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": 1.9419891834259033, + "learning_rate": 4.9500000000000004e-05, + "loss": 0.932, + "step": 106 + }, + { + "epoch": 1.0, + "eval_LOCATION_f1": 0.0, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.0, + "eval_LOCATION_recall": 0.0, + "eval_ORGANIZATION_f1": 0.0, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.0, + "eval_ORGANIZATION_recall": 0.0, + "eval_PERSON_f1": 0.0, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.0, + "eval_PERSON_recall": 0.0, + "eval_QUANTITY_f1": 0.0, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.0, + "eval_QUANTITY_recall": 0.0, + "eval_TIME_f1": 0.0, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.0, + "eval_TIME_recall": 0.0, + "eval_loss": 0.6092155575752258, + "eval_overall_accuracy": 0.8448836627470603, + "eval_overall_f1": 0.0, + "eval_overall_precision": 0.0, + "eval_overall_recall": 0.0, + "eval_runtime": 0.3355, + "eval_samples_per_second": 557.364, + "eval_steps_per_second": 8.942, + "step": 106 + }, + { + "epoch": 2.0, + "grad_norm": 1.316730260848999, + "learning_rate": 4.9e-05, + "loss": 0.5616, + "step": 212 + }, + { + "epoch": 2.0, + "eval_LOCATION_f1": 0.2391304347826087, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.55, + "eval_LOCATION_recall": 0.1527777777777778, + "eval_ORGANIZATION_f1": 0.11764705882352941, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.1590909090909091, + "eval_ORGANIZATION_recall": 0.09333333333333334, + "eval_PERSON_f1": 0.4430379746835444, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.4046242774566474, + "eval_PERSON_recall": 0.48951048951048953, + "eval_QUANTITY_f1": 0.0, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.0, + "eval_QUANTITY_recall": 0.0, + "eval_TIME_f1": 0.41860465116279066, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.5294117647058824, + "eval_TIME_recall": 0.34615384615384615, + "eval_loss": 0.4137322008609772, + "eval_overall_accuracy": 0.8864148111083312, + "eval_overall_f1": 0.3265993265993266, + "eval_overall_precision": 0.3803921568627451, + "eval_overall_recall": 0.2861356932153392, + "eval_runtime": 0.3298, + "eval_samples_per_second": 566.957, + "eval_steps_per_second": 9.096, + "step": 212 + }, + { + "epoch": 3.0, + "grad_norm": 1.2939311265945435, + "learning_rate": 4.85e-05, + "loss": 0.4058, + "step": 318 + }, + { + "epoch": 3.0, + "eval_LOCATION_f1": 0.606896551724138, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.6027397260273972, + "eval_LOCATION_recall": 0.6111111111111112, + "eval_ORGANIZATION_f1": 0.4195804195804196, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.4411764705882353, + "eval_ORGANIZATION_recall": 0.4, + "eval_PERSON_f1": 0.6629213483146068, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.5539906103286385, + "eval_PERSON_recall": 0.8251748251748252, + "eval_QUANTITY_f1": 0.12499999999999997, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.12, + "eval_QUANTITY_recall": 0.13043478260869565, + "eval_TIME_f1": 0.76, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.7916666666666666, + "eval_TIME_recall": 0.7307692307692307, + "eval_loss": 0.3135295808315277, + "eval_overall_accuracy": 0.9209407055291469, + "eval_overall_f1": 0.5768194070080862, + "eval_overall_precision": 0.5310173697270472, + "eval_overall_recall": 0.6312684365781711, + "eval_runtime": 0.3299, + "eval_samples_per_second": 566.824, + "eval_steps_per_second": 9.093, + "step": 318 + }, + { + "epoch": 4.0, + "grad_norm": 2.3784005641937256, + "learning_rate": 4.8e-05, + "loss": 0.3002, + "step": 424 + }, + { + "epoch": 4.0, + "eval_LOCATION_f1": 0.6832298136645962, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.6179775280898876, + "eval_LOCATION_recall": 0.7638888888888888, + "eval_ORGANIZATION_f1": 0.5341614906832299, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.5, + "eval_ORGANIZATION_recall": 0.5733333333333334, + "eval_PERSON_f1": 0.7710843373493976, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.6772486772486772, + "eval_PERSON_recall": 0.8951048951048951, + "eval_QUANTITY_f1": 0.2909090909090909, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.25, + "eval_QUANTITY_recall": 0.34782608695652173, + "eval_TIME_f1": 0.8461538461538461, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.8461538461538461, + "eval_TIME_recall": 0.8461538461538461, + "eval_loss": 0.24471472203731537, + "eval_overall_accuracy": 0.9344508381285964, + "eval_overall_f1": 0.6727989487516426, + "eval_overall_precision": 0.6066350710900474, + "eval_overall_recall": 0.7551622418879056, + "eval_runtime": 0.3307, + "eval_samples_per_second": 565.504, + "eval_steps_per_second": 9.072, + "step": 424 + }, + { + "epoch": 5.0, + "grad_norm": 1.4117958545684814, + "learning_rate": 4.75e-05, + "loss": 0.2302, + "step": 530 + }, + { + "epoch": 5.0, + "eval_LOCATION_f1": 0.6993865030674847, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.6263736263736264, + "eval_LOCATION_recall": 0.7916666666666666, + "eval_ORGANIZATION_f1": 0.6, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.6, + "eval_ORGANIZATION_recall": 0.6, + "eval_PERSON_f1": 0.810289389067524, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.75, + "eval_PERSON_recall": 0.8811188811188811, + "eval_QUANTITY_f1": 0.3870967741935484, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.3076923076923077, + "eval_QUANTITY_recall": 0.5217391304347826, + "eval_TIME_f1": 0.8235294117647058, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.84, + "eval_TIME_recall": 0.8076923076923077, + "eval_loss": 0.1956184059381485, + "eval_overall_accuracy": 0.9409557167875907, + "eval_overall_f1": 0.7082767978290366, + "eval_overall_precision": 0.6557788944723618, + "eval_overall_recall": 0.7699115044247787, + "eval_runtime": 0.3289, + "eval_samples_per_second": 568.494, + "eval_steps_per_second": 9.12, + "step": 530 + }, + { + "epoch": 6.0, + "grad_norm": 1.079532265663147, + "learning_rate": 4.7e-05, + "loss": 0.1943, + "step": 636 + }, + { + "epoch": 6.0, + "eval_LOCATION_f1": 0.7643312101910827, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7058823529411765, + "eval_LOCATION_recall": 0.8333333333333334, + "eval_ORGANIZATION_f1": 0.6049382716049383, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.5632183908045977, + "eval_ORGANIZATION_recall": 0.6533333333333333, + "eval_PERSON_f1": 0.8327868852459016, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.7839506172839507, + "eval_PERSON_recall": 0.8881118881118881, + "eval_QUANTITY_f1": 0.4482758620689655, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.37142857142857144, + "eval_QUANTITY_recall": 0.5652173913043478, + "eval_TIME_f1": 0.8148148148148148, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.7857142857142857, + "eval_TIME_recall": 0.8461538461538461, + "eval_loss": 0.17118220031261444, + "eval_overall_accuracy": 0.94520890668001, + "eval_overall_f1": 0.7364130434782609, + "eval_overall_precision": 0.6826196473551638, + "eval_overall_recall": 0.799410029498525, + "eval_runtime": 0.3295, + "eval_samples_per_second": 567.478, + "eval_steps_per_second": 9.104, + "step": 636 + }, + { + "epoch": 7.0, + "grad_norm": 1.1560226678848267, + "learning_rate": 4.6500000000000005e-05, + "loss": 0.1777, + "step": 742 + }, + { + "epoch": 7.0, + "eval_LOCATION_f1": 0.7792207792207793, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7317073170731707, + "eval_LOCATION_recall": 0.8333333333333334, + "eval_ORGANIZATION_f1": 0.6588235294117647, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.5894736842105263, + "eval_ORGANIZATION_recall": 0.7466666666666667, + "eval_PERSON_f1": 0.8327868852459016, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.7839506172839507, + "eval_PERSON_recall": 0.8881118881118881, + "eval_QUANTITY_f1": 0.49180327868852464, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.39473684210526316, + "eval_QUANTITY_recall": 0.6521739130434783, + "eval_TIME_f1": 0.8, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.7586206896551724, + "eval_TIME_recall": 0.8461538461538461, + "eval_loss": 0.16843655705451965, + "eval_overall_accuracy": 0.9467100325243933, + "eval_overall_f1": 0.7516778523489932, + "eval_overall_precision": 0.6896551724137931, + "eval_overall_recall": 0.8259587020648967, + "eval_runtime": 0.3312, + "eval_samples_per_second": 564.545, + "eval_steps_per_second": 9.057, + "step": 742 + }, + { + "epoch": 8.0, + "grad_norm": 1.1952399015426636, + "learning_rate": 4.600000000000001e-05, + "loss": 0.162, + "step": 848 + }, + { + "epoch": 8.0, + "eval_LOCATION_f1": 0.7547169811320755, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.6896551724137931, + "eval_LOCATION_recall": 0.8333333333333334, + "eval_ORGANIZATION_f1": 0.641025641025641, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.6172839506172839, + "eval_ORGANIZATION_recall": 0.6666666666666666, + "eval_PERSON_f1": 0.8571428571428572, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8164556962025317, + "eval_PERSON_recall": 0.9020979020979021, + "eval_QUANTITY_f1": 0.5084745762711865, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.4166666666666667, + "eval_QUANTITY_recall": 0.6521739130434783, + "eval_TIME_f1": 0.7719298245614036, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.7096774193548387, + "eval_TIME_recall": 0.8461538461538461, + "eval_loss": 0.15775252878665924, + "eval_overall_accuracy": 0.9492119089316988, + "eval_overall_f1": 0.7540983606557377, + "eval_overall_precision": 0.7022900763358778, + "eval_overall_recall": 0.8141592920353983, + "eval_runtime": 0.323, + "eval_samples_per_second": 579.007, + "eval_steps_per_second": 9.289, + "step": 848 + }, + { + "epoch": 9.0, + "grad_norm": 1.3559876680374146, + "learning_rate": 4.55e-05, + "loss": 0.154, + "step": 954 + }, + { + "epoch": 9.0, + "eval_LOCATION_f1": 0.7721518987341772, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7093023255813954, + "eval_LOCATION_recall": 0.8472222222222222, + "eval_ORGANIZATION_f1": 0.6748466257668712, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.625, + "eval_ORGANIZATION_recall": 0.7333333333333333, + "eval_PERSON_f1": 0.8609271523178808, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8176100628930818, + "eval_PERSON_recall": 0.9090909090909091, + "eval_QUANTITY_f1": 0.5614035087719297, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.47058823529411764, + "eval_QUANTITY_recall": 0.6956521739130435, + "eval_TIME_f1": 0.8070175438596492, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.7419354838709677, + "eval_TIME_recall": 0.8846153846153846, + "eval_loss": 0.1556292623281479, + "eval_overall_accuracy": 0.9522141606204654, + "eval_overall_f1": 0.7734056987788331, + "eval_overall_precision": 0.7160804020100503, + "eval_overall_recall": 0.8407079646017699, + "eval_runtime": 0.3293, + "eval_samples_per_second": 567.872, + "eval_steps_per_second": 9.11, + "step": 954 + }, + { + "epoch": 10.0, + "grad_norm": 2.670650005340576, + "learning_rate": 4.5e-05, + "loss": 0.1445, + "step": 1060 + }, + { + "epoch": 10.0, + "eval_LOCATION_f1": 0.7870967741935484, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7349397590361446, + "eval_LOCATION_recall": 0.8472222222222222, + "eval_ORGANIZATION_f1": 0.6792452830188679, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.6428571428571429, + "eval_ORGANIZATION_recall": 0.72, + "eval_PERSON_f1": 0.8646864686468647, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.81875, + "eval_PERSON_recall": 0.916083916083916, + "eval_QUANTITY_f1": 0.5762711864406781, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.4722222222222222, + "eval_QUANTITY_recall": 0.7391304347826086, + "eval_TIME_f1": 0.7586206896551724, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.6875, + "eval_TIME_recall": 0.8461538461538461, + "eval_loss": 0.1476029008626938, + "eval_overall_accuracy": 0.952964723542657, + "eval_overall_f1": 0.7765667574931878, + "eval_overall_precision": 0.7215189873417721, + "eval_overall_recall": 0.8407079646017699, + "eval_runtime": 0.3316, + "eval_samples_per_second": 563.983, + "eval_steps_per_second": 9.048, + "step": 1060 + }, + { + "epoch": 11.0, + "grad_norm": 1.2458206415176392, + "learning_rate": 4.4500000000000004e-05, + "loss": 0.1367, + "step": 1166 + }, + { + "epoch": 11.0, + "eval_LOCATION_f1": 0.7848101265822784, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7209302325581395, + "eval_LOCATION_recall": 0.8611111111111112, + "eval_ORGANIZATION_f1": 0.6956521739130435, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.6511627906976745, + "eval_ORGANIZATION_recall": 0.7466666666666667, + "eval_PERSON_f1": 0.8762541806020068, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8397435897435898, + "eval_PERSON_recall": 0.916083916083916, + "eval_QUANTITY_f1": 0.5862068965517241, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.4857142857142857, + "eval_QUANTITY_recall": 0.7391304347826086, + "eval_TIME_f1": 0.7586206896551724, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.6875, + "eval_TIME_recall": 0.8461538461538461, + "eval_loss": 0.14486071467399597, + "eval_overall_accuracy": 0.9544658493870403, + "eval_overall_f1": 0.784741144414169, + "eval_overall_precision": 0.7291139240506329, + "eval_overall_recall": 0.8495575221238938, + "eval_runtime": 0.3237, + "eval_samples_per_second": 577.663, + "eval_steps_per_second": 9.267, + "step": 1166 + }, + { + "epoch": 12.0, + "grad_norm": 21.2458438873291, + "learning_rate": 4.4000000000000006e-05, + "loss": 0.132, + "step": 1272 + }, + { + "epoch": 12.0, + "eval_LOCATION_f1": 0.781456953642384, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7468354430379747, + "eval_LOCATION_recall": 0.8194444444444444, + "eval_ORGANIZATION_f1": 0.6666666666666667, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.6419753086419753, + "eval_ORGANIZATION_recall": 0.6933333333333334, + "eval_PERSON_f1": 0.8637873754152824, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8227848101265823, + "eval_PERSON_recall": 0.9090909090909091, + "eval_QUANTITY_f1": 0.6071428571428571, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5151515151515151, + "eval_QUANTITY_recall": 0.7391304347826086, + "eval_TIME_f1": 0.8214285714285715, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.7666666666666667, + "eval_TIME_recall": 0.8846153846153846, + "eval_loss": 0.14586253464221954, + "eval_overall_accuracy": 0.9522141606204654, + "eval_overall_f1": 0.7805555555555554, + "eval_overall_precision": 0.7375328083989501, + "eval_overall_recall": 0.8289085545722714, + "eval_runtime": 0.329, + "eval_samples_per_second": 568.391, + "eval_steps_per_second": 9.119, + "step": 1272 + }, + { + "epoch": 13.0, + "grad_norm": 0.7611563205718994, + "learning_rate": 4.35e-05, + "loss": 0.1276, + "step": 1378 + }, + { + "epoch": 13.0, + "eval_LOCATION_f1": 0.7763157894736842, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7375, + "eval_LOCATION_recall": 0.8194444444444444, + "eval_ORGANIZATION_f1": 0.728395061728395, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.6781609195402298, + "eval_ORGANIZATION_recall": 0.7866666666666666, + "eval_PERSON_f1": 0.8762541806020068, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8397435897435898, + "eval_PERSON_recall": 0.916083916083916, + "eval_QUANTITY_f1": 0.6538461538461539, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5862068965517241, + "eval_QUANTITY_recall": 0.7391304347826086, + "eval_TIME_f1": 0.8518518518518519, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.8214285714285714, + "eval_TIME_recall": 0.8846153846153846, + "eval_loss": 0.13649380207061768, + "eval_overall_accuracy": 0.9572179134350763, + "eval_overall_f1": 0.803894297635605, + "eval_overall_precision": 0.7605263157894737, + "eval_overall_recall": 0.8525073746312685, + "eval_runtime": 0.3294, + "eval_samples_per_second": 567.689, + "eval_steps_per_second": 9.107, + "step": 1378 + }, + { + "epoch": 14.0, + "grad_norm": 1.2182074785232544, + "learning_rate": 4.3e-05, + "loss": 0.1248, + "step": 1484 + }, + { + "epoch": 14.0, + "eval_LOCATION_f1": 0.8079470198675496, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7721518987341772, + "eval_LOCATION_recall": 0.8472222222222222, + "eval_ORGANIZATION_f1": 0.7485380116959064, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.6666666666666666, + "eval_ORGANIZATION_recall": 0.8533333333333334, + "eval_PERSON_f1": 0.8762541806020068, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8397435897435898, + "eval_PERSON_recall": 0.916083916083916, + "eval_QUANTITY_f1": 0.6181818181818182, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.53125, + "eval_QUANTITY_recall": 0.7391304347826086, + "eval_TIME_f1": 0.7857142857142856, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.7333333333333333, + "eval_TIME_recall": 0.8461538461538461, + "eval_loss": 0.1439991444349289, + "eval_overall_accuracy": 0.9562171628721541, + "eval_overall_f1": 0.8060109289617485, + "eval_overall_precision": 0.7506361323155216, + "eval_overall_recall": 0.8702064896755162, + "eval_runtime": 0.3229, + "eval_samples_per_second": 579.183, + "eval_steps_per_second": 9.292, + "step": 1484 + }, + { + "epoch": 15.0, + "grad_norm": 0.442609965801239, + "learning_rate": 4.25e-05, + "loss": 0.1209, + "step": 1590 + }, + { + "epoch": 15.0, + "eval_LOCATION_f1": 0.7919463087248323, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7662337662337663, + "eval_LOCATION_recall": 0.8194444444444444, + "eval_ORGANIZATION_f1": 0.7425149700598803, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.6739130434782609, + "eval_ORGANIZATION_recall": 0.8266666666666667, + "eval_PERSON_f1": 0.8737201365187713, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8533333333333334, + "eval_PERSON_recall": 0.8951048951048951, + "eval_QUANTITY_f1": 0.6415094339622641, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5666666666666667, + "eval_QUANTITY_recall": 0.7391304347826086, + "eval_TIME_f1": 0.8148148148148148, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.7857142857142857, + "eval_TIME_recall": 0.8461538461538461, + "eval_loss": 0.13672932982444763, + "eval_overall_accuracy": 0.9577182887165374, + "eval_overall_f1": 0.8044692737430166, + "eval_overall_precision": 0.7639257294429708, + "eval_overall_recall": 0.8495575221238938, + "eval_runtime": 0.3298, + "eval_samples_per_second": 566.949, + "eval_steps_per_second": 9.095, + "step": 1590 + }, + { + "epoch": 16.0, + "grad_norm": 0.9648889899253845, + "learning_rate": 4.2e-05, + "loss": 0.1172, + "step": 1696 + }, + { + "epoch": 16.0, + "eval_LOCATION_f1": 0.7870967741935484, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7349397590361446, + "eval_LOCATION_recall": 0.8472222222222222, + "eval_ORGANIZATION_f1": 0.7710843373493976, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.7032967032967034, + "eval_ORGANIZATION_recall": 0.8533333333333334, + "eval_PERSON_f1": 0.8800000000000001, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8407643312101911, + "eval_PERSON_recall": 0.9230769230769231, + "eval_QUANTITY_f1": 0.6071428571428571, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5151515151515151, + "eval_QUANTITY_recall": 0.7391304347826086, + "eval_TIME_f1": 0.8, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.7586206896551724, + "eval_TIME_recall": 0.8461538461538461, + "eval_loss": 0.1445225328207016, + "eval_overall_accuracy": 0.9562171628721541, + "eval_overall_f1": 0.8087431693989071, + "eval_overall_precision": 0.7531806615776081, + "eval_overall_recall": 0.8731563421828908, + "eval_runtime": 0.3239, + "eval_samples_per_second": 577.418, + "eval_steps_per_second": 9.263, + "step": 1696 + }, + { + "epoch": 17.0, + "grad_norm": 0.29116642475128174, + "learning_rate": 4.15e-05, + "loss": 0.1115, + "step": 1802 + }, + { + "epoch": 17.0, + "eval_LOCATION_f1": 0.8, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7692307692307693, + "eval_LOCATION_recall": 0.8333333333333334, + "eval_ORGANIZATION_f1": 0.8, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.7333333333333333, + "eval_ORGANIZATION_recall": 0.88, + "eval_PERSON_f1": 0.8829431438127091, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8461538461538461, + "eval_PERSON_recall": 0.9230769230769231, + "eval_QUANTITY_f1": 0.5964912280701754, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5, + "eval_QUANTITY_recall": 0.7391304347826086, + "eval_TIME_f1": 0.8148148148148148, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.7857142857142857, + "eval_TIME_recall": 0.8461538461538461, + "eval_loss": 0.13870589435100555, + "eval_overall_accuracy": 0.957968476357268, + "eval_overall_f1": 0.8193103448275862, + "eval_overall_precision": 0.7694300518134715, + "eval_overall_recall": 0.8761061946902655, + "eval_runtime": 0.33, + "eval_samples_per_second": 566.697, + "eval_steps_per_second": 9.091, + "step": 1802 + }, + { + "epoch": 18.0, + "grad_norm": 0.9271082282066345, + "learning_rate": 4.1e-05, + "loss": 0.1103, + "step": 1908 + }, + { + "epoch": 18.0, + "eval_LOCATION_f1": 0.8, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7469879518072289, + "eval_LOCATION_recall": 0.8611111111111112, + "eval_ORGANIZATION_f1": 0.7826086956521738, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.7325581395348837, + "eval_ORGANIZATION_recall": 0.84, + "eval_PERSON_f1": 0.8851351351351351, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8562091503267973, + "eval_PERSON_recall": 0.916083916083916, + "eval_QUANTITY_f1": 0.631578947368421, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5294117647058824, + "eval_QUANTITY_recall": 0.782608695652174, + "eval_TIME_f1": 0.8, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.7586206896551724, + "eval_TIME_recall": 0.8461538461538461, + "eval_loss": 0.13877204060554504, + "eval_overall_accuracy": 0.9589692269201902, + "eval_overall_f1": 0.8176795580110496, + "eval_overall_precision": 0.7688311688311689, + "eval_overall_recall": 0.8731563421828908, + "eval_runtime": 0.3303, + "eval_samples_per_second": 566.086, + "eval_steps_per_second": 9.082, + "step": 1908 + }, + { + "epoch": 19.0, + "grad_norm": 1.1867777109146118, + "learning_rate": 4.05e-05, + "loss": 0.1071, + "step": 2014 + }, + { + "epoch": 19.0, + "eval_LOCATION_f1": 0.7919463087248323, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7662337662337663, + "eval_LOCATION_recall": 0.8194444444444444, + "eval_ORGANIZATION_f1": 0.7950310559006211, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.7441860465116279, + "eval_ORGANIZATION_recall": 0.8533333333333334, + "eval_PERSON_f1": 0.8903654485049833, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8481012658227848, + "eval_PERSON_recall": 0.9370629370629371, + "eval_QUANTITY_f1": 0.5964912280701754, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5, + "eval_QUANTITY_recall": 0.7391304347826086, + "eval_TIME_f1": 0.8, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.7586206896551724, + "eval_TIME_recall": 0.8461538461538461, + "eval_loss": 0.1341163069009781, + "eval_overall_accuracy": 0.9594696022016512, + "eval_overall_f1": 0.8188105117565698, + "eval_overall_precision": 0.7708333333333334, + "eval_overall_recall": 0.8731563421828908, + "eval_runtime": 0.3285, + "eval_samples_per_second": 569.33, + "eval_steps_per_second": 9.134, + "step": 2014 + }, + { + "epoch": 20.0, + "grad_norm": 1.6215965747833252, + "learning_rate": 4e-05, + "loss": 0.1028, + "step": 2120 + }, + { + "epoch": 20.0, + "eval_LOCATION_f1": 0.7625000000000001, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.6931818181818182, + "eval_LOCATION_recall": 0.8472222222222222, + "eval_ORGANIZATION_f1": 0.7597765363128492, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.6538461538461539, + "eval_ORGANIZATION_recall": 0.9066666666666666, + "eval_PERSON_f1": 0.8874172185430464, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8427672955974843, + "eval_PERSON_recall": 0.9370629370629371, + "eval_QUANTITY_f1": 0.6071428571428571, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5151515151515151, + "eval_QUANTITY_recall": 0.7391304347826086, + "eval_TIME_f1": 0.7118644067796609, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.6363636363636364, + "eval_TIME_recall": 0.8076923076923077, + "eval_loss": 0.15599055588245392, + "eval_overall_accuracy": 0.9519639729797348, + "eval_overall_f1": 0.7962962962962963, + "eval_overall_precision": 0.7218225419664268, + "eval_overall_recall": 0.887905604719764, + "eval_runtime": 0.3285, + "eval_samples_per_second": 569.337, + "eval_steps_per_second": 9.134, + "step": 2120 + }, + { + "epoch": 21.0, + "grad_norm": 1.221657633781433, + "learning_rate": 3.9500000000000005e-05, + "loss": 0.1016, + "step": 2226 + }, + { + "epoch": 21.0, + "eval_LOCATION_f1": 0.7898089171974522, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7294117647058823, + "eval_LOCATION_recall": 0.8611111111111112, + "eval_ORGANIZATION_f1": 0.8, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.7, + "eval_ORGANIZATION_recall": 0.9333333333333333, + "eval_PERSON_f1": 0.8910891089108911, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.84375, + "eval_PERSON_recall": 0.9440559440559441, + "eval_QUANTITY_f1": 0.6428571428571429, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5454545454545454, + "eval_QUANTITY_recall": 0.782608695652174, + "eval_TIME_f1": 0.8, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.7586206896551724, + "eval_TIME_recall": 0.8461538461538461, + "eval_loss": 0.1470298022031784, + "eval_overall_accuracy": 0.9567175381536153, + "eval_overall_f1": 0.8230563002680964, + "eval_overall_precision": 0.7542997542997543, + "eval_overall_recall": 0.9056047197640118, + "eval_runtime": 0.3301, + "eval_samples_per_second": 566.44, + "eval_steps_per_second": 9.087, + "step": 2226 + }, + { + "epoch": 22.0, + "grad_norm": 1.1022194623947144, + "learning_rate": 3.9000000000000006e-05, + "loss": 0.0999, + "step": 2332 + }, + { + "epoch": 22.0, + "eval_LOCATION_f1": 0.7866666666666666, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7564102564102564, + "eval_LOCATION_recall": 0.8194444444444444, + "eval_ORGANIZATION_f1": 0.8095238095238095, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.7311827956989247, + "eval_ORGANIZATION_recall": 0.9066666666666666, + "eval_PERSON_f1": 0.8896321070234114, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8525641025641025, + "eval_PERSON_recall": 0.9300699300699301, + "eval_QUANTITY_f1": 0.6071428571428571, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5151515151515151, + "eval_QUANTITY_recall": 0.7391304347826086, + "eval_TIME_f1": 0.8, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.7586206896551724, + "eval_TIME_recall": 0.8461538461538461, + "eval_loss": 0.1353609561920166, + "eval_overall_accuracy": 0.9597197898423818, + "eval_overall_f1": 0.8214285714285714, + "eval_overall_precision": 0.7686375321336761, + "eval_overall_recall": 0.8820058997050148, + "eval_runtime": 0.3274, + "eval_samples_per_second": 571.233, + "eval_steps_per_second": 9.164, + "step": 2332 + }, + { + "epoch": 23.0, + "grad_norm": 0.7958564758300781, + "learning_rate": 3.85e-05, + "loss": 0.0953, + "step": 2438 + }, + { + "epoch": 23.0, + "eval_LOCATION_f1": 0.802721088435374, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7866666666666666, + "eval_LOCATION_recall": 0.8194444444444444, + "eval_ORGANIZATION_f1": 0.8165680473372782, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.7340425531914894, + "eval_ORGANIZATION_recall": 0.92, + "eval_PERSON_f1": 0.8859060402684563, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8516129032258064, + "eval_PERSON_recall": 0.9230769230769231, + "eval_QUANTITY_f1": 0.5964912280701754, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5, + "eval_QUANTITY_recall": 0.7391304347826086, + "eval_TIME_f1": 0.8, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.7586206896551724, + "eval_TIME_recall": 0.8461538461538461, + "eval_loss": 0.13582873344421387, + "eval_overall_accuracy": 0.9597197898423818, + "eval_overall_f1": 0.8236914600550964, + "eval_overall_precision": 0.772609819121447, + "eval_overall_recall": 0.8820058997050148, + "eval_runtime": 0.3286, + "eval_samples_per_second": 569.058, + "eval_steps_per_second": 9.129, + "step": 2438 + }, + { + "epoch": 24.0, + "grad_norm": 1.488933801651001, + "learning_rate": 3.8e-05, + "loss": 0.0932, + "step": 2544 + }, + { + "epoch": 24.0, + "eval_LOCATION_f1": 0.7973856209150327, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7530864197530864, + "eval_LOCATION_recall": 0.8472222222222222, + "eval_ORGANIZATION_f1": 0.7976190476190477, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.7204301075268817, + "eval_ORGANIZATION_recall": 0.8933333333333333, + "eval_PERSON_f1": 0.8896321070234114, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8525641025641025, + "eval_PERSON_recall": 0.9300699300699301, + "eval_QUANTITY_f1": 0.5517241379310345, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.45714285714285713, + "eval_QUANTITY_recall": 0.6956521739130435, + "eval_TIME_f1": 0.7857142857142856, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.7333333333333333, + "eval_TIME_recall": 0.8461538461538461, + "eval_loss": 0.14096248149871826, + "eval_overall_accuracy": 0.9574681010758068, + "eval_overall_f1": 0.8147138964577656, + "eval_overall_precision": 0.7569620253164557, + "eval_overall_recall": 0.8820058997050148, + "eval_runtime": 0.3282, + "eval_samples_per_second": 569.836, + "eval_steps_per_second": 9.142, + "step": 2544 + }, + { + "epoch": 25.0, + "grad_norm": 1.2902323007583618, + "learning_rate": 3.7500000000000003e-05, + "loss": 0.0914, + "step": 2650 + }, + { + "epoch": 25.0, + "eval_LOCATION_f1": 0.7973856209150327, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7530864197530864, + "eval_LOCATION_recall": 0.8472222222222222, + "eval_ORGANIZATION_f1": 0.7953216374269007, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.7083333333333334, + "eval_ORGANIZATION_recall": 0.9066666666666666, + "eval_PERSON_f1": 0.8859060402684563, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8516129032258064, + "eval_PERSON_recall": 0.9230769230769231, + "eval_QUANTITY_f1": 0.5964912280701754, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5, + "eval_QUANTITY_recall": 0.7391304347826086, + "eval_TIME_f1": 0.8, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.7586206896551724, + "eval_TIME_recall": 0.8461538461538461, + "eval_loss": 0.13958008587360382, + "eval_overall_accuracy": 0.9584688516387291, + "eval_overall_f1": 0.8174386920980926, + "eval_overall_precision": 0.759493670886076, + "eval_overall_recall": 0.8849557522123894, + "eval_runtime": 0.3288, + "eval_samples_per_second": 568.764, + "eval_steps_per_second": 9.125, + "step": 2650 + }, + { + "epoch": 26.0, + "grad_norm": 1.1934072971343994, + "learning_rate": 3.7e-05, + "loss": 0.0898, + "step": 2756 + }, + { + "epoch": 26.0, + "eval_LOCATION_f1": 0.802721088435374, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7866666666666666, + "eval_LOCATION_recall": 0.8194444444444444, + "eval_ORGANIZATION_f1": 0.8072289156626506, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.7362637362637363, + "eval_ORGANIZATION_recall": 0.8933333333333333, + "eval_PERSON_f1": 0.8866666666666668, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8471337579617835, + "eval_PERSON_recall": 0.9300699300699301, + "eval_QUANTITY_f1": 0.5925925925925926, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5161290322580645, + "eval_QUANTITY_recall": 0.6956521739130435, + "eval_TIME_f1": 0.7777777777777779, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.75, + "eval_TIME_recall": 0.8076923076923077, + "eval_loss": 0.13738316297531128, + "eval_overall_accuracy": 0.9594696022016512, + "eval_overall_f1": 0.8210818307905686, + "eval_overall_precision": 0.774869109947644, + "eval_overall_recall": 0.8731563421828908, + "eval_runtime": 0.3283, + "eval_samples_per_second": 569.525, + "eval_steps_per_second": 9.137, + "step": 2756 + }, + { + "epoch": 27.0, + "grad_norm": 1.3343477249145508, + "learning_rate": 3.65e-05, + "loss": 0.0862, + "step": 2862 + }, + { + "epoch": 27.0, + "eval_LOCATION_f1": 0.7741935483870969, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7228915662650602, + "eval_LOCATION_recall": 0.8333333333333334, + "eval_ORGANIZATION_f1": 0.7878787878787877, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.7222222222222222, + "eval_ORGANIZATION_recall": 0.8666666666666667, + "eval_PERSON_f1": 0.8874172185430464, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8427672955974843, + "eval_PERSON_recall": 0.9370629370629371, + "eval_QUANTITY_f1": 0.6101694915254238, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5, + "eval_QUANTITY_recall": 0.782608695652174, + "eval_TIME_f1": 0.8, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.7586206896551724, + "eval_TIME_recall": 0.8461538461538461, + "eval_loss": 0.1477619707584381, + "eval_overall_accuracy": 0.957968476357268, + "eval_overall_f1": 0.8125000000000001, + "eval_overall_precision": 0.7531486146095718, + "eval_overall_recall": 0.8820058997050148, + "eval_runtime": 0.328, + "eval_samples_per_second": 570.203, + "eval_steps_per_second": 9.148, + "step": 2862 + }, + { + "epoch": 28.0, + "grad_norm": 0.5415946841239929, + "learning_rate": 3.6e-05, + "loss": 0.0847, + "step": 2968 + }, + { + "epoch": 28.0, + "eval_LOCATION_f1": 0.7894736842105262, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.75, + "eval_LOCATION_recall": 0.8333333333333334, + "eval_ORGANIZATION_f1": 0.8, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.7333333333333333, + "eval_ORGANIZATION_recall": 0.88, + "eval_PERSON_f1": 0.8859060402684563, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8516129032258064, + "eval_PERSON_recall": 0.9230769230769231, + "eval_QUANTITY_f1": 0.6037735849056605, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5333333333333333, + "eval_QUANTITY_recall": 0.6956521739130435, + "eval_TIME_f1": 0.8, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.7586206896551724, + "eval_TIME_recall": 0.8461538461538461, + "eval_loss": 0.13617141544818878, + "eval_overall_accuracy": 0.9599699774831123, + "eval_overall_f1": 0.8188105117565698, + "eval_overall_precision": 0.7708333333333334, + "eval_overall_recall": 0.8731563421828908, + "eval_runtime": 0.3285, + "eval_samples_per_second": 569.272, + "eval_steps_per_second": 9.133, + "step": 2968 + }, + { + "epoch": 29.0, + "grad_norm": 0.9468500018119812, + "learning_rate": 3.55e-05, + "loss": 0.0837, + "step": 3074 + }, + { + "epoch": 29.0, + "eval_LOCATION_f1": 0.8026315789473685, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7625, + "eval_LOCATION_recall": 0.8472222222222222, + "eval_ORGANIZATION_f1": 0.782122905027933, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.6730769230769231, + "eval_ORGANIZATION_recall": 0.9333333333333333, + "eval_PERSON_f1": 0.8821548821548821, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8506493506493507, + "eval_PERSON_recall": 0.916083916083916, + "eval_QUANTITY_f1": 0.6181818181818182, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.53125, + "eval_QUANTITY_recall": 0.7391304347826086, + "eval_TIME_f1": 0.8148148148148148, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.7857142857142857, + "eval_TIME_recall": 0.8461538461538461, + "eval_loss": 0.14570099115371704, + "eval_overall_accuracy": 0.9572179134350763, + "eval_overall_f1": 0.8168249660786975, + "eval_overall_precision": 0.7562814070351759, + "eval_overall_recall": 0.887905604719764, + "eval_runtime": 0.3283, + "eval_samples_per_second": 569.654, + "eval_steps_per_second": 9.139, + "step": 3074 + }, + { + "epoch": 30.0, + "grad_norm": 0.7011334896087646, + "learning_rate": 3.5e-05, + "loss": 0.0823, + "step": 3180 + }, + { + "epoch": 30.0, + "eval_LOCATION_f1": 0.7972972972972973, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7763157894736842, + "eval_LOCATION_recall": 0.8194444444444444, + "eval_ORGANIZATION_f1": 0.7951807228915663, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.7252747252747253, + "eval_ORGANIZATION_recall": 0.88, + "eval_PERSON_f1": 0.8918918918918919, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8627450980392157, + "eval_PERSON_recall": 0.9230769230769231, + "eval_QUANTITY_f1": 0.5660377358490566, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5, + "eval_QUANTITY_recall": 0.6521739130434783, + "eval_TIME_f1": 0.8148148148148148, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.7857142857142857, + "eval_TIME_recall": 0.8461538461538461, + "eval_loss": 0.13126899302005768, + "eval_overall_accuracy": 0.9617212909682261, + "eval_overall_f1": 0.8200836820083682, + "eval_overall_precision": 0.7777777777777778, + "eval_overall_recall": 0.8672566371681416, + "eval_runtime": 0.329, + "eval_samples_per_second": 568.471, + "eval_steps_per_second": 9.12, + "step": 3180 + }, + { + "epoch": 31.0, + "grad_norm": 1.1392722129821777, + "learning_rate": 3.45e-05, + "loss": 0.0818, + "step": 3286 + }, + { + "epoch": 31.0, + "eval_LOCATION_f1": 0.778523489932886, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7532467532467533, + "eval_LOCATION_recall": 0.8055555555555556, + "eval_ORGANIZATION_f1": 0.7672955974842767, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.7261904761904762, + "eval_ORGANIZATION_recall": 0.8133333333333334, + "eval_PERSON_f1": 0.8896321070234114, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8525641025641025, + "eval_PERSON_recall": 0.9300699300699301, + "eval_QUANTITY_f1": 0.5925925925925926, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5161290322580645, + "eval_QUANTITY_recall": 0.6956521739130435, + "eval_TIME_f1": 0.7924528301886792, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.7777777777777778, + "eval_TIME_recall": 0.8076923076923077, + "eval_loss": 0.13312852382659912, + "eval_overall_accuracy": 0.9599699774831123, + "eval_overall_f1": 0.8095238095238095, + "eval_overall_precision": 0.7706666666666667, + "eval_overall_recall": 0.8525073746312685, + "eval_runtime": 0.3227, + "eval_samples_per_second": 579.451, + "eval_steps_per_second": 9.296, + "step": 3286 + }, + { + "epoch": 32.0, + "grad_norm": 2.0022265911102295, + "learning_rate": 3.4000000000000007e-05, + "loss": 0.0791, + "step": 3392 + }, + { + "epoch": 32.0, + "eval_LOCATION_f1": 0.7792207792207793, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7317073170731707, + "eval_LOCATION_recall": 0.8333333333333334, + "eval_ORGANIZATION_f1": 0.7976190476190477, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.7204301075268817, + "eval_ORGANIZATION_recall": 0.8933333333333333, + "eval_PERSON_f1": 0.8918918918918919, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8627450980392157, + "eval_PERSON_recall": 0.9230769230769231, + "eval_QUANTITY_f1": 0.5964912280701754, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5, + "eval_QUANTITY_recall": 0.7391304347826086, + "eval_TIME_f1": 0.7777777777777779, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.75, + "eval_TIME_recall": 0.8076923076923077, + "eval_loss": 0.1390967071056366, + "eval_overall_accuracy": 0.960720540405304, + "eval_overall_f1": 0.8148148148148148, + "eval_overall_precision": 0.7615384615384615, + "eval_overall_recall": 0.8761061946902655, + "eval_runtime": 0.3226, + "eval_samples_per_second": 579.685, + "eval_steps_per_second": 9.3, + "step": 3392 + }, + { + "epoch": 33.0, + "grad_norm": 2.273530960083008, + "learning_rate": 3.35e-05, + "loss": 0.0767, + "step": 3498 + }, + { + "epoch": 33.0, + "eval_LOCATION_f1": 0.7919463087248323, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7662337662337663, + "eval_LOCATION_recall": 0.8194444444444444, + "eval_ORGANIZATION_f1": 0.7976190476190477, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.7204301075268817, + "eval_ORGANIZATION_recall": 0.8933333333333333, + "eval_PERSON_f1": 0.903010033444816, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8653846153846154, + "eval_PERSON_recall": 0.9440559440559441, + "eval_QUANTITY_f1": 0.5555555555555556, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.4838709677419355, + "eval_QUANTITY_recall": 0.6521739130434783, + "eval_TIME_f1": 0.7857142857142856, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.7333333333333333, + "eval_TIME_recall": 0.8461538461538461, + "eval_loss": 0.13481685519218445, + "eval_overall_accuracy": 0.961220915686765, + "eval_overall_f1": 0.8209366391184572, + "eval_overall_precision": 0.7700258397932817, + "eval_overall_recall": 0.8790560471976401, + "eval_runtime": 0.3255, + "eval_samples_per_second": 574.529, + "eval_steps_per_second": 9.217, + "step": 3498 + }, + { + "epoch": 34.0, + "grad_norm": 1.426448941230774, + "learning_rate": 3.3e-05, + "loss": 0.0747, + "step": 3604 + }, + { + "epoch": 34.0, + "eval_LOCATION_f1": 0.7792207792207793, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7317073170731707, + "eval_LOCATION_recall": 0.8333333333333334, + "eval_ORGANIZATION_f1": 0.7861271676300577, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.6938775510204082, + "eval_ORGANIZATION_recall": 0.9066666666666666, + "eval_PERSON_f1": 0.8874172185430464, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8427672955974843, + "eval_PERSON_recall": 0.9370629370629371, + "eval_QUANTITY_f1": 0.5964912280701754, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5, + "eval_QUANTITY_recall": 0.7391304347826086, + "eval_TIME_f1": 0.7857142857142856, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.7333333333333333, + "eval_TIME_recall": 0.8461538461538461, + "eval_loss": 0.14789237082004547, + "eval_overall_accuracy": 0.9569677257943457, + "eval_overall_f1": 0.8113207547169812, + "eval_overall_precision": 0.7468982630272953, + "eval_overall_recall": 0.887905604719764, + "eval_runtime": 0.3283, + "eval_samples_per_second": 569.537, + "eval_steps_per_second": 9.137, + "step": 3604 + }, + { + "epoch": 35.0, + "grad_norm": 3.439316749572754, + "learning_rate": 3.2500000000000004e-05, + "loss": 0.0748, + "step": 3710 + }, + { + "epoch": 35.0, + "eval_LOCATION_f1": 0.782051282051282, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7261904761904762, + "eval_LOCATION_recall": 0.8472222222222222, + "eval_ORGANIZATION_f1": 0.7225130890052356, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.5948275862068966, + "eval_ORGANIZATION_recall": 0.92, + "eval_PERSON_f1": 0.9016949152542374, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.875, + "eval_PERSON_recall": 0.9300699300699301, + "eval_QUANTITY_f1": 0.5964912280701754, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5, + "eval_QUANTITY_recall": 0.7391304347826086, + "eval_TIME_f1": 0.8, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.7586206896551724, + "eval_TIME_recall": 0.8461538461538461, + "eval_loss": 0.15996821224689484, + "eval_overall_accuracy": 0.9524643482611959, + "eval_overall_f1": 0.8010610079575596, + "eval_overall_precision": 0.727710843373494, + "eval_overall_recall": 0.8908554572271387, + "eval_runtime": 0.3302, + "eval_samples_per_second": 566.316, + "eval_steps_per_second": 9.085, + "step": 3710 + }, + { + "epoch": 36.0, + "grad_norm": 1.1632362604141235, + "learning_rate": 3.2000000000000005e-05, + "loss": 0.0748, + "step": 3816 + }, + { + "epoch": 36.0, + "eval_LOCATION_f1": 0.7843137254901961, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7407407407407407, + "eval_LOCATION_recall": 0.8333333333333334, + "eval_ORGANIZATION_f1": 0.788235294117647, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.7052631578947368, + "eval_ORGANIZATION_recall": 0.8933333333333333, + "eval_PERSON_f1": 0.8933333333333333, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8535031847133758, + "eval_PERSON_recall": 0.9370629370629371, + "eval_QUANTITY_f1": 0.576923076923077, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5172413793103449, + "eval_QUANTITY_recall": 0.6521739130434783, + "eval_TIME_f1": 0.8148148148148148, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.7857142857142857, + "eval_TIME_recall": 0.8461538461538461, + "eval_loss": 0.14071466028690338, + "eval_overall_accuracy": 0.9599699774831123, + "eval_overall_f1": 0.8175582990397804, + "eval_overall_precision": 0.764102564102564, + "eval_overall_recall": 0.8790560471976401, + "eval_runtime": 0.3292, + "eval_samples_per_second": 567.98, + "eval_steps_per_second": 9.112, + "step": 3816 + }, + { + "epoch": 37.0, + "grad_norm": 2.102585792541504, + "learning_rate": 3.15e-05, + "loss": 0.0694, + "step": 3922 + }, + { + "epoch": 37.0, + "eval_LOCATION_f1": 0.7755102040816326, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.76, + "eval_LOCATION_recall": 0.7916666666666666, + "eval_ORGANIZATION_f1": 0.7840909090909092, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.6831683168316832, + "eval_ORGANIZATION_recall": 0.92, + "eval_PERSON_f1": 0.8926174496644295, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8580645161290322, + "eval_PERSON_recall": 0.9300699300699301, + "eval_QUANTITY_f1": 0.5882352941176471, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5357142857142857, + "eval_QUANTITY_recall": 0.6521739130434783, + "eval_TIME_f1": 0.8, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.7586206896551724, + "eval_TIME_recall": 0.8461538461538461, + "eval_loss": 0.14074602723121643, + "eval_overall_accuracy": 0.9599699774831123, + "eval_overall_f1": 0.8143053645116919, + "eval_overall_precision": 0.7628865979381443, + "eval_overall_recall": 0.8731563421828908, + "eval_runtime": 0.3286, + "eval_samples_per_second": 569.138, + "eval_steps_per_second": 9.131, + "step": 3922 + }, + { + "epoch": 38.0, + "grad_norm": 1.254451870918274, + "learning_rate": 3.1e-05, + "loss": 0.0674, + "step": 4028 + }, + { + "epoch": 38.0, + "eval_LOCATION_f1": 0.7837837837837838, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7631578947368421, + "eval_LOCATION_recall": 0.8055555555555556, + "eval_ORGANIZATION_f1": 0.7928994082840236, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.7127659574468085, + "eval_ORGANIZATION_recall": 0.8933333333333333, + "eval_PERSON_f1": 0.8963210702341137, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8589743589743589, + "eval_PERSON_recall": 0.9370629370629371, + "eval_QUANTITY_f1": 0.5925925925925926, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5161290322580645, + "eval_QUANTITY_recall": 0.6956521739130435, + "eval_TIME_f1": 0.7777777777777779, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.75, + "eval_TIME_recall": 0.8076923076923077, + "eval_loss": 0.1403341442346573, + "eval_overall_accuracy": 0.9597197898423818, + "eval_overall_f1": 0.8176795580110496, + "eval_overall_precision": 0.7688311688311689, + "eval_overall_recall": 0.8731563421828908, + "eval_runtime": 0.3293, + "eval_samples_per_second": 567.86, + "eval_steps_per_second": 9.11, + "step": 4028 + }, + { + "epoch": 39.0, + "grad_norm": 0.8123182058334351, + "learning_rate": 3.05e-05, + "loss": 0.0677, + "step": 4134 + }, + { + "epoch": 39.0, + "eval_LOCATION_f1": 0.7837837837837838, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7631578947368421, + "eval_LOCATION_recall": 0.8055555555555556, + "eval_ORGANIZATION_f1": 0.7976878612716762, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.7040816326530612, + "eval_ORGANIZATION_recall": 0.92, + "eval_PERSON_f1": 0.8926174496644295, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8580645161290322, + "eval_PERSON_recall": 0.9300699300699301, + "eval_QUANTITY_f1": 0.6181818181818182, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.53125, + "eval_QUANTITY_recall": 0.7391304347826086, + "eval_TIME_f1": 0.8148148148148148, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.7857142857142857, + "eval_TIME_recall": 0.8461538461538461, + "eval_loss": 0.1457056701183319, + "eval_overall_accuracy": 0.9589692269201902, + "eval_overall_f1": 0.8214285714285714, + "eval_overall_precision": 0.7686375321336761, + "eval_overall_recall": 0.8820058997050148, + "eval_runtime": 0.3294, + "eval_samples_per_second": 567.661, + "eval_steps_per_second": 9.107, + "step": 4134 + }, + { + "epoch": 40.0, + "grad_norm": 1.4303356409072876, + "learning_rate": 3e-05, + "loss": 0.0667, + "step": 4240 + }, + { + "epoch": 40.0, + "eval_LOCATION_f1": 0.778523489932886, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7532467532467533, + "eval_LOCATION_recall": 0.8055555555555556, + "eval_ORGANIZATION_f1": 0.7664670658682634, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.6956521739130435, + "eval_ORGANIZATION_recall": 0.8533333333333334, + "eval_PERSON_f1": 0.8956228956228957, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8636363636363636, + "eval_PERSON_recall": 0.9300699300699301, + "eval_QUANTITY_f1": 0.6037735849056605, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5333333333333333, + "eval_QUANTITY_recall": 0.6956521739130435, + "eval_TIME_f1": 0.7777777777777779, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.75, + "eval_TIME_recall": 0.8076923076923077, + "eval_loss": 0.14027433097362518, + "eval_overall_accuracy": 0.960720540405304, + "eval_overall_f1": 0.8111111111111112, + "eval_overall_precision": 0.7664041994750657, + "eval_overall_recall": 0.8613569321533924, + "eval_runtime": 0.3293, + "eval_samples_per_second": 567.858, + "eval_steps_per_second": 9.11, + "step": 4240 + }, + { + "epoch": 41.0, + "grad_norm": 1.8948636054992676, + "learning_rate": 2.95e-05, + "loss": 0.0647, + "step": 4346 + }, + { + "epoch": 41.0, + "eval_LOCATION_f1": 0.7763157894736842, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7375, + "eval_LOCATION_recall": 0.8194444444444444, + "eval_ORGANIZATION_f1": 0.7630057803468209, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.673469387755102, + "eval_ORGANIZATION_recall": 0.88, + "eval_PERSON_f1": 0.9, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8598726114649682, + "eval_PERSON_recall": 0.9440559440559441, + "eval_QUANTITY_f1": 0.5882352941176471, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5357142857142857, + "eval_QUANTITY_recall": 0.6521739130434783, + "eval_TIME_f1": 0.7777777777777779, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.75, + "eval_TIME_recall": 0.8076923076923077, + "eval_loss": 0.14349956810474396, + "eval_overall_accuracy": 0.9602201651238429, + "eval_overall_f1": 0.810958904109589, + "eval_overall_precision": 0.7570332480818415, + "eval_overall_recall": 0.8731563421828908, + "eval_runtime": 0.3291, + "eval_samples_per_second": 568.246, + "eval_steps_per_second": 9.116, + "step": 4346 + }, + { + "epoch": 42.0, + "grad_norm": 1.1397053003311157, + "learning_rate": 2.9e-05, + "loss": 0.0625, + "step": 4452 + }, + { + "epoch": 42.0, + "eval_LOCATION_f1": 0.781456953642384, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7468354430379747, + "eval_LOCATION_recall": 0.8194444444444444, + "eval_ORGANIZATION_f1": 0.7745664739884393, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.6836734693877551, + "eval_ORGANIZATION_recall": 0.8933333333333333, + "eval_PERSON_f1": 0.903010033444816, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8653846153846154, + "eval_PERSON_recall": 0.9440559440559441, + "eval_QUANTITY_f1": 0.5818181818181818, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5, + "eval_QUANTITY_recall": 0.6956521739130435, + "eval_TIME_f1": 0.7777777777777779, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.75, + "eval_TIME_recall": 0.8076923076923077, + "eval_loss": 0.1473807841539383, + "eval_overall_accuracy": 0.9592194145609206, + "eval_overall_f1": 0.8142076502732241, + "eval_overall_precision": 0.7582697201017812, + "eval_overall_recall": 0.8790560471976401, + "eval_runtime": 0.3311, + "eval_samples_per_second": 564.763, + "eval_steps_per_second": 9.06, + "step": 4452 + }, + { + "epoch": 43.0, + "grad_norm": 0.21276895701885223, + "learning_rate": 2.8499999999999998e-05, + "loss": 0.0639, + "step": 4558 + }, + { + "epoch": 43.0, + "eval_LOCATION_f1": 0.7843137254901961, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7407407407407407, + "eval_LOCATION_recall": 0.8333333333333334, + "eval_ORGANIZATION_f1": 0.7784431137724551, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.7065217391304348, + "eval_ORGANIZATION_recall": 0.8666666666666667, + "eval_PERSON_f1": 0.903010033444816, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8653846153846154, + "eval_PERSON_recall": 0.9440559440559441, + "eval_QUANTITY_f1": 0.6071428571428571, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5151515151515151, + "eval_QUANTITY_recall": 0.7391304347826086, + "eval_TIME_f1": 0.7777777777777779, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.75, + "eval_TIME_recall": 0.8076923076923077, + "eval_loss": 0.15028373897075653, + "eval_overall_accuracy": 0.960720540405304, + "eval_overall_f1": 0.8175582990397804, + "eval_overall_precision": 0.764102564102564, + "eval_overall_recall": 0.8790560471976401, + "eval_runtime": 0.3289, + "eval_samples_per_second": 568.48, + "eval_steps_per_second": 9.12, + "step": 4558 + }, + { + "epoch": 44.0, + "grad_norm": 1.5534422397613525, + "learning_rate": 2.8000000000000003e-05, + "loss": 0.0615, + "step": 4664 + }, + { + "epoch": 44.0, + "eval_LOCATION_f1": 0.7922077922077924, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7439024390243902, + "eval_LOCATION_recall": 0.8472222222222222, + "eval_ORGANIZATION_f1": 0.793103448275862, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.696969696969697, + "eval_ORGANIZATION_recall": 0.92, + "eval_PERSON_f1": 0.9060402684563759, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8709677419354839, + "eval_PERSON_recall": 0.9440559440559441, + "eval_QUANTITY_f1": 0.5555555555555556, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.4838709677419355, + "eval_QUANTITY_recall": 0.6521739130434783, + "eval_TIME_f1": 0.8076923076923077, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.8076923076923077, + "eval_TIME_recall": 0.8076923076923077, + "eval_loss": 0.14987242221832275, + "eval_overall_accuracy": 0.961220915686765, + "eval_overall_f1": 0.8224043715846995, + "eval_overall_precision": 0.7659033078880407, + "eval_overall_recall": 0.887905604719764, + "eval_runtime": 0.3281, + "eval_samples_per_second": 569.908, + "eval_steps_per_second": 9.143, + "step": 4664 + }, + { + "epoch": 45.0, + "grad_norm": 0.7796268463134766, + "learning_rate": 2.7500000000000004e-05, + "loss": 0.0595, + "step": 4770 + }, + { + "epoch": 45.0, + "eval_LOCATION_f1": 0.7837837837837838, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7631578947368421, + "eval_LOCATION_recall": 0.8055555555555556, + "eval_ORGANIZATION_f1": 0.7745664739884393, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.6836734693877551, + "eval_ORGANIZATION_recall": 0.8933333333333333, + "eval_PERSON_f1": 0.8993288590604027, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.864516129032258, + "eval_PERSON_recall": 0.9370629370629371, + "eval_QUANTITY_f1": 0.5555555555555556, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.4838709677419355, + "eval_QUANTITY_recall": 0.6521739130434783, + "eval_TIME_f1": 0.8076923076923077, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.8076923076923077, + "eval_TIME_recall": 0.8076923076923077, + "eval_loss": 0.14581084251403809, + "eval_overall_accuracy": 0.9599699774831123, + "eval_overall_f1": 0.8137931034482759, + "eval_overall_precision": 0.7642487046632125, + "eval_overall_recall": 0.8702064896755162, + "eval_runtime": 0.329, + "eval_samples_per_second": 568.34, + "eval_steps_per_second": 9.118, + "step": 4770 + }, + { + "epoch": 46.0, + "grad_norm": 2.5624725818634033, + "learning_rate": 2.7000000000000002e-05, + "loss": 0.0608, + "step": 4876 + }, + { + "epoch": 46.0, + "eval_LOCATION_f1": 0.7866666666666666, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7564102564102564, + "eval_LOCATION_recall": 0.8194444444444444, + "eval_ORGANIZATION_f1": 0.7804878048780487, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.7191011235955056, + "eval_ORGANIZATION_recall": 0.8533333333333334, + "eval_PERSON_f1": 0.9023569023569024, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8701298701298701, + "eval_PERSON_recall": 0.9370629370629371, + "eval_QUANTITY_f1": 0.5882352941176471, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5357142857142857, + "eval_QUANTITY_recall": 0.6521739130434783, + "eval_TIME_f1": 0.8076923076923077, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.8076923076923077, + "eval_TIME_recall": 0.8076923076923077, + "eval_loss": 0.14081361889839172, + "eval_overall_accuracy": 0.9622216662496873, + "eval_overall_f1": 0.8207282913165266, + "eval_overall_precision": 0.7813333333333333, + "eval_overall_recall": 0.8643067846607669, + "eval_runtime": 0.323, + "eval_samples_per_second": 578.94, + "eval_steps_per_second": 9.288, + "step": 4876 + }, + { + "epoch": 47.0, + "grad_norm": 0.45547011494636536, + "learning_rate": 2.6500000000000004e-05, + "loss": 0.0612, + "step": 4982 + }, + { + "epoch": 47.0, + "eval_LOCATION_f1": 0.778523489932886, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7532467532467533, + "eval_LOCATION_recall": 0.8055555555555556, + "eval_ORGANIZATION_f1": 0.7885714285714287, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.69, + "eval_ORGANIZATION_recall": 0.92, + "eval_PERSON_f1": 0.903010033444816, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8653846153846154, + "eval_PERSON_recall": 0.9440559440559441, + "eval_QUANTITY_f1": 0.5714285714285715, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.48484848484848486, + "eval_QUANTITY_recall": 0.6956521739130435, + "eval_TIME_f1": 0.7924528301886792, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.7777777777777778, + "eval_TIME_recall": 0.8076923076923077, + "eval_loss": 0.1553085446357727, + "eval_overall_accuracy": 0.9589692269201902, + "eval_overall_f1": 0.8169398907103824, + "eval_overall_precision": 0.7608142493638677, + "eval_overall_recall": 0.8820058997050148, + "eval_runtime": 0.3224, + "eval_samples_per_second": 579.98, + "eval_steps_per_second": 9.304, + "step": 4982 + }, + { + "epoch": 48.0, + "grad_norm": 0.7590288519859314, + "learning_rate": 2.6000000000000002e-05, + "loss": 0.0602, + "step": 5088 + }, + { + "epoch": 48.0, + "eval_LOCATION_f1": 0.8026315789473685, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7625, + "eval_LOCATION_recall": 0.8472222222222222, + "eval_ORGANIZATION_f1": 0.7857142857142856, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.7096774193548387, + "eval_ORGANIZATION_recall": 0.88, + "eval_PERSON_f1": 0.9023569023569024, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8701298701298701, + "eval_PERSON_recall": 0.9370629370629371, + "eval_QUANTITY_f1": 0.5555555555555556, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.4838709677419355, + "eval_QUANTITY_recall": 0.6521739130434783, + "eval_TIME_f1": 0.7777777777777779, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.75, + "eval_TIME_recall": 0.8076923076923077, + "eval_loss": 0.1467842012643814, + "eval_overall_accuracy": 0.9617212909682261, + "eval_overall_f1": 0.8193103448275862, + "eval_overall_precision": 0.7694300518134715, + "eval_overall_recall": 0.8761061946902655, + "eval_runtime": 0.3286, + "eval_samples_per_second": 569.057, + "eval_steps_per_second": 9.129, + "step": 5088 + }, + { + "epoch": 49.0, + "grad_norm": 1.438230037689209, + "learning_rate": 2.5500000000000003e-05, + "loss": 0.0581, + "step": 5194 + }, + { + "epoch": 49.0, + "eval_LOCATION_f1": 0.7894736842105262, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.75, + "eval_LOCATION_recall": 0.8333333333333334, + "eval_ORGANIZATION_f1": 0.793103448275862, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.696969696969697, + "eval_ORGANIZATION_recall": 0.92, + "eval_PERSON_f1": 0.9, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8598726114649682, + "eval_PERSON_recall": 0.9440559440559441, + "eval_QUANTITY_f1": 0.576923076923077, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5172413793103449, + "eval_QUANTITY_recall": 0.6521739130434783, + "eval_TIME_f1": 0.8076923076923077, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.8076923076923077, + "eval_TIME_recall": 0.8076923076923077, + "eval_loss": 0.14886882901191711, + "eval_overall_accuracy": 0.9614711033274956, + "eval_overall_f1": 0.8219178082191781, + "eval_overall_precision": 0.7672634271099744, + "eval_overall_recall": 0.8849557522123894, + "eval_runtime": 0.328, + "eval_samples_per_second": 570.138, + "eval_steps_per_second": 9.147, + "step": 5194 + }, + { + "epoch": 50.0, + "grad_norm": 1.5559232234954834, + "learning_rate": 2.5e-05, + "loss": 0.0553, + "step": 5300 + }, + { + "epoch": 50.0, + "eval_LOCATION_f1": 0.7919463087248323, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7662337662337663, + "eval_LOCATION_recall": 0.8194444444444444, + "eval_ORGANIZATION_f1": 0.7630057803468209, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.673469387755102, + "eval_ORGANIZATION_recall": 0.88, + "eval_PERSON_f1": 0.9023569023569024, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8701298701298701, + "eval_PERSON_recall": 0.9370629370629371, + "eval_QUANTITY_f1": 0.6071428571428571, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5151515151515151, + "eval_QUANTITY_recall": 0.7391304347826086, + "eval_TIME_f1": 0.8076923076923077, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.8076923076923077, + "eval_TIME_recall": 0.8076923076923077, + "eval_loss": 0.15414132177829742, + "eval_overall_accuracy": 0.9594696022016512, + "eval_overall_f1": 0.8170563961485556, + "eval_overall_precision": 0.7654639175257731, + "eval_overall_recall": 0.8761061946902655, + "eval_runtime": 0.3287, + "eval_samples_per_second": 568.971, + "eval_steps_per_second": 9.128, + "step": 5300 + }, + { + "epoch": 51.0, + "grad_norm": 0.9792301058769226, + "learning_rate": 2.45e-05, + "loss": 0.0555, + "step": 5406 + }, + { + "epoch": 51.0, + "eval_LOCATION_f1": 0.8053691275167787, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7792207792207793, + "eval_LOCATION_recall": 0.8333333333333334, + "eval_ORGANIZATION_f1": 0.7666666666666667, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.6571428571428571, + "eval_ORGANIZATION_recall": 0.92, + "eval_PERSON_f1": 0.8970099667774087, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8544303797468354, + "eval_PERSON_recall": 0.9440559440559441, + "eval_QUANTITY_f1": 0.6415094339622641, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5666666666666667, + "eval_QUANTITY_recall": 0.7391304347826086, + "eval_TIME_f1": 0.8461538461538461, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.8461538461538461, + "eval_TIME_recall": 0.8461538461538461, + "eval_loss": 0.1572943478822708, + "eval_overall_accuracy": 0.9589692269201902, + "eval_overall_f1": 0.8244897959183674, + "eval_overall_precision": 0.7651515151515151, + "eval_overall_recall": 0.8938053097345132, + "eval_runtime": 0.3306, + "eval_samples_per_second": 565.692, + "eval_steps_per_second": 9.075, + "step": 5406 + }, + { + "epoch": 52.0, + "grad_norm": 0.6869613528251648, + "learning_rate": 2.4e-05, + "loss": 0.0562, + "step": 5512 + }, + { + "epoch": 52.0, + "eval_LOCATION_f1": 0.781456953642384, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7468354430379747, + "eval_LOCATION_recall": 0.8194444444444444, + "eval_ORGANIZATION_f1": 0.783625730994152, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.6979166666666666, + "eval_ORGANIZATION_recall": 0.8933333333333333, + "eval_PERSON_f1": 0.8993288590604027, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.864516129032258, + "eval_PERSON_recall": 0.9370629370629371, + "eval_QUANTITY_f1": 0.6296296296296297, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5483870967741935, + "eval_QUANTITY_recall": 0.7391304347826086, + "eval_TIME_f1": 0.7777777777777779, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.75, + "eval_TIME_recall": 0.8076923076923077, + "eval_loss": 0.1512700915336609, + "eval_overall_accuracy": 0.9602201651238429, + "eval_overall_f1": 0.8186813186813187, + "eval_overall_precision": 0.7660668380462725, + "eval_overall_recall": 0.8790560471976401, + "eval_runtime": 0.3294, + "eval_samples_per_second": 567.715, + "eval_steps_per_second": 9.108, + "step": 5512 + }, + { + "epoch": 53.0, + "grad_norm": 1.4329147338867188, + "learning_rate": 2.35e-05, + "loss": 0.0556, + "step": 5618 + }, + { + "epoch": 53.0, + "eval_LOCATION_f1": 0.7866666666666666, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7564102564102564, + "eval_LOCATION_recall": 0.8194444444444444, + "eval_ORGANIZATION_f1": 0.7906976744186046, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.7010309278350515, + "eval_ORGANIZATION_recall": 0.9066666666666666, + "eval_PERSON_f1": 0.8970099667774087, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8544303797468354, + "eval_PERSON_recall": 0.9440559440559441, + "eval_QUANTITY_f1": 0.6181818181818182, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.53125, + "eval_QUANTITY_recall": 0.7391304347826086, + "eval_TIME_f1": 0.7843137254901961, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.8, + "eval_TIME_recall": 0.7692307692307693, + "eval_loss": 0.15228261053562164, + "eval_overall_accuracy": 0.9604703527645734, + "eval_overall_f1": 0.8203017832647463, + "eval_overall_precision": 0.7666666666666667, + "eval_overall_recall": 0.8820058997050148, + "eval_runtime": 0.3284, + "eval_samples_per_second": 569.411, + "eval_steps_per_second": 9.135, + "step": 5618 + }, + { + "epoch": 54.0, + "grad_norm": 1.3528200387954712, + "learning_rate": 2.3000000000000003e-05, + "loss": 0.0549, + "step": 5724 + }, + { + "epoch": 54.0, + "eval_LOCATION_f1": 0.7947019867549668, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.759493670886076, + "eval_LOCATION_recall": 0.8333333333333334, + "eval_ORGANIZATION_f1": 0.7593582887700535, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.6339285714285714, + "eval_ORGANIZATION_recall": 0.9466666666666667, + "eval_PERSON_f1": 0.8896321070234114, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8525641025641025, + "eval_PERSON_recall": 0.9300699300699301, + "eval_QUANTITY_f1": 0.6071428571428571, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5151515151515151, + "eval_QUANTITY_recall": 0.7391304347826086, + "eval_TIME_f1": 0.7777777777777779, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.75, + "eval_TIME_recall": 0.8076923076923077, + "eval_loss": 0.16874687373638153, + "eval_overall_accuracy": 0.9554665999499625, + "eval_overall_f1": 0.8085676037483267, + "eval_overall_precision": 0.7401960784313726, + "eval_overall_recall": 0.8908554572271387, + "eval_runtime": 0.3286, + "eval_samples_per_second": 569.11, + "eval_steps_per_second": 9.13, + "step": 5724 + }, + { + "epoch": 55.0, + "grad_norm": 2.1224594116210938, + "learning_rate": 2.25e-05, + "loss": 0.0533, + "step": 5830 + }, + { + "epoch": 55.0, + "eval_LOCATION_f1": 0.7894736842105262, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.75, + "eval_LOCATION_recall": 0.8333333333333334, + "eval_ORGANIZATION_f1": 0.7759562841530054, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.6574074074074074, + "eval_ORGANIZATION_recall": 0.9466666666666667, + "eval_PERSON_f1": 0.8963210702341137, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8589743589743589, + "eval_PERSON_recall": 0.9370629370629371, + "eval_QUANTITY_f1": 0.6296296296296297, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5483870967741935, + "eval_QUANTITY_recall": 0.7391304347826086, + "eval_TIME_f1": 0.7924528301886792, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.7777777777777778, + "eval_TIME_recall": 0.8076923076923077, + "eval_loss": 0.16018283367156982, + "eval_overall_accuracy": 0.9589692269201902, + "eval_overall_f1": 0.8178137651821862, + "eval_overall_precision": 0.753731343283582, + "eval_overall_recall": 0.8938053097345132, + "eval_runtime": 0.3279, + "eval_samples_per_second": 570.252, + "eval_steps_per_second": 9.148, + "step": 5830 + }, + { + "epoch": 56.0, + "grad_norm": 0.7443162202835083, + "learning_rate": 2.2000000000000003e-05, + "loss": 0.0511, + "step": 5936 + }, + { + "epoch": 56.0, + "eval_LOCATION_f1": 0.8026315789473685, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7625, + "eval_LOCATION_recall": 0.8472222222222222, + "eval_ORGANIZATION_f1": 0.7719298245614036, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.6875, + "eval_ORGANIZATION_recall": 0.88, + "eval_PERSON_f1": 0.8993288590604027, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.864516129032258, + "eval_PERSON_recall": 0.9370629370629371, + "eval_QUANTITY_f1": 0.5964912280701754, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5, + "eval_QUANTITY_recall": 0.7391304347826086, + "eval_TIME_f1": 0.7777777777777779, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.75, + "eval_TIME_recall": 0.8076923076923077, + "eval_loss": 0.1552291363477707, + "eval_overall_accuracy": 0.9604703527645734, + "eval_overall_f1": 0.8169398907103824, + "eval_overall_precision": 0.7608142493638677, + "eval_overall_recall": 0.8820058997050148, + "eval_runtime": 0.3283, + "eval_samples_per_second": 569.654, + "eval_steps_per_second": 9.139, + "step": 5936 + }, + { + "epoch": 57.0, + "grad_norm": 1.7182573080062866, + "learning_rate": 2.15e-05, + "loss": 0.0497, + "step": 6042 + }, + { + "epoch": 57.0, + "eval_LOCATION_f1": 0.8, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7692307692307693, + "eval_LOCATION_recall": 0.8333333333333334, + "eval_ORGANIZATION_f1": 0.7790697674418604, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.6907216494845361, + "eval_ORGANIZATION_recall": 0.8933333333333333, + "eval_PERSON_f1": 0.9054054054054055, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8758169934640523, + "eval_PERSON_recall": 0.9370629370629371, + "eval_QUANTITY_f1": 0.5714285714285715, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.48484848484848486, + "eval_QUANTITY_recall": 0.6956521739130435, + "eval_TIME_f1": 0.7924528301886792, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.7777777777777778, + "eval_TIME_recall": 0.8076923076923077, + "eval_loss": 0.15355649590492249, + "eval_overall_accuracy": 0.961220915686765, + "eval_overall_f1": 0.8198074277854195, + "eval_overall_precision": 0.7680412371134021, + "eval_overall_recall": 0.8790560471976401, + "eval_runtime": 0.3298, + "eval_samples_per_second": 566.97, + "eval_steps_per_second": 9.096, + "step": 6042 + }, + { + "epoch": 58.0, + "grad_norm": 2.4079740047454834, + "learning_rate": 2.1e-05, + "loss": 0.0492, + "step": 6148 + }, + { + "epoch": 58.0, + "eval_LOCATION_f1": 0.8026315789473685, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7625, + "eval_LOCATION_recall": 0.8472222222222222, + "eval_ORGANIZATION_f1": 0.7701149425287357, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.6767676767676768, + "eval_ORGANIZATION_recall": 0.8933333333333333, + "eval_PERSON_f1": 0.8993288590604027, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.864516129032258, + "eval_PERSON_recall": 0.9370629370629371, + "eval_QUANTITY_f1": 0.6071428571428571, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5151515151515151, + "eval_QUANTITY_recall": 0.7391304347826086, + "eval_TIME_f1": 0.7924528301886792, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.7777777777777778, + "eval_TIME_recall": 0.8076923076923077, + "eval_loss": 0.1585557758808136, + "eval_overall_accuracy": 0.9599699774831123, + "eval_overall_f1": 0.8185538881309687, + "eval_overall_precision": 0.7614213197969543, + "eval_overall_recall": 0.8849557522123894, + "eval_runtime": 0.3293, + "eval_samples_per_second": 567.844, + "eval_steps_per_second": 9.11, + "step": 6148 + }, + { + "epoch": 59.0, + "grad_norm": 0.8430188298225403, + "learning_rate": 2.05e-05, + "loss": 0.0489, + "step": 6254 + }, + { + "epoch": 59.0, + "eval_LOCATION_f1": 0.8026315789473685, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7625, + "eval_LOCATION_recall": 0.8472222222222222, + "eval_ORGANIZATION_f1": 0.7816091954022988, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.6868686868686869, + "eval_ORGANIZATION_recall": 0.9066666666666666, + "eval_PERSON_f1": 0.9023569023569024, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8701298701298701, + "eval_PERSON_recall": 0.9370629370629371, + "eval_QUANTITY_f1": 0.5714285714285715, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.48484848484848486, + "eval_QUANTITY_recall": 0.6956521739130435, + "eval_TIME_f1": 0.7924528301886792, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.7777777777777778, + "eval_TIME_recall": 0.8076923076923077, + "eval_loss": 0.1573321521282196, + "eval_overall_accuracy": 0.960720540405304, + "eval_overall_f1": 0.819672131147541, + "eval_overall_precision": 0.7633587786259542, + "eval_overall_recall": 0.8849557522123894, + "eval_runtime": 0.3291, + "eval_samples_per_second": 568.242, + "eval_steps_per_second": 9.116, + "step": 6254 + }, + { + "epoch": 60.0, + "grad_norm": 0.6046874523162842, + "learning_rate": 2e-05, + "loss": 0.0494, + "step": 6360 + }, + { + "epoch": 60.0, + "eval_LOCATION_f1": 0.8079470198675496, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7721518987341772, + "eval_LOCATION_recall": 0.8472222222222222, + "eval_ORGANIZATION_f1": 0.793103448275862, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.696969696969697, + "eval_ORGANIZATION_recall": 0.92, + "eval_PERSON_f1": 0.8993288590604027, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.864516129032258, + "eval_PERSON_recall": 0.9370629370629371, + "eval_QUANTITY_f1": 0.5714285714285715, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.48484848484848486, + "eval_QUANTITY_recall": 0.6956521739130435, + "eval_TIME_f1": 0.7924528301886792, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.7777777777777778, + "eval_TIME_recall": 0.8076923076923077, + "eval_loss": 0.16140538454055786, + "eval_overall_accuracy": 0.9604703527645734, + "eval_overall_f1": 0.8224043715846995, + "eval_overall_precision": 0.7659033078880407, + "eval_overall_recall": 0.887905604719764, + "eval_runtime": 0.3222, + "eval_samples_per_second": 580.392, + "eval_steps_per_second": 9.311, + "step": 6360 + }, + { + "epoch": 61.0, + "grad_norm": 1.020039677619934, + "learning_rate": 1.9500000000000003e-05, + "loss": 0.0476, + "step": 6466 + }, + { + "epoch": 61.0, + "eval_LOCATION_f1": 0.7866666666666666, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7564102564102564, + "eval_LOCATION_recall": 0.8194444444444444, + "eval_ORGANIZATION_f1": 0.7719298245614036, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.6875, + "eval_ORGANIZATION_recall": 0.88, + "eval_PERSON_f1": 0.8933333333333333, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8535031847133758, + "eval_PERSON_recall": 0.9370629370629371, + "eval_QUANTITY_f1": 0.6071428571428571, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5151515151515151, + "eval_QUANTITY_recall": 0.7391304347826086, + "eval_TIME_f1": 0.830188679245283, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.8148148148148148, + "eval_TIME_recall": 0.8461538461538461, + "eval_loss": 0.15692444145679474, + "eval_overall_accuracy": 0.9609707280460346, + "eval_overall_f1": 0.8164383561643836, + "eval_overall_precision": 0.7621483375959079, + "eval_overall_recall": 0.8790560471976401, + "eval_runtime": 0.3296, + "eval_samples_per_second": 567.307, + "eval_steps_per_second": 9.101, + "step": 6466 + }, + { + "epoch": 62.0, + "grad_norm": 1.3080651760101318, + "learning_rate": 1.9e-05, + "loss": 0.0469, + "step": 6572 + }, + { + "epoch": 62.0, + "eval_LOCATION_f1": 0.7973856209150327, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7530864197530864, + "eval_LOCATION_recall": 0.8472222222222222, + "eval_ORGANIZATION_f1": 0.7810650887573964, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.7021276595744681, + "eval_ORGANIZATION_recall": 0.88, + "eval_PERSON_f1": 0.8933333333333333, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8535031847133758, + "eval_PERSON_recall": 0.9370629370629371, + "eval_QUANTITY_f1": 0.5818181818181818, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5, + "eval_QUANTITY_recall": 0.6956521739130435, + "eval_TIME_f1": 0.7924528301886792, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.7777777777777778, + "eval_TIME_recall": 0.8076923076923077, + "eval_loss": 0.1600496470928192, + "eval_overall_accuracy": 0.961220915686765, + "eval_overall_f1": 0.8164383561643836, + "eval_overall_precision": 0.7621483375959079, + "eval_overall_recall": 0.8790560471976401, + "eval_runtime": 0.3292, + "eval_samples_per_second": 568.119, + "eval_steps_per_second": 9.114, + "step": 6572 + }, + { + "epoch": 63.0, + "grad_norm": 1.0082831382751465, + "learning_rate": 1.85e-05, + "loss": 0.0482, + "step": 6678 + }, + { + "epoch": 63.0, + "eval_LOCATION_f1": 0.8157894736842106, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.775, + "eval_LOCATION_recall": 0.8611111111111112, + "eval_ORGANIZATION_f1": 0.7745664739884393, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.6836734693877551, + "eval_ORGANIZATION_recall": 0.8933333333333333, + "eval_PERSON_f1": 0.9016949152542374, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.875, + "eval_PERSON_recall": 0.9300699300699301, + "eval_QUANTITY_f1": 0.5555555555555556, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.4838709677419355, + "eval_QUANTITY_recall": 0.6521739130434783, + "eval_TIME_f1": 0.830188679245283, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.8148148148148148, + "eval_TIME_recall": 0.8461538461538461, + "eval_loss": 0.15659038722515106, + "eval_overall_accuracy": 0.9622216662496873, + "eval_overall_f1": 0.8225584594222835, + "eval_overall_precision": 0.770618556701031, + "eval_overall_recall": 0.8820058997050148, + "eval_runtime": 0.3292, + "eval_samples_per_second": 568.105, + "eval_steps_per_second": 9.114, + "step": 6678 + }, + { + "epoch": 64.0, + "grad_norm": 0.7832164168357849, + "learning_rate": 1.8e-05, + "loss": 0.0481, + "step": 6784 + }, + { + "epoch": 64.0, + "eval_LOCATION_f1": 0.7894736842105262, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.75, + "eval_LOCATION_recall": 0.8333333333333334, + "eval_ORGANIZATION_f1": 0.7734806629834254, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.660377358490566, + "eval_ORGANIZATION_recall": 0.9333333333333333, + "eval_PERSON_f1": 0.8970099667774087, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8544303797468354, + "eval_PERSON_recall": 0.9440559440559441, + "eval_QUANTITY_f1": 0.5714285714285715, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.48484848484848486, + "eval_QUANTITY_recall": 0.6956521739130435, + "eval_TIME_f1": 0.8148148148148148, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.7857142857142857, + "eval_TIME_recall": 0.8461538461538461, + "eval_loss": 0.17121106386184692, + "eval_overall_accuracy": 0.9589692269201902, + "eval_overall_f1": 0.814516129032258, + "eval_overall_precision": 0.7481481481481481, + "eval_overall_recall": 0.8938053097345132, + "eval_runtime": 0.3279, + "eval_samples_per_second": 570.252, + "eval_steps_per_second": 9.148, + "step": 6784 + }, + { + "epoch": 65.0, + "grad_norm": 0.6884886622428894, + "learning_rate": 1.75e-05, + "loss": 0.0437, + "step": 6890 + }, + { + "epoch": 65.0, + "eval_LOCATION_f1": 0.7947019867549668, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.759493670886076, + "eval_LOCATION_recall": 0.8333333333333334, + "eval_ORGANIZATION_f1": 0.7558139534883721, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.6701030927835051, + "eval_ORGANIZATION_recall": 0.8666666666666667, + "eval_PERSON_f1": 0.9060402684563759, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8709677419354839, + "eval_PERSON_recall": 0.9440559440559441, + "eval_QUANTITY_f1": 0.576923076923077, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5172413793103449, + "eval_QUANTITY_recall": 0.6521739130434783, + "eval_TIME_f1": 0.7924528301886792, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.7777777777777778, + "eval_TIME_recall": 0.8076923076923077, + "eval_loss": 0.1589990258216858, + "eval_overall_accuracy": 0.9617212909682261, + "eval_overall_f1": 0.815426997245179, + "eval_overall_precision": 0.7648578811369509, + "eval_overall_recall": 0.8731563421828908, + "eval_runtime": 0.3237, + "eval_samples_per_second": 577.733, + "eval_steps_per_second": 9.268, + "step": 6890 + }, + { + "epoch": 66.0, + "grad_norm": 0.4578692615032196, + "learning_rate": 1.7000000000000003e-05, + "loss": 0.0434, + "step": 6996 + }, + { + "epoch": 66.0, + "eval_LOCATION_f1": 0.7947019867549668, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.759493670886076, + "eval_LOCATION_recall": 0.8333333333333334, + "eval_ORGANIZATION_f1": 0.7630057803468209, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.673469387755102, + "eval_ORGANIZATION_recall": 0.88, + "eval_PERSON_f1": 0.9090909090909092, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8766233766233766, + "eval_PERSON_recall": 0.9440559440559441, + "eval_QUANTITY_f1": 0.6153846153846154, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5517241379310345, + "eval_QUANTITY_recall": 0.6956521739130435, + "eval_TIME_f1": 0.8461538461538461, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.8461538461538461, + "eval_TIME_recall": 0.8461538461538461, + "eval_loss": 0.16020527482032776, + "eval_overall_accuracy": 0.9629722291718789, + "eval_overall_f1": 0.8248275862068966, + "eval_overall_precision": 0.7746113989637305, + "eval_overall_recall": 0.8820058997050148, + "eval_runtime": 0.3226, + "eval_samples_per_second": 579.698, + "eval_steps_per_second": 9.3, + "step": 6996 + }, + { + "epoch": 67.0, + "grad_norm": 0.5099142789840698, + "learning_rate": 1.65e-05, + "loss": 0.0431, + "step": 7102 + }, + { + "epoch": 67.0, + "eval_LOCATION_f1": 0.7947019867549668, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.759493670886076, + "eval_LOCATION_recall": 0.8333333333333334, + "eval_ORGANIZATION_f1": 0.7790697674418604, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.6907216494845361, + "eval_ORGANIZATION_recall": 0.8933333333333333, + "eval_PERSON_f1": 0.9060402684563759, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8709677419354839, + "eval_PERSON_recall": 0.9440559440559441, + "eval_QUANTITY_f1": 0.6181818181818182, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.53125, + "eval_QUANTITY_recall": 0.7391304347826086, + "eval_TIME_f1": 0.830188679245283, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.8148148148148148, + "eval_TIME_recall": 0.8461538461538461, + "eval_loss": 0.16391661763191223, + "eval_overall_accuracy": 0.9619714786089567, + "eval_overall_f1": 0.8257887517146778, + "eval_overall_precision": 0.7717948717948718, + "eval_overall_recall": 0.887905604719764, + "eval_runtime": 0.3235, + "eval_samples_per_second": 578.034, + "eval_steps_per_second": 9.273, + "step": 7102 + }, + { + "epoch": 68.0, + "grad_norm": 0.38720881938934326, + "learning_rate": 1.6000000000000003e-05, + "loss": 0.0428, + "step": 7208 + }, + { + "epoch": 68.0, + "eval_LOCATION_f1": 0.8026315789473685, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7625, + "eval_LOCATION_recall": 0.8472222222222222, + "eval_ORGANIZATION_f1": 0.7727272727272727, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.6732673267326733, + "eval_ORGANIZATION_recall": 0.9066666666666666, + "eval_PERSON_f1": 0.903010033444816, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8653846153846154, + "eval_PERSON_recall": 0.9440559440559441, + "eval_QUANTITY_f1": 0.6296296296296297, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5483870967741935, + "eval_QUANTITY_recall": 0.7391304347826086, + "eval_TIME_f1": 0.8076923076923077, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.8076923076923077, + "eval_TIME_recall": 0.8076923076923077, + "eval_loss": 0.17015798389911652, + "eval_overall_accuracy": 0.9609707280460346, + "eval_overall_f1": 0.8240109140518419, + "eval_overall_precision": 0.766497461928934, + "eval_overall_recall": 0.8908554572271387, + "eval_runtime": 0.3289, + "eval_samples_per_second": 568.646, + "eval_steps_per_second": 9.123, + "step": 7208 + }, + { + "epoch": 69.0, + "grad_norm": 1.3377002477645874, + "learning_rate": 1.55e-05, + "loss": 0.0422, + "step": 7314 + }, + { + "epoch": 69.0, + "eval_LOCATION_f1": 0.8079470198675496, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7721518987341772, + "eval_LOCATION_recall": 0.8472222222222222, + "eval_ORGANIZATION_f1": 0.7816091954022988, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.6868686868686869, + "eval_ORGANIZATION_recall": 0.9066666666666666, + "eval_PERSON_f1": 0.903010033444816, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8653846153846154, + "eval_PERSON_recall": 0.9440559440559441, + "eval_QUANTITY_f1": 0.6415094339622641, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5666666666666667, + "eval_QUANTITY_recall": 0.7391304347826086, + "eval_TIME_f1": 0.8076923076923077, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.8076923076923077, + "eval_TIME_recall": 0.8076923076923077, + "eval_loss": 0.1659727841615677, + "eval_overall_accuracy": 0.9627220415311484, + "eval_overall_f1": 0.8285322359396433, + "eval_overall_precision": 0.7743589743589744, + "eval_overall_recall": 0.8908554572271387, + "eval_runtime": 0.3278, + "eval_samples_per_second": 570.439, + "eval_steps_per_second": 9.151, + "step": 7314 + }, + { + "epoch": 70.0, + "grad_norm": 1.1326639652252197, + "learning_rate": 1.5e-05, + "loss": 0.0443, + "step": 7420 + }, + { + "epoch": 70.0, + "eval_LOCATION_f1": 0.7973856209150327, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7530864197530864, + "eval_LOCATION_recall": 0.8472222222222222, + "eval_ORGANIZATION_f1": 0.7727272727272727, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.6732673267326733, + "eval_ORGANIZATION_recall": 0.9066666666666666, + "eval_PERSON_f1": 0.8963210702341137, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8589743589743589, + "eval_PERSON_recall": 0.9370629370629371, + "eval_QUANTITY_f1": 0.6037735849056605, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5333333333333333, + "eval_QUANTITY_recall": 0.6956521739130435, + "eval_TIME_f1": 0.8076923076923077, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.8076923076923077, + "eval_TIME_recall": 0.8076923076923077, + "eval_loss": 0.17037485539913177, + "eval_overall_accuracy": 0.960720540405304, + "eval_overall_f1": 0.8185538881309687, + "eval_overall_precision": 0.7614213197969543, + "eval_overall_recall": 0.8849557522123894, + "eval_runtime": 0.3313, + "eval_samples_per_second": 564.42, + "eval_steps_per_second": 9.055, + "step": 7420 + }, + { + "epoch": 71.0, + "grad_norm": 1.335763692855835, + "learning_rate": 1.45e-05, + "loss": 0.0422, + "step": 7526 + }, + { + "epoch": 71.0, + "eval_LOCATION_f1": 0.7973856209150327, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7530864197530864, + "eval_LOCATION_recall": 0.8472222222222222, + "eval_ORGANIZATION_f1": 0.783625730994152, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.6979166666666666, + "eval_ORGANIZATION_recall": 0.8933333333333333, + "eval_PERSON_f1": 0.9060402684563759, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8709677419354839, + "eval_PERSON_recall": 0.9440559440559441, + "eval_QUANTITY_f1": 0.6296296296296297, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5483870967741935, + "eval_QUANTITY_recall": 0.7391304347826086, + "eval_TIME_f1": 0.8076923076923077, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.8076923076923077, + "eval_TIME_recall": 0.8076923076923077, + "eval_loss": 0.1672845035791397, + "eval_overall_accuracy": 0.9622216662496873, + "eval_overall_f1": 0.8269230769230769, + "eval_overall_precision": 0.7737789203084833, + "eval_overall_recall": 0.887905604719764, + "eval_runtime": 0.3281, + "eval_samples_per_second": 569.916, + "eval_steps_per_second": 9.143, + "step": 7526 + }, + { + "epoch": 72.0, + "grad_norm": 1.0592600107192993, + "learning_rate": 1.4000000000000001e-05, + "loss": 0.0429, + "step": 7632 + }, + { + "epoch": 72.0, + "eval_LOCATION_f1": 0.8079470198675496, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7721518987341772, + "eval_LOCATION_recall": 0.8472222222222222, + "eval_ORGANIZATION_f1": 0.783625730994152, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.6979166666666666, + "eval_ORGANIZATION_recall": 0.8933333333333333, + "eval_PERSON_f1": 0.9060402684563759, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8709677419354839, + "eval_PERSON_recall": 0.9440559440559441, + "eval_QUANTITY_f1": 0.5818181818181818, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5, + "eval_QUANTITY_recall": 0.6956521739130435, + "eval_TIME_f1": 0.8076923076923077, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.8076923076923077, + "eval_TIME_recall": 0.8076923076923077, + "eval_loss": 0.1664479672908783, + "eval_overall_accuracy": 0.9622216662496873, + "eval_overall_f1": 0.8253094910591473, + "eval_overall_precision": 0.7731958762886598, + "eval_overall_recall": 0.8849557522123894, + "eval_runtime": 0.3286, + "eval_samples_per_second": 569.074, + "eval_steps_per_second": 9.13, + "step": 7632 + }, + { + "epoch": 73.0, + "grad_norm": 2.9215281009674072, + "learning_rate": 1.3500000000000001e-05, + "loss": 0.0433, + "step": 7738 + }, + { + "epoch": 73.0, + "eval_LOCATION_f1": 0.7919463087248323, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7662337662337663, + "eval_LOCATION_recall": 0.8194444444444444, + "eval_ORGANIZATION_f1": 0.7745664739884393, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.6836734693877551, + "eval_ORGANIZATION_recall": 0.8933333333333333, + "eval_PERSON_f1": 0.8993288590604027, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.864516129032258, + "eval_PERSON_recall": 0.9370629370629371, + "eval_QUANTITY_f1": 0.6153846153846154, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5517241379310345, + "eval_QUANTITY_recall": 0.6956521739130435, + "eval_TIME_f1": 0.8461538461538461, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.8461538461538461, + "eval_TIME_recall": 0.8461538461538461, + "eval_loss": 0.16273486614227295, + "eval_overall_accuracy": 0.9624718538904178, + "eval_overall_f1": 0.8232044198895028, + "eval_overall_precision": 0.7740259740259741, + "eval_overall_recall": 0.8790560471976401, + "eval_runtime": 0.328, + "eval_samples_per_second": 570.134, + "eval_steps_per_second": 9.147, + "step": 7738 + }, + { + "epoch": 74.0, + "grad_norm": 0.7708503007888794, + "learning_rate": 1.3000000000000001e-05, + "loss": 0.0403, + "step": 7844 + }, + { + "epoch": 74.0, + "eval_LOCATION_f1": 0.8026315789473685, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7625, + "eval_LOCATION_recall": 0.8472222222222222, + "eval_ORGANIZATION_f1": 0.7674418604651163, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.6804123711340206, + "eval_ORGANIZATION_recall": 0.88, + "eval_PERSON_f1": 0.8993288590604027, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.864516129032258, + "eval_PERSON_recall": 0.9370629370629371, + "eval_QUANTITY_f1": 0.6037735849056605, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5333333333333333, + "eval_QUANTITY_recall": 0.6956521739130435, + "eval_TIME_f1": 0.8461538461538461, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.8461538461538461, + "eval_TIME_recall": 0.8461538461538461, + "eval_loss": 0.16837267577648163, + "eval_overall_accuracy": 0.9614711033274956, + "eval_overall_f1": 0.8225584594222835, + "eval_overall_precision": 0.770618556701031, + "eval_overall_recall": 0.8820058997050148, + "eval_runtime": 0.3288, + "eval_samples_per_second": 568.797, + "eval_steps_per_second": 9.125, + "step": 7844 + }, + { + "epoch": 75.0, + "grad_norm": 0.7442087531089783, + "learning_rate": 1.25e-05, + "loss": 0.0415, + "step": 7950 + }, + { + "epoch": 75.0, + "eval_LOCATION_f1": 0.7763157894736842, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7375, + "eval_LOCATION_recall": 0.8194444444444444, + "eval_ORGANIZATION_f1": 0.7613636363636365, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.6633663366336634, + "eval_ORGANIZATION_recall": 0.8933333333333333, + "eval_PERSON_f1": 0.8993288590604027, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.864516129032258, + "eval_PERSON_recall": 0.9370629370629371, + "eval_QUANTITY_f1": 0.6037735849056605, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5333333333333333, + "eval_QUANTITY_recall": 0.6956521739130435, + "eval_TIME_f1": 0.8076923076923077, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.8076923076923077, + "eval_TIME_recall": 0.8076923076923077, + "eval_loss": 0.1693791300058365, + "eval_overall_accuracy": 0.9602201651238429, + "eval_overall_f1": 0.8125854993160055, + "eval_overall_precision": 0.7576530612244898, + "eval_overall_recall": 0.8761061946902655, + "eval_runtime": 0.3227, + "eval_samples_per_second": 579.537, + "eval_steps_per_second": 9.297, + "step": 7950 + }, + { + "epoch": 76.0, + "grad_norm": 2.1304056644439697, + "learning_rate": 1.2e-05, + "loss": 0.0405, + "step": 8056 + }, + { + "epoch": 76.0, + "eval_LOCATION_f1": 0.7973856209150327, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7530864197530864, + "eval_LOCATION_recall": 0.8472222222222222, + "eval_ORGANIZATION_f1": 0.752808988764045, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.6504854368932039, + "eval_ORGANIZATION_recall": 0.8933333333333333, + "eval_PERSON_f1": 0.8993288590604027, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.864516129032258, + "eval_PERSON_recall": 0.9370629370629371, + "eval_QUANTITY_f1": 0.6037735849056605, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5333333333333333, + "eval_QUANTITY_recall": 0.6956521739130435, + "eval_TIME_f1": 0.7924528301886792, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.7777777777777778, + "eval_TIME_recall": 0.8076923076923077, + "eval_loss": 0.17410139739513397, + "eval_overall_accuracy": 0.9597197898423818, + "eval_overall_f1": 0.8136054421768707, + "eval_overall_precision": 0.7550505050505051, + "eval_overall_recall": 0.8820058997050148, + "eval_runtime": 0.3294, + "eval_samples_per_second": 567.627, + "eval_steps_per_second": 9.106, + "step": 8056 + }, + { + "epoch": 77.0, + "grad_norm": 0.30889636278152466, + "learning_rate": 1.1500000000000002e-05, + "loss": 0.0399, + "step": 8162 + }, + { + "epoch": 77.0, + "eval_LOCATION_f1": 0.8026315789473685, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7625, + "eval_LOCATION_recall": 0.8472222222222222, + "eval_ORGANIZATION_f1": 0.7657142857142857, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.67, + "eval_ORGANIZATION_recall": 0.8933333333333333, + "eval_PERSON_f1": 0.8993288590604027, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.864516129032258, + "eval_PERSON_recall": 0.9370629370629371, + "eval_QUANTITY_f1": 0.6037735849056605, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5333333333333333, + "eval_QUANTITY_recall": 0.6956521739130435, + "eval_TIME_f1": 0.7777777777777779, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.75, + "eval_TIME_recall": 0.8076923076923077, + "eval_loss": 0.1697201281785965, + "eval_overall_accuracy": 0.9604703527645734, + "eval_overall_f1": 0.8169398907103824, + "eval_overall_precision": 0.7608142493638677, + "eval_overall_recall": 0.8820058997050148, + "eval_runtime": 0.3295, + "eval_samples_per_second": 567.594, + "eval_steps_per_second": 9.106, + "step": 8162 + }, + { + "epoch": 78.0, + "grad_norm": 0.9235987067222595, + "learning_rate": 1.1000000000000001e-05, + "loss": 0.0386, + "step": 8268 + }, + { + "epoch": 78.0, + "eval_LOCATION_f1": 0.8187919463087249, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7922077922077922, + "eval_LOCATION_recall": 0.8472222222222222, + "eval_ORGANIZATION_f1": 0.7469879518072289, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.6813186813186813, + "eval_ORGANIZATION_recall": 0.8266666666666667, + "eval_PERSON_f1": 0.9090909090909092, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8766233766233766, + "eval_PERSON_recall": 0.9440559440559441, + "eval_QUANTITY_f1": 0.6037735849056605, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5333333333333333, + "eval_QUANTITY_recall": 0.6956521739130435, + "eval_TIME_f1": 0.8461538461538461, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.8461538461538461, + "eval_TIME_recall": 0.8461538461538461, + "eval_loss": 0.1652342677116394, + "eval_overall_accuracy": 0.9629722291718789, + "eval_overall_f1": 0.8256624825662482, + "eval_overall_precision": 0.783068783068783, + "eval_overall_recall": 0.8731563421828908, + "eval_runtime": 0.3286, + "eval_samples_per_second": 569.056, + "eval_steps_per_second": 9.129, + "step": 8268 + }, + { + "epoch": 79.0, + "grad_norm": 1.591226577758789, + "learning_rate": 1.05e-05, + "loss": 0.0373, + "step": 8374 + }, + { + "epoch": 79.0, + "eval_LOCATION_f1": 0.8079470198675496, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7721518987341772, + "eval_LOCATION_recall": 0.8472222222222222, + "eval_ORGANIZATION_f1": 0.7709497206703911, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.6634615384615384, + "eval_ORGANIZATION_recall": 0.92, + "eval_PERSON_f1": 0.9060402684563759, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8709677419354839, + "eval_PERSON_recall": 0.9440559440559441, + "eval_QUANTITY_f1": 0.6037735849056605, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5333333333333333, + "eval_QUANTITY_recall": 0.6956521739130435, + "eval_TIME_f1": 0.8461538461538461, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.8461538461538461, + "eval_TIME_recall": 0.8461538461538461, + "eval_loss": 0.17290015518665314, + "eval_overall_accuracy": 0.961220915686765, + "eval_overall_f1": 0.8267394270122784, + "eval_overall_precision": 0.7690355329949239, + "eval_overall_recall": 0.8938053097345132, + "eval_runtime": 0.3291, + "eval_samples_per_second": 568.164, + "eval_steps_per_second": 9.115, + "step": 8374 + }, + { + "epoch": 80.0, + "grad_norm": 0.5225344300270081, + "learning_rate": 1e-05, + "loss": 0.0391, + "step": 8480 + }, + { + "epoch": 80.0, + "eval_LOCATION_f1": 0.8026315789473685, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7625, + "eval_LOCATION_recall": 0.8472222222222222, + "eval_ORGANIZATION_f1": 0.7630057803468209, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.673469387755102, + "eval_ORGANIZATION_recall": 0.88, + "eval_PERSON_f1": 0.8993288590604027, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.864516129032258, + "eval_PERSON_recall": 0.9370629370629371, + "eval_QUANTITY_f1": 0.6296296296296297, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5483870967741935, + "eval_QUANTITY_recall": 0.7391304347826086, + "eval_TIME_f1": 0.7777777777777779, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.75, + "eval_TIME_recall": 0.8076923076923077, + "eval_loss": 0.17415541410446167, + "eval_overall_accuracy": 0.9604703527645734, + "eval_overall_f1": 0.8180574555403557, + "eval_overall_precision": 0.7627551020408163, + "eval_overall_recall": 0.8820058997050148, + "eval_runtime": 0.3281, + "eval_samples_per_second": 569.982, + "eval_steps_per_second": 9.144, + "step": 8480 + }, + { + "epoch": 81.0, + "grad_norm": 0.4210955798625946, + "learning_rate": 9.5e-06, + "loss": 0.0375, + "step": 8586 + }, + { + "epoch": 81.0, + "eval_LOCATION_f1": 0.7947019867549668, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.759493670886076, + "eval_LOCATION_recall": 0.8333333333333334, + "eval_ORGANIZATION_f1": 0.7674418604651163, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.6804123711340206, + "eval_ORGANIZATION_recall": 0.88, + "eval_PERSON_f1": 0.903010033444816, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8653846153846154, + "eval_PERSON_recall": 0.9440559440559441, + "eval_QUANTITY_f1": 0.6415094339622641, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5666666666666667, + "eval_QUANTITY_recall": 0.7391304347826086, + "eval_TIME_f1": 0.7777777777777779, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.75, + "eval_TIME_recall": 0.8076923076923077, + "eval_loss": 0.17450082302093506, + "eval_overall_accuracy": 0.961220915686765, + "eval_overall_f1": 0.8203017832647463, + "eval_overall_precision": 0.7666666666666667, + "eval_overall_recall": 0.8820058997050148, + "eval_runtime": 0.3279, + "eval_samples_per_second": 570.361, + "eval_steps_per_second": 9.15, + "step": 8586 + }, + { + "epoch": 82.0, + "grad_norm": 1.8669618368148804, + "learning_rate": 9e-06, + "loss": 0.0388, + "step": 8692 + }, + { + "epoch": 82.0, + "eval_LOCATION_f1": 0.7947019867549668, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.759493670886076, + "eval_LOCATION_recall": 0.8333333333333334, + "eval_ORGANIZATION_f1": 0.7674418604651163, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.6804123711340206, + "eval_ORGANIZATION_recall": 0.88, + "eval_PERSON_f1": 0.903010033444816, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8653846153846154, + "eval_PERSON_recall": 0.9440559440559441, + "eval_QUANTITY_f1": 0.6415094339622641, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5666666666666667, + "eval_QUANTITY_recall": 0.7391304347826086, + "eval_TIME_f1": 0.7777777777777779, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.75, + "eval_TIME_recall": 0.8076923076923077, + "eval_loss": 0.173567995429039, + "eval_overall_accuracy": 0.961220915686765, + "eval_overall_f1": 0.8203017832647463, + "eval_overall_precision": 0.7666666666666667, + "eval_overall_recall": 0.8820058997050148, + "eval_runtime": 0.3284, + "eval_samples_per_second": 569.49, + "eval_steps_per_second": 9.136, + "step": 8692 + }, + { + "epoch": 83.0, + "grad_norm": 13.200667381286621, + "learning_rate": 8.500000000000002e-06, + "loss": 0.0378, + "step": 8798 + }, + { + "epoch": 83.0, + "eval_LOCATION_f1": 0.8026315789473685, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7625, + "eval_LOCATION_recall": 0.8472222222222222, + "eval_ORGANIZATION_f1": 0.7674418604651163, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.6804123711340206, + "eval_ORGANIZATION_recall": 0.88, + "eval_PERSON_f1": 0.9, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8598726114649682, + "eval_PERSON_recall": 0.9440559440559441, + "eval_QUANTITY_f1": 0.6296296296296297, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5483870967741935, + "eval_QUANTITY_recall": 0.7391304347826086, + "eval_TIME_f1": 0.7777777777777779, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.75, + "eval_TIME_recall": 0.8076923076923077, + "eval_loss": 0.1750916689634323, + "eval_overall_accuracy": 0.9609707280460346, + "eval_overall_f1": 0.819672131147541, + "eval_overall_precision": 0.7633587786259542, + "eval_overall_recall": 0.8849557522123894, + "eval_runtime": 0.3224, + "eval_samples_per_second": 580.046, + "eval_steps_per_second": 9.306, + "step": 8798 + }, + { + "epoch": 84.0, + "grad_norm": 1.3611278533935547, + "learning_rate": 8.000000000000001e-06, + "loss": 0.0381, + "step": 8904 + }, + { + "epoch": 84.0, + "eval_LOCATION_f1": 0.7919463087248323, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7662337662337663, + "eval_LOCATION_recall": 0.8194444444444444, + "eval_ORGANIZATION_f1": 0.7885714285714287, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.69, + "eval_ORGANIZATION_recall": 0.92, + "eval_PERSON_f1": 0.9060402684563759, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8709677419354839, + "eval_PERSON_recall": 0.9440559440559441, + "eval_QUANTITY_f1": 0.6415094339622641, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5666666666666667, + "eval_QUANTITY_recall": 0.7391304347826086, + "eval_TIME_f1": 0.8461538461538461, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.8461538461538461, + "eval_TIME_recall": 0.8461538461538461, + "eval_loss": 0.1736917495727539, + "eval_overall_accuracy": 0.9619714786089567, + "eval_overall_f1": 0.8308115543328748, + "eval_overall_precision": 0.7783505154639175, + "eval_overall_recall": 0.8908554572271387, + "eval_runtime": 0.3237, + "eval_samples_per_second": 577.629, + "eval_steps_per_second": 9.267, + "step": 8904 + }, + { + "epoch": 85.0, + "grad_norm": 1.1706650257110596, + "learning_rate": 7.5e-06, + "loss": 0.0363, + "step": 9010 + }, + { + "epoch": 85.0, + "eval_LOCATION_f1": 0.7947019867549668, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.759493670886076, + "eval_LOCATION_recall": 0.8333333333333334, + "eval_ORGANIZATION_f1": 0.757396449704142, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.6808510638297872, + "eval_ORGANIZATION_recall": 0.8533333333333334, + "eval_PERSON_f1": 0.8940397350993377, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8490566037735849, + "eval_PERSON_recall": 0.9440559440559441, + "eval_QUANTITY_f1": 0.6037735849056605, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5333333333333333, + "eval_QUANTITY_recall": 0.6956521739130435, + "eval_TIME_f1": 0.8461538461538461, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.8461538461538461, + "eval_TIME_recall": 0.8461538461538461, + "eval_loss": 0.17205749452114105, + "eval_overall_accuracy": 0.961220915686765, + "eval_overall_f1": 0.8170563961485556, + "eval_overall_precision": 0.7654639175257731, + "eval_overall_recall": 0.8761061946902655, + "eval_runtime": 0.3223, + "eval_samples_per_second": 580.279, + "eval_steps_per_second": 9.309, + "step": 9010 + }, + { + "epoch": 86.0, + "grad_norm": 2.736619234085083, + "learning_rate": 7.000000000000001e-06, + "loss": 0.0403, + "step": 9116 + }, + { + "epoch": 86.0, + "eval_LOCATION_f1": 0.8026315789473685, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7625, + "eval_LOCATION_recall": 0.8472222222222222, + "eval_ORGANIZATION_f1": 0.7885714285714287, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.69, + "eval_ORGANIZATION_recall": 0.92, + "eval_PERSON_f1": 0.903010033444816, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8653846153846154, + "eval_PERSON_recall": 0.9440559440559441, + "eval_QUANTITY_f1": 0.6415094339622641, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5666666666666667, + "eval_QUANTITY_recall": 0.7391304347826086, + "eval_TIME_f1": 0.7924528301886792, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.7777777777777778, + "eval_TIME_recall": 0.8076923076923077, + "eval_loss": 0.17720893025398254, + "eval_overall_accuracy": 0.9614711033274956, + "eval_overall_f1": 0.8278688524590163, + "eval_overall_precision": 0.7709923664122137, + "eval_overall_recall": 0.8938053097345132, + "eval_runtime": 0.3252, + "eval_samples_per_second": 575.026, + "eval_steps_per_second": 9.225, + "step": 9116 + }, + { + "epoch": 87.0, + "grad_norm": 1.3812779188156128, + "learning_rate": 6.5000000000000004e-06, + "loss": 0.0365, + "step": 9222 + }, + { + "epoch": 87.0, + "eval_LOCATION_f1": 0.8, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7692307692307693, + "eval_LOCATION_recall": 0.8333333333333334, + "eval_ORGANIZATION_f1": 0.7586206896551725, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.6666666666666666, + "eval_ORGANIZATION_recall": 0.88, + "eval_PERSON_f1": 0.9, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8598726114649682, + "eval_PERSON_recall": 0.9440559440559441, + "eval_QUANTITY_f1": 0.6037735849056605, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5333333333333333, + "eval_QUANTITY_recall": 0.6956521739130435, + "eval_TIME_f1": 0.7924528301886792, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.7777777777777778, + "eval_TIME_recall": 0.8076923076923077, + "eval_loss": 0.17484453320503235, + "eval_overall_accuracy": 0.9602201651238429, + "eval_overall_f1": 0.8164383561643836, + "eval_overall_precision": 0.7621483375959079, + "eval_overall_recall": 0.8790560471976401, + "eval_runtime": 0.3286, + "eval_samples_per_second": 569.045, + "eval_steps_per_second": 9.129, + "step": 9222 + }, + { + "epoch": 88.0, + "grad_norm": 1.9501535892486572, + "learning_rate": 6e-06, + "loss": 0.035, + "step": 9328 + }, + { + "epoch": 88.0, + "eval_LOCATION_f1": 0.8, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7692307692307693, + "eval_LOCATION_recall": 0.8333333333333334, + "eval_ORGANIZATION_f1": 0.7558139534883721, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.6701030927835051, + "eval_ORGANIZATION_recall": 0.8666666666666667, + "eval_PERSON_f1": 0.903010033444816, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8653846153846154, + "eval_PERSON_recall": 0.9440559440559441, + "eval_QUANTITY_f1": 0.6037735849056605, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5333333333333333, + "eval_QUANTITY_recall": 0.6956521739130435, + "eval_TIME_f1": 0.830188679245283, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.8148148148148148, + "eval_TIME_recall": 0.8461538461538461, + "eval_loss": 0.17459146678447723, + "eval_overall_accuracy": 0.9609707280460346, + "eval_overall_f1": 0.8198074277854195, + "eval_overall_precision": 0.7680412371134021, + "eval_overall_recall": 0.8790560471976401, + "eval_runtime": 0.3285, + "eval_samples_per_second": 569.214, + "eval_steps_per_second": 9.132, + "step": 9328 + }, + { + "epoch": 89.0, + "grad_norm": 0.9569804072380066, + "learning_rate": 5.500000000000001e-06, + "loss": 0.0355, + "step": 9434 + }, + { + "epoch": 89.0, + "eval_LOCATION_f1": 0.8079470198675496, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7721518987341772, + "eval_LOCATION_recall": 0.8472222222222222, + "eval_ORGANIZATION_f1": 0.7630057803468209, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.673469387755102, + "eval_ORGANIZATION_recall": 0.88, + "eval_PERSON_f1": 0.8970099667774087, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8544303797468354, + "eval_PERSON_recall": 0.9440559440559441, + "eval_QUANTITY_f1": 0.6415094339622641, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5666666666666667, + "eval_QUANTITY_recall": 0.7391304347826086, + "eval_TIME_f1": 0.7924528301886792, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.7777777777777778, + "eval_TIME_recall": 0.8076923076923077, + "eval_loss": 0.17787276208400726, + "eval_overall_accuracy": 0.960720540405304, + "eval_overall_f1": 0.8207934336525308, + "eval_overall_precision": 0.7653061224489796, + "eval_overall_recall": 0.8849557522123894, + "eval_runtime": 0.328, + "eval_samples_per_second": 570.159, + "eval_steps_per_second": 9.147, + "step": 9434 + }, + { + "epoch": 90.0, + "grad_norm": 0.9663112163543701, + "learning_rate": 5e-06, + "loss": 0.0371, + "step": 9540 + }, + { + "epoch": 90.0, + "eval_LOCATION_f1": 0.8079470198675496, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7721518987341772, + "eval_LOCATION_recall": 0.8472222222222222, + "eval_ORGANIZATION_f1": 0.757396449704142, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.6808510638297872, + "eval_ORGANIZATION_recall": 0.8533333333333334, + "eval_PERSON_f1": 0.903010033444816, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8653846153846154, + "eval_PERSON_recall": 0.9440559440559441, + "eval_QUANTITY_f1": 0.6415094339622641, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5666666666666667, + "eval_QUANTITY_recall": 0.7391304347826086, + "eval_TIME_f1": 0.830188679245283, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.8148148148148148, + "eval_TIME_recall": 0.8461538461538461, + "eval_loss": 0.1753513514995575, + "eval_overall_accuracy": 0.9614711033274956, + "eval_overall_f1": 0.8248275862068966, + "eval_overall_precision": 0.7746113989637305, + "eval_overall_recall": 0.8820058997050148, + "eval_runtime": 0.3275, + "eval_samples_per_second": 570.927, + "eval_steps_per_second": 9.159, + "step": 9540 + }, + { + "epoch": 91.0, + "grad_norm": 0.43051081895828247, + "learning_rate": 4.5e-06, + "loss": 0.0362, + "step": 9646 + }, + { + "epoch": 91.0, + "eval_LOCATION_f1": 0.8133333333333332, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.782051282051282, + "eval_LOCATION_recall": 0.8472222222222222, + "eval_ORGANIZATION_f1": 0.7664670658682634, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.6956521739130435, + "eval_ORGANIZATION_recall": 0.8533333333333334, + "eval_PERSON_f1": 0.903010033444816, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8653846153846154, + "eval_PERSON_recall": 0.9440559440559441, + "eval_QUANTITY_f1": 0.6037735849056605, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5333333333333333, + "eval_QUANTITY_recall": 0.6956521739130435, + "eval_TIME_f1": 0.8461538461538461, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.8461538461538461, + "eval_TIME_recall": 0.8461538461538461, + "eval_loss": 0.1716061383485794, + "eval_overall_accuracy": 0.9624718538904178, + "eval_overall_f1": 0.826629680998613, + "eval_overall_precision": 0.7801047120418848, + "eval_overall_recall": 0.8790560471976401, + "eval_runtime": 0.3284, + "eval_samples_per_second": 569.475, + "eval_steps_per_second": 9.136, + "step": 9646 + }, + { + "epoch": 92.0, + "grad_norm": 2.0720338821411133, + "learning_rate": 4.000000000000001e-06, + "loss": 0.0368, + "step": 9752 + }, + { + "epoch": 92.0, + "eval_LOCATION_f1": 0.8, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7692307692307693, + "eval_LOCATION_recall": 0.8333333333333334, + "eval_ORGANIZATION_f1": 0.744186046511628, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.6597938144329897, + "eval_ORGANIZATION_recall": 0.8533333333333334, + "eval_PERSON_f1": 0.903010033444816, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8653846153846154, + "eval_PERSON_recall": 0.9440559440559441, + "eval_QUANTITY_f1": 0.6037735849056605, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5333333333333333, + "eval_QUANTITY_recall": 0.6956521739130435, + "eval_TIME_f1": 0.830188679245283, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.8148148148148148, + "eval_TIME_recall": 0.8461538461538461, + "eval_loss": 0.1745324581861496, + "eval_overall_accuracy": 0.961220915686765, + "eval_overall_f1": 0.8170563961485556, + "eval_overall_precision": 0.7654639175257731, + "eval_overall_recall": 0.8761061946902655, + "eval_runtime": 0.3224, + "eval_samples_per_second": 580.026, + "eval_steps_per_second": 9.305, + "step": 9752 + }, + { + "epoch": 93.0, + "grad_norm": 0.6741355061531067, + "learning_rate": 3.5000000000000004e-06, + "loss": 0.036, + "step": 9858 + }, + { + "epoch": 93.0, + "eval_LOCATION_f1": 0.8079470198675496, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7721518987341772, + "eval_LOCATION_recall": 0.8472222222222222, + "eval_ORGANIZATION_f1": 0.7630057803468209, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.673469387755102, + "eval_ORGANIZATION_recall": 0.88, + "eval_PERSON_f1": 0.903010033444816, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8653846153846154, + "eval_PERSON_recall": 0.9440559440559441, + "eval_QUANTITY_f1": 0.6037735849056605, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5333333333333333, + "eval_QUANTITY_recall": 0.6956521739130435, + "eval_TIME_f1": 0.7924528301886792, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.7777777777777778, + "eval_TIME_recall": 0.8076923076923077, + "eval_loss": 0.1763608753681183, + "eval_overall_accuracy": 0.9609707280460346, + "eval_overall_f1": 0.8203017832647463, + "eval_overall_precision": 0.7666666666666667, + "eval_overall_recall": 0.8820058997050148, + "eval_runtime": 0.3211, + "eval_samples_per_second": 582.286, + "eval_steps_per_second": 9.341, + "step": 9858 + }, + { + "epoch": 94.0, + "grad_norm": 0.7255008816719055, + "learning_rate": 3e-06, + "loss": 0.0367, + "step": 9964 + }, + { + "epoch": 94.0, + "eval_LOCATION_f1": 0.8079470198675496, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7721518987341772, + "eval_LOCATION_recall": 0.8472222222222222, + "eval_ORGANIZATION_f1": 0.7602339181286549, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.6770833333333334, + "eval_ORGANIZATION_recall": 0.8666666666666667, + "eval_PERSON_f1": 0.903010033444816, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8653846153846154, + "eval_PERSON_recall": 0.9440559440559441, + "eval_QUANTITY_f1": 0.6037735849056605, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5333333333333333, + "eval_QUANTITY_recall": 0.6956521739130435, + "eval_TIME_f1": 0.830188679245283, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.8148148148148148, + "eval_TIME_recall": 0.8461538461538461, + "eval_loss": 0.17439185082912445, + "eval_overall_accuracy": 0.961220915686765, + "eval_overall_f1": 0.8225584594222835, + "eval_overall_precision": 0.770618556701031, + "eval_overall_recall": 0.8820058997050148, + "eval_runtime": 0.3287, + "eval_samples_per_second": 568.908, + "eval_steps_per_second": 9.127, + "step": 9964 + }, + { + "epoch": 95.0, + "grad_norm": 0.538052499294281, + "learning_rate": 2.5e-06, + "loss": 0.0346, + "step": 10070 + }, + { + "epoch": 95.0, + "eval_LOCATION_f1": 0.8079470198675496, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7721518987341772, + "eval_LOCATION_recall": 0.8472222222222222, + "eval_ORGANIZATION_f1": 0.7630057803468209, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.673469387755102, + "eval_ORGANIZATION_recall": 0.88, + "eval_PERSON_f1": 0.903010033444816, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8653846153846154, + "eval_PERSON_recall": 0.9440559440559441, + "eval_QUANTITY_f1": 0.6415094339622641, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5666666666666667, + "eval_QUANTITY_recall": 0.7391304347826086, + "eval_TIME_f1": 0.7924528301886792, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.7777777777777778, + "eval_TIME_recall": 0.8076923076923077, + "eval_loss": 0.179026797413826, + "eval_overall_accuracy": 0.961220915686765, + "eval_overall_f1": 0.823045267489712, + "eval_overall_precision": 0.7692307692307693, + "eval_overall_recall": 0.8849557522123894, + "eval_runtime": 0.3283, + "eval_samples_per_second": 569.674, + "eval_steps_per_second": 9.139, + "step": 10070 + }, + { + "epoch": 96.0, + "grad_norm": 0.13673511147499084, + "learning_rate": 2.0000000000000003e-06, + "loss": 0.0365, + "step": 10176 + }, + { + "epoch": 96.0, + "eval_LOCATION_f1": 0.8133333333333332, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.782051282051282, + "eval_LOCATION_recall": 0.8472222222222222, + "eval_ORGANIZATION_f1": 0.7701149425287357, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.6767676767676768, + "eval_ORGANIZATION_recall": 0.8933333333333333, + "eval_PERSON_f1": 0.903010033444816, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8653846153846154, + "eval_PERSON_recall": 0.9440559440559441, + "eval_QUANTITY_f1": 0.6415094339622641, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5666666666666667, + "eval_QUANTITY_recall": 0.7391304347826086, + "eval_TIME_f1": 0.7924528301886792, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.7777777777777778, + "eval_TIME_recall": 0.8076923076923077, + "eval_loss": 0.1783270239830017, + "eval_overall_accuracy": 0.9617212909682261, + "eval_overall_f1": 0.8257887517146778, + "eval_overall_precision": 0.7717948717948718, + "eval_overall_recall": 0.887905604719764, + "eval_runtime": 0.3285, + "eval_samples_per_second": 569.252, + "eval_steps_per_second": 9.132, + "step": 10176 + }, + { + "epoch": 97.0, + "grad_norm": 0.6854485869407654, + "learning_rate": 1.5e-06, + "loss": 0.0341, + "step": 10282 + }, + { + "epoch": 97.0, + "eval_LOCATION_f1": 0.8079470198675496, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.7721518987341772, + "eval_LOCATION_recall": 0.8472222222222222, + "eval_ORGANIZATION_f1": 0.7701149425287357, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.6767676767676768, + "eval_ORGANIZATION_recall": 0.8933333333333333, + "eval_PERSON_f1": 0.9, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8598726114649682, + "eval_PERSON_recall": 0.9440559440559441, + "eval_QUANTITY_f1": 0.6415094339622641, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5666666666666667, + "eval_QUANTITY_recall": 0.7391304347826086, + "eval_TIME_f1": 0.7924528301886792, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.7777777777777778, + "eval_TIME_recall": 0.8076923076923077, + "eval_loss": 0.17918910086154938, + "eval_overall_accuracy": 0.9609707280460346, + "eval_overall_f1": 0.823529411764706, + "eval_overall_precision": 0.7678571428571429, + "eval_overall_recall": 0.887905604719764, + "eval_runtime": 0.33, + "eval_samples_per_second": 566.696, + "eval_steps_per_second": 9.091, + "step": 10282 + }, + { + "epoch": 98.0, + "grad_norm": 0.28252169489860535, + "learning_rate": 1.0000000000000002e-06, + "loss": 0.0355, + "step": 10388 + }, + { + "epoch": 98.0, + "eval_LOCATION_f1": 0.8133333333333332, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.782051282051282, + "eval_LOCATION_recall": 0.8472222222222222, + "eval_ORGANIZATION_f1": 0.7692307692307693, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.6914893617021277, + "eval_ORGANIZATION_recall": 0.8666666666666667, + "eval_PERSON_f1": 0.903010033444816, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8653846153846154, + "eval_PERSON_recall": 0.9440559440559441, + "eval_QUANTITY_f1": 0.6037735849056605, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5333333333333333, + "eval_QUANTITY_recall": 0.6956521739130435, + "eval_TIME_f1": 0.8461538461538461, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.8461538461538461, + "eval_TIME_recall": 0.8461538461538461, + "eval_loss": 0.17533360421657562, + "eval_overall_accuracy": 0.9619714786089567, + "eval_overall_f1": 0.8271092669432918, + "eval_overall_precision": 0.7786458333333334, + "eval_overall_recall": 0.8820058997050148, + "eval_runtime": 0.3214, + "eval_samples_per_second": 581.777, + "eval_steps_per_second": 9.333, + "step": 10388 + }, + { + "epoch": 99.0, + "grad_norm": 1.7645491361618042, + "learning_rate": 5.000000000000001e-07, + "loss": 0.0369, + "step": 10494 + }, + { + "epoch": 99.0, + "eval_LOCATION_f1": 0.8133333333333332, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.782051282051282, + "eval_LOCATION_recall": 0.8472222222222222, + "eval_ORGANIZATION_f1": 0.7558139534883721, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.6701030927835051, + "eval_ORGANIZATION_recall": 0.8666666666666667, + "eval_PERSON_f1": 0.903010033444816, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8653846153846154, + "eval_PERSON_recall": 0.9440559440559441, + "eval_QUANTITY_f1": 0.6415094339622641, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5666666666666667, + "eval_QUANTITY_recall": 0.7391304347826086, + "eval_TIME_f1": 0.830188679245283, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.8148148148148148, + "eval_TIME_recall": 0.8461538461538461, + "eval_loss": 0.17752301692962646, + "eval_overall_accuracy": 0.9617212909682261, + "eval_overall_f1": 0.8253094910591473, + "eval_overall_precision": 0.7731958762886598, + "eval_overall_recall": 0.8849557522123894, + "eval_runtime": 0.3285, + "eval_samples_per_second": 569.299, + "eval_steps_per_second": 9.133, + "step": 10494 + }, + { + "epoch": 100.0, + "grad_norm": 0.791311502456665, + "learning_rate": 0.0, + "loss": 0.0366, + "step": 10600 + }, + { + "epoch": 100.0, + "eval_LOCATION_f1": 0.8133333333333332, + "eval_LOCATION_number": 72, + "eval_LOCATION_precision": 0.782051282051282, + "eval_LOCATION_recall": 0.8472222222222222, + "eval_ORGANIZATION_f1": 0.7558139534883721, + "eval_ORGANIZATION_number": 75, + "eval_ORGANIZATION_precision": 0.6701030927835051, + "eval_ORGANIZATION_recall": 0.8666666666666667, + "eval_PERSON_f1": 0.903010033444816, + "eval_PERSON_number": 143, + "eval_PERSON_precision": 0.8653846153846154, + "eval_PERSON_recall": 0.9440559440559441, + "eval_QUANTITY_f1": 0.6415094339622641, + "eval_QUANTITY_number": 23, + "eval_QUANTITY_precision": 0.5666666666666667, + "eval_QUANTITY_recall": 0.7391304347826086, + "eval_TIME_f1": 0.830188679245283, + "eval_TIME_number": 26, + "eval_TIME_precision": 0.8148148148148148, + "eval_TIME_recall": 0.8461538461538461, + "eval_loss": 0.17746011912822723, + "eval_overall_accuracy": 0.9617212909682261, + "eval_overall_f1": 0.8253094910591473, + "eval_overall_precision": 0.7731958762886598, + "eval_overall_recall": 0.8849557522123894, + "eval_runtime": 0.3267, + "eval_samples_per_second": 572.32, + "eval_steps_per_second": 9.182, + "step": 10600 + }, + { + "epoch": 100.0, + "step": 10600, + "total_flos": 4549660816698438.0, + "train_loss": 0.08814868762808026, + "train_runtime": 609.0931, + "train_samples_per_second": 276.969, + "train_steps_per_second": 17.403 + } + ], + "logging_steps": 500, + "max_steps": 10600, + "num_input_tokens_seen": 0, + "num_train_epochs": 100, + "save_steps": 500, + "total_flos": 4549660816698438.0, + "train_batch_size": 16, + "trial_name": null, + "trial_params": null +}