Conformal Language Model Reasoning with Coherent Factuality
This addresses the need for reliable language model outputs in decision pipelines, particularly for reasoning tasks, though it is incremental as it builds on existing conformal prediction techniques.
The paper tackled the problem of ensuring the correctness of language model outputs in reasoning tasks by defining 'coherent factuality' and developing a conformal-prediction-based method to guarantee it, achieving 90% factuality while retaining 80% or more of original claims on mathematical reasoning datasets.
Language models are increasingly being used in important decision pipelines, so ensuring the correctness of their outputs is crucial. Recent work has proposed evaluating the "factuality" of claims decomposed from a language model generation and applying conformal prediction techniques to filter out those claims that are not factual. This can be effective for tasks such as information retrieval, where constituent claims may be evaluated in isolation for factuality, but is not appropriate for reasoning tasks, as steps of a logical argument can be evaluated for correctness only within the context of the claims that precede them. To capture this, we define "coherent factuality" and develop a conformal-prediction-based method to guarantee coherent factuality for language model outputs. Our approach applies split conformal prediction to subgraphs within a "deducibility" graph" that represents the steps of a reasoning problem. We evaluate our method on mathematical reasoning problems from the MATH and FELM datasets and find that our algorithm consistently produces correct and substantiated orderings of claims, achieving coherent factuality across target coverage levels. Moreover, we achieve 90% factuality on our stricter definition while retaining 80% or more of the original claims, highlighting the utility of our deducibility-graph-guided approach.