Research

Paper

AI LLM February 20, 2026

On the Semantic and Syntactic Information Encoded in Proto-Tokens for One-Step Text Reconstruction

Authors

Ivan Bondarenko, Egor Palkin, Fedor Tikunov

Abstract

Autoregressive large language models (LLMs) generate text token-by-token, requiring n forward passes to produce a sequence of length n. Recent work, Exploring the Latent Capacity of LLMs for One-Step Text Reconstruction (Mezentsev and Oseledets), shows that frozen LLMs can reconstruct hundreds of tokens from only two learned proto-tokens in a single forward pass, suggesting a path beyond the autoregressive paradigm. In this paper, we study what information these proto-tokens encode and how they behave under reconstruction and controlled constraints. We perform a series of experiments aimed at disentangling semantic and syntactic content in the two proto-tokens, analyzing stability properties of the e-token, and visualizing attention patterns to the e-token during reconstruction. Finally, we test two regularization schemes for "imposing" semantic structure on the e-token using teacher embeddings, including an anchor-based loss and a relational distillation objective. Our results indicate that the m-token tends to capture semantic information more strongly than the e-token under standard optimization; anchor-based constraints trade off sharply with reconstruction accuracy; and relational distillation can transfer batch-level semantic relations into the proto-token space without sacrificing reconstruction quality, supporting the feasibility of future non-autoregressive seq2seq systems that predict proto-tokens as an intermediate representation.

Metadata

arXiv ID: 2602.18301
Provider: ARXIV
Primary Category: cs.LG
Published: 2026-02-20
Fetched: 2026-02-23 05:33

Related papers

Raw Data (Debug)
{
  "raw_xml": "<entry>\n    <id>http://arxiv.org/abs/2602.18301v1</id>\n    <title>On the Semantic and Syntactic Information Encoded in Proto-Tokens for One-Step Text Reconstruction</title>\n    <updated>2026-02-20T15:54:10Z</updated>\n    <link href='https://arxiv.org/abs/2602.18301v1' rel='alternate' type='text/html'/>\n    <link href='https://arxiv.org/pdf/2602.18301v1' rel='related' title='pdf' type='application/pdf'/>\n    <summary>Autoregressive large language models (LLMs) generate text token-by-token, requiring n forward passes to produce a sequence of length n. Recent work, Exploring the Latent Capacity of LLMs for One-Step Text Reconstruction (Mezentsev and Oseledets), shows that frozen LLMs can reconstruct hundreds of tokens from only two learned proto-tokens in a single forward pass, suggesting a path beyond the autoregressive paradigm. In this paper, we study what information these proto-tokens encode and how they behave under reconstruction and controlled constraints. We perform a series of experiments aimed at disentangling semantic and syntactic content in the two proto-tokens, analyzing stability properties of the e-token, and visualizing attention patterns to the e-token during reconstruction. Finally, we test two regularization schemes for \"imposing\" semantic structure on the e-token using teacher embeddings, including an anchor-based loss and a relational distillation objective. Our results indicate that the m-token tends to capture semantic information more strongly than the e-token under standard optimization; anchor-based constraints trade off sharply with reconstruction accuracy; and relational distillation can transfer batch-level semantic relations into the proto-token space without sacrificing reconstruction quality, supporting the feasibility of future non-autoregressive seq2seq systems that predict proto-tokens as an intermediate representation.</summary>\n    <category scheme='http://arxiv.org/schemas/atom' term='cs.LG'/>\n    <category scheme='http://arxiv.org/schemas/atom' term='cs.CL'/>\n    <published>2026-02-20T15:54:10Z</published>\n    <arxiv:primary_category term='cs.LG'/>\n    <author>\n      <name>Ivan Bondarenko</name>\n    </author>\n    <author>\n      <name>Egor Palkin</name>\n    </author>\n    <author>\n      <name>Fedor Tikunov</name>\n    </author>\n  </entry>"
}