Kenji-Endo: a BabyLM @EVALITA
- Calogero Jerik Scozzaro,
- Mattero Rinaldi,
- Gianluca Mittone,
- University of Torino,
- Aequa-tech
Research Output:
Conference Article in Proceeding or Book/Report chapter
Article in proceedings
Peer-review
Open access
Publication Information
Output type
Research Output:
Conference Article in Proceeding or Book/Report chapter
Article in proceedings
Peer-reviewOriginal language
Undefined/UnknownPages from-to (Number of pages)
Pages 16-26 (11 pages)Publication milestones
- Published - 2026
Publication status
Published - 2026
Publisher
CEUR Workshop ProceedingsPublication IDs
- ORCID: /0000-0001-9337-7250/work/220441813
- Scopus: 105040222824
Host publication title
Ceur Workshop ProceedingsAbstract
We present Kenji-Endo, a BabyLM pretrained on a dedicated Italian dataset, which participated in four tasks at the 9th edition of EVALITA: DeSegMa, MultiPRIDE, IMPOLS, and FadeIT. Kenji-Endo achieved competitive performance across all tasks, demonstrating that language modeling with limited data and compact model sizes
can represent a viable alternative to Large Language Models
can represent a viable alternative to Large Language Models
Publication metrics
PlumX
Citations
4
Access to documents
Final published version
License:CC BY, opens in new tab
Related Event
Title
9th Evaluation Campaign of Natural Language Processing and Speech Tools for Italian. Final Workshop
Event type
ConferenceDegree of recognition
International eventDate
26/02/2026 - 27/02/2026Location
BariItaly