# Benchmark Shadows: Data Alignment, Parameter Footprints, and Generalization in Large Language Models

arXiv: https://arxiv.org/abs/2604.07363 · HTML: https://mlsanity.com/paper/2604.07363/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 16 of its 20 references that are in the map.
- First posted on arXiv 1 Apr 2026.

## Builds on (most cited first, 16 of 16)

- [Scaling Up Visual and Vision-Language Representation Learning With Noisy Text Supervision](https://mlsanity.com/paper/2102.05918/) (2021)
- [Training Compute-Optimal Large Language Models](https://mlsanity.com/paper/2203.15556/) (2022)
- [Reproducible scaling laws for contrastive language-image learning](https://mlsanity.com/paper/2212.07143/) (2022)
- [Implicit Self-Regularization in Deep Neural Networks: Evidence from Random Matrix Theory and Implications for Learning](https://mlsanity.com/paper/1810.01075/) (2018)
- [Scaling Laws and Interpretability of Learning from Repeated Data](https://mlsanity.com/paper/2205.10487/) (2022)
- [Scaling Laws for Generative Mixed-Modal Language Models](https://mlsanity.com/paper/2301.03728/) (2023)
- [Traditional and Heavy-Tailed Self Regularization in Neural Network Models](https://mlsanity.com/paper/1901.08276/) (2019)
- [To Repeat or Not To Repeat: Insights from Scaling LLM under Token-Crisis](https://mlsanity.com/paper/2305.13230/) (2023)
- [Scalable Vision Language Model Training via High Quality Data Curation](https://mlsanity.com/paper/2501.05952/) (2025)
- [A Neural Scaling Law from the Dimension of the Data Manifold](https://mlsanity.com/paper/2004.10802/) (2020)
- [Learn from Downstream and Be Yourself in Multimodal Large Language Model Fine-Tuning](https://mlsanity.com/paper/2411.10928/) (2024)
- [Impact of classification difficulty on the weight matrices spectra in Deep Learning and application to early-stopping](https://mlsanity.com/paper/2111.13331/) (2021)
- [SETOL: A Semi-Empirical Theory of (Deep) Learning](https://mlsanity.com/paper/2507.17912/) (2025)
- [Both Text and Images Leaked! A Systematic Analysis of Data Contamination in Multimodal LLM](https://mlsanity.com/paper/2411.03823/) (2024)
- [Heavy-Tailed Regularization of Weight Matrices in Deep Neural Networks](https://mlsanity.com/paper/2304.02911/) (2023)
- [LongTail-Swap: benchmarking language models' abilities on rare words](https://mlsanity.com/paper/2510.04268/) (2025)

## Abstract (arXiv)

Large language models often achieve strong benchmark gains without corresponding improvements in broader capability. We hypothesize that this discrepancy arises from differences in training regimes induced by data distribution. To investigate this, we design controlled data interventions that isolate distributional effects under fixed training settings. We find that benchmark-aligned data improves narrow evaluation metrics while limiting broader representational development, whereas coverage-expanding data leads to more distributed parameter adaptation and better generalization. We further introduce parameter-space diagnostics based on spectral and rank analyses, which reveal distinct structural signatures of these regimes. Similar patterns are observed across diverse open-source model families, including multimodal models as a key case study, suggesting that these effects extend beyond controlled settings. A case study on prompt repetition shows that not all data artifacts induce regime shifts. These results indicate that benchmark performance alone is insufficient to characterize model capability, and highlight the importance of data distribution in shaping learning dynamics.