Back to feed
arXiv cs.CL·

Multi-Stage Training for Abusive Comment Detection in Indic Languages

Signal
45
Hype
15
In three linesMulti-stage pipeline for abusive comment detection in Indic languages. Combines linguistic preprocessing and model ensembles to minimize false-positive rates while preserving freedom of expression.
Read source
Your take?
AI safetyBenchmarksEvals

Summary generated by Claude — human-verified