Publication: Mitigating Hallucinations and Off-target Machine Translation with Source-Contrastive and Language-Contrastive Decoding
Mitigating Hallucinations and Off-target Machine Translation with Source-Contrastive and Language-Contrastive Decoding
Date
Date
Date
Citations
Sennrich, R., Vamvas, J., & Mohammadshahi, A. (2024). Mitigating Hallucinations and Off-target Machine Translation with Source-Contrastive and Language-Contrastive Decoding. 21–33. https://aclanthology.org/2024.eacl-short.4
Abstract
Abstract
Abstract
Hallucinations and off-target translation remain unsolved problems in MT, especially for low-resource languages and massively multilingual models. In this paper, we introduce two related methods to mitigate these failure cases with a modified decoding objective, without either requiring retraining or external models. In source-contrastive decoding, we search for a translation that is probable given the correct input, but improbable given a random input segment. In language-contrastive decoding, we search for a translation that is prob
Metrics
Downloads
Views
Additional indexing
Creators (Authors)
- Vamvas, Jannis
- Mohammadshahi, Alireza
Event Title
Event Title
Event Title
Event Location
Event Location
Event Location
Event Country
Event Country
Event Country
Event Start Date
Event Start Date
Event Start Date
Event End Date
Event End Date
Event End Date
Page range/Item number
Page range/Item number
Page range/Item number
Page end
Page end
Page end
Item Type
Item Type
Item Type
Dewey Decimal Classifikation
Dewey Decimal Classifikation
Dewey Decimal Classifikation
Language
Language
Language
Date available
Date available
Date available
OA Status
OA Status
OA Status
Free Access at
Free Access at
Free Access at
Metrics
Downloads
Views
Citations
Sennrich, R., Vamvas, J., & Mohammadshahi, A. (2024). Mitigating Hallucinations and Off-target Machine Translation with Source-Contrastive and Language-Contrastive Decoding. 21–33. https://aclanthology.org/2024.eacl-short.4
