Escaping Model Collapse via Synthetic Data Verification: Near-term Improvements and Long-term Convergence

2025/10/18 by Bingji Yi, Yi, Bingji, Qiyuan Liu +5
Computer Science · Physics and Astronomy · #Face recognition and analysis #Generative Adversarial Networks and Image Synthesis #Model Reduction and Neural Networks #cs.LG #stat.ML

paper · pdf · doi:10.48550/arxiv.2510.16657

openalex publication_date 2025/10/18 · openalex created_date 2025/10/22 · openalex updated_date 2026/07/28

Abstract

Synthetic data has been increasingly used to train frontier generative models. However, recent studies raise key concerns that iteratively retraining a generative model on its self-generated synthetic data may keep deteriorating model performance, a phenomenon often coined model collapse. In this paper, we investigate ways to modify the synthetic retraining process to avoid model collapse, and even possibly help reverse the trend from collapse to improvement. Our key finding is that by injecting information through an external synthetic data verifier, whether a human or a better model, synthetic retraining will not cause model collapse. Specifically, we situate our theoretical analysis in the fundamental linear regression setting, showing that verifier-guided retraining can yield near-term improvements, but ultimately drives the parameter estimate to the verifier's "knowledge center" in the long run. Our theory further predicts that, unless the verifier is perfectly reliable, these early gains will plateau and may even reverse. Indeed, our experiments across linear regression, Variational Autoencoders (VAEs) trained on MNIST, and fining-tuning SmolLM2-135M on the XSUM task confirm these theoretical insights.

Citations

• DeepSeek-R1 incentivizes reasoning in LLMs through reinforcement learning
• AI models collapse when trained on recursively generated data
• Is Model Collapse Inevitable? Breaking the Curse of Recursion by Accumulating Real and Synthetic Data
• Self-Consuming Generative Models Go MAD
• Textbooks Are All You Need
• Training language models to follow instructions with human feedback
• Learning from Simulated and Unsupervised Images through Adversarial\n Training

Related

• Collapse or Thrive? Perils and Promises of Synthetic Data in a Self-Generating World
• Is Model Collapse Inevitable? Breaking the Curse of Recursion by Accumulating Real and Synthetic Data
• Learning from Synthetic Data without Model Collapse in Iterative Instruction Tuning
• When Models Don't Collapse: On the Consistency of Iterative MLE
• ForTIFAI: Fending Off Recursive Training Induced Failure for AI Model Collapse