A Note on Statistically Accurate Tabular Data Generation Using Large Language Models

Andrey Sidorenko
5/5/2025
cs.LGcs.AI

No official code found. Best implementation: mostly-ai/paper-DataLLM-materials (0 stars, last commit 2025-05-05).

Abstract

Large language models (LLMs) have shown promise in synthetic tabular data generation, yet existing methods struggle to preserve complex feature dependencies, particularly among categorical variables. This work introduces a probability-driven prompting approach that leverages LLMs to estimate conditional distributions, enabling more accurate and scalable data synthesis. The results highlight the potential of prompting probability distributions to enhance the statistical fidelity of LLM-generated tabular data.

View on arXivView PDF

Code Implementations(1)

mostly-ai/paper-DataLLM-materialsCommunity100%
00Python, Jupyter NotebookMay 5, 20251 years ago

Similar Papers

JoLT: Joint Probabilistic Predictions on Tabular Data Using LLMs

86% match1 repo162025

FASTGEN: Fast and Cost-Effective Synthetic Tabular Data Generation with LLMs

85% match2025

Multi-level Diagnosis and Evaluation for Robust Tabular Feature Engineering with Large Language Models

85% match1 repo12025

Unleashing the Potential of Large Language Models for Predictive Tabular Tasks in Data Science

85% match2024

EPIC: Effective Prompting for Imbalanced-Class Data Synthesis in Tabular Data Classification via Large Language Models

85% match1 repo162026

EPIC: Effective Prompting for Imbalanced-Class Data Synthesis in Tabular Data Classification via Large Language Models

85% match1 repo162024

Cite this paper

@article{sidorenko2025note,
  title  = {A Note on Statistically Accurate Tabular Data Generation Using Large Language Models},
  author = {Andrey Sidorenko},
  year   = {2025},
  eprint = {2505.02659},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2505.02659v2}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2505.02659.svg)](https://codeofpaper.com/paper/2505.02659)

Discussion