Mathador-LM: A Dynamic Benchmark for Mathematical Reasoning on Large Language Models

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Kurtic, Eldar, Moeini, Amir, Alistarh, Dan
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913546482221056
author Kurtic, Eldar
Moeini, Amir
Alistarh, Dan
author_facet Kurtic, Eldar
Moeini, Amir
Alistarh, Dan
contents We introduce Mathador-LM, a new benchmark for evaluating the mathematical reasoning on large language models (LLMs), combining ruleset interpretation, planning, and problem-solving. This benchmark is inspired by the Mathador game, where the objective is to reach a target number using basic arithmetic operations on a given set of base numbers, following a simple set of rules. We show that, across leading LLMs, we obtain stable average performance while generating benchmark instances \emph{dynamically}, following a target difficulty level. Thus, our benchmark alleviates concerns about test-set leakage into training data, an issue that often undermines popular benchmarks. Additionally, we conduct a comprehensive evaluation of both open and closed-source state-of-the-art LLMs on Mathador-LM. Our findings reveal that contemporary models struggle with Mathador-LM, scoring significantly lower than average 3rd graders. This stands in stark contrast to their strong performance on popular mathematical reasoning benchmarks. The implementation of Mathador-LM benchmark is available at \href{https://github.com/IST-DASLab/Mathador-LM}{github.com/IST-DASLab/Mathador-LM}.
format Preprint
id arxiv_https___arxiv_org_abs_2406_12572
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Mathador-LM: A Dynamic Benchmark for Mathematical Reasoning on Large Language Models
Kurtic, Eldar
Moeini, Amir
Alistarh, Dan
Computation and Language
Artificial Intelligence
Machine Learning
I.2.7
We introduce Mathador-LM, a new benchmark for evaluating the mathematical reasoning on large language models (LLMs), combining ruleset interpretation, planning, and problem-solving. This benchmark is inspired by the Mathador game, where the objective is to reach a target number using basic arithmetic operations on a given set of base numbers, following a simple set of rules. We show that, across leading LLMs, we obtain stable average performance while generating benchmark instances \emph{dynamically}, following a target difficulty level. Thus, our benchmark alleviates concerns about test-set leakage into training data, an issue that often undermines popular benchmarks. Additionally, we conduct a comprehensive evaluation of both open and closed-source state-of-the-art LLMs on Mathador-LM. Our findings reveal that contemporary models struggle with Mathador-LM, scoring significantly lower than average 3rd graders. This stands in stark contrast to their strong performance on popular mathematical reasoning benchmarks. The implementation of Mathador-LM benchmark is available at \href{https://github.com/IST-DASLab/Mathador-LM}{github.com/IST-DASLab/Mathador-LM}.
title Mathador-LM: A Dynamic Benchmark for Mathematical Reasoning on Large Language Models
topic Computation and Language
Artificial Intelligence
Machine Learning
I.2.7
url https://arxiv.org/abs/2406.12572