ReTraceQA: Evaluating Reasoning Traces of Small Language Models in Commonsense Question Answering
This addresses the evaluation gap for SLMs in commonsense reasoning, which is incremental as it introduces a new benchmark but does not fundamentally change model capabilities.
The paper tackles the problem of evaluating small language models (SLMs) in commonsense question answering by showing that current metrics overestimate their capabilities, as 14-24% of correct answers come from flawed reasoning, and using large language models as judges reduces SLM performance by up to 25%.
While Small Language Models (SLMs) have demonstrated promising performance on an increasingly wide array of commonsense reasoning benchmarks, current evaluation practices rely almost exclusively on the accuracy of their final answers, neglecting the validity of the reasoning processes that lead to those answers. To address this issue, we introduce ReTraceQA, a novel benchmark that introduces process-level evaluation for commonsense reasoning tasks. Our expert-annotated dataset reveals that in a substantial portion of instances (14-24%), SLMs provide correct final answers despite flawed reasoning processes, suggesting that the capabilities of SLMs are often overestimated by evaluation metrics that focus only on comparing the final answer with the ground truth. Indeed, we show that when employing strong Large Language Models (LLMs) as automated judges for reasoning-aware evaluation rather than answer-only metrics, SLM performance drops significantly across all models and datasets, with scores decreasing by up to 25%.