2000 character limit reached
UCAM Biomedical translation at WMT19: Transfer learning multi-domain ensembles (1906.05786v1)
Published 13 Jun 2019 in cs.CL
Abstract: The 2019 WMT Biomedical translation task involved translating Medline abstracts. We approached this using transfer learning to obtain a series of strong neural models on distinct domains, and combining them into multi-domain ensembles. We further experiment with an adaptive language-model ensemble weighting scheme. Our submission achieved the best submitted results on both directions of English-Spanish.
- Danielle Saunders (16 papers)
- Felix Stahlberg (31 papers)
- Bill Byrne (57 papers)