Search papers, labs, and topics across Lattice.
This paper introduces a reference-free framework for evaluating reasoning in AI-generated answers, particularly in high-stakes domains where multi-step reasoning is crucial. By employing Natural Language Inference (NLI) to label local premise-target relations and organizing these into a hypergraph, the authors conduct a deterministic backward AND-OR search to assign audit labels to reasoning segments. The framework outperforms traditional LLM-as-judge baselines in both deductive mathematical reasoning and a new clinical reasoning benchmark, UroReason, highlighting the need for more nuanced evaluation of reasoning traces rather than just final answers.
LLMs often produce fluent but flawed reasoning, and our new framework reveals the hidden weaknesses in their outputs that traditional evaluation methods miss.
AI-generated answers in high-stakes domains are often fluent but difficult to verify, especially when they contain multi-step reasoning rather than a single final answer. We propose a reasoning-based, reference-free framework for auditing LLM-generated outputs. The method decomposes a generated reasoning trace into segments, labels local premise-target relations using Natural Language Inference (NLI), and organizes these relations into a hypergraph. A deterministic backward AND-OR search then assigns segment-level audit labels that indicate how each segment is grounded within the generated response. We evaluate the framework in two settings: deductive mathematical reasoning with Hard2Verify, and open-ended medical reasoning with UroReason, a new physician-annotated benchmark of LLM reasoning traces from real clinical cases. Across these settings, our NLI-hypergraph audit provides a more reliable reference-free evaluation signal than direct LLM-as-judge baselines. In the clinical setting, state-of-the-art LLM judges often fail to identify problematic reasoning segments, over-accepting fluent but weakly grounded responses. Our results show that QA evaluation should account for how inferential relations compose across a reasoning trace, rather than relying only on final answers or LLMs as verifiers. UroReason will be made available through an API, and our code will be released as open source.
