keyword

chains of reasoning

Chains of reasoning are structured sequences of intermediate logical steps or deductions generated by artificial intelligence systems to solve complex problems and justify their conclusions. Instead of producing an immediate answer in a single step, a system breaks down a primary task into smaller subproblems or intermediate assertions, addressing each sequentially to bridge the gap between initial premises and a final answer. This multistep approach improves problem-solving accuracy on complex reasoning tasks while providing an interpretable and verifiable explanation of how the conclusion was reached.

1 item

Entailer: Answering Questions with Faithful and Truthful Chains of Reasoning

Oyvind Tafjord, Bhavana Dalvi Mishra, Peter Clark

OrganizationsAllen Institute for AI

Why you should read this

Proposes a question-answering framework that couples backward-chaining entailment generation with self-query verification to construct multi-step reasoning trees directly reflecting a language model's internal beliefs.

Our goal is for a system to be able to answer questions and justify its answers with a faithful and truthful chain of reasoning. We present Entailer, a system that produces chains of reasoning by decomposing the question into subquestions and then answering each subquestion using a textual entailment model. The model is trained on a new dataset of 10K examples of multistep reasoning, and uses a novel method to generate the training data automatically from a large corpus. Entailer achieves state-of-the-art results on two multistep reasoning datasets, and its reasoning chains are more faithful and truthful than those of a large language model, as measured by human evaluation and a new automatic metric. We also show that Entailer can be used to improve the performance of a large language model by providing it with a faithful derivation of the answer.

Source

https://aclanthology.org/2022.emnlp-main.134.pdf

Added

2026-10-03