@inproceedings{xu-etal-2026-spinhash,
title = "{S}pin{H}ash: An {O}(1) Encoder for Any Vocabulary",
author = "Xu, Jia and
Wang, Jinhua and
Aliferis, Constantin F.",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.emnlp-main.574/",
pages = "13165--13180",
ISBN = "979-8-89176-459-0",
abstract = "The embedding table is a quiet ceiling of nearly every neural network model. Its size grows with the vocabulary, its rows are tied to a single modality, and unseen inputs require adding parameters. This makes the table prohibitively large at scale, impossible to share across modalities, and unable to represent new tokens without retraining.\par We remove this ceiling with SpinHash, an O(1) encoder in the vocabulary size that stays constant at about 0.2M parameters regardless of how many tokens or modalities it serves. Instead of storing a vocabulary dependent table, SpinHash hashes each token into a compact spin code and decodes it with one shared feedforward network. Our hash is not merely a compression mechanism, but also defines the geometry of the representation space, realized by a similarity preserving Spin-SimHash and an antiferromagnetic Ising Spin-Repulse.\par Across text, DNA, and clinical codes, SpinHash matches embedding tables with 17 to 283x fewer parameters, reaching parity on IMDB, AG-News, Yelp, and both clinical fusion tasks. For k=22 DNA, where 4{\textasciicircum}22 table entries are infeasible, SpinHash achieves 98.3{\%} promoter accuracy, outperforming same-size HyenaDNA-tiny and the 1000x larger Nucleotide Transformer.\par This work shows that representations need not be tied to a vocabulary, and that we can compute them once and reuse them across vocabularies, modalities, and domains."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="xu-etal-2026-spinhash">
<titleInfo>
<title>SpinHash: An O(1) Encoder for Any Vocabulary</title>
</titleInfo>
<name type="personal">
<namePart type="given">Jia</namePart>
<namePart type="family">Xu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jinhua</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Constantin</namePart>
<namePart type="given">F</namePart>
<namePart type="family">Aliferis</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-10</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing</title>
</titleInfo>
<name type="personal">
<namePart type="given">Hung-Yi</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sunipa</namePart>
<namePart type="family">Dev</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jessy</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Desmond</namePart>
<namePart type="family">Elliott</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Budapest, Hungary</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-459-0</identifier>
</relatedItem>
<abstract>The embedding table is a quiet ceiling of nearly every neural network model. Its size grows with the vocabulary, its rows are tied to a single modality, and unseen inputs require adding parameters. This makes the table prohibitively large at scale, impossible to share across modalities, and unable to represent new tokens without retraining.\par We remove this ceiling with SpinHash, an O(1) encoder in the vocabulary size that stays constant at about 0.2M parameters regardless of how many tokens or modalities it serves. Instead of storing a vocabulary dependent table, SpinHash hashes each token into a compact spin code and decodes it with one shared feedforward network. Our hash is not merely a compression mechanism, but also defines the geometry of the representation space, realized by a similarity preserving Spin-SimHash and an antiferromagnetic Ising Spin-Repulse.\par Across text, DNA, and clinical codes, SpinHash matches embedding tables with 17 to 283x fewer parameters, reaching parity on IMDB, AG-News, Yelp, and both clinical fusion tasks. For k=22 DNA, where 4⌃22 table entries are infeasible, SpinHash achieves 98.3% promoter accuracy, outperforming same-size HyenaDNA-tiny and the 1000x larger Nucleotide Transformer.\par This work shows that representations need not be tied to a vocabulary, and that we can compute them once and reuse them across vocabularies, modalities, and domains.</abstract>
<identifier type="citekey">xu-etal-2026-spinhash</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-main.574/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>13165</start>
<end>13180</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T SpinHash: An O(1) Encoder for Any Vocabulary
%A Xu, Jia
%A Wang, Jinhua
%A Aliferis, Constantin F.
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-459-0
%F xu-etal-2026-spinhash
%X The embedding table is a quiet ceiling of nearly every neural network model. Its size grows with the vocabulary, its rows are tied to a single modality, and unseen inputs require adding parameters. This makes the table prohibitively large at scale, impossible to share across modalities, and unable to represent new tokens without retraining.\par We remove this ceiling with SpinHash, an O(1) encoder in the vocabulary size that stays constant at about 0.2M parameters regardless of how many tokens or modalities it serves. Instead of storing a vocabulary dependent table, SpinHash hashes each token into a compact spin code and decodes it with one shared feedforward network. Our hash is not merely a compression mechanism, but also defines the geometry of the representation space, realized by a similarity preserving Spin-SimHash and an antiferromagnetic Ising Spin-Repulse.\par Across text, DNA, and clinical codes, SpinHash matches embedding tables with 17 to 283x fewer parameters, reaching parity on IMDB, AG-News, Yelp, and both clinical fusion tasks. For k=22 DNA, where 4⌃22 table entries are infeasible, SpinHash achieves 98.3% promoter accuracy, outperforming same-size HyenaDNA-tiny and the 1000x larger Nucleotide Transformer.\par This work shows that representations need not be tied to a vocabulary, and that we can compute them once and reuse them across vocabularies, modalities, and domains.
%U https://aclanthology.org/2026.emnlp-main.574/
%P 13165-13180
Markdown (Informal)
[SpinHash: An O(1) Encoder for Any Vocabulary](https://aclanthology.org/2026.emnlp-main.574/) (Xu et al., EMNLP 2026)
ACL
- Jia Xu, Jinhua Wang, and Constantin F. Aliferis. 2026. SpinHash: An O(1) Encoder for Any Vocabulary. In Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing, pages 13165–13180, Budapest, Hungary. Association for Computational Linguistics.