首个跨26语言的长文本模型评测基准，发现英语并非最强。
One ruler to measure them all: Benchmarking multilingual long-context language models
- 构建26语言长文本评测集，含7类合成任务
- 128K上下文下低资源语言性能差距扩大
- 跨语言指令时性能波动达20%，适合多语研究者
我们提出ONERULER，一个涵盖26种语言的多语言长上下文语言模型评测基准。该基准在英文RULER的基础上扩展，包含7个合成任务，测试检索与聚合能力，并引入无答案场景的'针在草堆'变体。通过先撰写英文指令再由母语者翻译至25种语言的两步流程构建。实验显示，当上下文长度从8K增至128K时，低资源语言与高资源语言间性能差距扩大；出人意料的是，英语在26种语言中仅排名第6，波兰语表现最佳。此外，许多模型（尤其是OpenAI的o3-mini-high）即使在高资源语言中也错误预测答案缺失。在跨语言场景下（指令与上下文语言不同），性能波动最高可达20%。ONERULER的发布旨在推动多语言与跨语言长上下文训练方法的研究。
原文摘要 · Abstract (English)
We present ONERULER, a multilingual benchmark designed to evaluate long-context language models across 26 languages. ONERULER adapts the English-only RULER benchmark (Hsieh et al., 2024) by including seven synthetic tasks that test both retrieval and aggregation, including new variations of the "needle-in-a-haystack" task that allow for the possibility of a nonexistent needle. We create ONERULER through a two-step process, first writing English instructions for each task and then collaborating with native speakers to translate them into 25 additional languages. Experiments with both open-weight and closed LLMs reveal a widening performance gap between low- and high-resource languages as context length increases from 8K to 128K tokens. Surprisingly, English is not the top-performing language on long-context tasks (ranked 6th out of 26), with Polish emerging as the top language. Our experiments also show that many LLMs (particularly OpenAI's o3-mini-high) incorrectly predict the absence of an answer, even in high-resource languages. Finally, in cross-lingual scenarios where instructions and context appear in different languages, performance can fluctuate by up to 20% depending on the instruction language. We hope the release of ONERULER will facilitate future research into improving multilingual and cross-lingual long-context training pipelines.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。