Constraint-based Pre-training: From Structured Constraints to Scalable Model Initialization

• URL: http://arxiv.org/abs/2604.14769v1
• Date: Thu, 16 Apr 2026 08:29:48 GMT
• Title: Constraint-based Pre-training: From Structured Constraints to Scalable Model Initialization
• Abstract summary: WeiT is a pre-training paradigm that disentangles size-agnostic knowledge into reusable weight templates.<n>We show that WeiT achieves state-of-the-art performance in initializing models with varying depths and widths.
• Score: 29.901255886542515
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: The pre-training and fine-tuning paradigm has become the dominant approach for model adaptation. However, conventional pre-training typically yields models at a fixed scale, whereas practical deployment often requires models of varying sizes, exposing its limitations when target model scales differ from those used during pre-training. To address this, we propose an innovative constraint-based pre-training paradigm that imposes structured constraints during pre-training to disentangle size-agnostic knowledge into reusable weight templates, while assigning size-specific adaptation to lightweight weight scalers, thereby reformulating variable-sized model initialization as a multi-task adaptation problem. Within this paradigm, we further introduce WeiT, which employs Kronecker-based constraints to regularize the pre-training process. Specifically, model parameters are represented as compositions of weight templates via concatenation and weighted aggregation, with adaptive connections governed by lightweight weight scalers whose parameters are learned from limited data. This design enables flexible and efficient construction of model weights across diverse downstream scales. Extensive experiments demonstrate the efficiency and effectiveness of WeiT, achieving state-of-the-art performance in initializing models with varying depths and widths across a broad range of perception and embodied learning tasks, including Image Classification, Image Generation, and Embodied Control. Moreover, its effectiveness generalizes to both Transformer-based and Convolution-based architectures, consistently enabling faster convergence and improved performance even under full training.

Related papers

• One-for-All Model Initialization with Frequency-Domain Knowledge [33.90471312176688]
  FRONT (FRequency dOmain kNowledge Transfer) is a novel framework that uses the Cosine Transform (DCT) to isolate the low-frequency "learngene"<n> FRONT achieves the state-of-the-art performance, accelerates convergence by up to 15 times in vision tasks, and reduces training FLOPs by an average of 40.5% in language tasks.
  arXiv  Detail & Related papers  (2026-03-08T08:25:49Z)
• SALAAD: Sparse And Low-Rank Adaptation via ADMM for Large Language Model Inference [38.037874715181964]
  We propose SALAAD, a plug-and-play framework that induces sparse and low-rank structures during training.<n>Experiments across model scales show that SALAAD substantially reduces memory consumption during deployment.<n>A single training run yields a continuous spectrum of model capacities, enabling smooth and elastic deployment across diverse memory budgets.
  arXiv  Detail & Related papers  (2026-02-01T00:00:11Z)
• Making Foundation Models Probabilistic via Singular Value Ensembles [56.4174499669573]
  Foundation models have become a dominant paradigm in machine learning, achieving remarkable performance across diverse tasks through large-scale pretraining.<n>Standard approach to quantifying uncertainty, training an ensemble of independent models, incurs prohibitive computational costs that scale linearly with ensemble size.<n>We propose Singular Value Ensemble (SVE), a parameter-efficient implicit ensemble method that builds on a simple, but powerful core assumption.<n>We show that SVE uncertainty quantification achieves comparable to explicit deep ensembles while increasing the parameter count of the base model by less than 1%.
  arXiv  Detail & Related papers  (2026-01-29T18:07:18Z)
• Self-Supervised Weight Templates for Scalable Vision Model Initialization [34.75805112986586]
  SWEET is a self-supervised framework that performs constraint-based pre-training to enable scalable initialization in vision tasks.<n>We introduce width-wise scaling, which regularizes the template along width-related dimensions and encourages robust, width-inwidth representations.<n>Experiments on textscclassification, textscsegmentation and textscgeneration tasks demonstrate the state-of-the-art performance of SWEET.
  arXiv  Detail & Related papers  (2026-01-27T15:15:17Z)
• An Integrated Fusion Framework for Ensemble Learning Leveraging Gradient Boosting and Fuzzy Rule-Based Models [59.13182819190547]
  Fuzzy rule-based models excel in interpretability and have seen widespread application across diverse fields.<n>They face challenges such as complex design specifications and scalability issues with large datasets.<n>This paper proposes an Integrated Fusion Framework that merges the strengths of both paradigms to enhance model performance and interpretability.
  arXiv  Detail & Related papers  (2025-11-11T10:28:23Z)
• Test-Time Model Adaptation for Quantized Neural Networks [37.84294929199108]
  Quantized models often suffer from severe performance degradation in dynamic environments with potential domain shifts.<n>Test-time adaptation (TTA) has emerged as an effective solution by enabling models to learn adaptively from test data.<n>We propose a continual zeroth-order adaptation (ZOA) framework that enables efficient model adaptation using only two forward passes.
  arXiv  Detail & Related papers  (2025-08-04T08:24:19Z)
• Transferable Post-training via Inverse Value Learning [83.75002867411263]
  We propose modeling changes at the logits level during post-training using a separate neural network (i.e., the value network)<n>After training this network on a small base model using demonstrations, this network can be seamlessly integrated with other pre-trained models during inference.<n>We demonstrate that the resulting value network has broad transferability across pre-trained models of different parameter sizes.
  arXiv  Detail & Related papers  (2024-10-28T13:48:43Z)
• SaRA: High-Efficient Diffusion Model Fine-tuning with Progressive Sparse Low-Rank Adaptation [52.6922833948127]
  In this work, we investigate the importance of parameters in pre-trained diffusion models.<n>We propose a novel model fine-tuning method to make full use of these ineffective parameters.<n>Our method enhances the generative capabilities of pre-trained models in downstream applications.
  arXiv  Detail & Related papers  (2024-09-10T16:44:47Z)
• WAVE: Weight Templates for Adaptive Initialization of Variable-sized Models [37.97945436202779]
  WAVE is a novel approach to initializing variable-sized models.<n>WAVE employs shared, size-agnostic weight templates alongside size-specific weight scalers.<n>WAVE achieves state-of-the-art performance in initializing models of various depth and width.
  arXiv  Detail & Related papers  (2024-06-25T12:43:33Z)
• An Emulator for Fine-Tuning Large Language Models using Small Language Models [91.02498576056057]
  We introduce emulated fine-tuning (EFT), a principled and practical method for sampling from a distribution that approximates the result of pre-training and fine-tuning at different scales. We show that EFT enables test-time adjustment of competing behavioral traits like helpfulness and harmlessness without additional training. Finally, a special case of emulated fine-tuning, which we call LM up-scaling, avoids resource-intensive fine-tuning of large pre-trained models by ensembling them with small fine-tuned models.
  arXiv  Detail & Related papers  (2023-10-19T17:57:16Z)
• Scaling Pre-trained Language Models to Deeper via Parameter-efficient Architecture [68.13678918660872]
  We design a more capable parameter-sharing architecture based on matrix product operator (MPO) MPO decomposition can reorganize and factorize the information of a parameter matrix into two parts. Our architecture shares the central tensor across all layers for reducing the model size.
  arXiv  Detail & Related papers  (2023-03-27T02:34:09Z)

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.