@inproceedings{debnath-etal-2026-cuet,
title = "Cuet Yet Another Baseline@{D}ravidian{L}ang{T}ech 2026: Shared Task on Prompt Recovery for {LLM} in {T}elugu",
author = "Debnath, Rotna Dipika and
Ruhi, Shahrin Afroz Hoque and
Labiba, Ayesha and
Mallik, Arpita and
Murad, Hasan",
editor = "Chakravarthi, Bharathi Raja and
Priyadharshini, Ruba and
Madasamy, Anand Kumar and
Thavareesan, Sajeetha and
Rajiakodi, Saranya and
Navaneethakrishnan, Subalalitha and
Chinnappa, Dhivya and
Palani, Balasubramanian and
Subramanian, Malliga and
Shanmugavadivel, Kogilavani and
Rajalakshmi, Ratnavel",
booktitle = "Proceedings of the Sixth Workshop on Speech, Vision, and Language Technologies for {D}ravidian Languages",
month = jul,
year = "2026",
address = "Underline (Virtual)",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.dravidianlangtech-1.25/",
pages = "191--195",
ISBN = "979-8-89176-401-9",
abstract = "Prompt recovery in large language models (LLMs) is the task of inferring the communicative intent and stylistic framing of the original instruction from model-generated output. This task is especially challenging for low-resource Dravidian languages such as Telugu, where agglutinative morphology, register variation, and scarce annotated data complicate stylistic modelling. In this paper, we present our system for the Shared Task on Prompt Recovery for LLM in Telugu at DravidianLangTech @ ACL{~}2026, which aims to classify Telugu transcript excerpts into nine communicative style categories: Formal, Informal, Optimistic, Pessimistic, Humorous, Serious, Inspiring, Authoritative, and Persuasive.We have implemented a transformer-based approach using ai4bharat/IndicBERTv2-MLM-only, MuRIL-base and Telugu-BERT for Telugu communicative style classification. Our system fine-tunes the pretrained Indic language training samples to capture stylistic patterns in Telugu transcripts. Our approach achieved a macro F1 score of 0.2993 on the evaluation set, demonstrating the potential of Indic-focused pretrained models for stylistic analysis in low-resource language settings.Controlled ablations reveal that label smoothing benefits stronger Indic backbones but degrades weaker ones, and that surface linguistic feature augmentation does not complement rich contextual representations on small datasets."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="debnath-etal-2026-cuet">
<titleInfo>
<title>Cuet Yet Another Baseline@DravidianLangTech 2026: Shared Task on Prompt Recovery for LLM in Telugu</title>
</titleInfo>
<name type="personal">
<namePart type="given">Rotna</namePart>
<namePart type="given">Dipika</namePart>
<namePart type="family">Debnath</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Shahrin</namePart>
<namePart type="given">Afroz</namePart>
<namePart type="given">Hoque</namePart>
<namePart type="family">Ruhi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ayesha</namePart>
<namePart type="family">Labiba</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Arpita</namePart>
<namePart type="family">Mallik</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hasan</namePart>
<namePart type="family">Murad</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-07</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the Sixth Workshop on Speech, Vision, and Language Technologies for Dravidian Languages</title>
</titleInfo>
<name type="personal">
<namePart type="given">Bharathi</namePart>
<namePart type="given">Raja</namePart>
<namePart type="family">Chakravarthi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ruba</namePart>
<namePart type="family">Priyadharshini</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Anand</namePart>
<namePart type="given">Kumar</namePart>
<namePart type="family">Madasamy</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sajeetha</namePart>
<namePart type="family">Thavareesan</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Saranya</namePart>
<namePart type="family">Rajiakodi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Subalalitha</namePart>
<namePart type="family">Navaneethakrishnan</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Dhivya</namePart>
<namePart type="family">Chinnappa</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Balasubramanian</namePart>
<namePart type="family">Palani</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Malliga</namePart>
<namePart type="family">Subramanian</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Kogilavani</namePart>
<namePart type="family">Shanmugavadivel</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ratnavel</namePart>
<namePart type="family">Rajalakshmi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Underline (Virtual)</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-401-9</identifier>
</relatedItem>
<abstract>Prompt recovery in large language models (LLMs) is the task of inferring the communicative intent and stylistic framing of the original instruction from model-generated output. This task is especially challenging for low-resource Dravidian languages such as Telugu, where agglutinative morphology, register variation, and scarce annotated data complicate stylistic modelling. In this paper, we present our system for the Shared Task on Prompt Recovery for LLM in Telugu at DravidianLangTech @ ACL 2026, which aims to classify Telugu transcript excerpts into nine communicative style categories: Formal, Informal, Optimistic, Pessimistic, Humorous, Serious, Inspiring, Authoritative, and Persuasive.We have implemented a transformer-based approach using ai4bharat/IndicBERTv2-MLM-only, MuRIL-base and Telugu-BERT for Telugu communicative style classification. Our system fine-tunes the pretrained Indic language training samples to capture stylistic patterns in Telugu transcripts. Our approach achieved a macro F1 score of 0.2993 on the evaluation set, demonstrating the potential of Indic-focused pretrained models for stylistic analysis in low-resource language settings.Controlled ablations reveal that label smoothing benefits stronger Indic backbones but degrades weaker ones, and that surface linguistic feature augmentation does not complement rich contextual representations on small datasets.</abstract>
<identifier type="citekey">debnath-etal-2026-cuet</identifier>
<location>
<url>https://aclanthology.org/2026.dravidianlangtech-1.25/</url>
</location>
<part>
<date>2026-07</date>
<extent unit="page">
<start>191</start>
<end>195</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Cuet Yet Another Baseline@DravidianLangTech 2026: Shared Task on Prompt Recovery for LLM in Telugu
%A Debnath, Rotna Dipika
%A Ruhi, Shahrin Afroz Hoque
%A Labiba, Ayesha
%A Mallik, Arpita
%A Murad, Hasan
%Y Chakravarthi, Bharathi Raja
%Y Priyadharshini, Ruba
%Y Madasamy, Anand Kumar
%Y Thavareesan, Sajeetha
%Y Rajiakodi, Saranya
%Y Navaneethakrishnan, Subalalitha
%Y Chinnappa, Dhivya
%Y Palani, Balasubramanian
%Y Subramanian, Malliga
%Y Shanmugavadivel, Kogilavani
%Y Rajalakshmi, Ratnavel
%S Proceedings of the Sixth Workshop on Speech, Vision, and Language Technologies for Dravidian Languages
%D 2026
%8 July
%I Association for Computational Linguistics
%C Underline (Virtual)
%@ 979-8-89176-401-9
%F debnath-etal-2026-cuet
%X Prompt recovery in large language models (LLMs) is the task of inferring the communicative intent and stylistic framing of the original instruction from model-generated output. This task is especially challenging for low-resource Dravidian languages such as Telugu, where agglutinative morphology, register variation, and scarce annotated data complicate stylistic modelling. In this paper, we present our system for the Shared Task on Prompt Recovery for LLM in Telugu at DravidianLangTech @ ACL 2026, which aims to classify Telugu transcript excerpts into nine communicative style categories: Formal, Informal, Optimistic, Pessimistic, Humorous, Serious, Inspiring, Authoritative, and Persuasive.We have implemented a transformer-based approach using ai4bharat/IndicBERTv2-MLM-only, MuRIL-base and Telugu-BERT for Telugu communicative style classification. Our system fine-tunes the pretrained Indic language training samples to capture stylistic patterns in Telugu transcripts. Our approach achieved a macro F1 score of 0.2993 on the evaluation set, demonstrating the potential of Indic-focused pretrained models for stylistic analysis in low-resource language settings.Controlled ablations reveal that label smoothing benefits stronger Indic backbones but degrades weaker ones, and that surface linguistic feature augmentation does not complement rich contextual representations on small datasets.
%U https://aclanthology.org/2026.dravidianlangtech-1.25/
%P 191-195
Markdown (Informal)
[Cuet Yet Another Baseline@DravidianLangTech 2026: Shared Task on Prompt Recovery for LLM in Telugu](https://aclanthology.org/2026.dravidianlangtech-1.25/) (Debnath et al., DravidianLangTech 2026)
ACL