@inproceedings{shrestha-srinivasan-2025-llm,
title = "{LLM} Bias Detection and Mitigation through the Lens of Desired Distributions",
author = "Shrestha, Ingroj and
Srinivasan, Padmini",
editor = "Christodoulopoulos, Christos and
Chakraborty, Tanmoy and
Rose, Carolyn and
Peng, Violet",
booktitle = "Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing",
month = nov,
year = "2025",
address = "Suzhou, China",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2025.emnlp-main.76/",
pages = "1464--1480",
ISBN = "979-8-89176-332-6",
abstract = "Although prior work on bias mitigation has focused on promoting social equality and demographic parity, less attention has been given to aligning LLM{'}s outputs to desired distributions. For example, we might want to align a model with real-world distributions to support factual grounding. Thus, we define bias as deviation from a desired distribution, which may be an equal or real-world distribution, depending on application goals. We propose a weighted adaptive loss based fine-tuning method that aligns LLM{'}s gender{--}profession output distribution with the desired distribution, while preserving language modeling capability. Using 3 profession sets{---}male-dominated, female-dominated, and gender-balanced{---}derived from U.S. labor statistics (2024), we assess both our adaptive method for reflecting reality and a non-adaptive variant for equality. Across three masked language models, bias is observed under both distributions. We achieve near-complete mitigation under equality and 30{--}75{\%} reduction under real-world settings. Autoregressive LLMs show no bias under equality but notable bias under real-world settings, with the Llama Instruct models (3.2-3B, 3.1-8B) achieving a 50{--}62{\%} reduction."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="shrestha-srinivasan-2025-llm">
<titleInfo>
<title>LLM Bias Detection and Mitigation through the Lens of Desired Distributions</title>
</titleInfo>
<name type="personal">
<namePart type="given">Ingroj</namePart>
<namePart type="family">Shrestha</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Padmini</namePart>
<namePart type="family">Srinivasan</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2025-11</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing</title>
</titleInfo>
<name type="personal">
<namePart type="given">Christos</namePart>
<namePart type="family">Christodoulopoulos</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Tanmoy</namePart>
<namePart type="family">Chakraborty</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Carolyn</namePart>
<namePart type="family">Rose</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Violet</namePart>
<namePart type="family">Peng</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Suzhou, China</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-332-6</identifier>
</relatedItem>
<abstract>Although prior work on bias mitigation has focused on promoting social equality and demographic parity, less attention has been given to aligning LLM’s outputs to desired distributions. For example, we might want to align a model with real-world distributions to support factual grounding. Thus, we define bias as deviation from a desired distribution, which may be an equal or real-world distribution, depending on application goals. We propose a weighted adaptive loss based fine-tuning method that aligns LLM’s gender–profession output distribution with the desired distribution, while preserving language modeling capability. Using 3 profession sets—male-dominated, female-dominated, and gender-balanced—derived from U.S. labor statistics (2024), we assess both our adaptive method for reflecting reality and a non-adaptive variant for equality. Across three masked language models, bias is observed under both distributions. We achieve near-complete mitigation under equality and 30–75% reduction under real-world settings. Autoregressive LLMs show no bias under equality but notable bias under real-world settings, with the Llama Instruct models (3.2-3B, 3.1-8B) achieving a 50–62% reduction.</abstract>
<identifier type="citekey">shrestha-srinivasan-2025-llm</identifier>
<location>
<url>https://aclanthology.org/2025.emnlp-main.76/</url>
</location>
<part>
<date>2025-11</date>
<extent unit="page">
<start>1464</start>
<end>1480</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T LLM Bias Detection and Mitigation through the Lens of Desired Distributions
%A Shrestha, Ingroj
%A Srinivasan, Padmini
%Y Christodoulopoulos, Christos
%Y Chakraborty, Tanmoy
%Y Rose, Carolyn
%Y Peng, Violet
%S Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing
%D 2025
%8 November
%I Association for Computational Linguistics
%C Suzhou, China
%@ 979-8-89176-332-6
%F shrestha-srinivasan-2025-llm
%X Although prior work on bias mitigation has focused on promoting social equality and demographic parity, less attention has been given to aligning LLM’s outputs to desired distributions. For example, we might want to align a model with real-world distributions to support factual grounding. Thus, we define bias as deviation from a desired distribution, which may be an equal or real-world distribution, depending on application goals. We propose a weighted adaptive loss based fine-tuning method that aligns LLM’s gender–profession output distribution with the desired distribution, while preserving language modeling capability. Using 3 profession sets—male-dominated, female-dominated, and gender-balanced—derived from U.S. labor statistics (2024), we assess both our adaptive method for reflecting reality and a non-adaptive variant for equality. Across three masked language models, bias is observed under both distributions. We achieve near-complete mitigation under equality and 30–75% reduction under real-world settings. Autoregressive LLMs show no bias under equality but notable bias under real-world settings, with the Llama Instruct models (3.2-3B, 3.1-8B) achieving a 50–62% reduction.
%U https://aclanthology.org/2025.emnlp-main.76/
%P 1464-1480
Markdown (Informal)
[LLM Bias Detection and Mitigation through the Lens of Desired Distributions](https://aclanthology.org/2025.emnlp-main.76/) (Shrestha & Srinivasan, EMNLP 2025)
ACL