Skip to content
AI Atlas
PaperActive

Data Scarcity and Model Sparsity: Mixtures-of-Experts Overfit More to Repeated Data

arxiv.org/abs/2609.11917

Updated 2 h ago · first seen 11 Sept 2026

paper_01M294FPJHCWSE05SGFZ6ZK7NC

Published
11 Sept 2026
T1 · 2 h ago
arXiv
2609.11917
T1 · 2 h ago
Category
cs.LG
T1 · 2 h ago

As of

Rewind the record: see this entity's attributes exactly as AI Atlas knew them on a given day.

Claim history

10 claims · 9 properties

Official pageofficial_url1

Claim history for Official page
ValueValid from → toStatusSourceConfidenceExtractor
https://arxiv.org/abs/2609.11917currentcurrentarXiv (Atom API + RSS)T1highdeterministic

Abstractabstract1

Claim history for Abstract
ValueValid from → toStatusSourceConfidenceExtractor
As the supply of human-written text is exhausted, it has become standard practice to repeat language model training data. Prior work has studied data repetition for densely activated Transformers, but the effects of data repetition remains largely unexplored for recently dominant sparse architectures such as Mixture-of-Experts (MoE), despite their increased compute efficiency. We vary data repetition rates across single- and multi-domain data mixes, and across MoE settings, including expert count and granularity. We consistently find, for models ranging from 80M to 1B active (8.5B total) parameters, that MoEs degrade more rapidly under data repetition. This effect increases with sparsity, dictated by total rather than active parameters. While 80M dense models can repeat data over 8x with minimal degradation, MoEs instead begin to suffer at 4x, and deteriorate rapidly, ceding their performance benefits in all-unique data settings to underperform dense models after 32x. We experiment with existing regularization methods as a potential remedy. We find that some methods, such as dropout, can mitigate overfitting. In particular, with strong masking-based regularization, MoEs are able to outperform dense models even when data is repeated more than 64 times. However, no method fully matches the performance of all-unique training data. Finally, we analyze internal mechanisms correlated with MoE overfitting in high repetition regimes, and find that MoE routing universally stabilizes early in training, and that expert specialization correlates with overfitting to repeated data. In sum, our work addresses the adverse interactions between sparsity and data repetition: we present evidence for the core mechanisms of overfitting and its potential remediation, and suggest promising avenues for future methods to reduce over-specialization in model parameters by disrupting memorization patterns.currentcurrentarXiv (Atom API + RSS)T1highdeterministic

Arxiv announce typearxiv_announce_type2

Claim history for Arxiv announce type
ValueValid from → toStatusSourceConfidenceExtractor
crosscurrentcurrentarXiv (Atom API + RSS)T1highdeterministic
newsupersededarXiv (Atom API + RSS)T1highdeterministic

arXiv idarxiv_id1

Claim history for arXiv id
ValueValid from → toStatusSourceConfidenceExtractor
2609.11917currentcurrentarXiv (Atom API + RSS)T1highdeterministic

Authorsauthors1

Claim history for Authors
ValueValid from → toStatusSourceConfidenceExtractor
Atindra Jha, Margaret Li, Jure Leskovec, Percy Liang, Luke ZettlemoyercurrentcurrentarXiv (Atom API + RSS)T1highdeterministic

Categoriescategories1

Claim history for Categories
ValueValid from → toStatusSourceConfidenceExtractor
cs.LG, cs.CLcurrentcurrentarXiv (Atom API + RSS)T1highdeterministic

PDFpdf_url1

Claim history for PDF
ValueValid from → toStatusSourceConfidenceExtractor
https://arxiv.org/pdf/2609.11917currentcurrentarXiv (Atom API + RSS)T1highdeterministic

Primary categoryprimary_category1

Claim history for Primary category
ValueValid from → toStatusSourceConfidenceExtractor
cs.LGcurrentcurrentarXiv (Atom API + RSS)T1highdeterministic

Publishedpublished_at1

Claim history for Published
ValueValid from → toStatusSourceConfidenceExtractor
11 Sept 2026currentcurrentarXiv (Atom API + RSS)T1highdeterministic

Claims are temporal and append-only: a new observation closes the previous claim (valid_to) instead of overwriting it. Conflicting claims from different sources are kept side by side and flagged — never averaged. Methodology →