BanglaShield: A Script-Robust Safety Classifier and Benchmark for Bangla Native, Romanized, and Code-Mixed Prompts
arXiv:2610.24288v1 Announce Type: new Abstract: Safety classifiers for large language models (LLMs) are typically built and evaluated on standard-script text, yet Bangla, like many lower-resource languages, is routinely written in native script, in Latin-script romanization, and code-mixed with Eng…
Read the full story at arXiv cs.CL ↗
Timeline · 1 report
- 2026-10-20 04:00 · arXiv cs.CL
BanglaShield: A Script-Robust Safety Classifier and Benchmark for Bangla Native, Romanized, and Code-Mixed Prompts