@inproceedings{ma-etal-2026-simlens,
title = "{S}im{L}ens: Two-Token Continuation for Early Exit in Single-Token Decision Tasks",
author = "Ma, Ming and
Zheng, Bowen and
Lin, Zhongqiao and
Yang, Tianming",
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.1568/",
pages = "35725--35741",
ISBN = "979-8-89176-459-0",
abstract = "Intermediate-layer representations in large language models (LLMs) contain task-relevant information, yet direct linear readouts often fail to recover the model{'}s eventual choice. This gap limits both interpretability and early-exit methods for candidate-scoring workloads. We study single-token decision settings {--} such as multiple-choice and boolean QA {--} where the answer is a single token at a position fixed by the task format, and we read out the model{'}s prediction there as its highest-probability token over the full vocabulary. We propose SimLens, a training-free readout that, at a chosen intermediate layer, keeps only the start token and the candidate answer token and forwards this reduced two-token state through the model{'}s remaining upper layers. To identify the layer at which the full sequence can be safely truncated, we distill LinearSimLens as a lightweight entropy monitor and combine the two in SimExit. Across different models, SimLens produces earlier and more faithful candidate readout than direct linear lenses under fair-compute accounting, while SimExit delivers a modest but stable speed{--}accuracy trade-off on the two models used end-to-end. Ablations show that the two retained tokens play complementary roles as global condition and semantic anchor."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="ma-etal-2026-simlens">
<titleInfo>
<title>SimLens: Two-Token Continuation for Early Exit in Single-Token Decision Tasks</title>
</titleInfo>
<name type="personal">
<namePart type="given">Ming</namePart>
<namePart type="family">Ma</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Bowen</namePart>
<namePart type="family">Zheng</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zhongqiao</namePart>
<namePart type="family">Lin</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Tianming</namePart>
<namePart type="family">Yang</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>Intermediate-layer representations in large language models (LLMs) contain task-relevant information, yet direct linear readouts often fail to recover the model’s eventual choice. This gap limits both interpretability and early-exit methods for candidate-scoring workloads. We study single-token decision settings – such as multiple-choice and boolean QA – where the answer is a single token at a position fixed by the task format, and we read out the model’s prediction there as its highest-probability token over the full vocabulary. We propose SimLens, a training-free readout that, at a chosen intermediate layer, keeps only the start token and the candidate answer token and forwards this reduced two-token state through the model’s remaining upper layers. To identify the layer at which the full sequence can be safely truncated, we distill LinearSimLens as a lightweight entropy monitor and combine the two in SimExit. Across different models, SimLens produces earlier and more faithful candidate readout than direct linear lenses under fair-compute accounting, while SimExit delivers a modest but stable speed–accuracy trade-off on the two models used end-to-end. Ablations show that the two retained tokens play complementary roles as global condition and semantic anchor.</abstract>
<identifier type="citekey">ma-etal-2026-simlens</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-main.1568/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>35725</start>
<end>35741</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T SimLens: Two-Token Continuation for Early Exit in Single-Token Decision Tasks
%A Ma, Ming
%A Zheng, Bowen
%A Lin, Zhongqiao
%A Yang, Tianming
%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 ma-etal-2026-simlens
%X Intermediate-layer representations in large language models (LLMs) contain task-relevant information, yet direct linear readouts often fail to recover the model’s eventual choice. This gap limits both interpretability and early-exit methods for candidate-scoring workloads. We study single-token decision settings – such as multiple-choice and boolean QA – where the answer is a single token at a position fixed by the task format, and we read out the model’s prediction there as its highest-probability token over the full vocabulary. We propose SimLens, a training-free readout that, at a chosen intermediate layer, keeps only the start token and the candidate answer token and forwards this reduced two-token state through the model’s remaining upper layers. To identify the layer at which the full sequence can be safely truncated, we distill LinearSimLens as a lightweight entropy monitor and combine the two in SimExit. Across different models, SimLens produces earlier and more faithful candidate readout than direct linear lenses under fair-compute accounting, while SimExit delivers a modest but stable speed–accuracy trade-off on the two models used end-to-end. Ablations show that the two retained tokens play complementary roles as global condition and semantic anchor.
%U https://aclanthology.org/2026.emnlp-main.1568/
%P 35725-35741
Markdown (Informal)
[SimLens: Two-Token Continuation for Early Exit in Single-Token Decision Tasks](https://aclanthology.org/2026.emnlp-main.1568/) (Ma et al., EMNLP 2026)
ACL