mlsanityData to 14 Oct 2026

On Support Samples of Next Word Prediction

arXiv 2506.04047 ↗First posted 4 Jun 2025cs.CLcited by 0

Not on the map. Why?

Abstract (arXiv)

Language models excel in various tasks by making complex decisions, yet understanding the rationale behind these decisions remains a challenge. This paper investigates data-centric interpretability in language models, focusing on the next-word prediction task. Using representer theorem, we identify two types of support samples-those that either promote or deter specific predictions. Our findings reveal that being a support sample is an intrinsic property, predictable even before training begins. Additionally, while non-support samples are less influential in direct predictions, they play a critical role in preventing overfitting and shaping generalization and representation learning. Notably, the importance of non-support samples increases in deeper layers, suggesting their significant role in intermediate representation formation. These insights shed light on the interplay between data and model decisions, offering a new dimension to understanding language model behavior and interpretability.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 14 of its 23 references that are in the map. First posted on arXiv 4 Jun 2025.

Builds on

14 of 23 references.

References · 14

  1. 2017Understanding Black-box Predictions via Influence Functionscited by 1,741
  2. 2018An Empirical Study of Example Forgetting during Deep Neural Network Learningcited by 678
  3. 2020Estimating Training Data Influence by Tracing Gradient Descentcited by 580
  4. 2021Deep Learning on a Data Diet: Finding Important Examples Early in Trainingcited by 533
  5. 2022Beyond neural scaling laws: beating power law scaling via data pruningcited by 528
  6. 2020Dataset Cartography: Mapping and Diagnosing Datasets with Training Dynamicscited by 425
  7. 2023Studying Large Language Model Generalization with Influence Functionscited by 326
  8. 2023TRAK: Attributing Model Behavior at Scalecited by 283
  9. 2020Influence Functions in Deep Learning Are Fragilecited by 252
  10. 2019On the Accuracy of Influence Functions for Measuring Group Effectscited by 147
  11. 2022If Influence Functions are the Answer, Then What is the Question?cited by 144
  12. 2020Characterizing Structural Regularities of Labeled Data in Overparameterized Modelscited by 93
2 more
  1. 2019Data Cleansing for Models Trained with SGDcited by 67
  2. 2024Dataset Difficulty and the Role of Inductive Biascited by 12

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2025TRACE: Training and Inference-Time Interpretability Analysis for Language Models
  2. 2026Is my model perplexed for the right reason? Contrasting LLMs' Benchmark Behavior with Token-Level Perplexity
  3. 2024Does Representation Matter? Exploring Intermediate Layers in Large Language Models
  4. 2026Scaling Inherently Interpretable Language Models
  5. 2026A Unifying Perspective on Language Model Representations: From Filler-Role Structure to Mechanistic Interpretability
  6. 2024Observational Scaling Laws and the Predictability of Language Model Performance
  7. 2026Learning to Read Out: Unembedding Dynamics in Language Model Pretraining
  8. 2024Towards Uncovering How Large Language Model Works: An Explainability Perspective