Efficient Architectures For Low-Resource Machine Translation
The result's identifiers
Result code in IS VaVaI
<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F00216224%3A14330%2F25%3A00143189" target="_blank" >RIV/00216224:14330/25:00143189 - isvavai.cz</a>
Result on the web
<a href="https://acl-bg.org/proceedings/2025/LowResNLP%202025/pdf/2025.lowresnlp-1.6.pdf" target="_blank" >https://acl-bg.org/proceedings/2025/LowResNLP%202025/pdf/2025.lowresnlp-1.6.pdf</a>
DOI - Digital Object Identifier
—
Alternative languages
Result language
angličtina
Original language name
Efficient Architectures For Low-Resource Machine Translation
Original language description
Low-resource Neural Machine Translation is highly sensitive to hyperparameters and needs careful tuning to achieve the best results with small amounts of training data. We focus on exploring the impact of changes in the Transformer architecture on downstream translation quality, and propose a metric to score the computational efficiency of such changes. By experimenting on English-Akkadian, German-Lower Sorbian, English-Italian, and English-Manipuri, we confirm previous finding in low-resource machine translation optimization, and show that smaller and more parameter-efficient models can achieve the same translation quality of larger and unwieldy ones at a fraction of the computational cost. Optimized models have around 95% less parameters, while dropping only up to 14.8% ChrF. We compile a list of optimal ranges for each hyperparameter.
Czech name
—
Czech description
—
Classification
Type
D - Article in proceedings
CEP classification
—
OECD FORD branch
10200 - Computer and information sciences
Result continuities
Project
<a href="/en/project/LM2023062" target="_blank" >LM2023062: Digital Research Infrastructure for Language Technologies, Arts and Humanities</a><br>
Continuities
P - Projekt vyzkumu a vyvoje financovany z verejnych zdroju (s odkazem do CEP)
Others
Publication year
2025
Confidentiality
S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů
Data specific for result type
Article name in the collection
Proceedings of the First Workshop on Advancing NLP for Low-Resource Languages associated with the International Conference RANLP 2025
ISBN
9789544521004
ISSN
—
e-ISSN
—
Number of pages
26
Pages from-to
39-64
Publisher name
INCOMA Ltd.
Place of publication
Shoumen, BULGARIA
Event location
Varna, Bulgaria
Event date
Sep 13, 2025
Type of event by nationality
WRD - Celosvětová akce
UT code for WoS article
—