SMT + iterative backtranslation (unsupervised)
Reported on 6 benchmarks across 1 task · 1 paper · 1 SOTA
Note: results are matched by exact model name. Different papers may use the same name for different model variants.
Natural Language Processing6 results
- BLEU score· 2018-09-04SOTA25.87best: 37.9 (FLAN 137B (few-shot, k=9))
- BLEU score· 2018-09-0418.23best: 40.68 (MADL)
- BLEU score· 2018-09-0423.05best: 40.7 (FLAN 137B (few-shot, k=11))
- BLEU score· 2018-09-0417.43best: 35.15 (Bi-SimCut)
- BLEU score· 2018-09-0414.08best: 35.14 (Transformer Cycle (Rev))
- BLEU score· 2018-09-0426.22best: 46.4 (Transformer+BT (ADMIN init))