AuLLMs

New methods tackle gender bias in audio-enabled LLMs

Researchers have developed new methods to mitigate gender bias in audio-enabled large language models (AuLLMs). These techniques involve adversarial training to unlearn gender perception without sacrificing performance …