Defines MIIT, constructs MIIT-dataset via automatic pipeline across seven risk categories, and trains MiShield-8B which outperforms commercial moderation services using progressively distilled reasoning supervision.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Safe Alone, Unsafe Together: Safeguarding Against Implicit Toxicity When Benign Images Combine
Defines MIIT, constructs MIIT-dataset via automatic pipeline across seven risk categories, and trains MiShield-8B which outperforms commercial moderation services using progressively distilled reasoning supervision.