HackerPapers

arXiv:2505.12540cs.LG

Harnessing the Universal Geometry of Embeddings

Rishi Jha, Collin Zhang, Vitaly Shmatikov, John X. Morris

Abstract

We introduce the first method for translating text embeddings from one vector space to another without any paired data, encoders, or predefined sets of matches. Our unsupervised approach translates any embedding to and from a universal latent representation (i.e., a universal semantic structure conjectured by the Platonic Representation Hypothesis). Our translations achieve high cosine similarity across model pairs with different architectures, parameter counts, and training datasets. The ability to translate unknown embeddings into a different space while preserving their geometry has serious implications for the security of vector databases. An adversary with access only to embedding vectors can extract sensitive information about the underlying documents, sufficient for classification and attribute inference.

arXiv comments
Journal reference
DOI
Subjects
cs.LGMachine Learning

On Hacker News

Linked by 5 identifiable commenters in 5 comments across 5 threads, from Dec 2025 to Aug 2026. Found in 3 HN submissions from May 2025 to Sep 2026. Open submission discussions.

Comments linking this paper, per year.

Discussed in comments5 threads

  1. Models Are Getting Dumber on Purpose 1 comment
  2. What Emily Bender meant by "stochastic parrots" 1 comment

    • Thank you for expanding! I come from more of an ML background so still learning a lot on these topics: Agreed that the neocortex uses fewer layers because of looping - I also suspect it's partly because neurons are more complex than the neurons used in ANNs, so in principle they…

      ainchView on HNParent context

  3. Show HN: Duplicate 3 layers in a 24B LLM, logical deduction .22→.76. No training 1 comment

    • https://arxiv.org/abs/2505.12540 https://arxiv.org/abs/2405.07987 These, other papers, and the lottery ticket phenomenon; what it boils down to is that any neural network like system which encodes some common mapping of a phenomenon in the context of the world - not necessarily…

      observationistView on HNParent context

  4. Show HN: Moltbook – A social network for moltbots (clawdbots) to hang out 1 comment

    • If you can access a models emebeddings then it is possible to retrieve what it knows using a model you have trained https://arxiv.org/html/2505.12540v2

      mnky9800nView on HNParent context

  5. The universal weight subspace hypothesis 1 comment

    • Interesting - I wonder if this ties into the Platonic Space Hypothesis recently being championed by computational biologist Mike Levin E.g https://youtu.be/Qp0rCU49lMs?si=UXbSBD3Xxpy9e3uY https://thoughtforms.life/symposium-on-the-platonic-space/ e.g see this paper on Universal…

      AIorNotView on HNParent context

Submitted to Hacker News

Connected through shared HN threads, not inferred similarity or a recommendation.