Conference paper · 2025 · Open access

Hyperbolic Fine-Tuning for Large Language Models

Menglin Yang, Ram B, Aosong Feng, Bo Xiong, J C Liu, Irwin King, Rex Ying

neural information processing systems, 33449–33484 · Published 1 January 2025 · DOI 10.52202/085713-0992

Open PDFView at the publisherCite this paper

In brief

“Hyperbolic Fine-Tuning for Large Language Models” is a 2025 conference paper by Menglin Yang and 6 co-authors, published in neural information processing systems. OpenAlex counts 47 citations as of 18 November 2026. It is open access under a CC BY-NC-SA licence: the full text is free to read at arXiv (Cornell University).

Citations
47
Most cited in
2026 (32)
Field-weighted impact
10.56
References
55
Authors
7

Abstract

Large language models (LLMs) have demonstrated remarkable performance across various tasks. However, it remains an open question whether the default Euclidean space is the most suitable choice for LLMs. In this study, we investigate the geometric characteristics of LLMs, focusing specifically on tokens and their embeddings. Our findings reveal that token frequency follows a power-law distribution, where high-frequency tokens (e.g., the, that ) constitute the minority, while low-frequency tokens (e.g., apple, dog) constitute the majority. Furthermore, high-frequency tokens cluster near the origin, whereas low-frequency tokens are positioned farther away in the embedding space. Additionally, token embeddings exhibit hyperbolic characteristics, indicating a latent tree-like structure within the embedding space. Motivated by these observations, we propose HypLoRA, an efficient fine-tuning approach that operates in hyperbolic space to exploit these underlying hierarchical structures better. HypLoRA performs low-rank adaptation directly in hyperbolic space, thereby preserving hyperbolic modeling capabilities throughout the fine-tuning process. Extensive experiments across various base models and reasoning benchmarks, specifically arithmetic and commonsense reasoning tasks, demonstrate that HypLoRA substantially improves LLM performance.

Cite this paper

Formatted from the catalogue record. Check names, capitals and page numbers against the paper before you submit.

APA 7

Yang, M., B, R., Feng, A., Xiong, B., Liu, J. C., King, I., & Ying, R. (2025). Hyperbolic Fine-Tuning for Large Language Models. neural information processing systems, 33449–33484. https://doi.org/10.52202/085713-0992

MLA 9

Yang, Menglin, et al. “Hyperbolic Fine-Tuning for Large Language Models.” neural information processing systems, 2025, pp. 33449–33484. https://doi.org/10.52202/085713-0992.

Chicago

Yang, Menglin, Ram B, Aosong Feng, Bo Xiong, J C Liu, Irwin King, and Rex Ying. 2025. “Hyperbolic Fine-Tuning for Large Language Models.” neural information processing systems: 33449–33484. https://doi.org/10.52202/085713-0992.

Harvard

Yang, M. et al. (2025) ‘Hyperbolic Fine-Tuning for Large Language Models’, neural information processing systems, pp. 33449–33484. doi:10.52202/085713-0992.

IEEE

M. Yang et al., “Hyperbolic Fine-Tuning for Large Language Models,” neural information processing systems, pp. 33449–33484, 2025, doi: 10.52202/085713-0992.

BibTeX

@inproceedings{yang2025hyperbolic,
  title     = {{Hyperbolic Fine-Tuning for Large Language Models}},
  author    = {Yang, Menglin and B, Ram and Feng, Aosong and Xiong, Bo and Liu, J C and King, Irwin and Ying, Rex},
  booktitle = {neural information processing systems},
  year      = {2025},
  pages     = {33449--33484},
  doi       = {10.52202/085713-0992},
  url       = {https://doi.org/10.52202/085713-0992}
}

Download for a reference manager: BibTeX (.bib) · RIS for Zotero, Mendeley and EndNote (.ris)

Citations by year

YearCitationsShare of the best year
202211
202422
20251212
20263232

Its field-weighted citation impact is 10.56: it has been cited 10.56 times as often as the average work of the same field, type and year (1.00 is the world average).

Topic and keywords

Physical Sciences › Computer Science › Artificial Intelligence › Topic Modeling

  • hyperbolic space
  • large language models
  • low-rank adaptation
  • token embeddings
  • power-law distribution
  • embedding space geometry
  • arithmetic reasoning
  • commonsense reasoning

Work with this paper

Most-cited papers in Computer Science

All highly cited Computer Science papers

Questions about this paper

How do I cite this paper in APA style?

Yang, M., B, R., Feng, A., Xiong, B., Liu, J. C., King, I., & Ying, R. (2025). Hyperbolic Fine-Tuning for Large Language Models. neural information processing systems, 33449–33484. https://doi.org/10.52202/085713-0992

How many times has this paper been cited?

47 times, according to OpenAlex on 18 November 2026. Google Scholar and Scopus count different sets of sources, so their numbers differ.

Is this paper free to read?

Yes. It is open access at arXiv (Cornell University): sign in and open the PDF from this page.

Who wrote this paper?

Menglin Yang, Ram B, Aosong Feng, Bo Xiong, J C Liu, Irwin King, Rex Ying.

What is the DOI of this paper?

10.52202/085713-0992, permanently at https://doi.org/10.52202/085713-0992. Sign in and open it from this page.

Bibliographic data from OpenAlex (CC0), last checked 18 November 2026. ResearchGuru does not host this paper; once you are signed in, the links open it at the publisher or the repository that does.