@inproceedings{maheswaran-desarkar-2026-beyond,
title = "Beyond {EOS}: Revealing the Internal Monologue of Language Models",
author = "Maheswaran, Aishwarya and
Desarkar, Maunendra Sankar",
editor = "Zhang, Yue and
Xu, Ruifeng and
Purwarianti, Ayu and
Rong, Tong and
Suzuki, Jun",
booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {IJCNLP}-{AACL} 2026",
month = nov,
year = "2026",
address = "Hengqin, China",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.findings-aacl.180/",
pages = "3556--3574",
ISBN = "979-8-89176-444-6",
abstract = "Outputs generated by LLMs reflect traits present in the training data. However, it is unclear how the model acquires these traits and steers itself in that direction. We demonstrate that allowing the model to continue generating beyond the {\ensuremath{<}}eos{\ensuremath{>}} token reveals its comprehension of the task in the form of specific `Notes' or Rules distinct from normal generation. We refer to this as the model{'}s Internal Monologue (IM) and demonstrate its utility to understand the model{'}s perception of the task it is being finetuned on. We find this monologue is transient and dissipates as training progresses. We hypothesize this is a inherent behavior of instruction tuned models that adapts to reflect task instructions until the latent space beyond {\ensuremath{<}}eos{\ensuremath{>}} is sufficiently altered. Importantly, we do not ask the model to reason or do anything special to invoke this. We perform analysis on four diverse dialogue datasets from the Counseling domain to compare how dataset differences affect the internal monologue during early finetuning. We then show this is a general phenomenon by testing with three models of varying sizes and families and other domain datasets. We demonstrate the utility of IM as a means to gain insight on the model{'}s perception of the dataset and task through clustering based visualizations."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="maheswaran-desarkar-2026-beyond">
<titleInfo>
<title>Beyond EOS: Revealing the Internal Monologue of Language Models</title>
</titleInfo>
<name type="personal">
<namePart type="given">Aishwarya</namePart>
<namePart type="family">Maheswaran</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Maunendra</namePart>
<namePart type="given">Sankar</namePart>
<namePart type="family">Desarkar</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-11</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Findings of the Association for Computational Linguistics: IJCNLP-AACL 2026</title>
</titleInfo>
<name type="personal">
<namePart type="given">Yue</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ruifeng</namePart>
<namePart type="family">Xu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ayu</namePart>
<namePart type="family">Purwarianti</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Tong</namePart>
<namePart type="family">Rong</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jun</namePart>
<namePart type="family">Suzuki</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Hengqin, China</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-444-6</identifier>
</relatedItem>
<abstract>Outputs generated by LLMs reflect traits present in the training data. However, it is unclear how the model acquires these traits and steers itself in that direction. We demonstrate that allowing the model to continue generating beyond the \ensuremath<eos\ensuremath> token reveals its comprehension of the task in the form of specific ‘Notes’ or Rules distinct from normal generation. We refer to this as the model’s Internal Monologue (IM) and demonstrate its utility to understand the model’s perception of the task it is being finetuned on. We find this monologue is transient and dissipates as training progresses. We hypothesize this is a inherent behavior of instruction tuned models that adapts to reflect task instructions until the latent space beyond \ensuremath<eos\ensuremath> is sufficiently altered. Importantly, we do not ask the model to reason or do anything special to invoke this. We perform analysis on four diverse dialogue datasets from the Counseling domain to compare how dataset differences affect the internal monologue during early finetuning. We then show this is a general phenomenon by testing with three models of varying sizes and families and other domain datasets. We demonstrate the utility of IM as a means to gain insight on the model’s perception of the dataset and task through clustering based visualizations.</abstract>
<identifier type="citekey">maheswaran-desarkar-2026-beyond</identifier>
<location>
<url>https://aclanthology.org/2026.findings-aacl.180/</url>
</location>
<part>
<date>2026-11</date>
<extent unit="page">
<start>3556</start>
<end>3574</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Beyond EOS: Revealing the Internal Monologue of Language Models
%A Maheswaran, Aishwarya
%A Desarkar, Maunendra Sankar
%Y Zhang, Yue
%Y Xu, Ruifeng
%Y Purwarianti, Ayu
%Y Rong, Tong
%Y Suzuki, Jun
%S Findings of the Association for Computational Linguistics: IJCNLP-AACL 2026
%D 2026
%8 November
%I Association for Computational Linguistics
%C Hengqin, China
%@ 979-8-89176-444-6
%F maheswaran-desarkar-2026-beyond
%X Outputs generated by LLMs reflect traits present in the training data. However, it is unclear how the model acquires these traits and steers itself in that direction. We demonstrate that allowing the model to continue generating beyond the \ensuremath<eos\ensuremath> token reveals its comprehension of the task in the form of specific ‘Notes’ or Rules distinct from normal generation. We refer to this as the model’s Internal Monologue (IM) and demonstrate its utility to understand the model’s perception of the task it is being finetuned on. We find this monologue is transient and dissipates as training progresses. We hypothesize this is a inherent behavior of instruction tuned models that adapts to reflect task instructions until the latent space beyond \ensuremath<eos\ensuremath> is sufficiently altered. Importantly, we do not ask the model to reason or do anything special to invoke this. We perform analysis on four diverse dialogue datasets from the Counseling domain to compare how dataset differences affect the internal monologue during early finetuning. We then show this is a general phenomenon by testing with three models of varying sizes and families and other domain datasets. We demonstrate the utility of IM as a means to gain insight on the model’s perception of the dataset and task through clustering based visualizations.
%U https://aclanthology.org/2026.findings-aacl.180/
%P 3556-3574
Markdown (Informal)
[Beyond EOS: Revealing the Internal Monologue of Language Models](https://aclanthology.org/2026.findings-aacl.180/) (Maheswaran & Desarkar, Findings 2026)
ACL