Fine-tuning open-source LLMs on a prover-filtered preference dataset improves whole-problem translation of natural-language reasoning into first-order logic, cutting syntax errors and increasing logical correctness.
URL https://aclanthology.org/2022.coling-1.481
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
LOGICPO: Efficient Translation of NL-based Logical Problems to FOL using LLMs and Preference Optimization
Fine-tuning open-source LLMs on a prover-filtered preference dataset improves whole-problem translation of natural-language reasoning into first-order logic, cutting syntax errors and increasing logical correctness.