Saved in:
Bibliographic Details
Main Authors: Ni, Zhiyu, Xiao, Yifeng, Liang, Zheng
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2603.02788
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912997414273024
author Ni, Zhiyu
Xiao, Yifeng
Liang, Zheng
author_facet Ni, Zhiyu
Xiao, Yifeng
Liang, Zheng
contents We present a framework for evaluating and benchmarking logical reasoning agents when assessment itself must be reproducible, auditable, and robust to execution failures. Building on agentified assessment, we use an assessor agent to issue tasks, enforce execution budgets, parse outputs, and record structured failure types, while the agent under test only needs to expose a standardized agent-to-agent interface. As a case study, we benchmark an auto-formalization agent for first-order logic (FOL) reasoning on a solver-verified and repaired split of FOLIO. The agent translates natural language premises and conclusions into executable Z3Py programs and employs satisfiability modulo theories (SMT) solving to determine logical entailment. On the cleaned FOLIO validation set, the auto-formalization agent achieves 86.70% accuracy under the assessor protocol, outperforming a chain-of-thought baseline (73.89%).
format Preprint
id arxiv_https___arxiv_org_abs_2603_02788
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Agentified Assessment of Logical Reasoning Agents
Ni, Zhiyu
Xiao, Yifeng
Liang, Zheng
Artificial Intelligence
We present a framework for evaluating and benchmarking logical reasoning agents when assessment itself must be reproducible, auditable, and robust to execution failures. Building on agentified assessment, we use an assessor agent to issue tasks, enforce execution budgets, parse outputs, and record structured failure types, while the agent under test only needs to expose a standardized agent-to-agent interface. As a case study, we benchmark an auto-formalization agent for first-order logic (FOL) reasoning on a solver-verified and repaired split of FOLIO. The agent translates natural language premises and conclusions into executable Z3Py programs and employs satisfiability modulo theories (SMT) solving to determine logical entailment. On the cleaned FOLIO validation set, the auto-formalization agent achieves 86.70% accuracy under the assessor protocol, outperforming a chain-of-thought baseline (73.89%).
title Agentified Assessment of Logical Reasoning Agents
topic Artificial Intelligence
url https://arxiv.org/abs/2603.02788