Influence Dynamics and Stagewise Data Attribution

• URL: http://arxiv.org/abs/2510.12071v1
• Date: Tue, 14 Oct 2025 02:21:23 GMT
• Title: Influence Dynamics and Stagewise Data Attribution
• Abstract summary: We introduce a framework for stagewise data attribution grounded in singular learning theory.<n>We predict that influence can change non-monotonically, including sign flips and sharp peaks at developmental transitions.
• Score: 5.002658006152838
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Current training data attribution (TDA) methods treat the influence one sample has on another as static, but neural networks learn in distinct stages that exhibit changing patterns of influence. In this work, we introduce a framework for stagewise data attribution grounded in singular learning theory. We predict that influence can change non-monotonically, including sign flips and sharp peaks at developmental transitions. We first validate these predictions analytically and empirically in a toy model, showing that dynamic shifts in influence directly map to the model's progressive learning of a semantic hierarchy. Finally, we demonstrate these phenomena at scale in language models, where token-level influence changes align with known developmental stages.

Related papers

• Interpreting learning dynamics of autoencoders: Transient scaling and emerging concepts of the Ising model [0.0]
  unsupervised autoencoders trained on microscopic spin configurations from the Ising model learn macroscopic, theory-relevant variables.<n>We quantify learning across multiple spatial scales and reveal two distinct dynamical regimes.<n>A dynamical viewpoint of learning is established in which intrinsic properties expose the effects of forced changes in representation.
  arXiv  Detail & Related papers  (2026-07-11T12:31:34Z)
• Reasoning as State Transition: A Representational Analysis of Reasoning Evolution in Large Language Models [50.39102836928242]
  We introduce a representational perspective to investigate the dynamics of the model's internal states.<n>We discover that post-training yields only limited improvement in static initial representation quality.
  arXiv  Detail & Related papers  (2026-01-31T15:23:33Z)
• Belief Dynamics Reveal the Dual Nature of In-Context Learning and Activation Steering [22.666436755894328]
  Large language models (LLMs) can be controlled at inference time through prompts (in-context learning) and internal activations (activation steering)<n>This work offers a unified account of prompt-based and activation-based control of LLM behavior, and a methodology for empirically predicting the effects of these interventions.
  arXiv  Detail & Related papers  (2025-11-01T16:46:03Z)
• What Do Latent Action Models Actually Learn? [26.1770533810057]
  Latent action models (LAMs) aim to learn action-relevant changes from unlabeled videos by compressing changes between frames as latents.<n>This paper studies this issue analytically, presenting a linear model that encapsulates the essence of LAM learning, while being tractable.
  arXiv  Detail & Related papers  (2025-05-27T02:45:30Z)
• On the Cone Effect in the Learning Dynamics [57.02319387815831]
  We take an empirical perspective to study the learning dynamics of neural networks in real-world settings.<n>Our key findings reveal a two-phase learning process: i) in Phase I, the eNTK evolves significantly, signaling the rich regime, and ii) in Phase II, the eNTK keeps evolving but is constrained in a narrow space.
  arXiv  Detail & Related papers  (2025-03-20T16:38:25Z)
• Latent State Models of Training Dynamics [51.88132043461152]
  We train models with different random seeds and compute a variety of metrics throughout training. We then fit a hidden Markov model (HMM) over the resulting sequences of metrics. We use the HMM representation to study phase transitions and identify latent "detour" states that slow down convergence.
  arXiv  Detail & Related papers  (2023-08-18T13:20:08Z)
• Loss Dynamics of Temporal Difference Reinforcement Learning [36.772501199987076]
  We study the case learning curves for temporal difference learning of a value function with linear function approximators. We study how learning dynamics and plateaus depend on feature structure, learning rate, discount factor, and reward function.
  arXiv  Detail & Related papers  (2023-07-10T18:17:50Z)
• Towards Robust and Adaptive Motion Forecasting: A Causal Representation Perspective [72.55093886515824]
  We introduce a causal formalism of motion forecasting, which casts the problem as a dynamic process with three groups of latent variables. We devise a modular architecture that factorizes the representations of invariant mechanisms and style confounders to approximate a causal graph. Experiment results on synthetic and real datasets show that our three proposed components significantly improve the robustness and reusability of the learned motion representations.
  arXiv  Detail & Related papers  (2021-11-29T18:59:09Z)
• The Grammar-Learning Trajectories of Neural Language Models [42.32479280480742]
  We show that neural language models acquire linguistic phenomena in a similar order, despite having different end performances over the data. Results suggest that NLMs exhibit consistent developmental'' stages.
  arXiv  Detail & Related papers  (2021-09-13T16:17:23Z)
• Learning Neural Models for Natural Language Processing in the Face of Distributional Shift [10.990447273771592]
  The dominating NLP paradigm of training a strong neural predictor to perform one task on a specific dataset has led to state-of-the-art performance in a variety of applications. It builds upon the assumption that the data distribution is stationary, ie. that the data is sampled from a fixed distribution both at training and test time. This way of training is inconsistent with how we as humans are able to learn from and operate within a constantly changing stream of information. It is ill-adapted to real-world use cases where the data distribution is expected to shift over the course of a model's lifetime
  arXiv  Detail & Related papers  (2021-09-03T14:29:20Z)
• Efficient Estimation of Influence of a Training Instance [56.29080605123304]
  We propose an efficient method for estimating the influence of a training instance on a neural network model. Our method is inspired by dropout, which zero-masks a sub-network and prevents the sub-network from learning each training instance. We demonstrate that the proposed method can capture training influences, enhance the interpretability of error predictions, and cleanse the training dataset for improving generalization.
  arXiv  Detail & Related papers  (2020-12-08T04:31:38Z)
• Multi-Stage Influence Function [97.19210942277354]
  We develop a multi-stage influence function score to track predictions from a finetuned model all the way back to the pretraining data. We study two different scenarios with the pretrained embeddings fixed or updated in the finetuning tasks.
  arXiv  Detail & Related papers  (2020-07-17T16:03:11Z)

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.