diff --git "a/trainer_state.json" "b/trainer_state.json" new file mode 100644--- /dev/null +++ "b/trainer_state.json" @@ -0,0 +1,3930 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 100.0, + "eval_steps": 500, + "global_step": 10600, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 1.0, + "grad_norm": 1.875795841217041, + "learning_rate": 4.9500000000000004e-05, + "loss": 0.9784, + "step": 106 + }, + { + "epoch": 1.0, + "eval_LOCATION_f1": 0.0, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.0, + "eval_LOCATION_recall": 0.0, + "eval_ORGANIZATION_f1": 0.0, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.0, + "eval_ORGANIZATION_recall": 0.0, + "eval_PERSON_f1": 0.024390243902439025, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.14285714285714285, + "eval_PERSON_recall": 0.013333333333333334, + "eval_QUANTITY_f1": 0.0, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.0, + "eval_QUANTITY_recall": 0.0, + "eval_TIME_f1": 0.0, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.0, + "eval_TIME_recall": 0.0, + "eval_loss": 0.5909550189971924, + "eval_overall_accuracy": 0.8392813192222496, + "eval_overall_f1": 0.010638297872340425, + "eval_overall_precision": 0.08, + "eval_overall_recall": 0.005698005698005698, + "eval_runtime": 0.3532, + "eval_samples_per_second": 529.423, + "eval_steps_per_second": 8.493, + "step": 106 + }, + { + "epoch": 2.0, + "grad_norm": 1.4692219495773315, + "learning_rate": 4.9e-05, + "loss": 0.4893, + "step": 212 + }, + { + "epoch": 2.0, + "eval_LOCATION_f1": 0.4444444444444445, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.5909090909090909, + "eval_LOCATION_recall": 0.3561643835616438, + "eval_ORGANIZATION_f1": 0.12328767123287672, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.1111111111111111, + "eval_ORGANIZATION_recall": 0.13846153846153847, + "eval_PERSON_f1": 0.6703910614525139, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.5769230769230769, + "eval_PERSON_recall": 0.8, + "eval_QUANTITY_f1": 0.2857142857142857, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.2962962962962963, + "eval_QUANTITY_recall": 0.27586206896551724, + "eval_TIME_f1": 0.6301369863013699, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.5897435897435898, + "eval_TIME_recall": 0.6764705882352942, + "eval_loss": 0.35233741998672485, + "eval_overall_accuracy": 0.9020428254984002, + "eval_overall_f1": 0.496, + "eval_overall_precision": 0.46616541353383456, + "eval_overall_recall": 0.5299145299145299, + "eval_runtime": 0.3497, + "eval_samples_per_second": 534.772, + "eval_steps_per_second": 8.579, + "step": 212 + }, + { + "epoch": 3.0, + "grad_norm": 0.853384792804718, + "learning_rate": 4.85e-05, + "loss": 0.2876, + "step": 318 + }, + { + "epoch": 3.0, + "eval_LOCATION_f1": 0.6242038216560509, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.5833333333333334, + "eval_LOCATION_recall": 0.6712328767123288, + "eval_ORGANIZATION_f1": 0.4090909090909091, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.40298507462686567, + "eval_ORGANIZATION_recall": 0.4153846153846154, + "eval_PERSON_f1": 0.8500000000000001, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8, + "eval_PERSON_recall": 0.9066666666666666, + "eval_QUANTITY_f1": 0.6268656716417911, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.5526315789473685, + "eval_QUANTITY_recall": 0.7241379310344828, + "eval_TIME_f1": 0.8571428571428571, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8333333333333334, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.19235990941524506, + "eval_overall_accuracy": 0.9369923701698253, + "eval_overall_f1": 0.7050938337801608, + "eval_overall_precision": 0.6658227848101266, + "eval_overall_recall": 0.7492877492877493, + "eval_runtime": 0.3493, + "eval_samples_per_second": 535.38, + "eval_steps_per_second": 8.589, + "step": 318 + }, + { + "epoch": 4.0, + "grad_norm": 0.8233245015144348, + "learning_rate": 4.8e-05, + "loss": 0.2003, + "step": 424 + }, + { + "epoch": 4.0, + "eval_LOCATION_f1": 0.7295597484276729, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.6744186046511628, + "eval_LOCATION_recall": 0.7945205479452054, + "eval_ORGANIZATION_f1": 0.6144578313253013, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.504950495049505, + "eval_ORGANIZATION_recall": 0.7846153846153846, + "eval_PERSON_f1": 0.8698412698412699, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8303030303030303, + "eval_PERSON_recall": 0.9133333333333333, + "eval_QUANTITY_f1": 0.7042253521126761, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.5952380952380952, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.8493150684931507, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7948717948717948, + "eval_TIME_recall": 0.9117647058823529, + "eval_loss": 0.16705787181854248, + "eval_overall_accuracy": 0.9414225941422594, + "eval_overall_f1": 0.7704081632653061, + "eval_overall_precision": 0.6974595842956121, + "eval_overall_recall": 0.8603988603988604, + "eval_runtime": 0.3492, + "eval_samples_per_second": 535.517, + "eval_steps_per_second": 8.591, + "step": 424 + }, + { + "epoch": 5.0, + "grad_norm": 1.155442476272583, + "learning_rate": 4.75e-05, + "loss": 0.1732, + "step": 530 + }, + { + "epoch": 5.0, + "eval_LOCATION_f1": 0.7672955974842768, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7093023255813954, + "eval_LOCATION_recall": 0.8356164383561644, + "eval_ORGANIZATION_f1": 0.6011560693641619, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.48148148148148145, + "eval_ORGANIZATION_recall": 0.8, + "eval_PERSON_f1": 0.8690095846645367, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8343558282208589, + "eval_PERSON_recall": 0.9066666666666666, + "eval_QUANTITY_f1": 0.7272727272727273, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6486486486486487, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.7594936708860759, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.6666666666666666, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.15311911702156067, + "eval_overall_accuracy": 0.94683731233079, + "eval_overall_f1": 0.7670886075949367, + "eval_overall_precision": 0.6902050113895216, + "eval_overall_recall": 0.8632478632478633, + "eval_runtime": 0.3497, + "eval_samples_per_second": 534.75, + "eval_steps_per_second": 8.579, + "step": 530 + }, + { + "epoch": 6.0, + "grad_norm": 1.4915152788162231, + "learning_rate": 4.7e-05, + "loss": 0.1556, + "step": 636 + }, + { + "epoch": 6.0, + "eval_LOCATION_f1": 0.7560975609756097, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.6813186813186813, + "eval_LOCATION_recall": 0.8493150684931506, + "eval_ORGANIZATION_f1": 0.631578947368421, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.5094339622641509, + "eval_ORGANIZATION_recall": 0.8307692307692308, + "eval_PERSON_f1": 0.8789808917197452, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8414634146341463, + "eval_PERSON_recall": 0.92, + "eval_QUANTITY_f1": 0.631578947368421, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.5106382978723404, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.6829268292682927, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.5833333333333334, + "eval_TIME_recall": 0.8235294117647058, + "eval_loss": 0.17121362686157227, + "eval_overall_accuracy": 0.9396997292640906, + "eval_overall_f1": 0.758364312267658, + "eval_overall_precision": 0.6710526315789473, + "eval_overall_recall": 0.8717948717948718, + "eval_runtime": 0.3488, + "eval_samples_per_second": 536.19, + "eval_steps_per_second": 8.602, + "step": 636 + }, + { + "epoch": 7.0, + "grad_norm": 0.9724661111831665, + "learning_rate": 4.6500000000000005e-05, + "loss": 0.1437, + "step": 742 + }, + { + "epoch": 7.0, + "eval_LOCATION_f1": 0.7924528301886793, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7325581395348837, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.6623376623376623, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.5730337078651685, + "eval_ORGANIZATION_recall": 0.7846153846153846, + "eval_PERSON_f1": 0.8761904761904761, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8363636363636363, + "eval_PERSON_recall": 0.92, + "eval_QUANTITY_f1": 0.6956521739130435, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8219178082191781, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7692307692307693, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.14172205328941345, + "eval_overall_accuracy": 0.9534826482894413, + "eval_overall_f1": 0.7948051948051948, + "eval_overall_precision": 0.7303102625298329, + "eval_overall_recall": 0.8717948717948718, + "eval_runtime": 0.3495, + "eval_samples_per_second": 534.982, + "eval_steps_per_second": 8.583, + "step": 742 + }, + { + "epoch": 8.0, + "grad_norm": 0.9051221609115601, + "learning_rate": 4.600000000000001e-05, + "loss": 0.1338, + "step": 848 + }, + { + "epoch": 8.0, + "eval_LOCATION_f1": 0.7590361445783131, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.6774193548387096, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.6883116883116883, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.5955056179775281, + "eval_ORGANIZATION_recall": 0.8153846153846154, + "eval_PERSON_f1": 0.887459807073955, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8571428571428571, + "eval_PERSON_recall": 0.92, + "eval_QUANTITY_f1": 0.6956521739130435, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.7532467532467532, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.6744186046511628, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.14630907773971558, + "eval_overall_accuracy": 0.9529904011813931, + "eval_overall_f1": 0.7902187902187902, + "eval_overall_precision": 0.7206572769953051, + "eval_overall_recall": 0.8746438746438746, + "eval_runtime": 0.3496, + "eval_samples_per_second": 534.867, + "eval_steps_per_second": 8.581, + "step": 848 + }, + { + "epoch": 9.0, + "grad_norm": 1.2815150022506714, + "learning_rate": 4.55e-05, + "loss": 0.1237, + "step": 954 + }, + { + "epoch": 9.0, + "eval_LOCATION_f1": 0.8079470198675496, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.782051282051282, + "eval_LOCATION_recall": 0.8356164383561644, + "eval_ORGANIZATION_f1": 0.670886075949367, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.5698924731182796, + "eval_ORGANIZATION_recall": 0.8153846153846154, + "eval_PERSON_f1": 0.8932038834951457, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8679245283018868, + "eval_PERSON_recall": 0.92, + "eval_QUANTITY_f1": 0.716417910447761, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.631578947368421, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8219178082191781, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7692307692307693, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.1243952140212059, + "eval_overall_accuracy": 0.9571745015998031, + "eval_overall_f1": 0.8073878627968336, + "eval_overall_precision": 0.7518427518427518, + "eval_overall_recall": 0.8717948717948718, + "eval_runtime": 0.3487, + "eval_samples_per_second": 536.261, + "eval_steps_per_second": 8.603, + "step": 954 + }, + { + "epoch": 10.0, + "grad_norm": 0.8149334788322449, + "learning_rate": 4.5e-05, + "loss": 0.1189, + "step": 1060 + }, + { + "epoch": 10.0, + "eval_LOCATION_f1": 0.8026315789473685, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7721518987341772, + "eval_LOCATION_recall": 0.8356164383561644, + "eval_ORGANIZATION_f1": 0.6751592356687899, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.5760869565217391, + "eval_ORGANIZATION_recall": 0.8153846153846154, + "eval_PERSON_f1": 0.8753993610223643, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8404907975460123, + "eval_PERSON_recall": 0.9133333333333333, + "eval_QUANTITY_f1": 0.716417910447761, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.631578947368421, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8333333333333333, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7894736842105263, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.13834965229034424, + "eval_overall_accuracy": 0.9532365247354172, + "eval_overall_f1": 0.8015768725361366, + "eval_overall_precision": 0.7439024390243902, + "eval_overall_recall": 0.8689458689458689, + "eval_runtime": 0.3499, + "eval_samples_per_second": 534.481, + "eval_steps_per_second": 8.575, + "step": 1060 + }, + { + "epoch": 11.0, + "grad_norm": 1.1204177141189575, + "learning_rate": 4.4500000000000004e-05, + "loss": 0.1139, + "step": 1166 + }, + { + "epoch": 11.0, + "eval_LOCATION_f1": 0.8235294117647058, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7875, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.6620689655172414, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6, + "eval_ORGANIZATION_recall": 0.7384615384615385, + "eval_PERSON_f1": 0.8996763754045306, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8742138364779874, + "eval_PERSON_recall": 0.9266666666666666, + "eval_QUANTITY_f1": 0.6666666666666666, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.575, + "eval_QUANTITY_recall": 0.7931034482758621, + "eval_TIME_f1": 0.7837837837837837, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.725, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.14040923118591309, + "eval_overall_accuracy": 0.9554516367216342, + "eval_overall_f1": 0.8053333333333333, + "eval_overall_precision": 0.7568922305764411, + "eval_overall_recall": 0.8603988603988604, + "eval_runtime": 0.351, + "eval_samples_per_second": 532.751, + "eval_steps_per_second": 8.547, + "step": 1166 + }, + { + "epoch": 12.0, + "grad_norm": 1.1477428674697876, + "learning_rate": 4.4000000000000006e-05, + "loss": 0.1056, + "step": 1272 + }, + { + "epoch": 12.0, + "eval_LOCATION_f1": 0.8076923076923077, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7590361445783133, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.7058823529411765, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6136363636363636, + "eval_ORGANIZATION_recall": 0.8307692307692308, + "eval_PERSON_f1": 0.888178913738019, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.852760736196319, + "eval_PERSON_recall": 0.9266666666666666, + "eval_QUANTITY_f1": 0.7272727272727273, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6486486486486487, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8055555555555555, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7631578947368421, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.13301722705364227, + "eval_overall_accuracy": 0.9579128722618755, + "eval_overall_f1": 0.8131578947368421, + "eval_overall_precision": 0.7555012224938875, + "eval_overall_recall": 0.8803418803418803, + "eval_runtime": 0.3491, + "eval_samples_per_second": 535.587, + "eval_steps_per_second": 8.592, + "step": 1272 + }, + { + "epoch": 13.0, + "grad_norm": 0.6786032319068909, + "learning_rate": 4.35e-05, + "loss": 0.1032, + "step": 1378 + }, + { + "epoch": 13.0, + "eval_LOCATION_f1": 0.8211920529801324, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7948717948717948, + "eval_LOCATION_recall": 0.8493150684931506, + "eval_ORGANIZATION_f1": 0.675, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.5684210526315789, + "eval_ORGANIZATION_recall": 0.8307692307692308, + "eval_PERSON_f1": 0.8945686900958466, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8588957055214724, + "eval_PERSON_recall": 0.9333333333333333, + "eval_QUANTITY_f1": 0.7058823529411765, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6153846153846154, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8285714285714286, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8055555555555556, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.1293208748102188, + "eval_overall_accuracy": 0.956928378045779, + "eval_overall_f1": 0.811023622047244, + "eval_overall_precision": 0.7518248175182481, + "eval_overall_recall": 0.8803418803418803, + "eval_runtime": 0.3485, + "eval_samples_per_second": 536.51, + "eval_steps_per_second": 8.607, + "step": 1378 + }, + { + "epoch": 14.0, + "grad_norm": 1.0159380435943604, + "learning_rate": 4.3e-05, + "loss": 0.1013, + "step": 1484 + }, + { + "epoch": 14.0, + "eval_LOCATION_f1": 0.7973856209150326, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7625, + "eval_LOCATION_recall": 0.8356164383561644, + "eval_ORGANIZATION_f1": 0.6666666666666666, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6301369863013698, + "eval_ORGANIZATION_recall": 0.7076923076923077, + "eval_PERSON_f1": 0.8717948717948718, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8395061728395061, + "eval_PERSON_recall": 0.9066666666666666, + "eval_QUANTITY_f1": 0.746268656716418, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6578947368421053, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.7777777777777778, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7368421052631579, + "eval_TIME_recall": 0.8235294117647058, + "eval_loss": 0.14084841310977936, + "eval_overall_accuracy": 0.9564361309377307, + "eval_overall_f1": 0.7978436657681941, + "eval_overall_precision": 0.7570332480818415, + "eval_overall_recall": 0.8433048433048433, + "eval_runtime": 0.3493, + "eval_samples_per_second": 535.324, + "eval_steps_per_second": 8.588, + "step": 1484 + }, + { + "epoch": 15.0, + "grad_norm": 1.4810669422149658, + "learning_rate": 4.25e-05, + "loss": 0.0926, + "step": 1590 + }, + { + "epoch": 15.0, + "eval_LOCATION_f1": 0.8211920529801324, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7948717948717948, + "eval_LOCATION_recall": 0.8493150684931506, + "eval_ORGANIZATION_f1": 0.6794871794871795, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.5824175824175825, + "eval_ORGANIZATION_recall": 0.8153846153846154, + "eval_PERSON_f1": 0.8945686900958466, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8588957055214724, + "eval_PERSON_recall": 0.9333333333333333, + "eval_QUANTITY_f1": 0.7647058823529411, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6666666666666666, + "eval_QUANTITY_recall": 0.896551724137931, + "eval_TIME_f1": 0.8285714285714286, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8055555555555556, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.13149809837341309, + "eval_overall_accuracy": 0.956928378045779, + "eval_overall_f1": 0.8179419525065963, + "eval_overall_precision": 0.7616707616707616, + "eval_overall_recall": 0.8831908831908832, + "eval_runtime": 0.35, + "eval_samples_per_second": 534.326, + "eval_steps_per_second": 8.572, + "step": 1590 + }, + { + "epoch": 16.0, + "grad_norm": 1.0576220750808716, + "learning_rate": 4.2e-05, + "loss": 0.0913, + "step": 1696 + }, + { + "epoch": 16.0, + "eval_LOCATION_f1": 0.8157894736842105, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7848101265822784, + "eval_LOCATION_recall": 0.8493150684931506, + "eval_ORGANIZATION_f1": 0.7083333333333334, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6455696202531646, + "eval_ORGANIZATION_recall": 0.7846153846153846, + "eval_PERSON_f1": 0.888888888888889, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8484848484848485, + "eval_PERSON_recall": 0.9333333333333333, + "eval_QUANTITY_f1": 0.6575342465753425, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.5454545454545454, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.767123287671233, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.717948717948718, + "eval_TIME_recall": 0.8235294117647058, + "eval_loss": 0.14829330146312714, + "eval_overall_accuracy": 0.9554516367216342, + "eval_overall_f1": 0.8058124174372523, + "eval_overall_precision": 0.7512315270935961, + "eval_overall_recall": 0.8689458689458689, + "eval_runtime": 0.349, + "eval_samples_per_second": 535.785, + "eval_steps_per_second": 8.595, + "step": 1696 + }, + { + "epoch": 17.0, + "grad_norm": 0.3293265998363495, + "learning_rate": 4.15e-05, + "loss": 0.0888, + "step": 1802 + }, + { + "epoch": 17.0, + "eval_LOCATION_f1": 0.794701986754967, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7692307692307693, + "eval_LOCATION_recall": 0.821917808219178, + "eval_ORGANIZATION_f1": 0.7210884353741497, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6463414634146342, + "eval_ORGANIZATION_recall": 0.8153846153846154, + "eval_PERSON_f1": 0.8917197452229298, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8536585365853658, + "eval_PERSON_recall": 0.9333333333333333, + "eval_QUANTITY_f1": 0.8253968253968255, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7647058823529411, + "eval_QUANTITY_recall": 0.896551724137931, + "eval_TIME_f1": 0.7567567567567567, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7, + "eval_TIME_recall": 0.8235294117647058, + "eval_loss": 0.13628779351711273, + "eval_overall_accuracy": 0.9584051193699237, + "eval_overall_f1": 0.8197596795727637, + "eval_overall_precision": 0.7713567839195979, + "eval_overall_recall": 0.8746438746438746, + "eval_runtime": 0.3508, + "eval_samples_per_second": 533.043, + "eval_steps_per_second": 8.551, + "step": 1802 + }, + { + "epoch": 18.0, + "grad_norm": 0.991680920124054, + "learning_rate": 4.1e-05, + "loss": 0.0861, + "step": 1908 + }, + { + "epoch": 18.0, + "eval_LOCATION_f1": 0.7692307692307693, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7228915662650602, + "eval_LOCATION_recall": 0.821917808219178, + "eval_ORGANIZATION_f1": 0.7114093959731543, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6309523809523809, + "eval_ORGANIZATION_recall": 0.8153846153846154, + "eval_PERSON_f1": 0.8980891719745222, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8597560975609756, + "eval_PERSON_recall": 0.94, + "eval_QUANTITY_f1": 0.7761194029850746, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6842105263157895, + "eval_QUANTITY_recall": 0.896551724137931, + "eval_TIME_f1": 0.7777777777777778, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7368421052631579, + "eval_TIME_recall": 0.8235294117647058, + "eval_loss": 0.1477837860584259, + "eval_overall_accuracy": 0.9559438838296825, + "eval_overall_f1": 0.8126649076517152, + "eval_overall_precision": 0.7567567567567568, + "eval_overall_recall": 0.8774928774928775, + "eval_runtime": 0.3475, + "eval_samples_per_second": 538.055, + "eval_steps_per_second": 8.632, + "step": 1908 + }, + { + "epoch": 19.0, + "grad_norm": 1.0064384937286377, + "learning_rate": 4.05e-05, + "loss": 0.0862, + "step": 2014 + }, + { + "epoch": 19.0, + "eval_LOCATION_f1": 0.8187919463087249, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8026315789473685, + "eval_LOCATION_recall": 0.8356164383561644, + "eval_ORGANIZATION_f1": 0.7210884353741497, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6463414634146342, + "eval_ORGANIZATION_recall": 0.8153846153846154, + "eval_PERSON_f1": 0.887459807073955, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8571428571428571, + "eval_PERSON_recall": 0.92, + "eval_QUANTITY_f1": 0.8125, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7428571428571429, + "eval_QUANTITY_recall": 0.896551724137931, + "eval_TIME_f1": 0.767123287671233, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.717948717948718, + "eval_TIME_recall": 0.8235294117647058, + "eval_loss": 0.12853632867336273, + "eval_overall_accuracy": 0.961112478464189, + "eval_overall_f1": 0.8225806451612903, + "eval_overall_precision": 0.7786259541984732, + "eval_overall_recall": 0.8717948717948718, + "eval_runtime": 0.3498, + "eval_samples_per_second": 534.521, + "eval_steps_per_second": 8.575, + "step": 2014 + }, + { + "epoch": 20.0, + "grad_norm": 0.54794842004776, + "learning_rate": 4e-05, + "loss": 0.0787, + "step": 2120 + }, + { + "epoch": 20.0, + "eval_LOCATION_f1": 0.8026315789473685, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7721518987341772, + "eval_LOCATION_recall": 0.8356164383561644, + "eval_ORGANIZATION_f1": 0.7027027027027027, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6265060240963856, + "eval_ORGANIZATION_recall": 0.8, + "eval_PERSON_f1": 0.8910256410256411, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8580246913580247, + "eval_PERSON_recall": 0.9266666666666666, + "eval_QUANTITY_f1": 0.7246376811594203, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.625, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.8285714285714286, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8055555555555556, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.1396997570991516, + "eval_overall_accuracy": 0.956928378045779, + "eval_overall_f1": 0.8149134487350199, + "eval_overall_precision": 0.765, + "eval_overall_recall": 0.8717948717948718, + "eval_runtime": 0.3498, + "eval_samples_per_second": 534.561, + "eval_steps_per_second": 8.576, + "step": 2120 + }, + { + "epoch": 21.0, + "grad_norm": 1.174849510192871, + "learning_rate": 3.9500000000000005e-05, + "loss": 0.0787, + "step": 2226 + }, + { + "epoch": 21.0, + "eval_LOCATION_f1": 0.8079470198675496, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.782051282051282, + "eval_LOCATION_recall": 0.8356164383561644, + "eval_ORGANIZATION_f1": 0.7007299270072994, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6666666666666666, + "eval_ORGANIZATION_recall": 0.7384615384615385, + "eval_PERSON_f1": 0.8846153846153846, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8518518518518519, + "eval_PERSON_recall": 0.92, + "eval_QUANTITY_f1": 0.6216216216216216, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.5111111111111111, + "eval_QUANTITY_recall": 0.7931034482758621, + "eval_TIME_f1": 0.8405797101449276, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8285714285714286, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.13471455872058868, + "eval_overall_accuracy": 0.9576667487078513, + "eval_overall_f1": 0.8048452220726783, + "eval_overall_precision": 0.7627551020408163, + "eval_overall_recall": 0.8518518518518519, + "eval_runtime": 0.349, + "eval_samples_per_second": 535.833, + "eval_steps_per_second": 8.596, + "step": 2226 + }, + { + "epoch": 22.0, + "grad_norm": 2.117086887359619, + "learning_rate": 3.9000000000000006e-05, + "loss": 0.0776, + "step": 2332 + }, + { + "epoch": 22.0, + "eval_LOCATION_f1": 0.7922077922077922, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7530864197530864, + "eval_LOCATION_recall": 0.8356164383561644, + "eval_ORGANIZATION_f1": 0.6751592356687899, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.5760869565217391, + "eval_ORGANIZATION_recall": 0.8153846153846154, + "eval_PERSON_f1": 0.8954248366013071, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8782051282051282, + "eval_PERSON_recall": 0.9133333333333333, + "eval_QUANTITY_f1": 0.7812500000000001, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7142857142857143, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.8450704225352113, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8108108108108109, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.13748787343502045, + "eval_overall_accuracy": 0.9556977602756583, + "eval_overall_f1": 0.8138297872340426, + "eval_overall_precision": 0.7630922693266833, + "eval_overall_recall": 0.8717948717948718, + "eval_runtime": 0.3484, + "eval_samples_per_second": 536.695, + "eval_steps_per_second": 8.61, + "step": 2332 + }, + { + "epoch": 23.0, + "grad_norm": 0.9372581839561462, + "learning_rate": 3.85e-05, + "loss": 0.0735, + "step": 2438 + }, + { + "epoch": 23.0, + "eval_LOCATION_f1": 0.8157894736842105, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7848101265822784, + "eval_LOCATION_recall": 0.8493150684931506, + "eval_ORGANIZATION_f1": 0.7412587412587412, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6794871794871795, + "eval_ORGANIZATION_recall": 0.8153846153846154, + "eval_PERSON_f1": 0.8817891373801918, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8466257668711656, + "eval_PERSON_recall": 0.92, + "eval_QUANTITY_f1": 0.8064516129032258, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7575757575757576, + "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": 0.13373343646526337, + "eval_overall_accuracy": 0.9618508491262614, + "eval_overall_f1": 0.8340080971659919, + "eval_overall_precision": 0.7923076923076923, + "eval_overall_recall": 0.8803418803418803, + "eval_runtime": 0.3486, + "eval_samples_per_second": 536.447, + "eval_steps_per_second": 8.606, + "step": 2438 + }, + { + "epoch": 24.0, + "grad_norm": 1.0222715139389038, + "learning_rate": 3.8e-05, + "loss": 0.073, + "step": 2544 + }, + { + "epoch": 24.0, + "eval_LOCATION_f1": 0.8026315789473685, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7721518987341772, + "eval_LOCATION_recall": 0.8356164383561644, + "eval_ORGANIZATION_f1": 0.6883116883116883, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.5955056179775281, + "eval_ORGANIZATION_recall": 0.8153846153846154, + "eval_PERSON_f1": 0.8996763754045306, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8742138364779874, + "eval_PERSON_recall": 0.9266666666666666, + "eval_QUANTITY_f1": 0.7692307692307692, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6944444444444444, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.8285714285714286, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8055555555555556, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.1394394040107727, + "eval_overall_accuracy": 0.9581589958158996, + "eval_overall_f1": 0.8186666666666667, + "eval_overall_precision": 0.7694235588972431, + "eval_overall_recall": 0.8746438746438746, + "eval_runtime": 0.3498, + "eval_samples_per_second": 534.589, + "eval_steps_per_second": 8.576, + "step": 2544 + }, + { + "epoch": 25.0, + "grad_norm": 0.7979212403297424, + "learning_rate": 3.7500000000000003e-05, + "loss": 0.0713, + "step": 2650 + }, + { + "epoch": 25.0, + "eval_LOCATION_f1": 0.7894736842105263, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.759493670886076, + "eval_LOCATION_recall": 0.821917808219178, + "eval_ORGANIZATION_f1": 0.7310344827586207, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6625, + "eval_ORGANIZATION_recall": 0.8153846153846154, + "eval_PERSON_f1": 0.8932038834951457, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8679245283018868, + "eval_PERSON_recall": 0.92, + "eval_QUANTITY_f1": 0.6956521739130435, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.7887323943661971, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7567567567567568, + "eval_TIME_recall": 0.8235294117647058, + "eval_loss": 0.14228081703186035, + "eval_overall_accuracy": 0.9598818606940684, + "eval_overall_f1": 0.8123324396782841, + "eval_overall_precision": 0.7670886075949367, + "eval_overall_recall": 0.8632478632478633, + "eval_runtime": 0.3501, + "eval_samples_per_second": 534.14, + "eval_steps_per_second": 8.569, + "step": 2650 + }, + { + "epoch": 26.0, + "grad_norm": 0.5613954663276672, + "learning_rate": 3.7e-05, + "loss": 0.0672, + "step": 2756 + }, + { + "epoch": 26.0, + "eval_LOCATION_f1": 0.7843137254901962, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.75, + "eval_LOCATION_recall": 0.821917808219178, + "eval_ORGANIZATION_f1": 0.7172413793103448, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.65, + "eval_ORGANIZATION_recall": 0.8, + "eval_PERSON_f1": 0.9067524115755627, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8757763975155279, + "eval_PERSON_recall": 0.94, + "eval_QUANTITY_f1": 0.6857142857142856, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.5853658536585366, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.7567567567567567, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7, + "eval_TIME_recall": 0.8235294117647058, + "eval_loss": 0.15085051953792572, + "eval_overall_accuracy": 0.956928378045779, + "eval_overall_f1": 0.8100929614873837, + "eval_overall_precision": 0.7587064676616916, + "eval_overall_recall": 0.8689458689458689, + "eval_runtime": 0.3515, + "eval_samples_per_second": 532.067, + "eval_steps_per_second": 8.536, + "step": 2756 + }, + { + "epoch": 27.0, + "grad_norm": 0.650170624256134, + "learning_rate": 3.65e-05, + "loss": 0.0651, + "step": 2862 + }, + { + "epoch": 27.0, + "eval_LOCATION_f1": 0.7866666666666666, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7662337662337663, + "eval_LOCATION_recall": 0.8082191780821918, + "eval_ORGANIZATION_f1": 0.7248322147651007, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6428571428571429, + "eval_ORGANIZATION_recall": 0.8307692307692308, + "eval_PERSON_f1": 0.8896103896103896, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8670886075949367, + "eval_PERSON_recall": 0.9133333333333333, + "eval_QUANTITY_f1": 0.716417910447761, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.631578947368421, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8169014084507041, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7837837837837838, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.14231501519680023, + "eval_overall_accuracy": 0.9581589958158996, + "eval_overall_f1": 0.8134228187919463, + "eval_overall_precision": 0.7690355329949239, + "eval_overall_recall": 0.8632478632478633, + "eval_runtime": 0.3486, + "eval_samples_per_second": 536.421, + "eval_steps_per_second": 8.606, + "step": 2862 + }, + { + "epoch": 28.0, + "grad_norm": 1.0213465690612793, + "learning_rate": 3.6e-05, + "loss": 0.0659, + "step": 2968 + }, + { + "epoch": 28.0, + "eval_LOCATION_f1": 0.8400000000000001, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8181818181818182, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.7142857142857142, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6666666666666666, + "eval_ORGANIZATION_recall": 0.7692307692307693, + "eval_PERSON_f1": 0.9019607843137256, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8846153846153846, + "eval_PERSON_recall": 0.92, + "eval_QUANTITY_f1": 0.7936507936507937, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7352941176470589, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.8571428571428571, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8333333333333334, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.13195978105068207, + "eval_overall_accuracy": 0.9616047255722373, + "eval_overall_f1": 0.8395061728395062, + "eval_overall_precision": 0.8095238095238095, + "eval_overall_recall": 0.8717948717948718, + "eval_runtime": 0.3507, + "eval_samples_per_second": 533.256, + "eval_steps_per_second": 8.555, + "step": 2968 + }, + { + "epoch": 29.0, + "grad_norm": 1.3771088123321533, + "learning_rate": 3.55e-05, + "loss": 0.0619, + "step": 3074 + }, + { + "epoch": 29.0, + "eval_LOCATION_f1": 0.8344370860927152, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8076923076923077, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.6956521739130436, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.5833333333333334, + "eval_ORGANIZATION_recall": 0.8615384615384616, + "eval_PERSON_f1": 0.9144736842105262, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.9025974025974026, + "eval_PERSON_recall": 0.9266666666666666, + "eval_QUANTITY_f1": 0.75, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6857142857142857, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8732394366197184, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8378378378378378, + "eval_TIME_recall": 0.9117647058823529, + "eval_loss": 0.1354289948940277, + "eval_overall_accuracy": 0.9616047255722373, + "eval_overall_f1": 0.833555259653795, + "eval_overall_precision": 0.7825, + "eval_overall_recall": 0.8917378917378918, + "eval_runtime": 0.3485, + "eval_samples_per_second": 536.594, + "eval_steps_per_second": 8.608, + "step": 3074 + }, + { + "epoch": 30.0, + "grad_norm": 1.0532158613204956, + "learning_rate": 3.5e-05, + "loss": 0.06, + "step": 3180 + }, + { + "epoch": 30.0, + "eval_LOCATION_f1": 0.8211920529801324, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7948717948717948, + "eval_LOCATION_recall": 0.8493150684931506, + "eval_ORGANIZATION_f1": 0.7083333333333334, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6455696202531646, + "eval_ORGANIZATION_recall": 0.7846153846153846, + "eval_PERSON_f1": 0.8925081433224756, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8726114649681529, + "eval_PERSON_recall": 0.9133333333333333, + "eval_QUANTITY_f1": 0.7741935483870968, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7272727272727273, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.7887323943661971, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7567567567567568, + "eval_TIME_recall": 0.8235294117647058, + "eval_loss": 0.1394852250814438, + "eval_overall_accuracy": 0.9613586020182131, + "eval_overall_f1": 0.8217687074829932, + "eval_overall_precision": 0.7864583333333334, + "eval_overall_recall": 0.8603988603988604, + "eval_runtime": 0.3484, + "eval_samples_per_second": 536.678, + "eval_steps_per_second": 8.61, + "step": 3180 + }, + { + "epoch": 31.0, + "grad_norm": 0.5543320178985596, + "learning_rate": 3.45e-05, + "loss": 0.0597, + "step": 3286 + }, + { + "epoch": 31.0, + "eval_LOCATION_f1": 0.8258064516129032, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7804878048780488, + "eval_LOCATION_recall": 0.8767123287671232, + "eval_ORGANIZATION_f1": 0.6861313868613138, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6527777777777778, + "eval_ORGANIZATION_recall": 0.7230769230769231, + "eval_PERSON_f1": 0.9114754098360656, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.896774193548387, + "eval_PERSON_recall": 0.9266666666666666, + "eval_QUANTITY_f1": 0.7575757575757576, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6756756756756757, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.8450704225352113, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8108108108108109, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.1364583671092987, + "eval_overall_accuracy": 0.9623430962343096, + "eval_overall_f1": 0.8310626702997275, + "eval_overall_precision": 0.7963446475195822, + "eval_overall_recall": 0.8689458689458689, + "eval_runtime": 0.3487, + "eval_samples_per_second": 536.339, + "eval_steps_per_second": 8.604, + "step": 3286 + }, + { + "epoch": 32.0, + "grad_norm": 0.2729957103729248, + "learning_rate": 3.4000000000000007e-05, + "loss": 0.0571, + "step": 3392 + }, + { + "epoch": 32.0, + "eval_LOCATION_f1": 0.8421052631578948, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.810126582278481, + "eval_LOCATION_recall": 0.8767123287671232, + "eval_ORGANIZATION_f1": 0.6962025316455697, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.5913978494623656, + "eval_ORGANIZATION_recall": 0.8461538461538461, + "eval_PERSON_f1": 0.9120521172638437, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.89171974522293, + "eval_PERSON_recall": 0.9333333333333333, + "eval_QUANTITY_f1": 0.7352941176470588, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6410256410256411, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.7999999999999999, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7777777777777778, + "eval_TIME_recall": 0.8235294117647058, + "eval_loss": 0.15985752642154694, + "eval_overall_accuracy": 0.9571745015998031, + "eval_overall_f1": 0.8264900662251655, + "eval_overall_precision": 0.7722772277227723, + "eval_overall_recall": 0.8888888888888888, + "eval_runtime": 0.3483, + "eval_samples_per_second": 536.94, + "eval_steps_per_second": 8.614, + "step": 3392 + }, + { + "epoch": 33.0, + "grad_norm": 1.3724058866500854, + "learning_rate": 3.35e-05, + "loss": 0.0567, + "step": 3498 + }, + { + "epoch": 33.0, + "eval_LOCATION_f1": 0.8496732026143791, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8125, + "eval_LOCATION_recall": 0.8904109589041096, + "eval_ORGANIZATION_f1": 0.7310344827586207, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6625, + "eval_ORGANIZATION_recall": 0.8153846153846154, + "eval_PERSON_f1": 0.9055374592833875, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8853503184713376, + "eval_PERSON_recall": 0.9266666666666666, + "eval_QUANTITY_f1": 0.7936507936507937, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7352941176470589, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.8285714285714286, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8055555555555556, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.14525863528251648, + "eval_overall_accuracy": 0.9633275904504061, + "eval_overall_f1": 0.8428184281842819, + "eval_overall_precision": 0.8036175710594315, + "eval_overall_recall": 0.886039886039886, + "eval_runtime": 0.3505, + "eval_samples_per_second": 533.595, + "eval_steps_per_second": 8.56, + "step": 3498 + }, + { + "epoch": 34.0, + "grad_norm": 1.019172191619873, + "learning_rate": 3.3e-05, + "loss": 0.0538, + "step": 3604 + }, + { + "epoch": 34.0, + "eval_LOCATION_f1": 0.794701986754967, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7692307692307693, + "eval_LOCATION_recall": 0.821917808219178, + "eval_ORGANIZATION_f1": 0.7222222222222222, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6582278481012658, + "eval_ORGANIZATION_recall": 0.8, + "eval_PERSON_f1": 0.8896103896103896, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8670886075949367, + "eval_PERSON_recall": 0.9133333333333333, + "eval_QUANTITY_f1": 0.8253968253968255, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7647058823529411, + "eval_QUANTITY_recall": 0.896551724137931, + "eval_TIME_f1": 0.8285714285714286, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8055555555555556, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.14389275014400482, + "eval_overall_accuracy": 0.9618508491262614, + "eval_overall_f1": 0.8260869565217391, + "eval_overall_precision": 0.7896103896103897, + "eval_overall_recall": 0.8660968660968661, + "eval_runtime": 0.3478, + "eval_samples_per_second": 537.643, + "eval_steps_per_second": 8.625, + "step": 3604 + }, + { + "epoch": 35.0, + "grad_norm": 0.2027391493320465, + "learning_rate": 3.2500000000000004e-05, + "loss": 0.055, + "step": 3710 + }, + { + "epoch": 35.0, + "eval_LOCATION_f1": 0.8051948051948052, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7654320987654321, + "eval_LOCATION_recall": 0.8493150684931506, + "eval_ORGANIZATION_f1": 0.7092198581560285, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6578947368421053, + "eval_ORGANIZATION_recall": 0.7692307692307693, + "eval_PERSON_f1": 0.896103896103896, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8734177215189873, + "eval_PERSON_recall": 0.92, + "eval_QUANTITY_f1": 0.71875, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6571428571428571, + "eval_QUANTITY_recall": 0.7931034482758621, + "eval_TIME_f1": 0.7887323943661971, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7567567567567568, + "eval_TIME_recall": 0.8235294117647058, + "eval_loss": 0.14834968745708466, + "eval_overall_accuracy": 0.9591434900319961, + "eval_overall_f1": 0.8157181571815719, + "eval_overall_precision": 0.7777777777777778, + "eval_overall_recall": 0.8575498575498576, + "eval_runtime": 0.3492, + "eval_samples_per_second": 535.458, + "eval_steps_per_second": 8.59, + "step": 3710 + }, + { + "epoch": 36.0, + "grad_norm": 1.058028221130371, + "learning_rate": 3.2000000000000005e-05, + "loss": 0.051, + "step": 3816 + }, + { + "epoch": 36.0, + "eval_LOCATION_f1": 0.8243243243243243, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8133333333333334, + "eval_LOCATION_recall": 0.8356164383561644, + "eval_ORGANIZATION_f1": 0.6896551724137931, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.625, + "eval_ORGANIZATION_recall": 0.7692307692307693, + "eval_PERSON_f1": 0.8938906752411575, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8633540372670807, + "eval_PERSON_recall": 0.9266666666666666, + "eval_QUANTITY_f1": 0.7692307692307692, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6944444444444444, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.8285714285714286, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8055555555555556, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.15179765224456787, + "eval_overall_accuracy": 0.9586512429239479, + "eval_overall_f1": 0.8227334235453315, + "eval_overall_precision": 0.7835051546391752, + "eval_overall_recall": 0.8660968660968661, + "eval_runtime": 0.35, + "eval_samples_per_second": 534.248, + "eval_steps_per_second": 8.571, + "step": 3816 + }, + { + "epoch": 37.0, + "grad_norm": 0.6165934205055237, + "learning_rate": 3.15e-05, + "loss": 0.0525, + "step": 3922 + }, + { + "epoch": 37.0, + "eval_LOCATION_f1": 0.8211920529801324, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7948717948717948, + "eval_LOCATION_recall": 0.8493150684931506, + "eval_ORGANIZATION_f1": 0.7042253521126761, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6493506493506493, + "eval_ORGANIZATION_recall": 0.7692307692307693, + "eval_PERSON_f1": 0.8990228013029316, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8789808917197452, + "eval_PERSON_recall": 0.92, + "eval_QUANTITY_f1": 0.75, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6857142857142857, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.7887323943661971, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7567567567567568, + "eval_TIME_recall": 0.8235294117647058, + "eval_loss": 0.1549917459487915, + "eval_overall_accuracy": 0.9586512429239479, + "eval_overall_f1": 0.8217687074829932, + "eval_overall_precision": 0.7864583333333334, + "eval_overall_recall": 0.8603988603988604, + "eval_runtime": 0.3479, + "eval_samples_per_second": 537.465, + "eval_steps_per_second": 8.622, + "step": 3922 + }, + { + "epoch": 38.0, + "grad_norm": 0.6908597350120544, + "learning_rate": 3.1e-05, + "loss": 0.0513, + "step": 4028 + }, + { + "epoch": 38.0, + "eval_LOCATION_f1": 0.8344370860927152, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8076923076923077, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.7407407407407408, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7142857142857143, + "eval_ORGANIZATION_recall": 0.7692307692307693, + "eval_PERSON_f1": 0.8925081433224756, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8726114649681529, + "eval_PERSON_recall": 0.9133333333333333, + "eval_QUANTITY_f1": 0.7575757575757576, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6756756756756757, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.8169014084507041, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7837837837837838, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.13987548649311066, + "eval_overall_accuracy": 0.9616047255722373, + "eval_overall_f1": 0.832876712328767, + "eval_overall_precision": 0.8021108179419525, + "eval_overall_recall": 0.8660968660968661, + "eval_runtime": 0.3495, + "eval_samples_per_second": 535.055, + "eval_steps_per_second": 8.584, + "step": 4028 + }, + { + "epoch": 39.0, + "grad_norm": 0.47891664505004883, + "learning_rate": 3.05e-05, + "loss": 0.0511, + "step": 4134 + }, + { + "epoch": 39.0, + "eval_LOCATION_f1": 0.8533333333333334, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8311688311688312, + "eval_LOCATION_recall": 0.8767123287671232, + "eval_ORGANIZATION_f1": 0.7162162162162162, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6385542168674698, + "eval_ORGANIZATION_recall": 0.8153846153846154, + "eval_PERSON_f1": 0.896774193548387, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.86875, + "eval_PERSON_recall": 0.9266666666666666, + "eval_QUANTITY_f1": 0.7619047619047619, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7058823529411765, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8055555555555555, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7631578947368421, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.14294342696666718, + "eval_overall_accuracy": 0.9613586020182131, + "eval_overall_f1": 0.8317631224764468, + "eval_overall_precision": 0.7882653061224489, + "eval_overall_recall": 0.8803418803418803, + "eval_runtime": 0.349, + "eval_samples_per_second": 535.775, + "eval_steps_per_second": 8.595, + "step": 4134 + }, + { + "epoch": 40.0, + "grad_norm": 0.5005225539207458, + "learning_rate": 3e-05, + "loss": 0.0479, + "step": 4240 + }, + { + "epoch": 40.0, + "eval_LOCATION_f1": 0.8289473684210525, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7974683544303798, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.6709677419354838, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.5777777777777777, + "eval_ORGANIZATION_recall": 0.8, + "eval_PERSON_f1": 0.9084967320261438, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8910256410256411, + "eval_PERSON_recall": 0.9266666666666666, + "eval_QUANTITY_f1": 0.716417910447761, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.631578947368421, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.7887323943661971, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7567567567567568, + "eval_TIME_recall": 0.8235294117647058, + "eval_loss": 0.16042561829090118, + "eval_overall_accuracy": 0.9566822544917548, + "eval_overall_f1": 0.8149134487350199, + "eval_overall_precision": 0.765, + "eval_overall_recall": 0.8717948717948718, + "eval_runtime": 0.3494, + "eval_samples_per_second": 535.212, + "eval_steps_per_second": 8.586, + "step": 4240 + }, + { + "epoch": 41.0, + "grad_norm": 1.4673404693603516, + "learning_rate": 2.95e-05, + "loss": 0.0478, + "step": 4346 + }, + { + "epoch": 41.0, + "eval_LOCATION_f1": 0.8533333333333334, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8311688311688312, + "eval_LOCATION_recall": 0.8767123287671232, + "eval_ORGANIZATION_f1": 0.7132867132867133, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6538461538461539, + "eval_ORGANIZATION_recall": 0.7846153846153846, + "eval_PERSON_f1": 0.8896103896103896, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8670886075949367, + "eval_PERSON_recall": 0.9133333333333333, + "eval_QUANTITY_f1": 0.8571428571428571, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7941176470588235, + "eval_QUANTITY_recall": 0.9310344827586207, + "eval_TIME_f1": 0.8571428571428571, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8333333333333334, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.14638105034828186, + "eval_overall_accuracy": 0.9620969726802855, + "eval_overall_f1": 0.8419618528610354, + "eval_overall_precision": 0.8067885117493473, + "eval_overall_recall": 0.8803418803418803, + "eval_runtime": 0.3474, + "eval_samples_per_second": 538.243, + "eval_steps_per_second": 8.635, + "step": 4346 + }, + { + "epoch": 42.0, + "grad_norm": 0.4796440303325653, + "learning_rate": 2.9e-05, + "loss": 0.0452, + "step": 4452 + }, + { + "epoch": 42.0, + "eval_LOCATION_f1": 0.8266666666666668, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8051948051948052, + "eval_LOCATION_recall": 0.8493150684931506, + "eval_ORGANIZATION_f1": 0.7222222222222222, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6582278481012658, + "eval_ORGANIZATION_recall": 0.8, + "eval_PERSON_f1": 0.9025974025974027, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.879746835443038, + "eval_PERSON_recall": 0.9266666666666666, + "eval_QUANTITY_f1": 0.7812500000000001, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7142857142857143, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.8169014084507041, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7837837837837838, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.14712657034397125, + "eval_overall_accuracy": 0.9616047255722373, + "eval_overall_f1": 0.8331071913161466, + "eval_overall_precision": 0.7953367875647669, + "eval_overall_recall": 0.8746438746438746, + "eval_runtime": 0.3492, + "eval_samples_per_second": 535.483, + "eval_steps_per_second": 8.591, + "step": 4452 + }, + { + "epoch": 43.0, + "grad_norm": 0.6836230754852295, + "learning_rate": 2.8499999999999998e-05, + "loss": 0.0444, + "step": 4558 + }, + { + "epoch": 43.0, + "eval_LOCATION_f1": 0.8441558441558442, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8024691358024691, + "eval_LOCATION_recall": 0.8904109589041096, + "eval_ORGANIZATION_f1": 0.7428571428571429, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6933333333333334, + "eval_ORGANIZATION_recall": 0.8, + "eval_PERSON_f1": 0.8918032786885245, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8774193548387097, + "eval_PERSON_recall": 0.9066666666666666, + "eval_QUANTITY_f1": 0.7812500000000001, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7142857142857143, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.8055555555555555, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7631578947368421, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.15814396739006042, + "eval_overall_accuracy": 0.9618508491262614, + "eval_overall_f1": 0.8353741496598639, + "eval_overall_precision": 0.7994791666666666, + "eval_overall_recall": 0.8746438746438746, + "eval_runtime": 0.3485, + "eval_samples_per_second": 536.57, + "eval_steps_per_second": 8.608, + "step": 4558 + }, + { + "epoch": 44.0, + "grad_norm": 1.3571970462799072, + "learning_rate": 2.8000000000000003e-05, + "loss": 0.0444, + "step": 4664 + }, + { + "epoch": 44.0, + "eval_LOCATION_f1": 0.8513513513513513, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.84, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.7272727272727272, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6666666666666666, + "eval_ORGANIZATION_recall": 0.8, + "eval_PERSON_f1": 0.8990228013029316, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8789808917197452, + "eval_PERSON_recall": 0.92, + "eval_QUANTITY_f1": 0.8253968253968255, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7647058823529411, + "eval_QUANTITY_recall": 0.896551724137931, + "eval_TIME_f1": 0.8823529411764706, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8823529411764706, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.14802251756191254, + "eval_overall_accuracy": 0.9643120846665025, + "eval_overall_f1": 0.8477366255144033, + "eval_overall_precision": 0.8174603174603174, + "eval_overall_recall": 0.8803418803418803, + "eval_runtime": 0.3488, + "eval_samples_per_second": 536.133, + "eval_steps_per_second": 8.601, + "step": 4664 + }, + { + "epoch": 45.0, + "grad_norm": 1.737742304801941, + "learning_rate": 2.7500000000000004e-05, + "loss": 0.0449, + "step": 4770 + }, + { + "epoch": 45.0, + "eval_LOCATION_f1": 0.8421052631578948, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.810126582278481, + "eval_LOCATION_recall": 0.8767123287671232, + "eval_ORGANIZATION_f1": 0.6845637583892618, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6071428571428571, + "eval_ORGANIZATION_recall": 0.7846153846153846, + "eval_PERSON_f1": 0.8925081433224756, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8726114649681529, + "eval_PERSON_recall": 0.9133333333333333, + "eval_QUANTITY_f1": 0.7272727272727273, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6486486486486487, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8695652173913043, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8571428571428571, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.1639709323644638, + "eval_overall_accuracy": 0.9584051193699237, + "eval_overall_f1": 0.8236877523553163, + "eval_overall_precision": 0.7806122448979592, + "eval_overall_recall": 0.8717948717948718, + "eval_runtime": 0.3494, + "eval_samples_per_second": 535.143, + "eval_steps_per_second": 8.585, + "step": 4770 + }, + { + "epoch": 46.0, + "grad_norm": 0.6632001996040344, + "learning_rate": 2.7000000000000002e-05, + "loss": 0.0415, + "step": 4876 + }, + { + "epoch": 46.0, + "eval_LOCATION_f1": 0.847682119205298, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8205128205128205, + "eval_LOCATION_recall": 0.8767123287671232, + "eval_ORGANIZATION_f1": 0.7313432835820897, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7101449275362319, + "eval_ORGANIZATION_recall": 0.7538461538461538, + "eval_PERSON_f1": 0.8794788273615636, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8598726114649682, + "eval_PERSON_recall": 0.9, + "eval_QUANTITY_f1": 0.7936507936507937, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7352941176470589, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.8285714285714286, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8055555555555556, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.14749835431575775, + "eval_overall_accuracy": 0.9620969726802855, + "eval_overall_f1": 0.833103448275862, + "eval_overall_precision": 0.8074866310160428, + "eval_overall_recall": 0.8603988603988604, + "eval_runtime": 0.3486, + "eval_samples_per_second": 536.455, + "eval_steps_per_second": 8.606, + "step": 4876 + }, + { + "epoch": 47.0, + "grad_norm": 0.8617376685142517, + "learning_rate": 2.6500000000000004e-05, + "loss": 0.0427, + "step": 4982 + }, + { + "epoch": 47.0, + "eval_LOCATION_f1": 0.8421052631578948, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.810126582278481, + "eval_LOCATION_recall": 0.8767123287671232, + "eval_ORGANIZATION_f1": 0.7285714285714285, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.68, + "eval_ORGANIZATION_recall": 0.7846153846153846, + "eval_PERSON_f1": 0.8846153846153846, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8518518518518519, + "eval_PERSON_recall": 0.92, + "eval_QUANTITY_f1": 0.7812500000000001, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7142857142857143, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.8450704225352113, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8108108108108109, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.1591494381427765, + "eval_overall_accuracy": 0.9606202313561408, + "eval_overall_f1": 0.8335588633288227, + "eval_overall_precision": 0.7938144329896907, + "eval_overall_recall": 0.8774928774928775, + "eval_runtime": 0.3508, + "eval_samples_per_second": 533.01, + "eval_steps_per_second": 8.551, + "step": 4982 + }, + { + "epoch": 48.0, + "grad_norm": 1.4177937507629395, + "learning_rate": 2.6000000000000002e-05, + "loss": 0.0424, + "step": 5088 + }, + { + "epoch": 48.0, + "eval_LOCATION_f1": 0.8289473684210525, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7974683544303798, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.7218045112781954, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7058823529411765, + "eval_ORGANIZATION_recall": 0.7384615384615385, + "eval_PERSON_f1": 0.8903225806451613, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8625, + "eval_PERSON_recall": 0.92, + "eval_QUANTITY_f1": 0.7936507936507937, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7352941176470589, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.7887323943661971, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7567567567567568, + "eval_TIME_recall": 0.8235294117647058, + "eval_loss": 0.16404277086257935, + "eval_overall_accuracy": 0.961112478464189, + "eval_overall_f1": 0.8285322359396434, + "eval_overall_precision": 0.798941798941799, + "eval_overall_recall": 0.8603988603988604, + "eval_runtime": 0.3479, + "eval_samples_per_second": 537.567, + "eval_steps_per_second": 8.624, + "step": 5088 + }, + { + "epoch": 49.0, + "grad_norm": 0.5694997310638428, + "learning_rate": 2.5500000000000003e-05, + "loss": 0.0408, + "step": 5194 + }, + { + "epoch": 49.0, + "eval_LOCATION_f1": 0.8441558441558442, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8024691358024691, + "eval_LOCATION_recall": 0.8904109589041096, + "eval_ORGANIZATION_f1": 0.7313432835820897, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7101449275362319, + "eval_ORGANIZATION_recall": 0.7538461538461538, + "eval_PERSON_f1": 0.8990228013029316, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8789808917197452, + "eval_PERSON_recall": 0.92, + "eval_QUANTITY_f1": 0.7575757575757576, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6756756756756757, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.8285714285714286, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8055555555555556, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.16149596869945526, + "eval_overall_accuracy": 0.9620969726802855, + "eval_overall_f1": 0.8372093023255812, + "eval_overall_precision": 0.8052631578947368, + "eval_overall_recall": 0.8717948717948718, + "eval_runtime": 0.3491, + "eval_samples_per_second": 535.625, + "eval_steps_per_second": 8.593, + "step": 5194 + }, + { + "epoch": 50.0, + "grad_norm": 1.3638904094696045, + "learning_rate": 2.5e-05, + "loss": 0.0403, + "step": 5300 + }, + { + "epoch": 50.0, + "eval_LOCATION_f1": 0.8441558441558442, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8024691358024691, + "eval_LOCATION_recall": 0.8904109589041096, + "eval_ORGANIZATION_f1": 0.7142857142857142, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6666666666666666, + "eval_ORGANIZATION_recall": 0.7692307692307693, + "eval_PERSON_f1": 0.8867313915857605, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8616352201257862, + "eval_PERSON_recall": 0.9133333333333333, + "eval_QUANTITY_f1": 0.75, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6857142857142857, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8169014084507041, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7837837837837838, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.1615622639656067, + "eval_overall_accuracy": 0.9598818606940684, + "eval_overall_f1": 0.8265582655826558, + "eval_overall_precision": 0.7881136950904393, + "eval_overall_recall": 0.8689458689458689, + "eval_runtime": 0.3494, + "eval_samples_per_second": 535.228, + "eval_steps_per_second": 8.587, + "step": 5300 + }, + { + "epoch": 51.0, + "grad_norm": 0.6612288951873779, + "learning_rate": 2.45e-05, + "loss": 0.0374, + "step": 5406 + }, + { + "epoch": 51.0, + "eval_LOCATION_f1": 0.8441558441558442, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8024691358024691, + "eval_LOCATION_recall": 0.8904109589041096, + "eval_ORGANIZATION_f1": 0.7424242424242424, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7313432835820896, + "eval_ORGANIZATION_recall": 0.7538461538461538, + "eval_PERSON_f1": 0.8774193548387097, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.85, + "eval_PERSON_recall": 0.9066666666666666, + "eval_QUANTITY_f1": 0.7812500000000001, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7142857142857143, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.8695652173913043, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8571428571428571, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.16601917147636414, + "eval_overall_accuracy": 0.9628353433423579, + "eval_overall_f1": 0.8367626886145405, + "eval_overall_precision": 0.8068783068783069, + "eval_overall_recall": 0.8689458689458689, + "eval_runtime": 0.3497, + "eval_samples_per_second": 534.82, + "eval_steps_per_second": 8.58, + "step": 5406 + }, + { + "epoch": 52.0, + "grad_norm": 1.5069667100906372, + "learning_rate": 2.4e-05, + "loss": 0.0382, + "step": 5512 + }, + { + "epoch": 52.0, + "eval_LOCATION_f1": 0.8211920529801324, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7948717948717948, + "eval_LOCATION_recall": 0.8493150684931506, + "eval_ORGANIZATION_f1": 0.7152317880794703, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.627906976744186, + "eval_ORGANIZATION_recall": 0.8307692307692308, + "eval_PERSON_f1": 0.8896103896103896, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8670886075949367, + "eval_PERSON_recall": 0.9133333333333333, + "eval_QUANTITY_f1": 0.7936507936507937, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7352941176470589, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.7777777777777778, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7368421052631579, + "eval_TIME_recall": 0.8235294117647058, + "eval_loss": 0.17699261009693146, + "eval_overall_accuracy": 0.9576667487078513, + "eval_overall_f1": 0.8214765100671142, + "eval_overall_precision": 0.7766497461928934, + "eval_overall_recall": 0.8717948717948718, + "eval_runtime": 0.3471, + "eval_samples_per_second": 538.682, + "eval_steps_per_second": 8.642, + "step": 5512 + }, + { + "epoch": 53.0, + "grad_norm": 1.0662306547164917, + "learning_rate": 2.35e-05, + "loss": 0.038, + "step": 5618 + }, + { + "epoch": 53.0, + "eval_LOCATION_f1": 0.846153846153846, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7951807228915663, + "eval_LOCATION_recall": 0.9041095890410958, + "eval_ORGANIZATION_f1": 0.7391304347826088, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6986301369863014, + "eval_ORGANIZATION_recall": 0.7846153846153846, + "eval_PERSON_f1": 0.887459807073955, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8571428571428571, + "eval_PERSON_recall": 0.92, + "eval_QUANTITY_f1": 0.7812500000000001, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7142857142857143, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.7999999999999999, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7777777777777778, + "eval_TIME_recall": 0.8235294117647058, + "eval_loss": 0.17185063660144806, + "eval_overall_accuracy": 0.9596357371400444, + "eval_overall_f1": 0.8335588633288227, + "eval_overall_precision": 0.7938144329896907, + "eval_overall_recall": 0.8774928774928775, + "eval_runtime": 0.3507, + "eval_samples_per_second": 533.285, + "eval_steps_per_second": 8.555, + "step": 5618 + }, + { + "epoch": 54.0, + "grad_norm": 0.5915600061416626, + "learning_rate": 2.3000000000000003e-05, + "loss": 0.0362, + "step": 5724 + }, + { + "epoch": 54.0, + "eval_LOCATION_f1": 0.8366013071895425, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8, + "eval_LOCATION_recall": 0.8767123287671232, + "eval_ORGANIZATION_f1": 0.7183098591549296, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6623376623376623, + "eval_ORGANIZATION_recall": 0.7846153846153846, + "eval_PERSON_f1": 0.8802588996763754, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8553459119496856, + "eval_PERSON_recall": 0.9066666666666666, + "eval_QUANTITY_f1": 0.8387096774193549, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7878787878787878, + "eval_QUANTITY_recall": 0.896551724137931, + "eval_TIME_f1": 0.8571428571428571, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8333333333333334, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.1768696904182434, + "eval_overall_accuracy": 0.958897366477972, + "eval_overall_f1": 0.8342391304347826, + "eval_overall_precision": 0.7974025974025974, + "eval_overall_recall": 0.8746438746438746, + "eval_runtime": 0.3484, + "eval_samples_per_second": 536.69, + "eval_steps_per_second": 8.61, + "step": 5724 + }, + { + "epoch": 55.0, + "grad_norm": 0.5453677773475647, + "learning_rate": 2.25e-05, + "loss": 0.034, + "step": 5830 + }, + { + "epoch": 55.0, + "eval_LOCATION_f1": 0.8311688311688312, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7901234567901234, + "eval_LOCATION_recall": 0.8767123287671232, + "eval_ORGANIZATION_f1": 0.7285714285714285, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.68, + "eval_ORGANIZATION_recall": 0.7846153846153846, + "eval_PERSON_f1": 0.8888888888888887, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8717948717948718, + "eval_PERSON_recall": 0.9066666666666666, + "eval_QUANTITY_f1": 0.8571428571428571, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7941176470588235, + "eval_QUANTITY_recall": 0.9310344827586207, + "eval_TIME_f1": 0.8571428571428571, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8333333333333334, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.16523310542106628, + "eval_overall_accuracy": 0.9616047255722373, + "eval_overall_f1": 0.840381991814461, + "eval_overall_precision": 0.806282722513089, + "eval_overall_recall": 0.8774928774928775, + "eval_runtime": 0.3497, + "eval_samples_per_second": 534.694, + "eval_steps_per_second": 8.578, + "step": 5830 + }, + { + "epoch": 56.0, + "grad_norm": 1.1855400800704956, + "learning_rate": 2.2000000000000003e-05, + "loss": 0.033, + "step": 5936 + }, + { + "epoch": 56.0, + "eval_LOCATION_f1": 0.8344370860927152, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8076923076923077, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.7480916030534351, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7424242424242424, + "eval_ORGANIZATION_recall": 0.7538461538461538, + "eval_PERSON_f1": 0.8838709677419355, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.85625, + "eval_PERSON_recall": 0.9133333333333333, + "eval_QUANTITY_f1": 0.7936507936507937, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7352941176470589, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.8450704225352113, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8108108108108109, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.16871850192546844, + "eval_overall_accuracy": 0.9623430962343096, + "eval_overall_f1": 0.837465564738292, + "eval_overall_precision": 0.8106666666666666, + "eval_overall_recall": 0.8660968660968661, + "eval_runtime": 0.3502, + "eval_samples_per_second": 534.014, + "eval_steps_per_second": 8.567, + "step": 5936 + }, + { + "epoch": 57.0, + "grad_norm": 0.6215670704841614, + "learning_rate": 2.15e-05, + "loss": 0.0357, + "step": 6042 + }, + { + "epoch": 57.0, + "eval_LOCATION_f1": 0.8456375838926175, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8289473684210527, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.7619047619047619, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7868852459016393, + "eval_ORGANIZATION_recall": 0.7384615384615385, + "eval_PERSON_f1": 0.8831168831168831, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8607594936708861, + "eval_PERSON_recall": 0.9066666666666666, + "eval_QUANTITY_f1": 0.8125, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7428571428571429, + "eval_QUANTITY_recall": 0.896551724137931, + "eval_TIME_f1": 0.8169014084507041, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7837837837837838, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.16778957843780518, + "eval_overall_accuracy": 0.9618508491262614, + "eval_overall_f1": 0.841225626740947, + "eval_overall_precision": 0.8228882833787466, + "eval_overall_recall": 0.8603988603988604, + "eval_runtime": 0.3498, + "eval_samples_per_second": 534.631, + "eval_steps_per_second": 8.577, + "step": 6042 + }, + { + "epoch": 58.0, + "grad_norm": 0.38038042187690735, + "learning_rate": 2.1e-05, + "loss": 0.0341, + "step": 6148 + }, + { + "epoch": 58.0, + "eval_LOCATION_f1": 0.8533333333333334, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8311688311688312, + "eval_LOCATION_recall": 0.8767123287671232, + "eval_ORGANIZATION_f1": 0.7258064516129032, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7627118644067796, + "eval_ORGANIZATION_recall": 0.6923076923076923, + "eval_PERSON_f1": 0.8823529411764707, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8653846153846154, + "eval_PERSON_recall": 0.9, + "eval_QUANTITY_f1": 0.8253968253968255, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7647058823529411, + "eval_QUANTITY_recall": 0.896551724137931, + "eval_TIME_f1": 0.7999999999999999, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7777777777777778, + "eval_TIME_recall": 0.8235294117647058, + "eval_loss": 0.16733954846858978, + "eval_overall_accuracy": 0.961112478464189, + "eval_overall_f1": 0.8359046283309959, + "eval_overall_precision": 0.8232044198895028, + "eval_overall_recall": 0.8490028490028491, + "eval_runtime": 0.3504, + "eval_samples_per_second": 533.675, + "eval_steps_per_second": 8.562, + "step": 6148 + }, + { + "epoch": 59.0, + "grad_norm": 0.5302915573120117, + "learning_rate": 2.05e-05, + "loss": 0.0324, + "step": 6254 + }, + { + "epoch": 59.0, + "eval_LOCATION_f1": 0.8366013071895425, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8, + "eval_LOCATION_recall": 0.8767123287671232, + "eval_ORGANIZATION_f1": 0.7375886524822695, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6842105263157895, + "eval_ORGANIZATION_recall": 0.8, + "eval_PERSON_f1": 0.8925081433224756, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8726114649681529, + "eval_PERSON_recall": 0.9133333333333333, + "eval_QUANTITY_f1": 0.7741935483870968, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7272727272727273, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8285714285714286, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8055555555555556, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.17023777961730957, + "eval_overall_accuracy": 0.9633275904504061, + "eval_overall_f1": 0.834924965893588, + "eval_overall_precision": 0.8010471204188482, + "eval_overall_recall": 0.8717948717948718, + "eval_runtime": 0.3498, + "eval_samples_per_second": 534.593, + "eval_steps_per_second": 8.576, + "step": 6254 + }, + { + "epoch": 60.0, + "grad_norm": 1.338559627532959, + "learning_rate": 2e-05, + "loss": 0.0333, + "step": 6360 + }, + { + "epoch": 60.0, + "eval_LOCATION_f1": 0.8609271523178808, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8333333333333334, + "eval_LOCATION_recall": 0.8904109589041096, + "eval_ORGANIZATION_f1": 0.7368421052631577, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7205882352941176, + "eval_ORGANIZATION_recall": 0.7538461538461538, + "eval_PERSON_f1": 0.9013157894736842, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8896103896103896, + "eval_PERSON_recall": 0.9133333333333333, + "eval_QUANTITY_f1": 0.8064516129032258, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7575757575757576, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.8571428571428571, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8333333333333334, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.16547058522701263, + "eval_overall_accuracy": 0.9657888259906473, + "eval_overall_f1": 0.8500000000000001, + "eval_overall_precision": 0.8292682926829268, + "eval_overall_recall": 0.8717948717948718, + "eval_runtime": 0.3501, + "eval_samples_per_second": 534.07, + "eval_steps_per_second": 8.568, + "step": 6360 + }, + { + "epoch": 61.0, + "grad_norm": 0.4147072434425354, + "learning_rate": 1.9500000000000003e-05, + "loss": 0.0327, + "step": 6466 + }, + { + "epoch": 61.0, + "eval_LOCATION_f1": 0.8400000000000001, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8181818181818182, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.7428571428571429, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6933333333333334, + "eval_ORGANIZATION_recall": 0.8, + "eval_PERSON_f1": 0.8817891373801918, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8466257668711656, + "eval_PERSON_recall": 0.92, + "eval_QUANTITY_f1": 0.75, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6857142857142857, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8695652173913043, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8571428571428571, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.17973226308822632, + "eval_overall_accuracy": 0.9608663549101649, + "eval_overall_f1": 0.8342391304347826, + "eval_overall_precision": 0.7974025974025974, + "eval_overall_recall": 0.8746438746438746, + "eval_runtime": 0.3496, + "eval_samples_per_second": 534.945, + "eval_steps_per_second": 8.582, + "step": 6466 + }, + { + "epoch": 62.0, + "grad_norm": 0.6705744862556458, + "learning_rate": 1.9e-05, + "loss": 0.0315, + "step": 6572 + }, + { + "epoch": 62.0, + "eval_LOCATION_f1": 0.8421052631578948, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.810126582278481, + "eval_LOCATION_recall": 0.8767123287671232, + "eval_ORGANIZATION_f1": 0.7352941176470589, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.704225352112676, + "eval_ORGANIZATION_recall": 0.7692307692307693, + "eval_PERSON_f1": 0.8802588996763754, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8553459119496856, + "eval_PERSON_recall": 0.9066666666666666, + "eval_QUANTITY_f1": 0.7936507936507937, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7352941176470589, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.7999999999999999, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7777777777777778, + "eval_TIME_recall": 0.8235294117647058, + "eval_loss": 0.1808030903339386, + "eval_overall_accuracy": 0.9601279842480925, + "eval_overall_f1": 0.8301369863013699, + "eval_overall_precision": 0.7994722955145118, + "eval_overall_recall": 0.8632478632478633, + "eval_runtime": 0.3485, + "eval_samples_per_second": 536.545, + "eval_steps_per_second": 8.608, + "step": 6572 + }, + { + "epoch": 63.0, + "grad_norm": 0.7228378653526306, + "learning_rate": 1.85e-05, + "loss": 0.0295, + "step": 6678 + }, + { + "epoch": 63.0, + "eval_LOCATION_f1": 0.8571428571428571, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8148148148148148, + "eval_LOCATION_recall": 0.9041095890410958, + "eval_ORGANIZATION_f1": 0.7153284671532848, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6805555555555556, + "eval_ORGANIZATION_recall": 0.7538461538461538, + "eval_PERSON_f1": 0.8903225806451613, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8625, + "eval_PERSON_recall": 0.92, + "eval_QUANTITY_f1": 0.8709677419354839, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.8181818181818182, + "eval_QUANTITY_recall": 0.9310344827586207, + "eval_TIME_f1": 0.8571428571428571, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8333333333333334, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.17379428446292877, + "eval_overall_accuracy": 0.9625892197883338, + "eval_overall_f1": 0.8458390177353343, + "eval_overall_precision": 0.8115183246073299, + "eval_overall_recall": 0.8831908831908832, + "eval_runtime": 0.3488, + "eval_samples_per_second": 536.14, + "eval_steps_per_second": 8.601, + "step": 6678 + }, + { + "epoch": 64.0, + "grad_norm": 0.5657730102539062, + "learning_rate": 1.8e-05, + "loss": 0.0296, + "step": 6784 + }, + { + "epoch": 64.0, + "eval_LOCATION_f1": 0.8366013071895425, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8, + "eval_LOCATION_recall": 0.8767123287671232, + "eval_ORGANIZATION_f1": 0.7424242424242424, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7313432835820896, + "eval_ORGANIZATION_recall": 0.7538461538461538, + "eval_PERSON_f1": 0.887459807073955, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8571428571428571, + "eval_PERSON_recall": 0.92, + "eval_QUANTITY_f1": 0.8571428571428571, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7941176470588235, + "eval_QUANTITY_recall": 0.9310344827586207, + "eval_TIME_f1": 0.8571428571428571, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8333333333333334, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.1676260232925415, + "eval_overall_accuracy": 0.963081466896382, + "eval_overall_f1": 0.8449931412894376, + "eval_overall_precision": 0.8148148148148148, + "eval_overall_recall": 0.8774928774928775, + "eval_runtime": 0.3485, + "eval_samples_per_second": 536.562, + "eval_steps_per_second": 8.608, + "step": 6784 + }, + { + "epoch": 65.0, + "grad_norm": 0.8568927645683289, + "learning_rate": 1.75e-05, + "loss": 0.0297, + "step": 6890 + }, + { + "epoch": 65.0, + "eval_LOCATION_f1": 0.8533333333333334, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8311688311688312, + "eval_LOCATION_recall": 0.8767123287671232, + "eval_ORGANIZATION_f1": 0.75, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7183098591549296, + "eval_ORGANIZATION_recall": 0.7846153846153846, + "eval_PERSON_f1": 0.8786885245901639, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.864516129032258, + "eval_PERSON_recall": 0.8933333333333333, + "eval_QUANTITY_f1": 0.7936507936507937, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7352941176470589, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.8115942028985507, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8, + "eval_TIME_recall": 0.8235294117647058, + "eval_loss": 0.1690235435962677, + "eval_overall_accuracy": 0.9620969726802855, + "eval_overall_f1": 0.8354080221300137, + "eval_overall_precision": 0.8118279569892473, + "eval_overall_recall": 0.8603988603988604, + "eval_runtime": 0.3507, + "eval_samples_per_second": 533.268, + "eval_steps_per_second": 8.555, + "step": 6890 + }, + { + "epoch": 66.0, + "grad_norm": 0.7192606329917908, + "learning_rate": 1.7000000000000003e-05, + "loss": 0.028, + "step": 6996 + }, + { + "epoch": 66.0, + "eval_LOCATION_f1": 0.8235294117647058, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7875, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.7575757575757576, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.746268656716418, + "eval_ORGANIZATION_recall": 0.7692307692307693, + "eval_PERSON_f1": 0.8932038834951457, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8679245283018868, + "eval_PERSON_recall": 0.92, + "eval_QUANTITY_f1": 0.8709677419354839, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.8181818181818182, + "eval_QUANTITY_recall": 0.9310344827586207, + "eval_TIME_f1": 0.8571428571428571, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8333333333333334, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.17469918727874756, + "eval_overall_accuracy": 0.9648043317745508, + "eval_overall_f1": 0.8484848484848485, + "eval_overall_precision": 0.8213333333333334, + "eval_overall_recall": 0.8774928774928775, + "eval_runtime": 0.3489, + "eval_samples_per_second": 535.899, + "eval_steps_per_second": 8.597, + "step": 6996 + }, + { + "epoch": 67.0, + "grad_norm": 0.2507493793964386, + "learning_rate": 1.65e-05, + "loss": 0.0278, + "step": 7102 + }, + { + "epoch": 67.0, + "eval_LOCATION_f1": 0.8400000000000001, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8181818181818182, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.7391304347826088, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6986301369863014, + "eval_ORGANIZATION_recall": 0.7846153846153846, + "eval_PERSON_f1": 0.8786885245901639, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.864516129032258, + "eval_PERSON_recall": 0.8933333333333333, + "eval_QUANTITY_f1": 0.8387096774193549, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7878787878787878, + "eval_QUANTITY_recall": 0.896551724137931, + "eval_TIME_f1": 0.8571428571428571, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8333333333333334, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.17670658230781555, + "eval_overall_accuracy": 0.9616047255722373, + "eval_overall_f1": 0.8386206896551723, + "eval_overall_precision": 0.8128342245989305, + "eval_overall_recall": 0.8660968660968661, + "eval_runtime": 0.3485, + "eval_samples_per_second": 536.521, + "eval_steps_per_second": 8.607, + "step": 7102 + }, + { + "epoch": 68.0, + "grad_norm": 0.48297932744026184, + "learning_rate": 1.6000000000000003e-05, + "loss": 0.0292, + "step": 7208 + }, + { + "epoch": 68.0, + "eval_LOCATION_f1": 0.8235294117647058, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7875, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.7246376811594203, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.684931506849315, + "eval_ORGANIZATION_recall": 0.7692307692307693, + "eval_PERSON_f1": 0.9042904290429042, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8954248366013072, + "eval_PERSON_recall": 0.9133333333333333, + "eval_QUANTITY_f1": 0.8253968253968255, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7647058823529411, + "eval_QUANTITY_recall": 0.896551724137931, + "eval_TIME_f1": 0.8169014084507041, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7837837837837838, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.18086454272270203, + "eval_overall_accuracy": 0.9625892197883338, + "eval_overall_f1": 0.8379120879120879, + "eval_overall_precision": 0.8090185676392573, + "eval_overall_recall": 0.8689458689458689, + "eval_runtime": 0.3491, + "eval_samples_per_second": 535.689, + "eval_steps_per_second": 8.594, + "step": 7208 + }, + { + "epoch": 69.0, + "grad_norm": 0.1779003143310547, + "learning_rate": 1.55e-05, + "loss": 0.0264, + "step": 7314 + }, + { + "epoch": 69.0, + "eval_LOCATION_f1": 0.8496732026143791, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8125, + "eval_LOCATION_recall": 0.8904109589041096, + "eval_ORGANIZATION_f1": 0.7092198581560285, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6578947368421053, + "eval_ORGANIZATION_recall": 0.7692307692307693, + "eval_PERSON_f1": 0.8954248366013071, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8782051282051282, + "eval_PERSON_recall": 0.9133333333333333, + "eval_QUANTITY_f1": 0.8709677419354839, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.8181818181818182, + "eval_QUANTITY_recall": 0.9310344827586207, + "eval_TIME_f1": 0.8571428571428571, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8333333333333334, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.17253029346466064, + "eval_overall_accuracy": 0.9640659611124784, + "eval_overall_f1": 0.8442622950819673, + "eval_overall_precision": 0.8110236220472441, + "eval_overall_recall": 0.8803418803418803, + "eval_runtime": 0.3497, + "eval_samples_per_second": 534.704, + "eval_steps_per_second": 8.578, + "step": 7314 + }, + { + "epoch": 70.0, + "grad_norm": 0.405724436044693, + "learning_rate": 1.5e-05, + "loss": 0.0259, + "step": 7420 + }, + { + "epoch": 70.0, + "eval_LOCATION_f1": 0.855263157894737, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8227848101265823, + "eval_LOCATION_recall": 0.8904109589041096, + "eval_ORGANIZATION_f1": 0.7445255474452555, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7083333333333334, + "eval_ORGANIZATION_recall": 0.7846153846153846, + "eval_PERSON_f1": 0.8954248366013071, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8782051282051282, + "eval_PERSON_recall": 0.9133333333333333, + "eval_QUANTITY_f1": 0.7812500000000001, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7142857142857143, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.8571428571428571, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8333333333333334, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.17524121701717377, + "eval_overall_accuracy": 0.9638198375584544, + "eval_overall_f1": 0.8449931412894376, + "eval_overall_precision": 0.8148148148148148, + "eval_overall_recall": 0.8774928774928775, + "eval_runtime": 0.3487, + "eval_samples_per_second": 536.201, + "eval_steps_per_second": 8.602, + "step": 7420 + }, + { + "epoch": 71.0, + "grad_norm": 1.9814850091934204, + "learning_rate": 1.45e-05, + "loss": 0.0274, + "step": 7526 + }, + { + "epoch": 71.0, + "eval_LOCATION_f1": 0.855263157894737, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8227848101265823, + "eval_LOCATION_recall": 0.8904109589041096, + "eval_ORGANIZATION_f1": 0.7445255474452555, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7083333333333334, + "eval_ORGANIZATION_recall": 0.7846153846153846, + "eval_PERSON_f1": 0.8810289389067525, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8509316770186336, + "eval_PERSON_recall": 0.9133333333333333, + "eval_QUANTITY_f1": 0.8125, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7428571428571429, + "eval_QUANTITY_recall": 0.896551724137931, + "eval_TIME_f1": 0.8285714285714286, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8055555555555556, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.19072633981704712, + "eval_overall_accuracy": 0.961112478464189, + "eval_overall_f1": 0.8392370572207084, + "eval_overall_precision": 0.804177545691906, + "eval_overall_recall": 0.8774928774928775, + "eval_runtime": 0.349, + "eval_samples_per_second": 535.864, + "eval_steps_per_second": 8.597, + "step": 7526 + }, + { + "epoch": 72.0, + "grad_norm": 0.3593089282512665, + "learning_rate": 1.4000000000000001e-05, + "loss": 0.0273, + "step": 7632 + }, + { + "epoch": 72.0, + "eval_LOCATION_f1": 0.8496732026143791, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8125, + "eval_LOCATION_recall": 0.8904109589041096, + "eval_ORGANIZATION_f1": 0.7518796992481205, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7352941176470589, + "eval_ORGANIZATION_recall": 0.7692307692307693, + "eval_PERSON_f1": 0.8737864077669903, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8490566037735849, + "eval_PERSON_recall": 0.9, + "eval_QUANTITY_f1": 0.8253968253968255, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7647058823529411, + "eval_QUANTITY_recall": 0.896551724137931, + "eval_TIME_f1": 0.7777777777777778, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7368421052631579, + "eval_TIME_recall": 0.8235294117647058, + "eval_loss": 0.1915425956249237, + "eval_overall_accuracy": 0.9598818606940684, + "eval_overall_f1": 0.832876712328767, + "eval_overall_precision": 0.8021108179419525, + "eval_overall_recall": 0.8660968660968661, + "eval_runtime": 0.3484, + "eval_samples_per_second": 536.761, + "eval_steps_per_second": 8.611, + "step": 7632 + }, + { + "epoch": 73.0, + "grad_norm": 0.8845640420913696, + "learning_rate": 1.3500000000000001e-05, + "loss": 0.0279, + "step": 7738 + }, + { + "epoch": 73.0, + "eval_LOCATION_f1": 0.8289473684210525, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7974683544303798, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.7323943661971831, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6753246753246753, + "eval_ORGANIZATION_recall": 0.8, + "eval_PERSON_f1": 0.8802588996763754, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8553459119496856, + "eval_PERSON_recall": 0.9066666666666666, + "eval_QUANTITY_f1": 0.8571428571428571, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7941176470588235, + "eval_QUANTITY_recall": 0.9310344827586207, + "eval_TIME_f1": 0.7999999999999999, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7777777777777778, + "eval_TIME_recall": 0.8235294117647058, + "eval_loss": 0.1847066581249237, + "eval_overall_accuracy": 0.9613586020182131, + "eval_overall_f1": 0.8315217391304348, + "eval_overall_precision": 0.7948051948051948, + "eval_overall_recall": 0.8717948717948718, + "eval_runtime": 0.3502, + "eval_samples_per_second": 534.048, + "eval_steps_per_second": 8.568, + "step": 7738 + }, + { + "epoch": 74.0, + "grad_norm": 0.45966747403144836, + "learning_rate": 1.3000000000000001e-05, + "loss": 0.0267, + "step": 7844 + }, + { + "epoch": 74.0, + "eval_LOCATION_f1": 0.8421052631578948, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.810126582278481, + "eval_LOCATION_recall": 0.8767123287671232, + "eval_ORGANIZATION_f1": 0.7480916030534351, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7424242424242424, + "eval_ORGANIZATION_recall": 0.7538461538461538, + "eval_PERSON_f1": 0.870967741935484, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.84375, + "eval_PERSON_recall": 0.9, + "eval_QUANTITY_f1": 0.8125, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7428571428571429, + "eval_QUANTITY_recall": 0.896551724137931, + "eval_TIME_f1": 0.8285714285714286, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8055555555555556, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.18399041891098022, + "eval_overall_accuracy": 0.9625892197883338, + "eval_overall_f1": 0.8335625859697385, + "eval_overall_precision": 0.8058510638297872, + "eval_overall_recall": 0.8632478632478633, + "eval_runtime": 0.3487, + "eval_samples_per_second": 536.311, + "eval_steps_per_second": 8.604, + "step": 7844 + }, + { + "epoch": 75.0, + "grad_norm": 0.6150727868080139, + "learning_rate": 1.25e-05, + "loss": 0.0257, + "step": 7950 + }, + { + "epoch": 75.0, + "eval_LOCATION_f1": 0.846153846153846, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7951807228915663, + "eval_LOCATION_recall": 0.9041095890410958, + "eval_ORGANIZATION_f1": 0.726027397260274, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.654320987654321, + "eval_ORGANIZATION_recall": 0.8153846153846154, + "eval_PERSON_f1": 0.8774193548387097, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.85, + "eval_PERSON_recall": 0.9066666666666666, + "eval_QUANTITY_f1": 0.8571428571428571, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7941176470588235, + "eval_QUANTITY_recall": 0.9310344827586207, + "eval_TIME_f1": 0.8285714285714286, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8055555555555556, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.18748342990875244, + "eval_overall_accuracy": 0.9618508491262614, + "eval_overall_f1": 0.8348993288590604, + "eval_overall_precision": 0.7893401015228426, + "eval_overall_recall": 0.886039886039886, + "eval_runtime": 0.3479, + "eval_samples_per_second": 537.546, + "eval_steps_per_second": 8.624, + "step": 7950 + }, + { + "epoch": 76.0, + "grad_norm": 0.12720619142055511, + "learning_rate": 1.2e-05, + "loss": 0.0257, + "step": 8056 + }, + { + "epoch": 76.0, + "eval_LOCATION_f1": 0.8421052631578948, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.810126582278481, + "eval_LOCATION_recall": 0.8767123287671232, + "eval_ORGANIZATION_f1": 0.7246376811594203, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.684931506849315, + "eval_ORGANIZATION_recall": 0.7692307692307693, + "eval_PERSON_f1": 0.8852459016393444, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8709677419354839, + "eval_PERSON_recall": 0.9, + "eval_QUANTITY_f1": 0.8571428571428571, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7941176470588235, + "eval_QUANTITY_recall": 0.9310344827586207, + "eval_TIME_f1": 0.8571428571428571, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8333333333333334, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.18419043719768524, + "eval_overall_accuracy": 0.9628353433423579, + "eval_overall_f1": 0.8406593406593407, + "eval_overall_precision": 0.8116710875331565, + "eval_overall_recall": 0.8717948717948718, + "eval_runtime": 0.3487, + "eval_samples_per_second": 536.289, + "eval_steps_per_second": 8.604, + "step": 8056 + }, + { + "epoch": 77.0, + "grad_norm": 1.0697262287139893, + "learning_rate": 1.1500000000000002e-05, + "loss": 0.0248, + "step": 8162 + }, + { + "epoch": 77.0, + "eval_LOCATION_f1": 0.8289473684210525, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7974683544303798, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.7142857142857142, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6666666666666666, + "eval_ORGANIZATION_recall": 0.7692307692307693, + "eval_PERSON_f1": 0.8766233766233766, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8544303797468354, + "eval_PERSON_recall": 0.9, + "eval_QUANTITY_f1": 0.8387096774193549, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7878787878787878, + "eval_QUANTITY_recall": 0.896551724137931, + "eval_TIME_f1": 0.8285714285714286, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8055555555555556, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.19783645868301392, + "eval_overall_accuracy": 0.958897366477972, + "eval_overall_f1": 0.8278688524590164, + "eval_overall_precision": 0.7952755905511811, + "eval_overall_recall": 0.8632478632478633, + "eval_runtime": 0.3502, + "eval_samples_per_second": 534.03, + "eval_steps_per_second": 8.567, + "step": 8162 + }, + { + "epoch": 78.0, + "grad_norm": 1.0857927799224854, + "learning_rate": 1.1000000000000001e-05, + "loss": 0.0248, + "step": 8268 + }, + { + "epoch": 78.0, + "eval_LOCATION_f1": 0.8129032258064516, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7682926829268293, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.7407407407407408, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7142857142857143, + "eval_ORGANIZATION_recall": 0.7692307692307693, + "eval_PERSON_f1": 0.8774193548387097, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.85, + "eval_PERSON_recall": 0.9066666666666666, + "eval_QUANTITY_f1": 0.8571428571428571, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7941176470588235, + "eval_QUANTITY_recall": 0.9310344827586207, + "eval_TIME_f1": 0.8285714285714286, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8055555555555556, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.19514693319797516, + "eval_overall_accuracy": 0.9606202313561408, + "eval_overall_f1": 0.8321964529331515, + "eval_overall_precision": 0.7984293193717278, + "eval_overall_recall": 0.8689458689458689, + "eval_runtime": 0.3504, + "eval_samples_per_second": 533.645, + "eval_steps_per_second": 8.561, + "step": 8268 + }, + { + "epoch": 79.0, + "grad_norm": 0.4629979431629181, + "learning_rate": 1.05e-05, + "loss": 0.0237, + "step": 8374 + }, + { + "epoch": 79.0, + "eval_LOCATION_f1": 0.8441558441558442, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8024691358024691, + "eval_LOCATION_recall": 0.8904109589041096, + "eval_ORGANIZATION_f1": 0.75, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7183098591549296, + "eval_ORGANIZATION_recall": 0.7846153846153846, + "eval_PERSON_f1": 0.8758169934640523, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8589743589743589, + "eval_PERSON_recall": 0.8933333333333333, + "eval_QUANTITY_f1": 0.8253968253968255, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7647058823529411, + "eval_QUANTITY_recall": 0.896551724137931, + "eval_TIME_f1": 0.8285714285714286, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8055555555555556, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.19404210150241852, + "eval_overall_accuracy": 0.9603741078021166, + "eval_overall_f1": 0.8367626886145405, + "eval_overall_precision": 0.8068783068783069, + "eval_overall_recall": 0.8689458689458689, + "eval_runtime": 0.3494, + "eval_samples_per_second": 535.278, + "eval_steps_per_second": 8.587, + "step": 8374 + }, + { + "epoch": 80.0, + "grad_norm": 0.49161767959594727, + "learning_rate": 1e-05, + "loss": 0.0261, + "step": 8480 + }, + { + "epoch": 80.0, + "eval_LOCATION_f1": 0.8366013071895425, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8, + "eval_LOCATION_recall": 0.8767123287671232, + "eval_ORGANIZATION_f1": 0.7703703703703704, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7428571428571429, + "eval_ORGANIZATION_recall": 0.8, + "eval_PERSON_f1": 0.870967741935484, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.84375, + "eval_PERSON_recall": 0.9, + "eval_QUANTITY_f1": 0.7692307692307692, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6944444444444444, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.8285714285714286, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8055555555555556, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.19337113201618195, + "eval_overall_accuracy": 0.9613586020182131, + "eval_overall_f1": 0.8321964529331515, + "eval_overall_precision": 0.7984293193717278, + "eval_overall_recall": 0.8689458689458689, + "eval_runtime": 0.3493, + "eval_samples_per_second": 535.38, + "eval_steps_per_second": 8.589, + "step": 8480 + }, + { + "epoch": 81.0, + "grad_norm": 1.3529359102249146, + "learning_rate": 9.5e-06, + "loss": 0.0234, + "step": 8586 + }, + { + "epoch": 81.0, + "eval_LOCATION_f1": 0.8441558441558442, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8024691358024691, + "eval_LOCATION_recall": 0.8904109589041096, + "eval_ORGANIZATION_f1": 0.7462686567164178, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7246376811594203, + "eval_ORGANIZATION_recall": 0.7692307692307693, + "eval_PERSON_f1": 0.8823529411764707, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8653846153846154, + "eval_PERSON_recall": 0.9, + "eval_QUANTITY_f1": 0.8307692307692308, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.75, + "eval_QUANTITY_recall": 0.9310344827586207, + "eval_TIME_f1": 0.8285714285714286, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8055555555555556, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.19133438169956207, + "eval_overall_accuracy": 0.9625892197883338, + "eval_overall_f1": 0.8395061728395062, + "eval_overall_precision": 0.8095238095238095, + "eval_overall_recall": 0.8717948717948718, + "eval_runtime": 0.3476, + "eval_samples_per_second": 537.954, + "eval_steps_per_second": 8.63, + "step": 8586 + }, + { + "epoch": 82.0, + "grad_norm": 0.8348244428634644, + "learning_rate": 9e-06, + "loss": 0.0217, + "step": 8692 + }, + { + "epoch": 82.0, + "eval_LOCATION_f1": 0.8441558441558442, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8024691358024691, + "eval_LOCATION_recall": 0.8904109589041096, + "eval_ORGANIZATION_f1": 0.7391304347826088, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6986301369863014, + "eval_ORGANIZATION_recall": 0.7846153846153846, + "eval_PERSON_f1": 0.8794788273615636, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8598726114649682, + "eval_PERSON_recall": 0.9, + "eval_QUANTITY_f1": 0.8709677419354839, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.8181818181818182, + "eval_QUANTITY_recall": 0.9310344827586207, + "eval_TIME_f1": 0.8285714285714286, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8055555555555556, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.19519662857055664, + "eval_overall_accuracy": 0.9623430962343096, + "eval_overall_f1": 0.8399452804377566, + "eval_overall_precision": 0.8078947368421052, + "eval_overall_recall": 0.8746438746438746, + "eval_runtime": 0.3492, + "eval_samples_per_second": 535.512, + "eval_steps_per_second": 8.591, + "step": 8692 + }, + { + "epoch": 83.0, + "grad_norm": 1.4528393745422363, + "learning_rate": 8.500000000000002e-06, + "loss": 0.0234, + "step": 8798 + }, + { + "epoch": 83.0, + "eval_LOCATION_f1": 0.8441558441558442, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8024691358024691, + "eval_LOCATION_recall": 0.8904109589041096, + "eval_ORGANIZATION_f1": 0.7846153846153847, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7846153846153846, + "eval_ORGANIZATION_recall": 0.7846153846153846, + "eval_PERSON_f1": 0.8896103896103896, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8670886075949367, + "eval_PERSON_recall": 0.9133333333333333, + "eval_QUANTITY_f1": 0.8571428571428571, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7941176470588235, + "eval_QUANTITY_recall": 0.9310344827586207, + "eval_TIME_f1": 0.8571428571428571, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8333333333333334, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.19463619589805603, + "eval_overall_accuracy": 0.9655427024366232, + "eval_overall_f1": 0.8551724137931035, + "eval_overall_precision": 0.8288770053475936, + "eval_overall_recall": 0.8831908831908832, + "eval_runtime": 0.3489, + "eval_samples_per_second": 536.036, + "eval_steps_per_second": 8.6, + "step": 8798 + }, + { + "epoch": 84.0, + "grad_norm": 0.33660465478897095, + "learning_rate": 8.000000000000001e-06, + "loss": 0.0237, + "step": 8904 + }, + { + "epoch": 84.0, + "eval_LOCATION_f1": 0.8387096774193549, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7926829268292683, + "eval_LOCATION_recall": 0.8904109589041096, + "eval_ORGANIZATION_f1": 0.7462686567164178, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7246376811594203, + "eval_ORGANIZATION_recall": 0.7692307692307693, + "eval_PERSON_f1": 0.8867313915857605, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8616352201257862, + "eval_PERSON_recall": 0.9133333333333333, + "eval_QUANTITY_f1": 0.7999999999999999, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7222222222222222, + "eval_QUANTITY_recall": 0.896551724137931, + "eval_TIME_f1": 0.8285714285714286, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8055555555555556, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.19823411107063293, + "eval_overall_accuracy": 0.9616047255722373, + "eval_overall_f1": 0.8376534788540245, + "eval_overall_precision": 0.8036649214659686, + "eval_overall_recall": 0.8746438746438746, + "eval_runtime": 0.3505, + "eval_samples_per_second": 533.564, + "eval_steps_per_second": 8.56, + "step": 8904 + }, + { + "epoch": 85.0, + "grad_norm": 0.6225272417068481, + "learning_rate": 7.5e-06, + "loss": 0.0247, + "step": 9010 + }, + { + "epoch": 85.0, + "eval_LOCATION_f1": 0.8441558441558442, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8024691358024691, + "eval_LOCATION_recall": 0.8904109589041096, + "eval_ORGANIZATION_f1": 0.7445255474452555, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7083333333333334, + "eval_ORGANIZATION_recall": 0.7846153846153846, + "eval_PERSON_f1": 0.8867313915857605, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8616352201257862, + "eval_PERSON_recall": 0.9133333333333333, + "eval_QUANTITY_f1": 0.7812500000000001, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7142857142857143, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.8285714285714286, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8055555555555556, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.2002585530281067, + "eval_overall_accuracy": 0.9618508491262614, + "eval_overall_f1": 0.8365122615803815, + "eval_overall_precision": 0.8015665796344648, + "eval_overall_recall": 0.8746438746438746, + "eval_runtime": 0.3497, + "eval_samples_per_second": 534.708, + "eval_steps_per_second": 8.578, + "step": 9010 + }, + { + "epoch": 86.0, + "grad_norm": 0.7242019772529602, + "learning_rate": 7.000000000000001e-06, + "loss": 0.0218, + "step": 9116 + }, + { + "epoch": 86.0, + "eval_LOCATION_f1": 0.8366013071895425, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8, + "eval_LOCATION_recall": 0.8767123287671232, + "eval_ORGANIZATION_f1": 0.7352941176470589, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.704225352112676, + "eval_ORGANIZATION_recall": 0.7692307692307693, + "eval_PERSON_f1": 0.8925081433224756, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8726114649681529, + "eval_PERSON_recall": 0.9133333333333333, + "eval_QUANTITY_f1": 0.8571428571428571, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7941176470588235, + "eval_QUANTITY_recall": 0.9310344827586207, + "eval_TIME_f1": 0.8285714285714286, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8055555555555556, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.19523298740386963, + "eval_overall_accuracy": 0.9640659611124784, + "eval_overall_f1": 0.8422496570644719, + "eval_overall_precision": 0.8121693121693122, + "eval_overall_recall": 0.8746438746438746, + "eval_runtime": 0.3519, + "eval_samples_per_second": 531.467, + "eval_steps_per_second": 8.526, + "step": 9116 + }, + { + "epoch": 87.0, + "grad_norm": 1.2844712734222412, + "learning_rate": 6.5000000000000004e-06, + "loss": 0.0224, + "step": 9222 + }, + { + "epoch": 87.0, + "eval_LOCATION_f1": 0.8496732026143791, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8125, + "eval_LOCATION_recall": 0.8904109589041096, + "eval_ORGANIZATION_f1": 0.7611940298507462, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7391304347826086, + "eval_ORGANIZATION_recall": 0.7846153846153846, + "eval_PERSON_f1": 0.8896103896103896, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8670886075949367, + "eval_PERSON_recall": 0.9133333333333333, + "eval_QUANTITY_f1": 0.7812500000000001, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7142857142857143, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.8285714285714286, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8055555555555556, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.1952124983072281, + "eval_overall_accuracy": 0.9640659611124784, + "eval_overall_f1": 0.8422496570644719, + "eval_overall_precision": 0.8121693121693122, + "eval_overall_recall": 0.8746438746438746, + "eval_runtime": 0.349, + "eval_samples_per_second": 535.792, + "eval_steps_per_second": 8.596, + "step": 9222 + }, + { + "epoch": 88.0, + "grad_norm": 0.27208393812179565, + "learning_rate": 6e-06, + "loss": 0.0211, + "step": 9328 + }, + { + "epoch": 88.0, + "eval_LOCATION_f1": 0.8366013071895425, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8, + "eval_LOCATION_recall": 0.8767123287671232, + "eval_ORGANIZATION_f1": 0.7338129496402878, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6891891891891891, + "eval_ORGANIZATION_recall": 0.7846153846153846, + "eval_PERSON_f1": 0.8925081433224756, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8726114649681529, + "eval_PERSON_recall": 0.9133333333333333, + "eval_QUANTITY_f1": 0.7812500000000001, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7142857142857143, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.8285714285714286, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8055555555555556, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.1970175951719284, + "eval_overall_accuracy": 0.9620969726802855, + "eval_overall_f1": 0.834924965893588, + "eval_overall_precision": 0.8010471204188482, + "eval_overall_recall": 0.8717948717948718, + "eval_runtime": 0.3477, + "eval_samples_per_second": 537.813, + "eval_steps_per_second": 8.628, + "step": 9328 + }, + { + "epoch": 89.0, + "grad_norm": 0.7367871999740601, + "learning_rate": 5.500000000000001e-06, + "loss": 0.021, + "step": 9434 + }, + { + "epoch": 89.0, + "eval_LOCATION_f1": 0.8289473684210525, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7974683544303798, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.75, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7183098591549296, + "eval_ORGANIZATION_recall": 0.7846153846153846, + "eval_PERSON_f1": 0.8802588996763754, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8553459119496856, + "eval_PERSON_recall": 0.9066666666666666, + "eval_QUANTITY_f1": 0.7812500000000001, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7142857142857143, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.8285714285714286, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8055555555555556, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.20013782382011414, + "eval_overall_accuracy": 0.9613586020182131, + "eval_overall_f1": 0.8317373461012312, + "eval_overall_precision": 0.8, + "eval_overall_recall": 0.8660968660968661, + "eval_runtime": 0.3491, + "eval_samples_per_second": 535.725, + "eval_steps_per_second": 8.595, + "step": 9434 + }, + { + "epoch": 90.0, + "grad_norm": 0.4959828555583954, + "learning_rate": 5e-06, + "loss": 0.0216, + "step": 9540 + }, + { + "epoch": 90.0, + "eval_LOCATION_f1": 0.8211920529801324, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7948717948717948, + "eval_LOCATION_recall": 0.8493150684931506, + "eval_ORGANIZATION_f1": 0.7611940298507462, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7391304347826086, + "eval_ORGANIZATION_recall": 0.7846153846153846, + "eval_PERSON_f1": 0.8896103896103896, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8670886075949367, + "eval_PERSON_recall": 0.9133333333333333, + "eval_QUANTITY_f1": 0.7936507936507937, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7352941176470589, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.8169014084507041, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7837837837837838, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.19662083685398102, + "eval_overall_accuracy": 0.963081466896382, + "eval_overall_f1": 0.8363136176066025, + "eval_overall_precision": 0.8085106382978723, + "eval_overall_recall": 0.8660968660968661, + "eval_runtime": 0.3491, + "eval_samples_per_second": 535.654, + "eval_steps_per_second": 8.593, + "step": 9540 + }, + { + "epoch": 91.0, + "grad_norm": 0.6600481867790222, + "learning_rate": 4.5e-06, + "loss": 0.0217, + "step": 9646 + }, + { + "epoch": 91.0, + "eval_LOCATION_f1": 0.8366013071895425, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8, + "eval_LOCATION_recall": 0.8767123287671232, + "eval_ORGANIZATION_f1": 0.7445255474452555, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7083333333333334, + "eval_ORGANIZATION_recall": 0.7846153846153846, + "eval_PERSON_f1": 0.8867313915857605, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8616352201257862, + "eval_PERSON_recall": 0.9133333333333333, + "eval_QUANTITY_f1": 0.8253968253968255, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7647058823529411, + "eval_QUANTITY_recall": 0.896551724137931, + "eval_TIME_f1": 0.8285714285714286, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8055555555555556, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.19578596949577332, + "eval_overall_accuracy": 0.9625892197883338, + "eval_overall_f1": 0.8387978142076502, + "eval_overall_precision": 0.8057742782152231, + "eval_overall_recall": 0.8746438746438746, + "eval_runtime": 0.3484, + "eval_samples_per_second": 536.773, + "eval_steps_per_second": 8.611, + "step": 9646 + }, + { + "epoch": 92.0, + "grad_norm": 0.1194983422756195, + "learning_rate": 4.000000000000001e-06, + "loss": 0.0198, + "step": 9752 + }, + { + "epoch": 92.0, + "eval_LOCATION_f1": 0.8366013071895425, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8, + "eval_LOCATION_recall": 0.8767123287671232, + "eval_ORGANIZATION_f1": 0.7611940298507462, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7391304347826086, + "eval_ORGANIZATION_recall": 0.7846153846153846, + "eval_PERSON_f1": 0.8896103896103896, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8670886075949367, + "eval_PERSON_recall": 0.9133333333333333, + "eval_QUANTITY_f1": 0.8253968253968255, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7647058823529411, + "eval_QUANTITY_recall": 0.896551724137931, + "eval_TIME_f1": 0.8169014084507041, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7837837837837838, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.19786879420280457, + "eval_overall_accuracy": 0.9635737140044303, + "eval_overall_f1": 0.8422496570644719, + "eval_overall_precision": 0.8121693121693122, + "eval_overall_recall": 0.8746438746438746, + "eval_runtime": 0.3495, + "eval_samples_per_second": 535.03, + "eval_steps_per_second": 8.583, + "step": 9752 + }, + { + "epoch": 93.0, + "grad_norm": 0.16003356873989105, + "learning_rate": 3.5000000000000004e-06, + "loss": 0.0205, + "step": 9858 + }, + { + "epoch": 93.0, + "eval_LOCATION_f1": 0.8289473684210525, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7974683544303798, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.7669172932330828, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.75, + "eval_ORGANIZATION_recall": 0.7846153846153846, + "eval_PERSON_f1": 0.8896103896103896, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8670886075949367, + "eval_PERSON_recall": 0.9133333333333333, + "eval_QUANTITY_f1": 0.8125, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7428571428571429, + "eval_QUANTITY_recall": 0.896551724137931, + "eval_TIME_f1": 0.8450704225352113, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8108108108108109, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.19753791391849518, + "eval_overall_accuracy": 0.9635737140044303, + "eval_overall_f1": 0.8434065934065934, + "eval_overall_precision": 0.8143236074270557, + "eval_overall_recall": 0.8746438746438746, + "eval_runtime": 0.3492, + "eval_samples_per_second": 535.578, + "eval_steps_per_second": 8.592, + "step": 9858 + }, + { + "epoch": 94.0, + "grad_norm": 0.8688186407089233, + "learning_rate": 3e-06, + "loss": 0.0206, + "step": 9964 + }, + { + "epoch": 94.0, + "eval_LOCATION_f1": 0.8289473684210525, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7974683544303798, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.7555555555555555, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7285714285714285, + "eval_ORGANIZATION_recall": 0.7846153846153846, + "eval_PERSON_f1": 0.8859934853420196, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8662420382165605, + "eval_PERSON_recall": 0.9066666666666666, + "eval_QUANTITY_f1": 0.8125, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7428571428571429, + "eval_QUANTITY_recall": 0.896551724137931, + "eval_TIME_f1": 0.8169014084507041, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7837837837837838, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.1996561884880066, + "eval_overall_accuracy": 0.9623430962343096, + "eval_overall_f1": 0.8367626886145405, + "eval_overall_precision": 0.8068783068783069, + "eval_overall_recall": 0.8689458689458689, + "eval_runtime": 0.3491, + "eval_samples_per_second": 535.588, + "eval_steps_per_second": 8.592, + "step": 9964 + }, + { + "epoch": 95.0, + "grad_norm": 0.6198663115501404, + "learning_rate": 2.5e-06, + "loss": 0.0214, + "step": 10070 + }, + { + "epoch": 95.0, + "eval_LOCATION_f1": 0.8387096774193549, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7926829268292683, + "eval_LOCATION_recall": 0.8904109589041096, + "eval_ORGANIZATION_f1": 0.7445255474452555, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7083333333333334, + "eval_ORGANIZATION_recall": 0.7846153846153846, + "eval_PERSON_f1": 0.8802588996763754, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8553459119496856, + "eval_PERSON_recall": 0.9066666666666666, + "eval_QUANTITY_f1": 0.8253968253968255, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7647058823529411, + "eval_QUANTITY_recall": 0.896551724137931, + "eval_TIME_f1": 0.8169014084507041, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7837837837837838, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.19923867285251617, + "eval_overall_accuracy": 0.9613586020182131, + "eval_overall_f1": 0.8353741496598639, + "eval_overall_precision": 0.7994791666666666, + "eval_overall_recall": 0.8746438746438746, + "eval_runtime": 0.3512, + "eval_samples_per_second": 532.477, + "eval_steps_per_second": 8.542, + "step": 10070 + }, + { + "epoch": 96.0, + "grad_norm": 0.5730480551719666, + "learning_rate": 2.0000000000000003e-06, + "loss": 0.0208, + "step": 10176 + }, + { + "epoch": 96.0, + "eval_LOCATION_f1": 0.8333333333333334, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7831325301204819, + "eval_LOCATION_recall": 0.8904109589041096, + "eval_ORGANIZATION_f1": 0.7669172932330828, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.75, + "eval_ORGANIZATION_recall": 0.7846153846153846, + "eval_PERSON_f1": 0.8802588996763754, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8553459119496856, + "eval_PERSON_recall": 0.9066666666666666, + "eval_QUANTITY_f1": 0.7936507936507937, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7352941176470589, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.8450704225352113, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8108108108108109, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.19816720485687256, + "eval_overall_accuracy": 0.9623430962343096, + "eval_overall_f1": 0.8387978142076502, + "eval_overall_precision": 0.8057742782152231, + "eval_overall_recall": 0.8746438746438746, + "eval_runtime": 0.3492, + "eval_samples_per_second": 535.457, + "eval_steps_per_second": 8.59, + "step": 10176 + }, + { + "epoch": 97.0, + "grad_norm": 0.7892258167266846, + "learning_rate": 1.5e-06, + "loss": 0.0217, + "step": 10282 + }, + { + "epoch": 97.0, + "eval_LOCATION_f1": 0.8441558441558442, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8024691358024691, + "eval_LOCATION_recall": 0.8904109589041096, + "eval_ORGANIZATION_f1": 0.7669172932330828, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.75, + "eval_ORGANIZATION_recall": 0.7846153846153846, + "eval_PERSON_f1": 0.8831168831168831, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8607594936708861, + "eval_PERSON_recall": 0.9066666666666666, + "eval_QUANTITY_f1": 0.7936507936507937, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7352941176470589, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.8169014084507041, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7837837837837838, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.19672809541225433, + "eval_overall_accuracy": 0.963081466896382, + "eval_overall_f1": 0.8395061728395062, + "eval_overall_precision": 0.8095238095238095, + "eval_overall_recall": 0.8717948717948718, + "eval_runtime": 0.348, + "eval_samples_per_second": 537.376, + "eval_steps_per_second": 8.621, + "step": 10282 + }, + { + "epoch": 98.0, + "grad_norm": 0.4192006289958954, + "learning_rate": 1.0000000000000002e-06, + "loss": 0.0205, + "step": 10388 + }, + { + "epoch": 98.0, + "eval_LOCATION_f1": 0.8387096774193549, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7926829268292683, + "eval_LOCATION_recall": 0.8904109589041096, + "eval_ORGANIZATION_f1": 0.7669172932330828, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.75, + "eval_ORGANIZATION_recall": 0.7846153846153846, + "eval_PERSON_f1": 0.8802588996763754, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8553459119496856, + "eval_PERSON_recall": 0.9066666666666666, + "eval_QUANTITY_f1": 0.7936507936507937, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7352941176470589, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.8169014084507041, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7837837837837838, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.19882169365882874, + "eval_overall_accuracy": 0.9620969726802855, + "eval_overall_f1": 0.8372093023255812, + "eval_overall_precision": 0.8052631578947368, + "eval_overall_recall": 0.8717948717948718, + "eval_runtime": 0.35, + "eval_samples_per_second": 534.343, + "eval_steps_per_second": 8.572, + "step": 10388 + }, + { + "epoch": 99.0, + "grad_norm": 0.21814487874507904, + "learning_rate": 5.000000000000001e-07, + "loss": 0.0204, + "step": 10494 + }, + { + "epoch": 99.0, + "eval_LOCATION_f1": 0.8441558441558442, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8024691358024691, + "eval_LOCATION_recall": 0.8904109589041096, + "eval_ORGANIZATION_f1": 0.7669172932330828, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.75, + "eval_ORGANIZATION_recall": 0.7846153846153846, + "eval_PERSON_f1": 0.8831168831168831, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8607594936708861, + "eval_PERSON_recall": 0.9066666666666666, + "eval_QUANTITY_f1": 0.7812500000000001, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7142857142857143, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.8169014084507041, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7837837837837838, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.19684545695781708, + "eval_overall_accuracy": 0.9628353433423579, + "eval_overall_f1": 0.8383561643835615, + "eval_overall_precision": 0.8073878627968337, + "eval_overall_recall": 0.8717948717948718, + "eval_runtime": 0.3493, + "eval_samples_per_second": 535.431, + "eval_steps_per_second": 8.59, + "step": 10494 + }, + { + "epoch": 100.0, + "grad_norm": 0.46098318696022034, + "learning_rate": 0.0, + "loss": 0.022, + "step": 10600 + }, + { + "epoch": 100.0, + "eval_LOCATION_f1": 0.8441558441558442, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8024691358024691, + "eval_LOCATION_recall": 0.8904109589041096, + "eval_ORGANIZATION_f1": 0.7669172932330828, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.75, + "eval_ORGANIZATION_recall": 0.7846153846153846, + "eval_PERSON_f1": 0.8802588996763754, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8553459119496856, + "eval_PERSON_recall": 0.9066666666666666, + "eval_QUANTITY_f1": 0.7936507936507937, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7352941176470589, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.8169014084507041, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7837837837837838, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.19717250764369965, + "eval_overall_accuracy": 0.9625892197883338, + "eval_overall_f1": 0.8383561643835615, + "eval_overall_precision": 0.8073878627968337, + "eval_overall_recall": 0.8717948717948718, + "eval_runtime": 0.3485, + "eval_samples_per_second": 536.575, + "eval_steps_per_second": 8.608, + "step": 10600 + }, + { + "epoch": 100.0, + "step": 10600, + "total_flos": 5043631411410816.0, + "train_loss": 0.06754140655949431, + "train_runtime": 594.8485, + "train_samples_per_second": 283.77, + "train_steps_per_second": 17.82 + } + ], + "logging_steps": 500, + "max_steps": 10600, + "num_input_tokens_seen": 0, + "num_train_epochs": 100, + "save_steps": 500, + "total_flos": 5043631411410816.0, + "train_batch_size": 16, + "trial_name": null, + "trial_params": null +}