Are We Done with MMLU?

• URL: http://arxiv.org/abs/2406.04127v2
• Date: Fri, 7 Jun 2024 15:19:06 GMT
• Title: Are We Done with MMLU?
• Authors: Aryo Pradipta Gema, Joshua Ong Jun Leang, Giwon Hong, Alessio Devoto, Alberto Carlo Maria Mancino, Rohit Saxena, Xuanli He, Yu Zhao, Xiaotang Du, Mohammad Reza Ghasemi Madani, Claire Barale, Robert McHardy, Joshua Harris, Jean Kaddour, Emile van Krieken, Pasquale Minervini,
• Abstract summary: We identify and analyse errors in the popular Massive Multitask Language Understanding benchmark. For example, we find that 57% of the analysed questions in the Virology subset contain errors. We create MMLU-Redux, which is a subset of 3,000 manually re-annotated questions across 30 MMLU subjects.
• Score: 18.740187299563473
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Maybe not. We identify and analyse errors in the popular Massive Multitask Language Understanding (MMLU) benchmark. Even though MMLU is widely adopted, our analysis demonstrates numerous ground truth errors that obscure the true capabilities of LLMs. For example, we find that 57% of the analysed questions in the Virology subset contain errors. To address this issue, we introduce a comprehensive framework for identifying dataset errors using a novel error taxonomy. Then, we create MMLU-Redux, which is a subset of 3,000 manually re-annotated questions across 30 MMLU subjects. Using MMLU-Redux, we demonstrate significant discrepancies with the model performance metrics that were originally reported. Our results strongly advocate for revising MMLU's error-ridden questions to enhance its future utility and reliability as a benchmark. Therefore, we open up MMLU-Redux for additional annotation https://huggingface.co/datasets/edinburgh-dawg/mmlu-redux.

Related papers

• SpecTool: A Benchmark for Characterizing Errors in Tool-Use LLMs [77.79172008184415]
  SpecTool is a new benchmark to identify error patterns in LLM output on tool-use tasks. We show that even the most prominent LLMs exhibit these error patterns in their outputs. Researchers can use the analysis and insights from SPECTOOL to guide their error mitigation strategies.
  arXiv  Detail & Related papers  (2024-11-20T18:56:22Z)
• MMLU-Pro+: Evaluating Higher-Order Reasoning and Shortcut Learning in LLMs [2.3603377248944017]
  Existing benchmarks for large language models (LLMs) increasingly struggle to differentiate between top-performing models. We introduce MMLU-Pro+, an enhanced benchmark building upon MMLU-Pro to assess shortcut learning and higher-order reasoning. Our results show that MMLU-Pro+ maintains MMLU-Pro's difficulty while providing a more rigorous test of model discrimination.
  arXiv  Detail & Related papers  (2024-09-03T19:31:03Z)
• Revisiting Multi-Modal LLM Evaluation [29.094387692681337]
  We pioneer evaluating recent MLLMs (LLaVA 1.5, LLaVA-NeXT, BLIP2, InstructBLIP, GPT-4V, and GPT-4o) on datasets designed to address weaknesses in earlier ones. Our code is integrated into the widely used LAVIS framework for MLLM evaluation, enabling the rapid assessment of future MLLMs.
  arXiv  Detail & Related papers  (2024-08-09T20:55:46Z)
• Inference-Time Decontamination: Reusing Leaked Benchmarks for Large Language Model Evaluation [61.350306618479365]
  Leakage of benchmarks can prevent the accurate assessment of large language models' true performance. We propose Inference-Time Decontamination (ITD) to address this issue. ITD reduces inflated accuracy by 22.9% on GSM8K and 19.0% on MMLU.
  arXiv  Detail & Related papers  (2024-06-20T04:35:59Z)
• WikiContradict: A Benchmark for Evaluating LLMs on Real-World Knowledge Conflicts from Wikipedia [59.96425443250666]
  Retrieval-augmented generation (RAG) has emerged as a promising solution to mitigate the limitations of large language models (LLMs) In this work, we conduct a comprehensive evaluation of LLM-generated answers to questions based on contradictory passages from Wikipedia. We benchmark a diverse range of both closed and open-source LLMs under different QA scenarios, including RAG with a single passage, and RAG with 2 contradictory passages.
  arXiv  Detail & Related papers  (2024-06-19T20:13:42Z)
• MMLU-Pro: A More Robust and Challenging Multi-Task Language Understanding Benchmark [44.840266648465054]
  This paper introduces MMLU-Pro, an enhanced dataset designed to extend the mostly knowledge-driven MMLU benchmark. With 24 different prompt styles tested, the sensitivity of model scores to prompt variations decreased from 4-5% in MMLU to just 2% in MMLU-Pro. Our assessments confirm that MMLU-Pro is a more discriminative benchmark to better track progress in the field.
  arXiv  Detail & Related papers  (2024-06-03T17:53:00Z)
• LLMs cannot find reasoning errors, but can correct them given the error location [0.9017736137562115]
  Poor self-correction performance stems from LLMs' inability to find logical mistakes, rather than their ability to correct a known mistake. We benchmark several state-of-the-art LLMs on their mistake-finding ability and demonstrate that they generally struggle with the task. We show that it is possible to obtain mistake location information without ground truth labels or in-domain training data.
  arXiv  Detail & Related papers  (2023-11-14T20:12:38Z)
• Learning From Mistakes Makes LLM Better Reasoner [106.48571828587728]
  Large language models (LLMs) recently exhibited remarkable reasoning capabilities on solving math problems. This work explores whether LLMs can LEarn from MistAkes (LEMA), akin to the human learning process.
  arXiv  Detail & Related papers  (2023-10-31T17:52:22Z)
• Can Large Language Models Infer Causation from Correlation? [104.96351414570239]
  We test the pure causal inference skills of large language models (LLMs) We formulate a novel task Corr2Cause, which takes a set of correlational statements and determines the causal relationship between the variables. We show that these models achieve almost close to random performance on the task.
  arXiv  Detail & Related papers  (2023-06-09T12:09:15Z)

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.