diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -10,399 +10,399 @@
   "log_history": [
     {
       "epoch": 1.0,
-      "grad_norm": 3.8788857460021973,
+      "grad_norm": 3.0579540729522705,
       "learning_rate": 4.9500000000000004e-05,
-      "loss": 0.3599,
+      "loss": 0.3605,
       "step": 106
     },
     {
       "epoch": 1.0,
-      "eval_LOCATION_f1": 0.8356164383561644,
+      "eval_LOCATION_f1": 0.8333333333333333,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.7922077922077922,
-      "eval_LOCATION_recall": 0.8840579710144928,
-      "eval_ORGANIZATION_f1": 0.512396694214876,
+      "eval_LOCATION_precision": 0.8,
+      "eval_LOCATION_recall": 0.8695652173913043,
+      "eval_ORGANIZATION_f1": 0.696969696969697,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.49206349206349204,
-      "eval_ORGANIZATION_recall": 0.5344827586206896,
-      "eval_PERSON_f1": 0.8625,
+      "eval_ORGANIZATION_precision": 0.6216216216216216,
+      "eval_ORGANIZATION_recall": 0.7931034482758621,
+      "eval_PERSON_f1": 0.8670886075949367,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8214285714285714,
-      "eval_PERSON_recall": 0.9078947368421053,
-      "eval_QUANTITY_f1": 0.6666666666666666,
+      "eval_PERSON_precision": 0.8353658536585366,
+      "eval_PERSON_recall": 0.9013157894736842,
+      "eval_QUANTITY_f1": 0.71875,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.6363636363636364,
-      "eval_QUANTITY_recall": 0.7,
-      "eval_TIME_f1": 0.8524590163934426,
+      "eval_QUANTITY_precision": 0.6764705882352942,
+      "eval_QUANTITY_recall": 0.7666666666666667,
+      "eval_TIME_f1": 0.8852459016393444,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.8125,
-      "eval_TIME_recall": 0.896551724137931,
-      "eval_loss": 0.1317242681980133,
-      "eval_overall_accuracy": 0.9569782330345711,
-      "eval_overall_f1": 0.7791842475386779,
-      "eval_overall_precision": 0.7426273458445041,
-      "eval_overall_recall": 0.8195266272189349,
-      "eval_runtime": 0.659,
-      "eval_samples_per_second": 283.783,
-      "eval_steps_per_second": 4.553,
+      "eval_TIME_precision": 0.84375,
+      "eval_TIME_recall": 0.9310344827586207,
+      "eval_loss": 0.14203932881355286,
+      "eval_overall_accuracy": 0.9562099871959027,
+      "eval_overall_f1": 0.8172942817294282,
+      "eval_overall_precision": 0.7730870712401056,
+      "eval_overall_recall": 0.8668639053254438,
+      "eval_runtime": 0.3189,
+      "eval_samples_per_second": 586.358,
+      "eval_steps_per_second": 9.407,
       "step": 106
     },
     {
       "epoch": 2.0,
-      "grad_norm": 2.0066733360290527,
+      "grad_norm": 2.4212687015533447,
       "learning_rate": 4.9e-05,
-      "loss": 0.1201,
+      "loss": 0.1203,
       "step": 212
     },
     {
       "epoch": 2.0,
-      "eval_LOCATION_f1": 0.8053691275167786,
+      "eval_LOCATION_f1": 0.8266666666666668,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.75,
-      "eval_LOCATION_recall": 0.8695652173913043,
-      "eval_ORGANIZATION_f1": 0.7230769230769231,
+      "eval_LOCATION_precision": 0.7654320987654321,
+      "eval_LOCATION_recall": 0.8985507246376812,
+      "eval_ORGANIZATION_f1": 0.6917293233082706,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.6527777777777778,
-      "eval_ORGANIZATION_recall": 0.8103448275862069,
-      "eval_PERSON_f1": 0.8860759493670887,
+      "eval_ORGANIZATION_precision": 0.6133333333333333,
+      "eval_ORGANIZATION_recall": 0.7931034482758621,
+      "eval_PERSON_f1": 0.8867924528301887,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8536585365853658,
-      "eval_PERSON_recall": 0.9210526315789473,
-      "eval_QUANTITY_f1": 0.6,
+      "eval_PERSON_precision": 0.8493975903614458,
+      "eval_PERSON_recall": 0.9276315789473685,
+      "eval_QUANTITY_f1": 0.6865671641791045,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.525,
-      "eval_QUANTITY_recall": 0.7,
+      "eval_QUANTITY_precision": 0.6216216216216216,
+      "eval_QUANTITY_recall": 0.7666666666666667,
       "eval_TIME_f1": 0.75,
       "eval_TIME_number": 29,
       "eval_TIME_precision": 0.6857142857142857,
       "eval_TIME_recall": 0.8275862068965517,
-      "eval_loss": 0.14421768486499786,
-      "eval_overall_accuracy": 0.9580025608194622,
-      "eval_overall_f1": 0.8010973936899862,
-      "eval_overall_precision": 0.7468030690537084,
-      "eval_overall_recall": 0.863905325443787,
-      "eval_runtime": 0.6779,
-      "eval_samples_per_second": 275.835,
-      "eval_steps_per_second": 4.425,
+      "eval_loss": 0.14541447162628174,
+      "eval_overall_accuracy": 0.9556978233034571,
+      "eval_overall_f1": 0.8087431693989071,
+      "eval_overall_precision": 0.751269035532995,
+      "eval_overall_recall": 0.8757396449704142,
+      "eval_runtime": 0.3201,
+      "eval_samples_per_second": 584.228,
+      "eval_steps_per_second": 9.373,
       "step": 212
     },
     {
       "epoch": 3.0,
-      "grad_norm": 4.826297760009766,
+      "grad_norm": 5.2256245613098145,
       "learning_rate": 4.85e-05,
-      "loss": 0.0789,
+      "loss": 0.0755,
       "step": 318
     },
     {
       "epoch": 3.0,
-      "eval_LOCATION_f1": 0.8051948051948051,
+      "eval_LOCATION_f1": 0.8609271523178808,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.7294117647058823,
-      "eval_LOCATION_recall": 0.8985507246376812,
-      "eval_ORGANIZATION_f1": 0.7218045112781956,
+      "eval_LOCATION_precision": 0.7926829268292683,
+      "eval_LOCATION_recall": 0.9420289855072463,
+      "eval_ORGANIZATION_f1": 0.71875,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.64,
-      "eval_ORGANIZATION_recall": 0.8275862068965517,
-      "eval_PERSON_f1": 0.852760736196319,
+      "eval_ORGANIZATION_precision": 0.6571428571428571,
+      "eval_ORGANIZATION_recall": 0.7931034482758621,
+      "eval_PERSON_f1": 0.8571428571428571,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.7988505747126436,
-      "eval_PERSON_recall": 0.9144736842105263,
-      "eval_QUANTITY_f1": 0.8307692307692307,
+      "eval_PERSON_precision": 0.8117647058823529,
+      "eval_PERSON_recall": 0.9078947368421053,
+      "eval_QUANTITY_f1": 0.78125,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.7714285714285715,
-      "eval_QUANTITY_recall": 0.9,
-      "eval_TIME_f1": 0.8813559322033899,
+      "eval_QUANTITY_precision": 0.7352941176470589,
+      "eval_QUANTITY_recall": 0.8333333333333334,
+      "eval_TIME_f1": 0.8852459016393444,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.8666666666666667,
-      "eval_TIME_recall": 0.896551724137931,
-      "eval_loss": 0.15757465362548828,
-      "eval_overall_accuracy": 0.954673495518566,
-      "eval_overall_f1": 0.819538670284939,
-      "eval_overall_precision": 0.7568922305764411,
-      "eval_overall_recall": 0.893491124260355,
-      "eval_runtime": 0.7577,
-      "eval_samples_per_second": 246.801,
-      "eval_steps_per_second": 3.959,
+      "eval_TIME_precision": 0.84375,
+      "eval_TIME_recall": 0.9310344827586207,
+      "eval_loss": 0.14322705566883087,
+      "eval_overall_accuracy": 0.9613316261203585,
+      "eval_overall_f1": 0.8292011019283746,
+      "eval_overall_precision": 0.7757731958762887,
+      "eval_overall_recall": 0.8905325443786982,
+      "eval_runtime": 0.3175,
+      "eval_samples_per_second": 588.914,
+      "eval_steps_per_second": 9.448,
       "step": 318
     },
     {
       "epoch": 4.0,
-      "grad_norm": 1.5358341932296753,
+      "grad_norm": 1.6368614435195923,
       "learning_rate": 4.8e-05,
-      "loss": 0.0497,
+      "loss": 0.0504,
       "step": 424
     },
     {
       "epoch": 4.0,
-      "eval_LOCATION_f1": 0.8243243243243245,
+      "eval_LOCATION_f1": 0.8456375838926175,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.7721518987341772,
-      "eval_LOCATION_recall": 0.8840579710144928,
-      "eval_ORGANIZATION_f1": 0.765625,
+      "eval_LOCATION_precision": 0.7875,
+      "eval_LOCATION_recall": 0.9130434782608695,
+      "eval_ORGANIZATION_f1": 0.6495726495726496,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.7,
-      "eval_ORGANIZATION_recall": 0.8448275862068966,
-      "eval_PERSON_f1": 0.8980891719745223,
+      "eval_ORGANIZATION_precision": 0.6440677966101694,
+      "eval_ORGANIZATION_recall": 0.6551724137931034,
+      "eval_PERSON_f1": 0.8734177215189874,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8703703703703703,
-      "eval_PERSON_recall": 0.9276315789473685,
-      "eval_QUANTITY_f1": 0.8125,
+      "eval_PERSON_precision": 0.8414634146341463,
+      "eval_PERSON_recall": 0.9078947368421053,
+      "eval_QUANTITY_f1": 0.8387096774193549,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.7647058823529411,
+      "eval_QUANTITY_precision": 0.8125,
       "eval_QUANTITY_recall": 0.8666666666666667,
-      "eval_TIME_f1": 0.75,
+      "eval_TIME_f1": 0.7761194029850746,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.6857142857142857,
-      "eval_TIME_recall": 0.8275862068965517,
-      "eval_loss": 0.15764029324054718,
-      "eval_overall_accuracy": 0.9628681177976952,
-      "eval_overall_f1": 0.8384401114206129,
-      "eval_overall_precision": 0.7921052631578948,
-      "eval_overall_recall": 0.8905325443786982,
-      "eval_runtime": 0.6908,
-      "eval_samples_per_second": 270.686,
-      "eval_steps_per_second": 4.343,
+      "eval_TIME_precision": 0.6842105263157895,
+      "eval_TIME_recall": 0.896551724137931,
+      "eval_loss": 0.14780984818935394,
+      "eval_overall_accuracy": 0.9597951344430218,
+      "eval_overall_f1": 0.8185654008438817,
+      "eval_overall_precision": 0.7801608579088471,
+      "eval_overall_recall": 0.8609467455621301,
+      "eval_runtime": 0.3217,
+      "eval_samples_per_second": 581.313,
+      "eval_steps_per_second": 9.326,
       "step": 424
     },
     {
       "epoch": 5.0,
-      "grad_norm": 0.8067348003387451,
+      "grad_norm": 2.231923818588257,
       "learning_rate": 4.75e-05,
-      "loss": 0.0304,
+      "loss": 0.0318,
       "step": 530
     },
     {
       "epoch": 5.0,
-      "eval_LOCATION_f1": 0.8591549295774648,
+      "eval_LOCATION_f1": 0.8472222222222223,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8356164383561644,
+      "eval_LOCATION_precision": 0.8133333333333334,
       "eval_LOCATION_recall": 0.8840579710144928,
-      "eval_ORGANIZATION_f1": 0.6554621848739496,
+      "eval_ORGANIZATION_f1": 0.7555555555555555,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.639344262295082,
-      "eval_ORGANIZATION_recall": 0.6724137931034483,
-      "eval_PERSON_f1": 0.8681672025723474,
+      "eval_ORGANIZATION_precision": 0.6623376623376623,
+      "eval_ORGANIZATION_recall": 0.8793103448275862,
+      "eval_PERSON_f1": 0.9079365079365079,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8490566037735849,
-      "eval_PERSON_recall": 0.8881578947368421,
-      "eval_QUANTITY_f1": 0.8125,
+      "eval_PERSON_precision": 0.8773006134969326,
+      "eval_PERSON_recall": 0.9407894736842105,
+      "eval_QUANTITY_f1": 0.7761194029850748,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.7647058823529411,
+      "eval_QUANTITY_precision": 0.7027027027027027,
       "eval_QUANTITY_recall": 0.8666666666666667,
-      "eval_TIME_f1": 0.8620689655172413,
+      "eval_TIME_f1": 0.8666666666666666,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.8620689655172413,
-      "eval_TIME_recall": 0.8620689655172413,
-      "eval_loss": 0.18078266084194183,
-      "eval_overall_accuracy": 0.9603072983354674,
-      "eval_overall_f1": 0.8242074927953891,
-      "eval_overall_precision": 0.8033707865168539,
-      "eval_overall_recall": 0.8461538461538461,
-      "eval_runtime": 0.7144,
-      "eval_samples_per_second": 261.763,
-      "eval_steps_per_second": 4.199,
+      "eval_TIME_precision": 0.8387096774193549,
+      "eval_TIME_recall": 0.896551724137931,
+      "eval_loss": 0.18356788158416748,
+      "eval_overall_accuracy": 0.9597951344430218,
+      "eval_overall_f1": 0.8515950069348128,
+      "eval_overall_precision": 0.8015665796344648,
+      "eval_overall_recall": 0.908284023668639,
+      "eval_runtime": 0.3175,
+      "eval_samples_per_second": 588.92,
+      "eval_steps_per_second": 9.448,
       "step": 530
     },
     {
       "epoch": 6.0,
-      "grad_norm": 2.896883010864258,
+      "grad_norm": 4.4022064208984375,
       "learning_rate": 4.7e-05,
-      "loss": 0.0258,
+      "loss": 0.0239,
       "step": 636
     },
     {
       "epoch": 6.0,
-      "eval_LOCATION_f1": 0.8289473684210527,
+      "eval_LOCATION_f1": 0.8450704225352113,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.7590361445783133,
-      "eval_LOCATION_recall": 0.9130434782608695,
-      "eval_ORGANIZATION_f1": 0.6166666666666667,
+      "eval_LOCATION_precision": 0.821917808219178,
+      "eval_LOCATION_recall": 0.8695652173913043,
+      "eval_ORGANIZATION_f1": 0.7014925373134329,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.5967741935483871,
-      "eval_ORGANIZATION_recall": 0.6379310344827587,
-      "eval_PERSON_f1": 0.8867313915857605,
+      "eval_ORGANIZATION_precision": 0.618421052631579,
+      "eval_ORGANIZATION_recall": 0.8103448275862069,
+      "eval_PERSON_f1": 0.8478964401294499,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8726114649681529,
-      "eval_PERSON_recall": 0.9013157894736842,
-      "eval_QUANTITY_f1": 0.8253968253968254,
+      "eval_PERSON_precision": 0.8343949044585988,
+      "eval_PERSON_recall": 0.8618421052631579,
+      "eval_QUANTITY_f1": 0.819672131147541,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.7878787878787878,
-      "eval_QUANTITY_recall": 0.8666666666666667,
-      "eval_TIME_f1": 0.8253968253968255,
+      "eval_QUANTITY_precision": 0.8064516129032258,
+      "eval_QUANTITY_recall": 0.8333333333333334,
+      "eval_TIME_f1": 0.8524590163934426,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.7647058823529411,
+      "eval_TIME_precision": 0.8125,
       "eval_TIME_recall": 0.896551724137931,
-      "eval_loss": 0.18976914882659912,
-      "eval_overall_accuracy": 0.9590268886043534,
+      "eval_loss": 0.1694508045911789,
+      "eval_overall_accuracy": 0.9577464788732394,
       "eval_overall_f1": 0.8175388967468176,
       "eval_overall_precision": 0.7831978319783198,
       "eval_overall_recall": 0.8550295857988166,
-      "eval_runtime": 0.7567,
-      "eval_samples_per_second": 247.118,
-      "eval_steps_per_second": 3.964,
+      "eval_runtime": 0.3257,
+      "eval_samples_per_second": 574.211,
+      "eval_steps_per_second": 9.212,
       "step": 636
     },
     {
       "epoch": 7.0,
-      "grad_norm": 0.08857329934835434,
+      "grad_norm": 0.2985347509384155,
       "learning_rate": 4.6500000000000005e-05,
-      "loss": 0.0151,
+      "loss": 0.0238,
       "step": 742
     },
     {
       "epoch": 7.0,
-      "eval_LOCATION_f1": 0.84,
+      "eval_LOCATION_f1": 0.853146853146853,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.7777777777777778,
-      "eval_LOCATION_recall": 0.9130434782608695,
-      "eval_ORGANIZATION_f1": 0.6491228070175439,
+      "eval_LOCATION_precision": 0.8243243243243243,
+      "eval_LOCATION_recall": 0.8840579710144928,
+      "eval_ORGANIZATION_f1": 0.75,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.6607142857142857,
-      "eval_ORGANIZATION_recall": 0.6379310344827587,
-      "eval_PERSON_f1": 0.8789808917197451,
+      "eval_ORGANIZATION_precision": 0.6857142857142857,
+      "eval_ORGANIZATION_recall": 0.8275862068965517,
+      "eval_PERSON_f1": 0.8616352201257862,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8518518518518519,
-      "eval_PERSON_recall": 0.9078947368421053,
-      "eval_QUANTITY_f1": 0.84375,
+      "eval_PERSON_precision": 0.8253012048192772,
+      "eval_PERSON_recall": 0.9013157894736842,
+      "eval_QUANTITY_f1": 0.7352941176470588,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.7941176470588235,
-      "eval_QUANTITY_recall": 0.9,
-      "eval_TIME_f1": 0.7692307692307692,
+      "eval_QUANTITY_precision": 0.6578947368421053,
+      "eval_QUANTITY_recall": 0.8333333333333334,
+      "eval_TIME_f1": 0.8333333333333334,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.6944444444444444,
+      "eval_TIME_precision": 0.8064516129032258,
       "eval_TIME_recall": 0.8620689655172413,
-      "eval_loss": 0.22696857154369354,
-      "eval_overall_accuracy": 0.9587708066581306,
-      "eval_overall_f1": 0.8203677510608204,
-      "eval_overall_precision": 0.7859078590785907,
-      "eval_overall_recall": 0.8579881656804734,
-      "eval_runtime": 0.7647,
-      "eval_samples_per_second": 244.536,
-      "eval_steps_per_second": 3.923,
+      "eval_loss": 0.19530388712882996,
+      "eval_overall_accuracy": 0.9608194622279129,
+      "eval_overall_f1": 0.8256624825662484,
+      "eval_overall_precision": 0.7810026385224275,
+      "eval_overall_recall": 0.8757396449704142,
+      "eval_runtime": 0.3219,
+      "eval_samples_per_second": 580.938,
+      "eval_steps_per_second": 9.32,
       "step": 742
     },
     {
       "epoch": 8.0,
-      "grad_norm": 0.24211041629314423,
+      "grad_norm": 0.04798915982246399,
       "learning_rate": 4.600000000000001e-05,
-      "loss": 0.0124,
+      "loss": 0.0134,
       "step": 848
     },
     {
       "epoch": 8.0,
-      "eval_LOCATION_f1": 0.8493150684931507,
+      "eval_LOCATION_f1": 0.8435374149659864,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8051948051948052,
+      "eval_LOCATION_precision": 0.7948717948717948,
       "eval_LOCATION_recall": 0.8985507246376812,
-      "eval_ORGANIZATION_f1": 0.706766917293233,
+      "eval_ORGANIZATION_f1": 0.768,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.6266666666666667,
-      "eval_ORGANIZATION_recall": 0.8103448275862069,
-      "eval_PERSON_f1": 0.8761904761904762,
+      "eval_ORGANIZATION_precision": 0.7164179104477612,
+      "eval_ORGANIZATION_recall": 0.8275862068965517,
+      "eval_PERSON_f1": 0.8652037617554859,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8466257668711656,
+      "eval_PERSON_precision": 0.8263473053892215,
       "eval_PERSON_recall": 0.9078947368421053,
-      "eval_QUANTITY_f1": 0.8253968253968254,
+      "eval_QUANTITY_f1": 0.7936507936507938,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.7878787878787878,
-      "eval_QUANTITY_recall": 0.8666666666666667,
-      "eval_TIME_f1": 0.8196721311475409,
+      "eval_QUANTITY_precision": 0.7575757575757576,
+      "eval_QUANTITY_recall": 0.8333333333333334,
+      "eval_TIME_f1": 0.8813559322033899,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.78125,
-      "eval_TIME_recall": 0.8620689655172413,
-      "eval_loss": 0.2596437335014343,
-      "eval_overall_accuracy": 0.9569782330345711,
-      "eval_overall_f1": 0.83008356545961,
-      "eval_overall_precision": 0.7842105263157895,
-      "eval_overall_recall": 0.8816568047337278,
-      "eval_runtime": 0.7892,
-      "eval_samples_per_second": 236.961,
-      "eval_steps_per_second": 3.802,
+      "eval_TIME_precision": 0.8666666666666667,
+      "eval_TIME_recall": 0.896551724137931,
+      "eval_loss": 0.23003453016281128,
+      "eval_overall_accuracy": 0.9633802816901409,
+      "eval_overall_f1": 0.8387096774193548,
+      "eval_overall_precision": 0.7973333333333333,
+      "eval_overall_recall": 0.8846153846153846,
+      "eval_runtime": 0.3389,
+      "eval_samples_per_second": 551.794,
+      "eval_steps_per_second": 8.852,
       "step": 848
     },
     {
       "epoch": 9.0,
-      "grad_norm": 0.5304147005081177,
+      "grad_norm": 0.5828952789306641,
       "learning_rate": 4.55e-05,
-      "loss": 0.0088,
+      "loss": 0.0097,
       "step": 954
     },
     {
       "epoch": 9.0,
-      "eval_LOCATION_f1": 0.8571428571428572,
+      "eval_LOCATION_f1": 0.8356164383561644,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8076923076923077,
-      "eval_LOCATION_recall": 0.9130434782608695,
+      "eval_LOCATION_precision": 0.7922077922077922,
+      "eval_LOCATION_recall": 0.8840579710144928,
       "eval_ORGANIZATION_f1": 0.7401574803149606,
       "eval_ORGANIZATION_number": 58,
       "eval_ORGANIZATION_precision": 0.6811594202898551,
       "eval_ORGANIZATION_recall": 0.8103448275862069,
-      "eval_PERSON_f1": 0.8535825545171339,
+      "eval_PERSON_f1": 0.8859934853420195,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8106508875739645,
-      "eval_PERSON_recall": 0.9013157894736842,
-      "eval_QUANTITY_f1": 0.8253968253968254,
+      "eval_PERSON_precision": 0.8774193548387097,
+      "eval_PERSON_recall": 0.8947368421052632,
+      "eval_QUANTITY_f1": 0.78125,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.7878787878787878,
-      "eval_QUANTITY_recall": 0.8666666666666667,
-      "eval_TIME_f1": 0.8333333333333334,
+      "eval_QUANTITY_precision": 0.7352941176470589,
+      "eval_QUANTITY_recall": 0.8333333333333334,
+      "eval_TIME_f1": 0.896551724137931,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.8064516129032258,
-      "eval_TIME_recall": 0.8620689655172413,
-      "eval_loss": 0.26296016573905945,
-      "eval_overall_accuracy": 0.9562099871959027,
-      "eval_overall_f1": 0.83008356545961,
-      "eval_overall_precision": 0.7842105263157895,
-      "eval_overall_recall": 0.8816568047337278,
-      "eval_runtime": 0.8059,
-      "eval_samples_per_second": 232.037,
-      "eval_steps_per_second": 3.723,
+      "eval_TIME_precision": 0.896551724137931,
+      "eval_TIME_recall": 0.896551724137931,
+      "eval_loss": 0.23057876527309418,
+      "eval_overall_accuracy": 0.9580025608194622,
+      "eval_overall_f1": 0.8404558404558404,
+      "eval_overall_precision": 0.8104395604395604,
+      "eval_overall_recall": 0.8727810650887574,
+      "eval_runtime": 0.3211,
+      "eval_samples_per_second": 582.432,
+      "eval_steps_per_second": 9.344,
       "step": 954
     },
     {
       "epoch": 10.0,
-      "grad_norm": 1.2787593603134155,
+      "grad_norm": 0.18654245138168335,
       "learning_rate": 4.5e-05,
-      "loss": 0.0083,
+      "loss": 0.0078,
       "step": 1060
     },
     {
       "epoch": 10.0,
-      "eval_LOCATION_f1": 0.8513513513513514,
+      "eval_LOCATION_f1": 0.8571428571428572,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.7974683544303798,
+      "eval_LOCATION_precision": 0.8076923076923077,
       "eval_LOCATION_recall": 0.9130434782608695,
-      "eval_ORGANIZATION_f1": 0.7703703703703704,
+      "eval_ORGANIZATION_f1": 0.7906976744186047,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.6753246753246753,
-      "eval_ORGANIZATION_recall": 0.896551724137931,
-      "eval_PERSON_f1": 0.9079365079365079,
+      "eval_ORGANIZATION_precision": 0.7183098591549296,
+      "eval_ORGANIZATION_recall": 0.8793103448275862,
+      "eval_PERSON_f1": 0.8753993610223643,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8773006134969326,
-      "eval_PERSON_recall": 0.9407894736842105,
-      "eval_QUANTITY_f1": 0.7500000000000001,
+      "eval_PERSON_precision": 0.8509316770186336,
+      "eval_PERSON_recall": 0.9013157894736842,
+      "eval_QUANTITY_f1": 0.78125,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.7058823529411765,
-      "eval_QUANTITY_recall": 0.8,
-      "eval_TIME_f1": 0.8135593220338982,
+      "eval_QUANTITY_precision": 0.7352941176470589,
+      "eval_QUANTITY_recall": 0.8333333333333334,
+      "eval_TIME_f1": 0.896551724137931,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.8,
-      "eval_TIME_recall": 0.8275862068965517,
-      "eval_loss": 0.23285891115665436,
-      "eval_overall_accuracy": 0.963124199743918,
-      "eval_overall_f1": 0.8488210818307906,
-      "eval_overall_precision": 0.7989556135770235,
-      "eval_overall_recall": 0.9053254437869822,
-      "eval_runtime": 0.8566,
-      "eval_samples_per_second": 218.314,
-      "eval_steps_per_second": 3.502,
+      "eval_TIME_precision": 0.896551724137931,
+      "eval_TIME_recall": 0.896551724137931,
+      "eval_loss": 0.2440958321094513,
+      "eval_overall_accuracy": 0.9664532650448143,
+      "eval_overall_f1": 0.849507735583685,
+      "eval_overall_precision": 0.8096514745308311,
+      "eval_overall_recall": 0.893491124260355,
+      "eval_runtime": 0.3238,
+      "eval_samples_per_second": 577.519,
+      "eval_steps_per_second": 9.265,
       "step": 1060
     },
     {
       "epoch": 11.0,
-      "grad_norm": 2.0766656398773193,
+      "grad_norm": 0.014143639244139194,
       "learning_rate": 4.4500000000000004e-05,
-      "loss": 0.0088,
+      "loss": 0.0057,
       "step": 1166
     },
     {
@@ -411,349 +411,349 @@
       "eval_LOCATION_number": 69,
       "eval_LOCATION_precision": 0.8076923076923077,
       "eval_LOCATION_recall": 0.9130434782608695,
-      "eval_ORGANIZATION_f1": 0.7154471544715447,
+      "eval_ORGANIZATION_f1": 0.6440677966101694,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.676923076923077,
-      "eval_ORGANIZATION_recall": 0.7586206896551724,
-      "eval_PERSON_f1": 0.875,
+      "eval_ORGANIZATION_precision": 0.6333333333333333,
+      "eval_ORGANIZATION_recall": 0.6551724137931034,
+      "eval_PERSON_f1": 0.8838709677419354,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8333333333333334,
-      "eval_PERSON_recall": 0.9210526315789473,
-      "eval_QUANTITY_f1": 0.78125,
+      "eval_PERSON_precision": 0.8670886075949367,
+      "eval_PERSON_recall": 0.9013157894736842,
+      "eval_QUANTITY_f1": 0.8387096774193549,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.7352941176470589,
-      "eval_QUANTITY_recall": 0.8333333333333334,
-      "eval_TIME_f1": 0.8387096774193549,
+      "eval_QUANTITY_precision": 0.8125,
+      "eval_QUANTITY_recall": 0.8666666666666667,
+      "eval_TIME_f1": 0.8524590163934426,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.7878787878787878,
+      "eval_TIME_precision": 0.8125,
       "eval_TIME_recall": 0.896551724137931,
-      "eval_loss": 0.23979602754116058,
-      "eval_overall_accuracy": 0.9610755441741358,
-      "eval_overall_f1": 0.8324022346368716,
-      "eval_overall_precision": 0.7883597883597884,
-      "eval_overall_recall": 0.8816568047337278,
-      "eval_runtime": 0.7972,
-      "eval_samples_per_second": 234.561,
-      "eval_steps_per_second": 3.763,
+      "eval_loss": 0.2606862783432007,
+      "eval_overall_accuracy": 0.9613316261203585,
+      "eval_overall_f1": 0.8309455587392551,
+      "eval_overall_precision": 0.8055555555555556,
+      "eval_overall_recall": 0.8579881656804734,
+      "eval_runtime": 0.32,
+      "eval_samples_per_second": 584.447,
+      "eval_steps_per_second": 9.376,
       "step": 1166
     },
     {
       "epoch": 12.0,
-      "grad_norm": 0.004628497175872326,
+      "grad_norm": 0.8590062260627747,
       "learning_rate": 4.4000000000000006e-05,
-      "loss": 0.007,
+      "loss": 0.0057,
       "step": 1272
     },
     {
       "epoch": 12.0,
-      "eval_LOCATION_f1": 0.8243243243243245,
+      "eval_LOCATION_f1": 0.8368794326241135,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.7721518987341772,
-      "eval_LOCATION_recall": 0.8840579710144928,
-      "eval_ORGANIZATION_f1": 0.7704918032786885,
+      "eval_LOCATION_precision": 0.8194444444444444,
+      "eval_LOCATION_recall": 0.855072463768116,
+      "eval_ORGANIZATION_f1": 0.7627118644067795,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.734375,
-      "eval_ORGANIZATION_recall": 0.8103448275862069,
-      "eval_PERSON_f1": 0.8758169934640522,
+      "eval_ORGANIZATION_precision": 0.75,
+      "eval_ORGANIZATION_recall": 0.7758620689655172,
+      "eval_PERSON_f1": 0.8706624605678233,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8701298701298701,
-      "eval_PERSON_recall": 0.881578947368421,
-      "eval_QUANTITY_f1": 0.7936507936507938,
+      "eval_PERSON_precision": 0.8363636363636363,
+      "eval_PERSON_recall": 0.9078947368421053,
+      "eval_QUANTITY_f1": 0.8064516129032259,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.7575757575757576,
+      "eval_QUANTITY_precision": 0.78125,
       "eval_QUANTITY_recall": 0.8333333333333334,
-      "eval_TIME_f1": 0.8813559322033899,
+      "eval_TIME_f1": 0.8620689655172413,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.8666666666666667,
-      "eval_TIME_recall": 0.896551724137931,
-      "eval_loss": 0.23579084873199463,
+      "eval_TIME_precision": 0.8620689655172413,
+      "eval_TIME_recall": 0.8620689655172413,
+      "eval_loss": 0.2562354505062103,
       "eval_overall_accuracy": 0.9623559539052496,
-      "eval_overall_f1": 0.8395415472779371,
-      "eval_overall_precision": 0.8138888888888889,
-      "eval_overall_recall": 0.8668639053254438,
-      "eval_runtime": 0.8766,
-      "eval_samples_per_second": 213.313,
-      "eval_steps_per_second": 3.422,
+      "eval_overall_f1": 0.8390804597701149,
+      "eval_overall_precision": 0.8156424581005587,
+      "eval_overall_recall": 0.863905325443787,
+      "eval_runtime": 0.3245,
+      "eval_samples_per_second": 576.335,
+      "eval_steps_per_second": 9.246,
       "step": 1272
     },
     {
       "epoch": 13.0,
-      "grad_norm": 0.023974481970071793,
+      "grad_norm": 0.005929162725806236,
       "learning_rate": 4.35e-05,
-      "loss": 0.0061,
+      "loss": 0.0041,
       "step": 1378
     },
     {
       "epoch": 13.0,
-      "eval_LOCATION_f1": 0.851063829787234,
+      "eval_LOCATION_f1": 0.8724832214765101,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8333333333333334,
-      "eval_LOCATION_recall": 0.8695652173913043,
-      "eval_ORGANIZATION_f1": 0.6019417475728155,
+      "eval_LOCATION_precision": 0.8125,
+      "eval_LOCATION_recall": 0.9420289855072463,
+      "eval_ORGANIZATION_f1": 0.7768595041322314,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.6888888888888889,
-      "eval_ORGANIZATION_recall": 0.5344827586206896,
-      "eval_PERSON_f1": 0.8932038834951456,
+      "eval_ORGANIZATION_precision": 0.746031746031746,
+      "eval_ORGANIZATION_recall": 0.8103448275862069,
+      "eval_PERSON_f1": 0.8690095846645369,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8789808917197452,
-      "eval_PERSON_recall": 0.9078947368421053,
-      "eval_QUANTITY_f1": 0.8,
+      "eval_PERSON_precision": 0.84472049689441,
+      "eval_PERSON_recall": 0.8947368421052632,
+      "eval_QUANTITY_f1": 0.819672131147541,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.7428571428571429,
-      "eval_QUANTITY_recall": 0.8666666666666667,
+      "eval_QUANTITY_precision": 0.8064516129032258,
+      "eval_QUANTITY_recall": 0.8333333333333334,
       "eval_TIME_f1": 0.8813559322033899,
       "eval_TIME_number": 29,
       "eval_TIME_precision": 0.8666666666666667,
       "eval_TIME_recall": 0.896551724137931,
-      "eval_loss": 0.24468651413917542,
-      "eval_overall_accuracy": 0.9613316261203585,
-      "eval_overall_f1": 0.8301329394387003,
-      "eval_overall_precision": 0.8289085545722714,
-      "eval_overall_recall": 0.8313609467455622,
-      "eval_runtime": 0.8154,
-      "eval_samples_per_second": 229.344,
-      "eval_steps_per_second": 3.679,
+      "eval_loss": 0.26208436489105225,
+      "eval_overall_accuracy": 0.964404609475032,
+      "eval_overall_f1": 0.8506401137980085,
+      "eval_overall_precision": 0.8191780821917808,
+      "eval_overall_recall": 0.8846153846153846,
+      "eval_runtime": 0.3219,
+      "eval_samples_per_second": 580.969,
+      "eval_steps_per_second": 9.32,
       "step": 1378
     },
     {
       "epoch": 14.0,
-      "grad_norm": 0.16095533967018127,
+      "grad_norm": 0.20337995886802673,
       "learning_rate": 4.3e-05,
-      "loss": 0.0073,
+      "loss": 0.0045,
       "step": 1484
     },
     {
       "epoch": 14.0,
-      "eval_LOCATION_f1": 0.8551724137931034,
+      "eval_LOCATION_f1": 0.8311688311688311,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8157894736842105,
-      "eval_LOCATION_recall": 0.8985507246376812,
-      "eval_ORGANIZATION_f1": 0.7500000000000001,
+      "eval_LOCATION_precision": 0.7529411764705882,
+      "eval_LOCATION_recall": 0.927536231884058,
+      "eval_ORGANIZATION_f1": 0.7241379310344829,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.7258064516129032,
-      "eval_ORGANIZATION_recall": 0.7758620689655172,
-      "eval_PERSON_f1": 0.8996763754045308,
+      "eval_ORGANIZATION_precision": 0.7241379310344828,
+      "eval_ORGANIZATION_recall": 0.7241379310344828,
+      "eval_PERSON_f1": 0.8717948717948718,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8853503184713376,
-      "eval_PERSON_recall": 0.9144736842105263,
-      "eval_QUANTITY_f1": 0.8307692307692307,
+      "eval_PERSON_precision": 0.85,
+      "eval_PERSON_recall": 0.8947368421052632,
+      "eval_QUANTITY_f1": 0.8253968253968254,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.7714285714285715,
-      "eval_QUANTITY_recall": 0.9,
-      "eval_TIME_f1": 0.8666666666666666,
+      "eval_QUANTITY_precision": 0.7878787878787878,
+      "eval_QUANTITY_recall": 0.8666666666666667,
+      "eval_TIME_f1": 0.8253968253968255,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.8387096774193549,
+      "eval_TIME_precision": 0.7647058823529411,
       "eval_TIME_recall": 0.896551724137931,
-      "eval_loss": 0.23856018483638763,
-      "eval_overall_accuracy": 0.9641485275288092,
-      "eval_overall_f1": 0.855507868383405,
-      "eval_overall_precision": 0.8282548476454293,
-      "eval_overall_recall": 0.8846153846153846,
-      "eval_runtime": 0.7969,
-      "eval_samples_per_second": 234.66,
-      "eval_steps_per_second": 3.765,
+      "eval_loss": 0.2810960114002228,
+      "eval_overall_accuracy": 0.9633802816901409,
+      "eval_overall_f1": 0.8305084745762711,
+      "eval_overall_precision": 0.7945945945945946,
+      "eval_overall_recall": 0.8698224852071006,
+      "eval_runtime": 0.3298,
+      "eval_samples_per_second": 566.967,
+      "eval_steps_per_second": 9.096,
       "step": 1484
     },
     {
       "epoch": 15.0,
-      "grad_norm": 3.3597922325134277,
+      "grad_norm": 1.0766212940216064,
       "learning_rate": 4.25e-05,
-      "loss": 0.0046,
+      "loss": 0.0048,
       "step": 1590
     },
     {
       "epoch": 15.0,
-      "eval_LOCATION_f1": 0.8493150684931507,
+      "eval_LOCATION_f1": 0.8666666666666666,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8051948051948052,
-      "eval_LOCATION_recall": 0.8985507246376812,
-      "eval_ORGANIZATION_f1": 0.75,
+      "eval_LOCATION_precision": 0.8024691358024691,
+      "eval_LOCATION_recall": 0.9420289855072463,
+      "eval_ORGANIZATION_f1": 0.7642276422764227,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.6857142857142857,
-      "eval_ORGANIZATION_recall": 0.8275862068965517,
-      "eval_PERSON_f1": 0.9055374592833876,
+      "eval_ORGANIZATION_precision": 0.7230769230769231,
+      "eval_ORGANIZATION_recall": 0.8103448275862069,
+      "eval_PERSON_f1": 0.8881789137380192,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.896774193548387,
+      "eval_PERSON_precision": 0.8633540372670807,
       "eval_PERSON_recall": 0.9144736842105263,
-      "eval_QUANTITY_f1": 0.8571428571428572,
+      "eval_QUANTITY_f1": 0.8125,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.8181818181818182,
-      "eval_QUANTITY_recall": 0.9,
+      "eval_QUANTITY_precision": 0.7647058823529411,
+      "eval_QUANTITY_recall": 0.8666666666666667,
       "eval_TIME_f1": 0.8666666666666666,
       "eval_TIME_number": 29,
       "eval_TIME_precision": 0.8387096774193549,
       "eval_TIME_recall": 0.896551724137931,
-      "eval_loss": 0.23872800171375275,
-      "eval_overall_accuracy": 0.9651728553137003,
-      "eval_overall_f1": 0.8579545454545454,
-      "eval_overall_precision": 0.825136612021858,
-      "eval_overall_recall": 0.893491124260355,
-      "eval_runtime": 0.8079,
-      "eval_samples_per_second": 231.459,
-      "eval_steps_per_second": 3.713,
+      "eval_loss": 0.26082485914230347,
+      "eval_overall_accuracy": 0.9626120358514725,
+      "eval_overall_f1": 0.8535211267605635,
+      "eval_overall_precision": 0.8145161290322581,
+      "eval_overall_recall": 0.8964497041420119,
+      "eval_runtime": 0.3242,
+      "eval_samples_per_second": 576.835,
+      "eval_steps_per_second": 9.254,
       "step": 1590
     },
     {
       "epoch": 16.0,
-      "grad_norm": 4.114704608917236,
+      "grad_norm": 0.31145572662353516,
       "learning_rate": 4.2e-05,
-      "loss": 0.003,
+      "loss": 0.0054,
       "step": 1696
     },
     {
       "epoch": 16.0,
-      "eval_LOCATION_f1": 0.8493150684931507,
+      "eval_LOCATION_f1": 0.8591549295774648,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8051948051948052,
-      "eval_LOCATION_recall": 0.8985507246376812,
-      "eval_ORGANIZATION_f1": 0.6363636363636364,
+      "eval_LOCATION_precision": 0.8356164383561644,
+      "eval_LOCATION_recall": 0.8840579710144928,
+      "eval_ORGANIZATION_f1": 0.7441860465116279,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.6730769230769231,
-      "eval_ORGANIZATION_recall": 0.603448275862069,
-      "eval_PERSON_f1": 0.9042904290429044,
+      "eval_ORGANIZATION_precision": 0.676056338028169,
+      "eval_ORGANIZATION_recall": 0.8275862068965517,
+      "eval_PERSON_f1": 0.8980891719745223,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.9072847682119205,
-      "eval_PERSON_recall": 0.9013157894736842,
-      "eval_QUANTITY_f1": 0.8571428571428572,
+      "eval_PERSON_precision": 0.8703703703703703,
+      "eval_PERSON_recall": 0.9276315789473685,
+      "eval_QUANTITY_f1": 0.8387096774193549,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.8181818181818182,
-      "eval_QUANTITY_recall": 0.9,
-      "eval_TIME_f1": 0.847457627118644,
+      "eval_QUANTITY_precision": 0.8125,
+      "eval_QUANTITY_recall": 0.8666666666666667,
+      "eval_TIME_f1": 0.8771929824561403,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.8333333333333334,
+      "eval_TIME_precision": 0.8928571428571429,
       "eval_TIME_recall": 0.8620689655172413,
-      "eval_loss": 0.2643202841281891,
-      "eval_overall_accuracy": 0.9610755441741358,
-      "eval_overall_f1": 0.8399412628487519,
-      "eval_overall_precision": 0.8338192419825073,
-      "eval_overall_recall": 0.8461538461538461,
-      "eval_runtime": 0.7994,
-      "eval_samples_per_second": 233.932,
-      "eval_steps_per_second": 3.753,
+      "eval_loss": 0.26623842120170593,
+      "eval_overall_accuracy": 0.9608194622279129,
+      "eval_overall_f1": 0.8551136363636364,
+      "eval_overall_precision": 0.8224043715846995,
+      "eval_overall_recall": 0.8905325443786982,
+      "eval_runtime": 0.3327,
+      "eval_samples_per_second": 562.013,
+      "eval_steps_per_second": 9.016,
       "step": 1696
     },
     {
       "epoch": 17.0,
-      "grad_norm": 0.06794843822717667,
+      "grad_norm": 0.1136045828461647,
       "learning_rate": 4.15e-05,
-      "loss": 0.0027,
+      "loss": 0.0048,
       "step": 1802
     },
     {
       "epoch": 17.0,
-      "eval_LOCATION_f1": 0.8724832214765101,
+      "eval_LOCATION_f1": 0.8666666666666666,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8125,
+      "eval_LOCATION_precision": 0.8024691358024691,
       "eval_LOCATION_recall": 0.9420289855072463,
-      "eval_ORGANIZATION_f1": 0.7538461538461538,
+      "eval_ORGANIZATION_f1": 0.7301587301587301,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.6805555555555556,
-      "eval_ORGANIZATION_recall": 0.8448275862068966,
-      "eval_PERSON_f1": 0.8701298701298701,
+      "eval_ORGANIZATION_precision": 0.6764705882352942,
+      "eval_ORGANIZATION_recall": 0.7931034482758621,
+      "eval_PERSON_f1": 0.8673139158576052,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8589743589743589,
+      "eval_PERSON_precision": 0.8535031847133758,
       "eval_PERSON_recall": 0.881578947368421,
-      "eval_QUANTITY_f1": 0.84375,
+      "eval_QUANTITY_f1": 0.7936507936507938,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.7941176470588235,
-      "eval_QUANTITY_recall": 0.9,
-      "eval_TIME_f1": 0.8666666666666666,
+      "eval_QUANTITY_precision": 0.7575757575757576,
+      "eval_QUANTITY_recall": 0.8333333333333334,
+      "eval_TIME_f1": 0.7575757575757576,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.8387096774193549,
-      "eval_TIME_recall": 0.896551724137931,
-      "eval_loss": 0.279176265001297,
+      "eval_TIME_precision": 0.6756756756756757,
+      "eval_TIME_recall": 0.8620689655172413,
+      "eval_loss": 0.28109654784202576,
       "eval_overall_accuracy": 0.9613316261203585,
-      "eval_overall_f1": 0.8466947960618846,
-      "eval_overall_precision": 0.806970509383378,
-      "eval_overall_recall": 0.8905325443786982,
-      "eval_runtime": 0.8067,
-      "eval_samples_per_second": 231.813,
-      "eval_steps_per_second": 3.719,
+      "eval_overall_f1": 0.8263305322128851,
+      "eval_overall_precision": 0.7845744680851063,
+      "eval_overall_recall": 0.8727810650887574,
+      "eval_runtime": 0.3218,
+      "eval_samples_per_second": 581.076,
+      "eval_steps_per_second": 9.322,
       "step": 1802
     },
     {
       "epoch": 18.0,
-      "grad_norm": 0.005153521429747343,
+      "grad_norm": 0.021722104400396347,
       "learning_rate": 4.1e-05,
-      "loss": 0.0037,
+      "loss": 0.0039,
       "step": 1908
     },
     {
       "epoch": 18.0,
-      "eval_LOCATION_f1": 0.8366013071895425,
+      "eval_LOCATION_f1": 0.8591549295774648,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.7619047619047619,
-      "eval_LOCATION_recall": 0.927536231884058,
-      "eval_ORGANIZATION_f1": 0.7154471544715447,
+      "eval_LOCATION_precision": 0.8356164383561644,
+      "eval_LOCATION_recall": 0.8840579710144928,
+      "eval_ORGANIZATION_f1": 0.7692307692307692,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.676923076923077,
-      "eval_ORGANIZATION_recall": 0.7586206896551724,
-      "eval_PERSON_f1": 0.8945686900958467,
+      "eval_ORGANIZATION_precision": 0.6944444444444444,
+      "eval_ORGANIZATION_recall": 0.8620689655172413,
+      "eval_PERSON_f1": 0.8375,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8695652173913043,
-      "eval_PERSON_recall": 0.9210526315789473,
-      "eval_QUANTITY_f1": 0.8571428571428572,
+      "eval_PERSON_precision": 0.7976190476190477,
+      "eval_PERSON_recall": 0.881578947368421,
+      "eval_QUANTITY_f1": 0.78125,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.8181818181818182,
-      "eval_QUANTITY_recall": 0.9,
-      "eval_TIME_f1": 0.8524590163934426,
+      "eval_QUANTITY_precision": 0.7352941176470589,
+      "eval_QUANTITY_recall": 0.8333333333333334,
+      "eval_TIME_f1": 0.8064516129032258,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.8125,
-      "eval_TIME_recall": 0.896551724137931,
-      "eval_loss": 0.27270039916038513,
-      "eval_overall_accuracy": 0.9620998719590269,
-      "eval_overall_f1": 0.8443197755960729,
-      "eval_overall_precision": 0.8026666666666666,
-      "eval_overall_recall": 0.8905325443786982,
-      "eval_runtime": 0.8021,
-      "eval_samples_per_second": 233.132,
-      "eval_steps_per_second": 3.74,
+      "eval_TIME_precision": 0.7575757575757576,
+      "eval_TIME_recall": 0.8620689655172413,
+      "eval_loss": 0.29267749190330505,
+      "eval_overall_accuracy": 0.9574903969270167,
+      "eval_overall_f1": 0.8217270194986072,
+      "eval_overall_precision": 0.7763157894736842,
+      "eval_overall_recall": 0.8727810650887574,
+      "eval_runtime": 0.3291,
+      "eval_samples_per_second": 568.172,
+      "eval_steps_per_second": 9.115,
       "step": 1908
     },
     {
       "epoch": 19.0,
-      "grad_norm": 0.6481520533561707,
+      "grad_norm": 0.02122313901782036,
       "learning_rate": 4.05e-05,
-      "loss": 0.0046,
+      "loss": 0.0038,
       "step": 2014
     },
     {
       "epoch": 19.0,
-      "eval_LOCATION_f1": 0.8378378378378378,
+      "eval_LOCATION_f1": 0.8767123287671235,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.7848101265822784,
-      "eval_LOCATION_recall": 0.8985507246376812,
-      "eval_ORGANIZATION_f1": 0.7258064516129032,
+      "eval_LOCATION_precision": 0.8311688311688312,
+      "eval_LOCATION_recall": 0.927536231884058,
+      "eval_ORGANIZATION_f1": 0.6666666666666667,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.6818181818181818,
-      "eval_ORGANIZATION_recall": 0.7758620689655172,
-      "eval_PERSON_f1": 0.9155844155844156,
+      "eval_ORGANIZATION_precision": 0.6176470588235294,
+      "eval_ORGANIZATION_recall": 0.7241379310344828,
+      "eval_PERSON_f1": 0.8974358974358975,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.9038461538461539,
-      "eval_PERSON_recall": 0.9276315789473685,
-      "eval_QUANTITY_f1": 0.8571428571428572,
+      "eval_PERSON_precision": 0.875,
+      "eval_PERSON_recall": 0.9210526315789473,
+      "eval_QUANTITY_f1": 0.8064516129032259,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.8181818181818182,
-      "eval_QUANTITY_recall": 0.9,
-      "eval_TIME_f1": 0.8620689655172413,
+      "eval_QUANTITY_precision": 0.78125,
+      "eval_QUANTITY_recall": 0.8333333333333334,
+      "eval_TIME_f1": 0.8524590163934426,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.8620689655172413,
-      "eval_TIME_recall": 0.8620689655172413,
-      "eval_loss": 0.24588538706302643,
-      "eval_overall_accuracy": 0.9636363636363636,
-      "eval_overall_f1": 0.855920114122682,
-      "eval_overall_precision": 0.8264462809917356,
-      "eval_overall_recall": 0.8875739644970414,
-      "eval_runtime": 0.8194,
-      "eval_samples_per_second": 228.23,
-      "eval_steps_per_second": 3.661,
+      "eval_TIME_precision": 0.8125,
+      "eval_TIME_recall": 0.896551724137931,
+      "eval_loss": 0.2553775906562805,
+      "eval_overall_accuracy": 0.9633802816901409,
+      "eval_overall_f1": 0.8401697312588403,
+      "eval_overall_precision": 0.8048780487804879,
+      "eval_overall_recall": 0.878698224852071,
+      "eval_runtime": 0.3254,
+      "eval_samples_per_second": 574.754,
+      "eval_steps_per_second": 9.221,
       "step": 2014
     },
     {
       "epoch": 20.0,
-      "grad_norm": 0.020622072741389275,
+      "grad_norm": 0.002756689442321658,
       "learning_rate": 4e-05,
-      "loss": 0.0048,
+      "loss": 0.0044,
       "step": 2120
     },
     {
@@ -762,1301 +762,1301 @@
       "eval_LOCATION_number": 69,
       "eval_LOCATION_precision": 0.7590361445783133,
       "eval_LOCATION_recall": 0.9130434782608695,
-      "eval_ORGANIZATION_f1": 0.7317073170731708,
+      "eval_ORGANIZATION_f1": 0.7107438016528926,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.6923076923076923,
-      "eval_ORGANIZATION_recall": 0.7758620689655172,
-      "eval_PERSON_f1": 0.9003215434083601,
+      "eval_ORGANIZATION_precision": 0.6825396825396826,
+      "eval_ORGANIZATION_recall": 0.7413793103448276,
+      "eval_PERSON_f1": 0.896774193548387,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8805031446540881,
-      "eval_PERSON_recall": 0.9210526315789473,
-      "eval_QUANTITY_f1": 0.8571428571428572,
+      "eval_PERSON_precision": 0.879746835443038,
+      "eval_PERSON_recall": 0.9144736842105263,
+      "eval_QUANTITY_f1": 0.8125,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.8181818181818182,
-      "eval_QUANTITY_recall": 0.9,
-      "eval_TIME_f1": 0.8125,
+      "eval_QUANTITY_precision": 0.7647058823529411,
+      "eval_QUANTITY_recall": 0.8666666666666667,
+      "eval_TIME_f1": 0.8253968253968255,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.7428571428571429,
+      "eval_TIME_precision": 0.7647058823529411,
       "eval_TIME_recall": 0.896551724137931,
-      "eval_loss": 0.2616201937198639,
-      "eval_overall_accuracy": 0.9613316261203585,
-      "eval_overall_f1": 0.8443197755960729,
-      "eval_overall_precision": 0.8026666666666666,
-      "eval_overall_recall": 0.8905325443786982,
-      "eval_runtime": 0.8054,
-      "eval_samples_per_second": 232.17,
-      "eval_steps_per_second": 3.725,
+      "eval_loss": 0.2622649371623993,
+      "eval_overall_accuracy": 0.9620998719590269,
+      "eval_overall_f1": 0.8366197183098592,
+      "eval_overall_precision": 0.7983870967741935,
+      "eval_overall_recall": 0.878698224852071,
+      "eval_runtime": 0.3319,
+      "eval_samples_per_second": 563.374,
+      "eval_steps_per_second": 9.038,
       "step": 2120
     },
     {
       "epoch": 21.0,
-      "grad_norm": 3.6123130321502686,
+      "grad_norm": 0.012515822425484657,
       "learning_rate": 3.9500000000000005e-05,
-      "loss": 0.0047,
+      "loss": 0.0033,
       "step": 2226
     },
     {
       "epoch": 21.0,
-      "eval_LOCATION_f1": 0.8666666666666666,
+      "eval_LOCATION_f1": 0.8513513513513514,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8024691358024691,
-      "eval_LOCATION_recall": 0.9420289855072463,
-      "eval_ORGANIZATION_f1": 0.736,
+      "eval_LOCATION_precision": 0.7974683544303798,
+      "eval_LOCATION_recall": 0.9130434782608695,
+      "eval_ORGANIZATION_f1": 0.735042735042735,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.6865671641791045,
-      "eval_ORGANIZATION_recall": 0.7931034482758621,
-      "eval_PERSON_f1": 0.8932038834951456,
+      "eval_ORGANIZATION_precision": 0.7288135593220338,
+      "eval_ORGANIZATION_recall": 0.7413793103448276,
+      "eval_PERSON_f1": 0.8853503184713376,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8789808917197452,
-      "eval_PERSON_recall": 0.9078947368421053,
-      "eval_QUANTITY_f1": 0.8253968253968254,
+      "eval_PERSON_precision": 0.8580246913580247,
+      "eval_PERSON_recall": 0.9144736842105263,
+      "eval_QUANTITY_f1": 0.7878787878787877,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.7878787878787878,
+      "eval_QUANTITY_precision": 0.7222222222222222,
       "eval_QUANTITY_recall": 0.8666666666666667,
-      "eval_TIME_f1": 0.8307692307692308,
+      "eval_TIME_f1": 0.8333333333333334,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.75,
-      "eval_TIME_recall": 0.9310344827586207,
-      "eval_loss": 0.24132052063941956,
-      "eval_overall_accuracy": 0.9646606914212548,
-      "eval_overall_f1": 0.848314606741573,
-      "eval_overall_precision": 0.8074866310160428,
-      "eval_overall_recall": 0.893491124260355,
-      "eval_runtime": 0.8121,
-      "eval_samples_per_second": 230.255,
-      "eval_steps_per_second": 3.694,
+      "eval_TIME_precision": 0.8064516129032258,
+      "eval_TIME_recall": 0.8620689655172413,
+      "eval_loss": 0.27664482593536377,
+      "eval_overall_accuracy": 0.9600512163892445,
+      "eval_overall_f1": 0.8397163120567377,
+      "eval_overall_precision": 0.8065395095367848,
+      "eval_overall_recall": 0.8757396449704142,
+      "eval_runtime": 0.3226,
+      "eval_samples_per_second": 579.696,
+      "eval_steps_per_second": 9.3,
       "step": 2226
     },
     {
       "epoch": 22.0,
-      "grad_norm": 0.009204844012856483,
+      "grad_norm": 0.001122328918427229,
       "learning_rate": 3.9000000000000006e-05,
-      "loss": 0.0041,
+      "loss": 0.0048,
       "step": 2332
     },
     {
       "epoch": 22.0,
-      "eval_LOCATION_f1": 0.8476821192052981,
+      "eval_LOCATION_f1": 0.8904109589041096,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.7804878048780488,
-      "eval_LOCATION_recall": 0.927536231884058,
-      "eval_ORGANIZATION_f1": 0.7384615384615385,
+      "eval_LOCATION_precision": 0.8441558441558441,
+      "eval_LOCATION_recall": 0.9420289855072463,
+      "eval_ORGANIZATION_f1": 0.7727272727272727,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.6666666666666666,
-      "eval_ORGANIZATION_recall": 0.8275862068965517,
-      "eval_PERSON_f1": 0.9032258064516129,
+      "eval_ORGANIZATION_precision": 0.6891891891891891,
+      "eval_ORGANIZATION_recall": 0.8793103448275862,
+      "eval_PERSON_f1": 0.8974358974358975,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8860759493670886,
+      "eval_PERSON_precision": 0.875,
       "eval_PERSON_recall": 0.9210526315789473,
-      "eval_QUANTITY_f1": 0.78125,
+      "eval_QUANTITY_f1": 0.8253968253968254,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.7352941176470589,
-      "eval_QUANTITY_recall": 0.8333333333333334,
-      "eval_TIME_f1": 0.8125,
+      "eval_QUANTITY_precision": 0.7878787878787878,
+      "eval_QUANTITY_recall": 0.8666666666666667,
+      "eval_TIME_f1": 0.8813559322033899,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.7428571428571429,
+      "eval_TIME_precision": 0.8666666666666667,
       "eval_TIME_recall": 0.896551724137931,
-      "eval_loss": 0.2953908145427704,
-      "eval_overall_accuracy": 0.9620998719590269,
-      "eval_overall_f1": 0.8428372739916552,
-      "eval_overall_precision": 0.7952755905511811,
-      "eval_overall_recall": 0.8964497041420119,
-      "eval_runtime": 0.8088,
-      "eval_samples_per_second": 231.216,
-      "eval_steps_per_second": 3.709,
+      "eval_loss": 0.26052865386009216,
+      "eval_overall_accuracy": 0.9651728553137003,
+      "eval_overall_f1": 0.8651685393258427,
+      "eval_overall_precision": 0.8235294117647058,
+      "eval_overall_recall": 0.9112426035502958,
+      "eval_runtime": 0.33,
+      "eval_samples_per_second": 566.654,
+      "eval_steps_per_second": 9.091,
       "step": 2332
     },
     {
       "epoch": 23.0,
-      "grad_norm": 0.007008604239672422,
+      "grad_norm": 0.01085362397134304,
       "learning_rate": 3.85e-05,
-      "loss": 0.0034,
+      "loss": 0.003,
       "step": 2438
     },
     {
       "epoch": 23.0,
-      "eval_LOCATION_f1": 0.8435374149659864,
+      "eval_LOCATION_f1": 0.8378378378378378,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.7948717948717948,
+      "eval_LOCATION_precision": 0.7848101265822784,
       "eval_LOCATION_recall": 0.8985507246376812,
-      "eval_ORGANIZATION_f1": 0.765625,
+      "eval_ORGANIZATION_f1": 0.7727272727272727,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.7,
-      "eval_ORGANIZATION_recall": 0.8448275862068966,
-      "eval_PERSON_f1": 0.9131832797427653,
+      "eval_ORGANIZATION_precision": 0.6891891891891891,
+      "eval_ORGANIZATION_recall": 0.8793103448275862,
+      "eval_PERSON_f1": 0.9032258064516129,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8930817610062893,
-      "eval_PERSON_recall": 0.9342105263157895,
-      "eval_QUANTITY_f1": 0.819672131147541,
+      "eval_PERSON_precision": 0.8860759493670886,
+      "eval_PERSON_recall": 0.9210526315789473,
+      "eval_QUANTITY_f1": 0.8387096774193549,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.8064516129032258,
-      "eval_QUANTITY_recall": 0.8333333333333334,
-      "eval_TIME_f1": 0.8125,
+      "eval_QUANTITY_precision": 0.8125,
+      "eval_QUANTITY_recall": 0.8666666666666667,
+      "eval_TIME_f1": 0.8666666666666666,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.7428571428571429,
+      "eval_TIME_precision": 0.8387096774193549,
       "eval_TIME_recall": 0.896551724137931,
-      "eval_loss": 0.26445022225379944,
-      "eval_overall_accuracy": 0.9661971830985916,
-      "eval_overall_f1": 0.8551336146272854,
-      "eval_overall_precision": 0.8150134048257373,
-      "eval_overall_recall": 0.8994082840236687,
-      "eval_runtime": 0.797,
-      "eval_samples_per_second": 234.637,
-      "eval_steps_per_second": 3.764,
+      "eval_loss": 0.2736039459705353,
+      "eval_overall_accuracy": 0.9646606914212548,
+      "eval_overall_f1": 0.8567415730337079,
+      "eval_overall_precision": 0.8155080213903744,
+      "eval_overall_recall": 0.9023668639053254,
+      "eval_runtime": 0.3239,
+      "eval_samples_per_second": 577.377,
+      "eval_steps_per_second": 9.263,
       "step": 2438
     },
     {
       "epoch": 24.0,
-      "grad_norm": 0.31136611104011536,
+      "grad_norm": 0.02887049876153469,
       "learning_rate": 3.8e-05,
-      "loss": 0.0021,
+      "loss": 0.0024,
       "step": 2544
     },
     {
       "epoch": 24.0,
-      "eval_LOCATION_f1": 0.8493150684931507,
+      "eval_LOCATION_f1": 0.8666666666666666,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8051948051948052,
-      "eval_LOCATION_recall": 0.8985507246376812,
-      "eval_ORGANIZATION_f1": 0.7874015748031497,
+      "eval_LOCATION_precision": 0.8024691358024691,
+      "eval_LOCATION_recall": 0.9420289855072463,
+      "eval_ORGANIZATION_f1": 0.7244094488188977,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.7246376811594203,
-      "eval_ORGANIZATION_recall": 0.8620689655172413,
-      "eval_PERSON_f1": 0.9067524115755626,
+      "eval_ORGANIZATION_precision": 0.6666666666666666,
+      "eval_ORGANIZATION_recall": 0.7931034482758621,
+      "eval_PERSON_f1": 0.9090909090909091,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8867924528301887,
-      "eval_PERSON_recall": 0.9276315789473685,
-      "eval_QUANTITY_f1": 0.8135593220338982,
+      "eval_PERSON_precision": 0.8974358974358975,
+      "eval_PERSON_recall": 0.9210526315789473,
+      "eval_QUANTITY_f1": 0.8125,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.8275862068965517,
-      "eval_QUANTITY_recall": 0.8,
-      "eval_TIME_f1": 0.8387096774193549,
+      "eval_QUANTITY_precision": 0.7647058823529411,
+      "eval_QUANTITY_recall": 0.8666666666666667,
+      "eval_TIME_f1": 0.8813559322033899,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.7878787878787878,
+      "eval_TIME_precision": 0.8666666666666667,
       "eval_TIME_recall": 0.896551724137931,
-      "eval_loss": 0.27145740389823914,
-      "eval_overall_accuracy": 0.9654289372599232,
-      "eval_overall_f1": 0.8595744680851063,
-      "eval_overall_precision": 0.8256130790190735,
+      "eval_loss": 0.28270941972732544,
+      "eval_overall_accuracy": 0.9646606914212548,
+      "eval_overall_f1": 0.8559322033898306,
+      "eval_overall_precision": 0.8189189189189189,
       "eval_overall_recall": 0.8964497041420119,
-      "eval_runtime": 0.8132,
-      "eval_samples_per_second": 229.963,
-      "eval_steps_per_second": 3.689,
+      "eval_runtime": 0.3467,
+      "eval_samples_per_second": 539.396,
+      "eval_steps_per_second": 8.653,
       "step": 2544
     },
     {
       "epoch": 25.0,
-      "grad_norm": 0.07211441546678543,
+      "grad_norm": 0.0016468241810798645,
       "learning_rate": 3.7500000000000003e-05,
-      "loss": 0.0022,
+      "loss": 0.0024,
       "step": 2650
     },
     {
       "epoch": 25.0,
-      "eval_LOCATION_f1": 0.8551724137931034,
+      "eval_LOCATION_f1": 0.8648648648648648,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8157894736842105,
-      "eval_LOCATION_recall": 0.8985507246376812,
-      "eval_ORGANIZATION_f1": 0.7500000000000001,
+      "eval_LOCATION_precision": 0.810126582278481,
+      "eval_LOCATION_recall": 0.927536231884058,
+      "eval_ORGANIZATION_f1": 0.7681159420289855,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.7258064516129032,
-      "eval_ORGANIZATION_recall": 0.7758620689655172,
-      "eval_PERSON_f1": 0.9084967320261438,
+      "eval_ORGANIZATION_precision": 0.6625,
+      "eval_ORGANIZATION_recall": 0.9137931034482759,
+      "eval_PERSON_f1": 0.8980891719745223,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.9025974025974026,
-      "eval_PERSON_recall": 0.9144736842105263,
-      "eval_QUANTITY_f1": 0.8064516129032259,
+      "eval_PERSON_precision": 0.8703703703703703,
+      "eval_PERSON_recall": 0.9276315789473685,
+      "eval_QUANTITY_f1": 0.8125,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.78125,
-      "eval_QUANTITY_recall": 0.8333333333333334,
-      "eval_TIME_f1": 0.8387096774193549,
+      "eval_QUANTITY_precision": 0.7647058823529411,
+      "eval_QUANTITY_recall": 0.8666666666666667,
+      "eval_TIME_f1": 0.896551724137931,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.7878787878787878,
+      "eval_TIME_precision": 0.896551724137931,
       "eval_TIME_recall": 0.896551724137931,
-      "eval_loss": 0.26712566614151,
-      "eval_overall_accuracy": 0.9638924455825865,
-      "eval_overall_f1": 0.8546762589928059,
-      "eval_overall_precision": 0.8319327731092437,
-      "eval_overall_recall": 0.878698224852071,
-      "eval_runtime": 0.8059,
-      "eval_samples_per_second": 232.043,
-      "eval_steps_per_second": 3.723,
+      "eval_loss": 0.28754085302352905,
+      "eval_overall_accuracy": 0.9646606914212548,
+      "eval_overall_f1": 0.8587257617728532,
+      "eval_overall_precision": 0.8072916666666666,
+      "eval_overall_recall": 0.9171597633136095,
+      "eval_runtime": 0.3225,
+      "eval_samples_per_second": 579.871,
+      "eval_steps_per_second": 9.303,
       "step": 2650
     },
     {
       "epoch": 26.0,
-      "grad_norm": 6.344851970672607,
+      "grad_norm": 0.22259019315242767,
       "learning_rate": 3.7e-05,
-      "loss": 0.0021,
+      "loss": 0.0025,
       "step": 2756
     },
     {
       "epoch": 26.0,
-      "eval_LOCATION_f1": 0.8311688311688311,
+      "eval_LOCATION_f1": 0.8378378378378378,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.7529411764705882,
-      "eval_LOCATION_recall": 0.927536231884058,
-      "eval_ORGANIZATION_f1": 0.7419354838709677,
+      "eval_LOCATION_precision": 0.7848101265822784,
+      "eval_LOCATION_recall": 0.8985507246376812,
+      "eval_ORGANIZATION_f1": 0.728813559322034,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.696969696969697,
-      "eval_ORGANIZATION_recall": 0.7931034482758621,
-      "eval_PERSON_f1": 0.9025974025974026,
+      "eval_ORGANIZATION_precision": 0.7166666666666667,
+      "eval_ORGANIZATION_recall": 0.7413793103448276,
+      "eval_PERSON_f1": 0.8924050632911392,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8910256410256411,
-      "eval_PERSON_recall": 0.9144736842105263,
-      "eval_QUANTITY_f1": 0.8333333333333334,
+      "eval_PERSON_precision": 0.8597560975609756,
+      "eval_PERSON_recall": 0.9276315789473685,
+      "eval_QUANTITY_f1": 0.8387096774193549,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.8333333333333334,
-      "eval_QUANTITY_recall": 0.8333333333333334,
-      "eval_TIME_f1": 0.8524590163934426,
+      "eval_QUANTITY_precision": 0.8125,
+      "eval_QUANTITY_recall": 0.8666666666666667,
+      "eval_TIME_f1": 0.8813559322033899,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.8125,
+      "eval_TIME_precision": 0.8666666666666667,
       "eval_TIME_recall": 0.896551724137931,
-      "eval_loss": 0.30937761068344116,
-      "eval_overall_accuracy": 0.9623559539052496,
-      "eval_overall_f1": 0.8486562942008485,
-      "eval_overall_precision": 0.8130081300813008,
-      "eval_overall_recall": 0.8875739644970414,
-      "eval_runtime": 0.8324,
-      "eval_samples_per_second": 224.639,
-      "eval_steps_per_second": 3.604,
+      "eval_loss": 0.2877175509929657,
+      "eval_overall_accuracy": 0.9641485275288092,
+      "eval_overall_f1": 0.8477951635846374,
+      "eval_overall_precision": 0.8164383561643835,
+      "eval_overall_recall": 0.8816568047337278,
+      "eval_runtime": 0.3224,
+      "eval_samples_per_second": 580.076,
+      "eval_steps_per_second": 9.306,
       "step": 2756
     },
     {
       "epoch": 27.0,
-      "grad_norm": 0.9656301736831665,
+      "grad_norm": 0.013442899100482464,
       "learning_rate": 3.65e-05,
-      "loss": 0.0033,
+      "loss": 0.0038,
       "step": 2862
     },
     {
       "epoch": 27.0,
-      "eval_LOCATION_f1": 0.8299319727891156,
+      "eval_LOCATION_f1": 0.8289473684210527,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.782051282051282,
-      "eval_LOCATION_recall": 0.8840579710144928,
-      "eval_ORGANIZATION_f1": 0.7603305785123967,
+      "eval_LOCATION_precision": 0.7590361445783133,
+      "eval_LOCATION_recall": 0.9130434782608695,
+      "eval_ORGANIZATION_f1": 0.6825396825396826,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.7301587301587301,
-      "eval_ORGANIZATION_recall": 0.7931034482758621,
-      "eval_PERSON_f1": 0.9137380191693291,
+      "eval_ORGANIZATION_precision": 0.6323529411764706,
+      "eval_ORGANIZATION_recall": 0.7413793103448276,
+      "eval_PERSON_f1": 0.9196141479099678,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8881987577639752,
+      "eval_PERSON_precision": 0.89937106918239,
       "eval_PERSON_recall": 0.9407894736842105,
-      "eval_QUANTITY_f1": 0.8387096774193549,
+      "eval_QUANTITY_f1": 0.8,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.8125,
+      "eval_QUANTITY_precision": 0.7428571428571429,
       "eval_QUANTITY_recall": 0.8666666666666667,
-      "eval_TIME_f1": 0.8253968253968255,
+      "eval_TIME_f1": 0.8387096774193549,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.7647058823529411,
+      "eval_TIME_precision": 0.7878787878787878,
       "eval_TIME_recall": 0.896551724137931,
-      "eval_loss": 0.2710714638233185,
-      "eval_overall_accuracy": 0.9651728553137003,
-      "eval_overall_f1": 0.8555240793201133,
-      "eval_overall_precision": 0.8206521739130435,
-      "eval_overall_recall": 0.893491124260355,
-      "eval_runtime": 0.802,
-      "eval_samples_per_second": 233.165,
-      "eval_steps_per_second": 3.741,
+      "eval_loss": 0.29557737708091736,
+      "eval_overall_accuracy": 0.9610755441741358,
+      "eval_overall_f1": 0.840782122905028,
+      "eval_overall_precision": 0.7962962962962963,
+      "eval_overall_recall": 0.8905325443786982,
+      "eval_runtime": 0.3217,
+      "eval_samples_per_second": 581.31,
+      "eval_steps_per_second": 9.326,
       "step": 2862
     },
     {
       "epoch": 28.0,
-      "grad_norm": 0.023881729692220688,
+      "grad_norm": 0.033651720732450485,
       "learning_rate": 3.6e-05,
-      "loss": 0.003,
+      "loss": 0.0046,
       "step": 2968
     },
     {
       "epoch": 28.0,
-      "eval_LOCATION_f1": 0.8344370860927153,
+      "eval_LOCATION_f1": 0.8609271523178808,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.7682926829268293,
-      "eval_LOCATION_recall": 0.9130434782608695,
-      "eval_ORGANIZATION_f1": 0.7804878048780489,
+      "eval_LOCATION_precision": 0.7926829268292683,
+      "eval_LOCATION_recall": 0.9420289855072463,
+      "eval_ORGANIZATION_f1": 0.7559055118110236,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.7384615384615385,
+      "eval_ORGANIZATION_precision": 0.6956521739130435,
       "eval_ORGANIZATION_recall": 0.8275862068965517,
-      "eval_PERSON_f1": 0.9161290322580645,
+      "eval_PERSON_f1": 0.9044585987261147,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8987341772151899,
+      "eval_PERSON_precision": 0.8765432098765432,
       "eval_PERSON_recall": 0.9342105263157895,
-      "eval_QUANTITY_f1": 0.78125,
+      "eval_QUANTITY_f1": 0.8253968253968254,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.7352941176470589,
-      "eval_QUANTITY_recall": 0.8333333333333334,
-      "eval_TIME_f1": 0.8387096774193549,
+      "eval_QUANTITY_precision": 0.7878787878787878,
+      "eval_QUANTITY_recall": 0.8666666666666667,
+      "eval_TIME_f1": 0.8813559322033899,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.7878787878787878,
+      "eval_TIME_precision": 0.8666666666666667,
       "eval_TIME_recall": 0.896551724137931,
-      "eval_loss": 0.2883354723453522,
-      "eval_overall_accuracy": 0.964404609475032,
-      "eval_overall_f1": 0.8563380281690142,
-      "eval_overall_precision": 0.8172043010752689,
-      "eval_overall_recall": 0.8994082840236687,
-      "eval_runtime": 0.8155,
-      "eval_samples_per_second": 229.295,
-      "eval_steps_per_second": 3.679,
+      "eval_loss": 0.2740738093852997,
+      "eval_overall_accuracy": 0.9649167733674776,
+      "eval_overall_f1": 0.8599439775910365,
+      "eval_overall_precision": 0.8164893617021277,
+      "eval_overall_recall": 0.908284023668639,
+      "eval_runtime": 0.3277,
+      "eval_samples_per_second": 570.688,
+      "eval_steps_per_second": 9.155,
       "step": 2968
     },
     {
       "epoch": 29.0,
-      "grad_norm": 0.004700132180005312,
+      "grad_norm": 0.0036596679128706455,
       "learning_rate": 3.55e-05,
-      "loss": 0.0021,
+      "loss": 0.0022,
       "step": 3074
     },
     {
       "epoch": 29.0,
-      "eval_LOCATION_f1": 0.8571428571428572,
+      "eval_LOCATION_f1": 0.8689655172413793,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8076923076923077,
+      "eval_LOCATION_precision": 0.8289473684210527,
       "eval_LOCATION_recall": 0.9130434782608695,
-      "eval_ORGANIZATION_f1": 0.7317073170731708,
+      "eval_ORGANIZATION_f1": 0.7936507936507937,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.6923076923076923,
-      "eval_ORGANIZATION_recall": 0.7758620689655172,
-      "eval_PERSON_f1": 0.8996763754045308,
+      "eval_ORGANIZATION_precision": 0.7352941176470589,
+      "eval_ORGANIZATION_recall": 0.8620689655172413,
+      "eval_PERSON_f1": 0.9126213592233009,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8853503184713376,
-      "eval_PERSON_recall": 0.9144736842105263,
-      "eval_QUANTITY_f1": 0.8387096774193549,
+      "eval_PERSON_precision": 0.8980891719745223,
+      "eval_PERSON_recall": 0.9276315789473685,
+      "eval_QUANTITY_f1": 0.8253968253968254,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.8125,
+      "eval_QUANTITY_precision": 0.7878787878787878,
       "eval_QUANTITY_recall": 0.8666666666666667,
-      "eval_TIME_f1": 0.8666666666666666,
+      "eval_TIME_f1": 0.896551724137931,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.8387096774193549,
+      "eval_TIME_precision": 0.896551724137931,
       "eval_TIME_recall": 0.896551724137931,
-      "eval_loss": 0.2531093657016754,
-      "eval_overall_accuracy": 0.9651728553137003,
-      "eval_overall_f1": 0.8530670470756063,
-      "eval_overall_precision": 0.8236914600550964,
-      "eval_overall_recall": 0.8846153846153846,
-      "eval_runtime": 0.8176,
-      "eval_samples_per_second": 228.73,
-      "eval_steps_per_second": 3.669,
+      "eval_loss": 0.280216246843338,
+      "eval_overall_accuracy": 0.9667093469910372,
+      "eval_overall_f1": 0.8730385164051355,
+      "eval_overall_precision": 0.8429752066115702,
+      "eval_overall_recall": 0.9053254437869822,
+      "eval_runtime": 0.3236,
+      "eval_samples_per_second": 577.801,
+      "eval_steps_per_second": 9.27,
       "step": 3074
     },
     {
       "epoch": 30.0,
-      "grad_norm": 0.1790827363729477,
+      "grad_norm": 0.09640705585479736,
       "learning_rate": 3.5e-05,
-      "loss": 0.0016,
+      "loss": 0.002,
       "step": 3180
     },
     {
       "epoch": 30.0,
-      "eval_LOCATION_f1": 0.8289473684210527,
+      "eval_LOCATION_f1": 0.8590604026845639,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.7590361445783133,
-      "eval_LOCATION_recall": 0.9130434782608695,
-      "eval_ORGANIZATION_f1": 0.6434782608695653,
+      "eval_LOCATION_precision": 0.8,
+      "eval_LOCATION_recall": 0.927536231884058,
+      "eval_ORGANIZATION_f1": 0.7903225806451614,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.6491228070175439,
-      "eval_ORGANIZATION_recall": 0.6379310344827587,
-      "eval_PERSON_f1": 0.910828025477707,
+      "eval_ORGANIZATION_precision": 0.7424242424242424,
+      "eval_ORGANIZATION_recall": 0.8448275862068966,
+      "eval_PERSON_f1": 0.9038461538461539,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8827160493827161,
-      "eval_PERSON_recall": 0.9407894736842105,
+      "eval_PERSON_precision": 0.88125,
+      "eval_PERSON_recall": 0.9276315789473685,
       "eval_QUANTITY_f1": 0.8387096774193549,
       "eval_QUANTITY_number": 30,
       "eval_QUANTITY_precision": 0.8125,
       "eval_QUANTITY_recall": 0.8666666666666667,
-      "eval_TIME_f1": 0.8253968253968255,
+      "eval_TIME_f1": 0.896551724137931,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.7647058823529411,
+      "eval_TIME_precision": 0.896551724137931,
       "eval_TIME_recall": 0.896551724137931,
-      "eval_loss": 0.29236629605293274,
-      "eval_overall_accuracy": 0.9626120358514725,
-      "eval_overall_f1": 0.8356940509915014,
-      "eval_overall_precision": 0.8016304347826086,
-      "eval_overall_recall": 0.8727810650887574,
-      "eval_runtime": 0.8083,
-      "eval_samples_per_second": 231.339,
-      "eval_steps_per_second": 3.711,
+      "eval_loss": 0.29095637798309326,
+      "eval_overall_accuracy": 0.9674775928297055,
+      "eval_overall_f1": 0.8680851063829786,
+      "eval_overall_precision": 0.8337874659400545,
+      "eval_overall_recall": 0.9053254437869822,
+      "eval_runtime": 0.3288,
+      "eval_samples_per_second": 568.665,
+      "eval_steps_per_second": 9.123,
       "step": 3180
     },
     {
       "epoch": 31.0,
-      "grad_norm": 0.0017360023921355605,
+      "grad_norm": 0.002530389465391636,
       "learning_rate": 3.45e-05,
-      "loss": 0.0048,
+      "loss": 0.0015,
       "step": 3286
     },
     {
       "epoch": 31.0,
-      "eval_LOCATION_f1": 0.8724832214765101,
+      "eval_LOCATION_f1": 0.8571428571428572,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8125,
-      "eval_LOCATION_recall": 0.9420289855072463,
-      "eval_ORGANIZATION_f1": 0.6885245901639345,
+      "eval_LOCATION_precision": 0.8076923076923077,
+      "eval_LOCATION_recall": 0.9130434782608695,
+      "eval_ORGANIZATION_f1": 0.768,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.65625,
-      "eval_ORGANIZATION_recall": 0.7241379310344828,
-      "eval_PERSON_f1": 0.9003215434083601,
+      "eval_ORGANIZATION_precision": 0.7164179104477612,
+      "eval_ORGANIZATION_recall": 0.8275862068965517,
+      "eval_PERSON_f1": 0.9049180327868853,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8805031446540881,
-      "eval_PERSON_recall": 0.9210526315789473,
-      "eval_QUANTITY_f1": 0.8253968253968254,
+      "eval_PERSON_precision": 0.9019607843137255,
+      "eval_PERSON_recall": 0.9078947368421053,
+      "eval_QUANTITY_f1": 0.8387096774193549,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.7878787878787878,
+      "eval_QUANTITY_precision": 0.8125,
       "eval_QUANTITY_recall": 0.8666666666666667,
-      "eval_TIME_f1": 0.8253968253968255,
+      "eval_TIME_f1": 0.896551724137931,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.7647058823529411,
+      "eval_TIME_precision": 0.896551724137931,
       "eval_TIME_recall": 0.896551724137931,
-      "eval_loss": 0.2867705523967743,
-      "eval_overall_accuracy": 0.9626120358514725,
-      "eval_overall_f1": 0.8446327683615819,
-      "eval_overall_precision": 0.8081081081081081,
-      "eval_overall_recall": 0.8846153846153846,
-      "eval_runtime": 0.8033,
-      "eval_samples_per_second": 232.803,
-      "eval_steps_per_second": 3.735,
+      "eval_loss": 0.29163551330566406,
+      "eval_overall_accuracy": 0.9661971830985916,
+      "eval_overall_f1": 0.8637015781922526,
+      "eval_overall_precision": 0.8384401114206128,
+      "eval_overall_recall": 0.8905325443786982,
+      "eval_runtime": 0.3204,
+      "eval_samples_per_second": 583.719,
+      "eval_steps_per_second": 9.364,
       "step": 3286
     },
     {
       "epoch": 32.0,
-      "grad_norm": 0.009486394003033638,
+      "grad_norm": 0.03596256300806999,
       "learning_rate": 3.4000000000000007e-05,
-      "loss": 0.0032,
+      "loss": 0.004,
       "step": 3392
     },
     {
       "epoch": 32.0,
-      "eval_LOCATION_f1": 0.8243243243243245,
+      "eval_LOCATION_f1": 0.881118881118881,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.7721518987341772,
-      "eval_LOCATION_recall": 0.8840579710144928,
-      "eval_ORGANIZATION_f1": 0.7142857142857142,
+      "eval_LOCATION_precision": 0.8513513513513513,
+      "eval_LOCATION_recall": 0.9130434782608695,
+      "eval_ORGANIZATION_f1": 0.7704918032786885,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.6617647058823529,
-      "eval_ORGANIZATION_recall": 0.7758620689655172,
-      "eval_PERSON_f1": 0.8974358974358975,
+      "eval_ORGANIZATION_precision": 0.734375,
+      "eval_ORGANIZATION_recall": 0.8103448275862069,
+      "eval_PERSON_f1": 0.9003215434083601,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.875,
+      "eval_PERSON_precision": 0.8805031446540881,
       "eval_PERSON_recall": 0.9210526315789473,
-      "eval_QUANTITY_f1": 0.7878787878787877,
+      "eval_QUANTITY_f1": 0.8387096774193549,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.7222222222222222,
+      "eval_QUANTITY_precision": 0.8125,
       "eval_QUANTITY_recall": 0.8666666666666667,
-      "eval_TIME_f1": 0.8125,
+      "eval_TIME_f1": 0.8135593220338982,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.7428571428571429,
-      "eval_TIME_recall": 0.896551724137931,
-      "eval_loss": 0.2836323380470276,
-      "eval_overall_accuracy": 0.9605633802816902,
-      "eval_overall_f1": 0.8324022346368716,
-      "eval_overall_precision": 0.7883597883597884,
-      "eval_overall_recall": 0.8816568047337278,
-      "eval_runtime": 0.8087,
-      "eval_samples_per_second": 231.245,
-      "eval_steps_per_second": 3.71,
+      "eval_TIME_precision": 0.8,
+      "eval_TIME_recall": 0.8275862068965517,
+      "eval_loss": 0.285200834274292,
+      "eval_overall_accuracy": 0.9649167733674776,
+      "eval_overall_f1": 0.860832137733142,
+      "eval_overall_precision": 0.8356545961002786,
+      "eval_overall_recall": 0.8875739644970414,
+      "eval_runtime": 0.333,
+      "eval_samples_per_second": 561.561,
+      "eval_steps_per_second": 9.009,
       "step": 3392
     },
     {
       "epoch": 33.0,
-      "grad_norm": 0.008940489031374454,
+      "grad_norm": 0.008540854789316654,
       "learning_rate": 3.35e-05,
-      "loss": 0.004,
+      "loss": 0.0038,
       "step": 3498
     },
     {
       "epoch": 33.0,
-      "eval_LOCATION_f1": 0.8611111111111112,
+      "eval_LOCATION_f1": 0.863013698630137,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8266666666666667,
-      "eval_LOCATION_recall": 0.8985507246376812,
-      "eval_ORGANIZATION_f1": 0.735042735042735,
+      "eval_LOCATION_precision": 0.8181818181818182,
+      "eval_LOCATION_recall": 0.9130434782608695,
+      "eval_ORGANIZATION_f1": 0.8032786885245901,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.7288135593220338,
-      "eval_ORGANIZATION_recall": 0.7413793103448276,
-      "eval_PERSON_f1": 0.896774193548387,
+      "eval_ORGANIZATION_precision": 0.765625,
+      "eval_ORGANIZATION_recall": 0.8448275862068966,
+      "eval_PERSON_f1": 0.9015873015873016,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.879746835443038,
-      "eval_PERSON_recall": 0.9144736842105263,
-      "eval_QUANTITY_f1": 0.8253968253968254,
+      "eval_PERSON_precision": 0.8711656441717791,
+      "eval_PERSON_recall": 0.9342105263157895,
+      "eval_QUANTITY_f1": 0.8387096774193549,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.7878787878787878,
+      "eval_QUANTITY_precision": 0.8125,
       "eval_QUANTITY_recall": 0.8666666666666667,
-      "eval_TIME_f1": 0.8253968253968255,
+      "eval_TIME_f1": 0.8196721311475409,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.7647058823529411,
-      "eval_TIME_recall": 0.896551724137931,
-      "eval_loss": 0.25650379061698914,
-      "eval_overall_accuracy": 0.9656850192061459,
-      "eval_overall_f1": 0.8493543758967002,
-      "eval_overall_precision": 0.8245125348189415,
-      "eval_overall_recall": 0.8757396449704142,
-      "eval_runtime": 0.8037,
-      "eval_samples_per_second": 232.687,
-      "eval_steps_per_second": 3.733,
+      "eval_TIME_precision": 0.78125,
+      "eval_TIME_recall": 0.8620689655172413,
+      "eval_loss": 0.28712666034698486,
+      "eval_overall_accuracy": 0.9649167733674776,
+      "eval_overall_f1": 0.8640226628895185,
+      "eval_overall_precision": 0.8288043478260869,
+      "eval_overall_recall": 0.9023668639053254,
+      "eval_runtime": 0.3198,
+      "eval_samples_per_second": 584.775,
+      "eval_steps_per_second": 9.381,
       "step": 3498
     },
     {
       "epoch": 34.0,
-      "grad_norm": 0.018895745277404785,
+      "grad_norm": 0.0038193664513528347,
       "learning_rate": 3.3e-05,
-      "loss": 0.0016,
+      "loss": 0.0042,
       "step": 3604
     },
     {
       "epoch": 34.0,
-      "eval_LOCATION_f1": 0.8366013071895425,
+      "eval_LOCATION_f1": 0.8749999999999999,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.7619047619047619,
-      "eval_LOCATION_recall": 0.927536231884058,
-      "eval_ORGANIZATION_f1": 0.6896551724137931,
+      "eval_LOCATION_precision": 0.84,
+      "eval_LOCATION_recall": 0.9130434782608695,
+      "eval_ORGANIZATION_f1": 0.7272727272727273,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.6896551724137931,
-      "eval_ORGANIZATION_recall": 0.6896551724137931,
-      "eval_PERSON_f1": 0.9015873015873016,
+      "eval_ORGANIZATION_precision": 0.6486486486486487,
+      "eval_ORGANIZATION_recall": 0.8275862068965517,
+      "eval_PERSON_f1": 0.8996763754045308,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8711656441717791,
-      "eval_PERSON_recall": 0.9342105263157895,
+      "eval_PERSON_precision": 0.8853503184713376,
+      "eval_PERSON_recall": 0.9144736842105263,
       "eval_QUANTITY_f1": 0.8387096774193549,
       "eval_QUANTITY_number": 30,
       "eval_QUANTITY_precision": 0.8125,
       "eval_QUANTITY_recall": 0.8666666666666667,
-      "eval_TIME_f1": 0.8524590163934426,
+      "eval_TIME_f1": 0.8666666666666666,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.8125,
+      "eval_TIME_precision": 0.8387096774193549,
       "eval_TIME_recall": 0.896551724137931,
-      "eval_loss": 0.2746643126010895,
-      "eval_overall_accuracy": 0.9651728553137003,
-      "eval_overall_f1": 0.8429985855728428,
-      "eval_overall_precision": 0.8075880758807588,
-      "eval_overall_recall": 0.8816568047337278,
-      "eval_runtime": 0.8043,
-      "eval_samples_per_second": 232.493,
-      "eval_steps_per_second": 3.73,
+      "eval_loss": 0.29013821482658386,
+      "eval_overall_accuracy": 0.9623559539052496,
+      "eval_overall_f1": 0.8543140028288543,
+      "eval_overall_precision": 0.8184281842818428,
+      "eval_overall_recall": 0.893491124260355,
+      "eval_runtime": 0.3361,
+      "eval_samples_per_second": 556.385,
+      "eval_steps_per_second": 8.926,
       "step": 3604
     },
     {
       "epoch": 35.0,
-      "grad_norm": 0.0035857518669217825,
+      "grad_norm": 0.014040589332580566,
       "learning_rate": 3.2500000000000004e-05,
-      "loss": 0.0018,
+      "loss": 0.0026,
       "step": 3710
     },
     {
       "epoch": 35.0,
-      "eval_LOCATION_f1": 0.8590604026845639,
+      "eval_LOCATION_f1": 0.8533333333333333,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8,
+      "eval_LOCATION_precision": 0.7901234567901234,
       "eval_LOCATION_recall": 0.927536231884058,
-      "eval_ORGANIZATION_f1": 0.7479674796747967,
+      "eval_ORGANIZATION_f1": 0.7936507936507937,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.7076923076923077,
-      "eval_ORGANIZATION_recall": 0.7931034482758621,
-      "eval_PERSON_f1": 0.9096774193548387,
+      "eval_ORGANIZATION_precision": 0.7352941176470589,
+      "eval_ORGANIZATION_recall": 0.8620689655172413,
+      "eval_PERSON_f1": 0.9161290322580645,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8924050632911392,
-      "eval_PERSON_recall": 0.9276315789473685,
-      "eval_QUANTITY_f1": 0.870967741935484,
+      "eval_PERSON_precision": 0.8987341772151899,
+      "eval_PERSON_recall": 0.9342105263157895,
+      "eval_QUANTITY_f1": 0.8387096774193549,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.84375,
-      "eval_QUANTITY_recall": 0.9,
-      "eval_TIME_f1": 0.8387096774193549,
+      "eval_QUANTITY_precision": 0.8125,
+      "eval_QUANTITY_recall": 0.8666666666666667,
+      "eval_TIME_f1": 0.8666666666666666,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.7878787878787878,
+      "eval_TIME_precision": 0.8387096774193549,
       "eval_TIME_recall": 0.896551724137931,
-      "eval_loss": 0.26389142870903015,
+      "eval_loss": 0.27812227606773376,
       "eval_overall_accuracy": 0.9659411011523688,
-      "eval_overall_f1": 0.8611898016997167,
-      "eval_overall_precision": 0.8260869565217391,
-      "eval_overall_recall": 0.8994082840236687,
-      "eval_runtime": 0.822,
-      "eval_samples_per_second": 227.49,
-      "eval_steps_per_second": 3.65,
+      "eval_overall_f1": 0.8700564971751412,
+      "eval_overall_precision": 0.8324324324324325,
+      "eval_overall_recall": 0.9112426035502958,
+      "eval_runtime": 0.3232,
+      "eval_samples_per_second": 578.539,
+      "eval_steps_per_second": 9.281,
       "step": 3710
     },
     {
       "epoch": 36.0,
-      "grad_norm": 0.0013648144667968154,
+      "grad_norm": 0.0036135606933385134,
       "learning_rate": 3.2000000000000005e-05,
-      "loss": 0.0032,
+      "loss": 0.0021,
       "step": 3816
     },
     {
       "epoch": 36.0,
-      "eval_LOCATION_f1": 0.8904109589041096,
+      "eval_LOCATION_f1": 0.8551724137931034,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8441558441558441,
-      "eval_LOCATION_recall": 0.9420289855072463,
-      "eval_ORGANIZATION_f1": 0.717948717948718,
+      "eval_LOCATION_precision": 0.8157894736842105,
+      "eval_LOCATION_recall": 0.8985507246376812,
+      "eval_ORGANIZATION_f1": 0.8091603053435115,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.711864406779661,
-      "eval_ORGANIZATION_recall": 0.7241379310344828,
-      "eval_PERSON_f1": 0.919093851132686,
+      "eval_ORGANIZATION_precision": 0.726027397260274,
+      "eval_ORGANIZATION_recall": 0.9137931034482759,
+      "eval_PERSON_f1": 0.9032258064516129,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.9044585987261147,
-      "eval_PERSON_recall": 0.9342105263157895,
-      "eval_QUANTITY_f1": 0.8064516129032259,
+      "eval_PERSON_precision": 0.8860759493670886,
+      "eval_PERSON_recall": 0.9210526315789473,
+      "eval_QUANTITY_f1": 0.8253968253968254,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.78125,
-      "eval_QUANTITY_recall": 0.8333333333333334,
-      "eval_TIME_f1": 0.8253968253968255,
+      "eval_QUANTITY_precision": 0.7878787878787878,
+      "eval_QUANTITY_recall": 0.8666666666666667,
+      "eval_TIME_f1": 0.8771929824561403,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.7647058823529411,
-      "eval_TIME_recall": 0.896551724137931,
-      "eval_loss": 0.2595536410808563,
-      "eval_overall_accuracy": 0.9654289372599232,
-      "eval_overall_f1": 0.860832137733142,
-      "eval_overall_precision": 0.8356545961002786,
-      "eval_overall_recall": 0.8875739644970414,
-      "eval_runtime": 0.8156,
-      "eval_samples_per_second": 229.278,
-      "eval_steps_per_second": 3.678,
+      "eval_TIME_precision": 0.8928571428571429,
+      "eval_TIME_recall": 0.8620689655172413,
+      "eval_loss": 0.29289114475250244,
+      "eval_overall_accuracy": 0.9641485275288092,
+      "eval_overall_f1": 0.8668555240793201,
+      "eval_overall_precision": 0.8315217391304348,
+      "eval_overall_recall": 0.9053254437869822,
+      "eval_runtime": 0.3214,
+      "eval_samples_per_second": 581.814,
+      "eval_steps_per_second": 9.334,
       "step": 3816
     },
     {
       "epoch": 37.0,
-      "grad_norm": 0.001226762542501092,
+      "grad_norm": 0.047017503529787064,
       "learning_rate": 3.15e-05,
-      "loss": 0.0024,
+      "loss": 0.002,
       "step": 3922
     },
     {
       "epoch": 37.0,
-      "eval_LOCATION_f1": 0.882758620689655,
+      "eval_LOCATION_f1": 0.84,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8421052631578947,
-      "eval_LOCATION_recall": 0.927536231884058,
-      "eval_ORGANIZATION_f1": 0.728813559322034,
+      "eval_LOCATION_precision": 0.7777777777777778,
+      "eval_LOCATION_recall": 0.9130434782608695,
+      "eval_ORGANIZATION_f1": 0.7967479674796748,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.7166666666666667,
-      "eval_ORGANIZATION_recall": 0.7413793103448276,
-      "eval_PERSON_f1": 0.9102564102564101,
+      "eval_ORGANIZATION_precision": 0.7538461538461538,
+      "eval_ORGANIZATION_recall": 0.8448275862068966,
+      "eval_PERSON_f1": 0.9078947368421053,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8875,
-      "eval_PERSON_recall": 0.9342105263157895,
-      "eval_QUANTITY_f1": 0.8125,
+      "eval_PERSON_precision": 0.9078947368421053,
+      "eval_PERSON_recall": 0.9078947368421053,
+      "eval_QUANTITY_f1": 0.8387096774193549,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.7647058823529411,
+      "eval_QUANTITY_precision": 0.8125,
       "eval_QUANTITY_recall": 0.8666666666666667,
-      "eval_TIME_f1": 0.847457627118644,
+      "eval_TIME_f1": 0.8333333333333334,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.8333333333333334,
+      "eval_TIME_precision": 0.8064516129032258,
       "eval_TIME_recall": 0.8620689655172413,
-      "eval_loss": 0.2782364785671234,
-      "eval_overall_accuracy": 0.9669654289372599,
-      "eval_overall_f1": 0.8595988538681948,
-      "eval_overall_precision": 0.8333333333333334,
-      "eval_overall_recall": 0.8875739644970414,
-      "eval_runtime": 0.8106,
-      "eval_samples_per_second": 230.693,
-      "eval_steps_per_second": 3.701,
+      "eval_loss": 0.2750895619392395,
+      "eval_overall_accuracy": 0.9641485275288092,
+      "eval_overall_f1": 0.8612303290414879,
+      "eval_overall_precision": 0.8337950138504155,
+      "eval_overall_recall": 0.8905325443786982,
+      "eval_runtime": 0.3236,
+      "eval_samples_per_second": 577.827,
+      "eval_steps_per_second": 9.27,
       "step": 3922
     },
     {
       "epoch": 38.0,
-      "grad_norm": 0.0014669048832729459,
+      "grad_norm": 0.002975118113681674,
       "learning_rate": 3.1e-05,
-      "loss": 0.002,
+      "loss": 0.0032,
       "step": 4028
     },
     {
       "epoch": 38.0,
-      "eval_LOCATION_f1": 0.8590604026845639,
+      "eval_LOCATION_f1": 0.8344370860927153,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8,
-      "eval_LOCATION_recall": 0.927536231884058,
-      "eval_ORGANIZATION_f1": 0.7704918032786885,
+      "eval_LOCATION_precision": 0.7682926829268293,
+      "eval_LOCATION_recall": 0.9130434782608695,
+      "eval_ORGANIZATION_f1": 0.7933884297520662,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.734375,
-      "eval_ORGANIZATION_recall": 0.8103448275862069,
-      "eval_PERSON_f1": 0.9084967320261438,
+      "eval_ORGANIZATION_precision": 0.7619047619047619,
+      "eval_ORGANIZATION_recall": 0.8275862068965517,
+      "eval_PERSON_f1": 0.9096774193548387,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.9025974025974026,
-      "eval_PERSON_recall": 0.9144736842105263,
-      "eval_QUANTITY_f1": 0.8064516129032259,
+      "eval_PERSON_precision": 0.8924050632911392,
+      "eval_PERSON_recall": 0.9276315789473685,
+      "eval_QUANTITY_f1": 0.8253968253968254,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.78125,
-      "eval_QUANTITY_recall": 0.8333333333333334,
-      "eval_TIME_f1": 0.8196721311475409,
+      "eval_QUANTITY_precision": 0.7878787878787878,
+      "eval_QUANTITY_recall": 0.8666666666666667,
+      "eval_TIME_f1": 0.847457627118644,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.78125,
+      "eval_TIME_precision": 0.8333333333333334,
       "eval_TIME_recall": 0.8620689655172413,
-      "eval_loss": 0.2879778742790222,
-      "eval_overall_accuracy": 0.9664532650448143,
-      "eval_overall_f1": 0.8571428571428571,
-      "eval_overall_precision": 0.8287292817679558,
-      "eval_overall_recall": 0.8875739644970414,
-      "eval_runtime": 0.8881,
-      "eval_samples_per_second": 210.566,
-      "eval_steps_per_second": 3.378,
+      "eval_loss": 0.292982280254364,
+      "eval_overall_accuracy": 0.9669654289372599,
+      "eval_overall_f1": 0.8607954545454546,
+      "eval_overall_precision": 0.8278688524590164,
+      "eval_overall_recall": 0.8964497041420119,
+      "eval_runtime": 0.3205,
+      "eval_samples_per_second": 583.486,
+      "eval_steps_per_second": 9.361,
       "step": 4028
     },
     {
       "epoch": 39.0,
-      "grad_norm": 0.008377453312277794,
+      "grad_norm": 0.048705920577049255,
       "learning_rate": 3.05e-05,
-      "loss": 0.0027,
+      "loss": 0.0014,
       "step": 4134
     },
     {
       "epoch": 39.0,
-      "eval_LOCATION_f1": 0.8724832214765101,
+      "eval_LOCATION_f1": 0.888888888888889,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8125,
-      "eval_LOCATION_recall": 0.9420289855072463,
-      "eval_ORGANIZATION_f1": 0.71875,
+      "eval_LOCATION_precision": 0.8533333333333334,
+      "eval_LOCATION_recall": 0.927536231884058,
+      "eval_ORGANIZATION_f1": 0.7833333333333333,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.6571428571428571,
-      "eval_ORGANIZATION_recall": 0.7931034482758621,
-      "eval_PERSON_f1": 0.8924050632911392,
+      "eval_ORGANIZATION_precision": 0.7580645161290323,
+      "eval_ORGANIZATION_recall": 0.8103448275862069,
+      "eval_PERSON_f1": 0.9009584664536742,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8597560975609756,
+      "eval_PERSON_precision": 0.8757763975155279,
       "eval_PERSON_recall": 0.9276315789473685,
-      "eval_QUANTITY_f1": 0.8000000000000002,
+      "eval_QUANTITY_f1": 0.8387096774193549,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.8,
-      "eval_QUANTITY_recall": 0.8,
-      "eval_TIME_f1": 0.8524590163934426,
+      "eval_QUANTITY_precision": 0.8125,
+      "eval_QUANTITY_recall": 0.8666666666666667,
+      "eval_TIME_f1": 0.847457627118644,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.8125,
-      "eval_TIME_recall": 0.896551724137931,
-      "eval_loss": 0.2859266996383667,
-      "eval_overall_accuracy": 0.9646606914212548,
-      "eval_overall_f1": 0.84593837535014,
-      "eval_overall_precision": 0.8031914893617021,
-      "eval_overall_recall": 0.893491124260355,
-      "eval_runtime": 0.8126,
-      "eval_samples_per_second": 230.123,
-      "eval_steps_per_second": 3.692,
+      "eval_TIME_precision": 0.8333333333333334,
+      "eval_TIME_recall": 0.8620689655172413,
+      "eval_loss": 0.26422762870788574,
+      "eval_overall_accuracy": 0.9677336747759283,
+      "eval_overall_f1": 0.8681948424068767,
+      "eval_overall_precision": 0.8416666666666667,
+      "eval_overall_recall": 0.8964497041420119,
+      "eval_runtime": 0.3204,
+      "eval_samples_per_second": 583.697,
+      "eval_steps_per_second": 9.364,
       "step": 4134
     },
     {
       "epoch": 40.0,
-      "grad_norm": 6.265191078186035,
+      "grad_norm": 0.0028175306506454945,
       "learning_rate": 3e-05,
-      "loss": 0.0028,
+      "loss": 0.0022,
       "step": 4240
     },
     {
       "epoch": 40.0,
-      "eval_LOCATION_f1": 0.8648648648648648,
+      "eval_LOCATION_f1": 0.84,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.810126582278481,
-      "eval_LOCATION_recall": 0.927536231884058,
-      "eval_ORGANIZATION_f1": 0.7704918032786885,
+      "eval_LOCATION_precision": 0.7777777777777778,
+      "eval_LOCATION_recall": 0.9130434782608695,
+      "eval_ORGANIZATION_f1": 0.7479674796747967,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.734375,
-      "eval_ORGANIZATION_recall": 0.8103448275862069,
-      "eval_PERSON_f1": 0.9166666666666667,
+      "eval_ORGANIZATION_precision": 0.7076923076923077,
+      "eval_ORGANIZATION_recall": 0.7931034482758621,
+      "eval_PERSON_f1": 0.9032258064516129,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.89375,
-      "eval_PERSON_recall": 0.9407894736842105,
-      "eval_QUANTITY_f1": 0.8064516129032259,
+      "eval_PERSON_precision": 0.8860759493670886,
+      "eval_PERSON_recall": 0.9210526315789473,
+      "eval_QUANTITY_f1": 0.8387096774193549,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.78125,
-      "eval_QUANTITY_recall": 0.8333333333333334,
-      "eval_TIME_f1": 0.8387096774193549,
+      "eval_QUANTITY_precision": 0.8125,
+      "eval_QUANTITY_recall": 0.8666666666666667,
+      "eval_TIME_f1": 0.8620689655172413,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.7878787878787878,
-      "eval_TIME_recall": 0.896551724137931,
-      "eval_loss": 0.27767661213874817,
-      "eval_overall_accuracy": 0.9674775928297055,
-      "eval_overall_f1": 0.8640226628895185,
-      "eval_overall_precision": 0.8288043478260869,
-      "eval_overall_recall": 0.9023668639053254,
-      "eval_runtime": 0.8112,
-      "eval_samples_per_second": 230.533,
-      "eval_steps_per_second": 3.698,
+      "eval_TIME_precision": 0.8620689655172413,
+      "eval_TIME_recall": 0.8620689655172413,
+      "eval_loss": 0.30111193656921387,
+      "eval_overall_accuracy": 0.9661971830985916,
+      "eval_overall_f1": 0.8534850640113798,
+      "eval_overall_precision": 0.821917808219178,
+      "eval_overall_recall": 0.8875739644970414,
+      "eval_runtime": 0.3358,
+      "eval_samples_per_second": 556.909,
+      "eval_steps_per_second": 8.934,
       "step": 4240
     },
     {
       "epoch": 41.0,
-      "grad_norm": 0.01230549905449152,
+      "grad_norm": 0.017286457121372223,
       "learning_rate": 2.95e-05,
-      "loss": 0.0013,
+      "loss": 0.0023,
       "step": 4346
     },
     {
       "epoch": 41.0,
-      "eval_LOCATION_f1": 0.8344370860927153,
+      "eval_LOCATION_f1": 0.8689655172413793,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.7682926829268293,
+      "eval_LOCATION_precision": 0.8289473684210527,
       "eval_LOCATION_recall": 0.9130434782608695,
-      "eval_ORGANIZATION_f1": 0.7540983606557378,
+      "eval_ORGANIZATION_f1": 0.752136752136752,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.71875,
-      "eval_ORGANIZATION_recall": 0.7931034482758621,
-      "eval_PERSON_f1": 0.910828025477707,
+      "eval_ORGANIZATION_precision": 0.7457627118644068,
+      "eval_ORGANIZATION_recall": 0.7586206896551724,
+      "eval_PERSON_f1": 0.8974358974358975,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8827160493827161,
-      "eval_PERSON_recall": 0.9407894736842105,
+      "eval_PERSON_precision": 0.875,
+      "eval_PERSON_recall": 0.9210526315789473,
       "eval_QUANTITY_f1": 0.8387096774193549,
       "eval_QUANTITY_number": 30,
       "eval_QUANTITY_precision": 0.8125,
       "eval_QUANTITY_recall": 0.8666666666666667,
-      "eval_TIME_f1": 0.8709677419354839,
+      "eval_TIME_f1": 0.8620689655172413,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.8181818181818182,
-      "eval_TIME_recall": 0.9310344827586207,
-      "eval_loss": 0.27420660853385925,
-      "eval_overall_accuracy": 0.9667093469910372,
-      "eval_overall_f1": 0.8579465541490858,
-      "eval_overall_precision": 0.8176943699731903,
-      "eval_overall_recall": 0.9023668639053254,
-      "eval_runtime": 0.804,
-      "eval_samples_per_second": 232.579,
-      "eval_steps_per_second": 3.731,
+      "eval_TIME_precision": 0.8620689655172413,
+      "eval_TIME_recall": 0.8620689655172413,
+      "eval_loss": 0.29203903675079346,
+      "eval_overall_accuracy": 0.9659411011523688,
+      "eval_overall_f1": 0.8587896253602305,
+      "eval_overall_precision": 0.8370786516853933,
+      "eval_overall_recall": 0.8816568047337278,
+      "eval_runtime": 0.3197,
+      "eval_samples_per_second": 584.864,
+      "eval_steps_per_second": 9.383,
       "step": 4346
     },
     {
       "epoch": 42.0,
-      "grad_norm": 0.0334417000412941,
+      "grad_norm": 0.021444261074066162,
       "learning_rate": 2.9e-05,
-      "loss": 0.0035,
+      "loss": 0.0026,
       "step": 4452
     },
     {
       "epoch": 42.0,
-      "eval_LOCATION_f1": 0.8904109589041096,
+      "eval_LOCATION_f1": 0.8648648648648648,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8441558441558441,
-      "eval_LOCATION_recall": 0.9420289855072463,
-      "eval_ORGANIZATION_f1": 0.7603305785123967,
+      "eval_LOCATION_precision": 0.810126582278481,
+      "eval_LOCATION_recall": 0.927536231884058,
+      "eval_ORGANIZATION_f1": 0.7868852459016394,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.7301587301587301,
-      "eval_ORGANIZATION_recall": 0.7931034482758621,
-      "eval_PERSON_f1": 0.9050632911392406,
+      "eval_ORGANIZATION_precision": 0.75,
+      "eval_ORGANIZATION_recall": 0.8275862068965517,
+      "eval_PERSON_f1": 0.8974358974358975,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8719512195121951,
-      "eval_PERSON_recall": 0.9407894736842105,
-      "eval_QUANTITY_f1": 0.7936507936507938,
+      "eval_PERSON_precision": 0.875,
+      "eval_PERSON_recall": 0.9210526315789473,
+      "eval_QUANTITY_f1": 0.8387096774193549,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.7575757575757576,
-      "eval_QUANTITY_recall": 0.8333333333333334,
-      "eval_TIME_f1": 0.8571428571428571,
+      "eval_QUANTITY_precision": 0.8125,
+      "eval_QUANTITY_recall": 0.8666666666666667,
+      "eval_TIME_f1": 0.8333333333333334,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.7941176470588235,
-      "eval_TIME_recall": 0.9310344827586207,
-      "eval_loss": 0.28786924481391907,
-      "eval_overall_accuracy": 0.9674775928297055,
-      "eval_overall_f1": 0.8631875881523273,
-      "eval_overall_precision": 0.8247978436657682,
-      "eval_overall_recall": 0.9053254437869822,
-      "eval_runtime": 0.9089,
-      "eval_samples_per_second": 205.735,
-      "eval_steps_per_second": 3.301,
+      "eval_TIME_precision": 0.8064516129032258,
+      "eval_TIME_recall": 0.8620689655172413,
+      "eval_loss": 0.2897973358631134,
+      "eval_overall_accuracy": 0.9646606914212548,
+      "eval_overall_f1": 0.8607954545454546,
+      "eval_overall_precision": 0.8278688524590164,
+      "eval_overall_recall": 0.8964497041420119,
+      "eval_runtime": 0.3355,
+      "eval_samples_per_second": 557.358,
+      "eval_steps_per_second": 8.942,
       "step": 4452
     },
     {
       "epoch": 43.0,
-      "grad_norm": 0.27700409293174744,
+      "grad_norm": 0.01654691994190216,
       "learning_rate": 2.8499999999999998e-05,
-      "loss": 0.0014,
+      "loss": 0.0015,
       "step": 4558
     },
     {
       "epoch": 43.0,
-      "eval_LOCATION_f1": 0.8767123287671235,
+      "eval_LOCATION_f1": 0.8533333333333333,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8311688311688312,
+      "eval_LOCATION_precision": 0.7901234567901234,
       "eval_LOCATION_recall": 0.927536231884058,
-      "eval_ORGANIZATION_f1": 0.7580645161290323,
+      "eval_ORGANIZATION_f1": 0.7819548872180451,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.7121212121212122,
-      "eval_ORGANIZATION_recall": 0.8103448275862069,
-      "eval_PERSON_f1": 0.877742946708464,
+      "eval_ORGANIZATION_precision": 0.6933333333333334,
+      "eval_ORGANIZATION_recall": 0.896551724137931,
+      "eval_PERSON_f1": 0.9067524115755626,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8383233532934131,
-      "eval_PERSON_recall": 0.9210526315789473,
-      "eval_QUANTITY_f1": 0.7741935483870969,
+      "eval_PERSON_precision": 0.8867924528301887,
+      "eval_PERSON_recall": 0.9276315789473685,
+      "eval_QUANTITY_f1": 0.8253968253968254,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.75,
-      "eval_QUANTITY_recall": 0.8,
-      "eval_TIME_f1": 0.8387096774193549,
+      "eval_QUANTITY_precision": 0.7878787878787878,
+      "eval_QUANTITY_recall": 0.8666666666666667,
+      "eval_TIME_f1": 0.847457627118644,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.7878787878787878,
-      "eval_TIME_recall": 0.896551724137931,
-      "eval_loss": 0.2879803478717804,
-      "eval_overall_accuracy": 0.9651728553137003,
-      "eval_overall_f1": 0.8443197755960729,
-      "eval_overall_precision": 0.8026666666666666,
-      "eval_overall_recall": 0.8905325443786982,
-      "eval_runtime": 0.8107,
-      "eval_samples_per_second": 230.668,
-      "eval_steps_per_second": 3.701,
+      "eval_TIME_precision": 0.8333333333333334,
+      "eval_TIME_recall": 0.8620689655172413,
+      "eval_loss": 0.31039759516716003,
+      "eval_overall_accuracy": 0.9649167733674776,
+      "eval_overall_f1": 0.8603351955307262,
+      "eval_overall_precision": 0.8148148148148148,
+      "eval_overall_recall": 0.9112426035502958,
+      "eval_runtime": 0.3294,
+      "eval_samples_per_second": 567.728,
+      "eval_steps_per_second": 9.108,
       "step": 4558
     },
     {
       "epoch": 44.0,
-      "grad_norm": 0.5553054213523865,
+      "grad_norm": 0.013932665809988976,
       "learning_rate": 2.8000000000000003e-05,
-      "loss": 0.002,
+      "loss": 0.0016,
       "step": 4664
     },
     {
       "epoch": 44.0,
-      "eval_LOCATION_f1": 0.8648648648648648,
+      "eval_LOCATION_f1": 0.8456375838926175,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.810126582278481,
-      "eval_LOCATION_recall": 0.927536231884058,
-      "eval_ORGANIZATION_f1": 0.796875,
+      "eval_LOCATION_precision": 0.7875,
+      "eval_LOCATION_recall": 0.9130434782608695,
+      "eval_ORGANIZATION_f1": 0.7999999999999999,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.7285714285714285,
-      "eval_ORGANIZATION_recall": 0.8793103448275862,
-      "eval_PERSON_f1": 0.9161290322580645,
+      "eval_ORGANIZATION_precision": 0.746268656716418,
+      "eval_ORGANIZATION_recall": 0.8620689655172413,
+      "eval_PERSON_f1": 0.9032258064516129,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8987341772151899,
-      "eval_PERSON_recall": 0.9342105263157895,
-      "eval_QUANTITY_f1": 0.8064516129032259,
+      "eval_PERSON_precision": 0.8860759493670886,
+      "eval_PERSON_recall": 0.9210526315789473,
+      "eval_QUANTITY_f1": 0.8387096774193549,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.78125,
-      "eval_QUANTITY_recall": 0.8333333333333334,
-      "eval_TIME_f1": 0.8387096774193549,
+      "eval_QUANTITY_precision": 0.8125,
+      "eval_QUANTITY_recall": 0.8666666666666667,
+      "eval_TIME_f1": 0.8421052631578947,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.7878787878787878,
-      "eval_TIME_recall": 0.896551724137931,
-      "eval_loss": 0.2945367693901062,
-      "eval_overall_accuracy": 0.9661971830985916,
-      "eval_overall_f1": 0.867605633802817,
-      "eval_overall_precision": 0.8279569892473119,
-      "eval_overall_recall": 0.9112426035502958,
-      "eval_runtime": 0.8052,
-      "eval_samples_per_second": 232.233,
-      "eval_steps_per_second": 3.726,
+      "eval_TIME_precision": 0.8571428571428571,
+      "eval_TIME_recall": 0.8275862068965517,
+      "eval_loss": 0.30845847725868225,
+      "eval_overall_accuracy": 0.9641485275288092,
+      "eval_overall_f1": 0.8620199146514936,
+      "eval_overall_precision": 0.8301369863013699,
+      "eval_overall_recall": 0.8964497041420119,
+      "eval_runtime": 0.3211,
+      "eval_samples_per_second": 582.336,
+      "eval_steps_per_second": 9.342,
       "step": 4664
     },
     {
       "epoch": 45.0,
-      "grad_norm": 0.12713950872421265,
+      "grad_norm": 0.011009820736944675,
       "learning_rate": 2.7500000000000004e-05,
-      "loss": 0.003,
+      "loss": 0.0018,
       "step": 4770
     },
     {
       "epoch": 45.0,
-      "eval_LOCATION_f1": 0.8311688311688311,
+      "eval_LOCATION_f1": 0.8590604026845639,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.7529411764705882,
+      "eval_LOCATION_precision": 0.8,
       "eval_LOCATION_recall": 0.927536231884058,
-      "eval_ORGANIZATION_f1": 0.7226890756302522,
+      "eval_ORGANIZATION_f1": 0.7596899224806202,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.7049180327868853,
-      "eval_ORGANIZATION_recall": 0.7413793103448276,
-      "eval_PERSON_f1": 0.9155844155844156,
+      "eval_ORGANIZATION_precision": 0.6901408450704225,
+      "eval_ORGANIZATION_recall": 0.8448275862068966,
+      "eval_PERSON_f1": 0.9003215434083601,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.9038461538461539,
-      "eval_PERSON_recall": 0.9276315789473685,
-      "eval_QUANTITY_f1": 0.819672131147541,
+      "eval_PERSON_precision": 0.8805031446540881,
+      "eval_PERSON_recall": 0.9210526315789473,
+      "eval_QUANTITY_f1": 0.8253968253968254,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.8064516129032258,
-      "eval_QUANTITY_recall": 0.8333333333333334,
-      "eval_TIME_f1": 0.8813559322033899,
+      "eval_QUANTITY_precision": 0.7878787878787878,
+      "eval_QUANTITY_recall": 0.8666666666666667,
+      "eval_TIME_f1": 0.847457627118644,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.8666666666666667,
-      "eval_TIME_recall": 0.896551724137931,
-      "eval_loss": 0.2897718548774719,
-      "eval_overall_accuracy": 0.9646606914212548,
-      "eval_overall_f1": 0.8530670470756063,
-      "eval_overall_precision": 0.8236914600550964,
-      "eval_overall_recall": 0.8846153846153846,
-      "eval_runtime": 0.8046,
-      "eval_samples_per_second": 232.414,
-      "eval_steps_per_second": 3.729,
+      "eval_TIME_precision": 0.8333333333333334,
+      "eval_TIME_recall": 0.8620689655172413,
+      "eval_loss": 0.3025805950164795,
+      "eval_overall_accuracy": 0.9615877080665813,
+      "eval_overall_f1": 0.8551336146272854,
+      "eval_overall_precision": 0.8150134048257373,
+      "eval_overall_recall": 0.8994082840236687,
+      "eval_runtime": 0.3247,
+      "eval_samples_per_second": 575.917,
+      "eval_steps_per_second": 9.239,
       "step": 4770
     },
     {
       "epoch": 46.0,
-      "grad_norm": 0.0031549884006381035,
+      "grad_norm": 13.414834022521973,
       "learning_rate": 2.7000000000000002e-05,
-      "loss": 0.0029,
+      "loss": 0.0033,
       "step": 4876
     },
     {
       "epoch": 46.0,
-      "eval_LOCATION_f1": 0.8441558441558441,
+      "eval_LOCATION_f1": 0.8366013071895425,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.7647058823529411,
-      "eval_LOCATION_recall": 0.9420289855072463,
-      "eval_ORGANIZATION_f1": 0.8125,
+      "eval_LOCATION_precision": 0.7619047619047619,
+      "eval_LOCATION_recall": 0.927536231884058,
+      "eval_ORGANIZATION_f1": 0.7479674796747967,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.7428571428571429,
-      "eval_ORGANIZATION_recall": 0.896551724137931,
-      "eval_PERSON_f1": 0.9067524115755626,
+      "eval_ORGANIZATION_precision": 0.7076923076923077,
+      "eval_ORGANIZATION_recall": 0.7931034482758621,
+      "eval_PERSON_f1": 0.8945686900958467,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8867924528301887,
-      "eval_PERSON_recall": 0.9276315789473685,
-      "eval_QUANTITY_f1": 0.870967741935484,
+      "eval_PERSON_precision": 0.8695652173913043,
+      "eval_PERSON_recall": 0.9210526315789473,
+      "eval_QUANTITY_f1": 0.8253968253968254,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.84375,
-      "eval_QUANTITY_recall": 0.9,
-      "eval_TIME_f1": 0.8125,
+      "eval_QUANTITY_precision": 0.7878787878787878,
+      "eval_QUANTITY_recall": 0.8666666666666667,
+      "eval_TIME_f1": 0.8333333333333334,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.7428571428571429,
-      "eval_TIME_recall": 0.896551724137931,
-      "eval_loss": 0.310619056224823,
-      "eval_overall_accuracy": 0.9654289372599232,
-      "eval_overall_f1": 0.8650904033379694,
-      "eval_overall_precision": 0.8162729658792651,
-      "eval_overall_recall": 0.9201183431952663,
-      "eval_runtime": 0.8131,
-      "eval_samples_per_second": 229.995,
-      "eval_steps_per_second": 3.69,
+      "eval_TIME_precision": 0.8064516129032258,
+      "eval_TIME_recall": 0.8620689655172413,
+      "eval_loss": 0.2862217128276825,
+      "eval_overall_accuracy": 0.963124199743918,
+      "eval_overall_f1": 0.8455056179775281,
+      "eval_overall_precision": 0.8048128342245989,
+      "eval_overall_recall": 0.8905325443786982,
+      "eval_runtime": 0.335,
+      "eval_samples_per_second": 558.225,
+      "eval_steps_per_second": 8.955,
       "step": 4876
     },
     {
       "epoch": 47.0,
-      "grad_norm": 0.037439510226249695,
+      "grad_norm": 0.003819537116214633,
       "learning_rate": 2.6500000000000004e-05,
-      "loss": 0.0033,
+      "loss": 0.0007,
       "step": 4982
     },
     {
       "epoch": 47.0,
-      "eval_LOCATION_f1": 0.8840579710144928,
+      "eval_LOCATION_f1": 0.888888888888889,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8840579710144928,
-      "eval_LOCATION_recall": 0.8840579710144928,
-      "eval_ORGANIZATION_f1": 0.6788990825688075,
+      "eval_LOCATION_precision": 0.8533333333333334,
+      "eval_LOCATION_recall": 0.927536231884058,
+      "eval_ORGANIZATION_f1": 0.7666666666666667,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.7254901960784313,
-      "eval_ORGANIZATION_recall": 0.6379310344827587,
-      "eval_PERSON_f1": 0.9215686274509804,
+      "eval_ORGANIZATION_precision": 0.7419354838709677,
+      "eval_ORGANIZATION_recall": 0.7931034482758621,
+      "eval_PERSON_f1": 0.9126213592233009,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.9155844155844156,
+      "eval_PERSON_precision": 0.8980891719745223,
       "eval_PERSON_recall": 0.9276315789473685,
-      "eval_QUANTITY_f1": 0.8524590163934426,
+      "eval_QUANTITY_f1": 0.8387096774193549,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.8387096774193549,
+      "eval_QUANTITY_precision": 0.8125,
       "eval_QUANTITY_recall": 0.8666666666666667,
       "eval_TIME_f1": 0.8620689655172413,
       "eval_TIME_number": 29,
       "eval_TIME_precision": 0.8620689655172413,
       "eval_TIME_recall": 0.8620689655172413,
-      "eval_loss": 0.26650553941726685,
-      "eval_overall_accuracy": 0.9638924455825865,
-      "eval_overall_f1": 0.8630952380952381,
-      "eval_overall_precision": 0.8682634730538922,
-      "eval_overall_recall": 0.8579881656804734,
-      "eval_runtime": 0.8111,
-      "eval_samples_per_second": 230.541,
-      "eval_steps_per_second": 3.699,
+      "eval_loss": 0.28280648589134216,
+      "eval_overall_accuracy": 0.9690140845070423,
+      "eval_overall_f1": 0.8715728715728714,
+      "eval_overall_precision": 0.8507042253521127,
+      "eval_overall_recall": 0.893491124260355,
+      "eval_runtime": 0.3231,
+      "eval_samples_per_second": 578.72,
+      "eval_steps_per_second": 9.284,
       "step": 4982
     },
     {
       "epoch": 48.0,
-      "grad_norm": 0.005395154468715191,
+      "grad_norm": 0.046515222638845444,
       "learning_rate": 2.6000000000000002e-05,
-      "loss": 0.0027,
+      "loss": 0.0018,
       "step": 5088
     },
     {
       "epoch": 48.0,
-      "eval_LOCATION_f1": 0.8311688311688311,
+      "eval_LOCATION_f1": 0.8767123287671235,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.7529411764705882,
+      "eval_LOCATION_precision": 0.8311688311688312,
       "eval_LOCATION_recall": 0.927536231884058,
-      "eval_ORGANIZATION_f1": 0.7538461538461538,
+      "eval_ORGANIZATION_f1": 0.75,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.6805555555555556,
-      "eval_ORGANIZATION_recall": 0.8448275862068966,
-      "eval_PERSON_f1": 0.896774193548387,
+      "eval_ORGANIZATION_precision": 0.6857142857142857,
+      "eval_ORGANIZATION_recall": 0.8275862068965517,
+      "eval_PERSON_f1": 0.9032258064516129,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.879746835443038,
-      "eval_PERSON_recall": 0.9144736842105263,
+      "eval_PERSON_precision": 0.8860759493670886,
+      "eval_PERSON_recall": 0.9210526315789473,
       "eval_QUANTITY_f1": 0.8387096774193549,
       "eval_QUANTITY_number": 30,
       "eval_QUANTITY_precision": 0.8125,
       "eval_QUANTITY_recall": 0.8666666666666667,
-      "eval_TIME_f1": 0.8387096774193549,
+      "eval_TIME_f1": 0.847457627118644,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.7878787878787878,
-      "eval_TIME_recall": 0.896551724137931,
-      "eval_loss": 0.3204702138900757,
-      "eval_overall_accuracy": 0.9605633802816902,
-      "eval_overall_f1": 0.8467966573816157,
-      "eval_overall_precision": 0.8,
-      "eval_overall_recall": 0.8994082840236687,
-      "eval_runtime": 0.8068,
-      "eval_samples_per_second": 231.778,
-      "eval_steps_per_second": 3.718,
+      "eval_TIME_precision": 0.8333333333333334,
+      "eval_TIME_recall": 0.8620689655172413,
+      "eval_loss": 0.2690236270427704,
+      "eval_overall_accuracy": 0.9664532650448143,
+      "eval_overall_f1": 0.8595744680851063,
+      "eval_overall_precision": 0.8256130790190735,
+      "eval_overall_recall": 0.8964497041420119,
+      "eval_runtime": 0.3232,
+      "eval_samples_per_second": 578.625,
+      "eval_steps_per_second": 9.283,
       "step": 5088
     },
     {
       "epoch": 49.0,
-      "grad_norm": 0.010527399368584156,
+      "grad_norm": 0.002773936837911606,
       "learning_rate": 2.5500000000000003e-05,
-      "loss": 0.0014,
+      "loss": 0.0005,
       "step": 5194
     },
     {
       "epoch": 49.0,
-      "eval_LOCATION_f1": 0.8724832214765101,
+      "eval_LOCATION_f1": 0.870748299319728,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8125,
-      "eval_LOCATION_recall": 0.9420289855072463,
-      "eval_ORGANIZATION_f1": 0.7419354838709677,
+      "eval_LOCATION_precision": 0.8205128205128205,
+      "eval_LOCATION_recall": 0.927536231884058,
+      "eval_ORGANIZATION_f1": 0.7317073170731708,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.696969696969697,
-      "eval_ORGANIZATION_recall": 0.7931034482758621,
-      "eval_PERSON_f1": 0.9131832797427653,
+      "eval_ORGANIZATION_precision": 0.6923076923076923,
+      "eval_ORGANIZATION_recall": 0.7758620689655172,
+      "eval_PERSON_f1": 0.8881789137380192,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8930817610062893,
-      "eval_PERSON_recall": 0.9342105263157895,
-      "eval_QUANTITY_f1": 0.8387096774193549,
-      "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.8125,
-      "eval_QUANTITY_recall": 0.8666666666666667,
-      "eval_TIME_f1": 0.8666666666666666,
-      "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.8387096774193549,
-      "eval_TIME_recall": 0.896551724137931,
-      "eval_loss": 0.28002333641052246,
-      "eval_overall_accuracy": 0.9656850192061459,
-      "eval_overall_f1": 0.8640226628895185,
-      "eval_overall_precision": 0.8288043478260869,
-      "eval_overall_recall": 0.9023668639053254,
-      "eval_runtime": 0.8074,
-      "eval_samples_per_second": 231.603,
-      "eval_steps_per_second": 3.716,
+      "eval_PERSON_precision": 0.8633540372670807,
+      "eval_PERSON_recall": 0.9144736842105263,
+      "eval_QUANTITY_f1": 0.870967741935484,
+      "eval_QUANTITY_number": 30,
+      "eval_QUANTITY_precision": 0.84375,
+      "eval_QUANTITY_recall": 0.9,
+      "eval_TIME_f1": 0.8620689655172413,
+      "eval_TIME_number": 29,
+      "eval_TIME_precision": 0.8620689655172413,
+      "eval_TIME_recall": 0.8620689655172413,
+      "eval_loss": 0.28763094544410706,
+      "eval_overall_accuracy": 0.9641485275288092,
+      "eval_overall_f1": 0.8534850640113798,
+      "eval_overall_precision": 0.821917808219178,
+      "eval_overall_recall": 0.8875739644970414,
+      "eval_runtime": 0.3215,
+      "eval_samples_per_second": 581.6,
+      "eval_steps_per_second": 9.33,
       "step": 5194
     },
     {
       "epoch": 50.0,
-      "grad_norm": 0.0005556904943659902,
+      "grad_norm": 0.001465336070396006,
       "learning_rate": 2.5e-05,
-      "loss": 0.0005,
+      "loss": 0.0017,
       "step": 5300
     },
     {
       "epoch": 50.0,
-      "eval_LOCATION_f1": 0.8590604026845639,
+      "eval_LOCATION_f1": 0.8648648648648648,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8,
+      "eval_LOCATION_precision": 0.810126582278481,
       "eval_LOCATION_recall": 0.927536231884058,
-      "eval_ORGANIZATION_f1": 0.7441860465116279,
+      "eval_ORGANIZATION_f1": 0.711864406779661,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.676056338028169,
-      "eval_ORGANIZATION_recall": 0.8275862068965517,
-      "eval_PERSON_f1": 0.9067524115755626,
+      "eval_ORGANIZATION_precision": 0.7,
+      "eval_ORGANIZATION_recall": 0.7241379310344828,
+      "eval_PERSON_f1": 0.9061488673139159,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8867924528301887,
-      "eval_PERSON_recall": 0.9276315789473685,
+      "eval_PERSON_precision": 0.89171974522293,
+      "eval_PERSON_recall": 0.9210526315789473,
       "eval_QUANTITY_f1": 0.8387096774193549,
       "eval_QUANTITY_number": 30,
       "eval_QUANTITY_precision": 0.8125,
       "eval_QUANTITY_recall": 0.8666666666666667,
-      "eval_TIME_f1": 0.8666666666666666,
+      "eval_TIME_f1": 0.8333333333333334,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.8387096774193549,
-      "eval_TIME_recall": 0.896551724137931,
-      "eval_loss": 0.30621612071990967,
-      "eval_overall_accuracy": 0.9633802816901409,
-      "eval_overall_f1": 0.8579465541490858,
-      "eval_overall_precision": 0.8176943699731903,
-      "eval_overall_recall": 0.9023668639053254,
-      "eval_runtime": 0.8061,
-      "eval_samples_per_second": 231.983,
-      "eval_steps_per_second": 3.722,
+      "eval_TIME_precision": 0.8064516129032258,
+      "eval_TIME_recall": 0.8620689655172413,
+      "eval_loss": 0.3069976568222046,
+      "eval_overall_accuracy": 0.964404609475032,
+      "eval_overall_f1": 0.8522238163558107,
+      "eval_overall_precision": 0.8272980501392758,
+      "eval_overall_recall": 0.878698224852071,
+      "eval_runtime": 0.3366,
+      "eval_samples_per_second": 555.511,
+      "eval_steps_per_second": 8.912,
       "step": 5300
     },
     {
       "epoch": 51.0,
-      "grad_norm": 0.002213704166933894,
+      "grad_norm": 0.0014170885551720858,
       "learning_rate": 2.45e-05,
       "loss": 0.0009,
       "step": 5406
     },
     {
       "epoch": 51.0,
-      "eval_LOCATION_f1": 0.8356164383561644,
+      "eval_LOCATION_f1": 0.870748299319728,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.7922077922077922,
-      "eval_LOCATION_recall": 0.8840579710144928,
-      "eval_ORGANIZATION_f1": 0.7200000000000001,
+      "eval_LOCATION_precision": 0.8205128205128205,
+      "eval_LOCATION_recall": 0.927536231884058,
+      "eval_ORGANIZATION_f1": 0.7812500000000001,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.6716417910447762,
-      "eval_ORGANIZATION_recall": 0.7758620689655172,
-      "eval_PERSON_f1": 0.8974358974358975,
+      "eval_ORGANIZATION_precision": 0.7142857142857143,
+      "eval_ORGANIZATION_recall": 0.8620689655172413,
+      "eval_PERSON_f1": 0.9061488673139159,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.875,
+      "eval_PERSON_precision": 0.89171974522293,
       "eval_PERSON_recall": 0.9210526315789473,
       "eval_QUANTITY_f1": 0.8387096774193549,
       "eval_QUANTITY_number": 30,
       "eval_QUANTITY_precision": 0.8125,
       "eval_QUANTITY_recall": 0.8666666666666667,
-      "eval_TIME_f1": 0.8387096774193549,
+      "eval_TIME_f1": 0.847457627118644,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.7878787878787878,
-      "eval_TIME_recall": 0.896551724137931,
-      "eval_loss": 0.3094327747821808,
-      "eval_overall_accuracy": 0.9615877080665813,
-      "eval_overall_f1": 0.8429985855728428,
-      "eval_overall_precision": 0.8075880758807588,
-      "eval_overall_recall": 0.8816568047337278,
-      "eval_runtime": 0.8118,
-      "eval_samples_per_second": 230.354,
-      "eval_steps_per_second": 3.696,
+      "eval_TIME_precision": 0.8333333333333334,
+      "eval_TIME_recall": 0.8620689655172413,
+      "eval_loss": 0.3090084195137024,
+      "eval_overall_accuracy": 0.9649167733674776,
+      "eval_overall_f1": 0.8652482269503546,
+      "eval_overall_precision": 0.8310626702997275,
+      "eval_overall_recall": 0.9023668639053254,
+      "eval_runtime": 0.3236,
+      "eval_samples_per_second": 577.785,
+      "eval_steps_per_second": 9.269,
       "step": 5406
     },
     {
       "epoch": 52.0,
-      "grad_norm": 0.12124722450971603,
+      "grad_norm": 0.009222417138516903,
       "learning_rate": 2.4e-05,
-      "loss": 0.0007,
+      "loss": 0.0005,
       "step": 5512
     },
     {
       "epoch": 52.0,
-      "eval_LOCATION_f1": 0.8493150684931507,
+      "eval_LOCATION_f1": 0.8590604026845639,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8051948051948052,
-      "eval_LOCATION_recall": 0.8985507246376812,
-      "eval_ORGANIZATION_f1": 0.7226890756302522,
+      "eval_LOCATION_precision": 0.8,
+      "eval_LOCATION_recall": 0.927536231884058,
+      "eval_ORGANIZATION_f1": 0.7377049180327868,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.7049180327868853,
-      "eval_ORGANIZATION_recall": 0.7413793103448276,
-      "eval_PERSON_f1": 0.9003215434083601,
+      "eval_ORGANIZATION_precision": 0.703125,
+      "eval_ORGANIZATION_recall": 0.7758620689655172,
+      "eval_PERSON_f1": 0.8996763754045308,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8805031446540881,
-      "eval_PERSON_recall": 0.9210526315789473,
-      "eval_QUANTITY_f1": 0.870967741935484,
+      "eval_PERSON_precision": 0.8853503184713376,
+      "eval_PERSON_recall": 0.9144736842105263,
+      "eval_QUANTITY_f1": 0.8387096774193549,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.84375,
-      "eval_QUANTITY_recall": 0.9,
-      "eval_TIME_f1": 0.8666666666666666,
+      "eval_QUANTITY_precision": 0.8125,
+      "eval_QUANTITY_recall": 0.8666666666666667,
+      "eval_TIME_f1": 0.8813559322033899,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.8387096774193549,
+      "eval_TIME_precision": 0.8666666666666667,
       "eval_TIME_recall": 0.896551724137931,
-      "eval_loss": 0.2926962077617645,
-      "eval_overall_accuracy": 0.9651728553137003,
-      "eval_overall_f1": 0.8538681948424068,
-      "eval_overall_precision": 0.8277777777777777,
-      "eval_overall_recall": 0.8816568047337278,
-      "eval_runtime": 0.8047,
-      "eval_samples_per_second": 232.378,
-      "eval_steps_per_second": 3.728,
+      "eval_loss": 0.3099108636379242,
+      "eval_overall_accuracy": 0.9656850192061459,
+      "eval_overall_f1": 0.855920114122682,
+      "eval_overall_precision": 0.8264462809917356,
+      "eval_overall_recall": 0.8875739644970414,
+      "eval_runtime": 0.3275,
+      "eval_samples_per_second": 570.979,
+      "eval_steps_per_second": 9.16,
       "step": 5512
     },
     {
       "epoch": 53.0,
-      "grad_norm": 0.007010570261627436,
+      "grad_norm": 0.0005486037116497755,
       "learning_rate": 2.35e-05,
-      "loss": 0.0005,
+      "loss": 0.001,
       "step": 5618
     },
     {
       "epoch": 53.0,
-      "eval_LOCATION_f1": 0.8413793103448277,
+      "eval_LOCATION_f1": 0.8533333333333333,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8026315789473685,
-      "eval_LOCATION_recall": 0.8840579710144928,
-      "eval_ORGANIZATION_f1": 0.7166666666666668,
+      "eval_LOCATION_precision": 0.7901234567901234,
+      "eval_LOCATION_recall": 0.927536231884058,
+      "eval_ORGANIZATION_f1": 0.7666666666666667,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.6935483870967742,
-      "eval_ORGANIZATION_recall": 0.7413793103448276,
-      "eval_PERSON_f1": 0.9038461538461539,
+      "eval_ORGANIZATION_precision": 0.7419354838709677,
+      "eval_ORGANIZATION_recall": 0.7931034482758621,
+      "eval_PERSON_f1": 0.8996763754045308,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.88125,
-      "eval_PERSON_recall": 0.9276315789473685,
+      "eval_PERSON_precision": 0.8853503184713376,
+      "eval_PERSON_recall": 0.9144736842105263,
       "eval_QUANTITY_f1": 0.8387096774193549,
       "eval_QUANTITY_number": 30,
       "eval_QUANTITY_precision": 0.8125,
@@ -2065,1857 +2065,1857 @@
       "eval_TIME_number": 29,
       "eval_TIME_precision": 0.8387096774193549,
       "eval_TIME_recall": 0.896551724137931,
-      "eval_loss": 0.3106415271759033,
-      "eval_overall_accuracy": 0.9638924455825865,
-      "eval_overall_f1": 0.8497854077253219,
-      "eval_overall_precision": 0.8227146814404432,
-      "eval_overall_recall": 0.878698224852071,
-      "eval_runtime": 0.8081,
-      "eval_samples_per_second": 231.4,
-      "eval_steps_per_second": 3.712,
+      "eval_loss": 0.31158438324928284,
+      "eval_overall_accuracy": 0.9649167733674776,
+      "eval_overall_f1": 0.8587731811697575,
+      "eval_overall_precision": 0.8292011019283747,
+      "eval_overall_recall": 0.8905325443786982,
+      "eval_runtime": 0.3209,
+      "eval_samples_per_second": 582.696,
+      "eval_steps_per_second": 9.348,
       "step": 5618
     },
     {
       "epoch": 54.0,
-      "grad_norm": 0.0005368912825360894,
+      "grad_norm": 0.0006028018542565405,
       "learning_rate": 2.3000000000000003e-05,
       "loss": 0.0006,
       "step": 5724
     },
     {
       "epoch": 54.0,
-      "eval_LOCATION_f1": 0.8472222222222223,
+      "eval_LOCATION_f1": 0.8533333333333333,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8133333333333334,
-      "eval_LOCATION_recall": 0.8840579710144928,
-      "eval_ORGANIZATION_f1": 0.7241379310344829,
+      "eval_LOCATION_precision": 0.7901234567901234,
+      "eval_LOCATION_recall": 0.927536231884058,
+      "eval_ORGANIZATION_f1": 0.7377049180327868,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.7241379310344828,
-      "eval_ORGANIZATION_recall": 0.7241379310344828,
-      "eval_PERSON_f1": 0.9102564102564101,
+      "eval_ORGANIZATION_precision": 0.703125,
+      "eval_ORGANIZATION_recall": 0.7758620689655172,
+      "eval_PERSON_f1": 0.9055374592833876,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8875,
-      "eval_PERSON_recall": 0.9342105263157895,
+      "eval_PERSON_precision": 0.896774193548387,
+      "eval_PERSON_recall": 0.9144736842105263,
       "eval_QUANTITY_f1": 0.8387096774193549,
       "eval_QUANTITY_number": 30,
       "eval_QUANTITY_precision": 0.8125,
       "eval_QUANTITY_recall": 0.8666666666666667,
-      "eval_TIME_f1": 0.8524590163934426,
+      "eval_TIME_f1": 0.847457627118644,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.8125,
-      "eval_TIME_recall": 0.896551724137931,
-      "eval_loss": 0.3165185749530792,
-      "eval_overall_accuracy": 0.9649167733674776,
-      "eval_overall_f1": 0.8546762589928059,
-      "eval_overall_precision": 0.8319327731092437,
-      "eval_overall_recall": 0.878698224852071,
-      "eval_runtime": 0.8064,
-      "eval_samples_per_second": 231.905,
-      "eval_steps_per_second": 3.72,
+      "eval_TIME_precision": 0.8333333333333334,
+      "eval_TIME_recall": 0.8620689655172413,
+      "eval_loss": 0.317971795797348,
+      "eval_overall_accuracy": 0.964404609475032,
+      "eval_overall_f1": 0.8542857142857142,
+      "eval_overall_precision": 0.8259668508287292,
+      "eval_overall_recall": 0.8846153846153846,
+      "eval_runtime": 0.3274,
+      "eval_samples_per_second": 571.109,
+      "eval_steps_per_second": 9.162,
       "step": 5724
     },
     {
       "epoch": 55.0,
-      "grad_norm": 0.0005507151363417506,
+      "grad_norm": 0.0006948837544769049,
       "learning_rate": 2.25e-05,
-      "loss": 0.0008,
+      "loss": 0.0004,
       "step": 5830
     },
     {
       "epoch": 55.0,
-      "eval_LOCATION_f1": 0.8920863309352518,
+      "eval_LOCATION_f1": 0.8590604026845639,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8857142857142857,
-      "eval_LOCATION_recall": 0.8985507246376812,
-      "eval_ORGANIZATION_f1": 0.6915887850467289,
+      "eval_LOCATION_precision": 0.8,
+      "eval_LOCATION_recall": 0.927536231884058,
+      "eval_ORGANIZATION_f1": 0.7500000000000001,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.7551020408163265,
-      "eval_ORGANIZATION_recall": 0.6379310344827587,
-      "eval_PERSON_f1": 0.9061488673139159,
+      "eval_ORGANIZATION_precision": 0.7258064516129032,
+      "eval_ORGANIZATION_recall": 0.7758620689655172,
+      "eval_PERSON_f1": 0.9032258064516129,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.89171974522293,
+      "eval_PERSON_precision": 0.8860759493670886,
       "eval_PERSON_recall": 0.9210526315789473,
-      "eval_QUANTITY_f1": 0.8064516129032259,
+      "eval_QUANTITY_f1": 0.8387096774193549,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.78125,
-      "eval_QUANTITY_recall": 0.8333333333333334,
-      "eval_TIME_f1": 0.8813559322033899,
+      "eval_QUANTITY_precision": 0.8125,
+      "eval_QUANTITY_recall": 0.8666666666666667,
+      "eval_TIME_f1": 0.8620689655172413,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.8666666666666667,
-      "eval_TIME_recall": 0.896551724137931,
-      "eval_loss": 0.28772225975990295,
-      "eval_overall_accuracy": 0.9677336747759283,
-      "eval_overall_f1": 0.8579881656804734,
-      "eval_overall_precision": 0.8579881656804734,
-      "eval_overall_recall": 0.8579881656804734,
-      "eval_runtime": 0.8028,
-      "eval_samples_per_second": 232.926,
-      "eval_steps_per_second": 3.737,
+      "eval_TIME_precision": 0.8620689655172413,
+      "eval_TIME_recall": 0.8620689655172413,
+      "eval_loss": 0.32179853320121765,
+      "eval_overall_accuracy": 0.9646606914212548,
+      "eval_overall_f1": 0.8583690987124463,
+      "eval_overall_precision": 0.8310249307479224,
+      "eval_overall_recall": 0.8875739644970414,
+      "eval_runtime": 0.3207,
+      "eval_samples_per_second": 583.185,
+      "eval_steps_per_second": 9.356,
       "step": 5830
     },
     {
       "epoch": 56.0,
-      "grad_norm": 0.004189824219793081,
+      "grad_norm": 0.0012037695851176977,
       "learning_rate": 2.2000000000000003e-05,
-      "loss": 0.0013,
+      "loss": 0.0006,
       "step": 5936
     },
     {
       "epoch": 56.0,
-      "eval_LOCATION_f1": 0.8965517241379309,
+      "eval_LOCATION_f1": 0.8533333333333333,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8552631578947368,
-      "eval_LOCATION_recall": 0.9420289855072463,
-      "eval_ORGANIZATION_f1": 0.6956521739130435,
+      "eval_LOCATION_precision": 0.7901234567901234,
+      "eval_LOCATION_recall": 0.927536231884058,
+      "eval_ORGANIZATION_f1": 0.7333333333333333,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.7017543859649122,
-      "eval_ORGANIZATION_recall": 0.6896551724137931,
-      "eval_PERSON_f1": 0.9126213592233009,
+      "eval_ORGANIZATION_precision": 0.7096774193548387,
+      "eval_ORGANIZATION_recall": 0.7586206896551724,
+      "eval_PERSON_f1": 0.9032258064516129,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8980891719745223,
-      "eval_PERSON_recall": 0.9276315789473685,
-      "eval_QUANTITY_f1": 0.8524590163934426,
+      "eval_PERSON_precision": 0.8860759493670886,
+      "eval_PERSON_recall": 0.9210526315789473,
+      "eval_QUANTITY_f1": 0.8387096774193549,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.8387096774193549,
+      "eval_QUANTITY_precision": 0.8125,
       "eval_QUANTITY_recall": 0.8666666666666667,
-      "eval_TIME_f1": 0.8387096774193549,
+      "eval_TIME_f1": 0.8333333333333334,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.7878787878787878,
-      "eval_TIME_recall": 0.896551724137931,
-      "eval_loss": 0.2933734357357025,
-      "eval_overall_accuracy": 0.9646606914212548,
-      "eval_overall_f1": 0.861271676300578,
-      "eval_overall_precision": 0.8418079096045198,
-      "eval_overall_recall": 0.8816568047337278,
-      "eval_runtime": 0.8184,
-      "eval_samples_per_second": 228.501,
-      "eval_steps_per_second": 3.666,
+      "eval_TIME_precision": 0.8064516129032258,
+      "eval_TIME_recall": 0.8620689655172413,
+      "eval_loss": 0.33105283975601196,
+      "eval_overall_accuracy": 0.9633802816901409,
+      "eval_overall_f1": 0.8518518518518519,
+      "eval_overall_precision": 0.8214285714285714,
+      "eval_overall_recall": 0.8846153846153846,
+      "eval_runtime": 0.3391,
+      "eval_samples_per_second": 551.529,
+      "eval_steps_per_second": 8.848,
       "step": 5936
     },
     {
       "epoch": 57.0,
-      "grad_norm": 0.00032704288605600595,
+      "grad_norm": 0.0022144129034131765,
       "learning_rate": 2.15e-05,
-      "loss": 0.0003,
+      "loss": 0.0009,
       "step": 6042
     },
     {
       "epoch": 57.0,
-      "eval_LOCATION_f1": 0.8904109589041096,
+      "eval_LOCATION_f1": 0.8476821192052981,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8441558441558441,
-      "eval_LOCATION_recall": 0.9420289855072463,
-      "eval_ORGANIZATION_f1": 0.6956521739130435,
+      "eval_LOCATION_precision": 0.7804878048780488,
+      "eval_LOCATION_recall": 0.927536231884058,
+      "eval_ORGANIZATION_f1": 0.7419354838709677,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.7017543859649122,
-      "eval_ORGANIZATION_recall": 0.6896551724137931,
-      "eval_PERSON_f1": 0.9096774193548387,
+      "eval_ORGANIZATION_precision": 0.696969696969697,
+      "eval_ORGANIZATION_recall": 0.7931034482758621,
+      "eval_PERSON_f1": 0.896774193548387,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8924050632911392,
-      "eval_PERSON_recall": 0.9276315789473685,
-      "eval_QUANTITY_f1": 0.8253968253968254,
+      "eval_PERSON_precision": 0.879746835443038,
+      "eval_PERSON_recall": 0.9144736842105263,
+      "eval_QUANTITY_f1": 0.8387096774193549,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.7878787878787878,
+      "eval_QUANTITY_precision": 0.8125,
       "eval_QUANTITY_recall": 0.8666666666666667,
-      "eval_TIME_f1": 0.8387096774193549,
+      "eval_TIME_f1": 0.8333333333333334,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.7878787878787878,
-      "eval_TIME_recall": 0.896551724137931,
-      "eval_loss": 0.31640374660491943,
-      "eval_overall_accuracy": 0.9638924455825865,
-      "eval_overall_f1": 0.8563218390804598,
-      "eval_overall_precision": 0.8324022346368715,
-      "eval_overall_recall": 0.8816568047337278,
-      "eval_runtime": 0.8095,
-      "eval_samples_per_second": 231.007,
-      "eval_steps_per_second": 3.706,
+      "eval_TIME_precision": 0.8064516129032258,
+      "eval_TIME_recall": 0.8620689655172413,
+      "eval_loss": 0.31793665885925293,
+      "eval_overall_accuracy": 0.964404609475032,
+      "eval_overall_f1": 0.8486562942008485,
+      "eval_overall_precision": 0.8130081300813008,
+      "eval_overall_recall": 0.8875739644970414,
+      "eval_runtime": 0.3249,
+      "eval_samples_per_second": 575.562,
+      "eval_steps_per_second": 9.234,
       "step": 6042
     },
     {
       "epoch": 58.0,
-      "grad_norm": 0.0005215631099417806,
+      "grad_norm": 0.0016710077179595828,
       "learning_rate": 2.1e-05,
-      "loss": 0.0004,
+      "loss": 0.002,
       "step": 6148
     },
     {
       "epoch": 58.0,
-      "eval_LOCATION_f1": 0.8843537414965987,
+      "eval_LOCATION_f1": 0.870748299319728,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8333333333333334,
-      "eval_LOCATION_recall": 0.9420289855072463,
-      "eval_ORGANIZATION_f1": 0.7107438016528926,
+      "eval_LOCATION_precision": 0.8205128205128205,
+      "eval_LOCATION_recall": 0.927536231884058,
+      "eval_ORGANIZATION_f1": 0.6725663716814159,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.6825396825396826,
-      "eval_ORGANIZATION_recall": 0.7413793103448276,
-      "eval_PERSON_f1": 0.9096774193548387,
+      "eval_ORGANIZATION_precision": 0.6909090909090909,
+      "eval_ORGANIZATION_recall": 0.6551724137931034,
+      "eval_PERSON_f1": 0.9155844155844156,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8924050632911392,
+      "eval_PERSON_precision": 0.9038461538461539,
       "eval_PERSON_recall": 0.9276315789473685,
-      "eval_QUANTITY_f1": 0.8387096774193549,
+      "eval_QUANTITY_f1": 0.8064516129032259,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.8125,
-      "eval_QUANTITY_recall": 0.8666666666666667,
-      "eval_TIME_f1": 0.8387096774193549,
+      "eval_QUANTITY_precision": 0.78125,
+      "eval_QUANTITY_recall": 0.8333333333333334,
+      "eval_TIME_f1": 0.896551724137931,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.7878787878787878,
+      "eval_TIME_precision": 0.896551724137931,
       "eval_TIME_recall": 0.896551724137931,
-      "eval_loss": 0.32982975244522095,
-      "eval_overall_accuracy": 0.9628681177976952,
-      "eval_overall_f1": 0.8575498575498575,
-      "eval_overall_precision": 0.8269230769230769,
-      "eval_overall_recall": 0.8905325443786982,
-      "eval_runtime": 0.8776,
-      "eval_samples_per_second": 213.09,
-      "eval_steps_per_second": 3.419,
+      "eval_loss": 0.27503037452697754,
+      "eval_overall_accuracy": 0.9638924455825865,
+      "eval_overall_f1": 0.8546511627906976,
+      "eval_overall_precision": 0.84,
+      "eval_overall_recall": 0.8698224852071006,
+      "eval_runtime": 0.3317,
+      "eval_samples_per_second": 563.712,
+      "eval_steps_per_second": 9.044,
       "step": 6148
     },
     {
       "epoch": 59.0,
-      "grad_norm": 0.13604550063610077,
+      "grad_norm": 0.03837126865983009,
       "learning_rate": 2.05e-05,
-      "loss": 0.0006,
+      "loss": 0.0018,
       "step": 6254
     },
     {
       "epoch": 59.0,
-      "eval_LOCATION_f1": 0.8965517241379309,
+      "eval_LOCATION_f1": 0.8648648648648648,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8552631578947368,
-      "eval_LOCATION_recall": 0.9420289855072463,
-      "eval_ORGANIZATION_f1": 0.6371681415929203,
+      "eval_LOCATION_precision": 0.810126582278481,
+      "eval_LOCATION_recall": 0.927536231884058,
+      "eval_ORGANIZATION_f1": 0.7540983606557378,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.6545454545454545,
-      "eval_ORGANIZATION_recall": 0.6206896551724138,
-      "eval_PERSON_f1": 0.9131832797427653,
-      "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8930817610062893,
-      "eval_PERSON_recall": 0.9342105263157895,
+      "eval_ORGANIZATION_precision": 0.71875,
+      "eval_ORGANIZATION_recall": 0.7931034482758621,
+      "eval_PERSON_f1": 0.8996763754045308,
+      "eval_PERSON_number": 152,
+      "eval_PERSON_precision": 0.8853503184713376,
+      "eval_PERSON_recall": 0.9144736842105263,
       "eval_QUANTITY_f1": 0.8253968253968254,
       "eval_QUANTITY_number": 30,
       "eval_QUANTITY_precision": 0.7878787878787878,
       "eval_QUANTITY_recall": 0.8666666666666667,
-      "eval_TIME_f1": 0.8524590163934426,
+      "eval_TIME_f1": 0.8771929824561403,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.8125,
-      "eval_TIME_recall": 0.896551724137931,
-      "eval_loss": 0.3474750220775604,
-      "eval_overall_accuracy": 0.9618437900128041,
-      "eval_overall_f1": 0.8513708513708513,
-      "eval_overall_precision": 0.8309859154929577,
-      "eval_overall_recall": 0.8727810650887574,
-      "eval_runtime": 0.812,
-      "eval_samples_per_second": 230.289,
-      "eval_steps_per_second": 3.694,
+      "eval_TIME_precision": 0.8928571428571429,
+      "eval_TIME_recall": 0.8620689655172413,
+      "eval_loss": 0.29145359992980957,
+      "eval_overall_accuracy": 0.9656850192061459,
+      "eval_overall_f1": 0.8583690987124463,
+      "eval_overall_precision": 0.8310249307479224,
+      "eval_overall_recall": 0.8875739644970414,
+      "eval_runtime": 0.3233,
+      "eval_samples_per_second": 578.416,
+      "eval_steps_per_second": 9.279,
       "step": 6254
     },
     {
       "epoch": 60.0,
-      "grad_norm": 0.0007290157955139875,
+      "grad_norm": 0.10238273441791534,
       "learning_rate": 2e-05,
-      "loss": 0.0009,
+      "loss": 0.0008,
       "step": 6360
     },
     {
       "epoch": 60.0,
-      "eval_LOCATION_f1": 0.8783783783783783,
+      "eval_LOCATION_f1": 0.8648648648648648,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8227848101265823,
-      "eval_LOCATION_recall": 0.9420289855072463,
-      "eval_ORGANIZATION_f1": 0.7731092436974789,
+      "eval_LOCATION_precision": 0.810126582278481,
+      "eval_LOCATION_recall": 0.927536231884058,
+      "eval_ORGANIZATION_f1": 0.7027027027027026,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.7540983606557377,
-      "eval_ORGANIZATION_recall": 0.7931034482758621,
-      "eval_PERSON_f1": 0.9096774193548387,
+      "eval_ORGANIZATION_precision": 0.7358490566037735,
+      "eval_ORGANIZATION_recall": 0.6724137931034483,
+      "eval_PERSON_f1": 0.9025974025974026,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8924050632911392,
-      "eval_PERSON_recall": 0.9276315789473685,
-      "eval_QUANTITY_f1": 0.8253968253968254,
+      "eval_PERSON_precision": 0.8910256410256411,
+      "eval_PERSON_recall": 0.9144736842105263,
+      "eval_QUANTITY_f1": 0.8387096774193549,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.7878787878787878,
+      "eval_QUANTITY_precision": 0.8125,
       "eval_QUANTITY_recall": 0.8666666666666667,
-      "eval_TIME_f1": 0.8524590163934426,
+      "eval_TIME_f1": 0.8771929824561403,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.8125,
-      "eval_TIME_recall": 0.896551724137931,
-      "eval_loss": 0.3445851802825928,
-      "eval_overall_accuracy": 0.9646606914212548,
-      "eval_overall_f1": 0.8673323823109843,
-      "eval_overall_precision": 0.837465564738292,
-      "eval_overall_recall": 0.8994082840236687,
-      "eval_runtime": 0.8135,
-      "eval_samples_per_second": 229.871,
-      "eval_steps_per_second": 3.688,
+      "eval_TIME_precision": 0.8928571428571429,
+      "eval_TIME_recall": 0.8620689655172413,
+      "eval_loss": 0.31033989787101746,
+      "eval_overall_accuracy": 0.9649167733674776,
+      "eval_overall_f1": 0.8542274052478134,
+      "eval_overall_precision": 0.8419540229885057,
+      "eval_overall_recall": 0.8668639053254438,
+      "eval_runtime": 0.3317,
+      "eval_samples_per_second": 563.83,
+      "eval_steps_per_second": 9.045,
       "step": 6360
     },
     {
       "epoch": 61.0,
-      "grad_norm": 0.01683318056166172,
+      "grad_norm": 0.002005122834816575,
       "learning_rate": 1.9500000000000003e-05,
-      "loss": 0.0005,
+      "loss": 0.0013,
       "step": 6466
     },
     {
       "epoch": 61.0,
-      "eval_LOCATION_f1": 0.8666666666666666,
+      "eval_LOCATION_f1": 0.8749999999999999,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8024691358024691,
-      "eval_LOCATION_recall": 0.9420289855072463,
-      "eval_ORGANIZATION_f1": 0.7200000000000001,
+      "eval_LOCATION_precision": 0.84,
+      "eval_LOCATION_recall": 0.9130434782608695,
+      "eval_ORGANIZATION_f1": 0.7868852459016394,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.6716417910447762,
-      "eval_ORGANIZATION_recall": 0.7758620689655172,
+      "eval_ORGANIZATION_precision": 0.75,
+      "eval_ORGANIZATION_recall": 0.8275862068965517,
       "eval_PERSON_f1": 0.9003215434083601,
       "eval_PERSON_number": 152,
       "eval_PERSON_precision": 0.8805031446540881,
       "eval_PERSON_recall": 0.9210526315789473,
-      "eval_QUANTITY_f1": 0.8387096774193549,
+      "eval_QUANTITY_f1": 0.8253968253968254,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.8125,
+      "eval_QUANTITY_precision": 0.7878787878787878,
       "eval_QUANTITY_recall": 0.8666666666666667,
-      "eval_TIME_f1": 0.8524590163934426,
+      "eval_TIME_f1": 0.896551724137931,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.8125,
+      "eval_TIME_precision": 0.896551724137931,
       "eval_TIME_recall": 0.896551724137931,
-      "eval_loss": 0.3661682605743408,
-      "eval_overall_accuracy": 0.9610755441741358,
-      "eval_overall_f1": 0.8519040902679831,
-      "eval_overall_precision": 0.8140161725067385,
-      "eval_overall_recall": 0.893491124260355,
-      "eval_runtime": 0.8073,
-      "eval_samples_per_second": 231.639,
-      "eval_steps_per_second": 3.716,
+      "eval_loss": 0.2831355929374695,
+      "eval_overall_accuracy": 0.9661971830985916,
+      "eval_overall_f1": 0.8681948424068767,
+      "eval_overall_precision": 0.8416666666666667,
+      "eval_overall_recall": 0.8964497041420119,
+      "eval_runtime": 0.3193,
+      "eval_samples_per_second": 585.653,
+      "eval_steps_per_second": 9.395,
       "step": 6466
     },
     {
       "epoch": 62.0,
-      "grad_norm": 0.0029261417221277952,
+      "grad_norm": 0.0019113136222586036,
       "learning_rate": 1.9e-05,
-      "loss": 0.002,
+      "loss": 0.0013,
       "step": 6572
     },
     {
       "epoch": 62.0,
-      "eval_LOCATION_f1": 0.8965517241379309,
+      "eval_LOCATION_f1": 0.870748299319728,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8552631578947368,
-      "eval_LOCATION_recall": 0.9420289855072463,
-      "eval_ORGANIZATION_f1": 0.7377049180327868,
+      "eval_LOCATION_precision": 0.8205128205128205,
+      "eval_LOCATION_recall": 0.927536231884058,
+      "eval_ORGANIZATION_f1": 0.7999999999999999,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.703125,
-      "eval_ORGANIZATION_recall": 0.7758620689655172,
-      "eval_PERSON_f1": 0.9067524115755626,
+      "eval_ORGANIZATION_precision": 0.746268656716418,
+      "eval_ORGANIZATION_recall": 0.8620689655172413,
+      "eval_PERSON_f1": 0.9032258064516129,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8867924528301887,
-      "eval_PERSON_recall": 0.9276315789473685,
-      "eval_QUANTITY_f1": 0.7936507936507938,
+      "eval_PERSON_precision": 0.8860759493670886,
+      "eval_PERSON_recall": 0.9210526315789473,
+      "eval_QUANTITY_f1": 0.8387096774193549,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.7575757575757576,
-      "eval_QUANTITY_recall": 0.8333333333333334,
-      "eval_TIME_f1": 0.8666666666666666,
+      "eval_QUANTITY_precision": 0.8125,
+      "eval_QUANTITY_recall": 0.8666666666666667,
+      "eval_TIME_f1": 0.896551724137931,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.8387096774193549,
+      "eval_TIME_precision": 0.896551724137931,
       "eval_TIME_recall": 0.896551724137931,
-      "eval_loss": 0.31876397132873535,
-      "eval_overall_accuracy": 0.9664532650448143,
-      "eval_overall_f1": 0.861626248216833,
-      "eval_overall_precision": 0.8319559228650137,
-      "eval_overall_recall": 0.893491124260355,
-      "eval_runtime": 0.8044,
-      "eval_samples_per_second": 232.462,
-      "eval_steps_per_second": 3.729,
+      "eval_loss": 0.29855355620384216,
+      "eval_overall_accuracy": 0.967989756722151,
+      "eval_overall_f1": 0.8717948717948717,
+      "eval_overall_precision": 0.8406593406593407,
+      "eval_overall_recall": 0.9053254437869822,
+      "eval_runtime": 0.327,
+      "eval_samples_per_second": 571.832,
+      "eval_steps_per_second": 9.174,
       "step": 6572
     },
     {
       "epoch": 63.0,
-      "grad_norm": 0.0014877787325531244,
+      "grad_norm": 0.0004677773977164179,
       "learning_rate": 1.85e-05,
-      "loss": 0.0005,
+      "loss": 0.0009,
       "step": 6678
     },
     {
       "epoch": 63.0,
-      "eval_LOCATION_f1": 0.8951048951048951,
+      "eval_LOCATION_f1": 0.8513513513513514,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8648648648648649,
-      "eval_LOCATION_recall": 0.927536231884058,
-      "eval_ORGANIZATION_f1": 0.7603305785123967,
+      "eval_LOCATION_precision": 0.7974683544303798,
+      "eval_LOCATION_recall": 0.9130434782608695,
+      "eval_ORGANIZATION_f1": 0.8064516129032258,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.7301587301587301,
-      "eval_ORGANIZATION_recall": 0.7931034482758621,
-      "eval_PERSON_f1": 0.9003215434083601,
+      "eval_ORGANIZATION_precision": 0.7575757575757576,
+      "eval_ORGANIZATION_recall": 0.8620689655172413,
+      "eval_PERSON_f1": 0.8881789137380192,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8805031446540881,
-      "eval_PERSON_recall": 0.9210526315789473,
-      "eval_QUANTITY_f1": 0.819672131147541,
+      "eval_PERSON_precision": 0.8633540372670807,
+      "eval_PERSON_recall": 0.9144736842105263,
+      "eval_QUANTITY_f1": 0.8387096774193549,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.8064516129032258,
-      "eval_QUANTITY_recall": 0.8333333333333334,
-      "eval_TIME_f1": 0.8666666666666666,
+      "eval_QUANTITY_precision": 0.8125,
+      "eval_QUANTITY_recall": 0.8666666666666667,
+      "eval_TIME_f1": 0.896551724137931,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.8387096774193549,
+      "eval_TIME_precision": 0.896551724137931,
       "eval_TIME_recall": 0.896551724137931,
-      "eval_loss": 0.3222368061542511,
-      "eval_overall_accuracy": 0.9669654289372599,
-      "eval_overall_f1": 0.864942528735632,
-      "eval_overall_precision": 0.840782122905028,
-      "eval_overall_recall": 0.8905325443786982,
-      "eval_runtime": 0.8114,
-      "eval_samples_per_second": 230.457,
-      "eval_steps_per_second": 3.697,
+      "eval_loss": 0.3033183217048645,
+      "eval_overall_accuracy": 0.9661971830985916,
+      "eval_overall_f1": 0.8624113475177306,
+      "eval_overall_precision": 0.8283378746594006,
+      "eval_overall_recall": 0.8994082840236687,
+      "eval_runtime": 0.3215,
+      "eval_samples_per_second": 581.707,
+      "eval_steps_per_second": 9.332,
       "step": 6678
     },
     {
       "epoch": 64.0,
-      "grad_norm": 0.000210550642805174,
+      "grad_norm": 0.005429080221801996,
       "learning_rate": 1.8e-05,
-      "loss": 0.0003,
+      "loss": 0.0017,
       "step": 6784
     },
     {
       "epoch": 64.0,
-      "eval_LOCATION_f1": 0.8767123287671235,
+      "eval_LOCATION_f1": 0.8476821192052981,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8311688311688312,
+      "eval_LOCATION_precision": 0.7804878048780488,
       "eval_LOCATION_recall": 0.927536231884058,
-      "eval_ORGANIZATION_f1": 0.7642276422764227,
+      "eval_ORGANIZATION_f1": 0.7906976744186047,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.7230769230769231,
-      "eval_ORGANIZATION_recall": 0.8103448275862069,
-      "eval_PERSON_f1": 0.9003215434083601,
+      "eval_ORGANIZATION_precision": 0.7183098591549296,
+      "eval_ORGANIZATION_recall": 0.8793103448275862,
+      "eval_PERSON_f1": 0.9038461538461539,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8805031446540881,
-      "eval_PERSON_recall": 0.9210526315789473,
-      "eval_QUANTITY_f1": 0.8064516129032259,
+      "eval_PERSON_precision": 0.88125,
+      "eval_PERSON_recall": 0.9276315789473685,
+      "eval_QUANTITY_f1": 0.8253968253968254,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.78125,
-      "eval_QUANTITY_recall": 0.8333333333333334,
-      "eval_TIME_f1": 0.8666666666666666,
+      "eval_QUANTITY_precision": 0.7878787878787878,
+      "eval_QUANTITY_recall": 0.8666666666666667,
+      "eval_TIME_f1": 0.8813559322033899,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.8387096774193549,
+      "eval_TIME_precision": 0.8666666666666667,
       "eval_TIME_recall": 0.896551724137931,
-      "eval_loss": 0.32881781458854675,
-      "eval_overall_accuracy": 0.9661971830985916,
-      "eval_overall_f1": 0.8603988603988605,
-      "eval_overall_precision": 0.8296703296703297,
-      "eval_overall_recall": 0.893491124260355,
-      "eval_runtime": 0.8114,
-      "eval_samples_per_second": 230.476,
-      "eval_steps_per_second": 3.697,
+      "eval_loss": 0.30157363414764404,
+      "eval_overall_accuracy": 0.9667093469910372,
+      "eval_overall_f1": 0.8627450980392156,
+      "eval_overall_precision": 0.8191489361702128,
+      "eval_overall_recall": 0.9112426035502958,
+      "eval_runtime": 0.3301,
+      "eval_samples_per_second": 566.553,
+      "eval_steps_per_second": 9.089,
       "step": 6784
     },
     {
       "epoch": 65.0,
-      "grad_norm": 0.0021329650189727545,
+      "grad_norm": 0.0004326178168412298,
       "learning_rate": 1.75e-05,
-      "loss": 0.0003,
+      "loss": 0.0007,
       "step": 6890
     },
     {
       "epoch": 65.0,
-      "eval_LOCATION_f1": 0.8767123287671235,
+      "eval_LOCATION_f1": 0.8590604026845639,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8311688311688312,
+      "eval_LOCATION_precision": 0.8,
       "eval_LOCATION_recall": 0.927536231884058,
-      "eval_ORGANIZATION_f1": 0.7704918032786885,
+      "eval_ORGANIZATION_f1": 0.7751937984496124,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.734375,
-      "eval_ORGANIZATION_recall": 0.8103448275862069,
-      "eval_PERSON_f1": 0.8987341772151899,
+      "eval_ORGANIZATION_precision": 0.704225352112676,
+      "eval_ORGANIZATION_recall": 0.8620689655172413,
+      "eval_PERSON_f1": 0.9038461538461539,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8658536585365854,
-      "eval_PERSON_recall": 0.9342105263157895,
-      "eval_QUANTITY_f1": 0.8387096774193549,
+      "eval_PERSON_precision": 0.88125,
+      "eval_PERSON_recall": 0.9276315789473685,
+      "eval_QUANTITY_f1": 0.8253968253968254,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.8125,
+      "eval_QUANTITY_precision": 0.7878787878787878,
       "eval_QUANTITY_recall": 0.8666666666666667,
-      "eval_TIME_f1": 0.8524590163934426,
+      "eval_TIME_f1": 0.896551724137931,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.8125,
+      "eval_TIME_precision": 0.896551724137931,
       "eval_TIME_recall": 0.896551724137931,
-      "eval_loss": 0.33311548829078674,
+      "eval_loss": 0.3076171875,
       "eval_overall_accuracy": 0.9661971830985916,
-      "eval_overall_f1": 0.8628005657708628,
-      "eval_overall_precision": 0.8265582655826558,
-      "eval_overall_recall": 0.9023668639053254,
-      "eval_runtime": 0.8037,
-      "eval_samples_per_second": 232.685,
-      "eval_steps_per_second": 3.733,
+      "eval_overall_f1": 0.8635724331926863,
+      "eval_overall_precision": 0.8230563002680965,
+      "eval_overall_recall": 0.908284023668639,
+      "eval_runtime": 0.3217,
+      "eval_samples_per_second": 581.291,
+      "eval_steps_per_second": 9.326,
       "step": 6890
     },
     {
       "epoch": 66.0,
-      "grad_norm": 0.0006951324758119881,
+      "grad_norm": 0.001978162443265319,
       "learning_rate": 1.7000000000000003e-05,
-      "loss": 0.0003,
+      "loss": 0.0013,
       "step": 6996
     },
     {
       "epoch": 66.0,
-      "eval_LOCATION_f1": 0.881118881118881,
+      "eval_LOCATION_f1": 0.8571428571428572,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8513513513513513,
+      "eval_LOCATION_precision": 0.8076923076923077,
       "eval_LOCATION_recall": 0.9130434782608695,
-      "eval_ORGANIZATION_f1": 0.6666666666666667,
+      "eval_ORGANIZATION_f1": 0.7999999999999999,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.6981132075471698,
-      "eval_ORGANIZATION_recall": 0.6379310344827587,
-      "eval_PERSON_f1": 0.9090909090909091,
+      "eval_ORGANIZATION_precision": 0.746268656716418,
+      "eval_ORGANIZATION_recall": 0.8620689655172413,
+      "eval_PERSON_f1": 0.9096774193548387,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8974358974358975,
-      "eval_PERSON_recall": 0.9210526315789473,
+      "eval_PERSON_precision": 0.8924050632911392,
+      "eval_PERSON_recall": 0.9276315789473685,
       "eval_QUANTITY_f1": 0.8253968253968254,
       "eval_QUANTITY_number": 30,
       "eval_QUANTITY_precision": 0.7878787878787878,
       "eval_QUANTITY_recall": 0.8666666666666667,
-      "eval_TIME_f1": 0.8666666666666666,
+      "eval_TIME_f1": 0.8387096774193549,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.8387096774193549,
+      "eval_TIME_precision": 0.7878787878787878,
       "eval_TIME_recall": 0.896551724137931,
-      "eval_loss": 0.31618252396583557,
-      "eval_overall_accuracy": 0.9661971830985916,
-      "eval_overall_f1": 0.8525547445255474,
-      "eval_overall_precision": 0.8414985590778098,
-      "eval_overall_recall": 0.863905325443787,
-      "eval_runtime": 0.8125,
-      "eval_samples_per_second": 230.14,
-      "eval_steps_per_second": 3.692,
+      "eval_loss": 0.2962900996208191,
+      "eval_overall_accuracy": 0.9638924455825865,
+      "eval_overall_f1": 0.8656294200848657,
+      "eval_overall_precision": 0.8292682926829268,
+      "eval_overall_recall": 0.9053254437869822,
+      "eval_runtime": 0.3273,
+      "eval_samples_per_second": 571.366,
+      "eval_steps_per_second": 9.166,
       "step": 6996
     },
     {
       "epoch": 67.0,
-      "grad_norm": 0.0028227961156517267,
+      "grad_norm": 0.000865288486238569,
       "learning_rate": 1.65e-05,
-      "loss": 0.0024,
+      "loss": 0.0011,
       "step": 7102
     },
     {
       "epoch": 67.0,
-      "eval_LOCATION_f1": 0.863013698630137,
+      "eval_LOCATION_f1": 0.8571428571428572,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8181818181818182,
+      "eval_LOCATION_precision": 0.8076923076923077,
       "eval_LOCATION_recall": 0.9130434782608695,
-      "eval_ORGANIZATION_f1": 0.7438016528925621,
+      "eval_ORGANIZATION_f1": 0.7317073170731708,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.7142857142857143,
+      "eval_ORGANIZATION_precision": 0.6923076923076923,
       "eval_ORGANIZATION_recall": 0.7758620689655172,
-      "eval_PERSON_f1": 0.9114754098360655,
+      "eval_PERSON_f1": 0.8974358974358975,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.9084967320261438,
-      "eval_PERSON_recall": 0.9144736842105263,
-      "eval_QUANTITY_f1": 0.8064516129032259,
+      "eval_PERSON_precision": 0.875,
+      "eval_PERSON_recall": 0.9210526315789473,
+      "eval_QUANTITY_f1": 0.8253968253968254,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.78125,
-      "eval_QUANTITY_recall": 0.8333333333333334,
-      "eval_TIME_f1": 0.8524590163934426,
+      "eval_QUANTITY_precision": 0.7878787878787878,
+      "eval_QUANTITY_recall": 0.8666666666666667,
+      "eval_TIME_f1": 0.8666666666666666,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.8125,
+      "eval_TIME_precision": 0.8387096774193549,
       "eval_TIME_recall": 0.896551724137931,
-      "eval_loss": 0.3060024380683899,
-      "eval_overall_accuracy": 0.9672215108834827,
-      "eval_overall_f1": 0.8575539568345324,
-      "eval_overall_precision": 0.834733893557423,
-      "eval_overall_recall": 0.8816568047337278,
-      "eval_runtime": 0.8036,
-      "eval_samples_per_second": 232.709,
-      "eval_steps_per_second": 3.733,
+      "eval_loss": 0.3187943994998932,
+      "eval_overall_accuracy": 0.9633802816901409,
+      "eval_overall_f1": 0.851063829787234,
+      "eval_overall_precision": 0.8174386920980926,
+      "eval_overall_recall": 0.8875739644970414,
+      "eval_runtime": 0.3211,
+      "eval_samples_per_second": 582.391,
+      "eval_steps_per_second": 9.343,
       "step": 7102
     },
     {
       "epoch": 68.0,
-      "grad_norm": 0.0007019271142780781,
+      "grad_norm": 0.00021025155729148537,
       "learning_rate": 1.6000000000000003e-05,
-      "loss": 0.002,
+      "loss": 0.0009,
       "step": 7208
     },
     {
       "epoch": 68.0,
-      "eval_LOCATION_f1": 0.8904109589041096,
+      "eval_LOCATION_f1": 0.8648648648648648,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8441558441558441,
-      "eval_LOCATION_recall": 0.9420289855072463,
-      "eval_ORGANIZATION_f1": 0.6851851851851852,
+      "eval_LOCATION_precision": 0.810126582278481,
+      "eval_LOCATION_recall": 0.927536231884058,
+      "eval_ORGANIZATION_f1": 0.784,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.74,
-      "eval_ORGANIZATION_recall": 0.6379310344827587,
-      "eval_PERSON_f1": 0.9155844155844156,
+      "eval_ORGANIZATION_precision": 0.7313432835820896,
+      "eval_ORGANIZATION_recall": 0.8448275862068966,
+      "eval_PERSON_f1": 0.8987341772151899,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.9038461538461539,
-      "eval_PERSON_recall": 0.9276315789473685,
-      "eval_QUANTITY_f1": 0.8387096774193549,
+      "eval_PERSON_precision": 0.8658536585365854,
+      "eval_PERSON_recall": 0.9342105263157895,
+      "eval_QUANTITY_f1": 0.8253968253968254,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.8125,
+      "eval_QUANTITY_precision": 0.7878787878787878,
       "eval_QUANTITY_recall": 0.8666666666666667,
-      "eval_TIME_f1": 0.8387096774193549,
+      "eval_TIME_f1": 0.8666666666666666,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.7878787878787878,
+      "eval_TIME_precision": 0.8387096774193549,
       "eval_TIME_recall": 0.896551724137931,
-      "eval_loss": 0.3085281252861023,
-      "eval_overall_accuracy": 0.9651728553137003,
-      "eval_overall_f1": 0.8600583090379009,
-      "eval_overall_precision": 0.8477011494252874,
-      "eval_overall_recall": 0.8727810650887574,
-      "eval_runtime": 0.8099,
-      "eval_samples_per_second": 230.884,
-      "eval_steps_per_second": 3.704,
+      "eval_loss": 0.33562034368515015,
+      "eval_overall_accuracy": 0.964404609475032,
+      "eval_overall_f1": 0.8623595505617977,
+      "eval_overall_precision": 0.820855614973262,
+      "eval_overall_recall": 0.908284023668639,
+      "eval_runtime": 0.3272,
+      "eval_samples_per_second": 571.432,
+      "eval_steps_per_second": 9.167,
       "step": 7208
     },
     {
       "epoch": 69.0,
-      "grad_norm": 13.608760833740234,
+      "grad_norm": 0.032114915549755096,
       "learning_rate": 1.55e-05,
-      "loss": 0.0014,
+      "loss": 0.001,
       "step": 7314
     },
     {
       "epoch": 69.0,
-      "eval_LOCATION_f1": 0.8904109589041096,
+      "eval_LOCATION_f1": 0.8783783783783783,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8441558441558441,
+      "eval_LOCATION_precision": 0.8227848101265823,
       "eval_LOCATION_recall": 0.9420289855072463,
-      "eval_ORGANIZATION_f1": 0.7086614173228347,
+      "eval_ORGANIZATION_f1": 0.7377049180327868,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.6521739130434783,
+      "eval_ORGANIZATION_precision": 0.703125,
       "eval_ORGANIZATION_recall": 0.7758620689655172,
-      "eval_PERSON_f1": 0.89171974522293,
+      "eval_PERSON_f1": 0.9260450160771704,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8641975308641975,
-      "eval_PERSON_recall": 0.9210526315789473,
-      "eval_QUANTITY_f1": 0.888888888888889,
+      "eval_PERSON_precision": 0.9056603773584906,
+      "eval_PERSON_recall": 0.9473684210526315,
+      "eval_QUANTITY_f1": 0.8253968253968254,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.8484848484848485,
-      "eval_QUANTITY_recall": 0.9333333333333333,
-      "eval_TIME_f1": 0.8524590163934426,
-      "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.8125,
-      "eval_TIME_recall": 0.896551724137931,
-      "eval_loss": 0.30335918068885803,
-      "eval_overall_accuracy": 0.9649167733674776,
-      "eval_overall_f1": 0.8551336146272854,
-      "eval_overall_precision": 0.8150134048257373,
-      "eval_overall_recall": 0.8994082840236687,
-      "eval_runtime": 0.8074,
-      "eval_samples_per_second": 231.614,
-      "eval_steps_per_second": 3.716,
+      "eval_QUANTITY_precision": 0.7878787878787878,
+      "eval_QUANTITY_recall": 0.8666666666666667,
+      "eval_TIME_f1": 0.8666666666666666,
+      "eval_TIME_number": 29,
+      "eval_TIME_precision": 0.8387096774193549,
+      "eval_TIME_recall": 0.896551724137931,
+      "eval_loss": 0.30885857343673706,
+      "eval_overall_accuracy": 0.9654289372599232,
+      "eval_overall_f1": 0.8693181818181819,
+      "eval_overall_precision": 0.8360655737704918,
+      "eval_overall_recall": 0.9053254437869822,
+      "eval_runtime": 0.3216,
+      "eval_samples_per_second": 581.431,
+      "eval_steps_per_second": 9.328,
       "step": 7314
     },
     {
       "epoch": 70.0,
-      "grad_norm": 0.00710656400769949,
+      "grad_norm": 0.0254677664488554,
       "learning_rate": 1.5e-05,
-      "loss": 0.0006,
+      "loss": 0.001,
       "step": 7420
     },
     {
       "epoch": 70.0,
-      "eval_LOCATION_f1": 0.882758620689655,
+      "eval_LOCATION_f1": 0.870748299319728,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8421052631578947,
+      "eval_LOCATION_precision": 0.8205128205128205,
       "eval_LOCATION_recall": 0.927536231884058,
-      "eval_ORGANIZATION_f1": 0.7244094488188977,
+      "eval_ORGANIZATION_f1": 0.8091603053435115,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.6666666666666666,
-      "eval_ORGANIZATION_recall": 0.7931034482758621,
-      "eval_PERSON_f1": 0.888888888888889,
+      "eval_ORGANIZATION_precision": 0.726027397260274,
+      "eval_ORGANIZATION_recall": 0.9137931034482759,
+      "eval_PERSON_f1": 0.9102564102564101,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8588957055214724,
-      "eval_PERSON_recall": 0.9210526315789473,
-      "eval_QUANTITY_f1": 0.8387096774193549,
+      "eval_PERSON_precision": 0.8875,
+      "eval_PERSON_recall": 0.9342105263157895,
+      "eval_QUANTITY_f1": 0.8253968253968254,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.8125,
+      "eval_QUANTITY_precision": 0.7878787878787878,
       "eval_QUANTITY_recall": 0.8666666666666667,
-      "eval_TIME_f1": 0.8064516129032258,
+      "eval_TIME_f1": 0.8666666666666666,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.7575757575757576,
-      "eval_TIME_recall": 0.8620689655172413,
-      "eval_loss": 0.3125246465206146,
-      "eval_overall_accuracy": 0.9646606914212548,
-      "eval_overall_f1": 0.8466947960618846,
-      "eval_overall_precision": 0.806970509383378,
-      "eval_overall_recall": 0.8905325443786982,
-      "eval_runtime": 0.8058,
-      "eval_samples_per_second": 232.071,
-      "eval_steps_per_second": 3.723,
+      "eval_TIME_precision": 0.8387096774193549,
+      "eval_TIME_recall": 0.896551724137931,
+      "eval_loss": 0.3140678107738495,
+      "eval_overall_accuracy": 0.9669654289372599,
+      "eval_overall_f1": 0.8723702664796634,
+      "eval_overall_precision": 0.8293333333333334,
+      "eval_overall_recall": 0.9201183431952663,
+      "eval_runtime": 0.339,
+      "eval_samples_per_second": 551.686,
+      "eval_steps_per_second": 8.851,
       "step": 7420
     },
     {
       "epoch": 71.0,
-      "grad_norm": 0.00047848603571765125,
+      "grad_norm": 0.0020287446677684784,
       "learning_rate": 1.45e-05,
-      "loss": 0.0009,
+      "loss": 0.0006,
       "step": 7526
     },
     {
       "epoch": 71.0,
-      "eval_LOCATION_f1": 0.8749999999999999,
+      "eval_LOCATION_f1": 0.8648648648648648,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.84,
-      "eval_LOCATION_recall": 0.9130434782608695,
-      "eval_ORGANIZATION_f1": 0.7301587301587301,
+      "eval_LOCATION_precision": 0.810126582278481,
+      "eval_LOCATION_recall": 0.927536231884058,
+      "eval_ORGANIZATION_f1": 0.7777777777777778,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.6764705882352942,
-      "eval_ORGANIZATION_recall": 0.7931034482758621,
-      "eval_PERSON_f1": 0.8952380952380953,
+      "eval_ORGANIZATION_precision": 0.7205882352941176,
+      "eval_ORGANIZATION_recall": 0.8448275862068966,
+      "eval_PERSON_f1": 0.9320388349514562,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8650306748466258,
-      "eval_PERSON_recall": 0.9276315789473685,
-      "eval_QUANTITY_f1": 0.8125,
+      "eval_PERSON_precision": 0.9171974522292994,
+      "eval_PERSON_recall": 0.9473684210526315,
+      "eval_QUANTITY_f1": 0.8253968253968254,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.7647058823529411,
+      "eval_QUANTITY_precision": 0.7878787878787878,
       "eval_QUANTITY_recall": 0.8666666666666667,
-      "eval_TIME_f1": 0.7666666666666667,
+      "eval_TIME_f1": 0.8813559322033899,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.7419354838709677,
-      "eval_TIME_recall": 0.7931034482758621,
-      "eval_loss": 0.3196626901626587,
-      "eval_overall_accuracy": 0.9638924455825865,
-      "eval_overall_f1": 0.843441466854725,
-      "eval_overall_precision": 0.8059299191374663,
-      "eval_overall_recall": 0.8846153846153846,
-      "eval_runtime": 0.8027,
-      "eval_samples_per_second": 232.972,
-      "eval_steps_per_second": 3.738,
+      "eval_TIME_precision": 0.8666666666666667,
+      "eval_TIME_recall": 0.896551724137931,
+      "eval_loss": 0.2926411032676697,
+      "eval_overall_accuracy": 0.9682458386683739,
+      "eval_overall_f1": 0.876595744680851,
+      "eval_overall_precision": 0.8419618528610354,
+      "eval_overall_recall": 0.9142011834319527,
+      "eval_runtime": 0.3255,
+      "eval_samples_per_second": 574.564,
+      "eval_steps_per_second": 9.218,
       "step": 7526
     },
     {
       "epoch": 72.0,
-      "grad_norm": 0.002629638649523258,
+      "grad_norm": 0.0008131638751365244,
       "learning_rate": 1.4000000000000001e-05,
-      "loss": 0.0009,
+      "loss": 0.0003,
       "step": 7632
     },
     {
       "epoch": 72.0,
-      "eval_LOCATION_f1": 0.882758620689655,
+      "eval_LOCATION_f1": 0.8590604026845639,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8421052631578947,
+      "eval_LOCATION_precision": 0.8,
       "eval_LOCATION_recall": 0.927536231884058,
-      "eval_ORGANIZATION_f1": 0.7317073170731708,
+      "eval_ORGANIZATION_f1": 0.7999999999999999,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.6923076923076923,
-      "eval_ORGANIZATION_recall": 0.7758620689655172,
-      "eval_PERSON_f1": 0.9073482428115016,
+      "eval_ORGANIZATION_precision": 0.746268656716418,
+      "eval_ORGANIZATION_recall": 0.8620689655172413,
+      "eval_PERSON_f1": 0.9255663430420712,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8819875776397516,
-      "eval_PERSON_recall": 0.9342105263157895,
-      "eval_QUANTITY_f1": 0.8125,
+      "eval_PERSON_precision": 0.910828025477707,
+      "eval_PERSON_recall": 0.9407894736842105,
+      "eval_QUANTITY_f1": 0.8253968253968254,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.7647058823529411,
+      "eval_QUANTITY_precision": 0.7878787878787878,
       "eval_QUANTITY_recall": 0.8666666666666667,
-      "eval_TIME_f1": 0.7868852459016394,
+      "eval_TIME_f1": 0.8666666666666666,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.75,
-      "eval_TIME_recall": 0.8275862068965517,
-      "eval_loss": 0.3206939697265625,
-      "eval_overall_accuracy": 0.9646606914212548,
-      "eval_overall_f1": 0.8526912181303117,
-      "eval_overall_precision": 0.8179347826086957,
-      "eval_overall_recall": 0.8905325443786982,
-      "eval_runtime": 0.8794,
-      "eval_samples_per_second": 212.64,
-      "eval_steps_per_second": 3.411,
+      "eval_TIME_precision": 0.8387096774193549,
+      "eval_TIME_recall": 0.896551724137931,
+      "eval_loss": 0.316145658493042,
+      "eval_overall_accuracy": 0.9669654289372599,
+      "eval_overall_f1": 0.8753541076487252,
+      "eval_overall_precision": 0.8396739130434783,
+      "eval_overall_recall": 0.9142011834319527,
+      "eval_runtime": 0.3261,
+      "eval_samples_per_second": 573.373,
+      "eval_steps_per_second": 9.198,
       "step": 7632
     },
     {
       "epoch": 73.0,
-      "grad_norm": 0.0006528229569084942,
+      "grad_norm": 0.000352115195710212,
       "learning_rate": 1.3500000000000001e-05,
-      "loss": 0.0006,
+      "loss": 0.0008,
       "step": 7738
     },
     {
       "epoch": 73.0,
-      "eval_LOCATION_f1": 0.8767123287671235,
+      "eval_LOCATION_f1": 0.8533333333333333,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8311688311688312,
+      "eval_LOCATION_precision": 0.7901234567901234,
       "eval_LOCATION_recall": 0.927536231884058,
-      "eval_ORGANIZATION_f1": 0.7049180327868853,
+      "eval_ORGANIZATION_f1": 0.7936507936507937,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.671875,
-      "eval_ORGANIZATION_recall": 0.7413793103448276,
-      "eval_PERSON_f1": 0.9061488673139159,
+      "eval_ORGANIZATION_precision": 0.7352941176470589,
+      "eval_ORGANIZATION_recall": 0.8620689655172413,
+      "eval_PERSON_f1": 0.9102564102564101,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.89171974522293,
-      "eval_PERSON_recall": 0.9210526315789473,
+      "eval_PERSON_precision": 0.8875,
+      "eval_PERSON_recall": 0.9342105263157895,
       "eval_QUANTITY_f1": 0.8253968253968254,
       "eval_QUANTITY_number": 30,
       "eval_QUANTITY_precision": 0.7878787878787878,
       "eval_QUANTITY_recall": 0.8666666666666667,
-      "eval_TIME_f1": 0.8196721311475409,
+      "eval_TIME_f1": 0.8813559322033899,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.78125,
-      "eval_TIME_recall": 0.8620689655172413,
-      "eval_loss": 0.32062897086143494,
-      "eval_overall_accuracy": 0.9646606914212548,
-      "eval_overall_f1": 0.8502139800285307,
-      "eval_overall_precision": 0.8209366391184573,
-      "eval_overall_recall": 0.8816568047337278,
-      "eval_runtime": 0.8113,
-      "eval_samples_per_second": 230.503,
-      "eval_steps_per_second": 3.698,
+      "eval_TIME_precision": 0.8666666666666667,
+      "eval_TIME_recall": 0.896551724137931,
+      "eval_loss": 0.31928354501724243,
+      "eval_overall_accuracy": 0.9674775928297055,
+      "eval_overall_f1": 0.867605633802817,
+      "eval_overall_precision": 0.8279569892473119,
+      "eval_overall_recall": 0.9112426035502958,
+      "eval_runtime": 0.3197,
+      "eval_samples_per_second": 584.939,
+      "eval_steps_per_second": 9.384,
       "step": 7738
     },
     {
       "epoch": 74.0,
-      "grad_norm": 0.0003340624098200351,
+      "grad_norm": 0.0021537726279348135,
       "learning_rate": 1.3000000000000001e-05,
-      "loss": 0.0004,
+      "loss": 0.0007,
       "step": 7844
     },
     {
       "epoch": 74.0,
-      "eval_LOCATION_f1": 0.8767123287671235,
+      "eval_LOCATION_f1": 0.8590604026845639,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8311688311688312,
+      "eval_LOCATION_precision": 0.8,
       "eval_LOCATION_recall": 0.927536231884058,
-      "eval_ORGANIZATION_f1": 0.7213114754098361,
+      "eval_ORGANIZATION_f1": 0.7868852459016394,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.6875,
-      "eval_ORGANIZATION_recall": 0.7586206896551724,
-      "eval_PERSON_f1": 0.8990228013029316,
+      "eval_ORGANIZATION_precision": 0.75,
+      "eval_ORGANIZATION_recall": 0.8275862068965517,
+      "eval_PERSON_f1": 0.9131832797427653,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8903225806451613,
-      "eval_PERSON_recall": 0.9078947368421053,
+      "eval_PERSON_precision": 0.8930817610062893,
+      "eval_PERSON_recall": 0.9342105263157895,
       "eval_QUANTITY_f1": 0.8253968253968254,
       "eval_QUANTITY_number": 30,
       "eval_QUANTITY_precision": 0.7878787878787878,
       "eval_QUANTITY_recall": 0.8666666666666667,
-      "eval_TIME_f1": 0.8196721311475409,
+      "eval_TIME_f1": 0.8813559322033899,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.78125,
-      "eval_TIME_recall": 0.8620689655172413,
-      "eval_loss": 0.32088541984558105,
-      "eval_overall_accuracy": 0.9641485275288092,
-      "eval_overall_f1": 0.8497854077253219,
-      "eval_overall_precision": 0.8227146814404432,
-      "eval_overall_recall": 0.878698224852071,
-      "eval_runtime": 0.8194,
-      "eval_samples_per_second": 228.212,
-      "eval_steps_per_second": 3.661,
+      "eval_TIME_precision": 0.8666666666666667,
+      "eval_TIME_recall": 0.896551724137931,
+      "eval_loss": 0.3125959038734436,
+      "eval_overall_accuracy": 0.9661971830985916,
+      "eval_overall_f1": 0.8693181818181819,
+      "eval_overall_precision": 0.8360655737704918,
+      "eval_overall_recall": 0.9053254437869822,
+      "eval_runtime": 0.334,
+      "eval_samples_per_second": 559.859,
+      "eval_steps_per_second": 8.982,
       "step": 7844
     },
     {
       "epoch": 75.0,
-      "grad_norm": 0.0001975786144612357,
+      "grad_norm": 0.0009629224659875035,
       "learning_rate": 1.25e-05,
-      "loss": 0.0003,
+      "loss": 0.0002,
       "step": 7950
     },
     {
       "epoch": 75.0,
-      "eval_LOCATION_f1": 0.8936170212765957,
+      "eval_LOCATION_f1": 0.8648648648648648,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.875,
-      "eval_LOCATION_recall": 0.9130434782608695,
-      "eval_ORGANIZATION_f1": 0.6991869918699186,
+      "eval_LOCATION_precision": 0.810126582278481,
+      "eval_LOCATION_recall": 0.927536231884058,
+      "eval_ORGANIZATION_f1": 0.7999999999999999,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.6615384615384615,
-      "eval_ORGANIZATION_recall": 0.7413793103448276,
-      "eval_PERSON_f1": 0.8996763754045308,
+      "eval_ORGANIZATION_precision": 0.7741935483870968,
+      "eval_ORGANIZATION_recall": 0.8275862068965517,
+      "eval_PERSON_f1": 0.9032258064516129,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8853503184713376,
-      "eval_PERSON_recall": 0.9144736842105263,
+      "eval_PERSON_precision": 0.8860759493670886,
+      "eval_PERSON_recall": 0.9210526315789473,
       "eval_QUANTITY_f1": 0.8253968253968254,
       "eval_QUANTITY_number": 30,
       "eval_QUANTITY_precision": 0.7878787878787878,
       "eval_QUANTITY_recall": 0.8666666666666667,
-      "eval_TIME_f1": 0.8196721311475409,
+      "eval_TIME_f1": 0.8813559322033899,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.78125,
-      "eval_TIME_recall": 0.8620689655172413,
-      "eval_loss": 0.32191890478134155,
-      "eval_overall_accuracy": 0.9636363636363636,
-      "eval_overall_f1": 0.8493543758967002,
-      "eval_overall_precision": 0.8245125348189415,
-      "eval_overall_recall": 0.8757396449704142,
-      "eval_runtime": 0.7893,
-      "eval_samples_per_second": 236.92,
-      "eval_steps_per_second": 3.801,
+      "eval_TIME_precision": 0.8666666666666667,
+      "eval_TIME_recall": 0.896551724137931,
+      "eval_loss": 0.31075483560562134,
+      "eval_overall_accuracy": 0.9656850192061459,
+      "eval_overall_f1": 0.8685714285714285,
+      "eval_overall_precision": 0.8397790055248618,
+      "eval_overall_recall": 0.8994082840236687,
+      "eval_runtime": 0.3244,
+      "eval_samples_per_second": 576.373,
+      "eval_steps_per_second": 9.247,
       "step": 7950
     },
     {
       "epoch": 76.0,
-      "grad_norm": 0.0004086096887476742,
+      "grad_norm": 0.0005110618658363819,
       "learning_rate": 1.2e-05,
       "loss": 0.0003,
       "step": 8056
     },
     {
       "epoch": 76.0,
-      "eval_LOCATION_f1": 0.8767123287671235,
+      "eval_LOCATION_f1": 0.870748299319728,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8311688311688312,
+      "eval_LOCATION_precision": 0.8205128205128205,
       "eval_LOCATION_recall": 0.927536231884058,
-      "eval_ORGANIZATION_f1": 0.732824427480916,
+      "eval_ORGANIZATION_f1": 0.8160000000000001,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.6575342465753424,
-      "eval_ORGANIZATION_recall": 0.8275862068965517,
-      "eval_PERSON_f1": 0.9038461538461539,
+      "eval_ORGANIZATION_precision": 0.7611940298507462,
+      "eval_ORGANIZATION_recall": 0.8793103448275862,
+      "eval_PERSON_f1": 0.910828025477707,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.88125,
-      "eval_PERSON_recall": 0.9276315789473685,
+      "eval_PERSON_precision": 0.8827160493827161,
+      "eval_PERSON_recall": 0.9407894736842105,
       "eval_QUANTITY_f1": 0.8253968253968254,
       "eval_QUANTITY_number": 30,
       "eval_QUANTITY_precision": 0.7878787878787878,
       "eval_QUANTITY_recall": 0.8666666666666667,
-      "eval_TIME_f1": 0.8196721311475409,
+      "eval_TIME_f1": 0.8813559322033899,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.78125,
-      "eval_TIME_recall": 0.8620689655172413,
-      "eval_loss": 0.3355537950992584,
-      "eval_overall_accuracy": 0.963124199743918,
-      "eval_overall_f1": 0.85273492286115,
-      "eval_overall_precision": 0.8106666666666666,
-      "eval_overall_recall": 0.8994082840236687,
-      "eval_runtime": 0.8123,
-      "eval_samples_per_second": 230.209,
-      "eval_steps_per_second": 3.693,
+      "eval_TIME_precision": 0.8666666666666667,
+      "eval_TIME_recall": 0.896551724137931,
+      "eval_loss": 0.3103707730770111,
+      "eval_overall_accuracy": 0.9667093469910372,
+      "eval_overall_f1": 0.8757062146892656,
+      "eval_overall_precision": 0.8378378378378378,
+      "eval_overall_recall": 0.9171597633136095,
+      "eval_runtime": 0.3262,
+      "eval_samples_per_second": 573.268,
+      "eval_steps_per_second": 9.197,
       "step": 8056
     },
     {
       "epoch": 77.0,
-      "grad_norm": 0.000383577193133533,
+      "grad_norm": 0.0007651924388483167,
       "learning_rate": 1.1500000000000002e-05,
-      "loss": 0.0002,
+      "loss": 0.0003,
       "step": 8162
     },
     {
       "epoch": 77.0,
-      "eval_LOCATION_f1": 0.8951048951048951,
+      "eval_LOCATION_f1": 0.8648648648648648,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8648648648648649,
+      "eval_LOCATION_precision": 0.810126582278481,
       "eval_LOCATION_recall": 0.927536231884058,
-      "eval_ORGANIZATION_f1": 0.7166666666666668,
+      "eval_ORGANIZATION_f1": 0.8188976377952756,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.6935483870967742,
-      "eval_ORGANIZATION_recall": 0.7413793103448276,
-      "eval_PERSON_f1": 0.9003215434083601,
+      "eval_ORGANIZATION_precision": 0.7536231884057971,
+      "eval_ORGANIZATION_recall": 0.896551724137931,
+      "eval_PERSON_f1": 0.9196141479099678,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8805031446540881,
-      "eval_PERSON_recall": 0.9210526315789473,
-      "eval_QUANTITY_f1": 0.8387096774193549,
+      "eval_PERSON_precision": 0.89937106918239,
+      "eval_PERSON_recall": 0.9407894736842105,
+      "eval_QUANTITY_f1": 0.8253968253968254,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.8125,
+      "eval_QUANTITY_precision": 0.7878787878787878,
       "eval_QUANTITY_recall": 0.8666666666666667,
-      "eval_TIME_f1": 0.8196721311475409,
+      "eval_TIME_f1": 0.8813559322033899,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.78125,
-      "eval_TIME_recall": 0.8620689655172413,
-      "eval_loss": 0.3264581859111786,
-      "eval_overall_accuracy": 0.9656850192061459,
-      "eval_overall_f1": 0.8550932568149211,
-      "eval_overall_precision": 0.83008356545961,
-      "eval_overall_recall": 0.8816568047337278,
-      "eval_runtime": 0.8006,
-      "eval_samples_per_second": 233.575,
-      "eval_steps_per_second": 3.747,
+      "eval_TIME_precision": 0.8666666666666667,
+      "eval_TIME_recall": 0.896551724137931,
+      "eval_loss": 0.3196667730808258,
+      "eval_overall_accuracy": 0.9682458386683739,
+      "eval_overall_f1": 0.8785310734463276,
+      "eval_overall_precision": 0.8405405405405405,
+      "eval_overall_recall": 0.9201183431952663,
+      "eval_runtime": 0.3219,
+      "eval_samples_per_second": 580.999,
+      "eval_steps_per_second": 9.321,
       "step": 8162
     },
     {
       "epoch": 78.0,
-      "grad_norm": 0.0004785690689459443,
+      "grad_norm": 0.0005060500116087496,
       "learning_rate": 1.1000000000000001e-05,
-      "loss": 0.0002,
+      "loss": 0.0005,
       "step": 8268
     },
     {
       "epoch": 78.0,
-      "eval_LOCATION_f1": 0.888888888888889,
+      "eval_LOCATION_f1": 0.8311688311688311,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8533333333333334,
+      "eval_LOCATION_precision": 0.7529411764705882,
       "eval_LOCATION_recall": 0.927536231884058,
-      "eval_ORGANIZATION_f1": 0.7213114754098361,
+      "eval_ORGANIZATION_f1": 0.7846153846153847,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.6875,
-      "eval_ORGANIZATION_recall": 0.7586206896551724,
-      "eval_PERSON_f1": 0.9003215434083601,
+      "eval_ORGANIZATION_precision": 0.7083333333333334,
+      "eval_ORGANIZATION_recall": 0.8793103448275862,
+      "eval_PERSON_f1": 0.9015873015873016,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8805031446540881,
-      "eval_PERSON_recall": 0.9210526315789473,
+      "eval_PERSON_precision": 0.8711656441717791,
+      "eval_PERSON_recall": 0.9342105263157895,
       "eval_QUANTITY_f1": 0.8253968253968254,
       "eval_QUANTITY_number": 30,
       "eval_QUANTITY_precision": 0.7878787878787878,
       "eval_QUANTITY_recall": 0.8666666666666667,
-      "eval_TIME_f1": 0.8196721311475409,
+      "eval_TIME_f1": 0.8666666666666666,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.78125,
-      "eval_TIME_recall": 0.8620689655172413,
-      "eval_loss": 0.32919907569885254,
-      "eval_overall_accuracy": 0.9649167733674776,
-      "eval_overall_f1": 0.8530670470756063,
-      "eval_overall_precision": 0.8236914600550964,
-      "eval_overall_recall": 0.8846153846153846,
-      "eval_runtime": 0.7937,
-      "eval_samples_per_second": 235.609,
-      "eval_steps_per_second": 3.78,
+      "eval_TIME_precision": 0.8387096774193549,
+      "eval_TIME_recall": 0.896551724137931,
+      "eval_loss": 0.3422912657260895,
+      "eval_overall_accuracy": 0.9636363636363636,
+      "eval_overall_f1": 0.8559556786703602,
+      "eval_overall_precision": 0.8046875,
+      "eval_overall_recall": 0.9142011834319527,
+      "eval_runtime": 0.3309,
+      "eval_samples_per_second": 565.193,
+      "eval_steps_per_second": 9.067,
       "step": 8268
     },
     {
       "epoch": 79.0,
-      "grad_norm": 0.8365729451179504,
+      "grad_norm": 0.0006566674564965069,
       "learning_rate": 1.05e-05,
-      "loss": 0.0003,
+      "loss": 0.0002,
       "step": 8374
     },
     {
       "epoch": 79.0,
-      "eval_LOCATION_f1": 0.888888888888889,
+      "eval_LOCATION_f1": 0.870748299319728,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8533333333333334,
+      "eval_LOCATION_precision": 0.8205128205128205,
       "eval_LOCATION_recall": 0.927536231884058,
-      "eval_ORGANIZATION_f1": 0.7154471544715447,
+      "eval_ORGANIZATION_f1": 0.8253968253968255,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.676923076923077,
-      "eval_ORGANIZATION_recall": 0.7586206896551724,
-      "eval_PERSON_f1": 0.891025641025641,
+      "eval_ORGANIZATION_precision": 0.7647058823529411,
+      "eval_ORGANIZATION_recall": 0.896551724137931,
+      "eval_PERSON_f1": 0.9073482428115016,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.86875,
-      "eval_PERSON_recall": 0.9144736842105263,
-      "eval_QUANTITY_f1": 0.8387096774193549,
+      "eval_PERSON_precision": 0.8819875776397516,
+      "eval_PERSON_recall": 0.9342105263157895,
+      "eval_QUANTITY_f1": 0.8253968253968254,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.8125,
+      "eval_QUANTITY_precision": 0.7878787878787878,
       "eval_QUANTITY_recall": 0.8666666666666667,
-      "eval_TIME_f1": 0.8196721311475409,
+      "eval_TIME_f1": 0.8813559322033899,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.78125,
-      "eval_TIME_recall": 0.8620689655172413,
-      "eval_loss": 0.333362877368927,
-      "eval_overall_accuracy": 0.9646606914212548,
-      "eval_overall_f1": 0.8490028490028488,
-      "eval_overall_precision": 0.8186813186813187,
-      "eval_overall_recall": 0.8816568047337278,
-      "eval_runtime": 0.8323,
-      "eval_samples_per_second": 224.68,
-      "eval_steps_per_second": 3.604,
+      "eval_TIME_precision": 0.8666666666666667,
+      "eval_TIME_recall": 0.896551724137931,
+      "eval_loss": 0.33246731758117676,
+      "eval_overall_accuracy": 0.967989756722151,
+      "eval_overall_f1": 0.8757062146892656,
+      "eval_overall_precision": 0.8378378378378378,
+      "eval_overall_recall": 0.9171597633136095,
+      "eval_runtime": 0.3216,
+      "eval_samples_per_second": 581.551,
+      "eval_steps_per_second": 9.33,
       "step": 8374
     },
     {
       "epoch": 80.0,
-      "grad_norm": 0.0004933515447191894,
+      "grad_norm": 0.00022042440832592547,
       "learning_rate": 1e-05,
-      "loss": 0.0006,
+      "loss": 0.0002,
       "step": 8480
     },
     {
       "epoch": 80.0,
-      "eval_LOCATION_f1": 0.8951048951048951,
+      "eval_LOCATION_f1": 0.8767123287671235,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8648648648648649,
+      "eval_LOCATION_precision": 0.8311688311688312,
       "eval_LOCATION_recall": 0.927536231884058,
-      "eval_ORGANIZATION_f1": 0.6890756302521007,
+      "eval_ORGANIZATION_f1": 0.7256637168141592,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.6721311475409836,
+      "eval_ORGANIZATION_precision": 0.7454545454545455,
       "eval_ORGANIZATION_recall": 0.7068965517241379,
-      "eval_PERSON_f1": 0.9055374592833876,
+      "eval_PERSON_f1": 0.9073482428115016,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.896774193548387,
-      "eval_PERSON_recall": 0.9144736842105263,
-      "eval_QUANTITY_f1": 0.870967741935484,
+      "eval_PERSON_precision": 0.8819875776397516,
+      "eval_PERSON_recall": 0.9342105263157895,
+      "eval_QUANTITY_f1": 0.8253968253968254,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.84375,
-      "eval_QUANTITY_recall": 0.9,
-      "eval_TIME_f1": 0.8196721311475409,
+      "eval_QUANTITY_precision": 0.7878787878787878,
+      "eval_QUANTITY_recall": 0.8666666666666667,
+      "eval_TIME_f1": 0.8813559322033899,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.78125,
-      "eval_TIME_recall": 0.8620689655172413,
-      "eval_loss": 0.33931657671928406,
-      "eval_overall_accuracy": 0.9646606914212548,
-      "eval_overall_f1": 0.8554913294797689,
-      "eval_overall_precision": 0.8361581920903954,
-      "eval_overall_recall": 0.8757396449704142,
-      "eval_runtime": 0.8001,
-      "eval_samples_per_second": 233.722,
-      "eval_steps_per_second": 3.75,
+      "eval_TIME_precision": 0.8666666666666667,
+      "eval_TIME_recall": 0.896551724137931,
+      "eval_loss": 0.33535081148147583,
+      "eval_overall_accuracy": 0.9654289372599232,
+      "eval_overall_f1": 0.861671469740634,
+      "eval_overall_precision": 0.8398876404494382,
+      "eval_overall_recall": 0.8846153846153846,
+      "eval_runtime": 0.3346,
+      "eval_samples_per_second": 558.874,
+      "eval_steps_per_second": 8.966,
       "step": 8480
     },
     {
       "epoch": 81.0,
-      "grad_norm": 0.0009226278634741902,
+      "grad_norm": 0.0007291326764971018,
       "learning_rate": 9.5e-06,
-      "loss": 0.0002,
+      "loss": 0.0004,
       "step": 8586
     },
     {
       "epoch": 81.0,
-      "eval_LOCATION_f1": 0.888888888888889,
+      "eval_LOCATION_f1": 0.8456375838926175,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8533333333333334,
-      "eval_LOCATION_recall": 0.927536231884058,
-      "eval_ORGANIZATION_f1": 0.7213114754098361,
+      "eval_LOCATION_precision": 0.7875,
+      "eval_LOCATION_recall": 0.9130434782608695,
+      "eval_ORGANIZATION_f1": 0.7741935483870968,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.6875,
-      "eval_ORGANIZATION_recall": 0.7586206896551724,
-      "eval_PERSON_f1": 0.8990228013029316,
+      "eval_ORGANIZATION_precision": 0.7272727272727273,
+      "eval_ORGANIZATION_recall": 0.8275862068965517,
+      "eval_PERSON_f1": 0.8974358974358975,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8903225806451613,
-      "eval_PERSON_recall": 0.9078947368421053,
+      "eval_PERSON_precision": 0.875,
+      "eval_PERSON_recall": 0.9210526315789473,
       "eval_QUANTITY_f1": 0.8253968253968254,
       "eval_QUANTITY_number": 30,
       "eval_QUANTITY_precision": 0.7878787878787878,
       "eval_QUANTITY_recall": 0.8666666666666667,
-      "eval_TIME_f1": 0.8196721311475409,
+      "eval_TIME_f1": 0.8666666666666666,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.78125,
-      "eval_TIME_recall": 0.8620689655172413,
-      "eval_loss": 0.3536657691001892,
-      "eval_overall_accuracy": 0.964404609475032,
-      "eval_overall_f1": 0.8522238163558107,
-      "eval_overall_precision": 0.8272980501392758,
-      "eval_overall_recall": 0.878698224852071,
-      "eval_runtime": 0.8023,
-      "eval_samples_per_second": 233.087,
-      "eval_steps_per_second": 3.739,
+      "eval_TIME_precision": 0.8387096774193549,
+      "eval_TIME_recall": 0.896551724137931,
+      "eval_loss": 0.3405434489250183,
+      "eval_overall_accuracy": 0.9641485275288092,
+      "eval_overall_f1": 0.8559322033898306,
+      "eval_overall_precision": 0.8189189189189189,
+      "eval_overall_recall": 0.8964497041420119,
+      "eval_runtime": 0.3228,
+      "eval_samples_per_second": 579.316,
+      "eval_steps_per_second": 9.294,
       "step": 8586
     },
     {
       "epoch": 82.0,
-      "grad_norm": 0.001014114823192358,
+      "grad_norm": 0.00018406176241114736,
       "learning_rate": 9e-06,
-      "loss": 0.0003,
+      "loss": 0.0002,
       "step": 8692
     },
     {
       "epoch": 82.0,
-      "eval_LOCATION_f1": 0.888888888888889,
+      "eval_LOCATION_f1": 0.8767123287671235,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8533333333333334,
+      "eval_LOCATION_precision": 0.8311688311688312,
       "eval_LOCATION_recall": 0.927536231884058,
-      "eval_ORGANIZATION_f1": 0.719298245614035,
+      "eval_ORGANIZATION_f1": 0.7563025210084033,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.7321428571428571,
-      "eval_ORGANIZATION_recall": 0.7068965517241379,
-      "eval_PERSON_f1": 0.8961038961038961,
+      "eval_ORGANIZATION_precision": 0.7377049180327869,
+      "eval_ORGANIZATION_recall": 0.7758620689655172,
+      "eval_PERSON_f1": 0.9102564102564101,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8846153846153846,
-      "eval_PERSON_recall": 0.9078947368421053,
-      "eval_QUANTITY_f1": 0.8064516129032259,
+      "eval_PERSON_precision": 0.8875,
+      "eval_PERSON_recall": 0.9342105263157895,
+      "eval_QUANTITY_f1": 0.8253968253968254,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.78125,
-      "eval_QUANTITY_recall": 0.8333333333333334,
-      "eval_TIME_f1": 0.8196721311475409,
+      "eval_QUANTITY_precision": 0.7878787878787878,
+      "eval_QUANTITY_recall": 0.8666666666666667,
+      "eval_TIME_f1": 0.8813559322033899,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.78125,
-      "eval_TIME_recall": 0.8620689655172413,
-      "eval_loss": 0.3606484830379486,
-      "eval_overall_accuracy": 0.9638924455825865,
-      "eval_overall_f1": 0.8505079825834543,
-      "eval_overall_precision": 0.8347578347578347,
-      "eval_overall_recall": 0.8668639053254438,
-      "eval_runtime": 0.8129,
-      "eval_samples_per_second": 230.048,
-      "eval_steps_per_second": 3.691,
+      "eval_TIME_precision": 0.8666666666666667,
+      "eval_TIME_recall": 0.896551724137931,
+      "eval_loss": 0.33905231952667236,
+      "eval_overall_accuracy": 0.9664532650448143,
+      "eval_overall_f1": 0.8669527896995709,
+      "eval_overall_precision": 0.8393351800554016,
+      "eval_overall_recall": 0.8964497041420119,
+      "eval_runtime": 0.3217,
+      "eval_samples_per_second": 581.317,
+      "eval_steps_per_second": 9.326,
       "step": 8692
     },
     {
       "epoch": 83.0,
-      "grad_norm": 0.0005941664567217231,
+      "grad_norm": 0.0005490140756592155,
       "learning_rate": 8.500000000000002e-06,
-      "loss": 0.0006,
+      "loss": 0.0003,
       "step": 8798
     },
     {
       "epoch": 83.0,
-      "eval_LOCATION_f1": 0.9090909090909092,
+      "eval_LOCATION_f1": 0.8648648648648648,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8783783783783784,
-      "eval_LOCATION_recall": 0.9420289855072463,
-      "eval_ORGANIZATION_f1": 0.6545454545454545,
+      "eval_LOCATION_precision": 0.810126582278481,
+      "eval_LOCATION_recall": 0.927536231884058,
+      "eval_ORGANIZATION_f1": 0.7479674796747967,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.6923076923076923,
-      "eval_ORGANIZATION_recall": 0.6206896551724138,
-      "eval_PERSON_f1": 0.9055374592833876,
+      "eval_ORGANIZATION_precision": 0.7076923076923077,
+      "eval_ORGANIZATION_recall": 0.7931034482758621,
+      "eval_PERSON_f1": 0.9038461538461539,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.896774193548387,
-      "eval_PERSON_recall": 0.9144736842105263,
-      "eval_QUANTITY_f1": 0.8064516129032259,
+      "eval_PERSON_precision": 0.88125,
+      "eval_PERSON_recall": 0.9276315789473685,
+      "eval_QUANTITY_f1": 0.8253968253968254,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.78125,
-      "eval_QUANTITY_recall": 0.8333333333333334,
-      "eval_TIME_f1": 0.8196721311475409,
+      "eval_QUANTITY_precision": 0.7878787878787878,
+      "eval_QUANTITY_recall": 0.8666666666666667,
+      "eval_TIME_f1": 0.8813559322033899,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.78125,
-      "eval_TIME_recall": 0.8620689655172413,
-      "eval_loss": 0.35767465829849243,
-      "eval_overall_accuracy": 0.9641485275288092,
-      "eval_overall_f1": 0.8491947291361641,
-      "eval_overall_precision": 0.8405797101449275,
-      "eval_overall_recall": 0.8579881656804734,
-      "eval_runtime": 0.8106,
-      "eval_samples_per_second": 230.7,
-      "eval_steps_per_second": 3.701,
+      "eval_TIME_precision": 0.8666666666666667,
+      "eval_TIME_recall": 0.896551724137931,
+      "eval_loss": 0.348821759223938,
+      "eval_overall_accuracy": 0.9636363636363636,
+      "eval_overall_f1": 0.8595744680851063,
+      "eval_overall_precision": 0.8256130790190735,
+      "eval_overall_recall": 0.8964497041420119,
+      "eval_runtime": 0.3214,
+      "eval_samples_per_second": 581.755,
+      "eval_steps_per_second": 9.333,
       "step": 8798
     },
     {
       "epoch": 84.0,
-      "grad_norm": 0.0007156149367801845,
+      "grad_norm": 0.0003215703763999045,
       "learning_rate": 8.000000000000001e-06,
-      "loss": 0.001,
+      "loss": 0.0002,
       "step": 8904
     },
     {
       "epoch": 84.0,
-      "eval_LOCATION_f1": 0.9027777777777778,
+      "eval_LOCATION_f1": 0.870748299319728,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8666666666666667,
-      "eval_LOCATION_recall": 0.9420289855072463,
-      "eval_ORGANIZATION_f1": 0.6548672566371682,
+      "eval_LOCATION_precision": 0.8205128205128205,
+      "eval_LOCATION_recall": 0.927536231884058,
+      "eval_ORGANIZATION_f1": 0.7563025210084033,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.6727272727272727,
-      "eval_ORGANIZATION_recall": 0.6379310344827587,
-      "eval_PERSON_f1": 0.8961038961038961,
+      "eval_ORGANIZATION_precision": 0.7377049180327869,
+      "eval_ORGANIZATION_recall": 0.7758620689655172,
+      "eval_PERSON_f1": 0.9137380191693291,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8846153846153846,
-      "eval_PERSON_recall": 0.9078947368421053,
-      "eval_QUANTITY_f1": 0.8064516129032259,
+      "eval_PERSON_precision": 0.8881987577639752,
+      "eval_PERSON_recall": 0.9407894736842105,
+      "eval_QUANTITY_f1": 0.8253968253968254,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.78125,
-      "eval_QUANTITY_recall": 0.8333333333333334,
-      "eval_TIME_f1": 0.8196721311475409,
+      "eval_QUANTITY_precision": 0.7878787878787878,
+      "eval_QUANTITY_recall": 0.8666666666666667,
+      "eval_TIME_f1": 0.8813559322033899,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.78125,
-      "eval_TIME_recall": 0.8620689655172413,
-      "eval_loss": 0.36741259694099426,
-      "eval_overall_accuracy": 0.9618437900128041,
-      "eval_overall_f1": 0.8430232558139535,
-      "eval_overall_precision": 0.8285714285714286,
-      "eval_overall_recall": 0.8579881656804734,
-      "eval_runtime": 0.799,
-      "eval_samples_per_second": 234.039,
-      "eval_steps_per_second": 3.755,
+      "eval_TIME_precision": 0.8666666666666667,
+      "eval_TIME_recall": 0.896551724137931,
+      "eval_loss": 0.343229740858078,
+      "eval_overall_accuracy": 0.9651728553137003,
+      "eval_overall_f1": 0.8673323823109843,
+      "eval_overall_precision": 0.837465564738292,
+      "eval_overall_recall": 0.8994082840236687,
+      "eval_runtime": 0.3342,
+      "eval_samples_per_second": 559.576,
+      "eval_steps_per_second": 8.977,
       "step": 8904
     },
     {
       "epoch": 85.0,
-      "grad_norm": 0.00027448913897387683,
+      "grad_norm": 0.00015092955436557531,
       "learning_rate": 7.5e-06,
-      "loss": 0.0006,
+      "loss": 0.0003,
       "step": 9010
     },
     {
       "epoch": 85.0,
-      "eval_LOCATION_f1": 0.9090909090909092,
+      "eval_LOCATION_f1": 0.8590604026845639,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8783783783783784,
-      "eval_LOCATION_recall": 0.9420289855072463,
-      "eval_ORGANIZATION_f1": 0.6725663716814159,
+      "eval_LOCATION_precision": 0.8,
+      "eval_LOCATION_recall": 0.927536231884058,
+      "eval_ORGANIZATION_f1": 0.7804878048780489,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.6909090909090909,
-      "eval_ORGANIZATION_recall": 0.6551724137931034,
-      "eval_PERSON_f1": 0.8932038834951456,
+      "eval_ORGANIZATION_precision": 0.7384615384615385,
+      "eval_ORGANIZATION_recall": 0.8275862068965517,
+      "eval_PERSON_f1": 0.9166666666666667,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8789808917197452,
-      "eval_PERSON_recall": 0.9078947368421053,
-      "eval_QUANTITY_f1": 0.8387096774193549,
+      "eval_PERSON_precision": 0.89375,
+      "eval_PERSON_recall": 0.9407894736842105,
+      "eval_QUANTITY_f1": 0.8253968253968254,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.8125,
+      "eval_QUANTITY_precision": 0.7878787878787878,
       "eval_QUANTITY_recall": 0.8666666666666667,
-      "eval_TIME_f1": 0.8196721311475409,
+      "eval_TIME_f1": 0.8813559322033899,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.78125,
-      "eval_TIME_recall": 0.8620689655172413,
-      "eval_loss": 0.36387643218040466,
-      "eval_overall_accuracy": 0.9615877080665813,
-      "eval_overall_f1": 0.8488372093023255,
-      "eval_overall_precision": 0.8342857142857143,
-      "eval_overall_recall": 0.863905325443787,
-      "eval_runtime": 0.8034,
-      "eval_samples_per_second": 232.767,
-      "eval_steps_per_second": 3.734,
+      "eval_TIME_precision": 0.8666666666666667,
+      "eval_TIME_recall": 0.896551724137931,
+      "eval_loss": 0.34420961141586304,
+      "eval_overall_accuracy": 0.9654289372599232,
+      "eval_overall_f1": 0.8696883852691217,
+      "eval_overall_precision": 0.8342391304347826,
+      "eval_overall_recall": 0.908284023668639,
+      "eval_runtime": 0.3181,
+      "eval_samples_per_second": 587.95,
+      "eval_steps_per_second": 9.432,
       "step": 9010
     },
     {
       "epoch": 86.0,
-      "grad_norm": 0.0002620484447106719,
+      "grad_norm": 0.0005039689131081104,
       "learning_rate": 7.000000000000001e-06,
-      "loss": 0.0004,
+      "loss": 0.0005,
       "step": 9116
     },
     {
       "epoch": 86.0,
-      "eval_LOCATION_f1": 0.8936170212765957,
+      "eval_LOCATION_f1": 0.8648648648648648,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.875,
-      "eval_LOCATION_recall": 0.9130434782608695,
-      "eval_ORGANIZATION_f1": 0.6851851851851852,
+      "eval_LOCATION_precision": 0.810126582278481,
+      "eval_LOCATION_recall": 0.927536231884058,
+      "eval_ORGANIZATION_f1": 0.7903225806451614,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.74,
-      "eval_ORGANIZATION_recall": 0.6379310344827587,
-      "eval_PERSON_f1": 0.8996763754045308,
+      "eval_ORGANIZATION_precision": 0.7424242424242424,
+      "eval_ORGANIZATION_recall": 0.8448275862068966,
+      "eval_PERSON_f1": 0.9196141479099678,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8853503184713376,
-      "eval_PERSON_recall": 0.9144736842105263,
-      "eval_QUANTITY_f1": 0.8387096774193549,
+      "eval_PERSON_precision": 0.89937106918239,
+      "eval_PERSON_recall": 0.9407894736842105,
+      "eval_QUANTITY_f1": 0.8253968253968254,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.8125,
+      "eval_QUANTITY_precision": 0.7878787878787878,
       "eval_QUANTITY_recall": 0.8666666666666667,
-      "eval_TIME_f1": 0.8196721311475409,
+      "eval_TIME_f1": 0.8813559322033899,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.78125,
-      "eval_TIME_recall": 0.8620689655172413,
-      "eval_loss": 0.3683052957057953,
-      "eval_overall_accuracy": 0.9633802816901409,
-      "eval_overall_f1": 0.8516886930983848,
-      "eval_overall_precision": 0.8454810495626822,
-      "eval_overall_recall": 0.8579881656804734,
-      "eval_runtime": 0.796,
-      "eval_samples_per_second": 234.923,
-      "eval_steps_per_second": 3.769,
+      "eval_TIME_precision": 0.8666666666666667,
+      "eval_TIME_recall": 0.896551724137931,
+      "eval_loss": 0.3417365849018097,
+      "eval_overall_accuracy": 0.9661971830985916,
+      "eval_overall_f1": 0.873758865248227,
+      "eval_overall_precision": 0.8392370572207084,
+      "eval_overall_recall": 0.9112426035502958,
+      "eval_runtime": 0.3234,
+      "eval_samples_per_second": 578.156,
+      "eval_steps_per_second": 9.275,
       "step": 9116
     },
     {
       "epoch": 87.0,
-      "grad_norm": 0.00031526130624115467,
+      "grad_norm": 0.00023389511625282466,
       "learning_rate": 6.5000000000000004e-06,
-      "loss": 0.0005,
+      "loss": 0.0002,
       "step": 9222
     },
     {
       "epoch": 87.0,
-      "eval_LOCATION_f1": 0.888888888888889,
+      "eval_LOCATION_f1": 0.8648648648648648,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8533333333333334,
+      "eval_LOCATION_precision": 0.810126582278481,
       "eval_LOCATION_recall": 0.927536231884058,
-      "eval_ORGANIZATION_f1": 0.6846846846846846,
+      "eval_ORGANIZATION_f1": 0.7777777777777778,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.7169811320754716,
-      "eval_ORGANIZATION_recall": 0.6551724137931034,
-      "eval_PERSON_f1": 0.8961038961038961,
+      "eval_ORGANIZATION_precision": 0.7205882352941176,
+      "eval_ORGANIZATION_recall": 0.8448275862068966,
+      "eval_PERSON_f1": 0.9166666666666667,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8846153846153846,
-      "eval_PERSON_recall": 0.9078947368421053,
-      "eval_QUANTITY_f1": 0.8387096774193549,
+      "eval_PERSON_precision": 0.89375,
+      "eval_PERSON_recall": 0.9407894736842105,
+      "eval_QUANTITY_f1": 0.8253968253968254,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.8125,
+      "eval_QUANTITY_precision": 0.7878787878787878,
       "eval_QUANTITY_recall": 0.8666666666666667,
-      "eval_TIME_f1": 0.8196721311475409,
+      "eval_TIME_f1": 0.8666666666666666,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.78125,
-      "eval_TIME_recall": 0.8620689655172413,
-      "eval_loss": 0.35311439633369446,
-      "eval_overall_accuracy": 0.9626120358514725,
-      "eval_overall_f1": 0.8483965014577258,
-      "eval_overall_precision": 0.8362068965517241,
-      "eval_overall_recall": 0.8609467455621301,
-      "eval_runtime": 0.7988,
-      "eval_samples_per_second": 234.112,
-      "eval_steps_per_second": 3.756,
+      "eval_TIME_precision": 0.8387096774193549,
+      "eval_TIME_recall": 0.896551724137931,
+      "eval_loss": 0.3444933593273163,
+      "eval_overall_accuracy": 0.9654289372599232,
+      "eval_overall_f1": 0.8688293370944994,
+      "eval_overall_precision": 0.8301886792452831,
+      "eval_overall_recall": 0.9112426035502958,
+      "eval_runtime": 0.319,
+      "eval_samples_per_second": 586.225,
+      "eval_steps_per_second": 9.405,
       "step": 9222
     },
     {
       "epoch": 88.0,
-      "grad_norm": 0.0015132308471947908,
+      "grad_norm": 0.0009348023450002074,
       "learning_rate": 6e-06,
       "loss": 0.0002,
       "step": 9328
     },
     {
       "epoch": 88.0,
-      "eval_LOCATION_f1": 0.881118881118881,
+      "eval_LOCATION_f1": 0.8648648648648648,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8513513513513513,
-      "eval_LOCATION_recall": 0.9130434782608695,
-      "eval_ORGANIZATION_f1": 0.6909090909090909,
+      "eval_LOCATION_precision": 0.810126582278481,
+      "eval_LOCATION_recall": 0.927536231884058,
+      "eval_ORGANIZATION_f1": 0.7777777777777778,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.7307692307692307,
-      "eval_ORGANIZATION_recall": 0.6551724137931034,
-      "eval_PERSON_f1": 0.9025974025974026,
+      "eval_ORGANIZATION_precision": 0.7205882352941176,
+      "eval_ORGANIZATION_recall": 0.8448275862068966,
+      "eval_PERSON_f1": 0.9131832797427653,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8910256410256411,
-      "eval_PERSON_recall": 0.9144736842105263,
-      "eval_QUANTITY_f1": 0.8387096774193549,
+      "eval_PERSON_precision": 0.8930817610062893,
+      "eval_PERSON_recall": 0.9342105263157895,
+      "eval_QUANTITY_f1": 0.8253968253968254,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.8125,
+      "eval_QUANTITY_precision": 0.7878787878787878,
       "eval_QUANTITY_recall": 0.8666666666666667,
-      "eval_TIME_f1": 0.8196721311475409,
+      "eval_TIME_f1": 0.8666666666666666,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.78125,
-      "eval_TIME_recall": 0.8620689655172413,
-      "eval_loss": 0.3533298671245575,
-      "eval_overall_accuracy": 0.9633802816901409,
-      "eval_overall_f1": 0.8508771929824561,
-      "eval_overall_precision": 0.8410404624277457,
-      "eval_overall_recall": 0.8609467455621301,
-      "eval_runtime": 0.8072,
-      "eval_samples_per_second": 231.657,
-      "eval_steps_per_second": 3.716,
+      "eval_TIME_precision": 0.8387096774193549,
+      "eval_TIME_recall": 0.896551724137931,
+      "eval_loss": 0.3449549674987793,
+      "eval_overall_accuracy": 0.9651728553137003,
+      "eval_overall_f1": 0.867231638418079,
+      "eval_overall_precision": 0.8297297297297297,
+      "eval_overall_recall": 0.908284023668639,
+      "eval_runtime": 0.3216,
+      "eval_samples_per_second": 581.43,
+      "eval_steps_per_second": 9.328,
       "step": 9328
     },
     {
       "epoch": 89.0,
-      "grad_norm": 0.00042143117752857506,
+      "grad_norm": 0.10062725096940994,
       "learning_rate": 5.500000000000001e-06,
       "loss": 0.0002,
       "step": 9434
     },
     {
       "epoch": 89.0,
-      "eval_LOCATION_f1": 0.881118881118881,
+      "eval_LOCATION_f1": 0.8648648648648648,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8513513513513513,
-      "eval_LOCATION_recall": 0.9130434782608695,
-      "eval_ORGANIZATION_f1": 0.719298245614035,
+      "eval_LOCATION_precision": 0.810126582278481,
+      "eval_LOCATION_recall": 0.927536231884058,
+      "eval_ORGANIZATION_f1": 0.7777777777777778,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.7321428571428571,
-      "eval_ORGANIZATION_recall": 0.7068965517241379,
-      "eval_PERSON_f1": 0.8961038961038961,
+      "eval_ORGANIZATION_precision": 0.7205882352941176,
+      "eval_ORGANIZATION_recall": 0.8448275862068966,
+      "eval_PERSON_f1": 0.9196141479099678,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8846153846153846,
-      "eval_PERSON_recall": 0.9078947368421053,
-      "eval_QUANTITY_f1": 0.8387096774193549,
+      "eval_PERSON_precision": 0.89937106918239,
+      "eval_PERSON_recall": 0.9407894736842105,
+      "eval_QUANTITY_f1": 0.8253968253968254,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.8125,
+      "eval_QUANTITY_precision": 0.7878787878787878,
       "eval_QUANTITY_recall": 0.8666666666666667,
-      "eval_TIME_f1": 0.8196721311475409,
+      "eval_TIME_f1": 0.8813559322033899,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.78125,
-      "eval_TIME_recall": 0.8620689655172413,
-      "eval_loss": 0.35247281193733215,
-      "eval_overall_accuracy": 0.9636363636363636,
-      "eval_overall_f1": 0.8517441860465117,
-      "eval_overall_precision": 0.8371428571428572,
-      "eval_overall_recall": 0.8668639053254438,
-      "eval_runtime": 0.8068,
-      "eval_samples_per_second": 231.777,
-      "eval_steps_per_second": 3.718,
+      "eval_TIME_precision": 0.8666666666666667,
+      "eval_TIME_recall": 0.896551724137931,
+      "eval_loss": 0.3450436592102051,
+      "eval_overall_accuracy": 0.9659411011523688,
+      "eval_overall_f1": 0.8712871287128714,
+      "eval_overall_precision": 0.8346883468834688,
+      "eval_overall_recall": 0.9112426035502958,
+      "eval_runtime": 0.3226,
+      "eval_samples_per_second": 579.728,
+      "eval_steps_per_second": 9.3,
       "step": 9434
     },
     {
       "epoch": 90.0,
-      "grad_norm": 0.0002251994446851313,
+      "grad_norm": 0.00016394459817092866,
       "learning_rate": 5e-06,
       "loss": 0.0002,
       "step": 9540
     },
     {
       "epoch": 90.0,
-      "eval_LOCATION_f1": 0.888888888888889,
+      "eval_LOCATION_f1": 0.8648648648648648,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8533333333333334,
+      "eval_LOCATION_precision": 0.810126582278481,
       "eval_LOCATION_recall": 0.927536231884058,
-      "eval_ORGANIZATION_f1": 0.7241379310344829,
+      "eval_ORGANIZATION_f1": 0.7777777777777778,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.7241379310344828,
-      "eval_ORGANIZATION_recall": 0.7241379310344828,
-      "eval_PERSON_f1": 0.9025974025974026,
-      "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8910256410256411,
-      "eval_PERSON_recall": 0.9144736842105263,
-      "eval_QUANTITY_f1": 0.8387096774193549,
+      "eval_ORGANIZATION_precision": 0.7205882352941176,
+      "eval_ORGANIZATION_recall": 0.8448275862068966,
+      "eval_PERSON_f1": 0.9196141479099678,
+      "eval_PERSON_number": 152,
+      "eval_PERSON_precision": 0.89937106918239,
+      "eval_PERSON_recall": 0.9407894736842105,
+      "eval_QUANTITY_f1": 0.8253968253968254,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.8125,
+      "eval_QUANTITY_precision": 0.7878787878787878,
       "eval_QUANTITY_recall": 0.8666666666666667,
-      "eval_TIME_f1": 0.8196721311475409,
+      "eval_TIME_f1": 0.8813559322033899,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.78125,
-      "eval_TIME_recall": 0.8620689655172413,
-      "eval_loss": 0.35829171538352966,
-      "eval_overall_accuracy": 0.9633802816901409,
-      "eval_overall_f1": 0.8567293777134587,
-      "eval_overall_precision": 0.8385269121813032,
-      "eval_overall_recall": 0.8757396449704142,
-      "eval_runtime": 0.813,
-      "eval_samples_per_second": 230.023,
-      "eval_steps_per_second": 3.69,
+      "eval_TIME_precision": 0.8666666666666667,
+      "eval_TIME_recall": 0.896551724137931,
+      "eval_loss": 0.3440031111240387,
+      "eval_overall_accuracy": 0.9661971830985916,
+      "eval_overall_f1": 0.8712871287128714,
+      "eval_overall_precision": 0.8346883468834688,
+      "eval_overall_recall": 0.9112426035502958,
+      "eval_runtime": 0.3218,
+      "eval_samples_per_second": 581.062,
+      "eval_steps_per_second": 9.322,
       "step": 9540
     },
     {
       "epoch": 91.0,
-      "grad_norm": 0.0003033955581486225,
+      "grad_norm": 0.00035388587275519967,
       "learning_rate": 4.5e-06,
       "loss": 0.0002,
       "step": 9646
     },
     {
       "epoch": 91.0,
-      "eval_LOCATION_f1": 0.8965517241379309,
+      "eval_LOCATION_f1": 0.8648648648648648,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8552631578947368,
-      "eval_LOCATION_recall": 0.9420289855072463,
-      "eval_ORGANIZATION_f1": 0.735042735042735,
+      "eval_LOCATION_precision": 0.810126582278481,
+      "eval_LOCATION_recall": 0.927536231884058,
+      "eval_ORGANIZATION_f1": 0.7777777777777778,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.7288135593220338,
-      "eval_ORGANIZATION_recall": 0.7413793103448276,
-      "eval_PERSON_f1": 0.8961038961038961,
+      "eval_ORGANIZATION_precision": 0.7205882352941176,
+      "eval_ORGANIZATION_recall": 0.8448275862068966,
+      "eval_PERSON_f1": 0.9196141479099678,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8846153846153846,
-      "eval_PERSON_recall": 0.9078947368421053,
-      "eval_QUANTITY_f1": 0.8387096774193549,
+      "eval_PERSON_precision": 0.89937106918239,
+      "eval_PERSON_recall": 0.9407894736842105,
+      "eval_QUANTITY_f1": 0.8253968253968254,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.8125,
+      "eval_QUANTITY_precision": 0.7878787878787878,
       "eval_QUANTITY_recall": 0.8666666666666667,
-      "eval_TIME_f1": 0.8196721311475409,
+      "eval_TIME_f1": 0.8813559322033899,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.78125,
-      "eval_TIME_recall": 0.8620689655172413,
-      "eval_loss": 0.3626328408718109,
-      "eval_overall_accuracy": 0.963124199743918,
-      "eval_overall_f1": 0.8571428571428573,
-      "eval_overall_precision": 0.8366197183098592,
-      "eval_overall_recall": 0.878698224852071,
-      "eval_runtime": 0.791,
-      "eval_samples_per_second": 236.421,
-      "eval_steps_per_second": 3.793,
+      "eval_TIME_precision": 0.8666666666666667,
+      "eval_TIME_recall": 0.896551724137931,
+      "eval_loss": 0.3439503610134125,
+      "eval_overall_accuracy": 0.9661971830985916,
+      "eval_overall_f1": 0.8712871287128714,
+      "eval_overall_precision": 0.8346883468834688,
+      "eval_overall_recall": 0.9112426035502958,
+      "eval_runtime": 0.3221,
+      "eval_samples_per_second": 580.542,
+      "eval_steps_per_second": 9.314,
       "step": 9646
     },
     {
       "epoch": 92.0,
-      "grad_norm": 0.00020177336409687996,
+      "grad_norm": 0.0005183388711884618,
       "learning_rate": 4.000000000000001e-06,
-      "loss": 0.0005,
+      "loss": 0.0002,
       "step": 9752
     },
     {
       "epoch": 92.0,
-      "eval_LOCATION_f1": 0.881118881118881,
+      "eval_LOCATION_f1": 0.8648648648648648,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8513513513513513,
-      "eval_LOCATION_recall": 0.9130434782608695,
-      "eval_ORGANIZATION_f1": 0.7304347826086957,
+      "eval_LOCATION_precision": 0.810126582278481,
+      "eval_LOCATION_recall": 0.927536231884058,
+      "eval_ORGANIZATION_f1": 0.7777777777777778,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.7368421052631579,
-      "eval_ORGANIZATION_recall": 0.7241379310344828,
-      "eval_PERSON_f1": 0.8961038961038961,
+      "eval_ORGANIZATION_precision": 0.7205882352941176,
+      "eval_ORGANIZATION_recall": 0.8448275862068966,
+      "eval_PERSON_f1": 0.9196141479099678,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8846153846153846,
-      "eval_PERSON_recall": 0.9078947368421053,
-      "eval_QUANTITY_f1": 0.8387096774193549,
+      "eval_PERSON_precision": 0.89937106918239,
+      "eval_PERSON_recall": 0.9407894736842105,
+      "eval_QUANTITY_f1": 0.8253968253968254,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.8125,
+      "eval_QUANTITY_precision": 0.7878787878787878,
       "eval_QUANTITY_recall": 0.8666666666666667,
-      "eval_TIME_f1": 0.8196721311475409,
+      "eval_TIME_f1": 0.8813559322033899,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.78125,
-      "eval_TIME_recall": 0.8620689655172413,
-      "eval_loss": 0.3607942759990692,
-      "eval_overall_accuracy": 0.9633802816901409,
-      "eval_overall_f1": 0.8534107402031931,
-      "eval_overall_precision": 0.8376068376068376,
-      "eval_overall_recall": 0.8698224852071006,
-      "eval_runtime": 0.8031,
-      "eval_samples_per_second": 232.851,
-      "eval_steps_per_second": 3.736,
+      "eval_TIME_precision": 0.8666666666666667,
+      "eval_TIME_recall": 0.896551724137931,
+      "eval_loss": 0.34474003314971924,
+      "eval_overall_accuracy": 0.9661971830985916,
+      "eval_overall_f1": 0.8712871287128714,
+      "eval_overall_precision": 0.8346883468834688,
+      "eval_overall_recall": 0.9112426035502958,
+      "eval_runtime": 0.3227,
+      "eval_samples_per_second": 579.512,
+      "eval_steps_per_second": 9.297,
       "step": 9752
     },
     {
       "epoch": 93.0,
-      "grad_norm": 0.0001372557453578338,
+      "grad_norm": 0.00015504320617765188,
       "learning_rate": 3.5000000000000004e-06,
-      "loss": 0.0002,
+      "loss": 0.0004,
       "step": 9858
     },
     {
       "epoch": 93.0,
-      "eval_LOCATION_f1": 0.888888888888889,
+      "eval_LOCATION_f1": 0.8648648648648648,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8533333333333334,
+      "eval_LOCATION_precision": 0.810126582278481,
       "eval_LOCATION_recall": 0.927536231884058,
-      "eval_ORGANIZATION_f1": 0.728813559322034,
+      "eval_ORGANIZATION_f1": 0.765625,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.7166666666666667,
-      "eval_ORGANIZATION_recall": 0.7413793103448276,
-      "eval_PERSON_f1": 0.9025974025974026,
+      "eval_ORGANIZATION_precision": 0.7,
+      "eval_ORGANIZATION_recall": 0.8448275862068966,
+      "eval_PERSON_f1": 0.9196141479099678,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8910256410256411,
-      "eval_PERSON_recall": 0.9144736842105263,
-      "eval_QUANTITY_f1": 0.8387096774193549,
+      "eval_PERSON_precision": 0.89937106918239,
+      "eval_PERSON_recall": 0.9407894736842105,
+      "eval_QUANTITY_f1": 0.8253968253968254,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.8125,
+      "eval_QUANTITY_precision": 0.7878787878787878,
       "eval_QUANTITY_recall": 0.8666666666666667,
-      "eval_TIME_f1": 0.8196721311475409,
+      "eval_TIME_f1": 0.8813559322033899,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.78125,
-      "eval_TIME_recall": 0.8620689655172413,
-      "eval_loss": 0.3570501208305359,
-      "eval_overall_accuracy": 0.9633802816901409,
-      "eval_overall_f1": 0.8571428571428573,
-      "eval_overall_precision": 0.8366197183098592,
-      "eval_overall_recall": 0.878698224852071,
-      "eval_runtime": 0.8126,
-      "eval_samples_per_second": 230.138,
-      "eval_steps_per_second": 3.692,
+      "eval_TIME_precision": 0.8666666666666667,
+      "eval_TIME_recall": 0.896551724137931,
+      "eval_loss": 0.34853395819664,
+      "eval_overall_accuracy": 0.9656850192061459,
+      "eval_overall_f1": 0.8688293370944994,
+      "eval_overall_precision": 0.8301886792452831,
+      "eval_overall_recall": 0.9112426035502958,
+      "eval_runtime": 0.3252,
+      "eval_samples_per_second": 575.06,
+      "eval_steps_per_second": 9.226,
       "step": 9858
     },
     {
       "epoch": 94.0,
-      "grad_norm": 0.00034642827813513577,
+      "grad_norm": 0.0016579556977376342,
       "learning_rate": 3e-06,
       "loss": 0.0002,
       "step": 9964
     },
     {
       "epoch": 94.0,
-      "eval_LOCATION_f1": 0.888888888888889,
+      "eval_LOCATION_f1": 0.870748299319728,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8533333333333334,
+      "eval_LOCATION_precision": 0.8205128205128205,
       "eval_LOCATION_recall": 0.927536231884058,
-      "eval_ORGANIZATION_f1": 0.728813559322034,
+      "eval_ORGANIZATION_f1": 0.7903225806451614,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.7166666666666667,
-      "eval_ORGANIZATION_recall": 0.7413793103448276,
-      "eval_PERSON_f1": 0.9025974025974026,
+      "eval_ORGANIZATION_precision": 0.7424242424242424,
+      "eval_ORGANIZATION_recall": 0.8448275862068966,
+      "eval_PERSON_f1": 0.9225806451612903,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8910256410256411,
-      "eval_PERSON_recall": 0.9144736842105263,
-      "eval_QUANTITY_f1": 0.8387096774193549,
+      "eval_PERSON_precision": 0.9050632911392406,
+      "eval_PERSON_recall": 0.9407894736842105,
+      "eval_QUANTITY_f1": 0.8253968253968254,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.8125,
+      "eval_QUANTITY_precision": 0.7878787878787878,
       "eval_QUANTITY_recall": 0.8666666666666667,
-      "eval_TIME_f1": 0.8196721311475409,
+      "eval_TIME_f1": 0.8813559322033899,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.78125,
-      "eval_TIME_recall": 0.8620689655172413,
-      "eval_loss": 0.3574298322200775,
-      "eval_overall_accuracy": 0.9633802816901409,
-      "eval_overall_f1": 0.8571428571428573,
-      "eval_overall_precision": 0.8366197183098592,
-      "eval_overall_recall": 0.878698224852071,
-      "eval_runtime": 0.8062,
-      "eval_samples_per_second": 231.955,
-      "eval_steps_per_second": 3.721,
+      "eval_TIME_precision": 0.8666666666666667,
+      "eval_TIME_recall": 0.896551724137931,
+      "eval_loss": 0.34736061096191406,
+      "eval_overall_accuracy": 0.9672215108834827,
+      "eval_overall_f1": 0.8762446657183499,
+      "eval_overall_precision": 0.8438356164383561,
+      "eval_overall_recall": 0.9112426035502958,
+      "eval_runtime": 0.324,
+      "eval_samples_per_second": 577.077,
+      "eval_steps_per_second": 9.258,
       "step": 9964
     },
     {
       "epoch": 95.0,
-      "grad_norm": 0.0005071151535958052,
+      "grad_norm": 0.0002204200136475265,
       "learning_rate": 2.5e-06,
-      "loss": 0.0004,
+      "loss": 0.0002,
       "step": 10070
     },
     {
       "epoch": 95.0,
-      "eval_LOCATION_f1": 0.8951048951048951,
+      "eval_LOCATION_f1": 0.870748299319728,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8648648648648649,
+      "eval_LOCATION_precision": 0.8205128205128205,
       "eval_LOCATION_recall": 0.927536231884058,
-      "eval_ORGANIZATION_f1": 0.7166666666666668,
+      "eval_ORGANIZATION_f1": 0.7903225806451614,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.6935483870967742,
-      "eval_ORGANIZATION_recall": 0.7413793103448276,
-      "eval_PERSON_f1": 0.9025974025974026,
+      "eval_ORGANIZATION_precision": 0.7424242424242424,
+      "eval_ORGANIZATION_recall": 0.8448275862068966,
+      "eval_PERSON_f1": 0.9225806451612903,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8910256410256411,
-      "eval_PERSON_recall": 0.9144736842105263,
-      "eval_QUANTITY_f1": 0.8387096774193549,
+      "eval_PERSON_precision": 0.9050632911392406,
+      "eval_PERSON_recall": 0.9407894736842105,
+      "eval_QUANTITY_f1": 0.8253968253968254,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.8125,
+      "eval_QUANTITY_precision": 0.7878787878787878,
       "eval_QUANTITY_recall": 0.8666666666666667,
-      "eval_TIME_f1": 0.8196721311475409,
+      "eval_TIME_f1": 0.8813559322033899,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.78125,
-      "eval_TIME_recall": 0.8620689655172413,
-      "eval_loss": 0.35781243443489075,
-      "eval_overall_accuracy": 0.9641485275288092,
-      "eval_overall_f1": 0.8559077809798271,
-      "eval_overall_precision": 0.8342696629213483,
-      "eval_overall_recall": 0.878698224852071,
-      "eval_runtime": 0.8038,
-      "eval_samples_per_second": 232.657,
-      "eval_steps_per_second": 3.732,
+      "eval_TIME_precision": 0.8666666666666667,
+      "eval_TIME_recall": 0.896551724137931,
+      "eval_loss": 0.3477398753166199,
+      "eval_overall_accuracy": 0.9669654289372599,
+      "eval_overall_f1": 0.8762446657183499,
+      "eval_overall_precision": 0.8438356164383561,
+      "eval_overall_recall": 0.9112426035502958,
+      "eval_runtime": 0.3352,
+      "eval_samples_per_second": 557.803,
+      "eval_steps_per_second": 8.949,
       "step": 10070
     },
     {
       "epoch": 96.0,
-      "grad_norm": 0.000192229199456051,
+      "grad_norm": 0.0003840687859337777,
       "learning_rate": 2.0000000000000003e-06,
-      "loss": 0.0003,
+      "loss": 0.0002,
       "step": 10176
     },
     {
       "epoch": 96.0,
-      "eval_LOCATION_f1": 0.8951048951048951,
+      "eval_LOCATION_f1": 0.870748299319728,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8648648648648649,
+      "eval_LOCATION_precision": 0.8205128205128205,
       "eval_LOCATION_recall": 0.927536231884058,
-      "eval_ORGANIZATION_f1": 0.7166666666666668,
+      "eval_ORGANIZATION_f1": 0.7903225806451614,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.6935483870967742,
-      "eval_ORGANIZATION_recall": 0.7413793103448276,
-      "eval_PERSON_f1": 0.8961038961038961,
+      "eval_ORGANIZATION_precision": 0.7424242424242424,
+      "eval_ORGANIZATION_recall": 0.8448275862068966,
+      "eval_PERSON_f1": 0.9196141479099678,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8846153846153846,
-      "eval_PERSON_recall": 0.9078947368421053,
-      "eval_QUANTITY_f1": 0.8387096774193549,
+      "eval_PERSON_precision": 0.89937106918239,
+      "eval_PERSON_recall": 0.9407894736842105,
+      "eval_QUANTITY_f1": 0.8253968253968254,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.8125,
+      "eval_QUANTITY_precision": 0.7878787878787878,
       "eval_QUANTITY_recall": 0.8666666666666667,
-      "eval_TIME_f1": 0.8196721311475409,
+      "eval_TIME_f1": 0.8813559322033899,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.78125,
-      "eval_TIME_recall": 0.8620689655172413,
-      "eval_loss": 0.35879892110824585,
-      "eval_overall_accuracy": 0.9638924455825865,
-      "eval_overall_f1": 0.8530259365994237,
-      "eval_overall_precision": 0.8314606741573034,
-      "eval_overall_recall": 0.8757396449704142,
-      "eval_runtime": 0.8034,
-      "eval_samples_per_second": 232.77,
-      "eval_steps_per_second": 3.734,
+      "eval_TIME_precision": 0.8666666666666667,
+      "eval_TIME_recall": 0.896551724137931,
+      "eval_loss": 0.34747499227523804,
+      "eval_overall_accuracy": 0.9669654289372599,
+      "eval_overall_f1": 0.875,
+      "eval_overall_precision": 0.8415300546448088,
+      "eval_overall_recall": 0.9112426035502958,
+      "eval_runtime": 0.3248,
+      "eval_samples_per_second": 575.753,
+      "eval_steps_per_second": 9.237,
       "step": 10176
     },
     {
       "epoch": 97.0,
-      "grad_norm": 0.00012044800678268075,
+      "grad_norm": 0.0001158128070528619,
       "learning_rate": 1.5e-06,
-      "loss": 0.0004,
+      "loss": 0.0002,
       "step": 10282
     },
     {
       "epoch": 97.0,
-      "eval_LOCATION_f1": 0.888888888888889,
+      "eval_LOCATION_f1": 0.870748299319728,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8533333333333334,
+      "eval_LOCATION_precision": 0.8205128205128205,
       "eval_LOCATION_recall": 0.927536231884058,
-      "eval_ORGANIZATION_f1": 0.7166666666666668,
+      "eval_ORGANIZATION_f1": 0.7903225806451614,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.6935483870967742,
-      "eval_ORGANIZATION_recall": 0.7413793103448276,
-      "eval_PERSON_f1": 0.9025974025974026,
+      "eval_ORGANIZATION_precision": 0.7424242424242424,
+      "eval_ORGANIZATION_recall": 0.8448275862068966,
+      "eval_PERSON_f1": 0.9196141479099678,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8910256410256411,
-      "eval_PERSON_recall": 0.9144736842105263,
-      "eval_QUANTITY_f1": 0.8387096774193549,
+      "eval_PERSON_precision": 0.89937106918239,
+      "eval_PERSON_recall": 0.9407894736842105,
+      "eval_QUANTITY_f1": 0.8253968253968254,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.8125,
+      "eval_QUANTITY_precision": 0.7878787878787878,
       "eval_QUANTITY_recall": 0.8666666666666667,
-      "eval_TIME_f1": 0.8196721311475409,
+      "eval_TIME_f1": 0.8813559322033899,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.78125,
-      "eval_TIME_recall": 0.8620689655172413,
-      "eval_loss": 0.3565698266029358,
-      "eval_overall_accuracy": 0.9638924455825865,
-      "eval_overall_f1": 0.8546762589928059,
-      "eval_overall_precision": 0.8319327731092437,
-      "eval_overall_recall": 0.878698224852071,
-      "eval_runtime": 0.8012,
-      "eval_samples_per_second": 233.392,
-      "eval_steps_per_second": 3.744,
+      "eval_TIME_precision": 0.8666666666666667,
+      "eval_TIME_recall": 0.896551724137931,
+      "eval_loss": 0.34782785177230835,
+      "eval_overall_accuracy": 0.9669654289372599,
+      "eval_overall_f1": 0.875,
+      "eval_overall_precision": 0.8415300546448088,
+      "eval_overall_recall": 0.9112426035502958,
+      "eval_runtime": 0.3347,
+      "eval_samples_per_second": 558.788,
+      "eval_steps_per_second": 8.965,
       "step": 10282
     },
     {
       "epoch": 98.0,
-      "grad_norm": 0.0002169125946238637,
+      "grad_norm": 0.0001685286551946774,
       "learning_rate": 1.0000000000000002e-06,
-      "loss": 0.0003,
+      "loss": 0.0001,
       "step": 10388
     },
     {
       "epoch": 98.0,
-      "eval_LOCATION_f1": 0.888888888888889,
+      "eval_LOCATION_f1": 0.870748299319728,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8533333333333334,
+      "eval_LOCATION_precision": 0.8205128205128205,
       "eval_LOCATION_recall": 0.927536231884058,
-      "eval_ORGANIZATION_f1": 0.7107438016528926,
+      "eval_ORGANIZATION_f1": 0.7903225806451614,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.6825396825396826,
-      "eval_ORGANIZATION_recall": 0.7413793103448276,
-      "eval_PERSON_f1": 0.9025974025974026,
+      "eval_ORGANIZATION_precision": 0.7424242424242424,
+      "eval_ORGANIZATION_recall": 0.8448275862068966,
+      "eval_PERSON_f1": 0.9196141479099678,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8910256410256411,
-      "eval_PERSON_recall": 0.9144736842105263,
-      "eval_QUANTITY_f1": 0.8387096774193549,
+      "eval_PERSON_precision": 0.89937106918239,
+      "eval_PERSON_recall": 0.9407894736842105,
+      "eval_QUANTITY_f1": 0.8253968253968254,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.8125,
+      "eval_QUANTITY_precision": 0.7878787878787878,
       "eval_QUANTITY_recall": 0.8666666666666667,
-      "eval_TIME_f1": 0.8196721311475409,
+      "eval_TIME_f1": 0.8813559322033899,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.78125,
-      "eval_TIME_recall": 0.8620689655172413,
-      "eval_loss": 0.3576311767101288,
-      "eval_overall_accuracy": 0.9636363636363636,
-      "eval_overall_f1": 0.8534482758620691,
-      "eval_overall_precision": 0.8296089385474861,
-      "eval_overall_recall": 0.878698224852071,
-      "eval_runtime": 0.7986,
-      "eval_samples_per_second": 234.172,
-      "eval_steps_per_second": 3.757,
+      "eval_TIME_precision": 0.8666666666666667,
+      "eval_TIME_recall": 0.896551724137931,
+      "eval_loss": 0.3480530083179474,
+      "eval_overall_accuracy": 0.9669654289372599,
+      "eval_overall_f1": 0.875,
+      "eval_overall_precision": 0.8415300546448088,
+      "eval_overall_recall": 0.9112426035502958,
+      "eval_runtime": 0.3236,
+      "eval_samples_per_second": 577.794,
+      "eval_steps_per_second": 9.269,
       "step": 10388
     },
     {
       "epoch": 99.0,
-      "grad_norm": 0.0006560708279721439,
+      "grad_norm": 0.0004627603921107948,
       "learning_rate": 5.000000000000001e-07,
-      "loss": 0.0004,
+      "loss": 0.0003,
       "step": 10494
     },
     {
       "epoch": 99.0,
-      "eval_LOCATION_f1": 0.8965517241379309,
+      "eval_LOCATION_f1": 0.870748299319728,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8552631578947368,
-      "eval_LOCATION_recall": 0.9420289855072463,
-      "eval_ORGANIZATION_f1": 0.7166666666666668,
+      "eval_LOCATION_precision": 0.8205128205128205,
+      "eval_LOCATION_recall": 0.927536231884058,
+      "eval_ORGANIZATION_f1": 0.7903225806451614,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.6935483870967742,
-      "eval_ORGANIZATION_recall": 0.7413793103448276,
-      "eval_PERSON_f1": 0.9025974025974026,
+      "eval_ORGANIZATION_precision": 0.7424242424242424,
+      "eval_ORGANIZATION_recall": 0.8448275862068966,
+      "eval_PERSON_f1": 0.9196141479099678,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8910256410256411,
-      "eval_PERSON_recall": 0.9144736842105263,
-      "eval_QUANTITY_f1": 0.8387096774193549,
+      "eval_PERSON_precision": 0.89937106918239,
+      "eval_PERSON_recall": 0.9407894736842105,
+      "eval_QUANTITY_f1": 0.8253968253968254,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.8125,
+      "eval_QUANTITY_precision": 0.7878787878787878,
       "eval_QUANTITY_recall": 0.8666666666666667,
-      "eval_TIME_f1": 0.8196721311475409,
+      "eval_TIME_f1": 0.8813559322033899,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.78125,
-      "eval_TIME_recall": 0.8620689655172413,
-      "eval_loss": 0.3574199676513672,
-      "eval_overall_accuracy": 0.9638924455825865,
-      "eval_overall_f1": 0.8563218390804598,
-      "eval_overall_precision": 0.8324022346368715,
-      "eval_overall_recall": 0.8816568047337278,
-      "eval_runtime": 0.7986,
-      "eval_samples_per_second": 234.152,
-      "eval_steps_per_second": 3.756,
+      "eval_TIME_precision": 0.8666666666666667,
+      "eval_TIME_recall": 0.896551724137931,
+      "eval_loss": 0.3480047881603241,
+      "eval_overall_accuracy": 0.9669654289372599,
+      "eval_overall_f1": 0.875,
+      "eval_overall_precision": 0.8415300546448088,
+      "eval_overall_recall": 0.9112426035502958,
+      "eval_runtime": 0.33,
+      "eval_samples_per_second": 566.586,
+      "eval_steps_per_second": 9.09,
       "step": 10494
     },
     {
       "epoch": 100.0,
-      "grad_norm": 0.0002968113112729043,
+      "grad_norm": 0.00016238813987001777,
       "learning_rate": 0.0,
       "loss": 0.0002,
       "step": 10600
     },
     {
       "epoch": 100.0,
-      "eval_LOCATION_f1": 0.8965517241379309,
+      "eval_LOCATION_f1": 0.870748299319728,
       "eval_LOCATION_number": 69,
-      "eval_LOCATION_precision": 0.8552631578947368,
-      "eval_LOCATION_recall": 0.9420289855072463,
-      "eval_ORGANIZATION_f1": 0.7166666666666668,
+      "eval_LOCATION_precision": 0.8205128205128205,
+      "eval_LOCATION_recall": 0.927536231884058,
+      "eval_ORGANIZATION_f1": 0.7903225806451614,
       "eval_ORGANIZATION_number": 58,
-      "eval_ORGANIZATION_precision": 0.6935483870967742,
-      "eval_ORGANIZATION_recall": 0.7413793103448276,
-      "eval_PERSON_f1": 0.9025974025974026,
+      "eval_ORGANIZATION_precision": 0.7424242424242424,
+      "eval_ORGANIZATION_recall": 0.8448275862068966,
+      "eval_PERSON_f1": 0.9225806451612903,
       "eval_PERSON_number": 152,
-      "eval_PERSON_precision": 0.8910256410256411,
-      "eval_PERSON_recall": 0.9144736842105263,
-      "eval_QUANTITY_f1": 0.8387096774193549,
+      "eval_PERSON_precision": 0.9050632911392406,
+      "eval_PERSON_recall": 0.9407894736842105,
+      "eval_QUANTITY_f1": 0.8253968253968254,
       "eval_QUANTITY_number": 30,
-      "eval_QUANTITY_precision": 0.8125,
+      "eval_QUANTITY_precision": 0.7878787878787878,
       "eval_QUANTITY_recall": 0.8666666666666667,
-      "eval_TIME_f1": 0.8196721311475409,
+      "eval_TIME_f1": 0.8813559322033899,
       "eval_TIME_number": 29,
-      "eval_TIME_precision": 0.78125,
-      "eval_TIME_recall": 0.8620689655172413,
-      "eval_loss": 0.35746997594833374,
-      "eval_overall_accuracy": 0.9638924455825865,
-      "eval_overall_f1": 0.8563218390804598,
-      "eval_overall_precision": 0.8324022346368715,
-      "eval_overall_recall": 0.8816568047337278,
-      "eval_runtime": 0.8145,
-      "eval_samples_per_second": 229.593,
-      "eval_steps_per_second": 3.683,
+      "eval_TIME_precision": 0.8666666666666667,
+      "eval_TIME_recall": 0.896551724137931,
+      "eval_loss": 0.34784960746765137,
+      "eval_overall_accuracy": 0.9669654289372599,
+      "eval_overall_f1": 0.8762446657183499,
+      "eval_overall_precision": 0.8438356164383561,
+      "eval_overall_recall": 0.9112426035502958,
+      "eval_runtime": 0.3263,
+      "eval_samples_per_second": 573.017,
+      "eval_steps_per_second": 9.193,
       "step": 10600
     },
     {
       "epoch": 100.0,
       "step": 10600,
       "total_flos": 4488439619319522.0,
-      "train_loss": 0.00870023393810217,
-      "train_runtime": 2597.4084,
-      "train_samples_per_second": 64.949,
-      "train_steps_per_second": 4.081
+      "train_loss": 0.00870069551422208,
+      "train_runtime": 967.5082,
+      "train_samples_per_second": 174.365,
+      "train_steps_per_second": 10.956
     }
   ],
   "logging_steps": 500,