ResearchBenchmarksBias Evaluation on BBQ GenderFollow47.2Ambiguity ScoreKLAAD33.6837.1940.744.21Mar 19, 2026Evaluation ResultsMethodMethodLinksAmbiguity ScoreAccuracyKLAADBackbone=LLaMA-3-8BBackbone=LLaMA-3-8B2026.0347.236.9CDABackbone=LLaMA-3-8BBackbone=LLaMA-3-8B2026.0341.938.1ORIGINALBackbone=LLaMA-3-8BBackbone=LLaMA-3-8B2026.0337.238.8UGIDBackbone=LLaMA-3-8BBackbone=LLaMA-3-8B2026.0334.239.8