diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -10,1659 +10,1659 @@
   "log_history": [
     {
       "epoch": 1.0,
-      "grad_norm": 4.555638790130615,
+      "grad_norm": 2.7315826416015625,
       "learning_rate": 4.9500000000000004e-05,
-      "loss": 0.3502,
+      "loss": 0.3629,
       "step": 106
     },
     {
       "epoch": 1.0,
-      "eval_LOCATION_f1": 0.751592356687898,
+      "eval_LOCATION_f1": 0.8,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.6941176470588235,
-      "eval_LOCATION_recall": 0.8194444444444444,
-      "eval_ORGANIZATION_f1": 0.7305389221556886,
+      "eval_LOCATION_precision": 0.7692307692307693,
+      "eval_LOCATION_recall": 0.8333333333333334,
+      "eval_ORGANIZATION_f1": 0.6941176470588235,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.6630434782608695,
-      "eval_ORGANIZATION_recall": 0.8133333333333334,
-      "eval_PERSON_f1": 0.8976897689768978,
+      "eval_ORGANIZATION_precision": 0.6210526315789474,
+      "eval_ORGANIZATION_recall": 0.7866666666666666,
+      "eval_PERSON_f1": 0.8888888888888888,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.85,
+      "eval_PERSON_precision": 0.8343558282208589,
       "eval_PERSON_recall": 0.951048951048951,
-      "eval_QUANTITY_f1": 0.5423728813559322,
+      "eval_QUANTITY_f1": 0.5396825396825397,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.4444444444444444,
-      "eval_QUANTITY_recall": 0.6956521739130435,
-      "eval_TIME_f1": 0.6779661016949152,
+      "eval_QUANTITY_precision": 0.425,
+      "eval_QUANTITY_recall": 0.7391304347826086,
+      "eval_TIME_f1": 0.721311475409836,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.6060606060606061,
-      "eval_TIME_recall": 0.7692307692307693,
-      "eval_loss": 0.16481448709964752,
-      "eval_overall_accuracy": 0.9522141606204654,
-      "eval_overall_f1": 0.7838926174496645,
-      "eval_overall_precision": 0.7192118226600985,
-      "eval_overall_recall": 0.8613569321533924,
-      "eval_runtime": 0.8792,
-      "eval_samples_per_second": 212.683,
-      "eval_steps_per_second": 3.412,
+      "eval_TIME_precision": 0.6285714285714286,
+      "eval_TIME_recall": 0.8461538461538461,
+      "eval_loss": 0.1598740667104721,
+      "eval_overall_accuracy": 0.9482111583687766,
+      "eval_overall_f1": 0.7839999999999999,
+      "eval_overall_precision": 0.7153284671532847,
+      "eval_overall_recall": 0.8672566371681416,
+      "eval_runtime": 0.3626,
+      "eval_samples_per_second": 515.737,
+      "eval_steps_per_second": 8.274,
       "step": 106
     },
     {
       "epoch": 2.0,
-      "grad_norm": 3.0741477012634277,
+      "grad_norm": 1.3484944105148315,
       "learning_rate": 4.9e-05,
-      "loss": 0.1178,
+      "loss": 0.128,
       "step": 212
     },
     {
       "epoch": 2.0,
-      "eval_LOCATION_f1": 0.8289473684210527,
+      "eval_LOCATION_f1": 0.7922077922077924,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7875,
-      "eval_LOCATION_recall": 0.875,
-      "eval_ORGANIZATION_f1": 0.7100000000000001,
+      "eval_LOCATION_precision": 0.7439024390243902,
+      "eval_LOCATION_recall": 0.8472222222222222,
+      "eval_ORGANIZATION_f1": 0.782122905027933,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.568,
-      "eval_ORGANIZATION_recall": 0.9466666666666667,
-      "eval_PERSON_f1": 0.9047619047619047,
+      "eval_ORGANIZATION_precision": 0.6730769230769231,
+      "eval_ORGANIZATION_recall": 0.9333333333333333,
+      "eval_PERSON_f1": 0.8993288590604027,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8807947019867549,
-      "eval_PERSON_recall": 0.9300699300699301,
-      "eval_QUANTITY_f1": 0.5614035087719297,
+      "eval_PERSON_precision": 0.864516129032258,
+      "eval_PERSON_recall": 0.9370629370629371,
+      "eval_QUANTITY_f1": 0.6545454545454547,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.47058823529411764,
-      "eval_QUANTITY_recall": 0.6956521739130435,
-      "eval_TIME_f1": 0.8461538461538461,
+      "eval_QUANTITY_precision": 0.5625,
+      "eval_QUANTITY_recall": 0.782608695652174,
+      "eval_TIME_f1": 0.8076923076923077,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8461538461538461,
-      "eval_TIME_recall": 0.8461538461538461,
-      "eval_loss": 0.14857080578804016,
-      "eval_overall_accuracy": 0.9537152864648486,
-      "eval_overall_f1": 0.8079470198675497,
-      "eval_overall_precision": 0.7331730769230769,
-      "eval_overall_recall": 0.8997050147492626,
-      "eval_runtime": 0.9505,
-      "eval_samples_per_second": 196.731,
-      "eval_steps_per_second": 3.156,
+      "eval_TIME_precision": 0.8076923076923077,
+      "eval_TIME_recall": 0.8076923076923077,
+      "eval_loss": 0.13682067394256592,
+      "eval_overall_accuracy": 0.9567175381536153,
+      "eval_overall_f1": 0.8238482384823848,
+      "eval_overall_precision": 0.7619047619047619,
+      "eval_overall_recall": 0.8967551622418879,
+      "eval_runtime": 0.3665,
+      "eval_samples_per_second": 510.258,
+      "eval_steps_per_second": 8.186,
       "step": 212
     },
     {
       "epoch": 3.0,
-      "grad_norm": 2.83402419090271,
+      "grad_norm": 2.2102744579315186,
       "learning_rate": 4.85e-05,
-      "loss": 0.0781,
+      "loss": 0.0841,
       "step": 318
     },
     {
       "epoch": 3.0,
-      "eval_LOCATION_f1": 0.8299319727891157,
+      "eval_LOCATION_f1": 0.8028169014084506,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.8133333333333334,
-      "eval_LOCATION_recall": 0.8472222222222222,
-      "eval_ORGANIZATION_f1": 0.7777777777777778,
+      "eval_LOCATION_precision": 0.8142857142857143,
+      "eval_LOCATION_recall": 0.7916666666666666,
+      "eval_ORGANIZATION_f1": 0.7810650887573964,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.6666666666666666,
-      "eval_ORGANIZATION_recall": 0.9333333333333333,
-      "eval_PERSON_f1": 0.8933333333333333,
+      "eval_ORGANIZATION_precision": 0.7021276595744681,
+      "eval_ORGANIZATION_recall": 0.88,
+      "eval_PERSON_f1": 0.8918032786885245,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8535031847133758,
-      "eval_PERSON_recall": 0.9370629370629371,
-      "eval_QUANTITY_f1": 0.7083333333333334,
+      "eval_PERSON_precision": 0.8395061728395061,
+      "eval_PERSON_recall": 0.951048951048951,
+      "eval_QUANTITY_f1": 0.6415094339622641,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.68,
+      "eval_QUANTITY_precision": 0.5666666666666667,
       "eval_QUANTITY_recall": 0.7391304347826086,
-      "eval_TIME_f1": 0.8461538461538461,
+      "eval_TIME_f1": 0.8333333333333333,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8461538461538461,
-      "eval_TIME_recall": 0.8461538461538461,
-      "eval_loss": 0.15726295113563538,
-      "eval_overall_accuracy": 0.9552164123092319,
-      "eval_overall_f1": 0.8363136176066025,
-      "eval_overall_precision": 0.7835051546391752,
-      "eval_overall_recall": 0.8967551622418879,
-      "eval_runtime": 0.8869,
-      "eval_samples_per_second": 210.857,
-      "eval_steps_per_second": 3.383,
+      "eval_TIME_precision": 0.9090909090909091,
+      "eval_TIME_recall": 0.7692307692307693,
+      "eval_loss": 0.1551402360200882,
+      "eval_overall_accuracy": 0.9547160370277709,
+      "eval_overall_f1": 0.8256624825662482,
+      "eval_overall_precision": 0.783068783068783,
+      "eval_overall_recall": 0.8731563421828908,
+      "eval_runtime": 0.3569,
+      "eval_samples_per_second": 523.957,
+      "eval_steps_per_second": 8.406,
       "step": 318
     },
     {
       "epoch": 4.0,
-      "grad_norm": 2.320517063140869,
+      "grad_norm": 8.499090194702148,
       "learning_rate": 4.8e-05,
-      "loss": 0.0572,
+      "loss": 0.0533,
       "step": 424
     },
     {
       "epoch": 4.0,
-      "eval_LOCATION_f1": 0.816326530612245,
+      "eval_LOCATION_f1": 0.8053691275167787,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.8,
+      "eval_LOCATION_precision": 0.7792207792207793,
       "eval_LOCATION_recall": 0.8333333333333334,
-      "eval_ORGANIZATION_f1": 0.7651006711409396,
+      "eval_ORGANIZATION_f1": 0.7719298245614036,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7702702702702703,
-      "eval_ORGANIZATION_recall": 0.76,
-      "eval_PERSON_f1": 0.9019607843137255,
+      "eval_ORGANIZATION_precision": 0.6875,
+      "eval_ORGANIZATION_recall": 0.88,
+      "eval_PERSON_f1": 0.8778877887788779,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8466257668711656,
-      "eval_PERSON_recall": 0.965034965034965,
-      "eval_QUANTITY_f1": 0.6415094339622641,
+      "eval_PERSON_precision": 0.83125,
+      "eval_PERSON_recall": 0.9300699300699301,
+      "eval_QUANTITY_f1": 0.6666666666666667,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.5666666666666667,
-      "eval_QUANTITY_recall": 0.7391304347826086,
-      "eval_TIME_f1": 0.7368421052631579,
+      "eval_QUANTITY_precision": 0.5806451612903226,
+      "eval_QUANTITY_recall": 0.782608695652174,
+      "eval_TIME_f1": 0.7540983606557377,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.6774193548387096,
-      "eval_TIME_recall": 0.8076923076923077,
-      "eval_loss": 0.17222364246845245,
-      "eval_overall_accuracy": 0.957968476357268,
-      "eval_overall_f1": 0.8230337078651685,
-      "eval_overall_precision": 0.7855227882037533,
-      "eval_overall_recall": 0.8643067846607669,
-      "eval_runtime": 0.9001,
-      "eval_samples_per_second": 207.754,
-      "eval_steps_per_second": 3.333,
+      "eval_TIME_precision": 0.6571428571428571,
+      "eval_TIME_recall": 0.8846153846153846,
+      "eval_loss": 0.17204634845256805,
+      "eval_overall_accuracy": 0.9552164123092319,
+      "eval_overall_f1": 0.8130081300813008,
+      "eval_overall_precision": 0.7518796992481203,
+      "eval_overall_recall": 0.8849557522123894,
+      "eval_runtime": 0.3715,
+      "eval_samples_per_second": 503.365,
+      "eval_steps_per_second": 8.075,
       "step": 424
     },
     {
       "epoch": 5.0,
-      "grad_norm": 0.44582077860832214,
+      "grad_norm": 10.976715087890625,
       "learning_rate": 4.75e-05,
       "loss": 0.0359,
       "step": 530
     },
     {
       "epoch": 5.0,
-      "eval_LOCATION_f1": 0.8258064516129032,
+      "eval_LOCATION_f1": 0.8079470198675496,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7710843373493976,
-      "eval_LOCATION_recall": 0.8888888888888888,
-      "eval_ORGANIZATION_f1": 0.7951807228915663,
+      "eval_LOCATION_precision": 0.7721518987341772,
+      "eval_LOCATION_recall": 0.8472222222222222,
+      "eval_ORGANIZATION_f1": 0.7784431137724551,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7252747252747253,
-      "eval_ORGANIZATION_recall": 0.88,
-      "eval_PERSON_f1": 0.9152542372881356,
+      "eval_ORGANIZATION_precision": 0.7065217391304348,
+      "eval_ORGANIZATION_recall": 0.8666666666666667,
+      "eval_PERSON_f1": 0.9049180327868852,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8881578947368421,
-      "eval_PERSON_recall": 0.9440559440559441,
-      "eval_QUANTITY_f1": 0.7450980392156864,
+      "eval_PERSON_precision": 0.8518518518518519,
+      "eval_PERSON_recall": 0.965034965034965,
+      "eval_QUANTITY_f1": 0.6785714285714286,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.6785714285714286,
+      "eval_QUANTITY_precision": 0.5757575757575758,
       "eval_QUANTITY_recall": 0.8260869565217391,
-      "eval_TIME_f1": 0.8799999999999999,
+      "eval_TIME_f1": 0.8,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.9166666666666666,
+      "eval_TIME_precision": 0.7586206896551724,
       "eval_TIME_recall": 0.8461538461538461,
-      "eval_loss": 0.15475203096866608,
-      "eval_overall_accuracy": 0.9642231673755316,
-      "eval_overall_f1": 0.8535564853556487,
-      "eval_overall_precision": 0.8095238095238095,
-      "eval_overall_recall": 0.9026548672566371,
-      "eval_runtime": 0.9396,
-      "eval_samples_per_second": 199.021,
-      "eval_steps_per_second": 3.193,
+      "eval_loss": 0.20843125879764557,
+      "eval_overall_accuracy": 0.9564673505128847,
+      "eval_overall_f1": 0.8310626702997276,
+      "eval_overall_precision": 0.7721518987341772,
+      "eval_overall_recall": 0.8997050147492626,
+      "eval_runtime": 0.3572,
+      "eval_samples_per_second": 523.556,
+      "eval_steps_per_second": 8.399,
       "step": 530
     },
     {
       "epoch": 6.0,
-      "grad_norm": 0.24779094755649567,
+      "grad_norm": 0.3230496346950531,
       "learning_rate": 4.7e-05,
-      "loss": 0.0236,
+      "loss": 0.0256,
       "step": 636
     },
     {
       "epoch": 6.0,
-      "eval_LOCATION_f1": 0.7808219178082192,
+      "eval_LOCATION_f1": 0.816326530612245,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7702702702702703,
-      "eval_LOCATION_recall": 0.7916666666666666,
-      "eval_ORGANIZATION_f1": 0.7564102564102564,
+      "eval_LOCATION_precision": 0.8,
+      "eval_LOCATION_recall": 0.8333333333333334,
+      "eval_ORGANIZATION_f1": 0.7597765363128492,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7283950617283951,
-      "eval_ORGANIZATION_recall": 0.7866666666666666,
-      "eval_PERSON_f1": 0.888888888888889,
+      "eval_ORGANIZATION_precision": 0.6538461538461539,
+      "eval_ORGANIZATION_recall": 0.9066666666666666,
+      "eval_PERSON_f1": 0.9006622516556291,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8571428571428571,
-      "eval_PERSON_recall": 0.9230769230769231,
-      "eval_QUANTITY_f1": 0.72,
+      "eval_PERSON_precision": 0.8553459119496856,
+      "eval_PERSON_recall": 0.951048951048951,
+      "eval_QUANTITY_f1": 0.6666666666666667,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.6666666666666666,
-      "eval_QUANTITY_recall": 0.782608695652174,
-      "eval_TIME_f1": 0.8148148148148148,
+      "eval_QUANTITY_precision": 0.5588235294117647,
+      "eval_QUANTITY_recall": 0.8260869565217391,
+      "eval_TIME_f1": 0.7368421052631579,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.7857142857142857,
-      "eval_TIME_recall": 0.8461538461538461,
-      "eval_loss": 0.18202847242355347,
-      "eval_overall_accuracy": 0.9604703527645734,
-      "eval_overall_f1": 0.8193456614509247,
-      "eval_overall_precision": 0.7912087912087912,
-      "eval_overall_recall": 0.8495575221238938,
-      "eval_runtime": 0.9076,
-      "eval_samples_per_second": 206.029,
-      "eval_steps_per_second": 3.305,
+      "eval_TIME_precision": 0.6774193548387096,
+      "eval_TIME_recall": 0.8076923076923077,
+      "eval_loss": 0.2402852177619934,
+      "eval_overall_accuracy": 0.9539654741055792,
+      "eval_overall_f1": 0.8194070080862534,
+      "eval_overall_precision": 0.7543424317617866,
+      "eval_overall_recall": 0.8967551622418879,
+      "eval_runtime": 0.3663,
+      "eval_samples_per_second": 510.497,
+      "eval_steps_per_second": 8.19,
       "step": 636
     },
     {
       "epoch": 7.0,
-      "grad_norm": 0.0966544821858406,
+      "grad_norm": 0.021450912579894066,
       "learning_rate": 4.6500000000000005e-05,
-      "loss": 0.0164,
+      "loss": 0.0188,
       "step": 742
     },
     {
       "epoch": 7.0,
-      "eval_LOCATION_f1": 0.8181818181818181,
+      "eval_LOCATION_f1": 0.8133333333333332,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7682926829268293,
-      "eval_LOCATION_recall": 0.875,
-      "eval_ORGANIZATION_f1": 0.744186046511628,
+      "eval_LOCATION_precision": 0.782051282051282,
+      "eval_LOCATION_recall": 0.8472222222222222,
+      "eval_ORGANIZATION_f1": 0.7624309392265193,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.6597938144329897,
-      "eval_ORGANIZATION_recall": 0.8533333333333334,
-      "eval_PERSON_f1": 0.9072847682119206,
+      "eval_ORGANIZATION_precision": 0.6509433962264151,
+      "eval_ORGANIZATION_recall": 0.92,
+      "eval_PERSON_f1": 0.8926174496644295,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8616352201257862,
-      "eval_PERSON_recall": 0.958041958041958,
-      "eval_QUANTITY_f1": 0.7547169811320754,
+      "eval_PERSON_precision": 0.8580645161290322,
+      "eval_PERSON_recall": 0.9300699300699301,
+      "eval_QUANTITY_f1": 0.6923076923076923,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.6666666666666666,
-      "eval_QUANTITY_recall": 0.8695652173913043,
-      "eval_TIME_f1": 0.7719298245614036,
+      "eval_QUANTITY_precision": 0.6206896551724138,
+      "eval_QUANTITY_recall": 0.782608695652174,
+      "eval_TIME_f1": 0.7118644067796609,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.7096774193548387,
-      "eval_TIME_recall": 0.8461538461538461,
-      "eval_loss": 0.2388230413198471,
-      "eval_overall_accuracy": 0.9567175381536153,
-      "eval_overall_f1": 0.8292682926829268,
-      "eval_overall_precision": 0.7669172932330827,
-      "eval_overall_recall": 0.9026548672566371,
-      "eval_runtime": 0.9186,
-      "eval_samples_per_second": 203.564,
-      "eval_steps_per_second": 3.266,
+      "eval_TIME_precision": 0.6363636363636364,
+      "eval_TIME_recall": 0.8076923076923077,
+      "eval_loss": 0.22656874358654022,
+      "eval_overall_accuracy": 0.9569677257943457,
+      "eval_overall_f1": 0.8162162162162162,
+      "eval_overall_precision": 0.7531172069825436,
+      "eval_overall_recall": 0.8908554572271387,
+      "eval_runtime": 0.356,
+      "eval_samples_per_second": 525.333,
+      "eval_steps_per_second": 8.428,
       "step": 742
     },
     {
       "epoch": 8.0,
-      "grad_norm": 6.724938869476318,
+      "grad_norm": 0.026271946728229523,
       "learning_rate": 4.600000000000001e-05,
-      "loss": 0.0131,
+      "loss": 0.0134,
       "step": 848
     },
     {
       "epoch": 8.0,
-      "eval_LOCATION_f1": 0.8157894736842106,
+      "eval_LOCATION_f1": 0.7891156462585034,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.775,
-      "eval_LOCATION_recall": 0.8611111111111112,
-      "eval_ORGANIZATION_f1": 0.7341772151898734,
+      "eval_LOCATION_precision": 0.7733333333333333,
+      "eval_LOCATION_recall": 0.8055555555555556,
+      "eval_ORGANIZATION_f1": 0.7530864197530864,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.6987951807228916,
-      "eval_ORGANIZATION_recall": 0.7733333333333333,
-      "eval_PERSON_f1": 0.8754208754208753,
+      "eval_ORGANIZATION_precision": 0.7011494252873564,
+      "eval_ORGANIZATION_recall": 0.8133333333333334,
+      "eval_PERSON_f1": 0.9066666666666667,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8441558441558441,
-      "eval_PERSON_recall": 0.9090909090909091,
-      "eval_QUANTITY_f1": 0.6923076923076923,
+      "eval_PERSON_precision": 0.8662420382165605,
+      "eval_PERSON_recall": 0.951048951048951,
+      "eval_QUANTITY_f1": 0.6415094339622641,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.6206896551724138,
-      "eval_QUANTITY_recall": 0.782608695652174,
-      "eval_TIME_f1": 0.6779661016949152,
+      "eval_QUANTITY_precision": 0.5666666666666667,
+      "eval_QUANTITY_recall": 0.7391304347826086,
+      "eval_TIME_f1": 0.7719298245614036,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.6060606060606061,
-      "eval_TIME_recall": 0.7692307692307693,
-      "eval_loss": 0.21797136962413788,
-      "eval_overall_accuracy": 0.9582186639979985,
-      "eval_overall_f1": 0.8022284122562674,
-      "eval_overall_precision": 0.7598944591029023,
-      "eval_overall_recall": 0.8495575221238938,
-      "eval_runtime": 0.906,
-      "eval_samples_per_second": 206.4,
-      "eval_steps_per_second": 3.311,
+      "eval_TIME_precision": 0.7096774193548387,
+      "eval_TIME_recall": 0.8461538461538461,
+      "eval_loss": 0.24617096781730652,
+      "eval_overall_accuracy": 0.9572179134350763,
+      "eval_overall_f1": 0.8178025034770514,
+      "eval_overall_precision": 0.7736842105263158,
+      "eval_overall_recall": 0.8672566371681416,
+      "eval_runtime": 0.3558,
+      "eval_samples_per_second": 525.577,
+      "eval_steps_per_second": 8.432,
       "step": 848
     },
     {
       "epoch": 9.0,
-      "grad_norm": 1.854614019393921,
+      "grad_norm": 0.6316163539886475,
       "learning_rate": 4.55e-05,
-      "loss": 0.0091,
+      "loss": 0.0135,
       "step": 954
     },
     {
       "epoch": 9.0,
-      "eval_LOCATION_f1": 0.8435374149659863,
+      "eval_LOCATION_f1": 0.7972027972027971,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.8266666666666667,
-      "eval_LOCATION_recall": 0.8611111111111112,
-      "eval_ORGANIZATION_f1": 0.788235294117647,
+      "eval_LOCATION_precision": 0.8028169014084507,
+      "eval_LOCATION_recall": 0.7916666666666666,
+      "eval_ORGANIZATION_f1": 0.8098159509202455,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7052631578947368,
-      "eval_ORGANIZATION_recall": 0.8933333333333333,
-      "eval_PERSON_f1": 0.8837209302325583,
+      "eval_ORGANIZATION_precision": 0.75,
+      "eval_ORGANIZATION_recall": 0.88,
+      "eval_PERSON_f1": 0.8829431438127091,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8417721518987342,
-      "eval_PERSON_recall": 0.9300699300699301,
+      "eval_PERSON_precision": 0.8461538461538461,
+      "eval_PERSON_recall": 0.9230769230769231,
       "eval_QUANTITY_f1": 0.6923076923076923,
       "eval_QUANTITY_number": 23,
       "eval_QUANTITY_precision": 0.6206896551724138,
       "eval_QUANTITY_recall": 0.782608695652174,
-      "eval_TIME_f1": 0.7777777777777779,
+      "eval_TIME_f1": 0.8148148148148148,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.75,
-      "eval_TIME_recall": 0.8076923076923077,
-      "eval_loss": 0.26150327920913696,
-      "eval_overall_accuracy": 0.9589692269201902,
-      "eval_overall_f1": 0.8314917127071824,
-      "eval_overall_precision": 0.7818181818181819,
-      "eval_overall_recall": 0.887905604719764,
-      "eval_runtime": 0.9105,
-      "eval_samples_per_second": 205.388,
-      "eval_steps_per_second": 3.295,
+      "eval_TIME_precision": 0.7857142857142857,
+      "eval_TIME_recall": 0.8461538461538461,
+      "eval_loss": 0.2374754250049591,
+      "eval_overall_accuracy": 0.9592194145609206,
+      "eval_overall_f1": 0.829817158931083,
+      "eval_overall_precision": 0.793010752688172,
+      "eval_overall_recall": 0.8702064896755162,
+      "eval_runtime": 0.3585,
+      "eval_samples_per_second": 521.613,
+      "eval_steps_per_second": 8.368,
       "step": 954
     },
     {
       "epoch": 10.0,
-      "grad_norm": 1.278331995010376,
+      "grad_norm": 0.20594312250614166,
       "learning_rate": 4.5e-05,
-      "loss": 0.0099,
+      "loss": 0.0082,
       "step": 1060
     },
     {
       "epoch": 10.0,
-      "eval_LOCATION_f1": 0.816326530612245,
+      "eval_LOCATION_f1": 0.7973856209150327,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.8,
-      "eval_LOCATION_recall": 0.8333333333333334,
-      "eval_ORGANIZATION_f1": 0.7500000000000001,
+      "eval_LOCATION_precision": 0.7530864197530864,
+      "eval_LOCATION_recall": 0.8472222222222222,
+      "eval_ORGANIZATION_f1": 0.7904191616766466,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7058823529411765,
-      "eval_ORGANIZATION_recall": 0.8,
-      "eval_PERSON_f1": 0.9183673469387755,
+      "eval_ORGANIZATION_precision": 0.717391304347826,
+      "eval_ORGANIZATION_recall": 0.88,
+      "eval_PERSON_f1": 0.8933333333333333,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8940397350993378,
-      "eval_PERSON_recall": 0.9440559440559441,
-      "eval_QUANTITY_f1": 0.6923076923076923,
+      "eval_PERSON_precision": 0.8535031847133758,
+      "eval_PERSON_recall": 0.9370629370629371,
+      "eval_QUANTITY_f1": 0.6415094339622641,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.6206896551724138,
-      "eval_QUANTITY_recall": 0.782608695652174,
-      "eval_TIME_f1": 0.8214285714285715,
+      "eval_QUANTITY_precision": 0.5666666666666667,
+      "eval_QUANTITY_recall": 0.7391304347826086,
+      "eval_TIME_f1": 0.8363636363636363,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.7666666666666667,
+      "eval_TIME_precision": 0.7931034482758621,
       "eval_TIME_recall": 0.8846153846153846,
-      "eval_loss": 0.24217499792575836,
-      "eval_overall_accuracy": 0.9619714786089567,
-      "eval_overall_f1": 0.8349788434414668,
-      "eval_overall_precision": 0.8,
-      "eval_overall_recall": 0.8731563421828908,
-      "eval_runtime": 0.9185,
-      "eval_samples_per_second": 203.602,
-      "eval_steps_per_second": 3.266,
+      "eval_loss": 0.27111825346946716,
+      "eval_overall_accuracy": 0.9574681010758068,
+      "eval_overall_f1": 0.8269230769230769,
+      "eval_overall_precision": 0.7737789203084833,
+      "eval_overall_recall": 0.887905604719764,
+      "eval_runtime": 0.3597,
+      "eval_samples_per_second": 519.941,
+      "eval_steps_per_second": 8.341,
       "step": 1060
     },
     {
       "epoch": 11.0,
-      "grad_norm": 0.09772255271673203,
+      "grad_norm": 0.010851857252418995,
       "learning_rate": 4.4500000000000004e-05,
-      "loss": 0.0059,
+      "loss": 0.0079,
       "step": 1166
     },
     {
       "epoch": 11.0,
-      "eval_LOCATION_f1": 0.7972972972972973,
+      "eval_LOCATION_f1": 0.7837837837837838,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7763157894736842,
-      "eval_LOCATION_recall": 0.8194444444444444,
-      "eval_ORGANIZATION_f1": 0.7619047619047621,
+      "eval_LOCATION_precision": 0.7631578947368421,
+      "eval_LOCATION_recall": 0.8055555555555556,
+      "eval_ORGANIZATION_f1": 0.7878787878787877,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.6881720430107527,
-      "eval_ORGANIZATION_recall": 0.8533333333333334,
-      "eval_PERSON_f1": 0.907849829351536,
+      "eval_ORGANIZATION_precision": 0.7222222222222222,
+      "eval_ORGANIZATION_recall": 0.8666666666666667,
+      "eval_PERSON_f1": 0.903010033444816,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8866666666666667,
-      "eval_PERSON_recall": 0.9300699300699301,
-      "eval_QUANTITY_f1": 0.6923076923076923,
+      "eval_PERSON_precision": 0.8653846153846154,
+      "eval_PERSON_recall": 0.9440559440559441,
+      "eval_QUANTITY_f1": 0.6792452830188679,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.6206896551724138,
+      "eval_QUANTITY_precision": 0.6,
       "eval_QUANTITY_recall": 0.782608695652174,
-      "eval_TIME_f1": 0.8518518518518519,
+      "eval_TIME_f1": 0.7636363636363636,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8214285714285714,
-      "eval_TIME_recall": 0.8846153846153846,
-      "eval_loss": 0.2546893060207367,
-      "eval_overall_accuracy": 0.9589692269201902,
-      "eval_overall_f1": 0.8307692307692307,
-      "eval_overall_precision": 0.7898936170212766,
+      "eval_TIME_precision": 0.7241379310344828,
+      "eval_TIME_recall": 0.8076923076923077,
+      "eval_loss": 0.27776339650154114,
+      "eval_overall_accuracy": 0.9574681010758068,
+      "eval_overall_f1": 0.825,
+      "eval_overall_precision": 0.7795275590551181,
       "eval_overall_recall": 0.8761061946902655,
-      "eval_runtime": 0.9121,
-      "eval_samples_per_second": 205.026,
-      "eval_steps_per_second": 3.289,
+      "eval_runtime": 0.3586,
+      "eval_samples_per_second": 521.408,
+      "eval_steps_per_second": 8.365,
       "step": 1166
     },
     {
       "epoch": 12.0,
-      "grad_norm": 0.023314962163567543,
+      "grad_norm": 4.9953837394714355,
       "learning_rate": 4.4000000000000006e-05,
-      "loss": 0.0062,
+      "loss": 0.0105,
       "step": 1272
     },
     {
       "epoch": 12.0,
-      "eval_LOCATION_f1": 0.7945205479452055,
+      "eval_LOCATION_f1": 0.7916666666666666,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7837837837837838,
-      "eval_LOCATION_recall": 0.8055555555555556,
-      "eval_ORGANIZATION_f1": 0.8074534161490684,
+      "eval_LOCATION_precision": 0.7916666666666666,
+      "eval_LOCATION_recall": 0.7916666666666666,
+      "eval_ORGANIZATION_f1": 0.8101265822784811,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7558139534883721,
-      "eval_ORGANIZATION_recall": 0.8666666666666667,
-      "eval_PERSON_f1": 0.8829431438127091,
+      "eval_ORGANIZATION_precision": 0.7710843373493976,
+      "eval_ORGANIZATION_recall": 0.8533333333333334,
+      "eval_PERSON_f1": 0.903010033444816,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8461538461538461,
-      "eval_PERSON_recall": 0.9230769230769231,
-      "eval_QUANTITY_f1": 0.6538461538461539,
+      "eval_PERSON_precision": 0.8653846153846154,
+      "eval_PERSON_recall": 0.9440559440559441,
+      "eval_QUANTITY_f1": 0.72,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.5862068965517241,
-      "eval_QUANTITY_recall": 0.7391304347826086,
-      "eval_TIME_f1": 0.8363636363636363,
+      "eval_QUANTITY_precision": 0.6666666666666666,
+      "eval_QUANTITY_recall": 0.782608695652174,
+      "eval_TIME_f1": 0.7857142857142856,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.7931034482758621,
-      "eval_TIME_recall": 0.8846153846153846,
-      "eval_loss": 0.2780083119869232,
-      "eval_overall_accuracy": 0.9602201651238429,
-      "eval_overall_f1": 0.8274894810659187,
-      "eval_overall_precision": 0.7887700534759359,
-      "eval_overall_recall": 0.8702064896755162,
-      "eval_runtime": 0.9675,
-      "eval_samples_per_second": 193.282,
-      "eval_steps_per_second": 3.101,
+      "eval_TIME_precision": 0.7333333333333333,
+      "eval_TIME_recall": 0.8461538461538461,
+      "eval_loss": 0.28447094559669495,
+      "eval_overall_accuracy": 0.961220915686765,
+      "eval_overall_f1": 0.8373408769448375,
+      "eval_overall_precision": 0.8043478260869565,
+      "eval_overall_recall": 0.8731563421828908,
+      "eval_runtime": 0.3599,
+      "eval_samples_per_second": 519.532,
+      "eval_steps_per_second": 8.335,
       "step": 1272
     },
     {
       "epoch": 13.0,
-      "grad_norm": 3.4945220947265625,
+      "grad_norm": 0.2857362926006317,
       "learning_rate": 4.35e-05,
-      "loss": 0.0037,
+      "loss": 0.0103,
       "step": 1378
     },
     {
       "epoch": 13.0,
-      "eval_LOCATION_f1": 0.782051282051282,
+      "eval_LOCATION_f1": 0.7891156462585034,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7261904761904762,
-      "eval_LOCATION_recall": 0.8472222222222222,
-      "eval_ORGANIZATION_f1": 0.7417218543046358,
+      "eval_LOCATION_precision": 0.7733333333333333,
+      "eval_LOCATION_recall": 0.8055555555555556,
+      "eval_ORGANIZATION_f1": 0.8170731707317072,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7368421052631579,
-      "eval_ORGANIZATION_recall": 0.7466666666666667,
-      "eval_PERSON_f1": 0.9163879598662208,
+      "eval_ORGANIZATION_precision": 0.7528089887640449,
+      "eval_ORGANIZATION_recall": 0.8933333333333333,
+      "eval_PERSON_f1": 0.8926174496644295,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8782051282051282,
-      "eval_PERSON_recall": 0.958041958041958,
-      "eval_QUANTITY_f1": 0.72,
+      "eval_PERSON_precision": 0.8580645161290322,
+      "eval_PERSON_recall": 0.9300699300699301,
+      "eval_QUANTITY_f1": 0.7037037037037037,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.6666666666666666,
-      "eval_QUANTITY_recall": 0.782608695652174,
-      "eval_TIME_f1": 0.8727272727272727,
+      "eval_QUANTITY_precision": 0.6129032258064516,
+      "eval_QUANTITY_recall": 0.8260869565217391,
+      "eval_TIME_f1": 0.9019607843137256,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8275862068965517,
-      "eval_TIME_recall": 0.9230769230769231,
-      "eval_loss": 0.2757226526737213,
-      "eval_overall_accuracy": 0.9627220415311484,
-      "eval_overall_f1": 0.8326300984528833,
-      "eval_overall_precision": 0.7956989247311828,
-      "eval_overall_recall": 0.8731563421828908,
-      "eval_runtime": 0.9689,
-      "eval_samples_per_second": 193.003,
-      "eval_steps_per_second": 3.096,
+      "eval_TIME_precision": 0.92,
+      "eval_TIME_recall": 0.8846153846153846,
+      "eval_loss": 0.2690028250217438,
+      "eval_overall_accuracy": 0.9602201651238429,
+      "eval_overall_f1": 0.8403361344537816,
+      "eval_overall_precision": 0.8,
+      "eval_overall_recall": 0.8849557522123894,
+      "eval_runtime": 0.3608,
+      "eval_samples_per_second": 518.338,
+      "eval_steps_per_second": 8.316,
       "step": 1378
     },
     {
       "epoch": 14.0,
-      "grad_norm": 0.025552373379468918,
+      "grad_norm": 0.11683321744203568,
       "learning_rate": 4.3e-05,
-      "loss": 0.005,
+      "loss": 0.0045,
       "step": 1484
     },
     {
       "epoch": 14.0,
-      "eval_LOCATION_f1": 0.8243243243243243,
+      "eval_LOCATION_f1": 0.7887323943661971,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.8026315789473685,
-      "eval_LOCATION_recall": 0.8472222222222222,
-      "eval_ORGANIZATION_f1": 0.7341772151898734,
+      "eval_LOCATION_precision": 0.8,
+      "eval_LOCATION_recall": 0.7777777777777778,
+      "eval_ORGANIZATION_f1": 0.809248554913295,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.6987951807228916,
-      "eval_ORGANIZATION_recall": 0.7733333333333333,
-      "eval_PERSON_f1": 0.9023569023569024,
+      "eval_ORGANIZATION_precision": 0.7142857142857143,
+      "eval_ORGANIZATION_recall": 0.9333333333333333,
+      "eval_PERSON_f1": 0.8933333333333333,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8701298701298701,
+      "eval_PERSON_precision": 0.8535031847133758,
       "eval_PERSON_recall": 0.9370629370629371,
-      "eval_QUANTITY_f1": 0.7169811320754716,
+      "eval_QUANTITY_f1": 0.6545454545454547,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.6333333333333333,
-      "eval_QUANTITY_recall": 0.8260869565217391,
-      "eval_TIME_f1": 0.8461538461538461,
+      "eval_QUANTITY_precision": 0.5625,
+      "eval_QUANTITY_recall": 0.782608695652174,
+      "eval_TIME_f1": 0.7636363636363636,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8461538461538461,
-      "eval_TIME_recall": 0.8461538461538461,
-      "eval_loss": 0.280028760433197,
-      "eval_overall_accuracy": 0.9594696022016512,
-      "eval_overall_f1": 0.8305084745762712,
-      "eval_overall_precision": 0.7967479674796748,
-      "eval_overall_recall": 0.8672566371681416,
-      "eval_runtime": 0.9073,
-      "eval_samples_per_second": 206.107,
-      "eval_steps_per_second": 3.307,
+      "eval_TIME_precision": 0.7241379310344828,
+      "eval_TIME_recall": 0.8076923076923077,
+      "eval_loss": 0.304815411567688,
+      "eval_overall_accuracy": 0.957968476357268,
+      "eval_overall_f1": 0.8248275862068966,
+      "eval_overall_precision": 0.7746113989637305,
+      "eval_overall_recall": 0.8820058997050148,
+      "eval_runtime": 0.3586,
+      "eval_samples_per_second": 521.439,
+      "eval_steps_per_second": 8.365,
       "step": 1484
     },
     {
       "epoch": 15.0,
-      "grad_norm": 0.192918598651886,
+      "grad_norm": 0.15561413764953613,
       "learning_rate": 4.25e-05,
-      "loss": 0.0058,
+      "loss": 0.005,
       "step": 1590
     },
     {
       "epoch": 15.0,
-      "eval_LOCATION_f1": 0.8356164383561645,
+      "eval_LOCATION_f1": 0.816326530612245,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.8243243243243243,
-      "eval_LOCATION_recall": 0.8472222222222222,
-      "eval_ORGANIZATION_f1": 0.7784431137724551,
+      "eval_LOCATION_precision": 0.8,
+      "eval_LOCATION_recall": 0.8333333333333334,
+      "eval_ORGANIZATION_f1": 0.8000000000000002,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7065217391304348,
-      "eval_ORGANIZATION_recall": 0.8666666666666667,
-      "eval_PERSON_f1": 0.8963210702341137,
+      "eval_ORGANIZATION_precision": 0.775,
+      "eval_ORGANIZATION_recall": 0.8266666666666667,
+      "eval_PERSON_f1": 0.9054054054054055,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8589743589743589,
+      "eval_PERSON_precision": 0.8758169934640523,
       "eval_PERSON_recall": 0.9370629370629371,
-      "eval_QUANTITY_f1": 0.6666666666666666,
+      "eval_QUANTITY_f1": 0.6792452830188679,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.64,
-      "eval_QUANTITY_recall": 0.6956521739130435,
-      "eval_TIME_f1": 0.8070175438596492,
+      "eval_QUANTITY_precision": 0.6,
+      "eval_QUANTITY_recall": 0.782608695652174,
+      "eval_TIME_f1": 0.8148148148148148,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.7419354838709677,
-      "eval_TIME_recall": 0.8846153846153846,
-      "eval_loss": 0.2716998755931854,
-      "eval_overall_accuracy": 0.9629722291718789,
-      "eval_overall_f1": 0.8340306834030684,
-      "eval_overall_precision": 0.791005291005291,
-      "eval_overall_recall": 0.8820058997050148,
-      "eval_runtime": 0.91,
-      "eval_samples_per_second": 205.503,
-      "eval_steps_per_second": 3.297,
+      "eval_TIME_precision": 0.7857142857142857,
+      "eval_TIME_recall": 0.8461538461538461,
+      "eval_loss": 0.27857497334480286,
+      "eval_overall_accuracy": 0.9597197898423818,
+      "eval_overall_f1": 0.8397163120567377,
+      "eval_overall_precision": 0.8087431693989071,
+      "eval_overall_recall": 0.8731563421828908,
+      "eval_runtime": 0.3743,
+      "eval_samples_per_second": 499.656,
+      "eval_steps_per_second": 8.016,
       "step": 1590
     },
     {
       "epoch": 16.0,
-      "grad_norm": 0.09364385157823563,
+      "grad_norm": 0.24679070711135864,
       "learning_rate": 4.2e-05,
-      "loss": 0.0046,
+      "loss": 0.0041,
       "step": 1696
     },
     {
       "epoch": 16.0,
-      "eval_LOCATION_f1": 0.816326530612245,
+      "eval_LOCATION_f1": 0.7866666666666666,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.8,
-      "eval_LOCATION_recall": 0.8333333333333334,
-      "eval_ORGANIZATION_f1": 0.735632183908046,
+      "eval_LOCATION_precision": 0.7564102564102564,
+      "eval_LOCATION_recall": 0.8194444444444444,
+      "eval_ORGANIZATION_f1": 0.8170731707317072,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.6464646464646465,
-      "eval_ORGANIZATION_recall": 0.8533333333333334,
-      "eval_PERSON_f1": 0.8993288590604027,
+      "eval_ORGANIZATION_precision": 0.7528089887640449,
+      "eval_ORGANIZATION_recall": 0.8933333333333333,
+      "eval_PERSON_f1": 0.8903654485049833,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.864516129032258,
+      "eval_PERSON_precision": 0.8481012658227848,
       "eval_PERSON_recall": 0.9370629370629371,
-      "eval_QUANTITY_f1": 0.6545454545454547,
+      "eval_QUANTITY_f1": 0.7169811320754716,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.5625,
-      "eval_QUANTITY_recall": 0.782608695652174,
-      "eval_TIME_f1": 0.7931034482758621,
+      "eval_QUANTITY_precision": 0.6333333333333333,
+      "eval_QUANTITY_recall": 0.8260869565217391,
+      "eval_TIME_f1": 0.830188679245283,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.71875,
-      "eval_TIME_recall": 0.8846153846153846,
-      "eval_loss": 0.30642953515052795,
-      "eval_overall_accuracy": 0.9539654741055792,
-      "eval_overall_f1": 0.8169398907103824,
-      "eval_overall_precision": 0.7608142493638677,
-      "eval_overall_recall": 0.8820058997050148,
-      "eval_runtime": 0.9036,
-      "eval_samples_per_second": 206.952,
-      "eval_steps_per_second": 3.32,
+      "eval_TIME_precision": 0.8148148148148148,
+      "eval_TIME_recall": 0.8461538461538461,
+      "eval_loss": 0.30642324686050415,
+      "eval_overall_accuracy": 0.9572179134350763,
+      "eval_overall_f1": 0.8349514563106795,
+      "eval_overall_precision": 0.7879581151832461,
+      "eval_overall_recall": 0.887905604719764,
+      "eval_runtime": 0.3572,
+      "eval_samples_per_second": 523.51,
+      "eval_steps_per_second": 8.399,
       "step": 1696
     },
     {
       "epoch": 17.0,
-      "grad_norm": 4.5664825439453125,
+      "grad_norm": 0.17553383111953735,
       "learning_rate": 4.15e-05,
-      "loss": 0.0063,
+      "loss": 0.0061,
       "step": 1802
     },
     {
       "epoch": 17.0,
-      "eval_LOCATION_f1": 0.7763157894736842,
+      "eval_LOCATION_f1": 0.7733333333333334,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7375,
-      "eval_LOCATION_recall": 0.8194444444444444,
-      "eval_ORGANIZATION_f1": 0.7922077922077922,
+      "eval_LOCATION_precision": 0.7435897435897436,
+      "eval_LOCATION_recall": 0.8055555555555556,
+      "eval_ORGANIZATION_f1": 0.7701863354037267,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7721518987341772,
-      "eval_ORGANIZATION_recall": 0.8133333333333334,
-      "eval_PERSON_f1": 0.8940397350993377,
+      "eval_ORGANIZATION_precision": 0.7209302325581395,
+      "eval_ORGANIZATION_recall": 0.8266666666666667,
+      "eval_PERSON_f1": 0.8910891089108911,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8490566037735849,
+      "eval_PERSON_precision": 0.84375,
       "eval_PERSON_recall": 0.9440559440559441,
-      "eval_QUANTITY_f1": 0.6792452830188679,
+      "eval_QUANTITY_f1": 0.7058823529411765,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.6,
+      "eval_QUANTITY_precision": 0.6428571428571429,
       "eval_QUANTITY_recall": 0.782608695652174,
-      "eval_TIME_f1": 0.830188679245283,
+      "eval_TIME_f1": 0.8148148148148148,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8148148148148148,
+      "eval_TIME_precision": 0.7857142857142857,
       "eval_TIME_recall": 0.8461538461538461,
-      "eval_loss": 0.28576722741127014,
-      "eval_overall_accuracy": 0.9617212909682261,
-      "eval_overall_f1": 0.8263305322128851,
-      "eval_overall_precision": 0.7866666666666666,
+      "eval_loss": 0.300820916891098,
+      "eval_overall_accuracy": 0.9574681010758068,
+      "eval_overall_f1": 0.8205841446453408,
+      "eval_overall_precision": 0.7763157894736842,
       "eval_overall_recall": 0.8702064896755162,
-      "eval_runtime": 0.9045,
-      "eval_samples_per_second": 206.753,
-      "eval_steps_per_second": 3.317,
+      "eval_runtime": 0.3595,
+      "eval_samples_per_second": 520.18,
+      "eval_steps_per_second": 8.345,
       "step": 1802
     },
     {
       "epoch": 18.0,
-      "grad_norm": 0.04714202508330345,
+      "grad_norm": 0.004105593077838421,
       "learning_rate": 4.1e-05,
-      "loss": 0.0033,
+      "loss": 0.0036,
       "step": 1908
     },
     {
       "epoch": 18.0,
-      "eval_LOCATION_f1": 0.8,
+      "eval_LOCATION_f1": 0.8133333333333332,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7692307692307693,
-      "eval_LOCATION_recall": 0.8333333333333334,
-      "eval_ORGANIZATION_f1": 0.779874213836478,
+      "eval_LOCATION_precision": 0.782051282051282,
+      "eval_LOCATION_recall": 0.8472222222222222,
+      "eval_ORGANIZATION_f1": 0.8129032258064516,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7380952380952381,
-      "eval_ORGANIZATION_recall": 0.8266666666666667,
-      "eval_PERSON_f1": 0.9006622516556291,
+      "eval_ORGANIZATION_precision": 0.7875,
+      "eval_ORGANIZATION_recall": 0.84,
+      "eval_PERSON_f1": 0.8993288590604027,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8553459119496856,
-      "eval_PERSON_recall": 0.951048951048951,
-      "eval_QUANTITY_f1": 0.6923076923076923,
+      "eval_PERSON_precision": 0.864516129032258,
+      "eval_PERSON_recall": 0.9370629370629371,
+      "eval_QUANTITY_f1": 0.7307692307692308,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.6206896551724138,
-      "eval_QUANTITY_recall": 0.782608695652174,
-      "eval_TIME_f1": 0.8421052631578947,
+      "eval_QUANTITY_precision": 0.6551724137931034,
+      "eval_QUANTITY_recall": 0.8260869565217391,
+      "eval_TIME_f1": 0.8148148148148148,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.7741935483870968,
-      "eval_TIME_recall": 0.9230769230769231,
-      "eval_loss": 0.30405738949775696,
-      "eval_overall_accuracy": 0.9619714786089567,
-      "eval_overall_f1": 0.8333333333333335,
-      "eval_overall_precision": 0.7874015748031497,
-      "eval_overall_recall": 0.8849557522123894,
-      "eval_runtime": 0.903,
-      "eval_samples_per_second": 207.088,
-      "eval_steps_per_second": 3.322,
+      "eval_TIME_precision": 0.7857142857142857,
+      "eval_TIME_recall": 0.8461538461538461,
+      "eval_loss": 0.29750457406044006,
+      "eval_overall_accuracy": 0.9627220415311484,
+      "eval_overall_f1": 0.8434414668547249,
+      "eval_overall_precision": 0.8081081081081081,
+      "eval_overall_recall": 0.8820058997050148,
+      "eval_runtime": 0.3588,
+      "eval_samples_per_second": 521.143,
+      "eval_steps_per_second": 8.361,
       "step": 1908
     },
     {
       "epoch": 19.0,
-      "grad_norm": 0.008477783761918545,
+      "grad_norm": 0.007032826077193022,
       "learning_rate": 4.05e-05,
-      "loss": 0.0033,
+      "loss": 0.0025,
       "step": 2014
     },
     {
       "epoch": 19.0,
-      "eval_LOCATION_f1": 0.8413793103448277,
+      "eval_LOCATION_f1": 0.8513513513513513,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.8356164383561644,
-      "eval_LOCATION_recall": 0.8472222222222222,
-      "eval_ORGANIZATION_f1": 0.8101265822784811,
+      "eval_LOCATION_precision": 0.8289473684210527,
+      "eval_LOCATION_recall": 0.875,
+      "eval_ORGANIZATION_f1": 0.834355828220859,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7710843373493976,
-      "eval_ORGANIZATION_recall": 0.8533333333333334,
-      "eval_PERSON_f1": 0.8979591836734694,
+      "eval_ORGANIZATION_precision": 0.7727272727272727,
+      "eval_ORGANIZATION_recall": 0.9066666666666666,
+      "eval_PERSON_f1": 0.8970099667774087,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8741721854304636,
-      "eval_PERSON_recall": 0.9230769230769231,
-      "eval_QUANTITY_f1": 0.6923076923076923,
+      "eval_PERSON_precision": 0.8544303797468354,
+      "eval_PERSON_recall": 0.9440559440559441,
+      "eval_QUANTITY_f1": 0.76,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.6206896551724138,
-      "eval_QUANTITY_recall": 0.782608695652174,
-      "eval_TIME_f1": 0.8148148148148148,
+      "eval_QUANTITY_precision": 0.7037037037037037,
+      "eval_QUANTITY_recall": 0.8260869565217391,
+      "eval_TIME_f1": 0.8461538461538461,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.7857142857142857,
+      "eval_TIME_precision": 0.8461538461538461,
       "eval_TIME_recall": 0.8461538461538461,
-      "eval_loss": 0.3013346493244171,
-      "eval_overall_accuracy": 0.960720540405304,
-      "eval_overall_f1": 0.844950213371266,
-      "eval_overall_precision": 0.8159340659340659,
-      "eval_overall_recall": 0.8761061946902655,
-      "eval_runtime": 0.9229,
-      "eval_samples_per_second": 202.623,
-      "eval_steps_per_second": 3.251,
+      "eval_loss": 0.3171651363372803,
+      "eval_overall_accuracy": 0.9622216662496873,
+      "eval_overall_f1": 0.8599439775910364,
+      "eval_overall_precision": 0.8186666666666667,
+      "eval_overall_recall": 0.9056047197640118,
+      "eval_runtime": 0.3602,
+      "eval_samples_per_second": 519.212,
+      "eval_steps_per_second": 8.33,
       "step": 2014
     },
     {
       "epoch": 20.0,
-      "grad_norm": 0.0049218675121665,
+      "grad_norm": 1.055253505706787,
       "learning_rate": 4e-05,
-      "loss": 0.0036,
+      "loss": 0.0039,
       "step": 2120
     },
     {
       "epoch": 20.0,
-      "eval_LOCATION_f1": 0.8219178082191781,
+      "eval_LOCATION_f1": 0.8187919463087249,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.8108108108108109,
-      "eval_LOCATION_recall": 0.8333333333333334,
-      "eval_ORGANIZATION_f1": 0.8,
+      "eval_LOCATION_precision": 0.7922077922077922,
+      "eval_LOCATION_recall": 0.8472222222222222,
+      "eval_ORGANIZATION_f1": 0.8098159509202455,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7333333333333333,
+      "eval_ORGANIZATION_precision": 0.75,
       "eval_ORGANIZATION_recall": 0.88,
-      "eval_PERSON_f1": 0.9023569023569024,
+      "eval_PERSON_f1": 0.8970099667774087,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8701298701298701,
-      "eval_PERSON_recall": 0.9370629370629371,
-      "eval_QUANTITY_f1": 0.6538461538461539,
+      "eval_PERSON_precision": 0.8544303797468354,
+      "eval_PERSON_recall": 0.9440559440559441,
+      "eval_QUANTITY_f1": 0.76,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.5862068965517241,
-      "eval_QUANTITY_recall": 0.7391304347826086,
-      "eval_TIME_f1": 0.830188679245283,
+      "eval_QUANTITY_precision": 0.7037037037037037,
+      "eval_QUANTITY_recall": 0.8260869565217391,
+      "eval_TIME_f1": 0.8461538461538461,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8148148148148148,
+      "eval_TIME_precision": 0.8461538461538461,
       "eval_TIME_recall": 0.8461538461538461,
-      "eval_loss": 0.2967695891857147,
-      "eval_overall_accuracy": 0.9609707280460346,
-      "eval_overall_f1": 0.8387096774193548,
-      "eval_overall_precision": 0.7994652406417112,
-      "eval_overall_recall": 0.8820058997050148,
-      "eval_runtime": 0.9818,
-      "eval_samples_per_second": 190.457,
-      "eval_steps_per_second": 3.055,
+      "eval_loss": 0.3097495436668396,
+      "eval_overall_accuracy": 0.9622216662496873,
+      "eval_overall_f1": 0.8475524475524475,
+      "eval_overall_precision": 0.8058510638297872,
+      "eval_overall_recall": 0.8938053097345132,
+      "eval_runtime": 0.3569,
+      "eval_samples_per_second": 524.015,
+      "eval_steps_per_second": 8.407,
       "step": 2120
     },
     {
       "epoch": 21.0,
-      "grad_norm": 0.03995713219046593,
+      "grad_norm": 0.4885401129722595,
       "learning_rate": 3.9500000000000005e-05,
-      "loss": 0.0038,
+      "loss": 0.0031,
       "step": 2226
     },
     {
       "epoch": 21.0,
-      "eval_LOCATION_f1": 0.7625899280575541,
+      "eval_LOCATION_f1": 0.781456953642384,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7910447761194029,
-      "eval_LOCATION_recall": 0.7361111111111112,
-      "eval_ORGANIZATION_f1": 0.7374999999999999,
+      "eval_LOCATION_precision": 0.7468354430379747,
+      "eval_LOCATION_recall": 0.8194444444444444,
+      "eval_ORGANIZATION_f1": 0.7974683544303797,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.6941176470588235,
-      "eval_ORGANIZATION_recall": 0.7866666666666666,
-      "eval_PERSON_f1": 0.9047619047619047,
+      "eval_ORGANIZATION_precision": 0.7590361445783133,
+      "eval_ORGANIZATION_recall": 0.84,
+      "eval_PERSON_f1": 0.8940397350993377,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8807947019867549,
-      "eval_PERSON_recall": 0.9300699300699301,
-      "eval_QUANTITY_f1": 0.7058823529411765,
+      "eval_PERSON_precision": 0.8490566037735849,
+      "eval_PERSON_recall": 0.9440559440559441,
+      "eval_QUANTITY_f1": 0.7307692307692308,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.6428571428571429,
-      "eval_QUANTITY_recall": 0.782608695652174,
-      "eval_TIME_f1": 0.830188679245283,
+      "eval_QUANTITY_precision": 0.6551724137931034,
+      "eval_QUANTITY_recall": 0.8260869565217391,
+      "eval_TIME_f1": 0.8,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8148148148148148,
+      "eval_TIME_precision": 0.7586206896551724,
       "eval_TIME_recall": 0.8461538461538461,
-      "eval_loss": 0.2894110381603241,
-      "eval_overall_accuracy": 0.9577182887165374,
-      "eval_overall_f1": 0.8177905308464849,
-      "eval_overall_precision": 0.7960893854748603,
-      "eval_overall_recall": 0.8407079646017699,
-      "eval_runtime": 0.9338,
-      "eval_samples_per_second": 200.266,
-      "eval_steps_per_second": 3.213,
+      "eval_loss": 0.32485198974609375,
+      "eval_overall_accuracy": 0.9597197898423818,
+      "eval_overall_f1": 0.83008356545961,
+      "eval_overall_precision": 0.7862796833773087,
+      "eval_overall_recall": 0.8790560471976401,
+      "eval_runtime": 0.3689,
+      "eval_samples_per_second": 506.94,
+      "eval_steps_per_second": 8.133,
       "step": 2226
     },
     {
       "epoch": 22.0,
-      "grad_norm": 0.001287546823732555,
+      "grad_norm": 0.03123115934431553,
       "learning_rate": 3.9000000000000006e-05,
-      "loss": 0.004,
+      "loss": 0.0037,
       "step": 2332
     },
     {
       "epoch": 22.0,
-      "eval_LOCATION_f1": 0.8266666666666667,
+      "eval_LOCATION_f1": 0.7972972972972973,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7948717948717948,
-      "eval_LOCATION_recall": 0.8611111111111112,
-      "eval_ORGANIZATION_f1": 0.7051282051282051,
+      "eval_LOCATION_precision": 0.7763157894736842,
+      "eval_LOCATION_recall": 0.8194444444444444,
+      "eval_ORGANIZATION_f1": 0.8263473053892216,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.6790123456790124,
-      "eval_ORGANIZATION_recall": 0.7333333333333333,
-      "eval_PERSON_f1": 0.8986486486486487,
+      "eval_ORGANIZATION_precision": 0.75,
+      "eval_ORGANIZATION_recall": 0.92,
+      "eval_PERSON_f1": 0.8910891089108911,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.869281045751634,
-      "eval_PERSON_recall": 0.9300699300699301,
-      "eval_QUANTITY_f1": 0.6792452830188679,
+      "eval_PERSON_precision": 0.84375,
+      "eval_PERSON_recall": 0.9440559440559441,
+      "eval_QUANTITY_f1": 0.7307692307692308,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.6,
-      "eval_QUANTITY_recall": 0.782608695652174,
-      "eval_TIME_f1": 0.8461538461538461,
+      "eval_QUANTITY_precision": 0.6551724137931034,
+      "eval_QUANTITY_recall": 0.8260869565217391,
+      "eval_TIME_f1": 0.8,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8461538461538461,
+      "eval_TIME_precision": 0.7586206896551724,
       "eval_TIME_recall": 0.8461538461538461,
-      "eval_loss": 0.29990673065185547,
-      "eval_overall_accuracy": 0.9554665999499625,
-      "eval_overall_f1": 0.8203677510608203,
-      "eval_overall_precision": 0.7880434782608695,
-      "eval_overall_recall": 0.855457227138643,
-      "eval_runtime": 0.9065,
-      "eval_samples_per_second": 206.288,
-      "eval_steps_per_second": 3.309,
+      "eval_loss": 0.3308863937854767,
+      "eval_overall_accuracy": 0.9602201651238429,
+      "eval_overall_f1": 0.8386206896551724,
+      "eval_overall_precision": 0.7875647668393783,
+      "eval_overall_recall": 0.8967551622418879,
+      "eval_runtime": 0.3591,
+      "eval_samples_per_second": 520.739,
+      "eval_steps_per_second": 8.354,
       "step": 2332
     },
     {
       "epoch": 23.0,
-      "grad_norm": 0.01974656991660595,
+      "grad_norm": 0.8348985314369202,
       "learning_rate": 3.85e-05,
-      "loss": 0.003,
+      "loss": 0.0015,
       "step": 2438
     },
     {
       "epoch": 23.0,
-      "eval_LOCATION_f1": 0.8211920529801324,
+      "eval_LOCATION_f1": 0.8079470198675496,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7848101265822784,
-      "eval_LOCATION_recall": 0.8611111111111112,
-      "eval_ORGANIZATION_f1": 0.8074534161490684,
+      "eval_LOCATION_precision": 0.7721518987341772,
+      "eval_LOCATION_recall": 0.8472222222222222,
+      "eval_ORGANIZATION_f1": 0.8176100628930819,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7558139534883721,
+      "eval_ORGANIZATION_precision": 0.7738095238095238,
       "eval_ORGANIZATION_recall": 0.8666666666666667,
-      "eval_PERSON_f1": 0.9084745762711864,
+      "eval_PERSON_f1": 0.8933333333333333,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.881578947368421,
+      "eval_PERSON_precision": 0.8535031847133758,
       "eval_PERSON_recall": 0.9370629370629371,
-      "eval_QUANTITY_f1": 0.6792452830188679,
+      "eval_QUANTITY_f1": 0.7692307692307693,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.6,
-      "eval_QUANTITY_recall": 0.782608695652174,
-      "eval_TIME_f1": 0.830188679245283,
+      "eval_QUANTITY_precision": 0.6896551724137931,
+      "eval_QUANTITY_recall": 0.8695652173913043,
+      "eval_TIME_f1": 0.8,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8148148148148148,
+      "eval_TIME_precision": 0.7586206896551724,
       "eval_TIME_recall": 0.8461538461538461,
-      "eval_loss": 0.3059156537055969,
-      "eval_overall_accuracy": 0.9622216662496873,
-      "eval_overall_f1": 0.8443197755960731,
-      "eval_overall_precision": 0.8048128342245989,
-      "eval_overall_recall": 0.887905604719764,
-      "eval_runtime": 0.9125,
-      "eval_samples_per_second": 204.922,
-      "eval_steps_per_second": 3.288,
+      "eval_loss": 0.32765087485313416,
+      "eval_overall_accuracy": 0.9609707280460346,
+      "eval_overall_f1": 0.8423988842398884,
+      "eval_overall_precision": 0.798941798941799,
+      "eval_overall_recall": 0.8908554572271387,
+      "eval_runtime": 0.3569,
+      "eval_samples_per_second": 523.957,
+      "eval_steps_per_second": 8.406,
       "step": 2438
     },
     {
       "epoch": 24.0,
-      "grad_norm": 2.60111141204834,
+      "grad_norm": 0.006436940282583237,
       "learning_rate": 3.8e-05,
-      "loss": 0.0038,
+      "loss": 0.0027,
       "step": 2544
     },
     {
       "epoch": 24.0,
-      "eval_LOCATION_f1": 0.8133333333333332,
+      "eval_LOCATION_f1": 0.7837837837837838,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.782051282051282,
-      "eval_LOCATION_recall": 0.8472222222222222,
-      "eval_ORGANIZATION_f1": 0.7471264367816093,
+      "eval_LOCATION_precision": 0.7631578947368421,
+      "eval_LOCATION_recall": 0.8055555555555556,
+      "eval_ORGANIZATION_f1": 0.8076923076923077,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.6565656565656566,
-      "eval_ORGANIZATION_recall": 0.8666666666666667,
-      "eval_PERSON_f1": 0.9152542372881356,
+      "eval_ORGANIZATION_precision": 0.7777777777777778,
+      "eval_ORGANIZATION_recall": 0.84,
+      "eval_PERSON_f1": 0.8933333333333333,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8881578947368421,
-      "eval_PERSON_recall": 0.9440559440559441,
-      "eval_QUANTITY_f1": 0.6792452830188679,
+      "eval_PERSON_precision": 0.8535031847133758,
+      "eval_PERSON_recall": 0.9370629370629371,
+      "eval_QUANTITY_f1": 0.72,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.6,
+      "eval_QUANTITY_precision": 0.6666666666666666,
       "eval_QUANTITY_recall": 0.782608695652174,
-      "eval_TIME_f1": 0.8,
+      "eval_TIME_f1": 0.830188679245283,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.7586206896551724,
+      "eval_TIME_precision": 0.8148148148148148,
       "eval_TIME_recall": 0.8461538461538461,
-      "eval_loss": 0.3071446716785431,
-      "eval_overall_accuracy": 0.9582186639979985,
-      "eval_overall_f1": 0.828060522696011,
-      "eval_overall_precision": 0.7757731958762887,
-      "eval_overall_recall": 0.887905604719764,
-      "eval_runtime": 0.9115,
-      "eval_samples_per_second": 205.145,
-      "eval_steps_per_second": 3.291,
+      "eval_loss": 0.3232249617576599,
+      "eval_overall_accuracy": 0.9609707280460346,
+      "eval_overall_f1": 0.8345120226308345,
+      "eval_overall_precision": 0.8016304347826086,
+      "eval_overall_recall": 0.8702064896755162,
+      "eval_runtime": 0.3705,
+      "eval_samples_per_second": 504.719,
+      "eval_steps_per_second": 8.097,
       "step": 2544
     },
     {
       "epoch": 25.0,
-      "grad_norm": 5.347325801849365,
+      "grad_norm": 0.9054812788963318,
       "learning_rate": 3.7500000000000003e-05,
-      "loss": 0.0039,
+      "loss": 0.0035,
       "step": 2650
     },
     {
       "epoch": 25.0,
-      "eval_LOCATION_f1": 0.7792207792207793,
+      "eval_LOCATION_f1": 0.8129032258064516,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7317073170731707,
-      "eval_LOCATION_recall": 0.8333333333333334,
-      "eval_ORGANIZATION_f1": 0.729559748427673,
+      "eval_LOCATION_precision": 0.7590361445783133,
+      "eval_LOCATION_recall": 0.875,
+      "eval_ORGANIZATION_f1": 0.8227848101265823,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.6904761904761905,
-      "eval_ORGANIZATION_recall": 0.7733333333333333,
-      "eval_PERSON_f1": 0.8993288590604027,
+      "eval_ORGANIZATION_precision": 0.7831325301204819,
+      "eval_ORGANIZATION_recall": 0.8666666666666667,
+      "eval_PERSON_f1": 0.8926174496644295,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.864516129032258,
-      "eval_PERSON_recall": 0.9370629370629371,
-      "eval_QUANTITY_f1": 0.6792452830188679,
+      "eval_PERSON_precision": 0.8580645161290322,
+      "eval_PERSON_recall": 0.9300699300699301,
+      "eval_QUANTITY_f1": 0.6428571428571429,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.6,
+      "eval_QUANTITY_precision": 0.5454545454545454,
       "eval_QUANTITY_recall": 0.782608695652174,
-      "eval_TIME_f1": 0.8148148148148148,
+      "eval_TIME_f1": 0.8,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.7857142857142857,
+      "eval_TIME_precision": 0.7586206896551724,
       "eval_TIME_recall": 0.8461538461538461,
-      "eval_loss": 0.3066267967224121,
-      "eval_overall_accuracy": 0.9572179134350763,
-      "eval_overall_f1": 0.8133704735376045,
-      "eval_overall_precision": 0.7704485488126649,
-      "eval_overall_recall": 0.8613569321533924,
-      "eval_runtime": 0.909,
-      "eval_samples_per_second": 205.719,
-      "eval_steps_per_second": 3.3,
+      "eval_loss": 0.35135143995285034,
+      "eval_overall_accuracy": 0.957968476357268,
+      "eval_overall_f1": 0.8337950138504155,
+      "eval_overall_precision": 0.7859007832898173,
+      "eval_overall_recall": 0.887905604719764,
+      "eval_runtime": 0.3594,
+      "eval_samples_per_second": 520.314,
+      "eval_steps_per_second": 8.347,
       "step": 2650
     },
     {
       "epoch": 26.0,
-      "grad_norm": 0.00547376973554492,
+      "grad_norm": 0.010295592248439789,
       "learning_rate": 3.7e-05,
-      "loss": 0.0031,
+      "loss": 0.0028,
       "step": 2756
     },
     {
       "epoch": 26.0,
-      "eval_LOCATION_f1": 0.8137931034482759,
+      "eval_LOCATION_f1": 0.8129032258064516,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.8082191780821918,
-      "eval_LOCATION_recall": 0.8194444444444444,
-      "eval_ORGANIZATION_f1": 0.7701863354037267,
+      "eval_LOCATION_precision": 0.7590361445783133,
+      "eval_LOCATION_recall": 0.875,
+      "eval_ORGANIZATION_f1": 0.7831325301204819,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7209302325581395,
-      "eval_ORGANIZATION_recall": 0.8266666666666667,
-      "eval_PERSON_f1": 0.8837209302325583,
+      "eval_ORGANIZATION_precision": 0.7142857142857143,
+      "eval_ORGANIZATION_recall": 0.8666666666666667,
+      "eval_PERSON_f1": 0.8979591836734694,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8417721518987342,
-      "eval_PERSON_recall": 0.9300699300699301,
-      "eval_QUANTITY_f1": 0.6923076923076923,
+      "eval_PERSON_precision": 0.8741721854304636,
+      "eval_PERSON_recall": 0.9230769230769231,
+      "eval_QUANTITY_f1": 0.7058823529411765,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.6206896551724138,
+      "eval_QUANTITY_precision": 0.6428571428571429,
       "eval_QUANTITY_recall": 0.782608695652174,
-      "eval_TIME_f1": 0.8679245283018868,
+      "eval_TIME_f1": 0.830188679245283,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8518518518518519,
-      "eval_TIME_recall": 0.8846153846153846,
-      "eval_loss": 0.3245824873447418,
-      "eval_overall_accuracy": 0.9597197898423818,
-      "eval_overall_f1": 0.8286516853932584,
-      "eval_overall_precision": 0.7908847184986595,
-      "eval_overall_recall": 0.8702064896755162,
-      "eval_runtime": 0.9241,
-      "eval_samples_per_second": 202.35,
-      "eval_steps_per_second": 3.246,
+      "eval_TIME_precision": 0.8148148148148148,
+      "eval_TIME_recall": 0.8461538461538461,
+      "eval_loss": 0.3200477361679077,
+      "eval_overall_accuracy": 0.9602201651238429,
+      "eval_overall_f1": 0.8344923504867873,
+      "eval_overall_precision": 0.7894736842105263,
+      "eval_overall_recall": 0.8849557522123894,
+      "eval_runtime": 0.3624,
+      "eval_samples_per_second": 516.068,
+      "eval_steps_per_second": 8.279,
       "step": 2756
     },
     {
       "epoch": 27.0,
-      "grad_norm": 0.009228230454027653,
+      "grad_norm": 0.0034800188150256872,
       "learning_rate": 3.65e-05,
-      "loss": 0.0026,
+      "loss": 0.0023,
       "step": 2862
     },
     {
       "epoch": 27.0,
-      "eval_LOCATION_f1": 0.8219178082191781,
+      "eval_LOCATION_f1": 0.8108108108108109,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.8108108108108109,
+      "eval_LOCATION_precision": 0.7894736842105263,
       "eval_LOCATION_recall": 0.8333333333333334,
-      "eval_ORGANIZATION_f1": 0.7407407407407408,
+      "eval_ORGANIZATION_f1": 0.7831325301204819,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.6896551724137931,
-      "eval_ORGANIZATION_recall": 0.8,
-      "eval_PERSON_f1": 0.903654485049834,
+      "eval_ORGANIZATION_precision": 0.7142857142857143,
+      "eval_ORGANIZATION_recall": 0.8666666666666667,
+      "eval_PERSON_f1": 0.8837209302325583,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8607594936708861,
-      "eval_PERSON_recall": 0.951048951048951,
-      "eval_QUANTITY_f1": 0.7169811320754716,
+      "eval_PERSON_precision": 0.8417721518987342,
+      "eval_PERSON_recall": 0.9300699300699301,
+      "eval_QUANTITY_f1": 0.6538461538461539,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.6333333333333333,
-      "eval_QUANTITY_recall": 0.8260869565217391,
-      "eval_TIME_f1": 0.8518518518518519,
+      "eval_QUANTITY_precision": 0.5862068965517241,
+      "eval_QUANTITY_recall": 0.7391304347826086,
+      "eval_TIME_f1": 0.8148148148148148,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8214285714285714,
-      "eval_TIME_recall": 0.8846153846153846,
-      "eval_loss": 0.32465118169784546,
-      "eval_overall_accuracy": 0.9584688516387291,
-      "eval_overall_f1": 0.8324022346368715,
-      "eval_overall_precision": 0.7904509283819628,
-      "eval_overall_recall": 0.8790560471976401,
-      "eval_runtime": 0.8984,
-      "eval_samples_per_second": 208.15,
-      "eval_steps_per_second": 3.339,
+      "eval_TIME_precision": 0.7857142857142857,
+      "eval_TIME_recall": 0.8461538461538461,
+      "eval_loss": 0.3428550064563751,
+      "eval_overall_accuracy": 0.9569677257943457,
+      "eval_overall_f1": 0.8238557558945908,
+      "eval_overall_precision": 0.7774869109947644,
+      "eval_overall_recall": 0.8761061946902655,
+      "eval_runtime": 0.3592,
+      "eval_samples_per_second": 520.564,
+      "eval_steps_per_second": 8.351,
       "step": 2862
     },
     {
       "epoch": 28.0,
-      "grad_norm": 3.7562034130096436,
+      "grad_norm": 0.001987306633964181,
       "learning_rate": 3.6e-05,
-      "loss": 0.0041,
+      "loss": 0.0031,
       "step": 2968
     },
     {
       "epoch": 28.0,
-      "eval_LOCATION_f1": 0.8082191780821918,
+      "eval_LOCATION_f1": 0.8026315789473685,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7972972972972973,
-      "eval_LOCATION_recall": 0.8194444444444444,
-      "eval_ORGANIZATION_f1": 0.7469879518072289,
+      "eval_LOCATION_precision": 0.7625,
+      "eval_LOCATION_recall": 0.8472222222222222,
+      "eval_ORGANIZATION_f1": 0.7777777777777777,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.6813186813186813,
-      "eval_ORGANIZATION_recall": 0.8266666666666667,
-      "eval_PERSON_f1": 0.9084745762711864,
+      "eval_ORGANIZATION_precision": 0.7241379310344828,
+      "eval_ORGANIZATION_recall": 0.84,
+      "eval_PERSON_f1": 0.8983606557377047,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.881578947368421,
-      "eval_PERSON_recall": 0.9370629370629371,
-      "eval_QUANTITY_f1": 0.6923076923076923,
+      "eval_PERSON_precision": 0.845679012345679,
+      "eval_PERSON_recall": 0.958041958041958,
+      "eval_QUANTITY_f1": 0.6666666666666666,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.6206896551724138,
-      "eval_QUANTITY_recall": 0.782608695652174,
-      "eval_TIME_f1": 0.830188679245283,
+      "eval_QUANTITY_precision": 0.6071428571428571,
+      "eval_QUANTITY_recall": 0.7391304347826086,
+      "eval_TIME_f1": 0.8070175438596492,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8148148148148148,
-      "eval_TIME_recall": 0.8461538461538461,
-      "eval_loss": 0.3042037785053253,
-      "eval_overall_accuracy": 0.9592194145609206,
-      "eval_overall_f1": 0.8286516853932584,
-      "eval_overall_precision": 0.7908847184986595,
-      "eval_overall_recall": 0.8702064896755162,
-      "eval_runtime": 0.9162,
-      "eval_samples_per_second": 204.094,
-      "eval_steps_per_second": 3.274,
+      "eval_TIME_precision": 0.7419354838709677,
+      "eval_TIME_recall": 0.8846153846153846,
+      "eval_loss": 0.3218757212162018,
+      "eval_overall_accuracy": 0.9599699774831123,
+      "eval_overall_f1": 0.828060522696011,
+      "eval_overall_precision": 0.7757731958762887,
+      "eval_overall_recall": 0.887905604719764,
+      "eval_runtime": 0.3703,
+      "eval_samples_per_second": 504.962,
+      "eval_steps_per_second": 8.101,
       "step": 2968
     },
     {
       "epoch": 29.0,
-      "grad_norm": 5.308003902435303,
+      "grad_norm": 0.021353095769882202,
       "learning_rate": 3.55e-05,
-      "loss": 0.0034,
+      "loss": 0.0028,
       "step": 3074
     },
     {
       "epoch": 29.0,
-      "eval_LOCATION_f1": 0.8413793103448277,
+      "eval_LOCATION_f1": 0.8057553956834532,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.8356164383561644,
-      "eval_LOCATION_recall": 0.8472222222222222,
-      "eval_ORGANIZATION_f1": 0.7975460122699386,
+      "eval_LOCATION_precision": 0.835820895522388,
+      "eval_LOCATION_recall": 0.7777777777777778,
+      "eval_ORGANIZATION_f1": 0.8076923076923077,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7386363636363636,
-      "eval_ORGANIZATION_recall": 0.8666666666666667,
-      "eval_PERSON_f1": 0.9146757679180888,
+      "eval_ORGANIZATION_precision": 0.7777777777777778,
+      "eval_ORGANIZATION_recall": 0.84,
+      "eval_PERSON_f1": 0.8800000000000001,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8933333333333333,
-      "eval_PERSON_recall": 0.9370629370629371,
-      "eval_QUANTITY_f1": 0.6923076923076923,
+      "eval_PERSON_precision": 0.8407643312101911,
+      "eval_PERSON_recall": 0.9230769230769231,
+      "eval_QUANTITY_f1": 0.6415094339622641,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.6206896551724138,
-      "eval_QUANTITY_recall": 0.782608695652174,
-      "eval_TIME_f1": 0.8461538461538461,
+      "eval_QUANTITY_precision": 0.5666666666666667,
+      "eval_QUANTITY_recall": 0.7391304347826086,
+      "eval_TIME_f1": 0.830188679245283,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8461538461538461,
+      "eval_TIME_precision": 0.8148148148148148,
       "eval_TIME_recall": 0.8461538461538461,
-      "eval_loss": 0.3232886493206024,
-      "eval_overall_accuracy": 0.9609707280460346,
-      "eval_overall_f1": 0.851063829787234,
-      "eval_overall_precision": 0.819672131147541,
-      "eval_overall_recall": 0.8849557522123894,
-      "eval_runtime": 0.9849,
-      "eval_samples_per_second": 189.867,
-      "eval_steps_per_second": 3.046,
+      "eval_loss": 0.35811248421669006,
+      "eval_overall_accuracy": 0.9562171628721541,
+      "eval_overall_f1": 0.8273894436519258,
+      "eval_overall_precision": 0.8011049723756906,
+      "eval_overall_recall": 0.855457227138643,
+      "eval_runtime": 0.3735,
+      "eval_samples_per_second": 500.719,
+      "eval_steps_per_second": 8.033,
       "step": 3074
     },
     {
       "epoch": 30.0,
-      "grad_norm": 0.0021002369467169046,
+      "grad_norm": 0.0009840091224759817,
       "learning_rate": 3.5e-05,
-      "loss": 0.0027,
+      "loss": 0.0031,
       "step": 3180
     },
     {
       "epoch": 30.0,
-      "eval_LOCATION_f1": 0.8356164383561645,
+      "eval_LOCATION_f1": 0.8,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.8243243243243243,
-      "eval_LOCATION_recall": 0.8472222222222222,
-      "eval_ORGANIZATION_f1": 0.732919254658385,
+      "eval_LOCATION_precision": 0.7692307692307693,
+      "eval_LOCATION_recall": 0.8333333333333334,
+      "eval_ORGANIZATION_f1": 0.830188679245283,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.686046511627907,
-      "eval_ORGANIZATION_recall": 0.7866666666666666,
-      "eval_PERSON_f1": 0.9121621621621621,
+      "eval_ORGANIZATION_precision": 0.7857142857142857,
+      "eval_ORGANIZATION_recall": 0.88,
+      "eval_PERSON_f1": 0.9163879598662208,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8823529411764706,
-      "eval_PERSON_recall": 0.9440559440559441,
-      "eval_QUANTITY_f1": 0.7272727272727273,
+      "eval_PERSON_precision": 0.8782051282051282,
+      "eval_PERSON_recall": 0.958041958041958,
+      "eval_QUANTITY_f1": 0.7307692307692308,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.625,
-      "eval_QUANTITY_recall": 0.8695652173913043,
-      "eval_TIME_f1": 0.7586206896551724,
+      "eval_QUANTITY_precision": 0.6551724137931034,
+      "eval_QUANTITY_recall": 0.8260869565217391,
+      "eval_TIME_f1": 0.8,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.6875,
+      "eval_TIME_precision": 0.7586206896551724,
       "eval_TIME_recall": 0.8461538461538461,
-      "eval_loss": 0.34255892038345337,
-      "eval_overall_accuracy": 0.9577182887165374,
-      "eval_overall_f1": 0.8296089385474861,
-      "eval_overall_precision": 0.7877984084880637,
-      "eval_overall_recall": 0.8761061946902655,
-      "eval_runtime": 0.9051,
-      "eval_samples_per_second": 206.618,
-      "eval_steps_per_second": 3.315,
+      "eval_loss": 0.3134097754955292,
+      "eval_overall_accuracy": 0.9619714786089567,
+      "eval_overall_f1": 0.8503496503496504,
+      "eval_overall_precision": 0.8085106382978723,
+      "eval_overall_recall": 0.8967551622418879,
+      "eval_runtime": 0.3742,
+      "eval_samples_per_second": 499.673,
+      "eval_steps_per_second": 8.016,
       "step": 3180
     },
     {
       "epoch": 31.0,
-      "grad_norm": 0.00587083725258708,
+      "grad_norm": 1.1641939878463745,
       "learning_rate": 3.45e-05,
-      "loss": 0.0049,
+      "loss": 0.0035,
       "step": 3286
     },
     {
       "epoch": 31.0,
-      "eval_LOCATION_f1": 0.8219178082191781,
+      "eval_LOCATION_f1": 0.751592356687898,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.8108108108108109,
-      "eval_LOCATION_recall": 0.8333333333333334,
-      "eval_ORGANIZATION_f1": 0.7305389221556886,
+      "eval_LOCATION_precision": 0.6941176470588235,
+      "eval_LOCATION_recall": 0.8194444444444444,
+      "eval_ORGANIZATION_f1": 0.8387096774193549,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.6630434782608695,
-      "eval_ORGANIZATION_recall": 0.8133333333333334,
-      "eval_PERSON_f1": 0.9152542372881356,
+      "eval_ORGANIZATION_precision": 0.8125,
+      "eval_ORGANIZATION_recall": 0.8666666666666667,
+      "eval_PERSON_f1": 0.9133333333333333,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8881578947368421,
-      "eval_PERSON_recall": 0.9440559440559441,
-      "eval_QUANTITY_f1": 0.6923076923076923,
+      "eval_PERSON_precision": 0.8726114649681529,
+      "eval_PERSON_recall": 0.958041958041958,
+      "eval_QUANTITY_f1": 0.6792452830188679,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.6206896551724138,
+      "eval_QUANTITY_precision": 0.6,
       "eval_QUANTITY_recall": 0.782608695652174,
-      "eval_TIME_f1": 0.8461538461538461,
+      "eval_TIME_f1": 0.8148148148148148,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8461538461538461,
+      "eval_TIME_precision": 0.7857142857142857,
       "eval_TIME_recall": 0.8461538461538461,
-      "eval_loss": 0.31757453083992004,
-      "eval_overall_accuracy": 0.9584688516387291,
-      "eval_overall_f1": 0.8314606741573034,
-      "eval_overall_precision": 0.7935656836461126,
-      "eval_overall_recall": 0.8731563421828908,
-      "eval_runtime": 0.9036,
-      "eval_samples_per_second": 206.959,
-      "eval_steps_per_second": 3.32,
+      "eval_loss": 0.31325557827949524,
+      "eval_overall_accuracy": 0.9614711033274956,
+      "eval_overall_f1": 0.8372739916550764,
+      "eval_overall_precision": 0.7921052631578948,
+      "eval_overall_recall": 0.887905604719764,
+      "eval_runtime": 0.3607,
+      "eval_samples_per_second": 518.406,
+      "eval_steps_per_second": 8.317,
       "step": 3286
     },
     {
       "epoch": 32.0,
-      "grad_norm": 0.03227420523762703,
+      "grad_norm": 0.0497661791741848,
       "learning_rate": 3.4000000000000007e-05,
-      "loss": 0.0027,
+      "loss": 0.0028,
       "step": 3392
     },
     {
       "epoch": 32.0,
-      "eval_LOCATION_f1": 0.8243243243243243,
+      "eval_LOCATION_f1": 0.802721088435374,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.8026315789473685,
-      "eval_LOCATION_recall": 0.8472222222222222,
-      "eval_ORGANIZATION_f1": 0.7625,
+      "eval_LOCATION_precision": 0.7866666666666666,
+      "eval_LOCATION_recall": 0.8194444444444444,
+      "eval_ORGANIZATION_f1": 0.8214285714285714,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7176470588235294,
-      "eval_ORGANIZATION_recall": 0.8133333333333334,
-      "eval_PERSON_f1": 0.9215017064846417,
+      "eval_ORGANIZATION_precision": 0.7419354838709677,
+      "eval_ORGANIZATION_recall": 0.92,
+      "eval_PERSON_f1": 0.9,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.9,
+      "eval_PERSON_precision": 0.8598726114649682,
       "eval_PERSON_recall": 0.9440559440559441,
-      "eval_QUANTITY_f1": 0.7083333333333334,
+      "eval_QUANTITY_f1": 0.7755102040816326,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.68,
-      "eval_QUANTITY_recall": 0.7391304347826086,
-      "eval_TIME_f1": 0.8846153846153846,
+      "eval_QUANTITY_precision": 0.7307692307692307,
+      "eval_QUANTITY_recall": 0.8260869565217391,
+      "eval_TIME_f1": 0.8461538461538461,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8846153846153846,
-      "eval_TIME_recall": 0.8846153846153846,
-      "eval_loss": 0.29548466205596924,
-      "eval_overall_accuracy": 0.9627220415311484,
-      "eval_overall_f1": 0.8473609129814551,
-      "eval_overall_precision": 0.8204419889502762,
-      "eval_overall_recall": 0.8761061946902655,
-      "eval_runtime": 0.9155,
-      "eval_samples_per_second": 204.26,
-      "eval_steps_per_second": 3.277,
+      "eval_TIME_precision": 0.8461538461538461,
+      "eval_TIME_recall": 0.8461538461538461,
+      "eval_loss": 0.32023829221725464,
+      "eval_overall_accuracy": 0.961220915686765,
+      "eval_overall_f1": 0.8491620111731844,
+      "eval_overall_precision": 0.8063660477453581,
+      "eval_overall_recall": 0.8967551622418879,
+      "eval_runtime": 0.3634,
+      "eval_samples_per_second": 514.539,
+      "eval_steps_per_second": 8.255,
       "step": 3392
     },
     {
       "epoch": 33.0,
-      "grad_norm": 0.016770539805293083,
+      "grad_norm": 0.0015366391744464636,
       "learning_rate": 3.35e-05,
-      "loss": 0.0017,
+      "loss": 0.0013,
       "step": 3498
     },
     {
       "epoch": 33.0,
-      "eval_LOCATION_f1": 0.8,
+      "eval_LOCATION_f1": 0.8219178082191781,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7692307692307693,
+      "eval_LOCATION_precision": 0.8108108108108109,
       "eval_LOCATION_recall": 0.8333333333333334,
-      "eval_ORGANIZATION_f1": 0.7974683544303797,
+      "eval_ORGANIZATION_f1": 0.7898089171974522,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7590361445783133,
-      "eval_ORGANIZATION_recall": 0.84,
-      "eval_PERSON_f1": 0.9054054054054055,
+      "eval_ORGANIZATION_precision": 0.7560975609756098,
+      "eval_ORGANIZATION_recall": 0.8266666666666667,
+      "eval_PERSON_f1": 0.8926174496644295,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8758169934640523,
-      "eval_PERSON_recall": 0.9370629370629371,
-      "eval_QUANTITY_f1": 0.6666666666666667,
+      "eval_PERSON_precision": 0.8580645161290322,
+      "eval_PERSON_recall": 0.9300699300699301,
+      "eval_QUANTITY_f1": 0.6923076923076923,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.5806451612903226,
+      "eval_QUANTITY_precision": 0.6206896551724138,
       "eval_QUANTITY_recall": 0.782608695652174,
-      "eval_TIME_f1": 0.830188679245283,
+      "eval_TIME_f1": 0.8461538461538461,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8148148148148148,
+      "eval_TIME_precision": 0.8461538461538461,
       "eval_TIME_recall": 0.8461538461538461,
-      "eval_loss": 0.3393411338329315,
-      "eval_overall_accuracy": 0.9592194145609206,
-      "eval_overall_f1": 0.8354430379746836,
-      "eval_overall_precision": 0.7983870967741935,
-      "eval_overall_recall": 0.8761061946902655,
-      "eval_runtime": 0.9082,
-      "eval_samples_per_second": 205.908,
-      "eval_steps_per_second": 3.303,
+      "eval_loss": 0.3315739035606384,
+      "eval_overall_accuracy": 0.961220915686765,
+      "eval_overall_f1": 0.8368794326241135,
+      "eval_overall_precision": 0.8060109289617486,
+      "eval_overall_recall": 0.8702064896755162,
+      "eval_runtime": 0.3575,
+      "eval_samples_per_second": 523.126,
+      "eval_steps_per_second": 8.392,
       "step": 3498
     },
     {
       "epoch": 34.0,
-      "grad_norm": 0.003996172454208136,
+      "grad_norm": 0.05237346515059471,
       "learning_rate": 3.3e-05,
-      "loss": 0.0017,
+      "loss": 0.0018,
       "step": 3604
     },
     {
       "epoch": 34.0,
-      "eval_LOCATION_f1": 0.8243243243243243,
+      "eval_LOCATION_f1": 0.8571428571428572,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.8026315789473685,
-      "eval_LOCATION_recall": 0.8472222222222222,
-      "eval_ORGANIZATION_f1": 0.8227848101265823,
+      "eval_LOCATION_precision": 0.84,
+      "eval_LOCATION_recall": 0.875,
+      "eval_ORGANIZATION_f1": 0.8048780487804879,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7831325301204819,
-      "eval_ORGANIZATION_recall": 0.8666666666666667,
-      "eval_PERSON_f1": 0.910958904109589,
+      "eval_ORGANIZATION_precision": 0.7415730337078652,
+      "eval_ORGANIZATION_recall": 0.88,
+      "eval_PERSON_f1": 0.8970099667774087,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8926174496644296,
-      "eval_PERSON_recall": 0.9300699300699301,
-      "eval_QUANTITY_f1": 0.7555555555555555,
+      "eval_PERSON_precision": 0.8544303797468354,
+      "eval_PERSON_recall": 0.9440559440559441,
+      "eval_QUANTITY_f1": 0.7169811320754716,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.7727272727272727,
-      "eval_QUANTITY_recall": 0.7391304347826086,
-      "eval_TIME_f1": 0.8461538461538461,
+      "eval_QUANTITY_precision": 0.6333333333333333,
+      "eval_QUANTITY_recall": 0.8260869565217391,
+      "eval_TIME_f1": 0.830188679245283,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8461538461538461,
+      "eval_TIME_precision": 0.8148148148148148,
       "eval_TIME_recall": 0.8461538461538461,
-      "eval_loss": 0.3279131352901459,
-      "eval_overall_accuracy": 0.9644733550162622,
-      "eval_overall_f1": 0.8575539568345324,
-      "eval_overall_precision": 0.8370786516853933,
-      "eval_overall_recall": 0.8790560471976401,
-      "eval_runtime": 0.9098,
-      "eval_samples_per_second": 205.55,
-      "eval_steps_per_second": 3.298,
+      "eval_loss": 0.36692577600479126,
+      "eval_overall_accuracy": 0.9622216662496873,
+      "eval_overall_f1": 0.8495821727019498,
+      "eval_overall_precision": 0.8047493403693932,
+      "eval_overall_recall": 0.8997050147492626,
+      "eval_runtime": 0.3644,
+      "eval_samples_per_second": 513.118,
+      "eval_steps_per_second": 8.232,
       "step": 3604
     },
     {
       "epoch": 35.0,
-      "grad_norm": 0.0200356375426054,
+      "grad_norm": 0.002315772697329521,
       "learning_rate": 3.2500000000000004e-05,
-      "loss": 0.002,
+      "loss": 0.0022,
       "step": 3710
     },
     {
       "epoch": 35.0,
-      "eval_LOCATION_f1": 0.8108108108108109,
+      "eval_LOCATION_f1": 0.8378378378378377,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7894736842105263,
-      "eval_LOCATION_recall": 0.8333333333333334,
-      "eval_ORGANIZATION_f1": 0.8,
+      "eval_LOCATION_precision": 0.8157894736842105,
+      "eval_LOCATION_recall": 0.8611111111111112,
+      "eval_ORGANIZATION_f1": 0.8198757763975156,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7333333333333333,
+      "eval_ORGANIZATION_precision": 0.7674418604651163,
       "eval_ORGANIZATION_recall": 0.88,
-      "eval_PERSON_f1": 0.9016949152542374,
+      "eval_PERSON_f1": 0.91156462585034,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.875,
-      "eval_PERSON_recall": 0.9300699300699301,
-      "eval_QUANTITY_f1": 0.6923076923076923,
+      "eval_PERSON_precision": 0.8874172185430463,
+      "eval_PERSON_recall": 0.9370629370629371,
+      "eval_QUANTITY_f1": 0.7307692307692308,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.6206896551724138,
-      "eval_QUANTITY_recall": 0.782608695652174,
-      "eval_TIME_f1": 0.8461538461538461,
+      "eval_QUANTITY_precision": 0.6551724137931034,
+      "eval_QUANTITY_recall": 0.8260869565217391,
+      "eval_TIME_f1": 0.9230769230769231,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8461538461538461,
-      "eval_TIME_recall": 0.8461538461538461,
-      "eval_loss": 0.33025962114334106,
-      "eval_overall_accuracy": 0.9597197898423818,
-      "eval_overall_f1": 0.8398876404494382,
-      "eval_overall_precision": 0.8016085790884718,
-      "eval_overall_recall": 0.8820058997050148,
-      "eval_runtime": 0.9088,
-      "eval_samples_per_second": 205.758,
-      "eval_steps_per_second": 3.301,
+      "eval_TIME_precision": 0.9230769230769231,
+      "eval_TIME_recall": 0.9230769230769231,
+      "eval_loss": 0.3101156949996948,
+      "eval_overall_accuracy": 0.9649737302977233,
+      "eval_overall_f1": 0.8628005657708627,
+      "eval_overall_precision": 0.8288043478260869,
+      "eval_overall_recall": 0.8997050147492626,
+      "eval_runtime": 0.3638,
+      "eval_samples_per_second": 514.005,
+      "eval_steps_per_second": 8.246,
       "step": 3710
     },
     {
       "epoch": 36.0,
-      "grad_norm": 0.04783305525779724,
+      "grad_norm": 0.0013785570627078414,
       "learning_rate": 3.2000000000000005e-05,
-      "loss": 0.0012,
+      "loss": 0.002,
       "step": 3816
     },
     {
       "epoch": 36.0,
-      "eval_LOCATION_f1": 0.8344370860927153,
+      "eval_LOCATION_f1": 0.8266666666666667,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7974683544303798,
-      "eval_LOCATION_recall": 0.875,
-      "eval_ORGANIZATION_f1": 0.7875000000000001,
+      "eval_LOCATION_precision": 0.7948717948717948,
+      "eval_LOCATION_recall": 0.8611111111111112,
+      "eval_ORGANIZATION_f1": 0.8198757763975156,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7411764705882353,
-      "eval_ORGANIZATION_recall": 0.84,
-      "eval_PERSON_f1": 0.9010238907849829,
+      "eval_ORGANIZATION_precision": 0.7674418604651163,
+      "eval_ORGANIZATION_recall": 0.88,
+      "eval_PERSON_f1": 0.8933333333333333,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.88,
-      "eval_PERSON_recall": 0.9230769230769231,
-      "eval_QUANTITY_f1": 0.72,
+      "eval_PERSON_precision": 0.8535031847133758,
+      "eval_PERSON_recall": 0.9370629370629371,
+      "eval_QUANTITY_f1": 0.6923076923076923,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.6666666666666666,
+      "eval_QUANTITY_precision": 0.6206896551724138,
       "eval_QUANTITY_recall": 0.782608695652174,
-      "eval_TIME_f1": 0.8461538461538461,
+      "eval_TIME_f1": 0.830188679245283,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8461538461538461,
+      "eval_TIME_precision": 0.8148148148148148,
       "eval_TIME_recall": 0.8461538461538461,
-      "eval_loss": 0.33322834968566895,
-      "eval_overall_accuracy": 0.9637227920940705,
-      "eval_overall_f1": 0.8441926345609065,
-      "eval_overall_precision": 0.8119891008174387,
-      "eval_overall_recall": 0.8790560471976401,
-      "eval_runtime": 0.9959,
-      "eval_samples_per_second": 187.774,
-      "eval_steps_per_second": 3.012,
+      "eval_loss": 0.3521296977996826,
+      "eval_overall_accuracy": 0.9592194145609206,
+      "eval_overall_f1": 0.8435754189944135,
+      "eval_overall_precision": 0.8010610079575596,
+      "eval_overall_recall": 0.8908554572271387,
+      "eval_runtime": 0.3743,
+      "eval_samples_per_second": 499.566,
+      "eval_steps_per_second": 8.014,
       "step": 3816
     },
     {
       "epoch": 37.0,
-      "grad_norm": 0.0011351928114891052,
+      "grad_norm": 0.002585264155641198,
       "learning_rate": 3.15e-05,
-      "loss": 0.001,
+      "loss": 0.0019,
       "step": 3922
     },
     {
       "epoch": 37.0,
-      "eval_LOCATION_f1": 0.816326530612245,
+      "eval_LOCATION_f1": 0.8211920529801324,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.8,
-      "eval_LOCATION_recall": 0.8333333333333334,
-      "eval_ORGANIZATION_f1": 0.7710843373493976,
+      "eval_LOCATION_precision": 0.7848101265822784,
+      "eval_LOCATION_recall": 0.8611111111111112,
+      "eval_ORGANIZATION_f1": 0.8148148148148148,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7032967032967034,
-      "eval_ORGANIZATION_recall": 0.8533333333333334,
-      "eval_PERSON_f1": 0.910958904109589,
+      "eval_ORGANIZATION_precision": 0.7586206896551724,
+      "eval_ORGANIZATION_recall": 0.88,
+      "eval_PERSON_f1": 0.8956228956228957,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8926174496644296,
+      "eval_PERSON_precision": 0.8636363636363636,
       "eval_PERSON_recall": 0.9300699300699301,
-      "eval_QUANTITY_f1": 0.72,
+      "eval_QUANTITY_f1": 0.7307692307692308,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.6666666666666666,
-      "eval_QUANTITY_recall": 0.782608695652174,
-      "eval_TIME_f1": 0.8235294117647058,
+      "eval_QUANTITY_precision": 0.6551724137931034,
+      "eval_QUANTITY_recall": 0.8260869565217391,
+      "eval_TIME_f1": 0.8148148148148148,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.84,
-      "eval_TIME_recall": 0.8076923076923077,
-      "eval_loss": 0.33810704946517944,
-      "eval_overall_accuracy": 0.9604703527645734,
-      "eval_overall_f1": 0.8385269121813032,
-      "eval_overall_precision": 0.8065395095367848,
-      "eval_overall_recall": 0.8731563421828908,
-      "eval_runtime": 0.9038,
-      "eval_samples_per_second": 206.905,
-      "eval_steps_per_second": 3.319,
+      "eval_TIME_precision": 0.7857142857142857,
+      "eval_TIME_recall": 0.8461538461538461,
+      "eval_loss": 0.34240958094596863,
+      "eval_overall_accuracy": 0.9587190392794596,
+      "eval_overall_f1": 0.8435754189944135,
+      "eval_overall_precision": 0.8010610079575596,
+      "eval_overall_recall": 0.8908554572271387,
+      "eval_runtime": 0.3616,
+      "eval_samples_per_second": 517.085,
+      "eval_steps_per_second": 8.295,
       "step": 3922
     },
     {
       "epoch": 38.0,
-      "grad_norm": 12.923250198364258,
+      "grad_norm": 4.613161563873291,
       "learning_rate": 3.1e-05,
-      "loss": 0.0021,
+      "loss": 0.0012,
       "step": 4028
     },
     {
       "epoch": 38.0,
-      "eval_LOCATION_f1": 0.816326530612245,
+      "eval_LOCATION_f1": 0.8104575163398693,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.8,
-      "eval_LOCATION_recall": 0.8333333333333334,
-      "eval_ORGANIZATION_f1": 0.7721518987341773,
+      "eval_LOCATION_precision": 0.7654320987654321,
+      "eval_LOCATION_recall": 0.8611111111111112,
+      "eval_ORGANIZATION_f1": 0.825,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7349397590361446,
-      "eval_ORGANIZATION_recall": 0.8133333333333334,
-      "eval_PERSON_f1": 0.9158249158249158,
+      "eval_ORGANIZATION_precision": 0.7764705882352941,
+      "eval_ORGANIZATION_recall": 0.88,
+      "eval_PERSON_f1": 0.8926174496644295,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8831168831168831,
-      "eval_PERSON_recall": 0.951048951048951,
-      "eval_QUANTITY_f1": 0.7058823529411765,
+      "eval_PERSON_precision": 0.8580645161290322,
+      "eval_PERSON_recall": 0.9300699300699301,
+      "eval_QUANTITY_f1": 0.6923076923076923,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.6428571428571429,
+      "eval_QUANTITY_precision": 0.6206896551724138,
       "eval_QUANTITY_recall": 0.782608695652174,
-      "eval_TIME_f1": 0.8461538461538461,
+      "eval_TIME_f1": 0.8148148148148148,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8461538461538461,
+      "eval_TIME_precision": 0.7857142857142857,
       "eval_TIME_recall": 0.8461538461538461,
-      "eval_loss": 0.3416782021522522,
-      "eval_overall_accuracy": 0.960720540405304,
-      "eval_overall_f1": 0.8425531914893618,
-      "eval_overall_precision": 0.8114754098360656,
-      "eval_overall_recall": 0.8761061946902655,
-      "eval_runtime": 0.9052,
-      "eval_samples_per_second": 206.591,
-      "eval_steps_per_second": 3.314,
+      "eval_loss": 0.3519201874732971,
+      "eval_overall_accuracy": 0.957968476357268,
+      "eval_overall_f1": 0.8396094839609484,
+      "eval_overall_precision": 0.7962962962962963,
+      "eval_overall_recall": 0.887905604719764,
+      "eval_runtime": 0.3618,
+      "eval_samples_per_second": 516.793,
+      "eval_steps_per_second": 8.291,
       "step": 4028
     },
     {
       "epoch": 39.0,
-      "grad_norm": 0.05247601494193077,
+      "grad_norm": 0.0017957445234060287,
       "learning_rate": 3.05e-05,
-      "loss": 0.0026,
+      "loss": 0.0012,
       "step": 4134
     },
     {
       "epoch": 39.0,
-      "eval_LOCATION_f1": 0.7891156462585034,
+      "eval_LOCATION_f1": 0.8266666666666667,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7733333333333333,
-      "eval_LOCATION_recall": 0.8055555555555556,
-      "eval_ORGANIZATION_f1": 0.7974683544303797,
+      "eval_LOCATION_precision": 0.7948717948717948,
+      "eval_LOCATION_recall": 0.8611111111111112,
+      "eval_ORGANIZATION_f1": 0.825,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7590361445783133,
-      "eval_ORGANIZATION_recall": 0.84,
-      "eval_PERSON_f1": 0.8963210702341137,
+      "eval_ORGANIZATION_precision": 0.7764705882352941,
+      "eval_ORGANIZATION_recall": 0.88,
+      "eval_PERSON_f1": 0.8926174496644295,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8589743589743589,
-      "eval_PERSON_recall": 0.9370629370629371,
+      "eval_PERSON_precision": 0.8580645161290322,
+      "eval_PERSON_recall": 0.9300699300699301,
       "eval_QUANTITY_f1": 0.7169811320754716,
       "eval_QUANTITY_number": 23,
       "eval_QUANTITY_precision": 0.6333333333333333,
       "eval_QUANTITY_recall": 0.8260869565217391,
-      "eval_TIME_f1": 0.830188679245283,
+      "eval_TIME_f1": 0.7777777777777779,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8148148148148148,
-      "eval_TIME_recall": 0.8461538461538461,
-      "eval_loss": 0.37723416090011597,
+      "eval_TIME_precision": 0.75,
+      "eval_TIME_recall": 0.8076923076923077,
+      "eval_loss": 0.3654736876487732,
       "eval_overall_accuracy": 0.9587190392794596,
-      "eval_overall_f1": 0.8338028169014085,
-      "eval_overall_precision": 0.7978436657681941,
-      "eval_overall_recall": 0.8731563421828908,
-      "eval_runtime": 0.9093,
-      "eval_samples_per_second": 205.644,
-      "eval_steps_per_second": 3.299,
+      "eval_overall_f1": 0.8419580419580419,
+      "eval_overall_precision": 0.800531914893617,
+      "eval_overall_recall": 0.887905604719764,
+      "eval_runtime": 0.3617,
+      "eval_samples_per_second": 517.022,
+      "eval_steps_per_second": 8.294,
       "step": 4134
     },
     {
       "epoch": 40.0,
-      "grad_norm": 0.0038078853394836187,
+      "grad_norm": 0.001596860121935606,
       "learning_rate": 3e-05,
-      "loss": 0.0021,
+      "loss": 0.001,
       "step": 4240
     },
     {
       "epoch": 40.0,
-      "eval_LOCATION_f1": 0.7891156462585034,
+      "eval_LOCATION_f1": 0.8,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7733333333333333,
+      "eval_LOCATION_precision": 0.7945205479452054,
       "eval_LOCATION_recall": 0.8055555555555556,
-      "eval_ORGANIZATION_f1": 0.7894736842105264,
+      "eval_ORGANIZATION_f1": 0.7948717948717949,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7792207792207793,
-      "eval_ORGANIZATION_recall": 0.8,
-      "eval_PERSON_f1": 0.9146757679180888,
+      "eval_ORGANIZATION_precision": 0.7654320987654321,
+      "eval_ORGANIZATION_recall": 0.8266666666666667,
+      "eval_PERSON_f1": 0.9090909090909092,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8933333333333333,
-      "eval_PERSON_recall": 0.9370629370629371,
-      "eval_QUANTITY_f1": 0.7272727272727272,
+      "eval_PERSON_precision": 0.8766233766233766,
+      "eval_PERSON_recall": 0.9440559440559441,
+      "eval_QUANTITY_f1": 0.7755102040816326,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.7619047619047619,
-      "eval_QUANTITY_recall": 0.6956521739130435,
-      "eval_TIME_f1": 0.8,
+      "eval_QUANTITY_precision": 0.7307692307692307,
+      "eval_QUANTITY_recall": 0.8260869565217391,
+      "eval_TIME_f1": 0.830188679245283,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.7586206896551724,
+      "eval_TIME_precision": 0.8148148148148148,
       "eval_TIME_recall": 0.8461538461538461,
-      "eval_loss": 0.331974595785141,
-      "eval_overall_accuracy": 0.960720540405304,
-      "eval_overall_f1": 0.8393632416787264,
-      "eval_overall_precision": 0.8238636363636364,
-      "eval_overall_recall": 0.855457227138643,
-      "eval_runtime": 0.902,
-      "eval_samples_per_second": 207.323,
-      "eval_steps_per_second": 3.326,
+      "eval_loss": 0.34077638387680054,
+      "eval_overall_accuracy": 0.961220915686765,
+      "eval_overall_f1": 0.8457142857142858,
+      "eval_overall_precision": 0.8199445983379502,
+      "eval_overall_recall": 0.8731563421828908,
+      "eval_runtime": 0.3642,
+      "eval_samples_per_second": 513.453,
+      "eval_steps_per_second": 8.237,
       "step": 4240
     },
     {
       "epoch": 41.0,
-      "grad_norm": 0.0014568913029506803,
+      "grad_norm": 0.0034711004700511694,
       "learning_rate": 2.95e-05,
-      "loss": 0.0013,
+      "loss": 0.0025,
       "step": 4346
     },
     {
       "epoch": 41.0,
-      "eval_LOCATION_f1": 0.8187919463087249,
+      "eval_LOCATION_f1": 0.8157894736842106,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7922077922077922,
-      "eval_LOCATION_recall": 0.8472222222222222,
-      "eval_ORGANIZATION_f1": 0.7425149700598803,
+      "eval_LOCATION_precision": 0.775,
+      "eval_LOCATION_recall": 0.8611111111111112,
+      "eval_ORGANIZATION_f1": 0.7898089171974522,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.6739130434782609,
+      "eval_ORGANIZATION_precision": 0.7560975609756098,
       "eval_ORGANIZATION_recall": 0.8266666666666667,
-      "eval_PERSON_f1": 0.9090909090909092,
+      "eval_PERSON_f1": 0.912751677852349,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8766233766233766,
-      "eval_PERSON_recall": 0.9440559440559441,
-      "eval_QUANTITY_f1": 0.76,
+      "eval_PERSON_precision": 0.8774193548387097,
+      "eval_PERSON_recall": 0.951048951048951,
+      "eval_QUANTITY_f1": 0.7058823529411765,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.7037037037037037,
-      "eval_QUANTITY_recall": 0.8260869565217391,
-      "eval_TIME_f1": 0.8627450980392156,
+      "eval_QUANTITY_precision": 0.6428571428571429,
+      "eval_QUANTITY_recall": 0.782608695652174,
+      "eval_TIME_f1": 0.8,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.88,
+      "eval_TIME_precision": 0.7586206896551724,
       "eval_TIME_recall": 0.8461538461538461,
-      "eval_loss": 0.3371780812740326,
-      "eval_overall_accuracy": 0.9619714786089567,
-      "eval_overall_f1": 0.8375350140056023,
-      "eval_overall_precision": 0.7973333333333333,
-      "eval_overall_recall": 0.8820058997050148,
-      "eval_runtime": 0.9175,
-      "eval_samples_per_second": 203.81,
-      "eval_steps_per_second": 3.27,
+      "eval_loss": 0.340951144695282,
+      "eval_overall_accuracy": 0.9614711033274956,
+      "eval_overall_f1": 0.8415147265077138,
+      "eval_overall_precision": 0.8021390374331551,
+      "eval_overall_recall": 0.8849557522123894,
+      "eval_runtime": 0.361,
+      "eval_samples_per_second": 518.055,
+      "eval_steps_per_second": 8.311,
       "step": 4346
     },
     {
       "epoch": 42.0,
-      "grad_norm": 0.01008665468543768,
+      "grad_norm": 0.02040654979646206,
       "learning_rate": 2.9e-05,
-      "loss": 0.0013,
+      "loss": 0.0012,
       "step": 4452
     },
     {
       "epoch": 42.0,
-      "eval_LOCATION_f1": 0.7837837837837838,
+      "eval_LOCATION_f1": 0.8322147651006712,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7631578947368421,
-      "eval_LOCATION_recall": 0.8055555555555556,
-      "eval_ORGANIZATION_f1": 0.779874213836478,
+      "eval_LOCATION_precision": 0.8051948051948052,
+      "eval_LOCATION_recall": 0.8611111111111112,
+      "eval_ORGANIZATION_f1": 0.8025477707006369,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7380952380952381,
-      "eval_ORGANIZATION_recall": 0.8266666666666667,
-      "eval_PERSON_f1": 0.9023569023569024,
+      "eval_ORGANIZATION_precision": 0.7682926829268293,
+      "eval_ORGANIZATION_recall": 0.84,
+      "eval_PERSON_f1": 0.903010033444816,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8701298701298701,
-      "eval_PERSON_recall": 0.9370629370629371,
-      "eval_QUANTITY_f1": 0.72,
+      "eval_PERSON_precision": 0.8653846153846154,
+      "eval_PERSON_recall": 0.9440559440559441,
+      "eval_QUANTITY_f1": 0.76,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.6666666666666666,
-      "eval_QUANTITY_recall": 0.782608695652174,
-      "eval_TIME_f1": 0.8518518518518519,
+      "eval_QUANTITY_precision": 0.7037037037037037,
+      "eval_QUANTITY_recall": 0.8260869565217391,
+      "eval_TIME_f1": 0.8076923076923077,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8214285714285714,
-      "eval_TIME_recall": 0.8846153846153846,
-      "eval_loss": 0.3549456298351288,
-      "eval_overall_accuracy": 0.9602201651238429,
-      "eval_overall_f1": 0.8333333333333334,
-      "eval_overall_precision": 0.7994579945799458,
-      "eval_overall_recall": 0.8702064896755162,
-      "eval_runtime": 0.9061,
-      "eval_samples_per_second": 206.389,
-      "eval_steps_per_second": 3.311,
+      "eval_TIME_precision": 0.8076923076923077,
+      "eval_TIME_recall": 0.8076923076923077,
+      "eval_loss": 0.3626217842102051,
+      "eval_overall_accuracy": 0.9614711033274956,
+      "eval_overall_f1": 0.8486562942008488,
+      "eval_overall_precision": 0.8152173913043478,
+      "eval_overall_recall": 0.8849557522123894,
+      "eval_runtime": 0.3631,
+      "eval_samples_per_second": 515.064,
+      "eval_steps_per_second": 8.263,
       "step": 4452
     },
     {
       "epoch": 43.0,
-      "grad_norm": 0.0012029644567519426,
+      "grad_norm": 0.001348801888525486,
       "learning_rate": 2.8499999999999998e-05,
-      "loss": 0.0003,
+      "loss": 0.003,
       "step": 4558
     },
     {
       "epoch": 43.0,
-      "eval_LOCATION_f1": 0.8,
+      "eval_LOCATION_f1": 0.816326530612245,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7469879518072289,
-      "eval_LOCATION_recall": 0.8611111111111112,
-      "eval_ORGANIZATION_f1": 0.7682926829268292,
+      "eval_LOCATION_precision": 0.8,
+      "eval_LOCATION_recall": 0.8333333333333334,
+      "eval_ORGANIZATION_f1": 0.7951807228915663,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7078651685393258,
-      "eval_ORGANIZATION_recall": 0.84,
+      "eval_ORGANIZATION_precision": 0.7252747252747253,
+      "eval_ORGANIZATION_recall": 0.88,
       "eval_PERSON_f1": 0.9060402684563759,
       "eval_PERSON_number": 143,
       "eval_PERSON_precision": 0.8709677419354839,
@@ -1671,25 +1671,25 @@
       "eval_QUANTITY_number": 23,
       "eval_QUANTITY_precision": 0.6428571428571429,
       "eval_QUANTITY_recall": 0.782608695652174,
-      "eval_TIME_f1": 0.8846153846153846,
+      "eval_TIME_f1": 0.7924528301886792,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8846153846153846,
-      "eval_TIME_recall": 0.8846153846153846,
-      "eval_loss": 0.36216506361961365,
-      "eval_overall_accuracy": 0.9604703527645734,
-      "eval_overall_f1": 0.836111111111111,
-      "eval_overall_precision": 0.7900262467191601,
-      "eval_overall_recall": 0.887905604719764,
-      "eval_runtime": 0.9049,
-      "eval_samples_per_second": 206.658,
-      "eval_steps_per_second": 3.315,
+      "eval_TIME_precision": 0.7777777777777778,
+      "eval_TIME_recall": 0.8076923076923077,
+      "eval_loss": 0.3544086813926697,
+      "eval_overall_accuracy": 0.9587190392794596,
+      "eval_overall_f1": 0.8391608391608392,
+      "eval_overall_precision": 0.7978723404255319,
+      "eval_overall_recall": 0.8849557522123894,
+      "eval_runtime": 0.3593,
+      "eval_samples_per_second": 520.389,
+      "eval_steps_per_second": 8.348,
       "step": 4558
     },
     {
       "epoch": 44.0,
-      "grad_norm": 0.04154082387685776,
+      "grad_norm": 4.9029974937438965,
       "learning_rate": 2.8000000000000003e-05,
-      "loss": 0.0032,
+      "loss": 0.0027,
       "step": 4664
     },
     {
@@ -1698,13 +1698,13 @@
       "eval_LOCATION_number": 72,
       "eval_LOCATION_precision": 0.7625,
       "eval_LOCATION_recall": 0.8472222222222222,
-      "eval_ORGANIZATION_f1": 0.7619047619047621,
+      "eval_ORGANIZATION_f1": 0.7999999999999999,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.6881720430107527,
-      "eval_ORGANIZATION_recall": 0.8533333333333334,
-      "eval_PERSON_f1": 0.9060402684563759,
+      "eval_ORGANIZATION_precision": 0.7157894736842105,
+      "eval_ORGANIZATION_recall": 0.9066666666666666,
+      "eval_PERSON_f1": 0.8940397350993377,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8709677419354839,
+      "eval_PERSON_precision": 0.8490566037735849,
       "eval_PERSON_recall": 0.9440559440559441,
       "eval_QUANTITY_f1": 0.7307692307692308,
       "eval_QUANTITY_number": 23,
@@ -1714,111 +1714,111 @@
       "eval_TIME_number": 26,
       "eval_TIME_precision": 0.7857142857142857,
       "eval_TIME_recall": 0.8461538461538461,
-      "eval_loss": 0.35855501890182495,
-      "eval_overall_accuracy": 0.957968476357268,
-      "eval_overall_f1": 0.8314917127071824,
-      "eval_overall_precision": 0.7818181818181819,
-      "eval_overall_recall": 0.887905604719764,
-      "eval_runtime": 0.9082,
-      "eval_samples_per_second": 205.905,
-      "eval_steps_per_second": 3.303,
+      "eval_loss": 0.3703295886516571,
+      "eval_overall_accuracy": 0.9587190392794596,
+      "eval_overall_f1": 0.8356164383561644,
+      "eval_overall_precision": 0.7800511508951407,
+      "eval_overall_recall": 0.8997050147492626,
+      "eval_runtime": 0.3623,
+      "eval_samples_per_second": 516.147,
+      "eval_steps_per_second": 8.28,
       "step": 4664
     },
     {
       "epoch": 45.0,
-      "grad_norm": 0.002047579735517502,
+      "grad_norm": 0.2568044662475586,
       "learning_rate": 2.7500000000000004e-05,
-      "loss": 0.001,
+      "loss": 0.0015,
       "step": 4770
     },
     {
       "epoch": 45.0,
-      "eval_LOCATION_f1": 0.8133333333333332,
+      "eval_LOCATION_f1": 0.8108108108108109,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.782051282051282,
-      "eval_LOCATION_recall": 0.8472222222222222,
-      "eval_ORGANIZATION_f1": 0.7950310559006211,
+      "eval_LOCATION_precision": 0.7894736842105263,
+      "eval_LOCATION_recall": 0.8333333333333334,
+      "eval_ORGANIZATION_f1": 0.8354430379746836,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7441860465116279,
-      "eval_ORGANIZATION_recall": 0.8533333333333334,
-      "eval_PERSON_f1": 0.9158249158249158,
+      "eval_ORGANIZATION_precision": 0.7951807228915663,
+      "eval_ORGANIZATION_recall": 0.88,
+      "eval_PERSON_f1": 0.8910891089108911,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8831168831168831,
-      "eval_PERSON_recall": 0.951048951048951,
-      "eval_QUANTITY_f1": 0.7307692307692308,
+      "eval_PERSON_precision": 0.84375,
+      "eval_PERSON_recall": 0.9440559440559441,
+      "eval_QUANTITY_f1": 0.72,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.6551724137931034,
-      "eval_QUANTITY_recall": 0.8260869565217391,
-      "eval_TIME_f1": 0.830188679245283,
+      "eval_QUANTITY_precision": 0.6666666666666666,
+      "eval_QUANTITY_recall": 0.782608695652174,
+      "eval_TIME_f1": 0.8076923076923077,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8148148148148148,
-      "eval_TIME_recall": 0.8461538461538461,
-      "eval_loss": 0.36997854709625244,
-      "eval_overall_accuracy": 0.9619714786089567,
-      "eval_overall_f1": 0.847124824684432,
-      "eval_overall_precision": 0.8074866310160428,
-      "eval_overall_recall": 0.8908554572271387,
-      "eval_runtime": 0.9062,
-      "eval_samples_per_second": 206.36,
-      "eval_steps_per_second": 3.311,
+      "eval_TIME_precision": 0.8076923076923077,
+      "eval_TIME_recall": 0.8076923076923077,
+      "eval_loss": 0.34526732563972473,
+      "eval_overall_accuracy": 0.9604703527645734,
+      "eval_overall_f1": 0.8438818565400844,
+      "eval_overall_precision": 0.8064516129032258,
+      "eval_overall_recall": 0.8849557522123894,
+      "eval_runtime": 0.3579,
+      "eval_samples_per_second": 522.551,
+      "eval_steps_per_second": 8.383,
       "step": 4770
     },
     {
       "epoch": 46.0,
-      "grad_norm": 0.0037038603331893682,
+      "grad_norm": 0.0012396867386996746,
       "learning_rate": 2.7000000000000002e-05,
-      "loss": 0.0034,
+      "loss": 0.0018,
       "step": 4876
     },
     {
       "epoch": 46.0,
-      "eval_LOCATION_f1": 0.8333333333333334,
+      "eval_LOCATION_f1": 0.816326530612245,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.8333333333333334,
+      "eval_LOCATION_precision": 0.8,
       "eval_LOCATION_recall": 0.8333333333333334,
-      "eval_ORGANIZATION_f1": 0.7924528301886793,
+      "eval_ORGANIZATION_f1": 0.8170731707317072,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.75,
-      "eval_ORGANIZATION_recall": 0.84,
-      "eval_PERSON_f1": 0.9121621621621621,
+      "eval_ORGANIZATION_precision": 0.7528089887640449,
+      "eval_ORGANIZATION_recall": 0.8933333333333333,
+      "eval_PERSON_f1": 0.9060402684563759,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8823529411764706,
+      "eval_PERSON_precision": 0.8709677419354839,
       "eval_PERSON_recall": 0.9440559440559441,
-      "eval_QUANTITY_f1": 0.7307692307692308,
+      "eval_QUANTITY_f1": 0.693877551020408,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.6551724137931034,
-      "eval_QUANTITY_recall": 0.8260869565217391,
-      "eval_TIME_f1": 0.830188679245283,
+      "eval_QUANTITY_precision": 0.6538461538461539,
+      "eval_QUANTITY_recall": 0.7391304347826086,
+      "eval_TIME_f1": 0.8235294117647058,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8148148148148148,
-      "eval_TIME_recall": 0.8461538461538461,
-      "eval_loss": 0.33895859122276306,
-      "eval_overall_accuracy": 0.9602201651238429,
-      "eval_overall_f1": 0.8494318181818181,
-      "eval_overall_precision": 0.8191780821917808,
-      "eval_overall_recall": 0.8820058997050148,
-      "eval_runtime": 0.9121,
-      "eval_samples_per_second": 205.028,
-      "eval_steps_per_second": 3.289,
+      "eval_TIME_precision": 0.84,
+      "eval_TIME_recall": 0.8076923076923077,
+      "eval_loss": 0.3448871970176697,
+      "eval_overall_accuracy": 0.961220915686765,
+      "eval_overall_f1": 0.846262341325811,
+      "eval_overall_precision": 0.8108108108108109,
+      "eval_overall_recall": 0.8849557522123894,
+      "eval_runtime": 0.3635,
+      "eval_samples_per_second": 514.485,
+      "eval_steps_per_second": 8.254,
       "step": 4876
     },
     {
       "epoch": 47.0,
-      "grad_norm": 0.0005634190747514367,
+      "grad_norm": 1.6906110048294067,
       "learning_rate": 2.6500000000000004e-05,
-      "loss": 0.0018,
+      "loss": 0.0028,
       "step": 4982
     },
     {
       "epoch": 47.0,
-      "eval_LOCATION_f1": 0.8211920529801324,
+      "eval_LOCATION_f1": 0.8137931034482759,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7848101265822784,
-      "eval_LOCATION_recall": 0.8611111111111112,
-      "eval_ORGANIZATION_f1": 0.7564102564102564,
+      "eval_LOCATION_precision": 0.8082191780821918,
+      "eval_LOCATION_recall": 0.8194444444444444,
+      "eval_ORGANIZATION_f1": 0.8148148148148148,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7283950617283951,
-      "eval_ORGANIZATION_recall": 0.7866666666666666,
+      "eval_ORGANIZATION_precision": 0.7586206896551724,
+      "eval_ORGANIZATION_recall": 0.88,
       "eval_PERSON_f1": 0.9060402684563759,
       "eval_PERSON_number": 143,
       "eval_PERSON_precision": 0.8709677419354839,
@@ -1827,236 +1827,236 @@
       "eval_QUANTITY_number": 23,
       "eval_QUANTITY_precision": 0.6666666666666666,
       "eval_QUANTITY_recall": 0.782608695652174,
-      "eval_TIME_f1": 0.8461538461538461,
+      "eval_TIME_f1": 0.830188679245283,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8461538461538461,
+      "eval_TIME_precision": 0.8148148148148148,
       "eval_TIME_recall": 0.8461538461538461,
-      "eval_loss": 0.3375060260295868,
-      "eval_overall_accuracy": 0.9632224168126094,
-      "eval_overall_f1": 0.8373408769448375,
-      "eval_overall_precision": 0.8043478260869565,
-      "eval_overall_recall": 0.8731563421828908,
-      "eval_runtime": 0.9199,
-      "eval_samples_per_second": 203.274,
-      "eval_steps_per_second": 3.261,
+      "eval_loss": 0.37111541628837585,
+      "eval_overall_accuracy": 0.9597197898423818,
+      "eval_overall_f1": 0.8474576271186441,
+      "eval_overall_precision": 0.8130081300813008,
+      "eval_overall_recall": 0.8849557522123894,
+      "eval_runtime": 0.3591,
+      "eval_samples_per_second": 520.775,
+      "eval_steps_per_second": 8.355,
       "step": 4982
     },
     {
       "epoch": 48.0,
-      "grad_norm": 0.0028315193485468626,
+      "grad_norm": 0.029315035790205002,
       "learning_rate": 2.6000000000000002e-05,
-      "loss": 0.0017,
+      "loss": 0.0014,
       "step": 5088
     },
     {
       "epoch": 48.0,
-      "eval_LOCATION_f1": 0.802721088435374,
+      "eval_LOCATION_f1": 0.8251748251748252,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7866666666666666,
+      "eval_LOCATION_precision": 0.8309859154929577,
       "eval_LOCATION_recall": 0.8194444444444444,
-      "eval_ORGANIZATION_f1": 0.7974683544303797,
+      "eval_ORGANIZATION_f1": 0.8198757763975156,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7590361445783133,
-      "eval_ORGANIZATION_recall": 0.84,
-      "eval_PERSON_f1": 0.9183673469387755,
+      "eval_ORGANIZATION_precision": 0.7674418604651163,
+      "eval_ORGANIZATION_recall": 0.88,
+      "eval_PERSON_f1": 0.9090909090909092,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8940397350993378,
+      "eval_PERSON_precision": 0.8766233766233766,
       "eval_PERSON_recall": 0.9440559440559441,
       "eval_QUANTITY_f1": 0.72,
       "eval_QUANTITY_number": 23,
       "eval_QUANTITY_precision": 0.6666666666666666,
       "eval_QUANTITY_recall": 0.782608695652174,
-      "eval_TIME_f1": 0.8461538461538461,
+      "eval_TIME_f1": 0.830188679245283,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8461538461538461,
+      "eval_TIME_precision": 0.8148148148148148,
       "eval_TIME_recall": 0.8461538461538461,
-      "eval_loss": 0.33102044463157654,
+      "eval_loss": 0.3521229922771454,
       "eval_overall_accuracy": 0.9632224168126094,
-      "eval_overall_f1": 0.8473609129814551,
-      "eval_overall_precision": 0.8204419889502762,
-      "eval_overall_recall": 0.8761061946902655,
-      "eval_runtime": 0.9017,
-      "eval_samples_per_second": 207.378,
-      "eval_steps_per_second": 3.327,
+      "eval_overall_f1": 0.8522727272727273,
+      "eval_overall_precision": 0.821917808219178,
+      "eval_overall_recall": 0.8849557522123894,
+      "eval_runtime": 0.3694,
+      "eval_samples_per_second": 506.192,
+      "eval_steps_per_second": 8.121,
       "step": 5088
     },
     {
       "epoch": 49.0,
-      "grad_norm": 0.009939000010490417,
+      "grad_norm": 0.02477916330099106,
       "learning_rate": 2.5500000000000003e-05,
-      "loss": 0.0014,
+      "loss": 0.0023,
       "step": 5194
     },
     {
       "epoch": 49.0,
-      "eval_LOCATION_f1": 0.8299319727891157,
+      "eval_LOCATION_f1": 0.8187919463087249,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.8133333333333334,
+      "eval_LOCATION_precision": 0.7922077922077922,
       "eval_LOCATION_recall": 0.8472222222222222,
-      "eval_ORGANIZATION_f1": 0.8074534161490684,
+      "eval_ORGANIZATION_f1": 0.8129032258064516,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7558139534883721,
-      "eval_ORGANIZATION_recall": 0.8666666666666667,
-      "eval_PERSON_f1": 0.9121621621621621,
+      "eval_ORGANIZATION_precision": 0.7875,
+      "eval_ORGANIZATION_recall": 0.84,
+      "eval_PERSON_f1": 0.8874172185430464,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8823529411764706,
-      "eval_PERSON_recall": 0.9440559440559441,
-      "eval_QUANTITY_f1": 0.7450980392156864,
+      "eval_PERSON_precision": 0.8427672955974843,
+      "eval_PERSON_recall": 0.9370629370629371,
+      "eval_QUANTITY_f1": 0.72,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.6785714285714286,
-      "eval_QUANTITY_recall": 0.8260869565217391,
-      "eval_TIME_f1": 0.8627450980392156,
+      "eval_QUANTITY_precision": 0.6666666666666666,
+      "eval_QUANTITY_recall": 0.782608695652174,
+      "eval_TIME_f1": 0.8148148148148148,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.88,
+      "eval_TIME_precision": 0.7857142857142857,
       "eval_TIME_recall": 0.8461538461538461,
-      "eval_loss": 0.32123759388923645,
-      "eval_overall_accuracy": 0.9624718538904178,
-      "eval_overall_f1": 0.8555240793201133,
-      "eval_overall_precision": 0.8228882833787466,
-      "eval_overall_recall": 0.8908554572271387,
-      "eval_runtime": 0.9187,
-      "eval_samples_per_second": 203.539,
-      "eval_steps_per_second": 3.265,
+      "eval_loss": 0.36508724093437195,
+      "eval_overall_accuracy": 0.9592194145609206,
+      "eval_overall_f1": 0.8394366197183099,
+      "eval_overall_precision": 0.8032345013477089,
+      "eval_overall_recall": 0.8790560471976401,
+      "eval_runtime": 0.3605,
+      "eval_samples_per_second": 518.761,
+      "eval_steps_per_second": 8.322,
       "step": 5194
     },
     {
       "epoch": 50.0,
-      "grad_norm": 0.7588798403739929,
+      "grad_norm": 0.017422163859009743,
       "learning_rate": 2.5e-05,
-      "loss": 0.0011,
+      "loss": 0.0027,
       "step": 5300
     },
     {
       "epoch": 50.0,
-      "eval_LOCATION_f1": 0.802721088435374,
+      "eval_LOCATION_f1": 0.8187919463087249,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7866666666666666,
-      "eval_LOCATION_recall": 0.8194444444444444,
-      "eval_ORGANIZATION_f1": 0.7924528301886793,
+      "eval_LOCATION_precision": 0.7922077922077922,
+      "eval_LOCATION_recall": 0.8472222222222222,
+      "eval_ORGANIZATION_f1": 0.8322981366459627,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.75,
-      "eval_ORGANIZATION_recall": 0.84,
-      "eval_PERSON_f1": 0.9152542372881356,
+      "eval_ORGANIZATION_precision": 0.7790697674418605,
+      "eval_ORGANIZATION_recall": 0.8933333333333333,
+      "eval_PERSON_f1": 0.9096989966555185,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8881578947368421,
-      "eval_PERSON_recall": 0.9440559440559441,
+      "eval_PERSON_precision": 0.8717948717948718,
+      "eval_PERSON_recall": 0.951048951048951,
       "eval_QUANTITY_f1": 0.72,
       "eval_QUANTITY_number": 23,
       "eval_QUANTITY_precision": 0.6666666666666666,
       "eval_QUANTITY_recall": 0.782608695652174,
-      "eval_TIME_f1": 0.8518518518518519,
+      "eval_TIME_f1": 0.9230769230769231,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8214285714285714,
-      "eval_TIME_recall": 0.8846153846153846,
-      "eval_loss": 0.3109734356403351,
-      "eval_overall_accuracy": 0.9622216662496873,
-      "eval_overall_f1": 0.8453900709219858,
-      "eval_overall_precision": 0.8142076502732241,
-      "eval_overall_recall": 0.8790560471976401,
-      "eval_runtime": 0.9093,
-      "eval_samples_per_second": 205.655,
-      "eval_steps_per_second": 3.299,
+      "eval_TIME_precision": 0.9230769230769231,
+      "eval_TIME_recall": 0.9230769230769231,
+      "eval_loss": 0.3326910436153412,
+      "eval_overall_accuracy": 0.9614711033274956,
+      "eval_overall_f1": 0.8607594936708861,
+      "eval_overall_precision": 0.8225806451612904,
+      "eval_overall_recall": 0.9026548672566371,
+      "eval_runtime": 0.3722,
+      "eval_samples_per_second": 502.368,
+      "eval_steps_per_second": 8.059,
       "step": 5300
     },
     {
       "epoch": 51.0,
-      "grad_norm": 0.0009810138726606965,
+      "grad_norm": 0.00688315462321043,
       "learning_rate": 2.45e-05,
-      "loss": 0.0005,
+      "loss": 0.0009,
       "step": 5406
     },
     {
       "epoch": 51.0,
-      "eval_LOCATION_f1": 0.8299319727891157,
+      "eval_LOCATION_f1": 0.8137931034482759,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.8133333333333334,
-      "eval_LOCATION_recall": 0.8472222222222222,
-      "eval_ORGANIZATION_f1": 0.825,
+      "eval_LOCATION_precision": 0.8082191780821918,
+      "eval_LOCATION_recall": 0.8194444444444444,
+      "eval_ORGANIZATION_f1": 0.8387096774193549,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7764705882352941,
-      "eval_ORGANIZATION_recall": 0.88,
-      "eval_PERSON_f1": 0.9246575342465754,
+      "eval_ORGANIZATION_precision": 0.8125,
+      "eval_ORGANIZATION_recall": 0.8666666666666667,
+      "eval_PERSON_f1": 0.912751677852349,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.9060402684563759,
-      "eval_PERSON_recall": 0.9440559440559441,
-      "eval_QUANTITY_f1": 0.7058823529411765,
+      "eval_PERSON_precision": 0.8774193548387097,
+      "eval_PERSON_recall": 0.951048951048951,
+      "eval_QUANTITY_f1": 0.7083333333333334,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.6428571428571429,
-      "eval_QUANTITY_recall": 0.782608695652174,
-      "eval_TIME_f1": 0.8518518518518519,
+      "eval_QUANTITY_precision": 0.68,
+      "eval_QUANTITY_recall": 0.7391304347826086,
+      "eval_TIME_f1": 0.9230769230769231,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8214285714285714,
-      "eval_TIME_recall": 0.8846153846153846,
-      "eval_loss": 0.30816784501075745,
+      "eval_TIME_precision": 0.9230769230769231,
+      "eval_TIME_recall": 0.9230769230769231,
+      "eval_loss": 0.33154383301734924,
       "eval_overall_accuracy": 0.9659744808606455,
-      "eval_overall_f1": 0.8607954545454546,
-      "eval_overall_precision": 0.8301369863013699,
-      "eval_overall_recall": 0.8938053097345132,
-      "eval_runtime": 0.914,
-      "eval_samples_per_second": 204.601,
-      "eval_steps_per_second": 3.282,
+      "eval_overall_f1": 0.8624641833810888,
+      "eval_overall_precision": 0.8384401114206128,
+      "eval_overall_recall": 0.887905604719764,
+      "eval_runtime": 0.358,
+      "eval_samples_per_second": 522.281,
+      "eval_steps_per_second": 8.379,
       "step": 5406
     },
     {
       "epoch": 52.0,
-      "grad_norm": 0.0026644645258784294,
+      "grad_norm": 0.0005034964415244758,
       "learning_rate": 2.4e-05,
-      "loss": 0.0003,
+      "loss": 0.001,
       "step": 5512
     },
     {
       "epoch": 52.0,
-      "eval_LOCATION_f1": 0.8157894736842106,
+      "eval_LOCATION_f1": 0.7945205479452055,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.775,
-      "eval_LOCATION_recall": 0.8611111111111112,
-      "eval_ORGANIZATION_f1": 0.8024691358024691,
+      "eval_LOCATION_precision": 0.7837837837837838,
+      "eval_LOCATION_recall": 0.8055555555555556,
+      "eval_ORGANIZATION_f1": 0.8322981366459627,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7471264367816092,
-      "eval_ORGANIZATION_recall": 0.8666666666666667,
-      "eval_PERSON_f1": 0.9183673469387755,
+      "eval_ORGANIZATION_precision": 0.7790697674418605,
+      "eval_ORGANIZATION_recall": 0.8933333333333333,
+      "eval_PERSON_f1": 0.912751677852349,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8940397350993378,
-      "eval_PERSON_recall": 0.9440559440559441,
-      "eval_QUANTITY_f1": 0.693877551020408,
+      "eval_PERSON_precision": 0.8774193548387097,
+      "eval_PERSON_recall": 0.951048951048951,
+      "eval_QUANTITY_f1": 0.68,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.6538461538461539,
+      "eval_QUANTITY_precision": 0.6296296296296297,
       "eval_QUANTITY_recall": 0.7391304347826086,
-      "eval_TIME_f1": 0.8518518518518519,
+      "eval_TIME_f1": 0.9056603773584906,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8214285714285714,
-      "eval_TIME_recall": 0.8846153846153846,
-      "eval_loss": 0.3176332414150238,
-      "eval_overall_accuracy": 0.9647235426569928,
-      "eval_overall_f1": 0.849507735583685,
-      "eval_overall_precision": 0.8118279569892473,
+      "eval_TIME_precision": 0.8888888888888888,
+      "eval_TIME_recall": 0.9230769230769231,
+      "eval_loss": 0.3466086983680725,
+      "eval_overall_accuracy": 0.960720540405304,
+      "eval_overall_f1": 0.8531073446327683,
+      "eval_overall_precision": 0.8184281842818428,
       "eval_overall_recall": 0.8908554572271387,
-      "eval_runtime": 0.9767,
-      "eval_samples_per_second": 191.467,
-      "eval_steps_per_second": 3.072,
+      "eval_runtime": 0.3715,
+      "eval_samples_per_second": 503.309,
+      "eval_steps_per_second": 8.074,
       "step": 5512
     },
     {
       "epoch": 53.0,
-      "grad_norm": 0.0603233277797699,
+      "grad_norm": 0.021436743438243866,
       "learning_rate": 2.35e-05,
-      "loss": 0.0004,
+      "loss": 0.0014,
       "step": 5618
     },
     {
       "epoch": 53.0,
-      "eval_LOCATION_f1": 0.8356164383561645,
+      "eval_LOCATION_f1": 0.8108108108108109,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.8243243243243243,
-      "eval_LOCATION_recall": 0.8472222222222222,
-      "eval_ORGANIZATION_f1": 0.8121212121212122,
+      "eval_LOCATION_precision": 0.7894736842105263,
+      "eval_LOCATION_recall": 0.8333333333333334,
+      "eval_ORGANIZATION_f1": 0.8271604938271604,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7444444444444445,
+      "eval_ORGANIZATION_precision": 0.7701149425287356,
       "eval_ORGANIZATION_recall": 0.8933333333333333,
-      "eval_PERSON_f1": 0.9215017064846417,
+      "eval_PERSON_f1": 0.9133333333333333,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.9,
-      "eval_PERSON_recall": 0.9440559440559441,
+      "eval_PERSON_precision": 0.8726114649681529,
+      "eval_PERSON_recall": 0.958041958041958,
       "eval_QUANTITY_f1": 0.68,
       "eval_QUANTITY_number": 23,
       "eval_QUANTITY_precision": 0.6296296296296297,
@@ -2065,333 +2065,333 @@
       "eval_TIME_number": 26,
       "eval_TIME_precision": 0.8214285714285714,
       "eval_TIME_recall": 0.8846153846153846,
-      "eval_loss": 0.3314068019390106,
-      "eval_overall_accuracy": 0.9647235426569928,
-      "eval_overall_f1": 0.8559322033898304,
-      "eval_overall_precision": 0.8211382113821138,
-      "eval_overall_recall": 0.8938053097345132,
-      "eval_runtime": 0.9085,
-      "eval_samples_per_second": 205.826,
-      "eval_steps_per_second": 3.302,
+      "eval_loss": 0.32782742381095886,
+      "eval_overall_accuracy": 0.9609707280460346,
+      "eval_overall_f1": 0.8515406162464986,
+      "eval_overall_precision": 0.8106666666666666,
+      "eval_overall_recall": 0.8967551622418879,
+      "eval_runtime": 0.3599,
+      "eval_samples_per_second": 519.612,
+      "eval_steps_per_second": 8.336,
       "step": 5618
     },
     {
       "epoch": 54.0,
-      "grad_norm": 0.00044817946036346257,
+      "grad_norm": 0.0009922637837007642,
       "learning_rate": 2.3000000000000003e-05,
-      "loss": 0.0002,
+      "loss": 0.0007,
       "step": 5724
     },
     {
       "epoch": 54.0,
-      "eval_LOCATION_f1": 0.8211920529801324,
+      "eval_LOCATION_f1": 0.8289473684210527,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7848101265822784,
-      "eval_LOCATION_recall": 0.8611111111111112,
-      "eval_ORGANIZATION_f1": 0.8050314465408804,
+      "eval_LOCATION_precision": 0.7875,
+      "eval_LOCATION_recall": 0.875,
+      "eval_ORGANIZATION_f1": 0.7976190476190477,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7619047619047619,
-      "eval_ORGANIZATION_recall": 0.8533333333333334,
-      "eval_PERSON_f1": 0.925170068027211,
+      "eval_ORGANIZATION_precision": 0.7204301075268817,
+      "eval_ORGANIZATION_recall": 0.8933333333333333,
+      "eval_PERSON_f1": 0.8910891089108911,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.9006622516556292,
-      "eval_PERSON_recall": 0.951048951048951,
-      "eval_QUANTITY_f1": 0.693877551020408,
+      "eval_PERSON_precision": 0.84375,
+      "eval_PERSON_recall": 0.9440559440559441,
+      "eval_QUANTITY_f1": 0.72,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.6538461538461539,
-      "eval_QUANTITY_recall": 0.7391304347826086,
-      "eval_TIME_f1": 0.8518518518518519,
+      "eval_QUANTITY_precision": 0.6666666666666666,
+      "eval_QUANTITY_recall": 0.782608695652174,
+      "eval_TIME_f1": 0.8363636363636363,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8214285714285714,
+      "eval_TIME_precision": 0.7931034482758621,
       "eval_TIME_recall": 0.8846153846153846,
-      "eval_loss": 0.3241828978061676,
-      "eval_overall_accuracy": 0.9657242932199149,
-      "eval_overall_f1": 0.8543140028288544,
-      "eval_overall_precision": 0.8206521739130435,
-      "eval_overall_recall": 0.8908554572271387,
-      "eval_runtime": 0.9117,
-      "eval_samples_per_second": 205.114,
-      "eval_steps_per_second": 3.291,
+      "eval_loss": 0.36466336250305176,
+      "eval_overall_accuracy": 0.9589692269201902,
+      "eval_overall_f1": 0.8406593406593406,
+      "eval_overall_precision": 0.7866323907455013,
+      "eval_overall_recall": 0.9026548672566371,
+      "eval_runtime": 0.36,
+      "eval_samples_per_second": 519.408,
+      "eval_steps_per_second": 8.333,
       "step": 5724
     },
     {
       "epoch": 55.0,
-      "grad_norm": 0.0010649035684764385,
+      "grad_norm": 0.003993849270045757,
       "learning_rate": 2.25e-05,
-      "loss": 0.0005,
+      "loss": 0.0017,
       "step": 5830
     },
     {
       "epoch": 55.0,
-      "eval_LOCATION_f1": 0.8289473684210527,
+      "eval_LOCATION_f1": 0.7947019867549668,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7875,
-      "eval_LOCATION_recall": 0.875,
-      "eval_ORGANIZATION_f1": 0.7875000000000001,
+      "eval_LOCATION_precision": 0.759493670886076,
+      "eval_LOCATION_recall": 0.8333333333333334,
+      "eval_ORGANIZATION_f1": 0.7948717948717949,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7411764705882353,
-      "eval_ORGANIZATION_recall": 0.84,
-      "eval_PERSON_f1": 0.9220338983050849,
+      "eval_ORGANIZATION_precision": 0.7654320987654321,
+      "eval_ORGANIZATION_recall": 0.8266666666666667,
+      "eval_PERSON_f1": 0.9072847682119206,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8947368421052632,
-      "eval_PERSON_recall": 0.951048951048951,
-      "eval_QUANTITY_f1": 0.68,
+      "eval_PERSON_precision": 0.8616352201257862,
+      "eval_PERSON_recall": 0.958041958041958,
+      "eval_QUANTITY_f1": 0.6274509803921569,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.6296296296296297,
-      "eval_QUANTITY_recall": 0.7391304347826086,
-      "eval_TIME_f1": 0.8518518518518519,
+      "eval_QUANTITY_precision": 0.5714285714285714,
+      "eval_QUANTITY_recall": 0.6956521739130435,
+      "eval_TIME_f1": 0.7857142857142856,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8214285714285714,
-      "eval_TIME_recall": 0.8846153846153846,
-      "eval_loss": 0.33954718708992004,
-      "eval_overall_accuracy": 0.9627220415311484,
-      "eval_overall_f1": 0.849507735583685,
-      "eval_overall_precision": 0.8118279569892473,
-      "eval_overall_recall": 0.8908554572271387,
-      "eval_runtime": 0.9361,
-      "eval_samples_per_second": 199.766,
-      "eval_steps_per_second": 3.205,
+      "eval_TIME_precision": 0.7333333333333333,
+      "eval_TIME_recall": 0.8461538461538461,
+      "eval_loss": 0.3635990619659424,
+      "eval_overall_accuracy": 0.9602201651238429,
+      "eval_overall_f1": 0.8296089385474861,
+      "eval_overall_precision": 0.7877984084880637,
+      "eval_overall_recall": 0.8761061946902655,
+      "eval_runtime": 0.3593,
+      "eval_samples_per_second": 520.494,
+      "eval_steps_per_second": 8.35,
       "step": 5830
     },
     {
       "epoch": 56.0,
-      "grad_norm": 0.0006854601670056581,
+      "grad_norm": 0.0005980022251605988,
       "learning_rate": 2.2000000000000003e-05,
-      "loss": 0.0001,
+      "loss": 0.0004,
       "step": 5936
     },
     {
       "epoch": 56.0,
-      "eval_LOCATION_f1": 0.8289473684210527,
+      "eval_LOCATION_f1": 0.778523489932886,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7875,
-      "eval_LOCATION_recall": 0.875,
-      "eval_ORGANIZATION_f1": 0.7826086956521738,
+      "eval_LOCATION_precision": 0.7532467532467533,
+      "eval_LOCATION_recall": 0.8055555555555556,
+      "eval_ORGANIZATION_f1": 0.8516129032258064,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7325581395348837,
-      "eval_ORGANIZATION_recall": 0.84,
-      "eval_PERSON_f1": 0.925170068027211,
+      "eval_ORGANIZATION_precision": 0.825,
+      "eval_ORGANIZATION_recall": 0.88,
+      "eval_PERSON_f1": 0.9042904290429041,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.9006622516556292,
-      "eval_PERSON_recall": 0.951048951048951,
-      "eval_QUANTITY_f1": 0.7083333333333334,
+      "eval_PERSON_precision": 0.85625,
+      "eval_PERSON_recall": 0.958041958041958,
+      "eval_QUANTITY_f1": 0.6792452830188679,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.68,
-      "eval_QUANTITY_recall": 0.7391304347826086,
-      "eval_TIME_f1": 0.8518518518518519,
+      "eval_QUANTITY_precision": 0.6,
+      "eval_QUANTITY_recall": 0.782608695652174,
+      "eval_TIME_f1": 0.8235294117647058,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8214285714285714,
-      "eval_TIME_recall": 0.8846153846153846,
-      "eval_loss": 0.3326149582862854,
-      "eval_overall_accuracy": 0.9647235426569928,
-      "eval_overall_f1": 0.8519040902679831,
-      "eval_overall_precision": 0.8162162162162162,
-      "eval_overall_recall": 0.8908554572271387,
-      "eval_runtime": 0.9055,
-      "eval_samples_per_second": 206.509,
-      "eval_steps_per_second": 3.313,
+      "eval_TIME_precision": 0.84,
+      "eval_TIME_recall": 0.8076923076923077,
+      "eval_loss": 0.3733396530151367,
+      "eval_overall_accuracy": 0.9597197898423818,
+      "eval_overall_f1": 0.8438818565400844,
+      "eval_overall_precision": 0.8064516129032258,
+      "eval_overall_recall": 0.8849557522123894,
+      "eval_runtime": 0.3652,
+      "eval_samples_per_second": 512.007,
+      "eval_steps_per_second": 8.214,
       "step": 5936
     },
     {
       "epoch": 57.0,
-      "grad_norm": 0.0004171407490503043,
+      "grad_norm": 0.0007011191919445992,
       "learning_rate": 2.15e-05,
-      "loss": 0.0008,
+      "loss": 0.0009,
       "step": 6042
     },
     {
       "epoch": 57.0,
-      "eval_LOCATION_f1": 0.8211920529801324,
+      "eval_LOCATION_f1": 0.8079470198675496,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7848101265822784,
-      "eval_LOCATION_recall": 0.8611111111111112,
-      "eval_ORGANIZATION_f1": 0.7950310559006211,
+      "eval_LOCATION_precision": 0.7721518987341772,
+      "eval_LOCATION_recall": 0.8472222222222222,
+      "eval_ORGANIZATION_f1": 0.858974358974359,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7441860465116279,
-      "eval_ORGANIZATION_recall": 0.8533333333333334,
-      "eval_PERSON_f1": 0.9090909090909092,
+      "eval_ORGANIZATION_precision": 0.8271604938271605,
+      "eval_ORGANIZATION_recall": 0.8933333333333333,
+      "eval_PERSON_f1": 0.8970099667774087,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8766233766233766,
+      "eval_PERSON_precision": 0.8544303797468354,
       "eval_PERSON_recall": 0.9440559440559441,
-      "eval_QUANTITY_f1": 0.68,
+      "eval_QUANTITY_f1": 0.7058823529411765,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.6296296296296297,
-      "eval_QUANTITY_recall": 0.7391304347826086,
-      "eval_TIME_f1": 0.8679245283018868,
+      "eval_QUANTITY_precision": 0.6428571428571429,
+      "eval_QUANTITY_recall": 0.782608695652174,
+      "eval_TIME_f1": 0.8148148148148148,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8518518518518519,
-      "eval_TIME_recall": 0.8846153846153846,
-      "eval_loss": 0.3388373553752899,
-      "eval_overall_accuracy": 0.9639729797348011,
-      "eval_overall_f1": 0.8455056179775281,
-      "eval_overall_precision": 0.806970509383378,
-      "eval_overall_recall": 0.887905604719764,
-      "eval_runtime": 0.9161,
-      "eval_samples_per_second": 204.134,
-      "eval_steps_per_second": 3.275,
+      "eval_TIME_precision": 0.7857142857142857,
+      "eval_TIME_recall": 0.8461538461538461,
+      "eval_loss": 0.3640208840370178,
+      "eval_overall_accuracy": 0.961220915686765,
+      "eval_overall_f1": 0.849929873772791,
+      "eval_overall_precision": 0.8101604278074866,
+      "eval_overall_recall": 0.8938053097345132,
+      "eval_runtime": 0.3606,
+      "eval_samples_per_second": 518.592,
+      "eval_steps_per_second": 8.32,
       "step": 6042
     },
     {
       "epoch": 58.0,
-      "grad_norm": 0.000792400271166116,
+      "grad_norm": 0.0015956854913383722,
       "learning_rate": 2.1e-05,
-      "loss": 0.0005,
+      "loss": 0.0003,
       "step": 6148
     },
     {
       "epoch": 58.0,
-      "eval_LOCATION_f1": 0.8344370860927153,
+      "eval_LOCATION_f1": 0.7973856209150327,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7974683544303798,
-      "eval_LOCATION_recall": 0.875,
-      "eval_ORGANIZATION_f1": 0.8121212121212122,
+      "eval_LOCATION_precision": 0.7530864197530864,
+      "eval_LOCATION_recall": 0.8472222222222222,
+      "eval_ORGANIZATION_f1": 0.8220858895705522,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7444444444444445,
+      "eval_ORGANIZATION_precision": 0.7613636363636364,
       "eval_ORGANIZATION_recall": 0.8933333333333333,
-      "eval_PERSON_f1": 0.8993288590604027,
+      "eval_PERSON_f1": 0.9023569023569024,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.864516129032258,
+      "eval_PERSON_precision": 0.8701298701298701,
       "eval_PERSON_recall": 0.9370629370629371,
-      "eval_QUANTITY_f1": 0.693877551020408,
+      "eval_QUANTITY_f1": 0.7058823529411765,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.6538461538461539,
-      "eval_QUANTITY_recall": 0.7391304347826086,
+      "eval_QUANTITY_precision": 0.6428571428571429,
+      "eval_QUANTITY_recall": 0.782608695652174,
       "eval_TIME_f1": 0.8518518518518519,
       "eval_TIME_number": 26,
       "eval_TIME_precision": 0.8214285714285714,
       "eval_TIME_recall": 0.8846153846153846,
-      "eval_loss": 0.3484926223754883,
-      "eval_overall_accuracy": 0.9622216662496873,
-      "eval_overall_f1": 0.8479776847977685,
-      "eval_overall_precision": 0.8042328042328042,
-      "eval_overall_recall": 0.8967551622418879,
-      "eval_runtime": 0.9061,
-      "eval_samples_per_second": 206.382,
-      "eval_steps_per_second": 3.311,
+      "eval_loss": 0.3814009130001068,
+      "eval_overall_accuracy": 0.9582186639979985,
+      "eval_overall_f1": 0.8440111420612812,
+      "eval_overall_precision": 0.7994722955145118,
+      "eval_overall_recall": 0.8938053097345132,
+      "eval_runtime": 0.3599,
+      "eval_samples_per_second": 519.642,
+      "eval_steps_per_second": 8.337,
       "step": 6148
     },
     {
       "epoch": 59.0,
-      "grad_norm": 0.0005213640397414565,
+      "grad_norm": 0.0003817932156380266,
       "learning_rate": 2.05e-05,
-      "loss": 0.0004,
+      "loss": 0.0005,
       "step": 6254
     },
     {
       "epoch": 59.0,
-      "eval_LOCATION_f1": 0.8211920529801324,
+      "eval_LOCATION_f1": 0.802721088435374,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7848101265822784,
-      "eval_LOCATION_recall": 0.8611111111111112,
-      "eval_ORGANIZATION_f1": 0.7950310559006211,
+      "eval_LOCATION_precision": 0.7866666666666666,
+      "eval_LOCATION_recall": 0.8194444444444444,
+      "eval_ORGANIZATION_f1": 0.8441558441558442,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7441860465116279,
-      "eval_ORGANIZATION_recall": 0.8533333333333334,
-      "eval_PERSON_f1": 0.9215017064846417,
+      "eval_ORGANIZATION_precision": 0.8227848101265823,
+      "eval_ORGANIZATION_recall": 0.8666666666666667,
+      "eval_PERSON_f1": 0.9096989966555185,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.9,
-      "eval_PERSON_recall": 0.9440559440559441,
-      "eval_QUANTITY_f1": 0.6666666666666666,
+      "eval_PERSON_precision": 0.8717948717948718,
+      "eval_PERSON_recall": 0.951048951048951,
+      "eval_QUANTITY_f1": 0.7058823529411765,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.6071428571428571,
-      "eval_QUANTITY_recall": 0.7391304347826086,
-      "eval_TIME_f1": 0.8627450980392156,
+      "eval_QUANTITY_precision": 0.6428571428571429,
+      "eval_QUANTITY_recall": 0.782608695652174,
+      "eval_TIME_f1": 0.888888888888889,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.88,
-      "eval_TIME_recall": 0.8461538461538461,
-      "eval_loss": 0.34474921226501465,
-      "eval_overall_accuracy": 0.9627220415311484,
-      "eval_overall_f1": 0.8486562942008488,
-      "eval_overall_precision": 0.8152173913043478,
-      "eval_overall_recall": 0.8849557522123894,
-      "eval_runtime": 0.9086,
-      "eval_samples_per_second": 205.807,
-      "eval_steps_per_second": 3.302,
+      "eval_TIME_precision": 0.8571428571428571,
+      "eval_TIME_recall": 0.9230769230769231,
+      "eval_loss": 0.3739457130432129,
+      "eval_overall_accuracy": 0.9619714786089567,
+      "eval_overall_f1": 0.8567375886524823,
+      "eval_overall_precision": 0.825136612021858,
+      "eval_overall_recall": 0.8908554572271387,
+      "eval_runtime": 0.3596,
+      "eval_samples_per_second": 520.068,
+      "eval_steps_per_second": 8.343,
       "step": 6254
     },
     {
       "epoch": 60.0,
-      "grad_norm": 0.0031596734188497066,
+      "grad_norm": 0.00037515757139772177,
       "learning_rate": 2e-05,
-      "loss": 0.0017,
+      "loss": 0.0005,
       "step": 6360
     },
     {
       "epoch": 60.0,
-      "eval_LOCATION_f1": 0.8289473684210527,
+      "eval_LOCATION_f1": 0.8219178082191781,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7875,
-      "eval_LOCATION_recall": 0.875,
-      "eval_ORGANIZATION_f1": 0.783625730994152,
+      "eval_LOCATION_precision": 0.8108108108108109,
+      "eval_LOCATION_recall": 0.8333333333333334,
+      "eval_ORGANIZATION_f1": 0.8354430379746836,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.6979166666666666,
-      "eval_ORGANIZATION_recall": 0.8933333333333333,
-      "eval_PERSON_f1": 0.8993288590604027,
+      "eval_ORGANIZATION_precision": 0.7951807228915663,
+      "eval_ORGANIZATION_recall": 0.88,
+      "eval_PERSON_f1": 0.9006622516556291,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.864516129032258,
-      "eval_PERSON_recall": 0.9370629370629371,
-      "eval_QUANTITY_f1": 0.6666666666666667,
+      "eval_PERSON_precision": 0.8553459119496856,
+      "eval_PERSON_recall": 0.951048951048951,
+      "eval_QUANTITY_f1": 0.72,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.5806451612903226,
+      "eval_QUANTITY_precision": 0.6666666666666666,
       "eval_QUANTITY_recall": 0.782608695652174,
-      "eval_TIME_f1": 0.8461538461538461,
+      "eval_TIME_f1": 0.8679245283018868,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8461538461538461,
-      "eval_TIME_recall": 0.8461538461538461,
-      "eval_loss": 0.3943552076816559,
-      "eval_overall_accuracy": 0.9559669752314236,
-      "eval_overall_f1": 0.8363136176066025,
-      "eval_overall_precision": 0.7835051546391752,
-      "eval_overall_recall": 0.8967551622418879,
-      "eval_runtime": 0.9068,
-      "eval_samples_per_second": 206.222,
-      "eval_steps_per_second": 3.308,
+      "eval_TIME_precision": 0.8518518518518519,
+      "eval_TIME_recall": 0.8846153846153846,
+      "eval_loss": 0.3766981363296509,
+      "eval_overall_accuracy": 0.9604703527645734,
+      "eval_overall_f1": 0.8547249647390691,
+      "eval_overall_precision": 0.8189189189189189,
+      "eval_overall_recall": 0.8938053097345132,
+      "eval_runtime": 0.3687,
+      "eval_samples_per_second": 507.159,
+      "eval_steps_per_second": 8.136,
       "step": 6360
     },
     {
       "epoch": 61.0,
-      "grad_norm": 0.0038910319563001394,
+      "grad_norm": 0.00191974185872823,
       "learning_rate": 1.9500000000000003e-05,
-      "loss": 0.0002,
+      "loss": 0.0008,
       "step": 6466
     },
     {
       "epoch": 61.0,
-      "eval_LOCATION_f1": 0.8322147651006712,
+      "eval_LOCATION_f1": 0.816326530612245,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.8051948051948052,
-      "eval_LOCATION_recall": 0.8611111111111112,
-      "eval_ORGANIZATION_f1": 0.8,
+      "eval_LOCATION_precision": 0.8,
+      "eval_LOCATION_recall": 0.8333333333333334,
+      "eval_ORGANIZATION_f1": 0.8333333333333334,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7333333333333333,
-      "eval_ORGANIZATION_recall": 0.88,
-      "eval_PERSON_f1": 0.8963210702341137,
+      "eval_ORGANIZATION_precision": 0.8024691358024691,
+      "eval_ORGANIZATION_recall": 0.8666666666666667,
+      "eval_PERSON_f1": 0.903654485049834,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8589743589743589,
-      "eval_PERSON_recall": 0.9370629370629371,
-      "eval_QUANTITY_f1": 0.7346938775510203,
+      "eval_PERSON_precision": 0.8607594936708861,
+      "eval_PERSON_recall": 0.951048951048951,
+      "eval_QUANTITY_f1": 0.7058823529411765,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.6923076923076923,
+      "eval_QUANTITY_precision": 0.6428571428571429,
       "eval_QUANTITY_recall": 0.782608695652174,
-      "eval_TIME_f1": 0.830188679245283,
+      "eval_TIME_f1": 0.8214285714285715,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8148148148148148,
-      "eval_TIME_recall": 0.8461538461538461,
-      "eval_loss": 0.38600656390190125,
-      "eval_overall_accuracy": 0.9599699774831123,
-      "eval_overall_f1": 0.8447552447552448,
-      "eval_overall_precision": 0.8031914893617021,
+      "eval_TIME_precision": 0.7666666666666667,
+      "eval_TIME_recall": 0.8846153846153846,
+      "eval_loss": 0.36986875534057617,
+      "eval_overall_accuracy": 0.9622216662496873,
+      "eval_overall_f1": 0.849507735583685,
+      "eval_overall_precision": 0.8118279569892473,
       "eval_overall_recall": 0.8908554572271387,
-      "eval_runtime": 0.9064,
-      "eval_samples_per_second": 206.303,
-      "eval_steps_per_second": 3.31,
+      "eval_runtime": 0.3616,
+      "eval_samples_per_second": 517.178,
+      "eval_steps_per_second": 8.297,
       "step": 6466
     },
     {
       "epoch": 62.0,
-      "grad_norm": 0.008979488164186478,
+      "grad_norm": 1.7503353357315063,
       "learning_rate": 1.9e-05,
-      "loss": 0.0001,
+      "loss": 0.0016,
       "step": 6572
     },
     {
@@ -2400,583 +2400,583 @@
       "eval_LOCATION_number": 72,
       "eval_LOCATION_precision": 0.7922077922077922,
       "eval_LOCATION_recall": 0.8472222222222222,
-      "eval_ORGANIZATION_f1": 0.8176100628930819,
+      "eval_ORGANIZATION_f1": 0.825,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7738095238095238,
-      "eval_ORGANIZATION_recall": 0.8666666666666667,
-      "eval_PERSON_f1": 0.9047619047619047,
+      "eval_ORGANIZATION_precision": 0.7764705882352941,
+      "eval_ORGANIZATION_recall": 0.88,
+      "eval_PERSON_f1": 0.8838709677419354,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8807947019867549,
-      "eval_PERSON_recall": 0.9300699300699301,
-      "eval_QUANTITY_f1": 0.7083333333333334,
+      "eval_PERSON_precision": 0.8203592814371258,
+      "eval_PERSON_recall": 0.958041958041958,
+      "eval_QUANTITY_f1": 0.7058823529411765,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.68,
-      "eval_QUANTITY_recall": 0.7391304347826086,
-      "eval_TIME_f1": 0.830188679245283,
+      "eval_QUANTITY_precision": 0.6428571428571429,
+      "eval_QUANTITY_recall": 0.782608695652174,
+      "eval_TIME_f1": 0.8679245283018868,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8148148148148148,
-      "eval_TIME_recall": 0.8461538461538461,
-      "eval_loss": 0.3512948453426361,
-      "eval_overall_accuracy": 0.9629722291718789,
-      "eval_overall_f1": 0.8477951635846371,
-      "eval_overall_precision": 0.8186813186813187,
-      "eval_overall_recall": 0.8790560471976401,
-      "eval_runtime": 0.9032,
-      "eval_samples_per_second": 207.035,
-      "eval_steps_per_second": 3.321,
+      "eval_TIME_precision": 0.8518518518518519,
+      "eval_TIME_recall": 0.8846153846153846,
+      "eval_loss": 0.34973686933517456,
+      "eval_overall_accuracy": 0.961220915686765,
+      "eval_overall_f1": 0.8437067773167358,
+      "eval_overall_precision": 0.7942708333333334,
+      "eval_overall_recall": 0.8997050147492626,
+      "eval_runtime": 0.364,
+      "eval_samples_per_second": 513.749,
+      "eval_steps_per_second": 8.242,
       "step": 6572
     },
     {
       "epoch": 63.0,
-      "grad_norm": 0.00276816263794899,
+      "grad_norm": 0.0005617731949314475,
       "learning_rate": 1.85e-05,
-      "loss": 0.0013,
+      "loss": 0.0004,
       "step": 6678
     },
     {
       "epoch": 63.0,
-      "eval_LOCATION_f1": 0.8289473684210527,
+      "eval_LOCATION_f1": 0.8299319727891157,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7875,
-      "eval_LOCATION_recall": 0.875,
-      "eval_ORGANIZATION_f1": 0.7904191616766466,
+      "eval_LOCATION_precision": 0.8133333333333334,
+      "eval_LOCATION_recall": 0.8472222222222222,
+      "eval_ORGANIZATION_f1": 0.834355828220859,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.717391304347826,
-      "eval_ORGANIZATION_recall": 0.88,
-      "eval_PERSON_f1": 0.9060402684563759,
+      "eval_ORGANIZATION_precision": 0.7727272727272727,
+      "eval_ORGANIZATION_recall": 0.9066666666666666,
+      "eval_PERSON_f1": 0.9210526315789473,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8709677419354839,
-      "eval_PERSON_recall": 0.9440559440559441,
-      "eval_QUANTITY_f1": 0.7083333333333334,
+      "eval_PERSON_precision": 0.8695652173913043,
+      "eval_PERSON_recall": 0.9790209790209791,
+      "eval_QUANTITY_f1": 0.7058823529411765,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.68,
-      "eval_QUANTITY_recall": 0.7391304347826086,
-      "eval_TIME_f1": 0.830188679245283,
+      "eval_QUANTITY_precision": 0.6428571428571429,
+      "eval_QUANTITY_recall": 0.782608695652174,
+      "eval_TIME_f1": 0.7796610169491526,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8148148148148148,
-      "eval_TIME_recall": 0.8461538461538461,
-      "eval_loss": 0.3540952205657959,
-      "eval_overall_accuracy": 0.9629722291718789,
-      "eval_overall_f1": 0.8440111420612812,
-      "eval_overall_precision": 0.7994722955145118,
-      "eval_overall_recall": 0.8938053097345132,
-      "eval_runtime": 0.9089,
-      "eval_samples_per_second": 205.74,
-      "eval_steps_per_second": 3.301,
+      "eval_TIME_precision": 0.696969696969697,
+      "eval_TIME_recall": 0.8846153846153846,
+      "eval_loss": 0.3623858094215393,
+      "eval_overall_accuracy": 0.9617212909682261,
+      "eval_overall_f1": 0.856353591160221,
+      "eval_overall_precision": 0.8051948051948052,
+      "eval_overall_recall": 0.9144542772861357,
+      "eval_runtime": 0.3601,
+      "eval_samples_per_second": 519.319,
+      "eval_steps_per_second": 8.331,
       "step": 6678
     },
     {
       "epoch": 64.0,
-      "grad_norm": 0.010789873078465462,
+      "grad_norm": 0.04498714208602905,
       "learning_rate": 1.8e-05,
-      "loss": 0.0007,
+      "loss": 0.0009,
       "step": 6784
     },
     {
       "epoch": 64.0,
-      "eval_LOCATION_f1": 0.779874213836478,
+      "eval_LOCATION_f1": 0.8435374149659863,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7126436781609196,
+      "eval_LOCATION_precision": 0.8266666666666667,
       "eval_LOCATION_recall": 0.8611111111111112,
-      "eval_ORGANIZATION_f1": 0.7848101265822784,
+      "eval_ORGANIZATION_f1": 0.8374999999999999,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7469879518072289,
-      "eval_ORGANIZATION_recall": 0.8266666666666667,
-      "eval_PERSON_f1": 0.9090909090909092,
+      "eval_ORGANIZATION_precision": 0.788235294117647,
+      "eval_ORGANIZATION_recall": 0.8933333333333333,
+      "eval_PERSON_f1": 0.9006622516556291,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8766233766233766,
-      "eval_PERSON_recall": 0.9440559440559441,
-      "eval_QUANTITY_f1": 0.7058823529411765,
+      "eval_PERSON_precision": 0.8553459119496856,
+      "eval_PERSON_recall": 0.951048951048951,
+      "eval_QUANTITY_f1": 0.72,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.6428571428571429,
+      "eval_QUANTITY_precision": 0.6666666666666666,
       "eval_QUANTITY_recall": 0.782608695652174,
       "eval_TIME_f1": 0.8518518518518519,
       "eval_TIME_number": 26,
       "eval_TIME_precision": 0.8214285714285714,
       "eval_TIME_recall": 0.8846153846153846,
-      "eval_loss": 0.34456735849380493,
-      "eval_overall_accuracy": 0.9619714786089567,
-      "eval_overall_f1": 0.8344923504867873,
-      "eval_overall_precision": 0.7894736842105263,
-      "eval_overall_recall": 0.8849557522123894,
-      "eval_runtime": 0.9096,
-      "eval_samples_per_second": 205.587,
-      "eval_steps_per_second": 3.298,
+      "eval_loss": 0.37043052911758423,
+      "eval_overall_accuracy": 0.9602201651238429,
+      "eval_overall_f1": 0.8583450210378682,
+      "eval_overall_precision": 0.8181818181818182,
+      "eval_overall_recall": 0.9026548672566371,
+      "eval_runtime": 0.36,
+      "eval_samples_per_second": 519.423,
+      "eval_steps_per_second": 8.333,
       "step": 6784
     },
     {
       "epoch": 65.0,
-      "grad_norm": 0.0003090931277256459,
+      "grad_norm": 0.0003102279151789844,
       "learning_rate": 1.75e-05,
-      "loss": 0.0008,
+      "loss": 0.001,
       "step": 6890
     },
     {
       "epoch": 65.0,
-      "eval_LOCATION_f1": 0.8181818181818181,
+      "eval_LOCATION_f1": 0.8266666666666667,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7682926829268293,
-      "eval_LOCATION_recall": 0.875,
-      "eval_ORGANIZATION_f1": 0.782051282051282,
+      "eval_LOCATION_precision": 0.7948717948717948,
+      "eval_LOCATION_recall": 0.8611111111111112,
+      "eval_ORGANIZATION_f1": 0.8220858895705522,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7530864197530864,
-      "eval_ORGANIZATION_recall": 0.8133333333333334,
-      "eval_PERSON_f1": 0.9084745762711864,
+      "eval_ORGANIZATION_precision": 0.7613636363636364,
+      "eval_ORGANIZATION_recall": 0.8933333333333333,
+      "eval_PERSON_f1": 0.9144736842105262,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.881578947368421,
-      "eval_PERSON_recall": 0.9370629370629371,
-      "eval_QUANTITY_f1": 0.7346938775510203,
+      "eval_PERSON_precision": 0.8633540372670807,
+      "eval_PERSON_recall": 0.972027972027972,
+      "eval_QUANTITY_f1": 0.72,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.6923076923076923,
+      "eval_QUANTITY_precision": 0.6666666666666666,
       "eval_QUANTITY_recall": 0.782608695652174,
-      "eval_TIME_f1": 0.830188679245283,
+      "eval_TIME_f1": 0.8679245283018868,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8148148148148148,
-      "eval_TIME_recall": 0.8461538461538461,
-      "eval_loss": 0.3443866968154907,
-      "eval_overall_accuracy": 0.9629722291718789,
-      "eval_overall_f1": 0.8429985855728431,
-      "eval_overall_precision": 0.8097826086956522,
-      "eval_overall_recall": 0.8790560471976401,
-      "eval_runtime": 0.906,
-      "eval_samples_per_second": 206.397,
-      "eval_steps_per_second": 3.311,
+      "eval_TIME_precision": 0.8518518518518519,
+      "eval_TIME_recall": 0.8846153846153846,
+      "eval_loss": 0.3582930862903595,
+      "eval_overall_accuracy": 0.96347260445334,
+      "eval_overall_f1": 0.8583333333333333,
+      "eval_overall_precision": 0.8110236220472441,
+      "eval_overall_recall": 0.911504424778761,
+      "eval_runtime": 0.3574,
+      "eval_samples_per_second": 523.156,
+      "eval_steps_per_second": 8.393,
       "step": 6890
     },
     {
       "epoch": 66.0,
-      "grad_norm": 0.008000018075108528,
+      "grad_norm": 0.023364154621958733,
       "learning_rate": 1.7000000000000003e-05,
-      "loss": 0.0007,
+      "loss": 0.0001,
       "step": 6996
     },
     {
       "epoch": 66.0,
-      "eval_LOCATION_f1": 0.823529411764706,
+      "eval_LOCATION_f1": 0.7894736842105262,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7777777777777778,
-      "eval_LOCATION_recall": 0.875,
-      "eval_ORGANIZATION_f1": 0.7564102564102564,
+      "eval_LOCATION_precision": 0.75,
+      "eval_LOCATION_recall": 0.8333333333333334,
+      "eval_ORGANIZATION_f1": 0.8387096774193549,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7283950617283951,
-      "eval_ORGANIZATION_recall": 0.7866666666666666,
-      "eval_PERSON_f1": 0.9121621621621621,
+      "eval_ORGANIZATION_precision": 0.8125,
+      "eval_ORGANIZATION_recall": 0.8666666666666667,
+      "eval_PERSON_f1": 0.9139072847682119,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8823529411764706,
-      "eval_PERSON_recall": 0.9440559440559441,
-      "eval_QUANTITY_f1": 0.7083333333333334,
+      "eval_PERSON_precision": 0.8679245283018868,
+      "eval_PERSON_recall": 0.965034965034965,
+      "eval_QUANTITY_f1": 0.72,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.68,
-      "eval_QUANTITY_recall": 0.7391304347826086,
-      "eval_TIME_f1": 0.830188679245283,
+      "eval_QUANTITY_precision": 0.6666666666666666,
+      "eval_QUANTITY_recall": 0.782608695652174,
+      "eval_TIME_f1": 0.8461538461538461,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8148148148148148,
+      "eval_TIME_precision": 0.8461538461538461,
       "eval_TIME_recall": 0.8461538461538461,
-      "eval_loss": 0.3399989902973175,
+      "eval_loss": 0.35251596570014954,
       "eval_overall_accuracy": 0.9627220415311484,
-      "eval_overall_f1": 0.8385269121813032,
-      "eval_overall_precision": 0.8065395095367848,
-      "eval_overall_recall": 0.8731563421828908,
-      "eval_runtime": 0.9215,
-      "eval_samples_per_second": 202.925,
-      "eval_steps_per_second": 3.255,
+      "eval_overall_f1": 0.8523206751054851,
+      "eval_overall_precision": 0.8145161290322581,
+      "eval_overall_recall": 0.8938053097345132,
+      "eval_runtime": 0.3673,
+      "eval_samples_per_second": 509.16,
+      "eval_steps_per_second": 8.168,
       "step": 6996
     },
     {
       "epoch": 67.0,
-      "grad_norm": 0.00037926252116449177,
+      "grad_norm": 0.0011218636063858867,
       "learning_rate": 1.65e-05,
-      "loss": 0.0004,
+      "loss": 0.0015,
       "step": 7102
     },
     {
       "epoch": 67.0,
-      "eval_LOCATION_f1": 0.8289473684210527,
+      "eval_LOCATION_f1": 0.823529411764706,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7875,
+      "eval_LOCATION_precision": 0.7777777777777778,
       "eval_LOCATION_recall": 0.875,
-      "eval_ORGANIZATION_f1": 0.8148148148148148,
+      "eval_ORGANIZATION_f1": 0.8121212121212122,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7586206896551724,
-      "eval_ORGANIZATION_recall": 0.88,
-      "eval_PERSON_f1": 0.9090909090909092,
-      "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8766233766233766,
-      "eval_PERSON_recall": 0.9440559440559441,
-      "eval_QUANTITY_f1": 0.7499999999999999,
+      "eval_ORGANIZATION_precision": 0.7444444444444445,
+      "eval_ORGANIZATION_recall": 0.8933333333333333,
+      "eval_PERSON_f1": 0.9078947368421052,
+      "eval_PERSON_number": 143,
+      "eval_PERSON_precision": 0.8571428571428571,
+      "eval_PERSON_recall": 0.965034965034965,
+      "eval_QUANTITY_f1": 0.72,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.72,
+      "eval_QUANTITY_precision": 0.6666666666666666,
       "eval_QUANTITY_recall": 0.782608695652174,
-      "eval_TIME_f1": 0.830188679245283,
+      "eval_TIME_f1": 0.8679245283018868,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8148148148148148,
-      "eval_TIME_recall": 0.8461538461538461,
-      "eval_loss": 0.3691372275352478,
-      "eval_overall_accuracy": 0.96347260445334,
-      "eval_overall_f1": 0.853932584269663,
-      "eval_overall_precision": 0.8150134048257373,
-      "eval_overall_recall": 0.8967551622418879,
-      "eval_runtime": 0.9131,
-      "eval_samples_per_second": 204.802,
-      "eval_steps_per_second": 3.286,
+      "eval_TIME_precision": 0.8518518518518519,
+      "eval_TIME_recall": 0.8846153846153846,
+      "eval_loss": 0.37641996145248413,
+      "eval_overall_accuracy": 0.9602201651238429,
+      "eval_overall_f1": 0.8524137931034482,
+      "eval_overall_precision": 0.8005181347150259,
+      "eval_overall_recall": 0.911504424778761,
+      "eval_runtime": 0.3613,
+      "eval_samples_per_second": 517.565,
+      "eval_steps_per_second": 8.303,
       "step": 7102
     },
     {
       "epoch": 68.0,
-      "grad_norm": 0.0003724870621226728,
+      "grad_norm": 0.0007424333016388118,
       "learning_rate": 1.6000000000000003e-05,
-      "loss": 0.0012,
+      "loss": 0.0006,
       "step": 7208
     },
     {
       "epoch": 68.0,
-      "eval_LOCATION_f1": 0.8322147651006712,
+      "eval_LOCATION_f1": 0.8356164383561645,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.8051948051948052,
-      "eval_LOCATION_recall": 0.8611111111111112,
-      "eval_ORGANIZATION_f1": 0.8280254777070064,
+      "eval_LOCATION_precision": 0.8243243243243243,
+      "eval_LOCATION_recall": 0.8472222222222222,
+      "eval_ORGANIZATION_f1": 0.7878787878787877,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7926829268292683,
+      "eval_ORGANIZATION_precision": 0.7222222222222222,
       "eval_ORGANIZATION_recall": 0.8666666666666667,
-      "eval_PERSON_f1": 0.9121621621621621,
+      "eval_PERSON_f1": 0.9133333333333333,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8823529411764706,
-      "eval_PERSON_recall": 0.9440559440559441,
-      "eval_QUANTITY_f1": 0.7083333333333334,
+      "eval_PERSON_precision": 0.8726114649681529,
+      "eval_PERSON_recall": 0.958041958041958,
+      "eval_QUANTITY_f1": 0.7058823529411765,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.68,
-      "eval_QUANTITY_recall": 0.7391304347826086,
-      "eval_TIME_f1": 0.8627450980392156,
+      "eval_QUANTITY_precision": 0.6428571428571429,
+      "eval_QUANTITY_recall": 0.782608695652174,
+      "eval_TIME_f1": 0.830188679245283,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.88,
+      "eval_TIME_precision": 0.8148148148148148,
       "eval_TIME_recall": 0.8461538461538461,
-      "eval_loss": 0.33306267857551575,
-      "eval_overall_accuracy": 0.9654741055791843,
-      "eval_overall_f1": 0.8587731811697574,
-      "eval_overall_precision": 0.8314917127071824,
-      "eval_overall_recall": 0.887905604719764,
-      "eval_runtime": 0.9838,
-      "eval_samples_per_second": 190.074,
-      "eval_steps_per_second": 3.049,
+      "eval_loss": 0.3496319651603699,
+      "eval_overall_accuracy": 0.9594696022016512,
+      "eval_overall_f1": 0.8475524475524475,
+      "eval_overall_precision": 0.8058510638297872,
+      "eval_overall_recall": 0.8938053097345132,
+      "eval_runtime": 0.3711,
+      "eval_samples_per_second": 503.904,
+      "eval_steps_per_second": 8.084,
       "step": 7208
     },
     {
       "epoch": 69.0,
-      "grad_norm": 0.0002074290969176218,
+      "grad_norm": 0.008528614416718483,
       "learning_rate": 1.55e-05,
-      "loss": 0.0003,
+      "loss": 0.0008,
       "step": 7314
     },
     {
       "epoch": 69.0,
-      "eval_LOCATION_f1": 0.8344370860927153,
+      "eval_LOCATION_f1": 0.8187919463087249,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7974683544303798,
-      "eval_LOCATION_recall": 0.875,
-      "eval_ORGANIZATION_f1": 0.8170731707317072,
+      "eval_LOCATION_precision": 0.7922077922077922,
+      "eval_LOCATION_recall": 0.8472222222222222,
+      "eval_ORGANIZATION_f1": 0.8000000000000002,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7528089887640449,
-      "eval_ORGANIZATION_recall": 0.8933333333333333,
-      "eval_PERSON_f1": 0.9054054054054055,
+      "eval_ORGANIZATION_precision": 0.775,
+      "eval_ORGANIZATION_recall": 0.8266666666666667,
+      "eval_PERSON_f1": 0.8970099667774087,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8758169934640523,
-      "eval_PERSON_recall": 0.9370629370629371,
-      "eval_QUANTITY_f1": 0.7499999999999999,
+      "eval_PERSON_precision": 0.8544303797468354,
+      "eval_PERSON_recall": 0.9440559440559441,
+      "eval_QUANTITY_f1": 0.693877551020408,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.72,
-      "eval_QUANTITY_recall": 0.782608695652174,
-      "eval_TIME_f1": 0.8461538461538461,
+      "eval_QUANTITY_precision": 0.6538461538461539,
+      "eval_QUANTITY_recall": 0.7391304347826086,
+      "eval_TIME_f1": 0.830188679245283,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8461538461538461,
+      "eval_TIME_precision": 0.8148148148148148,
       "eval_TIME_recall": 0.8461538461538461,
-      "eval_loss": 0.373975545167923,
-      "eval_overall_accuracy": 0.96347260445334,
-      "eval_overall_f1": 0.8551336146272857,
-      "eval_overall_precision": 0.8172043010752689,
-      "eval_overall_recall": 0.8967551622418879,
-      "eval_runtime": 0.9156,
-      "eval_samples_per_second": 204.238,
-      "eval_steps_per_second": 3.277,
+      "eval_loss": 0.34753233194351196,
+      "eval_overall_accuracy": 0.9594696022016512,
+      "eval_overall_f1": 0.8401697312588402,
+      "eval_overall_precision": 0.8070652173913043,
+      "eval_overall_recall": 0.8761061946902655,
+      "eval_runtime": 0.3588,
+      "eval_samples_per_second": 521.137,
+      "eval_steps_per_second": 8.36,
       "step": 7314
     },
     {
       "epoch": 70.0,
-      "grad_norm": 0.00042664690408855677,
+      "grad_norm": 0.014132479205727577,
       "learning_rate": 1.5e-05,
-      "loss": 0.0001,
+      "loss": 0.0,
       "step": 7420
     },
     {
       "epoch": 70.0,
-      "eval_LOCATION_f1": 0.8211920529801324,
+      "eval_LOCATION_f1": 0.8356164383561645,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7848101265822784,
-      "eval_LOCATION_recall": 0.8611111111111112,
-      "eval_ORGANIZATION_f1": 0.8025477707006369,
+      "eval_LOCATION_precision": 0.8243243243243243,
+      "eval_LOCATION_recall": 0.8472222222222222,
+      "eval_ORGANIZATION_f1": 0.8121212121212122,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7682926829268293,
-      "eval_ORGANIZATION_recall": 0.84,
-      "eval_PERSON_f1": 0.9054054054054055,
+      "eval_ORGANIZATION_precision": 0.7444444444444445,
+      "eval_ORGANIZATION_recall": 0.8933333333333333,
+      "eval_PERSON_f1": 0.8844884488448846,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8758169934640523,
+      "eval_PERSON_precision": 0.8375,
       "eval_PERSON_recall": 0.9370629370629371,
-      "eval_QUANTITY_f1": 0.7499999999999999,
+      "eval_QUANTITY_f1": 0.68,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.72,
-      "eval_QUANTITY_recall": 0.782608695652174,
-      "eval_TIME_f1": 0.8461538461538461,
+      "eval_QUANTITY_precision": 0.6296296296296297,
+      "eval_QUANTITY_recall": 0.7391304347826086,
+      "eval_TIME_f1": 0.8214285714285715,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8461538461538461,
-      "eval_TIME_recall": 0.8461538461538461,
-      "eval_loss": 0.37100422382354736,
-      "eval_overall_accuracy": 0.96347260445334,
-      "eval_overall_f1": 0.8494318181818181,
-      "eval_overall_precision": 0.8191780821917808,
-      "eval_overall_recall": 0.8820058997050148,
-      "eval_runtime": 0.9101,
-      "eval_samples_per_second": 205.482,
-      "eval_steps_per_second": 3.297,
+      "eval_TIME_precision": 0.7666666666666667,
+      "eval_TIME_recall": 0.8846153846153846,
+      "eval_loss": 0.3662912845611572,
+      "eval_overall_accuracy": 0.9592194145609206,
+      "eval_overall_f1": 0.8388888888888889,
+      "eval_overall_precision": 0.7926509186351706,
+      "eval_overall_recall": 0.8908554572271387,
+      "eval_runtime": 0.4097,
+      "eval_samples_per_second": 456.444,
+      "eval_steps_per_second": 7.323,
       "step": 7420
     },
     {
       "epoch": 71.0,
-      "grad_norm": 0.0007093641324900091,
+      "grad_norm": 0.0005066086887381971,
       "learning_rate": 1.45e-05,
       "loss": 0.0002,
       "step": 7526
     },
     {
       "epoch": 71.0,
-      "eval_LOCATION_f1": 0.8266666666666667,
+      "eval_LOCATION_f1": 0.8055555555555556,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7948717948717948,
-      "eval_LOCATION_recall": 0.8611111111111112,
-      "eval_ORGANIZATION_f1": 0.8101265822784811,
+      "eval_LOCATION_precision": 0.8055555555555556,
+      "eval_LOCATION_recall": 0.8055555555555556,
+      "eval_ORGANIZATION_f1": 0.8170731707317072,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7710843373493976,
-      "eval_ORGANIZATION_recall": 0.8533333333333334,
-      "eval_PERSON_f1": 0.91156462585034,
+      "eval_ORGANIZATION_precision": 0.7528089887640449,
+      "eval_ORGANIZATION_recall": 0.8933333333333333,
+      "eval_PERSON_f1": 0.9194630872483223,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8874172185430463,
-      "eval_PERSON_recall": 0.9370629370629371,
-      "eval_QUANTITY_f1": 0.7083333333333334,
+      "eval_PERSON_precision": 0.8838709677419355,
+      "eval_PERSON_recall": 0.958041958041958,
+      "eval_QUANTITY_f1": 0.68,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.68,
+      "eval_QUANTITY_precision": 0.6296296296296297,
       "eval_QUANTITY_recall": 0.7391304347826086,
-      "eval_TIME_f1": 0.8461538461538461,
+      "eval_TIME_f1": 0.8363636363636363,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8461538461538461,
-      "eval_TIME_recall": 0.8461538461538461,
-      "eval_loss": 0.3477381467819214,
-      "eval_overall_accuracy": 0.9647235426569928,
-      "eval_overall_f1": 0.8518518518518517,
-      "eval_overall_precision": 0.8236914600550964,
-      "eval_overall_recall": 0.8820058997050148,
-      "eval_runtime": 0.9072,
-      "eval_samples_per_second": 206.129,
-      "eval_steps_per_second": 3.307,
+      "eval_TIME_precision": 0.7931034482758621,
+      "eval_TIME_recall": 0.8846153846153846,
+      "eval_loss": 0.3715399205684662,
+      "eval_overall_accuracy": 0.9597197898423818,
+      "eval_overall_f1": 0.849507735583685,
+      "eval_overall_precision": 0.8118279569892473,
+      "eval_overall_recall": 0.8908554572271387,
+      "eval_runtime": 0.3561,
+      "eval_samples_per_second": 525.071,
+      "eval_steps_per_second": 8.424,
       "step": 7526
     },
     {
       "epoch": 72.0,
-      "grad_norm": 0.00021990617096889764,
+      "grad_norm": 0.00028915383154526353,
       "learning_rate": 1.4000000000000001e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 7632
     },
     {
       "epoch": 72.0,
-      "eval_LOCATION_f1": 0.823529411764706,
+      "eval_LOCATION_f1": 0.8435374149659863,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7777777777777778,
-      "eval_LOCATION_recall": 0.875,
-      "eval_ORGANIZATION_f1": 0.7950310559006211,
+      "eval_LOCATION_precision": 0.8266666666666667,
+      "eval_LOCATION_recall": 0.8611111111111112,
+      "eval_ORGANIZATION_f1": 0.8072289156626506,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7441860465116279,
-      "eval_ORGANIZATION_recall": 0.8533333333333334,
-      "eval_PERSON_f1": 0.9084745762711864,
+      "eval_ORGANIZATION_precision": 0.7362637362637363,
+      "eval_ORGANIZATION_recall": 0.8933333333333333,
+      "eval_PERSON_f1": 0.9163879598662208,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.881578947368421,
-      "eval_PERSON_recall": 0.9370629370629371,
-      "eval_QUANTITY_f1": 0.7499999999999999,
+      "eval_PERSON_precision": 0.8782051282051282,
+      "eval_PERSON_recall": 0.958041958041958,
+      "eval_QUANTITY_f1": 0.68,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.72,
-      "eval_QUANTITY_recall": 0.782608695652174,
-      "eval_TIME_f1": 0.830188679245283,
+      "eval_QUANTITY_precision": 0.6296296296296297,
+      "eval_QUANTITY_recall": 0.7391304347826086,
+      "eval_TIME_f1": 0.8214285714285715,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8148148148148148,
-      "eval_TIME_recall": 0.8461538461538461,
-      "eval_loss": 0.36287015676498413,
-      "eval_overall_accuracy": 0.9642231673755316,
-      "eval_overall_f1": 0.8478873239436621,
-      "eval_overall_precision": 0.8113207547169812,
-      "eval_overall_recall": 0.887905604719764,
-      "eval_runtime": 0.9141,
-      "eval_samples_per_second": 204.572,
-      "eval_steps_per_second": 3.282,
+      "eval_TIME_precision": 0.7666666666666667,
+      "eval_TIME_recall": 0.8846153846153846,
+      "eval_loss": 0.3853207230567932,
+      "eval_overall_accuracy": 0.9587190392794596,
+      "eval_overall_f1": 0.8523676880222841,
+      "eval_overall_precision": 0.8073878627968337,
+      "eval_overall_recall": 0.9026548672566371,
+      "eval_runtime": 0.362,
+      "eval_samples_per_second": 516.586,
+      "eval_steps_per_second": 8.287,
       "step": 7632
     },
     {
       "epoch": 73.0,
-      "grad_norm": 0.00021961626771371812,
+      "grad_norm": 0.0008943734574131668,
       "learning_rate": 1.3500000000000001e-05,
-      "loss": 0.0002,
+      "loss": 0.0007,
       "step": 7738
     },
     {
       "epoch": 73.0,
-      "eval_LOCATION_f1": 0.8181818181818181,
+      "eval_LOCATION_f1": 0.8275862068965517,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7682926829268293,
-      "eval_LOCATION_recall": 0.875,
-      "eval_ORGANIZATION_f1": 0.7643312101910827,
+      "eval_LOCATION_precision": 0.821917808219178,
+      "eval_LOCATION_recall": 0.8333333333333334,
+      "eval_ORGANIZATION_f1": 0.8198757763975156,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7317073170731707,
-      "eval_ORGANIZATION_recall": 0.8,
-      "eval_PERSON_f1": 0.91156462585034,
+      "eval_ORGANIZATION_precision": 0.7674418604651163,
+      "eval_ORGANIZATION_recall": 0.88,
+      "eval_PERSON_f1": 0.9060402684563759,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8874172185430463,
-      "eval_PERSON_recall": 0.9370629370629371,
-      "eval_QUANTITY_f1": 0.7499999999999999,
+      "eval_PERSON_precision": 0.8709677419354839,
+      "eval_PERSON_recall": 0.9440559440559441,
+      "eval_QUANTITY_f1": 0.68,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.72,
-      "eval_QUANTITY_recall": 0.782608695652174,
-      "eval_TIME_f1": 0.830188679245283,
+      "eval_QUANTITY_precision": 0.6296296296296297,
+      "eval_QUANTITY_recall": 0.7391304347826086,
+      "eval_TIME_f1": 0.8461538461538461,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8148148148148148,
+      "eval_TIME_precision": 0.8461538461538461,
       "eval_TIME_recall": 0.8461538461538461,
-      "eval_loss": 0.3637426495552063,
-      "eval_overall_accuracy": 0.9639729797348011,
-      "eval_overall_f1": 0.8413597733711049,
-      "eval_overall_precision": 0.8092643051771117,
-      "eval_overall_recall": 0.8761061946902655,
-      "eval_runtime": 0.909,
-      "eval_samples_per_second": 205.714,
-      "eval_steps_per_second": 3.3,
+      "eval_loss": 0.35061854124069214,
+      "eval_overall_accuracy": 0.9599699774831123,
+      "eval_overall_f1": 0.8498583569405098,
+      "eval_overall_precision": 0.8174386920980926,
+      "eval_overall_recall": 0.8849557522123894,
+      "eval_runtime": 0.357,
+      "eval_samples_per_second": 523.747,
+      "eval_steps_per_second": 8.402,
       "step": 7738
     },
     {
       "epoch": 74.0,
-      "grad_norm": 0.00018741752137430012,
+      "grad_norm": 0.0004187956510577351,
       "learning_rate": 1.3000000000000001e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 7844
     },
     {
       "epoch": 74.0,
-      "eval_LOCATION_f1": 0.8181818181818181,
+      "eval_LOCATION_f1": 0.8322147651006712,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7682926829268293,
-      "eval_LOCATION_recall": 0.875,
-      "eval_ORGANIZATION_f1": 0.751592356687898,
+      "eval_LOCATION_precision": 0.8051948051948052,
+      "eval_LOCATION_recall": 0.8611111111111112,
+      "eval_ORGANIZATION_f1": 0.8095238095238095,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7195121951219512,
-      "eval_ORGANIZATION_recall": 0.7866666666666666,
-      "eval_PERSON_f1": 0.9047619047619047,
+      "eval_ORGANIZATION_precision": 0.7311827956989247,
+      "eval_ORGANIZATION_recall": 0.9066666666666666,
+      "eval_PERSON_f1": 0.9163879598662208,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8807947019867549,
-      "eval_PERSON_recall": 0.9300699300699301,
-      "eval_QUANTITY_f1": 0.7499999999999999,
+      "eval_PERSON_precision": 0.8782051282051282,
+      "eval_PERSON_recall": 0.958041958041958,
+      "eval_QUANTITY_f1": 0.6538461538461539,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.72,
-      "eval_QUANTITY_recall": 0.782608695652174,
-      "eval_TIME_f1": 0.830188679245283,
+      "eval_QUANTITY_precision": 0.5862068965517241,
+      "eval_QUANTITY_recall": 0.7391304347826086,
+      "eval_TIME_f1": 0.8148148148148148,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8148148148148148,
+      "eval_TIME_precision": 0.7857142857142857,
       "eval_TIME_recall": 0.8461538461538461,
-      "eval_loss": 0.3637796938419342,
-      "eval_overall_accuracy": 0.9632224168126094,
-      "eval_overall_f1": 0.8356940509915013,
-      "eval_overall_precision": 0.8038147138964578,
-      "eval_overall_recall": 0.8702064896755162,
-      "eval_runtime": 0.9145,
-      "eval_samples_per_second": 204.477,
-      "eval_steps_per_second": 3.28,
+      "eval_loss": 0.4027647078037262,
+      "eval_overall_accuracy": 0.9584688516387291,
+      "eval_overall_f1": 0.847645429362881,
+      "eval_overall_precision": 0.7989556135770235,
+      "eval_overall_recall": 0.9026548672566371,
+      "eval_runtime": 0.3573,
+      "eval_samples_per_second": 523.424,
+      "eval_steps_per_second": 8.397,
       "step": 7844
     },
     {
       "epoch": 75.0,
-      "grad_norm": 0.0007877243915572762,
+      "grad_norm": 0.0005983862793073058,
       "learning_rate": 1.25e-05,
-      "loss": 0.0001,
+      "loss": 0.0,
       "step": 7950
     },
     {
       "epoch": 75.0,
-      "eval_LOCATION_f1": 0.8289473684210527,
+      "eval_LOCATION_f1": 0.816326530612245,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7875,
-      "eval_LOCATION_recall": 0.875,
-      "eval_ORGANIZATION_f1": 0.7919463087248321,
+      "eval_LOCATION_precision": 0.8,
+      "eval_LOCATION_recall": 0.8333333333333334,
+      "eval_ORGANIZATION_f1": 0.825,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7972972972972973,
-      "eval_ORGANIZATION_recall": 0.7866666666666666,
-      "eval_PERSON_f1": 0.9084745762711864,
+      "eval_ORGANIZATION_precision": 0.7764705882352941,
+      "eval_ORGANIZATION_recall": 0.88,
+      "eval_PERSON_f1": 0.9096989966555185,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.881578947368421,
-      "eval_PERSON_recall": 0.9370629370629371,
-      "eval_QUANTITY_f1": 0.7499999999999999,
+      "eval_PERSON_precision": 0.8717948717948718,
+      "eval_PERSON_recall": 0.951048951048951,
+      "eval_QUANTITY_f1": 0.6666666666666666,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.72,
-      "eval_QUANTITY_recall": 0.782608695652174,
+      "eval_QUANTITY_precision": 0.6071428571428571,
+      "eval_QUANTITY_recall": 0.7391304347826086,
       "eval_TIME_f1": 0.830188679245283,
       "eval_TIME_number": 26,
       "eval_TIME_precision": 0.8148148148148148,
       "eval_TIME_recall": 0.8461538461538461,
-      "eval_loss": 0.3778810501098633,
-      "eval_overall_accuracy": 0.9649737302977233,
-      "eval_overall_f1": 0.8493543758967002,
-      "eval_overall_precision": 0.8268156424581006,
-      "eval_overall_recall": 0.8731563421828908,
-      "eval_runtime": 0.9077,
-      "eval_samples_per_second": 206.004,
-      "eval_steps_per_second": 3.305,
+      "eval_loss": 0.3929165005683899,
+      "eval_overall_accuracy": 0.960720540405304,
+      "eval_overall_f1": 0.8478873239436621,
+      "eval_overall_precision": 0.8113207547169812,
+      "eval_overall_recall": 0.887905604719764,
+      "eval_runtime": 0.3576,
+      "eval_samples_per_second": 522.919,
+      "eval_steps_per_second": 8.389,
       "step": 7950
     },
     {
       "epoch": 76.0,
-      "grad_norm": 0.005742947105318308,
+      "grad_norm": 0.00021465642203111202,
       "learning_rate": 1.2e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 8056
     },
     {
       "epoch": 76.0,
-      "eval_LOCATION_f1": 0.8289473684210527,
+      "eval_LOCATION_f1": 0.8181818181818181,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7875,
+      "eval_LOCATION_precision": 0.7682926829268293,
       "eval_LOCATION_recall": 0.875,
-      "eval_ORGANIZATION_f1": 0.7894736842105264,
+      "eval_ORGANIZATION_f1": 0.8271604938271604,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7792207792207793,
-      "eval_ORGANIZATION_recall": 0.8,
-      "eval_PERSON_f1": 0.9054054054054055,
+      "eval_ORGANIZATION_precision": 0.7701149425287356,
+      "eval_ORGANIZATION_recall": 0.8933333333333333,
+      "eval_PERSON_f1": 0.9194630872483223,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8758169934640523,
-      "eval_PERSON_recall": 0.9370629370629371,
-      "eval_QUANTITY_f1": 0.7346938775510203,
+      "eval_PERSON_precision": 0.8838709677419355,
+      "eval_PERSON_recall": 0.958041958041958,
+      "eval_QUANTITY_f1": 0.6666666666666666,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.6923076923076923,
-      "eval_QUANTITY_recall": 0.782608695652174,
-      "eval_TIME_f1": 0.830188679245283,
+      "eval_QUANTITY_precision": 0.6071428571428571,
+      "eval_QUANTITY_recall": 0.7391304347826086,
+      "eval_TIME_f1": 0.8363636363636363,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8148148148148148,
-      "eval_TIME_recall": 0.8461538461538461,
-      "eval_loss": 0.3797122836112976,
-      "eval_overall_accuracy": 0.9644733550162622,
-      "eval_overall_f1": 0.8461538461538463,
-      "eval_overall_precision": 0.8181818181818182,
-      "eval_overall_recall": 0.8761061946902655,
-      "eval_runtime": 0.9011,
-      "eval_samples_per_second": 207.522,
-      "eval_steps_per_second": 3.329,
+      "eval_TIME_precision": 0.7931034482758621,
+      "eval_TIME_recall": 0.8846153846153846,
+      "eval_loss": 0.4091399908065796,
+      "eval_overall_accuracy": 0.960720540405304,
+      "eval_overall_f1": 0.8527777777777777,
+      "eval_overall_precision": 0.8057742782152231,
+      "eval_overall_recall": 0.9056047197640118,
+      "eval_runtime": 0.3609,
+      "eval_samples_per_second": 518.107,
+      "eval_steps_per_second": 8.312,
       "step": 8056
     },
     {
       "epoch": 77.0,
-      "grad_norm": 0.00030839102691970766,
+      "grad_norm": 0.00041957717621698976,
       "learning_rate": 1.1500000000000002e-05,
-      "loss": 0.0004,
+      "loss": 0.0,
       "step": 8162
     },
     {
@@ -2985,425 +2985,425 @@
       "eval_LOCATION_number": 72,
       "eval_LOCATION_precision": 0.7974683544303798,
       "eval_LOCATION_recall": 0.875,
-      "eval_ORGANIZATION_f1": 0.8024691358024691,
+      "eval_ORGANIZATION_f1": 0.8271604938271604,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7471264367816092,
-      "eval_ORGANIZATION_recall": 0.8666666666666667,
-      "eval_PERSON_f1": 0.8963210702341137,
+      "eval_ORGANIZATION_precision": 0.7701149425287356,
+      "eval_ORGANIZATION_recall": 0.8933333333333333,
+      "eval_PERSON_f1": 0.903010033444816,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8589743589743589,
-      "eval_PERSON_recall": 0.9370629370629371,
-      "eval_QUANTITY_f1": 0.76,
+      "eval_PERSON_precision": 0.8653846153846154,
+      "eval_PERSON_recall": 0.9440559440559441,
+      "eval_QUANTITY_f1": 0.6666666666666666,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.7037037037037037,
-      "eval_QUANTITY_recall": 0.8260869565217391,
+      "eval_QUANTITY_precision": 0.6071428571428571,
+      "eval_QUANTITY_recall": 0.7391304347826086,
       "eval_TIME_f1": 0.830188679245283,
       "eval_TIME_number": 26,
       "eval_TIME_precision": 0.8148148148148148,
       "eval_TIME_recall": 0.8461538461538461,
-      "eval_loss": 0.40759775042533875,
-      "eval_overall_accuracy": 0.9614711033274956,
-      "eval_overall_f1": 0.8475524475524475,
-      "eval_overall_precision": 0.8058510638297872,
-      "eval_overall_recall": 0.8938053097345132,
-      "eval_runtime": 0.9103,
-      "eval_samples_per_second": 205.428,
-      "eval_steps_per_second": 3.296,
+      "eval_loss": 0.3996489346027374,
+      "eval_overall_accuracy": 0.9602201651238429,
+      "eval_overall_f1": 0.8491620111731844,
+      "eval_overall_precision": 0.8063660477453581,
+      "eval_overall_recall": 0.8967551622418879,
+      "eval_runtime": 0.3555,
+      "eval_samples_per_second": 526.025,
+      "eval_steps_per_second": 8.439,
       "step": 8162
     },
     {
       "epoch": 78.0,
-      "grad_norm": 0.00025188998552039266,
+      "grad_norm": 0.0018605925142765045,
       "learning_rate": 1.1000000000000001e-05,
-      "loss": 0.0005,
+      "loss": 0.0,
       "step": 8268
     },
     {
       "epoch": 78.0,
-      "eval_LOCATION_f1": 0.8400000000000001,
+      "eval_LOCATION_f1": 0.8266666666666667,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.8076923076923077,
-      "eval_LOCATION_recall": 0.875,
-      "eval_ORGANIZATION_f1": 0.7950310559006211,
+      "eval_LOCATION_precision": 0.7948717948717948,
+      "eval_LOCATION_recall": 0.8611111111111112,
+      "eval_ORGANIZATION_f1": 0.8198757763975156,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7441860465116279,
-      "eval_ORGANIZATION_recall": 0.8533333333333334,
-      "eval_PERSON_f1": 0.8993288590604027,
+      "eval_ORGANIZATION_precision": 0.7674418604651163,
+      "eval_ORGANIZATION_recall": 0.88,
+      "eval_PERSON_f1": 0.9,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.864516129032258,
-      "eval_PERSON_recall": 0.9370629370629371,
-      "eval_QUANTITY_f1": 0.7346938775510203,
+      "eval_PERSON_precision": 0.8598726114649682,
+      "eval_PERSON_recall": 0.9440559440559441,
+      "eval_QUANTITY_f1": 0.6666666666666666,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.6923076923076923,
-      "eval_QUANTITY_recall": 0.782608695652174,
-      "eval_TIME_f1": 0.8679245283018868,
+      "eval_QUANTITY_precision": 0.6071428571428571,
+      "eval_QUANTITY_recall": 0.7391304347826086,
+      "eval_TIME_f1": 0.830188679245283,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8518518518518519,
-      "eval_TIME_recall": 0.8846153846153846,
-      "eval_loss": 0.39274299144744873,
-      "eval_overall_accuracy": 0.9624718538904178,
-      "eval_overall_f1": 0.849507735583685,
-      "eval_overall_precision": 0.8118279569892473,
+      "eval_TIME_precision": 0.8148148148148148,
+      "eval_TIME_recall": 0.8461538461538461,
+      "eval_loss": 0.3973155617713928,
+      "eval_overall_accuracy": 0.9597197898423818,
+      "eval_overall_f1": 0.8447552447552448,
+      "eval_overall_precision": 0.8031914893617021,
       "eval_overall_recall": 0.8908554572271387,
-      "eval_runtime": 0.9103,
-      "eval_samples_per_second": 205.436,
-      "eval_steps_per_second": 3.296,
+      "eval_runtime": 0.3588,
+      "eval_samples_per_second": 521.17,
+      "eval_steps_per_second": 8.361,
       "step": 8268
     },
     {
       "epoch": 79.0,
-      "grad_norm": 0.00023670213704463094,
+      "grad_norm": 0.0009037918644025922,
       "learning_rate": 1.05e-05,
-      "loss": 0.0003,
+      "loss": 0.0,
       "step": 8374
     },
     {
       "epoch": 79.0,
-      "eval_LOCATION_f1": 0.823529411764706,
+      "eval_LOCATION_f1": 0.8211920529801324,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7777777777777778,
-      "eval_LOCATION_recall": 0.875,
-      "eval_ORGANIZATION_f1": 0.7468354430379747,
+      "eval_LOCATION_precision": 0.7848101265822784,
+      "eval_LOCATION_recall": 0.8611111111111112,
+      "eval_ORGANIZATION_f1": 0.825,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7108433734939759,
-      "eval_ORGANIZATION_recall": 0.7866666666666666,
-      "eval_PERSON_f1": 0.9023569023569024,
+      "eval_ORGANIZATION_precision": 0.7764705882352941,
+      "eval_ORGANIZATION_recall": 0.88,
+      "eval_PERSON_f1": 0.903010033444816,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8701298701298701,
-      "eval_PERSON_recall": 0.9370629370629371,
-      "eval_QUANTITY_f1": 0.7450980392156864,
+      "eval_PERSON_precision": 0.8653846153846154,
+      "eval_PERSON_recall": 0.9440559440559441,
+      "eval_QUANTITY_f1": 0.6666666666666666,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.6785714285714286,
-      "eval_QUANTITY_recall": 0.8260869565217391,
-      "eval_TIME_f1": 0.8148148148148148,
+      "eval_QUANTITY_precision": 0.6071428571428571,
+      "eval_QUANTITY_recall": 0.7391304347826086,
+      "eval_TIME_f1": 0.830188679245283,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.7857142857142857,
+      "eval_TIME_precision": 0.8148148148148148,
       "eval_TIME_recall": 0.8461538461538461,
-      "eval_loss": 0.4000164866447449,
-      "eval_overall_accuracy": 0.960720540405304,
-      "eval_overall_f1": 0.8330995792426367,
-      "eval_overall_precision": 0.7941176470588235,
-      "eval_overall_recall": 0.8761061946902655,
-      "eval_runtime": 0.9149,
-      "eval_samples_per_second": 204.395,
-      "eval_steps_per_second": 3.279,
+      "eval_loss": 0.3981226086616516,
+      "eval_overall_accuracy": 0.9602201651238429,
+      "eval_overall_f1": 0.84593837535014,
+      "eval_overall_precision": 0.8053333333333333,
+      "eval_overall_recall": 0.8908554572271387,
+      "eval_runtime": 0.3579,
+      "eval_samples_per_second": 522.564,
+      "eval_steps_per_second": 8.383,
       "step": 8374
     },
     {
       "epoch": 80.0,
-      "grad_norm": 0.000326380948536098,
+      "grad_norm": 0.0010246345773339272,
       "learning_rate": 1e-05,
-      "loss": 0.0001,
+      "loss": 0.0,
       "step": 8480
     },
     {
       "epoch": 80.0,
-      "eval_LOCATION_f1": 0.8266666666666667,
+      "eval_LOCATION_f1": 0.8344370860927153,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7948717948717948,
-      "eval_LOCATION_recall": 0.8611111111111112,
-      "eval_ORGANIZATION_f1": 0.7483870967741936,
+      "eval_LOCATION_precision": 0.7974683544303798,
+      "eval_LOCATION_recall": 0.875,
+      "eval_ORGANIZATION_f1": 0.8198757763975156,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.725,
-      "eval_ORGANIZATION_recall": 0.7733333333333333,
-      "eval_PERSON_f1": 0.9084745762711864,
+      "eval_ORGANIZATION_precision": 0.7674418604651163,
+      "eval_ORGANIZATION_recall": 0.88,
+      "eval_PERSON_f1": 0.903010033444816,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.881578947368421,
-      "eval_PERSON_recall": 0.9370629370629371,
-      "eval_QUANTITY_f1": 0.7499999999999999,
+      "eval_PERSON_precision": 0.8653846153846154,
+      "eval_PERSON_recall": 0.9440559440559441,
+      "eval_QUANTITY_f1": 0.6666666666666666,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.72,
-      "eval_QUANTITY_recall": 0.782608695652174,
-      "eval_TIME_f1": 0.8679245283018868,
+      "eval_QUANTITY_precision": 0.6071428571428571,
+      "eval_QUANTITY_recall": 0.7391304347826086,
+      "eval_TIME_f1": 0.830188679245283,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8518518518518519,
-      "eval_TIME_recall": 0.8846153846153846,
-      "eval_loss": 0.3888261020183563,
-      "eval_overall_accuracy": 0.9629722291718789,
-      "eval_overall_f1": 0.8416547788873039,
-      "eval_overall_precision": 0.8149171270718232,
-      "eval_overall_recall": 0.8702064896755162,
-      "eval_runtime": 0.9033,
-      "eval_samples_per_second": 207.02,
-      "eval_steps_per_second": 3.321,
+      "eval_TIME_precision": 0.8148148148148148,
+      "eval_TIME_recall": 0.8461538461538461,
+      "eval_loss": 0.4011876583099365,
+      "eval_overall_accuracy": 0.9602201651238429,
+      "eval_overall_f1": 0.8475524475524475,
+      "eval_overall_precision": 0.8058510638297872,
+      "eval_overall_recall": 0.8938053097345132,
+      "eval_runtime": 0.38,
+      "eval_samples_per_second": 492.061,
+      "eval_steps_per_second": 7.894,
       "step": 8480
     },
     {
       "epoch": 81.0,
-      "grad_norm": 0.0005148330819793046,
+      "grad_norm": 0.00019219562818761915,
       "learning_rate": 9.5e-06,
-      "loss": 0.0001,
+      "loss": 0.0006,
       "step": 8586
     },
     {
       "epoch": 81.0,
-      "eval_LOCATION_f1": 0.8211920529801324,
+      "eval_LOCATION_f1": 0.8266666666666667,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7848101265822784,
+      "eval_LOCATION_precision": 0.7948717948717948,
       "eval_LOCATION_recall": 0.8611111111111112,
-      "eval_ORGANIZATION_f1": 0.7643312101910827,
+      "eval_ORGANIZATION_f1": 0.8271604938271604,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7317073170731707,
-      "eval_ORGANIZATION_recall": 0.8,
-      "eval_PERSON_f1": 0.8993288590604027,
+      "eval_ORGANIZATION_precision": 0.7701149425287356,
+      "eval_ORGANIZATION_recall": 0.8933333333333333,
+      "eval_PERSON_f1": 0.9,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.864516129032258,
-      "eval_PERSON_recall": 0.9370629370629371,
-      "eval_QUANTITY_f1": 0.72,
+      "eval_PERSON_precision": 0.8598726114649682,
+      "eval_PERSON_recall": 0.9440559440559441,
+      "eval_QUANTITY_f1": 0.6538461538461539,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.6666666666666666,
-      "eval_QUANTITY_recall": 0.782608695652174,
+      "eval_QUANTITY_precision": 0.5862068965517241,
+      "eval_QUANTITY_recall": 0.7391304347826086,
       "eval_TIME_f1": 0.8518518518518519,
       "eval_TIME_number": 26,
       "eval_TIME_precision": 0.8214285714285714,
       "eval_TIME_recall": 0.8846153846153846,
-      "eval_loss": 0.3981935977935791,
-      "eval_overall_accuracy": 0.9614711033274956,
-      "eval_overall_f1": 0.836619718309859,
-      "eval_overall_precision": 0.8005390835579514,
-      "eval_overall_recall": 0.8761061946902655,
-      "eval_runtime": 0.9078,
-      "eval_samples_per_second": 205.981,
-      "eval_steps_per_second": 3.305,
+      "eval_loss": 0.4144229590892792,
+      "eval_overall_accuracy": 0.9597197898423818,
+      "eval_overall_f1": 0.8467966573816156,
+      "eval_overall_precision": 0.8021108179419525,
+      "eval_overall_recall": 0.8967551622418879,
+      "eval_runtime": 0.3556,
+      "eval_samples_per_second": 525.889,
+      "eval_steps_per_second": 8.437,
       "step": 8586
     },
     {
       "epoch": 82.0,
-      "grad_norm": 0.0001784728665370494,
+      "grad_norm": 0.000247513729846105,
       "learning_rate": 9e-06,
-      "loss": 0.0003,
+      "loss": 0.0,
       "step": 8692
     },
     {
       "epoch": 82.0,
-      "eval_LOCATION_f1": 0.8378378378378377,
+      "eval_LOCATION_f1": 0.8211920529801324,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.8157894736842105,
+      "eval_LOCATION_precision": 0.7848101265822784,
       "eval_LOCATION_recall": 0.8611111111111112,
-      "eval_ORGANIZATION_f1": 0.7870967741935484,
+      "eval_ORGANIZATION_f1": 0.8227848101265823,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7625,
-      "eval_ORGANIZATION_recall": 0.8133333333333334,
-      "eval_PERSON_f1": 0.91156462585034,
+      "eval_ORGANIZATION_precision": 0.7831325301204819,
+      "eval_ORGANIZATION_recall": 0.8666666666666667,
+      "eval_PERSON_f1": 0.9,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8874172185430463,
-      "eval_PERSON_recall": 0.9370629370629371,
-      "eval_QUANTITY_f1": 0.7499999999999999,
+      "eval_PERSON_precision": 0.8598726114649682,
+      "eval_PERSON_recall": 0.9440559440559441,
+      "eval_QUANTITY_f1": 0.68,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.72,
-      "eval_QUANTITY_recall": 0.782608695652174,
-      "eval_TIME_f1": 0.8679245283018868,
+      "eval_QUANTITY_precision": 0.6296296296296297,
+      "eval_QUANTITY_recall": 0.7391304347826086,
+      "eval_TIME_f1": 0.830188679245283,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8518518518518519,
-      "eval_TIME_recall": 0.8846153846153846,
-      "eval_loss": 0.3737158179283142,
-      "eval_overall_accuracy": 0.9647235426569928,
-      "eval_overall_f1": 0.8538681948424068,
-      "eval_overall_precision": 0.83008356545961,
-      "eval_overall_recall": 0.8790560471976401,
-      "eval_runtime": 0.9108,
-      "eval_samples_per_second": 205.324,
-      "eval_steps_per_second": 3.294,
+      "eval_TIME_precision": 0.8148148148148148,
+      "eval_TIME_recall": 0.8461538461538461,
+      "eval_loss": 0.40510478615760803,
+      "eval_overall_accuracy": 0.9594696022016512,
+      "eval_overall_f1": 0.8455056179775281,
+      "eval_overall_precision": 0.806970509383378,
+      "eval_overall_recall": 0.887905604719764,
+      "eval_runtime": 0.3705,
+      "eval_samples_per_second": 504.754,
+      "eval_steps_per_second": 8.098,
       "step": 8692
     },
     {
       "epoch": 83.0,
-      "grad_norm": 0.00012157092714915052,
+      "grad_norm": 0.0001593719352968037,
       "learning_rate": 8.500000000000002e-06,
-      "loss": 0.0002,
+      "loss": 0.0006,
       "step": 8798
     },
     {
       "epoch": 83.0,
-      "eval_LOCATION_f1": 0.8322147651006712,
+      "eval_LOCATION_f1": 0.8104575163398693,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.8051948051948052,
+      "eval_LOCATION_precision": 0.7654320987654321,
       "eval_LOCATION_recall": 0.8611111111111112,
-      "eval_ORGANIZATION_f1": 0.7975460122699386,
+      "eval_ORGANIZATION_f1": 0.8048780487804879,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7386363636363636,
-      "eval_ORGANIZATION_recall": 0.8666666666666667,
-      "eval_PERSON_f1": 0.9047619047619047,
+      "eval_ORGANIZATION_precision": 0.7415730337078652,
+      "eval_ORGANIZATION_recall": 0.88,
+      "eval_PERSON_f1": 0.8970099667774087,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8807947019867549,
-      "eval_PERSON_recall": 0.9300699300699301,
-      "eval_QUANTITY_f1": 0.7499999999999999,
+      "eval_PERSON_precision": 0.8544303797468354,
+      "eval_PERSON_recall": 0.9440559440559441,
+      "eval_QUANTITY_f1": 0.6666666666666666,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.72,
-      "eval_QUANTITY_recall": 0.782608695652174,
+      "eval_QUANTITY_precision": 0.6071428571428571,
+      "eval_QUANTITY_recall": 0.7391304347826086,
       "eval_TIME_f1": 0.8518518518518519,
       "eval_TIME_number": 26,
       "eval_TIME_precision": 0.8214285714285714,
       "eval_TIME_recall": 0.8846153846153846,
-      "eval_loss": 0.3761243224143982,
-      "eval_overall_accuracy": 0.9632224168126094,
-      "eval_overall_f1": 0.8502824858757062,
-      "eval_overall_precision": 0.8157181571815718,
-      "eval_overall_recall": 0.887905604719764,
-      "eval_runtime": 0.9121,
-      "eval_samples_per_second": 205.019,
-      "eval_steps_per_second": 3.289,
+      "eval_loss": 0.4076941907405853,
+      "eval_overall_accuracy": 0.9602201651238429,
+      "eval_overall_f1": 0.8381742738589212,
+      "eval_overall_precision": 0.7890625,
+      "eval_overall_recall": 0.8938053097345132,
+      "eval_runtime": 0.3564,
+      "eval_samples_per_second": 524.736,
+      "eval_steps_per_second": 8.418,
       "step": 8798
     },
     {
       "epoch": 84.0,
-      "grad_norm": 0.01326628215610981,
+      "grad_norm": 0.0038378555327653885,
       "learning_rate": 8.000000000000001e-06,
-      "loss": 0.0001,
+      "loss": 0.0004,
       "step": 8904
     },
     {
       "epoch": 84.0,
-      "eval_LOCATION_f1": 0.8322147651006712,
+      "eval_LOCATION_f1": 0.8187919463087249,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.8051948051948052,
-      "eval_LOCATION_recall": 0.8611111111111112,
-      "eval_ORGANIZATION_f1": 0.8025477707006369,
+      "eval_LOCATION_precision": 0.7922077922077922,
+      "eval_LOCATION_recall": 0.8472222222222222,
+      "eval_ORGANIZATION_f1": 0.8280254777070064,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7682926829268293,
-      "eval_ORGANIZATION_recall": 0.84,
-      "eval_PERSON_f1": 0.9047619047619047,
+      "eval_ORGANIZATION_precision": 0.7926829268292683,
+      "eval_ORGANIZATION_recall": 0.8666666666666667,
+      "eval_PERSON_f1": 0.903010033444816,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8807947019867549,
-      "eval_PERSON_recall": 0.9300699300699301,
-      "eval_QUANTITY_f1": 0.7499999999999999,
+      "eval_PERSON_precision": 0.8653846153846154,
+      "eval_PERSON_recall": 0.9440559440559441,
+      "eval_QUANTITY_f1": 0.68,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.72,
-      "eval_QUANTITY_recall": 0.782608695652174,
-      "eval_TIME_f1": 0.8679245283018868,
+      "eval_QUANTITY_precision": 0.6296296296296297,
+      "eval_QUANTITY_recall": 0.7391304347826086,
+      "eval_TIME_f1": 0.830188679245283,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8518518518518519,
-      "eval_TIME_recall": 0.8846153846153846,
-      "eval_loss": 0.37578248977661133,
-      "eval_overall_accuracy": 0.9639729797348011,
-      "eval_overall_f1": 0.8530670470756063,
-      "eval_overall_precision": 0.8259668508287292,
-      "eval_overall_recall": 0.8820058997050148,
-      "eval_runtime": 0.9306,
-      "eval_samples_per_second": 200.941,
-      "eval_steps_per_second": 3.224,
+      "eval_TIME_precision": 0.8148148148148148,
+      "eval_TIME_recall": 0.8461538461538461,
+      "eval_loss": 0.39616209268569946,
+      "eval_overall_accuracy": 0.960720540405304,
+      "eval_overall_f1": 0.8474576271186441,
+      "eval_overall_precision": 0.8130081300813008,
+      "eval_overall_recall": 0.8849557522123894,
+      "eval_runtime": 0.359,
+      "eval_samples_per_second": 520.922,
+      "eval_steps_per_second": 8.357,
       "step": 8904
     },
     {
       "epoch": 85.0,
-      "grad_norm": 0.006039230152964592,
+      "grad_norm": 0.0006832346552982926,
       "learning_rate": 7.5e-06,
-      "loss": 0.0004,
+      "loss": 0.0005,
       "step": 9010
     },
     {
       "epoch": 85.0,
-      "eval_LOCATION_f1": 0.8322147651006712,
+      "eval_LOCATION_f1": 0.8243243243243243,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.8051948051948052,
-      "eval_LOCATION_recall": 0.8611111111111112,
-      "eval_ORGANIZATION_f1": 0.7975460122699386,
+      "eval_LOCATION_precision": 0.8026315789473685,
+      "eval_LOCATION_recall": 0.8472222222222222,
+      "eval_ORGANIZATION_f1": 0.8235294117647058,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7386363636363636,
-      "eval_ORGANIZATION_recall": 0.8666666666666667,
-      "eval_PERSON_f1": 0.9016949152542374,
+      "eval_ORGANIZATION_precision": 0.8076923076923077,
+      "eval_ORGANIZATION_recall": 0.84,
+      "eval_PERSON_f1": 0.9158249158249158,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.875,
-      "eval_PERSON_recall": 0.9300699300699301,
-      "eval_QUANTITY_f1": 0.7346938775510203,
+      "eval_PERSON_precision": 0.8831168831168831,
+      "eval_PERSON_recall": 0.951048951048951,
+      "eval_QUANTITY_f1": 0.68,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.6923076923076923,
-      "eval_QUANTITY_recall": 0.782608695652174,
-      "eval_TIME_f1": 0.8518518518518519,
+      "eval_QUANTITY_precision": 0.6296296296296297,
+      "eval_QUANTITY_recall": 0.7391304347826086,
+      "eval_TIME_f1": 0.8461538461538461,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8214285714285714,
-      "eval_TIME_recall": 0.8846153846153846,
-      "eval_loss": 0.3836413621902466,
-      "eval_overall_accuracy": 0.9629722291718789,
-      "eval_overall_f1": 0.8478873239436621,
-      "eval_overall_precision": 0.8113207547169812,
-      "eval_overall_recall": 0.887905604719764,
-      "eval_runtime": 0.9049,
-      "eval_samples_per_second": 206.647,
-      "eval_steps_per_second": 3.315,
+      "eval_TIME_precision": 0.8461538461538461,
+      "eval_TIME_recall": 0.8461538461538461,
+      "eval_loss": 0.38612908124923706,
+      "eval_overall_accuracy": 0.9617212909682261,
+      "eval_overall_f1": 0.8542857142857142,
+      "eval_overall_precision": 0.8282548476454293,
+      "eval_overall_recall": 0.8820058997050148,
+      "eval_runtime": 0.3569,
+      "eval_samples_per_second": 523.906,
+      "eval_steps_per_second": 8.405,
       "step": 9010
     },
     {
       "epoch": 86.0,
-      "grad_norm": 0.00042766937986016273,
+      "grad_norm": 0.0012918419670313597,
       "learning_rate": 7.000000000000001e-06,
-      "loss": 0.0005,
+      "loss": 0.0,
       "step": 9116
     },
     {
       "epoch": 86.0,
-      "eval_LOCATION_f1": 0.8322147651006712,
+      "eval_LOCATION_f1": 0.8187919463087249,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.8051948051948052,
-      "eval_LOCATION_recall": 0.8611111111111112,
-      "eval_ORGANIZATION_f1": 0.8076923076923077,
+      "eval_LOCATION_precision": 0.7922077922077922,
+      "eval_LOCATION_recall": 0.8472222222222222,
+      "eval_ORGANIZATION_f1": 0.8176100628930819,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7777777777777778,
-      "eval_ORGANIZATION_recall": 0.84,
-      "eval_PERSON_f1": 0.9016949152542374,
+      "eval_ORGANIZATION_precision": 0.7738095238095238,
+      "eval_ORGANIZATION_recall": 0.8666666666666667,
+      "eval_PERSON_f1": 0.9096989966555185,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.875,
-      "eval_PERSON_recall": 0.9300699300699301,
-      "eval_QUANTITY_f1": 0.7346938775510203,
+      "eval_PERSON_precision": 0.8717948717948718,
+      "eval_PERSON_recall": 0.951048951048951,
+      "eval_QUANTITY_f1": 0.693877551020408,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.6923076923076923,
-      "eval_QUANTITY_recall": 0.782608695652174,
-      "eval_TIME_f1": 0.8518518518518519,
+      "eval_QUANTITY_precision": 0.6538461538461539,
+      "eval_QUANTITY_recall": 0.7391304347826086,
+      "eval_TIME_f1": 0.830188679245283,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8214285714285714,
-      "eval_TIME_recall": 0.8846153846153846,
-      "eval_loss": 0.3791405260562897,
-      "eval_overall_accuracy": 0.9642231673755316,
-      "eval_overall_f1": 0.8506401137980086,
-      "eval_overall_precision": 0.8214285714285714,
-      "eval_overall_recall": 0.8820058997050148,
-      "eval_runtime": 0.904,
-      "eval_samples_per_second": 206.855,
-      "eval_steps_per_second": 3.319,
+      "eval_TIME_precision": 0.8148148148148148,
+      "eval_TIME_recall": 0.8461538461538461,
+      "eval_loss": 0.39533165097236633,
+      "eval_overall_accuracy": 0.9609707280460346,
+      "eval_overall_f1": 0.8490832157968972,
+      "eval_overall_precision": 0.8135135135135135,
+      "eval_overall_recall": 0.887905604719764,
+      "eval_runtime": 0.3669,
+      "eval_samples_per_second": 509.639,
+      "eval_steps_per_second": 8.176,
       "step": 9116
     },
     {
       "epoch": 87.0,
-      "grad_norm": 0.0006120882462710142,
+      "grad_norm": 0.00013653105997946113,
       "learning_rate": 6.5000000000000004e-06,
-      "loss": 0.0004,
+      "loss": 0.0003,
       "step": 9222
     },
     {
       "epoch": 87.0,
-      "eval_LOCATION_f1": 0.8266666666666667,
+      "eval_LOCATION_f1": 0.8211920529801324,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7948717948717948,
+      "eval_LOCATION_precision": 0.7848101265822784,
       "eval_LOCATION_recall": 0.8611111111111112,
-      "eval_ORGANIZATION_f1": 0.8025477707006369,
+      "eval_ORGANIZATION_f1": 0.8220858895705522,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7682926829268293,
-      "eval_ORGANIZATION_recall": 0.84,
-      "eval_PERSON_f1": 0.9023569023569024,
+      "eval_ORGANIZATION_precision": 0.7613636363636364,
+      "eval_ORGANIZATION_recall": 0.8933333333333333,
+      "eval_PERSON_f1": 0.9042904290429041,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8701298701298701,
-      "eval_PERSON_recall": 0.9370629370629371,
-      "eval_QUANTITY_f1": 0.7346938775510203,
+      "eval_PERSON_precision": 0.85625,
+      "eval_PERSON_recall": 0.958041958041958,
+      "eval_QUANTITY_f1": 0.6666666666666666,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.6923076923076923,
-      "eval_QUANTITY_recall": 0.782608695652174,
-      "eval_TIME_f1": 0.888888888888889,
+      "eval_QUANTITY_precision": 0.6071428571428571,
+      "eval_QUANTITY_recall": 0.7391304347826086,
+      "eval_TIME_f1": 0.7924528301886792,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8571428571428571,
-      "eval_TIME_recall": 0.9230769230769231,
-      "eval_loss": 0.38632336258888245,
-      "eval_overall_accuracy": 0.9639729797348011,
-      "eval_overall_f1": 0.8514851485148515,
-      "eval_overall_precision": 0.8179347826086957,
-      "eval_overall_recall": 0.887905604719764,
-      "eval_runtime": 0.9113,
-      "eval_samples_per_second": 205.193,
-      "eval_steps_per_second": 3.292,
+      "eval_TIME_precision": 0.7777777777777778,
+      "eval_TIME_recall": 0.8076923076923077,
+      "eval_loss": 0.41302046179771423,
+      "eval_overall_accuracy": 0.9602201651238429,
+      "eval_overall_f1": 0.8432732316227461,
+      "eval_overall_precision": 0.7958115183246073,
+      "eval_overall_recall": 0.8967551622418879,
+      "eval_runtime": 0.3613,
+      "eval_samples_per_second": 517.561,
+      "eval_steps_per_second": 8.303,
       "step": 9222
     },
     {
       "epoch": 88.0,
-      "grad_norm": 0.0006923701730556786,
+      "grad_norm": 0.0001936595799634233,
       "learning_rate": 6e-06,
       "loss": 0.0004,
       "step": 9328
@@ -3414,508 +3414,508 @@
       "eval_LOCATION_number": 72,
       "eval_LOCATION_precision": 0.7848101265822784,
       "eval_LOCATION_recall": 0.8611111111111112,
-      "eval_ORGANIZATION_f1": 0.7875000000000001,
+      "eval_ORGANIZATION_f1": 0.8220858895705522,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7411764705882353,
-      "eval_ORGANIZATION_recall": 0.84,
-      "eval_PERSON_f1": 0.907849829351536,
+      "eval_ORGANIZATION_precision": 0.7613636363636364,
+      "eval_ORGANIZATION_recall": 0.8933333333333333,
+      "eval_PERSON_f1": 0.9006622516556291,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8866666666666667,
-      "eval_PERSON_recall": 0.9300699300699301,
-      "eval_QUANTITY_f1": 0.7499999999999999,
+      "eval_PERSON_precision": 0.8553459119496856,
+      "eval_PERSON_recall": 0.951048951048951,
+      "eval_QUANTITY_f1": 0.6666666666666666,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.72,
-      "eval_QUANTITY_recall": 0.782608695652174,
-      "eval_TIME_f1": 0.9056603773584906,
+      "eval_QUANTITY_precision": 0.6071428571428571,
+      "eval_QUANTITY_recall": 0.7391304347826086,
+      "eval_TIME_f1": 0.7924528301886792,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8888888888888888,
-      "eval_TIME_recall": 0.9230769230769231,
-      "eval_loss": 0.3742952346801758,
-      "eval_overall_accuracy": 0.9647235426569928,
-      "eval_overall_f1": 0.851063829787234,
-      "eval_overall_precision": 0.819672131147541,
-      "eval_overall_recall": 0.8849557522123894,
-      "eval_runtime": 0.9099,
-      "eval_samples_per_second": 205.512,
-      "eval_steps_per_second": 3.297,
+      "eval_TIME_precision": 0.7777777777777778,
+      "eval_TIME_recall": 0.8076923076923077,
+      "eval_loss": 0.41551852226257324,
+      "eval_overall_accuracy": 0.9602201651238429,
+      "eval_overall_f1": 0.8416666666666667,
+      "eval_overall_precision": 0.7952755905511811,
+      "eval_overall_recall": 0.8938053097345132,
+      "eval_runtime": 0.3626,
+      "eval_samples_per_second": 515.716,
+      "eval_steps_per_second": 8.274,
       "step": 9328
     },
     {
       "epoch": 89.0,
-      "grad_norm": 0.0163017176091671,
+      "grad_norm": 0.000983737176284194,
       "learning_rate": 5.500000000000001e-06,
-      "loss": 0.0006,
+      "loss": 0.0,
       "step": 9434
     },
     {
       "epoch": 89.0,
-      "eval_LOCATION_f1": 0.816326530612245,
+      "eval_LOCATION_f1": 0.8211920529801324,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.8,
-      "eval_LOCATION_recall": 0.8333333333333334,
-      "eval_ORGANIZATION_f1": 0.7894736842105264,
+      "eval_LOCATION_precision": 0.7848101265822784,
+      "eval_LOCATION_recall": 0.8611111111111112,
+      "eval_ORGANIZATION_f1": 0.8220858895705522,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7792207792207793,
-      "eval_ORGANIZATION_recall": 0.8,
-      "eval_PERSON_f1": 0.907849829351536,
+      "eval_ORGANIZATION_precision": 0.7613636363636364,
+      "eval_ORGANIZATION_recall": 0.8933333333333333,
+      "eval_PERSON_f1": 0.9006622516556291,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8866666666666667,
-      "eval_PERSON_recall": 0.9300699300699301,
-      "eval_QUANTITY_f1": 0.7499999999999999,
+      "eval_PERSON_precision": 0.8553459119496856,
+      "eval_PERSON_recall": 0.951048951048951,
+      "eval_QUANTITY_f1": 0.6666666666666666,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.72,
-      "eval_QUANTITY_recall": 0.782608695652174,
-      "eval_TIME_f1": 0.9056603773584906,
+      "eval_QUANTITY_precision": 0.6071428571428571,
+      "eval_QUANTITY_recall": 0.7391304347826086,
+      "eval_TIME_f1": 0.7924528301886792,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8888888888888888,
-      "eval_TIME_recall": 0.9230769230769231,
-      "eval_loss": 0.3696248233318329,
-      "eval_overall_accuracy": 0.966224668501376,
-      "eval_overall_f1": 0.8513708513708513,
-      "eval_overall_precision": 0.8333333333333334,
-      "eval_overall_recall": 0.8702064896755162,
-      "eval_runtime": 0.9099,
-      "eval_samples_per_second": 205.507,
-      "eval_steps_per_second": 3.297,
+      "eval_TIME_precision": 0.7777777777777778,
+      "eval_TIME_recall": 0.8076923076923077,
+      "eval_loss": 0.4156357944011688,
+      "eval_overall_accuracy": 0.9602201651238429,
+      "eval_overall_f1": 0.8416666666666667,
+      "eval_overall_precision": 0.7952755905511811,
+      "eval_overall_recall": 0.8938053097345132,
+      "eval_runtime": 0.3581,
+      "eval_samples_per_second": 522.251,
+      "eval_steps_per_second": 8.378,
       "step": 9434
     },
     {
       "epoch": 90.0,
-      "grad_norm": 15.17189884185791,
+      "grad_norm": 0.001763033214956522,
       "learning_rate": 5e-06,
       "loss": 0.0002,
       "step": 9540
     },
     {
       "epoch": 90.0,
-      "eval_LOCATION_f1": 0.8,
+      "eval_LOCATION_f1": 0.8266666666666667,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7692307692307693,
-      "eval_LOCATION_recall": 0.8333333333333334,
-      "eval_ORGANIZATION_f1": 0.7894736842105264,
+      "eval_LOCATION_precision": 0.7948717948717948,
+      "eval_LOCATION_recall": 0.8611111111111112,
+      "eval_ORGANIZATION_f1": 0.8198757763975156,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7792207792207793,
-      "eval_ORGANIZATION_recall": 0.8,
-      "eval_PERSON_f1": 0.907849829351536,
+      "eval_ORGANIZATION_precision": 0.7674418604651163,
+      "eval_ORGANIZATION_recall": 0.88,
+      "eval_PERSON_f1": 0.9090909090909092,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8866666666666667,
-      "eval_PERSON_recall": 0.9300699300699301,
-      "eval_QUANTITY_f1": 0.7499999999999999,
+      "eval_PERSON_precision": 0.8766233766233766,
+      "eval_PERSON_recall": 0.9440559440559441,
+      "eval_QUANTITY_f1": 0.68,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.72,
-      "eval_QUANTITY_recall": 0.782608695652174,
-      "eval_TIME_f1": 0.9056603773584906,
+      "eval_QUANTITY_precision": 0.6296296296296297,
+      "eval_QUANTITY_recall": 0.7391304347826086,
+      "eval_TIME_f1": 0.830188679245283,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8888888888888888,
-      "eval_TIME_recall": 0.9230769230769231,
-      "eval_loss": 0.3719424605369568,
-      "eval_overall_accuracy": 0.966224668501376,
-      "eval_overall_f1": 0.8477011494252874,
-      "eval_overall_precision": 0.8263305322128851,
-      "eval_overall_recall": 0.8702064896755162,
-      "eval_runtime": 0.9059,
-      "eval_samples_per_second": 206.432,
-      "eval_steps_per_second": 3.312,
+      "eval_TIME_precision": 0.8148148148148148,
+      "eval_TIME_recall": 0.8461538461538461,
+      "eval_loss": 0.404066801071167,
+      "eval_overall_accuracy": 0.9614711033274956,
+      "eval_overall_f1": 0.849507735583685,
+      "eval_overall_precision": 0.8118279569892473,
+      "eval_overall_recall": 0.8908554572271387,
+      "eval_runtime": 0.3736,
+      "eval_samples_per_second": 500.522,
+      "eval_steps_per_second": 8.03,
       "step": 9540
     },
     {
       "epoch": 91.0,
-      "grad_norm": 0.00020513041818048805,
+      "grad_norm": 0.0002766927646007389,
       "learning_rate": 4.5e-06,
       "loss": 0.0002,
       "step": 9646
     },
     {
       "epoch": 91.0,
-      "eval_LOCATION_f1": 0.8187919463087249,
+      "eval_LOCATION_f1": 0.8322147651006712,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7922077922077922,
-      "eval_LOCATION_recall": 0.8472222222222222,
-      "eval_ORGANIZATION_f1": 0.8205128205128205,
+      "eval_LOCATION_precision": 0.8051948051948052,
+      "eval_LOCATION_recall": 0.8611111111111112,
+      "eval_ORGANIZATION_f1": 0.8322981366459627,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7901234567901234,
-      "eval_ORGANIZATION_recall": 0.8533333333333334,
-      "eval_PERSON_f1": 0.9084745762711864,
+      "eval_ORGANIZATION_precision": 0.7790697674418605,
+      "eval_ORGANIZATION_recall": 0.8933333333333333,
+      "eval_PERSON_f1": 0.9158249158249158,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.881578947368421,
-      "eval_PERSON_recall": 0.9370629370629371,
-      "eval_QUANTITY_f1": 0.7499999999999999,
+      "eval_PERSON_precision": 0.8831168831168831,
+      "eval_PERSON_recall": 0.951048951048951,
+      "eval_QUANTITY_f1": 0.6666666666666666,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.72,
-      "eval_QUANTITY_recall": 0.782608695652174,
-      "eval_TIME_f1": 0.9056603773584906,
+      "eval_QUANTITY_precision": 0.6071428571428571,
+      "eval_QUANTITY_recall": 0.7391304347826086,
+      "eval_TIME_f1": 0.830188679245283,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8888888888888888,
-      "eval_TIME_recall": 0.9230769230769231,
-      "eval_loss": 0.3863327205181122,
-      "eval_overall_accuracy": 0.9642231673755316,
-      "eval_overall_f1": 0.8587731811697574,
-      "eval_overall_precision": 0.8314917127071824,
-      "eval_overall_recall": 0.887905604719764,
-      "eval_runtime": 0.9044,
-      "eval_samples_per_second": 206.761,
-      "eval_steps_per_second": 3.317,
+      "eval_TIME_precision": 0.8148148148148148,
+      "eval_TIME_recall": 0.8461538461538461,
+      "eval_loss": 0.3980797231197357,
+      "eval_overall_accuracy": 0.9617212909682261,
+      "eval_overall_f1": 0.8551336146272857,
+      "eval_overall_precision": 0.8172043010752689,
+      "eval_overall_recall": 0.8967551622418879,
+      "eval_runtime": 0.3573,
+      "eval_samples_per_second": 523.342,
+      "eval_steps_per_second": 8.396,
       "step": 9646
     },
     {
       "epoch": 92.0,
-      "grad_norm": 0.0004068514099344611,
+      "grad_norm": 0.0008579469285905361,
       "learning_rate": 4.000000000000001e-06,
-      "loss": 0.0004,
+      "loss": 0.0,
       "step": 9752
     },
     {
       "epoch": 92.0,
-      "eval_LOCATION_f1": 0.8187919463087249,
+      "eval_LOCATION_f1": 0.8322147651006712,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7922077922077922,
-      "eval_LOCATION_recall": 0.8472222222222222,
-      "eval_ORGANIZATION_f1": 0.8129032258064516,
+      "eval_LOCATION_precision": 0.8051948051948052,
+      "eval_LOCATION_recall": 0.8611111111111112,
+      "eval_ORGANIZATION_f1": 0.8322981366459627,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7875,
-      "eval_ORGANIZATION_recall": 0.84,
-      "eval_PERSON_f1": 0.91156462585034,
+      "eval_ORGANIZATION_precision": 0.7790697674418605,
+      "eval_ORGANIZATION_recall": 0.8933333333333333,
+      "eval_PERSON_f1": 0.9158249158249158,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.8874172185430463,
-      "eval_PERSON_recall": 0.9370629370629371,
-      "eval_QUANTITY_f1": 0.7499999999999999,
+      "eval_PERSON_precision": 0.8831168831168831,
+      "eval_PERSON_recall": 0.951048951048951,
+      "eval_QUANTITY_f1": 0.693877551020408,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.72,
-      "eval_QUANTITY_recall": 0.782608695652174,
-      "eval_TIME_f1": 0.9056603773584906,
+      "eval_QUANTITY_precision": 0.6538461538461539,
+      "eval_QUANTITY_recall": 0.7391304347826086,
+      "eval_TIME_f1": 0.830188679245283,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8888888888888888,
-      "eval_TIME_recall": 0.9230769230769231,
-      "eval_loss": 0.38287997245788574,
-      "eval_overall_accuracy": 0.9644733550162622,
-      "eval_overall_f1": 0.8583690987124464,
-      "eval_overall_precision": 0.8333333333333334,
-      "eval_overall_recall": 0.8849557522123894,
-      "eval_runtime": 0.9043,
-      "eval_samples_per_second": 206.779,
-      "eval_steps_per_second": 3.317,
+      "eval_TIME_precision": 0.8148148148148148,
+      "eval_TIME_recall": 0.8461538461538461,
+      "eval_loss": 0.3970873951911926,
+      "eval_overall_accuracy": 0.9617212909682261,
+      "eval_overall_f1": 0.8575458392101551,
+      "eval_overall_precision": 0.8216216216216217,
+      "eval_overall_recall": 0.8967551622418879,
+      "eval_runtime": 0.3615,
+      "eval_samples_per_second": 517.255,
+      "eval_steps_per_second": 8.298,
       "step": 9752
     },
     {
       "epoch": 93.0,
-      "grad_norm": 0.00010082097287522629,
+      "grad_norm": 0.00014306257071439177,
       "learning_rate": 3.5000000000000004e-06,
-      "loss": 0.0002,
+      "loss": 0.0,
       "step": 9858
     },
     {
       "epoch": 93.0,
-      "eval_LOCATION_f1": 0.8187919463087249,
+      "eval_LOCATION_f1": 0.8266666666666667,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7922077922077922,
-      "eval_LOCATION_recall": 0.8472222222222222,
-      "eval_ORGANIZATION_f1": 0.8104575163398693,
+      "eval_LOCATION_precision": 0.7948717948717948,
+      "eval_LOCATION_recall": 0.8611111111111112,
+      "eval_ORGANIZATION_f1": 0.8322981366459627,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7948717948717948,
-      "eval_ORGANIZATION_recall": 0.8266666666666667,
-      "eval_PERSON_f1": 0.9084745762711864,
+      "eval_ORGANIZATION_precision": 0.7790697674418605,
+      "eval_ORGANIZATION_recall": 0.8933333333333333,
+      "eval_PERSON_f1": 0.9194630872483223,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.881578947368421,
-      "eval_PERSON_recall": 0.9370629370629371,
-      "eval_QUANTITY_f1": 0.7499999999999999,
+      "eval_PERSON_precision": 0.8838709677419355,
+      "eval_PERSON_recall": 0.958041958041958,
+      "eval_QUANTITY_f1": 0.6666666666666666,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.72,
-      "eval_QUANTITY_recall": 0.782608695652174,
-      "eval_TIME_f1": 0.9056603773584906,
+      "eval_QUANTITY_precision": 0.6071428571428571,
+      "eval_QUANTITY_recall": 0.7391304347826086,
+      "eval_TIME_f1": 0.8363636363636363,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8888888888888888,
-      "eval_TIME_recall": 0.9230769230769231,
-      "eval_loss": 0.37826383113861084,
-      "eval_overall_accuracy": 0.9652239179384539,
-      "eval_overall_f1": 0.8567335243553009,
-      "eval_overall_precision": 0.8328690807799443,
-      "eval_overall_recall": 0.8820058997050148,
-      "eval_runtime": 0.918,
-      "eval_samples_per_second": 203.7,
-      "eval_steps_per_second": 3.268,
+      "eval_TIME_precision": 0.7931034482758621,
+      "eval_TIME_recall": 0.8846153846153846,
+      "eval_loss": 0.4024323523044586,
+      "eval_overall_accuracy": 0.9614711033274956,
+      "eval_overall_f1": 0.8559440559440559,
+      "eval_overall_precision": 0.8138297872340425,
+      "eval_overall_recall": 0.9026548672566371,
+      "eval_runtime": 0.3607,
+      "eval_samples_per_second": 518.421,
+      "eval_steps_per_second": 8.317,
       "step": 9858
     },
     {
       "epoch": 94.0,
-      "grad_norm": 0.000542107445653528,
+      "grad_norm": 0.00042846077121794224,
       "learning_rate": 3e-06,
-      "loss": 0.0001,
+      "loss": 0.0,
       "step": 9964
     },
     {
       "epoch": 94.0,
-      "eval_LOCATION_f1": 0.8266666666666667,
+      "eval_LOCATION_f1": 0.8322147651006712,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7948717948717948,
+      "eval_LOCATION_precision": 0.8051948051948052,
       "eval_LOCATION_recall": 0.8611111111111112,
-      "eval_ORGANIZATION_f1": 0.8076923076923077,
+      "eval_ORGANIZATION_f1": 0.8322981366459627,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7777777777777778,
-      "eval_ORGANIZATION_recall": 0.84,
-      "eval_PERSON_f1": 0.9084745762711864,
+      "eval_ORGANIZATION_precision": 0.7790697674418605,
+      "eval_ORGANIZATION_recall": 0.8933333333333333,
+      "eval_PERSON_f1": 0.9194630872483223,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.881578947368421,
-      "eval_PERSON_recall": 0.9370629370629371,
-      "eval_QUANTITY_f1": 0.7499999999999999,
+      "eval_PERSON_precision": 0.8838709677419355,
+      "eval_PERSON_recall": 0.958041958041958,
+      "eval_QUANTITY_f1": 0.68,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.72,
-      "eval_QUANTITY_recall": 0.782608695652174,
-      "eval_TIME_f1": 0.9056603773584906,
+      "eval_QUANTITY_precision": 0.6296296296296297,
+      "eval_QUANTITY_recall": 0.7391304347826086,
+      "eval_TIME_f1": 0.8363636363636363,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8888888888888888,
-      "eval_TIME_recall": 0.9230769230769231,
-      "eval_loss": 0.38576415181159973,
-      "eval_overall_accuracy": 0.9642231673755316,
-      "eval_overall_f1": 0.8575498575498576,
-      "eval_overall_precision": 0.8292011019283747,
-      "eval_overall_recall": 0.887905604719764,
-      "eval_runtime": 0.9111,
-      "eval_samples_per_second": 205.24,
-      "eval_steps_per_second": 3.293,
+      "eval_TIME_precision": 0.7931034482758621,
+      "eval_TIME_recall": 0.8846153846153846,
+      "eval_loss": 0.3985014855861664,
+      "eval_overall_accuracy": 0.9619714786089567,
+      "eval_overall_f1": 0.8583450210378682,
+      "eval_overall_precision": 0.8181818181818182,
+      "eval_overall_recall": 0.9026548672566371,
+      "eval_runtime": 0.3651,
+      "eval_samples_per_second": 512.194,
+      "eval_steps_per_second": 8.217,
       "step": 9964
     },
     {
       "epoch": 95.0,
-      "grad_norm": 0.044874146580696106,
+      "grad_norm": 0.000827198673505336,
       "learning_rate": 2.5e-06,
-      "loss": 0.0001,
+      "loss": 0.0,
       "step": 10070
     },
     {
       "epoch": 95.0,
-      "eval_LOCATION_f1": 0.8266666666666667,
+      "eval_LOCATION_f1": 0.8322147651006712,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7948717948717948,
+      "eval_LOCATION_precision": 0.8051948051948052,
       "eval_LOCATION_recall": 0.8611111111111112,
-      "eval_ORGANIZATION_f1": 0.8076923076923077,
+      "eval_ORGANIZATION_f1": 0.8322981366459627,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7777777777777778,
-      "eval_ORGANIZATION_recall": 0.84,
-      "eval_PERSON_f1": 0.9084745762711864,
+      "eval_ORGANIZATION_precision": 0.7790697674418605,
+      "eval_ORGANIZATION_recall": 0.8933333333333333,
+      "eval_PERSON_f1": 0.9194630872483223,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.881578947368421,
-      "eval_PERSON_recall": 0.9370629370629371,
-      "eval_QUANTITY_f1": 0.7499999999999999,
+      "eval_PERSON_precision": 0.8838709677419355,
+      "eval_PERSON_recall": 0.958041958041958,
+      "eval_QUANTITY_f1": 0.6666666666666666,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.72,
-      "eval_QUANTITY_recall": 0.782608695652174,
-      "eval_TIME_f1": 0.888888888888889,
+      "eval_QUANTITY_precision": 0.6071428571428571,
+      "eval_QUANTITY_recall": 0.7391304347826086,
+      "eval_TIME_f1": 0.8363636363636363,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8571428571428571,
-      "eval_TIME_recall": 0.9230769230769231,
-      "eval_loss": 0.38630810379981995,
-      "eval_overall_accuracy": 0.9639729797348011,
-      "eval_overall_f1": 0.856330014224751,
-      "eval_overall_precision": 0.8269230769230769,
-      "eval_overall_recall": 0.887905604719764,
-      "eval_runtime": 0.8986,
-      "eval_samples_per_second": 208.113,
-      "eval_steps_per_second": 3.339,
+      "eval_TIME_precision": 0.7931034482758621,
+      "eval_TIME_recall": 0.8846153846153846,
+      "eval_loss": 0.39939987659454346,
+      "eval_overall_accuracy": 0.9617212909682261,
+      "eval_overall_f1": 0.8571428571428571,
+      "eval_overall_precision": 0.816,
+      "eval_overall_recall": 0.9026548672566371,
+      "eval_runtime": 0.358,
+      "eval_samples_per_second": 522.401,
+      "eval_steps_per_second": 8.381,
       "step": 10070
     },
     {
       "epoch": 96.0,
-      "grad_norm": 0.0001308424398303032,
+      "grad_norm": 0.0002911574556492269,
       "learning_rate": 2.0000000000000003e-06,
-      "loss": 0.0001,
+      "loss": 0.0003,
       "step": 10176
     },
     {
       "epoch": 96.0,
-      "eval_LOCATION_f1": 0.8266666666666667,
+      "eval_LOCATION_f1": 0.8243243243243243,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7948717948717948,
-      "eval_LOCATION_recall": 0.8611111111111112,
-      "eval_ORGANIZATION_f1": 0.8076923076923077,
+      "eval_LOCATION_precision": 0.8026315789473685,
+      "eval_LOCATION_recall": 0.8472222222222222,
+      "eval_ORGANIZATION_f1": 0.8322981366459627,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7777777777777778,
-      "eval_ORGANIZATION_recall": 0.84,
-      "eval_PERSON_f1": 0.9084745762711864,
+      "eval_ORGANIZATION_precision": 0.7790697674418605,
+      "eval_ORGANIZATION_recall": 0.8933333333333333,
+      "eval_PERSON_f1": 0.912751677852349,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.881578947368421,
-      "eval_PERSON_recall": 0.9370629370629371,
-      "eval_QUANTITY_f1": 0.76,
+      "eval_PERSON_precision": 0.8774193548387097,
+      "eval_PERSON_recall": 0.951048951048951,
+      "eval_QUANTITY_f1": 0.693877551020408,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.7037037037037037,
-      "eval_QUANTITY_recall": 0.8260869565217391,
-      "eval_TIME_f1": 0.8518518518518519,
+      "eval_QUANTITY_precision": 0.6538461538461539,
+      "eval_QUANTITY_recall": 0.7391304347826086,
+      "eval_TIME_f1": 0.8363636363636363,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8214285714285714,
+      "eval_TIME_precision": 0.7931034482758621,
       "eval_TIME_recall": 0.8846153846153846,
-      "eval_loss": 0.3916187286376953,
-      "eval_overall_accuracy": 0.9637227920940705,
-      "eval_overall_f1": 0.8539007092198581,
-      "eval_overall_precision": 0.8224043715846995,
-      "eval_overall_recall": 0.887905604719764,
-      "eval_runtime": 0.9154,
-      "eval_samples_per_second": 204.278,
-      "eval_steps_per_second": 3.277,
+      "eval_loss": 0.39506107568740845,
+      "eval_overall_accuracy": 0.961220915686765,
+      "eval_overall_f1": 0.8551336146272857,
+      "eval_overall_precision": 0.8172043010752689,
+      "eval_overall_recall": 0.8967551622418879,
+      "eval_runtime": 0.3613,
+      "eval_samples_per_second": 517.554,
+      "eval_steps_per_second": 8.303,
       "step": 10176
     },
     {
       "epoch": 97.0,
-      "grad_norm": 0.00018677054322324693,
+      "grad_norm": 0.0003046900383196771,
       "learning_rate": 1.5e-06,
-      "loss": 0.0001,
+      "loss": 0.0005,
       "step": 10282
     },
     {
       "epoch": 97.0,
-      "eval_LOCATION_f1": 0.8266666666666667,
+      "eval_LOCATION_f1": 0.8243243243243243,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7948717948717948,
-      "eval_LOCATION_recall": 0.8611111111111112,
-      "eval_ORGANIZATION_f1": 0.8101265822784811,
+      "eval_LOCATION_precision": 0.8026315789473685,
+      "eval_LOCATION_recall": 0.8472222222222222,
+      "eval_ORGANIZATION_f1": 0.8322981366459627,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7710843373493976,
-      "eval_ORGANIZATION_recall": 0.8533333333333334,
-      "eval_PERSON_f1": 0.9084745762711864,
+      "eval_ORGANIZATION_precision": 0.7790697674418605,
+      "eval_ORGANIZATION_recall": 0.8933333333333333,
+      "eval_PERSON_f1": 0.9133333333333333,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.881578947368421,
-      "eval_PERSON_recall": 0.9370629370629371,
-      "eval_QUANTITY_f1": 0.76,
+      "eval_PERSON_precision": 0.8726114649681529,
+      "eval_PERSON_recall": 0.958041958041958,
+      "eval_QUANTITY_f1": 0.68,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.7037037037037037,
-      "eval_QUANTITY_recall": 0.8260869565217391,
-      "eval_TIME_f1": 0.8518518518518519,
+      "eval_QUANTITY_precision": 0.6296296296296297,
+      "eval_QUANTITY_recall": 0.7391304347826086,
+      "eval_TIME_f1": 0.8363636363636363,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8214285714285714,
+      "eval_TIME_precision": 0.7931034482758621,
       "eval_TIME_recall": 0.8846153846153846,
-      "eval_loss": 0.3897801339626312,
-      "eval_overall_accuracy": 0.9642231673755316,
-      "eval_overall_f1": 0.8543140028288544,
-      "eval_overall_precision": 0.8206521739130435,
-      "eval_overall_recall": 0.8908554572271387,
-      "eval_runtime": 0.9174,
-      "eval_samples_per_second": 203.839,
-      "eval_steps_per_second": 3.27,
+      "eval_loss": 0.4083447754383087,
+      "eval_overall_accuracy": 0.9609707280460346,
+      "eval_overall_f1": 0.854341736694678,
+      "eval_overall_precision": 0.8133333333333334,
+      "eval_overall_recall": 0.8997050147492626,
+      "eval_runtime": 0.3599,
+      "eval_samples_per_second": 519.582,
+      "eval_steps_per_second": 8.336,
       "step": 10282
     },
     {
       "epoch": 98.0,
-      "grad_norm": 0.004196417052298784,
+      "grad_norm": 0.00015339584206230938,
       "learning_rate": 1.0000000000000002e-06,
-      "loss": 0.0001,
+      "loss": 0.0,
       "step": 10388
     },
     {
       "epoch": 98.0,
-      "eval_LOCATION_f1": 0.8266666666666667,
+      "eval_LOCATION_f1": 0.8243243243243243,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7948717948717948,
-      "eval_LOCATION_recall": 0.8611111111111112,
-      "eval_ORGANIZATION_f1": 0.8101265822784811,
+      "eval_LOCATION_precision": 0.8026315789473685,
+      "eval_LOCATION_recall": 0.8472222222222222,
+      "eval_ORGANIZATION_f1": 0.8322981366459627,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7710843373493976,
-      "eval_ORGANIZATION_recall": 0.8533333333333334,
-      "eval_PERSON_f1": 0.9084745762711864,
+      "eval_ORGANIZATION_precision": 0.7790697674418605,
+      "eval_ORGANIZATION_recall": 0.8933333333333333,
+      "eval_PERSON_f1": 0.9194630872483223,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.881578947368421,
-      "eval_PERSON_recall": 0.9370629370629371,
-      "eval_QUANTITY_f1": 0.76,
+      "eval_PERSON_precision": 0.8838709677419355,
+      "eval_PERSON_recall": 0.958041958041958,
+      "eval_QUANTITY_f1": 0.68,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.7037037037037037,
-      "eval_QUANTITY_recall": 0.8260869565217391,
-      "eval_TIME_f1": 0.8518518518518519,
+      "eval_QUANTITY_precision": 0.6296296296296297,
+      "eval_QUANTITY_recall": 0.7391304347826086,
+      "eval_TIME_f1": 0.8363636363636363,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8214285714285714,
+      "eval_TIME_precision": 0.7931034482758621,
       "eval_TIME_recall": 0.8846153846153846,
-      "eval_loss": 0.38807475566864014,
-      "eval_overall_accuracy": 0.9644733550162622,
-      "eval_overall_f1": 0.8543140028288544,
-      "eval_overall_precision": 0.8206521739130435,
-      "eval_overall_recall": 0.8908554572271387,
-      "eval_runtime": 0.9056,
-      "eval_samples_per_second": 206.492,
-      "eval_steps_per_second": 3.313,
+      "eval_loss": 0.407378226518631,
+      "eval_overall_accuracy": 0.9614711033274956,
+      "eval_overall_f1": 0.8567415730337079,
+      "eval_overall_precision": 0.8176943699731903,
+      "eval_overall_recall": 0.8997050147492626,
+      "eval_runtime": 0.3669,
+      "eval_samples_per_second": 509.703,
+      "eval_steps_per_second": 8.177,
       "step": 10388
     },
     {
       "epoch": 99.0,
-      "grad_norm": 0.0001258315023733303,
+      "grad_norm": 0.0002479134127497673,
       "learning_rate": 5.000000000000001e-07,
-      "loss": 0.0001,
+      "loss": 0.0,
       "step": 10494
     },
     {
       "epoch": 99.0,
-      "eval_LOCATION_f1": 0.8266666666666667,
+      "eval_LOCATION_f1": 0.8243243243243243,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7948717948717948,
-      "eval_LOCATION_recall": 0.8611111111111112,
-      "eval_ORGANIZATION_f1": 0.8101265822784811,
+      "eval_LOCATION_precision": 0.8026315789473685,
+      "eval_LOCATION_recall": 0.8472222222222222,
+      "eval_ORGANIZATION_f1": 0.8322981366459627,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7710843373493976,
-      "eval_ORGANIZATION_recall": 0.8533333333333334,
-      "eval_PERSON_f1": 0.9084745762711864,
+      "eval_ORGANIZATION_precision": 0.7790697674418605,
+      "eval_ORGANIZATION_recall": 0.8933333333333333,
+      "eval_PERSON_f1": 0.9194630872483223,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.881578947368421,
-      "eval_PERSON_recall": 0.9370629370629371,
-      "eval_QUANTITY_f1": 0.72,
+      "eval_PERSON_precision": 0.8838709677419355,
+      "eval_PERSON_recall": 0.958041958041958,
+      "eval_QUANTITY_f1": 0.68,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.6666666666666666,
-      "eval_QUANTITY_recall": 0.782608695652174,
-      "eval_TIME_f1": 0.8518518518518519,
+      "eval_QUANTITY_precision": 0.6296296296296297,
+      "eval_QUANTITY_recall": 0.7391304347826086,
+      "eval_TIME_f1": 0.8363636363636363,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8214285714285714,
+      "eval_TIME_precision": 0.7931034482758621,
       "eval_TIME_recall": 0.8846153846153846,
-      "eval_loss": 0.3856855630874634,
-      "eval_overall_accuracy": 0.9642231673755316,
-      "eval_overall_f1": 0.8514851485148515,
-      "eval_overall_precision": 0.8179347826086957,
-      "eval_overall_recall": 0.887905604719764,
-      "eval_runtime": 0.9072,
-      "eval_samples_per_second": 206.127,
-      "eval_steps_per_second": 3.307,
+      "eval_loss": 0.4076772630214691,
+      "eval_overall_accuracy": 0.9614711033274956,
+      "eval_overall_f1": 0.8567415730337079,
+      "eval_overall_precision": 0.8176943699731903,
+      "eval_overall_recall": 0.8997050147492626,
+      "eval_runtime": 0.3589,
+      "eval_samples_per_second": 521.056,
+      "eval_steps_per_second": 8.359,
       "step": 10494
     },
     {
       "epoch": 100.0,
-      "grad_norm": 0.00010668273171177134,
+      "grad_norm": 0.00025577095220796764,
       "learning_rate": 0.0,
       "loss": 0.0001,
       "step": 10600
     },
     {
       "epoch": 100.0,
-      "eval_LOCATION_f1": 0.8266666666666667,
+      "eval_LOCATION_f1": 0.8243243243243243,
       "eval_LOCATION_number": 72,
-      "eval_LOCATION_precision": 0.7948717948717948,
-      "eval_LOCATION_recall": 0.8611111111111112,
-      "eval_ORGANIZATION_f1": 0.8101265822784811,
+      "eval_LOCATION_precision": 0.8026315789473685,
+      "eval_LOCATION_recall": 0.8472222222222222,
+      "eval_ORGANIZATION_f1": 0.8322981366459627,
       "eval_ORGANIZATION_number": 75,
-      "eval_ORGANIZATION_precision": 0.7710843373493976,
-      "eval_ORGANIZATION_recall": 0.8533333333333334,
-      "eval_PERSON_f1": 0.9084745762711864,
+      "eval_ORGANIZATION_precision": 0.7790697674418605,
+      "eval_ORGANIZATION_recall": 0.8933333333333333,
+      "eval_PERSON_f1": 0.9194630872483223,
       "eval_PERSON_number": 143,
-      "eval_PERSON_precision": 0.881578947368421,
-      "eval_PERSON_recall": 0.9370629370629371,
-      "eval_QUANTITY_f1": 0.72,
+      "eval_PERSON_precision": 0.8838709677419355,
+      "eval_PERSON_recall": 0.958041958041958,
+      "eval_QUANTITY_f1": 0.68,
       "eval_QUANTITY_number": 23,
-      "eval_QUANTITY_precision": 0.6666666666666666,
-      "eval_QUANTITY_recall": 0.782608695652174,
-      "eval_TIME_f1": 0.8518518518518519,
+      "eval_QUANTITY_precision": 0.6296296296296297,
+      "eval_QUANTITY_recall": 0.7391304347826086,
+      "eval_TIME_f1": 0.8363636363636363,
       "eval_TIME_number": 26,
-      "eval_TIME_precision": 0.8214285714285714,
+      "eval_TIME_precision": 0.7931034482758621,
       "eval_TIME_recall": 0.8846153846153846,
-      "eval_loss": 0.3857269287109375,
-      "eval_overall_accuracy": 0.9642231673755316,
-      "eval_overall_f1": 0.8514851485148515,
-      "eval_overall_precision": 0.8179347826086957,
-      "eval_overall_recall": 0.887905604719764,
-      "eval_runtime": 0.9906,
-      "eval_samples_per_second": 188.783,
-      "eval_steps_per_second": 3.029,
+      "eval_loss": 0.40795865654945374,
+      "eval_overall_accuracy": 0.9614711033274956,
+      "eval_overall_f1": 0.8567415730337079,
+      "eval_overall_precision": 0.8176943699731903,
+      "eval_overall_recall": 0.8997050147492626,
+      "eval_runtime": 0.3623,
+      "eval_samples_per_second": 516.187,
+      "eval_steps_per_second": 8.281,
       "step": 10600
     },
     {
       "epoch": 100.0,
       "step": 10600,
       "total_flos": 4502314993613766.0,
-      "train_loss": 0.008494841067761815,
-      "train_runtime": 2623.6324,
-      "train_samples_per_second": 64.3,
-      "train_steps_per_second": 4.04
+      "train_loss": 0.008934246277599155,
+      "train_runtime": 970.249,
+      "train_samples_per_second": 173.873,
+      "train_steps_per_second": 10.925
     }
   ],
   "logging_steps": 500,