Scaling Laws for Hypernetwork-Based Knowledge Injection in Large Language Models

• URL: http://arxiv.org/abs/2607.19604v1
• Date: Tue, 21 Jul 2026 22:09:55 GMT
• Title: Scaling Laws for Hypernetwork-Based Knowledge Injection in Large Language Models
• Abstract summary: We investigate whether hypernetworks can be used to perform train-time knowledge injection.<n>We characterize how loss, reasoning accuracy, and out-of-distribution (OOD) generalization vary with hypernetwork depth, width, and target network size.
• Score: 6.546866759954739
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Injecting factual knowledge into large language models (LLMs) reliably and at scale remains an open challenge. Hypernetworks provide a promising solution to large-scale knowledge injection. Although hypernetworks are typically applied for test-time adaptation, we explore their use in train-time knowledge injection, where, given a large corpus of facts, we train a hypernetwork to generate a fixed LoRA adapter that, when inserted into the target model, enable the model to answer questions about those facts. In this work, we investigate whether hypernetworks can be used to perform train-time knowledge injection and how this ability varies with scale. The scaling behavior of hypernetworks remains largely unstudied. Our design decouples the hypernetwork's injection capacity from the target model's general capability, enabling, for the first time, a rigorous study of scaling laws for hypernetwork architectures. We characterize how loss, reasoning accuracy, and out-of-distribution (OOD) generalization vary with hypernetwork depth, width, and target network size. We construct a large-scale dataset, called MegaWikiQA, containing tens of millions of multi-hop question-answer examples across 39 domains constructed from examples in Wikidata5M. Our results reveal: (i) hypernetwork-based injection exhibits broadly predictive power law scaling along all architecture axes; and (ii) hypernetworks are capable of reliable OOD generalization at increasing scales, suggesting that hypernetwork provides a promising alternative to other train-time adaptation methods such as LoRA finetuning and full fine-tuning, exhibiting steeper scaling exponents in all OOD evaluations. Together, these results establish hypernetworks as a principled and scalable substrate for train-time adaptation, and provide the first empirically grounded scaling laws to guide hypernetworks for factual reasoning in large language models.

Related papers

• A Wireless Foundation Model for Multi-Task Prediction [50.21098141769079]
  We propose a unified foundation model for multi-task prediction in wireless networks that supports diverse prediction intervals.<n>After trained on large-scale datasets, the proposed foundation model demonstrates strong generalization to unseen scenarios and zero-shot performance on new tasks.
  arXiv  Detail & Related papers  (2025-07-08T12:37:55Z)
• Network Sparsity Unlocks the Scaling Potential of Deep Reinforcement Learning [57.3885832382455]
  We show that introducing static network sparsity alone can unlock further scaling potential beyond dense counterparts with state-of-the-art architectures.<n>Our analysis reveals that, in contrast to naively scaling up dense DRL networks, such sparse networks achieve both higher parameter efficiency for network expressivity.
  arXiv  Detail & Related papers  (2025-06-20T17:54:24Z)
• Identifying Super Spreaders in Multilayer Networks [0.6990493129893112]
  We introduce a novel approach to identifying super-spreaders in such networks by leveraging graph neural networks.<n>To this end, we construct a dataset by simulating information diffusion across hundreds of networks.<n>Our model, TopSpreadersNetwork, comprises a relationship-agnostic encoder and a custom aggregation layer.
  arXiv  Detail & Related papers  (2025-05-27T10:14:14Z)
• Unified Neural Network Scaling Laws and Scale-time Equivalence [10.918504301310753]
  We present a novel theoretical characterization of how three factors -- model size, training time, and data volume -- interact to determine the performance of deep neural networks. We first establish a theoretical and empirical equivalence between scaling the size of a neural network and increasing its training time proportionally. We then combine scale-time equivalence with a linear model analysis of double descent to obtain a unified theoretical scaling law.
  arXiv  Detail & Related papers  (2024-09-09T16:45:26Z)
• A method for quantifying the generalization capabilities of generative models for solving Ising models [5.699467840225041]
  We use a Hamming distance regularizer to quantify the generalization capabilities of various network architectures combined with VAN. We conduct numerical experiments on several network architectures combined with VAN, including feed-forward neural networks, recurrent neural networks, and graph neural networks. Our method is of great significance for assisting in the Neural Architecture Search field of searching for the optimal network architectures when solving large-scale Ising models.
  arXiv  Detail & Related papers  (2024-05-06T12:58:48Z)
• Generalization and Estimation Error Bounds for Model-based Neural Networks [78.88759757988761]
  We show that the generalization abilities of model-based networks for sparse recovery outperform those of regular ReLU networks. We derive practical design rules that allow to construct model-based networks with guaranteed high generalization.
  arXiv  Detail & Related papers  (2023-04-19T16:39:44Z)
• Magnitude Invariant Parametrizations Improve Hypernetwork Learning [0.0]
  Hypernetworks are powerful neural networks that predict the parameters of another neural network. Training typically converges far more slowly than for non-hypernetwork models. We identify a fundamental and previously unidentified problem that contributes to the challenge of training hypernetworks. We present a simple solution to this problem using a revised hypernetwork formulation that we call Magnitude Invariant Parametrizations (MIP)
  arXiv  Detail & Related papers  (2023-04-15T22:18:29Z)
• Semi-supervised Network Embedding with Differentiable Deep Quantisation [81.49184987430333]
  We develop d-SNEQ, a differentiable quantisation method for network embedding. d-SNEQ incorporates a rank loss to equip the learned quantisation codes with rich high-order information. It is able to substantially compress the size of trained embeddings, thus reducing storage footprint and accelerating retrieval speed.
  arXiv  Detail & Related papers  (2021-08-20T11:53:05Z)
• Hypernetwork Dismantling via Deep Reinforcement Learning [1.4877837830677472]
  We formulate the hypernetwork dismantling problem as a node sequence decision problem. We propose a deep reinforcement learning-based hypernetwork dismantling framework. Experimental results on five real-world hypernetworks demonstrate the effectiveness of our proposed framework.
  arXiv  Detail & Related papers  (2021-04-29T13:35:29Z)
• Deep Autoencoding Topic Model with Scalable Hybrid Bayesian Inference [55.35176938713946]
  We develop deep autoencoding topic model (DATM) that uses a hierarchy of gamma distributions to construct its multi-stochastic-layer generative network. We propose a Weibull upward-downward variational encoder that deterministically propagates information upward via a deep neural network, followed by a downward generative model. The efficacy and scalability of our models are demonstrated on both unsupervised and supervised learning tasks on big corpora.
  arXiv  Detail & Related papers  (2020-06-15T22:22:56Z)
• On Infinite-Width Hypernetworks [101.03630454105621]
  We show that hypernetworks do not guarantee to a global minima under descent. We identify the functional priors of these architectures by deriving their corresponding GP and NTK kernels. As part of this study, we make a mathematical contribution by deriving tight bounds on high order Taylor terms of standard fully connected ReLU networks.
  arXiv  Detail & Related papers  (2020-03-27T00:50:29Z)

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.