OLAC Record oai:lindat.mff.cuni.cz:11234/1-5847

Metadata
Title:		Debiasing Algorithm through Model Adaptation
Bibliographic Citation:		http://hdl.handle.net/11234/1-5847
Creator:		Limisiewicz, Tomasz
		Mareček, David
		Musil, Tomáš
Date (W3CDTF):		2025-03-10T10:47:40Z
Date Available:		2025-03-10T10:47:40Z
Description:		Debiasing Algorithm through Model Adaptation (DAMA) is based on guarding stereotypical gender signals and model editing. DAMA is performed on specific modules prone to convey gender bias, as shown by causal tracing. Our novel method effectively reduces gender bias in LLaMA models in three diagnostic tests: generation, coreference (WinoBias), and stereotypical sentence likelihood (StereoSet). The method does not change the model’s architecture, parameter count, or inference cost. We have also shown that the model’s performance in language modeling and a diverse set of downstream tasks is almost unaffected. This package contains both the source codes and English, English-to-Czech, and English-to-German datasets.
Identifier (URI):		http://hdl.handle.net/11234/1-5847
Language:		English
		Czech
		German
Language (ISO639):		eng
		ces
		deu
Publisher:		Charles University, Faculty of Mathematics and Physics, Institute of Formal and Applied Linguistics (UFAL)
Rights:		The MIT License (MIT)
		http://opensource.org/licenses/mit-license.php
Subject:		gender-bias
		large language models
		machine translation
Type:		toolService
Type (DCMI):		Software
OLAC Info
Archive:		LINDAT/CLARIAH-CZ digital library at the Institute of Formal and Applied Linguistics (ÚFAL), Faculty of Mathematics and Physics, Charles University
Description:		http://www.language-archives.org/archive/lindat.mff.cuni.cz
GetRecord:		OAI-PMH request for OLAC format
GetRecord:		Pre-generated XML file
OAI Info
OaiIdentifier:		oai:lindat.mff.cuni.cz:11234/1-5847
DateStamp:		2025-03-10
GetRecord:		OAI-PMH request for simple DC format
Search Info
Citation:		Limisiewicz, Tomasz; Mareček, David; Musil, Tomáš. 2025. Charles University, Faculty of Mathematics and Physics, Institute of Formal and Applied Linguistics (UFAL).
Terms:		area_Europe country_CZ country_DE country_GB dcmi_Software iso639_ces iso639_deu iso639_eng