# NanoFlux: Adversarial Dual-LLM Evaluation and Distillation For Multi-Domain Reasoning

arXiv: https://arxiv.org/abs/2509.23252 · HTML: https://mlsanity.com/paper/2509.23252/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Placed in Reasoning benchmarks and evaluation, one of 17 directions in LLM reasoning.
- Builds on 13 of its 31 references that are in the map.
- First posted on arXiv 27 Sep 2025.

## Placed in

- LLM reasoning / [Reasoning benchmarks and evaluation](https://mlsanity.com/lr/reasoning-benchmarks-and-evaluation/)

## Builds on (most cited first, 13 of 13)

- [Training Verifiers to Solve Math Word Problems](https://mlsanity.com/paper/2110.14168/) (2021)
- [Gemma 3 Technical Report](https://mlsanity.com/paper/2503.19786/) (2025)
- [PAL: Program-aided Language Models](https://mlsanity.com/paper/2211.10435/) (2022)
- [LIMO: Less is More for Reasoning](https://mlsanity.com/paper/2502.03387/) (2025)
- [MedGemma Technical Report](https://mlsanity.com/paper/2507.05201/) (2025)
- [R-Zero: Self-Evolving Reasoning LLM from Zero Data](https://mlsanity.com/paper/2508.05004/) (2025)
- [LLMs Can Easily Learn to Reason from Demonstrations Structure, not content, is what matters!](https://mlsanity.com/paper/2502.07374/) (2025)
- [Self-Questioning Language Models](https://mlsanity.com/paper/2508.03682/) (2025)
- [Straight to Zero: Why Linearly Decaying the Learning Rate to Zero Works Best for LLMs](https://mlsanity.com/paper/2502.15938/) (2025)
- [MedFuzz: Exploring the Robustness of Large Language Models in Medical Question Answering](https://mlsanity.com/paper/2406.06573/) (2024)
- [Adversarial Math Word Problem Generation](https://mlsanity.com/paper/2402.17916/) (2024)
- [VAULT: Vigilant Adversarial Updates via LLM-Driven Retrieval-Augmented Generation for NLI](https://mlsanity.com/paper/2508.00965/) (2025)
- [PLoRA: Efficient Concurrent LoRA Training for Large Language Models](https://mlsanity.com/paper/2508.02932/) (2025)

## Built on by (most cited first, 1 of 1)

- [FORT-Searcher: Synthesizing Shortcut-Resistant Search Tasks for Training Deep Search Agents](https://mlsanity.com/paper/2606.12087/) (2026)

## Abstract (arXiv)

We present NanoFlux, a novel adversarial framework for generating targeted training data to improve LLM reasoning, where adversarially-generated datasets containing fewer than 200 examples outperform conventional fine-tuning approaches. The framework employs a competitive dynamic between models alternating as Attacker and Defender, supervised by a tool-augmented Judge, synthesizing multi-step questions with explanatory annotations that target specific reasoning capabilities. Fine-tuning a 4B-parameter model on NanoFlux-generated data yields performance gains across diverse domains compared to full-benchmark fine-tuning: +5.9% on mathematical reasoning (GSMHard), +3.6% on scientific reasoning (GenomeBench), and +16.6% on medical reasoning (MultiMedQA), while reducing computational requirements by 3-14x. Ablation studies reveal a non-monotonic relationship between dataset characteristics and model performance, uncovering domain-specific optimal points for question complexity and reasoning quality. NanoFlux automates training data generation through embedding-based novelty filtering, tool-augmented evaluation, and multi-hop reasoning, suggesting that future model improvements may lie in the intelligent synthesis of small, precisely targeted training datasets.