Model collapse occurs in structured interactive learning if and only if the directed interaction graph satisfies a specific topological condition, with finite-sample guarantees for linear regression and asymptotic results for M-estimators.
arXiv preprint , volume =
7 Pith papers cite this work, alongside 9 external citations. Polarity classification is still indexing.
7
Pith papers citing it
9
external citations · external index
representative citing papers
TSGP uses a pre-trained transformer as a semantic variation operator in genetic programming, generalizes across d-dimensional problems, and outperforms standard GP, SLIM_GSGP, Deep Symbolic Regression, and Denoising Autoencoder GP on 24 datasets while producing more compact solutions.
Experiments across code LLMs show no-review collapses fastest, human-gated filters slow collapse, and AI self-gates lose effect over time, degenerating to ungated self-training under self-confirming acceptance as proven via gated distributional reweighting and spectral analysis.
Introduces a triangulation-based metric to quantify lexical shifts attributable to preference tuning without requiring manual curation of examples.
Iterated learning theory predicts and LLM experiments confirm non-monotonic compositionality during self-training, reframing model collapse as cultural transmission with matching human regularization patterns.
Self-training restructures language by amplifying surface markers and collapsing deep syntax according to structural depth rather than frequency, as evidenced by correlations across multiple models and a human fine-tuning control.
citing papers explorer
-
When Does Model Collapse Occur in Structured Interactive Learning?
Model collapse occurs in structured interactive learning if and only if the directed interaction graph satisfies a specific topological condition, with finite-sample guarantees for linear regression and asymptotic results for M-estimators.
-
Transformer Semantic Genetic Programming for d-dimensional Symbolic Regression Problems
TSGP uses a pre-trained transformer as a semantic variation operator in genetic programming, generalizes across d-dimensional problems, and outperforms standard GP, SLIM_GSGP, Deep Symbolic Regression, and Denoising Autoencoder GP on 24 datasets while producing more compact solutions.
-
When AI Reviews Its Own Code: Recursive Self-Training Collapse in Code LLMs
Experiments across code LLMs show no-review collapses fastest, human-gated filters slow collapse, and AI self-gates lose effect over time, degenerating to ungated self-training under self-confirming acceptance as proven via gated distributional reweighting and spectral analysis.
-
Isolating LLM Lexical Bias: A Curation-Free Triangulated Metric for Preference-Stage Learning
Introduces a triangulation-based metric to quantify lexical shifts attributable to preference tuning without requiring manual curation of examples.
-
Model Collapse as Cultural Evolution
Iterated learning theory predicts and LLM experiments confirm non-monotonic compositionality during self-training, reframing model collapse as cultural transmission with matching human regularization patterns.
-
Self-Training Doesn't Flatten Language -- It Restructures It: Surface Markers Amplify While Deep Syntax Dies
Self-training restructures language by amplifying surface markers and collapsing deep syntax according to structural depth rather than frequency, as evidenced by correlations across multiple models and a human fine-tuning control.
- The Economics of Model Collapse: Equilibrium, Welfare, and Optimal Provenance Subsidies in Synthetic Data Markets