Research map / Probabilistic and scientific modeling / Variational and generative sampling
Escaping Model Collapse via Synthetic Data Verification: Near-term Improvements and Long-term Convergence
Bingji Yi, Qiyuan Liu, Yuwei Cheng, Haifeng Xu
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add verifier filtering to synthetic retraining (ablation) | CVAE retraining · MNIST CVAE | FID: best synthetic model reaches 21.17 FID after 40 iterations vs 17.56 for a 60K real-data CVAE | Improved |
| use a stronger verifier trained on more real data (ablation) | CVAE retraining · MNIST CVAE (20K synthetic samples per round) | FID | Improved |
| add verifier filtering to iterative retraining (ablation) | SmolLM2-135M · XSUM | ROUGE-1 | Improved |
| make the verifier more selective (smaller radius r) (ablation) | linear regression retraining · simulated linear regression, biased verifier | convergence speed | Improved |
| use a biased verifier instead of an unbiased one (ablation) | linear regression retraining · simulated linear regression | estimator error | Hurt |
| add verifier filtering with an unbiased verifier (ablation) | linear regression retraining · simulated linear regression | estimation error | Improved |
Closest papers by meaning
- Collapse or Thrive: Perils and Promises of Synthetic Data in a Self-Generating World (ICML 2025 · similarity 0.93)
- A Theoretical Perspective: How to Prevent Model Collapse in Self-consuming Training Loops (ICLR 2025 · similarity 0.91)
- When Models Don’t Collapse: On the Consistency of Iterative MLE (NeurIPS 2025 · similarity 0.90)
- On the Stability of Iterative Retraining of Generative Models on their own Data (ICLR 2024 · similarity 0.90)
- A Closer Look at Model Collapse: From a Generalization-to-Memorization Perspective (NeurIPS 2025 · similarity 0.90)
- Learning with Synthetic Data via SGD in High-Dimensional Linear Regression (NeurIPS 2026 · similarity 0.90)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
