SomaliBench Eval: Measuring English-to-Somali Refusal Gaps in Open-Weight Language Models 事件

SomaliBench Eval: Measuring English-to-Somali Refusal Gaps in Open-Weight Language Models arXiv:2605.25420v1 Announce Type: new Abstract: Large language model safety evaluation remains heavily English-centered, leaving low-resource languages under-measured even when models are deployed globally. We evaluate four open-weight instruction-tuned models on SomaliBench v0, a native-author-verified benchmark of 100 harmful-intent prompts paired across English and Somali. Each of Llama-3.1-8B-Instruct,