Learning from Synthetic Data without Model Collapse in Iterative Instruction Tuning

• URL: http://arxiv.org/abs/2607.17043v1
• Date: Sun, 19 Jul 2026 03:20:01 GMT
• Title: Learning from Synthetic Data without Model Collapse in Iterative Instruction Tuning
• Abstract summary: We study the problem in synthetic data self-improving for instruction tuning.<n>We propose KITE (Knowledge-boundary Instruction Tuning via Exploration), a two-stage framework that combines failure-guided data generation with boundary-aware uncertainty curation.
• Score: 38.36757975609017
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Model collapse is a central challenge in learning from synthetic data: as later-generation large language models (LLMs) are trained on an increasing proportion of model-generated data, performance can degrade due to narrowed coverage and accumulated bias. Existing work mainly studies how to bound this degradation. In iterative model evolution, however, the more meaningful objective is to ensure that each successive model improves over its predecessor, which requires diagnosing collapse at a granularity that is actionable for data curation. We study this problem in synthetic data self-improving for instruction tuning. We show that collapse in this setting is not simply uniform performance degradation, but can appear as a polarization of competence, where synthetic training reinforces already strong skills while further degrading weak ones. Motivated by this observation, we propose KITE (Knowledge-boundary Instruction Tuning via Exploration), a two-stage framework that combines failure-guided data generation with boundary-aware uncertainty curation. Experiments across several datasets and multiple open-source LLMs show that KITE yields more stable improvement than strong synthetic-data baselines.

Related papers

• Model Inversion with Layer-Specific Modeling and Alignment for Data-Free Continual Learning [19.12792297140574]
  Continual learning aims to incrementally train a model on a sequence of tasks while retaining performance on prior ones.<n> storing and replaying data is often infeasible due to privacy or security constraints.<n>We propose Per-layer Model Inversion (PMI), inspired by faster convergence in single-layer optimization.
  arXiv  Detail & Related papers  (2025-10-30T09:58:48Z)
• Escaping Model Collapse via Synthetic Data Verification: Near-term Improvements and Long-term Convergence [31.751930228965467]
  We investigate ways to modify this synthetic retraining process to avoid model collapse.<n>Our key finding is that by injecting information through an external synthetic data verifier, synthetic retraining will not cause model collapse.
  arXiv  Detail & Related papers  (2025-10-18T22:39:39Z)
• Learning Robust Diffusion Models from Imprecise Supervision [75.53546939251146]
  DMIS is a unified framework for training robust Conditional Diffusion Models from Imprecise Supervision.<n>Our framework is derived from likelihood and decomposes the objective into generative and classification components.<n>Experiments on diverse forms of imprecise supervision, covering tasks covering image generation, weakly supervised learning, and dataset condensation demonstrate that DMIS consistently produces high-quality and class-discriminative samples.
  arXiv  Detail & Related papers  (2025-10-03T14:00:32Z)
• ForTIFAI: Fending Off Recursive Training Induced Failure for AI Model Collapse [13.096745830570944]
  We introduce the Truncated-Cross-Entropy (TCE) loss function to mitigate model collapse in synthetic data.<n>TCE mitigates collapse by selectively ignoring high-confidence tokens during training, effectively filtering out likely machine-generated artifacts.<n>Our results demonstrate that confidence-aware training objectives can substantially delay collapse onset, offering a practical and generalizable tool for model robustness under synthetic-data exposure.
  arXiv  Detail & Related papers  (2025-09-10T20:06:51Z)
• Data Curation Matters: Model Collapse and Spurious Shift Performance Prediction from Training on Uncurated Text Embeddings [0.0]
  Training models on uncurated Text Embeddings (TEs) can lead to a severe failure mode known as model collapse.<n>We introduce a set of metrics that capture the extent of model collapse, offering a new perspective on TE quality as a proxy for data curation.<n>These findings highlight the need for more nuanced curation and evaluation of embedding-based representations.
  arXiv  Detail & Related papers  (2025-06-22T11:01:41Z)
• EpiCoDe: Boosting Model Performance Beyond Training with Extrapolation and Contrastive Decoding [50.29046178980637]
  EpiCoDe is a method that boosts model performance in data-scarcity scenarios without extra training.<n>We show that EpiCoDe consistently outperforms existing methods with significant and robust improvement.
  arXiv  Detail & Related papers  (2025-06-04T02:11:54Z)
• A Theoretical Perspective: How to Prevent Model Collapse in Self-consuming Training Loops [55.07063067759609]
  High-quality data is essential for training large generative models, yet the vast reservoir of real data available online has become nearly depleted.<n>Models increasingly generate their own data for further training, forming Self-consuming Training Loops (STLs)<n>Some models degrade or even collapse, while others successfully avoid these failures, leaving a significant gap in theoretical understanding.
  arXiv  Detail & Related papers  (2025-02-26T06:18:13Z)
• Unveiling the Flaws: Exploring Imperfections in Synthetic Data and Mitigation Strategies for Large Language Models [89.88010750772413]
  Synthetic data has been proposed as a solution to address the issue of high-quality data scarcity in the training of large language models (LLMs) Our work delves into these specific flaws associated with question-answer (Q-A) pairs, a prevalent type of synthetic data, and presents a method based on unlearning techniques to mitigate these flaws. Our work has yielded key insights into the effective use of synthetic data, aiming to promote more robust and efficient LLM training.
  arXiv  Detail & Related papers  (2024-06-18T08:38:59Z)
• On the Stability of Iterative Retraining of Generative Models on their own Data [56.153542044045224]
  We study the impact of training generative models on mixed datasets. We first prove the stability of iterative training under the condition that the initial generative models approximate the data distribution well enough. We empirically validate our theory on both synthetic and natural images by iteratively training normalizing flows and state-of-the-art diffusion models.
  arXiv  Detail & Related papers  (2023-09-30T16:41:04Z)
• Generative Modeling Helps Weak Supervision (and Vice Versa) [87.62271390571837]
  We propose a model fusing weak supervision and generative adversarial networks. It captures discrete variables in the data alongside the weak supervision derived label estimate. It is the first approach to enable data augmentation through weakly supervised synthetic images and pseudolabels.
  arXiv  Detail & Related papers  (2022-03-22T20:24:21Z)
• Contrastive Model Inversion for Data-Free Knowledge Distillation [60.08025054715192]
  We propose Contrastive Model Inversion, where the data diversity is explicitly modeled as an optimizable objective. Our main observation is that, under the constraint of the same amount of data, higher data diversity usually indicates stronger instance discrimination. Experiments on CIFAR-10, CIFAR-100, and Tiny-ImageNet demonstrate that CMI achieves significantly superior performance when the generated data are used for knowledge distillation.
  arXiv  Detail & Related papers  (2021-05-18T15:13:00Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.