9216867469879518,
-      "eval_PERSON_f1": 0.9823321554770318,
+      "eval_ORGANIZATION_precision": 0.9047619047619048,
+      "eval_ORGANIZATION_recall": 0.9156626506024096,
+      "eval_PERSON_f1": 0.9787234042553192,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9858156028368794,
-      "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.06179361417889595,
-      "eval_overall_accuracy": 0.988477366255144,
-      "eval_overall_f1": 0.9477611940298507,
-      "eval_overall_precision": 0.9454094292803971,
-      "eval_overall_recall": 0.9501246882793017,
-      "eval_runtime": 0.641,
-      "eval_samples_per_second": 265.216,
-      "eval_steps_per_second": 4.68,
+      "eval_PERSON_precision": 0.9857142857142858,
+      "eval_PERSON_recall": 0.971830985915493,
+      "eval_loss": 0.07149485498666763,
+      "eval_overall_accuracy": 0.9871056241426612,
+      "eval_overall_f1": 0.9356435643564357,
+      "eval_overall_precision": 0.9287469287469288,
+      "eval_overall_recall": 0.942643391521197,
+      "eval_runtime": 0.2881,
+      "eval_samples_per_second": 590.06,
+      "eval_steps_per_second": 10.413,
       "step": 7680
     },
     {
       "epoch": 81.0,
-      "grad_norm": 0.23694337904453278,
+      "grad_norm": 1.0200001001358032,
       "learning_rate": 9.5e-06,
-      "loss": 0.0084,
+      "loss": 0.0087,
       "step": 7776
     },
     {
       "epoch": 81.0,
-      "eval_LOCATION_f1": 0.9368421052631579,
+      "eval_LOCATION_f1": 0.9166666666666667,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.9175257731958762,
-      "eval_LOCATION_recall": 0.956989247311828,
-      "eval_ORGANIZATION_f1": 0.9123867069486404,
+      "eval_LOCATION_precision": 0.8888888888888888,
+      "eval_LOCATION_recall": 0.946236559139785,
+      "eval_ORGANIZATION_f1": 0.9074626865671641,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9151515151515152,
-      "eval_ORGANIZATION_recall": 0.9096385542168675,
-      "eval_PERSON_f1": 0.9823321554770318,
+      "eval_ORGANIZATION_precision": 0.8994082840236687,
+      "eval_ORGANIZATION_recall": 0.9156626506024096,
+      "eval_PERSON_f1": 0.9787234042553192,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9858156028368794,
-      "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.06119789183139801,
-      "eval_overall_accuracy": 0.9882030178326474,
-      "eval_overall_f1": 0.9427860696517413,
-      "eval_overall_precision": 0.9404466501240695,
-      "eval_overall_recall": 0.9451371571072319,
-      "eval_runtime": 0.6315,
-      "eval_samples_per_second": 269.203,
-      "eval_steps_per_second": 4.751,
+      "eval_PERSON_precision": 0.9857142857142858,
+      "eval_PERSON_recall": 0.971830985915493,
+      "eval_loss": 0.06909173727035522,
+      "eval_overall_accuracy": 0.9871056241426612,
+      "eval_overall_f1": 0.9344870210135972,
+      "eval_overall_precision": 0.9264705882352942,
+      "eval_overall_recall": 0.942643391521197,
+      "eval_runtime": 0.2887,
+      "eval_samples_per_second": 588.812,
+      "eval_steps_per_second": 10.391,
       "step": 7776
     },
     {
       "epoch": 82.0,
-      "grad_norm": 0.3362814784049988,
+      "grad_norm": 0.21058472990989685,
       "learning_rate": 9e-06,
-      "loss": 0.0089,
+      "loss": 0.0081,
       "step": 7872
     },
     {
       "epoch": 82.0,
-      "eval_LOCATION_f1": 0.9319371727748691,
+      "eval_LOCATION_f1": 0.9166666666666667,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.9081632653061225,
-      "eval_LOCATION_recall": 0.956989247311828,
-      "eval_ORGANIZATION_f1": 0.9151515151515153,
+      "eval_LOCATION_precision": 0.8888888888888888,
+      "eval_LOCATION_recall": 0.946236559139785,
+      "eval_ORGANIZATION_f1": 0.906906906906907,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9207317073170732,
+      "eval_ORGANIZATION_precision": 0.9041916167664671,
       "eval_ORGANIZATION_recall": 0.9096385542168675,
-      "eval_PERSON_f1": 0.9823321554770318,
+      "eval_PERSON_f1": 0.9787234042553192,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9858156028368794,
-      "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.059507716447114944,
-      "eval_overall_accuracy": 0.9890260631001372,
-      "eval_overall_f1": 0.9427860696517413,
-      "eval_overall_precision": 0.9404466501240695,
-      "eval_overall_recall": 0.9451371571072319,
-      "eval_runtime": 0.6408,
-      "eval_samples_per_second": 265.312,
-      "eval_steps_per_second": 4.682,
+      "eval_PERSON_precision": 0.9857142857142858,
+      "eval_PERSON_recall": 0.971830985915493,
+      "eval_loss": 0.07346898317337036,
+      "eval_overall_accuracy": 0.9860082304526749,
+      "eval_overall_f1": 0.9343246592317225,
+      "eval_overall_precision": 0.9285714285714286,
+      "eval_overall_recall": 0.940149625935162,
+      "eval_runtime": 0.2908,
+      "eval_samples_per_second": 584.583,
+      "eval_steps_per_second": 10.316,
       "step": 7872
     },
     {
       "epoch": 83.0,
-      "grad_norm": 0.15785475075244904,
+      "grad_norm": 0.9036238789558411,
       "learning_rate": 8.500000000000002e-06,
-      "loss": 0.0089,
+      "loss": 0.0099,
       "step": 7968
     },
     {
       "epoch": 83.0,
-      "eval_LOCATION_f1": 0.946808510638298,
+      "eval_LOCATION_f1": 0.9214659685863874,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.9368421052631579,
-      "eval_LOCATION_recall": 0.956989247311828,
-      "eval_ORGANIZATION_f1": 0.924924924924925,
+      "eval_LOCATION_precision": 0.8979591836734694,
+      "eval_LOCATION_recall": 0.946236559139785,
+      "eval_ORGANIZATION_f1": 0.9189189189189191,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9221556886227545,
-      "eval_ORGANIZATION_recall": 0.927710843373494,
-      "eval_PERSON_f1": 0.9823321554770318,
+      "eval_ORGANIZATION_precision": 0.9161676646706587,
+      "eval_ORGANIZATION_recall": 0.9216867469879518,
+      "eval_PERSON_f1": 0.9752650176678446,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9858156028368794,
-      "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.05860907584428787,
-      "eval_overall_accuracy": 0.9890260631001372,
-      "eval_overall_f1": 0.9502487562189055,
-      "eval_overall_precision": 0.9478908188585607,
-      "eval_overall_recall": 0.9526184538653366,
-      "eval_runtime": 0.6371,
-      "eval_samples_per_second": 266.844,
-      "eval_steps_per_second": 4.709,
+      "eval_PERSON_precision": 0.9787234042553191,
+      "eval_PERSON_recall": 0.971830985915493,
+      "eval_loss": 0.06314818561077118,
+      "eval_overall_accuracy": 0.9882030178326474,
+      "eval_overall_f1": 0.9392812887236679,
+      "eval_overall_precision": 0.9334975369458128,
+      "eval_overall_recall": 0.9451371571072319,
+      "eval_runtime": 0.2899,
+      "eval_samples_per_second": 586.334,
+      "eval_steps_per_second": 10.347,
       "step": 7968
     },
     {
       "epoch": 84.0,
-      "grad_norm": 0.4708460569381714,
+      "grad_norm": 0.02120956964790821,
       "learning_rate": 8.000000000000001e-06,
-      "loss": 0.0084,
+      "loss": 0.0085,
       "step": 8064
     },
     {
       "epoch": 84.0,
-      "eval_LOCATION_f1": 0.9518716577540107,
+      "eval_LOCATION_f1": 0.9166666666666667,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.9468085106382979,
-      "eval_LOCATION_recall": 0.956989247311828,
-      "eval_ORGANIZATION_f1": 0.9244712990936556,
+      "eval_LOCATION_precision": 0.8888888888888888,
+      "eval_LOCATION_recall": 0.946236559139785,
+      "eval_ORGANIZATION_f1": 0.918429003021148,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9272727272727272,
-      "eval_ORGANIZATION_recall": 0.9216867469879518,
-      "eval_PERSON_f1": 0.9823321554770318,
+      "eval_ORGANIZATION_precision": 0.9212121212121213,
+      "eval_ORGANIZATION_recall": 0.9156626506024096,
+      "eval_PERSON_f1": 0.9787234042553192,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9858156028368794,
-      "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.05669812113046646,
-      "eval_overall_accuracy": 0.9909465020576131,
-      "eval_overall_f1": 0.951310861423221,
-      "eval_overall_precision": 0.9525,
-      "eval_overall_recall": 0.9501246882793017,
-      "eval_runtime": 0.6365,
-      "eval_samples_per_second": 267.075,
-      "eval_steps_per_second": 4.713,
+      "eval_PERSON_precision": 0.9857142857142858,
+      "eval_PERSON_recall": 0.971830985915493,
+      "eval_loss": 0.06590625643730164,
+      "eval_overall_accuracy": 0.9876543209876543,
+      "eval_overall_f1": 0.9391304347826087,
+      "eval_overall_precision": 0.9356435643564357,
+      "eval_overall_recall": 0.942643391521197,
+      "eval_runtime": 0.2882,
+      "eval_samples_per_second": 589.934,
+      "eval_steps_per_second": 10.411,
       "step": 8064
     },
     {
       "epoch": 85.0,
-      "grad_norm": 0.6606996059417725,
+      "grad_norm": 0.7899487018585205,
       "learning_rate": 7.5e-06,
-      "loss": 0.0088,
+      "loss": 0.0092,
       "step": 8160
     },
     {
       "epoch": 85.0,
-      "eval_LOCATION_f1": 0.9319371727748691,
+      "eval_LOCATION_f1": 0.922279792746114,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.9081632653061225,
+      "eval_LOCATION_precision": 0.89,
       "eval_LOCATION_recall": 0.956989247311828,
-      "eval_ORGANIZATION_f1": 0.918429003021148,
+      "eval_ORGANIZATION_f1": 0.9216867469879518,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9212121212121213,
-      "eval_ORGANIZATION_recall": 0.9156626506024096,
-      "eval_PERSON_f1": 0.9823321554770318,
+      "eval_ORGANIZATION_precision": 0.9216867469879518,
+      "eval_ORGANIZATION_recall": 0.9216867469879518,
+      "eval_PERSON_f1": 0.9752650176678446,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9858156028368794,
-      "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.06166466698050499,
-      "eval_overall_accuracy": 0.9890260631001372,
-      "eval_overall_f1": 0.9440993788819876,
-      "eval_overall_precision": 0.9405940594059405,
+      "eval_PERSON_precision": 0.9787234042553191,
+      "eval_PERSON_recall": 0.971830985915493,
+      "eval_loss": 0.06632930040359497,
+      "eval_overall_accuracy": 0.9873799725651577,
+      "eval_overall_f1": 0.9405940594059405,
+      "eval_overall_precision": 0.9336609336609336,
       "eval_overall_recall": 0.9476309226932669,
-      "eval_runtime": 0.6881,
-      "eval_samples_per_second": 247.07,
-      "eval_steps_per_second": 4.36,
+      "eval_runtime": 0.2876,
+      "eval_samples_per_second": 591.174,
+      "eval_steps_per_second": 10.432,
       "step": 8160
     },
     {
       "epoch": 86.0,
-      "grad_norm": 0.701899528503418,
+      "grad_norm": 1.0977509021759033,
       "learning_rate": 7.000000000000001e-06,
-      "loss": 0.0089,
+      "loss": 0.0076,
       "step": 8256
     },
     {
       "epoch": 86.0,
-      "eval_LOCATION_f1": 0.9361702127659575,
+      "eval_LOCATION_f1": 0.9166666666666667,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.9263157894736842,
+      "eval_LOCATION_precision": 0.8888888888888888,
       "eval_LOCATION_recall": 0.946236559139785,
-      "eval_ORGANIZATION_f1": 0.9212121212121213,
+      "eval_ORGANIZATION_f1": 0.906906906906907,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.926829268292683,
-      "eval_ORGANIZATION_recall": 0.9156626506024096,
-      "eval_PERSON_f1": 0.9823321554770318,
+      "eval_ORGANIZATION_precision": 0.9041916167664671,
+      "eval_ORGANIZATION_recall": 0.9096385542168675,
+      "eval_PERSON_f1": 0.9787234042553192,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9858156028368794,
-      "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.05820832401514053,
-      "eval_overall_accuracy": 0.9895747599451303,
-      "eval_overall_f1": 0.9463171036204744,
-      "eval_overall_precision": 0.9475,
-      "eval_overall_recall": 0.9451371571072319,
-      "eval_runtime": 0.6343,
-      "eval_samples_per_second": 268.032,
-      "eval_steps_per_second": 4.73,
+      "eval_PERSON_precision": 0.9857142857142858,
+      "eval_PERSON_recall": 0.971830985915493,
+      "eval_loss": 0.06624973565340042,
+      "eval_overall_accuracy": 0.9873799725651577,
+      "eval_overall_f1": 0.9343246592317225,
+      "eval_overall_precision": 0.9285714285714286,
+      "eval_overall_recall": 0.940149625935162,
+      "eval_runtime": 0.2903,
+      "eval_samples_per_second": 585.654,
+      "eval_steps_per_second": 10.335,
       "step": 8256
     },
     {
       "epoch": 87.0,
-      "grad_norm": 0.12563535571098328,
+      "grad_norm": 0.06188450753688812,
       "learning_rate": 6.5000000000000004e-06,
-      "loss": 0.0066,
+      "loss": 0.0073,
       "step": 8352
     },
     {
       "epoch": 87.0,
-      "eval_LOCATION_f1": 0.946808510638298,
+      "eval_LOCATION_f1": 0.9326424870466321,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.9368421052631579,
-      "eval_LOCATION_recall": 0.956989247311828,
-      "eval_ORGANIZATION_f1": 0.927710843373494,
+      "eval_LOCATION_precision": 0.9,
+      "eval_LOCATION_recall": 0.967741935483871,
+      "eval_ORGANIZATION_f1": 0.918429003021148,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.927710843373494,
-      "eval_ORGANIZATION_recall": 0.927710843373494,
-      "eval_PERSON_f1": 0.9823321554770318,
+      "eval_ORGANIZATION_precision": 0.9212121212121213,
+      "eval_ORGANIZATION_recall": 0.9156626506024096,
+      "eval_PERSON_f1": 0.9787234042553192,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9858156028368794,
-      "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.061063703149557114,
-      "eval_overall_accuracy": 0.9893004115226337,
-      "eval_overall_f1": 0.9514321295143213,
-      "eval_overall_precision": 0.9502487562189055,
-      "eval_overall_recall": 0.9526184538653366,
-      "eval_runtime": 0.6323,
-      "eval_samples_per_second": 268.857,
-      "eval_steps_per_second": 4.745,
+      "eval_PERSON_precision": 0.9857142857142858,
+      "eval_PERSON_recall": 0.971830985915493,
+      "eval_loss": 0.06855077296495438,
+      "eval_overall_accuracy": 0.9873799725651577,
+      "eval_overall_f1": 0.9429280397022332,
+      "eval_overall_precision": 0.9382716049382716,
+      "eval_overall_recall": 0.9476309226932669,
+      "eval_runtime": 0.2878,
+      "eval_samples_per_second": 590.656,
+      "eval_steps_per_second": 10.423,
       "step": 8352
     },
     {
       "epoch": 88.0,
-      "grad_norm": 0.18005654215812683,
+      "grad_norm": 0.21970857679843903,
       "learning_rate": 6e-06,
-      "loss": 0.0088,
+      "loss": 0.0068,
       "step": 8448
     },
     {
       "epoch": 88.0,
-      "eval_LOCATION_f1": 0.9368421052631579,
+      "eval_LOCATION_f1": 0.9166666666666667,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.9175257731958762,
-      "eval_LOCATION_recall": 0.956989247311828,
-      "eval_ORGANIZATION_f1": 0.9296636085626911,
+      "eval_LOCATION_precision": 0.8888888888888888,
+      "eval_LOCATION_recall": 0.946236559139785,
+      "eval_ORGANIZATION_f1": 0.9123867069486404,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9440993788819876,
-      "eval_ORGANIZATION_recall": 0.9156626506024096,
-      "eval_PERSON_f1": 0.9823321554770318,
+      "eval_ORGANIZATION_precision": 0.9151515151515152,
+      "eval_ORGANIZATION_recall": 0.9096385542168675,
+      "eval_PERSON_f1": 0.9787234042553192,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9858156028368794,
-      "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.058633558452129364,
-      "eval_overall_accuracy": 0.9893004115226337,
-      "eval_overall_f1": 0.95,
-      "eval_overall_precision": 0.9523809523809523,
-      "eval_overall_recall": 0.9476309226932669,
-      "eval_runtime": 0.6336,
-      "eval_samples_per_second": 268.325,
-      "eval_steps_per_second": 4.735,
+      "eval_PERSON_precision": 0.9857142857142858,
+      "eval_PERSON_recall": 0.971830985915493,
+      "eval_loss": 0.06893233209848404,
+      "eval_overall_accuracy": 0.9871056241426612,
+      "eval_overall_f1": 0.9366459627329192,
+      "eval_overall_precision": 0.9331683168316832,
+      "eval_overall_recall": 0.940149625935162,
+      "eval_runtime": 0.2916,
+      "eval_samples_per_second": 582.955,
+      "eval_steps_per_second": 10.287,
       "step": 8448
     },
     {
       "epoch": 89.0,
-      "grad_norm": 0.44412586092948914,
+      "grad_norm": 0.05854140594601631,
       "learning_rate": 5.500000000000001e-06,
-      "loss": 0.007,
+      "loss": 0.0076,
       "step": 8544
     },
     {
       "epoch": 89.0,
-      "eval_LOCATION_f1": 0.9417989417989417,
+      "eval_LOCATION_f1": 0.9166666666666667,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.9270833333333334,
-      "eval_LOCATION_recall": 0.956989247311828,
-      "eval_ORGANIZATION_f1": 0.9151515151515153,
+      "eval_LOCATION_precision": 0.8888888888888888,
+      "eval_LOCATION_recall": 0.946236559139785,
+      "eval_ORGANIZATION_f1": 0.9156626506024096,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9207317073170732,
-      "eval_ORGANIZATION_recall": 0.9096385542168675,
-      "eval_PERSON_f1": 0.9823321554770318,
+      "eval_ORGANIZATION_precision": 0.9156626506024096,
+      "eval_ORGANIZATION_recall": 0.9156626506024096,
+      "eval_PERSON_f1": 0.971830985915493,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9858156028368794,
-      "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.06018080934882164,
-      "eval_overall_accuracy": 0.9893004115226337,
-      "eval_overall_f1": 0.9451371571072319,
-      "eval_overall_precision": 0.9451371571072319,
-      "eval_overall_recall": 0.9451371571072319,
-      "eval_runtime": 0.6276,
-      "eval_samples_per_second": 270.894,
-      "eval_steps_per_second": 4.78,
+      "eval_PERSON_precision": 0.971830985915493,
+      "eval_PERSON_recall": 0.971830985915493,
+      "eval_loss": 0.06832458078861237,
+      "eval_overall_accuracy": 0.9868312757201646,
+      "eval_overall_f1": 0.9356435643564357,
+      "eval_overall_precision": 0.9287469287469288,
+      "eval_overall_recall": 0.942643391521197,
+      "eval_runtime": 0.2879,
+      "eval_samples_per_second": 590.424,
+      "eval_steps_per_second": 10.419,
       "step": 8544
     },
     {
       "epoch": 90.0,
-      "grad_norm": 0.5702086091041565,
+      "grad_norm": 0.20013177394866943,
       "learning_rate": 5e-06,
-      "loss": 0.0083,
+      "loss": 0.0071,
       "step": 8640
     },
     {
       "epoch": 90.0,
-      "eval_LOCATION_f1": 0.9518716577540107,
+      "eval_LOCATION_f1": 0.9270833333333334,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.9468085106382979,
+      "eval_LOCATION_precision": 0.898989898989899,
       "eval_LOCATION_recall": 0.956989247311828,
-      "eval_ORGANIZATION_f1": 0.9305135951661632,
+      "eval_ORGANIZATION_f1": 0.9189189189189191,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9333333333333333,
-      "eval_ORGANIZATION_recall": 0.927710843373494,
-      "eval_PERSON_f1": 0.9823321554770318,
+      "eval_ORGANIZATION_precision": 0.9161676646706587,
+      "eval_ORGANIZATION_recall": 0.9216867469879518,
+      "eval_PERSON_f1": 0.9787234042553192,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9858156028368794,
-      "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.057951029390096664,
-      "eval_overall_accuracy": 0.9901234567901235,
-      "eval_overall_f1": 0.9538077403245941,
-      "eval_overall_precision": 0.955,
-      "eval_overall_recall": 0.9526184538653366,
-      "eval_runtime": 0.6268,
-      "eval_samples_per_second": 271.204,
-      "eval_steps_per_second": 4.786,
+      "eval_PERSON_precision": 0.9857142857142858,
+      "eval_PERSON_recall": 0.971830985915493,
+      "eval_loss": 0.06415162235498428,
+      "eval_overall_accuracy": 0.9882030178326474,
+      "eval_overall_f1": 0.9417596034696407,
+      "eval_overall_precision": 0.9359605911330049,
+      "eval_overall_recall": 0.9476309226932669,
+      "eval_runtime": 0.2886,
+      "eval_samples_per_second": 589.092,
+      "eval_steps_per_second": 10.396,
       "step": 8640
     },
     {
       "epoch": 91.0,
-      "grad_norm": 0.6796770691871643,
+      "grad_norm": 0.0871826782822609,
       "learning_rate": 4.5e-06,
-      "loss": 0.0077,
+      "loss": 0.0069,
       "step": 8736
     },
     {
       "epoch": 91.0,
-      "eval_LOCATION_f1": 0.946808510638298,
+      "eval_LOCATION_f1": 0.9166666666666667,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.9368421052631579,
-      "eval_LOCATION_recall": 0.956989247311828,
-      "eval_ORGANIZATION_f1": 0.9161676646706587,
+      "eval_LOCATION_precision": 0.8888888888888888,
+      "eval_LOCATION_recall": 0.946236559139785,
+      "eval_ORGANIZATION_f1": 0.9101796407185628,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9107142857142857,
-      "eval_ORGANIZATION_recall": 0.9216867469879518,
-      "eval_PERSON_f1": 0.9823321554770318,
+      "eval_ORGANIZATION_precision": 0.9047619047619048,
+      "eval_ORGANIZATION_recall": 0.9156626506024096,
+      "eval_PERSON_f1": 0.9787234042553192,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9858156028368794,
-      "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.059122197329998016,
-      "eval_overall_accuracy": 0.9890260631001372,
-      "eval_overall_f1": 0.946583850931677,
-      "eval_overall_precision": 0.943069306930693,
-      "eval_overall_recall": 0.9501246882793017,
-      "eval_runtime": 0.6297,
-      "eval_samples_per_second": 269.99,
-      "eval_steps_per_second": 4.765,
+      "eval_PERSON_precision": 0.9857142857142858,
+      "eval_PERSON_recall": 0.971830985915493,
+      "eval_loss": 0.07019613683223724,
+      "eval_overall_accuracy": 0.9871056241426612,
+      "eval_overall_f1": 0.9356435643564357,
+      "eval_overall_precision": 0.9287469287469288,
+      "eval_overall_recall": 0.942643391521197,
+      "eval_runtime": 0.2912,
+      "eval_samples_per_second": 583.754,
+      "eval_steps_per_second": 10.302,
       "step": 8736
     },
     {
       "epoch": 92.0,
-      "grad_norm": 0.009496918879449368,
+      "grad_norm": 0.16851916909217834,
       "learning_rate": 4.000000000000001e-06,
-      "loss": 0.0078,
+      "loss": 0.0073,
       "step": 8832
     },
     {
       "epoch": 92.0,
-      "eval_LOCATION_f1": 0.9417989417989417,
+      "eval_LOCATION_f1": 0.9166666666666667,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.9270833333333334,
-      "eval_LOCATION_recall": 0.956989247311828,
-      "eval_ORGANIZATION_f1": 0.9189189189189191,
+      "eval_LOCATION_precision": 0.8888888888888888,
+      "eval_LOCATION_recall": 0.946236559139785,
+      "eval_ORGANIZATION_f1": 0.9156626506024096,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9161676646706587,
-      "eval_ORGANIZATION_recall": 0.9216867469879518,
-      "eval_PERSON_f1": 0.9823321554770318,
+      "eval_ORGANIZATION_precision": 0.9156626506024096,
+      "eval_ORGANIZATION_recall": 0.9156626506024096,
+      "eval_PERSON_f1": 0.9752650176678446,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9858156028368794,
-      "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.06082513928413391,
+      "eval_PERSON_precision": 0.9787234042553191,
+      "eval_PERSON_recall": 0.971830985915493,
+      "eval_loss": 0.06724745780229568,
       "eval_overall_accuracy": 0.9879286694101509,
-      "eval_overall_f1": 0.946583850931677,
-      "eval_overall_precision": 0.943069306930693,
-      "eval_overall_recall": 0.9501246882793017,
-      "eval_runtime": 0.6215,
-      "eval_samples_per_second": 273.552,
-      "eval_steps_per_second": 4.827,
+      "eval_overall_f1": 0.9368029739776952,
+      "eval_overall_precision": 0.9310344827586207,
+      "eval_overall_recall": 0.942643391521197,
+      "eval_runtime": 0.2954,
+      "eval_samples_per_second": 575.515,
+      "eval_steps_per_second": 10.156,
       "step": 8832
     },
     {
       "epoch": 93.0,
-      "grad_norm": 0.6363722085952759,
+      "grad_norm": 0.19750891625881195,
       "learning_rate": 3.5000000000000004e-06,
-      "loss": 0.0068,
+      "loss": 0.007,
       "step": 8928
     },
     {
       "epoch": 93.0,
-      "eval_LOCATION_f1": 0.9518716577540107,
+      "eval_LOCATION_f1": 0.9166666666666667,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.9468085106382979,
-      "eval_LOCATION_recall": 0.956989247311828,
-      "eval_ORGANIZATION_f1": 0.9101796407185628,
+      "eval_LOCATION_precision": 0.8888888888888888,
+      "eval_LOCATION_recall": 0.946236559139785,
+      "eval_ORGANIZATION_f1": 0.9156626506024096,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9047619047619048,
+      "eval_ORGANIZATION_precision": 0.9156626506024096,
       "eval_ORGANIZATION_recall": 0.9156626506024096,
-      "eval_PERSON_f1": 0.9823321554770318,
+      "eval_PERSON_f1": 0.9752650176678446,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9858156028368794,
-      "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.05810477212071419,
-      "eval_overall_accuracy": 0.9890260631001372,
-      "eval_overall_f1": 0.945273631840796,
-      "eval_overall_precision": 0.9429280397022333,
-      "eval_overall_recall": 0.9476309226932669,
-      "eval_runtime": 0.6304,
-      "eval_samples_per_second": 269.655,
-      "eval_steps_per_second": 4.759,
+      "eval_PERSON_precision": 0.9787234042553191,
+      "eval_PERSON_recall": 0.971830985915493,
+      "eval_loss": 0.06532972306013107,
+      "eval_overall_accuracy": 0.9879286694101509,
+      "eval_overall_f1": 0.9368029739776952,
+      "eval_overall_precision": 0.9310344827586207,
+      "eval_overall_recall": 0.942643391521197,
+      "eval_runtime": 0.2896,
+      "eval_samples_per_second": 587.055,
+      "eval_steps_per_second": 10.36,
       "step": 8928
     },
     {
       "epoch": 94.0,
-      "grad_norm": 0.42099490761756897,
+      "grad_norm": 0.07180120795965195,
       "learning_rate": 3e-06,
-      "loss": 0.0079,
+      "loss": 0.007,
       "step": 9024
     },
     {
       "epoch": 94.0,
-      "eval_LOCATION_f1": 0.946808510638298,
+      "eval_LOCATION_f1": 0.9166666666666667,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.9368421052631579,
-      "eval_LOCATION_recall": 0.956989247311828,
-      "eval_ORGANIZATION_f1": 0.9161676646706587,
+      "eval_LOCATION_precision": 0.8888888888888888,
+      "eval_LOCATION_recall": 0.946236559139785,
+      "eval_ORGANIZATION_f1": 0.9156626506024096,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9107142857142857,
-      "eval_ORGANIZATION_recall": 0.9216867469879518,
+      "eval_ORGANIZATION_precision": 0.9156626506024096,
+      "eval_ORGANIZATION_recall": 0.9156626506024096,
       "eval_PERSON_f1": 0.9823321554770318,
       "eval_PERSON_number": 142,
       "eval_PERSON_precision": 0.9858156028368794,
       "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.0605609156191349,
-      "eval_overall_accuracy": 0.988477366255144,
-      "eval_overall_f1": 0.946583850931677,
-      "eval_overall_precision": 0.943069306930693,
-      "eval_overall_recall": 0.9501246882793017,
-      "eval_runtime": 0.6284,
-      "eval_samples_per_second": 270.534,
-      "eval_steps_per_second": 4.774,
+      "eval_loss": 0.06773727387189865,
+      "eval_overall_accuracy": 0.9873799725651577,
+      "eval_overall_f1": 0.9392812887236679,
+      "eval_overall_precision": 0.9334975369458128,
+      "eval_overall_recall": 0.9451371571072319,
+      "eval_runtime": 0.288,
+      "eval_samples_per_second": 590.323,
+      "eval_steps_per_second": 10.417,
       "step": 9024
     },
     {
       "epoch": 95.0,
-      "grad_norm": 0.0327790230512619,
+      "grad_norm": 0.31866025924682617,
       "learning_rate": 2.5e-06,
-      "loss": 0.0072,
+      "loss": 0.0077,
       "step": 9120
     },
     {
       "epoch": 95.0,
-      "eval_LOCATION_f1": 0.9417989417989417,
+      "eval_LOCATION_f1": 0.9166666666666667,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.9270833333333334,
-      "eval_LOCATION_recall": 0.956989247311828,
-      "eval_ORGANIZATION_f1": 0.9189189189189191,
+      "eval_LOCATION_precision": 0.8888888888888888,
+      "eval_LOCATION_recall": 0.946236559139785,
+      "eval_ORGANIZATION_f1": 0.9129129129129129,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9161676646706587,
-      "eval_ORGANIZATION_recall": 0.9216867469879518,
-      "eval_PERSON_f1": 0.9823321554770318,
+      "eval_ORGANIZATION_precision": 0.9101796407185628,
+      "eval_ORGANIZATION_recall": 0.9156626506024096,
+      "eval_PERSON_f1": 0.971830985915493,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9858156028368794,
-      "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.06001435965299606,
-      "eval_overall_accuracy": 0.9895747599451303,
-      "eval_overall_f1": 0.946583850931677,
-      "eval_overall_precision": 0.943069306930693,
-      "eval_overall_recall": 0.9501246882793017,
-      "eval_runtime": 0.6347,
-      "eval_samples_per_second": 267.858,
-      "eval_steps_per_second": 4.727,
+      "eval_PERSON_precision": 0.971830985915493,
+      "eval_PERSON_recall": 0.971830985915493,
+      "eval_loss": 0.0693470910191536,
+      "eval_overall_accuracy": 0.9868312757201646,
+      "eval_overall_f1": 0.9344870210135972,
+      "eval_overall_precision": 0.9264705882352942,
+      "eval_overall_recall": 0.942643391521197,
+      "eval_runtime": 0.3012,
+      "eval_samples_per_second": 564.461,
+      "eval_steps_per_second": 9.961,
       "step": 9120
     },
     {
       "epoch": 96.0,
-      "grad_norm": 0.06773529201745987,
+      "grad_norm": 0.8485398292541504,
       "learning_rate": 2.0000000000000003e-06,
-      "loss": 0.007,
+      "loss": 0.0071,
       "step": 9216
     },
     {
       "epoch": 96.0,
-      "eval_LOCATION_f1": 0.9368421052631579,
+      "eval_LOCATION_f1": 0.9166666666666667,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.9175257731958762,
-      "eval_LOCATION_recall": 0.956989247311828,
-      "eval_ORGANIZATION_f1": 0.9161676646706587,
+      "eval_LOCATION_precision": 0.8888888888888888,
+      "eval_LOCATION_recall": 0.946236559139785,
+      "eval_ORGANIZATION_f1": 0.9101796407185628,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9107142857142857,
-      "eval_ORGANIZATION_recall": 0.9216867469879518,
-      "eval_PERSON_f1": 0.9823321554770318,
+      "eval_ORGANIZATION_precision": 0.9047619047619048,
+      "eval_ORGANIZATION_recall": 0.9156626506024096,
+      "eval_PERSON_f1": 0.971830985915493,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9858156028368794,
-      "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.05993801727890968,
-      "eval_overall_accuracy": 0.988477366255144,
-      "eval_overall_f1": 0.9442379182156134,
-      "eval_overall_precision": 0.9384236453201971,
-      "eval_overall_recall": 0.9501246882793017,
-      "eval_runtime": 0.6261,
-      "eval_samples_per_second": 271.504,
-      "eval_steps_per_second": 4.791,
+      "eval_PERSON_precision": 0.971830985915493,
+      "eval_PERSON_recall": 0.971830985915493,
+      "eval_loss": 0.07040981203317642,
+      "eval_overall_accuracy": 0.9871056241426612,
+      "eval_overall_f1": 0.9333333333333333,
+      "eval_overall_precision": 0.9242053789731052,
+      "eval_overall_recall": 0.942643391521197,
+      "eval_runtime": 0.2916,
+      "eval_samples_per_second": 583.084,
+      "eval_steps_per_second": 10.29,
       "step": 9216
     },
     {
       "epoch": 97.0,
-      "grad_norm": 0.022277064621448517,
+      "grad_norm": 0.03155836462974548,
       "learning_rate": 1.5e-06,
-      "loss": 0.0082,
+      "loss": 0.007,
       "step": 9312
     },
     {
       "epoch": 97.0,
-      "eval_LOCATION_f1": 0.9368421052631579,
+      "eval_LOCATION_f1": 0.9166666666666667,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.9175257731958762,
-      "eval_LOCATION_recall": 0.956989247311828,
-      "eval_ORGANIZATION_f1": 0.9216867469879518,
+      "eval_LOCATION_precision": 0.8888888888888888,
+      "eval_LOCATION_recall": 0.946236559139785,
+      "eval_ORGANIZATION_f1": 0.9129129129129129,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9216867469879518,
-      "eval_ORGANIZATION_recall": 0.9216867469879518,
-      "eval_PERSON_f1": 0.9823321554770318,
+      "eval_ORGANIZATION_precision": 0.9101796407185628,
+      "eval_ORGANIZATION_recall": 0.9156626506024096,
+      "eval_PERSON_f1": 0.9752650176678446,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9858156028368794,
-      "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.060069117695093155,
-      "eval_overall_accuracy": 0.988477366255144,
-      "eval_overall_f1": 0.946583850931677,
-      "eval_overall_precision": 0.943069306930693,
-      "eval_overall_recall": 0.9501246882793017,
-      "eval_runtime": 0.6293,
-      "eval_samples_per_second": 270.143,
-      "eval_steps_per_second": 4.767,
+      "eval_PERSON_precision": 0.9787234042553191,
+      "eval_PERSON_recall": 0.971830985915493,
+      "eval_loss": 0.069269560277462,
+      "eval_overall_accuracy": 0.9876543209876543,
+      "eval_overall_f1": 0.9356435643564357,
+      "eval_overall_precision": 0.9287469287469288,
+      "eval_overall_recall": 0.942643391521197,
+      "eval_runtime": 0.2886,
+      "eval_samples_per_second": 589.08,
+      "eval_steps_per_second": 10.396,
       "step": 9312
     },
     {
       "epoch": 98.0,
-      "grad_norm": 0.3417333662509918,
+      "grad_norm": 0.1880914568901062,
       "learning_rate": 1.0000000000000002e-06,
-      "loss": 0.0065,
+      "loss": 0.0062,
       "step": 9408
     },
     {
       "epoch": 98.0,
-      "eval_LOCATION_f1": 0.9368421052631579,
+      "eval_LOCATION_f1": 0.9166666666666667,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.9175257731958762,
-      "eval_LOCATION_recall": 0.956989247311828,
-      "eval_ORGANIZATION_f1": 0.9216867469879518,
+      "eval_LOCATION_precision": 0.8888888888888888,
+      "eval_LOCATION_recall": 0.946236559139785,
+      "eval_ORGANIZATION_f1": 0.9156626506024096,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9216867469879518,
-      "eval_ORGANIZATION_recall": 0.9216867469879518,
-      "eval_PERSON_f1": 0.9823321554770318,
+      "eval_ORGANIZATION_precision": 0.9156626506024096,
+      "eval_ORGANIZATION_recall": 0.9156626506024096,
+      "eval_PERSON_f1": 0.9752650176678446,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9858156028368794,
-      "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.06064934283494949,
-      "eval_overall_accuracy": 0.988477366255144,
-      "eval_overall_f1": 0.946583850931677,
-      "eval_overall_precision": 0.943069306930693,
-      "eval_overall_recall": 0.9501246882793017,
-      "eval_runtime": 0.6301,
-      "eval_samples_per_second": 269.806,
-      "eval_steps_per_second": 4.761,
+      "eval_PERSON_precision": 0.9787234042553191,
+      "eval_PERSON_recall": 0.971830985915493,
+      "eval_loss": 0.06927915662527084,
+      "eval_overall_accuracy": 0.9873799725651577,
+      "eval_overall_f1": 0.9368029739776952,
+      "eval_overall_precision": 0.9310344827586207,
+      "eval_overall_recall": 0.942643391521197,
+      "eval_runtime": 0.2886,
+      "eval_samples_per_second": 589.05,
+      "eval_steps_per_second": 10.395,
       "step": 9408
     },
     {
       "epoch": 99.0,
-      "grad_norm": 0.2305583655834198,
+      "grad_norm": 0.29024115204811096,
       "learning_rate": 5.000000000000001e-07,
-      "loss": 0.0075,
+      "loss": 0.0062,
       "step": 9504
     },
     {
       "epoch": 99.0,
-      "eval_LOCATION_f1": 0.9368421052631579,
+      "eval_LOCATION_f1": 0.9166666666666667,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.9175257731958762,
-      "eval_LOCATION_recall": 0.956989247311828,
-      "eval_ORGANIZATION_f1": 0.9189189189189191,
+      "eval_LOCATION_precision": 0.8888888888888888,
+      "eval_LOCATION_recall": 0.946236559139785,
+      "eval_ORGANIZATION_f1": 0.9156626506024096,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9161676646706587,
-      "eval_ORGANIZATION_recall": 0.9216867469879518,
-      "eval_PERSON_f1": 0.9823321554770318,
+      "eval_ORGANIZATION_precision": 0.9156626506024096,
+      "eval_ORGANIZATION_recall": 0.9156626506024096,
+      "eval_PERSON_f1": 0.9752650176678446,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9858156028368794,
-      "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.059974148869514465,
-      "eval_overall_accuracy": 0.9882030178326474,
-      "eval_overall_f1": 0.9454094292803972,
-      "eval_overall_precision": 0.9407407407407408,
-      "eval_overall_recall": 0.9501246882793017,
-      "eval_runtime": 0.6422,
-      "eval_samples_per_second": 264.724,
-      "eval_steps_per_second": 4.672,
+      "eval_PERSON_precision": 0.9787234042553191,
+      "eval_PERSON_recall": 0.971830985915493,
+      "eval_loss": 0.06901175528764725,
+      "eval_overall_accuracy": 0.9873799725651577,
+      "eval_overall_f1": 0.9368029739776952,
+      "eval_overall_precision": 0.9310344827586207,
+      "eval_overall_recall": 0.942643391521197,
+      "eval_runtime": 0.2901,
+      "eval_samples_per_second": 585.92,
+      "eval_steps_per_second": 10.34,
       "step": 9504
     },
     {
       "epoch": 100.0,
-      "grad_norm": 0.5298776030540466,
+      "grad_norm": 0.5472394824028015,
       "learning_rate": 0.0,
-      "loss": 0.0057,
+      "loss": 0.0068,
       "step": 9600
     },
     {
       "epoch": 100.0,
-      "eval_LOCATION_f1": 0.9368421052631579,
+      "eval_LOCATION_f1": 0.9166666666666667,
       "eval_LOCATION_number": 93,
-      "eval_LOCATION_precision": 0.9175257731958762,
-      "eval_LOCATION_recall": 0.956989247311828,
-      "eval_ORGANIZATION_f1": 0.9189189189189191,
+      "eval_LOCATION_precision": 0.8888888888888888,
+      "eval_LOCATION_recall": 0.946236559139785,
+      "eval_ORGANIZATION_f1": 0.9156626506024096,
       "eval_ORGANIZATION_number": 166,
-      "eval_ORGANIZATION_precision": 0.9161676646706587,
-      "eval_ORGANIZATION_recall": 0.9216867469879518,
-      "eval_PERSON_f1": 0.9823321554770318,
+      "eval_ORGANIZATION_precision": 0.9156626506024096,
+      "eval_ORGANIZATION_recall": 0.9156626506024096,
+      "eval_PERSON_f1": 0.9752650176678446,
       "eval_PERSON_number": 142,
-      "eval_PERSON_precision": 0.9858156028368794,
-      "eval_PERSON_recall": 0.9788732394366197,
-      "eval_loss": 0.05988968536257744,
-      "eval_overall_accuracy": 0.9882030178326474,
-      "eval_overall_f1": 0.9454094292803972,
-      "eval_overall_precision": 0.9407407407407408,
-      "eval_overall_recall": 0.9501246882793017,
-      "eval_runtime": 0.6423,
-      "eval_samples_per_second": 264.661,
-      "eval_steps_per_second": 4.67,
+      "eval_PERSON_precision": 0.9787234042553191,
+      "eval_PERSON_recall": 0.971830985915493,
+      "eval_loss": 0.06890885531902313,
+      "eval_overall_accuracy": 0.9876543209876543,
+      "eval_overall_f1": 0.9368029739776952,
+      "eval_overall_precision": 0.9310344827586207,
+      "eval_overall_recall": 0.942643391521197,
+      "eval_runtime": 0.2901,
+      "eval_samples_per_second": 586.027,
+      "eval_steps_per_second": 10.342,
       "step": 9600
     },
     {
       "epoch": 100.0,
       "step": 9600,
       "total_flos": 4208466347875278.0,
-      "train_loss": 0.037790342985341945,
-      "train_runtime": 1071.0881,
-      "train_samples_per_second": 142.939,
-      "train_steps_per_second": 8.963
+      "train_loss": 0.03755435147012273,
+      "train_runtime": 554.4341,
+      "train_samples_per_second": 276.137,
+      "train_steps_per_second": 17.315
     }
   ],
   "logging_steps": 500,