What Defines Good Reasoning in LLMs? Dissecting Reasoning Steps with Multi-Aspect Evaluation
This work addresses the need for more granular evaluation of reasoning in LLMs to improve model robustness, offering a scalable framework for analysis and debugging, though it is incremental in refining existing evaluation methods.
The paper tackles the problem of evaluating reasoning quality in large language models by decomposing it into relevance and coherence, and introduces causal stepwise evaluation (CaSE) to measure these aspects reliably, showing that using CaSE-evaluated data improves final task performance on benchmarks like MRa-GSM8K and MRa-MATH.
Evaluating large language models (LLMs) on final-answer correctness is the dominant paradigm. This approach, however, provides a coarse signal for model improvement and overlooks the quality of the underlying reasoning process. We argue that a more granular evaluation of reasoning offers a more effective path to building robust models. We decompose reasoning quality into two dimensions: relevance and coherence. Relevance measures if a step is grounded in the problem; coherence measures if it follows logically from prior steps. To measure these aspects reliably, we introduce causal stepwise evaluation (CaSE). This method assesses each reasoning step using only its preceding context, which avoids hindsight bias. We validate CaSE against human judgments on our new expert-annotated benchmarks, MRa-GSM8K and MRa-MATH. More importantly, we show that curating training data with CaSE-evaluated relevance and coherence directly improves final task performance. Our work provides a scalable framework for analyzing, debugging, and improving LLM reasoning, demonstrating the practical value of moving beyond validity checks.