Machine Translation on IWSLT En-Vi 2015 (test)
33.27BLEUBPE-dropout
Evaluation results
| Method | Links | |
|---|---|---|
| BPE-dropoutSegmentation=BPE-dropout2019.10 | 33.27 | |
| BPE-dropoutCase sensitivity=cased2019.10 | 32.7 | |
| Kudo (2018)Segmentation=Unigram Language Model2019.10 | 32.43 | |
| COKDcheckpoint_averaging=true, beam_size=52022.03 | 31.95 | |
| BPESegmentation=BPE2019.10 | 31.78 | |
| BPECase sensitivity=cased2019.10 | 31.44 | |
| MUSEModel size=base2019.11 | 31.3 | |
| MUSE-simpleModel size=base2019.11 | 30.7 | |
| NP2MT2019.11 | 30.6 | |
| Mutualcheckpoint_averaging=true2022.03 | 29.83 | |
| Seq-KDcheckpoint_averaging=true, level=sequence-level2022.03 | 29.69 | |
| Word-KDcheckpoint_averaging=true, level=word-level2022.03 | 29.54 | |
| ODCreported_in=Wei et al. (2019)2022.03 | 29.47 | |
| SACT2019.11 | 29.1 | |
| Transformercheckpoint_averaging=true2022.03 | 29.08 | |
| Transformerreported_in=Wei et al. (2019)2022.03 | 28.56 | |
| NBMT2019.11 | 28.1 |