{ "best_metric": 0.80243850868883, "best_model_checkpoint": "legal-bert-lora-no-grad/checkpoint-12217", "epoch": 21.0, "eval_steps": 500, "global_step": 13503, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.78, "learning_rate": 4.870399170554692e-05, "loss": 1.5473, "step": 500 }, { "epoch": 1.0, "eval_accuracy": 0.7172734314484895, "eval_f1_macro": 0.3920432121625908, "eval_f1_micro": 0.7172734314484894, "eval_f1_weighted": 0.6868869808248764, "eval_loss": 0.8484711050987244, "eval_macro_fpr": 0.027836504093424867, "eval_macro_sensitivity": 0.44215864388540455, "eval_macro_specificity": 0.9788744549045134, "eval_precision": 0.6896842532804647, "eval_precision_macro": 0.3892290687521846, "eval_recall": 0.7172734314484895, "eval_recall_macro": 0.44215864388540455, "eval_runtime": 32.4222, "eval_samples_per_second": 39.818, "eval_steps_per_second": 4.997, "eval_weighted_fpr": 0.027383899767424412, "eval_weighted_sensitivity": 0.7172734314484895, "eval_weighted_specificity": 0.9658433921192108, "step": 643 }, { "epoch": 1.56, "learning_rate": 4.740798341109383e-05, "loss": 0.7816, "step": 1000 }, { "epoch": 2.0, "eval_accuracy": 0.7544539116963594, "eval_f1_macro": 0.47513531679703136, "eval_f1_micro": 0.7544539116963594, "eval_f1_weighted": 0.7304002619527014, "eval_loss": 0.7113391757011414, "eval_macro_fpr": 0.023051270843305598, "eval_macro_sensitivity": 0.5054156753128263, "eval_macro_specificity": 0.9817050378554314, "eval_precision": 0.7491876907285084, "eval_precision_macro": 0.5281699878111706, "eval_recall": 0.7544539116963594, "eval_recall_macro": 0.5054156753128263, "eval_runtime": 33.6622, "eval_samples_per_second": 38.352, "eval_steps_per_second": 4.813, "eval_weighted_fpr": 0.02271912850283093, "eval_weighted_sensitivity": 0.7544539116963594, "eval_weighted_specificity": 0.9711216561351127, "step": 1286 }, { "epoch": 2.33, "learning_rate": 4.6111975116640746e-05, "loss": 0.6956, "step": 1500 }, { "epoch": 3.0, "eval_accuracy": 0.7986057319907048, "eval_f1_macro": 0.5439483491438576, "eval_f1_micro": 0.7986057319907048, "eval_f1_weighted": 0.7833451448204543, "eval_loss": 0.6460158824920654, "eval_macro_fpr": 0.01844352926124183, "eval_macro_sensitivity": 0.5700868724489073, "eval_macro_specificity": 0.9848453217277398, "eval_precision": 0.7793241511142787, "eval_precision_macro": 0.5435553848320287, "eval_recall": 0.7986057319907048, "eval_recall_macro": 0.5700868724489073, "eval_runtime": 33.205, "eval_samples_per_second": 38.88, "eval_steps_per_second": 4.879, "eval_weighted_fpr": 0.01769429699196951, "eval_weighted_sensitivity": 0.7986057319907048, "eval_weighted_specificity": 0.974074093925392, "step": 1929 }, { "epoch": 3.11, "learning_rate": 4.481596682218767e-05, "loss": 0.6117, "step": 2000 }, { "epoch": 3.89, "learning_rate": 4.3519958527734576e-05, "loss": 0.4942, "step": 2500 }, { "epoch": 4.0, "eval_accuracy": 0.8109992254066615, "eval_f1_macro": 0.641318903734146, "eval_f1_micro": 0.8109992254066615, "eval_f1_weighted": 0.8036671817127167, "eval_loss": 0.6429558992385864, "eval_macro_fpr": 0.016902452669854486, "eval_macro_sensitivity": 0.6670193163886222, "eval_macro_specificity": 0.9858199536379567, "eval_precision": 0.8014184747492699, "eval_precision_macro": 0.6315294283267076, "eval_recall": 0.8109992254066615, "eval_recall_macro": 0.6670193163886222, "eval_runtime": 34.2851, "eval_samples_per_second": 37.655, "eval_steps_per_second": 4.725, "eval_weighted_fpr": 0.016373641121997046, "eval_weighted_sensitivity": 0.8109992254066615, "eval_weighted_specificity": 0.9763000791626885, "step": 2572 }, { "epoch": 4.67, "learning_rate": 4.22239502332815e-05, "loss": 0.4088, "step": 3000 }, { "epoch": 5.0, "eval_accuracy": 0.8319132455460883, "eval_f1_macro": 0.7458887908906745, "eval_f1_micro": 0.8319132455460883, "eval_f1_weighted": 0.8254435175521364, "eval_loss": 0.7148057222366333, "eval_macro_fpr": 0.015014171873602745, "eval_macro_sensitivity": 0.7444010739541649, "eval_macro_specificity": 0.9871775908314393, "eval_precision": 0.8301095647411415, "eval_precision_macro": 0.795128331887366, "eval_recall": 0.8319132455460883, "eval_recall_macro": 0.7444010739541649, "eval_runtime": 32.6496, "eval_samples_per_second": 39.541, "eval_steps_per_second": 4.962, "eval_weighted_fpr": 0.014226709499770536, "eval_weighted_sensitivity": 0.8319132455460883, "eval_weighted_specificity": 0.9757506169255016, "step": 3215 }, { "epoch": 5.44, "learning_rate": 4.092794193882841e-05, "loss": 0.3722, "step": 3500 }, { "epoch": 6.0, "eval_accuracy": 0.8319132455460883, "eval_f1_macro": 0.7581556537237687, "eval_f1_micro": 0.8319132455460883, "eval_f1_weighted": 0.8284954274383471, "eval_loss": 0.7203006744384766, "eval_macro_fpr": 0.014862611036154928, "eval_macro_sensitivity": 0.7604143289637364, "eval_macro_specificity": 0.9873237068761225, "eval_precision": 0.8282196199964869, "eval_precision_macro": 0.7594204389852343, "eval_recall": 0.8319132455460883, "eval_recall_macro": 0.7604143289637364, "eval_runtime": 32.5293, "eval_samples_per_second": 39.687, "eval_steps_per_second": 4.98, "eval_weighted_fpr": 0.014226709499770536, "eval_weighted_sensitivity": 0.8319132455460883, "eval_weighted_specificity": 0.9779423575957491, "step": 3858 }, { "epoch": 6.22, "learning_rate": 3.963193364437533e-05, "loss": 0.3518, "step": 4000 }, { "epoch": 7.0, "learning_rate": 3.833592534992224e-05, "loss": 0.3088, "step": 4500 }, { "epoch": 7.0, "eval_accuracy": 0.8218435321456236, "eval_f1_macro": 0.748587043102148, "eval_f1_micro": 0.8218435321456236, "eval_f1_weighted": 0.8202672850983224, "eval_loss": 0.7796419858932495, "eval_macro_fpr": 0.015766934174909337, "eval_macro_sensitivity": 0.7436163865406156, "eval_macro_specificity": 0.986554823310482, "eval_precision": 0.8283195435686048, "eval_precision_macro": 0.7927343403715958, "eval_recall": 0.8218435321456236, "eval_recall_macro": 0.7436163865406156, "eval_runtime": 33.9886, "eval_samples_per_second": 37.983, "eval_steps_per_second": 4.766, "eval_weighted_fpr": 0.015247944842216919, "eval_weighted_sensitivity": 0.8218435321456236, "eval_weighted_specificity": 0.9764788175116066, "step": 4501 }, { "epoch": 7.78, "learning_rate": 3.7039917055469156e-05, "loss": 0.245, "step": 5000 }, { "epoch": 8.0, "eval_accuracy": 0.8187451587916343, "eval_f1_macro": 0.7408588571905798, "eval_f1_micro": 0.8187451587916342, "eval_f1_weighted": 0.8134714190372911, "eval_loss": 0.8731526136398315, "eval_macro_fpr": 0.016323302917471562, "eval_macro_sensitivity": 0.7403464639779053, "eval_macro_specificity": 0.98626453049897, "eval_precision": 0.8170725248278505, "eval_precision_macro": 0.7696333486311014, "eval_recall": 0.8187451587916343, "eval_recall_macro": 0.7403464639779053, "eval_runtime": 32.6647, "eval_samples_per_second": 39.523, "eval_steps_per_second": 4.959, "eval_weighted_fpr": 0.015566790846194785, "eval_weighted_sensitivity": 0.8187451587916343, "eval_weighted_specificity": 0.9752227986929156, "step": 5144 }, { "epoch": 8.55, "learning_rate": 3.574390876101607e-05, "loss": 0.2331, "step": 5500 }, { "epoch": 9.0, "eval_accuracy": 0.8264910921766073, "eval_f1_macro": 0.753016641591601, "eval_f1_micro": 0.8264910921766073, "eval_f1_weighted": 0.8260538424611459, "eval_loss": 0.871034562587738, "eval_macro_fpr": 0.015223198880351374, "eval_macro_sensitivity": 0.7521384355359189, "eval_macro_specificity": 0.9869379376488985, "eval_precision": 0.8280339150697505, "eval_precision_macro": 0.7595158765535532, "eval_recall": 0.8264910921766073, "eval_recall_macro": 0.7521384355359189, "eval_runtime": 32.6231, "eval_samples_per_second": 39.573, "eval_steps_per_second": 4.966, "eval_weighted_fpr": 0.014773776546629732, "eval_weighted_sensitivity": 0.8264910921766073, "eval_weighted_specificity": 0.9775779725568708, "step": 5787 }, { "epoch": 9.33, "learning_rate": 3.4447900466562985e-05, "loss": 0.1878, "step": 6000 }, { "epoch": 10.0, "eval_accuracy": 0.82571649883811, "eval_f1_macro": 0.7588436992559459, "eval_f1_micro": 0.82571649883811, "eval_f1_weighted": 0.8239752092365729, "eval_loss": 0.9866135120391846, "eval_macro_fpr": 0.015306627243977184, "eval_macro_sensitivity": 0.7603889662202937, "eval_macro_specificity": 0.9869690054538824, "eval_precision": 0.8260731585047043, "eval_precision_macro": 0.7672111037553849, "eval_recall": 0.82571649883811, "eval_recall_macro": 0.7603889662202937, "eval_runtime": 33.5564, "eval_samples_per_second": 38.473, "eval_steps_per_second": 4.828, "eval_weighted_fpr": 0.01485246550927454, "eval_weighted_sensitivity": 0.82571649883811, "eval_weighted_specificity": 0.9788185829701274, "step": 6430 }, { "epoch": 10.11, "learning_rate": 3.315189217210991e-05, "loss": 0.1934, "step": 6500 }, { "epoch": 10.89, "learning_rate": 3.1855883877656815e-05, "loss": 0.1627, "step": 7000 }, { "epoch": 11.0, "eval_accuracy": 0.82571649883811, "eval_f1_macro": 0.767473750579757, "eval_f1_micro": 0.82571649883811, "eval_f1_weighted": 0.8233713451484083, "eval_loss": 1.0529617071151733, "eval_macro_fpr": 0.015351918387195561, "eval_macro_sensitivity": 0.7711152769110098, "eval_macro_specificity": 0.9869581108070966, "eval_precision": 0.8268711216555169, "eval_precision_macro": 0.7706107626580333, "eval_recall": 0.82571649883811, "eval_recall_macro": 0.7711152769110098, "eval_runtime": 32.7851, "eval_samples_per_second": 39.378, "eval_steps_per_second": 4.941, "eval_weighted_fpr": 0.01485246550927454, "eval_weighted_sensitivity": 0.82571649883811, "eval_weighted_specificity": 0.9786551632683399, "step": 7073 }, { "epoch": 11.66, "learning_rate": 3.0559875583203736e-05, "loss": 0.1301, "step": 7500 }, { "epoch": 12.0, "eval_accuracy": 0.8264910921766073, "eval_f1_macro": 0.7575609282906195, "eval_f1_micro": 0.8264910921766073, "eval_f1_weighted": 0.8227831044737154, "eval_loss": 1.1041574478149414, "eval_macro_fpr": 0.015507747324292978, "eval_macro_sensitivity": 0.7586841377838723, "eval_macro_specificity": 0.986855053135477, "eval_precision": 0.8246142974152224, "eval_precision_macro": 0.7633072608154402, "eval_recall": 0.8264910921766073, "eval_recall_macro": 0.7586841377838723, "eval_runtime": 33.2755, "eval_samples_per_second": 38.797, "eval_steps_per_second": 4.868, "eval_weighted_fpr": 0.014773776546629732, "eval_weighted_sensitivity": 0.8264910921766073, "eval_weighted_specificity": 0.9763347048555474, "step": 7716 }, { "epoch": 12.44, "learning_rate": 2.926386728875065e-05, "loss": 0.1291, "step": 8000 }, { "epoch": 13.0, "eval_accuracy": 0.8233927188226181, "eval_f1_macro": 0.7580847580681433, "eval_f1_micro": 0.8233927188226181, "eval_f1_weighted": 0.8211060746900526, "eval_loss": 1.1460882425308228, "eval_macro_fpr": 0.015717446624001503, "eval_macro_sensitivity": 0.7612688033602836, "eval_macro_specificity": 0.9866762703086498, "eval_precision": 0.8215196709960565, "eval_precision_macro": 0.7582432160976473, "eval_recall": 0.8233927188226181, "eval_recall_macro": 0.7612688033602836, "eval_runtime": 32.7655, "eval_samples_per_second": 39.401, "eval_steps_per_second": 4.944, "eval_weighted_fpr": 0.015089344804765056, "eval_weighted_sensitivity": 0.8233927188226181, "eval_weighted_specificity": 0.9767513358071301, "step": 8359 }, { "epoch": 13.22, "learning_rate": 2.7967858994297562e-05, "loss": 0.1213, "step": 8500 }, { "epoch": 14.0, "learning_rate": 2.667185069984448e-05, "loss": 0.11, "step": 9000 }, { "epoch": 14.0, "eval_accuracy": 0.8226181254841208, "eval_f1_macro": 0.7627437208061313, "eval_f1_micro": 0.8226181254841208, "eval_f1_weighted": 0.819453013970647, "eval_loss": 1.1837199926376343, "eval_macro_fpr": 0.01587279795090152, "eval_macro_sensitivity": 0.7575723645459792, "eval_macro_specificity": 0.98654524509664, "eval_precision": 0.8182318569400476, "eval_precision_macro": 0.7716010744976882, "eval_recall": 0.8226181254841208, "eval_recall_macro": 0.7575723645459792, "eval_runtime": 32.1143, "eval_samples_per_second": 40.2, "eval_steps_per_second": 5.044, "eval_weighted_fpr": 0.01516857653838511, "eval_weighted_sensitivity": 0.8226181254841208, "eval_weighted_specificity": 0.9755605509654802, "step": 9002 }, { "epoch": 14.77, "learning_rate": 2.53758424053914e-05, "loss": 0.0863, "step": 9500 }, { "epoch": 15.0, "eval_accuracy": 0.8218435321456236, "eval_f1_macro": 0.7497899545528302, "eval_f1_micro": 0.8218435321456236, "eval_f1_weighted": 0.818300724101517, "eval_loss": 1.2020219564437866, "eval_macro_fpr": 0.01596050778819386, "eval_macro_sensitivity": 0.7457545376125337, "eval_macro_specificity": 0.9864813469122803, "eval_precision": 0.8185478073118013, "eval_precision_macro": 0.7615551574586096, "eval_recall": 0.8218435321456236, "eval_recall_macro": 0.7457545376125337, "eval_runtime": 32.9172, "eval_samples_per_second": 39.22, "eval_steps_per_second": 4.921, "eval_weighted_fpr": 0.015247944842216919, "eval_weighted_sensitivity": 0.8218435321456236, "eval_weighted_specificity": 0.9753766715385808, "step": 9645 }, { "epoch": 15.55, "learning_rate": 2.4079834110938313e-05, "loss": 0.0735, "step": 10000 }, { "epoch": 16.0, "eval_accuracy": 0.8187451587916343, "eval_f1_macro": 0.7516626482517895, "eval_f1_micro": 0.8187451587916342, "eval_f1_weighted": 0.8167711468473488, "eval_loss": 1.2490901947021484, "eval_macro_fpr": 0.01615903013896019, "eval_macro_sensitivity": 0.7463866263442519, "eval_macro_specificity": 0.9862838939338255, "eval_precision": 0.816010692122711, "eval_precision_macro": 0.7620383054800749, "eval_recall": 0.8187451587916343, "eval_recall_macro": 0.7463866263442519, "eval_runtime": 33.245, "eval_samples_per_second": 38.833, "eval_steps_per_second": 4.873, "eval_weighted_fpr": 0.015566790846194785, "eval_weighted_sensitivity": 0.8187451587916343, "eval_weighted_specificity": 0.9755132502157504, "step": 10288 }, { "epoch": 16.33, "learning_rate": 2.2783825816485228e-05, "loss": 0.0802, "step": 10500 }, { "epoch": 17.0, "eval_accuracy": 0.8164213787761425, "eval_f1_macro": 0.7483467346361333, "eval_f1_micro": 0.8164213787761425, "eval_f1_weighted": 0.8152034106224548, "eval_loss": 1.3287584781646729, "eval_macro_fpr": 0.016411109243053838, "eval_macro_sensitivity": 0.7530689671982778, "eval_macro_specificity": 0.986197275429763, "eval_precision": 0.8164986233001753, "eval_precision_macro": 0.7471296868733558, "eval_recall": 0.8164213787761425, "eval_recall_macro": 0.7530689671982778, "eval_runtime": 32.4097, "eval_samples_per_second": 39.834, "eval_steps_per_second": 4.999, "eval_weighted_fpr": 0.01580737677582872, "eval_weighted_sensitivity": 0.8164213787761425, "eval_weighted_specificity": 0.9765377526703035, "step": 10931 }, { "epoch": 17.11, "learning_rate": 2.1487817522032143e-05, "loss": 0.0517, "step": 11000 }, { "epoch": 17.88, "learning_rate": 2.0191809227579057e-05, "loss": 0.0525, "step": 11500 }, { "epoch": 18.0, "eval_accuracy": 0.8133230054221534, "eval_f1_macro": 0.7486434952959435, "eval_f1_micro": 0.8133230054221534, "eval_f1_weighted": 0.8105673102840433, "eval_loss": 1.3620411157608032, "eval_macro_fpr": 0.016771971366343907, "eval_macro_sensitivity": 0.747756174463169, "eval_macro_specificity": 0.9859165137366713, "eval_precision": 0.8127181738070712, "eval_precision_macro": 0.7556869343254475, "eval_recall": 0.8133230054221534, "eval_recall_macro": 0.747756174463169, "eval_runtime": 33.8455, "eval_samples_per_second": 38.144, "eval_steps_per_second": 4.786, "eval_weighted_fpr": 0.016130111772973698, "eval_weighted_sensitivity": 0.8133230054221534, "eval_weighted_specificity": 0.9754247006279183, "step": 11574 }, { "epoch": 18.66, "learning_rate": 1.8895800933125972e-05, "loss": 0.0474, "step": 12000 }, { "epoch": 19.0, "eval_accuracy": 0.8187451587916343, "eval_f1_macro": 0.80243850868883, "eval_f1_micro": 0.8187451587916342, "eval_f1_weighted": 0.8186882142398059, "eval_loss": 1.3783087730407715, "eval_macro_fpr": 0.01616653707794279, "eval_macro_sensitivity": 0.7971845339149931, "eval_macro_specificity": 0.9862841415250829, "eval_precision": 0.8219853895924313, "eval_precision_macro": 0.8153974354587618, "eval_recall": 0.8187451587916343, "eval_recall_macro": 0.7971845339149931, "eval_runtime": 34.0415, "eval_samples_per_second": 37.924, "eval_steps_per_second": 4.759, "eval_weighted_fpr": 0.015566790846194785, "eval_weighted_sensitivity": 0.8187451587916343, "eval_weighted_specificity": 0.9755169640846107, "step": 12217 }, { "epoch": 19.44, "learning_rate": 1.7599792638672887e-05, "loss": 0.0315, "step": 12500 }, { "epoch": 20.0, "eval_accuracy": 0.8226181254841208, "eval_f1_macro": 0.790748766871484, "eval_f1_micro": 0.8226181254841208, "eval_f1_weighted": 0.8231232301620584, "eval_loss": 1.400423288345337, "eval_macro_fpr": 0.015694178732213588, "eval_macro_sensitivity": 0.7910027660234611, "eval_macro_specificity": 0.986694893115232, "eval_precision": 0.8265938735422171, "eval_precision_macro": 0.8011102381814037, "eval_recall": 0.8226181254841208, "eval_recall_macro": 0.7910027660234611, "eval_runtime": 33.3314, "eval_samples_per_second": 38.732, "eval_steps_per_second": 4.86, "eval_weighted_fpr": 0.01516857653838511, "eval_weighted_sensitivity": 0.8226181254841208, "eval_weighted_specificity": 0.9778052712443601, "step": 12860 }, { "epoch": 20.22, "learning_rate": 1.6303784344219805e-05, "loss": 0.0379, "step": 13000 }, { "epoch": 21.0, "learning_rate": 1.500777604976672e-05, "loss": 0.0325, "step": 13500 }, { "epoch": 21.0, "eval_accuracy": 0.8187451587916343, "eval_f1_macro": 0.7920531234982356, "eval_f1_micro": 0.8187451587916342, "eval_f1_weighted": 0.8182432240494799, "eval_loss": 1.4682910442352295, "eval_macro_fpr": 0.016122147142737502, "eval_macro_sensitivity": 0.7876294399760013, "eval_macro_specificity": 0.9863487423677417, "eval_precision": 0.8197947092422727, "eval_precision_macro": 0.8030038288855238, "eval_recall": 0.8187451587916343, "eval_recall_macro": 0.7876294399760013, "eval_runtime": 34.2442, "eval_samples_per_second": 37.7, "eval_steps_per_second": 4.731, "eval_weighted_fpr": 0.015566790846194785, "eval_weighted_sensitivity": 0.8187451587916343, "eval_weighted_specificity": 0.9764859767244947, "step": 13503 } ], "logging_steps": 500, "max_steps": 19290, "num_train_epochs": 30, "save_steps": 500, "total_flos": 2.9284210564632576e+16, "trial_name": null, "trial_params": null }