1 paper · 1 filter
Jiahao Yuan, Zixiang Di, Shangzixin Zhao +4
Large language models (LLMs) face challenges in aligning with diverse cultural values despite their remarkable performance in generation, which stems from inherent monocultural bia…