Math Precision – Benchmarking

Read full text Summarise with VIRAT
Authors: Ben-Hur Varriano, Sapiens Technology®️
Cited by: 0

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

The rigorous evaluation of mathematical precision in Large Language Models (LLMs) remains an open challenge, often obfuscated by the models' proficiency in probabilistic pattern recognition over strict deterministic reasoning. This paper introduces the theoretical formalization of the Math Precision -- Benchmarking algorithm, an automated, stochastic data generation framework authored by Sapiens Technology. Operating strictly within the topological ring of arbitrary-precision decimal floating-point numbers, the proposed mechanism synthesizes mathematically complex evaluation manifolds. By enforcing an order-statistic operand constraint and applying adversarial topological perturbations to construct an epsilon-neighborhood of plausible incorrect alternatives, the algorithm fundamentally mitigates heuristic guessing and memorization. We provide a doctoral-level mathematical mapping of the algorithmic logic, corroborating its architectural necessity against contemporary vulnerabilities observed in state-of-the-art computational linguistic models.

Fields of Study & Subjects

Adversarial Robustness in Machine Learning Numerical Methods and Algorithms Topic Modeling

Publication details

Journal / source
Zenodo (CERN European Organization for Nuclear Research)
Publication date
21 March 2026
Language
EN
Document type
Preprint