VOYAGER: A Training Free Approach for Generating Diverse Datasets using LLMs

Avinash Amballa, Yashas Malur Saidutta, Chi-Heng Lin, Vivek Kulkarni, Srinivas Chappidi
12/12/2025
cs.CLcs.LG

No official code found. Best implementation: data2json/DataVOYAGER (6 stars, last commit 2025-12-21).

Abstract

Large language models (LLMs) are increasingly being used to generate synthetic datasets for the evaluation and training of downstream models. However, prior work has noted that such generated data lacks diversity. In this paper, we propose Voyager, a novel principled approach to generate diverse datasets. Our approach is iterative and directly optimizes a mathematical quantity that optimizes the diversity of the dataset using the machinery of determinantal point processes. Furthermore, our approach is training-free, applicable to closed-source models, and scalable. In addition to providing theoretical justification for the working of our method, we also demonstrate through comprehensive experiments that Voyager significantly outperforms popular baseline approaches by providing a 1.5-3x improvement in diversity.

View on arXivView PDF

Code Implementations(1)

data2json/DataVOYAGERCommunity62%

Faithful reproduction of "VOYAGER: A Training Free Approach for Generating Diverse Datasets using LLMs" https://arxiv.org/abs/2512.12072 https://arxiv.org/pdf/2512.12072

60Shell, PythonDec 21, 202510 months ago

Similar Papers

CorrSynth -- A Correlated Sampling Method for Diverse Dataset Generation from LLMs

86% match2024

Diversity as a Reward: Fine-Tuning LLMs on a Mixture of Domain-Undetermined Data

85% match1 repo7.1k2025

What Matters in LLM-generated Data: Diversity and Its Effect on Model Fine-Tuning

85% match2025

G2: Guided Generation for Enhanced Output Diversity in LLMs

85% match1 repo22025

Efficient Code LLM Training via Distribution-Consistent and Diversity-Aware Data Selection

84% match1 repo22025

How to Train Data-Efficient LLMs

84% match1 repo122024

Cite this paper

@article{amballa2025voyager,
  title  = {VOYAGER: A Training Free Approach for Generating Diverse Datasets using LLMs},
  author = {Avinash Amballa and Yashas Malur Saidutta and Chi-Heng Lin and Vivek Kulkarni and Srinivas Chappidi},
  year   = {2025},
  eprint = {2512.12072},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2512.12072v1}
}

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/2512.12072.svg)](https://codeofpaper.com/paper/2512.12072)

Discussion