{ "code": { "dirty": true, "git_revision": "a8b33432e3e0e292c5381b480ab9e6a74bd5a316" }, "corpus": "nemotron-pii", "corpus_diagnostics": { "dropped_labels": { "account_number": 16693, "age": 7167, "bank_routing_number": 8354, "biometric_identifier": 11379, "blood_type": 5539, "certificate_license_number": 3002, "city": 18347, "coordinate": 7659, "country": 23475, "county": 8465, "customer_id": 20502, "cvv": 4843, "date": 73867, "date_time": 13221, "device_identifier": 2507, "education_level": 9378, "employee_id": 8878, "employment_status": 11018, "gender": 7645, "health_plan_beneficiary_number": 10558, "http_cookie": 4892, "language": 8651, "license_plate": 4078, "mac_address": 4250, "medical_record_number": 11098, "occupation": 36888, "pin": 6456, "political_view": 6402, "postcode": 6279, "race_ethnicity": 9248, "religious_belief": 6257, "sexuality": 3312, "state": 18318, "swift_bic": 5559, "time": 24506, "unique_id": 1777, "url": 37847, "vehicle_identifier": 4561 }, "gold_bad_offset": 0, "gold_in_scope": 304548, "gold_out_of_scope": 72916, "gold_text_mismatch": 934, "rows": 100000, "rows_with_overlapping_gold": 1, "unknown_labels": {} }, "corpus_diagnostics_scope": "whole file, not the limited run", "corpus_license": "CC-BY-4.0", "corpus_revision": "b70ffaf5ff39e079776134c5bf4381f00a9fd1ed", "corpus_sha256": "1a4b0512ecb5370f0992d29d0f9c07351e6de13f0d7ea33bb18cecb984780247", "document_limit": 25, "documents_scored": 25, "environment": { "detectors": [ "RuleDetector", "HfTokenDetector" ], "devices": [ "cpu", "cuda" ], "executable": "/usr/local/bin/python", "machine": "x86_64", "packages": { "edgepii": "0.3.1", "faker": "40.44.0", "gliner2": "2.1.0", "huggingface-hub": "2.4.0", "peft": "0.21.2", "phonenumbers": "9.0.42", "pyarrow": "20.0.0", "pyyaml": "6.0.3", "tokenizers": "0.23.3", "torch": "2.8.0", "transformers": "5.22.1" }, "platform": "Linux-4.19.0-gvisor-x86_64-with-glibc2.31", "python": "3.11.0", "source_sha256": "53696ffc8e1fe460882917cbd2fd8c7c9dcbab1a978296377940e3d3fa8a8d09" }, "evaluation": { "annotation_policy": "source corpus policy", "fold": "confusables", "materialization_protocol": "source-v1", "min_score": null, "selection": "prefix_per_slice", "selection_sha256": "56a52feb9d8b4fa14087c5e4be7e6bd5eca609f8fe88ea17678855114fd38b3d", "spy_seed": null, "types": [ "ADDRESS", "EMAIL", "GOV_ID", "IP_ADDRESS", "PAYMENT", "PERSON", "PHONE", "SECRET", "USERNAME" ], "unique_documents": 75 }, "first_principles_verification": { "cases": 5, "contract_passed": true, "model_only_smoke_spans": 2, "scope": "inference and source/masking invariants; not an accuracy guarantee", "valid_original_source_spans": 8 }, "gates": { "failures": [], "passed": true, "policy": { "max_dropped": 0, "max_leaks": null, "min_recall": 0.0 } }, "latency_ms": { "calls": 75, "max": 41.52109199999998, "mean": 23.105201786666687, "p50": 22.511338000001047, "p95": 29.513382000001087, "scope": "locate only; excludes model loading and scoring" }, "model_only_ablation": { "corpus": "nemotron-pii", "corpus_diagnostics": { "dropped_labels": { "account_number": 16693, "age": 7167, "bank_routing_number": 8354, "biometric_identifier": 11379, "blood_type": 5539, "certificate_license_number": 3002, "city": 18347, "coordinate": 7659, "country": 23475, "county": 8465, "customer_id": 20502, "cvv": 4843, "date": 73867, "date_time": 13221, "device_identifier": 2507, "education_level": 9378, "employee_id": 8878, "employment_status": 11018, "gender": 7645, "health_plan_beneficiary_number": 10558, "http_cookie": 4892, "language": 8651, "license_plate": 4078, "mac_address": 4250, "medical_record_number": 11098, "occupation": 36888, "pin": 6456, "political_view": 6402, "postcode": 6279, "race_ethnicity": 9248, "religious_belief": 6257, "sexuality": 3312, "state": 18318, "swift_bic": 5559, "time": 24506, "unique_id": 1777, "url": 37847, "vehicle_identifier": 4561 }, "gold_bad_offset": 0, "gold_in_scope": 304548, "gold_out_of_scope": 72916, "gold_text_mismatch": 934, "rows": 100000, "rows_with_overlapping_gold": 1, "unknown_labels": {} }, "corpus_diagnostics_scope": "whole file, not the limited run", "corpus_license": "CC-BY-4.0", "corpus_revision": "b70ffaf5ff39e079776134c5bf4381f00a9fd1ed", "corpus_sha256": "1a4b0512ecb5370f0992d29d0f9c07351e6de13f0d7ea33bb18cecb984780247", "document_limit": 25, "documents_scored": 25, "environment": { "detectors": [ "HfTokenDetector" ], "devices": [ "cuda" ], "executable": "/usr/local/bin/python", "machine": "x86_64", "packages": { "edgepii": "0.3.1", "faker": "40.44.0", "gliner2": "2.1.0", "huggingface-hub": "2.4.0", "peft": "0.21.2", "phonenumbers": "9.0.42", "pyarrow": "20.0.0", "pyyaml": "6.0.3", "tokenizers": "0.23.3", "torch": "2.8.0", "transformers": "5.22.1" }, "platform": "Linux-4.19.0-gvisor-x86_64-with-glibc2.31", "python": "3.11.0", "source_sha256": "53696ffc8e1fe460882917cbd2fd8c7c9dcbab1a978296377940e3d3fa8a8d09" }, "evaluation": { "annotation_policy": "source corpus policy", "fold": "confusables", "materialization_protocol": "source-v1", "min_score": null, "selection": "prefix_per_slice", "selection_sha256": "56a52feb9d8b4fa14087c5e4be7e6bd5eca609f8fe88ea17678855114fd38b3d", "spy_seed": null, "types": [ "ADDRESS", "EMAIL", "GOV_ID", "IP_ADDRESS", "PAYMENT", "PERSON", "PHONE", "SECRET", "USERNAME" ], "unique_documents": 75 }, "latency_ms": { "calls": 75, "max": 33.32261900000333, "mean": 19.736251359999805, "p50": 19.05609299999611, "p95": 26.399070999993057, "scope": "locate only; excludes model loading and scoring" }, "preset": "custom", "preset_revision": "custom", "seconds": 1.504, "slices": { "all": { "containment": { "character_recall": 0.9408866995073891, "covered_characters": 573, "fully_contained_gold_spans": 51, "fully_contained_repeated_identifiers": 4, "gold_characters": 609, "high_severity_gold_spans": 12, "masked_characters": 839, "masking_precision": 0.6829558998808105, "repeated_identifiers": 5, "scope": "declared gold policy; zero gold does not establish PII-free text", "span_recall": 0.9622641509433962, "zero_gold_document_mask_rate": 1.0, "zero_gold_documents": 3, "zero_gold_masked_documents": 3 }, "documents": 25, "dropped_offsets": 0, "extra": 15, "f1": 0.7478260869565218, "f2": 0.7846715328467153, "fn": 10, "fp": 19, "gold_spans": 53, "high_severity_leaks": 2, "miss": 2, "per_type": { "ADDRESS": { "f1": 0.9411764705882353, "f2": 0.9756097560975611, "fn": 0, "fp": 1, "precision": 0.8888888888888888, "recall": 1.0, "tp": 8 }, "EMAIL": { "f1": 1.0, "f2": 1.0, "fn": 0, "fp": 0, "precision": 1.0, "recall": 1.0, "tp": 4 }, "GOV_ID": { "f1": 1.0, "f2": 1.0, "fn": 0, "fp": 0, "precision": 1.0, "recall": 1.0, "tp": 6 }, "PAYMENT": { "f1": 0.3076923076923077, "f2": 0.5263157894736843, "fn": 0, "fp": 9, "precision": 0.18181818181818182, "recall": 1.0, "tp": 2 }, "PERSON": { "f1": 0.7000000000000001, "f2": 0.660377358490566, "fn": 8, "fp": 4, "precision": 0.7777777777777778, "recall": 0.6363636363636364, "tp": 14 }, "PHONE": { "f1": 1.0, "f2": 1.0, "fn": 0, "fp": 0, "precision": 1.0, "recall": 1.0, "tp": 3 }, "SECRET": { "f1": 0.36363636363636365, "f2": 0.4347826086956521, "fn": 2, "fp": 5, "precision": 0.2857142857142857, "recall": 0.5, "tp": 2 }, "USERNAME": { "f1": 1.0, "f2": 1.0, "fn": 0, "fp": 0, "precision": 1.0, "recall": 1.0, "tp": 4 } }, "pre_policy_candidates": { "documents": 25, "fully_contained_gold_spans": 51, "gold_spans": 53, "high_severity_leaks": 2, "scope": "mapped requested-type candidates before overlap and score policy; null when unavailable", "span_recall": 0.9622641509433962 }, "precision": 0.6935483870967742, "predicted_spans": 62, "recall": 0.8113207547169812, "shift": 8, "slice": "all", "tp": 43, "unmapped_predictions": 54, "wrong_type": 0 }, "intl": { "containment": { "character_recall": 1.0, "covered_characters": 649, "fully_contained_gold_spans": 55, "fully_contained_repeated_identifiers": 8, "gold_characters": 649, "high_severity_gold_spans": 6, "masked_characters": 748, "masking_precision": 0.8676470588235294, "repeated_identifiers": 8, "scope": "declared gold policy; zero gold does not establish PII-free text", "span_recall": 1.0, "zero_gold_document_mask_rate": 0.6666666666666666, "zero_gold_documents": 3, "zero_gold_masked_documents": 2 }, "documents": 25, "dropped_offsets": 0, "extra": 11, "f1": 0.6306306306306306, "f2": 0.6340579710144928, "fn": 20, "fp": 21, "gold_spans": 55, "high_severity_leaks": 0, "miss": 0, "per_type": { "ADDRESS": { "f1": 0.923076923076923, "f2": 0.9677419354838709, "fn": 0, "fp": 1, "precision": 0.8571428571428571, "recall": 1.0, "tp": 6 }, "EMAIL": { "f1": 1.0, "f2": 1.0, "fn": 0, "fp": 0, "precision": 1.0, "recall": 1.0, "tp": 9 }, "GOV_ID": { "f1": 1.0, "f2": 1.0, "fn": 0, "fp": 0, "precision": 1.0, "recall": 1.0, "tp": 4 }, "PAYMENT": { "f1": 0.2222222222222222, "f2": 0.4166666666666667, "fn": 0, "fp": 7, "precision": 0.125, "recall": 1.0, "tp": 1 }, "PERSON": { "f1": 0.3111111111111111, "f2": 0.27777777777777773, "fn": 20, "fp": 11, "precision": 0.3888888888888889, "recall": 0.25925925925925924, "tp": 7 }, "PHONE": { "f1": 1.0, "f2": 1.0, "fn": 0, "fp": 0, "precision": 1.0, "recall": 1.0, "tp": 4 }, "SECRET": { "f1": 0.5, "f2": 0.7142857142857143, "fn": 0, "fp": 2, "precision": 0.3333333333333333, "recall": 1.0, "tp": 1 }, "USERNAME": { "f1": 1.0, "f2": 1.0, "fn": 0, "fp": 0, "precision": 1.0, "recall": 1.0, "tp": 3 } }, "pre_policy_candidates": { "documents": 25, "fully_contained_gold_spans": 55, "gold_spans": 55, "high_severity_leaks": 0, "scope": "mapped requested-type candidates before overlap and score policy; null when unavailable", "span_recall": 1.0 }, "precision": 0.625, "predicted_spans": 56, "recall": 0.6363636363636364, "shift": 20, "slice": "intl", "tp": 35, "unmapped_predictions": 35, "wrong_type": 0 }, "structured": { "containment": { "character_recall": 0.9438990182328191, "covered_characters": 673, "fully_contained_gold_spans": 68, "fully_contained_repeated_identifiers": 13, "gold_characters": 713, "high_severity_gold_spans": 8, "masked_characters": 915, "masking_precision": 0.7355191256830601, "repeated_identifiers": 14, "scope": "declared gold policy; zero gold does not establish PII-free text", "span_recall": 0.9577464788732394, "zero_gold_document_mask_rate": 1.0, "zero_gold_documents": 2, "zero_gold_masked_documents": 2 }, "documents": 25, "dropped_offsets": 0, "extra": 14, "f1": 0.7172413793103449, "f2": 0.7262569832402235, "fn": 19, "fp": 22, "gold_spans": 71, "high_severity_leaks": 2, "miss": 3, "per_type": { "ADDRESS": { "f1": 1.0, "f2": 1.0, "fn": 0, "fp": 0, "precision": 1.0, "recall": 1.0, "tp": 5 }, "EMAIL": { "f1": 1.0, "f2": 1.0, "fn": 0, "fp": 0, "precision": 1.0, "recall": 1.0, "tp": 5 }, "GOV_ID": { "f1": 1.0, "f2": 1.0, "fn": 0, "fp": 0, "precision": 1.0, "recall": 1.0, "tp": 3 }, "IP_ADDRESS": { "f1": 1.0, "f2": 1.0, "fn": 0, "fp": 0, "precision": 1.0, "recall": 1.0, "tp": 2 }, "PAYMENT": { "f1": 0.33333333333333337, "f2": 0.5555555555555556, "fn": 0, "fp": 8, "precision": 0.2, "recall": 1.0, "tp": 2 }, "PERSON": { "f1": 0.6987951807228915, "f2": 0.6561085972850679, "fn": 17, "fp": 8, "precision": 0.7837837837837838, "recall": 0.6304347826086957, "tp": 29 }, "PHONE": { "f1": 1.0, "f2": 1.0, "fn": 0, "fp": 0, "precision": 1.0, "recall": 1.0, "tp": 4 }, "SECRET": { "f1": 0.2, "f2": 0.2631578947368421, "fn": 2, "fp": 6, "precision": 0.14285714285714285, "recall": 0.3333333333333333, "tp": 1 }, "USERNAME": { "f1": 1.0, "f2": 1.0, "fn": 0, "fp": 0, "precision": 1.0, "recall": 1.0, "tp": 1 } }, "pre_policy_candidates": { "documents": 25, "fully_contained_gold_spans": 68, "gold_spans": 71, "high_severity_leaks": 2, "scope": "mapped requested-type candidates before overlap and score policy; null when unavailable", "span_recall": 0.9577464788732394 }, "precision": 0.7027027027027027, "predicted_spans": 74, "recall": 0.7323943661971831, "shift": 16, "slice": "structured", "tp": 52, "unmapped_predictions": 74, "wrong_type": 0 }, "unstructured": { "containment": { "character_recall": 0.9618055555555556, "covered_characters": 554, "fully_contained_gold_spans": 49, "fully_contained_repeated_identifiers": 3, "gold_characters": 576, "high_severity_gold_spans": 10, "masked_characters": 736, "masking_precision": 0.7527173913043478, "repeated_identifiers": 3, "scope": "declared gold policy; zero gold does not establish PII-free text", "span_recall": 0.98, "zero_gold_document_mask_rate": 1.0, "zero_gold_documents": 2, "zero_gold_masked_documents": 2 }, "documents": 25, "dropped_offsets": 0, "extra": 14, "f1": 0.6915887850467289, "f2": 0.7198443579766536, "fn": 13, "fp": 20, "gold_spans": 50, "high_severity_leaks": 0, "miss": 1, "per_type": { "ADDRESS": { "f1": 0.8571428571428571, "f2": 0.9375, "fn": 0, "fp": 2, "precision": 0.75, "recall": 1.0, "tp": 6 }, "EMAIL": { "f1": 1.0, "f2": 1.0, "fn": 0, "fp": 0, "precision": 1.0, "recall": 1.0, "tp": 5 }, "GOV_ID": { "f1": 1.0, "f2": 1.0, "fn": 0, "fp": 0, "precision": 1.0, "recall": 1.0, "tp": 5 }, "PAYMENT": { "f1": 0.47058823529411764, "f2": 0.6896551724137931, "fn": 0, "fp": 9, "precision": 0.3076923076923077, "recall": 1.0, "tp": 4 }, "PERSON": { "f1": 0.4864864864864865, "f2": 0.4368932038834952, "fn": 13, "fp": 6, "precision": 0.6, "recall": 0.4090909090909091, "tp": 9 }, "PHONE": { "f1": 1.0, "f2": 1.0, "fn": 0, "fp": 0, "precision": 1.0, "recall": 1.0, "tp": 4 }, "SECRET": { "f1": 0.4, "f2": 0.625, "fn": 0, "fp": 3, "precision": 0.25, "recall": 1.0, "tp": 1 }, "USERNAME": { "f1": 1.0, "f2": 1.0, "fn": 0, "fp": 0, "precision": 1.0, "recall": 1.0, "tp": 3 } }, "pre_policy_candidates": { "documents": 25, "fully_contained_gold_spans": 49, "gold_spans": 50, "high_severity_leaks": 0, "scope": "mapped requested-type candidates before overlap and score policy; null when unavailable", "span_recall": 0.98 }, "precision": 0.6491228070175439, "predicted_spans": 57, "recall": 0.74, "shift": 12, "slice": "unstructured", "tp": 37, "unmapped_predictions": 49, "wrong_type": 0 }, "us": { "containment": { "character_recall": 0.9408866995073891, "covered_characters": 573, "fully_contained_gold_spans": 51, "fully_contained_repeated_identifiers": 4, "gold_characters": 609, "high_severity_gold_spans": 12, "masked_characters": 839, "masking_precision": 0.6829558998808105, "repeated_identifiers": 5, "scope": "declared gold policy; zero gold does not establish PII-free text", "span_recall": 0.9622641509433962, "zero_gold_document_mask_rate": 1.0, "zero_gold_documents": 3, "zero_gold_masked_documents": 3 }, "documents": 25, "dropped_offsets": 0, "extra": 15, "f1": 0.7478260869565218, "f2": 0.7846715328467153, "fn": 10, "fp": 19, "gold_spans": 53, "high_severity_leaks": 2, "miss": 2, "per_type": { "ADDRESS": { "f1": 0.9411764705882353, "f2": 0.9756097560975611, "fn": 0, "fp": 1, "precision": 0.8888888888888888, "recall": 1.0, "tp": 8 }, "EMAIL": { "f1": 1.0, "f2": 1.0, "fn": 0, "fp": 0, "precision": 1.0, "recall": 1.0, "tp": 4 }, "GOV_ID": { "f1": 1.0, "f2": 1.0, "fn": 0, "fp": 0, "precision": 1.0, "recall": 1.0, "tp": 6 }, "PAYMENT": { "f1": 0.3076923076923077, "f2": 0.5263157894736843, "fn": 0, "fp": 9, "precision": 0.18181818181818182, "recall": 1.0, "tp": 2 }, "PERSON": { "f1": 0.7000000000000001, "f2": 0.660377358490566, "fn": 8, "fp": 4, "precision": 0.7777777777777778, "recall": 0.6363636363636364, "tp": 14 }, "PHONE": { "f1": 1.0, "f2": 1.0, "fn": 0, "fp": 0, "precision": 1.0, "recall": 1.0, "tp": 3 }, "SECRET": { "f1": 0.36363636363636365, "f2": 0.4347826086956521, "fn": 2, "fp": 5, "precision": 0.2857142857142857, "recall": 0.5, "tp": 2 }, "USERNAME": { "f1": 1.0, "f2": 1.0, "fn": 0, "fp": 0, "precision": 1.0, "recall": 1.0, "tp": 4 } }, "pre_policy_candidates": { "documents": 25, "fully_contained_gold_spans": 51, "gold_spans": 53, "high_severity_leaks": 2, "scope": "mapped requested-type candidates before overlap and score policy; null when unavailable", "span_recall": 0.9622641509433962 }, "precision": 0.6935483870967742, "predicted_spans": 62, "recall": 0.8113207547169812, "shift": 8, "slice": "us", "tp": 43, "unmapped_predictions": 54, "wrong_type": 0 } }, "taxonomy_version": "2026.11" }, "preset": "multi-encoder", "preset_revision": "a20835db9e671267b21a2951f5bf2c76e4901708", "runtime": { "deployment_source_sha256": "2d775313348a4b808205d2f4ff855c08edcd33190ee807973410ded392e56a7e", "device": "cuda", "gpu": "Tesla T4", "installed_packages": { "Faker": "40.44.0", "Jinja2": "3.1.6", "MarkupSafe": "3.0.4", "PyYAML": "6.0.3", "Pygments": "2.17.2", "accelerate": "1.15.0", "aiohttp": "3.8.3", "aiosignal": "1.3.1", "aiostream": "0.4.4", "annotated-doc": "0.0.5", "annotated-types": "0.8.0", "anyio": "4.16.0", "asgiref": "3.5.2", "asttokens": "2.4.1", "async-timeout": "4.0.3", "attrs": "23.2.0", "certifi": "2024.2.2", "charset-normalizer": "2.1.1", "click": "8.5.0", "cloudpickle": "2.2.0", "commonmark": "0.9.1", "decorator": "5.1.1", "edgepii": "0.3.1", "executing": "2.0.1", "fastapi": "0.162.0", "fastprogress": "1.0.0", "filelock": "4.2.9", "frozenlist": "1.4.1", "fsspec": "2026.9.0", "gliner2": "2.1.0", "grpclib": "0.4.3", "h11": "0.16.0", "h2": "4.1.0", "hf-xet": "1.8.0", "hpack": "4.0.0", "httpcore2": "2.16.0", "httpx2": "2.16.0", "huggingface_hub": "2.4.0", "hyperframe": "6.0.1", "idna": "3.20", "importlib-metadata": "4.8.1", "ipython": "8.23.0", "jedi": "0.19.1", "matplotlib-inline": "0.1.6", "mpmath": "1.3.0", "multidict": "6.0.5", "networkx": "3.6.1", "numpy": "1.26.4", "nvidia-cublas-cu12": "12.8.4.1", "nvidia-cuda-cupti-cu12": "12.8.90", "nvidia-cuda-nvrtc-cu12": "12.8.93", "nvidia-cuda-runtime-cu12": "12.8.90", "nvidia-cudnn-cu12": "9.10.2.21", "nvidia-cufft-cu12": "11.3.3.83", "nvidia-cufile-cu12": "1.13.1.3", "nvidia-curand-cu12": "10.3.9.90", "nvidia-cusolver-cu12": "11.7.3.90", "nvidia-cusparse-cu12": "12.5.8.93", "nvidia-cusparselt-cu12": "0.7.1", "nvidia-nccl-cu12": "2.27.3", "nvidia-nvjitlink-cu12": "12.8.93", "nvidia-nvtx-cu12": "12.8.90", "opentelemetry-api": "1.45.2", "opf": "0.1.0", "packaging": "26.3", "parso": "0.8.3", "peft": "0.21.2", "pexpect": "4.9.0", "phonenumbers": "9.0.42", "pip": "24.0", "prompt-toolkit": "3.0.43", "protobuf": "5.26.1", "psutil": "7.2.2", "ptyprocess": "0.7.0", "pure-eval": "0.2.2", "pyarrow": "20.0.0", "pydantic": "2.14.2", "pydantic_core": "2.50.2", "python-multipart": "0.0.9", "regex": "2026.9.29", "requests": "2.34.2", "rich": "12.3.0", "safetensors": "0.8.0", "sentencepiece": "0.2.1", "setuptools": "65.5.0", "six": "1.16.0", "sniffio": "1.3.1", "stack-data": "0.6.3", "starlette": "1.10.0", "sympy": "1.14.0", "tblib": "1.7.0", "tiktoken": "0.14.0", "tokenizers": "0.23.3", "toml": "0.10.2", "torch": "2.8.0", "tqdm": "4.70.1", "traitlets": "5.14.2", "transformers": "5.22.1", "triton": "3.4.0", "truststore": "0.10.5", "typeguard": "4.2.1", "typer": "0.6.1", "types-certifi": "2021.10.8.3", "types-toml": "0.10.4", "typing-inspection": "0.4.5", "typing_extensions": "4.16.0", "urllib3": "2.8.0", "wcwidth": "0.2.13", "wheel": "0.38.4", "yarl": "1.9.4", "zipp": "3.18.1" }, "model_load_scope": "download, initialization, and one warmup; excludes container startup", "model_load_seconds": 6.235802834, "provider": "modal", "warmup": "one synthetic document before timing" }, "seconds": 1.759, "slices": { "all": { "containment": { "character_recall": 0.9408866995073891, "covered_characters": 573, "fully_contained_gold_spans": 51, "fully_contained_repeated_identifiers": 4, "gold_characters": 609, "high_severity_gold_spans": 12, "masked_characters": 951, "masking_precision": 0.6025236593059937, "repeated_identifiers": 5, "scope": "declared gold policy; zero gold does not establish PII-free text", "span_recall": 0.9622641509433962, "zero_gold_document_mask_rate": 1.0, "zero_gold_documents": 3, "zero_gold_masked_documents": 3 }, "documents": 25, "dropped_offsets": 0, "extra": 27, "f1": 0.6771653543307086, "f2": 0.7517482517482518, "fn": 10, "fp": 31, "gold_spans": 53, "high_severity_leaks": 2, "miss": 2, "per_type": { "ADDRESS": { "f1": 0.9411764705882353, "f2": 0.9756097560975611, "fn": 0, "fp": 1, "precision": 0.8888888888888888, "recall": 1.0, "tp": 8 }, "EMAIL": { "f1": 1.0, "f2": 1.0, "fn": 0, "fp": 0, "precision": 1.0, "recall": 1.0, "tp": 4 }, "GOV_ID": { "f1": 1.0, "f2": 1.0, "fn": 0, "fp": 0, "precision": 1.0, "recall": 1.0, "tp": 6 }, "PAYMENT": { "f1": 0.3076923076923077, "f2": 0.5263157894736843, "fn": 0, "fp": 9, "precision": 0.18181818181818182, "recall": 1.0, "tp": 2 }, "PERSON": { "f1": 0.7000000000000001, "f2": 0.660377358490566, "fn": 8, "fp": 4, "precision": 0.7777777777777778, "recall": 0.6363636363636364, "tp": 14 }, "PHONE": { "f1": 0.375, "f2": 0.6000000000000001, "fn": 0, "fp": 10, "precision": 0.23076923076923078, "recall": 1.0, "tp": 3 }, "SECRET": { "f1": 0.30769230769230765, "f2": 0.4, "fn": 2, "fp": 7, "precision": 0.2222222222222222, "recall": 0.5, "tp": 2 }, "USERNAME": { "f1": 1.0, "f2": 1.0, "fn": 0, "fp": 0, "precision": 1.0, "recall": 1.0, "tp": 4 } }, "pre_policy_candidates": { "documents": 25, "fully_contained_gold_spans": 51, "gold_spans": 53, "high_severity_leaks": 2, "scope": "mapped requested-type candidates before overlap and score policy; null when unavailable", "span_recall": 0.9622641509433962 }, "precision": 0.581081081081081, "predicted_spans": 74, "recall": 0.8113207547169812, "shift": 8, "slice": "all", "tp": 43, "unmapped_predictions": 54, "wrong_type": 0 }, "intl": { "containment": { "character_recall": 1.0, "covered_characters": 649, "fully_contained_gold_spans": 55, "fully_contained_repeated_identifiers": 8, "gold_characters": 649, "high_severity_gold_spans": 6, "masked_characters": 815, "masking_precision": 0.7963190184049079, "repeated_identifiers": 8, "scope": "declared gold policy; zero gold does not establish PII-free text", "span_recall": 1.0, "zero_gold_document_mask_rate": 1.0, "zero_gold_documents": 3, "zero_gold_masked_documents": 3 }, "documents": 25, "dropped_offsets": 0, "extra": 18, "f1": 0.5932203389830508, "f2": 0.6183745583038869, "fn": 20, "fp": 28, "gold_spans": 55, "high_severity_leaks": 0, "miss": 0, "per_type": { "ADDRESS": { "f1": 0.923076923076923, "f2": 0.9677419354838709, "fn": 0, "fp": 1, "precision": 0.8571428571428571, "recall": 1.0, "tp": 6 }, "EMAIL": { "f1": 1.0, "f2": 1.0, "fn": 0, "fp": 0, "precision": 1.0, "recall": 1.0, "tp": 9 }, "GOV_ID": { "f1": 1.0, "f2": 1.0, "fn": 0, "fp": 0, "precision": 1.0, "recall": 1.0, "tp": 4 }, "PAYMENT": { "f1": 0.2222222222222222, "f2": 0.4166666666666667, "fn": 0, "fp": 7, "precision": 0.125, "recall": 1.0, "tp": 1 }, "PERSON": { "f1": 0.3111111111111111, "f2": 0.27777777777777773, "fn": 20, "fp": 11, "precision": 0.3888888888888889, "recall": 0.25925925925925924, "tp": 7 }, "PHONE": { "f1": 0.6153846153846153, "f2": 0.8, "fn": 0, "fp": 5, "precision": 0.4444444444444444, "recall": 1.0, "tp": 4 }, "SECRET": { "f1": 0.33333333333333337, "f2": 0.5555555555555556, "fn": 0, "fp": 4, "precision": 0.2, "recall": 1.0, "tp": 1 }, "USERNAME": { "f1": 1.0, "f2": 1.0, "fn": 0, "fp": 0, "precision": 1.0, "recall": 1.0, "tp": 3 } }, "pre_policy_candidates": { "documents": 25, "fully_contained_gold_spans": 55, "gold_spans": 55, "high_severity_leaks": 0, "scope": "mapped requested-type candidates before overlap and score policy; null when unavailable", "span_recall": 1.0 }, "precision": 0.5555555555555556, "predicted_spans": 63, "recall": 0.6363636363636364, "shift": 20, "slice": "intl", "tp": 35, "unmapped_predictions": 35, "wrong_type": 0 }, "structured": { "containment": { "character_recall": 0.9438990182328191, "covered_characters": 673, "fully_contained_gold_spans": 68, "fully_contained_repeated_identifiers": 13, "gold_characters": 713, "high_severity_gold_spans": 8, "masked_characters": 1074, "masking_precision": 0.6266294227188082, "repeated_identifiers": 14, "scope": "declared gold policy; zero gold does not establish PII-free text", "span_recall": 0.9577464788732394, "zero_gold_document_mask_rate": 1.0, "zero_gold_documents": 2, "zero_gold_masked_documents": 2 }, "documents": 25, "dropped_offsets": 0, "extra": 30, "f1": 0.6459627329192545, "f2": 0.6951871657754011, "fn": 19, "fp": 38, "gold_spans": 71, "high_severity_leaks": 2, "miss": 3, "per_type": { "ADDRESS": { "f1": 1.0, "f2": 1.0, "fn": 0, "fp": 0, "precision": 1.0, "recall": 1.0, "tp": 5 }, "EMAIL": { "f1": 1.0, "f2": 1.0, "fn": 0, "fp": 0, "precision": 1.0, "recall": 1.0, "tp": 5 }, "GOV_ID": { "f1": 1.0, "f2": 1.0, "fn": 0, "fp": 0, "precision": 1.0, "recall": 1.0, "tp": 3 }, "IP_ADDRESS": { "f1": 1.0, "f2": 1.0, "fn": 0, "fp": 0, "precision": 1.0, "recall": 1.0, "tp": 2 }, "PAYMENT": { "f1": 0.19999999999999998, "f2": 0.38461538461538464, "fn": 0, "fp": 16, "precision": 0.1111111111111111, "recall": 1.0, "tp": 2 }, "PERSON": { "f1": 0.6987951807228915, "f2": 0.6561085972850679, "fn": 17, "fp": 8, "precision": 0.7837837837837838, "recall": 0.6304347826086957, "tp": 29 }, "PHONE": { "f1": 0.5714285714285715, "f2": 0.7692307692307692, "fn": 0, "fp": 6, "precision": 0.4, "recall": 1.0, "tp": 4 }, "SECRET": { "f1": 0.16666666666666666, "f2": 0.2380952380952381, "fn": 2, "fp": 8, "precision": 0.1111111111111111, "recall": 0.3333333333333333, "tp": 1 }, "USERNAME": { "f1": 1.0, "f2": 1.0, "fn": 0, "fp": 0, "precision": 1.0, "recall": 1.0, "tp": 1 } }, "pre_policy_candidates": { "documents": 25, "fully_contained_gold_spans": 68, "gold_spans": 71, "high_severity_leaks": 2, "scope": "mapped requested-type candidates before overlap and score policy; null when unavailable", "span_recall": 0.9577464788732394 }, "precision": 0.5777777777777777, "predicted_spans": 90, "recall": 0.7323943661971831, "shift": 16, "slice": "structured", "tp": 52, "unmapped_predictions": 74, "wrong_type": 0 }, "unstructured": { "containment": { "character_recall": 0.9618055555555556, "covered_characters": 554, "fully_contained_gold_spans": 49, "fully_contained_repeated_identifiers": 3, "gold_characters": 576, "high_severity_gold_spans": 10, "masked_characters": 804, "masking_precision": 0.6890547263681592, "repeated_identifiers": 3, "scope": "declared gold policy; zero gold does not establish PII-free text", "span_recall": 0.98, "zero_gold_document_mask_rate": 1.0, "zero_gold_documents": 2, "zero_gold_masked_documents": 2 }, "documents": 25, "dropped_offsets": 0, "extra": 21, "f1": 0.6491228070175439, "f2": 0.7007575757575757, "fn": 13, "fp": 27, "gold_spans": 50, "high_severity_leaks": 0, "miss": 1, "per_type": { "ADDRESS": { "f1": 0.8571428571428571, "f2": 0.9375, "fn": 0, "fp": 2, "precision": 0.75, "recall": 1.0, "tp": 6 }, "EMAIL": { "f1": 1.0, "f2": 1.0, "fn": 0, "fp": 0, "precision": 1.0, "recall": 1.0, "tp": 5 }, "GOV_ID": { "f1": 1.0, "f2": 1.0, "fn": 0, "fp": 0, "precision": 1.0, "recall": 1.0, "tp": 5 }, "PAYMENT": { "f1": 0.4444444444444445, "f2": 0.6666666666666666, "fn": 0, "fp": 10, "precision": 0.2857142857142857, "recall": 1.0, "tp": 4 }, "PERSON": { "f1": 0.4864864864864865, "f2": 0.4368932038834952, "fn": 13, "fp": 6, "precision": 0.6, "recall": 0.4090909090909091, "tp": 9 }, "PHONE": { "f1": 0.5714285714285715, "f2": 0.7692307692307692, "fn": 0, "fp": 6, "precision": 0.4, "recall": 1.0, "tp": 4 }, "SECRET": { "f1": 0.4, "f2": 0.625, "fn": 0, "fp": 3, "precision": 0.25, "recall": 1.0, "tp": 1 }, "USERNAME": { "f1": 1.0, "f2": 1.0, "fn": 0, "fp": 0, "precision": 1.0, "recall": 1.0, "tp": 3 } }, "pre_policy_candidates": { "documents": 25, "fully_contained_gold_spans": 49, "gold_spans": 50, "high_severity_leaks": 0, "scope": "mapped requested-type candidates before overlap and score policy; null when unavailable", "span_recall": 0.98 }, "precision": 0.578125, "predicted_spans": 64, "recall": 0.74, "shift": 12, "slice": "unstructured", "tp": 37, "unmapped_predictions": 49, "wrong_type": 0 }, "us": { "containment": { "character_recall": 0.9408866995073891, "covered_characters": 573, "fully_contained_gold_spans": 51, "fully_contained_repeated_identifiers": 4, "gold_characters": 609, "high_severity_gold_spans": 12, "masked_characters": 951, "masking_precision": 0.6025236593059937, "repeated_identifiers": 5, "scope": "declared gold policy; zero gold does not establish PII-free text", "span_recall": 0.9622641509433962, "zero_gold_document_mask_rate": 1.0, "zero_gold_documents": 3, "zero_gold_masked_documents": 3 }, "documents": 25, "dropped_offsets": 0, "extra": 27, "f1": 0.6771653543307086, "f2": 0.7517482517482518, "fn": 10, "fp": 31, "gold_spans": 53, "high_severity_leaks": 2, "miss": 2, "per_type": { "ADDRESS": { "f1": 0.9411764705882353, "f2": 0.9756097560975611, "fn": 0, "fp": 1, "precision": 0.8888888888888888, "recall": 1.0, "tp": 8 }, "EMAIL": { "f1": 1.0, "f2": 1.0, "fn": 0, "fp": 0, "precision": 1.0, "recall": 1.0, "tp": 4 }, "GOV_ID": { "f1": 1.0, "f2": 1.0, "fn": 0, "fp": 0, "precision": 1.0, "recall": 1.0, "tp": 6 }, "PAYMENT": { "f1": 0.3076923076923077, "f2": 0.5263157894736843, "fn": 0, "fp": 9, "precision": 0.18181818181818182, "recall": 1.0, "tp": 2 }, "PERSON": { "f1": 0.7000000000000001, "f2": 0.660377358490566, "fn": 8, "fp": 4, "precision": 0.7777777777777778, "recall": 0.6363636363636364, "tp": 14 }, "PHONE": { "f1": 0.375, "f2": 0.6000000000000001, "fn": 0, "fp": 10, "precision": 0.23076923076923078, "recall": 1.0, "tp": 3 }, "SECRET": { "f1": 0.30769230769230765, "f2": 0.4, "fn": 2, "fp": 7, "precision": 0.2222222222222222, "recall": 0.5, "tp": 2 }, "USERNAME": { "f1": 1.0, "f2": 1.0, "fn": 0, "fp": 0, "precision": 1.0, "recall": 1.0, "tp": 4 } }, "pre_policy_candidates": { "documents": 25, "fully_contained_gold_spans": 51, "gold_spans": 53, "high_severity_leaks": 2, "scope": "mapped requested-type candidates before overlap and score policy; null when unavailable", "span_recall": 0.9622641509433962 }, "precision": 0.581081081081081, "predicted_spans": 74, "recall": 0.8113207547169812, "shift": 8, "slice": "us", "tp": 43, "unmapped_predictions": 54, "wrong_type": 0 } }, "taxonomy_version": "2026.11" }