WinoWhat: A Parallel Corpus of Paraphrased WinoGrande Sentences with Common Sense Categorization
This work addresses the problem of accurately assessing common sense reasoning in LLMs for researchers and practitioners, revealing potential overestimation in existing benchmarks.
The authors tackled the evaluation of common sense reasoning in LLMs using the WinoGrande benchmark, finding that models perform significantly worse on a new paraphrased corpus (WinoWhat), suggesting overestimation of reasoning capabilities, with minimal impact from benchmark memorization.
In this study, we take a closer look at how Winograd schema challenges can be used to evaluate common sense reasoning in LLMs. Specifically, we evaluate generative models of different sizes on the popular WinoGrande benchmark. We release WinoWhat, a new corpus, in which each instance of the WinoGrande validation set is paraphrased. Additionally, we evaluate the performance on the challenge across five common sense knowledge categories, giving more fine-grained insights on what types of knowledge are more challenging for LLMs. Surprisingly, all models perform significantly worse on WinoWhat, implying that LLM reasoning capabilities are overestimated on WinoGrande. To verify whether this is an effect of benchmark memorization, we match benchmark instances to LLM trainingdata and create two test-suites. We observe that memorization has a minimal effect on model performance on WinoGrande.