SampleMix: A Sample-wise Pre-training Data Mixing Strategey by Coordinating Data Quality and Diversity

• URL: http://arxiv.org/abs/2503.01506v1
• Date: Mon, 03 Mar 2025 13:22:11 GMT
• Title: SampleMix: A Sample-wise Pre-training Data Mixing Strategey by Coordinating Data Quality and Diversity
• Authors: Xiangyu Xi, Deyang Kong, Jian Yang, Jiawei Yang, Zhengyu Chen, Wei Wang, Jingang Wang, Xunliang Cai, Shikun Zhang, Wei Ye,
• Abstract summary: Existing pretraining data mixing methods for large language models (LLMs) typically follow a domain-wise methodology.<n>We propose a novel sample-wise data mixture approach based on a bottom-up paradigm.
• Score: 36.9096162214815
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Existing pretraining data mixing methods for large language models (LLMs) typically follow a domain-wise methodology, a top-down process that first determines domain weights and then performs uniform data sampling across each domain. However, these approaches neglect significant inter-domain overlaps and commonalities, failing to control the global diversity of the constructed training dataset. Further, uniform sampling within domains ignores fine-grained sample-specific features, potentially leading to suboptimal data distribution. To address these shortcomings, we propose a novel sample-wise data mixture approach based on a bottom-up paradigm. This method performs global cross-domain sampling by systematically evaluating the quality and diversity of each sample, thereby dynamically determining the optimal domain distribution. Comprehensive experiments across multiple downstream tasks and perplexity assessments demonstrate that SampleMix surpasses existing domain-based methods. Meanwhile, SampleMix requires 1.4x to 2.1x training steps to achieves the baselines' performance, highlighting the substantial potential of SampleMix to optimize pre-training data.

Related papers

• QuaDMix: Quality-Diversity Balanced Data Selection for Efficient LLM Pretraining [12.872792775510172]
  We introduce a unified data selection framework called QuaDMix, which automatically optimize the data distribution for large language models pretraining. Our experiments across diverse models and datasets demonstrate that QuaDMix achieves an average performance improvement of 7.2% across multiple benchmarks.
  arXiv  Detail & Related papers  (2025-04-23T08:36:50Z)
• Improving Distribution Alignment with Diversity-based Sampling [0.0]
  Domain shifts are ubiquitous in machine learning, and can substantially degrade a model's performance when deployed to real-world data. This paper proposes to improve these estimates by inducing diversity in each sampled minibatch. It simultaneously balances the data and reduces the variance of the gradients, thereby enhancing the model's generalisation ability.
  arXiv  Detail & Related papers  (2024-10-05T17:26:03Z)
• Deep Domain Isolation and Sample Clustered Federated Learning for Semantic Segmentation [2.515027627030043]
  In this paper, we explore for the first time the effect of covariate shifts between participants' data in 2D segmentation tasks. We develop Deep Domain Isolation (DDI) to isolate image domains directly in the gradient space of the model. We leverage this clustering algorithm through a Sample Clustered Federated Learning (SCFL) framework.
  arXiv  Detail & Related papers  (2024-10-04T12:43:07Z)
• Spanning Training Progress: Temporal Dual-Depth Scoring (TDDS) for Enhanced Dataset Pruning [50.809769498312434]
  We propose a novel dataset pruning method termed as Temporal Dual-Depth Scoring (TDDS) Our method achieves 54.51% accuracy with only 10% training data, surpassing random selection by 7.83% and other comparison methods by at least 12.69%.
  arXiv  Detail & Related papers  (2023-11-22T03:45:30Z)
• Tackling Diverse Minorities in Imbalanced Classification [80.78227787608714]
  Imbalanced datasets are commonly observed in various real-world applications, presenting significant challenges in training classifiers. We propose generating synthetic samples iteratively by mixing data samples from both minority and majority classes. We demonstrate the effectiveness of our proposed framework through extensive experiments conducted on seven publicly available benchmark datasets.
  arXiv  Detail & Related papers  (2023-08-28T18:48:34Z)
• Bidirectional Domain Mixup for Domain Adaptive Semantic Segmentation [73.3083304858763]
  This paper systematically studies the impact of mixup under the domain adaptaive semantic segmentation task. In specific, we achieve domain mixup in two-step: cut and paste. We provide extensive ablation experiments to empirically verify our main components of the framework.
  arXiv  Detail & Related papers  (2023-03-17T05:22:44Z)
• A Data Cartography based MixUp for Pre-trained Language Models [47.90235939359225]
  MixUp is a data augmentation strategy where additional samples are generated during training by combining random pairs of training samples and their labels. We propose TDMixUp, a novel MixUp strategy that leverages Training Dynamics and allows more informative samples to be combined for generating new data samples. We empirically validate that our method not only achieves competitive performance using a smaller subset of the training data compared with strong baselines, but also yields lower expected calibration error on the pre-trained language model, BERT, on both in-domain and out-of-domain settings in a wide range of NLP tasks.
  arXiv  Detail & Related papers  (2022-05-06T17:59:19Z)
• DRFLM: Distributionally Robust Federated Learning with Inter-client Noise via Local Mixup [58.894901088797376]
  federated learning has emerged as a promising approach for training a global model using data from multiple organizations without leaking their raw data. We propose a general framework to solve the above two challenges simultaneously. We provide comprehensive theoretical analysis including robustness analysis, convergence analysis, and generalization ability.
  arXiv  Detail & Related papers  (2022-04-16T08:08:29Z)
• Optimal Importance Sampling for Federated Learning [57.14673504239551]
  Federated learning involves a mixture of centralized and decentralized processing tasks. The sampling of both agents and data is generally uniform; however, in this work we consider non-uniform sampling. We derive optimal importance sampling strategies for both agent and data selection and show that non-uniform sampling without replacement improves the performance of the original FedAvg algorithm.
  arXiv  Detail & Related papers  (2020-10-26T14:15:33Z)

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.