Uncovering Cross-Linguistic Disparities in LLMs using Sparse Autoencoders

Open in new window