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": 1.9157154560089111, + "learning_rate": 4.9500000000000004e-05, + "loss": 0.965, + "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": 0.1188118811881188, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.23076923076923078, + "eval_PERSON_recall": 0.08, + "eval_QUANTITY_f1": 0.0, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.0, + "eval_QUANTITY_recall": 0.0, + "eval_TIME_f1": 0.05405405405405406, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.3333333333333333, + "eval_TIME_recall": 0.029411764705882353, + "eval_loss": 0.5562426447868347, + "eval_overall_accuracy": 0.8456805316268767, + "eval_overall_f1": 0.061465721040189124, + "eval_overall_precision": 0.18055555555555555, + "eval_overall_recall": 0.037037037037037035, + "eval_runtime": 0.3741, + "eval_samples_per_second": 499.863, + "eval_steps_per_second": 8.019, + "step": 106 + }, + { + "epoch": 2.0, + "grad_norm": 0.9464585781097412, + "learning_rate": 4.9e-05, + "loss": 0.4705, + "step": 212 + }, + { + "epoch": 2.0, + "eval_LOCATION_f1": 0.2752293577981651, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.4166666666666667, + "eval_LOCATION_recall": 0.2054794520547945, + "eval_ORGANIZATION_f1": 0.05660377358490566, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.07317073170731707, + "eval_ORGANIZATION_recall": 0.046153846153846156, + "eval_PERSON_f1": 0.6983240223463687, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.6009615384615384, + "eval_PERSON_recall": 0.8333333333333334, + "eval_QUANTITY_f1": 0.19999999999999998, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.23809523809523808, + "eval_QUANTITY_recall": 0.1724137931034483, + "eval_TIME_f1": 0.6233766233766234, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.5581395348837209, + "eval_TIME_recall": 0.7058823529411765, + "eval_loss": 0.3253639042377472, + "eval_overall_accuracy": 0.9030273197144967, + "eval_overall_f1": 0.49142857142857144, + "eval_overall_precision": 0.49283667621776506, + "eval_overall_recall": 0.49002849002849, + "eval_runtime": 0.3253, + "eval_samples_per_second": 574.813, + "eval_steps_per_second": 9.222, + "step": 212 + }, + { + "epoch": 3.0, + "grad_norm": 1.0709123611450195, + "learning_rate": 4.85e-05, + "loss": 0.2979, + "step": 318 + }, + { + "epoch": 3.0, + "eval_LOCATION_f1": 0.5930232558139534, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.5151515151515151, + "eval_LOCATION_recall": 0.6986301369863014, + "eval_ORGANIZATION_f1": 0.49645390070921985, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.4605263157894737, + "eval_ORGANIZATION_recall": 0.5384615384615384, + "eval_PERSON_f1": 0.8148148148148148, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.7586206896551724, + "eval_PERSON_recall": 0.88, + "eval_QUANTITY_f1": 0.5142857142857142, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.43902439024390244, + "eval_QUANTITY_recall": 0.6206896551724138, + "eval_TIME_f1": 0.6944444444444445, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.6578947368421053, + "eval_TIME_recall": 0.7352941176470589, + "eval_loss": 0.21049197018146515, + "eval_overall_accuracy": 0.934531134629584, + "eval_overall_f1": 0.6700898587933247, + "eval_overall_precision": 0.6098130841121495, + "eval_overall_recall": 0.7435897435897436, + "eval_runtime": 0.326, + "eval_samples_per_second": 573.557, + "eval_steps_per_second": 9.201, + "step": 318 + }, + { + "epoch": 4.0, + "grad_norm": 1.957574725151062, + "learning_rate": 4.8e-05, + "loss": 0.2203, + "step": 424 + }, + { + "epoch": 4.0, + "eval_LOCATION_f1": 0.6709677419354839, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.6341463414634146, + "eval_LOCATION_recall": 0.7123287671232876, + "eval_ORGANIZATION_f1": 0.5443786982248522, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.4423076923076923, + "eval_ORGANIZATION_recall": 0.7076923076923077, + "eval_PERSON_f1": 0.8517350157728706, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8083832335329342, + "eval_PERSON_recall": 0.9, + "eval_QUANTITY_f1": 0.5714285714285715, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.5294117647058824, + "eval_QUANTITY_recall": 0.6206896551724138, + "eval_TIME_f1": 0.8656716417910447, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8787878787878788, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.17120462656021118, + "eval_overall_accuracy": 0.9438838296825006, + "eval_overall_f1": 0.7263294422827496, + "eval_overall_precision": 0.6666666666666666, + "eval_overall_recall": 0.7977207977207977, + "eval_runtime": 0.3267, + "eval_samples_per_second": 572.428, + "eval_steps_per_second": 9.183, + "step": 424 + }, + { + "epoch": 5.0, + "grad_norm": 1.4644794464111328, + "learning_rate": 4.75e-05, + "loss": 0.19, + "step": 530 + }, + { + "epoch": 5.0, + "eval_LOCATION_f1": 0.7189542483660132, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.6875, + "eval_LOCATION_recall": 0.7534246575342466, + "eval_ORGANIZATION_f1": 0.641025641025641, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.5494505494505495, + "eval_ORGANIZATION_recall": 0.7692307692307693, + "eval_PERSON_f1": 0.8662420382165605, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8292682926829268, + "eval_PERSON_recall": 0.9066666666666666, + "eval_QUANTITY_f1": 0.6086956521739131, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.525, + "eval_QUANTITY_recall": 0.7241379310344828, + "eval_TIME_f1": 0.8219178082191781, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7692307692307693, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.15461339056491852, + "eval_overall_accuracy": 0.9512675363032242, + "eval_overall_f1": 0.7633986928104575, + "eval_overall_precision": 0.7053140096618358, + "eval_overall_recall": 0.8319088319088319, + "eval_runtime": 0.3274, + "eval_samples_per_second": 571.23, + "eval_steps_per_second": 9.164, + "step": 530 + }, + { + "epoch": 6.0, + "grad_norm": 1.5039018392562866, + "learning_rate": 4.7e-05, + "loss": 0.1692, + "step": 636 + }, + { + "epoch": 6.0, + "eval_LOCATION_f1": 0.6790123456790124, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.6179775280898876, + "eval_LOCATION_recall": 0.7534246575342466, + "eval_ORGANIZATION_f1": 0.6666666666666666, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.5436893203883495, + "eval_ORGANIZATION_recall": 0.8615384615384616, + "eval_PERSON_f1": 0.8769716088328076, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8323353293413174, + "eval_PERSON_recall": 0.9266666666666666, + "eval_QUANTITY_f1": 0.6486486486486487, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.5333333333333333, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.7368421052631577, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.6666666666666666, + "eval_TIME_recall": 0.8235294117647058, + "eval_loss": 0.169162318110466, + "eval_overall_accuracy": 0.9441299532365247, + "eval_overall_f1": 0.7578419071518193, + "eval_overall_precision": 0.6771300448430493, + "eval_overall_recall": 0.8603988603988604, + "eval_runtime": 0.3263, + "eval_samples_per_second": 573.162, + "eval_steps_per_second": 9.195, + "step": 636 + }, + { + "epoch": 7.0, + "grad_norm": 1.124490737915039, + "learning_rate": 4.6500000000000005e-05, + "loss": 0.1547, + "step": 742 + }, + { + "epoch": 7.0, + "eval_LOCATION_f1": 0.7096774193548389, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.6707317073170732, + "eval_LOCATION_recall": 0.7534246575342466, + "eval_ORGANIZATION_f1": 0.6575342465753424, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.5925925925925926, + "eval_ORGANIZATION_recall": 0.7384615384615385, + "eval_PERSON_f1": 0.8853503184713376, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8475609756097561, + "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": 0.7945205479452054, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7435897435897436, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.1434997171163559, + "eval_overall_accuracy": 0.9527442776273689, + "eval_overall_f1": 0.7824933687002652, + "eval_overall_precision": 0.7320099255583127, + "eval_overall_recall": 0.8404558404558404, + "eval_runtime": 0.3268, + "eval_samples_per_second": 572.222, + "eval_steps_per_second": 9.18, + "step": 742 + }, + { + "epoch": 8.0, + "grad_norm": 1.3524500131607056, + "learning_rate": 4.600000000000001e-05, + "loss": 0.1459, + "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": 0.6923076923076923, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.5934065934065934, + "eval_ORGANIZATION_recall": 0.8307692307692308, + "eval_PERSON_f1": 0.8910256410256411, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8580246913580247, + "eval_PERSON_recall": 0.9266666666666666, + "eval_QUANTITY_f1": 0.7384615384615385, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6666666666666666, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8108108108108107, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.75, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.14050818979740143, + "eval_overall_accuracy": 0.9549593896135861, + "eval_overall_f1": 0.8020698576972832, + "eval_overall_precision": 0.7345971563981043, + "eval_overall_recall": 0.8831908831908832, + "eval_runtime": 0.3272, + "eval_samples_per_second": 571.439, + "eval_steps_per_second": 9.167, + "step": 848 + }, + { + "epoch": 9.0, + "grad_norm": 4.653275012969971, + "learning_rate": 4.55e-05, + "loss": 0.1372, + "step": 954 + }, + { + "epoch": 9.0, + "eval_LOCATION_f1": 0.7999999999999999, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7560975609756098, + "eval_LOCATION_recall": 0.8493150684931506, + "eval_ORGANIZATION_f1": 0.6962025316455697, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.5913978494623656, + "eval_ORGANIZATION_recall": 0.8461538461538461, + "eval_PERSON_f1": 0.887459807073955, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8571428571428571, + "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": 0.1368522197008133, + "eval_overall_accuracy": 0.9559438838296825, + "eval_overall_f1": 0.8089005235602094, + "eval_overall_precision": 0.7481840193704601, + "eval_overall_recall": 0.8803418803418803, + "eval_runtime": 0.3269, + "eval_samples_per_second": 572.104, + "eval_steps_per_second": 9.178, + "step": 954 + }, + { + "epoch": 10.0, + "grad_norm": 1.1514642238616943, + "learning_rate": 4.5e-05, + "loss": 0.1324, + "step": 1060 + }, + { + "epoch": 10.0, + "eval_LOCATION_f1": 0.781456953642384, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7564102564102564, + "eval_LOCATION_recall": 0.8082191780821918, + "eval_ORGANIZATION_f1": 0.7105263157894737, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6206896551724138, + "eval_ORGANIZATION_recall": 0.8307692307692308, + "eval_PERSON_f1": 0.8817891373801918, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8466257668711656, + "eval_PERSON_recall": 0.92, + "eval_QUANTITY_f1": 0.676923076923077, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6111111111111112, + "eval_QUANTITY_recall": 0.7586206896551724, + "eval_TIME_f1": 0.8333333333333333, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7894736842105263, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.13584822416305542, + "eval_overall_accuracy": 0.9554516367216342, + "eval_overall_f1": 0.8047808764940239, + "eval_overall_precision": 0.753731343283582, + "eval_overall_recall": 0.8632478632478633, + "eval_runtime": 0.327, + "eval_samples_per_second": 571.849, + "eval_steps_per_second": 9.174, + "step": 1060 + }, + { + "epoch": 11.0, + "grad_norm": 1.072585105895996, + "learning_rate": 4.4500000000000004e-05, + "loss": 0.123, + "step": 1166 + }, + { + "epoch": 11.0, + "eval_LOCATION_f1": 0.7600000000000001, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7402597402597403, + "eval_LOCATION_recall": 0.7808219178082192, + "eval_ORGANIZATION_f1": 0.7222222222222222, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6582278481012658, + "eval_ORGANIZATION_recall": 0.8, + "eval_PERSON_f1": 0.8945686900958466, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8588957055214724, + "eval_PERSON_recall": 0.9333333333333333, + "eval_QUANTITY_f1": 0.6857142857142856, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.5853658536585366, + "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": 0.14104115962982178, + "eval_overall_accuracy": 0.9574206251538272, + "eval_overall_f1": 0.808, + "eval_overall_precision": 0.7593984962406015, + "eval_overall_recall": 0.8632478632478633, + "eval_runtime": 0.3267, + "eval_samples_per_second": 572.35, + "eval_steps_per_second": 9.182, + "step": 1166 + }, + { + "epoch": 12.0, + "grad_norm": 1.1150060892105103, + "learning_rate": 4.4000000000000006e-05, + "loss": 0.1175, + "step": 1272 + }, + { + "epoch": 12.0, + "eval_LOCATION_f1": 0.7547169811320754, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.6976744186046512, + "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": 0.8817891373801918, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8466257668711656, + "eval_PERSON_recall": 0.92, + "eval_QUANTITY_f1": 0.7272727272727273, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6486486486486487, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.7466666666666667, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.6829268292682927, + "eval_TIME_recall": 0.8235294117647058, + "eval_loss": 0.14132125675678253, + "eval_overall_accuracy": 0.9566822544917548, + "eval_overall_f1": 0.7973684210526317, + "eval_overall_precision": 0.7408312958435208, + "eval_overall_recall": 0.8632478632478633, + "eval_runtime": 0.329, + "eval_samples_per_second": 568.438, + "eval_steps_per_second": 9.119, + "step": 1272 + }, + { + "epoch": 13.0, + "grad_norm": 0.7552273869514465, + "learning_rate": 4.35e-05, + "loss": 0.1117, + "step": 1378 + }, + { + "epoch": 13.0, + "eval_LOCATION_f1": 0.7919463087248322, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7763157894736842, + "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": 0.8938906752411575, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8633540372670807, + "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": 0.7945205479452054, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7435897435897436, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.13354966044425964, + "eval_overall_accuracy": 0.9593896135860202, + "eval_overall_f1": 0.814419225634179, + "eval_overall_precision": 0.7663316582914573, + "eval_overall_recall": 0.8689458689458689, + "eval_runtime": 0.3249, + "eval_samples_per_second": 575.501, + "eval_steps_per_second": 9.233, + "step": 1378 + }, + { + "epoch": 14.0, + "grad_norm": 0.6877084970474243, + "learning_rate": 4.3e-05, + "loss": 0.1119, + "step": 1484 + }, + { + "epoch": 14.0, + "eval_LOCATION_f1": 0.7922077922077922, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7530864197530864, + "eval_LOCATION_recall": 0.8356164383561644, + "eval_ORGANIZATION_f1": 0.7417218543046358, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6511627906976745, + "eval_ORGANIZATION_recall": 0.8615384615384616, + "eval_PERSON_f1": 0.9032258064516129, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.875, + "eval_PERSON_recall": 0.9333333333333333, + "eval_QUANTITY_f1": 0.7384615384615385, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6666666666666666, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.7368421052631577, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.6666666666666666, + "eval_TIME_recall": 0.8235294117647058, + "eval_loss": 0.14005239307880402, + "eval_overall_accuracy": 0.9601279842480925, + "eval_overall_f1": 0.8174603174603174, + "eval_overall_precision": 0.762962962962963, + "eval_overall_recall": 0.8803418803418803, + "eval_runtime": 0.3271, + "eval_samples_per_second": 571.618, + "eval_steps_per_second": 9.17, + "step": 1484 + }, + { + "epoch": 15.0, + "grad_norm": 1.4625396728515625, + "learning_rate": 4.25e-05, + "loss": 0.1091, + "step": 1590 + }, + { + "epoch": 15.0, + "eval_LOCATION_f1": 0.8026315789473685, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7721518987341772, + "eval_LOCATION_recall": 0.8356164383561644, + "eval_ORGANIZATION_f1": 0.7142857142857143, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6179775280898876, + "eval_ORGANIZATION_recall": 0.8461538461538461, + "eval_PERSON_f1": 0.896774193548387, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.86875, + "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": 0.8169014084507041, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7837837837837838, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.12966011464595795, + "eval_overall_accuracy": 0.9601279842480925, + "eval_overall_f1": 0.8202396804260985, + "eval_overall_precision": 0.77, + "eval_overall_recall": 0.8774928774928775, + "eval_runtime": 0.3283, + "eval_samples_per_second": 569.563, + "eval_steps_per_second": 9.137, + "step": 1590 + }, + { + "epoch": 16.0, + "grad_norm": 1.1915158033370972, + "learning_rate": 4.2e-05, + "loss": 0.1012, + "step": 1696 + }, + { + "epoch": 16.0, + "eval_LOCATION_f1": 0.8104575163398693, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.775, + "eval_LOCATION_recall": 0.8493150684931506, + "eval_ORGANIZATION_f1": 0.7162162162162162, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6385542168674698, + "eval_ORGANIZATION_recall": 0.8153846153846154, + "eval_PERSON_f1": 0.8782051282051282, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.845679012345679, + "eval_PERSON_recall": 0.9133333333333333, + "eval_QUANTITY_f1": 0.6944444444444445, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.5813953488372093, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.7567567567567567, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7, + "eval_TIME_recall": 0.8235294117647058, + "eval_loss": 0.14699456095695496, + "eval_overall_accuracy": 0.9539748953974896, + "eval_overall_f1": 0.8036890645586298, + "eval_overall_precision": 0.7475490196078431, + "eval_overall_recall": 0.8689458689458689, + "eval_runtime": 0.3283, + "eval_samples_per_second": 569.593, + "eval_steps_per_second": 9.138, + "step": 1696 + }, + { + "epoch": 17.0, + "grad_norm": 0.5871163010597229, + "learning_rate": 4.15e-05, + "loss": 0.0994, + "step": 1802 + }, + { + "epoch": 17.0, + "eval_LOCATION_f1": 0.7973856209150326, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7625, + "eval_LOCATION_recall": 0.8356164383561644, + "eval_ORGANIZATION_f1": 0.738255033557047, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6547619047619048, + "eval_ORGANIZATION_recall": 0.8461538461538461, + "eval_PERSON_f1": 0.8938906752411575, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8633540372670807, + "eval_PERSON_recall": 0.9266666666666666, + "eval_QUANTITY_f1": 0.7301587301587301, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6764705882352942, + "eval_QUANTITY_recall": 0.7931034482758621, + "eval_TIME_f1": 0.7945205479452054, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7435897435897436, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.1347808837890625, + "eval_overall_accuracy": 0.9601279842480925, + "eval_overall_f1": 0.8197596795727637, + "eval_overall_precision": 0.7713567839195979, + "eval_overall_recall": 0.8746438746438746, + "eval_runtime": 0.3265, + "eval_samples_per_second": 572.774, + "eval_steps_per_second": 9.189, + "step": 1802 + }, + { + "epoch": 18.0, + "grad_norm": 1.2439115047454834, + "learning_rate": 4.1e-05, + "loss": 0.0963, + "step": 1908 + }, + { + "epoch": 18.0, + "eval_LOCATION_f1": 0.794701986754967, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7692307692307693, + "eval_LOCATION_recall": 0.821917808219178, + "eval_ORGANIZATION_f1": 0.691358024691358, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.5773195876288659, + "eval_ORGANIZATION_recall": 0.8615384615384616, + "eval_PERSON_f1": 0.8996763754045306, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8742138364779874, + "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": 0.7945205479452054, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7435897435897436, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.13867075741291046, + "eval_overall_accuracy": 0.9571745015998031, + "eval_overall_f1": 0.8115942028985507, + "eval_overall_precision": 0.7549019607843137, + "eval_overall_recall": 0.8774928774928775, + "eval_runtime": 0.3265, + "eval_samples_per_second": 572.731, + "eval_steps_per_second": 9.188, + "step": 1908 + }, + { + "epoch": 19.0, + "grad_norm": 0.8023453950881958, + "learning_rate": 4.05e-05, + "loss": 0.0929, + "step": 2014 + }, + { + "epoch": 19.0, + "eval_LOCATION_f1": 0.8181818181818182, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7777777777777778, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.7417218543046358, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6511627906976745, + "eval_ORGANIZATION_recall": 0.8615384615384616, + "eval_PERSON_f1": 0.8896103896103896, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8670886075949367, + "eval_PERSON_recall": 0.9133333333333333, + "eval_QUANTITY_f1": 0.7540983606557378, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.71875, + "eval_QUANTITY_recall": 0.7931034482758621, + "eval_TIME_f1": 0.8333333333333333, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7894736842105263, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.12837004661560059, + "eval_overall_accuracy": 0.9623430962343096, + "eval_overall_f1": 0.8284182305630027, + "eval_overall_precision": 0.7822784810126582, + "eval_overall_recall": 0.8803418803418803, + "eval_runtime": 0.3274, + "eval_samples_per_second": 571.175, + "eval_steps_per_second": 9.163, + "step": 2014 + }, + { + "epoch": 20.0, + "grad_norm": 0.7020606398582458, + "learning_rate": 4e-05, + "loss": 0.0925, + "step": 2120 + }, + { + "epoch": 20.0, + "eval_LOCATION_f1": 0.8051948051948052, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7654320987654321, + "eval_LOCATION_recall": 0.8493150684931506, + "eval_ORGANIZATION_f1": 0.7066666666666667, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6235294117647059, + "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": 0.7352941176470588, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6410256410256411, + "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": 0.14086037874221802, + "eval_overall_accuracy": 0.9591434900319961, + "eval_overall_f1": 0.8169761273209548, + "eval_overall_precision": 0.7642679900744417, + "eval_overall_recall": 0.8774928774928775, + "eval_runtime": 0.3282, + "eval_samples_per_second": 569.838, + "eval_steps_per_second": 9.142, + "step": 2120 + }, + { + "epoch": 21.0, + "grad_norm": 1.0593703985214233, + "learning_rate": 3.9500000000000005e-05, + "loss": 0.0864, + "step": 2226 + }, + { + "epoch": 21.0, + "eval_LOCATION_f1": 0.7922077922077922, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7530864197530864, + "eval_LOCATION_recall": 0.8356164383561644, + "eval_ORGANIZATION_f1": 0.7466666666666668, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6588235294117647, + "eval_ORGANIZATION_recall": 0.8615384615384616, + "eval_PERSON_f1": 0.8954248366013071, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8782051282051282, + "eval_PERSON_recall": 0.9133333333333333, + "eval_QUANTITY_f1": 0.7619047619047619, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7058823529411765, + "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": 0.13228537142276764, + "eval_overall_accuracy": 0.9625892197883338, + "eval_overall_f1": 0.8257372654155497, + "eval_overall_precision": 0.779746835443038, + "eval_overall_recall": 0.8774928774928775, + "eval_runtime": 0.3271, + "eval_samples_per_second": 571.774, + "eval_steps_per_second": 9.173, + "step": 2226 + }, + { + "epoch": 22.0, + "grad_norm": 1.3143647909164429, + "learning_rate": 3.9000000000000006e-05, + "loss": 0.0907, + "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": 0.7450980392156863, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6477272727272727, + "eval_ORGANIZATION_recall": 0.8769230769230769, + "eval_PERSON_f1": 0.896774193548387, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.86875, + "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": 0.7368421052631577, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.6666666666666666, + "eval_TIME_recall": 0.8235294117647058, + "eval_loss": 0.1397896260023117, + "eval_overall_accuracy": 0.9603741078021166, + "eval_overall_f1": 0.8190224570673711, + "eval_overall_precision": 0.7635467980295566, + "eval_overall_recall": 0.8831908831908832, + "eval_runtime": 0.3262, + "eval_samples_per_second": 573.244, + "eval_steps_per_second": 9.196, + "step": 2332 + }, + { + "epoch": 23.0, + "grad_norm": 1.1894057989120483, + "learning_rate": 3.85e-05, + "loss": 0.0852, + "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": 0.7417218543046358, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6511627906976745, + "eval_ORGANIZATION_recall": 0.8615384615384616, + "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": 0.8055555555555555, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7631578947368421, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.1318092942237854, + "eval_overall_accuracy": 0.9628353433423579, + "eval_overall_f1": 0.8284182305630027, + "eval_overall_precision": 0.7822784810126582, + "eval_overall_recall": 0.8803418803418803, + "eval_runtime": 0.3269, + "eval_samples_per_second": 572.048, + "eval_steps_per_second": 9.177, + "step": 2438 + }, + { + "epoch": 24.0, + "grad_norm": 1.077348232269287, + "learning_rate": 3.8e-05, + "loss": 0.0837, + "step": 2544 + }, + { + "epoch": 24.0, + "eval_LOCATION_f1": 0.8104575163398693, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.775, + "eval_LOCATION_recall": 0.8493150684931506, + "eval_ORGANIZATION_f1": 0.717948717948718, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6153846153846154, + "eval_ORGANIZATION_recall": 0.8615384615384616, + "eval_PERSON_f1": 0.8932038834951457, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8679245283018868, + "eval_PERSON_recall": 0.92, + "eval_QUANTITY_f1": 0.7741935483870968, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7272727272727273, + "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": 0.14012350142002106, + "eval_overall_accuracy": 0.9601279842480925, + "eval_overall_f1": 0.8218085106382977, + "eval_overall_precision": 0.770573566084788, + "eval_overall_recall": 0.8803418803418803, + "eval_runtime": 0.325, + "eval_samples_per_second": 575.406, + "eval_steps_per_second": 9.231, + "step": 2544 + }, + { + "epoch": 25.0, + "grad_norm": 1.0630335807800293, + "learning_rate": 3.7500000000000003e-05, + "loss": 0.0826, + "step": 2650 + }, + { + "epoch": 25.0, + "eval_LOCATION_f1": 0.8026315789473685, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7721518987341772, + "eval_LOCATION_recall": 0.8356164383561644, + "eval_ORGANIZATION_f1": 0.7891156462585034, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7073170731707317, + "eval_ORGANIZATION_recall": 0.8923076923076924, + "eval_PERSON_f1": 0.8996763754045306, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8742138364779874, + "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": 0.7945205479452054, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7435897435897436, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.14586162567138672, + "eval_overall_accuracy": 0.961112478464189, + "eval_overall_f1": 0.8326639892904953, + "eval_overall_precision": 0.7853535353535354, + "eval_overall_recall": 0.886039886039886, + "eval_runtime": 0.3256, + "eval_samples_per_second": 574.315, + "eval_steps_per_second": 9.214, + "step": 2650 + }, + { + "epoch": 26.0, + "grad_norm": 0.6446018815040588, + "learning_rate": 3.7e-05, + "loss": 0.082, + "step": 2756 + }, + { + "epoch": 26.0, + "eval_LOCATION_f1": 0.7866666666666666, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7662337662337663, + "eval_LOCATION_recall": 0.8082191780821918, + "eval_ORGANIZATION_f1": 0.7368421052631579, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6436781609195402, + "eval_ORGANIZATION_recall": 0.8615384615384616, + "eval_PERSON_f1": 0.8996763754045306, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8742138364779874, + "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": 0.8493150684931507, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7948717948717948, + "eval_TIME_recall": 0.9117647058823529, + "eval_loss": 0.13936233520507812, + "eval_overall_accuracy": 0.9608663549101649, + "eval_overall_f1": 0.8262032085561498, + "eval_overall_precision": 0.7783375314861462, + "eval_overall_recall": 0.8803418803418803, + "eval_runtime": 0.3272, + "eval_samples_per_second": 571.535, + "eval_steps_per_second": 9.169, + "step": 2756 + }, + { + "epoch": 27.0, + "grad_norm": 0.2514190077781677, + "learning_rate": 3.65e-05, + "loss": 0.0787, + "step": 2862 + }, + { + "epoch": 27.0, + "eval_LOCATION_f1": 0.8026315789473685, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7721518987341772, + "eval_LOCATION_recall": 0.8356164383561644, + "eval_ORGANIZATION_f1": 0.728476821192053, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6395348837209303, + "eval_ORGANIZATION_recall": 0.8461538461538461, + "eval_PERSON_f1": 0.9032258064516129, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.875, + "eval_PERSON_recall": 0.9333333333333333, + "eval_QUANTITY_f1": 0.6666666666666666, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.575, + "eval_QUANTITY_recall": 0.7931034482758621, + "eval_TIME_f1": 0.7466666666666667, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.6829268292682927, + "eval_TIME_recall": 0.8235294117647058, + "eval_loss": 0.14520442485809326, + "eval_overall_accuracy": 0.9576667487078513, + "eval_overall_f1": 0.8110964332892999, + "eval_overall_precision": 0.7561576354679803, + "eval_overall_recall": 0.8746438746438746, + "eval_runtime": 0.3263, + "eval_samples_per_second": 573.135, + "eval_steps_per_second": 9.195, + "step": 2862 + }, + { + "epoch": 28.0, + "grad_norm": 1.3224049806594849, + "learning_rate": 3.6e-05, + "loss": 0.0797, + "step": 2968 + }, + { + "epoch": 28.0, + "eval_LOCATION_f1": 0.8026315789473685, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7721518987341772, + "eval_LOCATION_recall": 0.8356164383561644, + "eval_ORGANIZATION_f1": 0.7832167832167831, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.717948717948718, + "eval_ORGANIZATION_recall": 0.8615384615384616, + "eval_PERSON_f1": 0.9061488673139159, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8805031446540881, + "eval_PERSON_recall": 0.9333333333333333, + "eval_QUANTITY_f1": 0.7540983606557378, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.71875, + "eval_QUANTITY_recall": 0.7931034482758621, + "eval_TIME_f1": 0.7466666666666667, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.6829268292682927, + "eval_TIME_recall": 0.8235294117647058, + "eval_loss": 0.14338870346546173, + "eval_overall_accuracy": 0.9613586020182131, + "eval_overall_f1": 0.8324324324324326, + "eval_overall_precision": 0.7917737789203085, + "eval_overall_recall": 0.8774928774928775, + "eval_runtime": 0.3258, + "eval_samples_per_second": 573.892, + "eval_steps_per_second": 9.207, + "step": 2968 + }, + { + "epoch": 29.0, + "grad_norm": 1.5945953130722046, + "learning_rate": 3.55e-05, + "loss": 0.0743, + "step": 3074 + }, + { + "epoch": 29.0, + "eval_LOCATION_f1": 0.8157894736842105, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7848101265822784, + "eval_LOCATION_recall": 0.8493150684931506, + "eval_ORGANIZATION_f1": 0.7682119205298013, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6744186046511628, + "eval_ORGANIZATION_recall": 0.8923076923076924, + "eval_PERSON_f1": 0.9025974025974027, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.879746835443038, + "eval_PERSON_recall": 0.9266666666666666, + "eval_QUANTITY_f1": 0.7868852459016394, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.75, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.7837837837837837, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.725, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.13935725390911102, + "eval_overall_accuracy": 0.9623430962343096, + "eval_overall_f1": 0.8364611260053618, + "eval_overall_precision": 0.789873417721519, + "eval_overall_recall": 0.8888888888888888, + "eval_runtime": 0.3259, + "eval_samples_per_second": 573.725, + "eval_steps_per_second": 9.204, + "step": 3074 + }, + { + "epoch": 30.0, + "grad_norm": 3.0987069606781006, + "learning_rate": 3.5e-05, + "loss": 0.0748, + "step": 3180 + }, + { + "epoch": 30.0, + "eval_LOCATION_f1": 0.8133333333333332, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7922077922077922, + "eval_LOCATION_recall": 0.8356164383561644, + "eval_ORGANIZATION_f1": 0.7432432432432433, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6626506024096386, + "eval_ORGANIZATION_recall": 0.8461538461538461, + "eval_PERSON_f1": 0.8802588996763754, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8553459119496856, + "eval_PERSON_recall": 0.9066666666666666, + "eval_QUANTITY_f1": 0.7419354838709677, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.696969696969697, + "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": 0.13977107405662537, + "eval_overall_accuracy": 0.9591434900319961, + "eval_overall_f1": 0.8183041722745626, + "eval_overall_precision": 0.7755102040816326, + "eval_overall_recall": 0.8660968660968661, + "eval_runtime": 0.3263, + "eval_samples_per_second": 573.058, + "eval_steps_per_second": 9.193, + "step": 3180 + }, + { + "epoch": 31.0, + "grad_norm": 1.0805306434631348, + "learning_rate": 3.45e-05, + "loss": 0.0749, + "step": 3286 + }, + { + "epoch": 31.0, + "eval_LOCATION_f1": 0.8133333333333332, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7922077922077922, + "eval_LOCATION_recall": 0.8356164383561644, + "eval_ORGANIZATION_f1": 0.7724137931034483, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7, + "eval_ORGANIZATION_recall": 0.8615384615384616, + "eval_PERSON_f1": 0.9061488673139159, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8805031446540881, + "eval_PERSON_recall": 0.9333333333333333, + "eval_QUANTITY_f1": 0.696969696969697, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6216216216216216, + "eval_QUANTITY_recall": 0.7931034482758621, + "eval_TIME_f1": 0.7466666666666667, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.6829268292682927, + "eval_TIME_recall": 0.8235294117647058, + "eval_loss": 0.14538507163524628, + "eval_overall_accuracy": 0.9596357371400444, + "eval_overall_f1": 0.8268456375838926, + "eval_overall_precision": 0.7817258883248731, + "eval_overall_recall": 0.8774928774928775, + "eval_runtime": 0.3282, + "eval_samples_per_second": 569.812, + "eval_steps_per_second": 9.141, + "step": 3286 + }, + { + "epoch": 32.0, + "grad_norm": 2.186657667160034, + "learning_rate": 3.4000000000000007e-05, + "loss": 0.0702, + "step": 3392 + }, + { + "epoch": 32.0, + "eval_LOCATION_f1": 0.8289473684210525, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7974683544303798, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.7692307692307692, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6593406593406593, + "eval_ORGANIZATION_recall": 0.9230769230769231, + "eval_PERSON_f1": 0.896103896103896, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8734177215189873, + "eval_PERSON_recall": 0.92, + "eval_QUANTITY_f1": 0.7619047619047619, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7058823529411765, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.7368421052631577, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.6666666666666666, + "eval_TIME_recall": 0.8235294117647058, + "eval_loss": 0.15071968734264374, + "eval_overall_accuracy": 0.9581589958158996, + "eval_overall_f1": 0.8291390728476822, + "eval_overall_precision": 0.7747524752475248, + "eval_overall_recall": 0.8917378917378918, + "eval_runtime": 0.3267, + "eval_samples_per_second": 572.334, + "eval_steps_per_second": 9.182, + "step": 3392 + }, + { + "epoch": 33.0, + "grad_norm": 0.7016972303390503, + "learning_rate": 3.35e-05, + "loss": 0.0687, + "step": 3498 + }, + { + "epoch": 33.0, + "eval_LOCATION_f1": 0.794701986754967, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7692307692307693, + "eval_LOCATION_recall": 0.821917808219178, + "eval_ORGANIZATION_f1": 0.7746478873239436, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7142857142857143, + "eval_ORGANIZATION_recall": 0.8461538461538461, + "eval_PERSON_f1": 0.896774193548387, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.86875, + "eval_PERSON_recall": 0.9266666666666666, + "eval_QUANTITY_f1": 0.8064516129032258, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7575757575757576, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.7368421052631577, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.6666666666666666, + "eval_TIME_recall": 0.8235294117647058, + "eval_loss": 0.14618100225925446, + "eval_overall_accuracy": 0.9601279842480925, + "eval_overall_f1": 0.8286099865047234, + "eval_overall_precision": 0.7871794871794872, + "eval_overall_recall": 0.8746438746438746, + "eval_runtime": 0.3282, + "eval_samples_per_second": 569.718, + "eval_steps_per_second": 9.14, + "step": 3498 + }, + { + "epoch": 34.0, + "grad_norm": 1.0766340494155884, + "learning_rate": 3.3e-05, + "loss": 0.0718, + "step": 3604 + }, + { + "epoch": 34.0, + "eval_LOCATION_f1": 0.8266666666666668, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8051948051948052, + "eval_LOCATION_recall": 0.8493150684931506, + "eval_ORGANIZATION_f1": 0.7682119205298013, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6744186046511628, + "eval_ORGANIZATION_recall": 0.8923076923076924, + "eval_PERSON_f1": 0.896103896103896, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8734177215189873, + "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": 0.7567567567567567, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7, + "eval_TIME_recall": 0.8235294117647058, + "eval_loss": 0.14720623195171356, + "eval_overall_accuracy": 0.9608663549101649, + "eval_overall_f1": 0.8299866131191433, + "eval_overall_precision": 0.7828282828282829, + "eval_overall_recall": 0.8831908831908832, + "eval_runtime": 0.3274, + "eval_samples_per_second": 571.163, + "eval_steps_per_second": 9.163, + "step": 3604 + }, + { + "epoch": 35.0, + "grad_norm": 0.6112431287765503, + "learning_rate": 3.2500000000000004e-05, + "loss": 0.0696, + "step": 3710 + }, + { + "epoch": 35.0, + "eval_LOCATION_f1": 0.7973856209150326, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7625, + "eval_LOCATION_recall": 0.8356164383561644, + "eval_ORGANIZATION_f1": 0.7866666666666665, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6941176470588235, + "eval_ORGANIZATION_recall": 0.9076923076923077, + "eval_PERSON_f1": 0.9032258064516129, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.875, + "eval_PERSON_recall": 0.9333333333333333, + "eval_QUANTITY_f1": 0.7384615384615385, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6666666666666666, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.7733333333333334, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7073170731707317, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.14851583540439606, + "eval_overall_accuracy": 0.9603741078021166, + "eval_overall_f1": 0.8313413014608234, + "eval_overall_precision": 0.7786069651741293, + "eval_overall_recall": 0.8917378917378918, + "eval_runtime": 0.3257, + "eval_samples_per_second": 574.18, + "eval_steps_per_second": 9.211, + "step": 3710 + }, + { + "epoch": 36.0, + "grad_norm": 0.9571568369865417, + "learning_rate": 3.2000000000000005e-05, + "loss": 0.0681, + "step": 3816 + }, + { + "epoch": 36.0, + "eval_LOCATION_f1": 0.8456375838926175, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8289473684210527, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.7792207792207791, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6741573033707865, + "eval_ORGANIZATION_recall": 0.9230769230769231, + "eval_PERSON_f1": 0.8996763754045306, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8742138364779874, + "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": 0.717948717948718, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.6363636363636364, + "eval_TIME_recall": 0.8235294117647058, + "eval_loss": 0.14458392560482025, + "eval_overall_accuracy": 0.9598818606940684, + "eval_overall_f1": 0.8306878306878307, + "eval_overall_precision": 0.7753086419753087, + "eval_overall_recall": 0.8945868945868946, + "eval_runtime": 0.3254, + "eval_samples_per_second": 574.594, + "eval_steps_per_second": 9.218, + "step": 3816 + }, + { + "epoch": 37.0, + "grad_norm": 1.0726656913757324, + "learning_rate": 3.15e-05, + "loss": 0.0666, + "step": 3922 + }, + { + "epoch": 37.0, + "eval_LOCATION_f1": 0.8079470198675496, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.782051282051282, + "eval_LOCATION_recall": 0.8356164383561644, + "eval_ORGANIZATION_f1": 0.7547169811320755, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6382978723404256, + "eval_ORGANIZATION_recall": 0.9230769230769231, + "eval_PERSON_f1": 0.8903225806451613, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8625, + "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": 0.7272727272727273, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.6511627906976745, + "eval_TIME_recall": 0.8235294117647058, + "eval_loss": 0.15450456738471985, + "eval_overall_accuracy": 0.9566822544917548, + "eval_overall_f1": 0.8173455978975033, + "eval_overall_precision": 0.7585365853658537, + "eval_overall_recall": 0.886039886039886, + "eval_runtime": 0.327, + "eval_samples_per_second": 571.922, + "eval_steps_per_second": 9.175, + "step": 3922 + }, + { + "epoch": 38.0, + "grad_norm": 0.4657502770423889, + "learning_rate": 3.1e-05, + "loss": 0.0666, + "step": 4028 + }, + { + "epoch": 38.0, + "eval_LOCATION_f1": 0.8299319727891157, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8243243243243243, + "eval_LOCATION_recall": 0.8356164383561644, + "eval_ORGANIZATION_f1": 0.7972027972027972, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7307692307692307, + "eval_ORGANIZATION_recall": 0.8769230769230769, + "eval_PERSON_f1": 0.896103896103896, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8734177215189873, + "eval_PERSON_recall": 0.92, + "eval_QUANTITY_f1": 0.7076923076923076, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6388888888888888, + "eval_QUANTITY_recall": 0.7931034482758621, + "eval_TIME_f1": 0.7945205479452054, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7435897435897436, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.1448453813791275, + "eval_overall_accuracy": 0.9618508491262614, + "eval_overall_f1": 0.8369565217391305, + "eval_overall_precision": 0.8, + "eval_overall_recall": 0.8774928774928775, + "eval_runtime": 0.3252, + "eval_samples_per_second": 575.07, + "eval_steps_per_second": 9.226, + "step": 4028 + }, + { + "epoch": 39.0, + "grad_norm": 2.360812187194824, + "learning_rate": 3.05e-05, + "loss": 0.0642, + "step": 4134 + }, + { + "epoch": 39.0, + "eval_LOCATION_f1": 0.8157894736842105, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7848101265822784, + "eval_LOCATION_recall": 0.8493150684931506, + "eval_ORGANIZATION_f1": 0.7947019867549668, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6976744186046512, + "eval_ORGANIZATION_recall": 0.9230769230769231, + "eval_PERSON_f1": 0.90032154340836, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8695652173913043, + "eval_PERSON_recall": 0.9333333333333333, + "eval_QUANTITY_f1": 0.8196721311475409, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.78125, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.7567567567567567, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7, + "eval_TIME_recall": 0.8235294117647058, + "eval_loss": 0.14753280580043793, + "eval_overall_accuracy": 0.9616047255722373, + "eval_overall_f1": 0.8411214953271028, + "eval_overall_precision": 0.7914572864321608, + "eval_overall_recall": 0.8974358974358975, + "eval_runtime": 0.3264, + "eval_samples_per_second": 572.953, + "eval_steps_per_second": 9.192, + "step": 4134 + }, + { + "epoch": 40.0, + "grad_norm": 1.1762744188308716, + "learning_rate": 3e-05, + "loss": 0.0653, + "step": 4240 + }, + { + "epoch": 40.0, + "eval_LOCATION_f1": 0.847682119205298, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8205128205128205, + "eval_LOCATION_recall": 0.8767123287671232, + "eval_ORGANIZATION_f1": 0.7361963190184049, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6122448979591837, + "eval_ORGANIZATION_recall": 0.9230769230769231, + "eval_PERSON_f1": 0.8918032786885245, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8774193548387097, + "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": 0.7466666666666667, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.6829268292682927, + "eval_TIME_recall": 0.8235294117647058, + "eval_loss": 0.1560455709695816, + "eval_overall_accuracy": 0.9579128722618755, + "eval_overall_f1": 0.8269484808454426, + "eval_overall_precision": 0.770935960591133, + "eval_overall_recall": 0.8917378917378918, + "eval_runtime": 0.3264, + "eval_samples_per_second": 572.926, + "eval_steps_per_second": 9.191, + "step": 4240 + }, + { + "epoch": 41.0, + "grad_norm": 0.4446096122264862, + "learning_rate": 2.95e-05, + "loss": 0.0621, + "step": 4346 + }, + { + "epoch": 41.0, + "eval_LOCATION_f1": 0.8079470198675496, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.782051282051282, + "eval_LOCATION_recall": 0.8356164383561644, + "eval_ORGANIZATION_f1": 0.7450980392156863, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6477272727272727, + "eval_ORGANIZATION_recall": 0.8769230769230769, + "eval_PERSON_f1": 0.8932038834951457, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8679245283018868, + "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": 0.7466666666666667, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.6829268292682927, + "eval_TIME_recall": 0.8235294117647058, + "eval_loss": 0.14887475967407227, + "eval_overall_accuracy": 0.958897366477972, + "eval_overall_f1": 0.824, + "eval_overall_precision": 0.7744360902255639, + "eval_overall_recall": 0.8803418803418803, + "eval_runtime": 0.3274, + "eval_samples_per_second": 571.082, + "eval_steps_per_second": 9.162, + "step": 4346 + }, + { + "epoch": 42.0, + "grad_norm": 0.19456997513771057, + "learning_rate": 2.9e-05, + "loss": 0.0621, + "step": 4452 + }, + { + "epoch": 42.0, + "eval_LOCATION_f1": 0.8211920529801324, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7948717948717948, + "eval_LOCATION_recall": 0.8493150684931506, + "eval_ORGANIZATION_f1": 0.7808219178082192, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7037037037037037, + "eval_ORGANIZATION_recall": 0.8769230769230769, + "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": 0.7272727272727273, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.6511627906976745, + "eval_TIME_recall": 0.8235294117647058, + "eval_loss": 0.14638268947601318, + "eval_overall_accuracy": 0.961112478464189, + "eval_overall_f1": 0.8299866131191433, + "eval_overall_precision": 0.7828282828282829, + "eval_overall_recall": 0.8831908831908832, + "eval_runtime": 0.3258, + "eval_samples_per_second": 573.908, + "eval_steps_per_second": 9.207, + "step": 4452 + }, + { + "epoch": 43.0, + "grad_norm": 0.6415175795555115, + "learning_rate": 2.8499999999999998e-05, + "loss": 0.059, + "step": 4558 + }, + { + "epoch": 43.0, + "eval_LOCATION_f1": 0.8157894736842105, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7848101265822784, + "eval_LOCATION_recall": 0.8493150684931506, + "eval_ORGANIZATION_f1": 0.7916666666666667, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7215189873417721, + "eval_ORGANIZATION_recall": 0.8769230769230769, + "eval_PERSON_f1": 0.8867313915857605, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8616352201257862, + "eval_PERSON_recall": 0.9133333333333333, + "eval_QUANTITY_f1": 0.71875, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6571428571428571, + "eval_QUANTITY_recall": 0.7931034482758621, + "eval_TIME_f1": 0.7466666666666667, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.6829268292682927, + "eval_TIME_recall": 0.8235294117647058, + "eval_loss": 0.14816471934318542, + "eval_overall_accuracy": 0.9601279842480925, + "eval_overall_f1": 0.825268817204301, + "eval_overall_precision": 0.7811704834605598, + "eval_overall_recall": 0.8746438746438746, + "eval_runtime": 0.3261, + "eval_samples_per_second": 573.371, + "eval_steps_per_second": 9.198, + "step": 4558 + }, + { + "epoch": 44.0, + "grad_norm": 1.0841795206069946, + "learning_rate": 2.8000000000000003e-05, + "loss": 0.061, + "step": 4664 + }, + { + "epoch": 44.0, + "eval_LOCATION_f1": 0.8266666666666668, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8051948051948052, + "eval_LOCATION_recall": 0.8493150684931506, + "eval_ORGANIZATION_f1": 0.8, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.725, + "eval_ORGANIZATION_recall": 0.8923076923076924, + "eval_PERSON_f1": 0.8938906752411575, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8633540372670807, + "eval_PERSON_recall": 0.9266666666666666, + "eval_QUANTITY_f1": 0.7384615384615385, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6666666666666666, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.7368421052631577, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.6666666666666666, + "eval_TIME_recall": 0.8235294117647058, + "eval_loss": 0.1557927429676056, + "eval_overall_accuracy": 0.9603741078021166, + "eval_overall_f1": 0.8326639892904953, + "eval_overall_precision": 0.7853535353535354, + "eval_overall_recall": 0.886039886039886, + "eval_runtime": 0.3272, + "eval_samples_per_second": 571.486, + "eval_steps_per_second": 9.168, + "step": 4664 + }, + { + "epoch": 45.0, + "grad_norm": 7.833734512329102, + "learning_rate": 2.7500000000000004e-05, + "loss": 0.0623, + "step": 4770 + }, + { + "epoch": 45.0, + "eval_LOCATION_f1": 0.8344370860927152, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8076923076923077, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.7857142857142856, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7333333333333333, + "eval_ORGANIZATION_recall": 0.8461538461538461, + "eval_PERSON_f1": 0.896774193548387, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.86875, + "eval_PERSON_recall": 0.9266666666666666, + "eval_QUANTITY_f1": 0.71875, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6571428571428571, + "eval_QUANTITY_recall": 0.7931034482758621, + "eval_TIME_f1": 0.7945205479452054, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7435897435897436, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.1500028669834137, + "eval_overall_accuracy": 0.963081466896382, + "eval_overall_f1": 0.8373983739837397, + "eval_overall_precision": 0.7984496124031008, + "eval_overall_recall": 0.8803418803418803, + "eval_runtime": 0.3263, + "eval_samples_per_second": 573.014, + "eval_steps_per_second": 9.193, + "step": 4770 + }, + { + "epoch": 46.0, + "grad_norm": 0.6496893763542175, + "learning_rate": 2.7000000000000002e-05, + "loss": 0.0585, + "step": 4876 + }, + { + "epoch": 46.0, + "eval_LOCATION_f1": 0.8400000000000001, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8181818181818182, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.7746478873239436, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7142857142857143, + "eval_ORGANIZATION_recall": 0.8461538461538461, + "eval_PERSON_f1": 0.8867313915857605, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8616352201257862, + "eval_PERSON_recall": 0.9133333333333333, + "eval_QUANTITY_f1": 0.71875, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6571428571428571, + "eval_QUANTITY_recall": 0.7931034482758621, + "eval_TIME_f1": 0.7567567567567567, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7, + "eval_TIME_recall": 0.8235294117647058, + "eval_loss": 0.1510009616613388, + "eval_overall_accuracy": 0.9620969726802855, + "eval_overall_f1": 0.8281461434370772, + "eval_overall_precision": 0.788659793814433, + "eval_overall_recall": 0.8717948717948718, + "eval_runtime": 0.3253, + "eval_samples_per_second": 574.888, + "eval_steps_per_second": 9.223, + "step": 4876 + }, + { + "epoch": 47.0, + "grad_norm": 1.4146417379379272, + "learning_rate": 2.6500000000000004e-05, + "loss": 0.0546, + "step": 4982 + }, + { + "epoch": 47.0, + "eval_LOCATION_f1": 0.8211920529801324, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7948717948717948, + "eval_LOCATION_recall": 0.8493150684931506, + "eval_ORGANIZATION_f1": 0.75, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6551724137931034, + "eval_ORGANIZATION_recall": 0.8769230769230769, + "eval_PERSON_f1": 0.8938906752411575, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8633540372670807, + "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": 0.7567567567567567, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7, + "eval_TIME_recall": 0.8235294117647058, + "eval_loss": 0.15528370440006256, + "eval_overall_accuracy": 0.9603741078021166, + "eval_overall_f1": 0.8255659121171771, + "eval_overall_precision": 0.775, + "eval_overall_recall": 0.8831908831908832, + "eval_runtime": 0.3262, + "eval_samples_per_second": 573.227, + "eval_steps_per_second": 9.196, + "step": 4982 + }, + { + "epoch": 48.0, + "grad_norm": 6.046639442443848, + "learning_rate": 2.6000000000000002e-05, + "loss": 0.0581, + "step": 5088 + }, + { + "epoch": 48.0, + "eval_LOCATION_f1": 0.8104575163398693, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.775, + "eval_LOCATION_recall": 0.8493150684931506, + "eval_ORGANIZATION_f1": 0.7651006711409396, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6785714285714286, + "eval_ORGANIZATION_recall": 0.8769230769230769, + "eval_PERSON_f1": 0.8838709677419355, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.85625, + "eval_PERSON_recall": 0.9133333333333333, + "eval_QUANTITY_f1": 0.7384615384615385, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6666666666666666, + "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": 0.15824827551841736, + "eval_overall_accuracy": 0.9593896135860202, + "eval_overall_f1": 0.8213333333333335, + "eval_overall_precision": 0.7719298245614035, + "eval_overall_recall": 0.8774928774928775, + "eval_runtime": 0.3269, + "eval_samples_per_second": 572.034, + "eval_steps_per_second": 9.177, + "step": 5088 + }, + { + "epoch": 49.0, + "grad_norm": 0.2904468774795532, + "learning_rate": 2.5500000000000003e-05, + "loss": 0.0569, + "step": 5194 + }, + { + "epoch": 49.0, + "eval_LOCATION_f1": 0.8211920529801324, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7948717948717948, + "eval_LOCATION_recall": 0.8493150684931506, + "eval_ORGANIZATION_f1": 0.7862068965517242, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7125, + "eval_ORGANIZATION_recall": 0.8769230769230769, + "eval_PERSON_f1": 0.9025974025974027, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.879746835443038, + "eval_PERSON_recall": 0.9266666666666666, + "eval_QUANTITY_f1": 0.7419354838709677, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.696969696969697, + "eval_QUANTITY_recall": 0.7931034482758621, + "eval_TIME_f1": 0.7733333333333334, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7073170731707317, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.16139063239097595, + "eval_overall_accuracy": 0.9618508491262614, + "eval_overall_f1": 0.8367071524966261, + "eval_overall_precision": 0.7948717948717948, + "eval_overall_recall": 0.8831908831908832, + "eval_runtime": 0.3275, + "eval_samples_per_second": 571.063, + "eval_steps_per_second": 9.161, + "step": 5194 + }, + { + "epoch": 50.0, + "grad_norm": 1.9106760025024414, + "learning_rate": 2.5e-05, + "loss": 0.0569, + "step": 5300 + }, + { + "epoch": 50.0, + "eval_LOCATION_f1": 0.8133333333333332, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7922077922077922, + "eval_LOCATION_recall": 0.8356164383561644, + "eval_ORGANIZATION_f1": 0.7894736842105263, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6896551724137931, + "eval_ORGANIZATION_recall": 0.9230769230769231, + "eval_PERSON_f1": 0.8932038834951457, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8679245283018868, + "eval_PERSON_recall": 0.92, + "eval_QUANTITY_f1": 0.7741935483870968, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7272727272727273, + "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": 0.15711840987205505, + "eval_overall_accuracy": 0.9598818606940684, + "eval_overall_f1": 0.8326639892904953, + "eval_overall_precision": 0.7853535353535354, + "eval_overall_recall": 0.886039886039886, + "eval_runtime": 0.3259, + "eval_samples_per_second": 573.738, + "eval_steps_per_second": 9.204, + "step": 5300 + }, + { + "epoch": 51.0, + "grad_norm": 0.6335307359695435, + "learning_rate": 2.45e-05, + "loss": 0.0534, + "step": 5406 + }, + { + "epoch": 51.0, + "eval_LOCATION_f1": 0.7999999999999999, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7560975609756098, + "eval_LOCATION_recall": 0.8493150684931506, + "eval_ORGANIZATION_f1": 0.7737226277372263, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7361111111111112, + "eval_ORGANIZATION_recall": 0.8153846153846154, + "eval_PERSON_f1": 0.8782051282051282, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.845679012345679, + "eval_PERSON_recall": 0.9133333333333333, + "eval_QUANTITY_f1": 0.7619047619047619, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7058823529411765, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.7945205479452054, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7435897435897436, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.15570282936096191, + "eval_overall_accuracy": 0.9616047255722373, + "eval_overall_f1": 0.8243243243243243, + "eval_overall_precision": 0.7840616966580977, + "eval_overall_recall": 0.8689458689458689, + "eval_runtime": 0.3279, + "eval_samples_per_second": 570.257, + "eval_steps_per_second": 9.149, + "step": 5406 + }, + { + "epoch": 52.0, + "grad_norm": 0.4868054687976837, + "learning_rate": 2.4e-05, + "loss": 0.0546, + "step": 5512 + }, + { + "epoch": 52.0, + "eval_LOCATION_f1": 0.8026315789473685, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7721518987341772, + "eval_LOCATION_recall": 0.8356164383561644, + "eval_ORGANIZATION_f1": 0.7643312101910829, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6521739130434783, + "eval_ORGANIZATION_recall": 0.9230769230769231, + "eval_PERSON_f1": 0.887459807073955, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8571428571428571, + "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": 0.7631578947368423, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.6904761904761905, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.17388612031936646, + "eval_overall_accuracy": 0.9544671425055378, + "eval_overall_f1": 0.8210526315789473, + "eval_overall_precision": 0.7628361858190709, + "eval_overall_recall": 0.8888888888888888, + "eval_runtime": 0.3262, + "eval_samples_per_second": 573.302, + "eval_steps_per_second": 9.197, + "step": 5512 + }, + { + "epoch": 53.0, + "grad_norm": 0.8087223172187805, + "learning_rate": 2.35e-05, + "loss": 0.0508, + "step": 5618 + }, + { + "epoch": 53.0, + "eval_LOCATION_f1": 0.8211920529801324, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7948717948717948, + "eval_LOCATION_recall": 0.8493150684931506, + "eval_ORGANIZATION_f1": 0.7564102564102564, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6483516483516484, + "eval_ORGANIZATION_recall": 0.9076923076923077, + "eval_PERSON_f1": 0.8910256410256411, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8580246913580247, + "eval_PERSON_recall": 0.9266666666666666, + "eval_QUANTITY_f1": 0.71875, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6571428571428571, + "eval_QUANTITY_recall": 0.7931034482758621, + "eval_TIME_f1": 0.8055555555555555, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7631578947368421, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.16712214052677155, + "eval_overall_accuracy": 0.9549593896135861, + "eval_overall_f1": 0.8264900662251655, + "eval_overall_precision": 0.7722772277227723, + "eval_overall_recall": 0.8888888888888888, + "eval_runtime": 0.3269, + "eval_samples_per_second": 572.089, + "eval_steps_per_second": 9.178, + "step": 5618 + }, + { + "epoch": 54.0, + "grad_norm": 1.31313157081604, + "learning_rate": 2.3000000000000003e-05, + "loss": 0.0534, + "step": 5724 + }, + { + "epoch": 54.0, + "eval_LOCATION_f1": 0.8157894736842105, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7848101265822784, + "eval_LOCATION_recall": 0.8493150684931506, + "eval_ORGANIZATION_f1": 0.7515923566878981, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6413043478260869, + "eval_ORGANIZATION_recall": 0.9076923076923077, + "eval_PERSON_f1": 0.8938906752411575, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8633540372670807, + "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": 0.7733333333333334, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7073170731707317, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.16759367287158966, + "eval_overall_accuracy": 0.9556977602756583, + "eval_overall_f1": 0.8274044795783927, + "eval_overall_precision": 0.7696078431372549, + "eval_overall_recall": 0.8945868945868946, + "eval_runtime": 0.3273, + "eval_samples_per_second": 571.327, + "eval_steps_per_second": 9.166, + "step": 5724 + }, + { + "epoch": 55.0, + "grad_norm": 1.2054860591888428, + "learning_rate": 2.25e-05, + "loss": 0.0543, + "step": 5830 + }, + { + "epoch": 55.0, + "eval_LOCATION_f1": 0.8026315789473685, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7721518987341772, + "eval_LOCATION_recall": 0.8356164383561644, + "eval_ORGANIZATION_f1": 0.7755102040816327, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6951219512195121, + "eval_ORGANIZATION_recall": 0.8769230769230769, + "eval_PERSON_f1": 0.8903225806451613, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8625, + "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": 0.7837837837837837, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.725, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.1556856632232666, + "eval_overall_accuracy": 0.9598818606940684, + "eval_overall_f1": 0.8273092369477911, + "eval_overall_precision": 0.7803030303030303, + "eval_overall_recall": 0.8803418803418803, + "eval_runtime": 0.3279, + "eval_samples_per_second": 570.31, + "eval_steps_per_second": 9.149, + "step": 5830 + }, + { + "epoch": 56.0, + "grad_norm": 2.491786241531372, + "learning_rate": 2.2000000000000003e-05, + "loss": 0.0517, + "step": 5936 + }, + { + "epoch": 56.0, + "eval_LOCATION_f1": 0.8211920529801324, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7948717948717948, + "eval_LOCATION_recall": 0.8493150684931506, + "eval_ORGANIZATION_f1": 0.7777777777777778, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7088607594936709, + "eval_ORGANIZATION_recall": 0.8615384615384616, + "eval_PERSON_f1": 0.8810289389067525, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8509316770186336, + "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": 0.7837837837837837, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.725, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.16153576970100403, + "eval_overall_accuracy": 0.9598818606940684, + "eval_overall_f1": 0.8279569892473119, + "eval_overall_precision": 0.7837150127226463, + "eval_overall_recall": 0.8774928774928775, + "eval_runtime": 0.3265, + "eval_samples_per_second": 572.734, + "eval_steps_per_second": 9.188, + "step": 5936 + }, + { + "epoch": 57.0, + "grad_norm": 0.5145077109336853, + "learning_rate": 2.15e-05, + "loss": 0.0512, + "step": 6042 + }, + { + "epoch": 57.0, + "eval_LOCATION_f1": 0.8026315789473685, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7721518987341772, + "eval_LOCATION_recall": 0.8356164383561644, + "eval_ORGANIZATION_f1": 0.7887323943661972, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7272727272727273, + "eval_ORGANIZATION_recall": 0.8615384615384616, + "eval_PERSON_f1": 0.8838709677419355, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.85625, + "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": 0.7733333333333334, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7073170731707317, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.163212388753891, + "eval_overall_accuracy": 0.9601279842480925, + "eval_overall_f1": 0.8263795423956932, + "eval_overall_precision": 0.7831632653061225, + "eval_overall_recall": 0.8746438746438746, + "eval_runtime": 0.3262, + "eval_samples_per_second": 573.237, + "eval_steps_per_second": 9.196, + "step": 6042 + }, + { + "epoch": 58.0, + "grad_norm": 0.5442776083946228, + "learning_rate": 2.1e-05, + "loss": 0.0522, + "step": 6148 + }, + { + "epoch": 58.0, + "eval_LOCATION_f1": 0.8157894736842105, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7848101265822784, + "eval_LOCATION_recall": 0.8493150684931506, + "eval_ORGANIZATION_f1": 0.7619047619047619, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6829268292682927, + "eval_ORGANIZATION_recall": 0.8615384615384616, + "eval_PERSON_f1": 0.8938906752411575, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8633540372670807, + "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": 0.7733333333333334, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7073170731707317, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.16096481680870056, + "eval_overall_accuracy": 0.9579128722618755, + "eval_overall_f1": 0.8277703604806409, + "eval_overall_precision": 0.7788944723618091, + "eval_overall_recall": 0.8831908831908832, + "eval_runtime": 0.326, + "eval_samples_per_second": 573.668, + "eval_steps_per_second": 9.203, + "step": 6148 + }, + { + "epoch": 59.0, + "grad_norm": 2.010756015777588, + "learning_rate": 2.05e-05, + "loss": 0.05, + "step": 6254 + }, + { + "epoch": 59.0, + "eval_LOCATION_f1": 0.8026315789473685, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7721518987341772, + "eval_LOCATION_recall": 0.8356164383561644, + "eval_ORGANIZATION_f1": 0.7482993197278912, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6707317073170732, + "eval_ORGANIZATION_recall": 0.8461538461538461, + "eval_PERSON_f1": 0.887459807073955, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8571428571428571, + "eval_PERSON_recall": 0.92, + "eval_QUANTITY_f1": 0.7272727272727273, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6486486486486487, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.7837837837837837, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.725, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.1725725382566452, + "eval_overall_accuracy": 0.9566822544917548, + "eval_overall_f1": 0.8186666666666667, + "eval_overall_precision": 0.7694235588972431, + "eval_overall_recall": 0.8746438746438746, + "eval_runtime": 0.3268, + "eval_samples_per_second": 572.211, + "eval_steps_per_second": 9.18, + "step": 6254 + }, + { + "epoch": 60.0, + "grad_norm": 1.3911691904067993, + "learning_rate": 2e-05, + "loss": 0.0505, + "step": 6360 + }, + { + "epoch": 60.0, + "eval_LOCATION_f1": 0.8079470198675496, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.782051282051282, + "eval_LOCATION_recall": 0.8356164383561644, + "eval_ORGANIZATION_f1": 0.7971014492753623, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7534246575342466, + "eval_ORGANIZATION_recall": 0.8461538461538461, + "eval_PERSON_f1": 0.8932038834951457, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8679245283018868, + "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": 0.7567567567567567, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7, + "eval_TIME_recall": 0.8235294117647058, + "eval_loss": 0.15419013798236847, + "eval_overall_accuracy": 0.9620969726802855, + "eval_overall_f1": 0.8353741496598639, + "eval_overall_precision": 0.7994791666666666, + "eval_overall_recall": 0.8746438746438746, + "eval_runtime": 0.3255, + "eval_samples_per_second": 574.46, + "eval_steps_per_second": 9.216, + "step": 6360 + }, + { + "epoch": 61.0, + "grad_norm": 1.3116302490234375, + "learning_rate": 1.9500000000000003e-05, + "loss": 0.0479, + "step": 6466 + }, + { + "epoch": 61.0, + "eval_LOCATION_f1": 0.8079470198675496, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.782051282051282, + "eval_LOCATION_recall": 0.8356164383561644, + "eval_ORGANIZATION_f1": 0.7814569536423841, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.686046511627907, + "eval_ORGANIZATION_recall": 0.9076923076923077, + "eval_PERSON_f1": 0.8903225806451613, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8625, + "eval_PERSON_recall": 0.92, + "eval_QUANTITY_f1": 0.7619047619047619, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7058823529411765, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.7945205479452054, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7435897435897436, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.16232585906982422, + "eval_overall_accuracy": 0.9598818606940684, + "eval_overall_f1": 0.8315508021390374, + "eval_overall_precision": 0.783375314861461, + "eval_overall_recall": 0.886039886039886, + "eval_runtime": 0.326, + "eval_samples_per_second": 573.55, + "eval_steps_per_second": 9.201, + "step": 6466 + }, + { + "epoch": 62.0, + "grad_norm": 0.7610757350921631, + "learning_rate": 1.9e-05, + "loss": 0.0481, + "step": 6572 + }, + { + "epoch": 62.0, + "eval_LOCATION_f1": 0.8026315789473685, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7721518987341772, + "eval_LOCATION_recall": 0.8356164383561644, + "eval_ORGANIZATION_f1": 0.7763157894736842, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6781609195402298, + "eval_ORGANIZATION_recall": 0.9076923076923077, + "eval_PERSON_f1": 0.8996763754045306, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8742138364779874, + "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": 0.7837837837837837, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.725, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.16829490661621094, + "eval_overall_accuracy": 0.9584051193699237, + "eval_overall_f1": 0.8308921438082557, + "eval_overall_precision": 0.78, + "eval_overall_recall": 0.8888888888888888, + "eval_runtime": 0.3277, + "eval_samples_per_second": 570.708, + "eval_steps_per_second": 9.156, + "step": 6572 + }, + { + "epoch": 63.0, + "grad_norm": 1.4961496591567993, + "learning_rate": 1.85e-05, + "loss": 0.047, + "step": 6678 + }, + { + "epoch": 63.0, + "eval_LOCATION_f1": 0.8157894736842105, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7848101265822784, + "eval_LOCATION_recall": 0.8493150684931506, + "eval_ORGANIZATION_f1": 0.7972972972972974, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7108433734939759, + "eval_ORGANIZATION_recall": 0.9076923076923077, + "eval_PERSON_f1": 0.8938906752411575, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8633540372670807, + "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": 0.767123287671233, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.717948717948718, + "eval_TIME_recall": 0.8235294117647058, + "eval_loss": 0.15628120303153992, + "eval_overall_accuracy": 0.9613586020182131, + "eval_overall_f1": 0.8380187416331994, + "eval_overall_precision": 0.7904040404040404, + "eval_overall_recall": 0.8917378917378918, + "eval_runtime": 0.3277, + "eval_samples_per_second": 570.71, + "eval_steps_per_second": 9.156, + "step": 6678 + }, + { + "epoch": 64.0, + "grad_norm": 0.4107072353363037, + "learning_rate": 1.8e-05, + "loss": 0.0483, + "step": 6784 + }, + { + "epoch": 64.0, + "eval_LOCATION_f1": 0.8266666666666668, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8051948051948052, + "eval_LOCATION_recall": 0.8493150684931506, + "eval_ORGANIZATION_f1": 0.7945205479452054, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7160493827160493, + "eval_ORGANIZATION_recall": 0.8923076923076924, + "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": 0.7945205479452054, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7435897435897436, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.16059257090091705, + "eval_overall_accuracy": 0.9616047255722373, + "eval_overall_f1": 0.8409703504043127, + "eval_overall_precision": 0.7979539641943734, + "eval_overall_recall": 0.8888888888888888, + "eval_runtime": 0.3257, + "eval_samples_per_second": 574.068, + "eval_steps_per_second": 9.21, + "step": 6784 + }, + { + "epoch": 65.0, + "grad_norm": 0.6396474838256836, + "learning_rate": 1.75e-05, + "loss": 0.0454, + "step": 6890 + }, + { + "epoch": 65.0, + "eval_LOCATION_f1": 0.8344370860927152, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8076923076923077, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.8169014084507042, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7532467532467533, + "eval_ORGANIZATION_recall": 0.8923076923076924, + "eval_PERSON_f1": 0.8996763754045306, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8742138364779874, + "eval_PERSON_recall": 0.9266666666666666, + "eval_QUANTITY_f1": 0.71875, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6571428571428571, + "eval_QUANTITY_recall": 0.7931034482758621, + "eval_TIME_f1": 0.7945205479452054, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7435897435897436, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.16208666563034058, + "eval_overall_accuracy": 0.9625892197883338, + "eval_overall_f1": 0.8443843031123139, + "eval_overall_precision": 0.8041237113402062, + "eval_overall_recall": 0.8888888888888888, + "eval_runtime": 0.3265, + "eval_samples_per_second": 572.829, + "eval_steps_per_second": 9.19, + "step": 6890 + }, + { + "epoch": 66.0, + "grad_norm": 0.5849748849868774, + "learning_rate": 1.7000000000000003e-05, + "loss": 0.046, + "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": 0.7919463087248322, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7023809523809523, + "eval_ORGANIZATION_recall": 0.9076923076923077, + "eval_PERSON_f1": 0.9025974025974027, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.879746835443038, + "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": 0.7945205479452054, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7435897435897436, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.16353175044059753, + "eval_overall_accuracy": 0.9618508491262614, + "eval_overall_f1": 0.8418230563002682, + "eval_overall_precision": 0.7949367088607595, + "eval_overall_recall": 0.8945868945868946, + "eval_runtime": 0.3257, + "eval_samples_per_second": 574.147, + "eval_steps_per_second": 9.211, + "step": 6996 + }, + { + "epoch": 67.0, + "grad_norm": 0.3327535092830658, + "learning_rate": 1.65e-05, + "loss": 0.0451, + "step": 7102 + }, + { + "epoch": 67.0, + "eval_LOCATION_f1": 0.8129032258064516, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7682926829268293, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.7866666666666665, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6941176470588235, + "eval_ORGANIZATION_recall": 0.9076923076923077, + "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": 0.7945205479452054, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7435897435897436, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.16600778698921204, + "eval_overall_accuracy": 0.9603741078021166, + "eval_overall_f1": 0.8362183754993342, + "eval_overall_precision": 0.785, + "eval_overall_recall": 0.8945868945868946, + "eval_runtime": 0.3263, + "eval_samples_per_second": 573.136, + "eval_steps_per_second": 9.195, + "step": 7102 + }, + { + "epoch": 68.0, + "grad_norm": 0.9939576983451843, + "learning_rate": 1.6000000000000003e-05, + "loss": 0.0436, + "step": 7208 + }, + { + "epoch": 68.0, + "eval_LOCATION_f1": 0.8266666666666668, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8051948051948052, + "eval_LOCATION_recall": 0.8493150684931506, + "eval_ORGANIZATION_f1": 0.7945205479452054, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7160493827160493, + "eval_ORGANIZATION_recall": 0.8923076923076924, + "eval_PERSON_f1": 0.896103896103896, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8734177215189873, + "eval_PERSON_recall": 0.92, + "eval_QUANTITY_f1": 0.7076923076923076, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6388888888888888, + "eval_QUANTITY_recall": 0.7931034482758621, + "eval_TIME_f1": 0.7567567567567567, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7, + "eval_TIME_recall": 0.8235294117647058, + "eval_loss": 0.16049455106258392, + "eval_overall_accuracy": 0.9613586020182131, + "eval_overall_f1": 0.8317631224764468, + "eval_overall_precision": 0.7882653061224489, + "eval_overall_recall": 0.8803418803418803, + "eval_runtime": 0.3279, + "eval_samples_per_second": 570.333, + "eval_steps_per_second": 9.15, + "step": 7208 + }, + { + "epoch": 69.0, + "grad_norm": 0.1674051731824875, + "learning_rate": 1.55e-05, + "loss": 0.043, + "step": 7314 + }, + { + "epoch": 69.0, + "eval_LOCATION_f1": 0.8235294117647058, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7875, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.7702702702702702, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6867469879518072, + "eval_ORGANIZATION_recall": 0.8769230769230769, + "eval_PERSON_f1": 0.8938906752411575, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8633540372670807, + "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": 0.7945205479452054, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7435897435897436, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.16724927723407745, + "eval_overall_accuracy": 0.9581589958158996, + "eval_overall_f1": 0.8342245989304813, + "eval_overall_precision": 0.7858942065491183, + "eval_overall_recall": 0.8888888888888888, + "eval_runtime": 0.3254, + "eval_samples_per_second": 574.709, + "eval_steps_per_second": 9.22, + "step": 7314 + }, + { + "epoch": 70.0, + "grad_norm": 1.958099603652954, + "learning_rate": 1.5e-05, + "loss": 0.0424, + "step": 7420 + }, + { + "epoch": 70.0, + "eval_LOCATION_f1": 0.8235294117647058, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7875, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.7755102040816327, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6951219512195121, + "eval_ORGANIZATION_recall": 0.8769230769230769, + "eval_PERSON_f1": 0.896774193548387, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.86875, + "eval_PERSON_recall": 0.9266666666666666, + "eval_QUANTITY_f1": 0.7419354838709677, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.696969696969697, + "eval_QUANTITY_recall": 0.7931034482758621, + "eval_TIME_f1": 0.7567567567567567, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7, + "eval_TIME_recall": 0.8235294117647058, + "eval_loss": 0.16436557471752167, + "eval_overall_accuracy": 0.9606202313561408, + "eval_overall_f1": 0.8310991957104557, + "eval_overall_precision": 0.7848101265822784, + "eval_overall_recall": 0.8831908831908832, + "eval_runtime": 0.3276, + "eval_samples_per_second": 570.858, + "eval_steps_per_second": 9.158, + "step": 7420 + }, + { + "epoch": 71.0, + "grad_norm": 1.849372386932373, + "learning_rate": 1.45e-05, + "loss": 0.0469, + "step": 7526 + }, + { + "epoch": 71.0, + "eval_LOCATION_f1": 0.8289473684210525, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7974683544303798, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.7682119205298013, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6744186046511628, + "eval_ORGANIZATION_recall": 0.8923076923076924, + "eval_PERSON_f1": 0.896103896103896, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8734177215189873, + "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": 0.7945205479452054, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7435897435897436, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.16665174067020416, + "eval_overall_accuracy": 0.9603741078021166, + "eval_overall_f1": 0.8342245989304813, + "eval_overall_precision": 0.7858942065491183, + "eval_overall_recall": 0.8888888888888888, + "eval_runtime": 0.3285, + "eval_samples_per_second": 569.213, + "eval_steps_per_second": 9.132, + "step": 7526 + }, + { + "epoch": 72.0, + "grad_norm": 0.11355619132518768, + "learning_rate": 1.4000000000000001e-05, + "loss": 0.0409, + "step": 7632 + }, + { + "epoch": 72.0, + "eval_LOCATION_f1": 0.8181818181818182, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7777777777777778, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.7567567567567568, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6746987951807228, + "eval_ORGANIZATION_recall": 0.8615384615384616, + "eval_PERSON_f1": 0.8903225806451613, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8625, + "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": 0.7837837837837837, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.725, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.17135433852672577, + "eval_overall_accuracy": 0.9593896135860202, + "eval_overall_f1": 0.8266666666666667, + "eval_overall_precision": 0.7769423558897243, + "eval_overall_recall": 0.8831908831908832, + "eval_runtime": 0.3271, + "eval_samples_per_second": 571.736, + "eval_steps_per_second": 9.172, + "step": 7632 + }, + { + "epoch": 73.0, + "grad_norm": 0.40124908089637756, + "learning_rate": 1.3500000000000001e-05, + "loss": 0.0437, + "step": 7738 + }, + { + "epoch": 73.0, + "eval_LOCATION_f1": 0.8157894736842105, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7848101265822784, + "eval_LOCATION_recall": 0.8493150684931506, + "eval_ORGANIZATION_f1": 0.7808219178082192, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7037037037037037, + "eval_ORGANIZATION_recall": 0.8769230769230769, + "eval_PERSON_f1": 0.8932038834951457, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8679245283018868, + "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": 0.7837837837837837, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.725, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.16818270087242126, + "eval_overall_accuracy": 0.9591434900319961, + "eval_overall_f1": 0.829530201342282, + "eval_overall_precision": 0.7842639593908629, + "eval_overall_recall": 0.8803418803418803, + "eval_runtime": 0.3264, + "eval_samples_per_second": 572.959, + "eval_steps_per_second": 9.192, + "step": 7738 + }, + { + "epoch": 74.0, + "grad_norm": 0.6592825651168823, + "learning_rate": 1.3000000000000001e-05, + "loss": 0.0411, + "step": 7844 + }, + { + "epoch": 74.0, + "eval_LOCATION_f1": 0.8104575163398693, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.775, + "eval_LOCATION_recall": 0.8493150684931506, + "eval_ORGANIZATION_f1": 0.7857142857142856, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7333333333333333, + "eval_ORGANIZATION_recall": 0.8461538461538461, + "eval_PERSON_f1": 0.896103896103896, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8734177215189873, + "eval_PERSON_recall": 0.92, + "eval_QUANTITY_f1": 0.7301587301587301, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6764705882352942, + "eval_QUANTITY_recall": 0.7931034482758621, + "eval_TIME_f1": 0.7945205479452054, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7435897435897436, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.1627454161643982, + "eval_overall_accuracy": 0.961112478464189, + "eval_overall_f1": 0.8331071913161466, + "eval_overall_precision": 0.7953367875647669, + "eval_overall_recall": 0.8746438746438746, + "eval_runtime": 0.3256, + "eval_samples_per_second": 574.25, + "eval_steps_per_second": 9.213, + "step": 7844 + }, + { + "epoch": 75.0, + "grad_norm": 0.21142633259296417, + "learning_rate": 1.25e-05, + "loss": 0.0432, + "step": 7950 + }, + { + "epoch": 75.0, + "eval_LOCATION_f1": 0.8211920529801324, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7948717948717948, + "eval_LOCATION_recall": 0.8493150684931506, + "eval_ORGANIZATION_f1": 0.7612903225806452, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6555555555555556, + "eval_ORGANIZATION_recall": 0.9076923076923077, + "eval_PERSON_f1": 0.8996763754045306, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8742138364779874, + "eval_PERSON_recall": 0.9266666666666666, + "eval_QUANTITY_f1": 0.7301587301587301, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6764705882352942, + "eval_QUANTITY_recall": 0.7931034482758621, + "eval_TIME_f1": 0.7945205479452054, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7435897435897436, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.1737181544303894, + "eval_overall_accuracy": 0.958897366477972, + "eval_overall_f1": 0.8308921438082557, + "eval_overall_precision": 0.78, + "eval_overall_recall": 0.8888888888888888, + "eval_runtime": 0.3268, + "eval_samples_per_second": 572.146, + "eval_steps_per_second": 9.179, + "step": 7950 + }, + { + "epoch": 76.0, + "grad_norm": 0.6738051176071167, + "learning_rate": 1.2e-05, + "loss": 0.0398, + "step": 8056 + }, + { + "epoch": 76.0, + "eval_LOCATION_f1": 0.8157894736842105, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7848101265822784, + "eval_LOCATION_recall": 0.8493150684931506, + "eval_ORGANIZATION_f1": 0.7733333333333334, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6823529411764706, + "eval_ORGANIZATION_recall": 0.8923076923076924, + "eval_PERSON_f1": 0.8932038834951457, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8679245283018868, + "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": 0.7945205479452054, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7435897435897436, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.16499055922031403, + "eval_overall_accuracy": 0.9608663549101649, + "eval_overall_f1": 0.8368983957219251, + "eval_overall_precision": 0.7884130982367759, + "eval_overall_recall": 0.8917378917378918, + "eval_runtime": 0.3263, + "eval_samples_per_second": 573.037, + "eval_steps_per_second": 9.193, + "step": 8056 + }, + { + "epoch": 77.0, + "grad_norm": 0.5994135141372681, + "learning_rate": 1.1500000000000002e-05, + "loss": 0.0392, + "step": 8162 + }, + { + "epoch": 77.0, + "eval_LOCATION_f1": 0.8104575163398693, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.775, + "eval_LOCATION_recall": 0.8493150684931506, + "eval_ORGANIZATION_f1": 0.7777777777777778, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7088607594936709, + "eval_ORGANIZATION_recall": 0.8615384615384616, + "eval_PERSON_f1": 0.896103896103896, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8734177215189873, + "eval_PERSON_recall": 0.92, + "eval_QUANTITY_f1": 0.7619047619047619, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7058823529411765, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.7837837837837837, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.725, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.1677028387784958, + "eval_overall_accuracy": 0.9603741078021166, + "eval_overall_f1": 0.8328840970350404, + "eval_overall_precision": 0.7902813299232737, + "eval_overall_recall": 0.8803418803418803, + "eval_runtime": 0.3272, + "eval_samples_per_second": 571.495, + "eval_steps_per_second": 9.168, + "step": 8162 + }, + { + "epoch": 78.0, + "grad_norm": 1.128940224647522, + "learning_rate": 1.1000000000000001e-05, + "loss": 0.0404, + "step": 8268 + }, + { + "epoch": 78.0, + "eval_LOCATION_f1": 0.8157894736842105, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7848101265822784, + "eval_LOCATION_recall": 0.8493150684931506, + "eval_ORGANIZATION_f1": 0.7702702702702702, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6867469879518072, + "eval_ORGANIZATION_recall": 0.8769230769230769, + "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": 0.7837837837837837, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.725, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.17405812442302704, + "eval_overall_accuracy": 0.9591434900319961, + "eval_overall_f1": 0.8326639892904953, + "eval_overall_precision": 0.7853535353535354, + "eval_overall_recall": 0.886039886039886, + "eval_runtime": 0.3258, + "eval_samples_per_second": 573.989, + "eval_steps_per_second": 9.208, + "step": 8268 + }, + { + "epoch": 79.0, + "grad_norm": 0.5310064554214478, + "learning_rate": 1.05e-05, + "loss": 0.0414, + "step": 8374 + }, + { + "epoch": 79.0, + "eval_LOCATION_f1": 0.8104575163398693, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.775, + "eval_LOCATION_recall": 0.8493150684931506, + "eval_ORGANIZATION_f1": 0.7724137931034483, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7, + "eval_ORGANIZATION_recall": 0.8615384615384616, + "eval_PERSON_f1": 0.8846153846153846, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8518518518518519, + "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": 0.7733333333333334, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7073170731707317, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.18120695650577545, + "eval_overall_accuracy": 0.9576667487078513, + "eval_overall_f1": 0.8251001335113485, + "eval_overall_precision": 0.7763819095477387, + "eval_overall_recall": 0.8803418803418803, + "eval_runtime": 0.3267, + "eval_samples_per_second": 572.4, + "eval_steps_per_second": 9.183, + "step": 8374 + }, + { + "epoch": 80.0, + "grad_norm": 0.45909491181373596, + "learning_rate": 1e-05, + "loss": 0.0401, + "step": 8480 + }, + { + "epoch": 80.0, + "eval_LOCATION_f1": 0.8181818181818182, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7777777777777778, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.7534246575342466, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6790123456790124, + "eval_ORGANIZATION_recall": 0.8461538461538461, + "eval_PERSON_f1": 0.8932038834951457, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8679245283018868, + "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": 0.7368421052631577, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.6666666666666666, + "eval_TIME_recall": 0.8235294117647058, + "eval_loss": 0.1721770316362381, + "eval_overall_accuracy": 0.9576667487078513, + "eval_overall_f1": 0.822429906542056, + "eval_overall_precision": 0.7738693467336684, + "eval_overall_recall": 0.8774928774928775, + "eval_runtime": 0.3276, + "eval_samples_per_second": 570.822, + "eval_steps_per_second": 9.158, + "step": 8480 + }, + { + "epoch": 81.0, + "grad_norm": 0.8190267086029053, + "learning_rate": 9.5e-06, + "loss": 0.041, + "step": 8586 + }, + { + "epoch": 81.0, + "eval_LOCATION_f1": 0.8235294117647058, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7875, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.7651006711409396, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6785714285714286, + "eval_ORGANIZATION_recall": 0.8769230769230769, + "eval_PERSON_f1": 0.8932038834951457, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8679245283018868, + "eval_PERSON_recall": 0.92, + "eval_QUANTITY_f1": 0.7741935483870968, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7272727272727273, + "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": 0.1708712875843048, + "eval_overall_accuracy": 0.9596357371400444, + "eval_overall_f1": 0.8299866131191433, + "eval_overall_precision": 0.7828282828282829, + "eval_overall_recall": 0.8831908831908832, + "eval_runtime": 0.3268, + "eval_samples_per_second": 572.165, + "eval_steps_per_second": 9.179, + "step": 8586 + }, + { + "epoch": 82.0, + "grad_norm": 1.201314091682434, + "learning_rate": 9e-06, + "loss": 0.0405, + "step": 8692 + }, + { + "epoch": 82.0, + "eval_LOCATION_f1": 0.8235294117647058, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7875, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.7832167832167831, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.717948717948718, + "eval_ORGANIZATION_recall": 0.8615384615384616, + "eval_PERSON_f1": 0.8932038834951457, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8679245283018868, + "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": 0.7368421052631577, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.6666666666666666, + "eval_TIME_recall": 0.8235294117647058, + "eval_loss": 0.16781871020793915, + "eval_overall_accuracy": 0.9593896135860202, + "eval_overall_f1": 0.829530201342282, + "eval_overall_precision": 0.7842639593908629, + "eval_overall_recall": 0.8803418803418803, + "eval_runtime": 0.3265, + "eval_samples_per_second": 572.664, + "eval_steps_per_second": 9.187, + "step": 8692 + }, + { + "epoch": 83.0, + "grad_norm": 2.120737314224243, + "learning_rate": 8.500000000000002e-06, + "loss": 0.0398, + "step": 8798 + }, + { + "epoch": 83.0, + "eval_LOCATION_f1": 0.8157894736842105, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7848101265822784, + "eval_LOCATION_recall": 0.8493150684931506, + "eval_ORGANIZATION_f1": 0.763888888888889, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6962025316455697, + "eval_ORGANIZATION_recall": 0.8461538461538461, + "eval_PERSON_f1": 0.8932038834951457, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8679245283018868, + "eval_PERSON_recall": 0.92, + "eval_QUANTITY_f1": 0.7741935483870968, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7272727272727273, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.7368421052631577, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.6666666666666666, + "eval_TIME_recall": 0.8235294117647058, + "eval_loss": 0.17089137434959412, + "eval_overall_accuracy": 0.9586512429239479, + "eval_overall_f1": 0.8263795423956932, + "eval_overall_precision": 0.7831632653061225, + "eval_overall_recall": 0.8746438746438746, + "eval_runtime": 0.3271, + "eval_samples_per_second": 571.7, + "eval_steps_per_second": 9.172, + "step": 8798 + }, + { + "epoch": 84.0, + "grad_norm": 0.8175678253173828, + "learning_rate": 8.000000000000001e-06, + "loss": 0.04, + "step": 8904 + }, + { + "epoch": 84.0, + "eval_LOCATION_f1": 0.8235294117647058, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7875, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.7919463087248322, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7023809523809523, + "eval_ORGANIZATION_recall": 0.9076923076923077, + "eval_PERSON_f1": 0.8932038834951457, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8679245283018868, + "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": 0.7733333333333334, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7073170731707317, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.17276740074157715, + "eval_overall_accuracy": 0.9579128722618755, + "eval_overall_f1": 0.8346666666666667, + "eval_overall_precision": 0.7844611528822055, + "eval_overall_recall": 0.8917378917378918, + "eval_runtime": 0.3268, + "eval_samples_per_second": 572.273, + "eval_steps_per_second": 9.181, + "step": 8904 + }, + { + "epoch": 85.0, + "grad_norm": 1.86958646774292, + "learning_rate": 7.5e-06, + "loss": 0.0404, + "step": 9010 + }, + { + "epoch": 85.0, + "eval_LOCATION_f1": 0.8129032258064516, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7682926829268293, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.7651006711409396, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6785714285714286, + "eval_ORGANIZATION_recall": 0.8769230769230769, + "eval_PERSON_f1": 0.8932038834951457, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8679245283018868, + "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": 0.7733333333333334, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7073170731707317, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.1715661585330963, + "eval_overall_accuracy": 0.9586512429239479, + "eval_overall_f1": 0.827127659574468, + "eval_overall_precision": 0.7755610972568578, + "eval_overall_recall": 0.886039886039886, + "eval_runtime": 0.3267, + "eval_samples_per_second": 572.31, + "eval_steps_per_second": 9.181, + "step": 9010 + }, + { + "epoch": 86.0, + "grad_norm": 0.4103405773639679, + "learning_rate": 7.000000000000001e-06, + "loss": 0.0364, + "step": 9116 + }, + { + "epoch": 86.0, + "eval_LOCATION_f1": 0.8129032258064516, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7682926829268293, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.7972027972027972, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7307692307692307, + "eval_ORGANIZATION_recall": 0.8769230769230769, + "eval_PERSON_f1": 0.896103896103896, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8734177215189873, + "eval_PERSON_recall": 0.92, + "eval_QUANTITY_f1": 0.7384615384615385, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6666666666666666, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.7733333333333334, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7073170731707317, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.16957244277000427, + "eval_overall_accuracy": 0.9608663549101649, + "eval_overall_f1": 0.8337801608579088, + "eval_overall_precision": 0.7873417721518987, + "eval_overall_recall": 0.886039886039886, + "eval_runtime": 0.3274, + "eval_samples_per_second": 571.12, + "eval_steps_per_second": 9.162, + "step": 9116 + }, + { + "epoch": 87.0, + "grad_norm": 3.4236576557159424, + "learning_rate": 6.5000000000000004e-06, + "loss": 0.0377, + "step": 9222 + }, + { + "epoch": 87.0, + "eval_LOCATION_f1": 0.8235294117647058, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7875, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.76, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6705882352941176, + "eval_ORGANIZATION_recall": 0.8769230769230769, + "eval_PERSON_f1": 0.896103896103896, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8734177215189873, + "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": 0.7368421052631577, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.6666666666666666, + "eval_TIME_recall": 0.8235294117647058, + "eval_loss": 0.17427609860897064, + "eval_overall_accuracy": 0.9581589958158996, + "eval_overall_f1": 0.8255659121171771, + "eval_overall_precision": 0.775, + "eval_overall_recall": 0.8831908831908832, + "eval_runtime": 0.3336, + "eval_samples_per_second": 560.496, + "eval_steps_per_second": 8.992, + "step": 9222 + }, + { + "epoch": 88.0, + "grad_norm": 0.8201001882553101, + "learning_rate": 6e-06, + "loss": 0.0379, + "step": 9328 + }, + { + "epoch": 88.0, + "eval_LOCATION_f1": 0.8104575163398693, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.775, + "eval_LOCATION_recall": 0.8493150684931506, + "eval_ORGANIZATION_f1": 0.7755102040816327, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6951219512195121, + "eval_ORGANIZATION_recall": 0.8769230769230769, + "eval_PERSON_f1": 0.896103896103896, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8734177215189873, + "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": 0.7733333333333334, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7073170731707317, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.17298519611358643, + "eval_overall_accuracy": 0.9593896135860202, + "eval_overall_f1": 0.8326639892904953, + "eval_overall_precision": 0.7853535353535354, + "eval_overall_recall": 0.886039886039886, + "eval_runtime": 0.3266, + "eval_samples_per_second": 572.521, + "eval_steps_per_second": 9.185, + "step": 9328 + }, + { + "epoch": 89.0, + "grad_norm": 0.4763755202293396, + "learning_rate": 5.500000000000001e-06, + "loss": 0.0375, + "step": 9434 + }, + { + "epoch": 89.0, + "eval_LOCATION_f1": 0.8235294117647058, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7875, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.7702702702702702, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6867469879518072, + "eval_ORGANIZATION_recall": 0.8769230769230769, + "eval_PERSON_f1": 0.896103896103896, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8734177215189873, + "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": 0.7466666666666667, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.6829268292682927, + "eval_TIME_recall": 0.8235294117647058, + "eval_loss": 0.17421159148216248, + "eval_overall_accuracy": 0.9584051193699237, + "eval_overall_f1": 0.8288770053475936, + "eval_overall_precision": 0.7808564231738035, + "eval_overall_recall": 0.8831908831908832, + "eval_runtime": 0.3259, + "eval_samples_per_second": 573.871, + "eval_steps_per_second": 9.206, + "step": 9434 + }, + { + "epoch": 90.0, + "grad_norm": 0.24638473987579346, + "learning_rate": 5e-06, + "loss": 0.0385, + "step": 9540 + }, + { + "epoch": 90.0, + "eval_LOCATION_f1": 0.8235294117647058, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7875, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.7777777777777778, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7088607594936709, + "eval_ORGANIZATION_recall": 0.8615384615384616, + "eval_PERSON_f1": 0.896103896103896, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8734177215189873, + "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": 0.7368421052631577, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.6666666666666666, + "eval_TIME_recall": 0.8235294117647058, + "eval_loss": 0.16978448629379272, + "eval_overall_accuracy": 0.9591434900319961, + "eval_overall_f1": 0.829530201342282, + "eval_overall_precision": 0.7842639593908629, + "eval_overall_recall": 0.8803418803418803, + "eval_runtime": 0.3272, + "eval_samples_per_second": 571.563, + "eval_steps_per_second": 9.169, + "step": 9540 + }, + { + "epoch": 91.0, + "grad_norm": 0.680090069770813, + "learning_rate": 4.5e-06, + "loss": 0.036, + "step": 9646 + }, + { + "epoch": 91.0, + "eval_LOCATION_f1": 0.8181818181818182, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7777777777777778, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.7466666666666668, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6588235294117647, + "eval_ORGANIZATION_recall": 0.8615384615384616, + "eval_PERSON_f1": 0.8932038834951457, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8679245283018868, + "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": 0.7733333333333334, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7073170731707317, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.17799106240272522, + "eval_overall_accuracy": 0.9576667487078513, + "eval_overall_f1": 0.824468085106383, + "eval_overall_precision": 0.773067331670823, + "eval_overall_recall": 0.8831908831908832, + "eval_runtime": 0.3268, + "eval_samples_per_second": 572.277, + "eval_steps_per_second": 9.181, + "step": 9646 + }, + { + "epoch": 92.0, + "grad_norm": 0.4846082925796509, + "learning_rate": 4.000000000000001e-06, + "loss": 0.0377, + "step": 9752 + }, + { + "epoch": 92.0, + "eval_LOCATION_f1": 0.8181818181818182, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7777777777777778, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.7808219178082192, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7037037037037037, + "eval_ORGANIZATION_recall": 0.8769230769230769, + "eval_PERSON_f1": 0.896103896103896, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8734177215189873, + "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": 0.7733333333333334, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7073170731707317, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.17370600998401642, + "eval_overall_accuracy": 0.9596357371400444, + "eval_overall_f1": 0.8353413654618473, + "eval_overall_precision": 0.7878787878787878, + "eval_overall_recall": 0.8888888888888888, + "eval_runtime": 0.3268, + "eval_samples_per_second": 572.206, + "eval_steps_per_second": 9.18, + "step": 9752 + }, + { + "epoch": 93.0, + "grad_norm": 0.28526270389556885, + "learning_rate": 3.5000000000000004e-06, + "loss": 0.0391, + "step": 9858 + }, + { + "epoch": 93.0, + "eval_LOCATION_f1": 0.8104575163398693, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.775, + "eval_LOCATION_recall": 0.8493150684931506, + "eval_ORGANIZATION_f1": 0.7702702702702702, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6867469879518072, + "eval_ORGANIZATION_recall": 0.8769230769230769, + "eval_PERSON_f1": 0.896103896103896, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8734177215189873, + "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": 0.7733333333333334, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7073170731707317, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.17698532342910767, + "eval_overall_accuracy": 0.9586512429239479, + "eval_overall_f1": 0.8288770053475936, + "eval_overall_precision": 0.7808564231738035, + "eval_overall_recall": 0.8831908831908832, + "eval_runtime": 0.3263, + "eval_samples_per_second": 573.061, + "eval_steps_per_second": 9.193, + "step": 9858 + }, + { + "epoch": 94.0, + "grad_norm": 0.6564857363700867, + "learning_rate": 3e-06, + "loss": 0.0362, + "step": 9964 + }, + { + "epoch": 94.0, + "eval_LOCATION_f1": 0.8235294117647058, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7875, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.7777777777777778, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7088607594936709, + "eval_ORGANIZATION_recall": 0.8615384615384616, + "eval_PERSON_f1": 0.896103896103896, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8734177215189873, + "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": 0.7368421052631577, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.6666666666666666, + "eval_TIME_recall": 0.8235294117647058, + "eval_loss": 0.17274247109889984, + "eval_overall_accuracy": 0.958897366477972, + "eval_overall_f1": 0.8333333333333334, + "eval_overall_precision": 0.7888040712468194, + "eval_overall_recall": 0.8831908831908832, + "eval_runtime": 0.3259, + "eval_samples_per_second": 573.776, + "eval_steps_per_second": 9.205, + "step": 9964 + }, + { + "epoch": 95.0, + "grad_norm": 0.7220708727836609, + "learning_rate": 2.5e-06, + "loss": 0.0372, + "step": 10070 + }, + { + "epoch": 95.0, + "eval_LOCATION_f1": 0.8181818181818182, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7777777777777778, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.7808219178082192, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7037037037037037, + "eval_ORGANIZATION_recall": 0.8769230769230769, + "eval_PERSON_f1": 0.896103896103896, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8734177215189873, + "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": 0.7733333333333334, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7073170731707317, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.173602893948555, + "eval_overall_accuracy": 0.9591434900319961, + "eval_overall_f1": 0.8364611260053618, + "eval_overall_precision": 0.789873417721519, + "eval_overall_recall": 0.8888888888888888, + "eval_runtime": 0.3265, + "eval_samples_per_second": 572.76, + "eval_steps_per_second": 9.189, + "step": 10070 + }, + { + "epoch": 96.0, + "grad_norm": 0.8609771132469177, + "learning_rate": 2.0000000000000003e-06, + "loss": 0.0358, + "step": 10176 + }, + { + "epoch": 96.0, + "eval_LOCATION_f1": 0.8104575163398693, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.775, + "eval_LOCATION_recall": 0.8493150684931506, + "eval_ORGANIZATION_f1": 0.7808219178082192, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7037037037037037, + "eval_ORGANIZATION_recall": 0.8769230769230769, + "eval_PERSON_f1": 0.896103896103896, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8734177215189873, + "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": 0.7733333333333334, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7073170731707317, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.17425887286663055, + "eval_overall_accuracy": 0.958897366477972, + "eval_overall_f1": 0.8310991957104557, + "eval_overall_precision": 0.7848101265822784, + "eval_overall_recall": 0.8831908831908832, + "eval_runtime": 0.3272, + "eval_samples_per_second": 571.576, + "eval_steps_per_second": 9.17, + "step": 10176 + }, + { + "epoch": 97.0, + "grad_norm": 1.0554527044296265, + "learning_rate": 1.5e-06, + "loss": 0.0363, + "step": 10282 + }, + { + "epoch": 97.0, + "eval_LOCATION_f1": 0.8181818181818182, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7777777777777778, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.7862068965517242, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7125, + "eval_ORGANIZATION_recall": 0.8769230769230769, + "eval_PERSON_f1": 0.896103896103896, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8734177215189873, + "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": 0.7733333333333334, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7073170731707317, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.17255622148513794, + "eval_overall_accuracy": 0.9598818606940684, + "eval_overall_f1": 0.8364611260053618, + "eval_overall_precision": 0.789873417721519, + "eval_overall_recall": 0.8888888888888888, + "eval_runtime": 0.3257, + "eval_samples_per_second": 574.162, + "eval_steps_per_second": 9.211, + "step": 10282 + }, + { + "epoch": 98.0, + "grad_norm": 1.4366707801818848, + "learning_rate": 1.0000000000000002e-06, + "loss": 0.0354, + "step": 10388 + }, + { + "epoch": 98.0, + "eval_LOCATION_f1": 0.8104575163398693, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.775, + "eval_LOCATION_recall": 0.8493150684931506, + "eval_ORGANIZATION_f1": 0.7808219178082192, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7037037037037037, + "eval_ORGANIZATION_recall": 0.8769230769230769, + "eval_PERSON_f1": 0.896103896103896, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8734177215189873, + "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": 0.7733333333333334, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7073170731707317, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.17508530616760254, + "eval_overall_accuracy": 0.9586512429239479, + "eval_overall_f1": 0.8310991957104557, + "eval_overall_precision": 0.7848101265822784, + "eval_overall_recall": 0.8831908831908832, + "eval_runtime": 0.3287, + "eval_samples_per_second": 568.834, + "eval_steps_per_second": 9.126, + "step": 10388 + }, + { + "epoch": 99.0, + "grad_norm": 0.7106881141662598, + "learning_rate": 5.000000000000001e-07, + "loss": 0.0347, + "step": 10494 + }, + { + "epoch": 99.0, + "eval_LOCATION_f1": 0.8181818181818182, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7777777777777778, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.7808219178082192, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7037037037037037, + "eval_ORGANIZATION_recall": 0.8769230769230769, + "eval_PERSON_f1": 0.896103896103896, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8734177215189873, + "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": 0.7733333333333334, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7073170731707317, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.17459198832511902, + "eval_overall_accuracy": 0.9591434900319961, + "eval_overall_f1": 0.8353413654618473, + "eval_overall_precision": 0.7878787878787878, + "eval_overall_recall": 0.8888888888888888, + "eval_runtime": 0.326, + "eval_samples_per_second": 573.649, + "eval_steps_per_second": 9.203, + "step": 10494 + }, + { + "epoch": 100.0, + "grad_norm": 4.75438928604126, + "learning_rate": 0.0, + "loss": 0.0355, + "step": 10600 + }, + { + "epoch": 100.0, + "eval_LOCATION_f1": 0.8181818181818182, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7777777777777778, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.7808219178082192, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7037037037037037, + "eval_ORGANIZATION_recall": 0.8769230769230769, + "eval_PERSON_f1": 0.896103896103896, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8734177215189873, + "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": 0.7733333333333334, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7073170731707317, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.17432990670204163, + "eval_overall_accuracy": 0.958897366477972, + "eval_overall_f1": 0.8326639892904953, + "eval_overall_precision": 0.7853535353535354, + "eval_overall_recall": 0.886039886039886, + "eval_runtime": 0.3275, + "eval_samples_per_second": 570.981, + "eval_steps_per_second": 9.16, + "step": 10600 + }, + { + "epoch": 100.0, + "step": 10600, + "total_flos": 5043631411410816.0, + "train_loss": 0.08183759390183215, + "train_runtime": 561.9207, + "train_samples_per_second": 300.398, + "train_steps_per_second": 18.864 + } + ], + "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 +}