SinoGlyphBench: A Diagnostic Benchmark for Chinese Glyph-Level Obfuscation in Language-Model Moderation
SinoGlyphBench, a diagnostic benchmark that identifies label-critical semantic anchors and creates matched original and glyph-obfuscated inputs in text and image modalities, is introduced, a diagnostic benchmark that identifies label-critical semantic anchors and creates matched original and glyph-obfuscated inputs in...
Yi-Fan Wang, Zi-Mu Wang, Su-Liu Qin et al.
· 0 citations