List your company
ResearchFormerly WizWand

Textual Style Transfer on Russian Detoxification Dataset

0.923STAmBART 3000 ENG
0.516360.621930.72750.83307Jun 5, 2022
Updated 8mo ago

Evaluation results

MethodLinks
mBART 3000 ENGEvaluation Setup=Cross-lingual Setup, Training Language=English, Training Iterations=30002022.060.9230.3950.5520.202
mBART 5000 ENGEvaluation Setup=Cross-lingual Setup, Training Language=English, Training Iterations=50002022.060.90.2990.5910.16
mT5 base ENGEvaluation Setup=Cross-lingual Setup, Training Language=English2022.060.8380.2760.5060.115
condBERTEvaluation Setup=Baselines2022.060.8190.7780.7440.422
mT5 small ENGEvaluation Setup=Cross-lingual Setup, Training Language=English2022.060.8050.2250.430.077
mT5 base*Evaluation Setup=Multilingual Setup, Training Language=All English and Russian data available (not equalized)2022.060.7730.6760.7950.43
mT5 baseEvaluation Setup=Multilingual Setup, Training Language=English and Russian (equalized)2022.060.7720.6760.7950.43
mT5 smallEvaluation Setup=Multilingual Setup, Training Language=English and Russian (equalized)2022.060.7450.7050.7940.428
mBART 5000 (FSMT)Evaluation Setup=Backtranslation Setup, Engine=FSMT2022.060.7370.6330.7310.348
mBART 5000 RUSEvaluation Setup=Cross-lingual Setup, Training Language=Russian, Training Iterations=50002022.060.7240.7460.8270.457
mBART 3000 RUSEvaluation Setup=Cross-lingual Setup, Training Language=Russian, Training Iterations=30002022.060.6990.7780.8580.475
mBART 5000Evaluation Setup=Multilingual Setup, Training Language=English and Russian (equalized), Training Iterations=50002022.060.6850.7780.8410.449
mT5 base RUSEvaluation Setup=Cross-lingual Setup, Training Language=Russian2022.060.6760.7940.8460.454
mBART 5000 (Google)Evaluation Setup=Backtranslation Setup, Engine=Google2022.060.6750.6690.6340.284
mT5 small RUSEvaluation Setup=Cross-lingual Setup, Training Language=Russian2022.060.5590.8220.8170.363
DeleteEvaluation Setup=Baselines2022.060.5320.8750.8340.364