Improving Rule-based Reasoning in LLMs using Neurosymbolic Representations

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Dhanraj, Varun, Eliasmith, Chris
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915624751464448
author Dhanraj, Varun
Eliasmith, Chris
author_facet Dhanraj, Varun
Eliasmith, Chris
contents Large language models (LLMs) continue to face challenges in reliably solving reasoning tasks, particularly those that require precise rule following, as often found in mathematical reasoning. This paper introduces a novel neurosymbolic method that improves LLM reasoning by encoding hidden states into neurosymbolic vectors, enabling problem-solving within a neurosymbolic vector space. The results are decoded and merged with the original hidden state, significantly boosting the model's performance on numerical reasoning tasks. By offloading computation through neurosymbolic representations, this method enhances efficiency, reliability, and interpretability. Experimental results demonstrate an average of 88.6% lower cross-entropy loss and 15.4 times more problems correctly solved on a suite of mathematical reasoning tasks compared to chain-of-thought prompting and supervised fine-tuning (LoRA), without degrading performance on other tasks. We make our code available at: https://github.com/vdhanraj/Neurosymbolic-LLM.
format Preprint
id arxiv_https___arxiv_org_abs_2502_01657
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Improving Rule-based Reasoning in LLMs using Neurosymbolic Representations
Dhanraj, Varun
Eliasmith, Chris
Machine Learning
Artificial Intelligence
Large language models (LLMs) continue to face challenges in reliably solving reasoning tasks, particularly those that require precise rule following, as often found in mathematical reasoning. This paper introduces a novel neurosymbolic method that improves LLM reasoning by encoding hidden states into neurosymbolic vectors, enabling problem-solving within a neurosymbolic vector space. The results are decoded and merged with the original hidden state, significantly boosting the model's performance on numerical reasoning tasks. By offloading computation through neurosymbolic representations, this method enhances efficiency, reliability, and interpretability. Experimental results demonstrate an average of 88.6% lower cross-entropy loss and 15.4 times more problems correctly solved on a suite of mathematical reasoning tasks compared to chain-of-thought prompting and supervised fine-tuning (LoRA), without degrading performance on other tasks. We make our code available at: https://github.com/vdhanraj/Neurosymbolic-LLM.
title Improving Rule-based Reasoning in LLMs using Neurosymbolic Representations
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2502.01657