Last released Jul 31, 2026
Inverse soft-Q learning (IQ-Learn) reward extraction for hate-speech analysis on the One Million Posts Corpus