diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -10,2856 +10,2856 @@
   "log_history": [
     {
       "epoch": 1.0,
-      "grad_norm": 4.311885356903076,
+      "grad_norm": 3.163893938064575,
       "learning_rate": 4.9500000000000004e-05,
-      "loss": 0.3177,
+      "loss": 0.3326,
       "step": 106
     },
     {
       "epoch": 1.0,
-      "eval_LOCATION_f1": 0.7453416149068322,
+      "eval_LOCATION_f1": 0.8266666666666668,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.6818181818181818,
-      "eval_LOCATION_recall": 0.821917808219178,
-      "eval_ORGANIZATION_f1": 0.6527777777777777,
+      "eval_LOCATION_precision": 0.8051948051948052,
+      "eval_LOCATION_recall": 0.8493150684931506,
+      "eval_ORGANIZATION_f1": 0.6666666666666667,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.5949367088607594,
+      "eval_ORGANIZATION_precision": 0.618421052631579,
       "eval_ORGANIZATION_recall": 0.7230769230769231,
-      "eval_PERSON_f1": 0.8782051282051282,
+      "eval_PERSON_f1": 0.8859934853420196,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.845679012345679,
-      "eval_PERSON_recall": 0.9133333333333333,
-      "eval_QUANTITY_f1": 0.716417910447761,
+      "eval_PERSON_precision": 0.8662420382165605,
+      "eval_PERSON_recall": 0.9066666666666666,
+      "eval_QUANTITY_f1": 0.7096774193548386,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.631578947368421,
-      "eval_QUANTITY_recall": 0.8275862068965517,
-      "eval_TIME_f1": 0.9428571428571428,
-      "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.9166666666666666,
-      "eval_TIME_recall": 0.9705882352941176,
-      "eval_loss": 0.12832513451576233,
-      "eval_overall_accuracy": 0.9576667487078513,
-      "eval_overall_f1": 0.7984084880636605,
-      "eval_overall_precision": 0.7468982630272953,
-      "eval_overall_recall": 0.8575498575498576,
-      "eval_runtime": 0.9602,
-      "eval_samples_per_second": 194.76,
-      "eval_steps_per_second": 3.124,
+      "eval_QUANTITY_precision": 0.6666666666666666,
+      "eval_QUANTITY_recall": 0.7586206896551724,
+      "eval_TIME_f1": 0.90625,
+      "eval_TIME_number": 34,
+      "eval_TIME_precision": 0.9666666666666667,
+      "eval_TIME_recall": 0.8529411764705882,
+      "eval_loss": 0.11882562935352325,
+      "eval_overall_accuracy": 0.9620969726802855,
+      "eval_overall_f1": 0.8176795580110497,
+      "eval_overall_precision": 0.7935656836461126,
+      "eval_overall_recall": 0.8433048433048433,
+      "eval_runtime": 0.359,
+      "eval_samples_per_second": 520.888,
+      "eval_steps_per_second": 8.356,
       "step": 106
     },
     {
       "epoch": 2.0,
-      "grad_norm": 2.1292572021484375,
+      "grad_norm": 2.463841438293457,
       "learning_rate": 4.9e-05,
-      "loss": 0.1156,
+      "loss": 0.1192,
       "step": 212
     },
     {
       "epoch": 2.0,
-      "eval_LOCATION_f1": 0.8,
+      "eval_LOCATION_f1": 0.7544910179640719,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.735632183908046,
-      "eval_LOCATION_recall": 0.8767123287671232,
-      "eval_ORGANIZATION_f1": 0.6540880503144655,
+      "eval_LOCATION_precision": 0.6702127659574468,
+      "eval_LOCATION_recall": 0.863013698630137,
+      "eval_ORGANIZATION_f1": 0.6031746031746031,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.5531914893617021,
-      "eval_ORGANIZATION_recall": 0.8,
-      "eval_PERSON_f1": 0.8690095846645367,
+      "eval_ORGANIZATION_precision": 0.4596774193548387,
+      "eval_ORGANIZATION_recall": 0.8769230769230769,
+      "eval_PERSON_f1": 0.896774193548387,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8343558282208589,
-      "eval_PERSON_recall": 0.9066666666666666,
-      "eval_QUANTITY_f1": 0.8125,
+      "eval_PERSON_precision": 0.86875,
+      "eval_PERSON_recall": 0.9266666666666666,
+      "eval_QUANTITY_f1": 0.7142857142857144,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.7428571428571429,
-      "eval_QUANTITY_recall": 0.896551724137931,
-      "eval_TIME_f1": 0.8571428571428571,
+      "eval_QUANTITY_precision": 0.6097560975609756,
+      "eval_QUANTITY_recall": 0.8620689655172413,
+      "eval_TIME_f1": 0.7631578947368423,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8333333333333334,
-      "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.12480369955301285,
-      "eval_overall_accuracy": 0.9579128722618755,
-      "eval_overall_f1": 0.804177545691906,
-      "eval_overall_precision": 0.7421686746987952,
-      "eval_overall_recall": 0.8774928774928775,
-      "eval_runtime": 0.9905,
-      "eval_samples_per_second": 188.796,
-      "eval_steps_per_second": 3.029,
+      "eval_TIME_precision": 0.6904761904761905,
+      "eval_TIME_recall": 0.8529411764705882,
+      "eval_loss": 0.17271338403224945,
+      "eval_overall_accuracy": 0.9396997292640906,
+      "eval_overall_f1": 0.7709359605911329,
+      "eval_overall_precision": 0.6789587852494577,
+      "eval_overall_recall": 0.8917378917378918,
+      "eval_runtime": 0.3677,
+      "eval_samples_per_second": 508.572,
+      "eval_steps_per_second": 8.159,
       "step": 212
     },
     {
       "epoch": 3.0,
-      "grad_norm": 3.7740285396575928,
+      "grad_norm": 4.771985054016113,
       "learning_rate": 4.85e-05,
-      "loss": 0.071,
+      "loss": 0.0772,
       "step": 318
     },
     {
       "epoch": 3.0,
-      "eval_LOCATION_f1": 0.8079470198675496,
+      "eval_LOCATION_f1": 0.8,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.782051282051282,
-      "eval_LOCATION_recall": 0.8356164383561644,
-      "eval_ORGANIZATION_f1": 0.6923076923076923,
+      "eval_LOCATION_precision": 0.735632183908046,
+      "eval_LOCATION_recall": 0.8767123287671232,
+      "eval_ORGANIZATION_f1": 0.7464788732394366,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6923076923076923,
-      "eval_ORGANIZATION_recall": 0.6923076923076923,
-      "eval_PERSON_f1": 0.8996763754045306,
+      "eval_ORGANIZATION_precision": 0.6883116883116883,
+      "eval_ORGANIZATION_recall": 0.8153846153846154,
+      "eval_PERSON_f1": 0.896774193548387,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8742138364779874,
+      "eval_PERSON_precision": 0.86875,
       "eval_PERSON_recall": 0.9266666666666666,
-      "eval_QUANTITY_f1": 0.7540983606557378,
+      "eval_QUANTITY_f1": 0.7999999999999999,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.71875,
-      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_QUANTITY_precision": 0.7741935483870968,
+      "eval_QUANTITY_recall": 0.8275862068965517,
       "eval_TIME_f1": 0.8529411764705882,
       "eval_TIME_number": 34,
       "eval_TIME_precision": 0.8529411764705882,
       "eval_TIME_recall": 0.8529411764705882,
-      "eval_loss": 0.121186263859272,
-      "eval_overall_accuracy": 0.9640659611124784,
-      "eval_overall_f1": 0.8261474269819192,
-      "eval_overall_precision": 0.8070652173913043,
-      "eval_overall_recall": 0.8461538461538461,
-      "eval_runtime": 0.9209,
-      "eval_samples_per_second": 203.06,
-      "eval_steps_per_second": 3.258,
+      "eval_loss": 0.12907364964485168,
+      "eval_overall_accuracy": 0.9625892197883338,
+      "eval_overall_f1": 0.835135135135135,
+      "eval_overall_precision": 0.794344473007712,
+      "eval_overall_recall": 0.8803418803418803,
+      "eval_runtime": 0.3624,
+      "eval_samples_per_second": 516.025,
+      "eval_steps_per_second": 8.278,
       "step": 318
     },
     {
       "epoch": 4.0,
-      "grad_norm": 0.12752270698547363,
+      "grad_norm": 0.4475593566894531,
       "learning_rate": 4.8e-05,
-      "loss": 0.0488,
+      "loss": 0.051,
       "step": 424
     },
     {
       "epoch": 4.0,
-      "eval_LOCATION_f1": 0.8051948051948052,
+      "eval_LOCATION_f1": 0.7999999999999999,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7654320987654321,
+      "eval_LOCATION_precision": 0.7560975609756098,
       "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.6973684210526315,
+      "eval_ORGANIZATION_f1": 0.708860759493671,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6091954022988506,
-      "eval_ORGANIZATION_recall": 0.8153846153846154,
-      "eval_PERSON_f1": 0.8825396825396825,
+      "eval_ORGANIZATION_precision": 0.6021505376344086,
+      "eval_ORGANIZATION_recall": 0.8615384615384616,
+      "eval_PERSON_f1": 0.9025974025974027,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8424242424242424,
+      "eval_PERSON_precision": 0.879746835443038,
       "eval_PERSON_recall": 0.9266666666666666,
-      "eval_QUANTITY_f1": 0.8709677419354839,
+      "eval_QUANTITY_f1": 0.8571428571428571,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.8181818181818182,
+      "eval_QUANTITY_precision": 0.7941176470588235,
       "eval_QUANTITY_recall": 0.9310344827586207,
-      "eval_TIME_f1": 0.8695652173913043,
+      "eval_TIME_f1": 0.8219178082191781,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8571428571428571,
+      "eval_TIME_precision": 0.7692307692307693,
       "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.16518306732177734,
-      "eval_overall_accuracy": 0.9566822544917548,
-      "eval_overall_f1": 0.827127659574468,
-      "eval_overall_precision": 0.7755610972568578,
-      "eval_overall_recall": 0.886039886039886,
-      "eval_runtime": 0.9276,
-      "eval_samples_per_second": 201.599,
-      "eval_steps_per_second": 3.234,
+      "eval_loss": 0.1436338722705841,
+      "eval_overall_accuracy": 0.958897366477972,
+      "eval_overall_f1": 0.8295904887714664,
+      "eval_overall_precision": 0.7733990147783252,
+      "eval_overall_recall": 0.8945868945868946,
+      "eval_runtime": 0.3718,
+      "eval_samples_per_second": 503.026,
+      "eval_steps_per_second": 8.07,
       "step": 424
     },
     {
       "epoch": 5.0,
-      "grad_norm": 1.2622820138931274,
+      "grad_norm": 0.9967474937438965,
       "learning_rate": 4.75e-05,
-      "loss": 0.0347,
+      "loss": 0.0341,
       "step": 530
     },
     {
       "epoch": 5.0,
-      "eval_LOCATION_f1": 0.832214765100671,
+      "eval_LOCATION_f1": 0.781456953642384,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.8157894736842105,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.75,
+      "eval_LOCATION_precision": 0.7564102564102564,
+      "eval_LOCATION_recall": 0.8082191780821918,
+      "eval_ORGANIZATION_f1": 0.763888888888889,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.7183098591549296,
-      "eval_ORGANIZATION_recall": 0.7846153846153846,
-      "eval_PERSON_f1": 0.8996763754045306,
+      "eval_ORGANIZATION_precision": 0.6962025316455697,
+      "eval_ORGANIZATION_recall": 0.8461538461538461,
+      "eval_PERSON_f1": 0.9049180327868852,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8742138364779874,
-      "eval_PERSON_recall": 0.9266666666666666,
-      "eval_QUANTITY_f1": 0.8709677419354839,
+      "eval_PERSON_precision": 0.8903225806451613,
+      "eval_PERSON_recall": 0.92,
+      "eval_QUANTITY_f1": 0.8571428571428571,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.8181818181818182,
+      "eval_QUANTITY_precision": 0.7941176470588235,
       "eval_QUANTITY_recall": 0.9310344827586207,
-      "eval_TIME_f1": 0.8857142857142858,
+      "eval_TIME_f1": 0.7894736842105262,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8611111111111112,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.15460582077503204,
-      "eval_overall_accuracy": 0.9662810730986956,
-      "eval_overall_f1": 0.8539944903581267,
-      "eval_overall_precision": 0.8266666666666667,
-      "eval_overall_recall": 0.8831908831908832,
-      "eval_runtime": 0.9741,
-      "eval_samples_per_second": 191.969,
-      "eval_steps_per_second": 3.08,
+      "eval_TIME_precision": 0.7142857142857143,
+      "eval_TIME_recall": 0.8823529411764706,
+      "eval_loss": 0.15582042932510376,
+      "eval_overall_accuracy": 0.9650504553285749,
+      "eval_overall_f1": 0.8362652232746955,
+      "eval_overall_precision": 0.7963917525773195,
+      "eval_overall_recall": 0.8803418803418803,
+      "eval_runtime": 0.3676,
+      "eval_samples_per_second": 508.665,
+      "eval_steps_per_second": 8.16,
       "step": 530
     },
     {
       "epoch": 6.0,
-      "grad_norm": 2.0014474391937256,
+      "grad_norm": 2.4357876777648926,
       "learning_rate": 4.7e-05,
-      "loss": 0.0238,
+      "loss": 0.0289,
       "step": 636
     },
     {
       "epoch": 6.0,
-      "eval_LOCATION_f1": 0.8211920529801324,
+      "eval_LOCATION_f1": 0.8152866242038217,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7948717948717948,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7870967741935484,
+      "eval_LOCATION_precision": 0.7619047619047619,
+      "eval_LOCATION_recall": 0.8767123287671232,
+      "eval_ORGANIZATION_f1": 0.6896551724137931,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6777777777777778,
-      "eval_ORGANIZATION_recall": 0.9384615384615385,
-      "eval_PERSON_f1": 0.9096774193548387,
+      "eval_ORGANIZATION_precision": 0.625,
+      "eval_ORGANIZATION_recall": 0.7692307692307693,
+      "eval_PERSON_f1": 0.8789808917197452,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.88125,
-      "eval_PERSON_recall": 0.94,
-      "eval_QUANTITY_f1": 0.7999999999999999,
+      "eval_PERSON_precision": 0.8414634146341463,
+      "eval_PERSON_recall": 0.92,
+      "eval_QUANTITY_f1": 0.8571428571428571,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.7222222222222222,
-      "eval_QUANTITY_recall": 0.896551724137931,
-      "eval_TIME_f1": 0.8985507246376812,
+      "eval_QUANTITY_precision": 0.7941176470588235,
+      "eval_QUANTITY_recall": 0.9310344827586207,
+      "eval_TIME_f1": 0.8450704225352113,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8857142857142857,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.18289177119731903,
-      "eval_overall_accuracy": 0.9648043317745508,
-      "eval_overall_f1": 0.856,
-      "eval_overall_precision": 0.8045112781954887,
-      "eval_overall_recall": 0.9145299145299145,
-      "eval_runtime": 1.0278,
-      "eval_samples_per_second": 181.945,
-      "eval_steps_per_second": 2.919,
+      "eval_TIME_precision": 0.8108108108108109,
+      "eval_TIME_recall": 0.8823529411764706,
+      "eval_loss": 0.18201519548892975,
+      "eval_overall_accuracy": 0.9606202313561408,
+      "eval_overall_f1": 0.824,
+      "eval_overall_precision": 0.7744360902255639,
+      "eval_overall_recall": 0.8803418803418803,
+      "eval_runtime": 0.3669,
+      "eval_samples_per_second": 509.621,
+      "eval_steps_per_second": 8.176,
       "step": 636
     },
     {
       "epoch": 7.0,
-      "grad_norm": 0.5132791996002197,
+      "grad_norm": 6.440542697906494,
       "learning_rate": 4.6500000000000005e-05,
-      "loss": 0.0178,
+      "loss": 0.0223,
       "step": 742
     },
     {
       "epoch": 7.0,
-      "eval_LOCATION_f1": 0.8157894736842105,
+      "eval_LOCATION_f1": 0.8129032258064516,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7848101265822784,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7132867132867133,
+      "eval_LOCATION_precision": 0.7682926829268293,
+      "eval_LOCATION_recall": 0.863013698630137,
+      "eval_ORGANIZATION_f1": 0.6906474820143885,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6538461538461539,
-      "eval_ORGANIZATION_recall": 0.7846153846153846,
-      "eval_PERSON_f1": 0.8996763754045306,
+      "eval_ORGANIZATION_precision": 0.6486486486486487,
+      "eval_ORGANIZATION_recall": 0.7384615384615385,
+      "eval_PERSON_f1": 0.9144736842105262,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8742138364779874,
+      "eval_PERSON_precision": 0.9025974025974026,
       "eval_PERSON_recall": 0.9266666666666666,
-      "eval_QUANTITY_f1": 0.8125,
+      "eval_QUANTITY_f1": 0.7999999999999999,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.7428571428571429,
-      "eval_QUANTITY_recall": 0.896551724137931,
-      "eval_TIME_f1": 0.7894736842105262,
+      "eval_QUANTITY_precision": 0.7741935483870968,
+      "eval_QUANTITY_recall": 0.8275862068965517,
+      "eval_TIME_f1": 0.8055555555555555,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.7142857142857143,
-      "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.2152390480041504,
-      "eval_overall_accuracy": 0.958897366477972,
-      "eval_overall_f1": 0.8279569892473119,
-      "eval_overall_precision": 0.7837150127226463,
-      "eval_overall_recall": 0.8774928774928775,
-      "eval_runtime": 0.9592,
-      "eval_samples_per_second": 194.95,
-      "eval_steps_per_second": 3.128,
+      "eval_TIME_precision": 0.7631578947368421,
+      "eval_TIME_recall": 0.8529411764705882,
+      "eval_loss": 0.18743985891342163,
+      "eval_overall_accuracy": 0.9620969726802855,
+      "eval_overall_f1": 0.8301369863013699,
+      "eval_overall_precision": 0.7994722955145118,
+      "eval_overall_recall": 0.8632478632478633,
+      "eval_runtime": 0.3646,
+      "eval_samples_per_second": 512.933,
+      "eval_steps_per_second": 8.229,
       "step": 742
     },
     {
       "epoch": 8.0,
-      "grad_norm": 0.8066347241401672,
+      "grad_norm": 1.5386182069778442,
       "learning_rate": 4.600000000000001e-05,
-      "loss": 0.0148,
+      "loss": 0.0154,
       "step": 848
     },
     {
       "epoch": 8.0,
-      "eval_LOCATION_f1": 0.8289473684210525,
+      "eval_LOCATION_f1": 0.8456375838926175,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7974683544303798,
+      "eval_LOCATION_precision": 0.8289473684210527,
       "eval_LOCATION_recall": 0.863013698630137,
-      "eval_ORGANIZATION_f1": 0.6811594202898551,
+      "eval_ORGANIZATION_f1": 0.7199999999999999,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6438356164383562,
-      "eval_ORGANIZATION_recall": 0.7230769230769231,
-      "eval_PERSON_f1": 0.8932038834951457,
+      "eval_ORGANIZATION_precision": 0.6352941176470588,
+      "eval_ORGANIZATION_recall": 0.8307692307692308,
+      "eval_PERSON_f1": 0.9019607843137256,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8679245283018868,
+      "eval_PERSON_precision": 0.8846153846153846,
       "eval_PERSON_recall": 0.92,
-      "eval_QUANTITY_f1": 0.8524590163934426,
+      "eval_QUANTITY_f1": 0.7868852459016394,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.8125,
-      "eval_QUANTITY_recall": 0.896551724137931,
-      "eval_TIME_f1": 0.8823529411764706,
+      "eval_QUANTITY_precision": 0.75,
+      "eval_QUANTITY_recall": 0.8275862068965517,
+      "eval_TIME_f1": 0.8333333333333333,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8823529411764706,
+      "eval_TIME_precision": 0.7894736842105263,
       "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.20222125947475433,
-      "eval_overall_accuracy": 0.9638198375584544,
-      "eval_overall_f1": 0.8351648351648352,
-      "eval_overall_precision": 0.8063660477453581,
-      "eval_overall_recall": 0.8660968660968661,
-      "eval_runtime": 0.9481,
-      "eval_samples_per_second": 197.234,
-      "eval_steps_per_second": 3.164,
+      "eval_loss": 0.22032785415649414,
+      "eval_overall_accuracy": 0.9623430962343096,
+      "eval_overall_f1": 0.8373983739837397,
+      "eval_overall_precision": 0.7984496124031008,
+      "eval_overall_recall": 0.8803418803418803,
+      "eval_runtime": 0.3655,
+      "eval_samples_per_second": 511.693,
+      "eval_steps_per_second": 8.209,
       "step": 848
     },
     {
       "epoch": 9.0,
-      "grad_norm": 0.36969929933547974,
+      "grad_norm": 0.07042641192674637,
       "learning_rate": 4.55e-05,
-      "loss": 0.0136,
+      "loss": 0.0122,
       "step": 954
     },
     {
       "epoch": 9.0,
-      "eval_LOCATION_f1": 0.7770700636942677,
+      "eval_LOCATION_f1": 0.8421052631578948,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7261904761904762,
-      "eval_LOCATION_recall": 0.8356164383561644,
-      "eval_ORGANIZATION_f1": 0.6710526315789473,
+      "eval_LOCATION_precision": 0.810126582278481,
+      "eval_LOCATION_recall": 0.8767123287671232,
+      "eval_ORGANIZATION_f1": 0.7943262411347517,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.5862068965517241,
-      "eval_ORGANIZATION_recall": 0.7846153846153846,
-      "eval_PERSON_f1": 0.8810289389067525,
+      "eval_ORGANIZATION_precision": 0.7368421052631579,
+      "eval_ORGANIZATION_recall": 0.8615384615384616,
+      "eval_PERSON_f1": 0.9155844155844156,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8509316770186336,
-      "eval_PERSON_recall": 0.9133333333333333,
-      "eval_QUANTITY_f1": 0.7692307692307692,
+      "eval_PERSON_precision": 0.8924050632911392,
+      "eval_PERSON_recall": 0.94,
+      "eval_QUANTITY_f1": 0.7999999999999999,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6944444444444444,
-      "eval_QUANTITY_recall": 0.8620689655172413,
-      "eval_TIME_f1": 0.8219178082191781,
+      "eval_QUANTITY_precision": 0.7741935483870968,
+      "eval_QUANTITY_recall": 0.8275862068965517,
+      "eval_TIME_f1": 0.8055555555555555,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.7692307692307693,
-      "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.22667749226093292,
-      "eval_overall_accuracy": 0.9559438838296825,
-      "eval_overall_f1": 0.8021108179419525,
-      "eval_overall_precision": 0.7469287469287469,
-      "eval_overall_recall": 0.8660968660968661,
-      "eval_runtime": 0.9515,
-      "eval_samples_per_second": 196.53,
-      "eval_steps_per_second": 3.153,
+      "eval_TIME_precision": 0.7631578947368421,
+      "eval_TIME_recall": 0.8529411764705882,
+      "eval_loss": 0.20473507046699524,
+      "eval_overall_accuracy": 0.9657888259906473,
+      "eval_overall_f1": 0.8567530695770805,
+      "eval_overall_precision": 0.8219895287958116,
+      "eval_overall_recall": 0.8945868945868946,
+      "eval_runtime": 0.3664,
+      "eval_samples_per_second": 510.372,
+      "eval_steps_per_second": 8.188,
       "step": 954
     },
     {
       "epoch": 10.0,
-      "grad_norm": 3.057220220565796,
+      "grad_norm": 0.1457027643918991,
       "learning_rate": 4.5e-05,
-      "loss": 0.0107,
+      "loss": 0.0125,
       "step": 1060
     },
     {
       "epoch": 10.0,
-      "eval_LOCATION_f1": 0.8157894736842105,
+      "eval_LOCATION_f1": 0.8493150684931505,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7848101265822784,
+      "eval_LOCATION_precision": 0.8493150684931506,
       "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7272727272727273,
+      "eval_ORGANIZATION_f1": 0.8,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6292134831460674,
-      "eval_ORGANIZATION_recall": 0.8615384615384616,
-      "eval_PERSON_f1": 0.896774193548387,
+      "eval_ORGANIZATION_precision": 0.725,
+      "eval_ORGANIZATION_recall": 0.8923076923076924,
+      "eval_PERSON_f1": 0.9245901639344262,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.86875,
-      "eval_PERSON_recall": 0.9266666666666666,
-      "eval_QUANTITY_f1": 0.7692307692307692,
+      "eval_PERSON_precision": 0.9096774193548387,
+      "eval_PERSON_recall": 0.94,
+      "eval_QUANTITY_f1": 0.7761194029850746,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6944444444444444,
-      "eval_QUANTITY_recall": 0.8620689655172413,
-      "eval_TIME_f1": 0.9117647058823528,
+      "eval_QUANTITY_precision": 0.6842105263157895,
+      "eval_QUANTITY_recall": 0.896551724137931,
+      "eval_TIME_f1": 0.7532467532467532,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.9117647058823529,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.24919411540031433,
-      "eval_overall_accuracy": 0.9584051193699237,
-      "eval_overall_f1": 0.835781041388518,
-      "eval_overall_precision": 0.7864321608040201,
-      "eval_overall_recall": 0.8917378917378918,
-      "eval_runtime": 0.9484,
-      "eval_samples_per_second": 197.164,
-      "eval_steps_per_second": 3.163,
+      "eval_TIME_precision": 0.6744186046511628,
+      "eval_TIME_recall": 0.8529411764705882,
+      "eval_loss": 0.23434974253177643,
+      "eval_overall_accuracy": 0.9618508491262614,
+      "eval_overall_f1": 0.8540540540540541,
+      "eval_overall_precision": 0.8123393316195373,
+      "eval_overall_recall": 0.9002849002849003,
+      "eval_runtime": 0.3705,
+      "eval_samples_per_second": 504.677,
+      "eval_steps_per_second": 8.096,
       "step": 1060
     },
     {
       "epoch": 11.0,
-      "grad_norm": 0.5814297795295715,
+      "grad_norm": 0.9365283846855164,
       "learning_rate": 4.4500000000000004e-05,
-      "loss": 0.0087,
+      "loss": 0.0086,
       "step": 1166
     },
     {
       "epoch": 11.0,
-      "eval_LOCATION_f1": 0.8344370860927152,
+      "eval_LOCATION_f1": 0.7261904761904762,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.8076923076923077,
-      "eval_LOCATION_recall": 0.863013698630137,
-      "eval_ORGANIZATION_f1": 0.6666666666666667,
+      "eval_LOCATION_precision": 0.6421052631578947,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.7417218543046358,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.618421052631579,
-      "eval_ORGANIZATION_recall": 0.7230769230769231,
-      "eval_PERSON_f1": 0.887459807073955,
+      "eval_ORGANIZATION_precision": 0.6511627906976745,
+      "eval_ORGANIZATION_recall": 0.8615384615384616,
+      "eval_PERSON_f1": 0.9102564102564102,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8571428571428571,
-      "eval_PERSON_recall": 0.92,
-      "eval_QUANTITY_f1": 0.7619047619047619,
+      "eval_PERSON_precision": 0.8765432098765432,
+      "eval_PERSON_recall": 0.9466666666666667,
+      "eval_QUANTITY_f1": 0.7575757575757576,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.7058823529411765,
-      "eval_QUANTITY_recall": 0.8275862068965517,
-      "eval_TIME_f1": 0.861111111111111,
+      "eval_QUANTITY_precision": 0.6756756756756757,
+      "eval_QUANTITY_recall": 0.8620689655172413,
+      "eval_TIME_f1": 0.7435897435897436,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8157894736842105,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.23406828939914703,
-      "eval_overall_accuracy": 0.9620969726802855,
-      "eval_overall_f1": 0.8211382113821137,
-      "eval_overall_precision": 0.7829457364341085,
-      "eval_overall_recall": 0.8632478632478633,
-      "eval_runtime": 0.9329,
-      "eval_samples_per_second": 200.457,
-      "eval_steps_per_second": 3.216,
+      "eval_TIME_precision": 0.6590909090909091,
+      "eval_TIME_recall": 0.8529411764705882,
+      "eval_loss": 0.3139881491661072,
+      "eval_overall_accuracy": 0.9520059069652966,
+      "eval_overall_f1": 0.8077419354838709,
+      "eval_overall_precision": 0.7382075471698113,
+      "eval_overall_recall": 0.8917378917378918,
+      "eval_runtime": 0.3653,
+      "eval_samples_per_second": 511.912,
+      "eval_steps_per_second": 8.212,
       "step": 1166
     },
     {
       "epoch": 12.0,
-      "grad_norm": 1.747200608253479,
+      "grad_norm": 0.04872719943523407,
       "learning_rate": 4.4000000000000006e-05,
-      "loss": 0.0061,
+      "loss": 0.0069,
       "step": 1272
     },
     {
       "epoch": 12.0,
-      "eval_LOCATION_f1": 0.8513513513513513,
+      "eval_LOCATION_f1": 0.8472222222222222,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.84,
-      "eval_LOCATION_recall": 0.863013698630137,
-      "eval_ORGANIZATION_f1": 0.7651006711409396,
+      "eval_LOCATION_precision": 0.8591549295774648,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.7972972972972974,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6785714285714286,
-      "eval_ORGANIZATION_recall": 0.8769230769230769,
-      "eval_PERSON_f1": 0.888888888888889,
+      "eval_ORGANIZATION_precision": 0.7108433734939759,
+      "eval_ORGANIZATION_recall": 0.9076923076923077,
+      "eval_PERSON_f1": 0.8974358974358974,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8484848484848485,
+      "eval_PERSON_precision": 0.8641975308641975,
       "eval_PERSON_recall": 0.9333333333333333,
-      "eval_QUANTITY_f1": 0.8064516129032258,
+      "eval_QUANTITY_f1": 0.7761194029850746,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.7575757575757576,
-      "eval_QUANTITY_recall": 0.8620689655172413,
-      "eval_TIME_f1": 0.8405797101449276,
+      "eval_QUANTITY_precision": 0.6842105263157895,
+      "eval_QUANTITY_recall": 0.896551724137931,
+      "eval_TIME_f1": 0.75,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8285714285714286,
-      "eval_TIME_recall": 0.8529411764705882,
-      "eval_loss": 0.25188830494880676,
-      "eval_overall_accuracy": 0.9638198375584544,
-      "eval_overall_f1": 0.8452220726783309,
-      "eval_overall_precision": 0.8010204081632653,
-      "eval_overall_recall": 0.8945868945868946,
-      "eval_runtime": 0.9458,
-      "eval_samples_per_second": 197.707,
-      "eval_steps_per_second": 3.172,
+      "eval_TIME_precision": 0.6521739130434783,
+      "eval_TIME_recall": 0.8823529411764706,
+      "eval_loss": 0.25979745388031006,
+      "eval_overall_accuracy": 0.9618508491262614,
+      "eval_overall_f1": 0.8415446071904128,
+      "eval_overall_precision": 0.79,
+      "eval_overall_recall": 0.9002849002849003,
+      "eval_runtime": 0.3741,
+      "eval_samples_per_second": 499.861,
+      "eval_steps_per_second": 8.019,
       "step": 1272
     },
     {
       "epoch": 13.0,
-      "grad_norm": 2.369873285293579,
+      "grad_norm": 0.055007077753543854,
       "learning_rate": 4.35e-05,
-      "loss": 0.0052,
+      "loss": 0.0069,
       "step": 1378
     },
     {
       "epoch": 13.0,
-      "eval_LOCATION_f1": 0.8101265822784809,
+      "eval_LOCATION_f1": 0.7894736842105263,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7529411764705882,
-      "eval_LOCATION_recall": 0.8767123287671232,
-      "eval_ORGANIZATION_f1": 0.7114093959731543,
+      "eval_LOCATION_precision": 0.759493670886076,
+      "eval_LOCATION_recall": 0.821917808219178,
+      "eval_ORGANIZATION_f1": 0.7536231884057971,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6309523809523809,
-      "eval_ORGANIZATION_recall": 0.8153846153846154,
-      "eval_PERSON_f1": 0.896103896103896,
+      "eval_ORGANIZATION_precision": 0.7123287671232876,
+      "eval_ORGANIZATION_recall": 0.8,
+      "eval_PERSON_f1": 0.9185667752442996,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8734177215189873,
-      "eval_PERSON_recall": 0.92,
-      "eval_QUANTITY_f1": 0.8253968253968255,
+      "eval_PERSON_precision": 0.8980891719745223,
+      "eval_PERSON_recall": 0.94,
+      "eval_QUANTITY_f1": 0.8387096774193549,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.7647058823529411,
+      "eval_QUANTITY_precision": 0.7878787878787878,
       "eval_QUANTITY_recall": 0.896551724137931,
-      "eval_TIME_f1": 0.7837837837837837,
+      "eval_TIME_f1": 0.7631578947368423,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.725,
+      "eval_TIME_precision": 0.6904761904761905,
       "eval_TIME_recall": 0.8529411764705882,
-      "eval_loss": 0.26311174035072327,
-      "eval_overall_accuracy": 0.9593896135860202,
-      "eval_overall_f1": 0.824468085106383,
-      "eval_overall_precision": 0.773067331670823,
-      "eval_overall_recall": 0.8831908831908832,
-      "eval_runtime": 0.9614,
-      "eval_samples_per_second": 194.505,
-      "eval_steps_per_second": 3.12,
+      "eval_loss": 0.25240910053253174,
+      "eval_overall_accuracy": 0.9623430962343096,
+      "eval_overall_f1": 0.8380952380952382,
+      "eval_overall_precision": 0.8020833333333334,
+      "eval_overall_recall": 0.8774928774928775,
+      "eval_runtime": 0.3642,
+      "eval_samples_per_second": 513.499,
+      "eval_steps_per_second": 8.238,
       "step": 1378
     },
     {
       "epoch": 14.0,
-      "grad_norm": 0.004293783102184534,
+      "grad_norm": 0.006796263158321381,
       "learning_rate": 4.3e-05,
-      "loss": 0.0045,
+      "loss": 0.0048,
       "step": 1484
     },
     {
       "epoch": 14.0,
-      "eval_LOCATION_f1": 0.8051948051948052,
+      "eval_LOCATION_f1": 0.7848101265822783,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7654320987654321,
+      "eval_LOCATION_precision": 0.7294117647058823,
       "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7464788732394366,
+      "eval_ORGANIZATION_f1": 0.7943262411347517,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6883116883116883,
-      "eval_ORGANIZATION_recall": 0.8153846153846154,
-      "eval_PERSON_f1": 0.8938906752411575,
+      "eval_ORGANIZATION_precision": 0.7368421052631579,
+      "eval_ORGANIZATION_recall": 0.8615384615384616,
+      "eval_PERSON_f1": 0.9281045751633987,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8633540372670807,
-      "eval_PERSON_recall": 0.9266666666666666,
-      "eval_QUANTITY_f1": 0.9180327868852458,
+      "eval_PERSON_precision": 0.9102564102564102,
+      "eval_PERSON_recall": 0.9466666666666667,
+      "eval_QUANTITY_f1": 0.7540983606557378,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.875,
-      "eval_QUANTITY_recall": 0.9655172413793104,
-      "eval_TIME_f1": 0.8266666666666665,
+      "eval_QUANTITY_precision": 0.71875,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8378378378378379,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.7560975609756098,
+      "eval_TIME_precision": 0.775,
       "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.2545875608921051,
-      "eval_overall_accuracy": 0.9643120846665025,
-      "eval_overall_f1": 0.8425302826379543,
-      "eval_overall_precision": 0.798469387755102,
-      "eval_overall_recall": 0.8917378917378918,
-      "eval_runtime": 0.9409,
-      "eval_samples_per_second": 198.736,
-      "eval_steps_per_second": 3.188,
+      "eval_loss": 0.2732875943183899,
+      "eval_overall_accuracy": 0.9633275904504061,
+      "eval_overall_f1": 0.8486486486486486,
+      "eval_overall_precision": 0.8071979434447301,
+      "eval_overall_recall": 0.8945868945868946,
+      "eval_runtime": 0.3699,
+      "eval_samples_per_second": 505.513,
+      "eval_steps_per_second": 8.11,
       "step": 1484
     },
     {
       "epoch": 15.0,
-      "grad_norm": 0.1012561097741127,
+      "grad_norm": 0.03723064064979553,
       "learning_rate": 4.25e-05,
-      "loss": 0.0064,
+      "loss": 0.0069,
       "step": 1590
     },
     {
       "epoch": 15.0,
-      "eval_LOCATION_f1": 0.8378378378378378,
+      "eval_LOCATION_f1": 0.8235294117647058,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.8266666666666667,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7428571428571429,
+      "eval_LOCATION_precision": 0.7875,
+      "eval_LOCATION_recall": 0.863013698630137,
+      "eval_ORGANIZATION_f1": 0.7671232876712328,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6933333333333334,
-      "eval_ORGANIZATION_recall": 0.8,
-      "eval_PERSON_f1": 0.9096774193548387,
+      "eval_ORGANIZATION_precision": 0.691358024691358,
+      "eval_ORGANIZATION_recall": 0.8615384615384616,
+      "eval_PERSON_f1": 0.9281045751633987,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.88125,
-      "eval_PERSON_recall": 0.94,
-      "eval_QUANTITY_f1": 0.888888888888889,
+      "eval_PERSON_precision": 0.9102564102564102,
+      "eval_PERSON_recall": 0.9466666666666667,
+      "eval_QUANTITY_f1": 0.7868852459016394,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.8235294117647058,
-      "eval_QUANTITY_recall": 0.9655172413793104,
-      "eval_TIME_f1": 0.9275362318840579,
+      "eval_QUANTITY_precision": 0.75,
+      "eval_QUANTITY_recall": 0.8275862068965517,
+      "eval_TIME_f1": 0.9014084507042254,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.9142857142857143,
+      "eval_TIME_precision": 0.8648648648648649,
       "eval_TIME_recall": 0.9411764705882353,
-      "eval_loss": 0.2393232136964798,
-      "eval_overall_accuracy": 0.9667733202067438,
-      "eval_overall_f1": 0.863013698630137,
-      "eval_overall_precision": 0.8311345646437994,
-      "eval_overall_recall": 0.8974358974358975,
-      "eval_runtime": 0.956,
-      "eval_samples_per_second": 195.598,
-      "eval_steps_per_second": 3.138,
+      "eval_loss": 0.2588309347629547,
+      "eval_overall_accuracy": 0.9665271966527197,
+      "eval_overall_f1": 0.8602442333785618,
+      "eval_overall_precision": 0.8212435233160622,
+      "eval_overall_recall": 0.9031339031339032,
+      "eval_runtime": 0.3676,
+      "eval_samples_per_second": 508.743,
+      "eval_steps_per_second": 8.162,
       "step": 1590
     },
     {
       "epoch": 16.0,
-      "grad_norm": 0.15641357004642487,
+      "grad_norm": 1.1037182807922363,
       "learning_rate": 4.2e-05,
-      "loss": 0.005,
+      "loss": 0.0047,
       "step": 1696
     },
     {
       "epoch": 16.0,
-      "eval_LOCATION_f1": 0.8472222222222222,
+      "eval_LOCATION_f1": 0.7643312101910829,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.8591549295774648,
-      "eval_LOCATION_recall": 0.8356164383561644,
-      "eval_ORGANIZATION_f1": 0.7210884353741497,
+      "eval_LOCATION_precision": 0.7142857142857143,
+      "eval_LOCATION_recall": 0.821917808219178,
+      "eval_ORGANIZATION_f1": 0.7586206896551724,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6463414634146342,
-      "eval_ORGANIZATION_recall": 0.8153846153846154,
-      "eval_PERSON_f1": 0.912621359223301,
+      "eval_ORGANIZATION_precision": 0.6875,
+      "eval_ORGANIZATION_recall": 0.8461538461538461,
+      "eval_PERSON_f1": 0.9210526315789475,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8867924528301887,
-      "eval_PERSON_recall": 0.94,
-      "eval_QUANTITY_f1": 0.888888888888889,
+      "eval_PERSON_precision": 0.9090909090909091,
+      "eval_PERSON_recall": 0.9333333333333333,
+      "eval_QUANTITY_f1": 0.8064516129032258,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.8235294117647058,
-      "eval_QUANTITY_recall": 0.9655172413793104,
-      "eval_TIME_f1": 0.9014084507042254,
+      "eval_QUANTITY_precision": 0.7575757575757576,
+      "eval_QUANTITY_recall": 0.8620689655172413,
+      "eval_TIME_f1": 0.9411764705882353,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8648648648648649,
+      "eval_TIME_precision": 0.9411764705882353,
       "eval_TIME_recall": 0.9411764705882353,
-      "eval_loss": 0.2432704120874405,
-      "eval_overall_accuracy": 0.9643120846665025,
-      "eval_overall_f1": 0.8583106267029973,
-      "eval_overall_precision": 0.8224543080939948,
-      "eval_overall_recall": 0.8974358974358975,
-      "eval_runtime": 0.9491,
-      "eval_samples_per_second": 197.036,
-      "eval_steps_per_second": 3.161,
+      "eval_loss": 0.2755008637905121,
+      "eval_overall_accuracy": 0.9638198375584544,
+      "eval_overall_f1": 0.8478260869565218,
+      "eval_overall_precision": 0.8103896103896104,
+      "eval_overall_recall": 0.8888888888888888,
+      "eval_runtime": 0.3643,
+      "eval_samples_per_second": 513.309,
+      "eval_steps_per_second": 8.235,
       "step": 1696
     },
     {
       "epoch": 17.0,
-      "grad_norm": 0.7005181312561035,
+      "grad_norm": 0.04473419114947319,
       "learning_rate": 4.15e-05,
-      "loss": 0.0059,
+      "loss": 0.0049,
       "step": 1802
     },
     {
       "epoch": 17.0,
-      "eval_LOCATION_f1": 0.7712418300653595,
+      "eval_LOCATION_f1": 0.8266666666666668,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7375,
-      "eval_LOCATION_recall": 0.8082191780821918,
-      "eval_ORGANIZATION_f1": 0.7450980392156863,
+      "eval_LOCATION_precision": 0.8051948051948052,
+      "eval_LOCATION_recall": 0.8493150684931506,
+      "eval_ORGANIZATION_f1": 0.7445255474452555,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6477272727272727,
-      "eval_ORGANIZATION_recall": 0.8769230769230769,
-      "eval_PERSON_f1": 0.9038461538461539,
+      "eval_ORGANIZATION_precision": 0.7083333333333334,
+      "eval_ORGANIZATION_recall": 0.7846153846153846,
+      "eval_PERSON_f1": 0.9067524115755627,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8703703703703703,
+      "eval_PERSON_precision": 0.8757763975155279,
       "eval_PERSON_recall": 0.94,
-      "eval_QUANTITY_f1": 0.9180327868852458,
+      "eval_QUANTITY_f1": 0.7812500000000001,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.875,
-      "eval_QUANTITY_recall": 0.9655172413793104,
-      "eval_TIME_f1": 0.8823529411764706,
+      "eval_QUANTITY_precision": 0.7142857142857143,
+      "eval_QUANTITY_recall": 0.8620689655172413,
+      "eval_TIME_f1": 0.8051948051948051,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8823529411764706,
-      "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.27076834440231323,
-      "eval_overall_accuracy": 0.9625892197883338,
-      "eval_overall_f1": 0.8433734939759037,
-      "eval_overall_precision": 0.7954545454545454,
-      "eval_overall_recall": 0.8974358974358975,
-      "eval_runtime": 0.955,
-      "eval_samples_per_second": 195.821,
-      "eval_steps_per_second": 3.142,
+      "eval_TIME_precision": 0.7209302325581395,
+      "eval_TIME_recall": 0.9117647058823529,
+      "eval_loss": 0.27417999505996704,
+      "eval_overall_accuracy": 0.9645582082205267,
+      "eval_overall_f1": 0.8389715832205684,
+      "eval_overall_precision": 0.7989690721649485,
+      "eval_overall_recall": 0.8831908831908832,
+      "eval_runtime": 0.3646,
+      "eval_samples_per_second": 512.878,
+      "eval_steps_per_second": 8.228,
       "step": 1802
     },
     {
       "epoch": 18.0,
-      "grad_norm": 0.009474368765950203,
+      "grad_norm": 0.27922680974006653,
       "learning_rate": 4.1e-05,
-      "loss": 0.0034,
+      "loss": 0.0049,
       "step": 1908
     },
     {
       "epoch": 18.0,
-      "eval_LOCATION_f1": 0.8194444444444445,
+      "eval_LOCATION_f1": 0.8157894736842105,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.8309859154929577,
-      "eval_LOCATION_recall": 0.8082191780821918,
-      "eval_ORGANIZATION_f1": 0.7499999999999999,
+      "eval_LOCATION_precision": 0.7848101265822784,
+      "eval_LOCATION_recall": 0.8493150684931506,
+      "eval_ORGANIZATION_f1": 0.8115942028985507,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.631578947368421,
-      "eval_ORGANIZATION_recall": 0.9230769230769231,
-      "eval_PERSON_f1": 0.8990228013029316,
+      "eval_ORGANIZATION_precision": 0.7671232876712328,
+      "eval_ORGANIZATION_recall": 0.8615384615384616,
+      "eval_PERSON_f1": 0.9225806451612903,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8789808917197452,
-      "eval_PERSON_recall": 0.92,
-      "eval_QUANTITY_f1": 0.8709677419354839,
+      "eval_PERSON_precision": 0.89375,
+      "eval_PERSON_recall": 0.9533333333333334,
+      "eval_QUANTITY_f1": 0.8253968253968255,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.8181818181818182,
-      "eval_QUANTITY_recall": 0.9310344827586207,
-      "eval_TIME_f1": 0.9142857142857143,
+      "eval_QUANTITY_precision": 0.7647058823529411,
+      "eval_QUANTITY_recall": 0.896551724137931,
+      "eval_TIME_f1": 0.8108108108108107,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8888888888888888,
-      "eval_TIME_recall": 0.9411764705882353,
-      "eval_loss": 0.2898458242416382,
-      "eval_overall_accuracy": 0.9601279842480925,
-      "eval_overall_f1": 0.8506056527590848,
-      "eval_overall_precision": 0.8061224489795918,
-      "eval_overall_recall": 0.9002849002849003,
-      "eval_runtime": 0.9476,
-      "eval_samples_per_second": 197.344,
-      "eval_steps_per_second": 3.166,
+      "eval_TIME_precision": 0.75,
+      "eval_TIME_recall": 0.8823529411764706,
+      "eval_loss": 0.27639302611351013,
+      "eval_overall_accuracy": 0.965296578882599,
+      "eval_overall_f1": 0.8602442333785618,
+      "eval_overall_precision": 0.8212435233160622,
+      "eval_overall_recall": 0.9031339031339032,
+      "eval_runtime": 0.3676,
+      "eval_samples_per_second": 508.694,
+      "eval_steps_per_second": 8.161,
       "step": 1908
     },
     {
       "epoch": 19.0,
-      "grad_norm": 3.1910598278045654,
+      "grad_norm": 0.017078474164009094,
       "learning_rate": 4.05e-05,
-      "loss": 0.0069,
+      "loss": 0.0033,
       "step": 2014
     },
     {
       "epoch": 19.0,
-      "eval_LOCATION_f1": 0.832214765100671,
+      "eval_LOCATION_f1": 0.8133333333333332,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.8157894736842105,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.717948717948718,
+      "eval_LOCATION_precision": 0.7922077922077922,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.7941176470588235,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6153846153846154,
-      "eval_ORGANIZATION_recall": 0.8615384615384616,
-      "eval_PERSON_f1": 0.9185667752442996,
+      "eval_ORGANIZATION_precision": 0.7605633802816901,
+      "eval_ORGANIZATION_recall": 0.8307692307692308,
+      "eval_PERSON_f1": 0.9096774193548387,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8980891719745223,
+      "eval_PERSON_precision": 0.88125,
       "eval_PERSON_recall": 0.94,
-      "eval_QUANTITY_f1": 0.746268656716418,
+      "eval_QUANTITY_f1": 0.8571428571428571,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6578947368421053,
-      "eval_QUANTITY_recall": 0.8620689655172413,
-      "eval_TIME_f1": 0.8823529411764706,
+      "eval_QUANTITY_precision": 0.7941176470588235,
+      "eval_QUANTITY_recall": 0.9310344827586207,
+      "eval_TIME_f1": 0.8493150684931507,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8823529411764706,
-      "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.28969457745552063,
-      "eval_overall_accuracy": 0.9601279842480925,
-      "eval_overall_f1": 0.8406961178045514,
-      "eval_overall_precision": 0.7929292929292929,
+      "eval_TIME_precision": 0.7948717948717948,
+      "eval_TIME_recall": 0.9117647058823529,
+      "eval_loss": 0.2768174409866333,
+      "eval_overall_accuracy": 0.9670194437607679,
+      "eval_overall_f1": 0.8579234972677595,
+      "eval_overall_precision": 0.8241469816272966,
       "eval_overall_recall": 0.8945868945868946,
-      "eval_runtime": 0.9561,
-      "eval_samples_per_second": 195.594,
-      "eval_steps_per_second": 3.138,
+      "eval_runtime": 0.3647,
+      "eval_samples_per_second": 512.741,
+      "eval_steps_per_second": 8.226,
       "step": 2014
     },
     {
       "epoch": 20.0,
-      "grad_norm": 3.0964884757995605,
+      "grad_norm": 0.4910053014755249,
       "learning_rate": 4e-05,
-      "loss": 0.0054,
+      "loss": 0.0042,
       "step": 2120
     },
     {
       "epoch": 20.0,
-      "eval_LOCATION_f1": 0.7973856209150326,
+      "eval_LOCATION_f1": 0.7875,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7625,
-      "eval_LOCATION_recall": 0.8356164383561644,
-      "eval_ORGANIZATION_f1": 0.7169811320754718,
+      "eval_LOCATION_precision": 0.7241379310344828,
+      "eval_LOCATION_recall": 0.863013698630137,
+      "eval_ORGANIZATION_f1": 0.7647058823529411,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6063829787234043,
-      "eval_ORGANIZATION_recall": 0.8769230769230769,
-      "eval_PERSON_f1": 0.9180327868852459,
+      "eval_ORGANIZATION_precision": 0.7323943661971831,
+      "eval_ORGANIZATION_recall": 0.8,
+      "eval_PERSON_f1": 0.9215686274509804,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.9032258064516129,
-      "eval_PERSON_recall": 0.9333333333333333,
-      "eval_QUANTITY_f1": 0.888888888888889,
+      "eval_PERSON_precision": 0.9038461538461539,
+      "eval_PERSON_recall": 0.94,
+      "eval_QUANTITY_f1": 0.8253968253968255,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.8235294117647058,
-      "eval_QUANTITY_recall": 0.9655172413793104,
-      "eval_TIME_f1": 0.8450704225352113,
+      "eval_QUANTITY_precision": 0.7647058823529411,
+      "eval_QUANTITY_recall": 0.896551724137931,
+      "eval_TIME_f1": 0.794871794871795,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8108108108108109,
-      "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.2853837013244629,
-      "eval_overall_accuracy": 0.9608663549101649,
-      "eval_overall_f1": 0.8415446071904128,
-      "eval_overall_precision": 0.79,
-      "eval_overall_recall": 0.9002849002849003,
-      "eval_runtime": 1.0257,
-      "eval_samples_per_second": 182.315,
-      "eval_steps_per_second": 2.925,
+      "eval_TIME_precision": 0.7045454545454546,
+      "eval_TIME_recall": 0.9117647058823529,
+      "eval_loss": 0.30328768491744995,
+      "eval_overall_accuracy": 0.9633275904504061,
+      "eval_overall_f1": 0.8425302826379543,
+      "eval_overall_precision": 0.798469387755102,
+      "eval_overall_recall": 0.8917378917378918,
+      "eval_runtime": 0.3736,
+      "eval_samples_per_second": 500.566,
+      "eval_steps_per_second": 8.03,
       "step": 2120
     },
     {
       "epoch": 21.0,
-      "grad_norm": 5.442500114440918,
+      "grad_norm": 1.6962295770645142,
       "learning_rate": 3.9500000000000005e-05,
-      "loss": 0.0064,
+      "loss": 0.0036,
       "step": 2226
     },
     {
       "epoch": 21.0,
-      "eval_LOCATION_f1": 0.7837837837837838,
+      "eval_LOCATION_f1": 0.8243243243243243,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7733333333333333,
-      "eval_LOCATION_recall": 0.7945205479452054,
-      "eval_ORGANIZATION_f1": 0.7215189873417722,
+      "eval_LOCATION_precision": 0.8133333333333334,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.7769784172661871,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6129032258064516,
-      "eval_ORGANIZATION_recall": 0.8769230769230769,
-      "eval_PERSON_f1": 0.9150326797385621,
+      "eval_ORGANIZATION_precision": 0.7297297297297297,
+      "eval_ORGANIZATION_recall": 0.8307692307692308,
+      "eval_PERSON_f1": 0.9250814332247558,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8974358974358975,
-      "eval_PERSON_recall": 0.9333333333333333,
-      "eval_QUANTITY_f1": 0.7761194029850746,
+      "eval_PERSON_precision": 0.9044585987261147,
+      "eval_PERSON_recall": 0.9466666666666667,
+      "eval_QUANTITY_f1": 0.8571428571428571,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.6842105263157895,
-      "eval_QUANTITY_recall": 0.896551724137931,
-      "eval_TIME_f1": 0.8333333333333333,
+      "eval_QUANTITY_precision": 0.7941176470588235,
+      "eval_QUANTITY_recall": 0.9310344827586207,
+      "eval_TIME_f1": 0.8767123287671232,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.7894736842105263,
-      "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.26608648896217346,
-      "eval_overall_accuracy": 0.9591434900319961,
-      "eval_overall_f1": 0.8282290279627165,
-      "eval_overall_precision": 0.7775,
-      "eval_overall_recall": 0.886039886039886,
-      "eval_runtime": 0.9644,
-      "eval_samples_per_second": 193.895,
-      "eval_steps_per_second": 3.111,
+      "eval_TIME_precision": 0.8205128205128205,
+      "eval_TIME_recall": 0.9411764705882353,
+      "eval_loss": 0.26916593313217163,
+      "eval_overall_accuracy": 0.9655427024366232,
+      "eval_overall_f1": 0.8657534246575342,
+      "eval_overall_precision": 0.8337730870712401,
+      "eval_overall_recall": 0.9002849002849003,
+      "eval_runtime": 0.3622,
+      "eval_samples_per_second": 516.23,
+      "eval_steps_per_second": 8.282,
       "step": 2226
     },
     {
       "epoch": 22.0,
-      "grad_norm": 0.017967188730835915,
+      "grad_norm": 0.29587170481681824,
       "learning_rate": 3.9000000000000006e-05,
-      "loss": 0.004,
+      "loss": 0.0073,
       "step": 2332
     },
     {
       "epoch": 22.0,
-      "eval_LOCATION_f1": 0.8157894736842105,
+      "eval_LOCATION_f1": 0.7682926829268293,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7848101265822784,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7516778523489933,
+      "eval_LOCATION_precision": 0.6923076923076923,
+      "eval_LOCATION_recall": 0.863013698630137,
+      "eval_ORGANIZATION_f1": 0.7248322147651007,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6666666666666666,
-      "eval_ORGANIZATION_recall": 0.8615384615384616,
+      "eval_ORGANIZATION_precision": 0.6428571428571429,
+      "eval_ORGANIZATION_recall": 0.8307692307692308,
       "eval_PERSON_f1": 0.912621359223301,
       "eval_PERSON_number": 150,
       "eval_PERSON_precision": 0.8867924528301887,
       "eval_PERSON_recall": 0.94,
-      "eval_QUANTITY_f1": 0.9032258064516129,
+      "eval_QUANTITY_f1": 0.8333333333333334,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.8484848484848485,
-      "eval_QUANTITY_recall": 0.9655172413793104,
-      "eval_TIME_f1": 0.861111111111111,
+      "eval_QUANTITY_precision": 0.8064516129032258,
+      "eval_QUANTITY_recall": 0.8620689655172413,
+      "eval_TIME_f1": 0.8157894736842106,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8157894736842105,
+      "eval_TIME_precision": 0.7380952380952381,
       "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.2437296360731125,
-      "eval_overall_accuracy": 0.9648043317745508,
-      "eval_overall_f1": 0.8548387096774193,
-      "eval_overall_precision": 0.8091603053435115,
-      "eval_overall_recall": 0.905982905982906,
-      "eval_runtime": 0.9549,
-      "eval_samples_per_second": 195.826,
-      "eval_steps_per_second": 3.142,
+      "eval_loss": 0.3261459171772003,
+      "eval_overall_accuracy": 0.9574206251538272,
+      "eval_overall_f1": 0.828496042216359,
+      "eval_overall_precision": 0.7714987714987716,
+      "eval_overall_recall": 0.8945868945868946,
+      "eval_runtime": 0.3664,
+      "eval_samples_per_second": 510.427,
+      "eval_steps_per_second": 8.189,
       "step": 2332
     },
     {
       "epoch": 23.0,
-      "grad_norm": 0.02713579311966896,
+      "grad_norm": 0.05311551317572594,
       "learning_rate": 3.85e-05,
-      "loss": 0.004,
+      "loss": 0.0024,
       "step": 2438
     },
     {
       "epoch": 23.0,
-      "eval_LOCATION_f1": 0.7973856209150326,
+      "eval_LOCATION_f1": 0.8211920529801324,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7625,
-      "eval_LOCATION_recall": 0.8356164383561644,
-      "eval_ORGANIZATION_f1": 0.7412587412587412,
+      "eval_LOCATION_precision": 0.7948717948717948,
+      "eval_LOCATION_recall": 0.8493150684931506,
+      "eval_ORGANIZATION_f1": 0.7862068965517242,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6794871794871795,
-      "eval_ORGANIZATION_recall": 0.8153846153846154,
-      "eval_PERSON_f1": 0.896103896103896,
+      "eval_ORGANIZATION_precision": 0.7125,
+      "eval_ORGANIZATION_recall": 0.8769230769230769,
+      "eval_PERSON_f1": 0.9114754098360656,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8734177215189873,
-      "eval_PERSON_recall": 0.92,
-      "eval_QUANTITY_f1": 0.9,
+      "eval_PERSON_precision": 0.896774193548387,
+      "eval_PERSON_recall": 0.9266666666666666,
+      "eval_QUANTITY_f1": 0.7812500000000001,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.8709677419354839,
-      "eval_QUANTITY_recall": 0.9310344827586207,
-      "eval_TIME_f1": 0.8571428571428571,
+      "eval_QUANTITY_precision": 0.7142857142857143,
+      "eval_QUANTITY_recall": 0.8620689655172413,
+      "eval_TIME_f1": 0.8533333333333334,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8333333333333334,
-      "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.23154456913471222,
-      "eval_overall_accuracy": 0.9660349495446714,
-      "eval_overall_f1": 0.8419618528610354,
-      "eval_overall_precision": 0.8067885117493473,
-      "eval_overall_recall": 0.8803418803418803,
-      "eval_runtime": 0.9493,
-      "eval_samples_per_second": 196.991,
-      "eval_steps_per_second": 3.16,
+      "eval_TIME_precision": 0.7804878048780488,
+      "eval_TIME_recall": 0.9411764705882353,
+      "eval_loss": 0.2862998843193054,
+      "eval_overall_accuracy": 0.9645582082205267,
+      "eval_overall_f1": 0.8513513513513514,
+      "eval_overall_precision": 0.8097686375321337,
+      "eval_overall_recall": 0.8974358974358975,
+      "eval_runtime": 0.3676,
+      "eval_samples_per_second": 508.763,
+      "eval_steps_per_second": 8.162,
       "step": 2438
     },
     {
       "epoch": 24.0,
-      "grad_norm": 0.04597410559654236,
+      "grad_norm": 0.008164864033460617,
       "learning_rate": 3.8e-05,
-      "loss": 0.0036,
+      "loss": 0.0031,
       "step": 2544
     },
     {
       "epoch": 24.0,
-      "eval_LOCATION_f1": 0.8533333333333334,
+      "eval_LOCATION_f1": 0.8076923076923077,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.8311688311688312,
-      "eval_LOCATION_recall": 0.8767123287671232,
-      "eval_ORGANIZATION_f1": 0.7755102040816327,
+      "eval_LOCATION_precision": 0.7590361445783133,
+      "eval_LOCATION_recall": 0.863013698630137,
+      "eval_ORGANIZATION_f1": 0.7581699346405228,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6951219512195121,
-      "eval_ORGANIZATION_recall": 0.8769230769230769,
-      "eval_PERSON_f1": 0.896774193548387,
+      "eval_ORGANIZATION_precision": 0.6590909090909091,
+      "eval_ORGANIZATION_recall": 0.8923076923076924,
+      "eval_PERSON_f1": 0.9255663430420712,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.86875,
-      "eval_PERSON_recall": 0.9266666666666666,
-      "eval_QUANTITY_f1": 0.8709677419354839,
+      "eval_PERSON_precision": 0.89937106918239,
+      "eval_PERSON_recall": 0.9533333333333334,
+      "eval_QUANTITY_f1": 0.8571428571428571,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.8181818181818182,
+      "eval_QUANTITY_precision": 0.7941176470588235,
       "eval_QUANTITY_recall": 0.9310344827586207,
-      "eval_TIME_f1": 0.8732394366197184,
+      "eval_TIME_f1": 0.8219178082191781,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8378378378378378,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.2499096691608429,
-      "eval_overall_accuracy": 0.9643120846665025,
-      "eval_overall_f1": 0.8594594594594593,
-      "eval_overall_precision": 0.8174807197943444,
-      "eval_overall_recall": 0.905982905982906,
-      "eval_runtime": 0.9457,
-      "eval_samples_per_second": 197.732,
-      "eval_steps_per_second": 3.172,
-      "step": 2544
-    },
-    {
+      "eval_TIME_precision": 0.7692307692307693,
+      "eval_TIME_recall": 0.8823529411764706,
+      "eval_loss": 0.30451545119285583,
+      "eval_overall_accuracy": 0.963081466896382,
+      "eval_overall_f1": 0.8514588859416445,
+      "eval_overall_precision": 0.7965260545905707,
+      "eval_overall_recall": 0.9145299145299145,
+      "eval_runtime": 0.3637,
+      "eval_samples_per_second": 514.093,
+      "eval_steps_per_second": 8.247,
+      "step": 2544
+    },
+    {
       "epoch": 25.0,
-      "grad_norm": 0.004746124614030123,
+      "grad_norm": 0.0027474737726151943,
       "learning_rate": 3.7500000000000003e-05,
-      "loss": 0.0042,
+      "loss": 0.0033,
       "step": 2650
     },
     {
       "epoch": 25.0,
-      "eval_LOCATION_f1": 0.8311688311688312,
+      "eval_LOCATION_f1": 0.7924528301886793,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7901234567901234,
-      "eval_LOCATION_recall": 0.8767123287671232,
-      "eval_ORGANIZATION_f1": 0.7368421052631579,
+      "eval_LOCATION_precision": 0.7325581395348837,
+      "eval_LOCATION_recall": 0.863013698630137,
+      "eval_ORGANIZATION_f1": 0.7517730496453899,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6436781609195402,
-      "eval_ORGANIZATION_recall": 0.8615384615384616,
-      "eval_PERSON_f1": 0.9025974025974027,
+      "eval_ORGANIZATION_precision": 0.6973684210526315,
+      "eval_ORGANIZATION_recall": 0.8153846153846154,
+      "eval_PERSON_f1": 0.912621359223301,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.879746835443038,
-      "eval_PERSON_recall": 0.9266666666666666,
-      "eval_QUANTITY_f1": 0.8253968253968255,
+      "eval_PERSON_precision": 0.8867924528301887,
+      "eval_PERSON_recall": 0.94,
+      "eval_QUANTITY_f1": 0.7936507936507937,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.7647058823529411,
-      "eval_QUANTITY_recall": 0.896551724137931,
-      "eval_TIME_f1": 0.8888888888888888,
+      "eval_QUANTITY_precision": 0.7352941176470589,
+      "eval_QUANTITY_recall": 0.8620689655172413,
+      "eval_TIME_f1": 0.8266666666666665,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8421052631578947,
-      "eval_TIME_recall": 0.9411764705882353,
-      "eval_loss": 0.2664180397987366,
-      "eval_overall_accuracy": 0.9650504553285749,
-      "eval_overall_f1": 0.8464619492656876,
-      "eval_overall_precision": 0.7964824120603015,
-      "eval_overall_recall": 0.9031339031339032,
-      "eval_runtime": 0.9509,
-      "eval_samples_per_second": 196.66,
-      "eval_steps_per_second": 3.155,
+      "eval_TIME_precision": 0.7560975609756098,
+      "eval_TIME_recall": 0.9117647058823529,
+      "eval_loss": 0.3293425738811493,
+      "eval_overall_accuracy": 0.9608663549101649,
+      "eval_overall_f1": 0.8380187416331994,
+      "eval_overall_precision": 0.7904040404040404,
+      "eval_overall_recall": 0.8917378917378918,
+      "eval_runtime": 0.3647,
+      "eval_samples_per_second": 512.757,
+      "eval_steps_per_second": 8.226,
       "step": 2650
     },
     {
       "epoch": 26.0,
-      "grad_norm": 0.010891449637711048,
+      "grad_norm": 0.0054503269493579865,
       "learning_rate": 3.7e-05,
-      "loss": 0.0026,
+      "loss": 0.0029,
       "step": 2756
     },
     {
       "epoch": 26.0,
-      "eval_LOCATION_f1": 0.8258064516129032,
+      "eval_LOCATION_f1": 0.8441558441558442,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7804878048780488,
-      "eval_LOCATION_recall": 0.8767123287671232,
-      "eval_ORGANIZATION_f1": 0.738255033557047,
+      "eval_LOCATION_precision": 0.8024691358024691,
+      "eval_LOCATION_recall": 0.8904109589041096,
+      "eval_ORGANIZATION_f1": 0.76,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6547619047619048,
-      "eval_ORGANIZATION_recall": 0.8461538461538461,
-      "eval_PERSON_f1": 0.8996763754045306,
+      "eval_ORGANIZATION_precision": 0.6705882352941176,
+      "eval_ORGANIZATION_recall": 0.8769230769230769,
+      "eval_PERSON_f1": 0.9250814332247558,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8742138364779874,
-      "eval_PERSON_recall": 0.9266666666666666,
+      "eval_PERSON_precision": 0.9044585987261147,
+      "eval_PERSON_recall": 0.9466666666666667,
       "eval_QUANTITY_f1": 0.8064516129032258,
       "eval_QUANTITY_number": 29,
       "eval_QUANTITY_precision": 0.7575757575757576,
       "eval_QUANTITY_recall": 0.8620689655172413,
-      "eval_TIME_f1": 0.8857142857142858,
+      "eval_TIME_f1": 0.8533333333333334,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8611111111111112,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.25709718465805054,
-      "eval_overall_accuracy": 0.963081466896382,
-      "eval_overall_f1": 0.8429530201342281,
-      "eval_overall_precision": 0.7969543147208121,
-      "eval_overall_recall": 0.8945868945868946,
-      "eval_runtime": 0.9562,
-      "eval_samples_per_second": 195.559,
-      "eval_steps_per_second": 3.137,
+      "eval_TIME_precision": 0.7804878048780488,
+      "eval_TIME_recall": 0.9411764705882353,
+      "eval_loss": 0.2977321743965149,
+      "eval_overall_accuracy": 0.9635737140044303,
+      "eval_overall_f1": 0.8582887700534759,
+      "eval_overall_precision": 0.8085642317380353,
+      "eval_overall_recall": 0.9145299145299145,
+      "eval_runtime": 0.3805,
+      "eval_samples_per_second": 491.399,
+      "eval_steps_per_second": 7.883,
       "step": 2756
     },
     {
       "epoch": 27.0,
-      "grad_norm": 0.005209265276789665,
+      "grad_norm": 8.443046569824219,
       "learning_rate": 3.65e-05,
-      "loss": 0.0017,
+      "loss": 0.0035,
       "step": 2862
     },
     {
       "epoch": 27.0,
-      "eval_LOCATION_f1": 0.821917808219178,
+      "eval_LOCATION_f1": 0.832214765100671,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.821917808219178,
-      "eval_LOCATION_recall": 0.821917808219178,
-      "eval_ORGANIZATION_f1": 0.7651006711409396,
+      "eval_LOCATION_precision": 0.8157894736842105,
+      "eval_LOCATION_recall": 0.8493150684931506,
+      "eval_ORGANIZATION_f1": 0.7532467532467532,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6785714285714286,
-      "eval_ORGANIZATION_recall": 0.8769230769230769,
-      "eval_PERSON_f1": 0.896103896103896,
+      "eval_ORGANIZATION_precision": 0.651685393258427,
+      "eval_ORGANIZATION_recall": 0.8923076923076924,
+      "eval_PERSON_f1": 0.9281045751633987,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8734177215189873,
-      "eval_PERSON_recall": 0.92,
-      "eval_QUANTITY_f1": 0.8666666666666666,
+      "eval_PERSON_precision": 0.9102564102564102,
+      "eval_PERSON_recall": 0.9466666666666667,
+      "eval_QUANTITY_f1": 0.8064516129032258,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.8387096774193549,
-      "eval_QUANTITY_recall": 0.896551724137931,
-      "eval_TIME_f1": 0.861111111111111,
+      "eval_QUANTITY_precision": 0.7575757575757576,
+      "eval_QUANTITY_recall": 0.8620689655172413,
+      "eval_TIME_f1": 0.8533333333333334,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8157894736842105,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.23790057003498077,
-      "eval_overall_accuracy": 0.9667733202067438,
-      "eval_overall_f1": 0.8489795918367347,
-      "eval_overall_precision": 0.8125,
-      "eval_overall_recall": 0.8888888888888888,
-      "eval_runtime": 0.9449,
-      "eval_samples_per_second": 197.905,
-      "eval_steps_per_second": 3.175,
+      "eval_TIME_precision": 0.7804878048780488,
+      "eval_TIME_recall": 0.9411764705882353,
+      "eval_loss": 0.3316052556037903,
+      "eval_overall_accuracy": 0.963081466896382,
+      "eval_overall_f1": 0.8552278820375334,
+      "eval_overall_precision": 0.8075949367088607,
+      "eval_overall_recall": 0.9088319088319088,
+      "eval_runtime": 0.3653,
+      "eval_samples_per_second": 511.94,
+      "eval_steps_per_second": 8.213,
       "step": 2862
     },
     {
       "epoch": 28.0,
-      "grad_norm": 0.02614201046526432,
+      "grad_norm": 0.8292134404182434,
       "learning_rate": 3.6e-05,
-      "loss": 0.0031,
+      "loss": 0.0076,
       "step": 2968
     },
     {
       "epoch": 28.0,
-      "eval_LOCATION_f1": 0.8211920529801324,
+      "eval_LOCATION_f1": 0.8366013071895425,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7948717948717948,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7913669064748202,
+      "eval_LOCATION_precision": 0.8,
+      "eval_LOCATION_recall": 0.8767123287671232,
+      "eval_ORGANIZATION_f1": 0.736,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.7432432432432432,
-      "eval_ORGANIZATION_recall": 0.8461538461538461,
-      "eval_PERSON_f1": 0.9061488673139159,
+      "eval_ORGANIZATION_precision": 0.7666666666666667,
+      "eval_ORGANIZATION_recall": 0.7076923076923077,
+      "eval_PERSON_f1": 0.9120521172638437,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8805031446540881,
+      "eval_PERSON_precision": 0.89171974522293,
       "eval_PERSON_recall": 0.9333333333333333,
       "eval_QUANTITY_f1": 0.8333333333333334,
       "eval_QUANTITY_number": 29,
       "eval_QUANTITY_precision": 0.8064516129032258,
       "eval_QUANTITY_recall": 0.8620689655172413,
-      "eval_TIME_f1": 0.9014084507042254,
+      "eval_TIME_f1": 0.8767123287671232,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8648648648648649,
+      "eval_TIME_precision": 0.8205128205128205,
       "eval_TIME_recall": 0.9411764705882353,
-      "eval_loss": 0.23548346757888794,
-      "eval_overall_accuracy": 0.9702190499630815,
-      "eval_overall_f1": 0.8602739726027397,
-      "eval_overall_precision": 0.8284960422163589,
-      "eval_overall_recall": 0.8945868945868946,
-      "eval_runtime": 0.9552,
-      "eval_samples_per_second": 195.775,
-      "eval_steps_per_second": 3.141,
+      "eval_loss": 0.26178210973739624,
+      "eval_overall_accuracy": 0.9645582082205267,
+      "eval_overall_f1": 0.8551532033426184,
+      "eval_overall_precision": 0.8365122615803815,
+      "eval_overall_recall": 0.8746438746438746,
+      "eval_runtime": 0.3641,
+      "eval_samples_per_second": 513.548,
+      "eval_steps_per_second": 8.239,
       "step": 2968
     },
     {
       "epoch": 29.0,
-      "grad_norm": 0.0011191533412784338,
+      "grad_norm": 0.1424623727798462,
       "learning_rate": 3.55e-05,
-      "loss": 0.0022,
+      "loss": 0.0027,
       "step": 3074
     },
     {
       "epoch": 29.0,
-      "eval_LOCATION_f1": 0.8101265822784809,
+      "eval_LOCATION_f1": 0.802547770700637,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7529411764705882,
-      "eval_LOCATION_recall": 0.8767123287671232,
-      "eval_ORGANIZATION_f1": 0.7194244604316546,
+      "eval_LOCATION_precision": 0.75,
+      "eval_LOCATION_recall": 0.863013698630137,
+      "eval_ORGANIZATION_f1": 0.75,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6756756756756757,
-      "eval_ORGANIZATION_recall": 0.7692307692307693,
-      "eval_PERSON_f1": 0.8990228013029316,
+      "eval_ORGANIZATION_precision": 0.6551724137931034,
+      "eval_ORGANIZATION_recall": 0.8769230769230769,
+      "eval_PERSON_f1": 0.9155844155844156,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8789808917197452,
-      "eval_PERSON_recall": 0.92,
-      "eval_QUANTITY_f1": 0.8852459016393444,
+      "eval_PERSON_precision": 0.8924050632911392,
+      "eval_PERSON_recall": 0.94,
+      "eval_QUANTITY_f1": 0.7868852459016394,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.84375,
-      "eval_QUANTITY_recall": 0.9310344827586207,
-      "eval_TIME_f1": 0.9411764705882353,
+      "eval_QUANTITY_precision": 0.75,
+      "eval_QUANTITY_recall": 0.8275862068965517,
+      "eval_TIME_f1": 0.7654320987654322,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.9411764705882353,
-      "eval_TIME_recall": 0.9411764705882353,
-      "eval_loss": 0.24776005744934082,
-      "eval_overall_accuracy": 0.9687423086389367,
-      "eval_overall_f1": 0.8485675306957708,
-      "eval_overall_precision": 0.8141361256544503,
-      "eval_overall_recall": 0.886039886039886,
-      "eval_runtime": 1.0172,
-      "eval_samples_per_second": 183.845,
-      "eval_steps_per_second": 2.949,
+      "eval_TIME_precision": 0.6595744680851063,
+      "eval_TIME_recall": 0.9117647058823529,
+      "eval_loss": 0.33089107275009155,
+      "eval_overall_accuracy": 0.958897366477972,
+      "eval_overall_f1": 0.8326745718050065,
+      "eval_overall_precision": 0.7745098039215687,
+      "eval_overall_recall": 0.9002849002849003,
+      "eval_runtime": 0.363,
+      "eval_samples_per_second": 515.091,
+      "eval_steps_per_second": 8.263,
       "step": 3074
     },
     {
       "epoch": 30.0,
-      "grad_norm": 0.001416489016264677,
+      "grad_norm": 0.010981522500514984,
       "learning_rate": 3.5e-05,
-      "loss": 0.0016,
+      "loss": 0.0025,
       "step": 3180
     },
     {
       "epoch": 30.0,
-      "eval_LOCATION_f1": 0.7870967741935484,
+      "eval_LOCATION_f1": 0.8366013071895425,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7439024390243902,
-      "eval_LOCATION_recall": 0.8356164383561644,
-      "eval_ORGANIZATION_f1": 0.757142857142857,
+      "eval_LOCATION_precision": 0.8,
+      "eval_LOCATION_recall": 0.8767123287671232,
+      "eval_ORGANIZATION_f1": 0.7755102040816327,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.7066666666666667,
-      "eval_ORGANIZATION_recall": 0.8153846153846154,
-      "eval_PERSON_f1": 0.90032154340836,
+      "eval_ORGANIZATION_precision": 0.6951219512195121,
+      "eval_ORGANIZATION_recall": 0.8769230769230769,
+      "eval_PERSON_f1": 0.9215686274509804,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8695652173913043,
-      "eval_PERSON_recall": 0.9333333333333333,
-      "eval_QUANTITY_f1": 0.8666666666666666,
+      "eval_PERSON_precision": 0.9038461538461539,
+      "eval_PERSON_recall": 0.94,
+      "eval_QUANTITY_f1": 0.8196721311475409,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.8387096774193549,
-      "eval_QUANTITY_recall": 0.896551724137931,
-      "eval_TIME_f1": 0.8888888888888888,
+      "eval_QUANTITY_precision": 0.78125,
+      "eval_QUANTITY_recall": 0.8620689655172413,
+      "eval_TIME_f1": 0.8450704225352113,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8421052631578947,
-      "eval_TIME_recall": 0.9411764705882353,
-      "eval_loss": 0.2568590044975281,
-      "eval_overall_accuracy": 0.9667733202067438,
-      "eval_overall_f1": 0.8455284552845528,
-      "eval_overall_precision": 0.8062015503875969,
-      "eval_overall_recall": 0.8888888888888888,
-      "eval_runtime": 0.9566,
-      "eval_samples_per_second": 195.488,
-      "eval_steps_per_second": 3.136,
+      "eval_TIME_precision": 0.8108108108108109,
+      "eval_TIME_recall": 0.8823529411764706,
+      "eval_loss": 0.30917105078697205,
+      "eval_overall_accuracy": 0.9648043317745508,
+      "eval_overall_f1": 0.8590785907859079,
+      "eval_overall_precision": 0.8191214470284238,
+      "eval_overall_recall": 0.9031339031339032,
+      "eval_runtime": 0.3731,
+      "eval_samples_per_second": 501.246,
+      "eval_steps_per_second": 8.041,
       "step": 3180
     },
     {
       "epoch": 31.0,
-      "grad_norm": 0.5795331597328186,
+      "grad_norm": 0.48626771569252014,
       "learning_rate": 3.45e-05,
-      "loss": 0.0027,
+      "loss": 0.003,
       "step": 3286
     },
     {
       "epoch": 31.0,
-      "eval_LOCATION_f1": 0.8194444444444445,
+      "eval_LOCATION_f1": 0.8157894736842105,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.8309859154929577,
-      "eval_LOCATION_recall": 0.8082191780821918,
-      "eval_ORGANIZATION_f1": 0.7733333333333334,
+      "eval_LOCATION_precision": 0.7848101265822784,
+      "eval_LOCATION_recall": 0.8493150684931506,
+      "eval_ORGANIZATION_f1": 0.7724137931034483,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6823529411764706,
-      "eval_ORGANIZATION_recall": 0.8923076923076924,
-      "eval_PERSON_f1": 0.9150326797385621,
+      "eval_ORGANIZATION_precision": 0.7,
+      "eval_ORGANIZATION_recall": 0.8615384615384616,
+      "eval_PERSON_f1": 0.9185667752442996,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8974358974358975,
-      "eval_PERSON_recall": 0.9333333333333333,
-      "eval_QUANTITY_f1": 0.8666666666666666,
+      "eval_PERSON_precision": 0.8980891719745223,
+      "eval_PERSON_recall": 0.94,
+      "eval_QUANTITY_f1": 0.8307692307692308,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.8387096774193549,
-      "eval_QUANTITY_recall": 0.896551724137931,
-      "eval_TIME_f1": 0.8450704225352113,
+      "eval_QUANTITY_precision": 0.75,
+      "eval_QUANTITY_recall": 0.9310344827586207,
+      "eval_TIME_f1": 0.7792207792207793,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8108108108108109,
+      "eval_TIME_precision": 0.6976744186046512,
       "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.26583045721054077,
-      "eval_overall_accuracy": 0.9648043317745508,
-      "eval_overall_f1": 0.8563611491108072,
-      "eval_overall_precision": 0.8236842105263158,
-      "eval_overall_recall": 0.8917378917378918,
-      "eval_runtime": 0.9478,
-      "eval_samples_per_second": 197.305,
-      "eval_steps_per_second": 3.165,
+      "eval_loss": 0.3233889043331146,
+      "eval_overall_accuracy": 0.9591434900319961,
+      "eval_overall_f1": 0.8471849865951743,
+      "eval_overall_precision": 0.8,
+      "eval_overall_recall": 0.9002849002849003,
+      "eval_runtime": 0.3687,
+      "eval_samples_per_second": 507.197,
+      "eval_steps_per_second": 8.137,
       "step": 3286
     },
     {
       "epoch": 32.0,
-      "grad_norm": 0.0013437798479571939,
+      "grad_norm": 0.002596928272396326,
       "learning_rate": 3.4000000000000007e-05,
-      "loss": 0.0022,
+      "loss": 0.0061,
       "step": 3392
     },
     {
       "epoch": 32.0,
-      "eval_LOCATION_f1": 0.861111111111111,
+      "eval_LOCATION_f1": 0.8344370860927152,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.8732394366197183,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7746478873239436,
+      "eval_LOCATION_precision": 0.8076923076923077,
+      "eval_LOCATION_recall": 0.863013698630137,
+      "eval_ORGANIZATION_f1": 0.738255033557047,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.7142857142857143,
+      "eval_ORGANIZATION_precision": 0.6547619047619048,
       "eval_ORGANIZATION_recall": 0.8461538461538461,
-      "eval_PERSON_f1": 0.9114754098360656,
+      "eval_PERSON_f1": 0.9120521172638437,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.896774193548387,
-      "eval_PERSON_recall": 0.9266666666666666,
-      "eval_QUANTITY_f1": 0.9333333333333333,
+      "eval_PERSON_precision": 0.89171974522293,
+      "eval_PERSON_recall": 0.9333333333333333,
+      "eval_QUANTITY_f1": 0.7936507936507937,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.9032258064516129,
-      "eval_QUANTITY_recall": 0.9655172413793104,
-      "eval_TIME_f1": 0.9014084507042254,
+      "eval_QUANTITY_precision": 0.7352941176470589,
+      "eval_QUANTITY_recall": 0.8620689655172413,
+      "eval_TIME_f1": 0.8857142857142858,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8648648648648649,
-      "eval_TIME_recall": 0.9411764705882353,
-      "eval_loss": 0.24495986104011536,
-      "eval_overall_accuracy": 0.9684961850849126,
-      "eval_overall_f1": 0.8753462603878116,
-      "eval_overall_precision": 0.8517520215633423,
-      "eval_overall_recall": 0.9002849002849003,
-      "eval_runtime": 0.9344,
-      "eval_samples_per_second": 200.127,
-      "eval_steps_per_second": 3.211,
+      "eval_TIME_precision": 0.8611111111111112,
+      "eval_TIME_recall": 0.9117647058823529,
+      "eval_loss": 0.2889486253261566,
+      "eval_overall_accuracy": 0.9638198375584544,
+      "eval_overall_f1": 0.8486486486486486,
+      "eval_overall_precision": 0.8071979434447301,
+      "eval_overall_recall": 0.8945868945868946,
+      "eval_runtime": 0.3766,
+      "eval_samples_per_second": 496.531,
+      "eval_steps_per_second": 7.966,
       "step": 3392
     },
     {
       "epoch": 33.0,
-      "grad_norm": 0.007274314295500517,
+      "grad_norm": 0.027849093079566956,
       "learning_rate": 3.35e-05,
-      "loss": 0.0026,
+      "loss": 0.0028,
       "step": 3498
     },
     {
       "epoch": 33.0,
-      "eval_LOCATION_f1": 0.861111111111111,
+      "eval_LOCATION_f1": 0.8571428571428572,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.8732394366197183,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7808219178082192,
+      "eval_LOCATION_precision": 0.8513513513513513,
+      "eval_LOCATION_recall": 0.863013698630137,
+      "eval_ORGANIZATION_f1": 0.7310344827586207,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.7037037037037037,
-      "eval_ORGANIZATION_recall": 0.8769230769230769,
-      "eval_PERSON_f1": 0.9306930693069307,
+      "eval_ORGANIZATION_precision": 0.6625,
+      "eval_ORGANIZATION_recall": 0.8153846153846154,
+      "eval_PERSON_f1": 0.9084967320261438,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.9215686274509803,
-      "eval_PERSON_recall": 0.94,
-      "eval_QUANTITY_f1": 0.9333333333333333,
+      "eval_PERSON_precision": 0.8910256410256411,
+      "eval_PERSON_recall": 0.9266666666666666,
+      "eval_QUANTITY_f1": 0.75,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.9032258064516129,
-      "eval_QUANTITY_recall": 0.9655172413793104,
-      "eval_TIME_f1": 0.8985507246376812,
+      "eval_QUANTITY_precision": 0.6857142857142857,
+      "eval_QUANTITY_recall": 0.8275862068965517,
+      "eval_TIME_f1": 0.9014084507042254,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8857142857142857,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.25383174419403076,
-      "eval_overall_accuracy": 0.9687423086389367,
-      "eval_overall_f1": 0.8836565096952907,
-      "eval_overall_precision": 0.8598382749326146,
-      "eval_overall_recall": 0.9088319088319088,
-      "eval_runtime": 0.9557,
-      "eval_samples_per_second": 195.672,
-      "eval_steps_per_second": 3.139,
+      "eval_TIME_precision": 0.8648648648648649,
+      "eval_TIME_recall": 0.9411764705882353,
+      "eval_loss": 0.26156508922576904,
+      "eval_overall_accuracy": 0.9660349495446714,
+      "eval_overall_f1": 0.8485675306957708,
+      "eval_overall_precision": 0.8141361256544503,
+      "eval_overall_recall": 0.886039886039886,
+      "eval_runtime": 0.3654,
+      "eval_samples_per_second": 511.807,
+      "eval_steps_per_second": 8.211,
       "step": 3498
     },
     {
       "epoch": 34.0,
-      "grad_norm": 0.0037260684184730053,
+      "grad_norm": 0.12765878438949585,
       "learning_rate": 3.3e-05,
-      "loss": 0.0019,
+      "loss": 0.0024,
       "step": 3604
     },
     {
       "epoch": 34.0,
-      "eval_LOCATION_f1": 0.8571428571428572,
+      "eval_LOCATION_f1": 0.8079470198675496,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.8513513513513513,
-      "eval_LOCATION_recall": 0.863013698630137,
-      "eval_ORGANIZATION_f1": 0.7769784172661871,
+      "eval_LOCATION_precision": 0.782051282051282,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.7417218543046358,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.7297297297297297,
-      "eval_ORGANIZATION_recall": 0.8307692307692308,
-      "eval_PERSON_f1": 0.9190938511326862,
+      "eval_ORGANIZATION_precision": 0.6511627906976745,
+      "eval_ORGANIZATION_recall": 0.8615384615384616,
+      "eval_PERSON_f1": 0.9245901639344262,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8930817610062893,
-      "eval_PERSON_recall": 0.9466666666666667,
-      "eval_QUANTITY_f1": 0.9180327868852458,
+      "eval_PERSON_precision": 0.9096774193548387,
+      "eval_PERSON_recall": 0.94,
+      "eval_QUANTITY_f1": 0.8253968253968255,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.875,
-      "eval_QUANTITY_recall": 0.9655172413793104,
-      "eval_TIME_f1": 0.8857142857142858,
+      "eval_QUANTITY_precision": 0.7647058823529411,
+      "eval_QUANTITY_recall": 0.896551724137931,
+      "eval_TIME_f1": 0.8985507246376812,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8611111111111112,
+      "eval_TIME_precision": 0.8857142857142857,
       "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.2427574247121811,
-      "eval_overall_accuracy": 0.9709574206251538,
-      "eval_overall_f1": 0.8760330578512396,
-      "eval_overall_precision": 0.848,
-      "eval_overall_recall": 0.905982905982906,
-      "eval_runtime": 0.9547,
-      "eval_samples_per_second": 195.877,
-      "eval_steps_per_second": 3.142,
+      "eval_loss": 0.2858164310455322,
+      "eval_overall_accuracy": 0.9657888259906473,
+      "eval_overall_f1": 0.8525033829499323,
+      "eval_overall_precision": 0.8118556701030928,
+      "eval_overall_recall": 0.8974358974358975,
+      "eval_runtime": 0.3765,
+      "eval_samples_per_second": 496.701,
+      "eval_steps_per_second": 7.968,
       "step": 3604
     },
     {
       "epoch": 35.0,
-      "grad_norm": 0.0030928810592740774,
+      "grad_norm": 0.005790454335510731,
       "learning_rate": 3.2500000000000004e-05,
-      "loss": 0.003,
+      "loss": 0.0016,
       "step": 3710
     },
     {
       "epoch": 35.0,
-      "eval_LOCATION_f1": 0.8243243243243243,
+      "eval_LOCATION_f1": 0.8400000000000001,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.8133333333333334,
-      "eval_LOCATION_recall": 0.8356164383561644,
-      "eval_ORGANIZATION_f1": 0.7972027972027972,
+      "eval_LOCATION_precision": 0.8181818181818182,
+      "eval_LOCATION_recall": 0.863013698630137,
+      "eval_ORGANIZATION_f1": 0.7755102040816327,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.7307692307692307,
+      "eval_ORGANIZATION_precision": 0.6951219512195121,
       "eval_ORGANIZATION_recall": 0.8769230769230769,
-      "eval_PERSON_f1": 0.9315960912052117,
+      "eval_PERSON_f1": 0.9306930693069307,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.910828025477707,
-      "eval_PERSON_recall": 0.9533333333333334,
-      "eval_QUANTITY_f1": 0.9,
+      "eval_PERSON_precision": 0.9215686274509803,
+      "eval_PERSON_recall": 0.94,
+      "eval_QUANTITY_f1": 0.7936507936507937,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.8709677419354839,
-      "eval_QUANTITY_recall": 0.9310344827586207,
-      "eval_TIME_f1": 0.9117647058823528,
+      "eval_QUANTITY_precision": 0.7352941176470589,
+      "eval_QUANTITY_recall": 0.8620689655172413,
+      "eval_TIME_f1": 0.8695652173913043,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.9117647058823529,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.25404903292655945,
-      "eval_overall_accuracy": 0.9699729264090573,
-      "eval_overall_f1": 0.8787878787878788,
-      "eval_overall_precision": 0.8506666666666667,
-      "eval_overall_recall": 0.9088319088319088,
-      "eval_runtime": 0.9451,
-      "eval_samples_per_second": 197.855,
-      "eval_steps_per_second": 3.174,
+      "eval_TIME_precision": 0.8571428571428571,
+      "eval_TIME_recall": 0.8823529411764706,
+      "eval_loss": 0.30194392800331116,
+      "eval_overall_accuracy": 0.9660349495446714,
+      "eval_overall_f1": 0.8633879781420766,
+      "eval_overall_precision": 0.8293963254593176,
+      "eval_overall_recall": 0.9002849002849003,
+      "eval_runtime": 0.3661,
+      "eval_samples_per_second": 510.813,
+      "eval_steps_per_second": 8.195,
       "step": 3710
     },
     {
       "epoch": 36.0,
-      "grad_norm": 0.0035446523688733578,
+      "grad_norm": 0.024567781016230583,
       "learning_rate": 3.2000000000000005e-05,
-      "loss": 0.0008,
+      "loss": 0.0017,
       "step": 3816
     },
     {
       "epoch": 36.0,
-      "eval_LOCATION_f1": 0.847682119205298,
+      "eval_LOCATION_f1": 0.8235294117647058,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.8205128205128205,
-      "eval_LOCATION_recall": 0.8767123287671232,
-      "eval_ORGANIZATION_f1": 0.7943262411347517,
+      "eval_LOCATION_precision": 0.7875,
+      "eval_LOCATION_recall": 0.863013698630137,
+      "eval_ORGANIZATION_f1": 0.7391304347826088,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.7368421052631579,
-      "eval_ORGANIZATION_recall": 0.8615384615384616,
-      "eval_PERSON_f1": 0.9185667752442996,
+      "eval_ORGANIZATION_precision": 0.6986301369863014,
+      "eval_ORGANIZATION_recall": 0.7846153846153846,
+      "eval_PERSON_f1": 0.934640522875817,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8980891719745223,
-      "eval_PERSON_recall": 0.94,
-      "eval_QUANTITY_f1": 0.9,
+      "eval_PERSON_precision": 0.9166666666666666,
+      "eval_PERSON_recall": 0.9533333333333334,
+      "eval_QUANTITY_f1": 0.7540983606557378,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.8709677419354839,
-      "eval_QUANTITY_recall": 0.9310344827586207,
+      "eval_QUANTITY_precision": 0.71875,
+      "eval_QUANTITY_recall": 0.7931034482758621,
       "eval_TIME_f1": 0.8985507246376812,
       "eval_TIME_number": 34,
       "eval_TIME_precision": 0.8857142857142857,
       "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.24802231788635254,
-      "eval_overall_accuracy": 0.9707112970711297,
-      "eval_overall_f1": 0.8763736263736264,
-      "eval_overall_precision": 0.8461538461538461,
-      "eval_overall_recall": 0.9088319088319088,
-      "eval_runtime": 0.9562,
-      "eval_samples_per_second": 195.566,
-      "eval_steps_per_second": 3.137,
+      "eval_loss": 0.2798221707344055,
+      "eval_overall_accuracy": 0.9665271966527197,
+      "eval_overall_f1": 0.8555708390646494,
+      "eval_overall_precision": 0.8271276595744681,
+      "eval_overall_recall": 0.886039886039886,
+      "eval_runtime": 0.3617,
+      "eval_samples_per_second": 517.047,
+      "eval_steps_per_second": 8.295,
       "step": 3816
     },
     {
       "epoch": 37.0,
-      "grad_norm": 0.0007106426637619734,
+      "grad_norm": 0.005459336563944817,
       "learning_rate": 3.15e-05,
-      "loss": 0.0016,
+      "loss": 0.0012,
       "step": 3922
     },
     {
       "epoch": 37.0,
-      "eval_LOCATION_f1": 0.8400000000000001,
+      "eval_LOCATION_f1": 0.802547770700637,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.8181818181818182,
+      "eval_LOCATION_precision": 0.75,
       "eval_LOCATION_recall": 0.863013698630137,
-      "eval_ORGANIZATION_f1": 0.8,
+      "eval_ORGANIZATION_f1": 0.7596899224806202,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.7466666666666667,
-      "eval_ORGANIZATION_recall": 0.8615384615384616,
-      "eval_PERSON_f1": 0.9090909090909091,
+      "eval_ORGANIZATION_precision": 0.765625,
+      "eval_ORGANIZATION_recall": 0.7538461538461538,
+      "eval_PERSON_f1": 0.9281045751633987,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8860759493670886,
-      "eval_PERSON_recall": 0.9333333333333333,
-      "eval_QUANTITY_f1": 0.8852459016393444,
+      "eval_PERSON_precision": 0.9102564102564102,
+      "eval_PERSON_recall": 0.9466666666666667,
+      "eval_QUANTITY_f1": 0.7741935483870968,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.84375,
-      "eval_QUANTITY_recall": 0.9310344827586207,
-      "eval_TIME_f1": 0.9117647058823528,
+      "eval_QUANTITY_precision": 0.7272727272727273,
+      "eval_QUANTITY_recall": 0.8275862068965517,
+      "eval_TIME_f1": 0.8493150684931507,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.9117647058823529,
+      "eval_TIME_precision": 0.7948717948717948,
       "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.2422735095024109,
-      "eval_overall_accuracy": 0.9714496677332021,
-      "eval_overall_f1": 0.8720770288858323,
-      "eval_overall_precision": 0.8430851063829787,
-      "eval_overall_recall": 0.9031339031339032,
-      "eval_runtime": 0.9429,
-      "eval_samples_per_second": 198.315,
-      "eval_steps_per_second": 3.182,
+      "eval_loss": 0.3006635010242462,
+      "eval_overall_accuracy": 0.9655427024366232,
+      "eval_overall_f1": 0.8500687757909217,
+      "eval_overall_precision": 0.8218085106382979,
+      "eval_overall_recall": 0.8803418803418803,
+      "eval_runtime": 0.3665,
+      "eval_samples_per_second": 510.29,
+      "eval_steps_per_second": 8.186,
       "step": 3922
     },
     {
       "epoch": 38.0,
-      "grad_norm": 0.004738088697195053,
+      "grad_norm": 0.08694141358137131,
       "learning_rate": 3.1e-05,
-      "loss": 0.0031,
+      "loss": 0.002,
       "step": 4028
     },
     {
       "epoch": 38.0,
-      "eval_LOCATION_f1": 0.8,
+      "eval_LOCATION_f1": 0.8266666666666668,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.8055555555555556,
-      "eval_LOCATION_recall": 0.7945205479452054,
-      "eval_ORGANIZATION_f1": 0.75,
+      "eval_LOCATION_precision": 0.8051948051948052,
+      "eval_LOCATION_recall": 0.8493150684931506,
+      "eval_ORGANIZATION_f1": 0.7482993197278912,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6551724137931034,
-      "eval_ORGANIZATION_recall": 0.8769230769230769,
-      "eval_PERSON_f1": 0.9114754098360656,
+      "eval_ORGANIZATION_precision": 0.6707317073170732,
+      "eval_ORGANIZATION_recall": 0.8461538461538461,
+      "eval_PERSON_f1": 0.9372937293729374,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.896774193548387,
-      "eval_PERSON_recall": 0.9266666666666666,
-      "eval_QUANTITY_f1": 0.8666666666666666,
+      "eval_PERSON_precision": 0.9281045751633987,
+      "eval_PERSON_recall": 0.9466666666666667,
+      "eval_QUANTITY_f1": 0.7936507936507937,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.8387096774193549,
-      "eval_QUANTITY_recall": 0.896551724137931,
-      "eval_TIME_f1": 0.8823529411764706,
+      "eval_QUANTITY_precision": 0.7352941176470589,
+      "eval_QUANTITY_recall": 0.8620689655172413,
+      "eval_TIME_f1": 0.8051948051948051,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8823529411764706,
-      "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.27414992451667786,
-      "eval_overall_accuracy": 0.9633275904504061,
-      "eval_overall_f1": 0.8493150684931507,
-      "eval_overall_precision": 0.8179419525065963,
-      "eval_overall_recall": 0.8831908831908832,
-      "eval_runtime": 0.9487,
-      "eval_samples_per_second": 197.117,
-      "eval_steps_per_second": 3.162,
+      "eval_TIME_precision": 0.7209302325581395,
+      "eval_TIME_recall": 0.9117647058823529,
+      "eval_loss": 0.32041510939598083,
+      "eval_overall_accuracy": 0.9625892197883338,
+      "eval_overall_f1": 0.8513513513513514,
+      "eval_overall_precision": 0.8097686375321337,
+      "eval_overall_recall": 0.8974358974358975,
+      "eval_runtime": 0.3702,
+      "eval_samples_per_second": 505.188,
+      "eval_steps_per_second": 8.105,
       "step": 4028
     },
     {
       "epoch": 39.0,
-      "grad_norm": 0.010891593992710114,
+      "grad_norm": 0.016844790428876877,
       "learning_rate": 3.05e-05,
-      "loss": 0.002,
+      "loss": 0.0017,
       "step": 4134
     },
     {
       "epoch": 39.0,
-      "eval_LOCATION_f1": 0.8104575163398693,
+      "eval_LOCATION_f1": 0.8378378378378378,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.775,
+      "eval_LOCATION_precision": 0.8266666666666667,
       "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7552447552447553,
+      "eval_ORGANIZATION_f1": 0.7152317880794703,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6923076923076923,
+      "eval_ORGANIZATION_precision": 0.627906976744186,
       "eval_ORGANIZATION_recall": 0.8307692307692308,
-      "eval_PERSON_f1": 0.9250814332247558,
+      "eval_PERSON_f1": 0.9220779220779222,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.9044585987261147,
+      "eval_PERSON_precision": 0.8987341772151899,
       "eval_PERSON_recall": 0.9466666666666667,
-      "eval_QUANTITY_f1": 0.7812500000000001,
+      "eval_QUANTITY_f1": 0.6885245901639345,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.7142857142857143,
-      "eval_QUANTITY_recall": 0.8620689655172413,
-      "eval_TIME_f1": 0.8108108108108107,
+      "eval_QUANTITY_precision": 0.65625,
+      "eval_QUANTITY_recall": 0.7241379310344828,
+      "eval_TIME_f1": 0.8985507246376812,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.75,
-      "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.27585744857788086,
-      "eval_overall_accuracy": 0.9633275904504061,
-      "eval_overall_f1": 0.844804318488529,
-      "eval_overall_precision": 0.8025641025641026,
-      "eval_overall_recall": 0.8917378917378918,
-      "eval_runtime": 0.9389,
-      "eval_samples_per_second": 199.164,
-      "eval_steps_per_second": 3.195,
+      "eval_TIME_precision": 0.8857142857142857,
+      "eval_TIME_recall": 0.9117647058823529,
+      "eval_loss": 0.2831703722476959,
+      "eval_overall_accuracy": 0.9638198375584544,
+      "eval_overall_f1": 0.8412483039348712,
+      "eval_overall_precision": 0.8031088082901554,
+      "eval_overall_recall": 0.8831908831908832,
+      "eval_runtime": 0.3629,
+      "eval_samples_per_second": 515.277,
+      "eval_steps_per_second": 8.266,
       "step": 4134
     },
     {
       "epoch": 40.0,
-      "grad_norm": 0.02131796069443226,
+      "grad_norm": 0.0038298426661640406,
       "learning_rate": 3e-05,
-      "loss": 0.0012,
+      "loss": 0.0019,
       "step": 4240
     },
     {
       "epoch": 40.0,
-      "eval_LOCATION_f1": 0.7999999999999999,
+      "eval_LOCATION_f1": 0.8152866242038217,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7560975609756098,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7552447552447553,
+      "eval_LOCATION_precision": 0.7619047619047619,
+      "eval_LOCATION_recall": 0.8767123287671232,
+      "eval_ORGANIZATION_f1": 0.7172413793103448,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6923076923076923,
-      "eval_ORGANIZATION_recall": 0.8307692307692308,
-      "eval_PERSON_f1": 0.9133333333333333,
+      "eval_ORGANIZATION_precision": 0.65,
+      "eval_ORGANIZATION_recall": 0.8,
+      "eval_PERSON_f1": 0.934640522875817,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.9133333333333333,
-      "eval_PERSON_recall": 0.9133333333333333,
-      "eval_QUANTITY_f1": 0.7936507936507937,
+      "eval_PERSON_precision": 0.9166666666666666,
+      "eval_PERSON_recall": 0.9533333333333334,
+      "eval_QUANTITY_f1": 0.75,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.7352941176470589,
-      "eval_QUANTITY_recall": 0.8620689655172413,
-      "eval_TIME_f1": 0.9117647058823528,
+      "eval_QUANTITY_precision": 0.6857142857142857,
+      "eval_QUANTITY_recall": 0.8275862068965517,
+      "eval_TIME_f1": 0.8055555555555555,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.9117647058823529,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.30947157740592957,
-      "eval_overall_accuracy": 0.963081466896382,
-      "eval_overall_f1": 0.8477366255144033,
-      "eval_overall_precision": 0.8174603174603174,
-      "eval_overall_recall": 0.8803418803418803,
-      "eval_runtime": 0.9475,
-      "eval_samples_per_second": 197.364,
-      "eval_steps_per_second": 3.166,
+      "eval_TIME_precision": 0.7631578947368421,
+      "eval_TIME_recall": 0.8529411764705882,
+      "eval_loss": 0.3074106276035309,
+      "eval_overall_accuracy": 0.9645582082205267,
+      "eval_overall_f1": 0.8387096774193549,
+      "eval_overall_precision": 0.7938931297709924,
+      "eval_overall_recall": 0.8888888888888888,
+      "eval_runtime": 0.3646,
+      "eval_samples_per_second": 512.836,
+      "eval_steps_per_second": 8.227,
       "step": 4240
     },
     {
       "epoch": 41.0,
-      "grad_norm": 2.5597097873687744,
+      "grad_norm": 0.004407234024256468,
       "learning_rate": 2.95e-05,
-      "loss": 0.0017,
+      "loss": 0.0007,
       "step": 4346
     },
     {
       "epoch": 41.0,
-      "eval_LOCATION_f1": 0.7682119205298013,
+      "eval_LOCATION_f1": 0.832214765100671,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7435897435897436,
-      "eval_LOCATION_recall": 0.7945205479452054,
-      "eval_ORGANIZATION_f1": 0.7199999999999999,
+      "eval_LOCATION_precision": 0.8157894736842105,
+      "eval_LOCATION_recall": 0.8493150684931506,
+      "eval_ORGANIZATION_f1": 0.7297297297297298,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6352941176470588,
+      "eval_ORGANIZATION_precision": 0.6506024096385542,
       "eval_ORGANIZATION_recall": 0.8307692307692308,
-      "eval_PERSON_f1": 0.9013157894736842,
+      "eval_PERSON_f1": 0.9315960912052117,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8896103896103896,
-      "eval_PERSON_recall": 0.9133333333333333,
-      "eval_QUANTITY_f1": 0.8253968253968255,
+      "eval_PERSON_precision": 0.910828025477707,
+      "eval_PERSON_recall": 0.9533333333333334,
+      "eval_QUANTITY_f1": 0.7384615384615385,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.7647058823529411,
-      "eval_QUANTITY_recall": 0.896551724137931,
-      "eval_TIME_f1": 0.9117647058823528,
+      "eval_QUANTITY_precision": 0.6666666666666666,
+      "eval_QUANTITY_recall": 0.8275862068965517,
+      "eval_TIME_f1": 0.9014084507042254,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.9117647058823529,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.2806366980075836,
+      "eval_TIME_precision": 0.8648648648648649,
+      "eval_TIME_recall": 0.9411764705882353,
+      "eval_loss": 0.3130010962486267,
       "eval_overall_accuracy": 0.9643120846665025,
-      "eval_overall_f1": 0.8315217391304348,
-      "eval_overall_precision": 0.7948051948051948,
-      "eval_overall_recall": 0.8717948717948718,
-      "eval_runtime": 0.9544,
-      "eval_samples_per_second": 195.941,
-      "eval_steps_per_second": 3.143,
+      "eval_overall_f1": 0.8513513513513514,
+      "eval_overall_precision": 0.8097686375321337,
+      "eval_overall_recall": 0.8974358974358975,
+      "eval_runtime": 0.3621,
+      "eval_samples_per_second": 516.375,
+      "eval_steps_per_second": 8.284,
       "step": 4346
     },
     {
       "epoch": 42.0,
-      "grad_norm": 0.008968453854322433,
+      "grad_norm": 0.0027904659509658813,
       "learning_rate": 2.9e-05,
-      "loss": 0.0022,
+      "loss": 0.0013,
       "step": 4452
     },
     {
       "epoch": 42.0,
-      "eval_LOCATION_f1": 0.7948717948717949,
+      "eval_LOCATION_f1": 0.8344370860927152,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7469879518072289,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.6896551724137931,
+      "eval_LOCATION_precision": 0.8076923076923077,
+      "eval_LOCATION_recall": 0.863013698630137,
+      "eval_ORGANIZATION_f1": 0.712121212121212,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.625,
-      "eval_ORGANIZATION_recall": 0.7692307692307693,
-      "eval_PERSON_f1": 0.9250814332247558,
+      "eval_ORGANIZATION_precision": 0.7014925373134329,
+      "eval_ORGANIZATION_recall": 0.7230769230769231,
+      "eval_PERSON_f1": 0.9377049180327869,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.9044585987261147,
-      "eval_PERSON_recall": 0.9466666666666667,
-      "eval_QUANTITY_f1": 0.8387096774193549,
+      "eval_PERSON_precision": 0.9225806451612903,
+      "eval_PERSON_recall": 0.9533333333333334,
+      "eval_QUANTITY_f1": 0.7096774193548386,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.7878787878787878,
-      "eval_QUANTITY_recall": 0.896551724137931,
-      "eval_TIME_f1": 0.9411764705882353,
+      "eval_QUANTITY_precision": 0.6666666666666666,
+      "eval_QUANTITY_recall": 0.7586206896551724,
+      "eval_TIME_f1": 0.8985507246376812,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.9411764705882353,
-      "eval_TIME_recall": 0.9411764705882353,
-      "eval_loss": 0.29396188259124756,
-      "eval_overall_accuracy": 0.9638198375584544,
-      "eval_overall_f1": 0.8455284552845528,
-      "eval_overall_precision": 0.8062015503875969,
-      "eval_overall_recall": 0.8888888888888888,
-      "eval_runtime": 0.9541,
-      "eval_samples_per_second": 195.989,
-      "eval_steps_per_second": 3.144,
+      "eval_TIME_precision": 0.8857142857142857,
+      "eval_TIME_recall": 0.9117647058823529,
+      "eval_loss": 0.28249990940093994,
+      "eval_overall_accuracy": 0.9682500615308886,
+      "eval_overall_f1": 0.8511821974965229,
+      "eval_overall_precision": 0.8315217391304348,
+      "eval_overall_recall": 0.8717948717948718,
+      "eval_runtime": 0.3687,
+      "eval_samples_per_second": 507.242,
+      "eval_steps_per_second": 8.138,
       "step": 4452
     },
     {
       "epoch": 43.0,
-      "grad_norm": 0.10193750262260437,
+      "grad_norm": 3.8555519580841064,
       "learning_rate": 2.8499999999999998e-05,
-      "loss": 0.002,
+      "loss": 0.0008,
       "step": 4558
     },
     {
       "epoch": 43.0,
-      "eval_LOCATION_f1": 0.8076923076923077,
+      "eval_LOCATION_f1": 0.7973856209150326,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7590361445783133,
-      "eval_LOCATION_recall": 0.863013698630137,
-      "eval_ORGANIZATION_f1": 0.7391304347826088,
+      "eval_LOCATION_precision": 0.7625,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.7857142857142856,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6986301369863014,
-      "eval_ORGANIZATION_recall": 0.7846153846153846,
-      "eval_PERSON_f1": 0.9144736842105262,
+      "eval_ORGANIZATION_precision": 0.7333333333333333,
+      "eval_ORGANIZATION_recall": 0.8461538461538461,
+      "eval_PERSON_f1": 0.934640522875817,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.9025974025974026,
-      "eval_PERSON_recall": 0.9266666666666666,
-      "eval_QUANTITY_f1": 0.8709677419354839,
+      "eval_PERSON_precision": 0.9166666666666666,
+      "eval_PERSON_recall": 0.9533333333333334,
+      "eval_QUANTITY_f1": 0.8125,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.8181818181818182,
-      "eval_QUANTITY_recall": 0.9310344827586207,
-      "eval_TIME_f1": 0.9411764705882353,
+      "eval_QUANTITY_precision": 0.7428571428571429,
+      "eval_QUANTITY_recall": 0.896551724137931,
+      "eval_TIME_f1": 0.8857142857142858,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.9411764705882353,
-      "eval_TIME_recall": 0.9411764705882353,
-      "eval_loss": 0.2899914085865021,
-      "eval_overall_accuracy": 0.9645582082205267,
-      "eval_overall_f1": 0.8571428571428572,
-      "eval_overall_precision": 0.8275862068965517,
-      "eval_overall_recall": 0.8888888888888888,
-      "eval_runtime": 0.9675,
-      "eval_samples_per_second": 193.282,
-      "eval_steps_per_second": 3.101,
+      "eval_TIME_precision": 0.8611111111111112,
+      "eval_TIME_recall": 0.9117647058823529,
+      "eval_loss": 0.31627556681632996,
+      "eval_overall_accuracy": 0.9672655673147921,
+      "eval_overall_f1": 0.8622100954979536,
+      "eval_overall_precision": 0.8272251308900523,
+      "eval_overall_recall": 0.9002849002849003,
+      "eval_runtime": 0.3621,
+      "eval_samples_per_second": 516.427,
+      "eval_steps_per_second": 8.285,
       "step": 4558
     },
     {
       "epoch": 44.0,
-      "grad_norm": 2.1496660709381104,
+      "grad_norm": 0.0015216912142932415,
       "learning_rate": 2.8000000000000003e-05,
-      "loss": 0.002,
+      "loss": 0.0035,
       "step": 4664
     },
     {
       "epoch": 44.0,
-      "eval_LOCATION_f1": 0.8513513513513513,
+      "eval_LOCATION_f1": 0.7948717948717949,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.84,
-      "eval_LOCATION_recall": 0.863013698630137,
-      "eval_ORGANIZATION_f1": 0.7633587786259541,
+      "eval_LOCATION_precision": 0.7469879518072289,
+      "eval_LOCATION_recall": 0.8493150684931506,
+      "eval_ORGANIZATION_f1": 0.7692307692307693,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.7575757575757576,
-      "eval_ORGANIZATION_recall": 0.7692307692307693,
-      "eval_PERSON_f1": 0.9342105263157895,
+      "eval_ORGANIZATION_precision": 0.7051282051282052,
+      "eval_ORGANIZATION_recall": 0.8461538461538461,
+      "eval_PERSON_f1": 0.9210526315789475,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.922077922077922,
-      "eval_PERSON_recall": 0.9466666666666667,
-      "eval_QUANTITY_f1": 0.8852459016393444,
+      "eval_PERSON_precision": 0.9090909090909091,
+      "eval_PERSON_recall": 0.9333333333333333,
+      "eval_QUANTITY_f1": 0.7096774193548386,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.84375,
-      "eval_QUANTITY_recall": 0.9310344827586207,
-      "eval_TIME_f1": 0.9275362318840579,
+      "eval_QUANTITY_precision": 0.6666666666666666,
+      "eval_QUANTITY_recall": 0.7586206896551724,
+      "eval_TIME_f1": 0.8219178082191781,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.9142857142857143,
-      "eval_TIME_recall": 0.9411764705882353,
-      "eval_loss": 0.2349400669336319,
-      "eval_overall_accuracy": 0.9707112970711297,
-      "eval_overall_f1": 0.8807854137447406,
-      "eval_overall_precision": 0.8674033149171271,
-      "eval_overall_recall": 0.8945868945868946,
-      "eval_runtime": 0.9468,
-      "eval_samples_per_second": 197.508,
-      "eval_steps_per_second": 3.169,
+      "eval_TIME_precision": 0.7692307692307693,
+      "eval_TIME_recall": 0.8823529411764706,
+      "eval_loss": 0.3363170623779297,
+      "eval_overall_accuracy": 0.9623430962343096,
+      "eval_overall_f1": 0.8373983739837397,
+      "eval_overall_precision": 0.7984496124031008,
+      "eval_overall_recall": 0.8803418803418803,
+      "eval_runtime": 0.3643,
+      "eval_samples_per_second": 513.278,
+      "eval_steps_per_second": 8.234,
       "step": 4664
     },
     {
       "epoch": 45.0,
-      "grad_norm": 0.005279150325804949,
+      "grad_norm": 0.1600458323955536,
       "learning_rate": 2.7500000000000004e-05,
-      "loss": 0.0025,
+      "loss": 0.0032,
       "step": 4770
     },
     {
       "epoch": 45.0,
-      "eval_LOCATION_f1": 0.7972972972972971,
+      "eval_LOCATION_f1": 0.8235294117647058,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7866666666666666,
-      "eval_LOCATION_recall": 0.8082191780821918,
-      "eval_ORGANIZATION_f1": 0.7131782945736435,
+      "eval_LOCATION_precision": 0.7875,
+      "eval_LOCATION_recall": 0.863013698630137,
+      "eval_ORGANIZATION_f1": 0.757142857142857,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.71875,
-      "eval_ORGANIZATION_recall": 0.7076923076923077,
-      "eval_PERSON_f1": 0.9144736842105262,
+      "eval_ORGANIZATION_precision": 0.7066666666666667,
+      "eval_ORGANIZATION_recall": 0.8153846153846154,
+      "eval_PERSON_f1": 0.9150326797385621,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.9025974025974026,
-      "eval_PERSON_recall": 0.9266666666666666,
-      "eval_QUANTITY_f1": 0.8852459016393444,
+      "eval_PERSON_precision": 0.8974358974358975,
+      "eval_PERSON_recall": 0.9333333333333333,
+      "eval_QUANTITY_f1": 0.7868852459016394,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.84375,
-      "eval_QUANTITY_recall": 0.9310344827586207,
-      "eval_TIME_f1": 0.9411764705882353,
+      "eval_QUANTITY_precision": 0.75,
+      "eval_QUANTITY_recall": 0.8275862068965517,
+      "eval_TIME_f1": 0.9253731343283583,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.9411764705882353,
-      "eval_TIME_recall": 0.9411764705882353,
-      "eval_loss": 0.26148721575737,
-      "eval_overall_accuracy": 0.9672655673147921,
-      "eval_overall_f1": 0.8535211267605634,
-      "eval_overall_precision": 0.8440111420612814,
-      "eval_overall_recall": 0.8632478632478633,
-      "eval_runtime": 0.946,
-      "eval_samples_per_second": 197.678,
-      "eval_steps_per_second": 3.171,
+      "eval_TIME_precision": 0.9393939393939394,
+      "eval_TIME_recall": 0.9117647058823529,
+      "eval_loss": 0.265531986951828,
+      "eval_overall_accuracy": 0.9660349495446714,
+      "eval_overall_f1": 0.8555708390646494,
+      "eval_overall_precision": 0.8271276595744681,
+      "eval_overall_recall": 0.886039886039886,
+      "eval_runtime": 0.3707,
+      "eval_samples_per_second": 504.483,
+      "eval_steps_per_second": 8.093,
       "step": 4770
     },
     {
       "epoch": 46.0,
-      "grad_norm": 29.016551971435547,
+      "grad_norm": 0.09002447128295898,
       "learning_rate": 2.7000000000000002e-05,
-      "loss": 0.0024,
+      "loss": 0.0029,
       "step": 4876
     },
     {
       "epoch": 46.0,
-      "eval_LOCATION_f1": 0.7777777777777778,
+      "eval_LOCATION_f1": 0.8187919463087249,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7078651685393258,
-      "eval_LOCATION_recall": 0.863013698630137,
-      "eval_ORGANIZATION_f1": 0.6956521739130435,
+      "eval_LOCATION_precision": 0.8026315789473685,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.7916666666666667,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6575342465753424,
-      "eval_ORGANIZATION_recall": 0.7384615384615385,
-      "eval_PERSON_f1": 0.9180327868852459,
+      "eval_ORGANIZATION_precision": 0.7215189873417721,
+      "eval_ORGANIZATION_recall": 0.8769230769230769,
+      "eval_PERSON_f1": 0.9220779220779222,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.9032258064516129,
-      "eval_PERSON_recall": 0.9333333333333333,
-      "eval_QUANTITY_f1": 0.8387096774193549,
+      "eval_PERSON_precision": 0.8987341772151899,
+      "eval_PERSON_recall": 0.9466666666666667,
+      "eval_QUANTITY_f1": 0.65625,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.7878787878787878,
-      "eval_QUANTITY_recall": 0.896551724137931,
-      "eval_TIME_f1": 0.8493150684931507,
+      "eval_QUANTITY_precision": 0.6,
+      "eval_QUANTITY_recall": 0.7241379310344828,
+      "eval_TIME_f1": 0.8450704225352113,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.7948717948717948,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.31196725368499756,
-      "eval_overall_accuracy": 0.961112478464189,
-      "eval_overall_f1": 0.8324324324324326,
-      "eval_overall_precision": 0.7917737789203085,
-      "eval_overall_recall": 0.8774928774928775,
-      "eval_runtime": 0.9587,
-      "eval_samples_per_second": 195.046,
-      "eval_steps_per_second": 3.129,
+      "eval_TIME_precision": 0.8108108108108109,
+      "eval_TIME_recall": 0.8823529411764706,
+      "eval_loss": 0.28975537419319153,
+      "eval_overall_accuracy": 0.963081466896382,
+      "eval_overall_f1": 0.8451086956521738,
+      "eval_overall_precision": 0.8077922077922078,
+      "eval_overall_recall": 0.886039886039886,
+      "eval_runtime": 0.3761,
+      "eval_samples_per_second": 497.155,
+      "eval_steps_per_second": 7.976,
       "step": 4876
     },
     {
       "epoch": 47.0,
-      "grad_norm": 0.0034174255561083555,
+      "grad_norm": 0.014626134186983109,
       "learning_rate": 2.6500000000000004e-05,
-      "loss": 0.0025,
+      "loss": 0.0011,
       "step": 4982
     },
     {
       "epoch": 47.0,
-      "eval_LOCATION_f1": 0.7999999999999999,
+      "eval_LOCATION_f1": 0.8243243243243243,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7560975609756098,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.6842105263157896,
+      "eval_LOCATION_precision": 0.8133333333333334,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.7857142857142856,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.5977011494252874,
-      "eval_ORGANIZATION_recall": 0.8,
-      "eval_PERSON_f1": 0.8983606557377048,
+      "eval_ORGANIZATION_precision": 0.7333333333333333,
+      "eval_ORGANIZATION_recall": 0.8461538461538461,
+      "eval_PERSON_f1": 0.9311475409836065,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8838709677419355,
-      "eval_PERSON_recall": 0.9133333333333333,
-      "eval_QUANTITY_f1": 0.8709677419354839,
+      "eval_PERSON_precision": 0.9161290322580645,
+      "eval_PERSON_recall": 0.9466666666666667,
+      "eval_QUANTITY_f1": 0.676923076923077,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.8181818181818182,
-      "eval_QUANTITY_recall": 0.9310344827586207,
-      "eval_TIME_f1": 0.9014084507042254,
+      "eval_QUANTITY_precision": 0.6111111111111112,
+      "eval_QUANTITY_recall": 0.7586206896551724,
+      "eval_TIME_f1": 0.8108108108108107,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8648648648648649,
-      "eval_TIME_recall": 0.9411764705882353,
-      "eval_loss": 0.3170968294143677,
-      "eval_overall_accuracy": 0.9618508491262614,
-      "eval_overall_f1": 0.8322147651006713,
-      "eval_overall_precision": 0.7868020304568528,
+      "eval_TIME_precision": 0.75,
+      "eval_TIME_recall": 0.8823529411764706,
+      "eval_loss": 0.29475802183151245,
+      "eval_overall_accuracy": 0.9648043317745508,
+      "eval_overall_f1": 0.8469945355191257,
+      "eval_overall_precision": 0.8136482939632546,
       "eval_overall_recall": 0.8831908831908832,
-      "eval_runtime": 0.9483,
-      "eval_samples_per_second": 197.192,
-      "eval_steps_per_second": 3.164,
+      "eval_runtime": 0.3643,
+      "eval_samples_per_second": 513.245,
+      "eval_steps_per_second": 8.234,
       "step": 4982
     },
     {
       "epoch": 48.0,
-      "grad_norm": 0.015307181514799595,
+      "grad_norm": 0.017472336068749428,
       "learning_rate": 2.6000000000000002e-05,
-      "loss": 0.0031,
+      "loss": 0.0019,
       "step": 5088
     },
     {
       "epoch": 48.0,
-      "eval_LOCATION_f1": 0.7898089171974523,
+      "eval_LOCATION_f1": 0.7948717948717949,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7380952380952381,
+      "eval_LOCATION_precision": 0.7469879518072289,
       "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7,
+      "eval_ORGANIZATION_f1": 0.7552447552447553,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6533333333333333,
-      "eval_ORGANIZATION_recall": 0.7538461538461538,
-      "eval_PERSON_f1": 0.9169435215946843,
+      "eval_ORGANIZATION_precision": 0.6923076923076923,
+      "eval_ORGANIZATION_recall": 0.8307692307692308,
+      "eval_PERSON_f1": 0.9285714285714286,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.9139072847682119,
-      "eval_PERSON_recall": 0.92,
-      "eval_QUANTITY_f1": 0.8437500000000001,
+      "eval_PERSON_precision": 0.9050632911392406,
+      "eval_PERSON_recall": 0.9533333333333334,
+      "eval_QUANTITY_f1": 0.7076923076923076,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.7714285714285715,
-      "eval_QUANTITY_recall": 0.9310344827586207,
+      "eval_QUANTITY_precision": 0.6388888888888888,
+      "eval_QUANTITY_recall": 0.7931034482758621,
       "eval_TIME_f1": 0.9014084507042254,
       "eval_TIME_number": 34,
       "eval_TIME_precision": 0.8648648648648649,
       "eval_TIME_recall": 0.9411764705882353,
-      "eval_loss": 0.3086757957935333,
-      "eval_overall_accuracy": 0.9635737140044303,
-      "eval_overall_f1": 0.840381991814461,
-      "eval_overall_precision": 0.806282722513089,
-      "eval_overall_recall": 0.8774928774928775,
-      "eval_runtime": 0.9663,
-      "eval_samples_per_second": 193.525,
-      "eval_steps_per_second": 3.105,
+      "eval_loss": 0.29780101776123047,
+      "eval_overall_accuracy": 0.9645582082205267,
+      "eval_overall_f1": 0.8452220726783309,
+      "eval_overall_precision": 0.8010204081632653,
+      "eval_overall_recall": 0.8945868945868946,
+      "eval_runtime": 0.3664,
+      "eval_samples_per_second": 510.398,
+      "eval_steps_per_second": 8.188,
       "step": 5088
     },
     {
       "epoch": 49.0,
-      "grad_norm": 0.0008989412453956902,
+      "grad_norm": 0.0013933341251686215,
       "learning_rate": 2.5500000000000003e-05,
-      "loss": 0.0016,
+      "loss": 0.0012,
       "step": 5194
     },
     {
       "epoch": 49.0,
-      "eval_LOCATION_f1": 0.8421052631578948,
+      "eval_LOCATION_f1": 0.832214765100671,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.810126582278481,
-      "eval_LOCATION_recall": 0.8767123287671232,
-      "eval_ORGANIZATION_f1": 0.7445255474452555,
+      "eval_LOCATION_precision": 0.8157894736842105,
+      "eval_LOCATION_recall": 0.8493150684931506,
+      "eval_ORGANIZATION_f1": 0.7448275862068966,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.7083333333333334,
-      "eval_ORGANIZATION_recall": 0.7846153846153846,
-      "eval_PERSON_f1": 0.9311475409836065,
+      "eval_ORGANIZATION_precision": 0.675,
+      "eval_ORGANIZATION_recall": 0.8307692307692308,
+      "eval_PERSON_f1": 0.9245901639344262,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.9161290322580645,
-      "eval_PERSON_recall": 0.9466666666666667,
-      "eval_QUANTITY_f1": 0.8571428571428571,
+      "eval_PERSON_precision": 0.9096774193548387,
+      "eval_PERSON_recall": 0.94,
+      "eval_QUANTITY_f1": 0.7692307692307692,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.7941176470588235,
-      "eval_QUANTITY_recall": 0.9310344827586207,
+      "eval_QUANTITY_precision": 0.6944444444444444,
+      "eval_QUANTITY_recall": 0.8620689655172413,
       "eval_TIME_f1": 0.8732394366197184,
       "eval_TIME_number": 34,
       "eval_TIME_precision": 0.8378378378378378,
       "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.27731239795684814,
-      "eval_overall_accuracy": 0.9665271966527197,
-      "eval_overall_f1": 0.8653846153846154,
-      "eval_overall_precision": 0.8355437665782494,
-      "eval_overall_recall": 0.8974358974358975,
-      "eval_runtime": 0.9482,
-      "eval_samples_per_second": 197.209,
-      "eval_steps_per_second": 3.164,
+      "eval_loss": 0.306385338306427,
+      "eval_overall_accuracy": 0.9628353433423579,
+      "eval_overall_f1": 0.8517006802721088,
+      "eval_overall_precision": 0.8151041666666666,
+      "eval_overall_recall": 0.8917378917378918,
+      "eval_runtime": 0.3627,
+      "eval_samples_per_second": 515.517,
+      "eval_steps_per_second": 8.27,
       "step": 5194
     },
     {
       "epoch": 50.0,
-      "grad_norm": 0.2832314670085907,
+      "grad_norm": 0.12874193489551544,
       "learning_rate": 2.5e-05,
-      "loss": 0.0014,
+      "loss": 0.0005,
       "step": 5300
     },
     {
       "epoch": 50.0,
-      "eval_LOCATION_f1": 0.8344370860927152,
+      "eval_LOCATION_f1": 0.8163265306122449,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.8076923076923077,
-      "eval_LOCATION_recall": 0.863013698630137,
-      "eval_ORGANIZATION_f1": 0.7361111111111112,
+      "eval_LOCATION_precision": 0.8108108108108109,
+      "eval_LOCATION_recall": 0.821917808219178,
+      "eval_ORGANIZATION_f1": 0.7432432432432433,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6708860759493671,
-      "eval_ORGANIZATION_recall": 0.8153846153846154,
-      "eval_PERSON_f1": 0.9306930693069307,
+      "eval_ORGANIZATION_precision": 0.6626506024096386,
+      "eval_ORGANIZATION_recall": 0.8461538461538461,
+      "eval_PERSON_f1": 0.9281045751633987,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.9215686274509803,
-      "eval_PERSON_recall": 0.94,
-      "eval_QUANTITY_f1": 0.8709677419354839,
+      "eval_PERSON_precision": 0.9102564102564102,
+      "eval_PERSON_recall": 0.9466666666666667,
+      "eval_QUANTITY_f1": 0.7692307692307692,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.8181818181818182,
-      "eval_QUANTITY_recall": 0.9310344827586207,
-      "eval_TIME_f1": 0.8857142857142858,
+      "eval_QUANTITY_precision": 0.6944444444444444,
+      "eval_QUANTITY_recall": 0.8620689655172413,
+      "eval_TIME_f1": 0.8219178082191781,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8611111111111112,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.277599960565567,
-      "eval_overall_accuracy": 0.9675116908688162,
-      "eval_overall_f1": 0.863013698630137,
-      "eval_overall_precision": 0.8311345646437994,
-      "eval_overall_recall": 0.8974358974358975,
-      "eval_runtime": 0.9467,
-      "eval_samples_per_second": 197.519,
-      "eval_steps_per_second": 3.169,
+      "eval_TIME_precision": 0.7692307692307693,
+      "eval_TIME_recall": 0.8823529411764706,
+      "eval_loss": 0.3278602063655853,
+      "eval_overall_accuracy": 0.9623430962343096,
+      "eval_overall_f1": 0.8443843031123139,
+      "eval_overall_precision": 0.8041237113402062,
+      "eval_overall_recall": 0.8888888888888888,
+      "eval_runtime": 0.3691,
+      "eval_samples_per_second": 506.611,
+      "eval_steps_per_second": 8.127,
       "step": 5300
     },
     {
       "epoch": 51.0,
-      "grad_norm": 0.0006746707949787378,
+      "grad_norm": 0.0021654218435287476,
       "learning_rate": 2.45e-05,
-      "loss": 0.0009,
+      "loss": 0.0022,
       "step": 5406
     },
     {
       "epoch": 51.0,
-      "eval_LOCATION_f1": 0.8344370860927152,
+      "eval_LOCATION_f1": 0.8493150684931505,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.8076923076923077,
-      "eval_LOCATION_recall": 0.863013698630137,
-      "eval_ORGANIZATION_f1": 0.7883211678832116,
+      "eval_LOCATION_precision": 0.8493150684931506,
+      "eval_LOCATION_recall": 0.8493150684931506,
+      "eval_ORGANIZATION_f1": 0.7943262411347517,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.75,
-      "eval_ORGANIZATION_recall": 0.8307692307692308,
+      "eval_ORGANIZATION_precision": 0.7368421052631579,
+      "eval_ORGANIZATION_recall": 0.8615384615384616,
       "eval_PERSON_f1": 0.9311475409836065,
       "eval_PERSON_number": 150,
       "eval_PERSON_precision": 0.9161290322580645,
       "eval_PERSON_recall": 0.9466666666666667,
-      "eval_QUANTITY_f1": 0.8852459016393444,
+      "eval_QUANTITY_f1": 0.7457627118644068,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.84375,
-      "eval_QUANTITY_recall": 0.9310344827586207,
-      "eval_TIME_f1": 0.8985507246376812,
+      "eval_QUANTITY_precision": 0.7333333333333333,
+      "eval_QUANTITY_recall": 0.7586206896551724,
+      "eval_TIME_f1": 0.8823529411764706,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8857142857142857,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.2767013907432556,
-      "eval_overall_accuracy": 0.969234555746985,
-      "eval_overall_f1": 0.8769017980636239,
-      "eval_overall_precision": 0.8521505376344086,
-      "eval_overall_recall": 0.9031339031339032,
-      "eval_runtime": 0.9471,
-      "eval_samples_per_second": 197.438,
-      "eval_steps_per_second": 3.167,
+      "eval_TIME_precision": 0.8823529411764706,
+      "eval_TIME_recall": 0.8823529411764706,
+      "eval_loss": 0.28876420855522156,
+      "eval_overall_accuracy": 0.9665271966527197,
+      "eval_overall_f1": 0.8678720445062587,
+      "eval_overall_precision": 0.8478260869565217,
+      "eval_overall_recall": 0.8888888888888888,
+      "eval_runtime": 0.3618,
+      "eval_samples_per_second": 516.847,
+      "eval_steps_per_second": 8.292,
       "step": 5406
     },
     {
       "epoch": 52.0,
-      "grad_norm": 0.011208178475499153,
+      "grad_norm": 6.50219202041626,
       "learning_rate": 2.4e-05,
-      "loss": 0.0016,
+      "loss": 0.0018,
       "step": 5512
     },
     {
       "epoch": 52.0,
-      "eval_LOCATION_f1": 0.8366013071895425,
+      "eval_LOCATION_f1": 0.8181818181818182,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.8,
-      "eval_LOCATION_recall": 0.8767123287671232,
-      "eval_ORGANIZATION_f1": 0.7445255474452555,
+      "eval_LOCATION_precision": 0.7777777777777778,
+      "eval_LOCATION_recall": 0.863013698630137,
+      "eval_ORGANIZATION_f1": 0.7346938775510204,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.7083333333333334,
-      "eval_ORGANIZATION_recall": 0.7846153846153846,
-      "eval_PERSON_f1": 0.9185667752442996,
+      "eval_ORGANIZATION_precision": 0.6585365853658537,
+      "eval_ORGANIZATION_recall": 0.8307692307692308,
+      "eval_PERSON_f1": 0.9311475409836065,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.8980891719745223,
-      "eval_PERSON_recall": 0.94,
-      "eval_QUANTITY_f1": 0.896551724137931,
+      "eval_PERSON_precision": 0.9161290322580645,
+      "eval_PERSON_recall": 0.9466666666666667,
+      "eval_QUANTITY_f1": 0.7619047619047619,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.896551724137931,
-      "eval_QUANTITY_recall": 0.896551724137931,
-      "eval_TIME_f1": 0.9411764705882353,
+      "eval_QUANTITY_precision": 0.7058823529411765,
+      "eval_QUANTITY_recall": 0.8275862068965517,
+      "eval_TIME_f1": 0.7142857142857143,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.9411764705882353,
-      "eval_TIME_recall": 0.9411764705882353,
-      "eval_loss": 0.2851575016975403,
-      "eval_overall_accuracy": 0.9672655673147921,
-      "eval_overall_f1": 0.8686030428769017,
-      "eval_overall_precision": 0.8440860215053764,
-      "eval_overall_recall": 0.8945868945868946,
-      "eval_runtime": 1.0171,
-      "eval_samples_per_second": 183.864,
-      "eval_steps_per_second": 2.95,
+      "eval_TIME_precision": 0.6,
+      "eval_TIME_recall": 0.8823529411764706,
+      "eval_loss": 0.34152865409851074,
+      "eval_overall_accuracy": 0.9579128722618755,
+      "eval_overall_f1": 0.8313413014608234,
+      "eval_overall_precision": 0.7786069651741293,
+      "eval_overall_recall": 0.8917378917378918,
+      "eval_runtime": 0.3789,
+      "eval_samples_per_second": 493.482,
+      "eval_steps_per_second": 7.917,
       "step": 5512
     },
     {
       "epoch": 53.0,
-      "grad_norm": 0.0011826782720163465,
+      "grad_norm": 0.015084599144756794,
       "learning_rate": 2.35e-05,
-      "loss": 0.0011,
+      "loss": 0.0024,
       "step": 5618
     },
     {
       "epoch": 53.0,
-      "eval_LOCATION_f1": 0.8051948051948052,
+      "eval_LOCATION_f1": 0.7820512820512822,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7654320987654321,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7737226277372263,
+      "eval_LOCATION_precision": 0.7349397590361446,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.7692307692307693,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.7361111111111112,
-      "eval_ORGANIZATION_recall": 0.8153846153846154,
-      "eval_PERSON_f1": 0.934640522875817,
+      "eval_ORGANIZATION_precision": 0.7051282051282052,
+      "eval_ORGANIZATION_recall": 0.8461538461538461,
+      "eval_PERSON_f1": 0.9276315789473684,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.9166666666666666,
-      "eval_PERSON_recall": 0.9533333333333334,
-      "eval_QUANTITY_f1": 0.8852459016393444,
+      "eval_PERSON_precision": 0.9155844155844156,
+      "eval_PERSON_recall": 0.94,
+      "eval_QUANTITY_f1": 0.7741935483870968,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.84375,
-      "eval_QUANTITY_recall": 0.9310344827586207,
-      "eval_TIME_f1": 0.8732394366197184,
+      "eval_QUANTITY_precision": 0.7272727272727273,
+      "eval_QUANTITY_recall": 0.8275862068965517,
+      "eval_TIME_f1": 0.8,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8378378378378378,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.2843783497810364,
-      "eval_overall_accuracy": 0.9672655673147921,
-      "eval_overall_f1": 0.8669410150891631,
-      "eval_overall_precision": 0.8359788359788359,
-      "eval_overall_recall": 0.9002849002849003,
-      "eval_runtime": 0.9535,
-      "eval_samples_per_second": 196.116,
-      "eval_steps_per_second": 3.146,
+      "eval_TIME_precision": 0.7317073170731707,
+      "eval_TIME_recall": 0.8823529411764706,
+      "eval_loss": 0.3337211012840271,
+      "eval_overall_accuracy": 0.9591434900319961,
+      "eval_overall_f1": 0.8405405405405405,
+      "eval_overall_precision": 0.7994858611825193,
+      "eval_overall_recall": 0.886039886039886,
+      "eval_runtime": 0.365,
+      "eval_samples_per_second": 512.392,
+      "eval_steps_per_second": 8.22,
       "step": 5618
     },
     {
       "epoch": 54.0,
-      "grad_norm": 2.3576104640960693,
+      "grad_norm": 0.005175034515559673,
       "learning_rate": 2.3000000000000003e-05,
-      "loss": 0.0007,
+      "loss": 0.0012,
       "step": 5724
     },
     {
       "epoch": 54.0,
-      "eval_LOCATION_f1": 0.7948717948717949,
+      "eval_LOCATION_f1": 0.8133333333333332,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7469879518072289,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7428571428571429,
+      "eval_LOCATION_precision": 0.7922077922077922,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.7464788732394366,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6933333333333334,
-      "eval_ORGANIZATION_recall": 0.8,
-      "eval_PERSON_f1": 0.9276315789473684,
+      "eval_ORGANIZATION_precision": 0.6883116883116883,
+      "eval_ORGANIZATION_recall": 0.8153846153846154,
+      "eval_PERSON_f1": 0.9250814332247558,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.9155844155844156,
-      "eval_PERSON_recall": 0.94,
-      "eval_QUANTITY_f1": 0.8125,
+      "eval_PERSON_precision": 0.9044585987261147,
+      "eval_PERSON_recall": 0.9466666666666667,
+      "eval_QUANTITY_f1": 0.7419354838709677,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.7428571428571429,
-      "eval_QUANTITY_recall": 0.896551724137931,
-      "eval_TIME_f1": 0.8732394366197184,
+      "eval_QUANTITY_precision": 0.696969696969697,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8378378378378379,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8378378378378378,
+      "eval_TIME_precision": 0.775,
       "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.3248312175273895,
-      "eval_overall_accuracy": 0.9633275904504061,
-      "eval_overall_f1": 0.8489795918367347,
-      "eval_overall_precision": 0.8125,
-      "eval_overall_recall": 0.8888888888888888,
-      "eval_runtime": 0.9541,
-      "eval_samples_per_second": 195.998,
-      "eval_steps_per_second": 3.144,
+      "eval_loss": 0.309725821018219,
+      "eval_overall_accuracy": 0.9643120846665025,
+      "eval_overall_f1": 0.8435374149659863,
+      "eval_overall_precision": 0.8072916666666666,
+      "eval_overall_recall": 0.8831908831908832,
+      "eval_runtime": 0.367,
+      "eval_samples_per_second": 509.552,
+      "eval_steps_per_second": 8.175,
       "step": 5724
     },
     {
       "epoch": 55.0,
-      "grad_norm": 0.004712595138698816,
+      "grad_norm": 0.0016396279679611325,
       "learning_rate": 2.25e-05,
-      "loss": 0.0011,
+      "loss": 0.0016,
       "step": 5830
     },
     {
       "epoch": 55.0,
-      "eval_LOCATION_f1": 0.8133333333333332,
+      "eval_LOCATION_f1": 0.7866666666666666,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7922077922077922,
-      "eval_LOCATION_recall": 0.8356164383561644,
-      "eval_ORGANIZATION_f1": 0.7432432432432433,
+      "eval_LOCATION_precision": 0.7662337662337663,
+      "eval_LOCATION_recall": 0.8082191780821918,
+      "eval_ORGANIZATION_f1": 0.75,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6626506024096386,
-      "eval_ORGANIZATION_recall": 0.8461538461538461,
-      "eval_PERSON_f1": 0.9215686274509804,
+      "eval_ORGANIZATION_precision": 0.7183098591549296,
+      "eval_ORGANIZATION_recall": 0.7846153846153846,
+      "eval_PERSON_f1": 0.9315960912052117,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.9038461538461539,
-      "eval_PERSON_recall": 0.94,
-      "eval_QUANTITY_f1": 0.8253968253968255,
+      "eval_PERSON_precision": 0.910828025477707,
+      "eval_PERSON_recall": 0.9533333333333334,
+      "eval_QUANTITY_f1": 0.7419354838709677,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.7647058823529411,
-      "eval_QUANTITY_recall": 0.896551724137931,
-      "eval_TIME_f1": 0.8266666666666665,
+      "eval_QUANTITY_precision": 0.696969696969697,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8333333333333333,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.7560975609756098,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.32550135254859924,
+      "eval_TIME_precision": 0.7894736842105263,
+      "eval_TIME_recall": 0.8823529411764706,
+      "eval_loss": 0.3206919729709625,
       "eval_overall_accuracy": 0.963081466896382,
-      "eval_overall_f1": 0.8463611859838276,
-      "eval_overall_precision": 0.80306905370844,
-      "eval_overall_recall": 0.8945868945868946,
-      "eval_runtime": 0.9634,
-      "eval_samples_per_second": 194.103,
-      "eval_steps_per_second": 3.114,
+      "eval_overall_f1": 0.8418156808803302,
+      "eval_overall_precision": 0.8138297872340425,
+      "eval_overall_recall": 0.8717948717948718,
+      "eval_runtime": 0.3659,
+      "eval_samples_per_second": 511.05,
+      "eval_steps_per_second": 8.199,
       "step": 5830
     },
     {
       "epoch": 56.0,
-      "grad_norm": 0.04717090353369713,
+      "grad_norm": 0.004443670157343149,
       "learning_rate": 2.2000000000000003e-05,
-      "loss": 0.0008,
+      "loss": 0.0013,
       "step": 5936
     },
     {
       "epoch": 56.0,
-      "eval_LOCATION_f1": 0.7749999999999999,
+      "eval_LOCATION_f1": 0.7999999999999999,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7126436781609196,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.712121212121212,
+      "eval_LOCATION_precision": 0.7792207792207793,
+      "eval_LOCATION_recall": 0.821917808219178,
+      "eval_ORGANIZATION_f1": 0.723076923076923,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.7014925373134329,
+      "eval_ORGANIZATION_precision": 0.7230769230769231,
       "eval_ORGANIZATION_recall": 0.7230769230769231,
-      "eval_PERSON_f1": 0.9144736842105262,
+      "eval_PERSON_f1": 0.9315960912052117,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.9025974025974026,
-      "eval_PERSON_recall": 0.9266666666666666,
-      "eval_QUANTITY_f1": 0.9032258064516129,
+      "eval_PERSON_precision": 0.910828025477707,
+      "eval_PERSON_recall": 0.9533333333333334,
+      "eval_QUANTITY_f1": 0.7540983606557378,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.8484848484848485,
-      "eval_QUANTITY_recall": 0.9655172413793104,
-      "eval_TIME_f1": 0.9253731343283583,
+      "eval_QUANTITY_precision": 0.71875,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8219178082191781,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.9393939393939394,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.3200584053993225,
-      "eval_overall_accuracy": 0.9625892197883338,
-      "eval_overall_f1": 0.8468965517241379,
-      "eval_overall_precision": 0.820855614973262,
-      "eval_overall_recall": 0.8746438746438746,
-      "eval_runtime": 0.9425,
-      "eval_samples_per_second": 198.4,
-      "eval_steps_per_second": 3.183,
+      "eval_TIME_precision": 0.7692307692307693,
+      "eval_TIME_recall": 0.8823529411764706,
+      "eval_loss": 0.31482699513435364,
+      "eval_overall_accuracy": 0.9635737140044303,
+      "eval_overall_f1": 0.8404993065187241,
+      "eval_overall_precision": 0.8189189189189189,
+      "eval_overall_recall": 0.8632478632478633,
+      "eval_runtime": 0.368,
+      "eval_samples_per_second": 508.191,
+      "eval_steps_per_second": 8.153,
       "step": 5936
     },
     {
       "epoch": 57.0,
-      "grad_norm": 0.006852170918136835,
+      "grad_norm": 0.0009756656363606453,
       "learning_rate": 2.15e-05,
-      "loss": 0.0005,
+      "loss": 0.0012,
       "step": 6042
     },
     {
       "epoch": 57.0,
-      "eval_LOCATION_f1": 0.7875,
+      "eval_LOCATION_f1": 0.8053691275167785,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7241379310344828,
-      "eval_LOCATION_recall": 0.863013698630137,
-      "eval_ORGANIZATION_f1": 0.7428571428571429,
+      "eval_LOCATION_precision": 0.7894736842105263,
+      "eval_LOCATION_recall": 0.821917808219178,
+      "eval_ORGANIZATION_f1": 0.762589928057554,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6933333333333334,
-      "eval_ORGANIZATION_recall": 0.8,
-      "eval_PERSON_f1": 0.9205298013245033,
+      "eval_ORGANIZATION_precision": 0.7162162162162162,
+      "eval_ORGANIZATION_recall": 0.8153846153846154,
+      "eval_PERSON_f1": 0.9315960912052117,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.9144736842105263,
-      "eval_PERSON_recall": 0.9266666666666666,
-      "eval_QUANTITY_f1": 0.8750000000000001,
+      "eval_PERSON_precision": 0.910828025477707,
+      "eval_PERSON_recall": 0.9533333333333334,
+      "eval_QUANTITY_f1": 0.6885245901639345,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.8,
-      "eval_QUANTITY_recall": 0.9655172413793104,
-      "eval_TIME_f1": 0.8571428571428571,
+      "eval_QUANTITY_precision": 0.65625,
+      "eval_QUANTITY_recall": 0.7241379310344828,
+      "eval_TIME_f1": 0.8656716417910447,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8333333333333334,
-      "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.31728222966194153,
+      "eval_TIME_precision": 0.8787878787878788,
+      "eval_TIME_recall": 0.8529411764705882,
+      "eval_loss": 0.3097487688064575,
       "eval_overall_accuracy": 0.9643120846665025,
-      "eval_overall_f1": 0.8478260869565218,
-      "eval_overall_precision": 0.8103896103896104,
-      "eval_overall_recall": 0.8888888888888888,
-      "eval_runtime": 0.956,
-      "eval_samples_per_second": 195.602,
-      "eval_steps_per_second": 3.138,
+      "eval_overall_f1": 0.846473029045643,
+      "eval_overall_precision": 0.8225806451612904,
+      "eval_overall_recall": 0.8717948717948718,
+      "eval_runtime": 0.3637,
+      "eval_samples_per_second": 514.156,
+      "eval_steps_per_second": 8.248,
       "step": 6042
     },
     {
       "epoch": 58.0,
-      "grad_norm": 0.0005476477672345936,
+      "grad_norm": 0.005117403343319893,
       "learning_rate": 2.1e-05,
-      "loss": 0.0002,
+      "loss": 0.0005,
       "step": 6148
     },
     {
       "epoch": 58.0,
-      "eval_LOCATION_f1": 0.7701863354037267,
+      "eval_LOCATION_f1": 0.8157894736842105,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7045454545454546,
+      "eval_LOCATION_precision": 0.7848101265822784,
       "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7361111111111112,
+      "eval_ORGANIZATION_f1": 0.7323943661971831,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6708860759493671,
-      "eval_ORGANIZATION_recall": 0.8153846153846154,
-      "eval_PERSON_f1": 0.9306930693069307,
+      "eval_ORGANIZATION_precision": 0.6753246753246753,
+      "eval_ORGANIZATION_recall": 0.8,
+      "eval_PERSON_f1": 0.9038461538461539,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.9215686274509803,
+      "eval_PERSON_precision": 0.8703703703703703,
       "eval_PERSON_recall": 0.94,
-      "eval_QUANTITY_f1": 0.8571428571428571,
+      "eval_QUANTITY_f1": 0.7619047619047619,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.7941176470588235,
-      "eval_QUANTITY_recall": 0.9310344827586207,
-      "eval_TIME_f1": 0.8732394366197184,
+      "eval_QUANTITY_precision": 0.7058823529411765,
+      "eval_QUANTITY_recall": 0.8275862068965517,
+      "eval_TIME_f1": 0.8108108108108107,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8378378378378378,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.33242636919021606,
-      "eval_overall_accuracy": 0.9648043317745508,
-      "eval_overall_f1": 0.8463611859838276,
-      "eval_overall_precision": 0.80306905370844,
-      "eval_overall_recall": 0.8945868945868946,
-      "eval_runtime": 0.949,
-      "eval_samples_per_second": 197.049,
-      "eval_steps_per_second": 3.161,
+      "eval_TIME_precision": 0.75,
+      "eval_TIME_recall": 0.8823529411764706,
+      "eval_loss": 0.33406224846839905,
+      "eval_overall_accuracy": 0.9633275904504061,
+      "eval_overall_f1": 0.8317631224764468,
+      "eval_overall_precision": 0.7882653061224489,
+      "eval_overall_recall": 0.8803418803418803,
+      "eval_runtime": 0.3652,
+      "eval_samples_per_second": 511.981,
+      "eval_steps_per_second": 8.214,
       "step": 6148
     },
     {
       "epoch": 59.0,
-      "grad_norm": 0.0010370915988460183,
+      "grad_norm": 0.002397208008915186,
       "learning_rate": 2.05e-05,
-      "loss": 0.0009,
+      "loss": 0.0013,
       "step": 6254
     },
     {
       "epoch": 59.0,
-      "eval_LOCATION_f1": 0.7922077922077922,
+      "eval_LOCATION_f1": 0.8243243243243243,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7530864197530864,
+      "eval_LOCATION_precision": 0.8133333333333334,
       "eval_LOCATION_recall": 0.8356164383561644,
-      "eval_ORGANIZATION_f1": 0.763888888888889,
+      "eval_ORGANIZATION_f1": 0.76056338028169,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6962025316455697,
-      "eval_ORGANIZATION_recall": 0.8461538461538461,
-      "eval_PERSON_f1": 0.9306930693069307,
+      "eval_ORGANIZATION_precision": 0.7012987012987013,
+      "eval_ORGANIZATION_recall": 0.8307692307692308,
+      "eval_PERSON_f1": 0.912621359223301,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.9215686274509803,
+      "eval_PERSON_precision": 0.8867924528301887,
       "eval_PERSON_recall": 0.94,
-      "eval_QUANTITY_f1": 0.8387096774193549,
+      "eval_QUANTITY_f1": 0.7419354838709677,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.7878787878787878,
-      "eval_QUANTITY_recall": 0.896551724137931,
-      "eval_TIME_f1": 0.8823529411764706,
+      "eval_QUANTITY_precision": 0.696969696969697,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8823529411764706,
+      "eval_TIME_precision": 0.7317073170731707,
       "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.30726489424705505,
-      "eval_overall_accuracy": 0.9667733202067438,
-      "eval_overall_f1": 0.8563611491108072,
-      "eval_overall_precision": 0.8236842105263158,
-      "eval_overall_recall": 0.8917378917378918,
-      "eval_runtime": 0.9456,
-      "eval_samples_per_second": 197.768,
-      "eval_steps_per_second": 3.173,
+      "eval_loss": 0.3231642544269562,
+      "eval_overall_accuracy": 0.9638198375584544,
+      "eval_overall_f1": 0.8396739130434784,
+      "eval_overall_precision": 0.8025974025974026,
+      "eval_overall_recall": 0.8803418803418803,
+      "eval_runtime": 0.3645,
+      "eval_samples_per_second": 512.977,
+      "eval_steps_per_second": 8.23,
       "step": 6254
     },
     {
       "epoch": 60.0,
-      "grad_norm": 0.0015386795857921243,
+      "grad_norm": 0.0004946636618115008,
       "learning_rate": 2e-05,
-      "loss": 0.0023,
+      "loss": 0.0012,
       "step": 6360
     },
     {
       "epoch": 60.0,
-      "eval_LOCATION_f1": 0.7999999999999999,
+      "eval_LOCATION_f1": 0.8211920529801324,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7560975609756098,
+      "eval_LOCATION_precision": 0.7948717948717948,
       "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7647058823529411,
+      "eval_ORGANIZATION_f1": 0.7285714285714285,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.7323943661971831,
-      "eval_ORGANIZATION_recall": 0.8,
-      "eval_PERSON_f1": 0.9180327868852459,
+      "eval_ORGANIZATION_precision": 0.68,
+      "eval_ORGANIZATION_recall": 0.7846153846153846,
+      "eval_PERSON_f1": 0.9250814332247558,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.9032258064516129,
-      "eval_PERSON_recall": 0.9333333333333333,
-      "eval_QUANTITY_f1": 0.888888888888889,
+      "eval_PERSON_precision": 0.9044585987261147,
+      "eval_PERSON_recall": 0.9466666666666667,
+      "eval_QUANTITY_f1": 0.7213114754098361,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.8235294117647058,
-      "eval_QUANTITY_recall": 0.9655172413793104,
-      "eval_TIME_f1": 0.861111111111111,
+      "eval_QUANTITY_precision": 0.6875,
+      "eval_QUANTITY_recall": 0.7586206896551724,
+      "eval_TIME_f1": 0.8450704225352113,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8157894736842105,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.31632086634635925,
-      "eval_overall_accuracy": 0.9643120846665025,
-      "eval_overall_f1": 0.8563611491108072,
-      "eval_overall_precision": 0.8236842105263158,
-      "eval_overall_recall": 0.8917378917378918,
-      "eval_runtime": 0.9522,
-      "eval_samples_per_second": 196.39,
-      "eval_steps_per_second": 3.151,
+      "eval_TIME_precision": 0.8108108108108109,
+      "eval_TIME_recall": 0.8823529411764706,
+      "eval_loss": 0.3058851659297943,
+      "eval_overall_accuracy": 0.965296578882599,
+      "eval_overall_f1": 0.841095890410959,
+      "eval_overall_precision": 0.8100263852242744,
+      "eval_overall_recall": 0.8746438746438746,
+      "eval_runtime": 0.3662,
+      "eval_samples_per_second": 510.646,
+      "eval_steps_per_second": 8.192,
       "step": 6360
     },
     {
       "epoch": 61.0,
-      "grad_norm": 0.0012541526230052114,
+      "grad_norm": 0.0031809015199542046,
       "learning_rate": 1.9500000000000003e-05,
-      "loss": 0.0004,
+      "loss": 0.0014,
       "step": 6466
     },
     {
       "epoch": 61.0,
-      "eval_LOCATION_f1": 0.7866666666666666,
+      "eval_LOCATION_f1": 0.8076923076923077,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7662337662337663,
-      "eval_LOCATION_recall": 0.8082191780821918,
-      "eval_ORGANIZATION_f1": 0.7424242424242424,
+      "eval_LOCATION_precision": 0.7590361445783133,
+      "eval_LOCATION_recall": 0.863013698630137,
+      "eval_ORGANIZATION_f1": 0.7534246575342466,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.7313432835820896,
-      "eval_ORGANIZATION_recall": 0.7538461538461538,
-      "eval_PERSON_f1": 0.9240924092409241,
+      "eval_ORGANIZATION_precision": 0.6790123456790124,
+      "eval_ORGANIZATION_recall": 0.8461538461538461,
+      "eval_PERSON_f1": 0.934640522875817,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.9150326797385621,
-      "eval_PERSON_recall": 0.9333333333333333,
-      "eval_QUANTITY_f1": 0.8852459016393444,
+      "eval_PERSON_precision": 0.9166666666666666,
+      "eval_PERSON_recall": 0.9533333333333334,
+      "eval_QUANTITY_f1": 0.7301587301587301,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.84375,
-      "eval_QUANTITY_recall": 0.9310344827586207,
-      "eval_TIME_f1": 0.8823529411764706,
+      "eval_QUANTITY_precision": 0.6764705882352942,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8888888888888888,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8823529411764706,
-      "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.30070415139198303,
-      "eval_overall_accuracy": 0.9670194437607679,
-      "eval_overall_f1": 0.8543417366946778,
-      "eval_overall_precision": 0.8402203856749312,
-      "eval_overall_recall": 0.8689458689458689,
-      "eval_runtime": 0.9574,
-      "eval_samples_per_second": 195.321,
-      "eval_steps_per_second": 3.133,
+      "eval_TIME_precision": 0.8421052631578947,
+      "eval_TIME_recall": 0.9411764705882353,
+      "eval_loss": 0.3144487142562866,
+      "eval_overall_accuracy": 0.9662810730986956,
+      "eval_overall_f1": 0.8506056527590848,
+      "eval_overall_precision": 0.8061224489795918,
+      "eval_overall_recall": 0.9002849002849003,
+      "eval_runtime": 0.3638,
+      "eval_samples_per_second": 514.057,
+      "eval_steps_per_second": 8.247,
       "step": 6466
     },
     {
       "epoch": 62.0,
-      "grad_norm": 1.634568691253662,
+      "grad_norm": 0.0013970229774713516,
       "learning_rate": 1.9e-05,
-      "loss": 0.0009,
+      "loss": 0.0002,
       "step": 6572
     },
     {
       "epoch": 62.0,
-      "eval_LOCATION_f1": 0.7999999999999999,
+      "eval_LOCATION_f1": 0.7721518987341772,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7792207792207793,
-      "eval_LOCATION_recall": 0.821917808219178,
-      "eval_ORGANIZATION_f1": 0.7887323943661972,
+      "eval_LOCATION_precision": 0.7176470588235294,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.7361111111111112,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.7272727272727273,
-      "eval_ORGANIZATION_recall": 0.8615384615384616,
-      "eval_PERSON_f1": 0.9281045751633987,
+      "eval_ORGANIZATION_precision": 0.6708860759493671,
+      "eval_ORGANIZATION_recall": 0.8153846153846154,
+      "eval_PERSON_f1": 0.9377049180327869,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.9102564102564102,
-      "eval_PERSON_recall": 0.9466666666666667,
-      "eval_QUANTITY_f1": 0.8387096774193549,
+      "eval_PERSON_precision": 0.9225806451612903,
+      "eval_PERSON_recall": 0.9533333333333334,
+      "eval_QUANTITY_f1": 0.6875,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.7878787878787878,
-      "eval_QUANTITY_recall": 0.896551724137931,
-      "eval_TIME_f1": 0.861111111111111,
+      "eval_QUANTITY_precision": 0.6285714285714286,
+      "eval_QUANTITY_recall": 0.7586206896551724,
+      "eval_TIME_f1": 0.8648648648648648,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8157894736842105,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.30260345339775085,
-      "eval_overall_accuracy": 0.9684961850849126,
-      "eval_overall_f1": 0.8606557377049181,
-      "eval_overall_precision": 0.8267716535433071,
-      "eval_overall_recall": 0.8974358974358975,
-      "eval_runtime": 0.9439,
-      "eval_samples_per_second": 198.118,
-      "eval_steps_per_second": 3.178,
+      "eval_TIME_precision": 0.8,
+      "eval_TIME_recall": 0.9411764705882353,
+      "eval_loss": 0.32296043634414673,
+      "eval_overall_accuracy": 0.965296578882599,
+      "eval_overall_f1": 0.8348993288590604,
+      "eval_overall_precision": 0.7893401015228426,
+      "eval_overall_recall": 0.886039886039886,
+      "eval_runtime": 0.3635,
+      "eval_samples_per_second": 514.438,
+      "eval_steps_per_second": 8.253,
       "step": 6572
     },
     {
       "epoch": 63.0,
-      "grad_norm": 0.42706361413002014,
+      "grad_norm": 0.0013390856329351664,
       "learning_rate": 1.85e-05,
-      "loss": 0.0014,
+      "loss": 0.0007,
       "step": 6678
     },
     {
       "epoch": 63.0,
-      "eval_LOCATION_f1": 0.8133333333333332,
+      "eval_LOCATION_f1": 0.8026315789473685,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7922077922077922,
+      "eval_LOCATION_precision": 0.7721518987341772,
       "eval_LOCATION_recall": 0.8356164383561644,
-      "eval_ORGANIZATION_f1": 0.7972027972027972,
+      "eval_ORGANIZATION_f1": 0.7862068965517242,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.7307692307692307,
+      "eval_ORGANIZATION_precision": 0.7125,
       "eval_ORGANIZATION_recall": 0.8769230769230769,
-      "eval_PERSON_f1": 0.9205298013245033,
+      "eval_PERSON_f1": 0.934640522875817,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.9144736842105263,
-      "eval_PERSON_recall": 0.9266666666666666,
-      "eval_QUANTITY_f1": 0.8852459016393444,
+      "eval_PERSON_precision": 0.9166666666666666,
+      "eval_PERSON_recall": 0.9533333333333334,
+      "eval_QUANTITY_f1": 0.6984126984126984,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.84375,
-      "eval_QUANTITY_recall": 0.9310344827586207,
-      "eval_TIME_f1": 0.8378378378378379,
+      "eval_QUANTITY_precision": 0.6470588235294118,
+      "eval_QUANTITY_recall": 0.7586206896551724,
+      "eval_TIME_f1": 0.8421052631578947,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.775,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.2926293909549713,
-      "eval_overall_accuracy": 0.9660349495446714,
-      "eval_overall_f1": 0.863013698630137,
-      "eval_overall_precision": 0.8311345646437994,
+      "eval_TIME_precision": 0.7619047619047619,
+      "eval_TIME_recall": 0.9411764705882353,
+      "eval_loss": 0.34889984130859375,
+      "eval_overall_accuracy": 0.9635737140044303,
+      "eval_overall_f1": 0.8490566037735849,
+      "eval_overall_precision": 0.8056265984654731,
       "eval_overall_recall": 0.8974358974358975,
-      "eval_runtime": 0.9623,
-      "eval_samples_per_second": 194.319,
-      "eval_steps_per_second": 3.117,
+      "eval_runtime": 0.3645,
+      "eval_samples_per_second": 513.037,
+      "eval_steps_per_second": 8.231,
       "step": 6678
     },
     {
       "epoch": 64.0,
-      "grad_norm": 0.004867079202085733,
+      "grad_norm": 0.001166167319752276,
       "learning_rate": 1.8e-05,
-      "loss": 0.0009,
+      "loss": 0.0001,
       "step": 6784
     },
     {
       "epoch": 64.0,
-      "eval_LOCATION_f1": 0.7530864197530865,
+      "eval_LOCATION_f1": 0.8026315789473685,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.6853932584269663,
+      "eval_LOCATION_precision": 0.7721518987341772,
       "eval_LOCATION_recall": 0.8356164383561644,
-      "eval_ORGANIZATION_f1": 0.7714285714285715,
+      "eval_ORGANIZATION_f1": 0.7832167832167831,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.72,
-      "eval_ORGANIZATION_recall": 0.8307692307692308,
-      "eval_PERSON_f1": 0.9302325581395349,
+      "eval_ORGANIZATION_precision": 0.717948717948718,
+      "eval_ORGANIZATION_recall": 0.8615384615384616,
+      "eval_PERSON_f1": 0.934640522875817,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.9271523178807947,
-      "eval_PERSON_recall": 0.9333333333333333,
-      "eval_QUANTITY_f1": 0.8524590163934426,
+      "eval_PERSON_precision": 0.9166666666666666,
+      "eval_PERSON_recall": 0.9533333333333334,
+      "eval_QUANTITY_f1": 0.6984126984126984,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.8125,
-      "eval_QUANTITY_recall": 0.896551724137931,
-      "eval_TIME_f1": 0.8378378378378379,
+      "eval_QUANTITY_precision": 0.6470588235294118,
+      "eval_QUANTITY_recall": 0.7586206896551724,
+      "eval_TIME_f1": 0.8219178082191781,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.775,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.32596632838249207,
-      "eval_overall_accuracy": 0.9633275904504061,
-      "eval_overall_f1": 0.8455284552845528,
-      "eval_overall_precision": 0.8062015503875969,
+      "eval_TIME_precision": 0.7692307692307693,
+      "eval_TIME_recall": 0.8823529411764706,
+      "eval_loss": 0.34576481580734253,
+      "eval_overall_accuracy": 0.9638198375584544,
+      "eval_overall_f1": 0.8466757123473541,
+      "eval_overall_precision": 0.8082901554404145,
       "eval_overall_recall": 0.8888888888888888,
-      "eval_runtime": 0.9542,
-      "eval_samples_per_second": 195.973,
-      "eval_steps_per_second": 3.144,
+      "eval_runtime": 0.3657,
+      "eval_samples_per_second": 511.325,
+      "eval_steps_per_second": 8.203,
       "step": 6784
     },
     {
       "epoch": 65.0,
-      "grad_norm": 0.28496435284614563,
+      "grad_norm": 0.3260734975337982,
       "learning_rate": 1.75e-05,
       "loss": 0.001,
       "step": 6890
     },
     {
       "epoch": 65.0,
-      "eval_LOCATION_f1": 0.7898089171974523,
+      "eval_LOCATION_f1": 0.7870967741935484,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7380952380952381,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7462686567164178,
+      "eval_LOCATION_precision": 0.7439024390243902,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.7272727272727272,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.7246376811594203,
-      "eval_ORGANIZATION_recall": 0.7692307692307693,
-      "eval_PERSON_f1": 0.9372937293729374,
+      "eval_ORGANIZATION_precision": 0.6666666666666666,
+      "eval_ORGANIZATION_recall": 0.8,
+      "eval_PERSON_f1": 0.934640522875817,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.9281045751633987,
-      "eval_PERSON_recall": 0.9466666666666667,
-      "eval_QUANTITY_f1": 0.8524590163934426,
+      "eval_PERSON_precision": 0.9166666666666666,
+      "eval_PERSON_recall": 0.9533333333333334,
+      "eval_QUANTITY_f1": 0.6875,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.8125,
-      "eval_QUANTITY_recall": 0.896551724137931,
-      "eval_TIME_f1": 0.861111111111111,
+      "eval_QUANTITY_precision": 0.6285714285714286,
+      "eval_QUANTITY_recall": 0.7586206896551724,
+      "eval_TIME_f1": 0.8219178082191781,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8157894736842105,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.31614455580711365,
-      "eval_overall_accuracy": 0.9620969726802855,
-      "eval_overall_f1": 0.8555708390646494,
-      "eval_overall_precision": 0.8271276595744681,
-      "eval_overall_recall": 0.886039886039886,
-      "eval_runtime": 0.9404,
-      "eval_samples_per_second": 198.851,
-      "eval_steps_per_second": 3.19,
+      "eval_TIME_precision": 0.7692307692307693,
+      "eval_TIME_recall": 0.8823529411764706,
+      "eval_loss": 0.340901255607605,
+      "eval_overall_accuracy": 0.9638198375584544,
+      "eval_overall_f1": 0.8313090418353576,
+      "eval_overall_precision": 0.7897435897435897,
+      "eval_overall_recall": 0.8774928774928775,
+      "eval_runtime": 0.3656,
+      "eval_samples_per_second": 511.525,
+      "eval_steps_per_second": 8.206,
       "step": 6890
     },
     {
       "epoch": 66.0,
-      "grad_norm": 0.00030941792647354305,
+      "grad_norm": 0.012995096854865551,
       "learning_rate": 1.7000000000000003e-05,
-      "loss": 0.0004,
+      "loss": 0.001,
       "step": 6996
     },
     {
       "epoch": 66.0,
-      "eval_LOCATION_f1": 0.8079470198675496,
+      "eval_LOCATION_f1": 0.8053691275167785,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.782051282051282,
-      "eval_LOCATION_recall": 0.8356164383561644,
-      "eval_ORGANIZATION_f1": 0.7746478873239436,
+      "eval_LOCATION_precision": 0.7894736842105263,
+      "eval_LOCATION_recall": 0.821917808219178,
+      "eval_ORGANIZATION_f1": 0.7692307692307693,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.7142857142857143,
+      "eval_ORGANIZATION_precision": 0.7051282051282052,
       "eval_ORGANIZATION_recall": 0.8461538461538461,
-      "eval_PERSON_f1": 0.9337748344370861,
+      "eval_PERSON_f1": 0.934640522875817,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.9276315789473685,
-      "eval_PERSON_recall": 0.94,
-      "eval_QUANTITY_f1": 0.8524590163934426,
+      "eval_PERSON_precision": 0.9166666666666666,
+      "eval_PERSON_recall": 0.9533333333333334,
+      "eval_QUANTITY_f1": 0.8064516129032258,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.8125,
-      "eval_QUANTITY_recall": 0.896551724137931,
-      "eval_TIME_f1": 0.8493150684931507,
+      "eval_QUANTITY_precision": 0.7575757575757576,
+      "eval_QUANTITY_recall": 0.8620689655172413,
+      "eval_TIME_f1": 0.8955223880597014,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.7948717948717948,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.30221542716026306,
-      "eval_overall_accuracy": 0.9662810730986956,
-      "eval_overall_f1": 0.8614540466392318,
-      "eval_overall_precision": 0.8306878306878307,
-      "eval_overall_recall": 0.8945868945868946,
-      "eval_runtime": 0.9538,
-      "eval_samples_per_second": 196.058,
-      "eval_steps_per_second": 3.145,
+      "eval_TIME_precision": 0.9090909090909091,
+      "eval_TIME_recall": 0.8823529411764706,
+      "eval_loss": 0.31374266743659973,
+      "eval_overall_accuracy": 0.9665271966527197,
+      "eval_overall_f1": 0.861072902338377,
+      "eval_overall_precision": 0.8324468085106383,
+      "eval_overall_recall": 0.8917378917378918,
+      "eval_runtime": 0.3667,
+      "eval_samples_per_second": 509.933,
+      "eval_steps_per_second": 8.181,
       "step": 6996
     },
     {
       "epoch": 67.0,
-      "grad_norm": 0.0024235707242041826,
+      "grad_norm": 0.0004249998601153493,
       "learning_rate": 1.65e-05,
-      "loss": 0.0005,
+      "loss": 0.0007,
       "step": 7102
     },
     {
       "epoch": 67.0,
-      "eval_LOCATION_f1": 0.7701863354037267,
+      "eval_LOCATION_f1": 0.8133333333333332,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7045454545454546,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7183098591549296,
+      "eval_LOCATION_precision": 0.7922077922077922,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.7586206896551724,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6623376623376623,
-      "eval_ORGANIZATION_recall": 0.7846153846153846,
-      "eval_PERSON_f1": 0.9276315789473684,
+      "eval_ORGANIZATION_precision": 0.6875,
+      "eval_ORGANIZATION_recall": 0.8461538461538461,
+      "eval_PERSON_f1": 0.9315960912052117,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.9155844155844156,
-      "eval_PERSON_recall": 0.94,
-      "eval_QUANTITY_f1": 0.8524590163934426,
+      "eval_PERSON_precision": 0.910828025477707,
+      "eval_PERSON_recall": 0.9533333333333334,
+      "eval_QUANTITY_f1": 0.7812500000000001,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.8125,
-      "eval_QUANTITY_recall": 0.896551724137931,
-      "eval_TIME_f1": 0.8266666666666665,
+      "eval_QUANTITY_precision": 0.7142857142857143,
+      "eval_QUANTITY_recall": 0.8620689655172413,
+      "eval_TIME_f1": 0.8,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.7560975609756098,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.3112208843231201,
-      "eval_overall_accuracy": 0.9638198375584544,
-      "eval_overall_f1": 0.8371467025572006,
-      "eval_overall_precision": 0.7933673469387755,
-      "eval_overall_recall": 0.886039886039886,
-      "eval_runtime": 0.9617,
-      "eval_samples_per_second": 194.442,
-      "eval_steps_per_second": 3.119,
+      "eval_TIME_precision": 0.7317073170731707,
+      "eval_TIME_recall": 0.8823529411764706,
+      "eval_loss": 0.3459141254425049,
+      "eval_overall_accuracy": 0.9633275904504061,
+      "eval_overall_f1": 0.8475033738191633,
+      "eval_overall_precision": 0.8051282051282052,
+      "eval_overall_recall": 0.8945868945868946,
+      "eval_runtime": 0.3626,
+      "eval_samples_per_second": 515.667,
+      "eval_steps_per_second": 8.273,
       "step": 7102
     },
     {
       "epoch": 68.0,
-      "grad_norm": 0.0005896908114664257,
+      "grad_norm": 0.0003220757935196161,
       "learning_rate": 1.6000000000000003e-05,
-      "loss": 0.0015,
+      "loss": 0.0004,
       "step": 7208
     },
     {
       "epoch": 68.0,
-      "eval_LOCATION_f1": 0.8311688311688312,
+      "eval_LOCATION_f1": 0.8133333333333332,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7901234567901234,
-      "eval_LOCATION_recall": 0.8767123287671232,
-      "eval_ORGANIZATION_f1": 0.7659574468085107,
+      "eval_LOCATION_precision": 0.7922077922077922,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.7285714285714285,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.7105263157894737,
-      "eval_ORGANIZATION_recall": 0.8307692307692308,
-      "eval_PERSON_f1": 0.9328859060402684,
+      "eval_ORGANIZATION_precision": 0.68,
+      "eval_ORGANIZATION_recall": 0.7846153846153846,
+      "eval_PERSON_f1": 0.9281045751633987,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.9391891891891891,
-      "eval_PERSON_recall": 0.9266666666666666,
-      "eval_QUANTITY_f1": 0.8524590163934426,
+      "eval_PERSON_precision": 0.9102564102564102,
+      "eval_PERSON_recall": 0.9466666666666667,
+      "eval_QUANTITY_f1": 0.7301587301587301,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.8125,
-      "eval_QUANTITY_recall": 0.896551724137931,
-      "eval_TIME_f1": 0.8732394366197184,
+      "eval_QUANTITY_precision": 0.6764705882352942,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8695652173913043,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8378378378378378,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.306359201669693,
-      "eval_overall_accuracy": 0.9665271966527197,
-      "eval_overall_f1": 0.8662068965517241,
-      "eval_overall_precision": 0.839572192513369,
-      "eval_overall_recall": 0.8945868945868946,
-      "eval_runtime": 0.9399,
-      "eval_samples_per_second": 198.964,
-      "eval_steps_per_second": 3.192,
+      "eval_TIME_precision": 0.8571428571428571,
+      "eval_TIME_recall": 0.8823529411764706,
+      "eval_loss": 0.31549710035324097,
+      "eval_overall_accuracy": 0.9648043317745508,
+      "eval_overall_f1": 0.8434065934065934,
+      "eval_overall_precision": 0.8143236074270557,
+      "eval_overall_recall": 0.8746438746438746,
+      "eval_runtime": 0.3751,
+      "eval_samples_per_second": 498.467,
+      "eval_steps_per_second": 7.997,
       "step": 7208
     },
     {
       "epoch": 69.0,
-      "grad_norm": 0.0005983751616440713,
+      "grad_norm": 0.013907487504184246,
       "learning_rate": 1.55e-05,
-      "loss": 0.0016,
+      "loss": 0.0004,
       "step": 7314
     },
     {
       "epoch": 69.0,
-      "eval_LOCATION_f1": 0.8344370860927152,
+      "eval_LOCATION_f1": 0.8026315789473685,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.8076923076923077,
-      "eval_LOCATION_recall": 0.863013698630137,
-      "eval_ORGANIZATION_f1": 0.7714285714285715,
+      "eval_LOCATION_precision": 0.7721518987341772,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.7222222222222222,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.72,
-      "eval_ORGANIZATION_recall": 0.8307692307692308,
-      "eval_PERSON_f1": 0.9311475409836065,
+      "eval_ORGANIZATION_precision": 0.6582278481012658,
+      "eval_ORGANIZATION_recall": 0.8,
+      "eval_PERSON_f1": 0.9220779220779222,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.9161290322580645,
+      "eval_PERSON_precision": 0.8987341772151899,
       "eval_PERSON_recall": 0.9466666666666667,
-      "eval_QUANTITY_f1": 0.8524590163934426,
+      "eval_QUANTITY_f1": 0.75,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.8125,
-      "eval_QUANTITY_recall": 0.896551724137931,
-      "eval_TIME_f1": 0.9117647058823528,
+      "eval_QUANTITY_precision": 0.6857142857142857,
+      "eval_QUANTITY_recall": 0.8275862068965517,
+      "eval_TIME_f1": 0.8695652173913043,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.9117647058823529,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.3030541241168976,
-      "eval_overall_accuracy": 0.9689884321929608,
-      "eval_overall_f1": 0.8717241379310345,
-      "eval_overall_precision": 0.8449197860962567,
-      "eval_overall_recall": 0.9002849002849003,
-      "eval_runtime": 0.9552,
-      "eval_samples_per_second": 195.774,
-      "eval_steps_per_second": 3.141,
+      "eval_TIME_precision": 0.8571428571428571,
+      "eval_TIME_recall": 0.8823529411764706,
+      "eval_loss": 0.3485003411769867,
+      "eval_overall_accuracy": 0.963081466896382,
+      "eval_overall_f1": 0.8385345997286296,
+      "eval_overall_precision": 0.8005181347150259,
+      "eval_overall_recall": 0.8803418803418803,
+      "eval_runtime": 0.3635,
+      "eval_samples_per_second": 514.499,
+      "eval_steps_per_second": 8.254,
       "step": 7314
     },
     {
       "epoch": 70.0,
-      "grad_norm": 0.00572817400097847,
+      "grad_norm": 0.007048820145428181,
       "learning_rate": 1.5e-05,
-      "loss": 0.0014,
+      "loss": 0.0003,
       "step": 7420
     },
     {
       "epoch": 70.0,
-      "eval_LOCATION_f1": 0.802547770700637,
+      "eval_LOCATION_f1": 0.794701986754967,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.75,
-      "eval_LOCATION_recall": 0.863013698630137,
-      "eval_ORGANIZATION_f1": 0.7338129496402878,
+      "eval_LOCATION_precision": 0.7692307692307693,
+      "eval_LOCATION_recall": 0.821917808219178,
+      "eval_ORGANIZATION_f1": 0.7310344827586207,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6891891891891891,
-      "eval_ORGANIZATION_recall": 0.7846153846153846,
-      "eval_PERSON_f1": 0.9144736842105262,
+      "eval_ORGANIZATION_precision": 0.6625,
+      "eval_ORGANIZATION_recall": 0.8153846153846154,
+      "eval_PERSON_f1": 0.9311475409836065,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.9025974025974026,
-      "eval_PERSON_recall": 0.9266666666666666,
-      "eval_QUANTITY_f1": 0.8524590163934426,
+      "eval_PERSON_precision": 0.9161290322580645,
+      "eval_PERSON_recall": 0.9466666666666667,
+      "eval_QUANTITY_f1": 0.7419354838709677,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.8125,
-      "eval_QUANTITY_recall": 0.896551724137931,
-      "eval_TIME_f1": 0.861111111111111,
+      "eval_QUANTITY_precision": 0.696969696969697,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8695652173913043,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8157894736842105,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.3171394467353821,
-      "eval_overall_accuracy": 0.9640659611124784,
-      "eval_overall_f1": 0.8458390177353343,
-      "eval_overall_precision": 0.8115183246073299,
-      "eval_overall_recall": 0.8831908831908832,
-      "eval_runtime": 0.9485,
-      "eval_samples_per_second": 197.155,
-      "eval_steps_per_second": 3.163,
+      "eval_TIME_precision": 0.8571428571428571,
+      "eval_TIME_recall": 0.8823529411764706,
+      "eval_loss": 0.3381772041320801,
+      "eval_overall_accuracy": 0.9638198375584544,
+      "eval_overall_f1": 0.8415300546448087,
+      "eval_overall_precision": 0.8083989501312336,
+      "eval_overall_recall": 0.8774928774928775,
+      "eval_runtime": 0.3678,
+      "eval_samples_per_second": 508.476,
+      "eval_steps_per_second": 8.157,
       "step": 7420
     },
     {
       "epoch": 71.0,
-      "grad_norm": 0.021534860134124756,
+      "grad_norm": 0.0006086269277147949,
       "learning_rate": 1.45e-05,
-      "loss": 0.0007,
+      "loss": 0.001,
       "step": 7526
     },
     {
       "epoch": 71.0,
-      "eval_LOCATION_f1": 0.7974683544303799,
+      "eval_LOCATION_f1": 0.8266666666666668,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7411764705882353,
-      "eval_LOCATION_recall": 0.863013698630137,
+      "eval_LOCATION_precision": 0.8051948051948052,
+      "eval_LOCATION_recall": 0.8493150684931506,
       "eval_ORGANIZATION_f1": 0.7375886524822695,
       "eval_ORGANIZATION_number": 65,
       "eval_ORGANIZATION_precision": 0.6842105263157895,
       "eval_ORGANIZATION_recall": 0.8,
-      "eval_PERSON_f1": 0.9210526315789475,
+      "eval_PERSON_f1": 0.9342105263157895,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.9090909090909091,
-      "eval_PERSON_recall": 0.9333333333333333,
-      "eval_QUANTITY_f1": 0.8387096774193549,
+      "eval_PERSON_precision": 0.922077922077922,
+      "eval_PERSON_recall": 0.9466666666666667,
+      "eval_QUANTITY_f1": 0.7301587301587301,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.7878787878787878,
-      "eval_QUANTITY_recall": 0.896551724137931,
-      "eval_TIME_f1": 0.8985507246376812,
+      "eval_QUANTITY_precision": 0.6764705882352942,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8571428571428571,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8857142857142857,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.3164141774177551,
-      "eval_overall_accuracy": 0.965296578882599,
-      "eval_overall_f1": 0.8501362397820162,
-      "eval_overall_precision": 0.814621409921671,
-      "eval_overall_recall": 0.8888888888888888,
-      "eval_runtime": 0.9499,
-      "eval_samples_per_second": 196.869,
-      "eval_steps_per_second": 3.158,
+      "eval_TIME_precision": 0.8333333333333334,
+      "eval_TIME_recall": 0.8823529411764706,
+      "eval_loss": 0.31484413146972656,
+      "eval_overall_accuracy": 0.9657888259906473,
+      "eval_overall_f1": 0.8489010989010989,
+      "eval_overall_precision": 0.8196286472148541,
+      "eval_overall_recall": 0.8803418803418803,
+      "eval_runtime": 0.3633,
+      "eval_samples_per_second": 514.677,
+      "eval_steps_per_second": 8.257,
       "step": 7526
     },
     {
       "epoch": 72.0,
-      "grad_norm": 0.0002935581433121115,
+      "grad_norm": 0.0002179538569180295,
       "learning_rate": 1.4000000000000001e-05,
-      "loss": 0.0007,
+      "loss": 0.0003,
       "step": 7632
     },
     {
       "epoch": 72.0,
-      "eval_LOCATION_f1": 0.7973856209150326,
+      "eval_LOCATION_f1": 0.832214765100671,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7625,
-      "eval_LOCATION_recall": 0.8356164383561644,
-      "eval_ORGANIZATION_f1": 0.7428571428571429,
+      "eval_LOCATION_precision": 0.8157894736842105,
+      "eval_LOCATION_recall": 0.8493150684931506,
+      "eval_ORGANIZATION_f1": 0.7462686567164178,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6933333333333334,
-      "eval_ORGANIZATION_recall": 0.8,
-      "eval_PERSON_f1": 0.9235880398671096,
+      "eval_ORGANIZATION_precision": 0.7246376811594203,
+      "eval_ORGANIZATION_recall": 0.7692307692307693,
+      "eval_PERSON_f1": 0.9372937293729374,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.9205298013245033,
-      "eval_PERSON_recall": 0.9266666666666666,
-      "eval_QUANTITY_f1": 0.8387096774193549,
+      "eval_PERSON_precision": 0.9281045751633987,
+      "eval_PERSON_recall": 0.9466666666666667,
+      "eval_QUANTITY_f1": 0.75,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.7878787878787878,
-      "eval_QUANTITY_recall": 0.896551724137931,
-      "eval_TIME_f1": 0.8695652173913043,
+      "eval_QUANTITY_precision": 0.6857142857142857,
+      "eval_QUANTITY_recall": 0.8275862068965517,
+      "eval_TIME_f1": 0.8571428571428571,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8571428571428571,
+      "eval_TIME_precision": 0.8333333333333334,
       "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.3045300543308258,
-      "eval_overall_accuracy": 0.9657888259906473,
-      "eval_overall_f1": 0.8496551724137932,
-      "eval_overall_precision": 0.8235294117647058,
+      "eval_loss": 0.3216892182826996,
+      "eval_overall_accuracy": 0.9660349495446714,
+      "eval_overall_f1": 0.8555555555555556,
+      "eval_overall_precision": 0.8346883468834688,
       "eval_overall_recall": 0.8774928774928775,
-      "eval_runtime": 1.093,
-      "eval_samples_per_second": 171.083,
-      "eval_steps_per_second": 2.745,
+      "eval_runtime": 0.377,
+      "eval_samples_per_second": 495.967,
+      "eval_steps_per_second": 7.957,
       "step": 7632
     },
     {
       "epoch": 73.0,
-      "grad_norm": 0.0001844457583501935,
+      "grad_norm": 0.0002243828057544306,
       "learning_rate": 1.3500000000000001e-05,
-      "loss": 0.0004,
+      "loss": 0.0007,
       "step": 7738
     },
     {
       "epoch": 73.0,
-      "eval_LOCATION_f1": 0.8266666666666668,
+      "eval_LOCATION_f1": 0.8157894736842105,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.8051948051948052,
+      "eval_LOCATION_precision": 0.7848101265822784,
       "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.759124087591241,
+      "eval_ORGANIZATION_f1": 0.7586206896551724,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.7222222222222222,
-      "eval_ORGANIZATION_recall": 0.8,
-      "eval_PERSON_f1": 0.9306930693069307,
+      "eval_ORGANIZATION_precision": 0.6875,
+      "eval_ORGANIZATION_recall": 0.8461538461538461,
+      "eval_PERSON_f1": 0.9276315789473684,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.9215686274509803,
+      "eval_PERSON_precision": 0.9155844155844156,
       "eval_PERSON_recall": 0.94,
-      "eval_QUANTITY_f1": 0.8387096774193549,
+      "eval_QUANTITY_f1": 0.7301587301587301,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.7878787878787878,
-      "eval_QUANTITY_recall": 0.896551724137931,
-      "eval_TIME_f1": 0.8823529411764706,
+      "eval_QUANTITY_precision": 0.6764705882352942,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8695652173913043,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8823529411764706,
+      "eval_TIME_precision": 0.8571428571428571,
       "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.3072561025619507,
-      "eval_overall_accuracy": 0.9667733202067438,
-      "eval_overall_f1": 0.8638888888888889,
-      "eval_overall_precision": 0.8428184281842819,
+      "eval_loss": 0.3233834207057953,
+      "eval_overall_accuracy": 0.9655427024366232,
+      "eval_overall_f1": 0.8485675306957708,
+      "eval_overall_precision": 0.8141361256544503,
       "eval_overall_recall": 0.886039886039886,
-      "eval_runtime": 0.9617,
-      "eval_samples_per_second": 194.445,
-      "eval_steps_per_second": 3.119,
+      "eval_runtime": 0.3624,
+      "eval_samples_per_second": 516.015,
+      "eval_steps_per_second": 8.278,
       "step": 7738
     },
     {
       "epoch": 74.0,
-      "grad_norm": 0.000534932769369334,
+      "grad_norm": 0.0005150302313268185,
       "learning_rate": 1.3000000000000001e-05,
-      "loss": 0.0005,
+      "loss": 0.0003,
       "step": 7844
     },
     {
@@ -2868,1054 +2868,1054 @@
       "eval_LOCATION_number": 73,
       "eval_LOCATION_precision": 0.782051282051282,
       "eval_LOCATION_recall": 0.8356164383561644,
-      "eval_ORGANIZATION_f1": 0.7375886524822695,
+      "eval_ORGANIZATION_f1": 0.7482993197278912,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6842105263157895,
-      "eval_ORGANIZATION_recall": 0.8,
-      "eval_PERSON_f1": 0.9215686274509804,
+      "eval_ORGANIZATION_precision": 0.6707317073170732,
+      "eval_ORGANIZATION_recall": 0.8461538461538461,
+      "eval_PERSON_f1": 0.9185667752442996,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.9038461538461539,
+      "eval_PERSON_precision": 0.8980891719745223,
       "eval_PERSON_recall": 0.94,
-      "eval_QUANTITY_f1": 0.8387096774193549,
+      "eval_QUANTITY_f1": 0.7301587301587301,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.7878787878787878,
-      "eval_QUANTITY_recall": 0.896551724137931,
-      "eval_TIME_f1": 0.8571428571428571,
+      "eval_QUANTITY_precision": 0.6764705882352942,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8695652173913043,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8333333333333334,
+      "eval_TIME_precision": 0.8571428571428571,
       "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.3145270049571991,
-      "eval_overall_accuracy": 0.9643120846665025,
-      "eval_overall_f1": 0.8493150684931507,
-      "eval_overall_precision": 0.8179419525065963,
+      "eval_loss": 0.317582905292511,
+      "eval_overall_accuracy": 0.9645582082205267,
+      "eval_overall_f1": 0.8412483039348712,
+      "eval_overall_precision": 0.8031088082901554,
       "eval_overall_recall": 0.8831908831908832,
-      "eval_runtime": 0.9509,
-      "eval_samples_per_second": 196.652,
-      "eval_steps_per_second": 3.155,
+      "eval_runtime": 0.3658,
+      "eval_samples_per_second": 511.223,
+      "eval_steps_per_second": 8.201,
       "step": 7844
     },
     {
       "epoch": 75.0,
-      "grad_norm": 0.000326872686855495,
+      "grad_norm": 0.0032255067490041256,
       "learning_rate": 1.25e-05,
-      "loss": 0.0006,
+      "loss": 0.0005,
       "step": 7950
     },
     {
       "epoch": 75.0,
-      "eval_LOCATION_f1": 0.8266666666666668,
+      "eval_LOCATION_f1": 0.8051948051948052,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.8051948051948052,
+      "eval_LOCATION_precision": 0.7654320987654321,
       "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7375886524822695,
+      "eval_ORGANIZATION_f1": 0.75,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6842105263157895,
-      "eval_ORGANIZATION_recall": 0.8,
-      "eval_PERSON_f1": 0.9302325581395349,
+      "eval_ORGANIZATION_precision": 0.6835443037974683,
+      "eval_ORGANIZATION_recall": 0.8307692307692308,
+      "eval_PERSON_f1": 0.9281045751633987,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.9271523178807947,
-      "eval_PERSON_recall": 0.9333333333333333,
-      "eval_QUANTITY_f1": 0.8524590163934426,
+      "eval_PERSON_precision": 0.9102564102564102,
+      "eval_PERSON_recall": 0.9466666666666667,
+      "eval_QUANTITY_f1": 0.7619047619047619,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.8125,
-      "eval_QUANTITY_recall": 0.896551724137931,
-      "eval_TIME_f1": 0.8823529411764706,
+      "eval_QUANTITY_precision": 0.7058823529411765,
+      "eval_QUANTITY_recall": 0.8275862068965517,
+      "eval_TIME_f1": 0.8333333333333333,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8823529411764706,
+      "eval_TIME_precision": 0.7894736842105263,
       "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.30685240030288696,
-      "eval_overall_accuracy": 0.9660349495446714,
-      "eval_overall_f1": 0.8599167822468793,
-      "eval_overall_precision": 0.8378378378378378,
-      "eval_overall_recall": 0.8831908831908832,
-      "eval_runtime": 0.9442,
-      "eval_samples_per_second": 198.046,
-      "eval_steps_per_second": 3.177,
+      "eval_loss": 0.34308016300201416,
+      "eval_overall_accuracy": 0.9645582082205267,
+      "eval_overall_f1": 0.8443843031123139,
+      "eval_overall_precision": 0.8041237113402062,
+      "eval_overall_recall": 0.8888888888888888,
+      "eval_runtime": 0.3678,
+      "eval_samples_per_second": 508.46,
+      "eval_steps_per_second": 8.157,
       "step": 7950
     },
     {
       "epoch": 76.0,
-      "grad_norm": 0.00039326981641352177,
+      "grad_norm": 0.0014211718225851655,
       "learning_rate": 1.2e-05,
-      "loss": 0.0003,
+      "loss": 0.0005,
       "step": 8056
     },
     {
       "epoch": 76.0,
-      "eval_LOCATION_f1": 0.8104575163398693,
+      "eval_LOCATION_f1": 0.8187919463087249,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.775,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7272727272727273,
+      "eval_LOCATION_precision": 0.8026315789473685,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.7586206896551724,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.7164179104477612,
-      "eval_ORGANIZATION_recall": 0.7384615384615385,
-      "eval_PERSON_f1": 0.9337748344370861,
+      "eval_ORGANIZATION_precision": 0.6875,
+      "eval_ORGANIZATION_recall": 0.8461538461538461,
+      "eval_PERSON_f1": 0.934640522875817,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.9276315789473685,
-      "eval_PERSON_recall": 0.94,
-      "eval_QUANTITY_f1": 0.8709677419354839,
+      "eval_PERSON_precision": 0.9166666666666666,
+      "eval_PERSON_recall": 0.9533333333333334,
+      "eval_QUANTITY_f1": 0.7301587301587301,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.8181818181818182,
-      "eval_QUANTITY_recall": 0.9310344827586207,
-      "eval_TIME_f1": 0.8695652173913043,
+      "eval_QUANTITY_precision": 0.6764705882352942,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8333333333333333,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8571428571428571,
+      "eval_TIME_precision": 0.7894736842105263,
       "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.30713656544685364,
-      "eval_overall_accuracy": 0.9662810730986956,
-      "eval_overall_f1": 0.8579387186629526,
-      "eval_overall_precision": 0.8392370572207084,
-      "eval_overall_recall": 0.8774928774928775,
-      "eval_runtime": 0.9561,
-      "eval_samples_per_second": 195.586,
-      "eval_steps_per_second": 3.138,
+      "eval_loss": 0.3415742814540863,
+      "eval_overall_accuracy": 0.9643120846665025,
+      "eval_overall_f1": 0.8489795918367347,
+      "eval_overall_precision": 0.8125,
+      "eval_overall_recall": 0.8888888888888888,
+      "eval_runtime": 0.371,
+      "eval_samples_per_second": 504.074,
+      "eval_steps_per_second": 8.087,
       "step": 8056
     },
     {
       "epoch": 77.0,
-      "grad_norm": 8.431855201721191,
+      "grad_norm": 0.00039511857903562486,
       "learning_rate": 1.1500000000000002e-05,
-      "loss": 0.001,
+      "loss": 0.0002,
       "step": 8162
     },
     {
       "epoch": 77.0,
-      "eval_LOCATION_f1": 0.7999999999999999,
+      "eval_LOCATION_f1": 0.7843137254901962,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7560975609756098,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7153284671532848,
+      "eval_LOCATION_precision": 0.75,
+      "eval_LOCATION_recall": 0.821917808219178,
+      "eval_ORGANIZATION_f1": 0.75,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6805555555555556,
-      "eval_ORGANIZATION_recall": 0.7538461538461538,
-      "eval_PERSON_f1": 0.9306930693069307,
+      "eval_ORGANIZATION_precision": 0.6835443037974683,
+      "eval_ORGANIZATION_recall": 0.8307692307692308,
+      "eval_PERSON_f1": 0.9250814332247558,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.9215686274509803,
-      "eval_PERSON_recall": 0.94,
-      "eval_QUANTITY_f1": 0.9180327868852458,
+      "eval_PERSON_precision": 0.9044585987261147,
+      "eval_PERSON_recall": 0.9466666666666667,
+      "eval_QUANTITY_f1": 0.7301587301587301,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.875,
-      "eval_QUANTITY_recall": 0.9655172413793104,
-      "eval_TIME_f1": 0.8695652173913043,
+      "eval_QUANTITY_precision": 0.6764705882352942,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8450704225352113,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8571428571428571,
+      "eval_TIME_precision": 0.8108108108108109,
       "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.29899826645851135,
-      "eval_overall_accuracy": 0.9677578144228403,
-      "eval_overall_f1": 0.8551724137931035,
-      "eval_overall_precision": 0.8288770053475936,
-      "eval_overall_recall": 0.8831908831908832,
-      "eval_runtime": 0.9455,
-      "eval_samples_per_second": 197.787,
-      "eval_steps_per_second": 3.173,
+      "eval_loss": 0.32952412962913513,
+      "eval_overall_accuracy": 0.9645582082205267,
+      "eval_overall_f1": 0.8373983739837397,
+      "eval_overall_precision": 0.7984496124031008,
+      "eval_overall_recall": 0.8803418803418803,
+      "eval_runtime": 0.3676,
+      "eval_samples_per_second": 508.663,
+      "eval_steps_per_second": 8.16,
       "step": 8162
     },
     {
       "epoch": 78.0,
-      "grad_norm": 0.0007179285166785121,
+      "grad_norm": 0.00022415521380025893,
       "learning_rate": 1.1000000000000001e-05,
-      "loss": 0.0005,
+      "loss": 0.0001,
       "step": 8268
     },
     {
       "epoch": 78.0,
-      "eval_LOCATION_f1": 0.8129032258064516,
+      "eval_LOCATION_f1": 0.8157894736842105,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7682926829268293,
-      "eval_LOCATION_recall": 0.863013698630137,
-      "eval_ORGANIZATION_f1": 0.7074829931972789,
+      "eval_LOCATION_precision": 0.7848101265822784,
+      "eval_LOCATION_recall": 0.8493150684931506,
+      "eval_ORGANIZATION_f1": 0.75,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6341463414634146,
-      "eval_ORGANIZATION_recall": 0.8,
-      "eval_PERSON_f1": 0.9271523178807947,
+      "eval_ORGANIZATION_precision": 0.6835443037974683,
+      "eval_ORGANIZATION_recall": 0.8307692307692308,
+      "eval_PERSON_f1": 0.9281045751633987,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.9210526315789473,
-      "eval_PERSON_recall": 0.9333333333333333,
-      "eval_QUANTITY_f1": 0.8852459016393444,
+      "eval_PERSON_precision": 0.9102564102564102,
+      "eval_PERSON_recall": 0.9466666666666667,
+      "eval_QUANTITY_f1": 0.7540983606557378,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.84375,
-      "eval_QUANTITY_recall": 0.9310344827586207,
-      "eval_TIME_f1": 0.8695652173913043,
+      "eval_QUANTITY_precision": 0.71875,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8450704225352113,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8571428571428571,
+      "eval_TIME_precision": 0.8108108108108109,
       "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.3076816499233246,
-      "eval_overall_accuracy": 0.9662810730986956,
-      "eval_overall_f1": 0.8501362397820162,
-      "eval_overall_precision": 0.814621409921671,
-      "eval_overall_recall": 0.8888888888888888,
-      "eval_runtime": 0.9591,
-      "eval_samples_per_second": 194.973,
-      "eval_steps_per_second": 3.128,
+      "eval_loss": 0.3368324041366577,
+      "eval_overall_accuracy": 0.9650504553285749,
+      "eval_overall_f1": 0.8474114441416893,
+      "eval_overall_precision": 0.8120104438642297,
+      "eval_overall_recall": 0.886039886039886,
+      "eval_runtime": 0.384,
+      "eval_samples_per_second": 487.008,
+      "eval_steps_per_second": 7.813,
       "step": 8268
     },
     {
       "epoch": 79.0,
-      "grad_norm": 0.0006159798358567059,
+      "grad_norm": 0.0002906069566961378,
       "learning_rate": 1.05e-05,
-      "loss": 0.0001,
+      "loss": 0.0004,
       "step": 8374
     },
     {
       "epoch": 79.0,
-      "eval_LOCATION_f1": 0.8129032258064516,
+      "eval_LOCATION_f1": 0.8133333333333332,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7682926829268293,
-      "eval_LOCATION_recall": 0.863013698630137,
-      "eval_ORGANIZATION_f1": 0.7310344827586207,
+      "eval_LOCATION_precision": 0.7922077922077922,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.76056338028169,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6625,
-      "eval_ORGANIZATION_recall": 0.8153846153846154,
-      "eval_PERSON_f1": 0.9306930693069307,
+      "eval_ORGANIZATION_precision": 0.7012987012987013,
+      "eval_ORGANIZATION_recall": 0.8307692307692308,
+      "eval_PERSON_f1": 0.9311475409836065,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.9215686274509803,
-      "eval_PERSON_recall": 0.94,
-      "eval_QUANTITY_f1": 0.8709677419354839,
+      "eval_PERSON_precision": 0.9161290322580645,
+      "eval_PERSON_recall": 0.9466666666666667,
+      "eval_QUANTITY_f1": 0.7540983606557378,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.8181818181818182,
-      "eval_QUANTITY_recall": 0.9310344827586207,
-      "eval_TIME_f1": 0.8732394366197184,
+      "eval_QUANTITY_precision": 0.71875,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8450704225352113,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8378378378378378,
-      "eval_TIME_recall": 0.9117647058823529,
-      "eval_loss": 0.31122517585754395,
-      "eval_overall_accuracy": 0.9665271966527197,
-      "eval_overall_f1": 0.8559782608695653,
-      "eval_overall_precision": 0.8181818181818182,
-      "eval_overall_recall": 0.8974358974358975,
-      "eval_runtime": 0.9466,
-      "eval_samples_per_second": 197.542,
-      "eval_steps_per_second": 3.169,
+      "eval_TIME_precision": 0.8108108108108109,
+      "eval_TIME_recall": 0.8823529411764706,
+      "eval_loss": 0.3211762309074402,
+      "eval_overall_accuracy": 0.9662810730986956,
+      "eval_overall_f1": 0.850480109739369,
+      "eval_overall_precision": 0.8201058201058201,
+      "eval_overall_recall": 0.8831908831908832,
+      "eval_runtime": 0.3664,
+      "eval_samples_per_second": 510.422,
+      "eval_steps_per_second": 8.189,
       "step": 8374
     },
     {
       "epoch": 80.0,
-      "grad_norm": 0.0002559934800956398,
+      "grad_norm": 0.00019364221952855587,
       "learning_rate": 1e-05,
       "loss": 0.0001,
       "step": 8480
     },
     {
       "epoch": 80.0,
-      "eval_LOCATION_f1": 0.8157894736842105,
+      "eval_LOCATION_f1": 0.8133333333333332,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7848101265822784,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7286821705426355,
+      "eval_LOCATION_precision": 0.7922077922077922,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.76056338028169,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.734375,
-      "eval_ORGANIZATION_recall": 0.7230769230769231,
-      "eval_PERSON_f1": 0.9337748344370861,
+      "eval_ORGANIZATION_precision": 0.7012987012987013,
+      "eval_ORGANIZATION_recall": 0.8307692307692308,
+      "eval_PERSON_f1": 0.9377049180327869,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.9276315789473685,
-      "eval_PERSON_recall": 0.94,
-      "eval_QUANTITY_f1": 0.8852459016393444,
+      "eval_PERSON_precision": 0.9225806451612903,
+      "eval_PERSON_recall": 0.9533333333333334,
+      "eval_QUANTITY_f1": 0.7540983606557378,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.84375,
-      "eval_QUANTITY_recall": 0.9310344827586207,
-      "eval_TIME_f1": 0.8695652173913043,
+      "eval_QUANTITY_precision": 0.71875,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8450704225352113,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8571428571428571,
+      "eval_TIME_precision": 0.8108108108108109,
       "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.3062417209148407,
-      "eval_overall_accuracy": 0.9680039379768643,
-      "eval_overall_f1": 0.8611500701262271,
-      "eval_overall_precision": 0.8480662983425414,
-      "eval_overall_recall": 0.8746438746438746,
-      "eval_runtime": 1.0069,
-      "eval_samples_per_second": 185.721,
-      "eval_steps_per_second": 2.979,
+      "eval_loss": 0.3226649761199951,
+      "eval_overall_accuracy": 0.9662810730986956,
+      "eval_overall_f1": 0.8532235939643347,
+      "eval_overall_precision": 0.8227513227513228,
+      "eval_overall_recall": 0.886039886039886,
+      "eval_runtime": 0.3657,
+      "eval_samples_per_second": 511.395,
+      "eval_steps_per_second": 8.204,
       "step": 8480
     },
     {
       "epoch": 81.0,
-      "grad_norm": 0.00021754943009000272,
+      "grad_norm": 0.00011983401782345027,
       "learning_rate": 9.5e-06,
-      "loss": 0.0005,
+      "loss": 0.0003,
       "step": 8586
     },
     {
       "epoch": 81.0,
-      "eval_LOCATION_f1": 0.8104575163398693,
+      "eval_LOCATION_f1": 0.8266666666666668,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.775,
+      "eval_LOCATION_precision": 0.8051948051948052,
       "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7482014388489208,
+      "eval_ORGANIZATION_f1": 0.7552447552447553,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.7027027027027027,
-      "eval_ORGANIZATION_recall": 0.8,
-      "eval_PERSON_f1": 0.9306930693069307,
+      "eval_ORGANIZATION_precision": 0.6923076923076923,
+      "eval_ORGANIZATION_recall": 0.8307692307692308,
+      "eval_PERSON_f1": 0.9250814332247558,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.9215686274509803,
-      "eval_PERSON_recall": 0.94,
-      "eval_QUANTITY_f1": 0.8709677419354839,
+      "eval_PERSON_precision": 0.9044585987261147,
+      "eval_PERSON_recall": 0.9466666666666667,
+      "eval_QUANTITY_f1": 0.7096774193548386,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.8181818181818182,
-      "eval_QUANTITY_recall": 0.9310344827586207,
-      "eval_TIME_f1": 0.8571428571428571,
+      "eval_QUANTITY_precision": 0.6666666666666666,
+      "eval_QUANTITY_recall": 0.7586206896551724,
+      "eval_TIME_f1": 0.861111111111111,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8333333333333334,
-      "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.3124181032180786,
-      "eval_overall_accuracy": 0.9670194437607679,
-      "eval_overall_f1": 0.858321870701513,
-      "eval_overall_precision": 0.8297872340425532,
-      "eval_overall_recall": 0.8888888888888888,
-      "eval_runtime": 0.9515,
-      "eval_samples_per_second": 196.533,
-      "eval_steps_per_second": 3.153,
+      "eval_TIME_precision": 0.8157894736842105,
+      "eval_TIME_recall": 0.9117647058823529,
+      "eval_loss": 0.3216117322444916,
+      "eval_overall_accuracy": 0.9662810730986956,
+      "eval_overall_f1": 0.8474114441416893,
+      "eval_overall_precision": 0.8120104438642297,
+      "eval_overall_recall": 0.886039886039886,
+      "eval_runtime": 0.3623,
+      "eval_samples_per_second": 516.203,
+      "eval_steps_per_second": 8.281,
       "step": 8586
     },
     {
       "epoch": 82.0,
-      "grad_norm": 0.002602573251351714,
+      "grad_norm": 0.001334535307250917,
       "learning_rate": 9e-06,
-      "loss": 0.0003,
+      "loss": 0.0004,
       "step": 8692
     },
     {
       "epoch": 82.0,
-      "eval_LOCATION_f1": 0.8104575163398693,
+      "eval_LOCATION_f1": 0.7999999999999999,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.775,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7482014388489208,
+      "eval_LOCATION_precision": 0.7792207792207793,
+      "eval_LOCATION_recall": 0.821917808219178,
+      "eval_ORGANIZATION_f1": 0.7552447552447553,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.7027027027027027,
-      "eval_ORGANIZATION_recall": 0.8,
-      "eval_PERSON_f1": 0.9306930693069307,
+      "eval_ORGANIZATION_precision": 0.6923076923076923,
+      "eval_ORGANIZATION_recall": 0.8307692307692308,
+      "eval_PERSON_f1": 0.9250814332247558,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.9215686274509803,
-      "eval_PERSON_recall": 0.94,
-      "eval_QUANTITY_f1": 0.9180327868852458,
+      "eval_PERSON_precision": 0.9044585987261147,
+      "eval_PERSON_recall": 0.9466666666666667,
+      "eval_QUANTITY_f1": 0.6984126984126984,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.875,
-      "eval_QUANTITY_recall": 0.9655172413793104,
-      "eval_TIME_f1": 0.8823529411764706,
+      "eval_QUANTITY_precision": 0.6470588235294118,
+      "eval_QUANTITY_recall": 0.7586206896551724,
+      "eval_TIME_f1": 0.8888888888888888,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8823529411764706,
-      "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.30511462688446045,
-      "eval_overall_accuracy": 0.9680039379768643,
-      "eval_overall_f1": 0.8646408839779005,
-      "eval_overall_precision": 0.839142091152815,
-      "eval_overall_recall": 0.8917378917378918,
-      "eval_runtime": 0.9547,
-      "eval_samples_per_second": 195.864,
-      "eval_steps_per_second": 3.142,
+      "eval_TIME_precision": 0.8421052631578947,
+      "eval_TIME_recall": 0.9411764705882353,
+      "eval_loss": 0.31339409947395325,
+      "eval_overall_accuracy": 0.9670194437607679,
+      "eval_overall_f1": 0.8435374149659863,
+      "eval_overall_precision": 0.8072916666666666,
+      "eval_overall_recall": 0.8831908831908832,
+      "eval_runtime": 0.3744,
+      "eval_samples_per_second": 499.428,
+      "eval_steps_per_second": 8.012,
       "step": 8692
     },
     {
       "epoch": 83.0,
-      "grad_norm": 0.0006645359680987895,
+      "grad_norm": 0.0002743491786532104,
       "learning_rate": 8.500000000000002e-06,
-      "loss": 0.0002,
+      "loss": 0.0003,
       "step": 8798
     },
     {
       "epoch": 83.0,
-      "eval_LOCATION_f1": 0.8289473684210525,
+      "eval_LOCATION_f1": 0.8456375838926175,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7974683544303798,
+      "eval_LOCATION_precision": 0.8289473684210527,
       "eval_LOCATION_recall": 0.863013698630137,
-      "eval_ORGANIZATION_f1": 0.759124087591241,
+      "eval_ORGANIZATION_f1": 0.76056338028169,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.7222222222222222,
-      "eval_ORGANIZATION_recall": 0.8,
-      "eval_PERSON_f1": 0.9306930693069307,
+      "eval_ORGANIZATION_precision": 0.7012987012987013,
+      "eval_ORGANIZATION_recall": 0.8307692307692308,
+      "eval_PERSON_f1": 0.9377049180327869,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.9215686274509803,
-      "eval_PERSON_recall": 0.94,
-      "eval_QUANTITY_f1": 0.8852459016393444,
+      "eval_PERSON_precision": 0.9225806451612903,
+      "eval_PERSON_recall": 0.9533333333333334,
+      "eval_QUANTITY_f1": 0.7419354838709677,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.84375,
-      "eval_QUANTITY_recall": 0.9310344827586207,
-      "eval_TIME_f1": 0.8823529411764706,
+      "eval_QUANTITY_precision": 0.696969696969697,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8888888888888888,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8823529411764706,
-      "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.2994389235973358,
-      "eval_overall_accuracy": 0.9694806793010091,
-      "eval_overall_f1": 0.868238557558946,
-      "eval_overall_precision": 0.845945945945946,
-      "eval_overall_recall": 0.8917378917378918,
-      "eval_runtime": 0.9473,
-      "eval_samples_per_second": 197.393,
-      "eval_steps_per_second": 3.167,
+      "eval_TIME_precision": 0.8421052631578947,
+      "eval_TIME_recall": 0.9411764705882353,
+      "eval_loss": 0.3101036846637726,
+      "eval_overall_accuracy": 0.9699729264090573,
+      "eval_overall_f1": 0.863013698630137,
+      "eval_overall_precision": 0.8311345646437994,
+      "eval_overall_recall": 0.8974358974358975,
+      "eval_runtime": 0.3642,
+      "eval_samples_per_second": 513.432,
+      "eval_steps_per_second": 8.237,
       "step": 8798
     },
     {
       "epoch": 84.0,
-      "grad_norm": 0.00037489261012524366,
+      "grad_norm": 0.0002648239897098392,
       "learning_rate": 8.000000000000001e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 8904
     },
     {
       "epoch": 84.0,
-      "eval_LOCATION_f1": 0.8289473684210525,
+      "eval_LOCATION_f1": 0.832214765100671,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7974683544303798,
-      "eval_LOCATION_recall": 0.863013698630137,
-      "eval_ORGANIZATION_f1": 0.759124087591241,
+      "eval_LOCATION_precision": 0.8157894736842105,
+      "eval_LOCATION_recall": 0.8493150684931506,
+      "eval_ORGANIZATION_f1": 0.7659574468085107,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.7222222222222222,
-      "eval_ORGANIZATION_recall": 0.8,
-      "eval_PERSON_f1": 0.9306930693069307,
+      "eval_ORGANIZATION_precision": 0.7105263157894737,
+      "eval_ORGANIZATION_recall": 0.8307692307692308,
+      "eval_PERSON_f1": 0.9342105263157895,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.9215686274509803,
-      "eval_PERSON_recall": 0.94,
-      "eval_QUANTITY_f1": 0.9180327868852458,
+      "eval_PERSON_precision": 0.922077922077922,
+      "eval_PERSON_recall": 0.9466666666666667,
+      "eval_QUANTITY_f1": 0.7419354838709677,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.875,
-      "eval_QUANTITY_recall": 0.9655172413793104,
-      "eval_TIME_f1": 0.8823529411764706,
+      "eval_QUANTITY_precision": 0.696969696969697,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8888888888888888,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8823529411764706,
-      "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.30111053586006165,
-      "eval_overall_accuracy": 0.9697268028550332,
-      "eval_overall_f1": 0.8710124826629682,
-      "eval_overall_precision": 0.8486486486486486,
-      "eval_overall_recall": 0.8945868945868946,
-      "eval_runtime": 0.9548,
-      "eval_samples_per_second": 195.855,
-      "eval_steps_per_second": 3.142,
+      "eval_TIME_precision": 0.8421052631578947,
+      "eval_TIME_recall": 0.9411764705882353,
+      "eval_loss": 0.31526586413383484,
+      "eval_overall_accuracy": 0.9682500615308886,
+      "eval_overall_f1": 0.8598901098901098,
+      "eval_overall_precision": 0.830238726790451,
+      "eval_overall_recall": 0.8917378917378918,
+      "eval_runtime": 0.3643,
+      "eval_samples_per_second": 513.338,
+      "eval_steps_per_second": 8.235,
       "step": 8904
     },
     {
       "epoch": 85.0,
-      "grad_norm": 0.0004415301082190126,
+      "grad_norm": 0.001205015112645924,
       "learning_rate": 7.5e-06,
-      "loss": 0.0002,
+      "loss": 0.0005,
       "step": 9010
     },
     {
       "epoch": 85.0,
-      "eval_LOCATION_f1": 0.8133333333333332,
+      "eval_LOCATION_f1": 0.8243243243243243,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7922077922077922,
+      "eval_LOCATION_precision": 0.8133333333333334,
       "eval_LOCATION_recall": 0.8356164383561644,
-      "eval_ORGANIZATION_f1": 0.7555555555555555,
+      "eval_ORGANIZATION_f1": 0.7671232876712328,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.7285714285714285,
-      "eval_ORGANIZATION_recall": 0.7846153846153846,
-      "eval_PERSON_f1": 0.9306930693069307,
+      "eval_ORGANIZATION_precision": 0.691358024691358,
+      "eval_ORGANIZATION_recall": 0.8615384615384616,
+      "eval_PERSON_f1": 0.9281045751633987,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.9215686274509803,
-      "eval_PERSON_recall": 0.94,
-      "eval_QUANTITY_f1": 0.8852459016393444,
+      "eval_PERSON_precision": 0.9102564102564102,
+      "eval_PERSON_recall": 0.9466666666666667,
+      "eval_QUANTITY_f1": 0.7301587301587301,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.84375,
-      "eval_QUANTITY_recall": 0.9310344827586207,
-      "eval_TIME_f1": 0.8955223880597014,
+      "eval_QUANTITY_precision": 0.6764705882352942,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8732394366197184,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.9090909090909091,
-      "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.2962562143802643,
-      "eval_overall_accuracy": 0.9699729264090573,
-      "eval_overall_f1": 0.8659217877094971,
-      "eval_overall_precision": 0.8493150684931506,
-      "eval_overall_recall": 0.8831908831908832,
-      "eval_runtime": 0.9501,
-      "eval_samples_per_second": 196.831,
-      "eval_steps_per_second": 3.158,
+      "eval_TIME_precision": 0.8378378378378378,
+      "eval_TIME_recall": 0.9117647058823529,
+      "eval_loss": 0.3358004689216614,
+      "eval_overall_accuracy": 0.9655427024366232,
+      "eval_overall_f1": 0.8528610354223433,
+      "eval_overall_precision": 0.8172323759791122,
+      "eval_overall_recall": 0.8917378917378918,
+      "eval_runtime": 0.3625,
+      "eval_samples_per_second": 515.798,
+      "eval_steps_per_second": 8.275,
       "step": 9010
     },
     {
       "epoch": 86.0,
-      "grad_norm": 0.06156329810619354,
+      "grad_norm": 0.0002704145445022732,
       "learning_rate": 7.000000000000001e-06,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 9116
     },
     {
       "epoch": 86.0,
-      "eval_LOCATION_f1": 0.8157894736842105,
+      "eval_LOCATION_f1": 0.8187919463087249,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7848101265822784,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7482014388489208,
+      "eval_LOCATION_precision": 0.8026315789473685,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.7777777777777778,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.7027027027027027,
-      "eval_ORGANIZATION_recall": 0.8,
-      "eval_PERSON_f1": 0.9306930693069307,
+      "eval_ORGANIZATION_precision": 0.7088607594936709,
+      "eval_ORGANIZATION_recall": 0.8615384615384616,
+      "eval_PERSON_f1": 0.9281045751633987,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.9215686274509803,
-      "eval_PERSON_recall": 0.94,
-      "eval_QUANTITY_f1": 0.9180327868852458,
+      "eval_PERSON_precision": 0.9102564102564102,
+      "eval_PERSON_recall": 0.9466666666666667,
+      "eval_QUANTITY_f1": 0.7301587301587301,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.875,
-      "eval_QUANTITY_recall": 0.9655172413793104,
-      "eval_TIME_f1": 0.8823529411764706,
+      "eval_QUANTITY_precision": 0.6764705882352942,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8732394366197184,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8823529411764706,
-      "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.30722203850746155,
-      "eval_overall_accuracy": 0.9682500615308886,
-      "eval_overall_f1": 0.8658367911479945,
-      "eval_overall_precision": 0.8413978494623656,
+      "eval_TIME_precision": 0.8378378378378378,
+      "eval_TIME_recall": 0.9117647058823529,
+      "eval_loss": 0.33567553758621216,
+      "eval_overall_accuracy": 0.9665271966527197,
+      "eval_overall_f1": 0.8540245566166439,
+      "eval_overall_precision": 0.819371727748691,
       "eval_overall_recall": 0.8917378917378918,
-      "eval_runtime": 0.9331,
-      "eval_samples_per_second": 200.413,
-      "eval_steps_per_second": 3.215,
+      "eval_runtime": 0.3643,
+      "eval_samples_per_second": 513.342,
+      "eval_steps_per_second": 8.235,
       "step": 9116
     },
     {
       "epoch": 87.0,
-      "grad_norm": 0.0020842724479734898,
+      "grad_norm": 0.00025789992650970817,
       "learning_rate": 6.5000000000000004e-06,
-      "loss": 0.0004,
+      "loss": 0.0002,
       "step": 9222
     },
     {
       "epoch": 87.0,
-      "eval_LOCATION_f1": 0.8157894736842105,
+      "eval_LOCATION_f1": 0.8053691275167785,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7848101265822784,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7517730496453899,
+      "eval_LOCATION_precision": 0.7894736842105263,
+      "eval_LOCATION_recall": 0.821917808219178,
+      "eval_ORGANIZATION_f1": 0.7692307692307693,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6973684210526315,
-      "eval_ORGANIZATION_recall": 0.8153846153846154,
-      "eval_PERSON_f1": 0.9306930693069307,
+      "eval_ORGANIZATION_precision": 0.7051282051282052,
+      "eval_ORGANIZATION_recall": 0.8461538461538461,
+      "eval_PERSON_f1": 0.9377049180327869,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.9215686274509803,
-      "eval_PERSON_recall": 0.94,
-      "eval_QUANTITY_f1": 0.8852459016393444,
+      "eval_PERSON_precision": 0.9225806451612903,
+      "eval_PERSON_recall": 0.9533333333333334,
+      "eval_QUANTITY_f1": 0.7419354838709677,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.84375,
-      "eval_QUANTITY_recall": 0.9310344827586207,
-      "eval_TIME_f1": 0.8695652173913043,
+      "eval_QUANTITY_precision": 0.696969696969697,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8333333333333333,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8571428571428571,
+      "eval_TIME_precision": 0.7894736842105263,
       "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.31957679986953735,
-      "eval_overall_accuracy": 0.9670194437607679,
-      "eval_overall_f1": 0.862258953168044,
-      "eval_overall_precision": 0.8346666666666667,
-      "eval_overall_recall": 0.8917378917378918,
-      "eval_runtime": 0.9537,
-      "eval_samples_per_second": 196.069,
-      "eval_steps_per_second": 3.145,
+      "eval_loss": 0.3370627760887146,
+      "eval_overall_accuracy": 0.9655427024366232,
+      "eval_overall_f1": 0.8508891928864568,
+      "eval_overall_precision": 0.8184210526315789,
+      "eval_overall_recall": 0.886039886039886,
+      "eval_runtime": 0.3668,
+      "eval_samples_per_second": 509.752,
+      "eval_steps_per_second": 8.178,
       "step": 9222
     },
     {
       "epoch": 88.0,
-      "grad_norm": 0.0001907886180561036,
+      "grad_norm": 0.0001500611542724073,
       "learning_rate": 6e-06,
       "loss": 0.0001,
       "step": 9328
     },
     {
       "epoch": 88.0,
-      "eval_LOCATION_f1": 0.8211920529801324,
+      "eval_LOCATION_f1": 0.8053691275167785,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7948717948717948,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7464788732394366,
+      "eval_LOCATION_precision": 0.7894736842105263,
+      "eval_LOCATION_recall": 0.821917808219178,
+      "eval_ORGANIZATION_f1": 0.7692307692307693,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6883116883116883,
-      "eval_ORGANIZATION_recall": 0.8153846153846154,
+      "eval_ORGANIZATION_precision": 0.7051282051282052,
+      "eval_ORGANIZATION_recall": 0.8461538461538461,
       "eval_PERSON_f1": 0.9306930693069307,
       "eval_PERSON_number": 150,
       "eval_PERSON_precision": 0.9215686274509803,
       "eval_PERSON_recall": 0.94,
-      "eval_QUANTITY_f1": 0.8852459016393444,
+      "eval_QUANTITY_f1": 0.7419354838709677,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.84375,
-      "eval_QUANTITY_recall": 0.9310344827586207,
-      "eval_TIME_f1": 0.8695652173913043,
+      "eval_QUANTITY_precision": 0.696969696969697,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8333333333333333,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8571428571428571,
+      "eval_TIME_precision": 0.7894736842105263,
       "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.320029079914093,
-      "eval_overall_accuracy": 0.9672655673147921,
-      "eval_overall_f1": 0.862258953168044,
-      "eval_overall_precision": 0.8346666666666667,
-      "eval_overall_recall": 0.8917378917378918,
-      "eval_runtime": 0.9489,
-      "eval_samples_per_second": 197.066,
-      "eval_steps_per_second": 3.161,
+      "eval_loss": 0.3302779197692871,
+      "eval_overall_accuracy": 0.9648043317745508,
+      "eval_overall_f1": 0.8477366255144033,
+      "eval_overall_precision": 0.8174603174603174,
+      "eval_overall_recall": 0.8803418803418803,
+      "eval_runtime": 0.3639,
+      "eval_samples_per_second": 513.91,
+      "eval_steps_per_second": 8.245,
       "step": 9328
     },
     {
       "epoch": 89.0,
-      "grad_norm": 0.03503809869289398,
+      "grad_norm": 0.0003504405322019011,
       "learning_rate": 5.500000000000001e-06,
       "loss": 0.0001,
       "step": 9434
     },
     {
       "epoch": 89.0,
-      "eval_LOCATION_f1": 0.8157894736842105,
+      "eval_LOCATION_f1": 0.8053691275167785,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7848101265822784,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7536231884057971,
+      "eval_LOCATION_precision": 0.7894736842105263,
+      "eval_LOCATION_recall": 0.821917808219178,
+      "eval_ORGANIZATION_f1": 0.76056338028169,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.7123287671232876,
-      "eval_ORGANIZATION_recall": 0.8,
-      "eval_PERSON_f1": 0.9306930693069307,
+      "eval_ORGANIZATION_precision": 0.7012987012987013,
+      "eval_ORGANIZATION_recall": 0.8307692307692308,
+      "eval_PERSON_f1": 0.9276315789473684,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.9215686274509803,
+      "eval_PERSON_precision": 0.9155844155844156,
       "eval_PERSON_recall": 0.94,
-      "eval_QUANTITY_f1": 0.8852459016393444,
+      "eval_QUANTITY_f1": 0.7741935483870968,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.84375,
-      "eval_QUANTITY_recall": 0.9310344827586207,
-      "eval_TIME_f1": 0.8823529411764706,
+      "eval_QUANTITY_precision": 0.7272727272727273,
+      "eval_QUANTITY_recall": 0.8275862068965517,
+      "eval_TIME_f1": 0.8493150684931507,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8823529411764706,
-      "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.30977243185043335,
-      "eval_overall_accuracy": 0.9684961850849126,
-      "eval_overall_f1": 0.8642659279778393,
-      "eval_overall_precision": 0.8409703504043127,
-      "eval_overall_recall": 0.8888888888888888,
-      "eval_runtime": 0.9534,
-      "eval_samples_per_second": 196.146,
-      "eval_steps_per_second": 3.147,
+      "eval_TIME_precision": 0.7948717948717948,
+      "eval_TIME_recall": 0.9117647058823529,
+      "eval_loss": 0.33001190423965454,
+      "eval_overall_accuracy": 0.9665271966527197,
+      "eval_overall_f1": 0.8493150684931507,
+      "eval_overall_precision": 0.8179419525065963,
+      "eval_overall_recall": 0.8831908831908832,
+      "eval_runtime": 0.3594,
+      "eval_samples_per_second": 520.316,
+      "eval_steps_per_second": 8.347,
       "step": 9434
     },
     {
       "epoch": 90.0,
-      "grad_norm": 0.00032746559008955956,
+      "grad_norm": 0.00030385254649445415,
       "learning_rate": 5e-06,
       "loss": 0.0001,
       "step": 9540
     },
     {
       "epoch": 90.0,
-      "eval_LOCATION_f1": 0.8235294117647058,
+      "eval_LOCATION_f1": 0.7999999999999999,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.7875,
-      "eval_LOCATION_recall": 0.863013698630137,
-      "eval_ORGANIZATION_f1": 0.7482014388489208,
+      "eval_LOCATION_precision": 0.7792207792207793,
+      "eval_LOCATION_recall": 0.821917808219178,
+      "eval_ORGANIZATION_f1": 0.76056338028169,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.7027027027027027,
-      "eval_ORGANIZATION_recall": 0.8,
-      "eval_PERSON_f1": 0.9306930693069307,
+      "eval_ORGANIZATION_precision": 0.7012987012987013,
+      "eval_ORGANIZATION_recall": 0.8307692307692308,
+      "eval_PERSON_f1": 0.9342105263157895,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.9215686274509803,
-      "eval_PERSON_recall": 0.94,
-      "eval_QUANTITY_f1": 0.8852459016393444,
+      "eval_PERSON_precision": 0.922077922077922,
+      "eval_PERSON_recall": 0.9466666666666667,
+      "eval_QUANTITY_f1": 0.7419354838709677,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.84375,
-      "eval_QUANTITY_recall": 0.9310344827586207,
-      "eval_TIME_f1": 0.8695652173913043,
+      "eval_QUANTITY_precision": 0.696969696969697,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8493150684931507,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8571428571428571,
-      "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.32293304800987244,
-      "eval_overall_accuracy": 0.9677578144228403,
-      "eval_overall_f1": 0.863448275862069,
-      "eval_overall_precision": 0.8368983957219251,
-      "eval_overall_recall": 0.8917378917378918,
-      "eval_runtime": 0.947,
-      "eval_samples_per_second": 197.475,
-      "eval_steps_per_second": 3.168,
+      "eval_TIME_precision": 0.7948717948717948,
+      "eval_TIME_recall": 0.9117647058823529,
+      "eval_loss": 0.33552029728889465,
+      "eval_overall_accuracy": 0.9660349495446714,
+      "eval_overall_f1": 0.8481532147742818,
+      "eval_overall_precision": 0.8157894736842105,
+      "eval_overall_recall": 0.8831908831908832,
+      "eval_runtime": 0.3647,
+      "eval_samples_per_second": 512.732,
+      "eval_steps_per_second": 8.226,
       "step": 9540
     },
     {
       "epoch": 91.0,
-      "grad_norm": 0.0002206874341936782,
+      "grad_norm": 0.00028406584169715643,
       "learning_rate": 4.5e-06,
       "loss": 0.0002,
       "step": 9646
     },
     {
       "epoch": 91.0,
-      "eval_LOCATION_f1": 0.8104575163398693,
+      "eval_LOCATION_f1": 0.7999999999999999,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.775,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7428571428571429,
+      "eval_LOCATION_precision": 0.7792207792207793,
+      "eval_LOCATION_recall": 0.821917808219178,
+      "eval_ORGANIZATION_f1": 0.76056338028169,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6933333333333334,
-      "eval_ORGANIZATION_recall": 0.8,
-      "eval_PERSON_f1": 0.9306930693069307,
+      "eval_ORGANIZATION_precision": 0.7012987012987013,
+      "eval_ORGANIZATION_recall": 0.8307692307692308,
+      "eval_PERSON_f1": 0.9276315789473684,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.9215686274509803,
+      "eval_PERSON_precision": 0.9155844155844156,
       "eval_PERSON_recall": 0.94,
-      "eval_QUANTITY_f1": 0.8852459016393444,
+      "eval_QUANTITY_f1": 0.7419354838709677,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.84375,
-      "eval_QUANTITY_recall": 0.9310344827586207,
-      "eval_TIME_f1": 0.8695652173913043,
+      "eval_QUANTITY_precision": 0.696969696969697,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.861111111111111,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8571428571428571,
-      "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.3312826454639435,
-      "eval_overall_accuracy": 0.9660349495446714,
-      "eval_overall_f1": 0.859504132231405,
-      "eval_overall_precision": 0.832,
-      "eval_overall_recall": 0.8888888888888888,
-      "eval_runtime": 0.9562,
-      "eval_samples_per_second": 195.558,
-      "eval_steps_per_second": 3.137,
+      "eval_TIME_precision": 0.8157894736842105,
+      "eval_TIME_recall": 0.9117647058823529,
+      "eval_loss": 0.3344913423061371,
+      "eval_overall_accuracy": 0.9657888259906473,
+      "eval_overall_f1": 0.8465753424657535,
+      "eval_overall_precision": 0.8153034300791556,
+      "eval_overall_recall": 0.8803418803418803,
+      "eval_runtime": 0.3611,
+      "eval_samples_per_second": 517.806,
+      "eval_steps_per_second": 8.307,
       "step": 9646
     },
     {
       "epoch": 92.0,
-      "grad_norm": 0.0002490660990588367,
+      "grad_norm": 0.00017178994312416762,
       "learning_rate": 4.000000000000001e-06,
-      "loss": 0.0005,
+      "loss": 0.0006,
       "step": 9752
     },
     {
       "epoch": 92.0,
-      "eval_LOCATION_f1": 0.832214765100671,
+      "eval_LOCATION_f1": 0.8053691275167785,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.8157894736842105,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7746478873239436,
+      "eval_LOCATION_precision": 0.7894736842105263,
+      "eval_LOCATION_recall": 0.821917808219178,
+      "eval_ORGANIZATION_f1": 0.76056338028169,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.7142857142857143,
-      "eval_ORGANIZATION_recall": 0.8461538461538461,
+      "eval_ORGANIZATION_precision": 0.7012987012987013,
+      "eval_ORGANIZATION_recall": 0.8307692307692308,
       "eval_PERSON_f1": 0.9306930693069307,
       "eval_PERSON_number": 150,
       "eval_PERSON_precision": 0.9215686274509803,
       "eval_PERSON_recall": 0.94,
-      "eval_QUANTITY_f1": 0.8852459016393444,
+      "eval_QUANTITY_f1": 0.7301587301587301,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.84375,
-      "eval_QUANTITY_recall": 0.9310344827586207,
-      "eval_TIME_f1": 0.8955223880597014,
+      "eval_QUANTITY_precision": 0.6764705882352942,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.861111111111111,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.9090909090909091,
-      "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.3042062222957611,
-      "eval_overall_accuracy": 0.9682500615308886,
-      "eval_overall_f1": 0.8725761772853187,
-      "eval_overall_precision": 0.8490566037735849,
-      "eval_overall_recall": 0.8974358974358975,
-      "eval_runtime": 0.9513,
-      "eval_samples_per_second": 196.564,
-      "eval_steps_per_second": 3.153,
+      "eval_TIME_precision": 0.8157894736842105,
+      "eval_TIME_recall": 0.9117647058823529,
+      "eval_loss": 0.32351088523864746,
+      "eval_overall_accuracy": 0.9660349495446714,
+      "eval_overall_f1": 0.8477366255144033,
+      "eval_overall_precision": 0.8174603174603174,
+      "eval_overall_recall": 0.8803418803418803,
+      "eval_runtime": 0.3663,
+      "eval_samples_per_second": 510.443,
+      "eval_steps_per_second": 8.189,
       "step": 9752
     },
     {
       "epoch": 93.0,
-      "grad_norm": 0.0001448757539037615,
+      "grad_norm": 0.0001485623506596312,
       "learning_rate": 3.5000000000000004e-06,
       "loss": 0.0002,
       "step": 9858
     },
     {
       "epoch": 93.0,
-      "eval_LOCATION_f1": 0.8266666666666668,
+      "eval_LOCATION_f1": 0.8133333333333332,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.8051948051948052,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7714285714285715,
+      "eval_LOCATION_precision": 0.7922077922077922,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.76056338028169,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.72,
+      "eval_ORGANIZATION_precision": 0.7012987012987013,
       "eval_ORGANIZATION_recall": 0.8307692307692308,
-      "eval_PERSON_f1": 0.93687707641196,
+      "eval_PERSON_f1": 0.9306930693069307,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.9337748344370861,
+      "eval_PERSON_precision": 0.9215686274509803,
       "eval_PERSON_recall": 0.94,
-      "eval_QUANTITY_f1": 0.8852459016393444,
+      "eval_QUANTITY_f1": 0.6984126984126984,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.84375,
-      "eval_QUANTITY_recall": 0.9310344827586207,
-      "eval_TIME_f1": 0.8955223880597014,
+      "eval_QUANTITY_precision": 0.6470588235294118,
+      "eval_QUANTITY_recall": 0.7586206896551724,
+      "eval_TIME_f1": 0.8732394366197184,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.9090909090909091,
-      "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.29598936438560486,
-      "eval_overall_accuracy": 0.9699729264090573,
-      "eval_overall_f1": 0.8734353268428373,
-      "eval_overall_precision": 0.8532608695652174,
-      "eval_overall_recall": 0.8945868945868946,
-      "eval_runtime": 0.9615,
-      "eval_samples_per_second": 194.482,
-      "eval_steps_per_second": 3.12,
+      "eval_TIME_precision": 0.8378378378378378,
+      "eval_TIME_recall": 0.9117647058823529,
+      "eval_loss": 0.32250866293907166,
+      "eval_overall_accuracy": 0.9665271966527197,
+      "eval_overall_f1": 0.8477366255144033,
+      "eval_overall_precision": 0.8174603174603174,
+      "eval_overall_recall": 0.8803418803418803,
+      "eval_runtime": 0.364,
+      "eval_samples_per_second": 513.677,
+      "eval_steps_per_second": 8.241,
       "step": 9858
     },
     {
       "epoch": 94.0,
-      "grad_norm": 0.0024158929008990526,
+      "grad_norm": 0.004287095740437508,
       "learning_rate": 3e-06,
       "loss": 0.0001,
       "step": 9964
     },
     {
       "epoch": 94.0,
-      "eval_LOCATION_f1": 0.8104575163398693,
+      "eval_LOCATION_f1": 0.8187919463087249,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.775,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7375886524822695,
+      "eval_LOCATION_precision": 0.8026315789473685,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.76056338028169,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6842105263157895,
-      "eval_ORGANIZATION_recall": 0.8,
+      "eval_ORGANIZATION_precision": 0.7012987012987013,
+      "eval_ORGANIZATION_recall": 0.8307692307692308,
       "eval_PERSON_f1": 0.9306930693069307,
       "eval_PERSON_number": 150,
       "eval_PERSON_precision": 0.9215686274509803,
       "eval_PERSON_recall": 0.94,
-      "eval_QUANTITY_f1": 0.8852459016393444,
+      "eval_QUANTITY_f1": 0.7419354838709677,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.84375,
-      "eval_QUANTITY_recall": 0.9310344827586207,
-      "eval_TIME_f1": 0.8695652173913043,
+      "eval_QUANTITY_precision": 0.696969696969697,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8732394366197184,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8571428571428571,
-      "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.30419912934303284,
-      "eval_overall_accuracy": 0.9680039379768643,
-      "eval_overall_f1": 0.858321870701513,
-      "eval_overall_precision": 0.8297872340425532,
-      "eval_overall_recall": 0.8888888888888888,
-      "eval_runtime": 0.9456,
-      "eval_samples_per_second": 197.761,
-      "eval_steps_per_second": 3.173,
+      "eval_TIME_precision": 0.8378378378378378,
+      "eval_TIME_recall": 0.9117647058823529,
+      "eval_loss": 0.32277625799179077,
+      "eval_overall_accuracy": 0.9665271966527197,
+      "eval_overall_f1": 0.8528198074277855,
+      "eval_overall_precision": 0.824468085106383,
+      "eval_overall_recall": 0.8831908831908832,
+      "eval_runtime": 0.3724,
+      "eval_samples_per_second": 502.201,
+      "eval_steps_per_second": 8.057,
       "step": 9964
     },
     {
       "epoch": 95.0,
-      "grad_norm": 0.000135577647597529,
+      "grad_norm": 0.0006665958790108562,
       "learning_rate": 2.5e-06,
       "loss": 0.0001,
       "step": 10070
     },
     {
       "epoch": 95.0,
-      "eval_LOCATION_f1": 0.8104575163398693,
+      "eval_LOCATION_f1": 0.8133333333333332,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.775,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7323943661971831,
+      "eval_LOCATION_precision": 0.7922077922077922,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.76056338028169,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6753246753246753,
-      "eval_ORGANIZATION_recall": 0.8,
-      "eval_PERSON_f1": 0.9342105263157895,
+      "eval_ORGANIZATION_precision": 0.7012987012987013,
+      "eval_ORGANIZATION_recall": 0.8307692307692308,
+      "eval_PERSON_f1": 0.9306930693069307,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.922077922077922,
-      "eval_PERSON_recall": 0.9466666666666667,
-      "eval_QUANTITY_f1": 0.8852459016393444,
+      "eval_PERSON_precision": 0.9215686274509803,
+      "eval_PERSON_recall": 0.94,
+      "eval_QUANTITY_f1": 0.7419354838709677,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.84375,
-      "eval_QUANTITY_recall": 0.9310344827586207,
-      "eval_TIME_f1": 0.8695652173913043,
+      "eval_QUANTITY_precision": 0.696969696969697,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8732394366197184,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8571428571428571,
-      "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.31631138920783997,
-      "eval_overall_accuracy": 0.9670194437607679,
-      "eval_overall_f1": 0.858710562414266,
-      "eval_overall_precision": 0.828042328042328,
-      "eval_overall_recall": 0.8917378917378918,
-      "eval_runtime": 0.9579,
-      "eval_samples_per_second": 195.219,
-      "eval_steps_per_second": 3.132,
+      "eval_TIME_precision": 0.8378378378378378,
+      "eval_TIME_recall": 0.9117647058823529,
+      "eval_loss": 0.32652491331100464,
+      "eval_overall_accuracy": 0.9662810730986956,
+      "eval_overall_f1": 0.8516483516483517,
+      "eval_overall_precision": 0.8222811671087533,
+      "eval_overall_recall": 0.8831908831908832,
+      "eval_runtime": 0.3673,
+      "eval_samples_per_second": 509.172,
+      "eval_steps_per_second": 8.169,
       "step": 10070
     },
     {
       "epoch": 96.0,
-      "grad_norm": 0.00016227008018177003,
+      "grad_norm": 0.0001760148152243346,
       "learning_rate": 2.0000000000000003e-06,
       "loss": 0.0001,
       "step": 10176
     },
     {
       "epoch": 96.0,
-      "eval_LOCATION_f1": 0.8104575163398693,
+      "eval_LOCATION_f1": 0.8133333333333332,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.775,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7323943661971831,
+      "eval_LOCATION_precision": 0.7922077922077922,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.76056338028169,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6753246753246753,
-      "eval_ORGANIZATION_recall": 0.8,
-      "eval_PERSON_f1": 0.9342105263157895,
+      "eval_ORGANIZATION_precision": 0.7012987012987013,
+      "eval_ORGANIZATION_recall": 0.8307692307692308,
+      "eval_PERSON_f1": 0.9276315789473684,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.922077922077922,
-      "eval_PERSON_recall": 0.9466666666666667,
-      "eval_QUANTITY_f1": 0.8852459016393444,
+      "eval_PERSON_precision": 0.9155844155844156,
+      "eval_PERSON_recall": 0.94,
+      "eval_QUANTITY_f1": 0.7419354838709677,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.84375,
-      "eval_QUANTITY_recall": 0.9310344827586207,
-      "eval_TIME_f1": 0.8695652173913043,
+      "eval_QUANTITY_precision": 0.696969696969697,
+      "eval_QUANTITY_recall": 0.7931034482758621,
+      "eval_TIME_f1": 0.8732394366197184,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8571428571428571,
-      "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.31874725222587585,
-      "eval_overall_accuracy": 0.9670194437607679,
-      "eval_overall_f1": 0.858710562414266,
-      "eval_overall_precision": 0.828042328042328,
-      "eval_overall_recall": 0.8917378917378918,
-      "eval_runtime": 0.9455,
-      "eval_samples_per_second": 197.774,
-      "eval_steps_per_second": 3.173,
+      "eval_TIME_precision": 0.8378378378378378,
+      "eval_TIME_recall": 0.9117647058823529,
+      "eval_loss": 0.3282625079154968,
+      "eval_overall_accuracy": 0.9667733202067438,
+      "eval_overall_f1": 0.850480109739369,
+      "eval_overall_precision": 0.8201058201058201,
+      "eval_overall_recall": 0.8831908831908832,
+      "eval_runtime": 0.3759,
+      "eval_samples_per_second": 497.539,
+      "eval_steps_per_second": 7.982,
       "step": 10176
     },
     {
       "epoch": 97.0,
-      "grad_norm": 0.0001934995234478265,
+      "grad_norm": 0.00015518975851591676,
       "learning_rate": 1.5e-06,
       "loss": 0.0002,
       "step": 10282
     },
     {
       "epoch": 97.0,
-      "eval_LOCATION_f1": 0.8104575163398693,
+      "eval_LOCATION_f1": 0.8133333333333332,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.775,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7323943661971831,
+      "eval_LOCATION_precision": 0.7922077922077922,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.76056338028169,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6753246753246753,
-      "eval_ORGANIZATION_recall": 0.8,
-      "eval_PERSON_f1": 0.9306930693069307,
+      "eval_ORGANIZATION_precision": 0.7012987012987013,
+      "eval_ORGANIZATION_recall": 0.8307692307692308,
+      "eval_PERSON_f1": 0.9377049180327869,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.9215686274509803,
-      "eval_PERSON_recall": 0.94,
-      "eval_QUANTITY_f1": 0.8852459016393444,
+      "eval_PERSON_precision": 0.9225806451612903,
+      "eval_PERSON_recall": 0.9533333333333334,
+      "eval_QUANTITY_f1": 0.7096774193548386,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.84375,
-      "eval_QUANTITY_recall": 0.9310344827586207,
-      "eval_TIME_f1": 0.8695652173913043,
+      "eval_QUANTITY_precision": 0.6666666666666666,
+      "eval_QUANTITY_recall": 0.7586206896551724,
+      "eval_TIME_f1": 0.8732394366197184,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8571428571428571,
-      "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.3178471326828003,
-      "eval_overall_accuracy": 0.9665271966527197,
-      "eval_overall_f1": 0.8571428571428572,
-      "eval_overall_precision": 0.8275862068965517,
-      "eval_overall_recall": 0.8888888888888888,
-      "eval_runtime": 0.9621,
-      "eval_samples_per_second": 194.357,
-      "eval_steps_per_second": 3.118,
+      "eval_TIME_precision": 0.8378378378378378,
+      "eval_TIME_recall": 0.9117647058823529,
+      "eval_loss": 0.33287057280540466,
+      "eval_overall_accuracy": 0.9670194437607679,
+      "eval_overall_f1": 0.8520547945205479,
+      "eval_overall_precision": 0.820580474934037,
+      "eval_overall_recall": 0.886039886039886,
+      "eval_runtime": 0.3628,
+      "eval_samples_per_second": 515.48,
+      "eval_steps_per_second": 8.27,
       "step": 10282
     },
     {
       "epoch": 98.0,
-      "grad_norm": 0.00015261530643329024,
+      "grad_norm": 0.00012976452126167715,
       "learning_rate": 1.0000000000000002e-06,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 10388
     },
     {
       "epoch": 98.0,
-      "eval_LOCATION_f1": 0.8104575163398693,
+      "eval_LOCATION_f1": 0.8133333333333332,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.775,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7323943661971831,
+      "eval_LOCATION_precision": 0.7922077922077922,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.76056338028169,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6753246753246753,
-      "eval_ORGANIZATION_recall": 0.8,
-      "eval_PERSON_f1": 0.9306930693069307,
+      "eval_ORGANIZATION_precision": 0.7012987012987013,
+      "eval_ORGANIZATION_recall": 0.8307692307692308,
+      "eval_PERSON_f1": 0.9377049180327869,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.9215686274509803,
-      "eval_PERSON_recall": 0.94,
-      "eval_QUANTITY_f1": 0.8852459016393444,
+      "eval_PERSON_precision": 0.9225806451612903,
+      "eval_PERSON_recall": 0.9533333333333334,
+      "eval_QUANTITY_f1": 0.7096774193548386,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.84375,
-      "eval_QUANTITY_recall": 0.9310344827586207,
-      "eval_TIME_f1": 0.8695652173913043,
+      "eval_QUANTITY_precision": 0.6666666666666666,
+      "eval_QUANTITY_recall": 0.7586206896551724,
+      "eval_TIME_f1": 0.8732394366197184,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8571428571428571,
-      "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.3165254294872284,
-      "eval_overall_accuracy": 0.9667733202067438,
-      "eval_overall_f1": 0.8571428571428572,
-      "eval_overall_precision": 0.8275862068965517,
-      "eval_overall_recall": 0.8888888888888888,
-      "eval_runtime": 0.9525,
-      "eval_samples_per_second": 196.335,
-      "eval_steps_per_second": 3.15,
+      "eval_TIME_precision": 0.8378378378378378,
+      "eval_TIME_recall": 0.9117647058823529,
+      "eval_loss": 0.3322244882583618,
+      "eval_overall_accuracy": 0.9670194437607679,
+      "eval_overall_f1": 0.8520547945205479,
+      "eval_overall_precision": 0.820580474934037,
+      "eval_overall_recall": 0.886039886039886,
+      "eval_runtime": 0.3827,
+      "eval_samples_per_second": 488.673,
+      "eval_steps_per_second": 7.84,
       "step": 10388
     },
     {
       "epoch": 99.0,
-      "grad_norm": 0.00038772751577198505,
+      "grad_norm": 0.001286118756979704,
       "learning_rate": 5.000000000000001e-07,
       "loss": 0.0001,
       "step": 10494
     },
     {
       "epoch": 99.0,
-      "eval_LOCATION_f1": 0.8104575163398693,
+      "eval_LOCATION_f1": 0.8133333333333332,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.775,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7323943661971831,
+      "eval_LOCATION_precision": 0.7922077922077922,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.76056338028169,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6753246753246753,
-      "eval_ORGANIZATION_recall": 0.8,
-      "eval_PERSON_f1": 0.9306930693069307,
+      "eval_ORGANIZATION_precision": 0.7012987012987013,
+      "eval_ORGANIZATION_recall": 0.8307692307692308,
+      "eval_PERSON_f1": 0.9377049180327869,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.9215686274509803,
-      "eval_PERSON_recall": 0.94,
-      "eval_QUANTITY_f1": 0.8852459016393444,
+      "eval_PERSON_precision": 0.9225806451612903,
+      "eval_PERSON_recall": 0.9533333333333334,
+      "eval_QUANTITY_f1": 0.7096774193548386,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.84375,
-      "eval_QUANTITY_recall": 0.9310344827586207,
-      "eval_TIME_f1": 0.8695652173913043,
+      "eval_QUANTITY_precision": 0.6666666666666666,
+      "eval_QUANTITY_recall": 0.7586206896551724,
+      "eval_TIME_f1": 0.8732394366197184,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8571428571428571,
-      "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.3173261284828186,
-      "eval_overall_accuracy": 0.9667733202067438,
-      "eval_overall_f1": 0.8571428571428572,
-      "eval_overall_precision": 0.8275862068965517,
-      "eval_overall_recall": 0.8888888888888888,
-      "eval_runtime": 0.9532,
-      "eval_samples_per_second": 196.185,
-      "eval_steps_per_second": 3.147,
+      "eval_TIME_precision": 0.8378378378378378,
+      "eval_TIME_recall": 0.9117647058823529,
+      "eval_loss": 0.33244797587394714,
+      "eval_overall_accuracy": 0.9670194437607679,
+      "eval_overall_f1": 0.8520547945205479,
+      "eval_overall_precision": 0.820580474934037,
+      "eval_overall_recall": 0.886039886039886,
+      "eval_runtime": 0.3638,
+      "eval_samples_per_second": 513.957,
+      "eval_steps_per_second": 8.245,
       "step": 10494
     },
     {
       "epoch": 100.0,
-      "grad_norm": 0.0001833343703765422,
+      "grad_norm": 0.00014845983241684735,
       "learning_rate": 0.0,
-      "loss": 0.0001,
+      "loss": 0.0003,
       "step": 10600
     },
     {
       "epoch": 100.0,
-      "eval_LOCATION_f1": 0.8104575163398693,
+      "eval_LOCATION_f1": 0.8133333333333332,
       "eval_LOCATION_number": 73,
-      "eval_LOCATION_precision": 0.775,
-      "eval_LOCATION_recall": 0.8493150684931506,
-      "eval_ORGANIZATION_f1": 0.7323943661971831,
+      "eval_LOCATION_precision": 0.7922077922077922,
+      "eval_LOCATION_recall": 0.8356164383561644,
+      "eval_ORGANIZATION_f1": 0.76056338028169,
       "eval_ORGANIZATION_number": 65,
-      "eval_ORGANIZATION_precision": 0.6753246753246753,
-      "eval_ORGANIZATION_recall": 0.8,
-      "eval_PERSON_f1": 0.9306930693069307,
+      "eval_ORGANIZATION_precision": 0.7012987012987013,
+      "eval_ORGANIZATION_recall": 0.8307692307692308,
+      "eval_PERSON_f1": 0.9377049180327869,
       "eval_PERSON_number": 150,
-      "eval_PERSON_precision": 0.9215686274509803,
-      "eval_PERSON_recall": 0.94,
-      "eval_QUANTITY_f1": 0.8852459016393444,
+      "eval_PERSON_precision": 0.9225806451612903,
+      "eval_PERSON_recall": 0.9533333333333334,
+      "eval_QUANTITY_f1": 0.7096774193548386,
       "eval_QUANTITY_number": 29,
-      "eval_QUANTITY_precision": 0.84375,
-      "eval_QUANTITY_recall": 0.9310344827586207,
-      "eval_TIME_f1": 0.8695652173913043,
+      "eval_QUANTITY_precision": 0.6666666666666666,
+      "eval_QUANTITY_recall": 0.7586206896551724,
+      "eval_TIME_f1": 0.8732394366197184,
       "eval_TIME_number": 34,
-      "eval_TIME_precision": 0.8571428571428571,
-      "eval_TIME_recall": 0.8823529411764706,
-      "eval_loss": 0.3181513547897339,
-      "eval_overall_accuracy": 0.9665271966527197,
-      "eval_overall_f1": 0.8571428571428572,
-      "eval_overall_precision": 0.8275862068965517,
-      "eval_overall_recall": 0.8888888888888888,
-      "eval_runtime": 0.9563,
-      "eval_samples_per_second": 195.536,
-      "eval_steps_per_second": 3.137,
+      "eval_TIME_precision": 0.8378378378378378,
+      "eval_TIME_recall": 0.9117647058823529,
+      "eval_loss": 0.3325321078300476,
+      "eval_overall_accuracy": 0.9670194437607679,
+      "eval_overall_f1": 0.8520547945205479,
+      "eval_overall_precision": 0.820580474934037,
+      "eval_overall_recall": 0.886039886039886,
+      "eval_runtime": 0.3655,
+      "eval_samples_per_second": 511.566,
+      "eval_steps_per_second": 8.207,
       "step": 10600
     },
     {
       "epoch": 100.0,
       "step": 10600,
       "total_flos": 4520246411442048.0,
-      "train_loss": 0.008314491887576879,
-      "train_runtime": 2597.8016,
-      "train_samples_per_second": 64.978,
-      "train_steps_per_second": 4.08
+      "train_loss": 0.008715969840952038,
+      "train_runtime": 962.9354,
+      "train_samples_per_second": 175.297,
+      "train_steps_per_second": 11.008
     }
   ],
   "logging_steps": 500,