{ "kind": "validation", "global_step": 8868, "aggregate_epoch": 4.0, "content_macro_f1": 0.7691223696148852, "content_f1_class2_3_mean": 0.7002175340967471, "dominated_in_both_objectives": false, "bad_streak": 0, "metrics": { "aggregate_equal_context_weight": { "selection_loss": 0.5140893343795243, "content_mae": 0.2737932315832202, "content_macro_mae": 0.3100500287727975, "content_exact_accuracy": 0.7986166666666666, "content_spearman": 0.8996877860492928, "content_gaussian_nll": -0.7082690140020707, "content_mean_predicted_variance": 0.1657321429377615, "integrity_accuracy": 0.9669555555555556, "integrity_macro_f1": 0.5279255895888997, "critical_average_precision": 0.18384277757309278, "critical_roc_auc": 0.9641408554278599, "critical_f1_at_0_5": 0.22842197035745423, "critical_brier": 0.004478420865291646, "final_mae": 0.2757862484826717, "final_macro_mae": 0.3203746614823401, "final_exact_accuracy": 0.7975944444444445, "final_spearman": 0.8992452079736484, "integrity_confusion_matrix": [ [ 172551, 3035, 574 ], [ 1757, 1296, 157 ], [ 300, 125, 205 ] ], "content_distribution": { "0": 17487, "1": 85223, "2": 44101, "3": 22566, "4": 10623 }, "final_distribution": { "0": 17487, "1": 85223, "2": 44690, "3": 22868, "4": 9732 }, "content_per_class": { "0": { "precision": 0.8771087093269285, "recall": 0.7591190299430834, "f1": 0.8138597049771835, "support": 20205 }, "1": { "precision": 0.8654705889255248, "recall": 0.8802090791923242, "f1": 0.8727776167176471, "support": 83796 }, "2": { "precision": 0.7085100111108592, "recall": 0.7307980166526336, "f1": 0.719481446515537, "support": 42756 }, "3": { "precision": 0.6948949747407605, "recall": 0.6675606641123882, "f1": 0.6809536216779573, "support": 23490 }, "4": { "precision": 0.7274781135272522, "recall": 0.7923715779760074, "f1": 0.7585394581861012, "support": 9753 } }, "content_macro_f1": 0.7691223696148852, "content_f1_class2_3_mean": 0.7002175340967471, "content_mean_variance_correct": 0.14257208054755013, "content_mean_variance_wrong": 0.25757694220547306, "content_wrong_to_correct_variance_ratio": 1.806643637493716, "content_variance_wrong_auroc": 0.7498307193383456, "content_variance_wrong_average_precision": 0.37411770726590243, "content_one_sigma_accuracy": 0.7989111111111111, "content_fixed_half_point_accuracy": 0.7986166666666666, "content_mean_predicted_sigma": 0.3627248205399464, "content_mean_absolute_calibration_gap": 0.19831773181999218 }, "by_length": { "512": { "selection_loss": 0.5347708140474868, "content_mae": 0.2877122715353697, "content_macro_mae": 0.3326414165132969, "content_exact_accuracy": 0.7846833333333333, "content_spearman": 0.8920195781597556, "content_gaussian_nll": -0.6381107757869618, "content_mean_predicted_variance": 0.1784611644987229, "integrity_accuracy": 0.9649666666666666, "integrity_macro_f1": 0.5231860914821642, "critical_average_precision": 0.1914824098632402, "critical_roc_auc": 0.9600141766022349, "critical_f1_at_0_5": 0.24, "critical_brier": 0.004935151331732246, "final_mae": 0.2900404591834574, "final_macro_mae": 0.345389726812951, "final_exact_accuracy": 0.7832666666666667, "final_spearman": 0.8915067951762922, "integrity_confusion_matrix": [ [ 57398, 1095, 227 ], [ 583, 424, 63 ], [ 96, 38, 76 ] ], "content_distribution": { "0": 5814, "1": 28407, "2": 14862, "3": 7542, "4": 3375 }, "final_distribution": { "0": 5814, "1": 28407, "2": 15133, "3": 7661, "4": 2985 }, "content_per_class": { "0": { "precision": 0.8753009975920193, "recall": 0.7556050482553823, "f1": 0.8110606422822536, "support": 6735 }, "1": { "precision": 0.857922343084451, "recall": 0.8725118144064156, "f1": 0.8651555760663129, "support": 27932 }, "2": { "precision": 0.6829498048714843, "recall": 0.712180746561886, "f1": 0.6972590506285635, "support": 14252 }, "3": { "precision": 0.6675948024396712, "recall": 0.6430395913154534, "f1": 0.6550871714806141, "support": 7830 }, "4": { "precision": 0.7217777777777777, "recall": 0.74930790525992, "f1": 0.735285239963779, "support": 3251 } }, "content_macro_f1": 0.7527695360843045, "content_f1_class2_3_mean": 0.6761731110545888, "content_mean_variance_correct": 0.1514430107925402, "content_mean_variance_wrong": 0.27692402498643776, "content_wrong_to_correct_variance_ratio": 1.8285691993128186, "content_variance_wrong_auroc": 0.7523747674967851, "content_variance_wrong_average_precision": 0.3968017995070952, "content_one_sigma_accuracy": 0.7935333333333333, "content_fixed_half_point_accuracy": 0.7846833333333333, "content_mean_predicted_sigma": 0.37587916885792183, "content_mean_absolute_calibration_gap": 0.20801088074152765 }, "1024": { "selection_loss": 0.5039974774694058, "content_mae": 0.26787705369252196, "content_macro_mae": 0.3007754810817112, "content_exact_accuracy": 0.8042833333333334, "content_spearman": 0.9028312134289629, "content_gaussian_nll": -0.7359567302086514, "content_mean_predicted_variance": 0.1592265743720674, "integrity_accuracy": 0.96835, "integrity_macro_f1": 0.530696196209282, "critical_average_precision": 0.18711201444922157, "critical_roc_auc": 0.9656394205114728, "critical_f1_at_0_5": 0.22589531680440772, "critical_brier": 0.004262811449088174, "final_mae": 0.26969235729647084, "final_macro_mae": 0.3098881766812208, "final_exact_accuracy": 0.8035666666666667, "final_spearman": 0.9024137669725919, "integrity_confusion_matrix": [ [ 57606, 938, 176 ], [ 593, 430, 47 ], [ 101, 44, 65 ] ], "content_distribution": { "0": 5802, "1": 28349, "2": 14688, "3": 7550, "4": 3611 }, "final_distribution": { "0": 5802, "1": 28349, "2": 14855, "3": 7639, "4": 3355 }, "content_per_class": { "0": { "precision": 0.8796966563254051, "recall": 0.7578322197475872, "f1": 0.8142298795565127, "support": 6735 }, "1": { "precision": 0.8693075593495362, "recall": 0.8822855506229415, "f1": 0.8757484763952311, "support": 27932 }, "2": { "precision": 0.71875, "recall": 0.7407381420151558, "f1": 0.7295784381478921, "support": 14252 }, "3": { "precision": 0.7054304635761589, "recall": 0.6802043422733078, "f1": 0.6925877763328999, "support": 7830 }, "4": { "precision": 0.7272223760731099, "recall": 0.8077514610888957, "f1": 0.7653745263771495, "support": 3251 } }, "content_macro_f1": 0.775503819361937, "content_f1_class2_3_mean": 0.711083107240396, "content_mean_variance_correct": 0.13785730128542895, "content_mean_variance_wrong": 0.24704203986997347, "content_wrong_to_correct_variance_ratio": 1.7920127375660806, "content_variance_wrong_auroc": 0.7488252815378621, "content_variance_wrong_average_precision": 0.365055072543468, "content_one_sigma_accuracy": 0.7999333333333334, "content_fixed_half_point_accuracy": 0.8042833333333334, "content_mean_predicted_sigma": 0.35608551753900974, "content_mean_absolute_calibration_gap": 0.19391285502780448 }, "2048": { "selection_loss": 0.5010323490574853, "content_mae": 0.265790369521769, "content_macro_mae": 0.2967331887233845, "content_exact_accuracy": 0.8068833333333333, "content_spearman": 0.9041704167079047, "content_gaussian_nll": -0.7507395360105985, "content_mean_predicted_variance": 0.15950868994249426, "integrity_accuracy": 0.96755, "integrity_macro_f1": 0.5300005982213699, "critical_average_precision": 0.182803358663597, "critical_roc_auc": 0.9674798301993486, "critical_f1_at_0_5": 0.21727019498607242, "critical_brier": 0.004237299815054516, "final_mae": 0.2676259289680868, "final_macro_mae": 0.3058460809528486, "final_exact_accuracy": 0.80595, "final_spearman": 0.903772047903423, "integrity_confusion_matrix": [ [ 57547, 1002, 171 ], [ 581, 442, 47 ], [ 103, 43, 64 ] ], "content_distribution": { "0": 5871, "1": 28467, "2": 14551, "3": 7474, "4": 3637 }, "final_distribution": { "0": 5871, "1": 28467, "2": 14702, "3": 7568, "4": 3392 }, "content_per_class": { "0": { "precision": 0.8763413387838528, "recall": 0.7639198218262806, "f1": 0.8162779628748215, "support": 6735 }, "1": { "precision": 0.8691818596971932, "recall": 0.8858298725476157, "f1": 0.8774269047323534, "support": 27932 }, "2": { "precision": 0.7242801182049343, "recall": 0.7394751613808588, "f1": 0.7317987709613581, "support": 14252 }, "3": { "precision": 0.7118009098207118, "recall": 0.6794380587484036, "f1": 0.6952430737062206, "support": 7830 }, "4": { "precision": 0.7330217211987902, "recall": 0.8200553675792064, "f1": 0.7740998838559814, "support": 3251 } }, "content_macro_f1": 0.7789693192261471, "content_f1_class2_3_mean": 0.7135209223337894, "content_mean_variance_correct": 0.13864480555917522, "content_mean_variance_wrong": 0.24668252567647403, "content_wrong_to_correct_variance_ratio": 1.77924102299806, "content_variance_wrong_auroc": 0.7460495586192982, "content_variance_wrong_average_precision": 0.35514109972948527, "content_one_sigma_accuracy": 0.8032666666666667, "content_fixed_half_point_accuracy": 0.8068833333333333, "content_mean_predicted_sigma": 0.35620977522290764, "content_mean_absolute_calibration_gap": 0.19302945969064433 } } }, "logged_at": "2026-10-04T00:33:26.173277+00:00" }