{ "updated": "2026-09-07", "reference": { "name": "Test300", "measured": "2026-09-06", "role": "Bekannte Regression, kein neuer unabhängiger Holdout", "sha256": "be83ad489aa60c56c1f221c72d5913a9dbda6608908af5e976e5e339242dd249", "documents": 300, "gold_spans": 9551, "gold_characters": 130996, "text_characters": 2003833, "gold_free_documents": 0 }, "models": [ { "id": "skira41", "name": "Skira 4.1", "profile": "Viterbi + Grundkette + QID-Grenze, O-Bias 0", "label_stage": 4, "mask_stage": 4, "labels": { "ADRESSE": { "gold": 608, "tp": 600, "fp": 8, "fn": 8, "precision": 98.68421052631578, "recall": 98.68421052631578, "f1": 98.68421052631578, "coverage": 100.0, "covered": 608, "uncovered": 0, "open_characters": 0 }, "ALTER": { "gold": 339, "tp": 337, "fp": 3, "fn": 2, "precision": 99.11764705882354, "recall": 99.41002949852508, "f1": 99.26362297496317, "coverage": 99.41002949852508, "covered": 337, "uncovered": 2, "open_characters": 34 }, "BERUF": { "gold": 503, "tp": 495, "fp": 129, "fn": 8, "precision": 79.32692307692307, "recall": 98.40954274353876, "f1": 87.84383318544809, "coverage": 98.80715705765407, "covered": 497, "uncovered": 6, "open_characters": 89 }, "DATUM": { "gold": 891, "tp": 868, "fp": 16, "fn": 23, "precision": 98.19004524886877, "recall": 97.41863075196409, "f1": 97.80281690140843, "coverage": 97.8675645342312, "covered": 872, "uncovered": 19, "open_characters": 272 }, "ID": { "gold": 638, "tp": 627, "fp": 8, "fn": 11, "precision": 98.74015748031496, "recall": 98.27586206896551, "f1": 98.50746268656717, "coverage": 99.21630094043887, "covered": 633, "uncovered": 5, "open_characters": 37 }, "KONTAKT": { "gold": 1251, "tp": 1251, "fp": 26, "fn": 0, "precision": 97.96397807361002, "recall": 100.0, "f1": 98.97151898734177, "coverage": 100.0, "covered": 1251, "uncovered": 0, "open_characters": 0 }, "ORG": { "gold": 396, "tp": 361, "fp": 39, "fn": 35, "precision": 90.25, "recall": 91.16161616161617, "f1": 90.70351758793971, "coverage": 94.6969696969697, "covered": 375, "uncovered": 21, "open_characters": 336 }, "ORT": { "gold": 772, "tp": 762, "fp": 17, "fn": 10, "precision": 97.81771501925546, "recall": 98.70466321243524, "f1": 98.25918762088975, "coverage": 99.87046632124353, "covered": 771, "uncovered": 1, "open_characters": 6 }, "PERSON": { "gold": 2887, "tp": 2845, "fp": 83, "fn": 42, "precision": 97.16530054644808, "recall": 98.54520263249047, "f1": 97.85038693035253, "coverage": 99.75753377208174, "covered": 2880, "uncovered": 7, "open_characters": 16 }, "PLZ": { "gold": 563, "tp": 563, "fp": 0, "fn": 0, "precision": 100.0, "recall": 100.0, "f1": 100.0, "coverage": 100.0, "covered": 563, "uncovered": 0, "open_characters": 0 }, "QID": { "gold": 252, "tp": 221, "fp": 21, "fn": 31, "precision": 91.32231404958677, "recall": 87.6984126984127, "f1": 89.47368421052632, "coverage": 91.66666666666666, "covered": 231, "uncovered": 21, "open_characters": 332 }, "WEB": { "gold": 451, "tp": 438, "fp": 17, "fn": 13, "precision": 96.26373626373626, "recall": 97.11751662971176, "f1": 96.68874172185431, "coverage": 99.77827050997783, "covered": 450, "uncovered": 1, "open_characters": 11 } }, "tp": 9368, "fp": 367, "fn": 183, "micro_precision": 96.23009758602979, "micro_recall": 98.08397026489372, "micro_f1": 97.1481903971793, "macro_precision": 95.4035022786569, "macro_recall": 97.11880724366462, "macro_f1": 96.17074861113393, "documents": 241, "open_spans": 83, "open_characters": 1133, "extra_characters": 2998, "predicted_characters": 132861, "covered_spans": 9468, "macro_coverage": 98.42257991648239 }, { "id": "skira5", "name": "Skira 5", "profile": "Klassenpipeline mit Bias 4; Maskierung pii-union-0p5-v1", "label_stage": 10, "mask_stage": 12, "labels": { "ADRESSE": { "gold": 608, "tp": 590, "fp": 66, "fn": 18, "precision": 89.9390243902439, "recall": 97.03947368421052, "f1": 93.35443037974683, "coverage": 100.0, "covered": 608, "uncovered": 0, "open_characters": 0 }, "ALTER": { "gold": 339, "tp": 305, "fp": 56, "fn": 34, "precision": 84.48753462603878, "recall": 89.97050147492625, "f1": 87.14285714285714, "coverage": 100.0, "covered": 339, "uncovered": 0, "open_characters": 0 }, "BERUF": { "gold": 503, "tp": 390, "fp": 1138, "fn": 113, "precision": 25.523560209424083, "recall": 77.53479125248509, "f1": 38.404726735598224, "coverage": 100.0, "covered": 503, "uncovered": 0, "open_characters": 0 }, "DATUM": { "gold": 891, "tp": 816, "fp": 614, "fn": 75, "precision": 57.06293706293706, "recall": 91.58249158249158, "f1": 70.31451960361913, "coverage": 100.0, "covered": 891, "uncovered": 0, "open_characters": 0 }, "ID": { "gold": 638, "tp": 614, "fp": 361, "fn": 24, "precision": 62.97435897435897, "recall": 96.23824451410658, "f1": 76.13143211407316, "coverage": 100.0, "covered": 638, "uncovered": 0, "open_characters": 0 }, "KONTAKT": { "gold": 1251, "tp": 1209, "fp": 69, "fn": 42, "precision": 94.60093896713614, "recall": 96.64268585131894, "f1": 95.6109134045077, "coverage": 100.0, "covered": 1251, "uncovered": 0, "open_characters": 0 }, "ORG": { "gold": 396, "tp": 296, "fp": 708, "fn": 100, "precision": 29.48207171314741, "recall": 74.74747474747475, "f1": 42.285714285714285, "coverage": 99.74747474747475, "covered": 395, "uncovered": 1, "open_characters": 9 }, "ORT": { "gold": 772, "tp": 686, "fp": 110, "fn": 86, "precision": 86.18090452261306, "recall": 88.860103626943, "f1": 87.5, "coverage": 100.0, "covered": 772, "uncovered": 0, "open_characters": 0 }, "PERSON": { "gold": 2887, "tp": 2569, "fp": 532, "fn": 318, "precision": 82.84424379232506, "recall": 88.98510564599931, "f1": 85.80494321977288, "coverage": 100.0, "covered": 2887, "uncovered": 0, "open_characters": 0 }, "PLZ": { "gold": 563, "tp": 561, "fp": 0, "fn": 2, "precision": 100.0, "recall": 99.64476021314387, "f1": 99.8220640569395, "coverage": 100.0, "covered": 563, "uncovered": 0, "open_characters": 0 }, "QID": { "gold": 252, "tp": 154, "fp": 411, "fn": 98, "precision": 27.256637168141594, "recall": 61.111111111111114, "f1": 37.69889840881273, "coverage": 99.60317460317461, "covered": 251, "uncovered": 1, "open_characters": 33 }, "WEB": { "gold": 451, "tp": 428, "fp": 42, "fn": 23, "precision": 91.06382978723404, "recall": 94.90022172949003, "f1": 92.94245385450597, "coverage": 100.0, "covered": 451, "uncovered": 0, "open_characters": 0 } }, "tp": 8618, "fp": 4107, "fn": 933, "micro_precision": 67.72495088408644, "micro_recall": 90.23138938331064, "micro_f1": 77.37475309750404, "macro_precision": 69.28467010113334, "macro_recall": 88.10474711947508, "macro_f1": 75.58441276717896, "documents": 298, "open_spans": 2, "open_characters": 42, "extra_characters": 58355, "predicted_characters": 189309, "covered_spans": 9549, "macro_coverage": 99.94588744588744 } ], "skira5_parameters": 140651161, "history": { "valravn": { "metric": "Historischer Span-F1", "reference": "Eigener fixer 14-Dokumente-Testsatz", "values": [ [ "v1", 87.0 ], [ "v2", 87.7 ], [ "v3", 92.1 ], [ "v4", 93.2 ], [ "v5", 91.3 ], [ "v12", 88.6 ], [ "v14", 88.2 ] ], "qid": [ [ "v12", 21 ], [ "v14", 42 ] ], "missing": [ "v6", "v7", "v8", "v9", "v10", "v11", "v13" ] }, "skira123": { "metric": "Vollständig übersehene Goldstellen", "reference": "bench1878 norm22", "documents": 1878, "gold_spans": 62023, "profile": "O-Bias 3, Kontext 2048", "values": [ [ "Skira 1", 531 ], [ "Skira 2", 270 ], [ "Skira 3", 233 ] ] }, "skira34": { "checkpoint_note": "Früher Skira-4-Checkpoint pool23, nicht der spätere 4k-Stand", "reference": "bench1878 norm24", "documents": 1878, "gold_spans": 62023, "profile": "O-Bias 3, Kontext 2048", "missed": [ [ "Skira 3", 233 ], [ "Skira 4", 222 ] ], "open_characters": [ [ "Skira 3", 6909 ], [ "Skira 4", 6306 ] ] } }, "sources": [ { "name": "Skira 4.1, comparison_v1.json", "sha256": "01b8e236b2b1b6a13a2e8003d284618f0f817241f4fa20c7d5bc74a4e0b121d5" }, { "name": "Skira 5, manifest.json", "sha256": "3998d7ea1aa4b5bdeedd5cbed644265fa828c1e7d837df3ba0b03633a15d295b" }, { "name": "Skira 6, test300_results.json", "sha256": "5cb2e39567275f315d3e381c91198e4d3faec968e4347269508ac12b9ebfcd0b" }, { "name": "Skira 6, validation_results.json", "sha256": "a415e0b0e0b33a34967c6e81520132ea89ac59716369c9bd562427375ba58f0a" }, { "name": "Skira 6, packaged_runtime_verification_v1.json", "sha256": "5537ead5c2c2eeba73d27e1cb119faa8b7a753ecdad3d9a33b4917f02e6aa343" }, { "name": "Skira 6 Labelkopf, manifest.json", "sha256": "0d30de2196427e09275390cd1d99925c6d74456e57f9d1f74e7a740ff0395ab6" }, { "name": "Skira 6 Labelkopf, evidence/test/results.json", "sha256": "c2f6d288c796d2a34f81d3bf0db08ac6f3c6fc56eac7c37181449ab9a779ee56" }, { "name": "Skira 6 Labelkopf, evidence/valid/results.json", "sha256": "03c228a8d2a4da92162ceefe638f30549bd4c4c16330672bd75c0c5d1a809e3d" } ], "decoder_preview": { "name": "Skira 6", "status": "Preview", "updated": "2026-09-07", "profile": "skira4_1_decoder", "labels": { "ADRESSE": { "gold": 608, "tp": 594, "fp": 30, "fn": 14, "precision": 95.1923076923077, "recall": 97.69736842105263, "f1": 96.42857142857143, "coverage": 100.0 }, "ALTER": { "gold": 339, "tp": 337, "fp": 21, "fn": 2, "precision": 94.1340782122905, "recall": 99.41002949852508, "f1": 96.70014347202296, "coverage": 100.0 }, "BERUF": { "gold": 503, "tp": 427, "fp": 586, "fn": 76, "precision": 42.15202369200395, "recall": 84.89065606361828, "f1": 56.332453825857506, "coverage": 100.0 }, "DATUM": { "gold": 891, "tp": 877, "fp": 234, "fn": 14, "precision": 78.93789378937895, "recall": 98.4287317620651, "f1": 87.61238761238762, "coverage": 100.0 }, "ID": { "gold": 638, "tp": 622, "fp": 110, "fn": 16, "precision": 84.97267759562843, "recall": 97.4921630094044, "f1": 90.8029197080292, "coverage": 100.0 }, "KONTAKT": { "gold": 1251, "tp": 1240, "fp": 52, "fn": 11, "precision": 95.97523219814241, "recall": 99.1207034372502, "f1": 97.52261108926466, "coverage": 100.0 }, "ORG": { "gold": 396, "tp": 357, "fp": 213, "fn": 39, "precision": 62.63157894736842, "recall": 90.15151515151516, "f1": 73.91304347826087, "coverage": 99.74747474747475 }, "ORT": { "gold": 772, "tp": 745, "fp": 47, "fn": 27, "precision": 94.06565656565657, "recall": 96.50259067357513, "f1": 95.26854219948848, "coverage": 100.0 }, "PERSON": { "gold": 2887, "tp": 2733, "fp": 281, "fn": 154, "precision": 90.67684140676842, "recall": 94.6657429857984, "f1": 92.62836807320794, "coverage": 100.0 }, "PLZ": { "gold": 563, "tp": 561, "fp": 0, "fn": 2, "precision": 100.0, "recall": 99.64476021314387, "f1": 99.8220640569395, "coverage": 100.0 }, "QID": { "gold": 252, "tp": 211, "fp": 149, "fn": 41, "precision": 58.611111111111114, "recall": 83.73015873015873, "f1": 68.95424836601308, "coverage": 99.60317460317461 }, "WEB": { "gold": 451, "tp": 433, "fp": 21, "fn": 18, "precision": 95.37444933920705, "recall": 96.0088691796009, "f1": 95.69060773480662, "coverage": 100.0 } }, "tp": 9137, "fp": 1744, "fn": 414, "micro_precision": 83.97206139141623, "micro_recall": 95.66537535336614, "micro_f1": 89.43813625685199, "macro_precision": 82.72698754582196, "macro_recall": 94.81194076047565, "macro_f1": 87.63966342040415, "macro_coverage": 99.94588744588745, "micro_coverage": 99.97905978431578, "validation_documents": 2799, "validation_macro_before": 72.78666636909283, "validation_macro_after": 83.76147654330751, "test_documents": 300, "mask_mismatches": 0, "validation_mask_mismatches": 0, "documents": 298, "open_characters": 42, "extra_characters": 58355, "packaged_identical_outputs": 300 }, "release": { "name": "Skira 6", "status": "Veröffentlicht", "release_id": "skira6-labelkopf-2026-09-07", "updated": "2026-09-07", "profile": "separate_label_head_bias0", "labels": { "ADRESSE": { "gold": 608, "tp": 601, "fp": 12, "fn": 7, "precision": 98.04241435562805, "recall": 98.84868421052632, "f1": 98.44389844389845, "coverage": 100.0, "covered": 608, "uncovered": 0, "open_characters": 0 }, "ALTER": { "gold": 339, "tp": 338, "fp": 10, "fn": 1, "precision": 97.12643678160919, "recall": 99.70501474926253, "f1": 98.39883551673945, "coverage": 100.0, "covered": 339, "uncovered": 0, "open_characters": 0 }, "BERUF": { "gold": 503, "tp": 486, "fp": 176, "fn": 17, "precision": 73.41389728096676, "recall": 96.62027833001989, "f1": 83.43347639484978, "coverage": 100.0, "covered": 503, "uncovered": 0, "open_characters": 0 }, "DATUM": { "gold": 891, "tp": 864, "fp": 42, "fn": 27, "precision": 95.36423841059603, "recall": 96.96969696969697, "f1": 96.16026711185309, "coverage": 100.0, "covered": 891, "uncovered": 0, "open_characters": 0 }, "ID": { "gold": 638, "tp": 629, "fp": 29, "fn": 9, "precision": 95.59270516717325, "recall": 98.58934169278997, "f1": 97.06790123456788, "coverage": 100.0, "covered": 638, "uncovered": 0, "open_characters": 0 }, "KONTAKT": { "gold": 1251, "tp": 1248, "fp": 31, "fn": 3, "precision": 97.57623143080532, "recall": 99.76019184652279, "f1": 98.65612648221345, "coverage": 100.0, "covered": 1251, "uncovered": 0, "open_characters": 0 }, "ORG": { "gold": 396, "tp": 362, "fp": 58, "fn": 34, "precision": 86.19047619047619, "recall": 91.41414141414141, "f1": 88.72549019607844, "coverage": 99.74747474747475, "covered": 395, "uncovered": 1, "open_characters": 9 }, "ORT": { "gold": 772, "tp": 757, "fp": 22, "fn": 15, "precision": 97.17586649550705, "recall": 98.05699481865285, "f1": 97.61444229529336, "coverage": 100.0, "covered": 772, "uncovered": 0, "open_characters": 0 }, "PERSON": { "gold": 2887, "tp": 2825, "fp": 138, "fn": 62, "precision": 95.34255821802228, "recall": 97.85244198129546, "f1": 96.5811965811966, "coverage": 100.0, "covered": 2887, "uncovered": 0, "open_characters": 0 }, "PLZ": { "gold": 563, "tp": 561, "fp": 0, "fn": 2, "precision": 100.0, "recall": 99.64476021314387, "f1": 99.8220640569395, "coverage": 100.0, "covered": 563, "uncovered": 0, "open_characters": 0 }, "QID": { "gold": 252, "tp": 214, "fp": 39, "fn": 38, "precision": 84.58498023715416, "recall": 84.92063492063492, "f1": 84.75247524752476, "coverage": 99.60317460317461, "covered": 251, "uncovered": 1, "open_characters": 33 }, "WEB": { "gold": 451, "tp": 435, "fp": 16, "fn": 16, "precision": 96.45232815964523, "recall": 96.45232815964523, "f1": 96.45232815964523, "coverage": 100.0, "covered": 451, "uncovered": 0, "open_characters": 0 } }, "tp": 9320, "fp": 573, "fn": 231, "micro_precision": 94.20802587688264, "micro_recall": 97.58140508847241, "micro_f1": 95.86504834396214, "macro_precision": 93.07184439396529, "macro_recall": 96.56954244219435, "macro_f1": 94.67570847673333, "macro_coverage": 99.94588744588745, "micro_coverage": 99.97905978431578, "validation_documents": 2799, "validation_macro_before": 83.76147654330751, "validation_macro_after": 92.48391774541122, "test_documents": 300, "mask_mismatches": 0, "validation_mask_mismatches": 0, "documents": 298, "open_characters": 42, "extra_characters": 58355, "counts": { "documents": 300, "documentsWithGold": 300, "goldFreeDocuments": 0, "fullyCoveredDocuments": 298, "uncoveredDocuments": 2, "totalEntities": 9551, "coveredEntities": 9549, "partiallyCoveredEntities": 0, "untouchedEntities": 2, "uncoveredEntities": 2, "textCharacters": 2003833, "goldCharacters": 130996, "coveredGoldCharacters": 130954, "uncoveredGoldCharacters": 42, "predictedCharacters": 189309, "extraCharacters": 58355 }, "validation_counts": { "documents": 2799, "documentsWithGold": 2799, "goldFreeDocuments": 0, "fullyCoveredDocuments": 2622, "uncoveredDocuments": 177, "totalEntities": 74064, "coveredEntities": 73854, "partiallyCoveredEntities": 177, "untouchedEntities": 33, "uncoveredEntities": 210, "textCharacters": 10560643, "goldCharacters": 1022641, "coveredGoldCharacters": 1019547, "uncoveredGoldCharacters": 3094, "predictedCharacters": 1697097, "extraCharacters": 677550 }, "head_parameters": 157465, "train_documents": 4000, "selected_epoch": 3, "training_epochs": 3, "encoder_calls": 390, "validation_encoder_calls": 3073 } }