BLISS: A Lightweight Bilevel Influence Scoring Method for Data Selection in Language Model Pretraining

• URL: http://arxiv.org/abs/2510.06048v2
• Date: Wed, 08 Oct 2025 17:49:49 GMT
• Title: BLISS: A Lightweight Bilevel Influence Scoring Method for Data Selection in Language Model Pretraining
• Authors: Jie Hao, Rui Yu, Wei Zhang, Huixia Wang, Jie Xu, Mingrui Liu,
• Abstract summary: BLISS is a lightweight data selection method that operates entirely emphfrom scratch, without relying on any external pretrained oracle models.<n>We validate BLISS by pretraining 410M/1B/2.8B Pythia and LLaMA-0.5B models on selected subsets of the C4 dataset.<n> BLISS achieves $1.7times$ speedup in reaching the same performance as the state-of-the-art method.
• Score: 28.32850393150554
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Effective data selection is essential for pretraining large language models (LLMs), enhancing efficiency and improving generalization to downstream tasks. However, existing approaches often require leveraging external pretrained models, making it difficult to disentangle the effects of data selection from those of the external pretrained models. In addition, they often overlook the long-term impact of selected data if the model is trained to convergence, primarily due to the prohibitive cost of full-scale LLM pretraining. In this paper, we introduce BLISS (\textbf{B}ileve\textbf{L} \textbf{I}nfluence \textbf{S}coring method for data \textbf{S}election): a lightweight data selection method that operates entirely \emph{from scratch}, without relying on any external pretrained oracle models, while explicitly accounting for the long-term impact of selected data. BLISS leverages a small proxy model as a surrogate for the LLM and employs a score model to estimate the long-term influence of training samples if the proxy model is trained to convergence. We formulate data selection as a bilevel optimization problem, where the upper-level objective optimizes the score model to assign importance weights to training samples, ensuring that minimizing the lower-level objective (i.e., training the proxy model over the weighted training loss until convergence) leads to best validation performance. Once optimized, the trained score model predicts influence scores for the dataset, enabling efficient selection of high-quality samples for LLM pretraining. We validate BLISS by pretraining 410M/1B/2.8B Pythia and LLaMA-0.5B models on selected subsets of the C4 dataset. Notably, under the 1B model setting, BLISS achieves $1.7\times$ speedup in reaching the same performance as the state-of-the-art method, demonstrating superior performance across multiple downstream tasks.

Related papers

• Influence-Preserving Proxies for Gradient-Based Data Selection in LLM Fine-tuning [51.87858735871145]
  We introduce Iprox, a framework that derives influence-preserving proxies directly from the target model.<n>Iprox consistently outperforms off-the-shelf proxies and baseline methods.
  arXiv  Detail & Related papers  (2026-02-19T20:57:30Z)
• SPaRFT: Self-Paced Reinforcement Fine-Tuning for Large Language Models [51.74498855100541]
  Large language models (LLMs) have shown strong reasoning capabilities when fine-tuned with reinforcement learning (RL)<n>We propose textbfSPaRFT, a self-paced learning framework that enables efficient learning based on the capability of the model being trained.
  arXiv  Detail & Related papers  (2025-08-07T03:50:48Z)
• LLM Data Selection and Utilization via Dynamic Bi-level Optimization [100.20933466418786]
  We propose a new Data Weighting Model (DWM) to adjust the weight of selected data within each batch to achieve a dynamic data utilization during training.<n>Our experiments demonstrate that DWM enhances the performance of models trained with randomly-selected data.<n>We further analyze how a model's data preferences evolve throughout training, providing new insights into the data preference of the model during training.
  arXiv  Detail & Related papers  (2025-07-22T02:47:12Z)
• Efficient Data Selection at Scale via Influence Distillation [53.03573620682107]
  This paper introduces Influence Distillation, a mathematicallyjustified framework for data selection.<n>By distilling each sample's influence on a target distribution, our method assigns model-specific weights that are used to select training data.<n>Experiments show that Influence Distillation matches or outperforms state-of-the-art performance while achieving up to $3.5times$ faster selection.
  arXiv  Detail & Related papers  (2025-05-25T09:08:00Z)
• Preference Curriculum: LLMs Should Always Be Pretrained on Their Preferred Data [19.221998577357713]
  Large language models (LLMs) generally utilize a consistent data distribution throughout the pretraining process.<n>As the model's capability improves, it is intuitive that its data preferences dynamically change, indicating the need for pretraining with different data at various training stages.<n>We propose the Perplexity Difference (PD) based Preference Curriculum learning framework, which always perceives and uses the data preferred by LLMs to train and boost them.
  arXiv  Detail & Related papers  (2025-01-21T13:12:13Z)
• A Self-Supervised Paradigm for Data-Efficient Medical Foundation Model Pre-training: V-information Optimization Framework [15.413974936297082]
  Self-supervised pre-training medical foundation models on large-scale datasets demonstrate exceptional performance.<n>Recent research challenges this common paradigm by introducing data-effective learning approaches.<n>We introduce V-information into self-supervised pre-training of foundation models to provide a theoretical foundation for sample selection.
  arXiv  Detail & Related papers  (2024-08-13T10:28:54Z)
• MATES: Model-Aware Data Selection for Efficient Pretraining with Data Influence Models [16.654859430784825]
  Current data selection methods, which rely on either hand-crafted rules or larger reference models, are conducted statically and do not capture the evolving data preferences during pretraining. We introduce model-aware data selection with data influence models (MATES), where a data influence model continuously adapts to the evolving data preferences of the pretraining model and then selects the data most effective for the current pretraining progress. Experiments of pretraining 410M and 1B models on the C4 dataset demonstrate that MATES significantly outperforms random data selection on extensive downstream tasks.
  arXiv  Detail & Related papers  (2024-06-10T06:27:42Z)
• How to Train Data-Efficient LLMs [56.41105687693619]
  We study data-efficient approaches for pre-training language models (LLMs) We find that Ask-LLM and Density sampling are the best methods in their respective categories. In our comparison of 19 samplers, involving hundreds of evaluation tasks and pre-training runs, we find that Ask-LLM and Density are the best methods in their respective categories.
  arXiv  Detail & Related papers  (2024-02-15T02:27:57Z)
• MILO: Model-Agnostic Subset Selection Framework for Efficient Model Training and Tuning [68.12870241637636]
  We propose MILO, a model-agnostic subset selection framework that decouples the subset selection from model training. Our empirical results indicate that MILO can train models $3times - 10 times$ faster and tune hyperparameters $20times - 75 times$ faster than full-dataset training or tuning without performance.
  arXiv  Detail & Related papers  (2023-01-30T20:59:30Z)

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.