Global MMLU: Understanding and Addressing Cultural and Linguistic Biases in Multilingual Evaluation

• URL: http://arxiv.org/abs/2412.03304v1
• Date: Wed, 04 Dec 2024 13:27:09 GMT
• Title: Global MMLU: Understanding and Addressing Cultural and Linguistic Biases in Multilingual Evaluation
• Authors: Shivalika Singh, Angelika Romanou, Clémentine Fourrier, David I. Adelani, Jian Gang Ngui, Daniel Vila-Suero, Peerat Limkonchotiwat, Kelly Marchisio, Wei Qi Leong, Yosephine Susanto, Raymond Ng, Shayne Longpre, Wei-Yin Ko, Madeline Smith, Antoine Bosselut, Alice Oh, Andre F. T. Martins, Leshem Choshen, Daphne Ippolito, Enzo Ferrante, Marzieh Fadaee, Beyza Ermis, Sara Hooker,
• Abstract summary: Cultural biases in multilingual datasets pose significant challenges for their effectiveness as global benchmarks.<n>We show that progress on MMLU depends heavily on learning Western-centric concepts, with 28% of all questions requiring culturally sensitive knowledge.<n>We release Global-MMLU, an improved MMLU with evaluation coverage across 42 languages.
• Score: 50.38159901496538
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Cultural biases in multilingual datasets pose significant challenges for their effectiveness as global benchmarks. These biases stem not only from language but also from the cultural knowledge required to interpret questions, reducing the practical utility of translated datasets like MMLU. Furthermore, translation often introduces artifacts that can distort the meaning or clarity of questions in the target language. A common practice in multilingual evaluation is to rely on machine-translated evaluation sets, but simply translating a dataset is insufficient to address these challenges. In this work, we trace the impact of both of these issues on multilingual evaluations and ensuing model performances. Our large-scale evaluation of state-of-the-art open and proprietary models illustrates that progress on MMLU depends heavily on learning Western-centric concepts, with 28% of all questions requiring culturally sensitive knowledge. Moreover, for questions requiring geographic knowledge, an astounding 84.9% focus on either North American or European regions. Rankings of model evaluations change depending on whether they are evaluated on the full portion or the subset of questions annotated as culturally sensitive, showing the distortion to model rankings when blindly relying on translated MMLU. We release Global-MMLU, an improved MMLU with evaluation coverage across 42 languages -- with improved overall quality by engaging with compensated professional and community annotators to verify translation quality while also rigorously evaluating cultural biases present in the original dataset. This comprehensive Global-MMLU set also includes designated subsets labeled as culturally sensitive and culturally agnostic to allow for more holistic, complete evaluation.

Related papers

• Evaluating the Quality of Benchmark Datasets for Low-Resource Languages: A Case Study on Turkish [1.59623393716069]
  This study addresses the need for robust and culturally appropriate benchmarks by evaluating the quality of 17 commonly used Turkish benchmark datasets. Our results reveal that 70% of the benchmark datasets fail to meet our quality standards. GPT-4o has stronger labeling capabilities for grammatical and technical tasks, while Llama3.3-70B excels at correctness and cultural knowledge evaluation.
  arXiv  Detail & Related papers  (2025-04-13T20:45:49Z)
• Do LLMs Understand Your Translations? Evaluating Paragraph-level MT with Question Answering [68.3400058037817]
  We introduce TREQA (Translation Evaluation via Question-Answering), a framework that extrinsically evaluates translation quality. We show that TREQA is competitive with and, in some cases, outperforms state-of-the-art neural and LLM-based metrics in ranking alternative paragraph-level translations.
  arXiv  Detail & Related papers  (2025-04-10T09:24:54Z)
• MMLU-ProX: A Multilingual Benchmark for Advanced Large Language Model Evaluation [60.52580061637301]
  MMLU-ProX is a comprehensive benchmark covering 13 typologically diverse languages with approximately 11,829 questions per language. We evaluate 25 state-of-the-art large language models (LLMs) using 5-shot chain-of-thought (CoT) and zero-shot prompting strategies, analyzing their performance across linguistic and cultural boundaries. Our experiments reveal consistent performance degradation from high-resource languages to lower-resource ones, with the best models achieving over 70% accuracy on English but dropping to around 40% for languages like Swahili.
  arXiv  Detail & Related papers  (2025-03-13T15:59:20Z)
• All Languages Matter: Evaluating LMMs on Culturally Diverse 100 Languages [73.93600813999306]
  ALM-bench is the largest and most comprehensive effort to date for evaluating LMMs across 100 languages.<n>It challenges existing models by testing their ability to understand and reason about culturally diverse images paired with text in various languages.<n>The benchmark offers a robust and nuanced evaluation framework featuring various question formats, including true/false, multiple choice, and open-ended questions.
  arXiv  Detail & Related papers  (2024-11-25T15:44:42Z)
• LLM-as-a-Judge & Reward Model: What They Can and Cannot Do [2.2469442203227863]
  We conduct a comprehensive analysis of automated evaluators, reporting several key findings on their behavior. We discover that English evaluation capabilities significantly influence language-specific evaluation capabilities, enabling evaluators trained in English to easily transfer their skills to other languages. We find that state-of-the-art evaluators struggle with challenging prompts, in either English or Korean, underscoring their limitations in assessing or generating complex reasoning questions.
  arXiv  Detail & Related papers  (2024-09-17T14:40:02Z)
• Beyond Metrics: Evaluating LLMs' Effectiveness in Culturally Nuanced, Low-Resource Real-World Scenarios [29.56889133557681]
  This research evaluates the performance of seven leading Large Language Models (LLMs) in sentiment analysis on a dataset derived from WhatsApp chats. We find that while Mistral-7b and Mixtral-8x7b achieved high F1 scores, they and other LLMs such as GPT-3.5-Turbo, Llama-2-70b, and Gemma-7b struggled with understanding linguistic and contextual nuances. GPT-4 and GPT-4-Turbo excelled in grasping diverse linguistic inputs and managing various contextual information.
  arXiv  Detail & Related papers  (2024-06-01T07:36:59Z)
• LLaMA Beyond English: An Empirical Study on Language Capability Transfer [49.298360366468934]
  We focus on how to effectively transfer the capabilities of language generation and following instructions to a non-English language. We analyze the impact of key factors such as vocabulary extension, further pretraining, and instruction tuning on transfer. We employ four widely used standardized testing benchmarks: C-Eval, MMLU, AGI-Eval, and GAOKAO-Bench.
  arXiv  Detail & Related papers  (2024-01-02T06:29:02Z)
• CMMLU: Measuring massive multitask language understanding in Chinese [133.70911295934746]
  This paper introduces a comprehensive Chinese benchmark that covers various subjects, including natural science, social sciences, engineering, and humanities. CMMLU fills the gap in evaluating the knowledge and reasoning capabilities of large language models within the Chinese context.
  arXiv  Detail & Related papers  (2023-06-15T15:49:51Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.