all AI news
How do languages influence each other? Studying cross-lingual data sharing during LM fine-tuning
May 22, 2024, 4:47 a.m. | Rochelle Choenni, Dan Garrette, Ekaterina Shutova
cs.CL updates on arXiv.org arxiv.org
Abstract: Multilingual large language models (MLLMs) are jointly trained on data from many different languages such that representation of individual languages can benefit from other languages' data. Impressive performance on zero-shot cross-lingual transfer shows that these models are capable of exploiting data from other languages. Yet, it remains unclear to what extent, and under which conditions, languages rely on each other's data. In this study, we use TracIn (Pruthi et al., 2020), a training data attribution …
abstract arxiv benefit cross-lingual cs.cl data data sharing fine-tuning influence language language models languages large language large language models mllms multilingual performance replace representation shows studying transfer type zero-shot
More from arxiv.org / cs.CL updates on arXiv.org
Dodo: Dynamic Contextual Compression for Decoder-only LMs
1 day, 17 hours ago |
arxiv.org
Active Learning for Multilingual Fingerspelling Corpora
1 day, 17 hours ago |
arxiv.org
Jobs in AI, ML, Big Data
Senior Data Engineer
@ Displate | Warsaw
Analyst, Data Analytics
@ T. Rowe Price | Owings Mills, MD - Building 4
Regulatory Data Analyst
@ Federal Reserve System | San Francisco, CA
Sr. Data Analyst
@ Bank of America | Charlotte
Data Analyst- Tech Refresh
@ CACI International Inc | 1J5 WASHINGTON DC (BOLLING AFB)
Senior AML/CFT & Data Analyst
@ Ocorian | Ebène, Mauritius