Saved in:
Bibliographic Details
Main Authors: Garg, Ishir, Kolhe, Neel, Zhao, Xuandong, Song, Dawn
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2601.00575
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911717404966912
author Garg, Ishir
Kolhe, Neel
Zhao, Xuandong
Song, Dawn
author_facet Garg, Ishir
Kolhe, Neel
Zhao, Xuandong
Song, Dawn
contents Large language models (LLMs) have demonstrated significant advancements in reasoning and code generation, but efficiently creating new benchmarks to evaluate these capabilities remains a challenge. Traditional benchmark creation relies on manual human effort, which is expensive and time-consuming. Furthermore, existing benchmarks often contaminate LLM training data, necessitating novel and diverse benchmarks to accurately assess their genuine capabilities. This work introduces InfoSynth, a novel framework for automatically generating and evaluating reasoning benchmarks guided by information-theoretic principles. We propose metrics based on KL-divergence and entropy to quantify benchmark novelty and diversity without relying on costly model evaluations. Building on this framework, we develop an end-to-end pipeline that synthesizes robust Python coding problems from seed datasets using genetic algorithms and iterative code feedback. Our method generates accurate test cases and solutions to new problems 97% of the time, and the synthesized benchmarks consistently exhibit higher difficulty compared to prior works. Moreover, our algorithm provides a method for controlling the novelty/diversity and difficulty of generated problems. InfoSynth offers a scalable, self-verifying pipeline for constructing high-quality, challenging coding benchmarks for LLMs. Project Page: https://ishirgarg.github.io/infosynth_web/
format Preprint
id arxiv_https___arxiv_org_abs_2601_00575
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle InfoSynth: Information-Guided Benchmark Synthesis for LLMs
Garg, Ishir
Kolhe, Neel
Zhao, Xuandong
Song, Dawn
Computation and Language
Large language models (LLMs) have demonstrated significant advancements in reasoning and code generation, but efficiently creating new benchmarks to evaluate these capabilities remains a challenge. Traditional benchmark creation relies on manual human effort, which is expensive and time-consuming. Furthermore, existing benchmarks often contaminate LLM training data, necessitating novel and diverse benchmarks to accurately assess their genuine capabilities. This work introduces InfoSynth, a novel framework for automatically generating and evaluating reasoning benchmarks guided by information-theoretic principles. We propose metrics based on KL-divergence and entropy to quantify benchmark novelty and diversity without relying on costly model evaluations. Building on this framework, we develop an end-to-end pipeline that synthesizes robust Python coding problems from seed datasets using genetic algorithms and iterative code feedback. Our method generates accurate test cases and solutions to new problems 97% of the time, and the synthesized benchmarks consistently exhibit higher difficulty compared to prior works. Moreover, our algorithm provides a method for controlling the novelty/diversity and difficulty of generated problems. InfoSynth offers a scalable, self-verifying pipeline for constructing high-quality, challenging coding benchmarks for LLMs. Project Page: https://ishirgarg.github.io/infosynth_web/
title InfoSynth: Information-Guided Benchmark Synthesis for LLMs
topic Computation and Language
url https://arxiv.org/abs/2601.00575