Matches in SemOpenAlex for { <https://semopenalex.org/work/W3200218906> ?p ?o ?g. }
- W3200218906 abstract "Reproducible benchmarks are crucial in driving progress of machine translation research. However, existing machine translation benchmarks have been mostly limited to high-resource or well-represented languages. Despite an increasing interest in low-resource machine translation, there are no standardized reproducible benchmarks for many African languages, many of which are used by millions of speakers but have less digitized textual data. To tackle these challenges, we propose AfroMT, a standardized, clean, and reproducible machine translation benchmark for eight widely spoken African languages. We also develop a suite of analysis tools for system diagnosis taking into account the unique properties of these languages. Furthermore, we explore the newly considered case of low-resource focused pretraining and develop two novel data augmentation-based strategies, leveraging word-level alignment information and pseudo-monolingual data for pretraining multilingual sequence-to-sequence models. We demonstrate significant improvements when pretraining on 11 languages, with gains of up to 2 BLEU points over strong baselines. We also show gains of up to 12 BLEU points over cross-lingual transfer baselines in data-constrained scenarios. All code and pretrained models will be released as further steps towards larger reproducible benchmarks for African languages." @default.
- W3200218906 created "2021-09-27" @default.
- W3200218906 creator A5068811427 @default.
- W3200218906 creator A5073571802 @default.
- W3200218906 creator A5074059447 @default.
- W3200218906 creator A5085263462 @default.
- W3200218906 date "2021-01-01" @default.
- W3200218906 modified "2023-09-25" @default.
- W3200218906 title "AfroMT: Pretraining Strategies and Reproducible Benchmarks for Translation of 8 African Languages" @default.
- W3200218906 cites W2101105183 @default.
- W3200218906 cites W2130942839 @default.
- W3200218906 cites W2148708890 @default.
- W3200218906 cites W2250342921 @default.
- W3200218906 cites W2251654371 @default.
- W3200218906 cites W2419539795 @default.
- W3200218906 cites W2538358357 @default.
- W3200218906 cites W2760656271 @default.
- W3200218906 cites W2887920589 @default.
- W3200218906 cites W2933138175 @default.
- W3200218906 cites W2944815030 @default.
- W3200218906 cites W2949303037 @default.
- W3200218906 cites W2958953787 @default.
- W3200218906 cites W2963088995 @default.
- W3200218906 cites W2963250244 @default.
- W3200218906 cites W2963341956 @default.
- W3200218906 cites W2963403868 @default.
- W3200218906 cites W2963506925 @default.
- W3200218906 cites W2963532001 @default.
- W3200218906 cites W2963807318 @default.
- W3200218906 cites W2964121744 @default.
- W3200218906 cites W2964308564 @default.
- W3200218906 cites W2976330679 @default.
- W3200218906 cites W2977458338 @default.
- W3200218906 cites W2982399380 @default.
- W3200218906 cites W3013122861 @default.
- W3200218906 cites W3032816972 @default.
- W3200218906 cites W3034469191 @default.
- W3200218906 cites W3035390927 @default.
- W3200218906 cites W3037332907 @default.
- W3200218906 cites W3090350559 @default.
- W3200218906 cites W3093517588 @default.
- W3200218906 cites W3105378761 @default.
- W3200218906 cites W3107826490 @default.
- W3200218906 cites W3176466730 @default.
- W3200218906 cites W3197536862 @default.
- W3200218906 cites W630532510 @default.
- W3200218906 doi "https://doi.org/10.18653/v1/2021.emnlp-main.99" @default.
- W3200218906 hasPublicationYear "2021" @default.
- W3200218906 type Work @default.
- W3200218906 sameAs 3200218906 @default.
- W3200218906 citedByCount "1" @default.
- W3200218906 countsByYear W32002189062023 @default.
- W3200218906 crossrefType "proceedings-article" @default.
- W3200218906 hasAuthorship W3200218906A5068811427 @default.
- W3200218906 hasAuthorship W3200218906A5073571802 @default.
- W3200218906 hasAuthorship W3200218906A5074059447 @default.
- W3200218906 hasAuthorship W3200218906A5085263462 @default.
- W3200218906 hasBestOaLocation W32002189061 @default.
- W3200218906 hasConcept C104317684 @default.
- W3200218906 hasConcept C105580179 @default.
- W3200218906 hasConcept C13280743 @default.
- W3200218906 hasConcept C138885662 @default.
- W3200218906 hasConcept C149364088 @default.
- W3200218906 hasConcept C154945302 @default.
- W3200218906 hasConcept C166957645 @default.
- W3200218906 hasConcept C177264268 @default.
- W3200218906 hasConcept C185592680 @default.
- W3200218906 hasConcept C185798385 @default.
- W3200218906 hasConcept C199360897 @default.
- W3200218906 hasConcept C203005215 @default.
- W3200218906 hasConcept C204321447 @default.
- W3200218906 hasConcept C205649164 @default.
- W3200218906 hasConcept C206345919 @default.
- W3200218906 hasConcept C2776760102 @default.
- W3200218906 hasConcept C31258907 @default.
- W3200218906 hasConcept C41008148 @default.
- W3200218906 hasConcept C41895202 @default.
- W3200218906 hasConcept C55493867 @default.
- W3200218906 hasConcept C79581498 @default.
- W3200218906 hasConcept C90805587 @default.
- W3200218906 hasConcept C95457728 @default.
- W3200218906 hasConceptScore W3200218906C104317684 @default.
- W3200218906 hasConceptScore W3200218906C105580179 @default.
- W3200218906 hasConceptScore W3200218906C13280743 @default.
- W3200218906 hasConceptScore W3200218906C138885662 @default.
- W3200218906 hasConceptScore W3200218906C149364088 @default.
- W3200218906 hasConceptScore W3200218906C154945302 @default.
- W3200218906 hasConceptScore W3200218906C166957645 @default.
- W3200218906 hasConceptScore W3200218906C177264268 @default.
- W3200218906 hasConceptScore W3200218906C185592680 @default.
- W3200218906 hasConceptScore W3200218906C185798385 @default.
- W3200218906 hasConceptScore W3200218906C199360897 @default.
- W3200218906 hasConceptScore W3200218906C203005215 @default.
- W3200218906 hasConceptScore W3200218906C204321447 @default.
- W3200218906 hasConceptScore W3200218906C205649164 @default.
- W3200218906 hasConceptScore W3200218906C206345919 @default.
- W3200218906 hasConceptScore W3200218906C2776760102 @default.
- W3200218906 hasConceptScore W3200218906C31258907 @default.
- W3200218906 hasConceptScore W3200218906C41008148 @default.
- W3200218906 hasConceptScore W3200218906C41895202 @default.