Moderation-03 / evaluation /coverage_audit.json
ifmain's picture
Publish epoch-5 Moderation 03, frozen presets and evaluation
666382f verified
Raw History Blame Contribute Delete
13.3 kB
{
"status": "coverage_expanded_ready_for_measurement",
"internal_rows": 1240,
"splits": {
"fit": 828,
"audit": 412
},
"challenge_rows": 265,
"category_language_positive_counts_at_reference_0_5": {
"fit": {
"en": {
"harassment": 10,
"harassment_threatening": 3,
"hate": 3,
"hate_threatening": 1,
"self_harm": 3,
"self_harm_instructions": 2,
"self_harm_intent": 3,
"sexual": 6,
"sexual_minors": 4,
"violence": 11,
"violence_graphic": 5
},
"de": {
"harassment": 17,
"harassment_threatening": 5,
"hate": 4,
"hate_threatening": 3,
"self_harm": 2,
"self_harm_instructions": 1,
"self_harm_intent": 2,
"sexual": 7,
"sexual_minors": 5,
"violence": 9,
"violence_graphic": 3
},
"fr": {
"harassment": 12,
"harassment_threatening": 4,
"hate": 3,
"hate_threatening": 2,
"self_harm": 4,
"self_harm_instructions": 3,
"self_harm_intent": 3,
"sexual": 10,
"sexual_minors": 5,
"violence": 12,
"violence_graphic": 5
},
"es": {
"harassment": 17,
"harassment_threatening": 3,
"hate": 4,
"hate_threatening": 2,
"self_harm": 5,
"self_harm_instructions": 3,
"self_harm_intent": 5,
"sexual": 7,
"sexual_minors": 5,
"violence": 8,
"violence_graphic": 3
},
"it": {
"harassment": 12,
"harassment_threatening": 4,
"hate": 4,
"hate_threatening": 2,
"self_harm": 4,
"self_harm_instructions": 2,
"self_harm_intent": 4,
"sexual": 10,
"sexual_minors": 5,
"violence": 10,
"violence_graphic": 4
},
"sv": {
"harassment": 14,
"harassment_threatening": 4,
"hate": 3,
"hate_threatening": 2,
"self_harm": 4,
"self_harm_instructions": 3,
"self_harm_intent": 4,
"sexual": 7,
"sexual_minors": 4,
"violence": 7,
"violence_graphic": 3
},
"fi": {
"harassment": 11,
"harassment_threatening": 4,
"hate": 4,
"hate_threatening": 2,
"self_harm": 3,
"self_harm_instructions": 2,
"self_harm_intent": 3,
"sexual": 9,
"sexual_minors": 5,
"violence": 10,
"violence_graphic": 4
},
"pl": {
"harassment": 14,
"harassment_threatening": 4,
"hate": 3,
"hate_threatening": 1,
"self_harm": 4,
"self_harm_instructions": 2,
"self_harm_intent": 4,
"sexual": 8,
"sexual_minors": 5,
"violence": 10,
"violence_graphic": 4
},
"cs": {
"harassment": 16,
"harassment_threatening": 5,
"hate": 7,
"hate_threatening": 2,
"self_harm": 5,
"self_harm_instructions": 2,
"self_harm_intent": 4,
"sexual": 7,
"sexual_minors": 4,
"violence": 7,
"violence_graphic": 3
},
"lv": {
"harassment": 17,
"harassment_threatening": 4,
"hate": 4,
"hate_threatening": 2,
"self_harm": 3,
"self_harm_instructions": 1,
"self_harm_intent": 3,
"sexual": 9,
"sexual_minors": 4,
"violence": 8,
"violence_graphic": 4
},
"zh": {
"harassment": 15,
"harassment_threatening": 4,
"hate": 3,
"hate_threatening": 2,
"self_harm": 5,
"self_harm_instructions": 3,
"self_harm_intent": 4,
"sexual": 7,
"sexual_minors": 4,
"violence": 9,
"violence_graphic": 3
},
"ja": {
"harassment": 13,
"harassment_threatening": 3,
"hate": 5,
"hate_threatening": 2,
"self_harm": 4,
"self_harm_instructions": 2,
"self_harm_intent": 4,
"sexual": 6,
"sexual_minors": 4,
"violence": 7,
"violence_graphic": 3
},
"ko": {
"harassment": 12,
"harassment_threatening": 2,
"hate": 3,
"hate_threatening": 1,
"self_harm": 4,
"self_harm_instructions": 2,
"self_harm_intent": 4,
"sexual": 9,
"sexual_minors": 5,
"violence": 8,
"violence_graphic": 4
},
"ru": {
"harassment": 17,
"harassment_threatening": 3,
"hate": 4,
"hate_threatening": 1,
"self_harm": 3,
"self_harm_instructions": 2,
"self_harm_intent": 3,
"sexual": 8,
"sexual_minors": 5,
"violence": 5,
"violence_graphic": 2
},
"uk": {
"harassment": 14,
"harassment_threatening": 3,
"hate": 3,
"hate_threatening": 1,
"self_harm": 4,
"self_harm_instructions": 1,
"self_harm_intent": 4,
"sexual": 6,
"sexual_minors": 5,
"violence": 7,
"violence_graphic": 3
},
"be": {
"harassment": 14,
"harassment_threatening": 3,
"hate": 5,
"hate_threatening": 2,
"self_harm": 4,
"self_harm_instructions": 2,
"self_harm_intent": 3,
"sexual": 12,
"sexual_minors": 5,
"violence": 9,
"violence_graphic": 3
},
"kk": {
"harassment": 12,
"harassment_threatening": 3,
"hate": 2,
"hate_threatening": 1,
"self_harm": 3,
"self_harm_instructions": 1,
"self_harm_intent": 3,
"sexual": 10,
"sexual_minors": 5,
"violence": 10,
"violence_graphic": 5
}
},
"audit": {
"en": {
"harassment": 5,
"harassment_threatening": 4,
"hate": 4,
"hate_threatening": 3,
"self_harm": 3,
"self_harm_instructions": 1,
"self_harm_intent": 3,
"sexual": 3,
"sexual_minors": 2,
"violence": 5,
"violence_graphic": 1
},
"de": {
"harassment": 6,
"harassment_threatening": 4,
"hate": 3,
"hate_threatening": 2,
"self_harm": 4,
"self_harm_instructions": 2,
"self_harm_intent": 4,
"sexual": 3,
"sexual_minors": 1,
"violence": 5,
"violence_graphic": 3
},
"fr": {
"harassment": 9,
"harassment_threatening": 3,
"hate": 4,
"hate_threatening": 2,
"self_harm": 2,
"self_harm_instructions": 1,
"self_harm_intent": 2,
"sexual": 3,
"sexual_minors": 2,
"violence": 3,
"violence_graphic": 1
},
"es": {
"harassment": 8,
"harassment_threatening": 4,
"hate": 3,
"hate_threatening": 2,
"self_harm": 2,
"self_harm_instructions": 1,
"self_harm_intent": 2,
"sexual": 4,
"sexual_minors": 2,
"violence": 5,
"violence_graphic": 2
},
"it": {
"harassment": 7,
"harassment_threatening": 3,
"hate": 3,
"hate_threatening": 2,
"self_harm": 3,
"self_harm_instructions": 2,
"self_harm_intent": 3,
"sexual": 5,
"sexual_minors": 2,
"violence": 4,
"violence_graphic": 1
},
"sv": {
"harassment": 9,
"harassment_threatening": 3,
"hate": 4,
"hate_threatening": 2,
"self_harm": 2,
"self_harm_instructions": 1,
"self_harm_intent": 2,
"sexual": 4,
"sexual_minors": 2,
"violence": 4,
"violence_graphic": 2
},
"fi": {
"harassment": 7,
"harassment_threatening": 3,
"hate": 3,
"hate_threatening": 2,
"self_harm": 4,
"self_harm_instructions": 2,
"self_harm_intent": 4,
"sexual": 4,
"sexual_minors": 2,
"violence": 5,
"violence_graphic": 1
},
"pl": {
"harassment": 7,
"harassment_threatening": 3,
"hate": 4,
"hate_threatening": 2,
"self_harm": 2,
"self_harm_instructions": 1,
"self_harm_intent": 2,
"sexual": 3,
"sexual_minors": 1,
"violence": 4,
"violence_graphic": 1
},
"cs": {
"harassment": 6,
"harassment_threatening": 3,
"hate": 3,
"hate_threatening": 2,
"self_harm": 3,
"self_harm_instructions": 2,
"self_harm_intent": 2,
"sexual": 3,
"sexual_minors": 1,
"violence": 4,
"violence_graphic": 2
},
"lv": {
"harassment": 7,
"harassment_threatening": 3,
"hate": 4,
"hate_threatening": 2,
"self_harm": 2,
"self_harm_instructions": 2,
"self_harm_intent": 2,
"sexual": 5,
"sexual_minors": 2,
"violence": 4,
"violence_graphic": 1
},
"zh": {
"harassment": 6,
"harassment_threatening": 3,
"hate": 4,
"hate_threatening": 2,
"self_harm": 2,
"self_harm_instructions": 1,
"self_harm_intent": 2,
"sexual": 4,
"sexual_minors": 2,
"violence": 6,
"violence_graphic": 2
},
"ja": {
"harassment": 6,
"harassment_threatening": 5,
"hate": 4,
"hate_threatening": 2,
"self_harm": 3,
"self_harm_instructions": 2,
"self_harm_intent": 3,
"sexual": 3,
"sexual_minors": 1,
"violence": 6,
"violence_graphic": 2
},
"ko": {
"harassment": 7,
"harassment_threatening": 4,
"hate": 3,
"hate_threatening": 2,
"self_harm": 2,
"self_harm_instructions": 1,
"self_harm_intent": 3,
"sexual": 5,
"sexual_minors": 1,
"violence": 4,
"violence_graphic": 1
},
"ru": {
"harassment": 6,
"harassment_threatening": 3,
"hate": 3,
"hate_threatening": 3,
"self_harm": 3,
"self_harm_instructions": 1,
"self_harm_intent": 3,
"sexual": 2,
"sexual_minors": 1,
"violence": 6,
"violence_graphic": 2
},
"uk": {
"harassment": 7,
"harassment_threatening": 3,
"hate": 3,
"hate_threatening": 2,
"self_harm": 3,
"self_harm_instructions": 2,
"self_harm_intent": 2,
"sexual": 3,
"sexual_minors": 1,
"violence": 4,
"violence_graphic": 2
},
"be": {
"harassment": 6,
"harassment_threatening": 4,
"hate": 3,
"hate_threatening": 2,
"self_harm": 2,
"self_harm_instructions": 1,
"self_harm_intent": 2,
"sexual": 2,
"sexual_minors": 1,
"violence": 7,
"violence_graphic": 2
},
"kk": {
"harassment": 9,
"harassment_threatening": 4,
"hate": 5,
"hate_threatening": 3,
"self_harm": 3,
"self_harm_instructions": 2,
"self_harm_intent": 3,
"sexual": 3,
"sexual_minors": 1,
"violence": 5,
"violence_graphic": 1
}
}
},
"all_11_categories_positive_in_all_17_languages_both_splits": true,
"external_rows": 1680,
"external_overlap_excluded_from_primary_test": 0,
"external_revision": "84e5cf3bcd6acb3dfc70b6760451645872218a3e",
"external_sha256": "ef3f7de86ee394fe337b0a566ba4b9c7db058f3caa8d6a4a0006138b663ab9a8",
"research_mapping": {
"pornography": [
"sexual"
],
"illegal_activity": [],
"disturbing_content": [
"violence_graphic",
"self_harm"
],
"hate": [
"hate",
"hate_threatening"
],
"violence": [
"violence",
"violence_graphic"
]
},
"limitations": [
"Coverage is not proof of detection. Illegal activity has no dedicated output.",
"Broad disturbing content is not equivalent to graphic violence.",
"Internal source labels are inherited teacher estimates; authored checks are not human-reviewed.",
"Translations share semantic groups; reported row counts are not independent sample counts.",
"Typography and 10 authored EN/RU semantic paraphrases do not reproduce the undisclosed researcher benchmark.",
"External data has unknown labels and no gold four-profile labels.",
"Challenge policy: fraud assistance, hateful content, praise of violence and gross imagery blocked at all levels; adult erotica allowed only at light."
]
}