MMLU-SR: A Benchmark for Stress-Testing Reasoning Capability of Large Language Models
This provides a rigorous benchmark for stress-testing reasoning in LLMs, addressing a critical issue for AI researchers, though it is incremental as it builds on existing MMLU datasets.
The authors tackled the problem of assessing true comprehension in Large Language Models by creating MMLU-SR, a benchmark that replaces key terms in questions with dummy words and definitions, and found a substantial reduction in model performance, indicating poor comprehension despite high scores on standard tests.
We propose MMLU-SR, a novel dataset designed to measure the true comprehension abilities of Large Language Models (LLMs) by challenging their performance in question-answering tasks with modified terms. We reasoned that an agent that "truly" understands a concept can still evaluate it when key terms are replaced by suitably defined alternate terms, and sought to differentiate such comprehension from mere text replacement. In our study, we modified standardized test questions by replacing a key term with a dummy word along with its definition. The key term could be in the context of questions, answers, or both questions and answers. Notwithstanding the high scores achieved by recent popular LLMs on the MMLU leaderboard, we found a substantial reduction in model performance after such replacement, suggesting poor comprehension. This new benchmark provides a rigorous benchmark for testing true model comprehension, and poses a challenge to the broader scientific community.