Saved in:
Bibliographic Details
Main Authors: Li, Xinye, Wan, Mingqi, Sui, Dianbo
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2505.12328
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912381148332032
author Li, Xinye
Wan, Mingqi
Sui, Dianbo
author_facet Li, Xinye
Wan, Mingqi
Sui, Dianbo
contents We present Team asdfo123's submission to the LLMSR@XLLM25 shared task, which evaluates large language models on producing fine-grained, controllable, and interpretable reasoning processes. Systems must extract all problem conditions, decompose a chain of thought into statement-evidence pairs, and verify the logical validity of each pair. Leveraging only the off-the-shelf Meta-Llama-3-8B-Instruct, we craft a concise few-shot, multi-turn prompt that first enumerates all conditions and then guides the model to label, cite, and adjudicate every reasoning step. A lightweight post-processor based on regular expressions normalises spans and enforces the official JSON schema. Without fine-tuning, external retrieval, or ensembling, our method ranks 5th overall, achieving macro F1 scores on par with substantially more complex and resource-consuming pipelines. We conclude by analysing the strengths and limitations of our approach and outlining directions for future research in structural reasoning with LLMs. Our code is available at https://github.com/asdfo123/LLMSR-asdfo123.
format Preprint
id arxiv_https___arxiv_org_abs_2505_12328
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle LLMSR@XLLM25: An Empirical Study of LLM for Structural Reasoning
Li, Xinye
Wan, Mingqi
Sui, Dianbo
Computation and Language
We present Team asdfo123's submission to the LLMSR@XLLM25 shared task, which evaluates large language models on producing fine-grained, controllable, and interpretable reasoning processes. Systems must extract all problem conditions, decompose a chain of thought into statement-evidence pairs, and verify the logical validity of each pair. Leveraging only the off-the-shelf Meta-Llama-3-8B-Instruct, we craft a concise few-shot, multi-turn prompt that first enumerates all conditions and then guides the model to label, cite, and adjudicate every reasoning step. A lightweight post-processor based on regular expressions normalises spans and enforces the official JSON schema. Without fine-tuning, external retrieval, or ensembling, our method ranks 5th overall, achieving macro F1 scores on par with substantially more complex and resource-consuming pipelines. We conclude by analysing the strengths and limitations of our approach and outlining directions for future research in structural reasoning with LLMs. Our code is available at https://github.com/asdfo123/LLMSR-asdfo123.
title LLMSR@XLLM25: An Empirical Study of LLM for Structural Reasoning
topic Computation and Language
url https://arxiv.org/abs/2505.12328