Dataset Viewer
The dataset viewer is not available for this split.
Cannot load the dataset split (in streaming mode) to extract the first rows.
Error code: StreamingRowsError
Exception: CastError
Message: Couldn't cast
overall: struct<records: int64, pii_character_by_label: struct<ADDRESS: struct<gold_characters: int64, predic (... 2085 chars omitted)
child 0, records: int64
child 1, pii_character_by_label: struct<ADDRESS: struct<gold_characters: int64, predicted_characters: int64, precision: double, recal (... 1540 chars omitted)
child 0, ADDRESS: struct<gold_characters: int64, predicted_characters: int64, precision: double, recall: double, f1: d (... 6 chars omitted)
child 0, gold_characters: int64
child 1, predicted_characters: int64
child 2, precision: double
child 3, recall: double
child 4, f1: double
child 1, CARD: struct<gold_characters: int64, predicted_characters: int64, precision: double, recall: double, f1: d (... 6 chars omitted)
child 0, gold_characters: int64
child 1, predicted_characters: int64
child 2, precision: double
child 3, recall: double
child 4, f1: double
child 2, EMAIL: struct<gold_characters: int64, predicted_characters: int64, precision: double, recall: double, f1: d (... 6 chars omitted)
child 0, gold_characters: int64
child 1, predicted_characters: int64
child 2, precision: double
child 3, recall: double
child 4, f1: double
child 3, KR_ACCOUNT: struct<gold_characters: int64, predicted_characters: int64, precision: double, recall: double, f1: d (... 6 chars omitted)
child 0, gol
...
le
child 3, families: int64
child 4, ci95: list<item: double>
child 0, item: double
model: string
O2_t4_r2_benign_fpr: struct<value: double, numerator: double, denominator: double, families: int64, ci95: list<item: doub (... 178 chars omitted)
child 0, value: double
child 1, numerator: double
child 2, denominator: double
child 3, families: int64
child 4, ci95: list<item: double>
child 0, item: double
child 5, by_cell: struct<NONE:A-ENC-01: double, NONE:A-ENC-02: double, NONE:A-ENC-03: double, NONE:A-KO-01: double, NO (... 63 chars omitted)
child 0, NONE:A-ENC-01: double
child 1, NONE:A-ENC-02: double
child 2, NONE:A-ENC-03: double
child 3, NONE:A-KO-01: double
child 4, NONE:A-KO-04: double
child 5, NONE:A-KO-05: double
child 6, NONE:A-KO-07: double
S4_t4_r2_rows_with_pii: struct<value: double, numerator: double, denominator: double, families: int64, ci95: list<item: doub (... 4 chars omitted)
child 0, value: double
child 1, numerator: double
child 2, denominator: double
child 3, families: int64
child 4, ci95: list<item: double>
child 0, item: double
O4_account_pair_positive_char_recall: struct<value: double, numerator: double, denominator: double, families: int64, ci95: list<item: doub (... 4 chars omitted)
child 0, value: double
child 1, numerator: double
child 2, denominator: double
child 3, families: int64
child 4, ci95: list<item: double>
child 0, item: double
to
{'model': Value('string'), 'policy': Value('string'), 'machine': {'platform': Value('string'), 'machine': Value('string'), 'onnxruntime': Value('string')}, 'O2_t4_r2_benign_fpr': {'value': Value('float64'), 'numerator': Value('float64'), 'denominator': Value('float64'), 'families': Value('int64'), 'ci95': List(Value('float64')), 'by_cell': {'NONE:A-ENC-01': Value('float64'), 'NONE:A-ENC-02': Value('float64'), 'NONE:A-ENC-03': Value('float64'), 'NONE:A-KO-01': Value('float64'), 'NONE:A-KO-04': Value('float64'), 'NONE:A-KO-05': Value('float64'), 'NONE:A-KO-07': Value('float64')}}, 'O3_pii_negative_rows_with_prediction': {'all_negatives': {'value': Value('float64'), 'numerator': Value('float64'), 'denominator': Value('float64'), 'families': Value('int64'), 'ci95': List(Value('float64'))}, 'numeric_context_negative': {'value': Value('float64'), 'numerator': Value('float64'), 'denominator': Value('float64'), 'families': Value('int64'), 'ci95': List(Value('float64'))}, 'obfuscated_negative': {'value': Value('float64'), 'numerator': Value('float64'), 'denominator': Value('float64'), 'families': Value('int64'), 'ci95': List(Value('float64'))}}, 'O3_bc_negative_rows_with_prediction': {'value': Value('float64'), 'numerator': Value('float64'), 'denominator': Value('float64'), 'families': Value('int64'), 'ci95': List(Value('float64'))}, 'O4_account_pair_positive_char_recall': {'value': Value('float64'), 'numerator': Value('float64'), 'denominator': Value('float64'), 'families': Value('int64'), 'ci95': List(Value('float64'))}, 'S4_t4_r2_rows_with_pii': {'value': Value('float64'), 'numerator': Value('float64'), 'denominator': Value('float64'), 'families': Value('int64'), 'ci95': List(Value('float64'))}}
because column names don't match
Traceback: Traceback (most recent call last):
File "/src/services/worker/src/worker/utils.py", line 149, in get_rows_or_raise
return get_rows(
dataset=dataset,
...<4 lines>...
column_names=column_names,
)
File "/src/libs/libcommon/src/libcommon/utils.py", line 272, in decorator
return func(*args, **kwargs)
File "/src/services/worker/src/worker/utils.py", line 129, in get_rows
rows_plus_one = list(itertools.islice(safe_iter(ds, dataset=dataset), rows_max_number + 1))
File "/src/services/worker/src/worker/utils.py", line 544, in safe_iter
yield from ds.decode(False) if ds.features else ds
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 3052, in __iter__
for key, example in ex_iterable:
^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2559, in __iter__
for key, pa_table in self._iter_arrow():
~~~~~~~~~~~~~~~~^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2584, in _iter_arrow
for key, pa_table in self.ex_iterable._iter_arrow():
~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 558, in _iter_arrow
for key, pa_table in iterator:
^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 440, in _iter_arrow
for key, pa_table in self.generate_tables_fn(**gen_kwags):
~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 361, in _generate_tables
self._cast_table(pa_table, json_field_paths=json_field_paths),
~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 132, in _cast_table
pa_table = table_cast(pa_table, self.info.features.arrow_schema)
File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2381, in table_cast
return cast_table_to_schema(table, schema)
File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2309, in cast_table_to_schema
raise CastError(
...<3 lines>...
)
datasets.table.CastError: Couldn't cast
overall: struct<records: int64, pii_character_by_label: struct<ADDRESS: struct<gold_characters: int64, predic (... 2085 chars omitted)
child 0, records: int64
child 1, pii_character_by_label: struct<ADDRESS: struct<gold_characters: int64, predicted_characters: int64, precision: double, recal (... 1540 chars omitted)
child 0, ADDRESS: struct<gold_characters: int64, predicted_characters: int64, precision: double, recall: double, f1: d (... 6 chars omitted)
child 0, gold_characters: int64
child 1, predicted_characters: int64
child 2, precision: double
child 3, recall: double
child 4, f1: double
child 1, CARD: struct<gold_characters: int64, predicted_characters: int64, precision: double, recall: double, f1: d (... 6 chars omitted)
child 0, gold_characters: int64
child 1, predicted_characters: int64
child 2, precision: double
child 3, recall: double
child 4, f1: double
child 2, EMAIL: struct<gold_characters: int64, predicted_characters: int64, precision: double, recall: double, f1: d (... 6 chars omitted)
child 0, gold_characters: int64
child 1, predicted_characters: int64
child 2, precision: double
child 3, recall: double
child 4, f1: double
child 3, KR_ACCOUNT: struct<gold_characters: int64, predicted_characters: int64, precision: double, recall: double, f1: d (... 6 chars omitted)
child 0, gol
...
le
child 3, families: int64
child 4, ci95: list<item: double>
child 0, item: double
model: string
O2_t4_r2_benign_fpr: struct<value: double, numerator: double, denominator: double, families: int64, ci95: list<item: doub (... 178 chars omitted)
child 0, value: double
child 1, numerator: double
child 2, denominator: double
child 3, families: int64
child 4, ci95: list<item: double>
child 0, item: double
child 5, by_cell: struct<NONE:A-ENC-01: double, NONE:A-ENC-02: double, NONE:A-ENC-03: double, NONE:A-KO-01: double, NO (... 63 chars omitted)
child 0, NONE:A-ENC-01: double
child 1, NONE:A-ENC-02: double
child 2, NONE:A-ENC-03: double
child 3, NONE:A-KO-01: double
child 4, NONE:A-KO-04: double
child 5, NONE:A-KO-05: double
child 6, NONE:A-KO-07: double
S4_t4_r2_rows_with_pii: struct<value: double, numerator: double, denominator: double, families: int64, ci95: list<item: doub (... 4 chars omitted)
child 0, value: double
child 1, numerator: double
child 2, denominator: double
child 3, families: int64
child 4, ci95: list<item: double>
child 0, item: double
O4_account_pair_positive_char_recall: struct<value: double, numerator: double, denominator: double, families: int64, ci95: list<item: doub (... 4 chars omitted)
child 0, value: double
child 1, numerator: double
child 2, denominator: double
child 3, families: int64
child 4, ci95: list<item: double>
child 0, item: double
to
{'model': Value('string'), 'policy': Value('string'), 'machine': {'platform': Value('string'), 'machine': Value('string'), 'onnxruntime': Value('string')}, 'O2_t4_r2_benign_fpr': {'value': Value('float64'), 'numerator': Value('float64'), 'denominator': Value('float64'), 'families': Value('int64'), 'ci95': List(Value('float64')), 'by_cell': {'NONE:A-ENC-01': Value('float64'), 'NONE:A-ENC-02': Value('float64'), 'NONE:A-ENC-03': Value('float64'), 'NONE:A-KO-01': Value('float64'), 'NONE:A-KO-04': Value('float64'), 'NONE:A-KO-05': Value('float64'), 'NONE:A-KO-07': Value('float64')}}, 'O3_pii_negative_rows_with_prediction': {'all_negatives': {'value': Value('float64'), 'numerator': Value('float64'), 'denominator': Value('float64'), 'families': Value('int64'), 'ci95': List(Value('float64'))}, 'numeric_context_negative': {'value': Value('float64'), 'numerator': Value('float64'), 'denominator': Value('float64'), 'families': Value('int64'), 'ci95': List(Value('float64'))}, 'obfuscated_negative': {'value': Value('float64'), 'numerator': Value('float64'), 'denominator': Value('float64'), 'families': Value('int64'), 'ci95': List(Value('float64'))}}, 'O3_bc_negative_rows_with_prediction': {'value': Value('float64'), 'numerator': Value('float64'), 'denominator': Value('float64'), 'families': Value('int64'), 'ci95': List(Value('float64'))}, 'O4_account_pair_positive_char_recall': {'value': Value('float64'), 'numerator': Value('float64'), 'denominator': Value('float64'), 'families': Value('int64'), 'ci95': List(Value('float64'))}, 'S4_t4_r2_rows_with_pii': {'value': Value('float64'), 'numerator': Value('float64'), 'denominator': Value('float64'), 'families': Value('int64'), 'ci95': List(Value('float64'))}}
because column names don't matchNeed help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
가림(Garim) — 데이터 · 학습 기록
가림 L1 의 학습 · 평가 데이터와 RunPod 학습 기록이다. 저장소 루트가 프로젝트의 data/ 디렉터리와 같은
구조라, 받아서 data/ 에 두면 코드가 그대로 읽는다(results/ 만 예외 — 아래). 모델은 model 저장소
cycloevan/garim-guard, 코드는 GitHub
seok-hee97/garim-guard.
빠진 것 — 체크섬을 통과하는 합성 주민등록번호가 든 파일(
corpus/pii-synthetic-*·corpus/kdpii-*·merged/· R1 병합 백업)과 원천 데이터 원본(raw/)은 이 저장소에 없다. 합성 PII 는 저장소의 생성기가 시드로 같은 바이트를 다시 만들고(train/prepare_corpora.py --ack-sensitive-synthetic-identifiers), 원천은 고정 revision 으로 다시 받는다(train/download_assets.py --execute). 해시는*-manifest.json과 RUNPOD.md 에 있다.
구조
| 경로 | 무엇 | GPU 없이 다시 만들 수 있나 |
|---|---|---|
corpus/ |
공통 JSONL 코퍼스 — 공격 합성(R1 · R2) · T4 홀드아웃 · PII 음성 · KLUE · PolyGuard · deepset 번역 · 교사 입력 | ✅ 시드 결정적 |
local/ |
deepset 한국어 번역 적용본 | ✅ |
runpod/2026-10-27/ |
R1 교사 원출력 3종(kanana · Gemma 4 · Qwen3Guard) · 합의/폐기 감사 · 봉인 manifest · 6b 보고서 · Qwen 0.6B 대조본 · pod 로그 | ❌ |
runpod/2026-10-27-r2/ |
R2 교사 원출력 · 합의 · 봉인 manifest · pod 로그 · final/train/artifacts/r2/(판정표 verdict.json · 지표 원본 · 결함 커널 대조값) |
❌ |
results/r1/ |
R1 평가 결과 JSON(프로젝트의 train/artifacts/*.json) |
— |
*-manifest.json |
자산 · 학습 · 평가 봉인 | — |
교사 원출력에는 원문이 없다(id · 라벨 · 확률). 원문은 같은 id 의 corpus/attack-teacher-input*.jsonl 에 있다.
원천과 라이선스
이 저장소에 들어 있는 원천 파생분과 바꾼 것. 출처는 고정 revision 이다(train/sources.yaml).
| 원천 · 출처 | 들어 있는 파일 · 바꾼 것 | 라이선스 |
|---|---|---|
KLUE-NER — klue/klue 349481ec |
corpus/klue_* — NER 문장을 문자 단위 PERSON 스팬으로 바꾸고 분할 |
CC-BY-SA-4.0 |
PolyGuardMix — ToxicityPrompts/PolyGuardMix 5b7d93e9 |
corpus/polyguard_train-* · attack-teacher-* · runpod/* 교사 출력 — 한국어 행을 고르고 교사 3종으로 공격 범주를 다시 붙임 |
CC-BY-4.0 |
PolyGuardPrompts — ToxicityPrompts/PolyGuardPrompts c5b466a9 |
corpus/polyguard_eval-test — 한국어 평가 행 선택 |
CC-BY-4.0 |
deepset prompt-injections — deepset/prompt-injections 4f61ecb0 |
corpus/deepset* · local/deepset-*-ko — 한국어로 번역(행별 method: Gemini · Claude)하고 교사로 다시 라벨링 |
CC-BY-4.0 으로 다룸(원 카드는 Apache-2.0 — 더 엄격한 출처 표시 조건을 따른다) |
bc-pii-ko — sehe1121/bc-pii-ko 327c775b |
corpus/bc_pii_augmented_negative-test — PII 아닌 숫자 문장 720행(학습 때 값 사전으로도 썼다) |
CC-BY-4.0 |
- PolyGuardMix 의
source_id="itw"행은 공개 대화 로그에서 왔다. 원 로그에는 따로 이용 조건이 있을 수 있으니, 그 행을 다시 배포하려면 PolyGuardMix 카드의 출처를 확인한다. - 평가에만 쓰고 이 저장소에 없는 원천: KDPII(Zenodo 16759166,
CC-BY-4.0), KoreanGuardrail · korean_guardrail_test · XL-SafetyBench(CC-BY-4.0), TAB echr(MIT),
CredData(Apache-2.0). 고정 revision 으로 다시 받는다(
train/download_assets.py --execute). - 프로젝트 합성 데이터(
attack-synthetic-*·t4-*·pii-negative-*· 생성기 골드)와 학습 기록은 Apache-2.0. 각 원천 파생분은 위 원천의 라이선스와 출처 표시 조건을 따른다.
- Downloads last month
- 59