@inproceedings{estienne-2023-unsupervised,
title = "Unsupervised Calibration through Prior Adaptation for Text Classification using Large Language Models",
author = "Estienne, Lautaro",
editor = "Hardalov, Momchil and
Kancheva, Zara and
Velichkov, Boris and
Nikolova-Koleva, Ivelina and
Slavcheva, Milena",
booktitle = "Proceedings of the 8th Student Research Workshop associated with the International Conference Recent Advances in Natural Language Processing",
month = sep,
year = "2023",
address = "Varna, Bulgaria",
publisher = "INCOMA Ltd., Shoumen, Bulgaria",
url = "https://aclanthology.org/2023.ranlp-stud.2",
pages = "13--22",
abstract = "A wide variety of natural language tasks are currently being addressed with large-scale language models (LLMs). These models are usually trained with a very large amount of unsupervised text data and adapted to perform a downstream natural language task using methods like fine-tuning, calibration or in-context learning. In this work, we propose an approach to adapt the prior class distribution to perform text classification tasks without the need for labelled samples and only a few in-domain sample queries. The proposed approach treats the LLM as a black box, adding a stage where the model posteriors are calibrated to the task. Results show that these methods outperform the un-adapted model for different number of training shots in the prompt and a previous approach where calibration is performed without using any adaptation data.",
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="estienne-2023-unsupervised">
<titleInfo>
<title>Unsupervised Calibration through Prior Adaptation for Text Classification using Large Language Models</title>
</titleInfo>
<name type="personal">
<namePart type="given">Lautaro</namePart>
<namePart type="family">Estienne</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2023-09</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 8th Student Research Workshop associated with the International Conference Recent Advances in Natural Language Processing</title>
</titleInfo>
<name type="personal">
<namePart type="given">Momchil</namePart>
<namePart type="family">Hardalov</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zara</namePart>
<namePart type="family">Kancheva</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Boris</namePart>
<namePart type="family">Velichkov</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ivelina</namePart>
<namePart type="family">Nikolova-Koleva</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Milena</namePart>
<namePart type="family">Slavcheva</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>INCOMA Ltd., Shoumen, Bulgaria</publisher>
<place>
<placeTerm type="text">Varna, Bulgaria</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>A wide variety of natural language tasks are currently being addressed with large-scale language models (LLMs). These models are usually trained with a very large amount of unsupervised text data and adapted to perform a downstream natural language task using methods like fine-tuning, calibration or in-context learning. In this work, we propose an approach to adapt the prior class distribution to perform text classification tasks without the need for labelled samples and only a few in-domain sample queries. The proposed approach treats the LLM as a black box, adding a stage where the model posteriors are calibrated to the task. Results show that these methods outperform the un-adapted model for different number of training shots in the prompt and a previous approach where calibration is performed without using any adaptation data.</abstract>
<identifier type="citekey">estienne-2023-unsupervised</identifier>
<location>
<url>https://aclanthology.org/2023.ranlp-stud.2</url>
</location>
<part>
<date>2023-09</date>
<extent unit="page">
<start>13</start>
<end>22</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Unsupervised Calibration through Prior Adaptation for Text Classification using Large Language Models
%A Estienne, Lautaro
%Y Hardalov, Momchil
%Y Kancheva, Zara
%Y Velichkov, Boris
%Y Nikolova-Koleva, Ivelina
%Y Slavcheva, Milena
%S Proceedings of the 8th Student Research Workshop associated with the International Conference Recent Advances in Natural Language Processing
%D 2023
%8 September
%I INCOMA Ltd., Shoumen, Bulgaria
%C Varna, Bulgaria
%F estienne-2023-unsupervised
%X A wide variety of natural language tasks are currently being addressed with large-scale language models (LLMs). These models are usually trained with a very large amount of unsupervised text data and adapted to perform a downstream natural language task using methods like fine-tuning, calibration or in-context learning. In this work, we propose an approach to adapt the prior class distribution to perform text classification tasks without the need for labelled samples and only a few in-domain sample queries. The proposed approach treats the LLM as a black box, adding a stage where the model posteriors are calibrated to the task. Results show that these methods outperform the un-adapted model for different number of training shots in the prompt and a previous approach where calibration is performed without using any adaptation data.
%U https://aclanthology.org/2023.ranlp-stud.2
%P 13-22
Markdown (Informal)
[Unsupervised Calibration through Prior Adaptation for Text Classification using Large Language Models](https://aclanthology.org/2023.ranlp-stud.2) (Estienne, RANLP 2023)
ACL