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

arXiv: https://arxiv.org/abs/2506.19262 · HTML: https://mlsanity.com/paper/2506.19262/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Placed in LLM synthetic data generation, one of 13 directions in Synthetic data and self-improvement for LLMs.
- Builds on 38 of its 41 references that are in the map.
- First posted on arXiv 24 Jun 2025.

## Placed in

- Synthetic data and self-improvement for LLMs / [LLM synthetic data generation](https://mlsanity.com/synth/llm-synthetic-data-generation/)

## Builds on (most cited first, 30 of 38)

- [GPT-4 Technical Report](https://mlsanity.com/paper/2303.08774/) (2023)
- [The Llama 3 Herd of Models](https://mlsanity.com/paper/2407.21783/) (2024)
- [Self-Instruct: Aligning Language Models with Self-Generated Instructions](https://mlsanity.com/paper/2212.10560/) (2022)
- [A Diversity-Promoting Objective Function for Neural Conversation Models](https://mlsanity.com/paper/1510.03055/) (2015)
- [ChatGPT Outperforms Crowd-Workers for Text-Annotation Tasks](https://mlsanity.com/paper/2303.15056/) (2023)
- [Phi-4 Technical Report](https://mlsanity.com/paper/2412.08905/) (2024)
- [The NarrativeQA Reading Comprehension Challenge](https://mlsanity.com/paper/1712.07040/) (2017)
- [A Corpus and Evaluation Framework for Deeper Understanding of Commonsense Stories](https://mlsanity.com/paper/1604.01696/) (2016)
- [The Curse of Recursion: Training on Generated Data Makes Models Forget](https://mlsanity.com/paper/2305.17493/) (2023)
- [Large Language Model as Attributed Training Data Generator: A Tale of Diversity and Bias](https://mlsanity.com/paper/2306.15895/) (2023)
- [ZeroGen: Efficient Zero-shot Learning via Dataset Generation](https://mlsanity.com/paper/2202.07922/) (2022)
- [Is GPT-3 a Good Data Annotator?](https://mlsanity.com/paper/2212.10450/) (2022)
- [Generating Training Data with Language Models: Towards Zero-Shot Language Understanding](https://mlsanity.com/paper/2202.04538/) (2022)
- [GPT3Mix: Leveraging Large-scale Language Models for Text Augmentation](https://mlsanity.com/paper/2104.08826/) (2021)
- [Is Model Collapse Inevitable? Breaking the Curse of Recursion by Accumulating Real and Synthetic Data](https://mlsanity.com/paper/2404.01413/) (2024)
- [AugGPT: Leveraging ChatGPT for Text Data Augmentation](https://mlsanity.com/paper/2302.13007/) (2023)
- [A Tale of Tails: Model Collapse as a Change of Scaling Laws](https://mlsanity.com/paper/2402.07043/) (2024)
- [Evaluating the Evaluation of Diversity in Natural Language Generation](https://mlsanity.com/paper/2004.02990/) (2020)
- [Synthetic Data Generation with Large Language Models for Text Classification: Potential and Limitations](https://mlsanity.com/paper/2310.07849/) (2023)
- [The Curious Decline of Linguistic Diversity: Training Language Models on Synthetic Text](https://mlsanity.com/paper/2311.09807/) (2023)
- [Language GANs Falling Short](https://mlsanity.com/paper/1811.02549/) (2018)
- [Model Collapse Demystified: The Case of Regression](https://mlsanity.com/paper/2402.07712/) (2024)
- [Large Language Models Suffer From Their Own Output: An Analysis of the Self-Consuming Training Loop](https://mlsanity.com/paper/2311.16822/) (2023)
- [How Bad is Training on Synthetic Data? A Statistical Analysis of Language Model Collapse](https://mlsanity.com/paper/2404.05090/) (2024)
- [ProGen: Progressive Zero-shot Dataset Generation via In-context Feedback](https://mlsanity.com/paper/2210.12329/) (2022)
- [Strong Model Collapse](https://mlsanity.com/paper/2410.04840/) (2024)
- [Smaller, Weaker, Yet Better: Training LLM Reasoners via Compute-Optimal Sampling](https://mlsanity.com/paper/2408.16737/) (2024)
- [On the Diversity of Synthetic Data and its Impact on Training Large Language Models](https://mlsanity.com/paper/2410.15226/) (2024)
- [Collapse or Thrive? Perils and Promises of Synthetic Data in a Self-Generating World](https://mlsanity.com/paper/2410.16713/) (2024)
- [Beyond Model Collapse: Scaling Up with Synthesized Data Requires Verification](https://mlsanity.com/paper/2406.07515/) (2024)

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

- [Verification Limits Code LLM Training](https://mlsanity.com/paper/2509.20837/) (2025)

## Abstract (arXiv)

With the remarkable generative capabilities of large language models (LLMs), using LLM-generated data to train downstream models has emerged as a promising approach to mitigate data scarcity in specific domains and reduce time-consuming annotations. However, recent studies have highlighted a critical issue: iterative training on self-generated data results in model collapse, where model performance degrades over time. Despite extensive research on the implications of LLM-generated data, these works often neglect the importance of data diversity, a key factor in data quality. In this work, we aim to understand the implications of the diversity of LLM-generated data on downstream model performance. Specifically, we explore how varying levels of diversity in LLM-generated data affect downstream model performance. Additionally, we investigate the performance of models trained on data that mixes different proportions of LLM-generated data, which we refer to as synthetic data. Our experimental results show that, with minimal distribution shift, moderately diverse LLM-generated data can enhance model performance in scenarios with insufficient labeled data, whereas highly diverse generated data has a negative impact. We hope our empirical findings will offer valuable guidance for future studies on LLMs as data generators.