GraSS: Scalable Data Attribution with Gradient Sparsification and Sparse Projection

• URL: http://arxiv.org/abs/2505.18976v3
• Date: Tue, 28 Oct 2025 15:46:13 GMT
• Title: GraSS: Scalable Data Attribution with Gradient Sparsification and Sparse Projection
• Authors: Pingbang Hu, Joseph Melkonian, Weijing Tang, Han Zhao, Jiaqi W. Ma,
• Abstract summary: gradient-based data attribution methods, such as influence functions, are critical for understanding the impact of individual training samples without requiring repeated model retraining.<n>GraSS, a novel gradient compression algorithm, and its variants FactGraSS for linear layers specifically, leverage the inherent sparsity of per-sample gradients to achieve sub-linear space and time complexity.<n>In particular, FactGraSS achieves up to 165% faster throughput on billion-scale models compared to the previous state-of-the-art baselines.
• Score: 11.516230371831774
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Gradient-based data attribution methods, such as influence functions, are critical for understanding the impact of individual training samples without requiring repeated model retraining. However, their scalability is often limited by the high computational and memory costs associated with per-sample gradient computation. In this work, we propose GraSS, a novel gradient compression algorithm and its variants FactGraSS for linear layers specifically, that explicitly leverage the inherent sparsity of per-sample gradients to achieve sub-linear space and time complexity. Extensive experiments demonstrate the effectiveness of our approach, achieving substantial speedups while preserving data influence fidelity. In particular, FactGraSS achieves up to 165% faster throughput on billion-scale models compared to the previous state-of-the-art baselines. Our code is publicly available at https://github.com/TRAIS-Lab/GraSS.

Related papers

• GradientSpace: Unsupervised Data Clustering for Improved Instruction Tuning [13.559381851907778]
  GradientSpace is a framework that clusters samples directly in full-dimensional gradient space.<n>We introduce an online SVD-based algorithm that operates on LoRA gradients to identify latent skills without the infeasible cost of storing all sample gradients.<n>We show that routing to a single, appropriate expert outperforms expert ensembles used in prior work, while significantly reducing inference latency.
  arXiv  Detail & Related papers  (2025-12-07T06:35:04Z)
• Nonparametric Data Attribution for Diffusion Models [57.820618036556084]
  Data attribution for generative models seeks to quantify the influence of individual training examples on model outputs.<n>We propose a nonparametric attribution method that operates entirely on data, measuring influence via patch-level similarity between generated and training images.
  arXiv  Detail & Related papers  (2025-10-16T03:37:16Z)
• Linearly Convergent Mixup Learning [0.0]
  We present two novel algorithms that extend to a broader range of binary classification models.<n>Unlike gradient-based approaches, our algorithms do not require hyper parameters like learning rates, simplifying their implementation and optimization.<n>Our algorithms achieve faster convergence to the optimal solution compared to descent gradient approaches, and that mixup data augmentation consistently improves the predictive performance across various loss functions.
  arXiv  Detail & Related papers  (2025-01-14T02:33:40Z)
• Zeroth-Order Fine-Tuning of LLMs in Random Subspaces [63.10833446782114]
  As language models grow in size, memory demands for backpropagation increase.<n>Zeroth-order (ZO) optimization methods offer a memory-efficient alternative.<n>In this paper, we propose Subspace Zero-order optimization to address the challenges posed by posed by high dimensionality perturbations.
  arXiv  Detail & Related papers  (2024-10-11T17:01:43Z)
• Language Models as Zero-shot Lossless Gradient Compressors: Towards General Neural Parameter Prior Models [56.00251589760559]
  Large language models (LLMs) can act as gradient priors in a zero-shot setting.<n>We introduce LM-GC, a novel method that integrates LLMs with arithmetic coding.<n>Experiments indicate that LM-GC surpasses existing state-of-the-art lossless compression methods.
  arXiv  Detail & Related papers  (2024-09-26T13:38:33Z)
• R-CONV: An Analytical Approach for Efficient Data Reconstruction via Convolutional Gradients [40.209183669098735]
  This paper introduces an advanced data leakage method to efficiently exploit convolutional layers' gradients. To the best of our knowledge, this is the first analytical approach that successfully reconstructs convolutional layer inputs directly from the gradients.
  arXiv  Detail & Related papers  (2024-06-06T16:28:04Z)
• Exploring Learning Complexity for Efficient Downstream Dataset Pruning [8.990878450631596]
  Existing dataset pruning methods require training on the entire dataset.<n>We propose a straightforward, novel, and training-free hardness score named Distorting-based Learning Complexity (DLC)<n>Our method is motivated by the observation that easy samples learned faster can also be learned with fewer parameters.
  arXiv  Detail & Related papers  (2024-02-08T02:29:33Z)
• Online Importance Sampling for Stochastic Gradient Optimization [33.42221341526944]
  We propose a practical algorithm that efficiently computes data importance on-the-fly during training.<n>We also introduce a novel metric based on the derivative of the loss w.r.t. the network output, designed for mini-batch importance sampling.
  arXiv  Detail & Related papers  (2023-11-24T13:21:35Z)
• Towards Memory- and Time-Efficient Backpropagation for Training Spiking Neural Networks [70.75043144299168]
  Spiking Neural Networks (SNNs) are promising energy-efficient models for neuromorphic computing. We propose the Spatial Learning Through Time (SLTT) method that can achieve high performance while greatly improving training efficiency. Our method achieves state-of-the-art accuracy on ImageNet, while the memory cost and training time are reduced by more than 70% and 50%, respectively, compared with BPTT.
  arXiv  Detail & Related papers  (2023-02-28T05:01:01Z)
• ScoreMix: A Scalable Augmentation Strategy for Training GANs with Limited Data [93.06336507035486]
  Generative Adversarial Networks (GANs) typically suffer from overfitting when limited training data is available. We present ScoreMix, a novel and scalable data augmentation approach for various image synthesis tasks.
  arXiv  Detail & Related papers  (2022-10-27T02:55:15Z)
• Condensing Graphs via One-Step Gradient Matching [50.07587238142548]
  We propose a one-step gradient matching scheme, which performs gradient matching for only one single step without training the network weights. Our theoretical analysis shows this strategy can generate synthetic graphs that lead to lower classification loss on real graphs. In particular, we are able to reduce the dataset size by 90% while approximating up to 98% of the original performance.
  arXiv  Detail & Related papers  (2022-06-15T18:20:01Z)
• Wyner-Ziv Gradient Compression for Federated Learning [4.619828919345114]
  Gradient compression is an effective method to reduce communication load by transmitting compressed gradients. This paper proposes a practical gradient compression scheme for federated learning, which uses historical gradients to compress gradients. We also implement our gradient quantization method on the real dataset, and the performance of our method is better than the previous schemes.
  arXiv  Detail & Related papers  (2021-11-16T07:55:43Z)
• Sparse Communication for Training Deep Networks [56.441077560085475]
  Synchronous gradient descent (SGD) is the most common method used for distributed training of deep learning models. In this algorithm, each worker shares its local gradients with others and updates the parameters using the average gradients of all workers. We study several compression schemes and identify how three key parameters affect the performance.
  arXiv  Detail & Related papers  (2020-09-19T17:28:11Z)
• Extrapolation for Large-batch Training in Deep Learning [72.61259487233214]
  We show that a host of variations can be covered in a unified framework that we propose. We prove the convergence of this novel scheme and rigorously evaluate its empirical performance on ResNet, LSTM, and Transformer.
  arXiv  Detail & Related papers  (2020-06-10T08:22:41Z)

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.