English

Learn-to-learn on Arbitrary Textual Conditioning: A Hypernetwork-Driven Meta-Gated LLM

Computation and Language 2026-05-05 v1 Machine Learning

Abstract

Conventional LLMs may suffer from corpus heterogeneity and subtle condition changes. While finetuning can create the catastrophe forgetting issue, application of meta-learning on LLMs is also limited due to its complexity and scalability. In this paper, we activate the meta-signal of β\beta within the SwiGLU blocks, resulting in a meta-gating mechanism that adaptively adjusts the nonlinearity of FFN. A hypernetwork is employed which dynamically produces β\beta on textual conditions, providing meta-controllability on LLMs. By testing on different condition types such as task, domain, persona, and style, our method outperforms finetuning and meta-learning baselines, and can generalize reasonably on unseen tasks, condition types, or instructions. Our code can be found in https://github.com/AaronJi/MeGan.

Keywords

Cite

@article{arxiv.2605.01973,
  title  = {Learn-to-learn on Arbitrary Textual Conditioning: A Hypernetwork-Driven Meta-Gated LLM},
  author = {Luo Ji and Qi Qin and Ningyuan Xi and Teng Chen and Qingqing Gu and Hongyan Li},
  journal= {arXiv preprint arXiv:2605.01973},
  year   = {2026}
}

Comments

Accepted by ICML2026

R2 v1 2026-07-01T12:47:36.491Z